We have shown previously that a selective metabotropic glutamate receptor (mGluR) agonist, 1S,3R-1-aminocyclo-pentane-1,3-dicarboxylate (1S,3R-ACPD), evokes an inward current in CA1 pyramidal neurons of rat hippocampal slices in the presence of K+ channel blockers (Crépel et al., 1994). This current has been characterized as a Ca2+-activated nonselective cationic (CAN) current. Using whole-cell patch-clamp recordings and intracellular dialysis, we now have identified the mGluR subtype and the mechanisms underlying the CAN current (I CAN) and report for the first time the presence of a synaptic I CANin the mammalian CNS. First, we have shown pharmacologically that activation of I CAN by 1S,3R-ACPD involves the group I mGluRs (and not the groups II and III) and a G-protein-dependent process. We also report that I CAN is modulated by the divalent cations (Mg2+, Cd2+, and Zn2+). Second, we have isolated a slow synaptic inward current evoked by a high-frequency stimulation in the presence of K+ channel blockers, ionotropic glutamate, and GABAA receptor antagonists. This current shows similar properties to the exogenously evoked I CAN: its reversal potential is close to the reversal potential of the 1S,3R-ACPD-evokedI CAN, and it is G-protein- and Ca2+-dependent. Because the amplitude and duration of I CAN increased in the presence of a glutamate uptake blocker, we suggest that this synaptic current is generated via the activation of mGluRs. We propose that the synapticI CAN, activated by a brief tetanic stimulation and leading to a long-lasting inward current, may be involved in neuronal plasticity and synchronized network-driven oscillations.
- Ca2+-activated nonselective cationic current
- slow synaptic inward current
- postsynaptic mGluRs
- intracellular perfusion
- whole-cell voltage clamp
- CA1 pyramidal neurons
Metabotropic glutamate receptors (mGluRs; Sugiyama et al., 1987) form a heterogenous family of G-protein-coupled receptors, which play a crucial role in controlling the cell excitability in the CNS. Multiple physiological roles for mGluRs have been identified, including neuronal excitation (increase of cell excitability, potentiation of glutamate release, coinduction, or facilitation of LTP) as well as neuronal inhibition (hyperpolarization, presynaptic inhibition of glutamate and GABA release, coinduction, or facilitation of LTD), and a possible implication in glutamate-induced neurotoxicity (for review, see Schoepp and Conn, 1993; Gallagher et al., 1994; Ben-Ari and Aniksztejn, 1995; Pin and Duvoisin, 1995). The mGluRs are coupled to a large variety of second messenger systems [including activation of phospholipase C (PLC) and inhibition of adenylyl cyclase] and modulate several ligand-gated ionic channels (for review, see Schoepp and Conn, 1993; Pin and Duvoisin, 1995).
In a previous report, using intracellular recordings, we showed that, in the presence of K+ channel blockers, ionotropic glutamate, and GABAA receptor antagonists, the selective mGluR agonist 1S,3R-1-aminocyclopentane-1,3-dicarboxylate (1S,3R-ACPD) evoked an inward current in CA1 pyramidal neurons of the rat hippocampus (Crépel et al., 1994). This current was characterized as a Ca2+-activated nonselective cationic current (CAN current) on the basis of its ionic properties and its blockade by intracellular injection of the Ca2+ chelating agent BAPTA. However, several questions remain unanswered. (1) Does the link between mGluRs and CAN current involve G-proteins? (2) What types of mGluRs are implicated in the activation of this CAN current (and consequently, which second messenger pathway)? (3) Can this current be activated synaptically?
We now have used whole-cell patch-clamp recordings and intracellular dialysis to study the properties of I CAN. Our observations show that (1) I CAN is activated by group I mGluRs via a G-protein- and Ca2+-dependent pathway, (2) I CAN is modulated by divalent cations, and (3) a high-frequency stimulation (HFS) synaptically generates a slow inward current through mGluRs, once glutamate and GABA ionotropic receptors have been blocked. This synaptic current demonstrates features specific to the CAN current; in particular, it is calcium-dependent (blocked by BAPTA) and G-protein-dependent (blocked by G-proteins inhibitors). Therefore, in CA1 pyramidal neurons, in addition to the well characterized fast excitatory currents resulting from the activation of ionotropic AMPA and NMDA receptors, brief tetani generate, via the activation of mGluRs, a long-lasting CAN current. Because of its unique features I CAN may play a major role in neuronal plasticity and network-driven oscillations in neurons of the mammalians CNS.
MATERIALS AND METHODS
Hippocampal slices preparation. Experiments were performed in CA1 hippocampal neurons in slices obtained from 100–150 gm male Wistar rats (20–40 d old). Rats were anesthetized with ether and decapitated. After decapitation, the brain was removed quickly from the skull, and hippocampi were dissected free on ice in a 0–5°C oxygenated artificial cerebrospinal fluid (ACSF) containing (in mm): 126 NaCl, 3.5 KCl, 1 CaCl2, 2 MgCl2, 1.2 NaH2PO4, 25 NaHCO3, and 11 glucose, equilibrated with 95% O2/5% CO2, pH 7.4. Transverse slices (500 μm) were prepared with a MacIlwain tissue chopper and were incubated in ACSF at room temperature, as previously described (Cherubini et al., 1987). After a 2 hr recovery period, hippocampal slices were transferred one at a time to a submerged recording chamber and superfused continuously (2.5–3 ml/min at 28–30°C) with a phosphate-free ACSF (P-free ACSF). The superfusing ACSF did not contain NaH2PO4 to avoid precipitation in the following conditions: (1) in the presence of the K+ channel blockers TEA (5–25 mm) and 4-AP (5 mm), (2) when the external Mg2+concentration was raised (from 2 to 10 mm), or (3) when Cd2+ (200 μm) or Zn2+ (200 μm) was added.
