Abstract
Although presynaptic localization of mGluR7 is well established, the mechanism by which the receptor may control Ca2+channels in neurons is still unknown. We show here that cultured cerebellar granule cells express native metabotropic glutamate receptor type 7 (mGluR7) in neuritic processes, whereas transfected mGluR7 was also expressed in cell bodies. This allowed us to study the effect of the transfected receptor on somatic Ca2+ channels. In transfected neurons, mGuR7 selectively inhibited P/Q-type Ca2+ channels. The effect was mimicked by GTPγS and blocked by pertussis toxin (PTX) or a selective antibody raised against the G-protein αo subunit, indicating the involvement of a Go-like protein. The mGuR7 effect did not display the characteristics of a direct interaction between G-protein βγ subunits and the α1A Ca2+ channel subunit, but was abolished by quenching βγ subunits with specific intracellular peptides. Intracellular dialysis of G-protein βγ subunits did not mimic the action of mGluR7, suggesting that both G-protein βγ and αo subunits were required to mediate the effect. Inhibition of phospholipase C (PLC) blocked the inhibitory action of mGluR7, suggesting that a coincident activation of PLC by the G-protein βγ with αo subunits was required. The Ca2+ chelator BAPTA, as well as inhibition of either the inositol trisphosphate (IP3) receptor or protein kinase C (PKC) abolished the mGluR7 effect. Moreover, activation of native mGluR7 induced a PTX-dependent IP3 formation. These results indicated that IP3-mediated intracellular Ca2+ release was required for PKC-dependent inhibition of the Ca2+ channels. Possible control of synaptic transmission by the present mechanisms is discussed.
The physiological actions of the neurotransmitter glutamate are mediated by ionotropic and metabotropic receptors (Nakanishi, 1992). Eight genes encoding mGluRs have been identified and classified into three groups. mGluR1 and mGluR5 belong to group I and activate phospholipase C (PLC) through stimulation of a Gq protein, in heterologous and homologous systems (Conn and Pin, 1997). The group II (mGluR2 and mGluR3) and group III (mGluR4, mGluR6, mGluR7, and mGluR8) mGluRs are coupled to Gi/o protein in neuron (Prezeau et al., 1994) and heterologous expressing cells (Conn and Pin, 1997). These receptors are widely distributed throughout the mammalian brain (Kinzie et al., 1995; Ohishi et al., 1995; Bradley et al., 1996;Kinoshita et al., 1998), but the mGluR7 subtype displays peculiar properties in that it is almost exclusively localized at presynaptic sites (Shigemoto et al., 1996, 1997; Kinzie et al., 1997). Because of a lack of specific pharmacology, functional discrimination between mGluR7 and the other group III mGluR subtypes can only be achieved according to their different affinity forl-2-amino-4-phosphonobutyrate (l-AP-4), a selective group III mGluR agonist. Indeed the affinity of mGluR7 forl-AP-4 is clearly lower (EC50 = 160–500 μm;Okamoto et al., 1994; Saugstad et al., 1994) than that of mGluR4, 6, and 8 (EC50 = 0.2–1.2, 0.9, and 0.06–0.60 μm, respectively; Pin et al., 1999).
In behavioral studies, young mGluR7 knock-out mice display deficits in the fear response and conditioned taste aversion, whereas the adult mutants develop lethal spontaneous epileptic seizures (Masugi et al., 1999). In vitro studies showed that mGluR7 stimulation mediates neuroprotective effects in cultured cerebellar granule cells by decreasing glutamate release (Lafon-Cazal et al., 1999a) and promotes excitotoxicity in cultured striatal neurons by inhibiting GABA release (Lafon-Cazal et al., 1999b). Group III mGluRs, presumably mGluR7, have been shown to inhibit glutamate autaptic currents in hippocampal neurons (O'Connor et al., 1999). These studies, together with those showing the presynaptic localization of the receptor in the murine adult brain, suggest that mGluR7 plays an important role in modulation and plasticity of synaptic transmission.
The mechanism by which mGluR7 may control neurotransmitter release is still unknown. Indeed, previous studies have shown thatl-AP-4 inhibits high-threshold voltage-gated Ca2+ channels in various neuronal preparations (Trombley and Westbrook, 1992; Rothe et al., 1994; Choi and Lovinger, 1996; Takahashi et al., 1996; Shen and Slaughter, 1998). Nevertheless, in these studies, the maximal inhibitions were obtained for relatively low concentrations of l-AP-4 (<100 μm) that should have selectively activated group III mGluRs, but with the exception of mGluR7. Moreover, inhibition of adenylyl cyclase by mGluR7 has only been shown in heterologous expression systems (Okamoto et al., 1994; Saugstad et al., 1994), and to our knowledge there is no clear study precluding that a different mechanism may function in neurons. Therefore, in the present study we investigated whether mGluR7 could modulate specific Ca2+ channel subtypes in cultured cerebellar granule cells and which coupling mechanism could be involved in this effect. We found that the receptor selectively inhibited P/Q-type Ca2+ channels by activating a Go-like protein and, unexpectedly, through a PLC-dependent pathway.
MATERIALS AND METHODS
Cell culture. Primary cultures of cerebellar cells were prepared as previously described (Van Vliet et al., 1989). Briefly, 1-week-old newborn mice were decapitated and cerebellum-dissected. The tissue was then gently triturated using fire-polished Pasteur pipettes, and the homogenate was centrifuged at 500 rpm. The pellet was resuspended and plated in tissue culture dishes previously coated with poly-l-ornithine. Cells were maintained in a 1:1 mixture of DMEM and F-12 nutrient (Life Technologies, Gaithersburg, MD), supplemented with glucose (30 mm), glutamine (2 mm), sodium bicarbonate (3 mm) and HEPES buffer (5 mm), decomplemented fetal calf serum (10%), and 25 mm KCl to improve neuronal survival. One-week-old cultures contained 125 × 103cells/cm2.
