Adenosine is an endogenous neuromodulator that decreases excitability of hippocampal circuits activating membrane-bound metabotropic A1 receptor (A1R). The presynaptic inhibitory action of adenosine A1R in glutamatergic synapses is well documented, but its influence on inhibitory GABAergic transmission is poorly known. We report that GABAA receptor (GABAAR)-mediated tonic, but not phasic, transmission is suppressed by A1R in hippocampal neurons. Adenosine A1R activation strongly inhibits GABAAR agonist (muscimol)-evoked currents in Cornu Ammonis 1 (CA1) pyramidal neurons and in a specific subpopulation of interneurons expressing axonal cannabinoid receptor type 1. In addition, A1R suppresses tonic GABAAR currents measured in the presence of elevated ambient GABA as well as in naïve slices. The inhibition of GABAergic currents involves both protein kinase A (PKA) and protein kinase C (PKC) signaling pathways and decreases GABAAR δ-subunit expression. On the contrary, no A1R-mediated modulation was detected in phasic inhibitory postsynaptic currents evoked either by afferent electrical stimulation or by spontaneous quantal release. The results show that A1R modulates extrasynaptic rather than synaptic GABAAR-mediated signaling, and that this modulation selectively occurs in hippocampal pyramidal neurons and in a specific subpopulation of inhibitory interneurons. We conclude that modulation of tonic GABAAR signaling by adenosine A1R in specific neuron types may regulate neuronal gain and excitability in the hippocampus.
GABA-releasing hippocampal interneurons regulate excitability of postsynaptic neurons via phasic and tonic GABAA receptor (GABAAR)-mediated signaling (McBain and Fisahn 2001; Klausberger and Somogyi 2008). GABAergic phasic transmission shows fast and precisely timed current kinetics generated by synaptic GABAAR. Tonic inhibition is generated by sustained or persistent activity of mainly extrasynaptic (Brickley et al. 1996; Salin and Prince 1996; Semyanov et al. 2003) high-affinity and slowly desensitizing GABAAR (Nusser et al. 1998; Haas and Macdonald 1999; Bianchi and Macdonald 2003; Caraiscos et al. 2004). In the hippocampus, tonic GABAAR-mediated currents have been characterized in pyramidal cells (Bai et al. 2001) and in inhibitory interneurons (Semyanov et al. 2003). Tonic and phasic GABAAR-mediated inhibition also exhibit distinct pharmacological properties (Semyanov et al. 2004; Farrant and Nusser 2005; Mann and Paulsen 2007), and hence these can be selectively modulated (see Farrant and Nusser 2005).
Adenosine, acting through high-affinity A1 receptor (A1R), is a well-characterized endogenous modulator of neuronal activity in the brain (Sebastião and Ribeiro 2009). Adenosine A1R modulates excitatory glutamatergic synapses at both the pre- and postsynaptic site (Boison 2012; Dias et al. 2013). On the contrary, phasic GABAergic transmission in pyramidal cells is not modulated by A1R (Burke and Nadler 1988; Kamiya 1991; Lambert and Teyler 1991; Yoon and Rothman 1991; Cunha and Ribeiro 2000). However, in pyramidal cells, immunohistochemical studies show intense labeling of A1R not only in dendritic glutamatergic synapses, but also in the perisomatic region where synapses are mainly GABAergic and inhibitory (Kasugai et al. 2010). Adenosine A1Rs are also expressed postsynaptically in GABAergic interneurons (Rivkees et al. 1995; Ochiishi et al. 1999). Although phasic GABAAR currents are unaffected by A1R activity, it is unknown whether tonic inhibitory currents (tonic-ICs) in pyramidal cells are modulated by the receptor. In addition, how adenosine A1R acts on disinhibitory signaling, that is, GABAergic transmission in inhibitory interneurons has not been studied.
We report that activation of adenosine A1R suppresses tonic, but not phasic GABAA currents in hippocampal pyramidal cells. In addition, similar suppression is present in a subpopulation of CA1 area inhibitory interneurons, with axonal cannabinoid receptor type 1 (CB1R). The results demonstrate that the A1R has a highly selective influence on GABAergic neurons. The target-specific modulation of tonic GABAAR conductance through A1R has implications in normal brain function as well as for the use of adenosine in antiepileptic therapies (Boison 2012; Duguid et al. 2012).
Materials and Methods
The procedures were identical to those previously used and described elsewhere (Dias et al. 2012). Three- to 5-week-old male Wistar rats (Harlan, Italy) were anesthetized with halothane (Sigma-Aldrich, St Louis, MO, USA) and sacrificed by decapitation in accordance with Portuguese law on animal care and the European Community guidelines (86/609/EEC). The brain was quickly removed and hemisected, and the hippocampus used to obtain transverse slices (300 µm thickness) cut on a Vibratome (Leica VT 1000S; Leica Microsystems, Germany) in ice-cold dissecting solution containing (in mM): 110 sucrose, 2.5 KCl, 0.5 CaCl2, 7 MgCl2, 25 NaHCO3, 1.25 NaH2PO4, 7 glucose, pH 7.4, bubbled with 95% O2/ 5% CO2. Slices were first incubated for 30 min at 35 °C in artificial cerebrospinal fluid (aCSF) that contained (in mM): 124 NaCl, 3 KCl, 1.25 NaH2PO4, 26 NaHCO3, 1 MgSO4, 2 CaCl2, 10 glucose, pH 7.4 (gassed with 95% O2/5% CO2), and used for experiment after recovering in a submerged storage chamber at room temperature (22–24 °C) for at least 60 min.
Individual slices were clamped with a grid in a recording chamber and continuously superfused by a gravitational superfusion system at 2–3 mL/min with aCSF at room temperature.