Whole-cell patch-clamp recordings. Whole-cell patch-clamp recordings were obtained from CA1 pyramidal neurons by using the “blind” patch-clamp technique. Membrane currents were recorded in the voltage-clamp mode with an Axopatch 200A amplifier (Axon Instruments, Foster City, CA). Patch electrodes were pulled from borosilicate thin-wall glass capillaries (GC150F-15, Clarck Electromedical Instruments, Pangbourne, UK) with a vertical puller (Glass Microelectrodes Puller PP83, Narishige, London, UK). Broad taper pipettes were used to optimize the solution exchange at their tips. These pipettes had a resistance of 5–7 MΩ when filled with a Cs gluconate (CsGlu) internal solution that contained (in mm): 120 CsGlu, 10 CsCl, 10 NaCl, 1 CaCl2, 2 MgATP, 0.5 GTP, 10 EGTA, and 10 HEPES, pH 7.25 (intracellular-free Ca2+ ≈ 100 nm). In some experiments CsGlu was substituted for equimolar Cs chloride (CsCl). Resting membrane potential was estimated from the potential observed on withdrawal of the electrode from the cell. Holding potential was maintained at −60 mV for all cells recorded. Membrane input conductance was monitored by application of hyperpolarizing voltage steps of 10 mV (0.03 Hz) throughout the experiments. Cells showing changes in intrinsic membrane input resistance during the experiments were discarded.
In all experiments, except during synaptic recordings (see below), 1 μm tetrodotoxin (TTX) and 50 μmmethoxyverapamil (D600), 10 μm bicuculline, and 1 mm kynurenate were present in the P-free ACSF to inhibit Na+ and voltage-dependent Ca2+currents, respectively, and to antagonize GABAA, AMPA, and NMDA receptor-mediated currents, respectively. In these experiments K+ currents were suppressed by concomitant intracellular perfusion of CsGlu-containing pipette solution and by the addition of 6 mm CsCl (to depressI Q), 5–25 mmtetraethylammonium chloride (TEACl; to depressI M, I K, and the fast Ca2+-dependent K+current I C), and 5 mm4-aminopyridine (4-AP; to block the early fast-inactivatingI A and the slow-inactivatingI D currents) to the P-free ACSF (Segal and Barker, 1984; Storm, 1988, 1990). This superfusing medium was called medium A. The glutamatergic [200 μm1S,3R-ACPD, 200 μm3,5-dihydroxyphenylglycine (DHPG), 10 μm(2S,1′R,2′R,3′R)-2-(2,3-dicarboxycyclopropyl) glycine (DCG IV), and 1 mm l-2-amino-4-phosphonobutyrate (l-AP4)] or muscarinic (60–120 μm carbachol) metabotropic receptor agonists, the glutamatergic [1 mm(S)-α-methyl-4-carboxyphenylglycine (S)-MCPG] or muscarinic (10 μmatropine) metabotropic receptor antagonists, and the activator of adenylyl cyclase (50 μm forskolin) or the glutamate uptake inhibitor dihydrokainic acid (DHK; 250–500 μm) were dissolved in this medium A and were bath-applied. The voltage dependence of these metabotropic receptor-induced currents was studied using (+50 to −100 mV, 8 sec) ramp voltage commands: the membrane potential was stepped from V H to +50 mV, held at +50 mV for 500 msec, and then ramped to −100 mV in 7.5 sec (ramp A).
Synaptic recordings were performed with recording pipettes containing 2 mm of N-(2,6-dimethylphenylcarbamoylmethyl) triethylammonium bromide (QX 314; diluted into the CsGlu solution) to block Na+ voltage-dependent channels. In these experiments the superfusing medium contained 20 μmbicuculline, 40 μm CNQX, 200 μm dl-2-amino-5-phosphono-valeric acid (dl-APV) (to block GABAA, AMPA, and NMDA receptor-mediated currents, respectively), K+ channel blockers (5–25 mm TEA, 5 mm 4-AP, and 6 mmCs+), and 1 mm Mg2+(medium B). The voltage dependence of the synaptically evoked current was studied using (−20 mV to −100 mV, 1 sec) ramp voltage commands: the membrane potential was stepped from V H to −20 mV, held at −20 mV for 100 msec, and then ramped to −100 mV in 1 sec (ramp B).
Pipette–whole-cell perfusion. Experiments requiring intracellular perfusion were performed with a 2PK+pipette–whole-cell perfusion kit combined with an MRC-6 multireservoir carrousel (ALA Scientific Instruments, Rockville Center, NY). The reservoirs contained various intracellular solutions (drugs were diluted to their final concentration in the CsGlu pipette solution) and were connected to a common pressure vessel. The outlet from the vessel was fed into a valve, and a thin polyethylene tube (PE 10) connected to the output of this valve was inserted into the recording pipette through the perfusion port of a specific pipette holder. The polyethylene tube ended by a connection to a polyimide-coated quartz microperfusion capillary, thinned down at the other end by a high flame pulling (cleaned and cut for a final tip size of 30–50 μm), extending close to the tip of the recording pipette in a position in which its orifice was approximately one-half of the internal diameter of the pipette. The inlet to the pressure vessel was fed into a valve and connected to a pressure/vacuum generator. Pipette–whole-cell perfusion was achieved by two successive 5 min periods of active perfusion (application of a well defined positive pressure on the inlet to the positive pressure vessel and a conversely negative pressure on the suction port of the pipette holder, both simultaneously generated by the pressure/vacuum generator), interspersed by 5 min of passive diffusion.
