Receptor-specific inhibition of GABAB-activated K+ currents by muscarinic and metabotropic glutamate receptors in immature rat hippocampus


Corresponding author W-K. Ho: Department of Physiology, Seoul National University College of Medicine, 28 Yonkeun-Dong, Jongno-gu, Seoul 110-799, Korea.  Email:


It has been shown that the activation of Gq-coupled receptors (GqPCRs) in cardiac myocytes inhibits the G protein-gated inwardly rectifying K+ current (IGIRK) via receptor-specific depletion of phosphatidylinositol 4,5-bisphosphate (PIP2). In this study, we investigated the mechanism of the receptor-mediated regulation of IGIRK in acutely isolated hippocampal CA1 neurons by the muscarinic receptor agonist, carbachol (CCh), and the group I metabotropic glutamate receptor (mGluR) agonist, 3,5-dihydroxyphenylglycine (DHPG). IGIRK was activated by the GABAB receptor agonist, baclofen. When baclofen was repetitively applied at intervals of 2–3 min, the amplitude of the second IGIRK was 92.3 ± 1.7% of the first IGIRK in control. Pretreatment of neurons with CCh or DHPG prior to the second application of baclofen caused a reduction in the amplitude of the second IGIRK to 54.8 ± 1.3% and 51.4 ± 0.6%, respectively. In PLCβ1 knockout mice, the effect of CCh on IGIRK was significantly reduced, whereas the effect of DHPG remained unchanged. The CCh-mediated inhibition of IGIRK was almost completely abolished by PKC inhibitors and pipette solutions containing BAPTA. The DHPG-mediated inhibition of IGIRK was attenuated by the inhibition of phospholipase A2 (PLA2), or the sequestration of arachidonic acid. We confirmed that DHPG eliminated the inhibition of IGIRK by arachidonic acid. These results indicate that muscarinic inhibition of IGIRK is mediated by the PLC/PKC signalling pathway, while group I mGluR inhibition of IGIRK occurs via the PLA2-dependent production of arachidonic acid. These results present a novel receptor-specific mechanism for crosstalk between GqPCRs and GABAB receptors.

The G protein-gated inwardly rectifying K+ (GIRK) channels are activated by stimulation of Gi/o-coupled receptors, such as M2 muscarinic and GABAB receptors, and provide an important mechanism for the slow inhibitory modulation of cellular excitability in the heart and brain (Luscher et al. 1997; Wickman et al. 1998). The activation of GIRK channels is accomplished by the direct binding of the Gβγ subunit to the channels. This causes an increase in the channel binding affinity of phosphatidylinositol-4,5-bisphosphate (PIP2) which is crucial for channel activity (Huang et al. 1998). Since PIP2 serves as a substrate for phospholipase C (PLC) that is activated by the Gq protein, we questioned whether PIP2 depletion, induced by the stimulation of Gq-coupled receptors (GqPCR), could inhibit GIRK currents (IGIRK) in physiological conditions. This was previously investigated in cardiac myocytes, and demonstrated that stimulation of the α1-adrenergic receptor resulted in the inhibition of IGIRK (Cho et al. 2001a). Subsequent experiments on the specificity of action of GqPCRs have shown that prostaglandin F and endothelin are potent inhibitors of IGIRK, but that bradykinin has little or no effect (Cho et al. 2005b). These results imply that crosstalk between GqPCRs and GIRK channels is specifically regulated, and therefore the results obtained in cardiac myocytes cannot be extrapolated to other cell types. However, there has been limited research conducted on the role of GqPCRs in neuronal GIRK channel modulation, and the mechanisms of GqPCR-mediated IGIRK modulation in the brain are poorly understood.

The role of GqPCRs in the regulation of neuronal ion channel has been intensively studied for M channels and Ca2+ channels in neurons of the superior cervical ganglion (SCG), which express various GqPCRs such as B2 bradykinin receptors (B2R) and M1 muscarinic receptors (M1R) (Delmas et al. 2005). These studies showed that M channels are regulated by both receptors, but use different mechanisms of action for their effect. The B2Rs regulate M channels by intracellular Ca2+/calmodulin signalling (Gamper & Shapiro, 2003), while the M1Rs achieve their regulatory effect by PIP2 depletion (Suh & Hille, 2002; Zhang et al. 2003). In contrast, Ca2+ channels are not affected by B2Rs, but regulated by M1Rs via PIP2 depletion (Gamper et al. 2004) or arachidonic acid release (Liu & Rittenhouse, 2003). These results emphasize the complexity of GqPCR-mediated signalling, and strongly support the idea that in native cells crosstalk between GqPCRs and ion channels is arranged in a receptor- and cell-specific manner.

In this study, we have investigated in hippocampal CA1 neurons whether GIRK channels are regulated by muscarinic and group I metabotropic glutamate receptors (mGluR), both of which are classified as GqPCRs (Kleppisch et al. 2001; Krause et al. 2002), and whether these regulations are receptor specific. Our results have shown that the GIRK channel was inhibited to a similar degree by both receptors, but with different mechanisms. The effect of muscarinic receptors was mediated by the PLC/PKC pathway, while the effect of mGluR was mediated by the PLA2/arachidonic acid pathway. These results demonstrate that in hippocampal CA1 neurons, the crosstalk between GqPCRs and GIRK channels is receptor specific.


