Metabotropic glutamate receptor modulation of voltage-gated Ca2+ channels involves multiple receptor subtypes in cortical neurons

Metabotropic glutamate receptor (mGluR) modulation of voltage-gated Ca2+ channels was examined in isolated deep layer frontoparietal cortical neurons under conditions designed to isolate calcium- independent modulatory pathways. Trans-1-aminocyclopentane-1,3- dicarboxylate (t-ACPD), a nonspecific mGluR agonist, produced rapid and reversible inhibition of Ca2+ channels. This effect was mimicked by agonists for group I and group II, but not group III, mGluRs. Effects of group I and II agonists often were observed in the same neurons, but separate subgroups of neurons were unresponsive to the group I agonist quisqualate or the group II agonist 2-(2,3-dicarboxycyclopropyl) glycine (DCG-IV). Inhibition by quisqualate and DCG-IV was nonocclusive in neurons responding to both agonists. These agonists thus appear to act on different mGluRs. The mGluR antagonist alpha-methyl-4- carboxylphenylglycine attenuated inhibition by t-ACPD, quisqualate, and DCG-IV. Inhibition by quisqualate and DCG-IV was voltage-dependent. Although the effects of both agonists were greatly reduced by N- ethylmaleimide (NEM), inhibition by DCG-IV was more sensitive to NEM than inhibition by quisqualate. t-ACPD-induced inhibition was reduced by omega-conotoxin GVIA (omega-CgTx) and omega-agatoxin IVA (omega- AgTx) but was affected little by nifedipine. Inhibition by DCG-IV and quisqualate also was reduced by omega-CgTx. We conclude that multiple mGluR subtypes inhibit Ca2+ channels in cortical neurons and that N- and possibly P-type channels are inhibited. Modulation is via a rapid- onset, voltage-dependent mechanism that likely involves a pertussis toxin (PTX)-sensitive G-protein. Type I mGluRs may work via additional PTX-insensitive pathways.


Metabotropic
glutamate receptor (mGluR) modulation of voltage-gated Cazf channels was examined in isolated deep layer frontoparietal cortical neurons under conditions designed to isolate calcium-independent modulatory pathways. Trans-laminocyclopentane-1,3-dicarboxylate (f-ACPD), a nonspecific mGluR agonist, produced rapid and reversible inhibition of Ca*+ channels. This effect was mimicked by agonists for group I and group II, but not group III, mGluRs. Effects of group I and II agonists often were observed in the same neurons, but separate subgroups of neurons were unresponsive to the group I agonist quisqualate or the group II agonist 2-(2,3-dicarboxycyclopropyl) glycine (DCG-IV). Inhibition by quisqualate and DCG-IV was nonocclusive in neurons responding to both agonists. These agonists thus appear to act on different mGluRs. The mGluR antagonist ol-methyl-4-carboxylphenylglycine attenuated inhibition by f-ACPD, quisqualate, and DCG-IV. Inhi-bition by quisqualate and DCG-IV was voltage-dependent. Although the effects of both agonists were greatly reduced by N-ethylmaleimide (NEM), inhibition by DCG-IV was more sensitive to NEM than inhibition by quisqualate.
f-ACPD-induced inhibition was reduced by o-conotoxin GVIA (w-CgTx) and o-agatoxin IVA (w-AgTx) but was affected little by nifedipine. Inhibition by DCG-IV and quisqualate also was reduced by o-CgTx. We conclude that multiple mGluR subtypes inhibit Ca"+ channels in cortical neurons and that N-and possibly P-type channels are inhibited. Modulation is via a rapid-onset, voltage-dependent mechanism that likely involves a pertussis toxin (PTX)-sensitive G-protein, Type I mGluRs may work via additional PTX-insensitive pathways.
Activation of mGluRs alters the function of a variety of subtypes of high-voltage-activated Ca2+ channels in neurons from the brain and nodose ganglion (Lester and Jahr, 1990;Sahara and Westbrook, 1993;Swartz and Bean, 1992;Trombley and Westbrook, 1992;Sayer et al., 1992;Hay and Kunze, 1994;Stefani et al., 1994;Chavis et al., 1994). mGluR inhibition of N-type, L-type, to be involved in this action (Swartz and Bean, 1992;Hay and Kunze, 1994). PTX-sensitive G-proteins are involved in modulation by mGluRs in some neurons (Sahara and Westbrook, 1993;Hay and Kunze, 1994;Chavis et al., 1994). mGluRs also act through other mechanisms, including slowly developing, possibly Ca" -dependent mechanisms in hippocampal and cortical neurons (Sayer et al., 1992;Sahara and Westbrook, 1993), and a slowly developing, possibly calcium-independent mechanism in cerebellar granule cells (Chavis et al., 1994).