Unless otherwise stated, drugs were added via the superfusion solution and their final concentration diluted from concentrated stocks.
N6-cyclopentyladenosine (CPA), 1,3-dipropyl-8-cyclopentylxanthine (DPCPX), and 1-[2-[tris(4-methoxyphenyl)methoxy]ethyl]-(S)-3-piperidinecarboxylic acid (SNAP5114) were obtained from Tocris Bioscience (Bristol, UK) and dissolved as 5, 5, and 100 mM stock solutions, respectively, in DMSO (maximal final concentration in aCSF was 0.036% v/v of DMSO and did not affected muscimol-evoked postsynaptic currents (muscimol-PSCs); change to 104.0 ± 2.3% of the baseline, n = 6, P = 0.140). 6-Cyano-7-nitroquinoxaline-2,3-dione (CNQX), dl-2-amino-5-phosphonopentanoic acid (dl-AP5), 2-(3-carboxypropyl)-3-amino-6-(4 methoxyphenyl)pyridazinium bromide (gabazine, SR-95531), tetrodotoxin citrate (TTX), and 1-(4,4-diphenyl-3-butenyl)-3-piperidinecarboxylic acid hydrochloride (SFK89976A) were obtained from Abcam Biochemicals (Cambridge, UK) and dissolved in water as 10, 50, 10, 1, and 100 mM, respectively. Muscimol was obtained from Sigma-Aldrich and dissolved as a 10-mM stock solution in NaOH (10 mM).
Visually guided whole-cell voltage-clamp recordings (Vh = −70 mV) were performed from CA1 neurons using a Carl Zeiss Axioskop 2FS upright microscope (Jena, Germany) equipped with a differential interference contrast-infrared (DIC-IR) CCD video camera (VX44, Till Photonics, Gräfelfing, Germany) and screen and recorded with an EPC-7 electrical amplifier (List Biologic, Campbell, CA, USA). Patch pipettes (4–9 MΩ) were pulled from borosilicate glass capillaries (1.5 mm outer diameter, 0.86 mm inner diameter, Harvard Apparatus, Holliston, MA, USA) with PC-10 Puller (Narishige Group, London, UK).
Whole-cell recordings of muscimol-PSCs were performed with an intracellular filling solution containing (in mM): 125 K-gluconate, 11 KCl, 0.1 CaCl2, 2 MgCl2, 1 EGTA, 10 HEPES, 2 MgATP, 0.3 NaGTP, 10 phosphocreatine, pH 7.3, adjusted with KOH (1 M), 280–290 mOsm; biocytin (Tocris Bioscience; 0.4%) was added in some experiments for post hoc analyses. Muscimol-PSCs were evoked through a micropipette (2–4 MΩ) containing muscimol (GABAAR agonist; 30 µM in aCSF) coupled to a pressure application system (Picopump PV820, World Precision Instruments, Stevenage, UK) and positioned close to the soma of the recorded cell. Single pulses of 10–15 ms and 6–8 psi were applied every 2 min.
Inhibitory postsynaptic currents (IPSCs), miniature IPSCs (mIPSCs), and tonic-ICs were recorded with a pipette solution containing (in mM): 125 CsCl, 8 NaCl, 1 CaCl2, 10 EGTA, 10 HEPES, 10 glucose, 5 MgATP, 0.4 NaGTP, pH 7.2, adjusted with CsOH (50 wt% in H2O), 280–290 mOsm; biocytin (0.4%) was added in some recordings for post hoc structural analyses. IPSCs were evoked as described elsewhere (Chevaleyre et al. 2007) with some alterations. Briefly, stimuli (0.067 Hz, 1–15 µA) were delivered via monopolar stimulation with a patch-type pipette filled with aCSF and positioned in Stratum radiatum, S. oriens, or S. pyramidale, 80–120 µm from the recorded cell. Recordings were performed in the continuous presence of N-methyl-d-aspartate (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)/kainate (KA) receptor antagonists (50 µM DL-AP5 and 10 µM CNQX, respectively).
The mIPSCs were recorded in the presence of NMDA (50 µM DL-AP5) and AMPA/KA (10 µM CNQX) receptor antagonists, as well as TTX (0.5 µM). The events were analyzed off-line using spontaneous event detection parameters of the Mini Analysis software (Synaptosoft, GA, USA).
For tonic-ICs, SFK89976A (GABA transporter (GAT)-1 inhibitor; 20 μM) and SNAP5114 (GAT-3 inhibitor; 20 μM) were added to the aCSF. GABA (5 µM) also added where mentioned. SR95531 (gabazine, a GABAAR inhibitor; 100 µM) was fast applied using a DAD-12 Superfusion System (ALA Scientific Instruments, Farmingdale, NY, USA). The tonic current measurements were performed as described in Glykys and Mody (2007a). Briefly, the digitized recording acquired at 10 kHz (0.1 ms) was binned to 5 ms. Binned data were loaded with Prism Version 5.00 for Windows (GraphPad Software, La Jolla, CA, USA) and an all-point histogram was plotted for every 200 points (every 1 s) and smoothed by Savitzky–Golay algorithm to obtain the peak value. A Gaussian was fitted to the part of the distribution from a point 3 pA to the left of the peak value to the rightmost (most positive) value of the histogram distribution. The mean of the fitted Gaussian was considered to be the mean holding current. This process was repeated for the entire recording. For statistical purposes, the 20- to 30-s period before applying gabazine (in control or CPA conditions) was compared with the 10- to 15-s period in the presence of gabazine (100 µM) under the same drug conditions. For a given neuron, we obtained the magnitude of the tonic current by subtracting the tonic current before perfusing gabazine from that recorded in the presence of gabazine. Slices were incubated for 50 min at room temperature with CPA (30 nM) for test conditions and with DMSO (0.0006%, v/v; same concentration of solvent as in test conditions) for control conditions.
