Metabotropic glutamate group II receptors activate a G protein-coupled inwardly rectifying K⁺ current in neurones of the rat cerebellum

Abstract

1. The effects of the group II metabotropic glutamate receptor (mGluR) agonists DCG-IV and LY354740 were examined in neurones freshly dissociated from the rat cerebellum and olfactory bulb, using the whole-cell configuration of the patch-clamp technique.

2. Under experimental conditions in which K⁺ currents would be inward, rapid application of DCG-IV and LY354740 to interneurones expressing the group II mGluRs induced an inward current in a subpopulation of interneurones of the cerebellum, the unipolar brush cells.

3. The currents induced by DCG-IV and LY354740 had the major characteristics of a G protein-coupled inwardly rectifying K⁺ channel (GIRK) current; namely, rapid activation and deactivation upon agonist application and removal, G protein dependence, strong inward rectification, Cs⁺ and Ba²⁺ sensitivity, and K⁺ selectivity.

4. In Golgi cells of the cerebellum and interneurones of the accessory olfactory bulb, which also express group II mGluRs, LY354740 did not induce GIRK activation but inhibited voltage-gated Ca²⁺ channel currents.

5. These results demonstrate that, in unipolar brush cells, native group II mGluRs can functionally couple to activation of GIRKs. Thus, the absence of coupling in the majority of CNS neurones examined to date may be due to restricted cellular co-localization or co-expression of the appropriate proteins.

Members of the G protein-coupled receptor superfamily that couple through G_o/G_i generally activate G protein-coupled inwardly rectifying K⁺ channels (GIRKs) in neurones through a fast, membrane-delimited pathway involving the Gβγ subunits of G proteins (Logothetis, Kurachi, Galper, Neer & Clapham, 1987; Reuveny et al. 1994; for review, see Hille, 1994). These include the M₂ muscarinic, μ, δ and κ opioid, α₂ adrenergic, 5-HT_1A serotonin, D₂ dopamine, GABA_B and somatostatin receptors (North, 1989). Gβγ subunits also modulate voltage-gated Ca²⁺ channels (VGCCs; Herlitze, Garcia, Mackie, Hille, Scheuer & Catterall, 1996; Ikeda, 1996), and most G_o/G_i protein-coupled receptors couple to both GIRKs and VGCCs. Coupling of metabotropic glutamate receptors (mGluRs) to VGCCs via G proteins has been shown in numerous neuronal preparations (for review, see Pin & Duvoisin, 1995). Activation of mGluRs also inhibits the voltage-dependent K⁺ current, I_K(M) (Charpak, Gahwiler, Do & Knopfel, 1990), the slow Ca²⁺-dependent K⁺ current, I_AHP (Baskys, Bernstein, Barolet & Carlen, 1990; Charpak et al. 1990; Desai & Conn, 1991), a leak K⁺ current, I_K(leak) (Guerineau, Gahwiler & Gerber, 1994), and a slow K⁺ current, I_K(slow) (Luthi, Gahwiler & Gerber, 1996). In dissociated rat CA1 neurones, group I mGluR agonists activated Ca²⁺-dependent K⁺ channels (Shirasaki, Harata & Akaike, 1994). Furthermore, non-specific cation channels can be activated by mGluR agonists (Guerineau, Bossu, Gahwiler & Gerber, 1995). However, GIRK activation by mGluRs still remains to be demonstrated in mammalian CNS neurones. The lack of mGluR coupling to GIRKs in neurones has been interpreted, beyond the fact that a coupling could have been overlooked, as a possible different compartmentalization of the respective proteins or by the absence of co-localization of the two receptor types. In two recent reports, evidence for the coupling of mGluRs to GIRKs was demonstrated in recombinant receptor studies. In the Xenopus oocyte expression system, coupling of recombinant mGluR1a, mGluR2 and mGluR7 with the GIRK subunits Kir3.1 and Kir3.4 was shown (Saugstad, Segerson & Westbrook, 1996). In another study, the G_o/G_i-coupled mGluRs (mGluR2, mGluR3, mGluR4, mGluR6 and mGluR7) expressed in Xenopus oocytes activated GIRKs, whereas the G_q-coupled mGluRs (mGluR1a, mGluR5) inhibited basal and ACh-evoked GIRK activity (Sharon, Vorobiov & Dascal, 1997). In both studies, the highest efficiency of coupling between mGluRs and GIRKs was obtained by expressing GIRK subunits with mGluR2.