We wished to characterize further mGluR modulation of Cazt channels involving calcium-independent pathways in deep-layer neocortical neurons by determining the following: (1) whether different mGluR subtypes inhibit Ca2+ channels in the same cortical neurons; (2) whether the modulation of Ca2+ channels by different mGluR subtypes involves similar mechanisms: and (3) what subtypes of Ca2+ channels are modulated.

MATERIALS AND METHODS
Cortical neuron preparation.
Brain slices were prepared from 9-to 20-dold Sprague-Dawley rats using previously described techniques (Lovinger, 1991;Lovinger et al., 1993). Rats were killed by decapitation, and the brains were cooled in ice-cold artificial CSF (aCSF) containing 124 mM NaCl, 4.5 mM KC], 2.0 mM CaCI,, 1.5 mM M&I,, 26 mM NaHCO,, 1.2 mM NaH,PO,. and 10 mM o-alucose adiusted to oH 7.4 bv bubbline '. 7. I , with 95% 0,/5% CO,. Corona1 sections (400 pm ihick) were cut in ice-cold medium using a manual vibroslice (World Precision Instruments, New Haven, CT) vibrating slicer. Slices were then transferred onto a nylon net submerged in aCSF at room temperature (2l-24"C), and the aCSF was constantly bubbled with 95% 0,/5% CO,. After 1 hr, a chosen slice was hemisected. Hemislices containing the cortex and striatum just anterior to the globus pallidus were incubated at 37°C for 20-Z min in aCSF containing 0.05 mg/ml pronase (Calbiochem, La Jolla, CA) bubbled with 95% 0,/5% CO,. Slices were transferred to a solution containing (in mM): 130 N-methylglucamine, 20 NaCI, 1 MgCl,, 10 HEPES, 10 o-glucose, adjusted to pH 7.4 with HCI, osmolarity adjusted to 328 mmol/kg with sucrose. The deep layer of dorsal cortex overlying the striatum was removed and transferred to another culture dish containing the above solution and triturated through a ~200 Eppendorf micropipette tip and then through flame-polished Pasteur pipettes of decreasing tip apertures. The cell suspension in the culture dish was placed onto the stage of an inverted microscope, and cells were allowed to settle for lo-15 min before normal external solution containing 150 mM NaCl, 2.5 mM KC], 1 mM MgCI,, 2.5 mM CaCl,, 10 mM HEPES, and 10 mM o-glucose (pH = 7.4 adjusted with NaOH, osmolarity adjusted to 340 mmol/kg with sucrose) was reintroduced to the culture dish.
Whole-cell recording.
Whole-cell voltage clamp recordings were obtained using pipettes made from borosilicate glass capillaries pulled on a Flaming-Brown micropipette puller. Pipette resistances ranged from 0.5 to 2 Ma when filled with internal solution. Two internal solutions were used. The first consisted of 117 mM tetraethylammonium (TEA) chloride, 4.5 mM MgCI,, 9 mM HEPES, 9 mM EGTA, 14 IIIM creatine phosphate (Tris salt), 4 tTIM ATP (magnesium salt), and 0.3 mM GTP (Tris salt), pH adjusted to 7.4 with TEA-OH, osmolarity adjusted to 315 mmol/kg with sucrose. The second contained 125 mM N-methyl-D-glucamine, 20 mM TEA-OH, 10 mM HEPES, 11 mM EGTA, 1 mM CaCI,, 14 mM creatine phosphate (Tris salt), 4 mM ATP (magnesium salt), and 0.3 mM GTP (Tris salt), pH adjusted to 7.4 with methanesulfonic acid, osmolarity adjusted to 315 mmolikg with sucrose. Creatine phosphate, ATP, and GTP were stored as aliquots at -20°C and added just before experiments, and solution was then kept ice-cold. There was no difference between cells filled with the two internal solutions in terms of I,, amplitude or modulation by mGluR agonists. The external recording solution contained 25 mM BaCI,, 145 mM TEACI, 10 mM HEPES, and 1 pM tetrodotoxin (TTX), pH adjusted to 7.4 with TEA-OH, osmolarity adjusted to 335 mmol/kg with sucrose. by the time course of Ca'+ channel inhibition by cadmium. Seals were formed and the whole-cell configuration was obtained in normal external solution before perfusing the cell with the external solution containing TEA and Ba'+.
