Caffeine is associated with procognitive effects in humans by counteracting overactivation of the adenosine A2A receptor (A2AR), which is upregulated in the human forebrain of aged and Alzheimer’s disease (AD) patients. We have previously shown that an anti-A2AR therapy reverts age-like memory deficits, by reestablishment of the hypothalamic-pituitary-adrenal (HPA) axis feedback and corticosterone circadian levels. These observations suggest that A2AR over-activation and glucocorticoid dysfunction are key events in age-related hippocampal deficits; but their direct connection has never been explored. We now show that inducing A2AR overexpression in an aging-like profile is sufficient to trigger HPA-axis dysfunction, namely loss of plasmatic corticosterone circadian oscillation, and promotes reduction of GR hippocampal levels. The synaptic plasticity and memory deficits triggered by GR in the hippocampus are amplified by A2AR over-activation and were rescued by anti-A2AR therapy; finally, we demonstrate that A2AR act on GR nuclear translocation and GR-dependent transcriptional regulation. We provide the first demonstration that A2AR is a major regulator of GR function and that this functional interconnection may be a trigger to age-related memory deficits. This supports the idea that the procognitive effects of A2AR antagonists, namely caffeine, on Alzheimer’s and age-related cognitive impairments may rely on its ability to modulate GR actions.
Excessive glucocorticoid production associated with chronic or severe stress impairs hippocampal neuronal function and predisposes the organism to neurodegeneration1. Release of cortisol from the adrenal cortex is under tight regulation of this hypothalamic–pituitary–adrenal (HPA) axis. The hippocampus plays a crucial role in regulating HPA axis2 and excessive glucocorticoid production disrupts the regulatory circuit that connects the hippocampus and the hypothalamus.
Age-related disorders are associated with downregulation of glucocorticoid receptors (GR) in the hippocampus, and subsequent desensitization of the regulatory feedback to the hypothalamus3. Accordingly, in a large study of elder humans aged 50–70 years, elevated salivary levels of cortisol were found to be correlated with poor cognitive function4. Increased glucocorticoid activity has also been associated with greater hippocampal atrophy and memory impairment in the elderly3. This is probably a consequence of dendritic retraction and hippocampal dysfunction that we have shown to occur upon chronic stress1. Moreover, higher cortisol levels have been also associated with more rapid Alzheimer’s disease (AD) progression5 and systemic administration of glucocorticoids or stress were shown to potentiate memory impairments, hippocampal damage, β-amyloid formation and Tau accumulation in transgenic AD mice6,7,8.
In the recent years, multiple lines of evidence have suggested an association between adenosine modulation and stress response. In particular, activation of the adenosine A2A receptor (A2AR) was shown to contribute to the stress response by inducing corticosterone secretion9 and by mimicking GR effects10. Moreover, we have recently shown that oral administration of an A2AR antagonist restores morphological, behavioral, and synaptic deficits induced by HPA-axis dysfunction in rodents1. As observed for HPA axis, we and others have demonstrated that A2A receptors are dysregulated in the rat or human brain upon aging and AD11,12,13.
There is a striking parallel between A2AR over-activation/over-expression and impaired GR receptor function, as evidenced by the similar ability of A2AR and GR antagonists to improve cognitive deficits as well as to mitigate amyloid and Tau pathologies reminiscent of AD14,15,16,17. Altogether, such observations strongly suggest that A2AR over-activation and GR dysfuntion are key events in age-related hippocampal deficits and raise the possibility that both pathways might be interconnected.
In the present study, we provide the first demonstration of the instrumental impact of A2AR modulation of GR function, a mechanism never hypothesized before. We specifically report that A2AR overexpression in forebrain neurons is sufficient to promote HPA-axis dysfunction, namely loss of plasmatic corticosterone circadian oscillation, and reduced GR hippocampal levels, both being age-related phenotypes18. Further, we show that A2AR activation modulates GR-induced deficits in hippocampal synaptic plasticity, increasing susceptibility to GR activation. Finally, we demonstrate that A2AR modulation impacts GR nuclear translocation and transcriptional activity.
Materials and Methods
All experimental procedures were carried strictly within the rules of the Portuguese official veterinary department, which complies with European Directive 2010/63/EC and the Portuguese law transposing this Directive (DL 113/2013); and approved by the Instituto de Medicina Molecular Internal Committee and the Portuguese Animal Ethics Committee (Direcção Geral de Veterinária). Environmental conditions were kept constant: food and water ad lib, 21 ± 0.5 °C, 60 ± 10% relative humidity, 12 h light/dark cycles, 2 to 3 animals per cage. The animals were killed by decapitation after anesthesia under halothane atmosphere. Male WT Sprague-Dawley and Tg(CaMKII-hA2AR) rats were used in the 8–14 week-old age range for all the experiments described.
