Endocannabinoids Link Feeding State and Auditory Perception-Related Gene Expression

Introduction

Because songbirds learn a form of vocal communication in a manner with similarities to human language acquisition, they have become an important model for studying neurobiology underlying vocal development (Mooney, 1999). We have been using a songbird, the zebra finch, to study cannabinoid effects on vocal learning (Soderstrom and Johnson, 2000, 2001, 2003; Whitney et al., 2003). Cannabinoids are a structurally diverse group of compounds and the active constituents of marijuana. One of these compounds, Δ⁹-tetrahydrocannabinol, is responsible for most of the psychotropic actions of marijuana and acts mostly through cannabinoid receptors of type 1 (CB₁) in the brain. Endocannabinoids, including anandamide and 2-arachidonoylglycerol (2-AG), are naturally occurring lipids capable of cannabinoid receptor activation (Di Marzo et al., 1998). Endocannabinoids produce diverse physiological effects ranging from peripheral modulation of immune function (Salzet et al., 2000; Di Marzo et al., 2004) to persistent changes in memory-related neuronal activity (Gerdeman and Lovinger, 2003). In humans, cannabinoids stimulate appetite and the intake of palatable food (Iversen, 2003), and accumulating evidence supports a relationship between cannabinoid signaling and feeding behavior (Di Marzo et al., 2001, 2004; Hanus et al., 2003; Matias et al., 2003; Matias and Di Marzo, 2004).

We found that cannabinoid treatment of adult male zebra finches, which express a CB₁ receptor ortholog (Soderstrom and Johnson, 2000), decreases the amount that they sing (Soderstrom and Johnson, 2001). A similar inhibition of singing has recently been shown to occur as a function of brief periods of limited food access (Rashotte et al., 2001; Johnson and Rashotte, 2002). These similar effects (singing reductions) produced by dissimilar manipulations (cannabinoid exposure and limited food access) led us to investigate the possibility of a common mechanism. We report that decreased singing produced by limited food is partially reversed by pretreatment with the cannabinoid antagonist SR141716A [N-piperidino-5-(4-chlorophenyl)-1-(2,4-dichlorophenyl)-4-methylpyrazole-3-carboxamide (SR)], implicating endocannabinoid involvement. Endocannabinoid involvement is further demonstrated by the ability of brief periods of limited food access to increase levels of the endocannabinoid 2-AG in the caudal telencephalon.

Zebra finch caudal telencephalon contains the caudal medial nidopallium (NCM), a region implicated in the integration of auditory information (Chew et al., 1996). The NCM is a secondary auditory region that receives projections from the primary region L2 (subfield of L) (Vates et al., 1996). We now report that L2 is distinctly stained by antisera directed against the zebra finch CB₁ cannabinoid receptor. Expression of the transcription factor ZENK within the NCM is stimulated by perception of novel song (Mello et al., 1995; Stripling et al., 1997). This type of immediate-early gene response may function as a persistent integrative signal important for processes related to learning and memory (Clayton, 2000). Novel song-stimulated ZENK expression within the NCM is selectively inhibited by exogenous cannabinoid pretreatment (Whitney et al., 2003). We show here that song-stimulated ZENK expression is also reduced by brief periods of limited food access. These reductions are cannabinoid antagonist reversible, implicating endocannabinoid involvement. Taken together, our results indicate that the feeding state and auditory perception-related ZENK expression within the NCM are linked through endocannabinoid signaling.

Materials and Methods

Except where noted, all materials and reagents were purchased from Sigma (St. Louis, MO) or Fisher Scientific (Pittsburgh, PA). Immunochemicals were purchased from Vector Laboratories (Burlingame, CA). We have used the recently revised system of nomenclature in descriptions of zebra finch neuroanatomy (Reiner, 2004).

Animals. Adult male zebra finches (>90 d old) from domestic stock were used. Before the start of the experiment, birds were housed in flight aviaries with mixed seeds (SunSeed VitaFinch), grit, water, and cuttle-bone available ad libitum. Each flight aviary contained several perches. The light/dark cycle was controlled at 14/10 hr, and the ambient temperature was maintained at 78° F.

