Abstract
Increasing data indicate that brain endocannabinoid system plays a role in the effects of antidepressant medications. Here we examined the effect of in vivo exposure to the selective serotonin uptake inhibitor fluoxetine on cannabinoid type 1 (CB1) receptor density and functionality in the rat prefrontal cortex (PFC) and cerebellum. Long-term treatment with fluoxetine (10 mg/kg/day) enhanced CB1 receptor inhibition of adenylyl cyclase (AC) in the PFC and reduced it in the cerebellum without altering receptor density and agonist stimulation of guanosine 5′-O-(3-[35S]thio) triphosphate ([35S]GTPγS) in either area. Analysis of [35S]GTPγS-labeled Gα subunits allowed for the detection of up-regulated CB1 receptor coupling to Gαi2, Gαi3 in the PFC, and reduced coupling to Gαi3 in the cerebellum of fluoxetine-treated rats. Concomitant administration of the 5-HT1A receptor antagonist N-[2-[4- (2-methoxyphenyl)-1-piperazinyl]ethyl]-N-2-pyridinylcyclohexanecarboxamide maleate (WAY100635; 0.1 mg/kg/day) reduced fluoxetine-induced modulation of CB1 receptor coupling to Gα subunits and AC in the PFC but not in the cerebellum. These results indicate that increased CB1 receptor signaling at the Gαi-AC transduction level is a long-term adaptation induced by fluoxetine in the PFC and point to a role for 5-HT1A receptors in this effect. Basal AC activity, protein kinase A (PKA) catalytic subunit expression, and phospho-cAMP response element-binding protein (pCREB)/CREB ratio were also up-regulated in the PFC of fluoxetine-treated animals, whereas no differences were detected in the cerebellum. It is interesting that long-term Δ9-tetrahydrocannabinol treatment did not elicit antidepressant-like effects or modulated behavioral responses of fluoxetine in an animal model of depression (olfactory bulbectomy). These data suggest that altered signal transduction through CB1 receptors in the PFC may participate in the regulation of the AC-PKA-CREB cascade induced by fluoxetine in this brain area.
Depression is a debilitating disease with a high prevalence and societal cost. Among antidepressant drugs (ADs), selective serotonin reuptake inhibitors (SSRIs) such as fluoxetine are the most widely prescribed. Despite their well established efficacy, the molecular mechanisms underlying the therapeutic action of SSRIs remain unclear. In the short term, SSRIs enhance the efficacy of 5-HT neurotransmission via the inhibition of 5-HT uptake. Nevertheless, clinical improvement results evident only after 2 to 3 weeks of treatment, suggesting a key role for long-term adaptations induced by these compounds (i.e., desensitization of presynaptic inhibitory 5-HT1A receptors) (Blier and de Montigny 1994; Elena Castro et al., 2003). Other neurobiological theories propose that the efficacy of SSRIs is due to alterations in various signaling pathways regulating cellular plasticity and survival. It has been demonstrated that long-term treatment with ADs up-regulates the cAMP-protein kinase A (PKA) cascade and increases neurogenesis in the hippocampus and prefrontal cortex (PFC) (Malberg and Blendy, 2005). 5-HT1A receptors could be involved in the latest effect, because both proliferative and behavioral effects of fluoxetine are abolished by genetic deletion of the 5-HT1A gene (Santarelli et al., 2003).
Cannabinoid CB1 receptors, negatively coupled to adenylyl cyclase (AC) via Gi/o proteins, mediate most of the actions of exogenous and endogenous cannabinoids in the central nervous system (Howlett et al., 2002). CB1 receptor knockout mice exhibit enhanced “depressive-like” behaviors in animal models (Martin et al., 2002; Mato et al., 2007). Pharmacological modulation of CB1 receptors has led to more conflicting results, because both CB1 receptor activation and blockade induce antidepressant-like responses in the same tasks (Griebel et al., 2005; Hill and Gorzalka, 2005; Witkin et al., 2005; Bambico et al., 2007; McLaughlin et al., 2007; Steiner et al., 2008; Morrish et al., 2009). These evidences strongly support the involvement of the brain endocannabinoid (EC) system in the modulation of mood, suggesting that CB1 receptors could play a role in the cause of major depression (Vinod and Hungund, 2006). In favor of this hypothesis, CB1 receptor mRNA (Bortolato et al., 2007), protein expression (Hill et al., 2008a), and signaling through Gi/o proteins (Rodríguez-Gaztelumendi et al., 2009) have been shown to be up-regulated in the PFC of rats subjected to different models of depression and in PFC samples from victims of major depression (Hungund et al., 2004).
In addition, evidence for the contribution of brain EC system to the long-term effects of ADs mainly comes from the fact that their long-term administration modulates CB1 receptor expression (Hill et al., 2008b) and coupling to Gi/o proteins (Rodríguez-Gaztelumendi et al., 2009), as well as the levels of the ECs anandamide and 2-arachidonoylglycerol (Hill et al., 2008b), in the PFC and/or hippocampus. It is noteworthy that information about the putative adaptations of brain CB1 receptors at the cAMP signaling level in response to long-term antidepressants treatment is lacking, despite the proposed importance of this transduction pathway for the long-term adaptations underlying antidepressants efficacy (Malberg and Blendy, 2005). Taking into account that 5-HT receptors are the primary targets of SSRIs and the fact that experimental data support the existence of cross-talk mechanisms between brain EC and 5-HT systems (Marco et al., 2004; Gobbi et al., 2005; Hill et al., 2006b; Mato et al., 2007; Aso et al., 2009), it is noteworthy that the possible implication of 5-HT receptors in the adaptations of the EC system induced by long-term AD treatment has not been addressed.
Thus, the main objective of this study was to investigate the possible contribution of 5-HT1A receptors to the modulation of CB1 receptor activity by long-term fluoxetine treatment. We carried out 2-week treatments with fluoxetine and/or the 5-HT1A receptor antagonist WAY100635, analyzing several steps of CB1-associated signaling pathway (receptor expression, coupling to Gi/o protein subunits, and modulation of AC activity) in the PFC, an area that contains important levels of both CB1 and 5-HT1A receptors, and in the cerebellum, which contains high levels of CB1 receptors but lacks 5-HT1A receptors (Pazos and Palacios, 1985; Pompeiano et al., 1992). We also analyzed the effect of the different treatments on the expression of several components of the cAMP-dependent pathway. Finally, we studied the influence of long-term Δ9-THC administration on the behavioral responses induced by fluoxetine in an animal model of depression (olfactory bulbectomy). Our data demonstrate that CB1 receptor coupling to specific Gαi proteins and to the inhibition of AC are regionally modulated by long-term fluoxetine administration and indicate a role for 5-HT1A receptors in these effects, strengthening the hypothesis that the EC system may play a role in the long-term adaptations induced by this SSRI.