The electrode series resistance was compensated 70-80%, and the leak current was determined by 10 mV hyperpolarizing steps from the holding potential. All data except the current traces shown in the figures are derived from leak-subtracted current values. Capacitance artifacts were not subtracted from the current traces shown in the figures. For display purposes, the full amplitude of capacitance artifacts was not shown in Figures L4,3A, and 5D. Whole-cell currents recorded with Axopatch 200 or 1D amplifiers were filtered at 5 kHz and digitized at up to 50 kHz and stored on a microcomputer using pC1amp software. The amplitude of Ba2+ currents was measured isochronally at the time point at which peak current was seen in the absence of drug. The data for the concentrationresponse curves were fit with KaleidaGraph (Synergy Software, Reading, PA) using the logistic equation to obtain IC,, values. Percent inhibition by mGluR agonists before and after depolarizing prepulses was calculated as the current amplitude in the presence of agonist divided by the amplitude of the response in the absence of drug before the prepulse. The percentage of total current inhibited by agonist before and after application of channel blockers was calculated by determining the percentage of the original total baseline current amplitude that was inhibited in the presence of agonist before and after channel blocker treatment. All averaged values are given as mean ? SEM. The statistical criterion for significance wasp < 0.05.

RESULTS
Pharmacological properties of mGluRs that modulate Ca*+ channels Barium current was recorded from 183 acutely isolated cortical neurons, which were identified as pyramidal neurons under phasecontrast microscopy. To investigate the involvement of mGluRs in the modulation of Ca2+ channels, t-ACPD, quisqualate, and DCG-IV were applied to acutely isolated cortical neurons. Inhibition of I,, developed within 2 set after the onset of application of each agonist. Inhibition was largely reversible 8-16 set after removal of agonist. Figure L4 shows current traces recorded immediately before, during, and after application of 200 PM t-ACPD, 200 nM quisqualate, and 5 PM DCG-IV to separate neurons. The rate of current onset was slowed considerably during agonist application. Maximal inhibition of I,, by 200 PM t-ACPD, 1 PM quisqualate, and 5 PM DCG-IV was 22.4 ? 2.0% (n = 15), 19.1 + 1.9% (n = 18), and 14.9 L 1.5% (n = 18), respectively (Fig. 1B). The concentration-response curves for t-ACPD-, quisqualate-, and DCG-IV-induced effects on I,, are shown in Figure  1C. The estimated IC,, values for t-ACPD, quisqualate, and DCG-IV were 11.5 PM, 0.12 pM, and 0.51 PM, respectively.
The effects of other mGluR agonists were subsequently examined. Ibotenate is a relatively nonselective agonist (see Pin and Duvoisin, 1995). DHPG is a selective agonist for group I mGluRs (Gereau and Conn, 1995a). 4C3HPG is known to be an agonist for mGluR2 and an antagonist for mGluR1 (Thomsen et al., Drug solutions were applied by gravity-driven perfusion from a linear 1994). All three agonists were effective at concentrations that have array of 12 microcapillary tubes (inner-diameter > 150 pm). Solution flow was controlled with a three-way stopcock. To assess the kinetics of been'shown to a&vate mGluRs. 4C3HPG (200 PM The concentrations tested were chosen to be at the high end of concentration-response curves from studies of mGluR clones. C, Concentration-response curves for t-ACPD, DCG-IV, and quisqualate inhibition of I,,. The data set includes only neurons tested with at least four agonist concentrations, and each data point is the average of data from five to seven neurons. Inhibition by each agonist is normalized to inhibition by the maximally effective agonist concentration (100 PM t-ACPD, 1 pM quisqualate, and 5 FM DCG-IV). I,, was evoked by 50 msec voltage steps from -70 to -10 mV in A, B, and C. Data points and error bars represent ?SEM.
vation. We also examined the actions of the selective group III mGluR agonists [D,L-AF'4 (100 PM) and L-SOP (100 PM)]. These agonists produced little inhibition of I,, (percent inhibition by D,L-AP~ = 3.1 +-0.5, n = 8; L-SOP = 1.7 ? 0.7, n = 6) in neurons in which t-ACPD produced >15% inhibition (Fig. 1B). There was no change in holding current during the application of any mGluR agonist at concentrations that produced maximal responses in these experiments in which NMG or TEA and Ba2+ were the major cations in the external and internal recording solutions.
