|Home | About | Journals | Submit | Contact Us | Français|
The neural circuitry that processes natural rewards converges with that engaged by addictive drugs. Because of this common neurocircuitry, drugs of abuse have been able to engage the hedonic mechanisms normally associated with the processing of natural rewards. Ghrelin is an orexigenic peptide that stimulates food intake by activating GHS-R1A receptors in the hypothalamus. However, ghrelin also activates GHS-R1A receptors on extrahypothalamic targets that mediate alcohol reward. The central nucleus of the amygdala (CeA) has a critical role in regulating ethanol consumption and the response to ethanol withdrawal. We previously demonstrated that rat CeA GABAergic transmission is enhanced by acute and chronic ethanol treatment. Here, we used quantitative RT-PCR (qRT-PCR) to detect Ghsr mRNA in the CeA and performed electrophysiological recordings to measure ghrelin effects on GABA transmission in this brain region. Furthermore, we examined whether acute or chronic ethanol treatment would alter these electrophysiological effects. Our qRT-PCR studies show the presence of Ghsr mRNA in the CeA. In naive animals, superfusion of ghrelin increased the amplitude of evoked inhibitory postsynaptic potentials (IPSPs) and the frequency of miniature inhibitory postsynaptic currents (mIPSCs). Coapplication of ethanol further increased the ghrelin-induced enhancement of IPSP amplitude, but to a lesser extent than ethanol alone. When applied alone, ethanol significantly increased IPSP amplitude, but this effect was attenuated by the application of ghrelin. In neurons from chronic ethanol-treated (CET) animals, the magnitude of ghrelin-induced increases in IPSP amplitude was not significantly different from that in naive animals, but the ethanol-induced increase in amplitude was abolished. Superfusion of the GHS-R1A antagonists -Lys3-GHRP-6 and JMV 3002 decreased evoked IPSP and mIPSC frequency, revealing tonic ghrelin activity in the CeA. -Lys3-GHRP-6 and JMV 3002 also blocked ghrelin-induced increases in GABAergic responses. Furthermore, -Lys3-GHRP-6 did not affect ethanol-induced increases in IPSP amplitude. These studies implicate a potential role for the ghrelin system in regulating GABAergic transmission and a complex interaction with ethanol at CeA GABAergic synapses.
The extended amygdala has a critical role in the motivational effects of drugs of abuse and the reinforcing properties of ethanol (Koob et al, 1998; Koob, 2003). The central nucleus of the amygdala (CeA) is a major component of the extended amygdala and is implicated in the behavioral and physiological responses associated with alcohol consumption (Nie et al, 1994; Funk et al, 2006) and anxiety (LeDoux et al, 1988; Walker and Davis, 1997). Behavioral studies have demonstrated that GABAergic transmission in the CeA is involved in the regulation of alcohol intake (Hyytia and Koob, 1995; McBride et al, 1998). Microinjection of GABAA receptor antagonists into the CeA blocks the reinforcing effect of alcohol (Hyytia and Koob, 1995), and injection of GABAA receptor agonists and benzodiazepines into the CeA decreases anxiety (Roberts et al, 1996), supporting the involvement of the CeA GABAergic system in ethanol reinforcement. We have previously reported that ethanol augments GABAergic transmission in CeA neurons via a presynaptic mechanism that enhances GABA release (Roberto et al, 2003, 2010; Nie et al, 2004; Bajo et al, 2008).
Ghrelin is a 28-amino acid orexigenic peptide that can induce hyperphagia (Kojima et al, 1999). This peptide is synthesized mainly in the gut, but it has been reported to be centrally produced in the brain (Cowley et al, 2003), mainly in the hypothalamus. Ghrelin-positive processes have also been reported outside of the hypothalamus, including the bed nucleus of stria terminalis, medial, central, and cortical amygdalae, the thalamic paraventricular nucleus, and the lateral habenula (Cowley et al, 2003). Some laboratories have also argued against the presence of ghrelin in the CNS (Furness et al, 2011).
Ghrelin is an endogenous ligand of the growth hormone secretagogue receptor (GHS-R1A), a G-protein-coupled receptor that binds Gαq and results in increased intracellular calcium when activated. GHS-R1As are expressed widely in the brain and peripheral tissues (eg, stomach, intestine, pancreas, thymus, gonads, thyroid, and heart). In the brain, relative expression of Ghsr mRNA encoding GHS-R1A differs across species (Landgren et al, 2011a). The most consistent brain sites of Ghsr mRNA expression across mice and rats include the arcuate nucleus, the suprachiasmatic nucleus, centrally projecting Edinger–Westphal nucleus (EWcp), area postrema, substantia nigra, and the ventral tegmental area (VTA) (Guan et al, 1997; Landgren et al, 2011a). Although initial characterizations did not report GHS-R1As in the amygdala, a recent study using quantitative reverse transcription-PCR (qRT-PCR) amplified Ghsr mRNA in dissections of amygdala at levels comparable to those in the VTA and higher than those in the nucleus accumbens or cortex of alcohol-experienced rats (Landgren et al, 2011b).