Plasmids and transfection. The N-terminal epitope-tagged mGluR7a receptor was constructed as follows. The Myc epitope was inserted in the extracellular domain, immediately downstream from the signal peptide. We used a mGluR5a-containing plasmid (pRKG5a-N-Myc) as the starting vector, in which the signal peptide was followed by the Myc-coding sequence at the N terminus of the protein, and then by aMluI site (Ango et al., 1999). The mGluR7a-coding sequence (except the signal peptide) was introduced into this vector using the MluI site and XbaI (3′ of the coding sequence), by following two steps: first we used a PCR and the oligonucleotide S, creating an Mlu I site in frame with the coding sequence of the vector (5′-gccAcgcgtatgtacgccccgcac-3′), and the oligonucleotide AS containing the XhoI site present in mGluR7 (5′-ttttctagaggaaggaatcaggcgggacca-3′); second the fragmentXhoI–XbaI was inserted by classical subcloning. The sequence was verified by sequencing. The resulting plasmid (pRKG7a-N-Myc) was referred as Myc-mGluR7. Functional coupling of Myc-mGluR7 was verified in human embryonic kidney 293 cells according to the protocol described elsewhere (Parmentier et al., 1998).
Immediately before plating, or 24 hr after plating, cerebellar cultures were transfected with the Myc-mGluR7 expression plasmid for immunocytochemical experiments or cotransfected with the transfection marker, green fluorescent protein (GFP)-containing plasmid, pEGFP-N1 (Clontech, Palo, Alto, CA), and non epitope-tagged mGluR7, for electrophysiological recordings, by using Transfast (Promega, Madison, WI), as described elsewhere (Ango et al., 1999).
Immunocytochemistry. Cultured cerebellar granule cells were fixed in a 4% paraformaldehyde and 0.1 mglucose-containing PBS solution. The culture was permeabilized with 0.05% Triton X-100, and fluorescent immunolabeling of native mGluR7 was performed by using a previously characterized anti-mGluR7a/b primary antibody (Shigemoto et al., 1996, 1997). The presence of Myc-mGluR7 protein at the cell surface of cultured neurons was examined in nonpermeabilized cerebellar cultures exposed to a monoclonal mouse anti-Myc primary antibody (a gift from B. Mouillac) diluted at 1:300 in a PBS–gelatin (0.2%) solution. After overnight incubation in the presence of either one of these primary antibodies at room temperature, cells were then rinsed and exposed to a goat Texas Red-conjugated anti-rabbit IgG secondary antibody or to a goat Texas Red-conjugated anti-mouse IgG secondary antibody (Jackson Immunoresearch, West Grove, PA; 1:1000 dilution), for 2 hr at room temperature. Then cells were rinsed again with PBS and mounted on glass coverslips for observation on an Axiophot 2 Zeiss microscope.
Electrophysiology. Whole-cell patch-clamp Ba2+ currents were recorded at room temperature from GFP and mGluR7 cotransfected cerebellar granule cells, after 9 ± 1 days in vitro as previously described (Ango et al., 1999). The bathing medium contained (in mm): BaCl2 20, HEPES 10, tetraethylammonium acetate 10, TTX 3 × 10−4, glucose 10, sodium acetate 120, and MK801 1 × 10−3, adjusted to pH 7.4 with NaOH and 330 mOsm with sodium acetate. Drug solutions were prepared in this bathing medium, and the pH was adjusted to 7.4. The NMDA receptor-channel blocker MK-801 (1 μm) was added to all the solutions to avoid activation of this receptor by the D isoform ofd,l-AP-4 (our unpublished observation). Patch pipettes were made from borosilicate glass, coated with Sylgard, and the tip was fire-polished. Pipettes had resistances of 3–5 MΩ when filled with the following internal solution (in mm): Cs-acetate 100, MgCl22, HEPES 10, glucose 15, CsCl 20, EGTA 20, Na2ATP 2, and cAMP 1, adjusted to pH 7.2 with CsOH and 300 mOsm with CsOH. In some experiments, intracellular EGTA was replaced by BAPTA.
Ba2+ currents were evoked by voltage-clamp pulses of 500 msec duration, from a holding potential of −80 mV to a test potential of 0 mV. Voltage pulses were applied at a rate of 0.1 Hz. Current signals were recorded with an Axopatch 200 amplifier, filtered at 1 kHz with an 8-pole Bessel filter, and sampled at 3 kHz on a Pentium II personal computer. Linear leak and capacitive currents were digitally subtracted from records before analysis by using the P/N procedure of the pClamp6 software of Axon Instruments(Foster City, CA). Analyses were performed by using the Clampfit subprogram of pClamp6. Ba2+ currents were measured at their peak amplitude and expressed as mean ± SEM of the indicated number (n) of experiments. In experiments in which neurons were dialyzed with compounds, current measurements were started at least 5 min after breaking the patch.
The intracellular I-II loop of the α1A Ca2+ channel subunit, which contained the binding site of G-protein βγ subunits and P/Q-type Ca2+ channel β subunit, was generated in the laboratory, according to the following procedure. The 68-mer peptide, corresponding to the sequence from 360 to 427 of the α1A (BI-2) subunit, was synthesized by the solid-phase method (Merrifield, 1986) by means of an automated peptide synthesizer (model 433A; Applied Biosystems, Foster City, CA). The peptide chain was assembled by a double-coupling strategy using Fmoc amino acid hydroxybenzotrioazol active esters. The crude peptide was purified to homogeneity by C18 reversed-phase HPLC and characterized by amino acid analysis after acidolysis, Edman sequencing, and mass spectrometry. The experimental values obtained were all in agreement with the theoretically deduced values.
