Microglial activation-mediated delayed and progressive degeneration of rat nigral dopaminergic neurons: relevance to Parkinson's disease

Authors

  • Hui-Ming Gao,

    1. Neuropharmacology Section, Laboratory of Pharmacology and Chemistry, National Institute of Environmental Health Sciences/National Institutes of Health, North Carolina, USA
    2. Dalian Medical University, Department of Physiology, Dalian, China
    Search for more papers by this author
  • Janwei Jiang,

    1. Neuropharmacology Section, Laboratory of Pharmacology and Chemistry, National Institute of Environmental Health Sciences/National Institutes of Health, North Carolina, USA
    Search for more papers by this author
  • Belinda Wilson,

    1. Neuropharmacology Section, Laboratory of Pharmacology and Chemistry, National Institute of Environmental Health Sciences/National Institutes of Health, North Carolina, USA
    Search for more papers by this author
  • Wanqin Zhang,

    1. Dalian Medical University, Department of Physiology, Dalian, China
    Search for more papers by this author
  • Jau-Shyong Hong,

    1. Neuropharmacology Section, Laboratory of Pharmacology and Chemistry, National Institute of Environmental Health Sciences/National Institutes of Health, North Carolina, USA
    Search for more papers by this author
  • Bin Liu

    1. Neuropharmacology Section, Laboratory of Pharmacology and Chemistry, National Institute of Environmental Health Sciences/National Institutes of Health, North Carolina, USA
    Search for more papers by this author

Address correspondence and reprint requests to Dr B. Liu, F1-01, PO Box 12233, Research Triangle Park, NC 27709, USA. E-mail: liu3@niehs.nih.gov

Abstract

The etiology of sporadic Parkinson's disease (PD) remains unknown. Increasing evidence has suggested a role for inflammation in the brain in the pathogenesis of PD. However, it has not been clearly demonstrated whether microglial activation, the most integral part of the brain inflammatory process, will result in a delayed and progressive degeneration of dopaminergic neurons in substantia nigra, a hallmark of PD. We report here that chronic infusion of an inflammagen lipopolysaccharide at 5 ng/h for 2 weeks into rat brain triggered a rapid activation of microglia that reached a plateau in 2 weeks, followed by a delayed and gradual loss of nigral dopaminergic neurons that began at between 4 and 6 weeks and reached 70% by 10 weeks. Further investigation of the underlying mechanism of action of microglia-mediated neurotoxicity using rat mesencephalic neuron-glia cultures demonstrated that low concentrations of lipopolysaccharide (0.1–10 ng/mL)-induced microglial activation and production of neurotoxic factors preceded the progressive and selective degeneration of dopaminergic neurons. Among the factors produced by activated microglia, the NADPH oxidase-mediated release of superoxide appeared to be a predominant effector of neurodegeneration, consistent with the notion that dopaminergic neurons are particularly vulnerable to oxidative insults. This is the first report that microglial activation induced by chronic exposure to inflammagen was capable of inducing a delayed and selective degeneration of nigral dopaminergic neurons and that microglia-originated free radicals play a pivotal role in dopaminergic neurotoxicity in this inflammation-mediated model of PD.

Abbreviations used:
DA

dopamine

DAB

3,3′-diaminobenzidine

DPI

diphenylene iodonium

FBS

fetal bovine serum

GAD

glutamic acid decarboxylase

GFAP

glial fibrillary acidic protein

GSH

glutathione

HBSS

Hank's balanced salt solution

HS

horse serum

IL-1β

interleukin-1β

LPS

lipopolysaccharide

MAP-2

microfilament-associated protein-2

MEM

minimum essential medium

l-NAME

NG-nitro-l-arginine methyl ester

NO

nitric oxide, PD, Parkinson's disease

ROS

reactive oxygen species

SN

substantia nigra

SOD

superoxide dismutase

TH-IR

tyrosine hydroxylase-immunoreactive

TNF-α

tumor necrosis factor-α

VΤΑ

ventral tegmental area.

Parkinson's disease (PD) is characterized by a progressive and selective degeneration of dopaminergic neurons of the substantia nigra (SN) pars compacta, leading to disorders in movements (Olanow and Tatton 1999). Mutations in several recently identified genes such as α-synuclein and parkin have been correlated with mostly early on-set and familial PD that however, only represents a very small fraction of the incidence of the disease (Polymeropoulos 1998; de Silva et al. 2000). The exact cause for the vast majority and idiopathic PD remains unknown.

