GABA_B receptor activation desensitizes postsynaptic GABA_B and A₁ adenosine responses in rat hippocampal neurones

Abstract

Whole-cell recordings of EPSCs and G-protein-activated inwardly rectifying (GIRK) currents were made from cultured hippocampal neurones to determine the effect of long-term agonist treatment on the presynaptic and postsynaptic responses mediated by GABA_B receptors (GABA_BRs). GABA_BR-mediated presynaptic inhibition was unaffected by agonist (baclofen) treatment for up to 48 h, and was desensitized by about one-half after 96 h. In contrast, GABA_BR-mediated GIRK currents were desensitized by a similar amount after only 2 h of agonist treatment. In addition, presynaptic inhibition mediated by A₁ adenosine receptors (A₁Rs) was unaffected by prolonged GABA_BR activation, whereas A₁R-mediated GIRK currents were desensitized. Desensitization of postsynaptic GABA_BR and A₁R responses was blocked by the GABA_BR antagonist (1-(S)-3,4-dichlorophenylethyl)amino-2-(S) hydroxypropyl-p-benzyl-phosphonic acid (CGP 55845A), but not by the A₁R antagonist cyclopentyldipropylxanthine (DPCPX). GIRK current amplitude could be partially restored after baclofen treatment by either coapplication of baclofen and adenosine, or intracellular infusion of the non-hydrolysable GTP analog 5′-guanylylimidodiphosphate (Gpp(NH)p). Short-term (4–24 h) baclofen treatment also significantly desensitized the inhibition of postsynaptic voltage-gated calcium channels by activation of GABA_BRs or A₁Rs. These results show that responses mediated by GABA_BRs and A₁Rs desensitize differently in presynaptic and postsynaptic compartments, and demonstrate the heterologous desensitization of postsynaptic A1R responses.

Neurones detect a wide variety of signals using G-protein-coupled receptors (GPCRs). Activation of a GPCR by agonist binding promotes nucleotide exchange and the dissociation of heterotrimeric G-proteins into α and βγ subunits. These subunits then bind to downstream effectors ranging from ion channels to protein kinases (Neer, 1995). Many neuronal GPCRs couple to the class of pertussis toxin (PTX)-sensitive G-proteins that inhibit adenylate cyclase, inhibit voltage-gated calcium channels, and activate inwardly rectifying potassium (GIRK) channels. Activation of these receptors on synaptic terminals inhibits neurotransmitter release by inhibiting calcium channels (Thompson et al. 1993; Wu & Saggau, 1997; Miller, 1998), whereas activation of the same receptors on cell bodies and dendrites decreases excitability by opening GIRK channels (Mark & Herlitze, 2000; Dascal, 2001). One such receptor is the GABA_B receptor (GABA_BR), which is widely expressed in the central nervous system (Billinton et al. 2001; Couve et al. 2000).

Responses mediated by GPCRs often desensitize after prolonged receptor activation, and desensitization is thought to, at least partly, underlie physiological tolerance (Ferguson & Caron, 1998; Bohn et al. 2000). Although the acute effects of GABA_BR activation are well known, relatively little is known about desensitization of the responses mediated by these receptors. Therefore, we have studied the agonist-induced desensitization of responses mediated by presynaptic and postsynaptic GABA_BRs. We found that responses mediated by postsynaptic GABA_BRs desensitize rapidly (2 h), whereas responses mediated by GABA_BRs located at presynaptic terminals desensitize very slowly (48 h). This differential desensitization is similar to that which occurs for presynaptic and postsynaptic A₁ adenosine receptors (A₁Rs) after prolonged activation of A₁Rs (Wetherington & Lambert, 2002). Surprisingly, prolonged activation of GABA_BRs also produced a heterologous desensitization of responses mediated by postsynaptic but not presynaptic A₁Rs. Activation of GIRK channels and inhibition of postsynaptic calcium channels were both blunted by chronic GABA_BR activation, suggesting a site of action upstream from these effector molecules. These results reinforce the idea that presynaptic and postsynaptic compartments differ substantially with respect to the regulation of GPCRs, and demonstrate unidirectional heterologous desensitization of postsynaptic GPCR-mediated responses.