Limited food access. To generate tissue for endocannabinoid analysis, eight pairs of adult male zebra finches (16 animals total) were assigned randomly to limited food access (No Food) or ad libitum (Food) groups and housed in isolation in separate rooms with ad libitum access to food and water. One day after isolation, the food cups of No Food group animals were replaced with empty containers immediately before the beginning of the dark phase. Because these animals do not feed in the dark, differences in food access between Food and No Food animals began with the start of the following light phase. Restricted food access in the No Food group continued for a total of 4 hr. Immediately after this treatment period, birds were either anesthetized with Equithesin (Suzuki et al., 1996) and perfused transcardially before immunocytochemistry experiments or rapidly decapitated, with brain tissue dissected for lipid extraction (see below). Note that this brief period of food access, even when implemented for weeks, does not change body mass (Johnson and Rashotte, 2002).

For investigations of singing incidence, four adult males were moved to auditory recording chambers to habituate 7 d before the initiation of experiments and maintained on 14/10 hr light/dark cycles. Food was available ad libitum. After the habituation period, birds were given vehicle injections 30 min before the beginning of light cycles for three consecutive days. Singing incidence was recorded for 4 hr, and mean song bouts per hour were calculated over the 3 d of recording. After a 4 d recovery period, birds were again injected with vehicle for three consecutive days as above, and, in addition, food was removed. Singing was again recorded for 4 hr, food was replaced, and the overall mean song bout incidence was calculated (see Fig. 1, No Food). After a second 4 d recovery period with ad libitum food, birds were injected with 3 mg/kg of the cannabinoid antagonist SR141716A 30 min before lights-on, and food was removed. Birds were recorded as described above, and food was replaced after hour 4 of light cycles. Means over three consecutive days were calculated (see Fig. 1, SR + No Food). After a final 4 d recovery period with ad libitum food, birds were again injected with vehicle and recorded for 4 hr for three consecutive days. These final vehicle injections were done to control for possible persistent effects of limited food and SR treatments. Significant differences were not detected (see Fig. 1, Post-Tx Food).

For ZENK immunocytochemistry experiments, a total of eight groups were used: (1) “No Food, No Song,” a control group in which animals were not provided food for a 4 hr period, as described above, and were not exposed to novel song stimulation (n = 2); (2) “SR, No Food, No Song,” a second control group in which birds were treated as in No Food, No Song, except that they were injected with SR141716A (3 mg/kg, i.m.) 3 hr into the light phase (n = 2); (3) “SR, Food, No Song,” a third control group receiving SR141716A under ad libitum food conditions (n = 3); (4) “Food, No Song,” a fourth control group in which animals received food ad libitum and were not exposed to a novel song (n = 4; as detailed below); (5) “Food, Song,” in which animals received food ad libitum and had activity within the NCM stimulated by exposure to a recording of novel song (n = 5); (6) “No Food, Song,” in which animals were not provided food for a 4 hr period, as described above, and had activity within the NCM stimulated by exposure to a recording of novel song (n = 7); (7) “SR, No Food, Song,” in which animals were treated as in the No Food, Song group, except that they were injected with the CB₁ cannabinoid receptor-selective antagonist SR141716A (3 mg/kg, i.m.) 30 min before song exposure (n = 5); (8) “SR, Food, Song,” in which birds received SR141716A under ad libitum food conditions (n = 3).