Preparation of isolated hippocampal neurons

Hippocampal CA1 neurons were isolated as previously described (Han et al. 1999), with minor modification. Protocols were approved by the Animal Care Committee at Seoul National University. Briefly 9–12-day-old Sprague-Dawley rats were decapitated under pentobarbital anaesthesia. The brain was quickly removed and submerged in ice-cold artificial cerebrospinal fluid (ACSF, see below) saturated with 95% O2 and 5% CO2. Transverse hippocampal slices (400 μm thick) were prepared using a vibratome (VT1000S, Leica, Germany). After a 30 min recovery period at 32°C, the slices were treated with protease type XIV (1 mg (5 ml)−1, Sigma, USA) for 30–60 min, and subsequently with protease type X (1 mg (5 ml)−1, Sigma) for 10–15 min at 32°C. The slices were allowed to recover during a 1 h incubation period at room temperature. The CA1 region was identified and punched out under a binocular microscope (SZ40, Olympus, Japan), placed in a recording chamber containing normal Tyrode solution (see below) and mechanically dissociated using a Pasteur pipette to release individual neurons. The dissociated neurons were allowed to adhere to the bottom of the recording chamber for 10–20 min. Cells identified as pyramidal neurons typically had a large pyramidal-shaped cell body, with a thick apical dendritic stump. Preparation of cells from knockout mice was performed using a similar method.

Knockout mice

The PLCβ1 mutation in mice has been maintained in two different genetic backgrounds, C57BL/6J and 129/sv. Heterozygous animals from the two backgrounds were mated to obtain homozygous mutant mice in the F1 background between C57BL/6J and 129/sv. The genotype of the progeny was determined by PCR as previously described (Kim et al. 1997). Wild-type littermates served as controls for the mutants. The mice had ad libitum access to food and water and were kept on a 14: 10 h light–dark cycle with lights on at 6 a.m. Animals were housed and cared for according to the guidelines of the Korea Institute of Science and Technology for the care of experimental animals.

Solutions and drugs

ACSF contained (mm): NaCl 125, NaHCO3 25, KCl 3, NaH2PO4 1.25, CaCl2 2, MgCl2 1, glucose 10, sucrose 5, vitamin C 0.3, bubbled with mixture of 95% O2 and 5% CO2 to a final pH of 7.4. Normal Tyrode solution contained (mm): NaCl 150, KCl 5, CaCl2 2, MgCl2 1, glucose 10, Hepes 10, adjusted to pH 7.4 with Tris-OH. The high-K+ normal Tyrode solution contained (mm): NaCl 95, KCl 60, CaCl2 2, MgCl2 1, glucose 10, Hepes 10, adjusted to pH 7.4 with Tris-OH. Nystatin perforation pipette solution contained (mm): KCl 40, K-methanesulphonate 120, Hepes 10, adjusted to pH 7.3 with KOH. For the results shown in Figure 4, conventional whole-cell configuration was used to deliver BAPTA into the cell with a pipette solution containing (mm): K-gluconate 110, KCl 30, Hepes 20, Mg-ATP 4, Na-vitamin C 4, Na-GTP 0.3, BAPTA 8, adjusted to pH 7.3 with KOH.

Figure 4.

PKC and intracellular Ca2+ are involved in the effects of CCh, but not DHPG
A, normalized average current traces of cells that were treated with PKC inhibitors (0.1 μm GF109203X, upper traces or 1 μm staurosporine, middle traces) before the application of CCh or DHPG. BAPTA (8 mm) was included in the pipette solution using conventional whole-cell configuration (lower traces). Note that I2,CCh (•, left) is almost completely overlapped with I1 (○), while I2,DHPG (•, right) remains significantly reduced. B, the peak amplitude of I2,CCh in the presence of GF109203X, staurosporine and BAPTA was 94.5 ± 1.9% (n= 7), 103.1 ± 3.2% (n= 6), and 95.1 ± 5.1% (n= 6) of I1,peak, respectively. The peak amplitude of I2,DHPG in the presence of GF109203X, staurosporine and BAPTA was 47.6 ± 1.4% (n= 4), 45.7 ± 0.9% (n= 3), and 53.0 ± 0.9% (n= 6) of I1,peak, respectively. *P < 0.05; GFX, GF109203X; STP, staurosporine. Error bars indicate s.e.m.

Stock solutions of baclofen (Tocris, UK), 3,5-dihydroxyphenylglycine (DHPG, Tocris), carbachol (CCh, Sigma), linopirdine (Tocris), and TTX (Sigma) were made by dissolution in de-ionized water and were stored stored at −20°C. On the day of the experiment, one aliquot was thawed and used. Stock solutions of bisindolylmaleimide I (GF109203X, Sigma), staurosporine (Tocris), wortmannin (Biomol, USA), arachidonyltrifluoromethyl ketone (AACOCF3, Calbiochem, USA), and arachidonic acid (Calbiochem) were made by dissolution in DMSO (Sigma). The concentration of DMSO in the external solution was maintained below 0.1%. Bovine serum albumin (BSA, Sigma) was directly dissolved in the external solution on the day of the experiment.

