Activation of ATP-sensitive K⁺ (K_ATP) channels by H₂O₂ underlies glutamate-dependent inhibition of striatal dopamine release

Abstract

In many cells, ATP-sensitive K⁺ channels (K_ATP channels) couple metabolic state to excitability. In pancreatic beta cells, for example, this coupling regulates insulin release. Although K_ATP channels are abundantly expressed in the brain, their physiological role and the factors that regulate them are poorly understood. One potential regulator is H₂O₂. We reported previously that dopamine (DA) release in the striatum is modulated by endogenous H₂O₂, generated downstream from glutamatergic α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)-receptor activation. Here we investigated whether H₂O₂-sensitive K_ATP channels contribute to DA-release modulation by glutamate and γ-aminobutyric acid (GABA). This question is important because DA–glutamate interactions underlie brain functions, including motor control and cognition. Synaptic DA release was evoked by using local electrical stimulation in slices of guinea pig striatum and monitored in real time with carbon-fiber microelectrodes and fast-scan cyclic voltammetry. The K_ATP-channel antagonist glibenclamide abolished the H₂O₂-dependent increase in DA release usually seen with AMPA-receptor blockade by GYKI-52466 [1-(4-aminophenyl)-4-methyl-7,8-methylenedioxy-5H-2,3-benzodiazepine hydrochloride] and the decrease in DA release seen with GABA-type-A-receptor blockade by picrotoxin. In contrast, 5-hydroxydecanoate, a mitochondrial K_ATP-channel blocker, was ineffective, as were sulpiride, a D₂-receptor antagonist, and tertiapin, a G protein-coupled K⁺-channel inhibitor. Diazoxide, a sulfonylurea receptor 1 (SUR1)selective K_ATP-channel opener, prevented DA modulation by H₂O₂, glutamate, and GABA, whereas cromakalim, a SUR2-selective opener, did not. Thus, endogenous H₂O₂ activates SUR1-containing K_ATP channels in the plasma membrane to inhibit DA release. These data not only demonstrate that K_ATP channels can modulate CNS transmitter release in response to fast-synaptic transmission but also introduce H₂O₂ as a K_ATP-channel regulator.

ATP-sensitive K⁺ channels (K_ATP channels) belong to a class of inwardly rectifying K⁺ channels that are activated by a decrease in the ATP/ADP ratio (1–4). The resulting hyperpolarization lowers cell activity and energy consumption and thereby links metabolic state to excitability (3). Importantly, K_ATP channels also play a role in the regulation of secretory events. A classic example is K_ATP-channel-dependent regulation of insulin secretion from pancreatic beta cells (5, 6). With normal extracellular levels of glucose, beta-cell K_ATP channels are open; when glucose levels increase, ATP levels rise, leading to K_ATP-channel closure, membrane depolarization, and consequent Ca²⁺-dependent insulin release (7).

The function of K_ATP channels in the CNS is less well defined. These channels are widely expressed in brain, with high levels in nigrostriatal dopamine (DA) cells and striatum (8–14). In contrast to pancreatic beta cells, however, K_ATP channels in many brain cell types are normally closed. For example, in DA neurons of the substantia nigra pars compacta, K_ATP channels do not contribute to resting membrane properties, although membrane hyperpolarization occurs when these channels are activated by selective openers or decreased levels of oxygen and/or glucose (15–20). Consistent with these data, activation of K_ATP channels by specific openers or by metabolic inhibition decreases the release of several neurotransmitters, including DA (21, 22), γ-aminobutyric acid (GABA) (15, 23–26), and glutamate (27–31). Together, these data have suggested that brain K_ATP channels may provide neuroprotection during metabolic stress. Significantly, however, these metabolically sensitive channels can be activated also under normal physiological conditions. For example, in the hypothalamus, K_ATP channels in glucose-sensitive and -responsive cells contribute to the regulation of systemic glucose homeostasis (32–34); in the brainstem, these channels control the excitability of vagal neurons (35) and medullary respiratory cells (36). Whether K_ATP channels modulate CNS-transmitter release under nonpathological conditions and what factors regulate channel opening, however, have not been established.