Measurement of inositol phosphate accumulation. The procedure we used to measure inositol triphosphate (IP3) accumulation in neurons was adapted from one previously described (Blahos et al., 1998). One-week-old cerebellar granule cell cultures were incubated for 14 hr in culture medium containing 2 μCi/mlmyo-(3H)inositol (23.4 Ci/mol) (NEN, Paris, France). Cells were then washed three times and incubated for 1 hr at 37°C, in 1 ml of HEPES saline buffer (in mm: NaCl 146, KCl 4.2, MgCl2 0.5, and HEPES 20, glucose 0.1%, pH 7.4) supplemented with 1 U/ml glutamate pyruvate transaminase (Boehringer Mannheim, Meylan, France) and 2 mm pyruvate (Sigma, Lisle d'Abeau, France). Cells were then washed again with the same buffer, and LiCl was added to a final concentration of 10 mm. The agonist was applied 15 min later and left for 5 min. The reaction was stopped by replacing the incubation medium with 0.5 ml of perchloric acid (5%) on ice. Supernatants were recovered, and IPs were purified on Dowex columns (Berridge et al., 1983). Total radioactivity remaining in the membrane fraction was counted after treatment with 10% Triton X-100 and 0.1 N NaOH for 30 min and used as a standard. Results were expressed as the ratio of [3H]IP production over radioactivity present in the membranes. Experiments were performed in triplicates for statistical analyses.
Materials.l-AP-4, Dihydroxy-phenyl-glycine (DHPG), and MK-801 were purchased from Tocris Cockson. PTX, Nimodipine, GF109203X, U73122, and U73343 were purchased from Research Biochemicals (Natick, MA). ω-Agatoxin-IVA and ω-Conotoxin-GVIA were from Alomone Labs (Jerusalem, Israel). The G-protein βγ subunits purified from bovine brain were from Calbiochem. The inhibitor peptide of the catalytic subunit and the competitive inhibitor of the regulatory subunit of protein kinase A, protein kinase A inhibitor peptide (PKI), and Rp-cAMPS respectively, were also from Calbiochem. GTPγS was from Sigma, and PDBu and PMA were from Fluka. An antibody raised against the Go-protein was a generous gift from V. Homburger. This antibody has been previously shown to specifically recognize the G-protein αo but not αi subunit (Lledo et al., 1992). The pcDNA3-CD8-βARK plasmid, which was composed of the CD8 antigen membrane receptor and a domain containing the G-protein βγ subunit-binding site of βARK, was a generous gift from Dr. J. Lang. The pEGFP-N1 expression plasmid was purchased from Clontech.
RESULTS
Absence of native mGluR7 expression in the soma of cultured cerebellar granule cells
Immunolabeling of native mGluR7 in permeabilized cerebellar granule cells revealed a somatic exclusion and a neuritic punctate pattern of distribution of the receptor (Fig.1A).d,l-AP-4, up to 1 mmconcentration, did not significantly alter the whole-cell Ba2+ current of neurons transfected (Fig.2A) or not (data not shown) with GFP alone (control). Together these results indicated that the native mGluR7 was absent at the surface of the soma of cultured cerebellar granule cells. Therefore these neurons were potentially a good model to study the effect of transfected mGluR7 on somatic Ca2+ channels in these neurons, providing that the transfected receptor would be expressed at the cell body membrane.
Localization of native and transfected mGluR7 in cultured cerebellar granule cells. A, Native mGluR7 immunolabeling in permeabilized cultured cerebellar granule cells.B, Nonpermeabilized cultured cerebellar granule cell transfected with the Myc-mGluR7 expression plasmid and labeled with an anti-Myc antibody. Note the presence of neuritic clusters inA and B and presence of somatic immunolabeling only in B.
Inhibitory effect of d,l-AP-4 on Ba2+ currents in mGluR7-transfected cerebellar granule cells. A, Each bar of the histogram represents the mean (± SEM; n = 10 to 18) of fractional reduction of whole-cell Ba2+ current induced byd,l-AP-4 (500 μm) applied alone, in cultured cerebellar granule cells transfected with GFP alone, or cotransfected with GFP + mGluR7. Note that d,l-AP-4 alone inhibited Ba2+ currents only in cotransfected cells.B, Ba2+ currents recorded in a mGluR7-transfected cell, in the absence and presence of 10 μm, 100 μm, 500 μm, or 1 mmd,l-AP-4. Please note the absence of change in activation kinetics in the presence of the agonist.C,D, Activation (C) and inactivation (D) curves of whole-cell Ba2+ currents obtained from two different granule cells, in the absence (control) and presence ofd,l-AP-4 (500 μm). Similar results were obtained from five other cells. E, Time course and concentration-dependent effect of d,l-AP-4 on Ba2+ currents in a mGluR7-transfected cerebellar granule cell. F, Inhibitory effects of ω-Agatoxin-IVA (250 nm), ω-Conotoxin-GVIA (1 μm), and nimodipine (1 μm) on Ba2+ currents obtained in nontransfected cultured cerebellar granule cells or transfected with GFP alone or cotransfected with GFP + mGluR7. Each bar of the histogram represents the mean (± SEM) of at least seven experiments. G, Inhibitory effects of ω-Agatoxin-IVA (250 nm), ω-Conotoxin-GVIA (1 μm), and nimodipine (1 μm) on Ba2+ currents obtained in the presence of d,l-AP-4 (500 μm), in cultured cerebellar granule cells transfected with GFP alone or cotransfected with GFP + mGluR7. Each bar of histogram represents the mean (± SEM) of at least 10 experiments. Note that the percentage of Ba2+ current inhibited by each toxin was similar in control and cotransfected cells, except for ω-Agatoxin-IVA, which was ineffective only in cotransfected cells.
