Unexpected inhibitory regulation of glutamate release from rat cerebrocortical nerve terminals by presynaptic 5-hydroxytryptamine-2A receptors

Materials

Fura-2-acetoxymethyl ester and 3,3′-dipropylthiadicarbocyanine iodide [DiSC₃(5)] were obtained from Molecular Probes (Eugene, OR). Percoll was obtained from Pharmacia. Ketanserin, RS102221, U73122, U73343, Ro320432, GF109203X, ω-conotoxin GVIA, and ω-conotoxin MVIIC were obtained from Tocris Cookson (Bristol, United Kingdom). Glutamate dehydrogenase, ω-agatoxin IVA, L-trans-pyrollidine-2,4-dicarboxylic acid, Ro320432, and all other reagents were obtained from Sigma (Poole, United Kingdom) or Merck (Poole, United Kingdom).

Synaptosomal Preparation

Synaptosomes were prepared as described previously (Dunkley et al., 1986; Sihra, 1997). Briefly, the cerebral cortex from 2-month-old male Sprague-Dawley rats was isolated and homogenized in a medium containing 320 mM sucrose, pH 7.4. The homogenate was spun for 2 min at 3,000g (5,000 rpm in a Beckman JA 25.5 rotor) at 4°C, and the supernatant was spun again at 14,500g (11,000 rpm in a JA 25.5 rotor) for 12 min. The pellet was gently resuspended in 8 ml of 320 mM sucrose, pH 7.4. Two millilitres of this synaptosomal suspension was placed into 3-ml Percoll discontinuous gradients containing 320 mM sucrose; 1 mM EDTA; 0.25 mM DL-dithiothreitol; and 3%, 10%, and 23% Percoll, pH 7.4. The gradients were centrifuged at 32,500g (16,500 rpm in a Beckman JA 20.1 rotor) for 7 min at 4°C. Synaptosomes placed between the 10% and the 23% Percoll bands were collected and diluted in a final volume of 30 ml HEPES buffer medium (HBM) consisting of 140 mM NaCl, 5 mM KCl, 5 mM NaHCO₃, 1 mM MgCl₂ · 6H₂O, 1.2 mM Na₂HPO₄, 10 mM glucose, 10 mM HEPES (pH 7.4) before centrifugation at 27,000g (15,000 rpm in a Beckman JA 25.5 rotor) for 10 min. The pellets thus formed were resuspended in 3 ml HBM, and the protein content was determined by the Bradford assay (Bradford, 1976); 0.5 mg of synaptosomal suspension was diluted in 10 ml of HBM and spun at 3,000g (5,000 rpm in a Beckman JA 25.5 rotor) for 10 min. The supernatants were discarded, and the synaptosomal pellets were stored on ice and used within 4–6 hr.

Glutamate Release Assay

Glutamate release from purified cerebrocortical synaptosomes was monitored online, with an assay employing exogenous glutamate dehydrogenase and NADP⁺ to couple the oxidative deamination of the released glutamate to the generation of NADPH detected fluorometrically (Nicholls and Sihra, 1986). Synaptosomal pellets (0.5 mg protein) were resuspended in 1.5 ml HBM containing 16 μM bovine serum albumin (BSA) and incubated in a stirred and thermostated cuvette maintained at 37°C in a Perkin-Elmer LS-50B spectrofluorimeter. NADP⁺ (2 mM), glutamate dehydrogenase (GDH; 50 U/ml), and CaCl₂ (1 mM) were added after 3 min. Glutamate release was monitored by measuring the increase of fluorescence (excitation and emission wavelengths of 340 nm and 460 nm, respectively) resulting from NADPH being produced by the oxidative deamination of released glutamate by GDH. In the absence of axonal connectivity, synaptosomes are not amenable to electrical stimulation, so several biochemical secretagogues have been developed, including the use of depolarization protocols involving potassium channel blockers such as 4-aminopyridine (4AP) or high external [K⁺] or direct mediation of Ca entry using Ca ionophores such as ionomycin (Sihra, 1997). 4AP destabilizes the membrane potential and is thought to cause repetitive spontaneous Na⁺ channel-dependent depolarization (Tibbs et al., 1989) that closely approximates in vivo depolarization of the synaptic terminal that leads to the activation of voltage-dependent Ca²⁺ channels (VDCCs) and neurotransmitter release. Elevated extracellular KCl concentrations depolarize the plasma membrane by shifting the K⁺ equilibrium potential above the threshold potential for activation of voltage-dependent ion channels. While Na⁺ channels are inactivated under these conditions (Barrie et al., 1991), VDCCs are activated nonetheless to mediate Ca²⁺ entry supporting neurotransmitter release. Comparison of the effects of modulators under 4AP and KCl stimulation protocols therefore makes it possible to distinguish between modulatory pathway that target: 1) ionic channels involved in maintaining the plasma membrane potential and shaping the action potential waveform vs. 2) the VDCCs coupled to glutamate release directly (Nicholls, 1998).

Secretagogues 4AP (3 mM), high external KCl (KCl; 15 mM or 30 mM), or ionomycin (5 μM) were added after 10 min of incubation to stimulate glutamate release. Data were accumulated at 2-sec intervals. A standard of exogenous glutamate (5 nmol) was added at the end of each experiment, and the fluorescence response used to calculate released glutamate was expressed as nanomoles glutamate per milligram synaptosomal protein (nmol/mg). Values quoted in the text and expressed in bar graphs represent levels of glutamate cumulatively release after 5 min of depolarization and are indicated as nmol/mg/5 min. Cumulative data were analysed in Lotus 1-2-3 and MicroCal Origin. Estimation of the IC₅₀ was based on a one-site model [Inhibition = (Inhibition_MAX × [DOI])/(IC₅₀ + [DOI])] using the nonlinear curve-fitting function in MicroCal Origin. Statistical analysis was performed by two-tailed Student's t-tests or one-way ANOVA, followed by post hoc two-tailed t-tests.