Fluorescence measurements. Fluorescence measurements were performed as previously described (Leinekugel et al., 1995) on neurons loaded with the Ca2+-sensitive dye Fluo-3 in the impermeant form (whole-cell configuration CsGlu pipette solution containing 0.01 mm Fluo-3), using a confocal laser scanning microscope (MRC Bio-Rad 600, Munich, Germany) equipped with an argon–krypton laser and photomultiplier. Excitation was delivered at 488 nm, and emission intensity was measured at wavelengths >500 nm. Images were acquired every 10 sec with the program SOM (Bio-Rad) and analyzed off-line with the program Fluo (Imstar, Paris, France). All results were expressed as ΔF/F 0, with Fequal to fluorescence from the defined portion of the image corresponding to the cell under investigation andF 0 equal to the mean base line fluorescence in the selected area from at least five consecutive images. Because Fluo-3 is a single wavelength chromophore and fluorescence is a function of the concentration of Ca2+ and dye (Kao et al., 1989), we have used this dye only for an approximate estimation of [Ca2+]i and included for analysis only experiments in which the fluorescence level recovered to control value after cell excitation. As described earlier (Leinekugel et al., 1995), individual neurons were selected by using the optics of an Axioscope Karl Zeiss microscope (water immersion objective 40×) that allows recognition of pyramidal neurons in slices.
Materials. DCG IV was generously supplied by Dr. K. Shimamoto (Suntori Institute for Bio-organic Research, Osaka, Japan). DHPG, l-AP4, [(S)-MCPG], 1S,3R-ACPD, bicuculline, 6-cyano-7-nitroquin-oxaline-2,3-dione (CNQX), anddl-APV were purchased from Tocris Cookson (Bristol, UK). Fluo-3 was purchased from Molecular Probes (Eugene, OR). Kynurenate and all other drugs were purchased from Sigma (St. Louis, MO).
Data analysis. Membrane responses were digitized and displayed simultaneously on a Nicolet digital oscilloscope and on a computer-driven chart recorder. Data were analyzed off-line on computer (programs G, Sadoc, France) and are presented as means ± SEM. Statistical significance (p ≤ 0.05) was assessed by the Student’s t test analysis (paired data).
Experiments were performed on a homogenous population of 78 pyramidal cells of the CA1 hippocampal region. On average, resting membrane potential was −62 ± 0.5 mV, and input resistance was 173 ± 5.7 MΩ.
In the presence of K+ channel blockers, 1S,3R-ACPD induced a Ca2+-activated nonselective cationic current:ICAN
As previously shown with intracellular recordings (Crépel et al., 1994), with continual application of K+ channel blockers (5–25 mm TEA, 5 mm 4-AP, and 6 mm Cs+ diluted in the superfusing medium A; see Materials and Methods), bath application of 1S,3R-ACPD (200 μm, 2 min) evoked a reversible inward current of −64.4 ± 3.4 pA, peak amplitude atV H = −60 mV (I ACPD; Fig.1 A-a). This current was associated with a significant increase in membrane conductance of 42.9 ± 3.4% (2.2 ± 0.1 and 3.1 ± 0.1 nS in the absence and in the presence of 1S,3R-ACPD, respectively;n = 55, p = 0.0001). The voltage dependence of I ACPD was studied by plottingI/V relations derived from hyperpolarizing ramp commands from +50 to −100 mV (ramp A; see Materials and Methods).I ACPD had a reversal potential of −17 ± 1.3 mV (n = 55), slightly more negative than that predicted by the Nernst equation (+3.5 mV) at the experimental conditions of [cations]o = 160.5 mm and [cations]i = 140 mm (but see Discussion). The I/V relation ofI ACPD, obtained by subtracting the current recorded in the presence of 1S,3R-ACPD from that recorded in the absence of 1S,3R-ACPD, was not linear but displayed an area of reduced slope conductance at voltages more negative than −40 mV (Fig. 1 D,E). The conductances calculated for the two representative membrane potential intervals, −100 to −60 mV (which will be referred to as negative conductance) and 0 to +40 mV (which will be referred to as positive conductance) were significantly different, 0.92 ± 0.08 and 2.1 ± 0.1 nS, respectively (n = 55,p = 0.0001), showing a negative rectification of −48 ± 5.7% (n = 55). These current properties were observed similarly in the presence of CsGlu-containing and CsCl-containing pipette solution (data not shown), confirming the chloride gradient independence of the current.
The dependence of I ACPD on [Ca2+]i was studied first by recording CA1 pyramidal cells in the whole-cell configuration in the presence of Fluo-3 diluted in the CsGlu pipette solution (see Materials and Methods). Changes of [Ca2+]i were monitored in the soma of pyramidal neurons with a confocal laser scanning microscope simultaneously to the recording ofI ACPD (performed in the presence of the superfusing medium A). As previously described in cultured hippocampal neurons and CA1 pyramidal cells (Mayer and Miller, 1990; Frenguelli et al., 1993; Jaffe and Brown, 1994; Petrozzino and Connor, 1994;Shirasaki et al., 1994), 1S,3R-ACPD induced a significant rise of [Ca2+]i of 49 ± 14% (V H = −60 mV; n = 4,p < 0.05) (Fig. 2). The increase of [Ca2+]i was correlated closely to theI ACPD simultaneously recorded, and both displayed similar latency (47.5 ± 10.3 and 60 ± 10.8 sec, respectively; n = 4, p = 0.15) and time to peak (127.5 ± 16 and 136.3 ± 9 sec, respectively;n = 4, p = 0.37) (Fig. 2).