In all recordings, data were low-pass filtered using a 3- and 10-kHz three-pole Bessel filter of an EPC-7 amplifier, digitized at 5 kHz (for muscimol-PSC and IPSCs) or 10 kHz (for mIPSCs and tonic-IC) using a Digidata 1322A board, and registered by the Clampex software version 10.2 (Molecular Devices, Sunnyvale, CA, USA). Series resistance was not compensated during voltage-clamp recordings, but was regularly monitored throughout each experiment with a −5 mV, 50 ms pulse, and cells with >20% change in series resistance were excluded from the data. All membrane potential values given in this study were corrected for liquid junction potential.
Morphologic and Immunohistochemical Analysis
The procedures were identical to those described previously by Oren et al. (2009), with some alterations. Briefly, interneurons were filled with biocytin (0.4%) during whole-cell recordings (at least 30 min). Slices were fixed overnight at 4 °C in 4% paraformaldehyde in 0.1 M sodium phosphate buffer (PB), pH 7.42. During fixation, slices were kept between 2 mixed cellulose ester membrane filter papers (Millipore, Durham, UK) to minimize deformation. Next day, slices were washed thoroughly in 0.1 M PB and stored in PB with 0.05% sodium azide at 4 °C. The permeabilization was made by 3 washes of 10 min each in 50 mM Tris-buffered saline (TBS) with 0.3% Triton X-100 (TBS-X). Slices were mounted in gelatin, re-sectioned to 60–70 µm thick, and neurons were visualized streptavidin conjugated with AlexaFluor 488 (diluted 1 : 1000, Invitrogen, Eugene, OR, USA) or Cy3 (diluted 1 : 2000, Jackson ImmunoResearch Laboratories, Inc., USA) in TBS-X (5 h of incubation) and mounted in Vectashield (Vector Laboratories, Peterborough, UK) under coverslips. Visualized cells were studied under an epifluorescence microscope [see Oren et al. (2009)] and illustrations made from collapsed z-stack images obtained with a laser scanning confocal microscope (Zeiss LSM 510 META, Jena, Germany) and reconstructed with the ImageJ software (v1.43u, NIH, MD, USA; NeuronJ plugin).
Postsynaptic pyramidal cells were identified by their characterized structure with mushroom-like spiny spines on dendrites, and CB1R-positve cells by co-localization of positive CB1R reaction signal in the Biocytin/Streptavidin reaction-visualized axon (Katona et al. 1999; Pawelzik et al. 2002). Basket cells were identified by their characteristic axon arborization inside S. pyramidale [see Nissen et al. (2010)].
Free-floating 60- to 70-µm-thick sections were washed in 50 mM TBS-TX, blocked in 20% normal horse serum (NHS, Vector Laboratories) in TBS-TX, and incubated in primary antibody (CB1R Guinea pig antibody, diluted 1 : 1000, Frontier Science Co., Ltd, Japan) at 4 °C for 48 h. Fluorochrome-conjugated secondary antibodies [indocarbocyanine (Cy3) or indodicarbocyanine (Cy5); Jackson ImmunoResearch Laboratories, Inc., USA] were applied overnight at 4 °C. After another wash in TBS-TX, sections were mounted in Vectashield (Vector Laboratories) under coverslips. Immunoreactivity was evaluated at ×40 objective using a laser scanning confocal microscope (Zeiss LSM 510 META, Jena, Germany) with the LSM software. Micrographs were adjusted for brightness and contrast only. Immunoreactivity was declared negative when fluorescence was not detected in relevant parts of the cell in an area where similar parts of unfilled cells were immunopositive.
Hippocampal slices were prepared as described for electrophysiological recordings and incubated with CPA as described for tonic-ICs. After the incubation period, the tissue (12–14 slices per condition) was stored at −80 °C. Samples were sonicated in 1% NP-40 lysis buffer containing (in mM): 50 Tris–HCl (pH 7.5), 150 NaCl, 5 ethylenediamine tetra-acetic acid (EDTA), 2 dithiothreitol (DTT), SDS 0.1%, and protease inhibitors (Roche). The lysate was incubated on ice and then the supernatant was collected following centrifugation at 16000 × g for 10 min at 4 °C. Protein concentrations were determined using a commercial Bradford assay (Sigma, MO, USA). Total protein (100 μg) was loaded onto a 10% SDS polyacrylamide gel, subjected to gel electrophoresis, transferred to a PVDF membrane (GE Healthcare), blocked in 10% nonfat milk, and probed with an antibody specific for the GABAAR δ subunit (1 : 500, PhosphoSolutions 868-GDN). After washing (3 × 5 min in TBST [10 mM Tris, 150 mM NaCl, and 0.05% Tween 20 in H2O]), blots were then incubated with secondary antibodies conjugated with horseradish peroxidase and bands were visualized with a commercial enhanced chemiluminescence detection method (ECL) kit (PerkinElmer Life Sciences, MA, USA). Values were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) loading control, and the relative intensities were normalized to the control sample. Densitometry of the bands was performed using the ImageJ processing software (NIH, MD, USA).
Data are expressed as the mean ± SEM of n cells from different slices (electrophysiological recordings) or n measurements from independent experiments (immunoblot assay). Statistical significance was either assessed by two-tailed Student's t-test, when comparing 2 groups, or by performing one-way ANOVA followed by Bonferroni's post hoc test for comparison between multiple experimental groups. A P-value of <0.05 was considered to account for significant differences. Analyses were conducted with the GraphPad Software.