Electrophysiological recordings

Voltage-clamp recordings of membrane currents were performed at room temperature using an EPC-10 amplifier (HEKA Elektronik, Germany), while the cells were superfused with the high-K+ external solutions (60 mm[K+]o) by gravity flow. To ensure a rapid solution turnover, the rate of superfusion was maintained at ∼5 ml min−1, which corresponded to 50 bath volumes (100 μl) per minute. The holding potential was −80 mV, and in most experiments the recording was performed in a perforated-patch configuration by using nystatin (200 μg ml−1, MP biomedicals, USA) at a sampling rate of 10 Hz filtered at 1 kHz. For the results shown in Fig. 4, the conventional whole-cell configuration was used for the perfusion of BAPTA into the cell. In experiments that tested the effects of CCh, linopirdine (10 μm) was added to the external solution to exclude the influence of muscarinic M current inhibition. Data were acquired using an IBM-compatible computer running Pulse software v8.67 (HEKA Elektronik). The patch pipettes were pulled from borosilicate capillaries (Hilgenberg-GmbH, Germany) using a Narishige puller (PP-83, Narishige, Japan). The patch pipettes had a resistance of 3–5 MΩ when filled with the above pipette solutions.

Data analysis

Data were analysed using IgorPro (version 4.1, WaveMetrics, USA) and Origin (version 6.0, Microcal, USA) software. Statistical data are expressed as mean ±s.e.m., where n represents the number of cells studied. The peak IGIRK amplitudes were determined as follows. Baseline currents were subtracted from IGIRK traces and normalized to the peak amplitude of the first IGIRK. The normalized IGIRK traces under each experimental condition were averaged to reduce noise and minimize artifacts. This method produced a representative IGIRK trace, as shown in Fig. 1B, from which we could determine the peak IGIRK amplitudes. The significance of differences between the peaks of the normalized average currents was evaluated using a Student's t test with a confidence level of P < 0.05.

Figure 1.

IGIRK was repetitively activated by baclofen
Nystatin-perforated whole-cell current traces at a holding potential of −80 mV in high-K+ normal Tyrode external solution ([K+]o= 60 mm). Voltage ramp pulses (arrowheads) were applied during the time course of the recording. 0.5 μm TTX was added to the external solution to obtain voltage-ramp responses. A, baclofen (100 μm) was applied for 40 s twice at a 2–3 min interval, as indicated by the horizontal lines above the trace. B, normalized average current traces obtained from the current traces in the dashed rectangle in A (I1 and I2). Note the similar peak amplitudes of I1 (○, left) and I2 (•, right). The peak amplitude of I2 was 92.3 ± 1.7% (n= 8) of that of I1. C, I–V curve for net IGIRK at peak. Data were calculated from I–V relationships at points indicated with arrowheads in A. The results were normalized to the current amplitude at −80 mV. The data from eight cells were averaged.


Activation of IGIRK by baclofen

It has previously been shown that baclofen, a selective GABAB receptor agonist, activates IGIRK in hippocampal neurons (Sodickson & Bean, 1996; Leaney, 2003) and in neocortical neurons (Sickmann & Alzheimer, 2003). From acutely isolated hippocampal CA1 neurons, we recorded membrane currents using a nystatin-perforated whole-cell patch-clamp technique, at a holding potential of −80 mV, while cells were perfused with a high-K+ bath solution (60 mm[K+]o). When holding current levels were stable and less than −100 pA, baclofen (100 μm) was applied to activate IGIRK. The application of baclofen triggered a fast increase in inward current (I1, the first IGIRK), followed by variable degrees of desensitization (Fig. 1A). Peak current amplitudes also varied among different cells, ranging from 50 pA to 500 pA. Following washout of baclofen, induced currents slowly returned to their basal levels. The re-application of baclofen at intervals of 2–3 min triggered an IGIRK (I2, the second IGIRK) with a similar amplitude to I1. To enable quantitative analysis, current traces in the dashed rectangles in Fig. 1A were normalized to the peak current amplitude of I1 and the normalized currents obtained from eight cells were averaged. For a comparison between I1 and I2, normalized average current traces of I1 (bsl00150, left) and I2 (bsl00085, right) were shown over a 30 s period following the application of baclofen (Fig. 1B). From these traces the peak value of I2 (I2,peak) was calculated to be 92.3 ± 1.7% (n= 8) of that of I1 (I1,peak) (Fig. 2D).

Figure 2.

IGIRK was inhibited by CCh and DHPG in a reversible manner
A, CCh (10 μm) or B, DHPG (50 μm) was applied prior to the application of baclofen as indicated by the horizontal lines above the trace. C, normalized average current traces obtained from the current traces in A (Ca) and B (Cb). Note that pretreatment with CCh (I2,CCh,•, Ca) or DHPG (I2,DHPG, •, Cb) significantly reduced the peak amplitudes of I1 (○). D, the peak amplitudes of I2, I2,CCh and I2,DHPG were 92.3 ± 1.7% (n= 8), 54.8 ± 1.3% (n= 8) and 51.4 ± 0.6% (n= 10) of the I1,peak, respectively. *P < 0.05. Error bars indicate s.e.m.