However, in the absence of a coupling of native mGluRs to GIRK activation in neurones, it remained a possibility that coupling in Xenopus oocytes was peculiar to the recombinant expression system. Many mGluR2s are presynaptically located and, thus, a coupling in cell bodies may have been missed. In order to optimize our chances of demonstrating a possible coupling of mGluRs to GIRK activation, we chose cells in which group II mGluR localization in cell bodies and dendrites had been shown by immunohistochemistry. These requirements are met in the dendrites and the cell bodies of interneurones of the cerebellum and the olfactory bulb (Neki, Ohishi, Kaneko, Shigemoto, Nakanishi & Mizuno, 1996; Petralia, Wang, Niedzielski & Wenthold, 1996). Because GIRK subunit mRNAs are also highly expressed in those brain regions (Karschin, Dissmann, Stuhmer & Karschin, 1996), we tested the ability of the potent and selective mGluR2/mGluR3 agonist, LY354740 (Monn et al. 1997), and of DCG-IV to activate GIRKs in freshly dissociated Golgi cells and unipolar brush cells of the cerebellum (Mugnaini & Floris, 1994) and neurones of the olfactory bulb using the whole-cell configuration of the patch-clamp technique.

METHODS

Cell preparation

Neurones from the vestibulocerebellum and accessory olfactory bulb of 10- to 14-day-old Sprague-Dawley rats were obtained according to standard methods. Briefly, animals were decapitated as approved by the local institutional animal welfare committee, and their brains removed. Sagittal slices (400 μm) were cut with a vibratome in an ice-cold solution that contained (mM): NaCl, 125; KCl, 2.5; CaCl₂, 2; MgCl₂, 1; NaH₂PO₄, 1.25; NaHCO₃, 26; and D-glucose, 25, bubbled with oxycarbon (95 % O₂, 5 % CO₂), pH 7.4, and were subsequently incubated at 20°C in the same solution. The accessory olfactory bulb and the vestibular part of the cerebellum were identified under a dissecting microscope. When neurones were needed for electrophysiological experiments, the desired regions of three slices were treated for 5 min at 37°C with a solution containing (mM): Na₂SO₄, 82; K₂SO₄, 30; MgCl₂, 3; Hepes, 2; and 1 mg ml⁻¹ papain, pH adjusted to 7.4 with NaOH. Neurones were isolated by gentle trituration with a Pasteur pipette in the same solution without enzyme, and plated on poly-L-ornithine-coated glass coverslips.

Electrophysiological experiments

Electrophysiological experiments were performed with the whole-cell configuration of the patch-clamp technique (Hamill, Marty, Neher, Sakmann & Sigworth, 1981). Whole-cell currents were amplified with an Axopatch 200A amplifier (Axon Instruments), filtered at 5 kHz and digitized at 10 kHz with a Digidata 1200A acquisition board (Axon Instruments) for subsequent storage on a Gateway 2000 P4D-66 personal computer. The data acquisition and analysis were performed with the pCLAMP6 software package (Axon Instruments). Pipettes were pulled from borosilicate glass with resistances ranging from 2 to 3 MΩ and filled with a solution containing (mM): potassium acetate, 80; KCl, 50; MgCl₂, 1; CaCl₂, 0.5; Hepes, 10; EGTA, 11; GTP, 0.3; ATP, 4; and phosphocreatine, 14, pH adjusted to 7.2 with KOH, osmolarity 330 mosmol l⁻¹ for K⁺ channel current measurements, and: TEA-Cl, 117; Hepes, 9; EGTA, 9; MgCl₂, 4.5; GTP, 0.3; ATP, 4; and phosphocreatine, 14, pH adjusted to 7.2 with TEA-OH, osmolarity 310 mosmol l⁻¹ for Ca²⁺ channel current measurements. The neurones were perfused with a standard salt solution that contained (mM): NaCl, 140; KCl, 5; CaCl₂, 2; MgCl₂, 1; Hepes, 10; and D-glucose, 10, pH adjusted to 7.4 with NaOH, osmolarity adjusted to 340 mosmol l⁻¹ with sucrose. For K⁺ channel current measurements, the standard salt solution was used with 25 mM KCl replacing 25 mM NaCl. To record Ba²⁺ currents through voltage-gated calcium channels, the neurones were perfused with a solution containing (mM): BaCl₂, 25; TEA-Cl, 145; Hepes, 10; EGTA, 0.1; and D-glucose, 10, pH adjusted to 7.4 with TEA-OH, osmolarity adjusted to 340 mosmol l⁻¹ with sucrose. For drug-testing experiments, substances were applied locally to the neurone by fast perfusion from a double-barrelled pipette assembly. The rate of solution exchange was assessed by measuring the decrease in Na⁺ current through AMPA/kainate receptors induced by exchanging the normal neurone perfusion solution with a solution containing a 10 times lower NaCl concentration. In these experiments, the rate of solution exchange was consistently around 30 ms.