To show directly that I ACPD depends on the rise in [Ca2+]i, we used pipette–whole-cell perfusion with the Ca2+chelating agent BAPTA (20 mm) diluted in the CsGlu pipette solution. Two series of preliminary experiments were performed to confirm the efficacy of the pipette–whole-cell perfusion system. The first consisted of three successive applications of 1S,3R-ACPD, one before and two after a perfusion with the same CsGlu pipette solution. We showed that the perfusion system by itself did not change I ACPDsignificantly (see Fig. 3), which could be evoked by up to six consecutive applications of 1S,3R-ACPD (data not shown). The second experiment consisted of testing the effects of whole-cell perfusion of BAPTA on the slow afterhyperpolarization (sAHP) after a depolarizing pulse. As previously described with intracellular recordings (Crépel et al., 1994), 10 min after BAPTA perfusion spike frequency accommodation and sAHP disappeared, confirming adequate BAPTA loading (data not shown). In control conditions (before intracellular perfusion of BAPTA), 1S,3R-ACPD generated a reversible inward current (peak amplitude = −86.2 ± 9 pA; reversal potential = −17.6 ± 4.4; V H = −60 mV;n = 5) (Fig. 1 A-a) associated with an increase in membrane conductance of 51.6 ± 13% (n = 5). After intracellular BAPTA perfusion (20 mm added to the CsGlu internal solution), subsequent applications of 1S,3R-ACPD failed to induce inward current (Fig. 1 A-b); under this condition the 1S,3R-ACPD-mediated current was nearly abolished (the peak amplitude was reduced by 93.7 ± 2.9%;n = 5, p = 0.0004) and was not associated with a significant increase in membrane conductance (0.8 ± 2.6%; n = 5, p = 0.004). Analysis of the I/V relations (ramp command A) confirm that BAPTA perfusion completely suppressed the 1S,3R-ACPD-mediated current (Fig.1 C–E). This showed that I ACPD is Ca2+-dependent, as previously reported (Crépel et al., 1994).
In conclusion, using whole-cell recordings, we have confirmed that, in the presence of K+ channel blockers, 1S,3R-ACPD induced a current with a reversal potential close to that of a nonselective cationic current, which was triggered by a rise in intracellular Ca2+concentration. This current will be referred to hereafter as a calcium-activated nonselective cationic current (I CAN), as per Crépel et al. (1994); its properties are not modified by whole-cell recordings.
ICAN is mediated via G-proteins
To clarify the transduction system leading to an increase in [Ca2+]i required for the activation of the CAN current, we used intracellular dialysis of the specific G-protein inhibitors GTPγS and GDPβS.
In the first set of experiments we investigated whether 1S,3R-ACPD still could generateI CAN when G-proteins were blocked in an activated state. GTPγS (500 μm) was diluted in the GTP-free CsGlu pipette solution to replace GTP at equimolar concentration. In five of five cells, in the presence of GTPγS, the first application of 1S,3R-ACPD (200 μm, 2 min) evoked an irreversibleI CAN (peak amplitude = −47 ± 5.5 pA; reversal potential = −19 ± 6.8 mV) (Fig.4 A-a); the second application of 1S,3R-ACPD failed to evoke any subsequent current, as illustrated in Figure 4 A-b(I CAN was reduced by 98.2 ± 1.8%;n = 5, p = 0.0001). The analysis of theI/V relations (ramp command A) confirm that GTPγS completely suppressed the second 1S,3R-ACPD-mediated current (Fig.4 B).
In the second set of experiments we examined whether 1S,3R-ACPD could induceI CAN when G-proteins were blocked in an inactivated state. In this set of experiments GDPβS was dialyzed into the cell via the pipette–whole-cell perfusion system; GDPβS (500 μm) was diluted in the GTP-free CsGlu pipette solution to replace GTP at equimolar concentration. In control conditions, before perfusion of GDPβS, a first application of 1S,3R-ACPD induced the expected fully reversibleI CAN (peak amplitude = −79.4 ± 10.9 pA; reversal potential = −16.43 ± 1.9 mV; n= 7). In the presence of GDPβS, a subsequent application of 1S,3R-ACPD failed to induceI CAN (Fig. 4 C): the second 1S,3R-ACPD-mediated current was nearly abolished (the peak amplitude of I CAN was reduced by 94.7 ± 6.4%; n = 7, p = 0.0007). The analysis of the I/V relations (ramp command A) confirmed that GDPβS completely suppressed the second 1S,3R-ACPD-mediated current (Fig.4 C).
Therefore, 1S,3R-ACPD activatesI CAN via a G-protein-dependent process.
ICAN is mediated by group I mGluRs
Glutamatergic metabotropic receptors are divided into three groups. Group I (including mGluRs 1 and 5) activates PLC [i.e., the inositol triphosphate (IP3) production pathway]; group II (including mGluRs 2 and 3) and group III (including mGluRs 4, 6, 7, and 8) inhibit adenylyl cyclase (i.e., the cAMP production pathway) (for review, see Schoepp and Conn, 1993; Pin and Duvoisin, 1995). To specify the type of mGluRs involved in the activation ofI CAN, we performed pharmacological experiments to test the ability of selective agonists for these three groups to induce I CAN. In each cell in which mGluRs agonists were tested, we subsequently applied 1S,3R-ACPD (200 μm, 2 min) to verify the ability of these cells to generateI CAN; only cells that displayedI CAN were kept for analysis. These data are summarized in Table 1.