Selective inhibition of P/Q-type Ba2+ current by transfected mGluR7 at the soma of cultured cerebellar granule cells
Eight to ten days after transfection of the Myc-mGluR7 expression plasmid in cultured cerebellar granule cells, Myc immunostaining revealed the presence of both somatic and neuritic cell surface clusters of the recombinant receptors (Fig. 1B). In these transfected neurons, d,l-AP-4 (500 μm) decreased the amplitude of the total Ba2+ current, without significantly affecting its activation and inactivation kinetics (Fig.2B). The effect started after a delay of ∼20 sec, developed slowly, and reached a plateau over a 1 min 30 sec application of the agonist. This inhibition was accompanied by a slight but not significant modification of voltage-dependent activation (Fig.2C) and no alteration of steady-state inactivation (Fig.2D) properties of the current. Thed,l-AP-4 effect was dose-dependent, the threshold effect being obtained for 100 μm, and the maximal effect (38% inhibition) for 500 μmconcentrations (Fig. 2A,B). The current inhibition lasted for at least 10 min after washout of the agonist (Fig.2E). This long-lastingd,l-AP-4-mediated inhibition of Ba2+ currents did not result from an agonist-independent run-down of the current, because the amount of inhibition was stable for several minutes during wash-out of the agonist (Fig. 2E). Moreover, no significant decrease of the current was observed over a period of 45 min, in the absence ofd,l-AP-4 (data not shown).
In cultured cerebellar granule cells transfected with GFP alone (control), P/Q-type (ω-Agatoxin-IVA; 250 nm), N-type (ω-Conotoxin-GVIA; 1 μm), and L-type (nimodipine; 1 μm) Ca2+ channel blockers inhibited the whole-cell Ba2+ current by 41, 10, and 22%, respectively. The remaining 27% of total Ba2+ current were of the R-type. Similar results were obtained in cultured cerebellar granule cells cotransfected with GFP and mGluR7 or nontransfected cultured cerebellar granule cells (Fig. 2F). Therefore, our transfection procedure did not alter functional expression of native Ca2+ channels in the studied cells.
To determine which types of Ca2+ channels were inhibited by transfected mGluR7, d,l-AP-4 (500 μm) was applied first, followed by perfusion of different selective Ca2+ channel blockers, on neurons cotransfected with mGluR7 and GFP. After application ofd,l-AP-4, the remaining Ba2+current was not significantly affected by application of ω-Agatoxin-IVA (250 nm), but was further depressed by ω-Conotoxin-GVIA (1 μm) or nimodipine (1 μm), and in similar proportions as those obtained in control cells (transfected with GFP alone; Fig. 2G). It is worth noting that the fractional inhibition induced byd,l-AP-4 in cotransfected neurons (38%; Fig.2A) was not significantly different from the fraction of ω-Agatoxin-IVA-sensitive current obtained in control neurons (41%; Fig. 2G).
In a second series of experiments, an initial application of a given channel blocker was immediately followed by application ofd,l-AP-4 (500 μm). When ω-Agatoxin-IVA was applied first, the ω-Agatoxin-IVA-resistant Ba2+ current was not affected by subsequent perfusion of d,l-AP-4 (500 μm; 3 ± 1% inhibition, n = 7, Fig.3A). On the other hand, when ω-Conotoxin-GVIA or nimodipine were applied first, the drug/toxin-insensitive Ba2+ current was further depressed by subsequent application ofd,l-AP-4 (37 ± 2% inhibition,n = 5, after ω-Conotoxin-GVIA, Fig. 3B; 36 ± 4% inhibition, n = 5, after nimodipine, Fig. 3C; 36 ± 2% inhibition, n = 5, after ω-Conotoxin-GVIA and nimodipine, Fig. 3D). Finally, after coapplication of all three Ca2+ channel blockers,d,l-AP-4 did not further inhibit the remaining Ba2+ current (3 ± 2% inhibition,n = 4). After application ofd,l-AP-4, ω-Conotoxin-GVIA (Fig.3A,C) and nimodipine (Fig. 3A,B) further inhibited the Ba2+ current (by 10 and 20%, respectively), whereas ω-Agatoxin-IVA did not (Fig.3B,D). Altogether these results demonstrated that mGluR7 selectively blocked the P/Q-type Ca2+channels, without significantly affecting N-, L-, and R-types.
Selective blockade of P/Q-type Ba2+ currents by d,l-AP-4. Absence of effect of d,l-AP-4 (500 μm) on ω-Agatoxin-IVA- (250 nm, A) and inhibitory effect of the agonist on ω-Conotoxin GVIA- (1 μm,B, D) and nimodipine- (1 μm,C, D) insensitive Ba2+ currents. Graphs A–D were obtained from four different mGluR7-transfected cerebellar granule cells. Similar results were obtained from at least five other cells, for each graph.
mGluR7-mediated activation of a Go-like protein
To examine if a G-protein was involved in the mGluR7-mediated inhibition of P/Q-type Ba2+ currents, we intracellularly applied the nonselective G-protein activator, GTPγS (100 μm). Under these conditions, thed,l-AP-4-mediated inhibition of Ba2+ current was highly reduced (Fig.4A). Indeed, ω-Agatoxin-IVA inhibited only 12 ± 2% (n = 5) of the current, indicating that the P/Q-type Ca2+ channels were already significantly inhibited by GTPγS. Overnight incubation of the culture in the presence of PTX (200 ng/ml) abolished the inhibitory effect ofd,l-AP-4 (Fig. 4A). Together these observations showed that a Gi/o-like protein was involved in the d,l-AP-4-mediated inhibition of P/Q-type Ca2+ channels.