Materials and Methods
Materials.
[3H]CP55,940 (165 Ci/mmol) and [35S]GTPγS (1250 Ci/mmol) were purchased from PerkinElmer Life and Analytical Sciences (Waltham, MA). cAMP 3H assay kit (1028 GBq/mmol) was obtained from GE Healthcare (Chalfont St. Giles, Buckinghamshire, UK). WAY100635, DTT, GDP, GTP, GTPγS, and Sigmacote were from Sigma-Aldrich (Madrid, Spain). WIN55,212-2 was obtained from Tocris Cookson (Bristol, UK). SR141716A was a generous gift from Sanofi Recherche (Montpellier, France). Δ9-Tetrahydrocannabinol (Δ9-THC) was purchased from THC Pharm GMBH The Health Concept (Frankfurt, Germany). Rabbit anti-Gαi1, anti-Gαi2, anti-Gαi3, anti-Gαz, anti-Gαo, and anti-extracellular signal regulated-kinase (ERK) 1/2, and mouse anti-glyceraldehyde-3-phosphate dehydrogenase and anti-histone H1 were purchased from Santa Cruz Biotechnology, Inc. (Heidelberg, Germany). Mouse anti-pERK1/2 and goat anti-rabbit and anti-mouse peroxidase-conjugated antibodies were purchased from Sigma-Aldrich. Rabbit anti-CREB and anti-pCREB (Ser133) were from Millipore (Billerica, MA). Fluoxetine-HCl was kindly donated by Eli Lilly and Co. (Barcelona, Spain). All other chemicals used were from the highest commercial grade available.
Animals.
Male Wistar rats (200–250 g) were maintained on a 12-h light/dark cycle, with access to food and water ad libitum. All experimental procedures were done in accordance with the Declaration of Helsinki, the Spanish legislation, and the European Communities Council Directive on “Protection of Animals Used in Experimental and Other Scientific Purposes” (86/609/EEC).
Drug Treatment and Tissue Preparation.
Rats were randomly assigned to one of the treatment groups, anesthetized with ether, and implanted the same day with an osmotic minipump Alzet 2002 (Alza Corporation, Palo Alto, CA), which delivered 0.5 μl/h. Animals (n = 9 per experimental group) were treated with fluoxetine HCl (10 mg/kg/day), WAY100635 (0.1 mg/kg/day), the combination of both drugs, or with vehicle (50% propylene glycol, 10% ethanol, and 40% distilled water), and minipumps were removed after 14 days of treatment. Rats were killed by decapitation 24 h after removing the minipumps. Two additional groups of rats (n = 4 rats randomly assigned to each experimental group) were treated with a single intraperitoneal injection of fluoxetine HCl (10 mg/kg) or vehicle and were killed 24 h later. In all cases, rat brains were rapidly removed and brain areas were dissected, frozen immediately in isopentane, and stored at −80°C until assay. For behavioral studies (olfactory bulbectomy), animals were treated with vehicle (5% ethanol, 5% Emulphor, and 90% saline), fluoxetine HCl (10 mg/kg/day s.c., 14 days, osmotic minipump), Δ9-THC (10 mg/kg/day i.p., 14 days), and the combination thereof, as described below.
Bilateral Olfactory Bulbectomy and Open Field Test.
The bilateral olfactory bulbectomy (OBX) procedure was performed as described previously (Rodríguez-Gaztelumendi et al., 2009). Animals (sham = 8, OBX = 32) were anesthetized, and the bulbs were excised and removed under stereotactic surgery. Sham-operated animals underwent the same procedure except for excision and aspiration of the olfactory bulbs. An open field test was conducted after a 15-day postsurgery recovery period, as described previously, recording for a 5-min period the number of ambulations, rearings, grooming episodes, and defecations. The next day, OBX animals were randomly divided into five subgroups to receive vehicle-treated sham-operated (n = 8), vehicle-treated OBX (n = 8), fluoxetine-treated OBX (n = 8), Δ9-THC-treated OBX (n = 8), or fluoxetine plus Δ9-THC-treated OBX (n = 8). An open field session was performed at the end of the treatment. At the end of the experiments, animals were killed, and the success of the operation was confirmed anatomically.
Protein Content Determination.
Membrane protein content was determined with the Bio-Rad Protein Assay Kit (Bio-Rad Laboratories, Munich, Germany), using γ-globulin as standard.
cAMP Assays.
For quantitation of endogenous cAMP levels, brain samples were homogenized [1:30 (w/v)] in ice-cold buffer (20 mM Tris-HCl, 300 mM sucrose, 1 mM EGTA, 5 mM EDTA, 5 mM DTT, and 25 μM leupeptin, pH 7.4) using a Teflon tissue grinder (10 strokes, 800 rpm) and centrifuged (1500 rpm, 5 min, 4°C). The resulting supernatants were incubated during 5 min at 100°C and then centrifuged (14,000 rpm, 5 min). cAMP accumulation was quantified in 50-μl supernatant aliquots by using a [3H]cAMP commercial kit (Amersham Signal Transduction Assays; GE Healthcare, Chalfont St. Giles, Buckinghamshire, UK).
To analyze AC activity, brain samples were homogenized [1:100 (w/v)] in ice-cold buffer and centrifuged (1500 rpm, 5 min, 4°C) as described above, and the resulting supernatants were centrifuged again (14,000 rpm, 15 min, 4°C). Pellets were resuspended (150 μg protein/ml) in ice-cold assay buffer (80 mM Tris-HCl, 60 mM sucrose, 2 mM MgCl2, 0.2 mM EGTA, 1 mM EDTA, 100 mM NaCl, 1 mM DTT, 10 μM GTP, 0.5% mg/ml BSA, 0.5 mM 3-isobutyl-1-methylxanthine, 5 mM phosphocreatine, 50 U/ml creatine phosphokinase, 5 U/ml myokinase, and 5 μM forskolin, pH 7.4) and incubated (5 min at 37°C) in the absence or presence of WIN55,212-2 (10 nM to 100 μM). The specificity of the cannabinoid agonist was verified by incubation of 10 μM WIN55,212-2 with 10 μM SR141716A. Experiments assessing 5-HT1A receptor-mediated inhibition of AC in the hippocampus were performed using 8-OH-DPAT (10 nM to 100 μM) as specific agonist and WAY100635 (10 μM) as antagonist. The enzymatic reaction was started by the addition of ATP to a final concentration of 200 μM. The mixture was then incubated (10 min at 37°C), and the reaction was rapidly terminated by 5-min incubation at 100°C. The samples were centrifuged (14,000 rpm, 5 min), and cAMP accumulation was quantified in 50-μl supernatant aliquots by using the [3H]cAMP commercial kit described above. The assays were performed in triplicate Sigmacote-treated borosilicate tubes, and the results were confirmed in two independent experiments.
