INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Adenosine is a naturally occurring purine nucleoside which has been believed to play modulatory roles in a variety of tissues and physiological circumstances (Dunwiddie and Masino 2001). In the CNS, endogenous adenosine acts as an extracellular signal molecule influencing synaptic transmission without acting as a neurotransmitter. The function of adenosine has been determined by occupying specific surface receptors. Four G protein-coupled adenosine receptors have been described, i.e., A₁, A_2A, A_2B, and A₃ receptors (for review, see Fredholm et al. 2001). A₁ receptor is ubiquitous within the CNS, with high levels being expressed in the hippocampus as well as in the cerebral cortex, the brain stem, and the spinal cord (Dixon et al. 1996; Ochiishi et al. 1999).

A₁ receptor activation leads to postsynaptic hyperpolarization, but more importantly, it inhibits the release of a variety of neurotransmitters including glutamate, GABA, and serotonin in the various brain regions (Bagley et al. 1999; Centonze et al. 2001; Chen and van den Pol 1997; Okada et al. 2001; Scanziani et al. 1992; Thompson et al. 1992; Uchimura and North 1991; Ulrich and Huguenard 1995; Wu et al. 1994a). In the hippocampus, presynaptic A₁ receptor activation is known to inhibit neurotransmitter release from glutamatergic synapses, but not GABAergic ones (Dolphin and Archer 1983; Lambert and Teyler 1991; Scanziani et al. 1992; Thompson et al. 1992; Yoon and Rothman 1991). However, most studies, which demonstrate GABAergic transmission is not under A₁ receptor modulation, have been performed in the adult hippocampal slices, slice culture, or cultured hippocampal neurons (Lambert and Teyler 1991; Scanziani et al. 1992; Thompson et al. 1992). In addition, there is convincing evidence for a profound change in the expression pattern of hippocampal A₁ receptors during postnatal development (Ochiishi et al. 1999). These findings tempt us to address whether presynaptic A₁ receptor activation can regulate the probability of neurotransmitter release from GABAergic synapses in immature hippocampal neurons and whether there is a developmental change in A₁ receptor-mediated modulation of GABAergic transmission.

To test this idea, we have used mechanically dissociated hippocampal CA1 neurons retaining functional GABAergic presynaptic nerve terminals, namely "synaptic bouton" preparation (Rhee et al. 1999). This preparation has some advantages such as a simple and reduced system while maintaining native presynaptic functions.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Preparation

Wistar rats (12- to 15-day-old, except where indicated) were decapitated under pentobarbital sodium anesthesia (50 mg/kg ip). The brain was quickly removed and transversely sliced at a thickness of 370 µm using a microslicer (VT1000S; Leica, Nussloch, Germany). Slices were kept in the incubation medium (see Solutions) saturated with 95% O₂-5% CO₂ at room temperature (21-24°C) for 1 h before the mechanical dissociation. For dissociation, slices were transferred into 35-mm culture dishes (Primaria 3801; Becton-Dickinson, Rutherford, NJ) containing the standard solution (see Solutions), and the region of the CA1 was identified under a binocular microscope (SMZ-1; Nikon, Tokyo). Details of the mechanical dissociation have been described previously (Rhee et al. 1999). Briefly, mechanical dissociation was accomplished using a custom-built vibration device and a fire-polished glass pipette oscillating at 50-60 Hz (0.1-0.2 mm). The tip of the fire-polished glass pipette was lightly placed on the surface of the hippocampal CA1 region with a micromanipulator. The tip of glass pipette was vibrated horizontally for approximately 2 min. Slices were removed, and the mechanically dissociated neurons were allowed to settle for 15 min to adhere to the bottom of the dish. Such neurons undergoing dissociation retained short portions of their proximal dendrites.

All experiments conformed to the guiding principles for the care and use of animals approved by the Council of the Physiological Society of Japan, and all efforts were made to minimize the number of animals and any suffering.

Electrical measurements

All electrical measurements were performed using the conventional whole-cell patch-clamp recording mode at a holding potential (V_H) of -60 mV. Membrane voltage was controlled and currents were recorded by a use of patch-clamp amplifier (CEZ-2300; Nihon Kohden, Tokyo). Patch pipettes were made from borosilicate capillary glass (1.5 mm OD; 0.9 mm ID; G-1.5; Narishige, Tokyo) in two stages on a vertical pipette puller (PB-7; Narishige). The resistance of the recording pipettes filled with internal solution was 5-6 M. Electrode capacitance and liquid junction potentials were compensated for, but series resistance was not. Neurons were visualized under phase contrast on an inverted microscope (Diapot; Nikon). Current and voltage were continuously monitored on an oscilloscope (VC-5-6023; Hitachi) and a pen recorder (RECTI-HORIT-8K; Sanei, Tokyo). Membrane currents were filtered at 1 kHz (E-3201A Decade Filter; NF Electronic Instruments, Tokyo), digitized at 4 kHz, and stored on a computer using pCLAMP 8.0 (Axon Instruments). All experiments were performed at room temperature (21-24°C).

