The neurotoxin 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine induces apoptosis in mouse nigrostriatal glia. Relevance to nigral neuronal death and striatal neurochemical changes.

Swiss mice were given 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), 25 mg/kg/day, for 5 consecutive days and killed at different days after MPTP discontinuance. Decreases in striatal tyrosine hydroxylase activity and levels of dopamine and its metabolites were observed 1 day after MPTP discontinuance. Ascorbic acid and glutamate levels had increased, dehydroascorbic acid and GSH decreased, whereas catabolites of high-energy phosphates (inosine, hypoxanthine, xanthine, and uric acid) were unchanged. In addition, gliosis was observed in both striatum and substantia nigra compacta (SNc). Sections of SNc showed some terminal deoxynucleotidyl transferase-mediated 2'-deoxyuridine 5'-triphosphate nick end labeling (TUNEL)-positive cells. Neurochemical parameters of dopaminergic activity showed a trend toward recovery 3 days after MPTP discontinuance. At this time point, TUNEL-positive cells were detected in SNc; some of them showed nuclei with neuronal morphology. A late (days 6-11) increase in striatal dopamine oxidative metabolism, ascorbic acid oxidative status, and catabolites of high-energy phosphates were observed concomitant with nigral neuron and nigrostriatal glial cell apoptotic death, as revealed by TUNEL, acridine orange, and Hoechst staining, and transmission electron microscopy. These data suggest that MPTP-induced activation/apoptotic death of glial cells plays a key role in the sequential linkage of neurochemical and cellular events leading to dopaminergic nigral neuron apoptotic death.

Parkinson's disease (PD) 1 is characterized by a selective loss of neurons in the substantia nigra compacta (SNc) and signif-icant diminution of neostriatal content of dopamine and its major acidic metabolites dihydroxyphenylacetic acid (DOPAC) and homovanillic acid (HVA). Consequently, the functioning of the nigrostriatal dopaminergic system is impaired. Although cellular and molecular pathways leading to neuronal death in PD are still unknown, major biochemical processes such as oxidative stress and impaired energy metabolism are involved (for review, see Ref. 1). The neurotoxin 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) induces in man a clinical syndrome closely resembling PD and is widely employed as a Parkinson's disease model in experimental animals (2). Astrocytes are the primary site of bioactivation of the innocuous MPTP to the toxic metabolite 1-methyl-4-phenyl-pyridinium ion (MPP ϩ ) via the monoamine oxidase-dependent formation of a dihydropyridinium intermediate (3). The dihydropyridinium intermediate is a reactive radical that undergoes autoxidation with formation of superoxide anion (O 2 . ) and MPP ϩ . In addition, the dihydropyridinium intermediate readily crosses the cellular membrane and undergoes autoxidation also in the extracellular space to form additional O 2 . and MPP ϩ (3). MPTPinduced hydroxyl radical formation in the striatum and substantia nigra of the mouse has been shown in vivo (4); in addition, MPP ϩ alone promotes reactive oxygen species (ROS) formation (5)(6)(7). MPTP/MPP ϩ -induced neuronal death has been widely demonstrated in experimental animals, including rodents (8,9). The mode of neuronal death has been extensively studied in vitro. MPTP and MPP ϩ induce apoptosis in neuroblastoma cells (10), dopaminergic cell cultures (11,12), and in PC12 cells (13). Only a few reports, however, have addressed the mode of MPTP-induced neuronal death in vivo. It has been demonstrated in mice (9) that MPTP can induce necrotic or apoptotic neuronal death according to the dosage schedule. Recent histological studies performed on brains from Parkinsonian patients suggest that nigral cell death could be apoptotic (14,15). However, apoptotic degeneration of nigral dopaminergic neurons in PD disease is still a controversial issue (for review, see Ref. 16).
MPTP, besides damaging nigrostriatal neurons, induces glial activation (17). Glia plays a key role in the antioxidant defense of the brain (18); in addition, dopaminergic neurons in SNc, which degenerate in PD, can be maintained by the glial cell line-derived neurotrophic factor (19). Glial cell line-derived neurotrophic factor induces both protective and restorative effects on the nigrostriatal dopaminergic system of MPTPtreated mice (20) and monkeys (21). All these findings suggest that glia plays an important role in protecting the functioning of the nigrostriatal dopaminergic system from MPTP toxicity. In reality, there is evidence suggesting that alterations in glial functioning may be an important contributor to the pathologic process that occurs in PD (22,23). The purpose of this study was to determine: (i) whether MPTP might produce cell damage in nigrostriatal glial cells of the mouse besides that in dopaminergic neurons; (ii) whether the eventual cellular damage might be characteristic of changes leading to apoptosis; (iii) whether neuronal and glial damage might be correlated with MPTP-induced striatal neurochemical changes.

EXPERIMENTAL PROCEDURES
Animals-Experiments were conducted on 3-month-old Swiss mice (Morini strain, Reggio Emilia, Italy), body weight 20 -25 g, maintained, under standard animal care conditions, on a 12-h day/night cycle and given food and water ad libitum. All studies were carried out in accordance with the European EC Directive 86/609 included in the Decreto number 116/1992 of the Italian Ministry of the Public Health.