Membrane Potential Measurement With DiSC₃(5)

The synaptosomal membrane potential can be monitored by positively charged membrane potential-sensitive carbocyanine dyes such as DiSC₃(5) (Enkvist et al., 1988). The dye becomes incorporated into the synaptosomal plasma membrane lipid bilayer. Upon depolarization with 4AP, the release of the dye from the membrane bilayer is indicated as an increase in fluorescence. Synaptosomes were preincubated and resuspended as described for the glutamate release experiments. After 3 min of incubation, 4 μM DiSC₃(5) was added and allowed to equilibrate before the addition of CaCl₂ (1 mM) after 4 min of incubation. 4AP was added to depolarize the synaptosomes at 10 min, and DiSC₃(5) fluorescence was monitored at excitation and emission wavelengths of 646 nm and 674 nm, respectively, and data accumulated at 2-sec intervals. Cumulative data were analyzed in Lotus 1-2-3, and results are expressed in fluorescence units.

Cytosolic Ca²⁺ Measurements

[Ca²⁺]_C was measured with the Ca²⁺ indicator Fura-2. Synaptosomes (0.5 mg/ml) were preincubated in HBM with 16 μM (1 mg/ml) BSA in the presence of 5 μM Fura-2-acetoxymethyl ester and 0.1 mM CaCl₂ for 30 min at 37°C in a stirred test tube. After Fura-2 loading, synaptosomes were centrifuged in a microcentrifuge for 30 sec at 3,000g (5,000 rpm in a Beckman JA 25.5 rotor). The synaptosomal pellets were resuspended in HBM with BSA and the synaptosomal suspension stirred in a thermostatted cuvette in a Perkin-Elmer LS-50B spectrofluorimeter. CaCl₂ (1 mM) was added after 3 min, and further additions were made after an additional 5 min, as described in the legends to the figures. Fluorescence data were accumulated at excitation wavelengths of 340 nm and 380 nm (emission wavelength 505 nm) at data accumulated at 7-sec intervals. Calibration procedures were performed as described previously (Sihra et al., 1992), using 0.1% sodium dodecyl sulfate to obtain the maximal fluorescence with Fura-2 saturation with Ca²⁺, followed by 10 mM EGTA (Tris buffered) to obtain minimum fluorescence in the absence of any Fura-2/Ca²⁺ complex. Cytosolic free Ca²⁺ concentration ([Ca²⁺]_C, nM) was calculated with equations described previously (Grynkiewicz et al., 1985). Cumulative data were analyzed in Lotus 1-2-3 and MicroCal Origin. Statistical analysis was performed by two-tailed Student's t-tests.

Effect of DOI on 4AP-Evoked Glutamate Release From Rat Cerebrocortical Synaptosomes

To investigate the influence of 5-HT₂ receptors on glutamate release, we examined whether 5-HT₂ receptor activation by the 5-HT_2A/2C receptor partial agonist DOI could affect the release of glutamate evoked by 4AP. Under control conditions, 4AP (3 mM) evoked a glutamate release of 18.6 ± 0.8 nmol/mg/5 min from synaptosomes incubated in the presence of 1 mM CaCl₂. Application of DOI (50 μM) produced an inhibition of 4AP-evoked glutamate release to 8.2 ± 0.3 nmol/mg/5 min (n = 8; P < 0.01), without altering the basal release of glutamate (Fig. 1A). The DOI-induced inhibition of 4AP-evoked glutamate release was concentration dependent, with an IC₅₀ value derived from a dose-response curve (Fig. 1C) of about 29 μM. Given the latter value and the robust 53% depression of 4AP-evoked glutamate release seen with 50 μM DOI, this concentration of DOI was used in subsequent experiments to evaluate the mechanisms underlying the ability of the agonist to reduce glutamate release.

Glutamate release produced by the depolarization of isolated nerve terminals is known to have two components: a physiologically relevant Ca²⁺-dependent component, which is produced by the exocytosis of synaptic vesicles containing glutamate, and a Ca²⁺-independent component that results from prolonged depolarization causing a membrane potential-mediated shift of the glutamate transporter steady-state toward the outward direction to effect cytosolic glutamate efflux (Nicholls, 1989). To examine the degree to which the observed modulation of total release by 5-HT₂ receptor activation reflected the Ca⁺-independent efflux of glutamate, we analyzed glutamate efflux by the addition of 0.1 mM EGTA to synaptosomes (incubated in the absence of external Ca²⁺) prior to depolarization with 4AP. This cytosolic release of glutamate amounted to less than 8 nmol/mg/5 min. Notably, this component of release was also affected by the prior addition of 50 μM DOI (4.1 ± 0.2 nmol/mg/5 min; n = 8; P < 0.01; Fig. 1B), suggesting that the overall inhibition of glutamate release by DOI reflects effects on both the Ca²⁺-dependent exocytosis of glutamate and the Ca²⁺-independent nonvesicular efflux. To examine the effect of DOI with the latter component abrogated, we looked at its effect in the presence of L-transpyrrolidine-2,4-dicarboxylic acid (L-trans-PDC), a potent inhibitor of the plasma membrane glutamate transporter, which should block or severely reduce the Ca²⁺-independent nonvesicular efflux by transporter reversal. In the presence of L-trans-PDC, although 4AP-evoked glutamate release was increased by the inhibitor (because of inhibition of reuptake of released glutamate), DOI continued to reduce significantly the secretagogue-induced release of glutamate. This indicates that a major proportion of the inhibition produced by DOI reflects an effect on the physiologically relevant, exocytotic pool of glutamate release. Similar results and conclusions could be drawn from analysis with the nontransportable glutamate uptake inhibitor DL-threo-beta-benzyloxyaspartate (DL-TBOA; Supplementary Fig. 1S).