We reported recently that glutamate inhibits DA release in the striatum by a pathway that involves H₂O₂ generated downstream from the activation of glutamatergic α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors (37). The discovery of this inhibitory action of glutamate resolves a long-standing conundrum about how glutamate modulates DA release in striatum in the absence of ionotropic glutamate receptors on DA terminals (38, 39). The mechanism of AMPA receptor–H₂O₂-dependent modulation is not yet known. Establishing this mechanism is important, however, because of the critical role of DA–glutamate interactions in movement control, cognitive processing, and the mediation of motivation and reward; correspondingly, DA–glutamate dysfunction has been implicated in Parkinson's disease (40, 41), schizophrenia (42, 43), and substance abuse (44, 45).

Preliminary studies suggested that the mechanism of glutamate–H₂O₂-dependent inhibition of DA release might involve K_ATP channels because the increase in DA release seen when AMPA receptors are blocked (and H₂O₂ generation is suppressed) can be prevented by the sulfonylurea tolbutamide (37). Consistent with this hypothesis, Krippeit-Drews et al. (46) have shown that exogenous H₂O₂ can lead to the opening of tolbutamide-sensitive K_ATP channels in pancreatic beta cells and thereby inhibit insulin release. On the other hand, it has been reported that tolbutamide can also inhibit K⁺ channels linked to dopaminergic D₂ receptors in both DA and non-DA cells (22, 47–49). The experiments described here were designed to resolve whether K_ATP channels or another K⁺ channel mediates glutamate- and GABA-dependent inhibition of DA release.

Materials and Methods

DA Recording in Striatal Slices. All animal handling procedures were in accordance with National Institutes of Health guidelines and were approved by the New York University School of Medicine Animal Care and Use Committee. Young adult guinea pigs (male, Hartley, 150–250 g) were deeply anesthetized with 40 mg·kg^–1 pentobarbital (i.p.) and decapitated. Coronal brain slices containing striatum (400-μm thickness) were prepared as described previously and then kept for at least 1 h in Hepesbuffered artificial cerebrospinal fluid (aCSF) before experimentation (50, 51). After transfer to a submersion recording chamber at 32°C, slices were allowed a 30-min equilibration before stimulation; the superfusing aCSF contained (in mM): 124 NaCl, 3.7 KCl, 26 NaHCO₃, 2.4 CaCl₂, 1.3 MgSO₄, 1.3 KH₂PO₄, and 10 glucose, equilibrated with 95% O₂/5% CO₂.

Fast-scan cyclic voltammetry was performed by a Millar Voltammeter (PD Systems International, West Molesey, U.K.); voltammetric procedures and data acquisition have been described (51). Carbon-fiber microelectrodes were made from single 8-μm carbon fibers; tip diameters were 2–4 μm (MPB Electrodes, London). Electrodes were calibrated in the recording chamber at 32°C after each experiment, and sensitivity to DA was determined in all experimental media used (37, 51). Release of DA was elicited in dorsolateral striatum by using a bipolar stimulating electrode on the slice surface; the carbon-fiber microelectrode was positioned between the poles, and the tip was inserted 50–100 μm beneath the slice surface (50, 51). The stimulus was a 10-Hz train of 30 pulses (pulse duration was 100 μs; pulse amplitude was 0.4–0.6 mA). Release of DA under these conditions is Ca²⁺-dependent and can be blocked by tetrodotoxin (51). Identification of released DA was based on characteristic DA voltammograms (Fig. 1B). Our previous studies confirmed that the evoked response in striatum is free from interference from other electroactive substances, including 5-hydroxytryptamine (serotonin), the DA metabolite DOPAC (3,4-dihydroxyphenylacetic acid), and ascorbate (52).

Fig. 1.

Glutamate–H₂O₂-dependent modulation of striatal DA release is blocked by glibenclamide. (A) Glibenclamide (Glib; 3 μM) caused a significant increase in evoked DA release (P < 0.01, glibenclamide vs. control; n = 5). (B) Applied voltage waveform and representative voltammograms of DA obtained during DA calibration (DA cal; 1 μM) and at maximum evoked [DA]_o during stimulation (10 Hz, 30 pulses) in normal aCSF (control) and in the presence of glibenclamide in the same striatal slice. Sampling interval was 100 ms; voltage scan rate was 800 V/s. (C) In the presence of glibenclamide, the usual effects of MCS (1 mM), GYKI-52466 (GYKI; 50 μM), and picrotoxin (100 μM) on DA release were prevented (P > 0.05, each agent vs. glibenclamide alone; n = 5). Data are given as mean ± SEM, illustrated as percentage of same-site control. Solid bars indicate the stimulation period.