Drug Administration-Groups of 8 mice per time period received an intraperitoneal injection of MPTP-HCl, 30 mg/kg/day, at 24-h intervals on 5 consecutive days. This dosage is below that producing necrotic cell death (9). The control group (n ϭ 11) was given a vehicle (saline, 2 ml/kg intraperitoneal). Mice were killed by decapitation at 1, 3, 6, 11, and 14 days after the last MPTP injection. Controls were killed after the last MPTP group. Heads were cooled by rapid immersion in liquid nitrogen. Thereafter, striata and SNc of both sides were rapidly removed as previously described (24). Whereas striata of the left side were frozen at Ϫ80°C for subsequent HPLC determination of neurochemicals (24), striata and SNc of the right side were fixed with a freshly prepared paraformaldehyde solution (4% in PBS, pH 7.4) for light microscopy. Left SNc were fixed with 2.5% glutaraldehyde in 0.1 M cacodilate buffer for transmission electron microscopy (TEM).
Light Microscopy-After extensive washing, the fixed tissue fragments were dehydrated through a graded series of ethanol, then passed in xylene and embedded in paraffin. Specimens (5-7 m) were washed in xylene and rehydrated according to standard protocols. Sections were either stained with hematoxylin/eosin, or used for terminal deoxynucleotidyl transferase-mediated 2Ј-deoxyuridine 5Ј-triphosphate nick end labeling (TUNEL); some sections were processed for fluorescence microscopy and therefore fluorochromized with acridine orange (AO) or Hoechst 33342.
TUNEL-Striatal and nigral sections were incubated, after rehydration, with proteinase K (20 g/ml in Tris-HCl buffer, pH 7.4 -8) for 15 min at 21°C. The blocking of endogenous peroxidase was performed with a solution of 0.3% H 2 O 2 in methanol. Sections were permeabilized by incubation with a solution of Triton X-100 0.1% in PBS and incubated with fluorescein-2Ј-deoxyuridine 5Ј-triphosphate and terminal deoxynucleotidyl transferase. Thereafter, sections were stained by an antifluorescein antibody conjugated with horseradish peroxidase and subsequently incubated with diaminobenzidine substrate to produce a dark brown precipitate. Stained sections were analyzed under light microscopy.
Glial Fibrillary Acidic Protein (GFAP) and TUNEL Double Staining-Astrocytes within the SNc were identified by means of GFAP immunocytochemistry. GFAP is the major component of intermediate filaments of astrocytes and is a common marker for astrocytes. Rehydrated sections were preincubated for 30 min with 0.1 M PBS containing 20% horse serum and 0.01 M Tween 20. Sections were incubated with: (i) a monoclonal anti-GFAP antibody (1:400, DAKO) overnight at room temperature; (ii) biotinylated anti-mouse IgG antibody (DAKO) for 30 min at room temperature; (iii) buffer containing 3% H 2 O 2 to inactivate the endogenous peroxidases; (iv) alkaline phosphatase-labeled avidin (1:100) for 30 min at room temperature. Between incubation steps, sections were washed 3 times in 0.1 M PBS containing 0.001 M Tween 20. Phosphatase was visualized by fast red. At this point, sections were incubated with fluorescein-2Ј-deoxyuridine 5Ј-triphosphate and termi-nal deoxynucleotidyl transferase to characterize the apoptotic nucleus. Thereafter, sections were stained by an anti-fluorescein antibody conjugated with horseradish peroxidase and subsequently incubated with diaminobenzidine substrate to produce a dark brown precipitate. A series of stained sections were analyzed under light microscopy.
AO and Hoechst 33342 Staining-Striatal and nigral sections were defatted (9) in Tris-buffered phenol for 25 min followed by 10 min in chloroform/methanol (1:1). Sections were rinsed in 100% ethanol for 2 min, air dried, and then rinsed in freshly made 1% glacial acetic acid for 1 min followed by distilled water for 1 min. Sections were kept for 30 s in a 0.01% solution of AO in PBS (pH 6.2), rinsed in PBS with mild agitation, and then mounted in 95% glycerol in PBS for immediate viewing under excitation at 488 nm on a Nikon Optiphot microscope. Double-stranded DNA complexes with AO to emit maximally at 530 nm as a bright green fluorescence, whereas single-stranded RNA emits maximally at 640 nm as a bright orange-red fluorescence. Rehydrated sections were rinsed in PBS and then fluorochromized with Hoechst 33342 (5 g/ml in PBS) for 45 min at room temperature. The specimens were observed with an epifluorescence microscope (Nikon Optiphot).
Hematoxylin/Eosin Staining-Gliosis was assessed by means of hematoxylin/eosin staining of rehydrated sections of striatum and SNc, using routine protocol. The presence of some morphological features, such as enlarged and vesicular nuclei with prominent nucleoli, and the presence of lengthened eosinophilic structures (Rosenthal fibers) were considered as a histopathological index of gliosis (25). The amount of both enlarged nuclei and Rosenthal fibers/microscopic field allowed us to quantify gliosis.
TEM-Fragments from SNc were fixed in 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4) for 3 h at 4°C and washed with the same buffer containing 7% sucrose; the specimens were then post-fixed in 1% OsO 4 for 2 h at 4°C and washed repeatedly in buffer. Thereafter, specimens were dehydrated in acetone, passed through propylene oxide, and finally embedded in epoxy resin (agar 100). Ultrathin section (60 nm) were collected on copper grids and stained with uranyl acetate and lead citrate. The specimens were observed with a Zeiss EM 109 at 50 kV.