[3H]CP55,940 Saturation Binding Assay.
Frozen brain samples were homogenized [1:100 (w/v] in ice-cold buffer (50 mM Tris-HCl, 250 mM sucrose, 3 mM MgCl2, and 1 mM EGTA, pH 7.4) using a motor-driven Teflon and glass tissue grinder (10 strokes, 1500 rpm). Homogenates were then centrifuged (1500 rpm, 5 min, 4°C), with the resulting supernatants centrifuged again (14,000 rpm, 15 min, 4°C). The obtained pellets were resuspended in assay buffer (50 mM Tris-HCl, 1 mM EDTA, and 3 mM MgCl2, pH 7.4) and centrifuged again (14,000g, 15 min, 4°C). Pellets were resuspended in assay buffer (50 μg protein/ml in the assay) containing 1 mg/ml BSA and incubated for 60 min at 30°C in the presence of nine different concentrations of the cannabinoid agonist [3H]CP55,940 (0.0125–3.2 nM). Nonspecific binding was determined with 1 μM WIN55,212-2. Bound radioactivity was determined using a Beckman LS6000 liquid scintillation counter (Beckman Coulter, Fullerton, CA) after overnight extraction in 5-ml Ecolite scintillation fluid. All assays were performed in duplicate Sigmacote-treated borosilicate tubes, and the results were confirmed in two independent experiments.
Agonist-Stimulated [35S]GTPγS Binding.
Tissue samples were homogenized [1:100 (w/v)] in ice-cold buffer (50 mM Tris-HCl, 250 mM sucrose, 3 mM MgCl2, 1 mM EGTA, and 1 mM DTT, pH 7.4) and processed as reported above for [3H]CP55,940 binding assays. Membrane aliquots (200 μg protein/ml in the assay) were incubated for 120 min at 30°C in assay buffer (100 mM NaCl, 50 mM Tris-HCl, 3 mM MgCl2, 1 mM EGTA, 1 mM DTT, 50 μM GDP, and 1 mg/ml BSA, pH 7.4) containing 0.1 nM [35S]GTPγS. Cannabinoid agonist stimulation of [35S]GTPγS binding was determined using 1 nM to 100 μM WIN55,212-2. The specificity of the CB1 receptor-mediated stimulation was verified by coincubation of 10 μM WIN55,212-2 with 1 μM SR141716A. Nonspecific binding was determined in the presence of 10 μM GTPγS. Experiments were terminated by sample dilution in ice-cold buffer (50 mM Tris-HCl and 1 mg/ml BSA, pH 7.4) and rapid filtration under vacuum (Cell Harvester M-12R; Brandel Inc., Gaithersburg, MD) through GF/C glass fiber filters. [35S]GTPγS binding assays were terminated and measured as described for [3H]CP55,940 experiments. All assays were performed in duplicate Sigmacote-treated borosilicate tubes, and the results were confirmed in two independent experiments.
Immunoprecipitation of [35S]GTPγS-Labeled Gα Subunits.
Membrane homogenates were obtained as reported for agonist-stimulated [35S]GTPγS binding assays. Resuspended pellets (500 μg protein/ml in the assay) were incubated with 2 nM [35S]GTPγS and 10 μM WIN55,212-2 in a final 100-μl assay volume for 30 min at 30°C. Nonspecific binding was determined in the presence of 10 μM GTPγS. Membrane suspensions were then solubilized on ice with 1% IGEPAL, 0.5% sodium deoxycholate, 0.1% SDS, 2.5 mM CHAPS, 0.1 mM phenylmethylsulfonyl fluoride, 0.01 M aprotinin, 1 μg/ml leupeptin, 1 μg/ml pepstatin, 1 μl/ml antipain, and 10 μg/ml chymostatin for 30 min. Solubilized membranes were incubated for 3 h at room temperature with 15 μl of specific rabbit anti-Gαi1, anti-Gαi2, anti-Gαi3, anti-Gαz, and anti-Gαo antibodies immobilized to superparamagnetic Dynabeads Protein A (overnight, 4°C). After three washes with 1 ml of phosphate-buffered saline, the beads were pelleted, and the entrapped radioactivity was counted in 4 ml of Ecolite scintillation cocktail. Antibody specificity was confirmed in our experimental conditions by Western blot as reported previously (Valdizán et al., 2009).
Western Blot.
Brain samples were homogenized [1:15 (w/v)] by using a Potter homogenizer provided with a loosely fitting Teflon pestle in ice-cold buffer (50 mM Tris-HCl, 250 mM sucrose, 3 mM MgCl2, 1 mM EGTA, and 1 mM DTT, pH 7.4) containing the following protease and phosphatase inhibitors: 1 mM phenylmethylsulfonyl fluoride, 10 μl/ml aprotinin, 10 μg/ml leupeptin, 10 μg/ml pepstatin A, 10 μg/ml antipain, 10 μg/ml chymostatin, 5 μg/ml trypsin inhibitor, 1 mM NaV, 1 mM NaF, 1 mM cantharidin, and 10 μM E-64. After homogenization, 100 μl of homogenate was lysed in homogenization buffer containing 1% IGEPAL, 0.1% sodium deoxycholate, 0.2% SDS, and 0.1% Triton X-100 for 30 min on ice for the total cell lysate (TCL). The remaining homogenate was centrifuged at 1500 rpm (5 min, 4°C), and the resulting pellets were used to obtain the nuclear fraction by homogenization in 20 mM HEPES, 0.45 M NaCl, and 1 mM EDTA, pH 7.9, containing protease and phosphatase inhibitors and incubation on ice for 30 min. Solubilized proteins were recovered in the supernatant after centrifugation (14,000 rpm, 10 min, 4°C). Protein preparations (30–50 μg per lane) were resolved on 12.5% SDS-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride (nonphosphorylated proteins) or to nitrocellulose (phosphorylated proteins) membranes. Membranes were incubated in the following primary antibodies: rabbit anti-ERK1/2 (1:2000), mouse anti-pERK1/2 (1:10,000), rabbit anti-CREB (1:1000), and rabbit anti-pCREB (1:1000). After extensive washings in Tris-buffered saline/0.05% Tween 20, membranes were incubated with horseradish peroxidase-conjugated secondary antibodies. Secondary antibodies were detected with ECL Advance kit (GE Healthcare). Blot quantitations were performed by densitometric scanning using Scion Image software (Scion Corporation, Frederick, MD). The densitometry values were normalized with respect to the values obtained with anti-glyceraldehyde-3-phosphate dehydrogenase (TCL) and/or anti-histone H1 (nuclear fraction) antibodies. Data for every sample were the mean of at least two independent experiments.