Adenosine A1R Inhibits Agonist-Evoked GABAAR-Mediated Currents in CA1 Pyramidal Cells
To investigate whether activation of adenosine A1R influences GABAAR-mediated responses in the postsynaptic neuron, we performed whole-cell patch-clamp recordings (Vh = −70 mV). In a first set of experiments, a selective GABAAR agonist, muscimol (30 µM), was pressure applied close to the soma of the recorded CA1 pyramidal cell (Fig. 1A) eliciting postsynaptic currents (muscimol-PSCs) that were blocked by GABAAR antagonist gabazine (10 µM; Fig. 1B).
We found that the adenosine A1R agonist CPA (30 nM, Moos et al. 1985) decreased muscimol-PSCs and the suppression reached a steady state within 40 min from wash-in of CPA (Fig. 1B). The amplitude of muscimol-PSCs was significantly reduced in 14 of 16 cells tested (effect showing a Gaussian distribution, Shapiro–Wilk test, n = 16), indicating consistency in pyramidal cells (average decrease to 62.1 ± 4.5% of the baseline, n = 16, P < 0.001, t-test; Fig. 1B,C). During CPA wash out, the suppression persisted for at least 40 min (Fig. 1B,C). Data from all tested pyramidal cells are plotted throughout the paper.
In a next set of experiments, we applied a high-affinity A1R antagonist, DPCPX (100 nM, Sebastião et al. 1990), to revert the suppressive effect of CPA on GABAergic currents. This restored muscimol-PSCs in all cells (average to 96.2 ± 3.7% of the original baseline, n = 7, P < 0.001, t-test; Fig. 1D,E), demonstrating that the CPA effect on GABAAR currents is reversible. A lower concentration of CPA (10 nM) was also capable of decreasing amplitude of muscimol-PSCs significantly in 12 of 17 cells (Fig. 1G). In addition, CPA (30 nM) failed to change muscimol-PSCs when washed in the presence of A1R antagonist DPCPX (100 nM; 103.7 ± 1.4% of the baseline, n = 6, P = 0.17, t-test; Fig. 1G). Interestingly, we found a significant increase in muscimol-PSCs following wash-in of DPCPX in naïve slices to 115.3 ± 4.9% of the baseline (n = 6, P < 0.05, t-test; Fig. 1F,G), which suggests tonically activated A1R and suppression of GABAAR-mediated currents in standard physiological conditions.
To confirm that the observed inhibitory action of adenosine A1R on GABAAR currents was not caused via an indirect effect on glutamatergic transmission or axonal GABAergic excitation (Alle and Geiger 2007; Ruiz et al. 2010), we reproduced the experiments in the continuous presence of NMDA and AMPA/KA receptor antagonists (50 µM DL-AP5 and 10 µM CNQX, respectively), and TTX (0.5 µM) to block action potential firing. Indeed, in these conditions, there was a similar suppression of muscimol-PSC by CPA (30 nM) as observed above (decrease in amplitude to 69.5 ± 8.0% of the baseline, n = 8, P < 0.001, t-test; Fig. 1G). Although previous studies have reported that GABAergic synapses may not be directly modulated by A1R (Lambert and Teyler 1991), our results show suppression of agonist-evoked postsynaptic GABAAR-mediated currents in pyramidal cells.
Phasic GABAAR-Mediated Currents Are Not Affected by Adenosine A1R in CA1 Pyramidal Cells
We next explored whether adenosine A1R modulates GABAAR-mediated IPSCs evoked by electrical afferent fiber stimulation. We stimulated in S. radiatum or S. oriens and recorded monosynaptic IPSCs in pyramidal cells in the presence of CNQX (10 µM) and DL-AP5 (50 µM). The IPSCs were fully blocked with gabazine (10 µM) at the end of experiment (Fig. 2A). We found that, in contrast to muscimol-PSCs, synaptic GABAAR IPSCs were not significantly modulated by CPA (30 nM) (89.3 ± 6.4% of the baseline, n = 9, P = 0.14, t-test; Fig. 2A,B). We also studied in separate experiments GABAergic mIPSCs in the presence of CNQX (10 µM), DL-AP5 (50 µM), and TTX (0.5 µM). Wash-in of CPA (30 nM for at least 50 min) failed to change either mIPSC frequency (99.4 ± 2.2% of the baseline, n = 13, P = 0.80, t-test; Fig. 2C,D) or amplitude (100.1 ± 1.2% of the baseline, n = 13, P = 0.96, t-test; Fig. 2C,E), confirming a lack of modulation of synaptic IPSCs by A1R.
Adenosine A1R Suppresses Tonic GABAergic Currents in CA1 Pyramidal Cells
Next, we hypothesized that A1R modulation could be selective to extrasynaptic GABAAR and studied adenosine A1R agonist effects on tonic-ICs in pyramidal cells. Glutamate receptor blockers (CNQX, 10 µM and DL-AP5, 50 µM) and TTX (0.5 µM) were added to the superfusion solution. In addition, to avoid any interference of adenosine receptors upon GAT activity (Cristóvão-Ferreira et al. 2009, 2013), which could indirectly affect tonic-ICs, the GABA transporters blockers, SFK89976A (20 μM; GAT-1 inhibitor) and SNAP5114 (20 μM; GAT-3 inhibitor), were also added to the superfusion solution. Tonic-IC was measured comparing the holding current before and in the presence of gabazine (100 µM; Fig. 3A,B; see Materials and Methods). Consistent with previous reports (Semyanov et al. 2003), pyramidal cells did not express measurable tonic GABAAR-mediated conductance (−3.1 ± 1.1 pA, n = 4), unless the extracellular concentration of GABA was enhanced (Glykys and Mody 2007a) to increase the signal-to-noise ratio. Therefore, in the remaining experiments aiming to evaluate tonic-ICs in pyramidal cells, GABA (5 µM) was added to the superfusion solution. Under such conditions, tonic-ICs were easily visualized (Fig. 3). Interestingly, in the presence of CPA (30 nM, incubated for at least 50 min), tonic-ICs were significantly lower than in control slices (−119.7 ± 12.5 pA, n = 8, for control compared with −57.7 ± 14.8 pA, n = 7, for CPA, P < 0.01, t-test; Fig. 3B–D). These results, taken together with the absence of effect of CPA on afferent-evoked IPSCs and mIPSCs, allow to conclude that adenosine A1R in pyramidal neurons selectively suppress tonic-ICs, known to be mediated by extra- and perisynaptically localized GABAAR (Glykys and Mody 2007b).