Increasing evidence points to the ghrelin system as having an important role in ethanol intake and reward. Genetic studies have shown an association between the single-nucleotide polymorphism (SNP) rs2232165 of the GHSR gene and heavy alcohol use (Landgren et al, 2008). Elevated plasma ghrelin levels have been reported in alcoholic patients (Kim et al, 2005; Kraus et al, 2005; Wurst et al, 2007), and among alcoholics, there is a strong positive correlation between ghrelin levels and craving (Addolorato et al, 2006; Leggio et al, 2012). However, other investigators found reduced ghrelin levels in alcoholics (Addolorato et al, 2006; Badaoui et al, 2008). These conflicting results have prompted studies investigating which neural substrates are involved in central ghrelin signaling and behavioral responses to ethanol intake. In mice, injection of ghrelin targeting the VTA and laterodorsal tegmental area (LDTg) produced an increase in alcohol consumption on the two-bottle choice test (Jerlhag et al, 2009). This increase in alcohol intake was reversed by administration of GHS-R1A antagonists and was absent in GHSR knockout mice (Jerlhag et al, 2009). Furthermore, injection of GHS-R1A antagonists targeting the VTA and LDTg attenuated both ethanol-induced locomotor stimulation and ethanol-induced dopamine release, measures of ethanol reward (Jerlhag et al, 2009). Recent studies suggest that the EWcp (also known as the perioculomotor urocortin-containing neurons) may be involved in the regulation of ethanol intake. Intraperitoneal injection of a GHS-R1A antagonist produced a reduction in alcohol preference and intake in mice, and this reduction in consumption was accompanied by a decrease of c-Fos immunopositive cells in the EWcp, but not the VTA or the arcuate nucleus (Kaur and Ryabinin, 2010).
The CeA is innervated by ghrelin-containing axon terminals (Cowley et al, 2003) and there is an increase in GHS-R1A gene expression in the amygdala of high-ethanol-preferring AA rats vs ethanol-avoiding ANA rats following alcohol self-administration (Landgren et al, 2011b). However, to our knowledge, there are no data available on the electrophysiological effects of ghrelin in CeA neurons. Here, we hypothesized that ghrelin signaling in the CeA modulates the effects of ethanol on GABAergic transmission and, consequently, may play a role in alcohol reinforcement and excessive drinking. We found that ghrelin acts through the GHS-R1A to increase GABA release. Tonic ghrelin signaling is present in CeA neurons, and the basal ghrelin signaling pathway in the CeA appears to interact with pathways activated by both acute and chronic ethanol exposure.
As previously described (Roberto et al, 2003, 2004), we prepared CeA slices from male Sprague-Dawley rats (150–300g; 7–9 weeks old; n=103) that were anesthetized with halothane in isoflurane (3%) and decapitated. We cut 400-μm transverse slices on a Vibratome Series 3000 (Technical Products International, St Louis, MO), incubated them in an interface configuration for about 30min, and then completely submerged and continuously superfused them (flow rate of 2–4ml/min) with warm (31°C), gassed ACSF of the following composition in mM: NaCl, 130; KCl, 3.5; NaH2PO4, 1.25; MgSO4·7H2O, 1.5; CaCl2, 2.0; NaHCO3, 24; glucose, 10. Drugs were added to the ACSF from stock solutions to obtain known concentrations in the superfusate.
We used the standard ethanol inhalation method of The Scripps Research Institute Alcohol Research Center to induce ethanol dependence (Rogers et al, 1979; Roberto et al, 2010). Briefly, male Sprague-Dawley rats were housed 2–4 per cage with a 0600h to 1800h light cycle and with free access to food and water. Rats were placed into either ethanol vapor chambers (n=20) or air-only chambers (naive controls; n=13). Chronic ethanol-treated (CET) rats were intermittently exposed (14h on, 10h off) to ethanol vapors for 3 to 4 weeks. On experiment days, CET rats were maintained in the ethanol vapor chamber until preparation of the CeA slices (under ethanol-free conditions). We recorded from slices of CET rats 2–8h after cutting the slices, as previously described (Roberto et al, 2004). Naive/control rats were exposed to air 24h/day.
Twice per week, we determined BALs of the CET animals from tail-blood samples. The target range for BALs in dependent rats during vapor exposure was 150–200mg%. Control animals were also routinely sampled to control for possible effects of handling.
We recorded from neurons in the medial subdivision of the CeA with sharp micropipettes filled with 3M KCl (Roberto et al, 2004). We held most neurons near their resting membrane potentials (RMPs). Data were acquired with an Axoclamp-2A preamplifier (Axon Instruments, Foster City, CA) and stored for later analysis using pClamp software (Axon Instruments). We evoked pharmacologically isolated GABAA receptor-mediated inhibitory postsynaptic potentials (IPSPs) by stimulating locally within the CeA through a bipolar stimulating electrode while superfusing the slices with the glutamate receptor blockers 6,7-dinitroquinoxaline-2,3-dione (DNQX; 20μM) and DL-2-amino-5-phosphonovalerate (APV; 30μM), and the GABAB receptor antagonist CGP 55845A (1μM). At the end of recording, we superfused 30μM bicuculline (or 50μM picrotoxin) to confirm the GABAAergic nature of the IPSPs; these antagonists completely blocked the IPSPs. To determine the synaptic response parameters for each cell, we performed an input/output (I/O) protocol (Roberto et al, 2003, 2010) consisting of a range of five current stimulations (50–250μA; 0.125Hz), starting at the threshold current required to elicit an IPSP up to the strength required to elicit the maximum amplitude. We normalized the three middle (omitting threshold and maximal) stimulus intensities of five equal steps as 1–3; these stimulus strengths and the half-maximal stimulus strength were maintained throughout the duration of the experiment. For some histograms (Figures 1c, 3a, 3d, and 5b) and in the time course studies (Figures 3f, 4e, and 4f), we applied four consecutive stimuli (at 30s intervals) at the half-maximal amplitude determined from the I/O relationship. We examined paired-pulse facilitation (PPF) in each neuron using paired stimuli at 50 and 100ms interstimulus intervals (Zucker, 1989; Bonci and Williams, 1997; Roberto et al, 2004). The stimulus strength was adjusted such that the amplitude of the first IPSP was half-maximal. We calculated the PPF ratio as the amplitude of the second IPSP over the amplitude of the first.