To ensure that the baclofen-induced inward currents were GIRK currents, I–V relationships were examined by applying voltage-ramp pulses from −130 mV to +10 mV (700 ms duration) before and during the application of baclofen. The I–V relationships for net IGIRK were obtained by subtracting the current response in the control condition from that recorded at peak IGIRK (indicated with arrowheads in Fig. 1A). The results were then normalized to the current amplitude at −80 mV. An averaged I–V relationship from eight cells is presented in Fig. 1C. This shows a typical inward rectification known for GIRK current with a reversal potential of around −27 mV, which is close to the equilibrium potential of the potassium ion (−25.2 mV) as calculated using the Nernst equation.

Both CCh and DHPG inhibited IGIRK in a reversible manner

The effects of CCh (muscarinic receptor agonist) or DHPG (group I mGluR agonist) on IGIRK were utilized in the study of the mechanisms involved in IGIRK regulation by different GqPCRs (Fig. 2). CCh (10 μm) was applied 30 s before the second application of baclofen, and I2 in the presence of CCh (I2,CCh) was compared with I1. The amplitude of I2,CCh was significantly reduced, and the subsequent application of baclofen after washout of CCh showed a partial recovery and an increased current compared to I2,CCh (Fig. 2A). The effect of DHPG (50 μm) on IGIRK was tested using the same experimental protocol (Fig. 2B), and showed a similar result to that obtained following the application of CCh. We also demonstrated that the decrease in current with CCh or DHPG occurred uniformly over the whole voltage range without significant changes in inward rectifying properties or reversal potentials (supplemental Fig. 1A). This suggests that the current inhibited by CCh or DHPG is IGIRK.

As noticed in the current trace shown in Fig. 2A, CCh induced a slight activation of inward currents before the application of baclofen in 11 out of 43 cells tested. We confirmed that the CCh-activated currents were distinctive from baclofen-activated GIRK currents, because these currents showed linear IV relationships characteristic for non-selective cation currents (NSC) (supplemental Fig. 1B). Furthermore, IV relationships for baclofen-activated current in the presence of CCh in cells that showed NSC activation did not differ from those without NSC, implying that the CCh-activated NSC did not affect the baclofen-activated current. So, we did not exclude the cells showing activations of NSC in the analysis of the effect of CCh on GIRK.

Normalized average current traces of I1 (○) were superimposed with I2 (bsl00085) in the presence of CCh (Fig. 2Ca) or DHPG (Fig. 2Cb) and indicated that CCh and DHPG inhibited IGIRK to a similar degree. The peak current amplitude of I2,CCh and I2,DHPG was 54.8 ± 1.3% (n= 8) and 51.4 ± 0.6% (n= 10) of the I1,peak, respectively, which was significantly smaller than the I2,peak in the absence of GqPCR agonists (P < 0.05) (Fig. 2D). Peak amplitudes of IGIRK following washout of CCh and DHPG were 72.5 ± 1.6% (n= 6) and 73.8 ± 2.8% (n= 9) of the I1,peak, respectively. The recovery following the washout of drugs was not complete, but was significant (P < 0.05).

PLCβ1 is involved in CCh-induced IGIRK inhibition

Muscarinic receptors and group I mGluRs are known to couple with Gq proteins (Kleppisch et al. 2001; Krause et al. 2002) and activate PLC to cause phosphoinositide (PI) hydrolysis in hippocampal neurons (Fisher & Bartus, 1985; Johnson et al. 1999). Among the isoforms of PLC, the PLCβ1 isoform was shown to be highly expressed in the hippocampus (Ross et al. 1989), and to be involved in the signalling pathway of muscarinic receptor activation (Kim et al. 1997) and group I mGluR activation (Chuang et al. 2001; Young et al. 2004). Therefore, we used PLCβ1 knockout mice to test whether PLC was involved in CCh- and DHPG-induced IGIRK inhibition. To confirm that IGIRK was inhibited by CCh or DHPG in wild-type mouse hippocampal CA1 neurons, we performed the same experiment as previously described (Fig. 2A and B) using the wild-type littermates of the knockout mice. We found that pretreatment with CCh or DHPG reduced the peak amplitude of IGIRK to a similar degree as that observed in rat CA1 neurons (Fig. 3C, open bars). When the same experiment was performed in cells isolated from the PLCβ1 knockout mice, CCh-induced IGIRK inhibition was significantly reduced (Fig. 3A), whereas DHPG-induced IGIRK inhibition was not affected (Fig. 3B). The peak amplitude of I2,CCh was 81.3 ± 2.0% (n= 8) of the I1,peak in PLCβ1 knockout mice which was significantly different from the result in wild-type mice (57.7 ± 2.2%, n= 3, P < 0.05). However, the peak amplitude of I2,DHPG was 50.2 ± 1.6% (n= 6) of the I1,peak in PLCβ1 knockout mice, which was not significantly different from the result with wild-type mice (56.6 ± 2.2%, n= 3, P > 0.05). These results indicate that CCh, but not DHPG, induces IGIRK inhibition via signalling pathways involving PLCβ1 in hippocampal CA1 neurons.