Data analysis

Spontaneous miniature inhibitory postsynaptic currents (mIPSCs) were detected and analyzed in preset epochs before, during, and after each experimental condition, using the MiniAnalysis Program (Synaptosoft, NJ). Briefly, the events were automatically screened using an amplitude threshold of 8 pA and were then visually accepted or rejected based on the rise and decay times. In complex waveforms where the event starts to rise before the previous event goes back to the baseline, the baseline for the second event was estimated by extrapolating the decay of the first peak at the location of the double peak. Then the peak amplitude of the second event was determined from this calculated baseline but not from the onset point of event. The average values of mIPSC frequency and amplitude during the control period (10-15 min) were calculated, and the frequency and amplitude of all the events during agonist application (5 min) were normalized to these values. The effect of the drug was quantified as a percentage decrease in mIPSC frequency compared with the control value. Numerical values were reported as means ± SE, using values normalized to the control levels. Possible differences in the amplitude and frequency distribution were tested by Student's paired two-tailed t-test using their absolute values but not normalized ones. Values of P < 0.05 were considered to be significant. On the other hand, the inter-event intervals and amplitudes of a large number of mIPSCs obtained from the same neuron were examined by constructing cumulative probability distributions and compared using Kolmogorov-Smirnov (K-S) test with Stat View software (SAS Institute).

Solutions

The incubation medium consisted of the following (in mM): 124 NaCl, 5 KCl, 1.2 KH₂PO₄, 24 NaHCO₃, 2.4 CaCl₂, 1.3 MgSO₄ and 10 glucose saturated with 95% CO₂-5% O₂. The pH was approximately 7.45. The standard external solution consisted of the following (in mM): 150 NaCl, 5 KCl, 2 CaCl₂, 1 MgCl₂, 10 glucose and 10 HEPES. The Ca²⁺-free external solution consisted of the following (in mM): 150 NaCl, 5 KCl, 5 MgCl₂, 2 EGTA, 10 glucose, and 10 HEPES. These external solutions were adjusted to pH 7.4 with Tris-base. For recording mIPSCs, these external solutions routinely contained 300 nM TTX to block voltage-dependent Na⁺ channels, and 10 µM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) and 20 µM D-2-amino-5-phosphonovaleric acid (AP5) to block ionotropic glutamatergic currents. The ionic composition of the internal (patch pipette) solution was (in mM) 80 Cs-methanesulfonate, 70 CsCl, 2 EGTA, 4 Mg-ATP, and 10 HEPES with pH adjusted to 7.2 with Tris-base.

Drugs

Drugs used in the present study were TTX, CNQX, AP5, 4-aminopyridine (4-AP), N-ethylmaleimide (NEM), EGTA, Mg-ATP, bicuculline, adenosine, forskolin, and 1,9-dideoxyforskolin (dideoxy-forskolin) from Sigma (St. Louis, MO); N⁶-cyclopentyladenosine (CPA), 8-cyclopentyl-1,3-dipropylxanthine (DPCPX), 2-[[2-[4-(2-caboxyethyl)phenyl]ethyl]amino]-N-ethylcarboxamido-adenosine (CGS 21680) and 8-(3-chlorostyryl)caffeine (CSC) from RBI (USA); (Rp)-Cyclic adenosine-3',5'-monophosphothioate sodium salt (Rp-cAMPS) from BIOLOG-Life Science Institute (Hayward, CA). All solutions containing drugs were applied by the Y-tube system which can achieve complete solution exchange within 20 ms (Akaike and Harata 1994).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

GABAergic miniature inhibitory postsynaptic currents

After the mechanical dissociation of the hippocampal CA1 region, we found that the individual neurons were either pyramidal or bipolar in a shape. In all following experiments, pyramidal neurons but not bipolar ones were used for electrical measurement. When a pyramidal neuron was held at a V_H of -60 mV, spontaneous inward currents were observed in the presence of 300 nM TTX, 10 µM CNQX, and 20 µM AP-5 (Fig. 1A). The currents were completely and reversibly blocked by 3 µM bicuculline, indicating that the spontaneous mIPSCs are GABAergic. Figure 1B shows typical spontaneous GABAergic mIPSCs at various V_H values. The chloride equilibrium potential (E_Cl) of these mIPSC, estimated from the I-V relationship, was about 15.5 mV (n = 4), which was almost identical to the theoretical Cl equilibrium potential (19.5 mV) calculated from the Nernst equation using extra- and intracellular Cl concentrations (161 and 70 mM, respectively). Thus the spontaneous events were identified as GABAergic mIPSCs mediated by -aminobutyric acid-A (GABA_A) receptors.