Drug and Chemicals. Glibenclamide, (–)-sulpiride, 5-hydroxydecanoic acid (5-HD), mercaptosuccinic acid (MCS), 1-(4-aminophenyl)-4-methyl-7,8-methylenedioxy-5H-2,3-benzodiazepine hydrochloride (GYKI-52466), picrotoxin, and all components of aCSF and Hepes-aCSF were obtained from Sigma. Tertiapin was obtained from Peptide Institute (Osaka); diazoxide and cromakalim were obtained from Tocris Cookson (Ellisville, MO). All experimental solutions were made immediately before use. Stock solutions of glibenclamide were made in ethanol (0.02% final concentration in aCSF); diazoxide and cromakalim stock solutions were made in DMSO (0.01% final concentration in aCSF). Control data for these agents were obtained in vehicle; at the concentrations used, neither ethanol nor DMSO altered evoked DA release compared with that in aCSF alone.

Experimental Design and Statistical Analysis. Pharmacological inhibition of glutathione peroxidase by MCS inhibits H₂O₂ metabolism and can thereby increase H₂O₂ availability (53, 54). We have used MCS previously as a tool to reveal the role of endogenous H₂O₂ generated during pulse-train stimulation in DA-release modulation (37, 55). Evoked DA release in the presence of MCS is significantly lower than control release in aCSF alone; the specific involvement of H₂O₂ was confirmed by the reversal of this release suppression by catalase in the continued presence of MCS (37). Moreover, generation of modulatory H₂O₂ requires AMPA-receptor activation: MCS has no effect on DA release when AMPA receptors were blocked with GYKI-52466. Conversely, modulation of DA release by GYKI-52466 or by the GABA type A (GABA_A) receptor antagonist picrotoxin is prevented by catalase, demonstrating that H₂O₂ is required for DA-release regulation by endogenous glutamate and GABA (37).

In the present experiments, therefore, we used either MCS (1 mM) or GYKI-52466 (50 μM) (37) to screen the ability of specific K⁺ blockers and openers to alter H₂O₂-dependent modulation of DA release. Key findings were confirmed by testing an agent against MCS, GYKI-52466, and picrotoxin (100 μM) (37) individually. In all experiments, the interval between stimulation trains was 10 min. Only slices with at least three consistent evoked increases in extracellular DA concentration ([DA]_o) under control conditions were used for these experiments; this criterion excluded ≈5% of slices tested. After consistent DA release was confirmed, a blocker or opener was superfused for 30–40 min, and continual recording of evoked [DA]_o was performed; MCS, GYKI-52466, or picrotoxin was then applied in the continued presence of the test agent.

Data are given as mean ± SEM (n = number of slices) and illustrated as percentage of control. For each slice, the last three control records obtained before drug application were averaged and the mean peak [DA]_o was taken as 100% for that slice. For studies in which the effect of a given K⁺-channel agent was tested for efficacy against DA-release modulation by MCS, GYKI-52466, and picrotoxin, statistical comparisons were made by using pooled control and test agent data. In all experiments, differences in peak evoked [DA]_o between conditions were assessed by using unpaired Student's t tests and were considered to be significant when P < 0.05.

Results

Glibenclamide Prevents AMPA Receptor–H₂O₂-Dependent Inhibition of DA Release. In various cell types, including neurons and pancreatic beta cells, exogenous H₂O₂ causes activation of a K⁺ channel that results in membrane hyperpolarization that can be prevented by the nonspecific K⁺ channel blocker, barium (Ba²⁺), or the sulfonylurea, tolbutamide (46, 56). In pilot studies, Ba²⁺ (100 μM) prevented the suppression of evoked DA release (10 Hz, 30 pulses) by exogenous H₂O₂ (1.5 mM) and by MCS (1 mM) applied sequentially in a given slice (n = 3; data not shown). These findings indicated a common mechanism involving K⁺ channels for both exogenous and endogenous H₂O₂. We found also that tolbutamide (200 μM) prevented inhibition of DA release by endogenous H₂O₂ in striatum (37).