All measures were taken before drug superfusion (control), during drug superfusion (10–30min, depending on the drug), and following washout (20–30min), and are expressed as mean±SEM. We used t-test analyses to evaluate differences in individual means between two groups (eg, when the effect of each dose of ghrelin or the GHS-R1A antagonists, -Lys3-GHRP-6 and JMV 3002, were compared with the respective baseline values). We also used within-subject one-way RM ANOVA to compare IPSPs in the same group of cells (naive or dependent rats) after application of different drugs. When appropriate, the Student Newman–Keuls post hoc test was used to assess significance between treatments with p<0.05 considered significant. To assess differences resulting from ethanol exposure (naive × dependent) and drug interaction between groups, we used a two-way RM ANOVA.
We recorded from a subset of CeA neurons using the whole-cell patch-clamp method in the presence of 20μM DNQX, 30μM APV, 1μM CGP 55845A, and 1μM tetrodotoxin (TTX) to isolate action potential-independent, GABAergic miniature inhibitory postsynaptic currents (mIPSCs). Patch pipettes (3–6MΩ) were pulled from borosilicate glass (Warner Instruments) and filled with an internal solution containing (in mM): 145 KCl, 10 HEPES, 2 MgCl2, 0.5 EGTA, 2 Na-ATP; 0.2 Na-GTP, pH 7.2–7.3, osmolarity 275–290mOsm. We visualized CeA neurons in brain slices (300μm) using infrared differential interference contrast (IR-DIC) optics and CCD camera (EXi Aqua, QImaging). A 60 × water immersion objective (Olympus) was used for identifying and approaching CeA neurons. Whole-cell voltage-clamp recordings of mIPSCs were made with a Multiclamp 700B amplifier (Molecular Devices), low-pass filtered at 2–5kHz, digitized (Digidata 1440A; Molecular Devices), and stored on a PC using pClamp 10 software (Axon Instruments). All cells were clamped at −60mV for the duration of the recording. All data are expressed as mean±SEM.
In all whole-cell patch-clamp recordings, series resistance (<10MΩ) was continuously monitored with a 10mV hyperpolarizing pulse and experiments with >20% change in series resistance were not included in final analysis. Frequency, amplitude, and kinetics of mIPSCs were analyzed using semiautomated threshold based mini detection software (Mini Analysis, Synaptosoft, Fort Lee, NJ) and were visually confirmed. To accurately determine the mIPSC amplitude, only mIPSCs that were >5pA were accepted for analysis. We evaluated results for mIPSCs and determined statistical significance using the Kolmogorov–Smirnov, nonparametric, two-sample test, with p<0.05 considered significant.
We purchased APV, DNQX, CGP 55845, ghrelin, and -Lys3-GHRP-6 from Tocris Bioscience (Ellisville, MO), JMV 3002 from Cayman Chemical (Ann Arbor, MI), and TTX and bicuculline from Sigma (St Louis, MO). Ethanol was purchased from Remet (La Mirada, CA).
Using a micro-knife, we isolated CeA, hippocampus, and cortex from brain slices of naive male Sprague-Dawley rats (n=3) prepared as described above. The samples were stored at −80°C until use. Total RNA was extracted using the Arcturus PicoPure RNA Isolation Kit (Applied Biosystems, Carlsbad, CA). First-strand cDNA was synthesized using SABiosciences RT2 First Strand Kit (Qiagen, Frederick, MD). Quantitative RT-PCR was performed on the Mx3000P light cycler (Stratagene, La Jolla, CA) using RT2 Primer Assays for Ghsr and 18S RNA (Qiagen). Because of relatively low animal numbers in the gene expression studies, nonparametric tests were used for statistical analyses. Comparison of mean cycle thresholds for 18S RNA by the Kruskall–Wallis test found no differences between brain regions (H=0.62, p=0.73). CT values for Ghsr were normalized by the equation 2−ΔCT, where ΔCT is the CT for Ghsr subtracted from CT for 18S. The normalized Ghsr values were compared between brain regions using the Kruskall–Wallis test followed by Mann–Whitney U-tests for post hoc analyses. Specificity of the Ghsr amplification was confirmed by automatic analysis of melting curves and subsequent agarose gel electrophoresis showing bands of predicted size. Lack of amplification of genomic DNA was confirmed by running the reactions without adding reverse transcriptase.
Quantitative RT-PCR performed on brain slices amplified Ghsr (the mRNA encoding GHS-R1A) from CeA, hippocampus, and cortex (Figure 1a). Lack of amplification in the no-reverse-transcriptase control indicated that genomic DNA was not amplified. Normalized Ghsr mRNA levels were 0.005±0.003 for hippocampus, 0.00031±0.00012 for CeA, and 0.00011±0.00025 for cortex. Kruskall–Wallis analysis confirmed significant differences between these values (H=7.2, p=0.027). Ghsr mRNA levels were significantly higher in the hippocampus than in CeA and significantly higher in CeA than in cortex (Mann–Whitney tests: U=9.0, p=0.0495 for both comparisons). Because of the presence of Ghsr mRNA in the CeA, we investigated the electrophysiological effects of ghrelin in this region.