HPLC Determination of Striatal Neurochemicals-Dopamine, DOPAC, HVA, 3-methoxytyramine (3-MT), ascorbic acid, dehydroascorbic acid (DHAA), glutamate, inosine, hypoxanthine, xanthine, and uric acid determinations were performed by HPLC according to methods previously described (24,26). GSH was determined according to the enzymatic recycling method of Anderson (27). Striata of the left side were weighed and homogenized in 1% meta-H 3 PO 4 containing 1 mM EDTA. After centrifugation (17,500 ϫ g for 10 min at 4°C), the supernatant was divided into five aliquots. The first was filtered and immediately injected into the HPLC system for dopamine, DOPAC, HVA, 3-MT, ascorbic acid, and uric acid determinations. HPLC with electrochemical detection was done with a high-pressure pump Varian 9001 with a Rheodyne injector, 15 cm ϫ 4.6 mm inner diameter TSK-ODS-80 TM column, electrochemical detector BAS LC-4B, and integrator Spectra Physics SP 4290. The mobile phase 0.1 M was citric acid, 0.1 M K 2 HPO 4 , 1 mM EDTA, 5% MeOH, and 70 mg/liter sodium octylsulfate (pH 3.0), the flow rate was 1.2 ml/min and 10 l of sample were injected. The second aliquot was adjusted to pH 7.0 with 45% K 2 HPO 4 and 1% DL-homocysteine was added to reduce DHAA to ascorbic acid. The sample was incubated for 30 min at 25°C, then adjusted to pH 3.0 with 30% meta-H 3 PO 4 , filtered, and injected (20 l) for total ascorbic acid determination. DHAA concentration was calculated from the difference in ascorbic acid content between the first and second aliquot. The third aliquot was processed for GSH determination (26). Glutamate determinations were performed, on 20 l of the fourth aliquot, by HPLC with fluorimetric detection with a Varian model 9010 gradient system (at 360 and 450 nm excitation and emission wavelengths) following derivatization with o-phthalaldehyde. Briefly, derivatized amino acids were separated on a Ultrasphere C18 high speed 75 ϫ 4.6-mm Beckman column. A gradient was run from 26 mM Na 2 HPO 4 /acetonitrile (95:5) (pH 6.6 with H 3 PO 4 ) to Na 2 HPO 4 /acetonitrile (70:30) in 20 min at a flow rate of 1.2 ml/min. On the fifth aliquot, inosine, hypoxanthine, and xanthine determinations were performed (26) directly by injecting 50 l of the filtered supernatant using the above high-pressure pump, column ODS-80 TM and a similar precolumn, UV detector (254 nm), and integrator. The mobile phase was composed of two eluants: K 2 HPO 4 /MeOH (97:3).
Tyrosine Hydroxylase (TH) Activity-TH activity was assessed by measuring tyrosine conversion to L-DOPA, according to Pothos et al. (28). Groups of 3 mice per time period were given MPTP according to the above treatment schedule. Controls were given saline intraperitoneally. Mice were killed by decapitation at 1, 3, and 6 days after the last MPTP injection. Heads were cooled by rapid immersion in liquid nitro-gen. Thereafter, striata of both sides of each animal were rapidly removed, pooled, weighed, and homogenized in 10 volumes of (w/v) ice-cold 0.05 M Tris-HCl buffer (pH 6.0), containing 0.2% Triton X-100 (v/v). Protein content was determined on a 50-l aliquot of the homogenate according to Lowry et al. (29), whereas the remainder was centrifuged at 10,000 ϫ g for 10 min. TH activity in the supernatant was determined at the incubation conditions as describe by Coyle (30). Fifty l of the supernatant were added to 16-ml tubes containing the following reaction mixture: 10 l of 1 M KH 2 PO 4 buffer (pH 5.5), 1.100 units of catalase in 10 l of distilled water, 5 l of 6.4 mM 2-amino-4hydroxy-6,7-dimethyltetrahydropteridine (freshly prepared in 0.005 M HCl), 10 l of dihydropteridine reductase, 5 l of 0.01 M nicotinamide dinucleotide phosphate, reduced form, 10 l of 0.1 M mercaptoethanol, and 10 l of 3 mM FeSO 4 . The reaction was initiated by the addition of 10 l of 2.0 mM tyrosine; the reaction mixtures was incubated for 10, 30, or 60 min at 37°C. Experiments were done in triplicate. TH activity was assessed by HPLC-EC determination (as described above) of L-DOPA formed. The percentage variation was calculated on the basis of the total amount of L-DOPA formed at the 3 incubation time points indicated above.
Statistical Analysis-All values were expressed in nanomole or picomole of milligrams/protein and given as mean Ϯ S.E. Biochemical data were analyzed using ANOVA with Student-Newman-Keuls's t test posthoc analysis. In some instances (TH activity) unpaired t test was also used.