Here we tested the suggested involvement of K_ATP channels by using a more potent K_ATP-channel blocker, glibenclamide. Consistent with a role for K_ATP channels, glibenclamide (3 μM) (6, 57) caused a 60% increase in evoked [DA]_o from a control level of 2.0 ± 0.1 μM (P < 0.01; n = 5) (Fig. 1 A). Identification and quantification of evoked DA release were made by comparing voltammograms obtained during local stimulation in striatum with voltammograms recorded during subsequent DA calibration (Fig. 1B). In the continued presence of glibenclamide, the usual suppression of DA release in MCS was prevented completely (P > 0.05; n = 5) (Fig. 1C). Modulation of DA release by AMPA- and GABA_A-receptor activation was also blocked: neither GYKI-52466 nor picrotoxin altered evoked [DA]_o in the presence of glibenclamide (P > 0.05; n = 5 for GYKI-52466 or picrotoxin) (Fig. 1C). In paired controls, MCS caused a 40% decrease in evoked [DA]_o, whereas GYKI-52466 induced a 100% increase in DA release and picrotoxin caused a 50% decrease (data not shown); these control responses were similar to those reported (37, 55).

Lack of Involvement of D₂ Receptors or G Protein-Activated K⁺ Channels. Several reports have suggested a G protein-coupled interaction between K_ATP channels and D₂ receptors in the nigrostriatal pathway (22, 47, 48, 58). Most relevant for the present studies, both tolbutamide and glibenclamide have been shown to inhibit D₂-receptor-activated K⁺ channels and the resultant membrane hyperpolarization (47, 48). We therefore examined whether such interactions might underlie the prevention of DA-release modulation by these sulfonylureas. The possible involvement of D₂ receptors was tested by using the specific D₂ receptor antagonist sulpiride (1 μM) (41). Consistent with previous findings (51, 59), sulpiride caused a marked increase (80%) in maximum evoked [DA]_o (Fig. 2A) (P < 0.001; n = 6). However, the D₂-receptor blockade did not prevent the expected inhibition of DA release that accompanies glutathione peroxidase inhibition by MCS (P < 0.001, sulpiride + MCS vs. sulpiride alone; n = 6) (Fig. 2 A).

Fig. 2.

Glutamate–H₂O₂-dependent inhibition of striatal DA release does not involve D₂ receptors or G protein-coupled K⁺ channels. (A) Sulpiride (1 μM), a D₂-receptor antagonist, increased evoked [DA]_o in striatum (P < 0.001, sulpiride vs. control; n = 6). The usual suppression of pulse-train-evoked DA release seen in MCS (1 mM) persisted in the presence of sulpiride (P < 0.001, sulpiride plus MCS vs. sulpiride; n = 6). (B) Blockade of G protein-coupled K⁺ channels with tertiapin (100 nM to 1 μM) caused a significant increase in striatal DA release (P < 0.001, tertiapin vs. control; n = 7) but did not alter the usual increase in evoked [DA]_o seen with AMPA receptor blockade by GYKI-52466 (GYKI; 50 μM) (P < 0.001, tertiapin + GYKI vs. tertiapin; n = 7). Solid bars indicate the stimulation period.

We then tested the possible involvement of other G protein-coupled K⁺ channels by using tertiapin (10 nM to 1 μM) (60, 61), which is a general inhibitor of G protein-activated inwardly rectifying K⁺ channels (62). The efficacy of tertiapin at each concentration tested was indicated by an increase in evoked [DA]_o. The lowest concentration (10 nM) produced an ≈50% increase in evoked [DA]_o but did not alter the further increase in evoked [DA]_o induced by GYKI-52466 (data not shown). A maximum effect of tertiapin was seen with 100 nM; no further increase was seen with 300 nM or 1 μM. The mean increase in evoked [DA]_o for pooled data for 100 nM to 1 μM tertiapin was 62% (P < 0.001; n = 7) (Fig. 2B). Even with these supramaximal concentrations of tertiapin, the usual effect of AMPA-receptor blockade persisted (P < 0.001, tertiapin + GYKI-52466 vs. tertiapin alone; n = 7) (Fig. 2B). Together, these data show that neither D₂-receptor-linked K⁺ channels nor other G protein-activated inwardly rectifying K⁺ channels contribute to the modulation of DA release by endogenously generated H₂O₂. They confirm also that the effects of glibenclamide and tolbutamide in this experimental paradigm are K_ATP-channel specific.

