Tetrahydrobiopterin scavenges superoxide in dopaminergic neurons

Increased oxidative stresses are implicated in the pathogenesis of Parkinson's disease (PD), and dopaminergic neurons may be intrinsically susceptible to oxidative damage. However, the selective presence of tetrahydrobiopterin (BH 4 ) makes dopaminergic neurons more resistant to oxidative stress caused by glutathione depletion. To further investigate the mechanisms of BH 4 protection, we examined the effects of BH 4 on superoxide levels in individual living mesencephalic neurons. Dopaminergic neurons have intrinsically lower levels of superoxide than nondopaminergic neurons. In addition, inhibiting BH 4 synthesis increased superoxide in dopaminergic neurons, while BH 4 supplementation decreased superoxide in nondopaminergic cells. BH 4 is also a cofactor in catecholamine and NO production. In order to exclude the possibility that the antioxidant effects of BH 4 are mediated by dopamine and NO, we used fibroblasts in which neither catecholamine nor NO production occurs. In fibroblasts, BH 4 decreased baseline reactive oxygen species (ROS), and attenuated ROS increase by rotenone and antimycin A. Physiologic concentrations of BH 4 directly scavenged superoxide generated by potassium superoxide in vitro . We hypothesize that BH 4 protects dopaminergic neurons from ordinary oxidative stresses generated by dopamine and its metabolites and that environmental insults or genetic defects may disrupt this intrinsic capacity of dopaminergic neurons and contribute to their degeneration in PD.


INTRODUCTION
Parkinson's Disease (PD) is characterized by the selective degeneration of dopaminergic neurons in the substantia nigra (SN) pars compacta. While the etiology and pathogenesis of this cell death in most cases of PD remain unknown, it has been widely hypothesized that increased oxidative stress and mitochondrial dysfunction contribute to dopaminergic neuronal degeneration. Autopsy studies have noted numerous abnormalities indicative of increased oxidative damage in the SN (1)(2)(3)(4)(5). In particular, decreased glutathione in the SN has been found in autopsy brains from individuals with incidental Lewy bodies, presumed to represent presymptomatic PD (6). In addition, mitochondrial complex I activity is also decreased in the SN of PD patients (7,8), and complex I inhibition can increase the formation of reactive oxygen species (9,10).
Although these findings suggest that dopaminergic neurons are susceptible to oxidative stress, studies in vivo have not observed dopaminergic neuronal death following the oxidative stress generated by glutathione depletion (11)(12)(13). Furthermore, mesencephalic dopaminergic neurons in primary monolayer (14) and reaggregate (15) cultures are less susceptible to the toxicity of glutathione depletion than nondopaminergic neurons.
To explain this unanticipated resistance of dopaminergic neurons to oxidative stress, we hypothesized that dopaminergic neurons may contain additional antioxidants for the metabolism of ROS. One candidate is tetrahydrobiopterin (BH 4 ), which is present abundantly in dopaminergic neurons where it functions as a cofactor for tyrosine hydroxylase in catecholamine synthesis. Recently, BH 4 has also been found to decrease superoxide formed by xanthine/xanthine oxidase (16), a macrophage/phorbol myristate acetate reaction system (17), and by NOS (18,19), and to protect against a variety of oxidative stressors (20)(21)(22)(23). BH 4 also mediates the marked preferential resistance of dopaminergic neurons to glutathione depletion (15). Therefore, BH 4 may provide dopaminergic neurons with an additional pathway for the metabolism of reactive oxygen species (ROS).
In this paper, we delineate a mechanism underlying the increased resistance of dopaminergic neurons to oxidative stress. First, using microfluorimetry of individual living neurons, we show that dopaminergic neurons have lower levels of superoxide than nondopaminerigic neurons. We then show that the presence of BH 4 is both sufficient and necessary to maintain low superoxide levels in dopaminergic neurons, and in fibroblasts that have been engineered to synthesize BH 4 but do not produce dopamine or NO. Lastly, we show that BH 4 decreases superoxide in vitro through a direct scavenging mechanism.
After 48 h, the medium was changed, and 10 µg/ml 5-fluoro-2-deoxyuridine was added to inhibit glial growth. Cells were grown an additional 4-6 d prior to study. Glia represented less than 15% of the total cells in cultures, based on the number of cells that lacked immunoreactivity to neurofilament protein 200 (NFP) or neuron-specific nuclear protein (NeuN) antibodies.