Figure 3.

PLCβ1 is involved in CCh, but not DHPG-induced IGIRK inhibition
A, CCh or B, DHPG was applied prior to the application of baclofen as indicated by the horizontal lines above the trace in the PLCβ1 knockout (KO) mice. CCh-induced inhibition was significantly suppressed, while DHPG-induced inhibition was not. C, in wild-type mice, the peak amplitudes of I2,CCh and I2,DHPG were 57.7 ± 2.2% (n= 3) and 56.6 ± 2.2% (n= 3) of the I1,peak, while in PLCβ1 knockout mice they were 81.3 ± 2.0% (n= 8) and 50.2 ± 1.6% (n= 6) of I1,peak, respectiely. *P < 0.05. Error bars indicate s.e.m.

PKC and intracellular Ca2+ are involved in the effect of CCh

The involvement of PLCβ1 in CCh-induced IGIRK inhibition led us to examine the downstream pathways responsible for the observed effect. Since it has been reported that the neuronal GIRK channel is regulated by PKC (Takano et al. 1995; Sharon et al. 1997; Leaney et al. 2001), we tested whether PKC inhibitors could block the effects of CCh on IGIRK. Normalized average current traces of I1 (○) and I2,CCh (•) recorded in the presence of PKC inhibitors were superimposed (Fig. 4A, left). The application of GF109203X (0.1 μm) prior to the activation of I2 eliminated the CCh-induced inhibition of IGIRK, so that I1 and I2,CCh were almost completely overlapped. The protein kinase inhibitor, staurosporine (1 μm), also blocked the effect of CCh (Fig. 4A, left). The relative peak amplitudes of IGIRK (I2,CCh/I1) after treatment with GF109203X and staurosporine were 94.5 ± 1.9% (n= 7) and 103.1 ± 3.2% (n= 6), respectively (Fig. 4B). We also examined whether CCh-induced IGIRK inhibition could be abolished by the addition of a high concentration of BAPTA (8 mm) in the pipette solution (Fig. 4A, left). To deliver BAPTA into the cytosol, a conventional whole-cell configuration was established rather than the nystatin-perforated patch used in preceding experiments. In this configuration, I1 was similarly activated by baclofen, and the I2,peak after the pretreatment of CCh was 95.1 ± 5.1% (n= 6) of the I1,peak which is not significantly different from I2 in the absence of CCh. These results suggest that Ca2+-dependent PKC is responsible for CCh-induced IGIRK inhibition. The observed values are significantly different from the CCh-induced inhibition of IGIRK (54.8 ± 1.3%, n= 8, P < 0.05).

When the same experiment was performed to examine the involvement of PKC and intracellular Ca2+ in the effect of DHPG on IGIRK, the results were very different from those obtained using CCh (Fig. 4A, right). The relative peak amplitude of IGIRK (I2,DHPG/I1) in the presence of BAPTA was 53.0 ± 0.9% (n= 6), and the values after treatment with GF109203X and staurosporine were 47.6 ± 1.4% (n= 4) and 45.7 ± 0.9% (n= 3), respectively (Fig. 4B). These values are not significantly different from those obtained by DHPG-induced inhibition in IGIRK(51.4 ± 0.6%, n= 10), indicating that the DHPG effect occurs via a different pathway than PLC/PKC activation.

The PLA2/arachidonic acid pathway is involved in DHPG-induced IGIRK inhibition

Arachidonic acid, which is liberated from the plasma membrane mainly by phospholipase A2 (PLA2), has been reported to inhibit the ATP-dependent gating of GIRK channels (Kim & Pleumsamran, 2000). The PLA2 phospholipase has also been implicated in receptor-mediated IGIRK inhibition (Rogalski et al. 1999). Moreover, glutamate has been reported to activate PLA2 (Kim et al. 1995) and enhance the liberation of arachidonic acid in neurons (Allen et al. 2001) and astrocytes (Stella et al. 1994). We therefore examined the effect of the addition of 50 μm AACOCF3, a PLA2 inhibitor, to the bath prior to the second application of baclofen with DHPG. As shown in Fig. 5A, the application of AACOCF3 significantly attenuated the effect of DHPG. The addition of BSA to experimental systems has been shown to sequester and reduce the available pool of free fatty acids, including arachidonic acid, and produce a concentration gradient across the plasma membrane (Kamp et al. 1993). This causes the net outward movement of arachidonic acid and reduces its intracellular availability. The addition of BSA significantly reduced the effect of DHPG on IGIRK (Fig. 5B). In the presence of AACOCF3 or BSA, the peak amplitudes of I2,DHPG were 82.9 ± 1.6% (n= 8) and 83.2 ± 1.3% (n= 10) of I1, respectively. This was significantly different from I2,DHPG/I1 in the control experiments (51.4 ± 0.6%, n= 10, P < 0.05) (Fig. 5C). These results indicate that the inhibitory effect of DHPG on IGIRK can be attributed to PLA2-dependent signalling pathways.