We performed experiments 2–8h after preparation of CeA slices from naive rats. Using sharp micropipettes, we recorded intracellularly from neurons in the medial subdivision of the CeA. Using a stimulating electrode, we evoked synaptic GABAergic responses near the recording site. The mean RMP of these CeA neurons was −78.3±0.7mV with input resistance of 139.9±4.6MΩ (n=104).
We tested six different doses (50–1000nM) of ghrelin (Figure 1b). Overall ANOVA demonstrated that 30min of ghrelin superfusion increased the mean amplitude of evoked GABAergic IPSPs in CeA neurons (F(5, 58)=4.06, p<0.01). As shown in Figure 1c, 75, 100, 150, 200, and 1000nM ghrelin increased IPSP amplitude to 121.2±6.2% (p<0.05, n=8), 136.0±4.0% (p<0.01, n=24), 134.1±4.4% (p<0.01, n=11), 129.2±6.6% (p<0.01, n=12), and 128.5±6.1% (p<0.05, n=4), respectively. The lowest concentration of ghrelin tested (50nM) did not affect IPSP amplitude (96.2±9.6% of control; p>0.05, n=5). The higher doses tested significantly and equally increased IPSP amplitude. Although we did not find the effect of these doses to be dose-dependent, our results are consistent with previous studies that found dose-dependence at concentrations <100nM and a plateau effect at concentrations >100nM (Yanagida et al, 2008; Kim et al, 2009). The effect of ghrelin was partially reversible upon 20-min washout (Figure 1c). Because the higher concentrations of ghrelin produced increases in IPSP amplitude similar in magnitude to the 100nM dose, we chose to use the 100nM concentration throughout the rest of the study, unless specified otherwise.
We examined the acute effects of 100nM ghrelin on the basic membrane properties of neurons in the medial subnucleus of the CeA. Analysis of current–voltage (I–V) relationships shows that ghrelin had no effect on membrane potential or conductance (Figure 2a). Furthermore, ghrelin had no significant effect on spike amplitudes (data not shown). In our analyses, we classified neurons as Type A or Type B according to the electrophysiological characteristics described by Schiess et al (1999). Based on the firing pattern and lack of accommodation of action potentials, we found a predominance (85%) of Type A neurons in our recorded neuronal population, which are medium-sized spiny projecting neurons (Schiess et al, 1999). Our cell typing analysis does not reveal any significant difference in the effect of ghrelin on membrane characteristics (Supplementary Figure 1), suggesting that ghrelin may act on all cell types, and the GHS-R1A receptor may be present in all cell types.
Overall ANOVA revealed a significant effect (F(2, 41)=36.52, p<0.05) and post hoc analysis revealed that superfusion of 100nM ghrelin increased evoked GABA-mediated IPSP amplitudes to 136.9±4.3% of control (averaged over all stimulus strengths; p<0.05; n=17) with partial recovery following 20-min washout, at which point IPSP amplitudes returned to 117.9±4.2% of control (Figure 2b). Ghrelin-induced enhancement of IPSPs was measurable 5–8min after the onset of ghrelin superfusion (see, eg, Figure 4e). In six cells, ghrelin had no effect on IPSPs.
Ghrelin could act at a pre- and/or post-synaptic site to enhance IPSP amplitude. To determine whether ghrelin alters GABA release at CeA synapses, we carried out two types of experiments. In the first set of experiments, we examined PPF at 50 and 100ms interstimulus intervals, a phenomenon whereby a secondary synaptic response is influenced by a preceding primary stimulus of equal intensity (Manabe et al, 1993; Andreasen and Hablitz, 1994). Changes in the PPF ratio (second IPSP/first IPSP) are inversely related to transmitter release such that a reduction of the PPF ratio is associated with an increased probability of transmitter release. Overall, t-test analyses of our paired-pulse studies showed highly variable or no effects of ghrelin on PPF of IPSPs in most CeA neurons. During control recording, the 50 and 100ms PPF ratios were 0.92±0.07 (n=26) and 1.02±0.06 (n=27), respectively (Figure 2c). Overall, ghrelin application did not significantly alter the 50 and 100ms PPF ratios (1.06±0.10; p>0.05 and 1.08±0.07; p>0.05, respectively; Figure 2c), suggesting that ghrelin could be acting both pre- and post-synaptically. However, in 10 out of the 27 cells, ghrelin decreased PPF ratios, suggesting a presynaptic effect.
To further pursue the synaptic site of ghrelin action, we studied mIPSCs in 13 CeA neurons using whole-cell patch-clamp recording in the presence of 1μM TTX. Six of the 13 cells did not respond to ghrelin. Generally, a change in mIPSC frequency indicates an altered probability of transmitter release, and a change in mIPSC amplitude reflects alterations in the sensitivity of postsynaptic GABAA receptors (De Koninck and Mody, 1994; Otis et al, 1994). Paired t-test analyses demonstrate that superfusion of ghrelin increased the mean frequency of mIPSCs to 167.4±17.4% of control (p<0.05, n=7), and also increased the mean mIPSC amplitude to 124.0±16.4% of control (p>0.05, n=7; Figure 2d). Although the mean change in mIPSC amplitude did not reach statistical significance, it was significantly increased in three of the seven cells tested. Thus, acute superfusion of 100nM ghrelin increased mIPSC frequency in all seven cells and increased mIPSC amplitude in some cells, suggesting increased presynaptic release of GABA and postsynaptic ghrelin activity as well. Together, the presynaptic and postsynaptic effects may account for the lack of a significant effect on PPF due to masking/blunting effects. After 10-min washout of the drug, mIPSC frequency was 96.8±20.1% of control (n=4) and amplitude was 102.4±9.0% of control (n=4).