Generation and maintenance of transgenic animals
Transgenic rats overexpressing the human adenosine A2A receptor (A2AR) under the control of the CaMKIIα promotor, tg(CaMKII-hA2AR), were generated by microinjection of a linearized DNA construct into the male pronucleus of Sprague–Dawley rat zygotes with established methods19. The construct contained a full-length human A2AR cDNA cloned into an expression vector with the 8.5 kb mouse CaMKIIα promoter20 and a polyadenylation cassette of bovine growth hormone (see Fig. 1a). Sprague Dawley wild type (WT) male rats were used as controls. Genotyping: Transgenic rats were identified by PCR (30 cycles, 58 °C annealing temperature) of their genomic DNA isolated from ear biopsies by the use of the transgene-specific primers CaMKII-hA2A and rat β-actin primers as an internal control (Invitrogen, see Table 1). Breeding efficiency and litter size was not affected in tg(CaMKII-hA2AR) animals. The average weight of the animals was also similar between WT 282.9 ± 37.7 g and tg(CaMKII-hA2AR) 286.7 ± 22.8 g at the age tested.
RNA extraction and quantitative real-time PCR analysis (RT-qPCR)
Total RNA was extracted and purified using the RNeasy Lipid Tissue Mini Kit (Qiagen) for tissue samples and with NucleoSpin RNA kit (Macherey-Nagel) for neuronal cultures. RNA quality was assessed by NanoDrop 2000 (Thermo Scientific) analysis (A260/A280 ≈ 2; 260/235 >1.8). Total RNA (2 μg) was reverse-transcribed using random primers and SuperScript™ First-Strand Synthesis System for RT-PCR (Invitrogen). RT-qPCR analysis was performed on a ABI 7900 HT (Applied Biosystems) using a home mix consisting of 100 mM Tris pH 8.3, 6 mM MgCl2, 1 mg/ml BSA, 4 mM dNTPs, 0.66x SYBR Green and 1x ROX reference dye. PPIA (cyclophilin A) and β-actin were used as reference genes for human tissues whereas PPIA, Rpl13A (ribosomal protein L13A) and Pgk1 (phosphoglycerate kinase 1) were used for rat tissues and PPIA (cyclophilin A) for primary cultures. The relative expression of target genes was determined by the comparative CT method21.
Blood was collected from the tail in animals previously handled to minimize stress and without anesthesia at two different time points, 8 AM, and 8 PM as in1. The plasma was isolated by centrifugation at 2000 g, 4 °C for 15 min and corticosterone quantified by radioimmunoassay using the rat corticosterone 3H kit (MP Biomedicals), according to the manufacturer’s protocol.
Dissection and tissue collection
After decapitation the brain was rapidly removed and the hippocampi were dissected free in ice-cold Krebs solution composed of (mM): NaCl 124; KCl 3; NaH2PO4 1.25; NaHCO3 26; MgSO4 1; CaCl2 2; and glucose 10, previously gassed with 95% O2 and 5% CO2, pH 7.4). One hippocampus was used for electrophysiological recordings, the remaining brain was separated by areas (Hippocampus, cortex, cerebellum, striatum, hypoyhalamaus) and rapidly frozen in liquid nitrogen for further analysis.
10–14 weeks old WT and tg(CaMKII-hA2AR) rats treated either with vehicle or KW 6002, were first handled for 5 days before testing in the behavior assays. The Y-maze was performed in a two-trial recognition test in a Y-shaped maze with 3 arms (each with 35 cm length x 10 cm width x 20 cm height), angled at 120°; on the first trial (learning trial), the animal explored the maze for 10 min with only two arms opened (start and other arm); after 1 h, the animal is re-exposed to the maze for 5 min (test trial) with the novel arm available, the preference for the novel arm is considered a measure of short-term reference memory. The number of transitions was used to evaluate motor performance. The maze was cleaned with a 70% ethanol solution between each animal. Rat tracings during the learning task were continuously monitored by an automated tracking system (Smart 2.5, PanLab, Barcelona) and the time spent exploring each arm was quantified.
Slices (400 μm thick) were obtained from the same animals used for CORT analysis and behavior testing, with a McIlwain tissue chopper, left to recover for at least 1 h in Krebs solution and field excitatory postsynaptic potentials (fEPSPs) were recorded as previously described1 in the CA1 stratum radiatum. Long term potentiation (LTP, 100 Hz, 1s) was recorded as previously described at 32 °C with a constant flux of 3 mL/min1. Whenever indicated, drugs were preincubated at 32 °C. The intensity of the stimulus was maintained during the induction protocol. LTP was quantified as the % of change in the average slope of the fEPSP measured during the 5 data points immediately preceeding the induction of LTP, comparing to the fEPSP measured from 46 to 60 min (5 data points of 8 averages each) after LTP induction. In each individual experiment, the same LTP-inducing paradigm was delivered to each pathway. Hippocampal slices were incubated with dexamethasone 100 nM (at 32 °C), for the time periods indicated (20′ and 60′), always followed by 60 min resting prior to recording, time necessary to ensure nuclear translocation (as controled in cell cultures) and gene-dependent effects. Antagonists (50 nM SCH58261; 100 nM RU486) were applied 15–20 min before treatment and agonists (CGS21680, 30 nM) at the same time.