Inhibition by DOI Is Mediated by 5-HT_2A Receptors: Pharmacology

The somewhat surprising finding that DOI causes an inhibition of glutamate release in synaptosomes, in contrast to facilitatory effects observed in slice and in vivo studies, led us to next ask whether the effects were indeed receptor mediated. We pursued this question by examining the effect of the potent 5-HT_2A receptor antagonist ketanserin (which has moderate affinity for the 5-HT_2C receptor type; Baxter et al., 1995) and the 5-HT_2C receptor antagonist RS102221 on the action of DOI. As shown in Figure 2, the inhibitory effect of 50 μM DOI on 4AP-evoked glutamate release was completely prevented when synaptosomes were pretreated with 10 μM ketanserin, previously shown to be sufficient for blocking synaptosomal 5-HT₂ receptor responses (Maura et al., 1991; Garro et al., 2001). Thus, compared with the approximately 51% decrease of 4AP-evoked release produced by DOI (control 18.1 ± 0.5 nmol/ mg/5 min; DOI 8.8 ± 0.6 nmol/mg/5 min), in the presence of 10 μM ketanserin, the effect of DOI was completely blocked (ketanserin + DOI 17.5 ± 0.5 nmol/ mg/5 min). The application of 10 μM ketanserin alone elicited no significant decrease in the release of glutamate evoked by 4AP (ketanserin, 17.7 ± 0.5 nmol/mg/5 min). In contrast to the effect of 5-HT_2A receptor antagonism by ketanserin, the 5-HT_2C receptor antagonist 5 μM RS102221 influenced neither 4AP-evoked glutamate release, per se, nor the ability of DOI to depress this release (control 4AP 18.6 ± 1.1 nmol/mg/5 min; RS102221 18.3 ± 0.9 nmol/mg/5 min; RS102221 + DOI, 7.9 ± 1.2 nmol/mg/5 min) by 58% (Fig. 2), a value comparable that observed in the absence of RS102221 (Figs. 1A, 2). These results suggest first that DOI-mediated inhibition of 4AP-mediated release is indeed receptor-mediated and not a nonspecific effect of the compound. Second, these experiments invoke regulatory 5-HT_2A receptors, rather than 5-HT_2C receptors, in the DOI-mediated inhibition of glutamate release. The other G_q-coupled member in the 5HT₂ class, the 5-HT_2B receptor, is absent from the cerebrocortex (Duxon et al., 1997).

Effect 5-HT_2A Receptor Activation on Nerve Terminal Excitability and Ca²⁺ Influx

We next addressed the potential mechanisms underlying the 5-HT_2A receptor-mediated inhibition of glutamate release by assessing the effects of DOI on synaptosomal plasma membrane potential and Ca²⁺ influx. Because inhibition of Na⁺ channels or activation of K⁺ channels is known to stabilize membrane excitability and consequently cause a reduction in the evoked entry of Ca²⁺ and neurotransmitter release (Rehm and Tempel, 1991; Li et al., 1993; Pongs et al., 1999), we reasoned that the observed inhibitory effect of DOI on 4AP-evoked glutamate release could be due to an alteration of nerve terminal excitability. To test this possibility, we examined the effects of DOI on synaptosomal plasma membrane potential under resting conditions and on depolarization, with the membrane potential-sensitive dye DiSC₃(5). As shown in Figure 3A, 4AP (3 mM) caused an increase in DiSC₃(5) fluorescence of 3.1 ± 0.05 fluorescence units/5 min (n = 7). Preincubation of synaptosomes with 50 μM DOI did not alter the resting plasma membrane potential; however, in the presence of DOI, the 4AP-induced elevation of DiSC₃(5) fluorescence was significantly decreased (2.1 ± 0.08 fluorescence units/5 min; n = 7; P < 0.01; Fig. 3A). This suggests that an attenuation of synaptosomal plasma membrane potential depolarization produced by 4AP, at least in part, underlies the observed inhibition of 4AP-evoked glutamate release by 5-HT_2A receptor activation.

To examine whether the attenuation of depolarization-induced synaptosomal membrane potential reflected a decreased entry of Ca²⁺ through the VDCCs, we used the Ca²⁺ indicator Fura-2 to assess what effect DOI has on the 4AP-induced increase in cytosolic Ca²⁺ levels ([Ca²⁺]_C). After addition of 4AP (3 mM), [Ca²⁺]_C in synaptosomes was elevated to a plateau level of 205.1 ± 6.2 nM (Fig. 3B). Preincubation of synaptosomes with 50 μM DOI reduced the 4AP-evoked [Ca²⁺]_C increase by 19% (165.3 ± 5.7 nM; n = 8; P < 0.01; Fig. 3C). Importantly, the inhibition of 4AP-evoked increase of [Ca²⁺]_C by DOI was suppressed by 10 μM ketanserin preincubation (ketanserin 201.1 ± 4.2 nM; ketanserin + DOI 199.1 ± 5.1 nM). In contrast to this effect of 5-HT_2A receptor antagonism on DOI-mediated Ca²⁺ responses, 5-HT_2C receptor blockade with 5 μM RS102221 was without effect on the DOI-mediated suppression of 4AP-evoked Ca²⁺ influx (RS102221 158.9 ± 7.2 nM; RS102221 + DOI 202.8 ± 6.1 nM; Fig. 3C). These data indicate that DOI reduces 4AP-evoked changes in [Ca²⁺]_C with a pharmacological profile similar to that observed for the reduction of release (Fig. 2).