To determine whether GHS-R1As regulate baseline evoked and spontaneous GABAergic transmission, and to test their involvement in ghrelin effects in the CeA, we tested -Lys3-GHRP-6, a putative selective GHS-R1A antagonist. Overall ANOVA revealed that superfusion of 50μM -Lys3-GHRP-6 for 30min decreased evoked IPSP amplitudes to 49.2±3.1% of control (F(2, 26)=75.87; p<0.01; n=10; Figure 3a). Washout (20min) only partially reversed this effect (63.9±6.4% of control; n=7). Paired t-test analyses show that application of 50μM -Lys3-GHRP-6 also decreased mIPSC frequency (64.98±9.2% of control; p<0.05; n=5) and amplitude (69.97±1.8% of control; p<0.05; n=5; Figure 3c, left panel), and also decreased mIPSC decay (72.5±4.0% of control; n=5; data not shown). The decrease in mIPSC frequency and amplitude suggest the involvement of both pre- and post-synaptic mechanisms, congruent with the ghrelin effects we observed (Figure 2c). The change in mIPSC decay suggests direct effects at the GHS-R1A or possibly nonspecific effects, perhaps due to the high dose of the antagonist (Depoortere et al, 2006; Benso et al, 2007). Thus, we also tested a lower dose (25μM) of -Lys3-GHRP-6, which reduced IPSP amplitude to 65.8±3.9% of control (F(2, 39)=25.61; p<0.01; n=15; Figure 3a) and only partially recovered after 20-min washout (82.2±6.5% of control; p<0.05; n=10). Similar to ghrelin application (Figures 2c and and4c),4c), superfusion of -Lys3-GHRP-6 did not alter the 50 or 100ms PPF ratio at either dose (Figure 3b), suggesting both pre- and post-synaptic effects. During control recording, the 50 and 100ms PPF ratios were 0.97±0.10 (n=14) and 1.04±0.05 (n=16), respectively. Following -Lys3-GHRP-6 (25μM) application, the PPF ratios were 0.90±0.09 (n=14) and 0.89±0.06 (n=16). In experiments using 50μM -Lys3-GHRP-6, the baseline 50 and 100ms PPF ratios were 1.09±0.09 (n=6) and 1.04±0.15 (n=6), respectively, and with the addition of -Lys3-GHRP-6, the PPF ratios were 1.18±0.13 (n=6) and 1.12±0.14 (n=6). The tonic GHS-R1A activity observed in our -Lys3-GHRP-6 experiments prompted us to test another GHS-R1A antagonist (JMW 3002), to confirm these findings on both baseline evoked and spontaneous GABAergic transmission. Like -Lys3-GHRP-6, superfusion of JMV 3002 (3nM; Salome et al, 2009) for 30min decreased evoked IPSP amplitudes (at half-max stimulus intensity) to 84.2±5.9% of control (p<0.05, n=8; Figure 3d). JMV 3002 had no effect on PPF ratios. During control recording, the 50 and 100ms PPF ratios were 0.79±0.10 (n=7) and 0.91±0.07 (n=7), respectively. During JMV 3002 application, the PPF ratios were 0.78±0.16 (n=7) and 0.82±0.09 (n=7).
To verify that ghrelin increased IPSPs in CeA by activating the GHS-R1A receptor, we treated the slices with 25μM -Lys3-GHRP-6 and then applied 100nM ghrelin in the continued presence of the antagonist. Overall ANOVA revealed a significant effect of -Lys3-GHRP-6 application (F(2, 18)=18.2; p<0.05). -Lys3-GHRP-6 decreased IPSPs to 63.7±6.9% of control (averaged over all stimulus strengths; p<0.05; n=7; Figure 3e), and subsequent addition of ghrelin failed to increase IPSP amplitude; it persisted at 57.8±6.2% of control (n=7; Figure 3d and e). The effect of the antagonist was measurable 12min after the onset of -Lys3-GHRP-6 superfusion and reached a maximum effect after 20min of superfusion (Figure 3f). Following 20-min washout, IPSP amplitudes remained below baseline levels. In five neurons, we tested 100nM ghrelin in the presence of JMV 3002. Ghrelin did not significantly change IPSP amplitude (76.9±10.3% of control; n=5; Figure 3d). The IPSP amplitudes returned to control (100.9±10.4%, n=5) levels after 20min of washout of the two drugs. Like -Lys-GHRP-6, JMV 3002 blocked the ghrelin-induced facilitation of evoked IPSPs. Paired t-test analyses show that application of 3nM JMV 3002 also decreased mIPSC frequency (55.5±10.7% of control; p<0.05; n=7) and amplitude (76.7±5.5% of control; p<0.05; n=7; Figure 3c, right panel), and also decreased mIPSC decay (78.0±5.5% of control; n=7). In one cell, JMV 3002 did not alter mIPSCs. Taken together, these results indicate that ghrelin exerts its effect through GHS-R1As and these receptors may tonically regulate CeA GABAergic activity at both pre- and post-synaptic sites.