d,l-AP-4 inhibited Ba2+ current through an indirect action of Go-protein. A, Mean (± SEM;n = 6–10) fractional reduction of the whole-cell Ba2+ current induced by d,l-AP-4 (500 μm) in mGluR7-transfected cerebellar granule cells, under different conditions (from left toright): control condition (CT), in the presence of intracellular GTPγS (100 μm), after an overnight PTX treatment (200 ng/ml), after intracellular dialysis of an antibody raised against the G-protein αo subunit (1:100 dilution;anti-αoAb), after intracellular dialysis of the α1A I-II loop peptide (10 μm;I-II loop), in cells cotransfected with CD8-βARK chimera (βARK), after intracellular dialysis of purified G-protein βγ subunits (50 μg/ml; βγ). NS, Not significantly different from control. B, Whole-cell Ba2+ currents evoked by depolarizing steps to +10 mV, from a holding potential of −80 mV, preceded (right) or not (left) by a prepulse to +80 mV. Note that d,l-AP-4 (500 μm) induced similar Ba2+ current inhibition in the presence or absence of prepulse depolarization. Similar results were observed in eight other mGluR7-transfected neurons.
A specific antibody raised against the G-protein αo subunit, which did not recognize the G-protein αi subunit (see Materials and Methods), was used to determine which type of G-protein was involved in the l-AP-4 effect. This antibody significantly inhibited the effect of d,l-AP-4 on the Ba2+ current (Fig. 4A) without significantly altering the current density in the absence of agonist (89 ± 7 pA/pF, n = 10 without antibody; 79 ± 8 pA/pF, n = 6, with antibody). The boiled antibody was without effect (36 ± 4%,d,l-AP-4-induced inhibtion of Ba2+ current, n = 5). These results showed that a Go-, rather than a Gi-like protein, was involved in thed,l-AP-4-mediated effect.
We then investigated the involvement of the G-protein αo and βγ subunits. In heterologous expression systems, it has been shown that the P/Q-type Ca2+ channels can be blocked by a direct binding of G-protein βγ subunits on the I-II loop of the α1A Ca2+ channel subunit (De Waard et al., 1997; Bourinet et al., 1999). To test the implication of the native G-protein βγ subunits in the mGluR7-mediated inhibition of Ba2+ current in transfected cerebellar granule cells, these subunits were quenched in two ways. First, a peptide derived from the I-II loop of the α1A Ca2+ channel subunit (10 μm) was applied into the cell via the recording electrode. Second, the cDNA coding for a chimera composed of the CD8 membrane receptor antigen and of a domain containing the G-protein βγ subunit binding site of βARK, was cotransfected with mGluR7 in the cerebellar neurons. Under both conditions, the inhibitory effect of d,l-AP-4 was strongly reduced (Fig. 4A). We verified that after dialysis of the I-II loop of the α1A or transfection of the CD8-βARK chimera, ω-Agatoxin IVA still inhibited the whole-cell Ba2+ current (35 ± 6% inhibition,n = 5, in dialyzed neurons; 37 ± 2% in transfected neurons), indicating that the P/Q-type Ca2+ channels were not significantly affected by the tested peptides. Also, perfusion of a peptide-free solution or transfection of the CD8-βARK chimera deleted of its binding site for G-protein βγ subunits did not affect the action ofd,l-AP-4 on Ba2+currents (33 ± 3% inhibition, n = 5 in dialyzed neurons; 39 ± 3%, n = 5 in transfected neurons). These results indicated that mGluR7-mediated Ca2+ channel inhibition involved G-protein βγ subunits.
However, dialysis of G-protein βγ subunits (both at 50 μg/ml) did not significantly modify activation and inactivation kinetics of the whole-cell Ba2+ current (data not shown), and neither significantly altered the fraction of Ba2+ current that was inhibited byd,l-AP-4 (Fig. 4A), in mGluR7/GFP-cotransfected cerebellar granule cells. The tested G-protein βγ subunits inhibited and slowed activation kinetics of P/Q-type Ca2+ channels expressed inXenopus oocytes, indicating that the absence of effect of the tested G-protein βγ subunits in cultured cerebellar granule cells did not result from a lack of activity of these molecules (data not shown). Together, these observations indicated that G-protein βγ subunits were required, but not sufficient to mediate the mGluR7-induced inhibition of Ba2+current.
We further examined whether the mGluR7-mediated Ba2+ current inhibition could result from the well characterized direct interaction of Go-protein βγ subunits with the I-II loop of α1A Ca2+ channel subunit (De Waard et al., 1997; Bourinet et al., 1999). A positive prepulse relieves this interaction and inhibition (voltage-dependent facilitation) of the P/Q-type Ba2+ current. In mGluR7-transfected cerebellar granule cells, no such voltage-dependent facilitation was observed in the absence or presence ofd,l-AP-4 (Fig. 4B). Thus the ratio,R, between amplitudes of Ba2+currents evoked with and without depolarizing prepulse was measured in the absence and presence of d,l-AP-4. We obtained values of R that were not significantly different from 1 (R = 1.10 ± 0.07%, n = 9, in the absence of agonist; R = 1.04 ± 0.06%,n = 9, in the presence ofd,l-AP-4). This result indicated the absence of both tonic and mGluR7-mediated inhibition of Ca2+ channels through direct interaction of Go-protein βγ subunits in mGluR7-transfected cerebellar granule cells.