Nuclear/cytoplasmic fraction enrichment was performed by differential centrifugation. Samples were homogenized with a 29G syringe and centrifuged at 1000 g for 10 min. The supernatant is the cytoplasmic fraction; the pellet was resuspended in 100 μL of sucrose buffer (0.32 M sucrose, 50 mM Tris, pH 7.6), homogenized and centrifuged again to ensure a minimum contamination with cytoplasm. 150 μL of 1.5x sample buffer (350 mM Tris, 30% glycerol, 10% SDS, 600 mM dithiothreitol and 0.012% bromophenol blue, pH 6.8) were added to the nuclear fraction and 15 μL were used for immunoblot detection. The cytoplasmic fraction was prepared with 20 μL of sample and 5 μL of 5x sample buffer. Tissue homogenates of WT and tg(CaMKII-hA2AR) were prepared from frozen samples. Briefly samples were homogenized by sonication in immunoprecipitation-assay (RIPA) buffer (50 mM Tris, 1 mM EDTA, 150 mM NaCl 0.1% SDS, 1% NP 40, pH 8.0)22. Protein was quantified using the BioRad Protein DC assay based on Lowry23. The appropriate volume of sample was completed with sample buffer.
Samples were denatured by heating to 95 °C for 5 min or at 70 °C for 30 min for A2AR detection. Samples and molecular weight markers were resolved by SDS-PAGE (8% or 10% for resolving and a 5% for stacking gels) in denaturing conditions and electro-transferred to PVDF membranes (Millipore). Membranes were blocked with 5% non-fat dry milk in TBS-T (Tris buffer saline with 0.1% Tween-20, 200 nM Tris, 1.5 M NaCl). After washing with TBS-T, membranes were incubated with primary antibody in TBS-T with 3% BSA. Secondary antibody incubation was in 5% non-fat dry milk in TBS-T. Primary antibodies were rabbit GR specific M20 (1:750/1:1000 sc-1004, Santa Cruz Biotechnology), rabbit lamin A/C specific (1:2000, #2032, Cell Signaling), rabbit pan-cadherin specific (1:20000, abcam ab6529) rabbit αTubulin specific (1:2000, abcam, ab4074), mouse GAPDH specific (1:1000, ambion, AM4300) and mouse A2AR specific (1:2000, Upstate/Millipore - 05-717), secondary antibodies conjugated with horseradish peroxidase were goat, rabbit, or mouse specific antibodies (Santa Cruz Biotechnology, Heidelberg, Germany). Chemiluminescence detection was performed with ECL-PLUS western blotting detection reagent (GE Healthcare) using X-Ray films (Fujifilm). Optical density was determined with Image-J software.
N1E-115 mouse neuroblastoma cells (CRL-2263) were cultured in Dulbecco’s modified Eagle’s medium (DMEM) without pyruvate supplemented with 10% (v/v) fetal bovine serum (FBS), 100 U/ml penicillin-streptomycin, and 2 mM L-glutamine (Gibco). Cells were plated into 6-well plates for 24 h to reach 60% confluence before transfection with Exgene 500 (Euromedex). Briefly 4 μg of pGL3(GRE)3_TK_Luc (GRE_Luc) plasmid (kindly given by Dr. Philippe Lefevbre, Inserm U1011) were mixed in 400 μL of non-supplemented DMEM with 20 μL of Exgene 500 (the mix volume/well) and incubated for 15 min at RT. Cells were incubated for 3 h with the transfection mix before completing the volume to 3 mL. Drug treatments were performed 24 h after transfection in two technical replicates. Primary neuronal cultures. Cortical neurons from 18 days Sprague Dawley rat embryos (Harlan, Barcelona) were cultured according to Valadas24. Briefly, the embryos were collected in Hanks’ balanced salt solution (HBSS) and rapidly decapitated. Meninges and white mater were removed and whole cortices were fragmented and cells were isolated by trypsinization in HBSS Ca2+/Mg2+ (1 mM/1 mM, 0.025% trypsin) and centrifuged at 200 rpm. Cells were washed with HBSS Ca2+/Mg2+ supplemented with 10% FBS and resuspended in Neurobasal medium. Cells were plated on poly-L-lysine-coated coverslips in 6-well plates at density of 1 × 106 cells/well. Neurons were grown in Neurobasal medium with 2% B-27 supplemented with 25 μM glutamate, 0.5 mM glutamine and 2 U/mL penicillin/streptomycin, in the absence of any positive selection for neurons. The medium was replaced at day 4 (without glutamate). Drug treatments were performed at day 8, 1 h after replacing the medium by Neurobasal without B27. All cells were kept in a 5% CO2 humidified incubator at 37 °C.
Cell treatments were performed as in Valadas24. Drug treatments were vehicle matched to drugs, with a max DMSO of 0.001%. Briefly, mouse neuroblastoma N1E115 cells were treated with dexamethasone 100 nM for 24 h; antagonists (SCH58261 10–100 nM, KW6002 30 nM and RU486 100 nM) were applied 15–20 min before treatment and agonists (CGS21680, 10–50 nM) were co-applied with dexamethasone. After treatment cells were washed in ice-cold PBS and processed for luciferase assay. Primary neuronal cultures were treated with dexamethasone 100 nM for different periods of time 0, 5, 10, 15, 30, 60, 90 min; the A2AR antagonist SCH58261 (50 nM) was applied 15–20 min before dexamethasone. After treatment, cells were washed in ice-cold PBS and resuspended in 200 μL of sucrose solution (0.32 M sucrose, 50 mM Tris, pH 7.6) supplemented with protease inhibitors (Roche). Hippocampal slices incubated with dexamethasone 100 nM (at 32 °C), for the time periods indicated (20′ and 60′), always followed by 60 min resting prior to recording, time necessary for nuclear translocation (as controled in cell cultures). Antagonists (50 nM SCH58261; 100 nM RU486) were applied 15–20 min before treatment and agonists (CGS21680, 30 nM) at the same time. In vivo therapy: KW6002 (istradefylline, a selective A2AR antagonist) or vehicle were orally administered in the drinking water (5 mg/kg/day) to WT male rats as described1.