Involvement of Presynaptic P/Q- and N-Type Ca²⁺ Channels in DOI-Induced Inhibition of Glutamate Release

In the rat cerebrocortical nerve terminal preparation, the release of glutamate is supported by N- and P/Q-type VDCCs (Vazquez and Sanchez-Prieto, 1997). To determine whether a specific VDCCs was involved in the inhibition of 4AP-evoked glutamate release by DOI, we examined the effects of DOI in the presence of Ca²⁺ channel blockers/toxins. As shown in Figure 4A, DOI was ineffective following the prior block of VDCCs with ω-CgTX GVIA (N-type) and ω-Aga IVA (P/Q-type). Specifically, in the absence of Ca²⁺ channel blockers, DOI inhibited the 4AP-evoked release from 18.9 ± 0.2 nmol/mg/5 min to 8.8 ± 0.3 nmol/mg/5 min (n = 10; P < 0.05; Fig. 4B). Application of 2 μM ω-CgTX GVIA or 500 nM ω-Aga IVA reduced 4AP-evoked glutamate release to 16.1 ± 0.2 (n = 5; P < 0.05) or 11.7 ± 0.4 nmol/mg/5 min (n = 6; P < 0.01), respectively, and to 8.3 ± 0.4 nmol/mg/5 min when both toxins were coapplied. In the presence of ω-CgTX GVIA alone, DOI could still effectively reduced release (from 16.1 ± 0.2 to 5.8 ± 0.4 nmol/mg/5 min); in the presence ω-Aga IVA, the inhibition by DOI (from 11.7 ± 0.4 to 9.2 ± 0.5 nmol/mg/5 min respectively) was somewhat diminished (Fig. 4B). After the combined application of ω-Aga IVA and ω-CgTX GVIA, when glutamate release was reduced to 8.3 ± 0.4 nmol/mg/5 min (n = 6; P < 0.001, Fig. 4B), DOI caused a marginal further decrease to 7.9 ± 0.5 nmol/mg/5 min, which represented a statistically insignificant change in 4AP-evoked glutamate release resulting from DOI in the presence of P/Q- and N-type VDCC block (4.8% ± 2.3%, cf. DOI alone, 53.4% ± 3.2%; n = 8; P < 0.01; Fig. 4B). Similarly, we found that 1 μM ω-CgTX MVIIC, a wide-spectrum blocker of N-, P-, and Q-type Ca²⁺ channels, reduced the 4AP-evoked glutamate release to 6.4 ± 0.8 nmol/mg/5 min and completely prevented any further inhibition of glutamate release by 50 μM DOI (MVIIC + DOI 6.7 ± 0.8 nmol/mg/5 min; n = 5; Fig. 4B). Similarly to the lack of effect of DOI in the presence of ω-CgTX MVIIC, the agonist had no effect on the residual release in the presence of all three channel toxins (see Supplementary Fig. 2S). Overall, these data point to the possibility that the effects of DOI are manifest through the direct or indirect attenuation of P/Q-type VDCC activity.

Locus of 5-HT_2A Receptor-Induced Inhibition of Glutamate Release

The data given above are consistent with the activation of 5-HT_2A receptors reducing glutamate release by the attenuation of nerve terminal excitability, leading to a reduction in the influx of Ca²⁺ through largely P/Q-type VDCCs. The possibility remains that 5-HT_2A receptor activation also directly impinges on these VDCCs to modulate their activity. Comparison of modulatory effects obtained with 4AP- and KCl-mediated depolarization has been used extensively to gain insight into the precise locus of modulation because of the different ways in which these paradigms achieve stimulation (Nicholls, 1993; see also Materials and Methods). In short, given that 4AP- and KCl-mediated depolarizations activate the same VDCCs (Vazquez and Sanchez-Prieto, 1997), DOI should arguably inhibit the Ca²⁺ influx mediated by either depolarizing agent and thereby inhibit glutamate release equally well. To examine this issue, we looked at the effect of DOI on the release evoked by high external KCl concentrations. In notable contrast to the modulation of 4AP-induced release by DOI, KCl (30 mM)-evoked glutamate release was not significantly affected by 50 μM DOI (control 25.8 ± 0.7 nmol/mg/5 min; DOI 25.4 ± 0.4 nmol/mg/5 min; n = 6; P > 0.05; Fig. 5A). This lack of effect of 5-HT_2A receptor activation on KCl-evoked glutamate release was borne out in experiments looking at KCl-evoked Ca²⁺ entry. Again, in contrast to 4AP-evoked VDCC activation, 30 mM KCl-induced Ca²⁺ influx was not affected by DOI (Fig. 5B). The lack of effect of DOI on KCl-evoked Ca²⁺ influx is therefore in congruence with the ineffectiveness of the agonist with KCl evoked glutamate release (Fig. 5A). From the higher release obtained with 30 mM KCl compared with 3 mM 4AP, a plausible explanation for the lack of effect of DOI may be that the stimulation with 30 mM KCl is perhaps too strong to be down-modulated, as has been observed with de facto inhibitory receptors (Nicholls, 1998; Perkinton and Sihra, 1998). However, our observation that DOI failed to have any effect, even when 15 mM KCl was used to evoke synaptosomal depolarization (Fig. 5A, inset), argues against this possibility.