View larger version (30K): [in this window] [in a new window]   Fig. 1. GABAergic miniature inhibitory postsynaptic currents (mIPSCs) recorded from mechanically dissociated hippocampal CA1 neurons. A: typical traces of mIPSCs observed before, during, and after the application of 10 µM bicuculline in the presence of 300 nM TTX, 10 µM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX), and 20 µM D-2-amino-5-phosphonovaleric acid (AP5). Insets: typical traces with an expanded time scale. B: traces of mIPSCs recorded at each VH (a) and their mean amplitude I-V curve (b). In b, each point is the mean from 4 neurons. [Cl]i and [Cl]o represent the intra- and extracellular Cl concentrations, respectively. A straight line is the least-squared linear fit to the mean mIPSC amplitude values at each VH. Each point and error bar shows the mean ± SE.

As shown in later, these GABAergic mIPSCs recorded in the presence of TTX was greatly reduced by adding Cd²⁺ or by removal of extracellular Ca²⁺. This observation is consistent with the possible involvement of Ca²⁺ influx from the external solution even during Na⁺ channel blockade. One possibility to explain this phenomenon is that the GABAergic presynaptic nerve terminals on CA1 neurons may have a somewhat depolarized membrane potential. At depolarized potentials, the spontaneous activation of voltage-dependent Ca²⁺ channels (VDCCs) may result in the spontaneous release of GABA. Thus events that remain in the Ca²⁺-free solution may be classical miniature currents as shown in the previous studies (Capogna et al. 1993; Scanziani et al. 1992). Alternatively, since Ca²⁺-dependent mIPSCs have been also reported in central neurons (Doze et al. 1995; Soltesz and Mody 1995), the synaptic events observed in the present preparation might be Ca²⁺-dependent mIPSCs. However, the reason for the dependency of mIPSCs on extracellular Ca²⁺ and/or VDCCs remains incompletely understood at this time.

Modulation of GABAergic mIPSCs by adenosine

Application of adenosine (10 µM) or CPA (1 µM) decreased the mIPSC frequency in the majority of hippocampal CA1 neurons tested (88 of 133; 66.2%). On washing out adenosine, the mIPSC frequency completely recovered to control levels within 5 min (Fig. 2, Aa and Ab). Mean responses show a sustained decrease in mIPSC frequency (Fig. 2Ab). Figure 2B shows cumulative probability plots for inter-event interval and current amplitude of mIPSCs. Adenosine shifted the distribution curve of mIPSCs frequency to the right, indicating the reduction of mIPSC frequency. The amplitude distribution was not affected. The pooled data (n = 12) show that adenosine decreased the mean mIPSCs frequency to 71.1 ± 2.3% of the control (P < 0.05), but the mean amplitude was not affected (99.9 ± 3.3% of the control, P = 0.79; Fig. 2, Ca and Cb). Together, the results suggest that adenosine acts presynaptically to inhibit the release probability of GABA at these synapses.

View larger version (37K): [in this window] [in a new window]   Fig. 2. Adenosine-induced inhibition of GABAergic mIPSCs. Aa: typical traces of mIPSCs observed before, during, and after the application of 10 µM adenosine. Insets: current traces with an expanded time scale. b, the time course of mIPSC frequency before, during, and after adding adenosine. The number of events in every 10-s period (open circle, presence of adenosine; closed circle, absence of adenosine) was summed and plotted. Each point is the mean ± SE from 12 neurons. B: cumulative distributions for inter-event interval (a; P < 0.01; K-S test) and current amplitude (b; P = 0.36; K-S test) of mIPSCs recorded from the same neuron (1,276 events for the control, and 435 events for adenosine). C: each column is the mean from 12 and 9 cases for P12-15 and P30 hippocampal CA1 neurons, respectively. All amplitudes and frequencies are normalized to those of control mIPSCs. *P < 0.05, n.s; not significant. These definitions are applied to all subsequent figures.