GTP, ATP and creatine phosphate were obtained from Boehringer Mannheim, DCG-IV from Tocris Cookson (Bristol, UK) and LY354740 ((+)-2-aminobicyclo[3.1.0]hexane-2,6-dicarboxylic acid) was synthesized in our laboratories by Dr R. Jakob-Røtne and Dr G. Adam. All other chemicals were obtained from Sigma.

RESULTS

Identification of neurones

After the dissociation procedure, some morphological structures of neurones were preserved which allowed their identification. Interneurones of the cerebellum had an intermediate size between that of granule cells and Purkinje cells. Putative unipolar brush cells (UBCs), which are prominently represented in the granular layer of the vestibulocerebellum (Mugnaini & Floris, 1994), were identified by their single, large dendritic trunk that terminated in a tightly packed cluster of branchlets. (Fig. 1A). Putative Golgi cells were bigger in size and possessed at least two thinner dendritic trunks (Fig. 1B). Putative UBCs and Golgi cells have been identified as being positively stained with an antibody to mGluR2/mGluR3 in an immunohistochemical study performed with sections of rat cerebellum (Petralia et al. 1996). We also performed immunohistochemical staining of freshly dissociated cerebellar interneurones with a polyclonal antibody (Chemicon, Temacula, CA, USA) to mGluR2/mGluR3. Whereas the soma of Golgi cells was densely stained, the soma of UBCs was moderately stained, whilst the brush-like structure was highly stained. In contrast, granule cells and Purkinje cells were not stained (R. Pink, personal communication).

Figure 1. Phase contrast photographs of a freshly dissociated unipolar brush cell (A) and Golgi cell (B) of the rat cerebellum.

Note the tip of the single dendrite of the unipolar brush cell forming a tightly packed cluster of branchlets (A). The smaller round cells in A and B are granule cells. Scale bar, 10 μm.

DCG-IV- and LY354740-induced inward current

Freshly dissociated UBCs were voltage-clamped at a potential of -90 mV, under conditions in which currents through K⁺ channels would be inward ([K⁺]_i= 130 mM; [K⁺]_o= 30 mM). Application of saturating concentrations of LY354740 (1 μM) and DCG-IV (1 μM) induced inward currents in all UBCs tested. As represented in Fig. 2A and B from a representative UBC, the LY354740- and DCG-IV-induced currents activated rapidly, reached a plateau and persisted until removal of the drug from the perfusing solution. The mean maximum current amplitudes were 664 ± 198 pA (n= 7) and 488 ± 210 pA (n= 5) for LY354740 and DCG-IV, respectively. To evaluate the kinetics of activation (t_on) and deactivation (t_off) of the current induced by a high concentration of LY354740 (10 μM), the time course of the currents was fitted with a single exponential function for the time interval where the current amounted to 10 and 90 % of its maximum amplitude. The values for t_on and t_off were 313 ± 31 ms and 1.20 ± 0.14 s (n= 5), respectively.

Figure 2. Group II mGluR agonists activate a G protein-coupled inward current in unipolar brush cells.

LY354740 (A) and DCG-IV (B) were applied at the indicated concentrations with a rapid application system for the times indicated by the bars. The neurone was held at -90 mV and the recording was made under conditions in which K⁺ currents would be inward ([K⁺]_i= 130 mM; [K⁺]_o= 30 mM). C and D, LY354740 (10 μM) was applied for the time indicated by the bar in the presence of GTP-γ-S (300 μM) in the recording pipette while the neurone was clamped at -90 mV and the bathing solution contained 30 mM K⁺. The inward current induced by LY354740 was largely irreversible (C) and a subsequent application of LY354740 (D) had only a small effect.