METHODS

Cell culture and chronic drug application

All procedures were carried out in strict accordance with a detailed protocol approved by the Committee on Animal Use for Research and Education (CUARE) of the Medical College of Georgia, and in accordance with National Institutes of Health guidelines for euthanasia of rat neonates (http://oacu.od.nih.gov/arac/euthmous.htm). The animal care facility and programme at the Medical College of Georgia are accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International. The Medical College of Georgia is a registered research facility with the United States Department of Agriculture (USDA No. 57-R-0002), and has an Office of Laboratory Animal Welfare OLAW Assurance Statement (A3307-01; dated 3rd April 2000) on file. Briefly, newborn (postnatal day 0) Sprague-Dawley rat pups were decapitated, and their brains were rapidly dissected into ice-cold saline solution. The hippocampi were dissected and then digested with papain, and dissociated neurones were plated onto collagen/polylysine microislands, as described previously (Wetherington & Lambert, 2002; Bekkers & Stevens, 1991). Growth medium contained minimal essential medium supplemented with B-27 (GIBCO), serum extender (Becton Dickinson), 5 % defined fetal bovine serum (Hyclone), 0.6 % glucose, 1 mm pyruvate and 0.5 mm glutamine. Neurones were treated chronically with drugs (or vehicle) by adding sterile-filtered stock solutions directly to culture dishes, which were then returned to the incubator (37 °C; 5 % CO₂) for the appropriate time. Recordings were made less than 40 min after dishes were removed from the incubator and washed with drug-free external solution and no more than two recordings were made from any one culture dish.

Recording solutions and electrophysiology

Whole-cell patch-clamp recordings were made from isolated (one neurone per microisland) neurones after 14-18 days in vitro. For synaptic (autaptic) and GIRK current recordings, electrodes were filled with a solution containing (mm): 140 potassium gluconate, 5 KCl, 0.2 EGTA, 10 Hepes, 3 MgATP and 0.3 Na₂GTP (pH 7.2, ≈295 mosmol (kg H₂O)⁻¹). In some experiments GTP was replaced with 0.6 mm 5′-guanylylimidodiphosphate (Gpp(NH)p). The external solution contained (mm): 150 NaCl, 2.5 KCl, 10 Hepes, 10 glucose, 1.5 CaCl₂ and 2.5 MgCl₂ (pH 7.2, ≈310 mosmol (kg H₂O)⁻¹). In some experiments the external potassium concentration was increased (to 6 or 30 mm) by replacing NaCl with KCl. For calcium current recordings, electrodes were filled with a solution containing (mm): 100 CsCl, 40 TEA-Cl, 0.2 EGTA, 10 Hepes, 3 MgATP and 0.3 Na₂GTP (pH 7.2, ≈295 mosmol (kg H₂O)⁻¹), and the external solution contained (mm): 150 NaCl, 2.5 KCl, 10 Hepes, 10 glucose, 3 CaCl₂, 2 MgCl₂, 0.2 BaCl₂ and (m): 0.5 TTX, 10 6,7-dinitroquinoxaline-2,3-dione (DNQX) and 10 d(−)-2-amino-5-phosphonopentanoic acid (pH 7.2, ≈310 mosmol (kg H₂O)⁻¹). All recordings were made at room temperature. For synaptic recordings neurones were held at −60 mV and depolarized above 0 mV with 2 ms square commands every 5 s. For GIRK current recordings, the membrane potential was ramped from −100 to −10 mV at a rate of 0.18 mV ms⁻¹, or stepped to −90 mV. For calcium current recordings, neurones were held at −80 mV and stepped to 0 mV for 40 ms. Currents evoked by this protocol were subjected to P/4 leak subtraction prior to analysis. Currents were digitized and recorded with a multifunction input/output board and WinWCP software (provided by Dr J. Dempster, Strathclyde University, Glasgow, UK). Drugs were applied via a fused silica tube (i.d. 200 μm) connected to multiple reservoirs. Numerical values, plots and bar graphs are expressed as means ± s.e.m., and statistical comparisons were made using Student's unpaired t test or ANOVA, unless specified otherwise. Concentration-response curves were fitted to the Hill equation.

RESULTS

Differential desensitization of presynaptic and postsynaptic GABA_BR-mediated responses

EPSCs were evoked in hippocampal neurones grown on substrate microislands by triggering unclamped action potentials. Adenosine (100 μM) or the selective GABA_BR agonist baclofen (50 μM) reversibly decreased the amplitude of EPSCs by ≈80 % (Fig. 1A). Previous studies have shown that A₁Rs and GABA_BRs inhibit EPSCs in hippocampal neurones by decreasing neurotransmitter release, largely by inhibiting presynaptic calcium channels (Thompson et al. 1993; Wu & Saggau, 1997; Miller, 1998). In addition to these presynaptic effects, application of either adenosine or baclofen reversibly activated postsynaptic inwardly rectifying potassium (GIRK) channels, which produced an inward current during voltage commands to −100 mV (Fig. 1B; 6 mm external potassium). We used this system to study the effect of chronic GABA_BR activation on responses mediated by presynaptic and postsynaptic GABA_BR and A₁Rs.