Although many previous attempts have been made to determine whether adenosine can modulate hippocampal GABAergic transmission, the results have been negative (Dolphin and Archer 1983; Lambert and Teyler 1991). However, most of these studies were performed with adult hippocampal neurons. Thus we also tested the effects of adenosine receptor agonists on GABAergic transmission in adult hippocampal CA1 neurons (postnatal 30-day-old, 100-120 g) and found that adenosine (10 µM, n = 9, Fig. 2C) or CGS-21680, a selective A_2A receptor agonist (30 nM, n = 4, data not shown), had no effect on GABAergic mIPSCs.

Effects of adenosine receptor agonists and antagonists

To identify the subtypes of adenosine receptors participating in the decrease of mIPSC frequency, the effects of alternative agonists and antagonists were examined. DPCPX (100 nM), a selective A₁ receptor antagonist, did not alter the GABAergic mIPSCs (Fig. 3B). Although the lack of effect of DPCPX on the basal GABAergic mIPSCs suggests the lack of apparent tonic inhibition of GABAergic transmission by endogenous adenosine, it is difficult to evaluate the tonic inhibition of endogenous adenosine because the present study was performed with dissociated neurons. DPCPX completely blocked the inhibitory action of adenosine on GABAergic mIPSCs (n = 6, Fig. 3, Ab and B). Likewise, CPA (1 µM), a selective A₁ receptor agonist, inhibited mIPSCs (Fig. 3, Ac and C) by reducing the mean frequency to 71.6 ± 6.5% of the control (P < 0.05, n = 6) without affecting the mean amplitude (105.1 ± 4.3% of the control, P = 0.35, n = 6, Fig. 3C). At lower concentrations, however, CPA inhibition was less potent (10 and 100 nM; respectively, data not shown). CGS-21680 (30 nM, n = 4), and CSC (3 µM, n = 4), a selective A_2A receptor antagonist, did not affect either the frequency or the amplitude of mIPSCs (data not shown). Such results indicate that adenosine modulation of GABAergic synaptic transmission might be mediated by presynaptic adenosine A₁ receptors. Therefore given its selectivity for A₁ receptors, CPA was used in the following experiments.

View larger version (45K): [in this window] [in a new window]   Fig. 3. A1 receptor-mediated presynaptic inhibition of GABAergic mIPSCs. A: typical traces of mIPSCs observed before, during, and after the application of 10 µM adenosine in the absence (a) or presence (b) of 100 nM 8-cyclopentyl-1,3-dipropylxanthine (DPCPX), and the application of 1 µM N6-cyclopentyladenosine (CPA) (c). a and b were obtained from the same neuron. B: each column is the mean from 6 neurons. All frequencies and amplitudes of mIPSCs are normalized to the control. C: cumulative distributions for inter-event interval (a; P < 0.01; K-S test) and current amplitude (b; P = 0.36; K-S test) of mIPSCs obtained from the same neuron shown Ac (373 events for the control, and 179 events for CPA). Insets: each column is the mean from 6 neurons. All frequencies and amplitudes of mIPSCs are normalized to the control.

Mechanisms of A₁ receptor-mediated inhibition of GABAergic transmission

Adenosine A₁ receptors are known to be coupled to pertussis toxin sensitive G-proteins (G_i or G_o) (Chen and Lambert 2000; Dunwiddie and Masino 2001). Accordingly, to examine whether the adenosine-mediated inhibition of mIPSC frequency is coupled to G_i/G_o proteins, we utilized NEM, a sulfhydryl alkylating agent (Asano and Ogasawara 1986). Pretreatment of 10 µM NEM for 15 min increased mIPSC frequency to 342.4 ± 55.8% of the control (P < 0.01, n = 6) without affecting the mean amplitude (120.6 ± 15.5% of the control, P = 0.41, Fig. 4, Ab and B). In the presence of NEM, however, the inhibitory effect of CPA on GABAergic mIPSCs was completely occluded to 110.2 ± 3.2% of the NEM condition (P < 0.05, n = 6, Fig. 4, A and B). The results suggest that adenosine A₁ receptors on the GABAergic presynaptic nerve terminals projecting to CA1 pyramidal neurons seem to be coupled to G_i/G_o protein. Functionally, A₁ receptors are known to be associated with inhibition of adenylyl cyclase, inhibition of Ca²⁺ influx, as well as activation of K⁺ channels (Dunwiddie and Masino 2001). In the following study, therefore, we examined possible signal transduction pathways between A₁ receptor activation and the inhibition of spontaneous GABA release.