Measurement of ROS
Relative superoxide levels were determined by measuring the rate of increase in ethidium (Et) fluorescence generated by the oxidation of hydroethidine (HEt) by superoxide (24). All measurements were made with a Nikon Diaphot epifluorescence microscope, and illumination with a 150 W Xe arc (attenuated by a neutral density 1.5, ultraviolet-grade filter; Omega Optical, Brattleboro, VT). Et imaging used standard rhodamine optics (excitation 510-560 nm; dichroic mirror 580 nm; emission >590 nm) (Nikon, Melville, NY). Images were formed using a 40X Fluor NA 0.85 objective (Nikon) and collected on a Hamamatsu ICCD (sensitivity set at 7.0). Images were 8-bit (256 intensity levels), and 16 frames were averaged every 10 seconds.
Background subtraction was made using the first image obtained when HEt solution was added.
Data acquisition was controlled by MetaFluor or MetaMorph software (Universal Imaging Corp.), and average intensity over identified regions of interest were logged to hard disk and displayed in real time. The relative superoxide level for each cell was determined from the slope of the increase in ethidium fluorescence over time, fit by linear regression. HEt, a dye freely permeable to cells (25,26), was freshly prepared and used at a final concentration of 3.2 µM.
We also measured the oxidation of 2,7-dichlorodihydrofluorescein (DCHF) to 2,7dichlorofluorescein (DCF) which can be caused by various ROS and reactive nitrogen species  (27). Images were obtained in the presence of 33 µM DCHF at 5 sec intervals for 15 min. Images were corrected by subtracting an autofluoresence image obtained before dye addition.
Optics utilized excitation at 460-490, dichroic mirror 500, emission >515nm. Rate of fluorescence increase was determined by fitting a line to the final 5 min by linear regression.
To quantify fluorescence, cell somata were circled, and the average cytosolic fluorescence computed. Because systematic differences in somal volumes between dopaminergic and nondopaminergic neurons would alter fluorescence intensities, any volume difference could confound our interpretation of ROS differences. Accordingly, mean cross sectional somal areas (and hence cell volume) were compared between dopaminergic and nondopaminergic neurons and were found to be comparable (data not shown).

Immunohistochemitry and post hoc identification of dopaminergic neurons
Neurons in which fluorescence measurements were made were identified as either dopaminergic or nondopaminergic based on post-hoc TH-immunohistochemistry. After each study, the location of the field of cells was marked on the slide. Cells were then washed with 0.1 M PBS, and fixed in 4% paraformaldehyde/0.1 M phosphate buffer for 15 minutes. Cells were exposed to 0.6% H 2 O 2 in PBS for 15 minutes to remove endogenous peroxidase activity, and preincubated with 0.1 M PBS containing 1% BSA and 0.2% Triton X-100 for 30 minutes. Cells were incubated at room temperature overnight in primary polyclonal anti-TH antiserum (1:1000).
Cultures were next incubated for 1 hr with anti-rabbit biotinylated secondary antibody at 1:200 dilution, and an additional 1 hr with an avidin-biotin conjugate of peroxidase (Vectastain ABC kit). Dopaminergic neurons were visualized as a black reaction product, using diaminobenzidine Following immunohistochemistry, the original field of imaged cells field was found and re-imaged. The orientation and location of stained cells were matched, and neurons were identified as dopaminergic or nondopaminergic based on immunostaining.

Western blot analysis
Fibroblasts were grown to confluence in T162 flasks prior to harvest. Cells were resuspended in 200 µl PBS, sonicated and centrifuged at 16,000 g for 2 minutes. The supernatant was frozen at -20 °C.
A total of 25 µl of supernatant from each sample were electrophoresed on 10% SDS-PAGE gels, and transferred to a polyvinylidene difluoride (PVDF) membrane. Following transfer, membranes were incubated overnight in 5% nonfat milk, and then incubated 1 h with primary antibodies against endothelial NOS (eNOS) at 1:500, inducible NOS (iNOS) at 1:1000, or constitutive NOS (cNOS, neuronal or brain NOS) at 1:175 in 3% milk. Membranes were then incubated for 1 h with the appropriate secondary (anti-rabbit horseradish peroxidase or antimouse horseradish peroxidase, 1:2000) in 3% milk, and then developed using ECL.