Data Analysis.
[3H]CP55,940 saturation binding data were transformed using the method described by Scatchard (1949), and GraphPad Prism computer software for Windows (GraphPad Software Inc., San Diego, CA) was used to estimate maximal [3H]CP55,940 binding sites (Bmax) and dissociation constants (Kd). Kd values were normalized for comparison as −logKd (pKd). The effect of each concentration of cannabinoid agonist was expressed as a percentage of stimulation: [% = (agonist effect × 100)/basal activity] in [35S]GTPγS assays and the percentage of inhibition in cAMP assays [% = (agonist effect × 100)/(forskolin effect − 100)]. Analysis of concentration-effect curves was conducted by nonlinear regression using Prism software to estimate the theoretical maximal effect (Emax) and the potency (EC50) in [35S]GTPγS binding assays and the (Imax) and (IC50) values in cAMP assays of the cannabinoid agonist in each assay. EC50 and IC50 values were normalized as −logEC50 and −logIC50 (pEC50 and pIC50) for comparison. Levels of coupling of CB1 receptors by WIN55,212-2 (10−5 M) to the diverse G protein subunits were obtained as the percentage over the value in the absence of agonist. The efficiency of coupling of CB1 receptors was obtained from the control group. Statistical comparison of experimental groups was made using one-way ANOVA with Newman-Keuls post hoc tests. Differences were taken as statistically significant when p < 0.05. Data are presented as mean ± S.E.M.
Results
Long-Term Fluoxetine Treatment Modulates Basal and CB1 Receptor-Dependent Inhibition of Adenylyl Cyclase Activity but Not Endogenous cAMP Levels.
Endogenous cAMP levels were unchanged in the PFC or in the cerebellum after long-term treatment with fluoxetine, the combination of fluoxetine plus WAY100635, or WAY100635 alone (in picomoles per milligram of protein; vehicle: PFC, 1.76 ± 0.43, cerebellum, 3.34 ± 0.44; fluoxetine: PFC, 1.78 ± 0.20, cerebellum, 3.57 ± 0.46; fluoxetine + WAY100635: PFC, 1.79 ± 0.47, cerebellum, 3.23 ± 0.18; WAY100635: PFC, 1.81 ± 0.42, cerebellum, 3.38 ± 0.23).
The effects of long-term treatment with fluoxetine, the combination of fluoxetine plus WAY100635, or WAY100635 alone on basal, forskolin-stimulated, and CB1 receptor-mediated inhibition of AC in the PFC and cerebellum are depicted in Fig. 1. Long-term fluoxetine treatment significantly increased basal AC activity in the PFC, and this effect was not observed in the animals treated with fluoxetine + WAY100635 or with WAY100635 alone (Fig. 1A). One-way ANOVA for independent measures revealed an effect of the group (F = 3.2, p < 0.05; Fig. 1A) with significant differences between the vehicle and fluoxetine groups (Fig. 1A; *, p < 0.05) and between the fluoxetine versus WAY100635 alone groups (Fig. 1A, $, p < 0.05). Differences in the ability of the AC activator forskolin to increase cAMP levels were no detected in any of the treated groups (vehicle, 164 ± 34; fluoxetine, 207 ± 42; fluoxetine + WAY100635, 181 ± 36; and WAY100635, 172 ± 22 pmol/min/mg protein). Incubation with the cannabinoid agonist WIN55,212-2 induced a concentration-dependent decrease of forskolin-stimulated AC in the PFC of vehicle-treated rats, with a maximal inhibition (Imax) of 30.0 ± 3.0% and a pIC50 of 5.7 ± 0.08. Long-term treatment with fluoxetine induced a significant increase in the maximal ability of WIN55,212-2 to inhibit AC in the PFC (Imax = 41.8 ± 3.7%) that was absent in those rats treated with fluoxetine plus the 5-HT1A receptor antagonist WAY100635 (Imax = 25.5 ± 3.2%) or WAY100635 alone (Imax = 25.3 ± 2.5%) (Fig. 1B). One-way ANOVA revealed a significant effect of the group (F = 6.11, p < 0.01; Fig. 1B), with significant differences between the vehicle and fluoxetine groups (Fig. 1B; *, p < 0.05) and between the fluoxetine versus fluoxetine plus WAY100635 (Fig. 1B; $$, p < 0.01) and WAY100635 alone groups (Fig. 1B; $$, p < 0.01). The potency of the cannabinoid agonist to inhibit AC was not altered by long-term treatment with fluoxetine, fluoxetine plus WAY100635, or WAY100635 alone (Fig. 1B). It is noteworthy that the observed modifications in basal AC activity and WIN55,212-2 efficacy to inhibit AC in the PFC of rats continuously treated with fluoxetine were not detected after a single exposure to the SSRI (Table 1). In the cerebellum of the same groups of rats, we did not observe any modifications of basal (Fig. 1A) or forskolin-stimulated AC activity (vehicle, 76 ± 9; fluoxetine, 94 ± 21; fluoxetine + WAY100635, 78 ± 11; and WAY100635, 88 ± 14 pmol/min/mg protein). In contrast to the observed up-regulation of CB1 receptor coupling to AC in the PFC of fluoxetine-treated rats, the ability of WIN55,212-2 to inhibit cAMP production was reduced in the cerebellum after long-term fluoxetine treatment (Imax = 23.5 ± 0.9% in fluoxetine-treated rats versus 38.7 ± 2.1% in vehicle-treated rats). This reduced coupling ability of CB1 receptors to AC in the cerebellum of fluoxetine-treated rats was not prevented by cotreatment with the 5-HT1A receptor antagonist WAY100635 (Imax = 18.1 ± 1.9%), which when administered alone did not modify WIN55,212-2 efficacy in cAMP assays (Imax = 39.6 ± 2.5%) (Fig. 1C). One-way ANOVA indicated a significant effect of the group (F = 25.2, p < 0.001; Fig. 1C), with significant differences between the vehicle versus fluoxetine (Fig. 1C; **, p < 0.01) and fluoxetine + WAY100635 groups (Fig. 1C; **, p < 0.01) and between the fluoxetine versus WAY100635 alone groups (Fig. 1C; p < 0.01). None of the treatments altered the potency of the cannabinoid agonist to inhibit AC in the cerebellum (Fig. 1C).