Adenosine A1R-Mediated Effect on GABAA Currents Is PKA/PKC-Dependent
Adenosine A1R is Gi/o coupled (Freissmuth et al. 1991; Jockers et al. 1994; Nanoff et al. 1995) and involves signaling cascades that require PKA and in some cases, PKC (Akbar et al. 1994; Cascalheira and Sebastião 1998). GABAAR-mediated currents are affected by activity of both PKA (Kano and Konnerth 1992; Kano et al. 1992; Moss et al. 1992; Robello et al. 1993; Nusser et al. 1999; Poisbeau et al. 1999) and PKC signaling pathways (Poisbeau et al. 1999; Brandon, Jovanovic, Smart, et al. 2002; Bright and Smart 2013). We tested whether activity of those kinases could be involved in A1R suppression of tonic GABAAR currents (Fig. 4A). The PKC or the PKA blockers (GF109203x, 1 μM, or Rp-cAMPs, 100 μM, respectively) were added intracellularly through the whole-cell patch-pipette filling solution. In either situation (intracellular inhibition of PKA or PKC), the effect of CPA (30 nM) on muscimol-PSC was blocked. Amplitude of muscimol-PSCs in the presence of CPA and GF109203x was 97.1 ± 4.3% (n = 6, P = 0.53, t-test; Fig. 4B,C) and in the presence of CPA and Rp-cAMPs 101.0 ± 4.0% (n = 6, P = 0.80, t-test; Fig. 4B,C) of the pre-CPA values. These results show the involvement of both kinases in A1R modulation of GABAergic currents. We then asked if we could uncover a sequence of kinase activation cascade. We measured muscimol-PSC modulation when one of the signaling pathways was activated in the presence of a blocker of the other pathway. First, the adenylate cyclase (AC) activator, forskolin (5 μM, Seamon et al. 1981), was bath applied to activate cAMP/PKA signaling. Forskolin increased the amplitude of muscimol-PSC to 117.5 ± 4.4% of the baseline (n = 4, P = 0.029, t-test; Fig. 4D,F). The effect was similar to the blockade of A1R in naïve slices with DCPCX (see Fig. 1G). Loading the patch pipette with the PKC inhibitor, GF109203x (1 µM), completely prevented forskolin effect on muscimol-PCSs (96.0 ± 4.1% of the baseline, n = 5, P = 0.38, t-test; Fig. 4D,F). These results suggest that PKA signaling is upstream of PKC in the GABAAR current suppression cascade. To further test this idea, we washed-in an activator of PKC (phorbol 12,13 didecanoate, PDD, 250 nM). This suppressed muscimol-PSCs to 54.4 ± 4.8% of the baseline (n = 4, P = 0.002, t-test; Fig. 4E,F), akin to the generated by A1R activation with CPA (see Fig. 1B,C). Adding a PKA inhibitor, Rp-cAMPs to the pipette filling solution, failed to prevent the suppression of muscimol-PSCs by PDD (60.5 ± 8.6% of the baseline; n = 3, P = 0.04, t-test; Fig. 4E,F). Altogether these results show that PKC is downstream to PKA activation in the GABAAR current suppression cascade.
Knowing that GABAARs are substrate for kinases and that PKC activity decreases extrasynaptic GABAAR expression (Bright and Smart 2013), we decided to evaluate whether A1R actions on tonic inhibition could be associated with decreased expression of GABAAR. We performed immunoblot assays against the δ-subunit of GABAAR, a subunit present exclusively in extra- and perisynaptic GABAARs in the hippocampus (Nusser et al. 1998; Wei et al. 2003; Sun et al. 2004; Glykys et al. 2007), therefore most relevant for tonic-ICs. We found that, in slices that had been incubated with CPA (30 nM, for at least 50 min), GABAAR δ-subunit immunoreactivity was significantly decreased to 68.5 ± 9.5% when compared with the control slices (n = 4, P = 0.04, paired t-test; Fig. 4H).
Taken together, these results demonstrate that A1R actions upon GABAergic currents involve postsynaptic signaling requiring both PKA and PKC pathways and suggest that A1R activation leads to inhibition of PKA signaling, releasing PKC activity which then suppresses GABAAR currents (Fig. 4G). Results from immunoblot assays fit this idea, suggesting that A1R-mediated decrease in tonic inhibition is associated with decreased expression of extrasynaptic GABAAR δ-subunit.
Adenosine A1R Suppresses Tonic GABAAR Currents in a Specific Subpopulation of Hippocampal Interneurons
Next, we investigated A1R effects on GABAAR responses in hippocampal interneurons (Fig. 5D). We recorded muscimol-PSCs in CA1 area interneurons whose soma was located in S. radiatum or S. oriens. The interneuron population showed nonparametric distribution in response to CPA (30 nM; Shapiro–Wilk test, n = 17; Fig. 5A), and in fact we found 2 different populations of cells. A subset of interneurons showed a significant and robust suppression of muscimol-PSCs following CPA application (average reduction to 66.3 ± 2.2% of the baseline, n = 7, P < 0.001, t-test; Fig. 5B) similar to that observed in pyramidal cells (see Fig. 1C,D). In the remaining tested interneurons, muscimol-PSC was unchanged by CPA (amplitude 101.2 ± 2.0% of the baseline, n = 10, P = 0.58, t-test; Fig. 5C).
