|Home | About | Journals | Submit | Contact Us | Français|
Hydrogen peroxide (H2O2) is emerging as a ubiquitous small-molecule messenger in biology, particularly in the brain, but underlying mechanisms of peroxide signaling remain an open frontier for study. For example, dynamic dopamine transmission in dorsolateral striatum is regulated on a subsecond timescale by glutamate via H2O2 signaling, which activates ATP-sensitive potassium (KATP) channels to inhibit dopamine release. However, the origin of this modulatory H2O2 has been elusive. Here we addressed three possible sources of H2O2 produced for rapid neuronal signaling in striatum: mitochondrial respiration; monoamine oxidase (MAO); and NADPH oxidase (Nox). Evoked dopamine release in guinea-pig striatal slices was monitored with carbon-fiber microelectrodes and fast-scan cyclic voltammetry. Using direct fluorescence imaging of H2O2 and tissue analysis of ATP, we found that co-application of rotenone (50 nM), a mitochondrial complex I inhibitor, and succinate (5 mM), a complex II substrate, limited H2O2 production, but maintained tissue ATP content. Strikingly, co-application of rotenone and succinate also prevented glutamate-dependent regulation of dopamine release, implicating mitochondrial H2O2 in release modulation. By contrast, inhibitors of MAO or Nox had no effect on dopamine release, suggesting a limited role for these metabolic enzymes in rapid H2O2 production in the striatum. These data provide the first demonstration that respiring mitochondria are the primary source of H2O2 generation for dynamic neuronal signaling.
Beginning with Ramón y Cajal's discovery of gaps between neurons (Ramón y Cajal, 1909), neurotransmission has been considered to be “hard-wired,” with point-to-point synaptic connections providing interneuronal communication. However, nonsynaptic communication by diffusion-based volume transmission (Fuxe and Agnati, 1991; Vizi, 2000) is also increasingly appreciated as playing a critical role. For example, dopamine, a key motor-system transmitter in the striatum, acts by volume transmission to activate predominantly extrasynaptic receptors after synaptic release (Sesack et al., 1994; Yung et al., 1995; Cragg and Rice, 2004; Rice and Cragg, 2008). In this context, an emerging diffusible messenger is the reactive oxygen species (ROS), hydrogen peroxide (H2O2) (Atkins and Sweat, 1999; Kamsler and Segal, 2005; Avshalumov and Rice, 2003; Avshalumov et al., 2003, 2007). Importantly, H2O2 mediates the regulation of striatal dopamine release by the classical synaptic transmitter, glutamate (Avshalumov et al., 2003, 2008), in the absence of glutamate synapses or ionotropic receptors on dopaminergic axons (Smith and Bolam, 1990; Bernard et al., 1997, Bernard and Bolam, 1998; Chen et al., 1998).
Evidence for H2O2 involvement in modulation of striatal dopamine release by glutamate comes from several avenues. Blockade of glutamatergic AMPA receptors (AMPARs) causes an increase in locally evoked dopamine release, which is prevented by the H2O2-metabolizing enzymes glutathione (GSH) peroxidase or catalase (Avshalumov et al., 2003). Conversely, locally evoked dopamine release is suppressed when H2O2 levels are amplified by inhibition of GSH peroxidase; this suppression is lost when AMPARs are blocked, demonstrating that modulatory H2O2 generation requires glutamate (Avshalumov et al., 2003). The mechanism of release inhibition is H2O2-dependent activation of ATP-sensitive K+ (KATP) channels (Avshalumov and Rice, 2003) and a key cellular source of modulatory H2O2 is striatal medium spiny neurons (MSNs) (Avshalumov et al., 2008).