Analysis of zebra finch brain endocannabinoid content. Fourteen adult male zebra finches were assigned randomly to control or limited food access groups (n = 7 animals per treatment). Limited food access was administered as described above. Brains were removed, blocked down the midline, and tissue from each half was extracted and analyzed independently resulting in two data points per sample per bird, with a final n = 14 per treatment. Tissue from each brain half was rapidly dissected into rostral [containing the song regions lMAN (lateral magnocellular nucleus of anterior nidopallium) and Area X] and caudal [containing the song regions HVC and RA (robust nucleus of arcopallium) and auditory regions L2 and NCM] telencephalon, cerebellum, and midbrain and immediately frozen in liquid nitrogen. Frozen tissue was stored at -80°C until extracted, as described previously (Matias et al., 2003). Briefly, tissues were homogenized in and extracted with 50 mm chloroform: methanol:Tris-HCl, pH 7.5 [2:1:1 (v/v)] containing internal standards (10 pmol of [²H]₈ anandamide, [²H]₄ N-palmitoylethanolamine, and 100 pmol of [²H]₅-2-arachidonyl glycerol; obtained from Cayman Chemical, Ann Arbor, MI). The lipid-containing organic phase was collected and dried. The mass of dried extract was determined and prepurified by open-bed chromatography on silica gel. Fractions eluted with 9:1 chloroform:methanol were analyzed by liquid chromatography-atmospheric pressure chemical ionization-mass spectrometry (LC-APCI-MS) using a Shimadzu HPLC apparatus (LC-APCI-MS) coupled with a LCMS-2010 quadruple MS via an APCI interface (both from Shimadzu). MS analyses were performed in the selected ion monitoring, as described previously (Di Marzo et al., 2000). The temperature of the APCI source was 400°C, and the HPLC column was a Phenomenex (5 mm, 150 × 4.5 mm) reverse-phase column, eluted as described previously (Di Marzo et al., 2000). Anandamide (retention time, 13.5 min), N-palmitoylethanolamine (retention time, 17.5 min), and 2-AG (retention time, 15.0 min) quasi-molecular ions (m/z = 348.0, 300.0, and 379.3) were quantified by isotope dilution with the above-mentioned deuterated standards (same retention times and m/z = 356.0, 304.0, and 384.3), and their amounts in picomoles or nanomoles were normalized per milligram of lipid extract. Two LC-MS peaks for both deuterated and undeuterated mono-arachidonoylglycerol were found at retention times of 15.0 and 16.9 min, respectively, corresponding to 2-AG and 1(3)-AG, in agreement with the previous observation that 2-AG undergoes isomerization during the purification procedure (Di Marzo et al., 1999). Therefore, the amounts of 2-AG were calculated by adding the amounts of the two isomers.

Silica column fractions eluted with 1:1 chloroform:methanol, which contain the N-acylphosphatidylethanolamines, were processed with Streptomyces chromofuscus phospholipase D to yield the corresponding N-acylethanolamines in a quantitative manner, as described previously (Di Marzo et al., 2001). Therefore, the anandamide biosynthetic precursor N-arachidonoyl-phosphatidylethanolamine was quantified by quantifying, by means of LC-APCI-MS, the anandamide produced from its hydrolysis by the phospholipase D.

Induction of ZENK expression within NCM. Increased expression of the immediate-early gene ZENK within the NCM was produced using an established protocol (Mello and Ribeiro, 1998). Three hours, 30 min into the light phase, a digital recording of a novel zebra finch song consisting of three bouts of ∼5 sec each, followed by 45 sec of silence, was played 30 times. This procedure results in increased ZENK expression selectively within the NCM. Note that novel song-stimulated ZENK expression is inhibited by pretreatment with the cannabinoid agonist WIN55212-2 (R(+)[2,3-dihydro-5-methyl-3-([morpholinyl]methyl)pyrrolo(1,2,3-de)-1,4-benzoxazin-yl]-(1-naphthalenyl)methanone) in an antagonist-reversible manner, clearly demonstrating CB₁ receptor involvement (Whitney et al., 2003). Thirty minutes after the end of song stimulation, birds were overdosed with Equithesin and perfused transcardially for preparation of brain tissue for ZENK immunohistochemistry experiments described below.

Cannabinoid antagonist administration. The CB₁ receptor-selective antagonist SR141716A (Pertwee, 1999) was obtained from the Research Triangle Institute (Raleigh, NC) through the National Institute on Drug Abuse program. Akamuls was a gift from Rhodia (Cranberry, NJ). Stocks of SR141716A (20 mm) were prepared in DMSO and stored at -80°C until diluted for injection. Dilutions were made to produce a suspension of DMSO:Alkamuls:PBS (1:1:18). The concentration of SR141716A was adjusted to deliver 3 mg/kg in a volume of 50 μl. Treatments were given 30 min before initiation of song exposure (see above) by intramuscular injection into the large pectoralis flight muscle.