The observation that both quisqualate and DCG-IV modulated Ca2+ channel function was interesting given that these agonists have selective effects on different mGluR subtypes. It thus appeared possible that more than one mGluR subtype was involved in Ca2+ channel modulation. The first piece of information that was consistent with this hypothesis was the observation that DCG-IV was effective in fewer neurons than either t-ACPD or quisqualate. Inhibition of current amplitude of ~5% was observed in 95% of neurons in which 100 PM t-ACPD was tested (n = 63), 93% of neurons in which 1 pM quisqualate was tested (n = 42), but only 75% of neurons in which 5 PM DCG-IV was tested (n = 48). In 11 of these neurons, DCG-IV failed to produce significant inhibition, whereas t-ACPD or quisqualate produced >15% inhibition (Fig. 24). In addition, three neurons were sensitive to t-ACPD but did not respond to quisqualate (Fig. 2B). Thus, a proportion of the neurons tested did not respond to particular mGluR agonists even though effects of other agonists could be demonstrated in the same neuron.
We next correlated the inhibition produced by a maximally effective concentration of a given mGluR agonist with that produced by other agonists in the same neurons. The correlation between the percent inhibition produced by quisqualate and DCG-IV was quite low (Y = 0.11, n = lo), as was the correlation between inhibition by t-ACPD and quisqualate (r = 0.01, n = 12) or t-ACPD and DCG-IV (r = 0.28, n = 10) (t test,p > 0.1).
We then examined the effect of combined application of maximally effective concentrations of the selective agonists DCG-IV Choi and Lovinger .   and quisqualate. We hypothesized that if these agonists act on the same receptor, then inhibition by the combination of both agonists should be similar in magnitude to that produced by either agonist alone. If, however, the agonists act on separate receptors, one would expect that inhibition in the combined presence of both agonists would be greater than that observed with either agonist alone. Inhibition by the combination of 5 pM DCG-IV + 1 pM quisqualate was greater than that by either DCG-IV or quisqualate alone in every neuron in which both agonists produced >5% inhibition when applied separately (n = 5) (Fig. 2C). Furthermore, we observed that inhibition by 100 PM t-ACPD was well correlated with inhibition during the combined application of DCG-IV and quisqualate (Y = 0.96, n = 6) (t test, p < 0.001).
Antagonism by MCPG MCPG has been reported to be a competitive antagonist of mGluRs (Birse et al., 1993;Hayashi et al., 1994). The ability of MCPG to block mGluR agonist modulation of I,, was examined, as shown in Figure  partially blocked I,, inhibition by t-ACPD, quisqualate, and DCG-IV at agonist concentrations that produced submaximal inhibition. Percent inhibition by 2.5 PM t-ACPD, 200 nM quisqualate, or 600 nM DCG-IV averaged 18.6 + 2.9, 16.7 2 5.1, and 16.7 t 4.1%, respectively, in the absence of MCPG and 3.7 -C 0.2, 6.1 t 1.9, and 2.9 t 0.6, respectively, in the presence of MCPG. Taken together, these results suggest that t-ACPD, DCG-IV, and quisqualate inhibit I,, through mGluR activation in acutely isolated cortical neurons.