Inflammation in the brain, in particular, activation of microglia, has been increasingly associated with the pathogenesis of PD, as well as several other neurodegenerative disorders including Alzheimer's disease, multiple sclerosis, and the AIDS dementia complex (McGeer et al. 1988; Dickson et al. 1993; Raine 1994; Rogers and Shen 2000). Microglia, the resident immune cells in the brain, play a role of immune surveillance under normal conditions. However, microglia become readily activated in response to injuries in the brain and immunological challenges (Kreutzberg 1996; Aloisi 1999; Gonzalez-Scarano and Baltuch 1999; Streit 2000). Activated microglia produce a wide array of cytotoxic factors, including tumor necrosis factor-α (TNF-α), interleukin-1β (IL-1β), eicosanoids, nitric oxide (NO), and reactive oxygen species (ROS), that impact on neurons to induce neurodegeneration (Minghetti and Levi 1998; Hirsch 2000; Liu et al. 2002). Evidence of microglial activation has mostly derived from post mortem analysis of the SN of PD patients which appears to suggest that microglia are mostly involved in the late stage of the pathogenetic process of PD (Vawter et al. 1996; Cassarino et al. 1997; Banati et al. 1998; Hunot et al. 1999). In contrast, epidemiological studies and case reports strongly implicate a correlation between antecedent traumatic brain injury, life episodes of encephalitis, or intrauterine brain inflammation and later-life development of PD (Ravenholt and Foege 1982; Factor et al. 1988; Mattock et al. 1988; Williams et al. 1991; Casals et al. 1998; Plassman et al. 2000). Combined with the observation that the SN region of the brain is particularly enriched in microglia (Lawson et al. 1990; Kim et al. 2000), it is therefore logical to hypothesize that microglial activation plays an important role in the initiation and early progression stage of the degeneration of nigral dopaminergic neurons.

To test the hypothesis, we chronically infused an inflammagen, lipopolysaccharide (LPS) at 5 ng/h for 2 weeks into an area right above the SN of rat brain. The temporal relationship between the activation of microglia and degeneration of nigral dopaminergic neurons was determined. The mechanism of microglial activation-mediated dopaminergic neurodegeneration was further studied in primary mesencephalic neuron-glia cultures following stimulation with low concentrations of LPS (0.1–10 ng/mL). We report here that LPS induced a rapid activation of microglia followed by a delayed, progressive and selective destruction of nigral dopaminergic neurons both in vivo and in vitro. The NADPH oxidase-mediated release of superoxide free radical from activated microglia appeared to be a major contributor to the degeneration of dopaminergic neurons.

Materials and methods

Animal studies

All animals were treated in strict accordance with the National Institutes of Health (Bethesda, MD, USA) Guide for Humane Care and Use of Laboratory Animal. All efforts were made to minimize the number of animals and their suffering. Adult male Fischer 344 rats (220–250 g, Charles River, Raleigh, NC, USA) were anesthetized with sodium pentobarbital (50 mg/kg) and positioned in a small-animal stereotaxic apparatus. For infusion of LPS or vehicle into an area above the SN pars compacta, a hole in the skull was drilled using the following coordinates: 4.8 mm posterior to the bregma and 1.7 mm lateral to the midline (Paxinos and Watson 1986). Inserted at 8.0 mm ventral to the surface of skull was a cannula (30 gauge, Plastics One, Roanoke, VA, USA) connected, via polyethylene tubing, to a Model 2002 osmotic minipump (0.5 µL/h, 200 µL total capacity, Alza Corp., Palo Alto, CA, USA) preloaded with LPS (Escherichia coli 0111:B4, Sigma, St Louis, MO, USA) dissolved in phosphate-buffered saline (PBS) or PBS alone as the control. The cannula was secured with the aid of small screws and dental cement, and the minipump was implanted under the skin on the back of the animal. After the surgery, the wounds were sutured. At desired time points, rats were anesthetized with sodium pentobarbital and then transcardially perfused with saline (0.85% NaCl) followed by ice-cold 4% paraformaldehyde in PBS as previously described (Liu et al. 2000a). The brains were removed, post-fixed for 2 days at 4°C in 4% paraformaldehyde in PBS, and cryoprotected for 2–3 days at 4°C in 30% sucrose in PBS. Coronal sections (35 µm in thickness) were cut with a microtome through the nigral complex and stored in 0.05% sodium azide in PBS. The delivery of the content of the retrieved minipump was confirmed by measuring the remaining volume of fluid in the chamber of the pump.

Rat ventral mesencephalic neuron-glia cultures

Primary mixed neuron-glia cultures were prepared following our previously published protocol (Gao et al. 2002). In brief, dispersed cells from ventral mesencephalic tissues of embryonic day 13/14 Fischer 344 rats were seeded to 24-well (5 × 105/well) or 96-well (105/well) culture plates precoated with poly-d-lysine. The cultures were maintained at 37°C in a humidified atmosphere of 5% CO2 and 95% air in minimum essential medium (MEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS) and 10% heat-inactivated horse serum (HS), 1 g/L glucose, 2 mm l-glutamine, 1 mm sodium pyruvate, 100 µm non-essential amino acids, 50 U/mL penicillin, and 50 µg/mL streptomycin. For treatment, 7-day-old cultures were switched to MEM containing 2% FBS, 2% HS, 2 mm l-glutamine, 1 mm sodium pyruvate, 50 U/mL penicillin, and 50 µg/mL streptomycin. Immunocytochemical analysis with cell-type specific antibodies (see below) indicated that, at the time of treatment, cultures were made up of approximately 11% microglia, 48% astrocytes, and 41% neurons of which 2.3–3.5% were tyrosine hydroxylase-immunoreactive (TH-IR) neurons. Cultures were either treated with vehicle or LPS or pretreated with various agents prior to treatment with LPS as specified in the legends to figures.