Figure 1. Differential desensitization of presynaptic and postsynaptic responses mediated by GABA_BRs and A₁Rs after prolonged GABA_BR activation.

A, examples of EPSCs evoked under control conditions, in the presence of a submaximal concentration of the GABA_B receptor (GABA_BR) agonist baclofen (5 μM), and a maximal concentration of baclofen (50 μM) shown superimposed. Recordings were made from control neurones (top panel), or neurones treated with baclofen (50 μM) for 4 h. Presynaptic inhibition mediated by GABA_BRs was unaffected by baclofen pretreatment. Calibration bars are 20 ms and 1 nA. B, examples of current traces evoked by hyperpolarizing voltage commands (from −60 to −90 mV) in the same cells as A. Activation of the inwardly rectifying potassium (GIRK) current was substantially desensitized by baclofen treatment. The calibration bars are 20 ms and 0.5 nA. C, summary of data from experiments like those shown in A (n = 6-28). Only presynaptic inhibition produced by 5 μM baclofen after 4 or 24 h and that produced by 50 μM baclofen after 24 h was significantly different (P < 0.01) from control presynaptic inhibition. D, summary of data from experiments like those shown in B, from the same cells as C. The GIRK current evoked by baclofen or adenosine was significantly different (P < 0.01) from control at all times after baclofen treatment. Baclofen at 5 μM produced more than half the maximal GIRK current in control neurones, and less than half of the maximal current in baclofen-treated neurones.

In the first set of experiments cultures were treated with either 50 μM baclofen or vehicle and returned to the incubator at 37 °C for 2-24 h. We then tested presynaptic and postsynaptic receptor function using saturating concentrations of adenosine and baclofen (50 μM and 100 μM, respectively), and a subsaturating concentration of baclofen (5 μM). Presynaptic inhibition mediated by activation of GABA_BRs was modestly but significantly decreased after prolonged GABA_BR activation (Fig. 1A and C). For example, the presynaptic inhibition produced by 50 μM baclofen was 77 ± 1 % (n = 14) after 24 h of baclofen treatment compared to 87 ± 1 % (n = 28) in control cells (P < 0.01). The presynaptic inhibition produced by 5 μM baclofen was also significantly smaller (P < 0.01) after 4 and 24 h of baclofen treatment (Fig. 1C), although once again the difference was relatively small. Presynaptic inhibition was not significantly smaller (P > 0.01) at all other timepoints (< 24 h), and the presynaptic inhibition produced by adenosine was unaffected at all timepoints after baclofen treatment.

In contrast to these small effects on presynaptic inhibition, activation of postsynaptic GIRK channels was greatly reduced after as little as 2 h of baclofen treatment (Fig. 1B and D). At all time points measured, baclofen treatment significantly reduced the amount of GIRK current activated by both GABA_BRs and A₁Rs (P < 0.01; Fig. 1D). For example, 50 μM baclofen activated 252 ± 21 pA of GIRK current in control cells (n = 28) compared to 70 ± 17 pA (n = 16) in cells treated with baclofen for 24 h. In addition to decreasing the maximal baclofen-induced GIRK current, chronic GABA_BR activation also decreased the sensitivity of this response. As shown in Fig. 1D, 5 μM baclofen is greater than the EC₅₀ in control cells, and less than the EC₅₀ in treated cells. Contrary to our expectations, prolonged GABA_BR activation also produced a heterologous desensitization of postsynaptic A₁R-mediated responses. This desensitization was less robust than the homologous desensitization of GABA_BR responses, but was still highly significant (e.g. 168 ± 13 pA, n = 26vs. 80 ± 15 pA, n = 16 after 24 h, P < 0.01). Thus, prolonged baclofen treatment produced both homologous desensitization of postsynaptic GABA_BR responses, and heterologous desensitization of postsynaptic A₁R responses.