In another set of experiments, we tested whether blockade of GHS-R1As with 25μM -Lys3-GHRP-6 would affect ethanol-induced increases in GABAergic response. Overall ANOVA showed a significant effect of drug application (F(2, 9)=12.18; p<0.05). -Lys3-GHRP-6 decreased IPSPs to 57.5±7.9% of control (averaged over all stimulus strengths; p<0.05; n=4; Figure 3g), and the addition of ethanol increased IPSP amplitudes to 102.0±9.7% of control (n=4; Figure 3g). Thus, application of -Lys3-GHRP-6 did not alter ethanol-induced augmentation of IPSPs, ruling out mediation of the ethanol effects by the GHS-R1A.
We have previously shown that ethanol increases GABAergic transmission in the CeA. Given the recent evidence that central ghrelin signaling influences alcohol intake (Jerlhag et al, 2009; Kaur and Ryabinin, 2010), we sought to determine whether activation of CeA GHS-R1As by ghrelin would alter the effects of ethanol on GABAergic transmission. To test the influence of GHS-R1A activation on ethanol-induced augmentation of IPSPs, we first superfused a maximal concentration of ghrelin (100–200nM) onto CeA slices and then added 44mM ethanol (a dose producing maximal effects; Roberto et al, 2003; Bajo et al, 2008) in the continued presence of ghrelin. We pooled the data from different doses of ghrelin because our dose-response experiments demonstrated that there was no change in the magnitude of effect at these different concentrations (Figure 1c). Overall ANOVA demonstrated that superfusion of ghrelin significantly increased IPSP amplitudes to 132.7±4.4% of control (averaged over all stimulus strengths; F(2, 30)=43.67; p<0.05; n=11; Figure 4a) and subsequent addition of ethanol in the presence of ghrelin further increased IPSP amplitudes to 153.9±5.6% of control (p<0.05 compared with ghrelin; n=11). Again, ghrelin had no effect on 50 or 100ms PPF ratio (control: 0.98±0.09 (n=11) and 1.02±0.07 (n=14); ghrelin: 1.06±0.09 (n=11) and 1.05±0.07 (n=14); Figure 4c), but superfusion of ethanol after ghrelin produced a significant decrease in 50 and 100ms PPF ratio (0.83±0.07; p<0.05; n=11 and 0.83±0.07; p<0.05; n=14, respectively; Figure 4c), suggesting enhanced presynaptic release of GABA. The additional increase in IPSP amplitude induced by 44mM ethanol was observable 6–8min after ethanol application (Figure 4e). To assess the effect of concomitant application of ghrelin and ethanol, we had to compare this with the effect of ethanol alone. Thus, we inverted the order of drug superfusion. Superfusion of ethanol significantly increased IPSP amplitude to 130.2±4.2% of control (averaged across all stimulus strengths; F(2, 18)=4.213; p<0.05, n=7; Figure 4b). Ethanol also decreased the 50 and 100ms PPF ratio (control: 0.92±0.10 (n=7) and 1.14±0.11 (n=7); ethanol: 0.74±0.04 and 0.88±0.10 (p<0.05), respectively; Figure 4d). We observed the effects of ethanol within 5–6min of superfusion (Figure 4f). The magnitude of the increase in IPSP amplitude induced by ethanol alone (30%) was comparable to the magnitude of the additional increase in amplitude produced by applying ethanol after ghrelin (20%) (Figure 4a and b). We further explored this aspect by applying ghrelin after the superfusion of ethanol alone. In the presence of ethanol, ghrelin did not further increase the mean IPSP amplitude, but actually diminished the effect of ethanol (113.6±12.0% of control, n=7; Figure 4b). Superfusion of ghrelin following ethanol increased the 50 and 100ms PPF ratio (0.96±0.12 and 1.07±0.09, respectively; Figure 4d), but this effect did not reach statistical significance.
To determine whether chronic ethanol exposure could alter the ghrelin signaling system in the CeA, we applied ghrelin and ethanol to neurons from CET rats. Overall ANOVA showed a significant drug effect (F(2, 18)=13.48; p<0.01). As in naive rats, superfusion of ghrelin (100nM) in CET CeA neurons increased IPSP amplitudes to 128.2±3.8% of control (averaged across all stimulus strengths; p<0.05; n=7; Figure 5a), suggesting that ethanol dependence does not alter basal ghrelin signaling in the CeA. In contrast to naive rats, subsequent application of 44mM ethanol in CET rats did not further increase IPSP amplitude (it remained at 137.1±8.3% of control; n=7; Figure 5a) or affect the 50 or 100ms PPF ratio (control: 1.04±0.10 (n=7) and 1.07±0.09 (n=10); ghrelin: 1.14±0.17 and 1.18±0.13; ghrelin with ethanol: 1.09±0.13 and 1.04±0.11, respectively; Figure 5c). Thus, in ethanol-dependent rats, the ghrelin-induced enhancement of GABA transmission completely preempted the effect of ethanol.