The subcellular source of H2O2 generation has been elusive, however. Three potential sources might contribute. The first is mitochondrial respiration, which produces superoxide anion (•O2-) by the one-electron reduction of molecular oxygen (O2), with subsequent conversion of •O2- to H2O2 by superoxide dismutase or spontaneous dismutation (Boveris et al., 1973; Adam-Vizi, 2007). The second is monoamine oxidase (MAO), which catalyzes deamination of dopamine through a two-electron reduction of O2 to H2O2 (Maker et al., 1981) and is expressed abundantly in striatum (Azzaro et al., 1985). The third is NADPH oxidase (Nox), a family of enzymes that catalyze the one-electron reduction of O2 to form •O2- and consequently H2O2 (Lambeth, 2004; Infanger et al., 2006; Rhee, 2006; Bedard and Krause, 2007). Nox has been implicated in a variety of signaling pathways and is also found in striatum (Infanger et al., 2006; Kishida and Klann, 2007).
Here we examined contributions from these subcellular sources to rapid H2O2-dependent signaling. Synaptic release of dopamine was elicited by pulse-train stimulation in guinea-pig striatal slices; manipulation of mitochondrial H2O2 generation was monitored in MSNs using fluorescence imaging. The data show that mitochondrial respiration is the primary source of modulatory H2O2 and reveal an exquisite interplay among neuronal activity, mitochondrial respiration, and transmitter release, bridged by a unique signaling molecule, H2O2.
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 50 mg/kg pentobarbital (i.p.) and decapitated. For voltammetric recording, coronal brain slices (400 μm) containing striatum were prepared as described previously (Chen and Rice, 2001; Avshalumov et al., 2003). In some experiments, ROS generation was monitored in striatal MSNs. Brain slices for these studies were prepared from animals that were perfused intracardially with ice-cold modified artificial CSF (ACSF) (Bao et al., 2005; Avshalumov et al., 2008). Slices were maintained in a holding chamber for at least 1 h at room temperature before experimentation in HEPES-buffered ACSF containing (in mM): 120 NaCl; 5 KCl; 20 NaHCO3; 6.7 HEPES acid; 3.3 HEPES salt; 2 CaCl2; 2 MgSO4; and 10 glucose, equilibrated with 95% O2/5% CO2 (Rice et al., 1994). For recording, slices were transferred to a submersion chamber at 32°C and superfused at 1.2 mL/min with bicarbonate-buffered ACSF containing (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 (Rice et al., 1994; Avshalumov et al., 2003). Media osmolarity was ~300 mOsmol/L (Osmette osmometer; Precision System Inc., MA, USA); in experiments with disodium succinate, osmolarity was maintained by decreasing NaCl.
Axonal dopamine release in dorsolateral striatum was evoked using a surface bipolar stimulating electrode. The resulting increase in extracellular dopamine concentration ([DA]o) was monitored using fast-scan cyclic voltammetry (Millar Voltammeter, Queen Mary, University of London, UK) with 8 μm carbon-fiber microelectrodes that were either made in-house or purchased (MPB Electrodes, London, UK; or WPI, Sarasota, FL, USA). The stimulus was a 30-pulse train (10 Hz) applied at 10-min intervals; pulse duration was 100 μs and amplitude was 0.6-0.9 mA. Under these conditions, dopamine release is tetrodotoxin (TTX) sensitive and Ca2+ dependent (Chen and Rice, 2001). Scan rate for fast-scan cyclic voltammetry was 800 V/s, voltage range was -0.7 to +1.3 V versus Ag/AgCl, and sampling interval was 100 ms. Data acquisition and analysis were as described previously (Chen and Rice, 2001).
After a 30 min slice equilibration period in the recording chamber, evoked [DA]o was monitored until release was stable for at least three stimulations; once stable, evoked [DA]o was constant for at least two hours in ACSF alone. Maximal drug effects were typically seen within 20-40 min of application with stable release levels recorded subsequently. Released dopamine was identified by characteristic oxidation and reduction peak potentials (Chen and Rice, 2001; Bao et al., 2005) (Fig. 1b); [DA]o was calculated from post-experimental electrode calibration in the recording chamber at 32°C in all media used in a given experiment (e.g., ACSF and ACSF plus succinate). All agents used were screened initially for possible interference with dopamine detection. No interference was seen with any agents reported here, although several Nox inhibitors did interfere, as noted in Results, and were excluded from further testing.