We then performed experiments to determine whether baclofen-induced desensitization of postsynaptic GABA_BR and A₁R responses was mediated by persistent activation of GABA_BRs. Neurone cultures were treated as before with baclofen or the selective GABA_BR antagonist (1-(S)-3,4-dichlorophenylethyl)amino-2-(S)hydroxy propyl-p-benzyl-phosphonic acid (CGP 55845A; 1 μM) together with baclofen for 4 h. As shown in Fig. 2, baclofen alone desensitized postsynaptic GABA_BR and A₁R responses as before, and this desensitization was completely prevented by the addition of CGP 55845A (n = 14). Pretreatment with CGP 55845A alone for 4 h had no effect on postsynaptic GIRK responses (n = 6; Fig. 2). We specifically wanted to rule out the possibility that endogenous adenosine activated (and desensitized) postsynaptic A₁R responses in the presence of baclofen. We therefore treated cultures with baclofen and the selective A₁R antagonist cyclopentyldipropylxanthine (DPCPX; 1 μM). DPCPX had no effect on the baclofen-induced desensitization of GABA_BR or A₁R responses (n = 6; Fig. 2). These results suggest that the desensitization of GABA_BR and A₁R responses after baclofen treatment are the result of GABA_BR activation, and do not involve activation of A₁Rs. Finally, we wanted to test the reversibility of baclofen-induced desensitization of postsynaptic GIRK responses following agonist removal. Cultures were treated with baclofen for 24 h, at which time the agonist-containing medium was removed and replaced with drug-free conditioned medium for 24 h. GABA_BR- and A₁R-mediated GIRK currents recorded from these cells (n = 8) were not significantly different from controls, suggesting that postsynaptic desensitization can be completely reversed within 24 h (Fig. 2).

Figure 2. Desensitization of postsynaptic GABA_BR and A₁ adenosine receptor (A₁R) responses is mediated by activation of GABA_BRs.

Summary of data plotting the GIRK current amplitude evoked by baclofen (50 μM) or adenosine (100 μM) in control neurones, neurones treated with 50 μM baclofen for 4 h, neurones treated with the GABABR antagonist 1-(S)-3,4-dichlorophenylethyl)amino-2-(S) hydroxypropyl-p-benzyl-phosphonic acid (CGP 55845A; 1 μM) alone, neurones treated with baclofen and CGP 55845A for 4 h, neurones treated with baclofen and the A₁R antagonist cyclopentyldipropylxanthine (DPCPX; 1 μM), and neurones treated with baclofen for 24 h then washed with baclofen-free medium for an additional 24 h. Homologous and heterologous desensitizations were blocked by preventing activation of GABA_BRs, but not by preventing activation of A₁Rs. The number of experiments in each group is given in parentheses. * P < 0.05, ** P < 0.01.

To determine whether the presynaptic inhibition mediated by GABA_BRs or A₁Rs was completely resistant to chronic activation of GABA_BRs, neurones were exposed to baclofen for extended periods of time (up to 96 h). In this series of experiments, more than 48 h of baclofen treatment was required to significantly (P < 0.01) reduce presynaptic inhibition. After 48 h of baclofen treatment, the presynaptic inhibition mediated by GABA_BRs (50 μM baclofen) was not different from control (75 ± 5 vs. 84 ± 2 %, respectively, n = 15, P > 0.05; Fig. 3A). This result was confirmed in a separate set of cells using a range of baclofen concentrations (0.1-100 μM). The EC₅₀ for GABA_BR-mediated presynaptic inhibition was 4.28 μM (n ≥ 8) in control cells, and 6.69 μM (n ≥ 8) in cells treated with baclofen for 48 h (Fig. 3B). This result suggests that the lack of observable desensitization at this time point was not the result of a receptor reserve. However, after 96 h of baclofen treatment, presynaptic inhibition was reduced to 46 ± 6 % (n = 5; P < 0.01). In contrast, baclofen treatment did not significantly alter inhibition of EPSCs by A₁Rs, even after 96 h (Fig. 3A). Taken together, these experiments and those shown in Fig. 1 demonstrate that GABA_BR-mediated presynaptic inhibition is largely unaffected by chronic receptor activation for up to 48 h. The presynaptic inhibition mediated by GABA_BRs is decreased by more prolonged receptor activation, but presynaptic inhibition mediated by A1Rs is unaffected. Thus, presynaptic and postsynaptic GABA_BR and A₁R responses are differentially regulated by persistent GABA_BR activation.

Figure 3. Homologous desensitization of presynaptic inhibition mediated by GABA_BRs occurs after 72 h.

A, percentage presynaptic inhibition produced by baclofen (50 μM) and adenosine (100 μM) plotted as a function of baclofen treatment time. GABA_BR-mediated presynaptic inhibition was significantly desensitized after 72 h (* P < 0.01), whereas A₁R-mediated presynaptic inhibition was not significantly diminished. The number of experiments in each group is given in parentheses. B, concentration-response curves plotting percentage presynaptic inhibition versus baclofen concentration generated from control neurones (n ≥ 8; EC₅₀ = 4.28 μM), and neurones treated with baclofen for 48 h (n ≥ 8; EC₅₀ = 6.69 μM). The lack of observable desensitization after 48 h was not due to the presence of a receptor reserve.