We performed a separate set of experiments in which we superfused 44mM ethanol onto slices from naive and ethanol-dependent rats. Consistent with our previous reports (Roberto et al, 2004, 2010), here we found that the increase in IPSP amplitude in the two groups of animals was similar (naive: 145.9±2.3% of control; p<0.05, n=16; CET: 137.8±7.1% of control; p<0.05, n=10; Figure 5b) and it was associated with a similar decrease in 50 and 100ms PPF ratio in both naive (control: 1.02±0.07 (n=10) and 1.27±0.06 (n=12); ethanol: 0.87±0.06 (p<0.05) and 1.03±0.08 (p<0.05), respectively) and CET rats (control: 1.08±0.11 (n=8) and 1.31±0.09 (n=9); ethanol: 0.85±0.09 (p<0.05) and 0.99±0.09 (p<0.05), respectively; Figure 5d).
Recent studies have implicated ghrelin and the GHS-R1A in alcohol consumption, and there is critical overlap in the neural pathways responsible for food consumption and those involved in excessive alcohol use (Thiele et al, 2003; Landgren et al, 2008; Landgren et al, 2010; Dickson et al, 2011; Leggio et al, 2011). Central ghrelin signaling mediates ethanol intake by activating GHS-R1As in the VTA, LDTg, and EWcp (Jerlhag et al, 2009; Kaur and Ryabinin, 2010). However, the functional role of ghrelin in the CeA, an area critical for alcohol reward and reinforcement, has until now remained unknown. Here, we hypothesized that ghrelin signaling in the CeA modulates the effects of ethanol on GABAergic transmission, and may consequently play a role in alcohol reinforcement and excessive drinking. This is the first study to characterize the electrophysiological effects of ghrelin on GABAergic transmission and its interaction with ethanol in CeA. The results presented here reveal that ghrelin modulates GABAergic transmission in the rat CeA. Like ethanol (Roberto et al, 2003), ghrelin increases IPSP amplitudes in CeA neurons from both naive and ethanol-dependent rats via pre- and post-synaptic mechanisms. When applied subsequent to ghrelin superfusion, ethanol further enhanced IPSP amplitudes in naive rats, but not in ethanol-dependent rats (Figure 5b). However, when applied first, ethanol occluded the ghrelin-induced increase in GABAergic transmission (Figure 4b). Finally, application of GHS-R1A antagonists revealed tonic facilitatory activity of the GHS-R1A on CeA GABAergic transmission. Analysis of mIPSCs confirms both pre- and post-synaptic sites of action. Taken together, these findings suggest an important role for the ghrelin system in regulating GABAergic transmission and shared mechanisms with ethanol in the CeA.
GHS-R1As have not been identified in the amygdala previously (Zigman et al, 2006), although Ghsr mRNA has been recently identified in the amygdalae of alcohol-consuming rats (Landgren et al, 2011b). Our qRT-PCR data revealing the presence of Ghsr mRNA in the CeA combined with our electrophysiological data indicate that functional GHS-R1As are present in the CeA and that these receptors are tonically active. Interestingly, the differences in expression levels are in agreement with the previous study showing higher levels of Ghsr mRNA in hippocampus than in VTA and amygdala, which are in turn higher than in nucleus accumbens and prefrontal cortex (Landgren et al, 2011b).
Blockade of GHS-R1A receptors with -Lys3-GHRP-6 and JMV 3002 had a significant inhibitory effect on both evoked and spontaneous GABAergic activity, suggesting constitutive activation of GHS-R1As or tonic activity of endogenous ghrelin signaling in rat CeA. Pretreatment of CeA neurons with the GHS-R1A antagonists completely blocked the ghrelin-induced facilitation of IPSP amplitudes (Figure 3d), indicating that ghrelin exerts its effect through GHS-R1As. However, GHS-R1A antagonism did not alter the ethanol-induced augmentation of IPSPs (Figure 3g), suggesting that these receptors do not mediate ethanol-induced IPSP enhancement and that ethanol effects in naive rats occur independently of GHS-R1A activation or blockade. In addition, in naive animals, ghrelin increased IPSP amplitude and ethanol produced a further increase, suggesting partially additive effects. Thus, the additive effect of ethanol may indicate that ghrelin and ethanol activate different intracellular mechanisms to increase GABAergic transmission. Ghrelin-induced increase in evoked IPSP amplitudes was not associated with a significant change in PPF ratio, but ethanol applied in the presence of ghrelin further increased the amplitude of evoked IPSPs and also significantly decreased PPF ratio, suggesting presynaptic effects of ethanol on GABA release. Interestingly, when ethanol is applied first, the facilitatory effects of ghrelin are completely abolished. The addition of ghrelin to the bath diminished the ethanol-induced enhancement of IPSP amplitude (Figure 4b and f), suggesting a preemptive effect of ethanol on ghrelin. This ethanol-induced facilitation of IPSPs was associated with a decrease in PPF (indicating increased GABA release) and blocked the ghrelin effects, suggesting that ethanol may have already activated ghrelin-dependent (shared) mechanisms.