G protein involvement

The G protein involvement in the LY354740 activation of inward currents in UBCs was assessed by testing the effect of GTP-γ-S, the non-hydrolysable analogue of GTP that irreversibly activates G proteins. When the recording pipette contained GTP-γ-S (300 μM) as a replacement for GTP, LY354740 (10 μM) still activated inward currents with a maximum amplitude amounting to 395 ± 96 pA (n= 4) when the neurones were clamped at -90 mV in the presence of 30 mM [K⁺]_o. However, as shown for a representative neurone in Fig. 2C and D, the currents were largely irreversible, and a second application of LY354740 had only a small effect. The maximum amplitude of the current induced by the second application of LY354740 recorded from the same neurones was 74 ± 16 pA (n= 4).

K⁺ dependence and inward rectification

To assess the K⁺ dependence of the inward current induced by LY354740 in UBCs, current-voltage relationships were generated in 5 and 30 mM [K⁺]_o, with a pipette solution containing 130 mM K⁺. The holding potential of the neurones was first changed from -70 to -150 mV for 40 ms, and then altered according to a voltage ramp from -150 to -15 mV lasting 270 ms. Five successive current responses were generated and then averaged. The average current responses represented in Fig. 3A were recorded in the presence and absence of LY354740 (1 μM) in 5 mM [K⁺]_o, and in Fig. 3B in 30 mM [K⁺]_o. To resolve the LY354740-induced current, I-V curves in the presence and absence of agonist were subtracted. In Fig. 3C, the LY354740-induced currents in 5 mM and 30 mM external K⁺ were strongly inwardly rectifying with only a small outward current at potentials more positive than the reversal potential. The reversal potentials of the LY354740-induced currents were -79.5 ± 4.1 mV (n= 4) and -38.3 ± 1.1 mV (n= 9) in 5 and 30 mM [K⁺]_o, respectively. The shift in reversal potential (-41.2 mV) is close to that predicted for a purely K⁺-selective conductance according to the Nernst equation (-45.1 mV).

Figure 3. K⁺ dependence of the LY354740-induced current in a unipolar brush cell.

Currents measured in the absence (Control) and presence of LY354740 (1 μM) with bathing solutions containing 5 mM K⁺ (A) and 30 mM K⁺ (B). The holding potential of the neurone was varied from -150 to -15 mV over 270 ms. The traces represent the average currents of 5 voltage ramps. C, traces obtained by subtraction of the averaged currents represented in A (5 mM K⁺) and B (30 mM K⁺), which correspond to the I-V relationships for the LY354740-induced current.

Ba²⁺ and Cs⁺ sensitivities

The Ba²⁺ and Cs⁺ sensitivities of the LY354740-induced inward current in UBCs were determined by generating current-voltage relationships in the presence and absence of BaCl₂ and CsCl, with 30 mM [K⁺]_o. As shown from the I-V relationships in Fig. 4A, the LY354740-induced inward current was blocked by 100 μM Cs⁺ in a strongly voltage-dependent manner. At a holding potential of -150 mV, the block by Cs⁺ amounted to 94 ± 0.5 % (n= 4). However, at -50 mV, the block by Cs⁺ was almost relieved and amounted only to 12 ± 5.0 % (n= 4). In contrast, at all concentrations tested, Ba²⁺ blocked the LY354740-induced current with no apparent voltage dependence (Fig. 4B). At a holding potential of -90 mV, the degree of Ba²⁺ inhibition amounted to 43.8 ± 1.5 and 93 ± 2.5 % of control for 10 and 100 μM Ba²⁺, respectively (n= 3). As shown for Ba²⁺ in Fig. 4C and D, the inhibition of the LY354740-induced currents by Cs⁺ and Ba²⁺ was fully reversible.

Figure 4. Cs⁺ and Ba²⁺ sensitivity of the LY354740-induced current in unipolar brush cells.