Intracellular H2O2 generation was monitored using fluorescence imaging with Redoxfluor-1 (RF1), which is a reversible fluorescein-based dye coupled to a disulfide/dithiol switch that permits reversible activation of the dye by ROS (Miller et al., 2007a). This ROS-sensitive dye was loaded into individual cells via a patch pipette used for whole-cell recording (Avshalumov et al., 2005, 2008; Bao et al., 2005). Striatal MSNs were identified morphologically (e.g., Fig. 2b) and by their electrophysiological characteristics, as previously (Bao et al., 2005; Avshalumov et al., 2008). The intracellular solution contained (in mM): 120 K-gluconate; 20 KCl; 2 MgCl; 10 Na-HEPES; 10 EGTA; 3 Na2-ATP; 0.2 GTP; pH adjusted to 7.2-7.3 with KOH, as well as Alexa Red (0.1%) for cell visualization and RF1 (5 μM) for H2O2 imaging. After whole-cell configuration was obtained, cells were held for ~20 min to allow dye infiltration. Images were acquired using a Cascade CCD camera (Roper Scientific, Tucson, AZ, USA) and an illuminator and monochrometer from Photon Technology International (Lawrenceville, NJ, USA) and were analyzed using ImageMaster software (Photon Technology International), as described previously (Avshalumov et al 2005, 2008; Bao et al. 2005). Fluorescence intensity for 535 nm emission in background-subtracted frame averages were normalized; average data are presented as [(intensity - basal)/(basal)] × 100%.
Striatal ATP content was determined in separate experiments using a TD-20/20 Luminometer (Turner Biosystems, Sunnyvale, CA, USA) with an Enlighten ATP assay kit (Promega, Madison, WI, USA), as previously (Bao et al., 2005). This method is based on the luciferase-catalyzed reaction of ATP with luciferin with light emission at 560 nm. Tissue ATP content was determined in unfrozen tissue samples from striatal slices; ATP content is given as μmol/g tissue wet weight (Bao et al., 2005). To mirror the paradigm used for dopamine release studies, slices were incubated in ACSF for 60 min, then transferred to chambers containing 50 nM rotenone, 5 mM succinate, or 50 nM rotenone plus 5 mM succinate in ACSF for 90 min, all at 32°C. Paired control samples were taken from the contralateral striatum after incubation in ACSF alone for the same time periods.
Disodium succinate, glibenclamide, rotenone, mercaptosuccinic acid, pargyline hydrochloride, clorgyline, phenylarsine oxide (PAO), and all ACSF components were from Sigma (St. Louis, MO, USA); GYKI-52466 was from Tocris (Ellisville, MO, USA) and catalase (bovine liver) was from Calbiochem (La Jolla, CA, USA). Stock solutions of glibenclamide, rotenone and PAO were made in DMSO (≤0.1% final DMSO concentration in ACSF) and stock solution of GYKI-52466 were made in 0.1 M HCl; neither vehicle alone altered evoked [DA]o.
Voltammetric data are given as mean ± SEM (n = number of slices) and illustrated as percent 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. Differences in peak evoked [DA]o between the effect of a single test agent and same-site controls were assessed using paired Student's t-test. One-way ANOVA with Dunnett's post-hoc analysis was used to assess the effect of multiple agents, to assess the time course of drug effects, and to compare drug-induced changes under different conditions. Paired t-tests were used to assess ATP content data and one-way ANOVA was used to assess RF1 fluorescence data. Differences were considered to be significant for p < 0.05.
In isolated mitochondria, partial inhibition of mitochondrial respiration by rotenone, a complex I inhibitor, causes an increase in H2O2 production by complex I (Servais et al., 2003; Sherer et al., 2003). Succinate, a complex II substrate, also causes an increase in H2O2 production by complex I by reverse electron flow from complex II to complex I; this increase in H2O2 generation is blocked by rotenone at nanomolar concentrations (Liu et al., 2002; Servais et al., 2003; Gyulkhandanyan and Pennefather, 2004). Our previous work showed that nanomolar rotenone also produces functionally relevant levels of H2O2 in MSNs in the complex neuron-glial microenvironment of striatal slices (Bao et al., 2005). This unstimulated increase in mitochondrial H2O2 generation leads to suppression of evoked dopamine release in striatal slices that is prevented by catalase and by KATP-channel blockade.