Anti-zebra finch CB₁ cannabinoid receptor immunocytochemistry. Previous molecular cloning of cDNA encoding the zebra finch CB₁ cannabinoid receptor (Soderstrom and Johnson, 2001) has allowed us to produce a species-specific polyclonal antibody with Spring Valley Laboratories (Woodbine, MD). We have targeted a 16-amino acid region within the intracellular tail portion of the zebra finch CB₁ cannabinoid receptor: HAFRSMFPTSEGTAQP-C-OH. This sequence encompasses His 408 to Pro 423. Cys 417 was changed to a Ser residue (Ser shares the same physicochemical properties of Cys, and replacement of Cys minimizes complications attributable to oxidation). The predicted epitope is FRSMFP. The antisera were produced in rabbit. The immunochemical specificity of these antisera is demonstrated by the absence of staining after preincubation with 20 μm of the immunizing peptide (see Fig. 4, compare A, B). Western blotting indicates staining of bands with molecular weights consistent with isoforms of zebra finch CB₁ (see Fig. 3). Masses of these isoforms are similar to those reported by Egertova and Elphick (2000) and Wager-Miller et al. (2002) and may represent products of alternate splicing, glycosylation, and receptor dimerization. For immunohistochemistry experiments, 30 μm sections of adult male zebra finch medial telencephalon were reacted with affinity-purified anti-zebra finch CB₁ antibody (1:3000), as described below for ZENK, except that an HRP-conjugated anti-rabbit antibody was used in place of avidin-biotin-peroxidase complex.

ZENK immunocytochemistry. Polyclonal anti-Egr-1 antibody was purchased from Santa Cruz Biotechnology (catalog #C-19; Santa Cruz, CA). The specificity of the interaction of this reagent with zebra finch ZENK has been well established (Mello and Ribeiro, 1998). Other reagents were purchased from Vector Laboratories. As indicated above, birds were killed by Equithesin overdose and perfused transcardially with phosphate-buffered 4% paraformaldehyde, pH 7.0. After brains were removed and immersed overnight in 4% paraformaldehyde, the left telencephalic hemisphere was blocked at its lateral edge and sectioned parasagitally (medial to lateral) at 30 μm on a vibrating microtome. ZENK immuocytochemistry was performed using a standard protocol reported by Whitney et al. (2000). Briefly, tissue sections were rinsed in 0.1% H₂O₂ for 30 min, blocked with 5% goat serum for 30 min, and incubated overnight in blocking solution containing anti-Egr-1 antibody (1:1000). After antibody exposure, sections were rinsed in saline, incubated in blocking solution containing biotinylated anti-rabbit solution (1:500) for 1 hr, rinsed with PBS again, and then submerged in avidin-biotin-peroxidase complex solution for 1 hr. Antibody labeling was visualized with DAB solution. Control sections that were not incubated in primary antibody were not immunoreactive.

Data analysis and statistics. Data analyses and statistical procedures were performed using Prism GraphPad, SigmaStat, and Microsoft Excel personal computer software. All quantification procedures on individual birds were completed without knowledge of the experimental group. The density of ZENK-immunolabeled cells was measured using previously established procedures (Whitney et al., 2000). Immunoreactive nuclei within the NCM was the unit of count within a high-power grid defined by 60× objective (volume, 0.0009 mm³). As described by Mello and Ribeiro (1998), an inclusive counting criterion was used (all lightly and darkly stained cells were counted). For each bird, three randomly selected high-power fields were counted per NCM in four different tissue sections. High-power fields were counted independently by two persons blind to treatment. These duplicate counts were pooled for analysis. The density of ZENK-labeled cells was defined as the total number of immunoreactive cells counted divided by the volume of the 12 high-power fields. Parasagittal sections of the caudal telencephalon ∼0.15 mm from midline were used to quantify NCM ZENK induction. Previous studies show that in this region robust uniform levels of song-induced ZENK expression occurs within clearly defined boundaries of the NCM (Mello and Clayton, 1994; Mello and Ribeiro, 1998; Whitney et al., 2003). ZENK expression data are reported as mean labeled cells per cubed millimeter of NCM ± SEM. Differences between treatment and control groups were evaluated by one-way ANOVA. Between-group differences in the number of ZENK-expressing cells in the NCM were assessed with the Student-Newman-Keuls (SNK) test. Differences in anandamide and 2-AG content across groups and brain regions were assessed with two-factor ANOVA, followed by SNK post-tests.