Mechanism of inhibition of Ca2+ channels by different mGluR subtypes
Hille and co-workers have identified different pathways linking receptor activation and Ca2+ channel modulation. These pathways can be distinguished by the time course of onset of inhibition with "fast" modulation developing within a few seconds, and "slow" modulation developing over tens of seconds (Hille, 1992;Shapiro et al., 1994). To determine how fast the inhibition by different mGluR agonists was achieved, we compared the time course of inhibition by DCG-IV and quisqualate. I,, was evoked by 16 msec voltage steps from -70 to -10 mV every 2 set (this protocol allowed for maximal current activation without measurable inactivation). Using this pulse protocol, no sign of run-down of I,, amplitude was seen. As shown in Figure 4, substantial inhibition by 5 pM DCG-IV and 1 PM quisqualate was achieved within 2 set, and saturation of inhibition was observed in <4 set (n = 5). This time course is comparable with that previously reported for "fast" inhibition and is considerably faster than "slow" modulation (Hille, 1992). As mentioned earlier, mGluR agonists including t-ACPD, quisqualate, and DCG-IV slowed I,, activation kinetics. This slowing has been proposed to represent a voltage-dependent relief of inhibition during the depolarizing test pulse (Bean, 1989). To determine whether inhibition by mGluR agonists was voltagedependent, we measured the current-voltage relationship of I,, before, during, and after application of 100 PM t-ACPD, 5 PM DCG-IV, and 1 pM quisqualate (Fig. 5&C). Inhibition of I,, appeared to be voltage-dependent as previously reported for other G-protein-coupled neurotransmitter receptors (Bean, 1989;Elmslie et al., 1990). It has been reported that depolarizing prepulses transiently facilitate Ca2+ channel current by relieving G-protein-mediated inhibition induced by receptor activation (Jones, 1991;Ikeda, 1991Ikeda, , 1992Boland and Bean, 1993). The Choi and Lovinger . mGluRs Inhibit Ca2+ Channel s i n Isolated Cortical Neurons relief of inhibition by depolarizing prep&es has been reported for both N-and non-N-type, presumably P-type, channels (Mintz and Bean, 1993;Swartz, 1993). To determine whether depolarizing prepulses relieved inhibition by mGluR agonists in isolated cortical neurons, we measured I,, amplitude during short, moderately depolarizing voltage steps separated by a large transient depolarization. We observed that depolarizing prepulses facilitated I,, to a greater extent in the presence than in the absence of mGluR agonists (Fig. 5D), suggesting that the inhibition by mGluR agonists was relieved by the prepulse. To compare the depolarization relief of Ca*+ channel inhibition by DCG-IV and quisqualate, we measured the percent inhibition by each agonist before and after a depolarizing prepulse. The percent inhibition by 5 FM DCG-IV (n = 7) and 1 pM quisqualate (n = 14) averaged 20.8 t 3.9 and 19.7 2 1.5 before and 8.9 2 2.3 and 8.5 f 1.4 after a depolarizing prepulse (Fig. 5&F). Thus, no significant difference was found in the depolarization relief of the inhibition by DCG-IV and quisqualate. Previous studies have suggested that calcium channel modulation by mGluRs is G-protein-mediated (Lester and Jahr, 1990;Swartz and Bean, 1992;Sayer et al., 1992;Sahara and Westbrook, 1993;Hay and Kunze, 1994;Stefani et al., 1994;Chavis et al., 1994). Thus, we predicted that introduction of the nonhydrolyzable GTP analog GTPyS into the cell would mimic the effect of agonists and/or render agonist-induced inhibition irreversible. Dialysis of the cell interior with 100-300 PM GTPyS produced gradual inhibition that resembled that produced by mGluR agonists and eliminated fast, reversible inhibition by agonists, including DCG-IV, quisqualate, and t-ACPD (n = 10) (data not shown). However, in some GTPyS-filled neurons (n = 8), mGluR agonists produced a slowly developing, irreversible inhibition (data not shown).
Because of the difficulty in using PTX in experiments on acutely isolated neurons, we used NEM, a sulfhydryl-alkylating agent (Jakobs et al., 1982), which has been shown to uncouple PTXsensitive G-proteins from receptors (Nakajima et al., 1990;Shapiro et al., 1994). A recent study suggests that NEM blocks Ca*+ channel modulation by PTX-sensitive G-proteins in neurons (Shapiro et al., 1994). To determine the NEM sensitivity of Ca*+ channel inhibition by DCG-IV and quisqualate in isolated cortical neurons, we briefly applied mGluR agonists before and after a 120 set application of 50 PM NEM (Fig. 6). To determine how much of the inhibition by mGluR agonists was sensitive to NEM, we measured the percent inhibition by DCG-IV and quisqualate before and after application of NEM. The percent inhibition by 1 pM quisqualate (n = 14) and 5 pM DCG-IV (n = 8) averaged 20.2 2 1.6 and 20.7 ? 2.7, respectively, before and 8.6 ? 0.7 and 4.1 2 1.0, respectively, after application of 50 pM NEM. Thus, inhibition by DCG-IV appeared to be affected by NEM to a greater extent than inhibition by quisqualate. These data suggest that a PTXsensitive G-protein mediates the majority of DCG-IV and quisqualate-induced inhibition, However, part of the inhibition by quisqualate might be mediated by a PTX-insensitive G-protein(s).