To determine the specificity of fluoxetine effect on CB1 receptor coupling to AC, we analyzed the inhibition of forskolin-stimulated cAMP levels by a Gi/o protein-coupled receptor with a well known implication in mood disorders, as is the case of 5-HT1A receptors. Inhibition of AC by 5-HT1A receptors was studied in the hippocampus of the same treated rats, given the high expression of 5-HT1A protein in this brain area, and its involvement in the long-term effects of ADs. As reported above for the PFC, long-term fluoxetine treatment increased basal AC activity in the hippocampus, and this effect was not observed in the animals treated with fluoxetine + WAY100635 or with WAY100635 alone (Table 2). Statistical comparison using one-way ANOVA indicated an effect of the group (Table 2; F = 4.1; p < 0.05), with significant differences between the vehicle and fluoxetine groups (Table 2; *, p < 0.05) and between the fluoxetine versus fluoxetine + WAY100635 and WAY100635 alone groups (Table 2; #, p < 0.05 in both cases). Differences in the stimulatory effect of forskolin on cAMP levels were not detected in any of the treatment groups (Table 2). In contrast to the reported effect of fluoxetine on CB1 receptor coupling to AC in the PFC, the Imax for AC inhibition by the 5-HT1A receptor agonist 8-OH-DPAT was reduced in the hippocampus of the same treated rats. One-way ANOVA revealed a significant effect of the group (Table 2; F = 4; *, p < 0.05), with significant differences between the vehicle and fluoxetine groups (Table 2; **, p < 0.01). The potency of 8-OH-DPAT to inhibit AC was not altered by any of the treatments (Table 2).
Long-Term Fluoxetine Treatment Does Not Alter CB1 Receptor Density or Agonist-Stimulated [35S]GTPγS Binding.
The observed alterations in the efficacy of WIN55,212-2 to inhibit AC in the PFC and cerebellum of fluoxetine-treated rats could reflect changes in CB1 receptor expression and/or coupling to Gi/o proteins. To test this hypothesis, we carried out saturation binding experiments with the cannabinoid agonist [3H]CP55, 940 and with WIN55,212-2-stimulated [35S]GTPγS binding assays after long-term treatment with fluoxetine, fluoxetine + WAY100635, or WAY100635 alone. [3H]CP55,940 binding to rat PFC and cerebellum membranes was saturated at 3 nM, with the nonspecific binding being approximately 30% of the total radioligand binding at a concentration close to the estimated apparent dissociation constant (Kd). Long-term in vivo treatment with fluoxetine, WAY100635, or the combination of both did not modify CB1 receptor density or agonist affinity in either brain area (Table 3). In addition, none of treatments altered the maximal effect or the potency of WIN55,212-2 to stimulate [35S]GTPγS binding (Table 3).
Long-Term Fluoxetine Treatment Modulates CB1 Receptor Coupling to Specific Gα Protein Subunits.
Results from [35S]GTPγS binding experiments suggest that the observed changes in CB1 receptor signaling at the AC level induced by long-term fluoxetine treatment in the PFC and cerebellum take place in the absence of any modification in the receptor coupling to Gi/o proteins. Nevertheless, the possibility also exists that the administration of this SSRI affects the coupling ability of CB1 receptors to specific Gα subunits in a complex manner, so that these modifications are overlooked when assessing global coupling to Gi/o proteins in [35S]GTPγS binding assays. To further clarify this point, we performed immunoprecipitation of WIN55,212-2-stimulated [35S]GTPγS-labeled Gα protein subunits after long-term treatment with fluoxetine, WAY100635, or the combination of both compounds. The coupling efficiency of CB1 receptors to Gα protein subunits induced by the agonist WIN55,212-2 in PFC was Gαi3 (173 ± 18%) ≅ Gαi2 (169 ± 6%) > Gαo (145 ± 8%) > Gαz (131 ± 14%). In the cerebellum, coupling efficiency values were Gαi3 (335 ± 28%) > Gαo (210 ± 36%) ≅ Gαz (210 ± 20%) > Gαi2 (203 ± 3%) ≅ Gα1 (202 ± 15%) (Fig. 2). No significant coupling of CB1 receptors to Gαi1 was observed in PFC after activation with WIN55, 212-2.
In the PFC, one-way ANOVA indicated a significant difference in the efficacy of the cannabinoid agonist to activate Gαi2 and Gαi3 subunits among groups (Fig. 2A; F = 3.5, *, p < 0.05; and F = 3.03, *, p < 0.05, respectively), whereas no difference was detected regarding stimulation of Gα0 and Gαz proteins. Post hoc test revealed that the level of Gαi2 and Gαi3 subunits activation by WIN55,212-2 was significantly increased in the PFC of fluoxetine-treated rats (Fig. 2A; *p < 0.05 versus vehicle). As reported above for the fluoxetine effect on CB1 receptor coupling to AC in this brain area, the 5-HT1A receptor antagonist WAY100635 did not alter the profile of Gα protein activation by WIN55,212-2 when administered alone but prevented the increased stimulation of Gαi2 protein subunits when coadministered with the SSRI (Fig. 2A; $, p < 0.05 versus fluoxetine).
In the cerebellum of fluoxetine-treated rats, we detected a selective down-regulation of CB1 receptor coupling to Gαi2 subunits. This effect of the SSRI in the cerebellum was not prevented by concomitant administration of WAY100635; that by itself did not modify WIN55,212-2 activation of Gα subunits (Fig. 2B). One-way ANOVA indicated a significant effect of the group (F = 21.9, p < 0.001; Fig. 2B), with significant differences between the vehicle versus fluoxetine (Fig. 2B; ***, p < 0.001) and fluoxetine + WAY100635 groups (Fig. 2B; ***, p < 0.001).
Long-Term Fluoxetine Treatment Modulates PKA Expression and CREB Phosphorylation in the PFC.