RESULTS
One Day after MPTP Discontinuance-Striatal dopamine and DOPAC ϩ HVA levels had significantly decreased by about 58 and 60%, respectively, as compared with controls ( Fig. 1, panel A); the (DOPAC ϩ HVA)/dopamine ratio, which is a reliable index of dopamine turnover, did not statistically differ from controls ( Fig. 1, panel C). 3-MT levels, which can be assumed to be an index of dopamine release (31), had significantly decreased by about 38% (Fig. 1, panel B). All these changes denote an impairment of the functioning of the dopaminergic system and may be related to the MPTP-induced decrease of TH activity (32). In fact, striatal TH activity, assessed by measuring tyrosine conversion to L-DOPA, had reduced by about 80%, as compared with controls ( Fig. 2). Ascorbic acid (ϩ15%) and glutamate (ϩ33%) levels had significantly increased, whereas GSH (Ϫ28%) and DHAA (Ϫ33%) levels decreased. The DHAA/ascorbic acid ratio, which can be assumed to be a reliable index of the ascorbic acid oxidative status and ROS formation (33), had decreased by about 38% (Fig. 3, panels A-D). Inosine, hypoxanthine, xanthine, and uric acid levels, as well as the inosine/hypoxanthine ratio, were in the range of control values (Fig. 4, panels A-D). It is widely accepted that increases in regional brain inosine, hypoxanthine, xanthine, and uric acid levels reflect degradation of intracellular high-energy phosphates (26,34). Therefore, the lack of changes in compounds associated with high-energy phosphate degradation denotes that at this time point MPTP did not yet induce energy failure.
At this time point (day 1), a moderate reactive gliosis, as revealed by the presence of some enlarged and vesicular nuclei with prominent nucleoli, as well as of eosinophilic Rosenthal fibers, was observed in both striatum and SNc (Fig. 5, panels c and d). The analysis of DNA fragmentation in striatal sections showed no signs of TUNEL positivity (Fig. 6, panel a), whereas some TUNEL-positive cells were detected in sections of SNc (Fig. 6, panel b). AO and Hoechst 33342 staining of SNc sections showed few pyknotic nuclei.
Three Days after MPTP Discontinuance-Striatal dopamine levels ( Fig. 1, panel A) and TH activity (Fig. 2) showed a significant trend to recovery (p Ͻ 0.05, compared with day 1 values); DOPAC ϩ HVA levels ( Fig. 1, panel A) and the (DOPAC ϩ HVA)/dopamine ratio (Fig. 1, panel C) were within the range of control values, whereas 3-MT levels were still significantly lower by about 35%, compared with controls ( Fig.   1, panel B). Ascorbic acid and DHAA levels, as well as the DHAA/AA ratio, were within the range of control values (Fig. 3, panels A, B, and D). GSH (Fig. 3, panel A) and glutamate (Fig.  3, panel C) levels had significantly decreased by about 24% and increased by about 35%, respectively. Inosine, hypoxanthine, xanthine, and uric acid levels, as well as the inosine/hypoxanthine ratio, were still in the range of control values (Fig. 4, panels A-D).
At this time point (day 3), striatal gliosis was more intense than on day 1 in both striatum and SNc, as revealed by the presence of several enlarged and vesicular cell nuclei with prominent nucleoli and eosinophilic Rosenthal fibers thicker than those seen on day 1. No signs of TUNEL positivity were detected in striatum sections (Fig. 6, panel c). In SNc sections several TUNEL-and GFAP-positive cells were observed. In addition, some nuclei with neuronal morphology showing light TUNEL positivity were detected (Fig. 6, panel d). Hoechst 33342 staining of SNc sections showed glial cells with pyknotic nuclei or condensed chromatin (apoptotic features), and neurons with presumptive normal morphology (Fig. 7, panel c). AO staining of SNc sections showed several glial cells with con- densed chromatin (apopoptic feature), and some neuronal cells with bright orange cytoplasmic fluorescence (presumptive of cellular integrity) (Fig. 7, panel d).
Six Days after MPTP Discontinuance-Striatal dopamine levels ( Fig. 1, panel A) and TH activity (Fig. 2) did not recover further. DOPAC ϩ HVA levels ( Fig. 1, panel A) and the (DOPAC ϩ HVA)/dopamine ratio (Fig. 1, panel C) significantly increased by about 25 and 76%, respectively, compared with controls. 3-MT levels were still significantly lower by about 44% (Fig. 1, panel B). Ascorbic levels were within the control range, whereas the DHAA and DHAA/ascorbic acid ratio increased by 26 and 33%, respectively (Fig. 3, panels A, B, and D). GSH (Fig. 3, panel A) and glutamate (Fig. 3, panel C) levels still showed a decrease of about 24% and an increase of about 44%, respectively. In addition, all compounds associated with highenergy phosphate degradation were significantly increased: inosine by 55%, hypoxanthine by 110%, xanthine by 195%, and uric acid by 71% (Fig. 4, panels A, C, and D). The increase in all compounds associated with high-energy phosphate degradation denotes that at this time point MPTP did indeed induce energy failure. The inosine/hypoxanthine ratio was significantly decreased, as a consequence of an increase in hypoxanthine levels greater than inosine (Fig. 4, panel B). Under physiological conditions, hypoxanthine is converted to inosinic acid by the salvage purine enzyme hypoxanthine-guanine phosphoribosyltransferase (EC 2.4.2.8); the proportion of hypoxanthine converted to inosinic acid is greater than that degraded to xanthine and uric acid by xanthine oxidase (EC 1.1.3.22) (35); in addition, hypoxanthine and inosine are incorporated into nucleotides at a similar rate (36). In some conditions, for instance in the aging brain (37), reutilization of preformed hypoxanthine prevails on de novo production of new purines to allow the cells to maintain their nucleotide pool. The increase in hypoxanthine greater than inosine levels denotes that MPTP might impair the incorporation in nucleotides of inosinic acid formed by the salvage purine enzyme hypoxanthine-guanine phosphoribosyltransferase; consequently, inosinic acid might be degraded again to inosine and then to hypoxanthine (26,37). In addition, all of the above striatal neurochemical changes denote a condition of increases in ROS formation, as revealed by increases in the DHAA/ascorbic acid ratio (33), dopamine oxidative metabolism (38 -40), as well as hypoxanthine and xanthine oxidative metabolism (41,42).