In light of those data, we asked whether succinate-dependent H2O2 generation also had the functional consequence of suppressing dopamine release in striatal slices. Importantly, Yodoya et al. (2006) have shown that succinate can enter neurons via Na+-coupled carboxylate transporters, making this experiment feasible. Under control conditions, the average [DA]o evoked by local pulse-train stimulation (30 pulses, 10 Hz) was 1.18 ± 0.07 μM and was stable for at least 2 h when elicited at 10 min intervals. When slices were exposed to 5 mM succinate, the concentration examined previously with isolated mitochondria (Liu et al., 2002; Gyulkhandanyan and Pennefather, 2004), however, evoked [DA]o fell by 20-30% (p < 0.01 vs. same site control; n = 6) within 10 min, with a maximum decrease of 30-40% (p < 0.01; n = 6) seen after 30 min (Fig. 1a-c; see also Fig. 2d). Importantly, succinate-induced DA release inhibition was reversible, with a return of evoked [DA]o to control levels after succinate washout (typically 40 min wash) (p > 0.05 washout vs. control; n = 6) (Fig. 1a-c). A lower concentration of succinate (1 mM) had no effect on evoked [DA]o after 60 min exposure (p > 0.05 vs. control; n = 5), whereas higher concentrations of succinate (10-50 mM) caused a progressive and irreversible decline in evoked [DA]o. Consequently, 5 mM succinate was used in all subsequent studies.
Demonstrating that succinate-induced DA release suppression required H2O2 generation, the effect of 5 mM succinate on evoked [DA]o was prevented by catalase (500 units/mL) (n = 6, p > 0.05) (Fig 1d,f), which has been shown previously to prevent H2O2-dependent suppression of DA release (Avshalumov et al., 2003, 2007). Heat-inactivated catalase did not alter the effect of succinate on evoked [DA]o (not shown). To ascertain whether the H2O2-dependent effect of succinate was mediated by KATP channels, we examined whether DA release suppression could be prevented by glibenclamide, a KATP-channel blocker. In the presence of glibenclamide (3 μM; Avshalumov and Rice, 2003), the effect of 5 mM succinate on evoked [DA]o was lost (Fig. 1e,f) (p > 0.05; n = 6).
To confirm that exogenously applied succinate, like rotenone (Bao et al., 2005), indeed increase H2O2 generation in brain slices, we directly visualized intracellular H2O2 produced in striatal MSNs using RF1, a recently described imaging dye that is reversibly activated by H2O2 and other ROS (Miller et al., 2007a). The reversibility of RF1 offers a distinct advantage over other available dyes like DCF (dichlorodihydrofluorescein) that undergo irreversible activation by ROS (Avshalumov et al., 2007). Succinate (5 mM) caused a robust increase in RF1 fluorescence, indicating increased ROS levels (p < 0.01; n = 7) (Fig. 2a-c). The succinate-induced increase in RF1 fluorescence was prevented by catalase (500 units/mL), confirming the H2O2 dependence of the RF1 response (not illustrated).
Moreover, the succinate-induced increase in RF1 fluorescence was reversed by the addition of 50 nM rotenone, which is near the IC50 for inhibition of mitochondrial complex I (Betarbet et al., 2000; Votyakova and Reynolds, 2001) (p > 0.05 succinate + rotenone vs. control; n = 7) (Fig. 2a,c). No decrement in RF1 fluorescence was seen over the same time frame in succinate alone. These data demonstrate that, as in isolated mitochondria (Liu et al., 2002; Servais et al., 2003; Gyulkhandanyan and Pennefather, 2004), rotenone plus succinate inhibits mitochondrial H2O2 generation in striatal neurons.