mGluR7-mediated PKC-dependent blockade of Ba2+ current
Because mGluR7 appeared to inhibit P/Q-type Ca2+ channels via an indirect action of a Go-like protein, we searched for any involvement of additional intracellular factors. The protein kinase A inhibitors, Rp-cAMPS (10 μm) and PKI (1 μm), added to the recording pipette solution, failed to inhibit Ba2+ currents in the absence or presence of d,l-AP-4, in nontransfected as well as mGluR7-transfected cerebellar granule cells (data not shown). On the other hand, the PKC activator PDBu (1 μm) inhibited the whole-cell Ba2+ current of cultured cerebellar granule cells by 27 ± 5% (n = 7). Similar results were obtained with the other PKC agonist, PMA (200 nm; 21 ± 7% inhibition, n= 4). These effects were not additive to the inhibitory effect of ω-Agatoxin IVA (5 ± 3% inhibition after PDBu application,n = 7; Fig.5A), indicating that P/Q-type Ca2+ channels in these neurons were PKC-sensitive. The hypothesis that mGluR7 activated a PKC-dependent pathway was therefore tested. In mGluR7-transfected granule neurons, a 30 min pretreatment with the selective PKC inhibitor GF109203X (10 μm) almost abolished the inhibitory effect ofd,l-AP-4 (Fig. 5B). We then studied whether PLC was also involved. Thed,l-AP-4-mediated inhibitory effect was altered by the PLC inhibitor U73122 (2 μm), whereas the inactive analog U73343 (same concentration) was without effect (Fig.5B). These results indicated that mGluR7 blocked P/Q-type Ca2+ channels via a PLC/PKC-dependent pathway.
d,l-AP-4-mediated inhibition of Ba2+ currents involved a PLC/PKC pathway in mGluR7-transfected cerebellar granule cells. A, Time course of the inhibitory effect of the PKC activator PDBu (1 μm, 30 min) on whole-cell Ba2+currents and absence of effect of ω-Agatoxin-IVA (250 nm) after the PDBu-mediated inhibition. B, Mean (± SEM;n = 5–10) fractional reduction of the whole-cell Ba2+ current recorded in mGluR7-transfected cerebellar granule cells induced by d,l-AP-4 (500 μm), under the following conditions (fromleft to right): in control cells (CT), in cells pretreated for 30 min with the PKC antagonist GF109203X (10 μm), and after 5 min dialysis of the PLC antagonist U73122 (2 μm) or the inactive analog U73343 (same concentration).
Because inhibition of Ba2+ currents by transfected mGluR7 was PLC-dependent, we tested whether the native receptor was able to induce IP3 formation in the studied neurons. In nontransfected cerebellar cultures,d,l-AP-4 at concentrations that stimulated mGluR7 (500 μm or 1 mm), increased IP3 formation by more than twofold, whereas lower concentrations of d,l-AP-4 (100 μm and 200 μm) had no significant effect on basal IP3 formation (Fig.6A). Thed,l-AP-4-induced formation of IP3 was abolished by an overnight pretreatment of the cultures with PTX (200 ng/ml; Fig. 6A). Therefore, native mGluR7 in cerebellar granule cells induced formation of IP3 via a Go-protein-dependent pathway. This result was in agreement with our electrophysiological data. Similar increases of IP3 formation were obtained in mGluR7 transfected cerebellar cultures (Fig. 6A). This apparent absence of effect of transfected mGluR7 can be explained by the low rate of transfection (2–5%) obtained with our method (Ango et al., 1999).
d,l-AP-4-induced IP3formation in nontransfected or mGluR7-transfected cultured cerebellar granule cells. A, IP formation was determined in nontransfected or mGluR7-transfected cerebellar cultures (fromleft to right) in the absence (Basal) and presence of the mGluR1 agonist DHPG (positive control), or different concentrations ofd,l-AP-4. The last bar of the histogram on theright was obtained in PTX-treated cells. Each bar of the histogram represents the mean ± SEM of four independent experiments performed in triplicate. B, Mean (± SEM;n = 5–10) fractional reduction of the whole-cell Ba2+ current recorded in mGluR7-transfected cerebellar granule cells induced by d,l-AP-4 (500 μm), under the following conditions (fromleft to right): in control cells, in cells recorded with an intracellular medium containing 20 mm BAPTA, and after 5 min dialysis of the IP3receptor antagonist heparin (400 μg/μl).
The observed mGluR7-induced IP3 formation should lead to diacylglycerol synthesis and intracellular Ca2+ release from IP3-sensitive stores. The released Ca2+, together with diacylglycerol, should then activate PKC, which in turn blocks P/Q-type Ca2+ channels. In agreement with this hypothesis, the inhibitory effect of d,l-AP-4 on Ba2+ currents was antagonized by the IP3 receptor blocker heparin (400 μg/ml; Fig.6B). Although resistant to 20 mm EGTA, the l-AP-4 induced P/Q type Ca2+ channel inhibition was abolished by the faster Ca2+ chelator BAPTA (20 mm), added to the intracellular recording medium (Fig. 6B).
DISCUSSION
The present results indicated that activation of mGluR7 selectively inhibited P/Q-type Ca2+channels in cultured cerebellar granule cells. This blockade involved a Go-protein and unexpectedly for a group III mGluR, PLC, intracellular Ca2+, and PKC activation. Consistent with these results, we found that mGluR7 stimulated neuronal IP3 formation in a PTX-dependent manner. We therefore propose a model in which mGluR7 activates a Go-protein, the βγ subunits of which directly stimulated PLC, likely in combination with the αo subunit, and induced IP3 and diacylglycerol formation. This in turn results in intracellular Ca2+ release from IP3-sensitive Ca2+stores, PKC activation, and P/Q-type Ca2+channel inhibition (Fig. 7).
Model for mGluR7-induced inhibition of the P/Q-type Ca2+ channels in cerebellar granule cells. mGluR7 activates a Go protein, the αo and βγ subunits of which stimulates a PLC. This results in IP3 and diacylglycerol (DAG) formation. IP3-induced Ca2+ release and DAG stimulate PKC, which in turn blocks the P/Q-type Ca2+ channel. Whether PKC directly phosphorylates the channel or acts on an intermediate protein is not determined.