At this time point (day 6), gliosis was more intense than on day 3 in both striatum and SNc. Sections of both regions, besides several enlarged and vesicular cell nuclei with prominent nucleoli, showed extensive areas of eosinophilic structures (Rosenthal fibers), indicative of astrocytes arranged in fascicles (Fig. 5, panels e and f). In contrast to day 3, striatal sections showed several TUNEL-positive cells (Fig. 6, panel e). Thus, the detection of TUNEL-positive cells in the striatum was concomitant with increases in all compounds associated with high-energy phosphate degradation. Furthermore, the number of TUNEL-positive cell nuclei of heterogeneous size in SNc greatly increased as compared with day 3 (Fig. 6, panel f).
Hoechst 33342 staining of SNc sections showed glial cells with apparent clumps of hypercondensed chromatin and neurons with nuclear fragmentation (apoptotic features) (Fig. 7, panel e). AO staining of SNc sections showed several glial cells with condensed chromatin (apopoptic feature), and neurons with condensed chromatin margination at the nuclear periphery associated with pericellular space formation created by cell shrinkage (apoptotic features); in addition, the lack of orange cytoplasmic fluorescence is presumptive of cellular impairment (Fig. 7, panel f). TEM revealed nuclear chromatin condensation with apoptotic features in pigmented neurons (Fig. 8) and glial cells (Fig. 9). GFAP/TUNEL double staining revealed the astrocytic involvement in the apoptotic process (Fig. 10).
At this time point (day 11), both striatal and nigral gliosis showed a trend toward subsidence, compared with day 6, as revealed by the apparent reduction of extensive areas of eosinophilic structures. The number of TUNEL-positive cells in the striatum (Fig. 6, panel g) was somewhat greater than on day 6. Again, the increase in the number of TUNEL-positive cells in the striatum was concomitant with a further increase in all compounds associated with high-energy phosphate degradation. The number of TUNEL-positive cells in the SNc was in the range of that on day 6. Some TUNEL-positive cells showed nuclei with neuronal morphology (Fig. 6, panel h). AO and Hoechst 33342 staining of SNc sections showed glial and neuronal cells with features similar to those shown in Fig. 7, panels e and f (day 6).
Fourteen Days after MPTP Discontinuance-At this time point, the decrease in striatal dopamine and 3-MT levels appeared to be stable (by about 33 and 27%, respectively, com- pared with controls) (Fig. 1, panels A and B). DOPAC ϩ HVA levels and the (DOPAC ϩ HVA)/dopamine ratio (Fig. 1, panels A, C) were increased further. Ascorbic acid levels had decreased by about 19% (Fig. 3, panel A), whereas DHAA levels increased by about 55% (Fig. 3, panel B); consequently, the DHAA/AA ratio increased by about 100% (Fig. 3, panel D). Changes in GSH (Fig. 3, panel A) and glutamate (Fig. 3, panel C) levels did not differ from those of day 11. The increase in inosine, hypoxanthine, xanthine, and uric acid levels tended to subside (Fig. 4, panels A, C, and D). In addition, the inosine/ hypoxanthine ratio was in the control range (Fig. 4, panel B).
At this time point (day 14), both striatal and nigral sections showed a clear regression of gliosis (Fig. 5, panels g and h). In addition, both striatal and nigral sections showed a noticeable reduction in the number of TUNEL-positive cells, as compared with day 11. Thus, the reduction of the number of TUNELpositive cells in the striatum was concomitant with the tendency to return to control values of all compounds associated with high-energy phosphate degradation. Hoechst 33342 staining of SNc sections showed few glial cells with condensed chromatin and neurons with presumptive normal morphology (Fig.  7, panel g). AO staining of SNc sections showed some glial cells with clumps of hypercondensed chromatin (apopoptic feature), and some neuronal cells with bright orange cytoplasmic fluorescence (presumptive of cellular integrity) (Fig. 7, panel h). DISCUSSION The results of this study show that a sequential linkage of striatal neurochemical and nigrostriatal cellular events occurs following the discontinuation of a MPTP treatment schedule that is unable to induce necrotic cell death (9). Early (1 day) after MPTP discontinuance, striatal dopaminergic functioning appears to be impaired, as revealed by decreases in striatal dopamine and 3-MT levels and TH activity. However, no clear sign of nigral cell loss, as well as of striatal energy failure, could be detected at this time point. Thus, early MPTP-induced impairment of striatal dopaminergic functioning appears to be unrelated to both nigral neuronal loss and energy failure. The decline of striatal dopamine content in Parkinson's disease has been thought to arise solely from the severe loss of nigrostriatal neurons. However, the striatal dopamine deficit is greater than that presumable on the basis of the loss of nigral neurons (43), suggesting either impaired dopamine synthesis before neuronal degeneration (32), or an increase in striatal dopamine nonenzymatic oxidation and catecholamine conjugates formation (44,45), or both. Studies in animal models of Parkinson's disease demonstrate that the decrease in the markers of the striatal dopamine metabolism is far greater than the loss of nigral cell bodies (46). TH is a selective target for nitration following exposure to MPTP with a consequent loss of activity (32); in addition, striatal TH is not nitrated in mice overexpressing copper/zinc superoxide dismutase, thus supporting a critical role for superoxide radicals in TH nitration. Dopamine itself, besides TH, is a direct target of MPTP toxicity, because it has been shown that MPP ϩ triggers intracellular dopamine nonenzymatic oxidation (7).