Previous immunohistochemical studies have shown that CA1 pyramidal neurons prominently express mGluR5 (Abe et al., 1992; Shigemoto et al., 1993; Luján et al., 1996). Therefore, we first tested the most selective agonist of the group I mGluRs, the DHPG (Ito et al., 1992;Schoepp et al., 1994). With superfusion of medium A (see Materials and Methods), bath application of 200 μm DHPG evoked a current identical to that evoked by 1S,3R-ACPD. The currents evoked by DHPG and 1S,3R-ACPD had similar peak amplitudes, reversal potentials, and slope conductances; they both exhibited an area of reduced slope conductance at potentials more negative than −40 mV (see Fig. 5, Table 1). Second, we tested the ability of specific agonists of the group II (DCG IV; Ishida et al., 1993) and the group III (l-AP4; (Thomsen et al., 1992) mGluRs to induce I CAN. Under perfusion with medium A (see Materials and Methods), DCG IV (10 μm; n = 6) and l-AP4 (1 mm; n = 5) were bath-applied for 2–10 min at concentrations 10–50 times higher than their reported EC50 in hippocampal slices (Gereau and Conn, 1995). We found that DCG IV and l-AP4 failed to evoke any current (Table 1), whereas in the same cells 1S,3R-ACPD evoked I CAN (data not shown).
Finally, we tested the ability of the competitive nonselective mGluR antagonist (S)-MCPG (Eaton et al., 1993) to antagonize the DHPG-evoked I CAN. (S)-MCPG (1 mm; n = 3) was bath-applied after recording a first DHPG-evokedI CAN in a CA1 pyramidal neuron at least 15–30 min before a second application of DHPG (100–200 μm) on the same neuron. (S)-MCPG failed to induce a significant reduction of the DHPG-evokedI CAN, which displayed similar peak amplitudes (−62.8 ± 13.6 and −68.6 ± 20.3 pA;n = 3, p = 0.3) and slope conductances (negative conductances: 0.2 ± 0.47 and 0.3 ± 0.7 nS;n = 3, p = 0.35; positive conductances: 2.4 ± 0.4 and 1.9 ± 0.65 nS; n = 3,p = 0.11), respectively, in the absence and in the presence of (S)-MCPG (data not shown).
These results strongly suggest that I CAN was evoked by metabotropic receptors positively linked to the IP3 production pathway (i.e., mGluRs of the group I) and not by metabotropic receptors negatively linked to the adenylyl cyclase pathway (i.e., mGluRs of the group II and III). If so, other types of metabotropic receptors linked to the IP3 production pathway (for review, see Schoepp and Conn, 1993) should activateI CAN also. In keeping with this hypothesis and in agreement with previous data (Shen and North, 1992; Colino and Halliwell, 1993; Fraser and MacVicar, 1996), we found that activation of muscarinic receptors induced a current showing similar features to those of I CAN: under perfusion with medium A, bath application of 60–120 μm carbachol (a selective agonist of muscarinic receptors linked to IP3 production pathway) (Fisher et al., 1983, 1984; Dutar and Nicoll, 1988) generated a CAN current that was smaller but otherwise comparable toI CAN induced by 1S,3R-ACPD (n = 9; Fig. 6 C, Table 1). To demonstrate further that I CAN is not attributable to a change of intracellular cAMP content, we tested the ability of 1S,3R-ACPD to induceI CAN in the presence of forskolin, an activator of adenylyl cyclase (which increases the intracellular concentration of cAMP). In this set of experiments 50 μm forskolin (diluted in medium A) was bath-applied before (10 min), during (2 min), and after (5 min) the application of 1S,3R-ACPD (2 min, 200 μm). First, we observed that forskolin itself did not generate any significant current (−0.21 ± 6 pA;n = 8); second, in the presence of forskolin, the current evoked by 1S,3R-ACPD was not significantly different from that evoked in the absence of forskolin (Fig. 6 A): neither the peak amplitude (96.5 ± 27% of the control; n = 8, p = 0.15) nor the reversal potential (−8.5 ± 3.3 and −8.3 ± 3.7 mV, respectively, in the presence and in absence of forskolin;n = 8, p = 0.3) nor the I/Vrelation of I CAN was altered by forskolin (Fig.6 B).
Therefore, in CA1 pyramidal cells the CAN current can be activated by group I mGluRs and muscarinic receptors, which are positively linked to PLC pathway. Our results also demonstrate that the transduction system involved in the activation of I CAN does not implicate a change of cAMP production.
ICAN is sensitive to external divalent cations
As shown in Figures 1, 2, 3, 4, 5, 6, the I/V relation ofI CAN exhibited an area of reduced slope conductance at potentials more negative than −40 mV. We reasoned that this negative rectification could be attributable to a voltage-dependent block by external divalent cations, as has been demonstrated for NMDA channels (Ascher and Nowak, 1988; Mayer et al., 1989) or cyclic nucleotide-gated nonselective channels (for review, seeZufall et al., 1994).