In Vitro Superoxide Measurements
Saturated potassium superoxide (KO 2 ) solution was made by mixing KO 2 in dimethyl sulfoxide, centrifuging at 1500 rpm for 5 min, and then decanting the supernatant (24). This procedure yields about 3.6 mM solution (28). The stock solution of KO 2 was then added to buffer Cu/Zn-SOD (9 U/mL) was used as a control.
The effects of BH 4 on KO 2 -generated superoxide levels were also measured by electron paramagnetic resonance (EPR) spectrometry using a spin trapping agent 5-diethoxy-phosphoryl- (19,29). DEPMPO was prepared as a 40 mM solution in 50 mM sodium phosphate buffer, pH 7.4, treated with an iron chelator, Catalase (300 U/ml) was also added to prevent other oxygen radicals from contaminating the EPR spectra. Spectra were recorded within 1 min after mixing, with a Varian E-12 EPR spectrometer equipped with a century series bridge (E-102) and a TM011 cavity mounted horizontally in the magnet. EPR spectra were recorded at X-band (9.  PFG produce BH 4 but not L-DOPA or dopamine (32).

BH 4 decreases superoxide in primary mesencephalic neurons
PFG produced BH 4 and PF did not (Table 1). PFG also had much lower levels of superoxide compared with PF ( Fig 2K). Addition of sepiapterin (20 µM) to PF markedly increased BH 4 levels ( Table 1) and also decreased Et fluorescence dramatically compared with controls (Fig. 2K). The decreased superoxide level in sepiapterin-treated PF was comparable to superoxide levels in PFG (Fig. 2K). Inhibition of BH 4 synthesis in PFG by DAHP (1 mM) and NAS (1 mM) decreased BH 4 PFG to undetectable levels (Table 1), and increased superoxide levels. These data indicate that BH 4 is sufficient to decrease superoxide, independent of its role as a cofactor for dopamine synthesis.
To determine if BH 4 lowers ROS through its effect on NOS, we examined whether fibroblasts express NOS. NOS has been shown to produce superoxide instead of NO in the absence of BH 4 and NO can also scavenge superoxide (33). Western blot analyses did not detect eNOS, iNOS or cNOS expression in either PF or PFG (Fig. 3). To provide further evidence for the absence of NOS effect on superoxide levels in our system, we examined the effect of L-NAME on superoxide levels. Treatment of fibroblasts with L-NAME (1 mM) for 2 h had no significant effect on basal superoxide levels or following treatment with sepiapterin (20 µM) ( To further investigate the effects of BH 4 on ROS, we employed DCF fluorescence. Oxidation of DCFH to DCF by various ROS and reactive nitrogen species has been noted (27).
Sepiapterin (20 µM) decreased DCF fluorescence in PF (Fig. 2L), and PFG had lower levels of DCF fluorescence than PF (Fig. 2L). In addition, inhibition of BH 4 synthesis in PFG by DAHP (1 mM) and NAS (1 mM) increased DCF fluorescence. These DCF data were consistent with our Et fluorescence data. Taken together, our data using Et and DCF fluorescence suggest that BH 4 decreases both superoxide and downstream superoxide metabolites such as peroxides and other ROS.

BH 4 decreases mitochondrial ROS
In order to assess the effects of BH 4 on mitochondrial production of superoxide, rotenone and antimycin A were used to block electron transport distal to the primary sites of superoxide formation in complexes I and III, respectively (10,(34)(35)(36). Both rotenone (20 nM) and antimycin A (1 µM) significantly increased superoxide levels in control fibroblasts. In both cases, this increase in superoxide was significantly attenuated in cells that either expressed BH 4 or were pretreated with sepiapterin (Fig. 4). These data indicate that BH 4 either interferes with mitochondrial ROS generation, directly scavenges ROS formed within mitochondria, or both.
Mitochodrial depolarization-induced release of ethidium and subsequent fluorescence enhancement following binding to nucleic acids can complicate the interpretation of ethidiumbased methods of superoxide measurement (37). To test for this complication in our system, we used myxothiazole to depolarize mitochondria without increasing superoxide production.