Desensitization of postsynaptic GIRK currents is partly overcome by liberation of additional βγ subunits

Homologous and heterologous desensitization of GABA_BR- and A₁R-mediated GIRK currents following baclofen exposure could be the result of a change in the number or function of the receptors, G-proteins, or the effector ion channels. However, many of the possible changes in GIRK channel function would decrease the maximal GIRK current without shifting agonist sensitivity. A shift in sensitivity such as we observed is consistent with a defect in either βγ delivery to the channels (due to a change in receptor or G-protein number or function), or a decrease in the sensitivity of GIRK channels to βγ subunits, but not with a change in the number of GIRK channels or their gating. GIRK currents mediated by saturating concentrations of adenosine and baclofen occlude each other in cultured hippocampal neurones, suggesting that these receptors couple to a common pool of channels (Sodickson & Bean, 1998; Wetherington & Lambert, 2002). We therefore reasoned that if chronic baclofen treatment impaired the delivery of βγ subunits to the channels, or decreased the sensitivity of GIRK channels to βγ subunits, then baclofen- and adenosine-evoked currents should become additive. We therefore examined GIRK currents evoked by saturating concentrations of baclofen or baclofen and adenosine combined in control and baclofen-treated neurones. In agreement with previous studies, the ratio of baclofen-induced current to that induced by baclofen together with adenosine was 0.94 ± 0.05 (n = 9) in control neurones; currents evoked by baclofen and baclofen plus adenosine were not significantly different (P > 0.05; paired t test). However, this ratio was 0.65 ± 0.05 (n = 6) in cultures treated with baclofen for 4 h, and 0.41 ± 0.05 in cultures treated for 24 h (n = 6); currents evoked by baclofen and baclofen plus adenosine were significantly different in baclofen-treated cells (P < 0.01; paired t test). The GIRK current activated by combined application of baclofen and adenosine was smaller in cells treated for 4 h compared to controls (205 ± 52 vs. 241 ± 29 pA), but this difference was not statistically significant (P > 0.05). However, the combined GIRK current was significantly smaller in cells treated for 24 h (107 ± 40 pA; P < 0.01). These results suggest that after chronic baclofen treatment, neither activation of GABA_BRs nor A₁Rs is able to completely activate postsynaptic GIRK channels, and that activation of both receptors can partially overcome desensitization. These results are consistent with a defect in βγ delivery to GIRK channels, or the sensitivity of these channels to βγ subunits.

In an attempt to discriminate between a change in receptor activation of G-proteins and G-protein activation of GIRK channels, we next measured receptor-independent activation of GIRK currents in control and baclofen-treated cells. If desensitization was mediated solely by a change in receptor function, we reasoned that the receptor-independent activation of GIRK channels would be unaffected in baclofen-treated cells. If, on the other hand, desensitization was mediated solely downstream of receptor function, we predicted that receptor-independent activation of GIRK channels would be impaired. To activate GIRK channels in a receptor-independent manner, the GTP in the intracellular solution was replaced with 600 μM Gpp(NH)p, a non-hydrolysable analogue of GTP. Neurones were held in whole-cell voltage-clamp mode for approximately 15 min in an external solution containing 30 mm potassium. Receptors were activated with agonist infrequently in these experiments, so that the activation of GIRK channels resulted from spontaneous (rather than receptor-activated) nucleotide exchange. In neurones perfused internally with a GTP-containing internal solution, the holding current remained constant throughout the recording, as did baclofen-evoked current and the current blocked by 100 μM barium, a GIRK channel blocker (Fig. 4A). In contrast, in neurones perfused internally with a Gpp(NH)p-containing internal solution, the holding current increased throughout the recording, baclofen-evoked current decreased, and barium-sensitive current increased (Fig. 4B). We then compared currents that were reversibly evoked by baclofen at the outset of these recordings (presumably before Gpp(NH)p had replaced endogenous GTP) and barium-sensitive current at the end of these experiments in control and baclofen-treated neurones. As predicted based on the results shown above, baclofen-evoked currents were significantly desensitized in baclofen-treated neurones (316 ± 44 pA, n = 7) compared with controls (999 ± 148 pA, n = 11; P < 0.01). In the same cells, the barium-sensitive GIRK current (following Gpp(NH)p infusion) was also significantly reduced in baclofen-treated neurones (1026 ± 201 pA) compared with controls (1765 ± 291 pA; P = 0.046; Fig. 4C). The latter result suggests that the baclofen-induced desensitization of GIRK currents is not solely due to a defect in receptor function. However, it was notable that the receptor-activated GIRK current was desensitized to a far greater extent than was receptor-independent GIRK current. Baclofen activated 69 ± 14 % of the current activated by Gpp(NH)p in control neurones, but only 39 ± 6 % of the current activated by Gpp(NH)p in baclofen-treated cells, suggesting there may also be a defect in receptor activation of G-proteins.