Three days after MPTP discontinuance, clear signs of nigral neuronal loss could be detected. At this time point, however, both striatal dopamine levels and TH activity had significantly recovered up to 80% of control values. It has been shown that following MPTP systemic administration, MPTP and MPP ϩ accumulate neither in the striatum nor in the brainstem, because both are cleared off within 8 -24 h (24). Therefore, the transient inhibition of striatal TH activity and the consequent decrease in dopamine levels appear to be linked to the presence of MPP ϩ within dopaminergic terminals. These findings further demonstrate that different mechanisms underlie nigral and striatal MPTP toxicity, i.e. an early transient impairment of dopaminergic functioning at striatal sites in contrast with delayed nuclear damage at nigral sites. From 6 to 14 days after MPTP discontinuance, both dopamine levels and TH activity failed to recover further; moreover, additional signs of nigral neuronal loss were detected together with massive apoptotic death of nigrostriatal glial cells and striatal neurochemical changes denoting energy failure and increases in ROS formation. It is most likely that these neurochemical changes occur also at the nigral sites (47). The increase in striatal dopamine turnover, as revealed by the increase in the (DOPAC ϩ HVA)/ dopamine ratio, suggests that dopamine synthesis in surviving neurons was no longer affected. In reality, it has been shown that MPTP-induced partial loss of striatal dopaminergic THimmunoreactive fibers results in compensatory sprouting from intact fibers and compensatory changes in the synthesis and metabolism of dopamine (48). Taken together, these findings allow us to speculate that, at these time points, the decline of striatal dopamine content and TH activity might arise solely from the loss of dopaminergic nigral neurons.
The question arises about the mechanism by which MPTP induces loss of nigral neurons at different days after its discontinuance. In fact, TUNEL selectivity in detecting apoptosis in the MPTP in vivo model of PD is still a controversial issue (9). In the present study, TUNEL-positive and AO-and Hoechst 33342-stained cells were judged to be apoptotic because of the supportive physical evidence of nuclear condensation obtained with TEM (10).
The late MPTP-induced nigral neuron loss might have been conditioned by the massive glial apoptotic death and increases in reactive radicals formation. A pivotal role of glial pathology in PD disease has recently been outlined (22). There is evidence suggesting that oxidative stress might originate in glial cells rather than in neurons (23) and alterations in glial functioning may be an important contributor to the pathologic process that occurs in PD (22,49). Glia play a key role in the antioxidant defense of the brain (18) and maintains dopaminergic neuron functioning by supplying trophic factors (19). Today, MPTPinduced activation of resting microglia, and the consequent activation of inducible nitric-oxide (NO) synthase, is believed to be the main culprit of MPTP-induced dopaminergic neuron death (50). Blockade of MPTP-induced microglial activation by means of the tetracycline derivative minocycline mitigates the demise of dopaminergic neurons (50). In PD, activated microglia are found in proximity to damaged nigral cells, suggesting their role in triggering or amplifying neuronal injury as well as in removing the debris of injured cells (51). Lipopolysaccharideinduced activation of the rat nigral microglia is also known to induce the death of dopaminergic neurons (52); in this model, the opioid receptor antagonist naloxone protects nigral neurons without inhibiting microglial activation (52) microglia-induced dopaminergic cell injury (53). Interestingly, lipopolysaccharide-induced activation of astrocytic cultures, coupled with GSH depletion (by means of buthionine-SR-sulfoximine) and complex I inhibition (by means of MPP ϩ ), resulted in extracellular accumulation of NO, hydrogen peroxide, and glutamate (54). In the present study, GFAP immunocytochemistry showed that MPTP-induced reactive gliosis had an astrocytic component, as expected, because astrocytes represent the majority of all glial cells. The activation/apoptotic death of astrocytes, and the presence of nigral neurons with apoptotic features (as revealed by TUNEL positivity, AO and Hoechst 33342 staining, and TEM images of chromatin condensation), occurred in concomitance with decreases in GSH levels, increases in glutamate levels, and increases in compounds associated with high-energy phosphate degradation. Therefore, it is most likely that MPTP-induced activation/apoptotic death of astrocytes may contribute to microglial activation-mediated dopaminergic neuron death, because activated microglia and astrocytes share common cytotoxic mediators (mainly NO and hydrogen peroxide, which are notoriously noted as second messengers in the oxidative stress-induced apoptotic cell death).