BH 4 directly scavenges ROS
To investigate whether BH 4 can scavenge superoxide directly, we used an in vitro model of KO 2generated superoxide (24,38). Consistent with previous observations, KO 2  Although KO 2 generates primarily superoxide, other ROS may be subsequently generated and scavenged by BH 4 . Therefore, to provide more direct evidence that BH 4 scavenges superoxide, we also used EPR with the spin trapping agent DEPMPO. DEPMPO is more sensitive and specific for superoxide detection than other spin traps, due to the formation of a more stable adduct (19,29). The mean signal height of the 5th line of the DEPMPO spectrum corresponding to superoxide was 275.9 ± 7.5 (SEM, an arbitrary unit, n=27) in the KO 2 (50 µM) group. This value was attenuated to 127.7 ± 2.7 (n=30, p< 0.01 by Fisher's LSD protected t-Test) when 200 µM BH 4 was added to the mixture (Fig. 6). There was no detectable superoxide signal when SOD (100 U/ml) was added as a positive control to scavenge superoxide (Fig. 6C) by TH through partial uncoupling of the hydroxylation reaction (39). Therefore, in order for normal dopaminergic neurons to survive, they may require an intrinsically increased antioxidant capacity.
Our results suggest that BH 4 provides an intrinsic antioxidant capacity unique to dopaminergic neurons. Thus, BH 4 protects against BSO (15), H 2 O 2 , UV light (22), and paraquat toxicity (40). Compared with nondopaminergic neurons, dopaminergic neurons are also more resistant to NO toxicity, but equally vulnerable to peroxynitrite toxicity (41,42). This arginine are suboptimal (19,33). Conversely, NOS can also decrease superoxide levels, since NO formed by NOS reacts rapidly with superoxide to form peroxynitrite (47). Therefore, BH 4 could lower superoxide levels both by decreasing production by NOS (48,49), and by increasing superoxide scavenging by NO. To examine whether there are direct effects of BH 4 on superoxide independent of NOS, we used fibroblasts that do not express NOS (Fig. 3). These fibroblasts had lower superoxide levels than control fibroblasts when genetically modified to produce BH 4 or provided with the BH 4 precursor sepiapterin (Fig. 2K). Thus, the superoxide lowering effect of BH 4 is not mediated through NO.
We provide the first direct evidence that BH 4 scavenges superoxide, by showing its effect on superoxide generated nonenzymatically by KO 2 (24) using both Et fluorescence and EPR measurements. Although BH 4 has previously been hypothesized to scavenge superoxide directly, these studies generated superoxide by xanthine/xanthine oxidase or a macrophage/phorbol myristate acetate reaction system (16,17,19,22,40), and hence BH 4 could have inhibited these superoxide generating enzymes. We saw a BH 4  in the dorsal raphe nucleus (52), and are therefore presumed to have comparatively low levels of BH 4 . Low levels of BH 4 could contribute to the greater susceptibility of dopaminergic neurons in the SN versus other catecholaminergic neurons in PD. In addition, the strongest known risk factor for PD is age, and BH 4 levels are known to decrease significantly with increasing age in mice (53) and humans (54). In particular, Chen et al. (55) reported that the number of GTPCHimmunoreactive nigral neurons were decreased more than TH-immunoreactive neurons in aged monkeys.
In conclusion, BH 4 directly scavenges superoxide, and thereby provides dopaminergic neurons with an enhanced antioxidant capacity. We propose that this intrinsic resistance of dopaminergic neurons to oxidative stress is critical to their survival, and disturbances in this capacity through genetic mutations or exposure to exogenous toxins could underlie their demise in PD. Understanding these processes may lead to therapeutic interventions to protect both dopaminergic and nondopaminergic neurons from a variety of oxidative stressors.     The effect of KO 2 (O) on Et fluorescence was examined in the presence of HEt (H, 10 µM) and

Figure Legends
increasing concentrations of BH 4 (B), as described in methods. SOD (9 U/mL) was used as a control. Data are expressed as means ± SEM. a p< 0.05, b p < 0.01 versus HO, by Newman-Keuls post hoc test, n = 3 per group.