Luciferase activity was evaluated with the luciferase assay system (Promega) according to the manufacture’s procedure. Briefly, N1E115 cells were lysed in 150 μL luciferase cell culture lysis reagent for 15 min at 4 °C. The supernatant was collected after 2 min, centrifuged at 12,000 g at 4 °C and 5 μL were used for the assay, each technical replicate was assayed in duplicate. Luciferase activity was measured on a Mithras Microplate Reader LB 940 (Berthold Technologies).
The A2AR selective antagonist 2-(2-furanyl)-7-(2-phenylethyl)-7H-pyrazolo[4,3-e][1,2,4]triazolo[1,5-c]pyrimidin-5-amine (SCH58261), the A2AR selective agonist 4-[2-[[6-amino-9-(N-ethyl-β-D-ribofuranuronamidosyl)-9H-purin-2-yl]amino]ethyl]benzene propanoic acid (CGS21680) and the GR antagonist (11β,17β)-11-[4-(dimethylamino)phenyl]-17-hydroxy-17-(1-propynyl)-estra-4,9-dien-3-one (RU486) were purchased from Tocris Cookson, UK. These drugs were diluted in the assay solution from 5 mM or 10 mM (for RU486) stock aliquots made in DMSO stored at −20 °C. The GR agonist (11β,16α)-9-fluoro-11,17,21-trihydroxy-16-methylpregna-1,4-diene-3,20-dione,9α-fluoro-16α-methyl-11β,17α,21-trihydroxy-1,4-pregnadiene-3,20-dione,9α-fluoro-16α-methylprednisolone (dexamethasone) was purchased from Sigma (Spain), diluted from 10 mM stock in DMSO and stored at −20 °C. The A2AR selective antagonist, (E)-8-[2-(3,4-dimethoxyphenyl)vinyl]-1,3-diethyl-7-methyl-3,7-dihydropurine-2,6-dione (KW6002, istradefylline) was synthesized according to a published procedure25. The purity of the product was determined by HPLC analysis coupled to electrospray ionization mass spectrometry and was greater than 98%. For in vitro assays a fresh 10 mM stock solution in DMSO was prepared, used only for 1 week and stored at −20 °C. All other reagents used were of the highest purity available either from Merck, Germany or Sigma Aldrich, Spain.
Values presented are mean ± SEM of n experiments. To test the significance of the differences between groups in Western blotting experiments, unpaired Student’s t-test was used. To compare within slice drug treatments in LTP, paired Student’s t-test was used. In all other experiments, when comparing 3 or more groups a one-way ANOVA was used, followed by a Bonferroni’s Multiple Comparison post hoc test. For the analysis of the primary neuronal cultures and corticosterone levels, a two-way ANOVA repeated measures test was used. Values of P < 0.05 were considered to be statistically significant.
Overexpression of A2AR in forebrain neurons impairs HPA-axis and plasticity, rescued by an in vivo anti-A2AR therapy
We have previously demonstrated that HPA-axis function and GR hippocampal levels could be restored through blockade of A2AR activation1. We now tested if the forebrain A2AR overexpression – similar to what is found in aged and AD human brain - could be sufficient to drive HPA-axis dysfunction and hippocampal synaptic impairments. We generated transgenic rats that selectively overexpress human A2AR in neurons under the control of the CaMKIIα promoter [Tg(CaMKII-hA2AR); Fig. 1a] that display cognitive impairments26. As expected, expression of A2AR was achieved in forebrain areas (Fig. 1b), mainly in the hippocampus and cortex, though we also detected a slight increase A2AR mRNA levels in other areas of the nervous system (Fig. 1b). Peripheral expression assessed in the spleen and adrenal glands did not reveal significant changes in A2AR protein levels (Fig. 1c). We have observed a temporal profile of protein overexpression in the hippocampus from 2 week-old onwards reaching a plateau at 12 week-old with a 4.9 fold increase in immunoreactivity (Fig. 1d), which is comparable to levels found in physiological aging27. Importantly, there was no change in adenosine A1 receptor levels in the hippocampus of Tg(CaMKII-hA2AR) animals (Fig. 1e).
We found that A2AR overexpression triggered a decrease in GR protein and mRNA levels in the hippocampus (Fig. 2a,b), as well as an increase in CRH mRNA in the hypothalamus, compared to WT animals (Fig. 2c). Additionally, transgenic animals present higher basal levels of corticosterone in the morning (AM) and a loss of the corticosterone circadian oscillation characteristic of WT rats (Fig. 2d). These features are typical of aged animals18. In order to understand the impact of A2AR overexpression in hippocampal function, fEPSP recordings were performed in the CA1 pyramidal neurons, while stimulating the Schaffer collaterals projections.