H₂O₂-Dependent Inhibition of DA Release Does Not Involve Mitochondrial K_ATP Channels. Glibenclamide and tolbutamide can inhibit mitochondrial K_ATP (mitK_ATP) channels, as well as those in the plasma membrane (63, 64). To distinguish the site of K_ATP-channel activation by H₂O₂, we tested the efficacy of 5-HD, a selective inhibitor of mitK_ATP-channel function (65). Like the sulfonylureas, 5-HD (100 μM) (66) caused an increase (43%) in peak [DA]_o (P < 0.01; n = 4) (Fig. 3). However, 5-HD did not prevent the usual increase in DA release caused by blockade of AMPA receptors by GYKI-52466 (P < 0.001, 5-HD + GYKI-52466 vs. 5-HD alone; n = 4) (Fig. 3), which demonstrated a lack of involvement of mitK_ATP channels in glutamate–H₂O₂-dependent inhibition.

Fig. 3.

mitK_ATP channels do not mediate glutamate–H₂O₂-dependent modulation of striatal DA release. The mitK_ATP-channel inhibitor 5-HD (100 μM) caused an increase in evoked [DA]_o (P < 0.01, 5-HD vs. control; n = 4). However, 5-HD had no effect on the further increase in DA release that accompanied AMPA-receptor blockade by GYKI-52466 (GYKI; 50 μM) (P < 0.001, 5-HD + GYKI vs. 5-HD; n = 4). Solid bars indicate the stimulation period.

Effects of Selective K_ATP Channel Openers on DA Release. K_ATP channels are multimeric proteins containing two distinct subunits, which are an inwardly rectifying pore-forming subunit (Kir6.1 or Kir6.2) and a sulfonylurea receptor subunit 1 or 2 (SUR1/2) (7, 67–69). In brain, most channels contain Kir6.2 (70, 71). SUR1- and SUR2-based channels can be distinguished by their differential sensitivity to K_ATP-channel openers; preferential selectivity of diazoxide corresponds to SUR1 and selectivity of cromakalim corresponds to SUR2 (72, 73). We therefore compared the efficacy of these openers on evoked DA release. Activation of K_ATP channels by either diazoxide-suppressed (30 μM) (74) or cromakalim-suppressed (30 μM) (72) DA release, with a 42% decrease in peak [DA]_o in diazoxide (P < 0.01; n = 5) and a 32% decrease in peak [DA]_o in cromakalim (P < 0.01; n = 5) (Fig. 4). Diazoxide completely prevented the usual effects of MCS, GYKI-52466, and picrotoxin on DA release (Fig. 4 Upper; P > 0.05 vs. diazoxide alone for each agent; n = 5 for each). In sharp contrast, cromakalim did not alter the expected effect of any of these agents (Fig. 4 Lower): inhibition of glutathione peroxidase by MCS in the presence of cromakalim resulted in a further decrease in evoked [DA]_o (P < 0.05, cromakalim + MCS vs. cromakalim alone; n = 5). GYKI-52466 caused the usual increase in evoked [DA]_o (P < 0.05; n = 5), and PTX caused the usual decrease (P < 0.05; n = 5). These data specifically implicate H₂O₂-sensitive SUR1-containing K_ATP channels in the modulation of striatal DA release by glutamate and GABA.

Fig. 4.

Differential effects of K_ATP-channel openers on striatal DA release. (Upper) Diazoxide (30 μM), a SUR1-selective K_ATP-channel opener, decreased evoked [DA]_o in striatum (P < 0.01, diazoxide vs. control; n = 5). Moreover, diazoxide abolished the effects of MCS, GYKI-52466 (GYKI), and picrotoxin (PTX) on DA release (P > 0.05, diazoxide + each agent vs. diazoxide; n = 5). (Lower) Cromakalim (30 μM), a SUR2-selective K_ATP-channel opener, also caused a significant decrease in evoked [DA]_o (P < 0.01, cromakalim vs. control; n = 5) but did not alter the usual pattern of DA-release modulation seen with MCS, GYKI-52466, and PTX (P < 0.05, cromakalim + each agent vs. cromakalim; n = 5). Solid bars indicate the stimulation period.