We then examined the effect of 50 nM rotenone on the H2O2-dependent inhibition of dopamine release by succinate. Again, exposure of striatal slices to 5 mM succinate caused a significant suppression of evoked [DA]o within 10 min (p < 0.01; n = 6) (Fig. 2d). Consistent with our RF1 imaging data (Fig. 2a,c), rotenone reversed the succinate-induced inhibition of evoked [DA]o to control levels in the continued presence of succinate (p > 0.05 succinate + rotenone vs. control; n = 6) (Fig. 2d-f), demonstrating the loss of H2O2-dependent dopamine release suppression in the presence of rotenone plus succinate. It should be noted that the time course of dopamine release suppression by succinate and recovery in succinate + rotenone is prolonged compared to changes in H2O2 generation indicated by RF1 fluorescence recorded in single MSNs Fig. 2a,d). This is due primarily to the difference in recording site for the two measurements. Evoked [DA]o reflects a population response so that the time course of pharmacological effects on dopamine release necessarily includes restricted diffusion of agents into a slice. By contrast, imaging data are taken from superficial cells so that little, if any, tissue penetration is required, leading to a faster response.
We reported previously that striatal ATP content is maintained during 30 min exposure to 50 nM rotenone (Bao et al., 2005). In the present studies, we found that when the period of exposure was increased to 90 min, the typical duration of dopamine release experiments, 50 nM rotenone did cause a small, but significant decrease in striatal ATP content compared to paired controls (1.05 ± 0.04 μmol/g rotenone vs. 1.28 ± 0.05 μmol/g ACSF; p < 0.001; n = 10 slice pairs) (Fig 3). By contrast, 90 min exposure to 5 mM succinate caused no change in striatal ATP content (1.26 ± 0.04 μmol/g succinate vs. 1.27 ± 0.04 μmol/g ACSF; p > 0.05; n = 6 slice pairs) (Fig. 3). Moreover, when succinate was included with rotenone, striatal ATP content was also maintained during 90 min incubation (1.29 ± 0.05 μmol/g rotenone + succinate vs. 1.26 ± 0.05 μmol/g ACSF; p > 0.05; n = 27 slice pairs) (Fig. 3). Thus, the rotenone-succinate cocktail not only suppresses H2O2 generation (Fig. 2), but also maintains ATP production.
Having established that 50 nM rotenone plus 5 mM succinate can suppress mitochondrial ROS generation, we examined whether the effect of blocking glutamatergic AMPARs on evoked [DA]o was altered in the presence of the rotenone-succinate cocktail. Under control conditions, AMPAR blockade by a selective antagonist, GYKI-52466 (50 μM; Avshalumov et al. 2003), caused an expected ~75% increase in evoked [DA]o (p < 0.001 vs. same site control; n = 5) (Fig. 4a,c), which previous work has shown to be entirely H2O2-dependent (Avshalumov et al., 2003). However, when mitochondrial H2O2 generation was inhibited by rotenone-succinate, the effect of GYKI-52466 on evoked [DA]o was lost (p > 0.05 rotenone-succinate + GYKI vs. rotenonesuccinate alone; n = 6) (Fig. 4b,c).
In contrast to the enhancement of pulse-train evoked [DA]o seen with AMPAR blockade, a decrease in evoked [DA]o is seen when endogenous H2O2 levels are amplified by inhibition of GSH peroxidase with MCS (1 mM; Avshalumov et al., 2003, 2005). The effect of MCS is prevented by AMPAR blockade, showing glutamate dependence, and by catalase, confirming H2O2 dependence (Avshalumov et al., 2003). Here, MCS caused a typical ~35% decrease in evoked [DA]o under control conditions (p < 0.01 vs. same-site control; n = 5) (Fig. 4d,f). This effect was prevented when MCS was applied in the presence of the rotenone-succinate cocktail (p > 0.05 rotenone-succinate + MCS vs. rotenone-succinate alone; n = 5) (Fig. 4e,f). These data indicate a primary role for the electron transport chain of mitochondria in the generation of modulatory H2O2 on a subsecond timescale.