Ca2+ channel subtypes modulated by mGluRs
We examined the effects of Ca*+ channel blockers including nifedipine (an L-type channel blocker) (Fox et al., 1987a,b), o-CgTx (an N-type channel blocker) (Aosaki and Kasai, 1989;Plummer et al., 1989), and o-AgTx (a blocker of P-and other, channels) (Mintz et al., 1992). Application of each of these agonists inhibited some proportion of I,, activated by voltage steps to -10 mV from a holding potential of -70 mV. w-CgTx (1 PM Wash .-/ 0 before after before after prepulse prep&e prepulse prepulse Figure 5. mGluR agonists inhibit I,, in a voltage-dependent manner. A, Current-voltage relationships determined before, during, and after application of 100 pM t-ACPD. Holding potential was -70 mV. Test potentials were given every 5 sec. B and C, Plot of I,, inhibition versus test pulse potential derived from current-voltage relationships taken before, during, and after 5 pM DCG-IV (B) or 1 pM quisqualate (C) treatment. Holding potential was -70 mV. Test potentials were given every 5 sec. D, Relief of t-ACPD-induced inhibition by depolarizing prepulses. Current activated by the indicated voltage steps before and during application of 100 pM t-ACPD. Note the increase in the ratio of the response to the second test pulse relative to the first test pulse in the presence of t-ACPD. E, Graph showing the percent inhibition by 5 pM DCG-IV and 1 pM quisqualate before and after a depolarizing prepulse. A single denominator (control response before the prepulse) was chosen to calculate percent inhibition before and after the prepulse. Prepulses increase the ratio of the response to the second test pulse relative to the first test pulse by 5.1 + 2.3 and 22.3 -C 4.1% in the absence or presence of 5 PM DCG-IV (n = 7), respectively, and by 7.9 t 1.9 and 23.3 t 3.6% in the absence or the presence of 1 FM quisqualate (n = 14), respectively. Current was evoked by test pulses (-70 to -5 mV, 25 msec duration) given before and after a 50 msec prepulse to +80 mV, as in D.
blocked 31.5 t 7.0% (n = 5) of Ina, nifedipine (5 pM) blocked 22.8 + 1.1% (n = 5) of Ina, and o-AgTx with 1 mg/ml cytochrome C (100 nM) blocked 33.5 t 4.3% (n = 5) of I,, (Fig. 70). We estimated that a substantial proportion of current (lo-20% of total current amplitude) was not affected by any of these blockers. However, the proportion of the Cazi channel types varied among cells. In some cells, apparently unequal proportions of the different Cazi channels (e.g., <lo% of L-type, 50% of N-or P-type) were present.
To determine the Ca2+ channel subtypes that were inhibited by t-ACPD, I,, was elicited before and during block of N-, P-, and L-type currents with 1 FM o-CgTx, 100 nM w-AgTx, and 5 PM nifedipine (Fig. 7). The percent of total current inhibited by t-ACPD was 18.3 t 1.9 before and 6.4 t 1.6 after application of 1 PM w-CgTx (n = 5) (paired t test,p < 0.001). In another group of cells (n = 5), the percent of total current inhibited by t-ACPD was 28.5 + 5.7 before and 16.7 C 5.4 after application of 100 nM o-AgTx (paired t test, p < O.OOl), whereas t-ACPD inhibited I,, by 30.6 ? 5.9% before and by 24.5 t 4.1% after application by 5 PM nifedipine in yet another group of cells (n = 5) (paired t test, p < 0.05) (Fig. 7E). We also examined the involvement of N-type channels in I,, modulation by DCG-IV and quisqualate. The percent of total current inhibited by 5 pM DCG-IV (n = 4) and 1 FM quisqualate (n = 5) was 26.6 -C 5.4 and 15.7 t 2.6, respec-tively, before and 11.9 + 4.2 and 7.36 + 2.2, respectively, after application of 1 pM o-CgTx (paired t test, p <-0.01 for both agonists). Thus, the effect of o-CgTx on inhibition by DCG-IV and quisqualate appeared to be similar to that of w-CgTx on inhibition by t-ACPD.