Figure 5.

Inhibition of PLA2-mediated signalling pathways significantly blocked DHPG-induced IGIRK reduction
A, cells were pretreated with 50 μm of the PLA2 inhibitor AACOCF3, or B, 1 mg ml−1 BSA, a free fatty acid-binding protein, before the application of DHPG. DHPG-induced inhibition was significantly suppressed by both AACOCF3 and BSA. C, the peak amplitudes of I2,DHPG in the presence of AACOCF3 and BSA were 82.9 ± 1.6% (n= 8) and 83.2 ± 1.3% (n= 10) of I1,peak, respectively. *P < 0.05. ACF3, AACOCF3. Error bars indicate s.e.m.

To determine whether IGIRK inhibition by DHPG is mediated by the PLA2-dependent release of arachidonic acid, we tested whether arachidonic acid could directly inhibit IGIRK. The amplitude of IGIRK was reversibly reduced by around 25% following the pretreatment of cells with 4 μm arachidonic acid (Fig. 6A). In the presence of 7 μm of arachidonic acid, IGIRK was decreased by around 50%. At concentrations above 7 μm, achidonic acid often caused an increase in the background current (data not shown), suggesting that it activates background conductance, possibly via the two-pore K+ channels. To investigate whether the effects of arachidonic acid could be reduced by DHPG, we applied 7 μm arachidonic acid to DHPG-pretreated cells and control cells, and compared the effects on IGIRK. In order to distinguish between the inhibitory effects of arachidonic acid on IGIRK and desensitization kinetics, we applied arachidonic acid for a short period following IGIRK activation by baclofen. The application of arachidonic acid caused a rapid decrease in IGIRK, which could be reversed by wash-out in control cells (Fig. 6B). In contrast, arachidonic acid did not cause a decrease in IGIRK in DHPG-pretreated cells where IGIRK activation was already reduced by around 50% (Fig. 6C). Measurements of normalized average current traces pooled from 10 control cells (○) and from 7 DHPG-pretreated cells (•) were superimposed (Fig. 6D). The time of arachidonic acid application was considered to be 0 s. The relative amplitude of IGIRK just prior to arachidonic acid application in DHPG-pretreated cells was considered to be 51.4% of the control IGIRK, which was the average magnitude of IGIRK following pretreatment with DHPG (Fig. 2D). These pooled traces demonstrate that the application of arachidonic acid significantly inhibited IGIRK, and that DHPG pretreatment abolished this effect. This indicates that DHPG and arachidonic acid inhibit IGIRK by the common mechanism. From these data, it can be concluded that the effect of DHPG on IGIRK is mediated by activation of PLA2 and the consequent release of arachidonic acid.

Figure 6.

DHPG eliminated the inhibition of IGIRK by arachidonic acid
A, arachidonic acid (4 μm) was applied prior to the activation of the second IGIRK. B, arachidonic acid (7 μm) was applied to the bath after baclofen-induced IGIRK activation. C, cells were pretreated with DHPG followed by baclofen-induced IGIRK activation and application of arachidonic acid (7 μm). D, current traces obtained from 10 control (○), and 7 DHPG-pretreated (•) cells were collected and portions of the traces, as indicated in B and C, were normalized against the amplitude of IGIRK just prior to arachidonic acid application. The time of arachidonic acid application was designated as 0 s. The effect of arachidonic acid was blocked by pretreatment of cells with DHPG. AA, arachidonic acid. Error bars indicate s.e.m.

Wortmannin does not significantly affect the effect of CCh or DHPG

It was previously reported that PIP2 depletion induced by PLC activation was involved in GqPCR-mediated inhibition of GIRK channels in cardiac myocytes (Cho et al. 2001a, 2005b; Meyer et al. 2001). Since both GqPCRs cause PI hydrolysis in hippocampal neurons (Fisher & Bartus, 1985; Johnson et al. 1999), it is possible that PIP2 depletion plays a role in IGIRK inhibition. To investigate this possibility in hippocampal CA1 neurons, we designed an experiment to potentiate PIP2 depletion by blocking its re-synthesis. Wortmannin can be used to block the synthesis of PI-4-phosphate from PI, and is commonly used to deplete PIP2 in the plasma membrane (Nakanishi et al. 1995). If CCh or DHPG exert their effect on IGIRK by depleting PIP2, the addition of wortmannin should theoretically potentiate the effect and suppress IGIRK recovery when the agonist is washed out. To investigate this possibility, we tested the effect of CCh on IGIRK and the recovery of IGIRK in the presence of wortmannin (10–100 μm). Wortmannin did not potentiate the effect of CCh, and the subsequent application of baclofen activated IGIRK with partial recovery even in the continued presence of wortmannin (Fig. 7A). The peak amplitude of I2,CCh and IGIRK after washout (I3) in the presence of wortmannin was 47.2 ± 1.7% (n= 5) and 68.1 ± 1.9% (n= 4) of I1,peak, respectively. We performed the same experiment using DHPG in place of CCh and obtained similar results (Fig. 7B). The peak amplitude of I2,DHPG and I3 in the presence of wortmannin was 49.7 ± 1.4% (n= 9) and 90.7 ± 1.7% (n= 7) of I1,peak, respectively. These results suggest that PIP2 depletion is not a major mechanism in the inhibition of IGIRK by CCh or DHPG (Fig. 7C).