View larger version (33K): [in this window] [in a new window]   Fig. 4. A1 receptor-mediated presynaptic inhibition of GABAergic mIPSCs is coupled to Gi/Go protein. A: typical traces of mIPSCs observed before, during, and after the application of 1 µM CPA in the absence (a) and presence (b) of 10 µM N-ethylmaleimide (NEM). B: each column is the mean from 6 neurons. All amplitude and frequencies are normalized to the control.

Involvement of K⁺ channels

A₁ receptor activation has been previously suggested to reduce neuronal excitability by activating G protein-coupled inwardly rectifying K⁺ (GIRK) channels (Dunwiddie and Masino 2001). To determine whether presynaptic activation of GIRK channels is responsible for the inhibitory effect of CPA on GABAergic mIPSC frequency, we tested the effects of Ba²⁺, which is known to block the GIRK channels (Birnstiel et al. 1992; Gerber et al. 1989), on CPA-induced inhibition of mIPSCs. Ba²⁺ (1 mM) significantly increased mIPSC frequency to 260.3 ± 29.6% of the control (P < 0.01, n = 6, Fig. 5, Ac and Ba), but also slightly decreased the mean amplitude (80.5 ± 8.9% of the control, P < 0.05, Fig. 5Bb). This increase in mIPSC frequency is consistent with the expected depolarization of presynaptic nerve terminals, which should activate VDCCs. In the presence of Ba²⁺, however, CPA still decreased mIPSC frequency to 64.4 ± 6.3% of the Ba²⁺ condition (P < 0.05, n = 6), without affecting the mean amplitude (Fig. 5B). We also examined the effect of 4-AP, a K⁺ channel blocker (Bagley et al. 1999; Harvey and Marshall 1977), on CPA-induced inhibition of mIPSCs. Application of 100 µM 4-AP greatly increased mIPSC frequency to 310.1 ± 40.6% of the control (P < 0.05, n = 4, Fig. 5, Ab and Ba), without altering the mIPSC amplitude (95.9 ± 7.7% of the control, Fig. 5Bb). In the presence of 4-AP, CPA also effectively depressed mIPSC frequency to 75.9 ± 5.6% of the 4-AP condition (P < 0.05, n = 4) without affecting mIPSCs amplitude (94.7 ± 2.5% of the 4-AP condition, P = 0.80, Fig. 5B). Together, these results suggest that activation of presynaptic K⁺ channels is therefore unlikely to contribute to the A₁ receptor-mediated inhibition of GABAergic mIPSCs.

View larger version (36K): [in this window] [in a new window]   Fig. 5. A1 receptor-mediated presynaptic inhibition of GABAergic mIPSCs is not related to the increase of K+ conductance. A: typical traces of mIPSCs observed before, during, and after the application of 1 µM CPA (a) in the standard solution with 100 µM 4-aminopyridine (4-AP) (b) or 1 mM Ba2+ (c). a and b were obtained from the same neuron. B: each column is the mean from measured neurons (4-AP, n = 4; Ba2+, n = 5). All amplitudes and frequencies are normalized to the respective control. **P < 0.01.

Effect of the Ca²⁺-free external solution

Although the postsynaptic membrane potential in our dissociated neurons can be accurately controlled by voltage clamping, the presynaptic nerve terminals are not under direct control. Since Ca²⁺ influx through VDCCs plays an important part in the release of neurotransmitter from presynaptic nerve terminals (Wu and Saggau 1994b), we tested whether the A₁ receptor-mediated inhibition of GABAergic mIPSCs requires extracellular Ca²⁺ entry. Ca²⁺-free external solution markedly decreased not only mIPSC frequency to 43.4 ± 6.2% of the control (P < 0.05, n = 5), but also mIPSC amplitude (62.2 ± 2.3% of the control, P < 0.05, Fig. 6, A and B). The result suggests that about 60% of GABAergic mIPSCs depend on Ca²⁺ influx from the extracellular sites. In the Ca²⁺-free external solution, CPA failed to alter mIPSC frequency (101.3 ± 3.3% of the Ca²⁺-free condition, P = 0.58, Fig. 6Ba). Thus the A₁ receptor-mediated inhibitory action strongly depends on extracellular Ca²⁺, suggesting a possible involvement of VDCCs.