Although the above-mentioned experiments argue for 5-HT_2A receptor-mediated effects on 4AP-induced Ca²⁺ influx and glutamate release being mediated through an effect on synaptosomal excitability, GPCRs have been reported to modulate the exocytotic process directly (Blackmer et al., 2001, 2005). To determine whether the regulation of glutamate release by DOI also reflects putative actions downstream of Ca²⁺ influx, we examined the effects of DOI on glutamate release induced by ionomycin (5 μM) or hypertonic solution (500 mM sucrose). The Ca²⁺-selective ionophore ionomycin causes a direct increase in intrasynaptosomal Ca²⁺ levels and triggers release of neurotransmitter without depolarization and Ca²⁺ channel activation (Sihra et al., 1992), whereas hypertonic solution increases the probability of release by emptying the readily releasable pool through a Ca²⁺-independent mechanism (Rosenmund and Stevens, 1996). These treatments therefore allow the assessment of only those modulatory influences directly affecting synaptic vesicle trafficking and exocytosis, without the involvement of upstream ion-channel function. As shown in Figure 6A, ionomycin (5 μM) caused a control glutamate release of 19.9 ± 0.6 nmol/mg/5 min. In the presence of 50 μM DOI, ionomycin-induced release of glutamate was not significantly affected (20.4 ± 0.8 nmol/mg/5 min; Fig. 6A). Similarly, the release evoked by 500 mM sucrose (23.9 ± 0.9 nmol/mg/5 min) was also unaltered by pretreatment with DOI (24.1 ± 1.1 nmol/mg/5 min; n = 5 Fig. 6B). Therefore, these results suggest that the DOI-mediated inhibition of glutamate release is not a consequence of a direct action of 5-HT_2A receptor signaling on release events downstream of Ca²⁺ influx.

Involvement of a PLC/PKC Pathway in DOI-Induced Inhibition of 4AP-Evoked Glutamate Release

Molecular cloning has established that all 5-HT receptor subtypes, except 5-HT₃ receptors, belong to the superfamily of the GPCRs (Hoyer and Martin, 1997). We therefore examined the signal transduction mechanism involved in the DOI-mediated inhibition of glutamate release through 5-HT_2A receptor activation. For a number of systems, 5-HT_2A receptors have been suggested to be coupled to PLC activation through the G_q proteins. As shown in Figure 7A, application of the PLC inhibitor U73122 (10 μM) itself reduced the 4AP-evoked glutamate release from 20.6 ± 0.7 nmol/mg/5 min to 12.7 ± 0.3 nmol/mg/5 min, a substantial inhibition (63.2 ± 3.2% control; Fig. 7C) indicating a basal PLC activity. Notably, however, when U73122 (10 μM) was applied together with DOI (50 μM), the inhibitory effect of DOI on 4AP-evoked glutamate release was significantly occluded (DOI 57.7% ± 2.1% control; U73122 + DOI 60.7% ± 2.8% control; Fig. 7A,C). The specificity of the effect of U73122 was evaluated by using the inactive analog U73343. U73343 (10 μM) had no effect on either control 4AP-evoked glutamate release (U73343 97% ± 1.9% control; Fig. 7C) or inhibition thereof by DOI (U73343 + DOI 59.2% ± 2.5% control; Fig. 7C).

Downstream of PLC activation, the two possibilities of diacylglycerol-mediated activation of PKC- and/or IP₃-mediated liberation of Ca²⁺ stores exist as mediators of the modulation by 5-HT_2A receptor/PLC activation. Although there is a paucity of data supporting the involvement intrasynaptosomal Ca²⁺ stores in glutamate release, there is a plethora of data implicating PKC as a key modulator of presynaptic function (Barrie et al., 1991; Sihra and Nichols, 1993). To examine the involvement of PKC in the modulation seen with 5-HT_2A receptor activation, we performed experiments to test the effect of DOI following inhibition of PKC. Treatment of synaptosomes with the PKC inhibitor Ro320432 (10 μM for 30 min) itself attenuated control 4AP-evoked glutamate release from 20.1 ± 0.8 nmol/mg/5 min to 16.2 ± 0.4 nmol/mg/5 min (n = 6; P < 0.05; Fig. 7A,C). Under this condition, DOI (50 μM) further reduced release to only 15.3 ± 1.0 nmol/mg/5 min (Fig. 7C). In the presence of Ro320432, therefore, DOI induced a statistically insignificant inhibition of glutamate release of 4.9% ± 4.5% compared with the 42.3% ± 3.6% inhibition produced by DOI alone. This occlusion of the effect of DOI by PKC inhibition with Ro320432 was also obtained with another PKC inhibitor, GF109203X. As with Ro320432, although GF109203X (10 μM) attenuated control 4AP-evoked glutamate release by 23.4% ± 3.3%, it also occluded the inhibitory effect of DOI to 4.5% ± 3.1%, a statistically insignificant change (Fig. 7C).

These results with PLC and PKC inhibitors together indicate that 5-HT_2A receptor-inhibited glutamate release involves a PLC/PKC-dependent pathway. However, these results and those described above are at odds with previous reports invoking a facilitatory function of 5-HT_2A receptors operating through PLC/PKC and, indeed, our own studies over the years showing that activation of this signaling cascade largely facilitates the release of glutamate from rat cerebrocortical nerve terminals (Coffey et al., 1993, 1994; Wang and Sihra, 2004). Inhibitory modulation has largely been attributed to inhibitory GPCRs coupled to G_i/o (Wang et al., 2002; Wang and Sihra, 2003) that are pertussis (PTX) sensitive. We could, however, find no evidence for the inhibitory coupling of 5-HT_2A receptor observed herein being through a PTX-sensitive G-protein (data not shown).

Inhibition of Glutamate Release by Presynaptic 5-HT_2A Receptors

Although many previous studies have found that glutamatergic synaptic transmission and glutamate release in the cerebral cortex are increased by 5-HT_2A receptors (Aghajanian and Marek, 1997, 1999b; Marek et al., 2000; Scruggs et al., 2003), activation of 5-HT_2A receptors has also been reported to inhibit synaptic plasticity in the visual cortex (Edagawa et al., 2000) as well as glutamate release from cerebellar mossy fibers (Marcoli et al., 2001). These disparities and the relatively few reports of effects of 5-HT_2A receptors on glutamate release studied directly in nerve terminals (Marcoli et al., 2001) prompted this investigation. In the present study, the principal finding was that isolated cerebrocortical glutamatergic nerve terminals have functional 5-HT_2A receptors. Intriguingly, the activation of these receptors by DOI inhibited 4AP-evoked glutamate release through a mechanism that involves the suppression of synaptosomal excitability and voltage-dependent Ca²⁺-influx. These observations stand against the predominance of studies showing facilitatory effects 5-HT_2A receptors on synaptic transmission, but perhaps even more surprising was our observation that the inhibitory effects of 5-HT_2A receptor activation are also contingent on PLC/PKC activity.