Hippocampal slices from [tg(CaMKII-hA2AR)] have a decreased post-tetanic potentiation (PTP), whereas long-term potentiation (LTP) was not significantly changed (Fig. 2e,g) compared to WT animals (at 32 °C). To unmask any effect hidden by endogenous adenosine release - which inhibits LTP- we have decreased the adenosine endogenous levels28, by lowering the perfusion temperature to 30.5 °C. In these conditions, we were able to reveal an overexcitability of LTP (Fig. 2f,g) in [tg(CaMKII-hA2AR)] animals, similar to what we previously described in the aged hippocampus29. Accordingly, the input-output curve from Tg(CaMKII-hA2AR) rats is shifted to the left compared to WT animals (Fig. 2h). We then investigated if in vivo long-term therapy with an A2AR blocker would revert synaptic impairments induced by A2AR overexpression. The treatment of [tg(CaMKII-hA2AR)] rats for 1 month with the selective A2AR antagonist, KW6002 [5 mg/kg/day, orally, in the same dose we have shown to rescue GR expression upon stress1], rescued LTP amplitude in [tg(CaMKII-hA2AR)] to values close to those of WT animals (Fig. 2e,g). Accordingly, when tested for short-term reference memory, using the modified Y-maze test, Tg(CaMKII-hA2AR) animals performed worse than wild type (WT), revealing no preference for the novel arm. This effect was rescued by one-month oral tratment with KW6002 (5 mg/kg/day) (Fig. 2i). We found no changes on the total number of transitions, which could compromise the tests and indicate striatum-associated secondary effects.
Overexpression of A2AR increases hippocampal susceptibility to dexamethasone
We tested if A2ARs were involved in the known GR gene-dependent effects on synaptic plasticity by applying dexamethasone for the time periods indicated, followed by one-hour resting prior to recording, to allow gene-dependent effects. Prolonged exposure (for 60 min) of WT hippocampal slices to a synthetic GR agonist, dexamethasone (100 nM), abolished long-term potentiation (LTP) (Fig. 3a,b) an effect prevented by the GR antagonist, RU486 (Fig. 3b). Interestingly, the blockade of A2AR with SCH58261 (50 nM) prevented these GR-induced effects as efficiently as RU486 (Fig. 3a,b), whereas neither GR nor A2AR antagonists alone had an effect on control (vehicle-treated) slices (Fig. 3b). We then tested whether, conversely, A2AR overexpression increased susceptibility to dexamethasone. Using a shorter exposure - 20 min - to dexamethasone (100 nM), we found that dexamethasone treatment had no impact on LTP magnitude in WT animals (Fig. 3c,d). In contrast, such shorter exposure time was sufficient to decrease LTP magnitude in tg(CaMKII-hA2AR) animals (Fig. 3d,e), an effect completely prevented by the GR antagonist RU486 (Fig. 3f).
We further explored whether the A2AR signaling in these animals followed an aging-like pattern. In Tg(CaMKII-hA2AR) animals, A2AR tonically increase excitatory transmission, an effect revealed by the inhibitory effect of the A2AR selective antagonist SCH58261 (50 nM) on basal synaptic transmission (Fig. 4c), that was not observed in WT animals. The effect of CGS 21680 (30 nM), a selective A2AR agonist on basal synaptic transmission was much higher in tg(CaMKII-hA2AR) than in WT animals (Fig. 4a,b). This effect was abolished by H89, a protein kinase A (PKA) inhibitor, but not by GF 109203x, a protein kinase C (PKC) inhibitor (Fig. 4d). To evaluate if the tonic adenosine inhibitory tonus was altered in tg(CaMKII-hA2AR) animals, we activated A2AR while blocking A1R with a selective antagonist, DPCPX (50 nM). A1R blockade did not prevent A2AR effects on basal synaptic transmission (Fig. 4e). Finally we explored if the A2AR- A1R crosstalk, shown to disappear in normal aging30, is lost in the rats that overexpress A2AR. While in WT animals, A1R activation by CPA (30 nM) causes a strong inhibition of synaptic transmission that is attenuated when A2AR are simultaneously activated with CGS 21680; in tg(CaMKII-hA2AR) animals, the A2AR activation did not modify A1R mediated effects (Fig. 4f,g).
Adenosine A2AR regulates GR transcriptional activity and translocation
We then evaluated if this increased sensitivity to dexamethasone in tg(CaMKII-hA2AR) animals is linked to an ability of A2AR to alter transcriptional regulation by GR. Therefore, mouse neuroblastoma N1E115 cells were transiently transfected with the plasmid pGL3(GRE)3_TK_Luc (GRE_Luc) that contains the glucocorticoid response elements (GRE, at which GR binds to regulate gene transcription) coupled to the luciferase gene. Exposure to dexamethasone (100 nM for 24 h) triggered the expected increase in luciferase expression, an effect blocked by the GR selective antagonist RU486 (100 nM; Fig. 5a). Notably, A2AR blockade with SCH58261 (10–100 nM) or KW6002, (30 nM) reduced the dexamethasone-induced increase of luciferase expression (Fig. 5b). Conversely, A2AR activation with the selective agonist CGS21680 (10–50 nM), increased the dexamethasone-induced luciferase expression (Fig. 5c). Finally, even in the absence of an exogenous GR activation, A2AR blockade decreased luciferase expression (Fig. 5d) while A2AR activation increased it (Fig. 5e); this effect of A2AR on the endogenous GR activity was prevented by RU486 (Fig. 5f).