Hydrogen peroxide (H 2 O 2) is not a free radical, but in the presence of iron it is converted via Fenton reactions to highly reactive hydroxyl radicals, according to the following reactions.  on left side, panels a, c, e, and g) and SNc (micrographs on right side, panels b, d, f, and h) of mice at different days after MPTP discontinuance. Panels a and b, controls. Panels c and d, 1 day after MPTP discontinuance; a moderate reactive gliosis is revealed by the presence of some enlarged and vesicular nuclei with prominent nucleoli (arrows), as well as of eosinophilic Rosenthal fibers (arrowheads). Panels e and f, 6 days after MPTP discontinuance: intense gliosis, as revealed by extensive areas of eosinophilic structures (Rosenthal fibers), indicative of astrocytes arranged in fascicles (arrowheads). Panels g and h, 14 days after MPTP discontinuance: regression of gliosis. Scale bar: 20 m. 6. TUNEL staining of paraffin sections from striatum (micrographs on left side, panels a, c, e, and g) and SNc (micro- graphs on right side, panels b, d, f, and h)   The above reactions outline the potential role for iron in the pathophysiological cascade of the MPTP-induced neuronal death (56). In reality, misregulation of the iron metabolism and iron-induced oxidative stress are widely believed to be an im-portant pathogenetic mechanism of neuronal death in PD (for review, see Ref. 55).

FIG.
In addition to glial activation-induced generation of NO and ROS, MPTP-induced increases in both dopamine and hypoxanthine/xanthine enzymatic oxidative metabolisms represent a further mechanism of ROS generation. MPTP-induced increases in dopamine and hypoxanthine/xanthine metabolism paralleled the increase in ascorbic acid oxidative status. Enzymatic oxidation of dopamine notoriously generates H 2 O 2 (38 -40), whereas the enzymatic oxidation of hypoxanthine and xanthine generates O 2 . (41). So far, GSH has been considered the main scavenger of H 2 O 2 generated by enzymatic dopamine oxidation (38 -40). According to the "radical sink hypothesis," GSH and superoxide dismutase act in a concerted effort to eliminate biologically generated radicals (58). However, the physiological relevance of GSH as an antioxidant has been questioned, because the ultimate formation of GSSG is preceded by the generation of highly reactive radicals, such as the glutathione thiol radical (GS ⅐ ) and the disulfide radical anion (59). Ascorbic acid can also interact nonenzymatically with H 2 O 2 and organic peroxides, but the interaction could also be enzyme-catalyzed (60). In addition, the ascorbate radical is quite unreactive and, unlike GS ⅐ , is not known to cause any biological damage (61). Ascorbic acid and GSH compartmentalization studies in the brain (62) show an ascorbic acid-to-GSH ratio of 4:1 in neurons (10 versus 2.5 mM, respectively), with an opposite ratio of 1:4 in glia (1 versus 4 mM, respectively). Sturgeon et al. (59) demonstrated that in the presence of physiological concentrations of ascorbic acid and GSH, ascorbic acid, but not GSH, was involved in the detoxification pathway of oxidizing free radicals formed by peroxidase; therefore, ascorbic acid was considered the ultimate radical sink. In the present study, whereas the time course of changes in the ascorbic acid oxidative status paralleled the time course of those of both dopamine and hypoxanthine/xanthine oxidative metabolisms, changes in GSH levels resulted unrelated. In fact, 1 day after MPTP discontinuance, GSH levels had significantly decreased; in contrast, the hypoxanthine/xanthine oxidative metabolism was unchanged, whereas that of dopamine had greatly decreased; in addition, the ascorbic acid oxidative status was significantly decreased. From 3 to 14 days after MPTP discontinuance, the increase in dopamine and hypoxanthine/xanthine oxidative metabolisms strictly paralleled the increase in ascorbic acid oxidative status; in contrast, the decrease in GSH levels was in the range of those observed 1 day after MPTP discontinuance. In a previous study (63), we demonstrated that morphineinduced increases in striatal dopamine and hypoxanthine/xanthine oxidative metabolisms were positively correlated with increases in the ascorbic acid oxidative status. Taken together, these findings strongly support the hypothesis that ascorbic acid may be the ultimate neuronal sink for radicals generated from dopamine and hypoxanthine/xanthine oxidative metabolisms.
The link between dopamine and xanthine metabolisms is as follows (39,63).
The above link is supported by the finding that allopurinol, a well known inhibitor of xanthine oxidase, inhibits in vivo both striatal xanthine and dopamine enzymatic oxidative metabolisms (64 -66). Reports during the past two decades emphasize a key role for GSH in the pathogenesis of Parkinson's disease (23). In PD, postmortem findings show an early decrease (by about 40%) in the nigral levels of GSH (67). However, because dopaminergic neurons make up only 1-2% of the total nigral cell population, and even less in PD, the extent of GSH decreases is too great to take place solely in dopaminergic neurons. This suggests that GSH depletion may occur primarily, if not exclusively, in glial cells (astrocytes and microglia) (54). Much less attention has been paid to ascorbic acid in PD. Postmortem findings showed a significant mean decrease of about 26% of ascorbic acid levels in caudate ϩ globus pallidus ϩ substantia nigra (68). As recalled above, ascorbic acid and GSH compartmentalization studies in the brain give an ascorbic acid-to-GSH ratio of 4:1 in neurons, with an opposite ratio of 1:4 in glia (62). Therefore, if The arrowhead indicates a TUNEL-positive/GFAP-negative cell, which is a neuron, because the nucleus is relatively larger than that of labeled glial cells, and it clearly shows condensed chromatin associated with the nucleolus. Scale bar: 20 m.