View larger version (35K): [in this window] [in a new window]   Fig. 6. A1 receptor-mediated presynaptic inhibition of GABAergic mIPSCs is closely related to the Ca2+ influx. A: typical traces of mIPSCs observed before, during, and after the application 1 µM CPA in the standard solution (a) and in the Ca2+-free external solution (b). B: each column is the mean from 5 neurons. All amplitudes and frequencies are normalized to the control.

To address this, the effects of Cd²⁺, a general VDCC blocker, was tested on CPA-induced inhibition of GABAergic mIPSCs. As shown in Fig. 6B, Cd²⁺ (100 µM) also decreased not only mIPSC frequency to 45.6 ± 12.8% of the control (P < 0.05, n = 5), but also mIPSC amplitude (64.7 ± 5.1% of the control, P < 0.05). In the presence of Cd²⁺, CPA again failed to decrease mIPSC frequency (88.0 ± 3.4% of the Cd²⁺ condition, P = 0.1, Fig. 6Ba).

Involvement of adenylyl cyclase-cAMP and PKA pathways

Since A₁ receptor activation is negatively coupled to cAMP formation by inhibiting adenylyl cyclase (AC) in some brain region (Ebersolt et al. 1983), we tested the effect of forskolin, an AC activator (Seamon et al. 1981), on A₁ receptor-mediated inhibition of GABAergic mIPSCs. Forskolin (10 µM) significantly increased mIPSC frequency 156.3 ± 27.5% of the control (P < 0.05, n = 5, Fig. 7, Ab and Ba) without affecting the mean mIPSC amplitude (92.8 ± 7.5% of the control, Fig. 7Bb). Forskolin completely prevented CPA-induced inhibition of GABAergic mIPSC frequency (94.4 ± 9.7% of the forskolin condition, P = 0.43, n = 5, Fig. 7Ba) without altering the current amplitude (96.1 ± 2.6% to the forskolin condition, Fig. 7Bb). To conform the negative coupling between A₁ receptor activation and cAMP formation, we tested the effect of dideoxy-forskolin, an inactive form of forskolin, on CPA-induced inhibition of GABAergic mIPSC frequency. Dideoxy-forskolin (10 µM) did not change mIPSC frequency (92.5 ± 10.3% of the control, P = 0.29, n = 5), nor did it alter the inhibitory effect of CPA on GABAergic mIPSC frequency (73.4 ± 2.2% of the dideoxy-forskolin condition, P < 0.05, n = 5, Fig. 7, Ac and Ba). The results suggest that the reduction of cAMP formation by A₁ receptor activation is closely related to CPA-induced inhibition of GABAergic mIPSCs.

View larger version (36K): [in this window] [in a new window]   Fig. 7. A1 receptor-mediated presynaptic inhibition of GABAergic mIPSCs is coupled to the AC-cAMP signal transduction pathway. A: typical traces of mIPSCs observed before, during, and after the application of 1 µM CPA in the absence (a) or presence (b) of 10 µM forskolin, and in the presence of 10 µM dideoxy-forskolin (c). a and b were obtained from the same neuron. B: each column is the mean from measured neurons (forskolin, n = 5; dideoxy-forskolin, n = 5). All amplitudes and frequencies are normalized to the respective control.

It is well known that a change in intracellular cAMP concentration affects a variety of cellular signaling via cAMP-dependent protein kinase A (PKA) in the hippocampus (Bouron 2001). In addition, the PKA-dependent modulation of VDCCs can directly regulate neurotransmitter release (Okada et al. 2001). Thus to address the possibility that PKA was also involved in the modulation of GABA release observed in the present study, we examined the effects of Rp-cAMPS, a selective PKA inhibitor, on CPA-induced inhibition of GABAergic mIPSCs. Rp-cAMPS (100 µM) significantly decreased basal GABAergic mIPSC frequency to 74.9 ± 7.5% of the control (P < 0.05, Fig. 8, Ab and Ba), without affecting the mean mIPSC amplitude (93.2 ± 3.1% of the control, P = 0.61, n = 7, Fig. 8Bb). In the presence of Rp-cAMPS, however, CPA-induced inhibition of GABA release was completely occluded, with mean mIPSC frequency in the presence of both CPA and Rp-cAMPS being 99.9 ± 14.3% of the Rp-cAMPS control value (P = 0.15, n = 7, Fig. Ba). Therefore the A₁ receptor-mediated inhibition of GABAergic mIPSCs is likely to be dependent on PKA activity.