I-V relationships for the LY354740-induced current were generated in the absence (Control) and presence of the indicated concentrations of Cs⁺ (A) and Ba²⁺ (B) in the bathing solution. The holding potential of the neurone was varied from -150 to -15 mV over 270 ms, and 5 current responses were averaged. The traces in A and B represent the LY354740-induced current obtained by subtraction of the averaged current responses in the presence and absence of LY354740. [K⁺]_o was 30 mM. C and D, the block by Ba²⁺ is reversible. C, LY354740 (10 μM) was applied with a rapid application system for the time indicated by the bar. The unipolar brush cell was held at -90 mV. D, the inward current was activated by LY354740 (10 μM, bar), and Ba²⁺ (100 μM) was co-applied with LY354740 for the time shown by the bar. [K⁺]_o= 30 mM.

Effect of LY354740 in Golgi cells of the cerebellum

When LY354740 (10 μM) was applied to putative Golgi cells, none of them responded to the agonist with an inward current (n= 10). We tested, therefore, whether LY354740 would inhibit VGCCs in these cells. Figure 5 shows a representative result from such an experiment. Ba²⁺ currents through VGCCs were evoked by 25 ms step potentials to 0 mV from a holding potential of -80 mV every 5 s (inset). The mean maximum current amplitude recorded from four cells was 238 ± 36 pA. When LY354740 (1 μM) was applied, the amplitudes and the activation kinetics of voltage-evoked Ba²⁺ currents were decreased (inset). The mean inhibition of the peak current amplitudes by LY354740 (1 μM) was 43 ± 5.0 % of control (n= 4). This suggests that, in our morphologically identified presumptive Golgi cells, functional group II mGluRs are expressed and couple to VGCCs but do not activate K⁺ channels.

Figure 5. LY354740 inhibits VGCCs in Golgi cells of the cerebellum.

High-voltage-gated calcium channel currents were evoked at 5 s intervals according to the stimulation protocol shown in the inset. The currents measured at the peak of the control response and their inhibition by LY354740 are plotted versus time. LY354740 (1 μM) was applied for the time indicated by the bar. The inset shows superimposed current traces evoked in the absence (1) and presence (2) of LY354740.

Another nucleus densely stained with group II mGluR antibody is the rostal end of the accessory olfactory bulb (Petralia et al. 1996). Therefore, we tested whether in interneurones of the accessory olfactory bulb LY354740 would also induce an inward current. These interneurones were identified by their intermediate size between that of granule cells and mitral cells. In interneurones dissociated from this brain region, LY354740 (10 μM) did not induce an inward current (n= 22) but, as in Golgi cells, inhibited VGCCs (not shown). The mean maximum current amplitude was 681 ± 131 pA (n= 4), and the mean inhibition of the peak current amplitudes by LY354740 (10 μM) was 17 ± 2.0 % of control (n= 4).

DISCUSSION

The results in this study demonstrate for the first time that native group II mGluRs can functionally couple to activation of GIRKs in mammalian neurones, albeit in a subpopulation of interneurones from the rat cerebellum, the unipolar brush cells. Previous investigations have shown that mGluRs inhibit K⁺ channels or activate non-specific cation channels in mammalian neurones (Baskys et al. 1990; Charpak et al. 1990; Desai & Conn, 1991; Guerineau et al. 1994, 1995; Luthi et al. 1996). However, to our knowledge, the only reports of a glutamate-activated, G protein-sensitive K⁺ current were described in invertebrate neurones, i.e. mollusc (Bolshakov, Gapon & Magazanik, 1991) and crustacean (Miwa, Ui & Kawai, 1990).