In addition, we explored the ability of A2AR to control the nuclear translocation of GR in primary cortical neuronal cultures. As expected, dexamethasone (100 nM) induced a significant enrichment of GR in the nuclear versus cytoplasmic fraction in a time-dependent manner, which was maximal after 90 minutes; this effect was completely inhibited upon A2AR blockade (Fig. 5g), in line with data obtained in N1E115 cells.
Overexpression of A2AR receptors downregulates the expression of GR target genes
In order to test whether A2ARs are impacting on GR-dependent regulation of endogenous target genes, we evaluated the effect of selective A2AR inhibition on GR-dependent gene regulation in primary neuronal cultures. Quantitative PCR experiments targeting GR-activated genes (GILZ; PER1 and Bcl2)31,32,33, containing GRE motifs34,35,36 were run following 1h incubation with dexamethasone (100 nM), in the presence or absence of the A2AR selective antagonist SCH51280 (50 nM). Arguing for a physiological role for A2AR in modulating GR activity, we found that the increased GILZ, PER1 and Bcl2 mRNA expression induced by dexamethasone was significantly reduced whenever A2AR are blocked with SCH 58261 (Fig. 6a–c).
Finally, we assessed the impact of the age-like GR downregulation and A2AR overactivation seen in the hippocampus of [tg(CaMKII-hA2AR)] (see Fig. 1), on the expression of the same GR-target genes (Fig. 6d). As predicted, we found that expression levels for both GILZ and Bcl2 were decreased in [tg(CaMKII-hA2AR)] compared to WT animals, whereas for PER1 we could not detect significant changes.
Our findings demonstrate for the first time that GR transcriptional activity and nuclear localization are modulated by adenosine A2A receptors, thereby affecting GR function. Importantly, we here show that neuronal A2AR overexpression is sufficient to impair the HPA-axis function and decrease GR hippocampal levels. Furthermore, the combined evidence that A2AR over-expression increases the susceptibility to GR agonists on one hand, and that A2AR blockade prevents the detrimental synaptic effects of GR activation on the other hand suggests that A2AR play a critical role in the control of memory dysfunction by modulating GR expression and activation.
This is the first time that the link between the well-documented hippocampal increase in A2AR expression and the decrease in GR density and associated HPA-axis dysfuntion - features of aging and Alzheimer’s disease (AD) - is established. This also sustains the novel hypothesis that A2AR up-regulation through modulation of GR function is sufficient to trigger synaptic dysfunction and subsequent memory impairments. This novel A2AR-GR interaction has far-reaching implications in multiple pathologies in which corticosteroids play a pivotal role and that are alleviated by A2AR antagonists, notably caffeine.
Stress hormones and HPA-axis dysfunction have long been recognized as a critical feature underlying brain aging and pathology37. Indeed, altered cortisol levels are observed in post-traumatic stress syndrome or major depression38 and elevated salivary levels of cortisol were found to be correlated with poor cognitive function in a large study of humans aged 50–70 years4. Increased glucocorticoid activity has a predominant impact on the hippocampus, which plays an inhibitory role in regulating the HPA axis2 and controls mood and memory39. Thus, chronic exposure to glucocorticoids leads to cell death and hippocampal atrophy40,41 and is associated with memory impairment in the elderly3. Accordingly, recent evidence supports a pivotal role of stress hormones in neurodegenerative diseases, namely in AD42. This is re-enforced by the following observations: 1) administration of the GR antagonist, RU486, reverts multiple features of AD pathology15,16; 2) repeated stress worsens AD-induced deficits43; 3) elevated cortisol levels are associated with a faster disease progression in AD5; 4) systemic administration of glucocorticoids or stress potentiate memory impairments, hippocampal damage, β-amyloid formation and Tau pathology in transgenic AD mice6,8,44.
Interestingly, in aging or in other brain pathologies where a dysfunction of the HPA-axis is present, there is also an upsurge of A2AR in the hippocampus1,11 and their blockade has proven to be beneficial, but the mechanisms remained unknown1,14. We now provide the first demonstration that A2AR modulation of GR may be essential for the efficacy of A2AR blockade, a mechanism never hypothesized before.
The data we present clearly show that a specific increase in A2AR in forebrain neurons is able to impair the stress response system. A2AR neuronal overexpression disturbed HPA-axis and increased plasma corticosterone levels, providing a tentative connection between the adenosine neuromodulation system and the control of GR signaling.
Our observations are in agreement with previous reports that A2AR activation in a model of spinal cord injury mimicked the effects of GR activation in attenuating neuronal damage10. In particular, we showed that A2ARs modulate GR transcriptional activity, an effect reversed by a GR antagonist (Fig. 5a–c). This decrease in GR-dependent transcriptional regulation could be a consequence of the observed effect of A2AR on the sub-cellular localization of GR (Fig. 5g). The reduced nuclear localization upon A2AR inhibition provides a straightforward explanation for the reduced GR-dependent regulation of genes involved in different functions such as apoptosis (Bcl2) and the circadian clock (PER1)32,33. Our results indeed show that the mRNA levels for both GILZ and Bcl-2 are decreased in the hippocampus of the transgenic rats as compared to WT animals, while PER1 levels remain unchanged (Fig. 6d).