Aiming to identify the characteristics of the CPA responsive interneurons, we discovered that the A1R effect on GABAAR currents correlated with the expression of a specific marker, axonal CB1R, in the studied cells. Recorded interneurons were filled with biocytin and visualized with streptavidin-fluorophore. All successfully visualized cells were tested in immunohistochemical reaction for axonal CB1R expression (Katona et al. 1999; Klausberger et al. 2005; Nissen et al. 2010). Importantly, we found that 9 of 10 cells responding to CPA in muscimol-PSCs were immunopositive for CB1R (CB1R-positive). In CB1R-positive interneurons, average muscimol-PSC inhibition by CPA was to 58.8 ± 5.0% of baseline responses (n = 10, P < 0.001, t-test; Fig. 5E,H,J). Analyses on the laminar distribution of CB1R-positive interneuron axon revealed basket cells (n = 4; Fig. 5F) and dendritic targeting Schaffer collateral-associated cells (Fig. 5G), indicating that GABAAR current modulation by A1R occurs in various types of CB1R-positive interneurons (Somogyi and Klausberger 2005; Lee et al. 2010). Interestingly, the A1R agonist (CPA, 30 nM) failed to significantly suppress muscimol-PSCs in any CB1R immunonegative (CB1R-negative) interneuron. Indeed, muscimol-PSCs in CB1R-negative interneurons were 99.0 ± 1.4% of the baseline (n = 10, P = 0.60, t-test; Fig. 5E,I,J) in the presence of CPA. This population of CB1R-negative neurons included 3 basket cells.
To directly assess A1R-mediated actions on tonic inhibitory responses, we recorded tonic-IC in immuhistochemical-identified CB1R-positive and CB1R-negative interneurons. In the first set of experiments, to allow better comparison with results from pyramidal cells, GABA (5 µM) was added to the aCSF together with GABA transport blockers (SFK89976A, 20 μM and SNAP5114, 20 μM), glutamate receptor antagonists (CNQX, 10 µM and DL-AP5, 50 µM), and TTX (0.5 µM). In these experiments, averaged tonic-ICs recorded from interneurons in control slices were −153.3 ± 10.8 pA (n = 5). In slices incubated with CPA (30 nM for at least 50 min), tonic-ICs were significantly lower than control in 4 of 5 CB1R-positive interneurons (−47.9 ± 7.0 pA, n = 4, P < 0.001, t-test; Fig. 6A–C), but not in CB1R-negative interneurons (−144.1 ± 8.7 pA, n = 5, CB1R-negative in CPA, P = 0.53, t-test; Fig. 6A,C).
We then evaluated if adenosine A1R could also affect tonic transmission in the presence of endogenous concentrations of GABA and recorded tonic-ICs in interneurons without supplying the aCSF with GABA. Contrary to what was observed for pyramidal cells, naïve interneurons showed a significant tonic-IC (−15.4 ± 1.4 pA, n = 4, Fig. 6D–F). Upon incubation with CPA, tonic-IC was clearly smaller in 5 of 7 anatomically identified interneurons (−8.8 ± 1.0 pA, n = 5, in CPA, P < 0.05, t-test; Fig. 6D–F).
Finally, we tested whether, similar to that observed in pyramidal cells, A1R modulation of inhibitory currents in interneurons was restricted to extrasynaptic GABAAR-mediated currents. We recorded electrical stimulation-evoked IPSCs in the CA1 area interneurons. Cells were visualized post hoc and tested for axonal CB1R immunoreaction. Similar to the results obtained with pyramidal cells, A1R activation failed to significantly modulate IPSCs in either CB1R-positive (84.0 ± 5.7% of the baseline, n = 3, P = 0.10, t-test; Fig. 7A–C) or CB1R-negative (96.1 ± 3.6% of the baseline, n = 11, P = 0.3, t-test; Fig. 7A–D) interneurons, indicating a lack of modulation of phasic interneuron inhibition by A1R.
Taken together, the above results show A1R modulation of tonic GABAAR currents in a specific subpopulation of GABAergic interneurons expressing axonal CB1Rs.
The results show that adenosine A1R selectively modulates tonic GABAAR currents generated by extrasynaptic receptors, but has no effect on phasic synaptic GABAAR currents. The modulation is consistent with CA1 pyramidal cells, but present only in a specific population of postsynaptic CA1 GABAergic inhibitory interneurons with axonal CB1R. A1R-mediated modulation requires intracellular PKA/PKC signaling. Sustained A1R activity results in a decreased expression of GABAAR δ-subunit, a key component of extrasynaptic receptors mediating tonic GABAAR currents [see Farrant and Nusser (2005)].
Adenosine has a broad spectrum of modulatory actions in the brain. Through A1R, it acts as an anticonvulsant agent with neuroprotective effects (Sebastião and Ribeiro 2009; Boison 2012). These actions are partly based on suppression of glutamatergic transmission either by presynaptically reducing calcium influx (Scanziani et al. 1992; Yawo and Chuhma 1993) and neurotransmitter release (Schubert et al. 1986; Proctor and Dunwiddie 1987; Barrie and Nicholls 1993) or postsynaptically facilitating potassium currents (Gerber et al. 1989; Thompson et al. 1992) and inhibiting ionotropic glutamatergic receptors (de Mendonça et al. 1995; Li and Henry 2000). Thus, the effect of adenosine via A1R on glutamatergic transmission is well known. A role of adenosine in regulation of inhibitory GABAergic transmission has received much less attention and is much less investigated. This is surprising because already in early 90s, it was demonstrated that adenosine strongly modulates dysynaptic inhibition in the hippocampus, although it has no direct effect on GABAergic synapses to pyramidal cells (Kamiya 1991; Lambert and Teyler 1991; Yoon and Rothman 1991; Thompson et al. 1992).