GSH depletion is thought to occur in PD mainly in glia, that of ascorbic acid may most likely occur in neurons. In the present study, low striatal GSH levels were detected at all time points after MPTP discontinuance. The role of these GSH changes in the pathophysiological cascade of the MPTP-induced neuronal death has yet to be elucidated. GSH depletion by means of buthionine sulfoximine reportedly potentiates MPTP/MPP ϩ toxicity to nigral neurons (69,70); in addition, MPTP is known to induce an early increase in vivo in the release of glial GSH, which undergoes hydrolysis in the extracellular compartment (56). This mechanism might account in part for the early MPTP-induced decrease in GSH levels. GSH early depletion, however, may also be related to MPTP bioactivation in astrocytes, a process that generates O 2 .
(3). It is quite likely that, in such a case, astrocytic SOD (71) and GSH did act in a concerted effort (58) to eliminate nonphysiologically generated radicals; in addition, it is most likely that astrocytic ascorbic acid did not participate, because 1 day after MPTP discontinuance the ascorbic acid oxidative status had even decreased. However, because MPTP and MPP ϩ accumulate neither in the striatum nor in the brainstem, both being cleared off within 8 -24 h (24), the mechanism of the late decrease in GSH levels seems to be unrelated to the aforesaid MPTP effects. According to Martenson and Meister (72), one of the important physiological roles of GSH in the central nervous system is to maintain ascorbic acid in its reduced state, because ascorbic acid may have reductive functions that are not efficiently performed by GSH. GSH-dependent reduction of DHAA to ascorbic acid is catalyzed by mammalian glutaredoxin and protein-disulfide isomerase (73). In addition, DHAA may be directly reduced to ascorbic acid by GSH (74). Therefore, GSH-dependent reduction of DHAA to ascorbic acid may be one of the mechanisms underlying late GSH depletion. The above hypothesis is supported by the finding that GSH depletion in newborn rats results in decreases in brain ascorbic acid levels and great increases in DHAA levels (72). The findings of a MPTP-induced decrease in GSH levels, coupled with increases in DHAA and glutamate levels, also suggest a possible MPTP disrupting effect on the concerted roles of GSH and glutamate in maintaining neuronal ascorbic acid levels. The mammalian brain does not synthesize ascorbic acid, which is supplied from the periphery. Ascorbic acid exists primarily in its reduced form, because DHAA can be readily reduced by enzymatic mechanisms involving semi-DHAA-and DHAA reductase (75), GSH-dependent glutaredoxin, and protein-disulfide isomerase (73). A homeostatic mechanism, demonstrated in vivo, keeps striatal extracellular ascorbic acid concentrations (0.2-0.4 mM) constant (76). Extracellular ascorbic acid is actively transported into neurons by means of Na ϩdependent transporters (77), to maintain the estimated concentration of 10 mM. Glia do not have detectable levels of SVTC2, therefore the source of glial ascorbic acid is still uncertain. Glial ascorbic acid is released into the extracellular compartment by means of the glutamate-ascorbate hetero-exchange mechanism, demonstrated in vivo by Miele et al. (78). Accordingly, glial glutamate uptake results in glial ascorbic acid release into the extracellular compartment. In turn, glutamate taken up by glia may act as a substrate for the synthesis of GSH, which, in turn, ensures glial DHAA reduction back to ascorbic acid by GSH-dependent glutaredoxin and protein-disulfide isomerase (19,73). Extra-glial DHAA (either released from neurons, or arising from extracellular ascorbic acid oxidation, or both) may be taken up by glia by means of glucose transporters, which are efficient transporters of DHAA (79). MPTP-induced increases in DHAA and glutamate, coupled with decreases in GSH levels, might indicate an impairment of the above recycling mechanism, which is considered an important brain neuroprotective mechanism (80). In the present study, the late decrease in ascorbic acid levels (Ϫ13% at day 11, and Ϫ19% at day 14), which presumably occurs in neurons, may denote an impairment of those mechanisms ensuring the estimated 10 mM neuronal ascorbic acid concentration.
A glutamatergic mechanism is claimed to be involved in MPTP-induced neurotoxicity (81). NO-and ROS-mediated oxidative stress has long been implicated in the pathogenesis of both acute and chronic N-methyl-D-aspartic acid-mediated neurotoxic effects of glutamate (70). In the present study, high tissue levels of glutamate were detected at all time points after MPTP discontinuance. We have no direct evidence of activation of N-methyl-D-aspartic acid receptors as a consequence of high tissue levels of glutamate, although it is most likely to occur (70), possibly potentiated by the increased ascorbic acid oxidative status (82); our data allow us to speculate only about the source of glutamate: an early MPTP-induced increase in the release/hydrolysis of glial GSH (56) and a late oxidative-stress mediated inhibition of glial uptake (81).
In conclusion, the results of the present study suggest a sequential linkage of cellular and neurochemical events in MPTP-induced nigral and striatal neurotoxic effects. Early impairment of striatal dopaminergic functioning is transient and unrelated to nigral cell loss. Late nigral cell loss appears to be related to energy failure and seems to recognize interconnected mechanisms in which glia play key mediating roles: NO-and ROS-induced oxidative stress, impairment of cellular antioxidant defense mechanisms, excitoxicity, i.e. all those mechanisms currently associated with neuronal death in Parkinson's disease.