DISCUSSION
Previous studies have demonstrated that mGluR activation modulates Ca2+ channels in a variety of neurons (Lester and Jahr, 1990;Sayer et al., 1992;Swartz and Bean, 1992;Trombley and Westbrook, 1992;Sahara and Westbrook, 1993;Hay and Kunze, 1994;Stefani et al., 1994;Chavis et al., 1994;Chavis et al., 1995). The majority of studies have shown that t-ACPD and quisqualate activate Ca2+ channel modulation (Swartz and Bean, 1992;Sayer et al., 1992;Hay and Kunze, 1994;Chavis et al., 1995). In addition, some evidence for modulatory effects on Ca2' channels of the group III mGluR agonist L-AP4 (Trombley and Westbrook, 1992;Sahara and Westbrook, 1993) and L-CCG-I (a moderately selective group II mGluR agonist) has been presented (Chavis et al., 1994). Interestingly, t-ACPD was able to inhibit Ca2+ channel function in hippocampal neurons from mutant mice lacking mGluR1, suggesting that this receptor cannot account for all of the Ca2+ channel-modulatory actions of mGluRs (Aiba et al., 1994). From these studies it has been hypothesized that several Time (set) mGluR subtypes can participate in Ca2+ channel modulation (Pin and Duvoisin, 1995). However, there is very little evidence that different mGluR subtypes participate in modulation in the same population of neurons. In the present study, we have obtained evidence that agonists selective for different mGluRs inhibit calcium current in some neurons, whereas in other neurons only one subtype-selective agonist was effective. In addition, inhibition during the combined application of DCG-IV and quisqualate was greater than that elicited by application of either agonist alone. Finally, inhibition by the nonspecific mGluR agonist t-ACPD was well correlated with the combined actions of DCG-IV and quisqualate in individual neurons but was not well correlated with the actions of either agonist alone. These findings suggest that more than one mGluR mediates Ca2+ channel inhibition in these neurons. The mGluR agonists that elicited Ca2+ channel modulation in cortical neurons were nonspecific, selective for group I mGluRs (Manzoni et al., 1991;Nakanishi, 1992;Gereau and Conn, 1995a), or selective for group II mGluRs (Hayashi et al., 1993;Thomsen et al., 1994;Gereau and Conn, 1995a). The findings with MCPG suggest that each of these agonists works through mGluRs, and appear to exclude the involvement of mGluR4 because this subtype is not blocked by MCPG (Hayashi et al., 1994). We observed no evidence of inhibition by agonists selective for group III mGluRs. However, LAP4 modulates Ca2+ channel function in olfactory bulb and hippocampal neurons (Trombley and Westbrook, 1992;Sahara and Westbrook, 1993). Thus, there is evidence for participation of all subgroups of mGluRs in Cazt channel modulation. However, members of subgroups I and II appear to be the best candidates for modulating Ca2' channels in cortical neurons, at least under conditions used in the present study.
Activation of different mGluRs appears to modulate Ca2+ channels via a similar fast, reversible, G-protein-mediated, voltage-dependent mechanism in isolated cortical neurons. The only difference in inhibition mediated by group I and group II mGluRs was the sensitivity to NEM, a compound previously demonstrated to mimic the actions of PTX (Hille, 1992;Shapiro 20 Ins I 5pMDCGIV et al., 1994). Thus, inhibition by group I mGluRs may involve both PTX-sensitive and -insensitive G-proteins whereas group II mGluRs use mainly PTX-sensitive G-proteins. We have evaluated further the specificity of NEM for antagonizing PTX-sensitive G-protein-mediated modulatory pathways. NEM appears to be a fairly specific blocker of PTX-sensitive G-proteins under conditions similar to those used in the present study (Choi et al., 1995). Previous results suggest that group II mGluRs couple negatively to adenylate cyclase through PTX-sensitive G-proteins. Thus, involvement of PTX-sensitive G-proteins might be expected in the case of DCG-IV-induced Ca2+ channel inhibition. It has been reported that quisqualate-activated mGluRs and group I mGluRs can modulate Ca2+ channels, PI hydrolysis, and Ca2+ mobilization through PTX-sensitive or -insensitive G-proteins (Abe et al., 1992;Aramori and Nakanishi, 1992;Linden et al., 1994;Hay and Kunze, 1994) and, thus, the observed involvement of both PTXsensitive and -insensitive G-proteins in the actions of quisqualate is also not surprising. Our findings suggest that mGluRs, particularly a group I mGluR, inhibit Ca2+ channels through multiple G-proteins in the same neuron, consistent with previous studies in which 012 adrenergic and pancreatic polypeptide receptors inhibit Ca2+ channels through multiple G-proteins in superior cervical ganglion neurons (see Shapiro et al., 1994).