During the past 2 decades, tonic GABAAR-mediated inhibition has been described in neurons in the hippocampus and in many other brain areas [for review see Semyanov et al. (2004); Farrant and Nusser (2005); Glykys and Mody (2007b)]. Tonic GABAAR-mediated membrane conductance plays a role in regulation of synaptic integration, input to output signal transformation, and firing rate of individual neurons and ultimately overall excitability of the hippocampus (Hamann et al. 2002; Mitchell and Silver 2003; Semyanov et al. 2003; Bright et al. 2007; Rothman et al. 2009). Deregulation of tonic inhibition has also been implicated in pathophysiological conditions including schizophrenia (Damgaard et al. 2011; Gill et al. 2011; Hines et al. 2012), stroke (Clarkson et al. 2010), and epilepsy (Dibbens et al. 2004; Peng et al. 2004; Naylor et al. 2005; Scimemi et al. 2005; Feng et al. 2006; Zhang et al. 2007). This makes tonic GABAergic responses an important target to modulation via endogenous or exogenous drugs. Indeed, neuroactive steroids, ethanol, and some anticonvulsant drugs act on extrasynaptic GABAAR and modulate tonic GABAergic conductance (Stell et al. 2003; Cope et al. 2005; Ferando and Mody 2012). Interestingly, GABAAR responsible for tonic currents and postsynaptic adenosine A1R mainly locate in extra- and perisynaptic areas (Rivkees et al. 1995; Swanson et al. 1995; Ochiishi et al. 1999; Glykys and Mody 2007a), which makes them potential candidates to interact. This idea is further supported by A1R coupling to Gi/o signaling pathways since GABAAR is strongly modulated by PKA- and PKC-mediated phosphorylation (Kano and Konnerth 1992; Kano et al. 1992; Moss et al. 1992; Robello et al. 1993; Nusser et al. 1999; Poisbeau et al. 1999; Brandon, Jovanovic, Smart, et al. 2002; Bright and Smart 2013). We evaluated this possibility by recording afferent-evoked synaptic IPSCs and agonist-evoked GABAAR currents in hippocampal neurons. These 2 ways to generate postsynaptic GABAergic currents allowed us to discriminate responses mediated by synaptic and extrasynaptic GABAAR. Local application of muscimol (a selective GABAAR agonist) through a micropipette positioned close to the recorded cell soma predominantly activates extrasynaptic GABAAR, which are prominent in the perisomatic postsynaptic area (Kasugai et al. 2010). Accordingly, the resulting muscimol-PSC exhibited slow current kinetics characteristic of extrasynaptic GABAAR-mediated responses (Pearce 1993; Banks et al. 1998; Banks and Pearce 2000). As we here report, in all studied pyramidal cells and in a subpopulation of interneurons, the muscimol-evoked GABAAR currents were inhibited by the A1R agonist. In contrast, the A1R agonist failed to change phasic synaptic GABAAR currents generated either by quantal release or by afferent stimulation [see also Kamiya (1991); Lambert and Teyler (1991); Yoon and Rothman (1991); Thompson et al. 1992]. Such selective modulation of tonic GABAAR signaling might be important in controlling neuronal synchronization (Maex and De Schutter 1998; Glykys and Mody 2007b). Our data on the facilitation of muscimol-PSCs by the A1R antagonist in naïve slices demonstrate that endogenous adenosine can tonically suppress extrasynaptic GABAAR conductance. Because adenosine is paracrinally released from neurons and astrocytes (Boison 2006; Haydon and Carmignoto 2006), changes in ambient levels of endogenous adenosine are likely to occur and, therefore, tune peri- and extrasynaptic GABAAR activity. Interestingly, when compared with glutamatergic neurons, interneurons are easily disconnected by hypoxia due to A1R activation (Khazipov et al. 1995), an indication that adenosine release onto GABAergic neurons is higher.
Many signaling mechanisms are involved in the modulation of GABAAR that are relevant to both phasic and tonic inhibition. Various protein kinases phosphorylate serine residues of GABAAR subunits (Brandon, Jovanovic, and Moss 2002), including PKA and PKC phosphorylation mechanism (Moss et al. 1995; Brandon et al. 2001; Brandon, Jovanovic, Smart, et al. 2002). Adenosine A1Rs are coupled to Gi/o proteins (Freissmuth et al. 1991; Jockers et al. 1994; Nanoff et al. 1995), but also affect phospholipase C and phosphoinositol-3-kinase activity (Akbar et al. 1994; Dickenson and Hill 1998; Schulte and Fredholm 2000; Cascalheira and Sebastiäo 1998; Cascalheira et al. 2002). We found that PKA and PKC signaling cascades were responsible for A1R-mediated inhibition of tonic GABAA currents. The results also indicated that A1R-mediated inhibition of AC activity relieves a negative regulation of PKA over PKC. Disinhibition of PKC then promotes suppression of tonic GABAA currents in hippocampal neurons (see Fig. 4G). PKC-mediated phosphorylation of extrasynaptic GABAAR in the hippocampus causes a decrease in their expression level and function (Bright and Smart 2013). Accordingly, we detected that, upon incubation with an A1R agonist, there is a decrease in the expression of a marker of extrasynaptic GABAAR.