View larger version (31K): [in this window] [in a new window]   Fig. 8. A1 receptor-mediated presynaptic inhibition of GABAergic mIPSCs is related to PKA. A: typical traces of mIPSCs observed before, during, and after the application of 1 µM CPA in the absence (a) or presence (b) of 100 µM Rp-cAMPS. a and b were obtained from the same neuron. B: each column is the mean from 7 neurons. All amplitudes and frequencies are normalized to the control.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Adenosine A₁ receptor-mediated inhibition of spontaneous GABA release

In immature CA1 neurons, adenosine reversibly decreased GABAergic mIPSC frequency without affecting the amplitude distribution, indicating that adenosine acts presynaptically to inhibit spontaneous GABA release from the presynaptic nerve terminals. This effect was completely blocked by DPCPX, a selective A₁ receptor antagonist, and was mimicked by CPA, a selective A₁ receptor agonist. Thus adenosine primarily seems to act presynaptic A₁ receptors. On the other hand, since adenosine did not change GABAergic mIPSC frequency after the blockade of A₁ receptors with DPCPX, GABAergic presynaptic nerve terminals projecting to CA1 neurons might express only A₁ receptors.

The present results clearly suggest the existence of functional adenosine A₁ receptors on the GABAergic presynaptic nerve terminals, but are not consistent with previous studies showing that adenosine via presynaptic A₁ receptors inhibits glutamatergic transmission but not the GABAergic one in hippocampal neurons (Dolphin and Archer 1983; Lambert and Teyler 1991; Scanziani et al. 1992; Thompson et al. 1992; Yoon and Rothman 1991). The discrepancy might be explained by the preparation used. That is, most studies were performed with the adult hippocampal slice (Dolphin and Archer 1983; Lambert and Teyler 1991) and cultured hippocampal neurons (Scanziani et al. 1992; Thompson et al. 1992; Yoon and Rothman 1991), whereas the present study was performed with acutely dissociated immature hippocampal neurons. In addition, it should be also noted that presynaptic A₁ receptor activation could inhibit GABAergic transmission in immature but not adult CA1 neurons. The results are closely consistent with not only the expression pattern of A₁ receptors in the hippocampus during postnatal development (Ochiishi et al. 1999) but also the previous electrophysiological evidence showing that adenosine does not affect GABAergic synaptic transmission in the adult hippocampus (Dolphin and Archer 1983; Lambert and Teyler 1991).

On the other hand, the present results suggest that A_2A receptors are not involved in the modulation of GABAergic transmission in either immature or adult hippocampal CA1 neurons, although a recent study demonstrated that A_2A receptor-mediated facilitation of GABA release from hippocampal synaptosomal preparations (Cunha and Rebeiro 2000). One explanation for this discrepancy is that Cunha and Ribeiro used a synaptosomal fraction from the whole hippocampus, which includes CA1, CA2, CA3, and dentate gyrus, whereas we used only acutely dissociated CA1 neurons. Although the present study demonstrated A₁ receptor-mediated inhibition of spontaneous GABAergic transmission in immature hippocampal CA1 neurons, more studies would be critically needed to evaluate the validity of adenosine-mediated inhibition of evoked GABAergic transmission in the slice preparation.

Intracellular signal transduction pathway of A₁ receptor-mediated inhibition of GABAergic transmission

Activation of A₁ receptors inhibits spontaneous neurotransmitter release via pertussis toxin-sensitive G protein in hippocampal neurons form embryonic and neonatal rats (Bouron and Reuter 1997; Scholz and Miller 1992). In the present study, A₁ receptor-mediated inhibition of mIPSC frequency was completely attenuated by adding NEM. The results are consistent with the previous findings showing that A₁ receptor-mediated presynaptic inhibition is coupled to NEM-sensitive G_i/G_o proteins in the hippocampus (Greif et al. 2000). G protein-coupled receptors have three possible modes of action in causing presynaptic inhibition for neurotransmitter release: inhibition of VDCCs, an increase in K⁺ conductance, or direct modulation of synaptic release machinery downstream of Ca²⁺ influx (Wu and Saggau 1997). Functionally, the inhibitory action of G protein-coupled A₁ receptors on neurotransmitter release is mediated by inhibition of adenylyl cyclase, inhibition of Ca²⁺ influx, as well as activation of K⁺ channels (Dunwiddie and Masino 2001). In hippocampal CA1 GABAergic synapses, however, K⁺ channels are unlikely to contribute to the presynaptic adenosine modulation of GABAergic mIPSCs, because K⁺ channel blockers, such as Ba²⁺ and 4-AP, did not prevent the inhibitory effect of CPA. This observation is consistent with the recent result showing that 4-AP-sensitive K⁺ channels are not related to adenosine inhibition of GABAergic mIPSCs in periaqueductal gray neurons (Bagley et al. 1999).