PER1, being the only clock gene in the group, exhibits a robust amplitude of rhythmic expression, initiated in the SCN and PVN45. In all other brain regions, such as the hippocampus while still rhythmic, driven by CORT and GRE-dependent effects, have attenuated amplitudes45. Accordingly, other authors failed to see significant rhythmic expression of Per1 mRNA in whole hippocampus of rodents46. One can especulate that although we observe GR downregulation in Tg(CaMKII-hA2AR), the levels are not sufficient to disturb Per1 expression.
The recently described ability of caffeine to impact on the circadian clock47 strengthens our hypothesis since corticosterone circadian oscillation is tightly controlled by clock gene expression. Furthermore, we not only show that A2AR overexpression impairs circadian corticosterone, we also show that A2AR blockade was sufficient to prevent the deleterious impact of a synthetic GR agonist (dexamethasone) on hippocampal synaptic plasticity (Fig. 3a,b). This tight A2AR-GR interaction discloses a new view on how HPA-axis dysfunction emerges, and also supports the therapeutic utility of A2AR antagonists as an important alternative to GR antagonists in reestablishing HPA-axis dysfunction, which occurs in multiple clinical conditions1. In fact, the therapeutic interest of using selective A2AR antagonists against multiple pathologies is increasing and A2AR antagonists have been recently approved as co-adjuvant therapy for Parkinson’s disease48. Various studies also support the ability of caffeine and A2AR blockade to prevent memory impairment in various conditions49, and recent work revealed that caffeine can even have pro-cognitive effects50. A2AR antagonism was also proposed for the treatment of depression and anxiety-like disorders51 in agreement with the decreased incidence of depression in individuals consuming caffeine52. However, the lack of knowledge with regarding the mechanism of action of A2AR antagonists compromised their acceptance for clinical use. The present report shows that A2AR not only regulate HPA-axis function, but also directly modulate GR, which represent key findings for understanding the mechanisms by which A2AR antagonism is effective. Our data complete and strengthen our previous demonstration that A2AR blockade overcame stress effects by reestablishing the HPA-axis and GR levels in the hippocampus1. These findings are critical, not only for possible treatment strategies of the memory dysfunction associated with psychopathologies, but also in the context of aging and other circumstances in which the glucocorticoid response is impaired.
We have also observed that the A2AR synaptic responses in tg(CaMKII-hA2AR) are dependent on PKA activation and that in vivo administration of an A2A blocker rescues the higher LTP recorded in these animals back to WT levels (panel f), Fig. 2). This supports the crucial contribution of the A2A-PKA interplay to LTP induction mechanisms, as proposed recently53. Furthermore, the features of adenosine neuromodulation evaluated in Tg(CaMKII-hA2AR) follow an aging-like profile. Namely, A2AR activation in Tg(CaMKII-hA2AR) animals has a direct effect in basal synaptic transmission, mediated by PKA and not PKC, and independent of A1R inhibition. This reproduces what is observed in aged animals30. Since A1 receptor levels (Fig. 1e) are not changed in transgenic animals, and we obtain the same A2AR effect on synaptic transmission regardless of A1R blockade with DPCPX (Fig. 4e), this does not seem to depend on endogenous adenosine levels. Taken together, the results now presented suggest that an increase in neuronal A2AR is sufficient to drive age-like changes also in adenosine modulation and consequent LTP impairment. In accordance, the LTP magnitude that we obtained from young adult Tg(CaMKII-hA2AR) animals (Fig. 2g) is similar to the LTP observed previously in aged animals29,54,55 which typically present memory deficits in a variety of tasks. The fact that this ‘aged LTP’ was normalized upon A2AR blockade29,54, in a similar way we have shown to occur now in Tg(CaMKII-hA2AR) animals, further reinforces this relationship.
Previous studies already hinted on a direct A2AR-HPA axis link, but data were contradictory. Global knockout of the A2AR was shown to impact on both plasma corticosterone levels and melanocyte stimulating hormone levels of pro-opiomelacortin expression56. This was mainly mediated by a profound effect on adenosine tonic modulation by the deletion of the gene during embryonic and postnatal development, which occurs both centrally and systemically. Exposure to higher levels of corticosterone in early-life, as occurs in these KO animals, induces an independent disruption of the inhibitory hippocampal-hypothalamic feedback control of corticosterone release resulting in chronic higher levels of this hormone in plasma, as we have demonstrated previously1. Due to the intrinsic limitations of that constitutive model, the authors could not prove whether the central effects would simply derive from knocking out A2AR KO in the adrenal glands. Our current model overcomes this confounding factor, since CAMKIIa activity which drives A2AR expression occurs only from postnatal day 4 onwards, mainly in glutamatergic neurons of the forebrain57, thus inducing an age-like physiological pattern of A2AR overexpression (Fig. 1d).