Long-term administration of ADs, including SSRIs, has been consistently shown to up-regulate different components of the cAMP pathway, including PKA and CREB (Malberg and Blendy, 2005). On the other hand, CB1 receptors are coupled to the activation of the ERK-CREB transduction cascade in several brain areas, including the PFC and the cerebellum (Rubino et al., 2004). In an attempt to elucidate the functional consequences of the observed modulation of CB1 receptor coupling to Gαi proteins-AC by long-term fluoxetine treatment, we analyzed the effects of the SSRI on the expression levels of PKA catalytic and regulatory 1α domains (PKA_c and PKA_r1α) and of pCREB/CREB, ERK1/2, and pERK1/2. One-way ANOVA indicated that PKA_c levels in TCLs of PFC varied among treated groups (F = 3.2; *, p < 0.05), whereas PKA_r1α expression remained unaltered. Post hoc tests revealed increased PKA_c levels in fluoxetine-treated rats (*, p < 0.05 versus vehicle) that were not detected in rats treated with the combination of fluoxetine + WAY100635 or with the 5-HT1A receptor antagonist alone (Fig. 3, A and B). Long-term fluoxetine treatment also modulated CREB and pCREB expression in TCL of rat PFC (F = 2.8, p < 0.05 and F = 22.2, p < 0.001, respectively; Fig. 3) without altering ERK1/2 and pERK1/2 levels in the same area (Fig. 3A and Table 4). The expression of CREB was significantly enhanced in PFC lysates from fluoxetine-treated rats (Table 4; *, p < 0.05 versus vehicle), and this effect was not observed in the fluoxetine + WAY100635 group. Increased pCREB levels were also detected in the TCL of rat PFC after long-term fluoxetine treatment (Table 4; ***, p < 0.001 versus vehicle) and after fluoxetine + WAY100635 (Table 4; ***, p < 0.001 versus vehicle). Consistently, pCREB/CREB ratio was found to be increased in PFC lysates from rats continuously treated with fluoxetine (Fig. 3C; F = 4.7, p < 0.01; *, p < 0.05 versus vehicle) and fluoxetine + WAY100635 (Fig. 3C; *, p < 0.05 versus vehicle). Consistently, we detected enhanced pCREB expression in nuclear fractions of the same treated rats (Fig. 3D; F = 6.5; **, p < 0.01). Post hoc analysis indicated that long-term treatment with fluoxetine and with fluoxetine + WAY100635 enhanced pCREB levels in nuclear fractions of PFC (Fig. 3D; **, p < 0.01; *, p < 0.05 versus vehicle, respectively) (Fig. 3D). Treatment with WAY100635 alone did not modulate CREB, pCREB, ERK1/2, or pERK1/2 levels in the TCL and/or nuclear fractions of rat PFC (Fig. 3 and Table 4). In contrast to this observed up-regulation of the PKA-CREB transduction cascade in the PFC by long-term fluoxetine treatment, no modifications in PKA_c, PKA_r1α, pCREB/CREB, or pERK/ERK levels were detected in the cerebellum of treated rats (Table 5).
Role of CB1 Receptors in the Antidepressant Efficacy of Long-Term Fluoxetine Treatment in the OBX Depression Model.
To test whether the observed modulation of CB1 receptor coupling to specific Gαi protein subunits/AC in the PFC by long-term fluoxetine treatment could participate in the therapeutic efficacy of this SSRI, we studied the behavioral consequences of long-term CB1 receptor activation in the OBX depression model. Two weeks after sham (n = 8) or OBX (n = 32) surgery and before the initiation of the long-term treatments, OBX-induced hyperactivity was confirmed on the open field test (data not shown). The next day, sham animals were treated with vehicle, and OBX animals were treated with either vehicle, fluoxetine (10 mg/kg/day), the CB1 receptor agonist Δ9-THC (10 mg/kg/day), or the combination of both compounds for another 14 days, and the behavioral effects of the treatments were evaluated in the open field (Table 6). In agreement with our previous observations (Rodríguez-Gaztelumendi et al., 2009), long-term fluoxetine treatment induced a complete reversion of the increased number of ambulations in OBX rats relative to the vehicle-treated OBX group (Table 6; #, p < 0.05). In contrast, we detected no difference in total ambulation scores between Δ9-THC-treated and vehicle-treated OBX rats (Table 6). Finally, long-term treatment with Δ9-THC did not modulate the ability of fluoxetine to reduce hyperactivity in OBX rats (Table 6). Thus, long-term fluoxetine treatment rectified the locomotor hyperactivity in the OBX rats, whereas long-term Δ9-THC was without effect and did not affect the behavioral effects of fluoxetine in this depression model.
Discussion
The present results constitute the first evidence that long-term treatment with an antidepressant compound enhances CB1 receptor coupling to specific Gαi subunits and to AC inhibition in a brain area involved the pathogenesis of mood disorders and in the long-term effects of ADs, as is the case with the PFC. Among the few studies that have addressed the effects of long-term antidepressant exposure on the activity of brain EC system, most of them have focused on CB1 receptor density. Some of them suggest an enhancement of CB1 receptor expression in brain areas with a well established role in depression (Hill et al., 2006a,b; Hill et al., 2007). Increased CB1 receptor density in the rat PFC has been reported after long-term administration of the monoamine oxidase inhibitor tranylcypromine or with fluoxetine (Hill et al., 2008b) but not with the tricyclic desipramine (Hill et al., 2006a). Although the absence of CB1 receptor expression modulation by fluoxetine reported here is in contrast to the findings of Hill et al. (2008b), probably because of differences in the treatment regimen between both studies, our data demonstrating up-regulated CB1 receptor coupling to Gαi proteins/AC strengthen the idea that long-term administration of chemical antidepressants enhance CB1 receptor activity in this brain area. When interpreting these findings, it must also be considered that reduced CB1 receptor density has also been reported in the PFC after electroconvulsive shock (Hill et al., 2007). These data suggest that forebrain EC system may be important for the efficacy of antidepressant treatments, reinforcing the interest of simultaneously analyzing the effects of AD administration on the different levels of the CB1 receptor transduction cascade.