Discussion

The data reported here demonstrate that K_ATP channels can modulate CNS transmitter release as a response to fast synaptic transmission by glutamate and GABA. Additionally, these studies show that H₂O₂ is the second messenger that mediates this response. Exogenous H₂O₂ has been shown to lead to K_ATP-channel opening in beta cells and subsequent inhibition of insulin release (36); however, a role for endogenous H₂O₂ in K_ATP-channel regulation has not been reported previously. The normally closed state of K_ATP channels in the CNS and their known activation by deprivation of oxygen and/or glucose have suggested that a primary function is to provide neuroprotection during ischemia or other brain injury (8, 17, 27, 29, 31, 75–77). However, these channels also have significant roles in normal physiology, including maintenance of glucose homeostasis by glucose-sensing cells of the hypothalamus (32–34) and regulation of excitability in metabolically sensitive cells of the vagal and respiratory centers (35, 36). The present studies reveal a regulatory function for K_ATP channels that requires H₂O₂ as a key signaling molecule.

Are D₂ Receptors Involved? Our previous studies with tolbutamide suggested that sulfonylurea-sensitive K⁺ channels mediate AMPA-receptor–H₂O₂-dependent modulation of DA release (37). This finding was confirmed by the present results with glibenclamide (Fig. 1). These data alone fall short of proving K_ATP-channel involvement, however, because of the potential interaction between sulfonylureas and D₂ receptors, which activate G protein-coupled K⁺ channels (78, 79). In the substantia nigra pars compacta, tolbutamide has been shown to reverse D₂-autoreceptor-mediated inhibition of DA neurons (47), and in the striatum, both tolbutamide and glibenclamide can antagonize D₂-mediated inhibition in non-DA neurons (48). These and other results have suggested a link between D₂ receptors and K_ATP channels in which signal transduction is mediated by G proteins (22, 47–49, 58), although such coupling is not seen under all experimental conditions (80, 81). The H₂O₂-sensitive K_ATP channels described in the present report appear to be distinct from striatal K_ATP channels examined previously. In the studies of Lin et al. (48), for example, tolbutamide was 10- to 100-fold more potent than glibenclamide in blocking D₂-receptor-mediated inhibition in striatal neurons. This relative potency is opposite to that required for prevention of AMPA-receptor–H₂O₂-mediated inhibition of DA release: glibenclamide was effective at 3 μM (Fig. 1), whereas 200 μM tolbutamide was required (37).

Moreover, we found no evidence for the involvement of D₂ receptors or G protein coupling in the modulation of striatal DA release by H₂O₂-sensitive K_ATP channels. Like tolbutamide or glibenclamide, the D₂ antagonist sulpiride caused a significant increase in evoked [DA]_o; unlike the sulfonylureas, however, sulpiride did not alter the usual decrease in DA release when H₂O₂ availability was increased by MCS (Fig. 2 A). Similarly, tertiapin, a G protein-activated, inwardly rectifying K⁺-channel inhibitor, caused a significant increase in evoked [DA]_o but had no effect AMPA-receptor-dependent modulation of DA release (Fig. 2B). Thus, these data demonstrate the independence of H₂O₂-sensitive K_ATP channels and D₂-receptor-linked or other G protein-coupled K⁺ channels in striatum.

Which K_ATP Channels Are Involved in DA-Release Modulation by H₂O₂? Recent evidence suggests that endogenous H₂O₂ can activate mitK_ATP channels in cardiac myocytes during ischemic preconditioning (82). A necessary experiment, therefore, was to assess the role of mitK_ATP channels in glutamate–H₂O₂-dependent modulation of striatal DA release. This assessment required the use of a selective mitK_ATP-channel blocker, 5-HD (65), because neither glibenclamide nor tolbutamide can distinguish between mitK_ATP channels and those in plasma membranes (63, 64). Interestingly, 5-HD alone caused an increase in evoked [DA]_o; this finding suggests some involvement of mitK_ATP channels in DA-release regulation (Fig. 3). One possible explanation for this increase is that 5-HD caused a decrease in mitochondrial production of reactive oxygen species, which can accompany a mitK_ATP-channel blockade (83); mitochondrial respiration is a significant source of cellular reactive oxygen species, including H₂O₂ (84). Despite this effect of 5-HD under control conditions, this mitK_ATP-channel blocker did not prevent a further increase in evoked [DA]_o when AMPA receptors were blocked, demonstrating that mitK_ATP channels are not involved in the dynamic modulation of DA release by H₂O₂.