Figure 4. A non-hydrolysable GTP analogue partially restores activation of GIRK current in baclofen-treated neurones.

A, continuous recording of membrane current in a neurone loaded with intracellular GTP (30 mm external potassium). Application of 50 μM baclofen (up arrows) produced an inward GIRK current, whereas application of 100 μM barium (down arrows) produced an outward current, presumably by blocking the tonic inward current. B, continuous recording of membrane current in a neurone loaded with intracellular 5′-guanylylimidodiphosphate (Gpp(NH)p; 0.6 mm), a non-hydrolysable GTP analogue. Neurones loaded with Gpp(NH)p developed a barium-sensitive inward current that occluded baclofen-induced currents over the course of 10-15 min. Calibration bars are 0.5 nA and 2 min. Both neurones in A and B were untreated controls. C, summary of baclofen- and Gpp(NH)p-induced currents in control (n = 7) and baclofen-treated (24 h; n = 11) cells. Gpp(NH)p-induced current was defined as the barium-sensitive current after Gpp(NH)p infusion. Gpp(NH)p evoked a proportionally larger current in baclofen-treated neurones, but the absolute magnitude of this current was significantly smaller than that evoked in control neurones. * P < 0.05.

Desensitization of postsynaptic GIRK currents is upstream of GIRK channels: GABA_BR and A₁R inhibition of calcium channels

The results of the experiments described above suggest that baclofen treatment impairs either delivery of βγ subunits to GIRK channels, or the sensitivity of these channels to βγ subunits. Since postsynaptic voltage-gated calcium channels also bind (and are inhibited by) βγ subunits (Dascal, 2001), we measured inhibition of whole-cell calcium currents by activation of somatodendritic GABA_BRs and A₁Rs in baclofen-treated and control neurones. If desensitization of GIRK currents resulted from a defect in GIRK channel sensitivity to βγ subunits, we predicted that inhibition of postsynaptic voltage-gated calcium channels would be unaffected by chronic baclofen treatment. Using a caesium-based intracellular solution and an external solution designed to isolate calcium currents, neurones were held at −80 mV and stepped to 0 mV for 40 ms. The currents evoked by this protocol were completely abolished by cadmium (100 μM; data not shown). Although voltage clamp was clearly compromised in these neurones, we were able to reliably observe reversible receptor-mediated inhibition of calcium currents. In agreement with numerous previous studies in cultured hippocampal neurones, baclofen and adenosine inhibited peak calcium currents by 25 ± 2 and 21 ± 2 %, respectively, in control neurones (n = 16; Fig. 5). In neurones treated with baclofen for 4 h, baclofen inhibited calcium currents by 16 ± 3 % and adenosine inhibited calcium currents 17 ± 2.1 % (n = 12). Inhibition of calcium currents by GABA_BRs was significantly reduced compared to controls (P < 0.05), but inhibition by A₁Rs was not significantly different (P = 0.18). As was the case with activation of GIRK channels, inhibition of calcium currents by baclofen and adenosine was occlusive. The ratio of inhibition by baclofen to inhibition by both baclofen and adenosine was 0.95 ± 0.5 in control cells (n = 9), and a paired t test showed no significant difference between the inhibition produced by baclofen and that produced by baclofen and adenosine combined (P = 0.25). In contrast, the ratio of inhibition by baclofen to inhibition by combined baclofen and adenosine was 0.77 ± 0.5 in cells treated with baclofen for 4 h (n = 9), and a paired t test showed a significant difference between inhibition produced by baclofen and inhibition produced by baclofen and adenosine combined (P < 0.05). Baclofen treatment for 24 h resulted in no additional desensitization of inhibition of calcium currents by GABA_BRs (16 ± 2 %), but produced significant desensitization of calcium current inhibition by A₁Rs (12 ± 2 %; P < 0.01, n = 8). These results suggest that delivery of βγ subunits to both GIRK channels and voltage-gated calcium channels is impaired after prolonged GABA_BR activation. They also suggest that inhibition of voltage-gated calcium channels by GABA_BRs and A₁Rs is regulated differently in somatodendritic and axonal compartments, as a similar degree of desensitization at presynaptic terminals would have manifested as an observable desensitization of presynaptic inhibition.