We next examined whether dopamine metabolism by MAO might also be a source of dynamically generated H2O2. Two MAO isoforms, type A (MAO-A) and type B (MAO-B) are found in guinea-pig striatum, consistent with their key roles in dopamine metabolism (Azzaro et al., 1985). We therefore tested a cocktail of clorgyline, a selective inhibitor of MAO-A, and pargyline, a selective inhibitor of MAO-B at concentrations of 10 μM each, based on the use of similar concentrations in brain slices (Rice et al. 1994; Bonnet et al., 2000) that are ≥100-fold higher than that found necessary to inhibit MAO activity in synaptosomes (Azzaro et al., 1985). Application of clorgyline-pargyline alone for 90 min did not alter evoked [DA]o (p > 0.05 vs. same site control; n = 6) (Fig. 5a,b). Moreover, the presence of this MAO inhibitor cocktail had no effect on the usual increase in evoked [DA]o seen with GYKI-52466 (p < 0.001 vs. clorgylinepargyline alone, n = 6; p > 0.05 for the GYKI-52466-induced increase in evoked [DA]o in clorgyline-pargyline versus that in ACSF alone; n = 5-6) (Fig. 5c,d). Similarly, the usual ~35% decrease in evoked [DA]o seen when GSH peroxidase is inhibited by MCS (1 mM) was also unaltered in the presence of the clorgyline-pargyline cocktail (p < 0.01 vs. clorgyline-pargyline alone, n = 6; p > 0.05 for the MCS-induced decrease in evoked [DA]o in clorgyline-pargyline versus that in ACSF alone; n = 5-6) (Fig. 5e,f). These results indicate that MAO does not contribute to dynamic glutamate and H2O2-dependent modulation of evoked [DA]o.
Lastly, we examined whether Nox family enzymes might contribute to modulatory H2O2 generation. Of several Nox inhibitors screened, including DPI (diphenyleneiodonium chloride), cadmium(II), AEBSF (4-(2-aminoethyl)-benzenesulfonyl fluoride), and apocynin, only PAO did not interfere with voltammetric detection of evoked dopamine release, making it the only suitable drug for these studies. Significant inhibition of Nox activity in neutrophils is seen with 2 μM PAO, with complete inhibition at 10 μM (Doussiere et al., 2001). We limited the concentration of PAO tested to 10 μM because of other reported effects of this thiol-modifying agent, including inhibition of catecholamine release from adrenal chromaffin cells reported for 20 μM PAO (Schaefer et al., 1994). Indeed, we found that 100 μM PAO caused a precipitous fall in evoked [DA]o in striatal slices. In contrast, 10 μM PAO caused a slight, but non-significant increase in evoked [DA]o that persisted over 60 min of exposure (p > 0.05 vs. same site control; n = 7) (Fig. 6a,b), with a fall in evoked [DA]o sometimes seen with longer exposure times. Within a 60 min window, however, the presence of 10 μM PAO did not prevent the usual increase in evoked [DA]o seen with GYKI-52466 (p < 0.001 vs. PAO alone, n = 5; p > 0.05 for the GYKI-52466-induced increase in evoked [DA]o in PAO versus that in ACSF alone, n = 5 for each group) or the decrease in evoked [DA]o by MCS (p < 0.01 PAO + MCS vs. PAO alone; n = 7; p > 0.05 for the MCS-induced decrease in evoked [DA]o in clorgyline-pargyline versus that in ACSF alone, n = 5-7) (Fig. 6). Together with the efficacy of the rotenone-succinate cocktail in preventing the consequences of AMPAR blockade and GSH peroxidase inhibition on dopamine release, these results suggest that Nox does not contribute significantly to the generation of modulatory H2O2 in the striatum.