We thus investigated the possible regulatory role of external Mg2+ on I CAN evoked by 1S,3R-ACPD in seven cells superfused with medium A containing the following external Mg2+concentrations: 0.1, 2, and 10 mm. Attempts to investigate effects of medium A containing zero external Mg2+were abandoned, because we failed to obtain stable long-lasting recordings with such superfusing solution (K+channel blockers, low Ca2+, and 0 Mg2+). The results of these experiments are illustrated in Figure 7 A and summarized in Table 1. In the presence of 0.1 mm Mg2+,I CAN exhibited a larger peak amplitude than that previously observed with physiological Mg2+concentration; its negative and positive conductances were nearly identical, and consequently its I/V relation was linear. When the external Mg2+ concentration was raised from 0.1 to 2 mm, we observed a significant decrease of the peak amplitude of I CAN associated with a significant reduction of the negative conductance (−58.8 ± 3.7%;p = 0.0001). The positive conductance was unchanged, leading to a nonlinear I/V curve. The reversal potential ofI CAN in 2 mm Mg2+was similar to that observed in 0.1 mmMg2+. When the external Mg2+concentration was raised from 2 to 6–10 mm, the peak amplitude and the negative conductance of I CANwere decreased further, but the positive conductance also was strongly reduced, and the reversal potential was shifted to a more positive value.
Thus, as for NMDA and cyclic nucleotide-gated currents, there is a voltage-dependent regulation of the CAN current by physiological external Mg2+ concentrations (between 1 and 2 mm) at negative membrane potentials. Higher external Mg2+ concentration also shifted the reversal potential to more positive values and reduced CAN current at positive membrane potentials.
As for NMDA currents and cyclic nucleotide-gated currents (Ascher and Nowak, 1988; Mayer et al., 1989; Zufall et al., 1994), other divalent cations such as Cd2+ and Zn2+also modulate I CAN. In the presence of 200 μm of external Cd2+ the peak amplitude of I CAN was reduced significantly, and its reversal potential shifted to more positive values; both negative and positive conductances were strongly reduced (n = 14; Fig. 7 B, Table 1). Similarly, in the presence of 200 μm of external Zn2+ peak amplitude ofI CAN was reduced, and its reversal potential shifted to more positive values. As with Cd2+ and high concentrations of Mg2+, both negative and positive conductances were strongly reduced by Zn2+(n = 12; Fig. 7 C, Table1).
In conclusion, external divalent cations regulate the CAN current by reducing its slope conductance. Their effects can be classified in two groups: the voltage-dependent blockers, such as Mg2+(at physiological concentrations), and the non- (or weak) voltage-dependent blockers, such as Cd2+ and Zn2+ (or high concentrations of Mg2+).
ICAN can be evoked synaptically
We determined the conditions required for synaptic activation ofI CAN to elucidate its possible role in synaptic transmission. In these experiments the CA1 region was isolated surgically from CA3. Synaptic responses were evoked by a bipolar stimulating electrode placed in the stratum radiatum or at the border between stratum radiatum and stratum lacunosum moleculare and recorded in the presence of K+ channel blockers. Low-frequency stimulations (0.05 Hz) evoked postsynaptic responses (Fig. 8 A), which were abolished completely by bicuculline, CNQX, and dl-APV (medium B; see Materials and Methods) (Fig. 8 A-b,A-c). However, HFS (25–100 Hz, 1 sec) evoked a slow inward current that was maximal at the frequency of 100 Hz (as illustrated in Fig.8 B-a). The current evoked by a 100 Hz HFS had a peak amplitude of −42.9 ± 3.5 pA (V H = −60 mV; n = 36) and a total duration of 11 ± 1.3 sec (n = 36). The amplitude of this postsynaptic current was dependent on the frequency of stimulation (Fig.8 B-a), the stimulus intensity (Fig.8 B-c), and the duration of the HFS (Fig.8 B-b), suggesting its dependence on the level of transmitter release. Furthermore, this current was blocked completely by bath application of TTX (1 μm), confirming its synaptic origin (n = 5; Fig. 9 A-a,A-c). The voltage dependence of synaptically evoked currents was studied using ramp voltage commands applied at the peak amplitude of the evoked response. To avoid the voltage-dependent activation of calcium channels, we limited the ramp voltage commands to negative potentials (−20 mV to −100 mV, ramp B; see Materials and Methods). The recorded synaptically evoked current showed a reversal potential of −22.8 ± 8.2 mV (close to the reversal potential ofI ACPD) and a conductance of 0.7 ± 0.11 nS (calculated between −100 and −60 mV; n = 13) (Fig. 9 C). Thus HFS evoked a slow inward postsynaptic current resistant to ionotropic glutamate and GABA receptor antagonists.
Previous studies have described slow excitatory synaptic responses mediated by muscarinic receptors in CA1 and activated by HFS (Cole and Nicoll, 1984; Madison et al., 1987). Therefore, we tested the ability of the muscarinic receptor antagonist atropine to block the HFS-evoked response. In five of five cells, atropine (up to 20 μm) changed neither the peak amplitude (−2.2 ± 2.2%; paired data;p = 0.2) nor the duration (−8.2 ± 3.4%; paired data; p = 0.09) of the response (Fig.9 A-a,A-b).
To study the possible activation of the synaptically evoked inward current by nonionotropic glutamate receptors, we tested its sensitivity to the glutamate uptake inhibitor DHK (Watkins and Evans, 1981). As shown in Figure 9 B, bath application of DHK (250 to 500 μm for 15 min) significantly increased the duration of the HFS-evoked synaptic current (by 120 ± 18.5%;n = 5, p = 0.0002; paired data). DHK also slightly increased the peak amplitude of the current in two of the five cells. Thus, in CA1 pyramidal cells HFS evoked a nonionotropic glutamate-mediated excitatory postsynaptic current.