Stimulation of CB1 receptors by WIN 55212-2 resulted in the activation of at least five different Gαi/o proteins subunits in PFC and cerebellum, with slight differences in the efficacy of subunit activation across brain regions, as described previously (Prather et al., 2000). Our results also indicate increased CB1 receptor coupling to the activation of Gαi2 and Gαi3 subunits, and not to Gαo subunits, in the PFC of fluoxetine-treated rats. The absence of fluoxetine effect on CB1 receptor-stimulated [35S]GTPγS binding strengthens this idea, because these assays mainly detect the activation of Gαo subunits (Jiang et al., 2001), which might be related to the fact that Gαo are in significant excess over Gαi in the brain (Sternweis and Robishaw, 1984; Spicher et al., 1992). It is noteworthy that both CB1 receptor stimulation of Gαi2 and AC inhibition were significantly reduced in the cerebellum of the same fluoxetine-treated rats. Altogether, these data strongly suggest that fluoxetine-induced modulation of CB1 receptor coupling to Gαi proteins may contribute to the observed changes in the level of CB1 receptor-dependent modulation of AC activity. The fact that opposite modulation of CB1 receptor coupling to Gαi proteins/AC by long-term fluoxetine treatment was detected in the PFC and cerebellum of the same animals strengthens the emerging idea that the effects of long-term AD treatment on the activity of EC system differ among brain structures (Hill et al., 2008b).
A major finding of the present study is that pharmacological blockade of 5-HT1A receptors with WAY100635 suppresses fluoxetine-induced up-regulation of CB1 receptor coupling to Gαi2 proteins and to AC in the PFC. It is noteworthy that WAY100635 did not prevent desensitization of CB1 receptor signaling at the Gαi2 protein/AC level in the cerebellum, a brain area devoid of 5-HT1A receptors (Pazos and Palacios, 1985; Pompeiano et al., 1992). These results reinforce the hypothesis that fluoxetine-induced modulation of CB1 receptor signaling through Gαi2 proteins underlies the observed changes at the AC level. In addition, these findings strongly suggest that the enhanced CB1 signaling at the Gαi2 protein/AC in the PFC by long-term fluoxetine treatment results from the activation of 5-HT1A receptors. The efficacy of our WAY100635 administration protocol to block 5-HT1A receptors during fluoxetine administration is supported by the ability of this compound to prevent the desensitization of 5-HT1A receptor-mediated responses in the dorsal raphe nucleus (DRN) (Castro et al., 2008) and in the hippocampus (present study). Another important question is whether long-term administration of other antidepressant compounds that inhibit 5-HT uptake induces a similar effect in brain areas expressing 5-HT1A receptors. In this regard, data from our laboratory suggest that this is indeed the case, because administration of the dual 5-HT and noradrenaline inhibitor venlafaxine (40 mg/kg/day s.c., 14 days, minipumps) also enhances CB1 receptor-mediated inhibition of AC in the rat PFC (E. M. Valdizán, E. Castro, and A. Pazos, unpublished observations). Although the molecular mechanisms by which fluoxetine up-regulates CB1 receptor functionality remain to be elucidated, a growing body of evidence supports the existence of interactions between brain EC and 5-HT systems, with our results pointing out to a special involvement of 5-HT1A receptors. It is noteworthy that adequate CB1 receptor functionality seems to be important for 5-HT1A receptor-mediated biochemical and behavioral responses, because decreased efficacy of the 8-OH-DPAT to activate Gi/o proteins (Mato et al., 2007), impaired anxiolytic (Urigüen et al., 2004) and hypothermic (Mato et al., 2007) effects of 5-HT1A agonists and reduced functionality of 5-HT1A autoreceptors in the DRN (Aso et al., 2009) have been detected in CB1 knockout animals. In this context, the present report uncovers the possibility that modulation of CB1 receptors participates in those effects of long-term fluoxetine treatment that are triggered by 5-HT1A receptors.
Regarding the observed increase in basal AC activity (without modification of endogenous cAMP levels), PKA_c subunit expression and pCREB/CREB ratio in the PFC by long-term fluoxetine treatment, our findings are in good agreement with research carried out during the past decade, which provides strong support for the up-regulation of the cAMP pathway after long-term antidepressant administration (Malberg and Blendy, 2005). Results include increased coupling of stimulatory G protein (Gαs) to AC (Donati and Rasenick, 2003) in the presence of persistent activation of various Gi/o-coupled receptors, consistent with the partial reversion of the increase of basal AC activity and PKA_c subunit observed after association of fluoxetine with the 5-HT1A receptor antagonist in the PFC and/or hippocampus. Although the modulation of CB1 receptor functionality could be involved in the activation of the cAMP-CREB cascade by long-term fluoxetine treatment, our methodological approach does not allow for the clarification of this issue. It is noteworthy that WAY100635 completely prevented a fluoxetine-induced increase in CB1 receptor coupling to Gαi2 protein/AC and partially reversed the up-regulation of both basal AC activity and PKA-c subunit expression induced by the SSRI in the PFC. These findings suggest that up-regulated CB1 receptor signaling contributes to some extent to 5-HT1A receptor-mediated modulation of the cAMP cascade after fluoxetine administration, although the involvement of other mechanisms cannot be discarded. In this scenario, the lack of ability of WAY100635 to prevent fluoxetine-induced increase in pCREB/CREB ratio would indicate that the up-regulation of CB1 receptor signaling does not contribute to this effect. Indeed, although studies addressing the effects of CB1 receptor antagonists on the modulation of cAMP-CREB cascade by long-term AD administration are lacking, we have observed no significant differences between control and CB1 knockout animals with respect to pCREB up-regulation by long-term fluoxetine treatment (E. M. Valdizán, A. Ozaita, E. Aso, R. Maldonado, A. Pazos, and O. Valverde, unpublished data). On the other hand, we found no significant modification in pERK/ERK expression in the PFC after long-term fluoxetine treatment.