It must be noted that there is strong evidence for mGluR inhibition of Ca2+ channels in cortical neurons via Ca2+-dependent mechanisms excluded in the present study (Sayer et al., 1992). The mechanism described in this previous study appears to be involved in modulation of L-type currents. Thus, mGluRs likely inhibit different calcium channels using diverse mechanisms in cortical neurons.
The different mGluRs appear to modulate similar Ca2+ channel subtypes. Inhibition of N-type channels accounted for the bulk of the effects of t-ACPD, DCG-IV, and quisqualate. The effects of t-ACPD and quisqualate are similar to those observed in previous studies (Swartz and 'Bean, 1992;Sahara and Westbrook, 1993;Swartz et al., 1993;Hay and Kunze, 1994;Stefani et al., 1994). However, inhibition of non-N-type channels accounted for a sizable proportion of receptor-mediated modulation by all three agonists. The observation that Graph illustrating the percent inhibition of I,, by each Ca *+ channel blocker. E, Graph illustrating the percent inhibition by 100 PM t-ACPD before and after application of each channel blocker. A single denominator (control response before the application of channel blockers) was chosen to calculate the percent of the total current inhibited in the presence of mGluR agonists before and during the application of channel blockers. I,, was evoked by 50 msec voltage steps from -70 to -10 mV. Voltage pulses were given every 8 sec.
some of the current inhibited by t-ACPD was o-AgTx-sensitive suggests that one or more of the mGluR subtypes inhibits non-Ntype channels affected by this blocker. The concentration of o-AgTx used was in the range that is fairly selective for P-type channels, suggesting the involvement of this channel type. We were not able to distinguish further the Ca2+ channel subtypes modulated by DCG-IV because of depletion of our limited supply of this agonist.
GABAa receptors modulate P-type Cazt channels in cerebellar Purkinje and spinal cord neurons through a G-protein-mediated mechanism similar to the mechanism described at present (Mintz and Bean, 1993). Thus, it is becoming evident that non-N-type, non-L-type channels, including the P-type, can be modulated by G-protein-coupled receptors via a fast modulatory pathway. Inhibition of Ca2' channels has been suggested as a mechanism for regulation of transmitter release by presynaptic autoreceptors (Lipscombe et al., 1989;Bley and Tsien, 1990). Also, N-and non-Ntype Ca2+ channels are responsible for synaptic transmission at corticostriatal and hippocampal synapses (Luebke et al., 1993;Lovinger et al., 1994;Wheeler et al., 1994;Wu and Saggau, 1994) and P-type channels appear to be involved in synaptic transmission at hippocampal synapses (Luebke et al., 1993;Wu and Saggau, 1994;Castillo et al., 1994). N-and P-type channels involved in excitationsecretion coupling may be modulated by mGluRs in presynaptic terminals, leading to synaptic depression. Our results indicate that a group II mGluR inhibits glutamatergic synaptic transmission presynaptically at corticostriatal synapses - . Thus, it is conceivable that modulation of Ca2' channels by group II mGluR agonists represents the mechanism by which synaptic transmission is presynaptically inhibited. However, low concentrations of quisqualate that do not activate group II mGluRs have been shown to inhibit synaptic transmission in hippocampus and striatum (Baskys and Malenka, 1991;Calabresi et al., 1992). It is thus possible that a group I mGluR subtype also functions as a presynaptic inhibitory receptor and that inhibition of Ca 2c channels is a viable mechanism for the action of this receptor on transmission. It should be noted, however, that some neurotransmitter receptors can produce presynaptic inhibition independent of the involvement of Ca2+ channels (Scholz and Miller, 1992;Scanziani et al., 1992Scanziani et al., , 1993. Furthermore, recent evidence indicates that mechanisms downstream from calcium entry contribute to modulation of synaptic transmission by mGluR autoreceptors in corticostriatal cocultures and in hippocampus (Tyler and Lovinger, 1995;Gereau and Conn, 1995b). Clearly, further work is needed to assess the importance of Ca2+ channel inhibition in the action of inhibitory presynaptic glutamate receptors.