Given the contentious nature of our primary observation of an inhibition of glutamate release by DOI, it was important to obviate the possibility that the agonist has nonspecific effects. Two lines of evidence argue against this parsimonious explanation. First, the lack of effect of DOI with alternative secretagogues such as KCl, ionomycin, or hypertonic sucrose argues against any spurious effect of the compound on nerve terminal function. Second, and most important, the DOI inhibition of glutamate release and Ca²⁺ influx was completely blockable by ketanserin, a selective 5-HT_2A receptor antagonist with moderate affinity for 5-HT_2C receptor (Baxter et al., 1995), strongly implying a receptor-mediated regulation.

DOI displays partial agonist activity at 5-HT₂ receptors, which mainly reflects actions at 5-HT_2A (Krebs-Thomson et al., 1998; Smith et al., 1998), although some activity of the agonist has also been attributed to 5-HT_2C receptor stimulation (Chen et al., 2003). In the present study, the abrogation of the inhibitory effect DOI on 4AP-evoked glutamate release by ketanserin (a selective 5-HT_2A receptor antagonist), together with the lack of effect of RS100221 (a 5-HT_2C receptor antagonist), favors the presence of 5-HT_2A- rather than 5-HT_2C-receptor subtypes on glutamatergic nerve terminals in the rat cerebral cortex. Anatomical studies examining the presence of cortical 5-HT_2A receptors have indicated a somatodendritic localization of 5-HT_2A receptors on pyramidal and GABAergic neuron (Willins et al., 1997; Jakab and Goldman-Rakic, 1998; Santana et al., 2004), with a small proportion of receptors on axonal varicosities, which were apparently nonglutamatergic (Miner et al., 2003). Notwithstanding these observations, the present functional analysis argues for the presence of 5-HT_2A receptors on glutamategic nerve terminals, in agreement with previous reports in which nerve terminals were studied in isolation (Marcoli et al., 2001). The apparent discrepancy between the receptor localization studies and the present report may reflect the relative high sensitivity of functional assays of neurotransmitter release modulation compared with immunocytochemical detection of expression levels of receptors. This is somewhat borne out by the functional demonstration of 5-HT_2A receptors on glutamatergic terminals revealed by sensitive electrophysiological analyses of excitatory postsynaptic potentials in cortical slices (Aghajanian and Marek, 1997, 1999b; Marek and Aghajanian, 1999; Marek et al., 2000) and cortical microdialysis (Scruggs et al., 2003).

Mechanism of DOI-Induced Inhibition of Glutamate Release

Addressing the cellular mechanism of action of 5-HT_2A receptors in the DOI-mediated inhibition of glutamate release, we considered three possibilities that might underlie the modulation: 1) alteration of synaptosomal excitability and downstream modulation of Ca²⁺ influx into the terminal, 2) direct regulation of VDCCs to affecting Ca²⁺ entry, and/or 3) direct modulation of the vesicular release machinery. The last of these possibilities was obviated by our observation that DOI did not affect ionomycin- or hypertonic sucrose-induced release. On the other hand, our demonstration, using the Ca²⁺ indicator Fura-2, that the 4AP-evoked increase in [Ca²⁺]_C was reduced by DOI in a ketanserin-sensitive manner suggests that 5-HT_2A receptor activation may be attenuating glutamate release by indirectly or directly reducing Ca²⁺ influx into cerebrocortical nerve terminals. We examined whether Ca²⁺ entry specifically through P/Q- and N- type VDCCs subserves the effect of DOI by assessing the effect of the agonist in the presence of Ca²⁺ channel toxins, singly or in combination. Our observation that the inhibitory effect of DOI on 4AP-evoked glutamate release was completely prevented only under conditions in which all of the release-coupled VDCCs had been blocked indicates that 5-HT_2A receptors effectively suppress glutamate release components supported by P/Q- and N-type VDCCs. However, the fact that DOI remains largely effective in the presence of ω-CgTx GVIA (to block N-type VDCCs) seems to implicate P/Q-type VDCCs as being predominantly targeted by 5-HT_2A receptor activation.

Given these results, the question remained of whether the suppression of release-coupled VDCCs by 5-HT_2A receptor action reflected a direct effect on VDCC function. To address this issue, we compared the relative effects of DOI on the 4AP- vs. KCl-evoked activation of synaptosomal Ca²⁺ entry. Both of these depolarizing treatments are thought to activate P/Q- and N-type VDCCs coupled to glutamate similarly and thus should reflect this by qualitatively similar modulation, if this occurs at the level of VDCCs. Where the two depolarizing paradigms differ is in that, whereas 4AP-mediated depolarization/VDCC activation involves upstream ion (K⁺ and Na⁺)-channel activity, 30 mM external KCl effects experimental clamping of the K⁺ electrochemical potential and hence membrane potential (Nicholls, 1998). As such, the latter type of depolarization bypasses regulation at the level of K⁺ channels and also effectively removes voltage-dependent Na⁺ channel involvement, in that these channels desensitize in response to strong and continued depolarization (Barrie et al., 1991). Comparison of the two depolarizing agents revealed that 4AP- but not 30 mM KCl-evoked VDCC activation was inhibited by DOI stimulation of 5-HT_2A receptors. The lack of effect of DOI on 30 mM KCl-induced Ca²⁺ influx was paralleled by the similar inability of DOI to affect 30 mM KCl-induced glutamate release.