The role of the observed up-regulation of CB1 receptor coupling to AC in the behavioral adaptations induced by long-term fluoxetine treatment is also difficult to interpret at the present time. In addition to the already mentioned possibility that CB1 receptors participate in these effects via the modulation of fluoxetine-induced adaptations in cAMP cascade, recent data also suggest that up-regulated CB1 receptor signaling in the PFC could also elicit antidepressant effects by enhancing the activity of 5-HT neurons in the DRN (Gobbi et al., 2005; Bambico et al., 2007). Consistent with the idea that enhanced signaling through CB1 receptors may result in antidepressant effects, CB1 receptor knockout mice exhibit enhanced depressive-like behaviors (Martin et al., 2002; Mato et al., 2007), and low doses of cannabinoid agonists (Bambico et al., 2007; McLaughlin et al., 2007; Morrish et al., 2009) or inhibitors of EC degradation (Gobbi et al., 2005; Bortolato et al., 2007) produce antidepressant-like effects in rodents. Nevertheless, it should be noted that CB1 receptor antagonists also behave as antidepressants in behavioral models (Griebel et al., 2005; Witkin et al., 2005; Steiner et al., 2008). When interpreting these controversial data, an important consideration is that, although studies specifically addressing the antidepressant effects of CB1 cannabinoid agonists and antagonists in humans are lacking, the high incidence of depression and anxiety in clinical trials with the CB1 receptor antagonist rimonabant for the treatment of obesity, despite depressed mood being an exclusion criterion in these trials (Mitchell and Morris, 2007), has motivated its recent withdrawal from use or lack of approval. Along with preclinical data, these clinical findings strongly suggest that CB1 receptors may represent a novel target for the treatment of mood disorders. In support of this, we have recently demonstrated in the PFC of OBX rats, an animal model of depression, a significant increase in CB1 receptor density and coupling to G proteins in the PFC, together with the characteristic hyperactivity in the open field test: both neurochemical and behavioral responses were absent after fluoxetine (Rodríguez-Gaztelumendi et al., 2009). It is noteworthy that exposure to a single Δ9-THC injection before the initiation of long-term treatment with fluoxetine significantly decreased hyperactivity in OBX rats, indicating that direct activation of CB1 receptors elicits antidepressant-like effects in this depression model (Rodríguez-Gaztelumendi et al., 2009). In contrast, here we report that long-term Δ9-THC administration neither elicits antidepressant-like effects nor modulates the behavioral efficacy of long-term fluoxetine treatment in this model. It could be hypothesized that sustained administration of Δ9-THC is expected to induce adaptive changes in the CB1 receptor signaling axis, including down-regulation in many brain areas (Breivogel et al., 1999). These changes could strongly modify those reported here induced by fluoxetine, resulting in the lack of influence of Δ9-THC on the antidepressant responses. The full clarification of this issue goes beyond the scope of the present study.
In summary, this study demonstrates that long-term fluoxetine treatment modulates the functionality of CB1 receptors in the rat brain in a region-specific manner. Up-regulated CB1 receptor signaling at the Gαi protein-AC transduction level in the PFC, even in the absence of significant changes in receptor density, is one of the long-term molecular adaptations triggered by long-term fluoxetine treatment. This response is mediated through the activation of 5-HT circuits implicating 5-HT1A receptors, because it is prevented by coadministration of WAY100635 and does not take place in the cerebellum, a brain area devoid of these receptors. These results unveil the possible relevance of the EC/5-HT interactions for the therapeutic responses of ADs that inhibit the uptake of this monoamine. Further research is necessary to fully clarify the extent to which modulation of the EC system may play a role in the therapeutic efficacy of these compounds.
Acknowledgments
The technical assistance of Beatriz Romero and Rebeca Madureira is kindly acknowledged.
Footnotes
This work was supported by the Ministerio de Education y Ciencia [Grants CICYT SAF04-00941, SAF07-61862]; Plan Nacional sobre Drogas; and Fundación Marqués de Valdecilla [Grant API 05/23].
Article, publication date, and citation information can be found at http://molpharm.aspetjournals.org.
doi:10.1124/mol.109.060079
-
ABBREVIATIONS:
- AD
- antidepressant drug
- AC
- adenylyl cyclase
- BSA
- bovine serum albumin
- CB1
- cannabinoid type 1
- [3H]CP55,940
- [3H](1R,3R,4R)-3-[2-hydroxy-4-(1,1-dimethylheptyl)phenyl]-4-(3-hydroxypropyl)cyclohexan-1-ol
- 8-OH-DPAT
- 8-hydroxydi-n-propylaminotetralin
- DTT
- dl-dithiothreitol
- [35S]GTPγS
- guanosine 5′-O-(3-[35S]thio)triphosphate
- PFC
- prefrontal cortex
- SR141716A
- N-(piperidin-1-yl)-5-(4-chlorophenyl)-1-(2,4-dichlorophenyl)-4-methyl-1H-pyrazole-3-carboxamide hydrochloride
- SSRI
- selective serotonin reuptake inhibitor
- Δ9-THC
- Δ9-tetrahydrocannabinol
- WAY100635
- N-[2-[4-(2-methoxyphenyl)-1-piperazinyl]ethyl]-N-2-pyridinylcyclohexanecarboxamide maleate
- WIN55,212-2
- (R)-(+)-[2,3-dihydro-5-methyl-3-(4-morpholinylmethyl)pyrrolo-[1,2,3-de]-1,4-benzoxazin-6-yl]-1-naphthalenyl-methanonemesylate
- PKA
- protein kinase A
- EC
- endocannabinoid
- ERK
- extracellular signal regulated-kinase
- OBX
- olfactory bulbectomy
- TCL
- total cell lysate
- ANOVA
- analysis of variance
- DRN
- dorsal raphe nucleus
- CHAPS
- 3-[(3-cholamidopropyl)dimethylammonio]propanesulfonate
- CREB
- cAMP response element-binding protein
- 5-HT
- 5-hydroxytryptamine
- E-64
- N-(trans-epoxysuccinyl)-l-leucine 4-guanidinobutylamide.
- Received August 4, 2009.
- Accepted December 7, 2009.
- Copyright © 2010 The American Society for Pharmacology and Experimental Therapeutics
References
- Aso et al., 2009.↵
- Bambico et al., 2007.↵
- Bortolato et al., 2007.↵
- Blier and de Montigny, 1994.↵
- Breivogel et al., 1999.↵
- Castro et al., 2008.↵
- Donati and Rasenick, 2003.↵
- Elena Castro et al., 2003.↵
- Gobbi et al., 2005.↵
- Griebel et al., 2005.↵
- Hill and Gorzalka, 2005.↵
- Hill et al., 2006a.↵
- Hill et al., 2007.↵
- Hill et al., 2008a.↵
- Hill et al., 2008b.↵
- Hill et al., 2006b.↵
- Howlett et al., 2002.↵
- Hungund et al., 2004.↵
- Jiang et al., 2001.↵
- Malberg and Blendy, 2005.↵
- Marco et al., 2004.↵
- Martin et al., 2002.↵
- Mato et al., 2007.↵
- McLaughlin et al., 2007.↵
- Mitchell and Morris, 2007.↵
- Morrish et al., 2009.↵
- Pazos and Palacios, 1985.↵
- Pompeiano et al., 1992.↵
- Prather et al., 2000.↵
- Rodríguez-Gaztelumendi et al., 2009.↵
- Rubino et al., 2004.↵
- Santarelli et al., 2003.↵
- Scatchard, 1949.↵
- Spicher et al., 1992.↵
- Steiner et al., 2008.↵
- Sternweis and Robishaw, 1984.↵
- Urigüen et al., 2004.↵
- Valdizán et al., 2009.↵
- Vinod and Hungund, 2006.↵
- Witkin et al., 2005.↵