It is increasingly recognized that H2O2 can act as a signaling molecule in the CNS and other tissues via a variety of H2O2 sensing systems, including transcription factors, phosphatases, kinases, and ion channels (Rhee, 2006; Rhee et al., 2005; Kishida and Klann, 2007; Avshalumov et al., 2007; Veal et al., 2007). Many of the pathways are relatively slow, on the order of minutes to hours, with the time course of H2O2 generation occurring on a similar scale, as seen in cultured neurons exposed to a growth factor (EGF) that activates intra-neuronal, Nox-dependent generation of H2O2 (Miller et al., 2007b). On the other hand, H2O2 can also activate ion channels to mediate inter-neuronal signaling on a subsecond time scale, as seen in the inhibition of striatal dopamine release by glutamatergic activation of AMPARs, consequent H2O2 generation and KATP-channel opening, which occurs within 500 ms after initiation of a train of stimuli (Avshalumov et al., 2003; Avshalumov and Rice, 2003, 2008). The subcellular source of neuromodulatory H2O2 must therefore operate on a time scale commensurate with this rapid signaling mechanism.
We show here for the first time that the source of dynamic, glutamate-dependent H2O2 generation in the striatum is mitochondrial respiration. The effect of AMPAR blockade on evoked dopamine release was completely prevented by a cocktail of rotenone and succinate that suppressed mitochondrial H2O2 generation, yet maintained tissue ATP content. Strikingly, the effects of rotenone and succinate on H2O2 generation in striatal slices (Fig. 2 and Bao et al., 2005) mirror the effects of these agents in isolated mitochondria (Liu et al., 2002; Servais et al., 2003; Gyulkhandanyan and Pennefather, 2004). These data demonstrate that manipulation of mitochondrial complexes I and II leads to functionally relevant H2O2 production, even in the intricate neuron-glia microenvironment of brain slices. Thus, the brain antioxidant network permits signaling by H2O2 (Avshalumov et al., 2004; Rhee et al., 2005; D'Autreaux and Toledano, 2007), yet at the same time prevents oxidative damage, indicted by the reversibility of dynamic, H2O2-dependent inhibition of dopamine release (e.g., Figs. 1a and 2b2b).
Of the three possible sources of modulatory H2O2 examined, mitochondria, MAO, and Nox, the mitochondrial electron transport chain could be predicted to6 be the most likely source, given the rapid responsiveness of mitochondrial respiration to increased cellular activity. Studies in isolated mitochondria suggest that a small but significant percentage of O2 metabolized produces •O2- and H2O2 (Boveris and Chance, 1973), with increased O2 metabolism causing increased ROS production (Dykens, 1994). The cellular distribution of MAO in the striatum also argues against a significant role in the generation of modulatory H2O2. Most obviously, MAO-B is found predominantly in glia, rather than neurons (Azarro et al., 1985; Levitt et al., 1985). The predominant neuronal isoform, MAO-A, is expressed in MSNs, but at low levels (Francis et al., 1985). Given that the dopamine transporter (DAT) is expressed by dopaminergic axons but not other cellular elements in the striatum (Nirenberg et al., 1996; Hersch et al., 1997), lack of a specific dopamine uptake mechanism in either astrocytes or MSNs would further limit dopamine metabolism and H2O2 production by MAO in these cell types. Importantly, suppression of dopamine release by modulatory H2O2 persists when the DAT is inhibited (Avshalumov et al., 2003), indicating a lack of contribution from metabolism of released dopamine by MAO-A in dopaminergic axons, as well. In contrast to neuron-glia segregation of MAO isoforms, there is evidence for Nox expression in striatal MSNs and in surrounding neuropil, although the specific distribution remains unresolved from possible species or methodological differences in previous studies (Serrano et al., 2003; Kim et al., 2005).