Transfected granule cells as a model to study the mGluR7 signaling
The cultured cerebellar granule cell preparation provided a more physiological environment than the classical heterologous expression system to study the transduction signaling of transfected neuronal receptors. At the concentrations presently used (0.5 and 1 mm), d,l-AP-4 could not distinguish between the different group III mGluRs. Because the agonist did not affect Ba2+ currents in the absence of transfected mGluR7, native functional group III mGluRs were likely absent at the somatic plasma membrane of cultured cerebellar granule cells. This was confirmed, at least for mGluR7, by our immunolabeling experiments. Indeed, the native receptor was strictly localized in neuritic processes, whereas the transfected receptor was also detected at the somatic membrane. Together, these observations indicated that the d,l-AP-4 effects that we observed on Ba2+ currents in mGluR7-transfected cells certainly resulted from selective activation of this receptor, with the exclusion of any other group III mGluRs.
It could be argued that the unexpected coupling of the transfected mGluR7 with PLC pathway resulted from overexpression of the receptor in the cell body. Although not definitively proved, the following results argued against this hypothesis. First, the native neuritic mGluR7 also activated PLC, as indicated by the d,l-AP-4-mediated IP3 formation observed in nontransfected cerebellar cultures. Second, cotransfection and overexpression of mGluR7 with mGluR2 did not change the coupling characteristics of the latter receptor (our unpublished observations), i.e., selective inhibition of N- and L-type Ca2+ channels (Chavis et al., 1995, 1998).
Indirect Go-protein-mediated inhibition of P/Q-type Ca2+ channels
The mechanisms by which mGluRs block Ca2+ channels in neurons remain controversial. On one hand, a membrane delimited action mediated by direct interaction between Go-protein βγ subunits and Ca2+ channels has been proposed (Trombley and Westbrook, 1992; Choi and Lovinger, 1996; Ikeda, 1996). On the other hand, a slow inhibition of Ca2+ channels recorded in cell-attached patches has also been found and was consistent with the involvement of a soluble intracellular messenger (Chavis et al., 1994). The nature of this messenger remains however to be determined. In the present study, inhibition of P/Q-type Ca2+ channels involved a Go-like protein, the βγ subunits of which did not seem to directly interact with the Ca2+ channel, because this inhibition was neither accompanied by slow activation kinetics of the Ba2+ current, nor removed by a depolarizing prepulses. A direct voltage-insensitive action of G-protein αo subunit on Ca2+ channels has been reported in sympathetic neurons (Delmas et al., 1998). However, such a mechanism did not seem to be involved in cerebellar granule cells because inhibition of PLC or PKC completely abolished the effect of d,l-AP-4 onICa (Fig. 5B).
Now, cloned P/Q-types Ca2+ channels are generally inhibited by direct interaction of G-protein βγ subunits with the I-II loop of the α1A Ca2+ channel subunit (De Waard et al., 1997). However, it has been recently reported that alternative splicing of the α1A gene generates α1A-a and α1A-b subunits with distinct properties. Thus, the presence of a Val421residue in the I-II linker domain of the α1A-b (absent in the α1A-a) splice variant subunit confers to the Ca2+ channel the faculty of being directly inhibited by G-protein βγ subunits (Bourinet et al., 1999). The authors' data also predict that most of α1A transcript in cerebellar neurons should be of the α1A-a subtype. Together these data are consistent with the hypothesis that cultured cerebellar granule cells may predominantly express the α1A-a splice variant subunit, which could explain the absence of direct effect of Go-protein βγ subunits on P/Q-type Ca2+ currents in these neurons on application of mGluR7 agonist.
Activation of Go-protein generally leads to inhibition of N-type Ca2+ channels (Hille, 1994) through a direct action of the G-protein βγ subunits on the channels (Ikeda, 1996), whereas in the present study N-type Ca2+ channels were spared. It is worth noting that in cultured cerebellar granule cells, inhibition of N-type Ca2+ channels was mediated by the Go-protein-coupled mGluR2/3, and this effect did not display the fast kinetics and membrane-delimited voltage-dependent characteristics of a direct action of Go-protein βγ subunits on these channels (Chavis et al., 1994, 1995). This suggests that other mechanisms are involved in our preparation and mediate a selective inhibition of P/Q-type versus N-type Ca2+ channels by mGluR7 and mGluR2/3, respectively. Two tentative hypotheses can be proposed to explain such a selectivity. mGluRs may be colocalized with specific Ca2+ channels in functional microdomains, probably through interaction with scaffold proteins, the nature of which remains however to be identified. Alternatively, particular G-protein subunit combinations may activate specific signaling pathways. For instance, it has been reported that distinct αo, β, and γ subunit combinations display different efficacy in modulating βARK (Muller et al., 1993) or voltage-gated Ca2+ channel activity (Kleuss et al., 1993; Kalkbrenner et al., 1995).
Cellular determinants of the GluR7-mediated activation of PLC
Unexpectedly, we found that the mGluR7-mediated inhibition of the Ca2+ channels was PLC-dependent. The effect was abolished by treatments with PTX or a specific antibody raised against the G-protein αo subunit, indicating that a Go protein activated PLC in cultured cerebellar granule cells. This finding was reminiscent of the Go-mediated activation of PKC in enteric neurons (Pan et al., 1997). The effect of mGluR7 was mimicked by the nonselective G-protein activator GTPγS and blocked by quenching the G-protein βγ subunits with intracellular specific peptides. A likely hypothesis is that G-protein βγ subunits directly acted on a PLCβ in our preparation, as it is the case in various heterologous expression systems (Blank et al., 1992; Boyer et al., 1992, 1994; Camps et al., 1992; Blitzer et al., 1993). However, it is worth noting that intracellular dialysis of purified G-protein βγ subunits did not mimic the inhibitory effect of mGluR7 on Ba2+ currents. Together these observations indicated that G-protein βγ subunits were required, but not sufficient to activate PLC in neurons, and that G-protein αo and βγ subunits were both involved. Although G-protein αo subunit reconstituted in phospholipid vesicles was not required to activate PLC, this subunit shifted to the left the concentration-effect curve for βγ-mediated activation of PLCβ and increased the maximal activity of PLCβ (Boyer et al., 1992). It is therefore possible that under our experimental conditions, such a synergistic effect of G-protein αo subunits on activation of a PLCβ by the βγ subunits was required to reach a level of PLC activity sufficient to activate PKC and inhibit P/Q-type Ca2+channels.