The transducing pathways by which A2AR trigger GR/GRE transcriptional activity remain to be elucidated. Given the complexicity of A2AR signaling58 and its engagement in numerous signalosome protein complexes59, this has been difficult to dissect. A2AR can recruit multiple signaling pathways, being the most common in the hippocampus the cAMP/PKA/CREB, PKC and MAPK pathways60. Our data indicate that A2AR effects are probably associated to the activation of PKA rather than PKC. The effects observed in GILZ and Bcl-2 expression levels in Tg(CaMKII-hA2AR) (Fig. 6d) may thus result either from the modulation of GR actions on the glucocorticoid response element (GRE), or reflect an independent action following A2AR-driven CREB activation and binding to cyclic AMP response elements (CREs). It is known that CREB is a positive activator of Bcl-2 expression in rat61. On the other hand, Per1 gene contains CRE responsive elements and its levels are elevated by activation of CREB62. If A2AR would act directly through CRE in our model, one would expect an up-regulation of both Bcl-2 and Per1 in Tg(CaMKII-hA2AR) animals. However, we observed a clear down-regulation of Bcl-2 and no differences in Per1 expression, which is consistent with a response to the GR downregulation and less GRE-mediated gene expression, rather than a CRE- activation for these two genes. Furthermore, there is no evidence of putative binding sites for CRE in the rat GILZ, neither in the literature nor in databases such as TRED (Transcriptional Regulatory Element Database). Alternatively, a direct protein kinase A (PKA) modulation of GR binding to GRE is also possible, as previously described63. PKA exerts a regulatory role in the activation of multiple nuclear hormone receptors. For example, it has been shown that PKA activates GR-dependent DNA binding in cotransfection studies. In addition, PKA can directly phosphorylate GR in vitro, enhances transcription by the retinoic acid receptor, and regulates dimerization of human estrogen receptor-α (revised in64) It has been shown that PKA associates with GR and potentiates GR-dependent transcription, namely that PKA attenuates GR cross-repression64. Our observation that tg(CaMKII-hA2AR) have an enhanced A2AR-PKA activation associated to an increased sensitivity to dexamethasone favours the latter hypothesis.
Finally, there is an apparent paradox that arises from the present study: the fact that both stress and A2AR upregulation decrease GR levels in the hippocampus, while simultaneously potentiating GR activation. The former is however reconciled by the fact that GR activation is an important pathway to decrease GR expression and activation effects65,66,67, particularly upon chronic exposure to glucocorticoids68 as in chronic stress or aging. Indeed, this combination of GR downregulation and increased levels of cortisol is observed upon aging: while GR mRNA is significantly reduced in several hippocampal subfields (i.e. stratum granulosum and temporal hippocampus proper) of aged cognitively impaired rats compared to young animals, their cortisol levels take significantly longer to return to baseline following an acute stressor, and this was significantly correlated with poorer spatial learning ability18. Therefore, by increasing GR nuclear location and GR-mediated transcription, A2AR not only increase the susceptibility to stress, but also, through the same pathway, contribute to the downregulation of GR. Whether GR are more prone to activation in these conditions due to increased nuclear localization, a higher affinity for DNA, or to faster kinetics, remains to be clarified. Additionally, a redistribution of GR receptors in the hippocampus was also observed after exposure to corticosteroids68 which may also account for a modified susceptibility upon higher circulating corticosterone levels.
In summary, our results provide the first evidence for a functional interaction between GR and A2ARs, revealing that A2AR modulate GR transcriptional activity and nuclear localization directly affecting GR target genes. Moreover, we demonstrate that this interaction impacts GR-mediated effects on synaptic plasticity, and that A2AR blockade prevents the deleterious effects associated with GR activation/function. Our data also reveal that this interaction may be particularly important in situations where A2AR are over-activated, such as in chronic stress or aging, possibly due to an activation of the PKA pathway.
The beneficial effects of A2AR antagonists, namely caffeine, against cognitive impairments may be, at least partially, due to the now reported effects on GR. The expansion of this interaction to the immune response, cell proliferation, tumor response and other cellular functions that imply GR or corticosteroids use in therapeutics, could have an enormous clinical impact.
How to cite this article: Batalha, V. L. et al. The caffeine-binding adenosine A2A receptor induces age-like HPA-axis dysfunction by targeting glucocorticoid receptor function. Sci. Rep. 6, 31493; doi: 10.1038/srep31493 (2016).
The authors acknowledge J. Baião for technical assistance in the animal house and C. Batalha for the figures layout. VLB and DGF were supported by a grant from Fundação para a Ciência e Tecnologia and an EMBO fellowship (Harvard Medical School-GSV lab- exchange period); LVL is an Investigator FCT, funded by Fundação para a Ciência e Tecnologia (PTDC-099853/2009). TFO is supported by the DFG Center for Nanoscale Microscopy and Molecular Physiology of the Brain. MH, LB and DB thank the following supports: LabEx (excellence laboratory) DISTALZ (Development of Innovative Strategies for a Transdisciplinary approach to Alzheimer’s disease), Inserm, CNRS, Université Lille 2, Région Nord/Pas-de-Calais, DN2M, ANR (ADORATAU) and FUI MEDIALZ. DB, CEM and YB are grateful for support by the LECMA/Alzheimer Forschung Initiative (AFI). DB and LVL were recipients of PHC FCT/Pessoa program and are currently recipients of an international exchange program (AAP Internationalisation / LIA) funded by the University of Lille.
This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/