Mitochondria are abundant in neuronal axons and dendrites, including dopaminergic axons in striatum. However, the absence of ionotropic glutamate receptors on dopaminergic axons (Bernard and Bolam, 1998; Chen et al., 1998) and the lack of effect of GYKI-42466 or MCS on dopamine release evoked by selective dopamine pathway stimulation (Avshalumov et al., 2008) suggest that these are not the site of glutamate-dependent, mitochondrial H2O2 generation. Rather, our previous studies suggest that generation of modulatory H2O2 occurs predominantly in MSNs (Avshalumov et al., 2008), which are the principal output cells of the striatum that compose 90-95% of the striatal neuron population (Kemp and Powell, 1971). Interestingly, glutamate synapses are closely apposed to dopamine synapses on the dendritic spines of these neurons (Smith and Bolam, 1990; Bernard and Bolam, 1998; Chen et al., 1998), placing them in an ideal position to modulate axonal dopamine release via diffusible H2O2. This anatomical organization and the dependence of H2O2 on synaptic glutamate release help ensure temporally and spatially discrete modulation of dopamine release by this process, which adds another dimension to the current understanding of signaling specificity for H2O2 and other ROS (e.g., D'Autreaux and Toledano, 2007).
The site of •O2-, and hence H2O2, generation within mitochondria is currently under debate. Studies in isolated mitochondria have shown that blockade of either complex I or III can increase ROS generation (Cadenas et al., 1977; Votyakova and Reynolds, 2001; Liu et al., 2002; Servais et al., 2003, Gyulkhandanyan and Pennefather, 2004; Adam-Vizi 2005). Consistent with such data, partial inhibition of complex I by nanomolar concentrations of rotenone also increases ROS generation in MSNs in striatal slices (Bao et al., 2005). We show here that nanomolar rotenone reversed succinate-induced H2O2 generation and consequent suppression of dopamine release, presumably by blocking succinate-driven H2O2 generation at complex I. These data support a role for complex I in the generation of modulatory H2O2. The findings may also have relevance for understanding the pathophysiology of Parkinson disease, as postmortem tissues from individuals with Parkinson disease show a deficiency in complex I activity (Schapira et al., 1990), as well as oxidative damage (for review see Lin and Beal, 2006).
The data presented show that mitochondria are the source of modulatory H2O2 in the striatum, a key brain region in motor control. Additional evidence indicates that the two other sources of H2O2 generation, MAO and Nox, do not contribute to dynamic, subsecond H2O2 signaling in the striatum. Of course, these enzymes have other critical roles in the CNS, including the key role of MAO in dopamine metabolism (Azzaro et al., 1985; Cohen et al., 1997) and the wide range of functional consequences of Nox-dependent generation of H2O2 and other ROS over longer time scales than those examined here (Lambeth, 2004; Rhee et al., 2005; Infanger et al., 2006; Miller et al., 2007b; Bedard and Krause, 2007; Kishida and Klann, 2007). The lack of contribution of MAO and Nox to rapid H2O2 signaling indicates that mitochondria alone provide a temporally linked response to neuronal activity that generates a sufficiently rapid, but controlled elevation in H2O2 required for ion channel activation. The present studies mark the first step in elucidating different layers of regulation by H2O2 that can be defined by the time course of generation.
The findings reported here also help define characteristics of physiological versus pathophysiological roles of H2O2 in the CNS. Chronic exposure to low levels of H2O2 formed from dopamine metabolism by MAO has been proposed to contribute to oxidative stress in the slow degeneration of dopaminergic neurons in the substantia nigra pars compacta in Parkinson disease (Cohen et al., 1997). Similarly, Nox-derived superoxide contributes to the pathological consequences of exposure to rotenone and to other neurotoxins that target the nigrostriatal dopamine pathway (Wu et al., 2003; Gao et al., 2003). The time frame required for the development of such pathophysiological consequences is days to years. This contrasts sharply with the subsecond time frame of the transient mitochondrial H2O2 elevation and consequent KATP-channel opening that mediate the reversible inhibition of striatal dopamine release by glutamate.
These studies were supported by NIH/NINDS grants NS-36362 and NS-45325 (M.E.R.), the National Parkinson Foundation (M.E.R.) and NIH/NIGMS grant GM 79465 (C.J.C.). C.J.C. is an Investigator with the Howard Hughes Medical Institute. We are grateful to the Turner Biosystems Luminometer Grants Program for the grant of a TD-20/20 Luminometer.