Figure 7.

Wortmannin did not significantly affect the effect of CCh or DHPG
The PI-4-kinase inhibitor, wortmannin (10 μm), was applied with CCh (A) or DHPG (B). The extent of I2,CCh or I2,DHPG suppression was not potentiated, and the recovery of IGIRK after washout (I3) was not suppressed. C, the peak amplitudes of I2,CCh or I2,DHPG in the presence of wortmannin were 47.2 ± 1.7% (n= 5) and 49.7 ± 1.4% (n= 9) of I1,peak, respectively. The peak amplitudes of I3 after washout of CCh or DHPG were 68.1 ± 1.9% (n= 4) and 90.7 ± 1.7% (n= 7) of I1,peak, respectively. Error bars indicate s.e.m.


This study demonstrates that GABAB receptor-activated IGIRK is inhibited by the muscarinic receptor agonist CCh and the group I mGluR agonist DHPG in hippocampal CA1 neurons, and that the mechanism of inhibition is receptor-specific. The muscarinic effect is mediated via the PLC/PKC signalling pathway, and the mGluR effect is mediated by the PLA2-dependent production of arachidonic acid. This is the first report to describe the receptor-specific mechanism underlying muscarinic receptor- and mGluR-mediated inhibition of GIRK channels in native neurons.

Cell- and receptor- specificity in GqPCR-mediated ion channel regulation

The mechanism of receptor-mediated GIRK current regulation has been investigated thoroughly in cardiac myocytes, and it has been shown that GIRK current inhibition by GqPCR agonists, such as phenylephrine, prostaglandin F, endothelin, and angiotensin, is mediated by PIP2 depletion (Cho et al. 2001a, 2005b). However, not all GqPCRs inhibit GIRK currents, and the bradykinin and M1/M3 muscarinic receptors have little effect on IGIRK (Cho et al. 2002, 2005b). To explain this receptor-specific effect, we hypothesized that PIP2 depletion was localized adjacent to the GqPCRs if its mobility was low, and that spatial proximity between the GqPCR and its target protein is necessary for receptor-mediated PIP2 signalling (Cho et al. 2005a).

In this study we have shown that muscarinic receptor- or mGluR-mediated inhibition of GIRK currents in hippocampal CA1 neurons is mediated by mechanisms other than PIP2 depletion. Our data suggest that PIP2 depletion and/or colocalization characteristics of GqPCR and GIRK channels in CA1 neurons may differ from those in cardiac myocytes. Currently we do not have a complete understanding of the mechanisms involved in the differential use of PIP2 signals in different cell types. However, it is possible that GqPCR stimulation in neurons may not deplete PIP2 enough to inhibit GIRK channels. This possibility has been proposed to explain the insensitivity of Ca2+ channels to bradykinin in SCG neurons (Gamper et al. 2004), where a bradykinin-induced intracellular Ca2+ increase stimulates PIP2 re-synthesis which compensates for PIP2 hydrolysis. Considering that both CCh and DHPG induce an intracellular Ca2+ increase in hippocampal neurons (Wakamori et al. 1993; Seymour-Laurent & Barish, 1995; Irving & Collingridge, 1998; our unpublished observation), this possibility appears to be very likely. Also, the possibility that the difference in PIP2 mobility is involved in the differential use of PIP2 signals in different cell types remains to be tested.

Receptor-specific mechanisms in the regulation of neuronal ion channels were intensively studied for inhibition of M currents and Ca2+ currents by M1 muscarinic receptors (M1R) and B2 bradykinin receptors (B2R) in SCG neurons. The M1Rs inhibit both M currents and Ca2+ currents, and it is generally accepted that this inhibition is mediated by membrane PIP2 depletion (Zhang et al. 2003; Gamper et al. 2004; Winks et al. 2005). In contrast, B2Rs inhibit M currents, but not Ca2+ currents (Gamper et al. 2004). Since M1Rs and B2Rs induce different Ca2+ responses in SCG neurons, and PIP2 synthesis is affected by Ca2+-dependent PI-4-kinase via the NCS-1, it has been suggested that the difference in Ca2+ signals that leads to the differential use of PIP2 signal underlies receptor specificity of the GqPCR actions on SCG neuron channels (Delmas et al. 2002; Gamper et al. 2004). However, this idea does not appear to be directly applicable to the receptor-specific mechanism of GIRK current inhibition in hippocampal neurons, since CCh and DHPG can induce a similar Ca2+ increase. It is possible that signalling microdomains responsible for the induction of selective responses may vary among different neuron types. We could not exclude the possibility that the signalling microdomain underlying the receptor-specific signalling mechanism may undergo developmental change, as our study used animals under the age of 2 weeks, whereas SCG neurons are usually isolated from 2- to 6-week-old-rats.