Figure 5. GABA_BR activation desensitizes inhibition of the postsynaptic calcium currents mediated by GABA_BRs and A₁Rs.

A, calcium currents evoked by depolarizing voltage commands (−80 to 0 mV) in a control neurone. Traces evoked under control conditions and in the presence of adenosine (100 μM) are superimposed (left), and traces evoked after washing and in the presence of baclofen (50 μM) are shown superimposed (right). B, identical to A, recorded from a neurone that had been treated with baclofen (50 μM) for 24 h. Calibration bars are 50 ms and 1 nA. C, plot of peak calcium current amplitude versus time in the same neurone as A. D, plot of peak calcium current amplitude versus time in the same neurone as B. For panels C and D, the following drugs were applied where indicated by the horizontal bars: 1, 5 μM baclofen; 2, 5 μM baclofen and 0.75 μM adenosine; 3, 100 μM adenosine; 4, 50 μM baclofen. E, summary of experiments identical to those shown in A-D with baclofen treatment times of 4 and 24 h. The number of experiments is given in parentheses; * P < 0.05, ** P < 0.01.

DISCUSSION

We have examined the effect of prolonged GABA_BR activation on presynaptic and postsynaptic GABA_BR- and A₁R-mediated responses. We found that responses mediated by postsynaptic GABA_BRs desensitize much more rapidly than those mediated by presynaptic GABA_BRs. Postsynaptic activation of GIRK channels and inhibition of voltage-gated calcium channels was greatly diminished after a few hours of baclofen treatment, whereas presynaptic inhibition was unaffected for up to 48 h, and only modestly decreased thereafter. This finding is analogous to our previous observation that postsynaptic A₁Rs undergo homologous desensitization more quickly than presynaptic A₁Rs (Wetherington & Lambert, 2002). Spare presynaptic receptors are present at many synapses, and can support unchanged maximal responses despite substantial desensitization (North & Vitek, 1980; Kouznetsova et al. 2002). Indeed, presynaptic and postsynaptic GPCR responses (including those mediated by GABA_BRs) have been shown to be differentially sensitive to PTX (Dutar & Nicoll, 1988; Colmers & Pittman, 1989), consistent with idea that excess signalling machinery is present at presynaptic terminals. However, the present results show that the potency of baclofen to produce presynaptic inhibition was unchanged (Fig. 1 and Fig. 3), ruling out the possibility that desensitization of presynaptic GABA_BR signalling was masked by a receptor reserve. Our results are in agreement with those of Blanchet & Lüscher (2002), who recently reported differential desensitization of presynaptic and postsynaptic responses mediated by μ-opioid receptors, although this study did not rule out the presence of a presynaptic receptor reserve. Presynaptic inhibition resulting from activation of GABA_BRs or A₁Rs is mediated largely by inhibition of voltage-gated calcium channels, rather than activation of GIRK channels (Wu & Saggau, 1997). Therefore, a difference in desensitization of presynaptic and postsynaptic responses could be due to a difference in effector ion channel, or a difference in the neuronal compartment. However, since the inhibition of postsynaptic voltage-gated calcium channels by GABA_BRs and A₁Rs desensitized rapidly (Fig. 5), it appears that location is more important than the effector (cf. Blanchet & Lüscher, 2002). Finally, recent studies of presynaptic inhibition mediated by CB1 receptors and μ-opioid receptors in hippocampal neurones have indicated that desensitization of these responses occurs much more slowly than would be predicted from studies in heterologous expression systems (Bushell et al. 2002; Kouznetsova et al. 2002). Taken together, these results suggest that GPCRs in general may be regulated differently in the presynaptic and postsynaptic compartments.