These combined results indicate a complex cellular mechanism that underlies ethanol and ghrelin interactions in which the order of drug application plays an important role in these drug effects on GABAergic transmission in CeA. We speculate that ethanol activates intracellular mechanisms different from those activated by ghrelin only when ethanol is applied after ghrelin. These mechanisms may be presynaptic. Overall, ghrelin increases evoked GABA IPSP amplitudes without altering PPF ratios. However, ghrelin consistently increased the frequency and sometimes the amplitude of mIPSCs, suggesting an increase in both GABA release and GABAA receptor activation. In accord with the ghrelin data, GHS-R1A antagonists decrease evoked IPSPs without altering PPF ratios, and they also decrease both the frequencies and amplitudes of mIPSCs, pointing to both pre- and post-synaptic localization of GHS-R1As, and also account for the pre- and post-synaptic effects of ghrelin. We interpret the absence of change in PPF ratios of evoked IPSPs by ghrelin and the GHS-R1A antagonists as the result of this combination of both pre- and post-synaptic effects, in which the postsynaptic effect may blunt or mask the presynaptic effect. However, we cannot rule out the fact that evoked and spontaneous GABAergic transmission may recruit different GABAergic synapses and also possibly different GABA vesicles. Perhaps, spontaneous GABAergic transmission is more sensitive than evoked transmission to the presynaptic action of ghrelin, and hence both ghrelin and its antagonists significantly alter the frequencies of mIPSCs but not the PPF ratio of evoked IPSPs. Ethanol increases GABAergic transmission via both pre- and post-synaptic mechanisms (Roberto et al, 2003), and when ethanol is applied after ghrelin pretreatment, it still increased evoked IPSP amplitude and decreased PPF ratio (Figure 4a and c), suggesting that the presynaptic effects of ethanol on GABA release are still active and may involve different GABA synapses or/and GABA vesicles.
The GHS-R1A is a G-protein-coupled receptor, and its activation results in cAMP signaling activity (Howard et al, 1996; Kojima et al, 1999). Our previous studies showed that the adenylate cyclase pathway is one of the second messenger pathways engaged by ethanol and CRF receptor activation (Cruz et al, 2011; Cruz et al, 2012). We speculate that the adenylate cyclase pathway may represent a common signaling pathway targeted in the same directions by ghrelin and ethanol in regulating GABAergic transmission, thus playing an essential role in their interactions. Ethanol- and ghrelin-dependent adenylate cyclase activation, in turn, could modulate downstream effector molecules such as voltage-sensitive Ca++ channels and synaptotagmin, resulting in an increase in presynaptic GABA release. In CET rats, the ghrelin-induced increase in GABA was similar to that observed in naive animals, but ghrelin blocked the additional ethanol-induced increase typically seen in naive rats (Figure 5b). These results point to two critical considerations: (1) during the development of ethanol dependence, neuroadaptative changes occur in the GABAergic system, including those of presynaptic GABA function and presynaptic signaling engaged by ghrelin; (2) because an occlusive interaction occurred in CeA neurons from CET rats, we speculate that there are shared mechanisms between the two agents, likely at a presynaptic site. Ethanol increased GABA responses and reduced the PPF ratio in both naive (Figure 4b and d) and dependent rats (Figure 5b and d), which is consistent with enhanced GABA release elicited by ethanol (Roberto et al, 2003, 2004; Cruz et al, 2012). Our finding that ghrelin prevents both the ethanol-induced increase in GABA response (Figure 5b) and decrease in PPF ratio (Figure 5c) in CET animals suggests that those shared intracellular mechanisms may be presynaptic.
The fact that we observed no significant difference of ghrelin effect on evoked GABAergic transmission in CET animals (Figure 5b) suggests no neuroadaptation of GHS-R1As in CeA. This is in agreement with the involvement of ghrelin in food consumption, as mechanisms involved in food consumption should not adapt with prolonged experience with natural rewards. Based on our data, we cannot exclude the possibility that GHS-R1A levels did not change after ethanol dependence, and that CeA neurons adapt via other mechanisms, such as changes in signal transduction machinery. Further work is required to clarify possible synaptic mechanisms and signal transduction machinery affected by ethanol and ghrelin, and to determine if this tonic GHS-R1A activation is altered in CET animals or perhaps in different prandial states.
Activation of the ghrelin-GHS-R1A system during the development of alcohol dependence may have motivational significance. In mice, peripheral and central administration of ghrelin reduced the total number of entries and total time spent in the open arms in an elevated plus maze test (Asakawa et al, 2001a). Furthermore, starvation stress increased ghrelin gene expression (Asakawa et al, 2001b) and tail pinch stress significantly increased ghrelin mRNA expression (Asakawa et al, 2001a). These studies suggest that ghrelin induces potent anxiogenic activities, most likely via the CRF system. Ghrelin stimulates the HPA axis via stimulation of CRH release from the hypothalamus. Blockade of CRF receptors has been shown to attenuate ghrelin-induced anxiogenesis in the elevated plus maze test (Asakawa et al, 2001a), and injection of astressin (CRF2 receptor antagonist) attenuated ghrelin-induced plasma corticosterone increase in neonatal chicks (Saito et al, 2005). We have demonstrated that CRF, like ethanol, augments GABAergic transmission, and that ethanol dependence heightens the sensitivity to CRF and CRF1 antagonists in CeA (Roberto et al, 2010). Taken together with the data presented here, these studies suggest that central ghrelin signaling could possibly play a role in behavioral responses to alcohol withdrawal. The ghrelin system may constitute part of a brain pathway modulating reinforcement of alcohol consumption, and uncovering the role of ghrelin/GHS-R1A and its interaction with ethanol in the CeA may provide a therapeutic target for the treatment of alcoholism.
This is manuscript number 20852 from The Scripps Research Institute. We thank Paul Schweitzer and Michal Bajo for their valuable comments. This work was supported by the Pearson Center for Alcoholism and Addiction Research and NIAAA Grants AA015566, AA06420, AA016985, AA017447, AA016647, AA013498, and AA010760.
The authors declare no conflict of interest.
Supplementary Information accompanies the paper on the Neuropsychopharmacology website (http://www.nature.com/npp)