As shown for the 1S,3R-ACPD-evoked CAN current, the synaptically evoked current also involves Ca2+- and G-protein-dependent processes. In the presence of BAPTA, GDPβS, or GTPγS added to the pipette solution, the first responses evoked by the HFS within the 5 min after the passage to whole-cell configuration were similar to the responses evoked in the absence of BAPTA, GDPβS (Fig. 10 A-a, B-a, respectively), or GTPγS (data not shown). After 20 min of cell dialysis with BAPTA (20 mm; n = 5; Fig.10 A-b) or GDPβS (500 μm;n = 7; Fig. 10 B-b), a second HFS failed to evoked a response. With GTPγS, after 20 min of cell dialysis the first response evoked by HFS was significantly longer in four of six cells (29 ± 3.1 vs 11 ± 1.3 sec in control,p = 0.001) and irreversible in the two other cells (Fig. 10 C-a); subsequent HFS-evoked responses were nearly abolished (Fig. 10 C-b).
Therefore HFS generates a mGluRs-dependent slow inward current that is triggered by a rise in [Ca2+]i, mediated by G-proteins. This current exhibit properties of a synapticI CAN.
The present report describes a slow nonselective cationic current (I CAN) triggered by a [Ca2+]i rise, generated by glutamate acting on group I mGluRs via a G-protein-dependent process. It provides the first characterization in the mammalian CNS of a synapticI CAN generated by HFS. The presence of a mGluR-mediated I CAN in the CNS has several important physiological implications.
Properties of ICAN evoked by 1S,3R-ACPD in CA1 pyramidal neurons
In the CNS several studies have described an inward current associated with an increase in membrane conductance, mediated by exogenous activation of glutamate metabotropic receptors, in the presence of K+ channel blockers (Glaum and Miller, 1992; Staub et al., 1992; Mercuri et al., 1993). A nonselective cationic current induced by 1S,3R-ACPD was, however, first described in CA1 pyramidal neurons of adult rat hippocampal slices by Crépel et al. (1994). In that work the authors characterized a Ca2+-dependent, weakly temperature-sensitive, nonselective cationic current corresponding to the CAN cationic current defined in invertebrate preparations (Swandulla and Partridge, 1990). The 1S,3R-ACPD-induced current described in the present study using whole-cell patch-clamp recordings has similar features to those currents described above, namely in terms of nonselectivity for monovalent cations, reversal potential, and dependence on [Ca2+]i.
CAN channels have been found principally to be permeable to Na+ and K+ (for review, seeSwandulla and Partridge, 1990) and occasionally to Cs+ ions (Yellen, 1982; Lipton, 1986; Simmoneau et al., 1987; Teulon et al., 1987). In our experimental conditions we have observed that the reversal potential of the CAN current was more negative than that expected from the Nernst equation. We suspect that it may be attributable to a better permeability of CAN channels for Cs+, which will tend to displace the reversal potential to more negative values. Using the Goldman–Hodgkin–Katz equation (Goldman, 1943; Hodgkin and Katz, 1949), we estimated, under our experimental conditions, a permeability ratioP Cs +/P Na += 3.
Additionally, hippocampal I CAN is modulated by the external divalent cations Mg2+, Cd2+, and Zn2+ via two pathways: a voltage-dependent block (Mg2+) and a voltage-independent one (Cd2+, Zn2+). These properties, and in particular the voltage-dependent block by external Mg2+, are reminiscent of other types of cationic channels, including NMDA channels (Ascher and Nowak, 1988; Mayer et al., 1989), cyclic nucleotide-gated channels (for review, see Zufall et al., 1994), and a nonselective cationic conductance recently described in CA3 pyramidal cells (Caeser et al., 1993; Guérineau et al., 1995). The voltage-dependent modulation of I CAN suggests that CAN channels are regulated directly by external Mg2+. It is not clear, however, whether this property will enable mGluRs to exert a coincident detection property that typically is associated with NMDARs. Future studies will have to determine the consequences of this feature.
ICAN is generated via a G-protein-dependent process by group I mGluRs
Previous studies have shown clearly that mGluRs are linked to phospholipase C and adenylyl cyclase via G-proteins (for review, seeSchoepp and Conn, 1993; Pin and Duvoisin, 1995). As expected, we found that activation of I CAN by 1S,3R-ACPD involved a G-protein-dependent process. In the presence of GTPγS, which blocks the G-protein in its activated state, 1S,3R-ACPD irreversibly activated I CAN, whereas in the presence of GDPβS, which blocks the G-protein in its inactivated state, 1S,3R-ACPD did not evoke any current. These two complementary results clearly demonstrate the role of a G-protein-dependent process in the 1S,3R-ACPD-induced I CANand further confirm that this current is activated by mGluRs via a metabotropic pathway (Sladeczek et al., 1985; Sugiyama et al., 1987).
Previous immunohistological studies have shown that group I mGluRs are expressed prominently in the CA1 region of hippocampus (Abe et al., 1992; Shigemoto et al., 1993; Luján et al., 1996), whereas group II mGluRs are not (Ohishi et al., 1994). In keeping with this (also seeGereau and Conn, 1995), in CA1 pyramidal neuronsI CAN is mediated by group I mGluRs, which are known to be positively linked to the IP3 production pathway and do not implicate a change of adenylyl cyclase activity, because (1) DHPG, a selective agonist of group I mGluRs (mGluR 1 and 5) (Ito et al., 1992; Schoepp et al., 1994), generates a current identical to that evoked by 1S,3R-ACPD; (2) DCG IV andl-AP4, selective agonists of group II and III mGluRs, respectively, which are both negatively linked to the adenylyl cyclase pathway (Tanabe et al., 1992, 1993), do not evokeI CAN; and (3) forskolin, an activator of adenylyl cyclase, does not modify I CAN.