A second important finding of the present study is that prolonged GABA_BR activation desensitized postsynaptic A₁R-mediated responses. This result is in marked contrast to our previous observation that prolonged activation of A₁Rs has no effect on postsynaptic GABA_BR responses, even though the two receptors share common downstream G-proteins and effectors (Wetherington & Lambert, 2002). Heterologous desensitization of GPCR-mediated responses has been observed often, including responses mediated by GABA_BRs and A₁Rs. For example, Nomura et al. (1994) reported that baclofen treatment for 15 min desensitized the inhibition of calcium currents by GABA_BRs and μ-opioid receptors in neonatal dorsal root ganglion neurones. Similarly, inhibition of low-magnesium-induced spiking in hippocampal neurones by activation of A₁Rs is desensitized by prolonged (24 h) activation of CB1 cannabinoid receptors (Kouznetsova et al. 2002). A number of possible mechanisms could account for the heterologous desensitization of GPCR-mediated responses. In the present study, desensitization both of GIRK channel activation and calcium channel inhibition suggests that the site of action is upstream of the effector ion channels, although we cannot rule out an independent regulation of both ion channels. Thus, prolonged activation of GABA_BRs most likely changes either the function of PTX-sensitive G-proteins or the A₁Rs. Our results do not distinguish clearly between these two possibilities. After baclofen treatment, postsynaptic GABA_BR and A₁R responses become additive, and a non-hydrolysable GTP analogue activates a proportionally greater amount of GIRK current. These results suggest that G-protein capacity does not limit postsynaptic responses, and are consistent with a defect in receptor-G-protein coupling. On the other hand, Gpp(NH)p activated less absolute GIRK current in baclofen-treated neurones, consistent with a defect in G-protein-effector coupling, or simply a decrease in the number of GIRK channels. Additional experiments will be required to identify unambiguously the sites of homologous and heterologous desensitization of GABA_BR and A₁R responses.

Relatively little is known about the mechanisms that regulate GABA_BR and A₁R function. However, Couve et al. (2002) recently demonstrated that GABA_BRs in the brain are constitutively phosphorylated by protein kinase A (PKA), that phosphorylation promotes coupling to GIRK channels in hippocampal neurones, and that activation of GABA_BRs promotes dephosphorylation. It is certainly possible that this mechanism contributes to the homologous desensitization of postsynaptic GABA_BR-mediated responses observed here. It is worth noting that this desensitization mechanism depends on the ability of GABA_BR activation to inhibit adenylate cyclase, and that this ability may not be equal in different neuronal compartments due to unequal distribution of PTX-sensitive G-proteins that inhibit (Gα_i) adenylate cyclase and have no effect on (Gα_o) adenylate cyclase (Aoki et al. 1992). As for regulation of A1R function, the role of phosphorylation is controversial (Palmer & Stiles, 1997), although the present results suggest that GABA_BR- and A₁R-mediated responses may be regulated by common mechanisms. Clearly additional experiments will be required to determine the mechanisms of homologous and heterologous desensitization of presynaptic and postsynaptic GABA_BRs and A₁Rs in hippocampal neurones.

When might GABA_BRs be desensitized under physiological conditions? One possibility is that normal patterns of neuronal activity in the intact brain activate GABA_BRs to the extent that responses mediated by these receptors are ‘tonically’ desensitized to some extent, despite the presence of mechanisms that limit the extracellular availability of GABA. This idea is supported by observations that GABA_BR-mediated responses and receptor binding are enhanced in the brain and spinal cord after chronic GABA_BR blockade (Pratt & Bowery, 1993; Malcangio et al. 1993, 1995). In light of the present results, it would be interesting to determine whether these changes occur differentially at presynaptic and postsynaptic sites. Synchronized neuronal activity (e.g. burst firing in the thalamus) is known to strongly activate GABA_BRs (Kim et al. 1997; Jacobsen et al. 2001), and thus periods of such activity might be expected to produce (or enhance) desensitization similar to that studied here. In addition to the possibility that endogenous GABA might desensitize GABA_BR-mediated responses, it is also possible that similar desensitization mechanisms underlie tolerance to the antinociceptive effects of baclofen in animal models (Vaught et al. 1985; Malcangio et al. 1992), as well as the tolerance that develops in patients receiving chronic intrathecal baclofen to alleviate spinal spasticity (Nielsen et al. 2002).

Finally, the functional significance of differential desensitization of GABA_BRs is unknown. Indeed, the precise functional roles of presynaptic and postsynaptic GABA_BRs are largely unknown, in part due to the fact that these receptors appear to be identical and cannot be distinguished using pharmacological tools. Since GABA_BRs are located both at presynaptic and postsynaptic sites on both excitatory (glutamatergic) and inhibitory (GABAergic) neurones in the forebrain, activation of these receptors can produce complex effects on network activity. It will be interesting to determine whether the differential densensitization we have observed in glutamatergic neurones also occurs in GABAergic neurones. It is also important to note that presynaptic and postsynaptic GABA_BRs may be exposed to different amounts of GABA in the intact brain, which could also produce variable desensitization at the two sites.

In summary, we have demonstrated differential agonist-induced desensitization of presynaptic and postsynaptic responses mediated by GABA_BRs. Such desensitization may contribute to the physiological tolerance that occurs during the prolonged administration of baclofen. In addition, activation of GABA_BRs produces a heterologous desensitization of responses mediated by A₁Rs at a site upstream of the effector molecules.