It has been shown that PKC phosphorylates a site located in the domain I-II linker of the cloned α1A Ca2+channel subunit, which results in upregulation of the P/Q-type Ca2+ current (Bourinet et al., 1999). Because phorbol esters or mGluR7 inhibited native P/Q-type Ca2+ channels in cultured cerebellar granule cells, a different PKC phosphorylation site was involved in this effect. An alternative hypothesis is that PKC phosphorylated an intermediate protein that in turn downregulated the Ca2+ channel.
A reason why native P/Q-type Ca2+ channels in cultured cerebellar granule cells were not sensitive to the facilitatory effect of PKC could be that these neurons express the α1A-a Ca2+ channel subunit isoform, as suggested above. Indeed, in addition to be little sensitive to G-protein βγ subunits interaction, this subunit isoform harbors low sensitivity to upregulation mediated by PKC (Bourinet et al., 1999).
Possible physiological consequences of the mGluR7-mediated inhibition of P/Q-type Ca2+ channels
Activation of PLC by mGluR7 was sufficient to inhibit Ca2+ channels in cultured cerebellar granule cells. This observation was consistent with the absence of effect of PKA inhibitors on Ba2+ currents, in the absence or presence of d,l-AP-4, in mGluR7-transfected cerebellar granule cells. Moreover, it has been reported that l-AP-4 activates cAMP-independent and PLC-dependent pathways in mitral olfactory bulb (Schoppa and Westbrook, 1997) and retinal ganglion cells (Shen and Slaughter, 1998) respectively. Also, the efficiency of the coupling between mGluR7 and adenylyl cyclase in BHK cells is relatively low (Saugstad et al., 1994). Altogether these results suggest that the classical inhibition of adenylyl cyclase by group III mGuRs, which has been described in heterologous expression systems (Okamoto et al., 1994; Saugstad et al., 1994; Wu et al., 1998), would not be the primary transduction pathway by which mGluR7 triggers its physiological effects in neurons. We therefore propose that in natural systems this receptor, like group I mGluRs, acts through a PLC-dependent cascade.
To further understand the role of mGluR7 in synaptic transmission, one needs to transpose our results to neuronal synaptic terminals. Like mGluR7 (Shigemoto et al., 1996, 1997; Kinzie et al., 1997), IP3-sensitive Ca2+stores, PKC activity (Rodriguez-Moreno et al., 1998), and P/Q-type Ca2+ channels (Turner et al., 1992;Takahashi and Momiyama, 1993; Regehr and Mintz, 1994; Dunlap et al., 1995) have been found at presynaptic sites and control neurotransmitter release. Therefore, our results anticipate that mGluR7 would downregulate synaptic transmission through activation of PKC. This hypothesis is in apparent discrepancy with the classical phorbol ester-mediated facilitation of transmitter release, but it has been shown that these compounds act independently of PKCγ, the major brain isoform of PKC (Goda et al., 1996).
This does not exclude the possibility that other presynaptic receptors can mediate direct effects of Go-protein on N-type Ca2+ channels. Indeed, it has been shown that although N-type Ca2+ channels can be inhibited by direct and indirect effects of G-proteins in sympathetic neuron somata, only the direct pathway seems to mediate inhibition of transmitter release (Koh and Hille, 1997). The relative importance of a direct versus indirect inhibition of presynaptic Ca2+ channels may depend on the colocalization of Ca2+ channels with Go-protein-coupled receptors. Thus, although the model we propose here (Fig. 7) provides a possible physiological mechanism by which mGluR7 could dampen synaptic transmission, it may not be predominant under low-frequency synaptic activity. Indeed, whereas neurotransmitter glutamate normally lasts shortly in the synaptic cleft, the mGluR7-induced inhibition of Ba2+ current needed a long time to develop and lasted long after washout. Because of these kinetic properties, the mGluR7-activated pathway may be more important under sustained synaptic activity such as during induction of synaptic plasticity or subthreshold epileptic neuronal discharges. In agreement with this hypothesis, inhibition of excitatory synaptic transmission by group III mGluRs, during high- but not low-frequency synaptic activity, has been observed in locus coeruleus (Dube and Marshall, 2000). This hypothesis provides the mGluR7 pathway as neuroprotective and is consistent with physiological studies showing that adult mGluR7 knock-out mice died from epileptic seizures (Masugi et al., 1999).
Footnotes
This work was supported by Centre National de la Recherche Scientifique and grants from Association Française contre les Myopathies, Fondation pour la Recherche Médicale, Bayer (France), and Hoechst-Marrion-Roussel (FRHMR1/9702). We thank J. P. Pin and F. Ango for constructive discussion of this work. We also thank Dr. J. Saugstad (Atlanta, GA) for the rat mGluR7a cDNA, J. M. Sabatier (Marseille, France) for the synthesis of the 68 AA peptide, V. Homburger (Montpellier, France) for the anti-Gαo antibody, and B. Mouillac (Montpellier, France) for the anti-cMyc monoclonal antibody.
Correspondence should be addressed to Dr. L. Fagni, Centre National de la Recherche Scientifique, Unité Propre Recherche 9023, CNRS-INSERM de Pharmacologie et d'Endocrinologie, 141 Rue de la Cardonille, 34000 Montpellier, France. E-mail:fagni{at}bacchus.montp.inserm.fr.