Because mGluR5s are expressed prominently in CA1 pyramidal neurons (Abe et al., 1992; Shigemoto et al., 1993; Luján et al., 1996), whereas mGluR1 subtypes are localized in CA1 interneurons (Martin et al., 1992; Baude et al., 1993; Luján et al., 1996), the CAN current described in the present report probably is mediated by mGluR5. In addition, the absence of effect of MCPG provides further evidence that I CAN involved the activation of mGluR5 and not mGluR1, because MCPG antagonizes mGluR1-mediated, but not mGluR5-mediated, responses (Joly et al., 1995) and has no effect on CA1 pyramidal neurons (Chinestra et al., 1993; Izumi and Zorumski, 1994;Manzoni et al., 1994; Selig et al., 1995) (but see Bashir et al., 1993;Brown et al., 1994). Direct evidence in support of this must await the development of a selective mGluR5 antagonist.
Our pharmacological study shows that I CAN is evoked by group I mGluRs, which are known to be positively linked to the IP3 production pathway. We then suggest thatI CAN can be activated by other receptors linked to the same messenger pathway. In keeping with this and in agreement with previous data (Shen and North, 1992; Colino and Halliwell, 1993;Fraser and MacVicar, 1996), we found that activation of muscarinic receptors, linked to the IP3 production pathway in the CA1 area (Fisher et al., 1983; Dutar and Nicoll, 1988), also activates a current with the same feature as I CAN.
Properties and functional significance of a synapticICAN generated by mGluRs
Several lines of evidence show that the current activated synaptically by a brief HFS of afferent fibers in the stratum lacunosum moleculare–radiatum area was similar to theI CAN evoked by exogenous application of 1S,3R-ACPD. HFS induced an excitatory postsynaptic response composed of a fast ionotropic component and a slow metabotropic component that persist in the presence of ionotropic glutamate and GABAA receptor antagonists. This slow inward current increases in amplitude and duration in the presence of glutamate uptake inhibitor and is not blocked by muscarinic receptor antagonists. In addition, it is blocked by the Ca2+chelating agent BAPTA and by G-protein inhibitors (GTPγS and GDPβS), and it displayed a reversal potential similar to that of the 1S,3R-ACPD-activated I CAN. Taken together, these data strongly suggest that this slow synaptic inward current is generated, as I ACPD, through mGluRs via G-protein- and Ca2+-dependent activation of an excitatory postsynaptic nonselective cationic currentI CAN. Although synaptic currents evoked in the presence of ionotropic glutamate receptor antagonists have been reported in earlier studies (Charpak and Gähwiler, 1991; Glaum and Miller, 1992; Batchelor et al., 1994; Pozzo Miller et al., 1995;Batchelor and Garthwaite, 1997), to the best of our knowledge the present report is the first description of a synaptically triggered CAN current in mammalian CNS neurons. This synapticI CAN exhibited a slow kinetic and may mediate a long-lasting excitation outlasting the action of most voltage-dependent ionic currents (as NMDA, for instance). This property is in keeping with the specific slow kinetics of the CAN single channel observed in several invertebrate preparations (for review, see Swandulla and Partridge, 1990). In the present study the synapticI CAN required relatively strong HFS to be generated. This is likely the consequence of the perisynaptic location of mGluR5 in CA1 pyramidal neurons (Baude et al., 1993; Luján et al., 1996). An accumulation of glutamate in the synaptic cleft and a “spillover” may be required to reach the perisynaptic receptors and trigger a mGluR-mediated response. Therefore, at least in CA1 pyramidal neurons I CAN probably does not participate in ongoing spontaneous activity; a different situation may prevail in interneurons (see Miles and Poncer, 1993).
The demonstration of a synaptically activated CAN current mediated by mGluRs adds a novel facet to the pleiotropic repertoire of modulations exerted by glutamate metabotropic receptors in the CNS. In the hippocampus, activation of cationic currents through metabotropic receptors generates long-lasting plateau potentials (Fraser and MacVicar, 1996) and provides the basis for synchronized activities (Yaari and Jensen, 1989; Miles and Poncer, 1993; Bianchi and Wong, 1995). The synaptically activated I CAN described in the present report may play an important role in synaptic integration, notably as frequency sensors, in regard to its strong frequency dependence. Interestingly, Batchelor and Garthwaite (1997)recently have described, in the cerebellum, an inward current generated through mGluRs that integrates temporally dispersed signals from two different inputs. It will be important to establish whether this synaptic response is mediated by a CAN current.
We suggest that CAN currents activated in the hippocampus, presumably as a consequence of an important accumulation of glutamate, may contribute not only to the normal physiological functions, such as the neuronal plasticity, but also to the long-lasting excitotoxic depolarizations observed in pathological conditions, such as epilepsy or ischemia.
Financial support from Institut National de la Santé et de la Recherche Médicale, Centre National pour la Recherche Scientifique, and Direction des Recherches, Etudes et Techniques, is acknowledged. We are grateful to Dr. K. Schimamoto for her generous gift of (2S,1′R,2′R,3′R)-2-(2,3-dicarboxycyclopropyl) glycine and to Drs. H. McLean, C. Hammond, S. Williams, and D. D. Fraser for their helpful comments and discussions.
Correspondence should be addressed to Dr. Patrice Congar, Institut National de la Santé et de la Recherche Médicale Unité 29, 123 Boulevard de Port Royal, 75674 Paris Cedex 14, France.