All tested pyramidal cells were sensitive to A1R-mediated modulation of tonic GABAergic currents, somehow contrasting what occurs in pyramidal neurons from the somatosensory cortex, which are heterogeneous for the sensitivity to postsynaptic A1R-mediated modulation (van Aerde et al. 2015). Among the interneurons, we show that those that exhibit modulation of tonic GABAA currents by A1R are also immunopositive for CB1R, whereas CB1R-negative interneurons are insensitive to A1R activation. Similar to the pyramidal neurons, A1R-mediated suppression of GABAergic responses in interneurons was significant only for tonic GABAA currents. In the hippocampus, axonal expression of CB1R strongly correlates with expression of cholecystokinin (CCK) in interneurons (Katona et al. 1999). The modulation of tonic GABAAR allows regulation of excitability and signaling through these interneurons (Mitchell and Silver 2003). In fact, low concentration of picrotoxin (1 µM), aimed to predominantly inhibit tonic currents in interneurons, increases spontaneous output from GABAergic cells to pyramidal cells, seen as the increased frequency of spontaneous IPSCs (Semyanov et al. 2003). Discharge of interneurons expressing CCK is coupled to coordinated oscillatory activities in hippocampus in vivo (Klausberger and Somogyi 2008). Firing of hippocampal CCK-positive inhibitory neurons is coupled to synchronous network oscillations in theta (4–8 Hz) and gamma (30–80 Hz) rhythms, which occur during cognitive processes in the hippocampus (Klausberger et al. 2005; Tukker et al. 2007; Lasztóczi et al. 2011). Controlling excitability and discharge by robust tonic GABAAR conductance in these neurons (Pietersen et al. 2009; Oke et al. 2010; Schulz et al. 2012) could allow adenosine A1R modulation of hippocampal rhythm generation and information processing associated with coordinated rhythmic activities.
Adenosine A1R actions decrease hippocampal excitability, and hence adenosine is a suitable endogenous anticonvulsant compound (Boison 2012; Dias et al. 2013). Most documented actions of A1R as an anticonvulsant substance rely on its ability to refrain glutamatergic transmission (Khan et al. 2001; Boison 2012). Here, we demonstrate a direct suppression of tonic GABAergic inhibition by A1R in inhibitory interneurons, therefore highlighting another target for A1R-mediated neuromodulation and excitability control. The resulting reduction in the disinhibition of interneurons caused by A1R-mediated suppression of tonic GABAergic inhibition can increase inhibitory GABAergic output to the hippocampal principal cell population. In parallel, adenosine A1R also reduce tonic GABAergic inhibition in pyramidal cells. However, in low ambient GABA levels, tonic GABAAR inhibition is likely to be more pronounced in interneurons than in pyramidal cells (Bai et al. 2001; Semyanov et al. 2003). Therefore, the net effect of A1R-mediated modulation of tonic GABAAR on hippocampal pyramidal cell excitability may depend on ambient GABA concentrations as well as other conditions that control extrasynaptic GABAAR activation levels in the 2 cell populations (Scimemi et al. 2005; Wlodarczyk et al. 2013). (see Fig. 8).
Ambient GABA and adenosine levels are dynamic in the brain and both are increased during episodes of epileptiform activity (Chin et al. 1995; Berman et al. 2000; Pavlov and Walker 2013). Decreasing tonic GABAAR conductance in pyramidal cells during high ambient GABA levels should increase pyramidal cell excitability. However, during epileptiform discharges when ambient GABA concentrations reach peak, GABAAR currents can turn to depolarizing and excitatory (Köhling et al. 2000; Cohen et al. 2002; Ellender et al. 2014). This means that A1R-mediated suppression of tonic GABAAR conductance in pyramidal cells can also have an antiepileptic effect (Ilie et al. 2012). In contrast, adenosine A2A and A3 receptors may promote excitability in epileptic tissues by exacerbating use-dependent run-down of phasic GABAA currents (Roseti et al. 2009). These opposite actions of adenosine receptors are particularly relevant when planning adenosine-mediated therapies in pathological conditions such as epilepsy.
In conclusion, we here propose that adenosine A1Rs, by changing the inhibitory tonus of neurons without affecting phasic inhibitory synaptic transmission, can homeostatically regulate inhibition and control neuronal gain without disrupting fidelity of synaptic GABAergic inhibition (Pouille and Scanziani 2001; Lamsa et al. 2005). Its selectivity to specific interneuron populations may confer to adenosine an important modulatory action on hippocampal network oscillations that are the critical bases for hippocampal-dependent behavior and cognitive processes.
This work was supported by Fundação para a Ciência e Tecnologia (FCT), Portugal fellowships (to D.M.R.—SFRH/BD/60386/2009 and R.B.D.—SFRH/BPD/89057/2012), and project grant (PTDC/SAU-NMC/110838/2009 to D.M.R., R.B.D., J.A.R., and A.M.S. and EXPL/bim-mec/0009/2013 to D.M.R., R.B.D., S.T.D., J.A.R., and A.M.S.); Faculdade de Medicina, Universidade de Lisboa (to D.M.R., R.B.D., J.A.R., and A.M.S.); the Medical Research Council UK (to K.P.L.); the John Fell OUP Research Fund (to K.P.L.), and the Department of Pharmacology University of Oxford (to K.P.L.). S.T.D. integrates the Portuguese Program for Advanced Medical Education, sponsored by Fundação Calouste Gulbenkian and FCT (SFRH/BDINT/51548/2011).
We acknowledge Dr Wiebke Nissen and Kathryn Newton from the Department of Pharmacology, University of Oxford, for all support with the immunohistochemistry of interneurons. Conflict of Interest: None declared.