Based on the mechanistic differences between the 4AP- and the KCl-mediated depolarization discussed above, the data presented are consistent with the regulation of glutamate release mediated by 5-HT_2A receptor activation impinging upstream of VDCCs, possibly at the level of synaptosomal excitability. In line with this, 4AP-mediated depolarization, probed using the membrane potential-sensitive dye DiSC₃(5), was found to be inhibited by DOI. Such an effect of DOI on synaptosomal excitability could be brought about as result of a 5-HT_2A receptor-mediated regulation of hyperpolarizing K⁺ channels, as has been proposed previously (Bartrup and Newberry, 1994), or, alternatively, through a mechanism involving 5-HT_2A receptor stabilization of Na⁺ channels to limit 4AP-mediated depolarization of the synaptosomal membrane potential.

Notwithstanding the rationale for the use of differential modulation of 4AP- and KCl-mediated as an indicator of the locus of regulatory action being upstream of Ca²⁺ entry, it is prudent to consider an important alternative scenario. It is possible that the modulatory effects of presynaptic metabotropic receptors impinge on the activation kinetics of presynaptic ion channels. In such a situation, modulation by 5-HT_2A receptor activation would be apparent only with 4AP-depolarization, during which channels would be repetitively opening and closing, and not with KCl depolarization, when constantly clamped/activated. In the absence of any means to address this directly in synaptosomes, we would therefore be circumspect about stating that 5-HT_2A receptor activation affects only loci upstream of VDCC activation rather than VDCCs directly. Having said this, direct metabotropic receptor inhibition of the VDCC is generally thought to involve G_i/o coupling. Not only are 5-HT_2A receptors proposed to tranduce via G_q but we could find no evidence for G_i/o involvement on the basis of pertussis toxin sensitivity of the 5-HT_2A receptor-mediated inhibition reported herein (data not shown). Our tentative conclusion is therefore that the inhibition of glutamate release by 5-HT_2A receptor activation occurs primarily by suppression of synaptosomal excitability.

In synaptic terminals, the activation of K⁺ channels is recognized to result in presynaptic inhibition resulting from the enhancement of K⁺ conductance bringing about nerve terminal hyperpolarization. This causes a subsequent decrease in voltage-dependent presynaptic Ca²⁺ influx and a consequent reduction of neurotransmitter release (Nicoll, 1988). Such a mechanism could potentially subserve the 5-HT_2A receptor-mediated inhibition 4AP-mediated glutamate release from cerebrocortical synaptosomes. Apart from the directly observed attenuation of 4AP-evoked synaptosomal plasma membrane depolarization by DOI, our observation that DOI inhibits Ca²⁺-independent glutamate release also points to a suppression of synaptosomal excitability. Together with the Ca²⁺-dependent exocytosis of glutamate, Ca²⁺-independent, nonvesicular/cytosolic efflux can also be elicited directly as a result of the depolarization of the nerve terminals by using biochemical methods (Nicholls et al., 1987). It is known that, under such conditions (possibly resembling those occurring during brain ischaemia), the uptake carrier can work in reverse to cause the efflux of glutamate out of the neuron and thus essentially serve to invoke a Ca²⁺-independent, nonvesicular mechanism for glutamate release (Nicholls, 1993; Szatkowski and Attwell, 1994). The magnitude of this efflux of glutamate is a direct reflection of the extent of depolarization-dependent reversal of the electrogenic glutamate transporter produced by biochemical secretagogues (Nicholls et al., 1987). Thus our observation that DOI inhibits this efflux supports the hypothesis that 5-HT_2A receptor activation leads to a suppressive effect on nerve terminal excitability, although a direct receptor-mediated inhibition of the glutamate carrier itself cannot be ruled out at this point. Experiments using the glutamate transport blockers L-trans-PDC and DL-TBOA, though inconclusive on the latter point, do serve to show that, even in the absence of transporter activity, an effect of DOI on the physiological, Ca²⁺-dependent release of glutamate persists, albeit at a reduced level.

The inhibitory effect of 5-HT_2A receptor activation on glutamate release through a reduction in synaptosomal excitability and/or alteration in VDCC activity is surprising in that it contrasts with several previous reports utilizing slice and in vivo preparations. Thus, although 5-HT_2A receptors have been proposed to exist on glutamatergic terminals from whole-cell patch clamp studies of rat prefrontal cortical slices (Aghajanian and Marek, 1997, 1999b; Marek and Aghajanian, 1999; Marek et al., 2000) and cortical microdialysis analysis (Scruggs et al., 2003), the modulation observed has been found to be facilitatory. Moreover, the regulation was suggested to operate by the inhibition of voltage-gated K⁺ channels, which would increase neuronal excitability (Lambe and Aghajanian, 2001). Where inhibitory responses to 5-HT_2A receptors have been observed in the past, they have been attributed to 5-HT_2A receptors on GABAergic interneurons facilitating GABA release to hyperpolarize distal pyramidal neurons (Tanaka and North, 1993; Marek and Aghajanian, 1994; Zhou and Hablitz, 1999). Although the latter mechanism clearly cannot operate within synaptosomal preparation and under the conditions of our experiments, a previous study has shown direct inhibitory 5-HT_2A receptor coupling to hyperpolarizing K⁺ channels (Bartrup and Newberry, 1994). The latter type of influence of 5-HT_2A receptors would be in line with our observation with synaptosomes, but, notably, the aforementioned study was carried out with C6 glioma cells and without looking at neurotransmitter release. In another notable case in which isolated nerve terminals were utilized, in congruence with the current observations, DOI was found to inhibit synaptosomal glutamate release, albeit through an undefined mechanism (Marcoli et al., 2001). What then might be the reason(s) for the aforementioned diametrically conflicting observations? Nothwithstanding the contrasting techniques used to demonstrate the facilatatory effects of 5-HT_2A receptor in previous studies and the inhibitory activity reported herein, we sought to understand the reasons for the discrepancy by examining whether this could be due to differential signaling mechanisms evoked by 5-HT_2A receptor activation.