Results

Discussion

Our results demonstrate that brief periods of limited food availability increase the levels of the endocannabinoid 2-AG in zebra finch brain. At the same time, limited food availability decreases song-induced expression of the immediate-early gene ZENK within the NCM, a region important for integration of auditory information (Mello and Ribeiro, 1998). Because food-related decreases in NCM activity were reversed by pretreatment with the cannabinoid antagonist SR141716A, we suggest that this effect is attributable to increased endocannabinoid signaling via CB₁ receptors. This suggestion was confirmed by the finding of elevated 2-AG, but not anandamide, levels in the caudal telencephalon, which contains the NCM. It is not unusual for the two major endocannabinoids to be differentially regulated during physiological and pathological conditions (for examples, see Giuffrida et al., 1999; Di Marzo et al., 2000; Marsicano et al., 2003; Valenti et al., 2004). This phenomenon is attributable to distinct enzymes underlying biosynthetic and metabolic processes for anandamide and 2-AG (Di Marzo et al., 1998), and these enzymes are likely subject to differential regulation by physiological stimuli.

Increased 2-AG levels were also found in two other brain regions of zebra finches subjected to limited food access, the rostral telecephalon and the midbrain, but not the cerebellum, and are consistent with elevated 2-AG levels in the limbic forebrain and hypothalamus, but not the cerebellum, of fasting rats (Kirkham et al., 2002). It is becoming clear that cannabinoid signaling is an important mediator of ingestive behavior (Berger et al., 2001; Berry and Mechoulam, 2002). For example, CB₁ receptor-deficient mice are leaner than controls (Cota et al., 2003). In humans, CB₁ cannabinoid receptor activation increases appetite, and the CB₁-selective antagonist SR141716A exhibits hypophagic efficacy (Iversen, 2003). A part of endocannabinoid effects on ingestive behavior appears to be under the control of the leptin signaling system, which reduces endocannabinoid levels in mouse hypothalamus and is associated with elevated endocannabinoid levels in hypothalami of genetically obese, leptin-deficient rodents (Di Marzo et al., 2001). The endocannabinoid system is also involved in reinforcing the incentive value of food, as demonstrated by the finding that, under certain conditions, SR141716A inhibits the intake of palatable food selectively versus normal food (Higgs et al., 2003). In this case, endocannabinoids seem to act mostly in the limbic forebrain and more particularly in the nucleus accumbens, possibly to stimulate dopamine release from terminals of neurons originating in the ventral tegmental area. This is supported by demonstration that infusion of 2-AG directly into the nucleus accumbens shell dose dependently stimulates feeding (Kirkham et al., 2002).

Elevated 2-AG levels after short periods without food are now well documented in rat and mouse brain (Kirkham et al., 2002; Hanus et al., 2003). Interestingly, longer, perhaps starvation-relevant, periods of dietary restriction produce significant decreases in 2-AG levels (Hanus et al., 2003). This may reflect a beneficial adaptation to extended periods of food shortage, mitigating behavior-altering, and as our results indicate, perception-altering endocannabinoid effects that may distract from other important activities during difficult times. It will be important to determine whether similarly extended periods of food restriction will reduce brain 2-AG in our avian model. Our current results suggest that such reductions should augment auditory perception-related ZENK responsiveness in the NCM.

Insight into brain regions involved in endocannabinoid regulation of NCM activity has been gained through development of a species-specific anti-zebra finch CB₁ antibody. We were generously provided anti-rat CB₁ receptor antibodies by Dr. Maurice Elphick (Queen Mary College of London, London, UK) and Dr. Ken Mackie (University of Washington, Seattle, WA) but could not demonstrate specific interaction of these reagents with the avian receptor, likely because of a combination of reduced affinity for zebra finch CB₁ and increased nonspecific interactions with avian proteins (Fig. 3). Specificity of the new anti-zebra finch CB₁ antibody is supported by the following: (1) elimination of staining after preincubation with 20 μm of the immunizing peptide (Fig. 4, compare A, B); (2) Western blotting results demonstrating interaction with isoforms consistent in mass with those labeled by well established anti-rat reagents (Fig. 3) and; (3) consistency of the pattern of antibody labeling with that observed previously with cDNA probes directed to CB₁ mRNA within the caudal telencephalon (Soderstrom and Johnson, 2000). The distinct neuropil staining and negative staining of large neurons within distinct regions of the telencephalon obtained with the anti-zebra finch CB₁ antibody is consistent with the staining patterns obtained in rat brain using antibodies directed against both N-terminal (Tsou et al., 1998) and C-terminal (Egertova and Elphick, 2000) regions of the rat CB₁ receptor. Distinct staining of somata was not observed in rat brain with the C-terminal-directed antibody, and therefore our reagent (which is directed against a more proximal region of the C-terminal tail domain of CB₁) shares more similarity in staining pattern with the N-terminal-directed rat reagent. These similarities further extend to intense interaction with proteins of ∼63 kDa on Western blots (Fig. 3) (Tsou et al., 1998, compare their Fig. 1A), suggesting that this isoform may be that associated with somal expression.