Both LY35470, a selective and potent group II mGluR agonist, and DCG-IV, a group II mGluR agonist, activated an inward current in UBCs. The inward current activated by LY354740 presented the characteristics of a GIRK current; namely, G protein dependence, strong inward rectification, K⁺ selectivity, and block by Cs⁺ and Ba²⁺. The activation and deactivation kinetics of the inward current induced by a saturating concentration of LY354740 were similar to those found for a GABA_B-induced GIRK current obtained with a saturating concentration of baclofen (t_on= 225 ms, t_off= 1.1 s; Sodickson & Bean, 1996). The reversal potentials of the LY354740-induced current in UBCs found for two different concentrations of [K⁺]_o were -80 and -38 mV for 5 and 30 mM [K⁺]_o, respectively. These values were in agreement with the values predicted from the Nernst equation for a purely selective K⁺ conductance: -82 mV for 5 mM and -37 mV for 30 mM [K⁺]_o. Therefore, the current induced by LY354740 was essentially carried by K⁺, similar to the previously described K⁺ currents activated by baclofen (Sodickson & Bean, 1996), 5-HT (Penington, Kelly & Fox, 1993) and other neurotransmitters (North, 1989). The inward current induced by LY354740 in the present study was strongly inwardly rectifying. In addition, a slight outward current was found when generating I-V relationships in the presence of 5 mM [K⁺]_o. This property is shared with a GABA_B-activated K⁺ current which showed a substantial outward current at depolarized potentials (Sodickson & Bean, 1996) but not with homomeric recombinant GIRKs formed from the GIRK1 subunit which is expressed in the cerebellum (Dascal et al. 1993; Kubo, Reuveny, Slesinger, Jan & Jan, 1993).

Cs⁺ blocked the LY354740-induced current in a voltage-dependent manner. However, the block by Ba²⁺ was not voltage dependent. This is similar to the GIRK current activated by baclofen (Sodickson & Bean, 1996). Furthermore, since an approximately 50 % block of the current was observed with 10 μM Ba²⁺, and a complete block by 100 μM Ba²⁺, a high sensitivity to Ba²⁺ was demonstrated, again consistent with the GABA_B-activated K⁺ current (Sodickson & Bean, 1996) and the 5-HT-induced K⁺ current (Penington et al. 1993).

In the cerebellum, mGluR2-like immunoreactivity is seen in Golgi cell soma and dendrites (Neki et al. 1996). Furthermore, a polyclonal mGluR2/mGluR3 antibody stained Golgi cell somas in the granular layer, and their dendrites were stained up into the molecular layer (Petralia et al. 1996). In the present study, VGCCs from freshly dissociated Golgi cells were modulated by LY354740. Therefore, functional mGluR2/mGluR3 are expressed on the soma of these cells. However, application of LY354740 to freshly dissociated Golgi cells did not induce a K⁺ current. Another cell type examined in the present study were interneurones of the accessory olfactory bulb. Similar to Golgi cells, these neurones express mGluR2/mGluR3 highly (Petralia et al. 1996), which coupled to and inhibited VGCCs but did not activate GIRKs. A possible explanation for the lack of LY354740-induced inward current is that, in contrast to VGCCs, GIRKs are not expressed on the soma of neurones from these two preparations. Alternatively, GIRKs might be expressed but do not couple to mGluR due to a different compartmentalization of the proteins. Nevertheless, since freshly dissociated neurones have lost the main part of their dendrites after the dissociation procedure, a possible coupling of mGluRs to GIRKs in dendrites of Golgi cells or neurones of the olfactory bulb cannot be excluded.

Like Golgi cells, UBCs were stained by a polyclonal mGluR2/mGluR3 antibody (Petralia et al. 1996). In all UBCs tested in the present study, LY354740 induced an inward current. The main morphological difference between freshly dissociated Golgi cells and UBCs, beyond the different soma size, is that the latter cells possess a tightly packed group of branchlets connected to the cell soma by a single dendrite. In this group of branchlets, mGlu2/mGlu3 receptors are highly expressed (Petralia et al. 1996).

The properties of synaptic transmission at the mossy fibre (MF)-UBC synapse have been characterized in detail (Rossi, Alford, Mugnaini & Slater, 1995). In particular, monosynaptic activation of MFs produces an unusually slow EPSC, mainly mediated by AMPA/kainate and NMDA receptors. It was hypothesized that the very large postsynaptic densities and relatively large volume of the synaptic cleft at the MF-UBC synapse could prevent the rapid clearance of glutamate. This would allow extensive rebinding of glutamate to postsynaptic AMPA/kainate and NMDA receptors and, thus, prolong the EPSC (Rossi et al. 1995). Although UBCs appear to receive an inhibitory GABAergic input from Golgi cells, mGluR activation of GIRKs by glutamate may be an additional mechanism to regulate the strong excitatory input from MFs to UBCs.

In conclusion, the data presented here indicate that native group II mGluRs can functionally couple to GIRKs. Therefore, the absence of coupling in the majority of CNS neurones observed to date may be due to restricted cellular co-localization or co-expression of all of the appropriate proteins required for this to occur.