We excluded PIP2 depletion as a major mechanism involved in in CCh- or DHPG-mediated inhibition of GIRK currents in hippocampal CA1 neurons, but it should be noted that this was not based on direct experimental evidence. The exclusion was inferred from experiments in which CCh- or DHPG-mediated inhibition of GIRK currents was not significantly affected by the presence of wortmannin. It is generally accepted that PIP2 depletion is the mechanism responsible for the receptor-mediated effect when it is potentiated, and its recovery is blocked by the addition of wortmannin. However, there remains uncertainty, since it is not yet proved whether wortmannin indeed potentiates agonist-induced PIP2 depletion and inhibits its recovery in CA1 neurons.

Signalling pathways for muscarinic receptor-mediated GIRK channel regulation

We have shown that muscarinic inhibition of GIRK currents can be suppressed by blocking Ca2+- and PKC- signalling pathways, suggesting the involvement of Ca2+-dependent PKC. In contrast, GIRK currents in cardiac myocytes are not affected by PKC activation (Cho et al. 2001b), and PKC is not involved in GqPCR-induced inhibition of cardiac GIRK currents (Cho et al. 2001a). Experiments that used heterologous expression systems have shown involvement of PKC in GqPCR-induced inhibition of GIRK channels, but the possible involvement of Ca2+ is controversial. The involvement of Ca2+-dependent canonical PKC in M1-induced inhibition of GIRK1/4 channels expressed in Xenopus oocytes has been demonstrated (Hill & Peralta, 2001), while the role of Ca2+-independent novel PKC in M3-induced inhibition of GIRK1/2 channels expressed in HEK293 cells has also been shown (Leaney et al. 2001).

Recently, Brown et al. (2005) showed that the muscarinic inhibition of GIRK currents was attributable to PKC, while PIP2 in the pipette attenuated this effect, and the addition of wortmannin blocked the recovery after washout. This suggests dual regulation of the channel by both PIP2 and PKC. Data from this study suggest that GIRK channel activity is dependent on membrane PIP2 levels, while PKC phosphorylates the channel causing a reduction in PIP2–channel interaction. Although we suggested that PIP2 depletion is not a key mechanism in CCh-mediated inhibition of GIRK currents, we have not excluded the possibility of the involvement of PIP2 itself. It is feasible that PIP2 acts downstream of PKC in such a way that affinity of GIRK channels to PIP2 is reduced by PKC, thereby causing a decrease in channel activity.

Arachidonic acid, signalling molecules in the group I mGluR-mediated pathway

Previous studies on group I mGluR have shown involvement of PLC and IP3 in their signalling pathways (Abdul-Ghani et al. 1996; Chuang et al. 2001; Young et al. 2004). In this study we have shown that mGluR-mediated inhibition of IGIRK in CA1 neurons occurs in a Ca2+-independent manner and involves the PLA2/arachidonic acid pathway. Several studies have suggested the possibility of coupling between mGluR and PLA2. In cultured mouse striatal astrocytes, glutamate can evoke arachidonic acid release which can be inhibited by PLA2 inhibitors, but not by DAG lipase inhibitors (Stella et al. 1994). Consistent with this finding, the addition of glutamate caused an increase in cytosolic PLA2 activity in rat cortical neurons (Kim et al. 1995), and the addition of DHPG directly enhanced arachidonic acid release in rat neurons (Allen et al. 2001). These studies, and the data presented here, strongly support a coupling between the mGluR and the PLA2/arachidonic acid pathway.

The involvement of PLA2/arachidonic acid pathway in receptor-mediated ion channel regulation was suggested to be responsible for the inhibition of GIRK channels expressed in Xenopus oocytes by endothelin receptor stimulation (Rogalski et al. 1999). In a subsequent work, the site within the channel required for arachidonic acid sensitivity was identified and it was found that a critical aspartate residue for arachidonic acid sensitivity is the same residue required for Nai regulation of PIP2 gating (Rogalski & Chavkin, 2001). These results suggest that GIRK channel gating involves a series of regulatory steps including Gβγ, PIP2, Nai and arachidonic acid binding to the channel gating domain. Arachidonic acid might inhibit channel activity by altering interactions between the channels and PIP2. Our study shows that hippocampal GIRK channels are also responsive to dual modulation by arachidonic acid and PIP2. The converging but antagonistic actions of arachidonic acid and PIP2 have also been associated with Ca2+ channels (Delmas et al. 2005) and A-type K+ channels (Oliver et al. 2004).

Receptor-mediated ion channel regulation exemplifies the complexity of signalling networks. A single receptor can activate many different pathways; however, targets are regulated by their own specific signalling cascade. The mechanism by which a single receptor can link to a specific target and how this is regulated in a cell- and receptor-specific way remains to be understood.



This work was supported by the Korea Science and Engineering Foundation (KOSEF) grant funded by the Korea government (MOST) (No. 2004-015-E00023). D. Lee was supported by the BK21 Program from the Korean Ministry of Education.