Availability of this anti-zebra finch CB₁ antibody will facilitate determination of the pattern of cannabinoid receptor expression within avian CNS and during the distinct stages of late postnatal development when zebra finch vocal learning occurs. These experiments are currently underway. Immunochemical results already obtained support distinct, high-level CB₁ expression in brain regions important for vocal learning (Area X and lMAN) and production (HVC and RA) (Fig. 4). These results confirm and extend our previous finding of distinct, high-level CB₁ mRNA expression within HVC and RA (Soderstrom and Johnson, 2000) and provide insight into brain regions likely to be involved in cannabinoid-altered vocal learning (Soderstrom and Johnson, 2003).

The dense anti-CB₁ labeling within L2 suggests a possible location for food-related changes in endocannabinoid signaling (Fig. 2). L2 is the primary recipient of thalamic auditory input and is characterized electrophysiologically by simple tone responses (Bonke et al., 1979). L2 projects to L1, L3, and the NCM, and these secondary auditory regions are characterized by increased response selectivity consistent with integration of auditory information (Muller and Scheich, 1987). Within L1 and L3, neurons are more likely to depolarize in response to complex sounds than to simple tones (Muller and Scheich, 1988). In the case of zebra finch NCM, activity evidenced by ZENK expression is selectively elicited by exposure to novel song patterns indicating a high degree of integration (Mello et al., 1995). Because complex sound selectivity in other secondary auditory regions (L1 and L3) is dependent on GABAergic signaling at the level of L2 (Muller and Scheich, 1987), it is possible that similar integrative responsiveness within NCM is also dependent on L2 GABAergic activity. The high density of CB₁ receptors within L2 raises the possibility that their activation may disrupt this signaling through mechanisms similar to CB₁ inhibition of GABA release in rat hippocampus and the nucleus accumbens (Hoffman and Lupica, 2000, 2001). Additional electrophysiological experiments will be required to understand the circuitry involved in the effects that we have measured. The well studied, tonotopic anatomy of the avian auditory telencephalon will facilitate these experiments.

Our results extend previous work demonstrating that exogenously administered cannabinoids reduce song perception-related ZENK expression (Whitney et al., 2003) and demonstrate for the first time endocannabinoid mediation of processes related to integration of auditory information. These results also indicate a relationship between auditory perception and feeding state.

The ZENK expression that we have studied has been characterized as part of a “genomic action potential,” wherein integrative neurons receive suprathreshold activation and respond through immediate-early gene expression rather than, or in addition to, depolarization (Clayton, 2000). Products of immediate-early genes such as ZENK are transcription factors capable of altering patterns of gene expression. Such changes in gene expression patterns are responsible for persistent physiological changes, making them candidates for involvement in processes related to learning and memory. For example, ZENK expression is enhanced with production of long-term potentiation of synaptic activity in rat hippocampus (Cole et al., 1989). In our zebra finch system, because ZENK expression within the NCM only occurs in response to unfamiliar song patterns, memory retrieval processes must be involved. Interference with memory is a well characterized effect of cannabinoids in mammalian species (Iversen, 2003). Together, these facts support the possibility that impaired memory retrieval plays a role in the endocannabinoid-associated decreased NCM ZENK responsiveness that we have measured. Use of a songbird model, based on an animal that produces and responds to a learned form of vocal communication, should prove uniquely useful in studying cannabinoid effects on learning and memory, especially within the context of late postnatal CNS development when song patterns are memorized and practiced (Soderstrom and Johnson, 2003).