Final confirmation that K_ATP channels are required for DA-release modulation by H₂O₂, glutamate, and GABA was provided by the loss of efficacy of MCS, GYKI-52466, and picrotoxin when K_ATP channels had already been opened by diazoxide (Fig. 4). Together with the inability of SUR2-selective cromakalim to alter H₂O₂-dependent DA modulation by these agents, this result specifically implicates SUR1-based K_ATP channels (71, 72). These findings do not exclude other roles for SUR2-based channels in DA modulation because cromakalim, like diazoxide, did cause a significant decrease in control-evoked [DA]_o (Fig. 4). However, the data indicate that the K_ATP channels targeted by AMPA-receptor-dependent H₂O₂ in striatum contain SUR1 subunits. The question of whether the effect of H₂O₂ is direct or mediated by additional pathways [e.g., by altering ATP levels as exogenous H₂O₂ does in pancreatic beta cells (46)] remains open.

The most logical location for H₂O₂-sensitive K_ATP channels to modulate striatal DA release is on the presynaptic membrane of DA release sites. Such localization is supported by the high expression of sulfonylurea binding sites in DA cells of the substantia nigra pars compacta (8, 10, 11, 13), which project to dorsal striatum (85). Moreover, a majority of K_ATP channels in DA cells express Kir6.2 subunits combined with SUR1 rather than SUR2 subunits because of the greater metabolic sensitivity of SUR1- vs. SUR2-based channels (86). The most likely site of AMPA-receptor-dependent H₂O₂ generation, on the other hand, is in medium spiny neurons, given the absence of AMPA and GABA_A receptors on DA terminals and the abundance of these receptors on spiny neurons (37–39, 87). Although K_ATP channels are also expressed in medium spiny neurons (88), the present findings argue against significant involvement of those channels. We have already shown that activation of inhibitory GABA_A receptors opposes AMPA-receptor-dependent H₂O₂ generation and thereby increases DA release (37). K_ATP-channel opening in medium spiny neurons would be expected to have a similar effect by opposing AMPA-receptor-dependent excitation and H₂O₂ generation. Thus, if K_ATP channels in medium spiny neurons were the primary targets of H₂O₂, one would expect enhanced DA release in the presence of K_ATP openers and suppressed DA release in the presence of K_ATP-channel blockers. However, the opposite behaviors were observed (Figs. 1 and 4), consistent with direct presynaptic regulation of DA release. Together, these data support a model of striatal DA-release regulation (37) in which AMPA-receptor-dependent H₂O₂ diffuses from postsynaptic sites in medium spiny neurons to inhibit DA release by activating H₂O₂-sensitive K_ATP channels in presynaptic DA terminals.

Conclusions

The data reported here demonstrate a functional role for K_ATP channels in the regulation of transmitter release in dorsal striatum. Moreover, they reveal that these channels are normally activated by H₂O₂ generated downstream from AMPA-receptor activation and modulated by GABA_A-receptor input. The specific K_ATP channels involved are plasmalemmal SUR1-based channels, with probable location on DA axon terminals. Given that endogenous H₂O₂ also inhibits synaptic DA release in ventral striatum (nucleus accumbens) and somatodendritic DA release in the substantia nigra pars compacta (55), we anticipate that K_ATP channels will be shown to mediate H₂O₂-dependent modulation of DA release throughout motor and reward centers of the brain. The present findings should therefore help to clarify normal DA–glutamate interactions in these regions. Because DA–glutamate dysfunction has been implicated as a causal factor in Parkinson's disease, schizophrenia, and addiction (40–45, 89), these data may also suggest novel pathways through which such dysfunction could occur.

As noted above, K_ATP channels are expressed abundantly in DA cells; the most metabolically sensitive of these channels are composed of SUR1 and Kir6.2 subunits (86). The characteristics of DA-cell K_ATP channels (e.g., whether they are G protein coupled and the extent to which they interact with D₂ receptors) are unresolved. The K_ATP channels shown here to mediate inhibition of axonal DA release in striatum are sensitive to endogenous H₂O₂, are not G protein coupled, and are independent from D₂ receptors. Whether these are the defining characteristics of known K_ATP channels expressed in DA cells or whether they describe a novel class of K_ATP channel awaits further investigation.