In dorsal root ganglion cells (Holz et al. 1986) and in hippocampal cultures (Scholz and Miller 1992), a pertussis toxin-sensitive inhibition of somatic Ca²⁺-current has been demonstrated but this mechanism of action has not been found in hippocampal slices (Greene and Haas 1985) and organotypic cultures nor is it responsible for the powerful presynaptic adenosine actions observed in this and other brain regions (Thompson et al. 1992). Nevertheless, activation of A₁ receptors inhibits the N-type Ca²⁺ current in isolated CA3 pyramidal cells (Mogul et al. 1993) and the presynaptic A₁ receptor-mediated effects are ascribed mainly to a reduction of Ca²⁺ influx (Okada et al. 2001; Wu and Saggau 1994a). In the present study, because the inhibitory action of CPA on GABAergic mIPSCs was completely occluded either in the Ca²⁺-free external solution or in the presence of Cd²⁺, A₁ receptor activation might be not related to the modulation of synaptic release machinery, but is most likely used to reduce the Ca²⁺ influx from extracellular sites, indicating the possible involvement of VDCCs. Such a conclusion is consistent with the previous reports showing that A₁ receptor-mediated presynaptic inhibition mostly results from the decrease in Ca²⁺ influx through VDCCs (Okada et al. 2001; Wu and Saggau 1994a).

cAMP- and PKA-dependent inhibition of GABAergic mIPSCs

Activation of the AC-cAMP signal transduction pathway directly facilitates the presynaptic neurotransmitter release in rat central neurons (Bouron 2001). In the hippocampus, an increase in intracellular cAMP concentration increases the number of readily releasable vesicles, without affecting either the number of morphologically docked vesicle or the number of active synaptic terminals (Trudeau et al. 1996). In addition, cAMP-dependent PKA activation is known to directly modulate both the secretory apparatus and/or the VDCCs (Bouron 2001). Considering these findings, our results (Figs. 7 and 8) suggest that the adenosine-induced inhibition of GABAergic mIPSCs is mediated by AC-cAMP- and PKA-dependent pathways. This conclusion is also closely consistent with the recent finding of Bagley et al. (1999). However, it should be noted that A₁ receptor-mediated inhibitory effect on GABAergic mIPSCs was completely occluded either in the Ca²⁺-free external solution or after the PKA blockade. This implies that A₁ receptor activation might not cause two independent pathways, but is likely to lead a subsequent signal transduction pathway that is a reduction of cAMP formation, a decrease in PKA activity, and the inhibition of Ca²⁺ influx. This conclusion is partly supported by the previous findings showing that cAMP-dependent PKA directly modulates L- or N-type Ca²⁺ channels (Okada et al. 2001). However, more studies are be needed to reveal the definite signal transduction mechanism of A₁ receptor-mediated inhibition of GABAergic transmission.

Physiological implications

Consistent with its role as an inhibitory neuromodulator for excitatory synaptic transmission, adenosine exhibits anticonvulsant effects in experimental models of epilepsy. Exogenously administered adenosine receptor agonists reduce seizure activity (Zhang et al. 1990), whereas adenosine receptor antagonists have proconvulsant effects (Dunwiddie 1980), which in the hippocampus are mediated by A₁ receptors (Alzheimer et al. 1989). During hypoxia and ischemic conditions, excitatory synaptic transmission is generally suppressed by an increase in the extracellular levels of adenosine, and this synaptic depression is also mediated by presynaptic A₁ receptor activation (Heron et al. 1993; Katchman and Hershkowitz 1993). Therefore A₁ receptors are believed to have neuroprotective roles against the excessive excitotoxicity in the hippocampus. This conclusion seems to be incompatible with the present results demonstrating A₁ receptor-mediated inhibition of GABAergic transmission onto immature hippocampal CA1 neurons. It should be noted that, however, during 1-2 wk of postnatal development, GABA-induced postsynaptic responses are converted from depolarization to hyperpolarization by increased expression levels of the K-Cl cotransporter, which is a major chloride extrusion system (Rivera et al. 1999; Vu et al. 2000). In addition, we found that GABA_A receptor activation indeed depolarizes immature CA1 neurons (Jang, Jeong, and Akalke, unpublished observation). Consequently, in immature CA1 neurons, A₁ receptors might exert their neuroprotective effect against the excitotoxicity by the inhibition of depolarizing GABAergic transmission.