Second Messenger Involvement in 5-HT_2A Receptor-Mediated Inhibition of Glutamate Release

5-HT_2A receptors are representative of a family of neurotransmitter receptors known to couple functionally to G_q/G₁₁ proteins to stimulate PLC and phosphatidylinositide turnover, resulting in the activation of PKC (Hoyer et al., 1994). Consistently with this, in the present study, a G-protein coupled to the PLC/PKC pathway seems to be involved in the inhibitory action of DOI on 4AP-evoked glutamate release from cerebrocortical synaptosomes. Two lines of evidence support this notion. First, the inhibitory effect of DOI on 4AP-evoked glutamate release was occluded by inhibiting PLC with the specific inhibitor U73122, but not by the inactive analogue U73343. Second, in synaptosomes treated with the specific PKC inhibitors Ro320432 or GF109203X, the inhibition of release by DOI was substantially reduced (∼4% of inhibition remained). The lack of complete suppression of the DOI-induced inhibitory action may be explained by the incomplete block of PKC by 10 μM Ro320432 or GF 109203X. These results with PLC and PKC inhibitors, taken together with the possibility that DOI reduced 4AP-induced synaptosomal excitability, suggest that a 5-HT_2A receptor-PLC/PKC activity either increases K⁺ conductance or decreases Na⁺ conductance to reduce Ca²⁺ influx through P/Q- and N-type Ca²⁺ channels and, in turn, reduces glutamate release. Despite the inhibitory coupling, we found no evidence for the 5-HT_2A receptor modulation of 4AP-evoked glutamate release being PTX-sensitive (data not shown), as has been the case for number of other inhibitory presynaptic receptors that we and others have shown to modulate glutamate release (Sanchez-Prieto et al., 1996; Wang et al., 2002; Wang and Sihra, 2003).

The conclusion from this analysis is that the presynaptic inhibitory action of 5-HT_2A receptors on glutamate release from synaptosomes is mediated by a G-protein-coupled PLC/PKC-dependent pathway. The question is how the inhibitory effect of 5-HT_2A receptor activation reported herein can be consolidated with the previous nonsynaptosomal studies. In these earlier studies, using electrophysiological techniques in rat cortical neurons and in vivo microdialysis, 5-HT_2A receptor activation was reported to facilitate excitatory synaptic transmission and glutamate release, respectively, again, through G_q/PLC/PKC coupling (Aghajanian and Marek, 1999a; Lambe and Aghajanian, 2001; Scruggs et al., 2003). Given that the slice and in vivo preparations would not delineate pre- and postsynaptic receptor responses, one possibility is that these reports reflect the overall effect of 5-HT_2A receptors at multiple localizations, whereas the synaptosomal studies exclusively reflect the behavior of a presynaptic form of the receptor. However, given that there is little evidence for alternative forms of the 5-HT_2A receptor in different neuronal localizations, this situation seems unlikely. Another tenable hypothesis to explain the divergent data would be that the coupling of presynaptically localized 5-HT_2A receptor through a G_q/PLC/PKC mechanism is switched to an inhibitory state. There is some precedence for switching of G_q/PLC/PKC signaling from facilitatory to inhibitory modes from work with metabotropic glutamate receptors (mGluRs). Thus, although type 1 mGluRs are well-established facilitators of glutamate release through G_q/PLC/PKC coupling (Sanchez-Prieto et al., 1996), following prolonged receptor activation they are able to undergo a temporal switch to effect release inhibition by the same G_q/PLC/PKC pathway (Rodriguez-Moreno et al., 1998; Herrero et al., 1998). The switching appears to be contingent on the balance between kinase and phosphatase activity (Sistiaga and Sanchez-Prieto, 2000). It remains to be seen whether the reported discrepancies in the direction of 5-HT_2A receptor-mediated regulation of glutamate release reflect such a mode switching of a G_q/PLC/PKC pathway, depending on the particular experimental scenario.

One final possibility to explain the dichotomy of modulation seen pertains to the potential specific pharmacological actions of DOI. It is well established that basal/constitutive PLC/PKC in isolated nerve terminals is high in the absence of any agonist (Barrie et al., 1991; Coffey et al., 1993). If, as has be indicated for several G-protein-coupled receptors, this constitutive activity is in part produced by the unliganded stimulatory activity of 5-HT_2A receptor/G_q/PLC/PKC cascade, then there is the potential for an agonist to produce a paradoxical suppression of this signaling pathway. Given this, one possible mechanism by which DOI might cause an inhibition of glutamate release is by acting on the 5-HT_2A receptor/G_q/PLC/PKC cascade as an inverse agonist. The paucity of specific agonists at 5-HT_2A receptors makes the examination of this possibility problematic at the present time, but it is interesting to note, in this regard, that a 5-HT_2A receptor ligand, LY367265, purported to be a receptor antagonist has an inhibitory effect similar to that of DOI with respect to glutamate release (Wang, 2005).

Therapeutic Considerations

Better knowledge of the 5-HT_2A receptor-induced inhibition of glutamate release in cerebrocortical nerve terminals is of potential interest for a number of pathological conditions including neurodegenerative and psychiatric disorders. In fact, excessive glutamate release has been proposed to be involved in the pathogenesis of these diseases (Beal, 1992), and suppression of glutamate release from nerve terminals is considered to be a novel treatment strategy for CNS disorders. On the basis of the results presented, we would suggest that presynaptic 5-HT_2A receptors might have a physiological role in preventing the excessive accumulation of extracellular glutamate occurring as a result of repetitive activity and thereby potentially act in a neuroprotective manner in cerebral cortex.