Microglia-enriched cultures

Primary microglia were prepared from whole brains of 1-day-old Fisher 344 rats as previously described (Liu et al. 2001b). The enriched microglia were greater than 98% pure as determined by immunostaining for microglia- and astrocyte-specific markers (see below). For superoxide assays, microglia (5 × 104/well) were seeded in 96-well culture plates and maintained overnight in DMEM supplemented with 10% FBS before use.

Immunohistochemistry and immunocytochemistry

Paraformaldehyde-fixed floating brain sections or formaldehyde-fixed cell cultures were immunostained following our previously published protocols (Liu et al. 2000a; Gao et al. 2002). Microglia were stained with a monoclonal anti-CR3 complement receptor antibody (OX-42, 5 µg/mL, BD Pharmingen, San Diego, CA, USA). Astrocytes were identified with anti-glial fibrillary acidic protein (GFAP, 1 : 1000) antiserum from Dako (Carpinteria, CA, USA). Neurons were recognized with a monoclonal antibody against the microfilament-associated protein-2 (MAP-2, 1 : 400, BD Pharmingen) to detect both perikarya and neurites and a monoclonal antibody against a neuron-specific nuclear protein (Neu-N, 1 : 2000, Chemicon, Temecula, CA, USA) to detect neuronal nuclei. Dopaminergic neurons were detected with anti-tyrosine hydroxylase antiserum (1 : 20 000, a generous gift from GlaxoWellcome, RTP, NC, USA). GABAergic neurons were identified with a monoclonal anti-glutamic acid decarboxylase (GAD) antibody (1 : 1000, Chemicon). Briefly, brain sections or cell cultures were blocked with appropriate normal serum followed by incubation overnight at 4°C with a primary antibody diluted in antibody diluent (Dako). After incubation with an appropriate biotinylated secondary antibody and then the Vectastain ABC reagents (Vector Laboratory, Burlingame, CA, USA), the bound complex was visualized by color development with 3,3′-diaminobenzidine (DAB). For double label immunocytochemical staining, cultures were first stained with the anti-TH antibody using DAB as a chromophore (brown) followed with the anti-Neu-N antibody using DAB as a chromophore and nickel sulfate as an intensifying agent (dark blue). Images were recorded with a Zeiss upright or a Nikon inverted microscope connected to a charge-coupled device camera (DAGE-MTI, Michigan City, IN, USA) operated through the MetaMorph software (Universal Image and Co., West Chester, PA, USA).

Nissl staining of brain sections

Brain sections were dehydrated by processing first through increasing and then decreasing concentrations of ethanol (50, 70, 85, 95, 100%, 1 min each). After rinsing in water, the sections were stained for 30 s with 0.1% cresyl violet in 2.5% acetic acid. After extensive rinsing in water followed by dehydration in increasing concentrating of ethanol and finally xylene, the sections were mounted and sealed under coverslips.

Confocal double-label immunofluorescence

Brain sections were first stained with the anti-Neu-N antibody (1 : 2000) followed by the Alexa-488 conjugated goat anti-mouse secondary antibody (1 : 1000, Molecular Probes, Eugene, OR, USA) and then stained with the anti-TH antibody (1 : 20 000) followed by the Alexa-546 conjugated goat anti-rabbit secondary antibody (1 : 1000). After mounting the sections onto glass slides with the Prolong Antifade reagents (Molecular Probes), fluorescent images were obtained with a Zeiss LSM 510 NLO laser scanning confocal microscope fitted with an Argon ion laser (488 nm) and a HeNe laser (543 nm) and recorded with the Zeiss LSM510 software.

Counting of immunostained neurons and measurement of neuronal dendrites

As described previously (Liu et al. 2000a), every effort was made to ensure even cutting of the brain sections. Brains with unevenly sliced sections were excluded. Following immunostaining, of the 24 sections covering a range of the nigral complex approximately 4.5–5.4 mm posterior to bregma (rostral to cordial) for each animal, 12 evenly spaced sections (i.e. every other one) were selected for visual counting of the number of nigral TH-IR neurons (Liu et al. 2000a). A second and immediately adjacent set of equally spaced sections from the same animal was selected to count the nigral Neu-N-IR neurons. Sets of brain sections from four to 10 animals in each treatment group were used. Counting was always performed by 2–3 individuals in a blind manner. To count the number of IR-neurons in cell cultures under a microscope (40 × power), 10 evenly distributed areas in each well of the 24-well plate were selected (Gao et al. 2002). For comparison, sister culture wells on the same plate that received identical treatment were immunostained with different cell-type specific antibodies (TH-, Neu-N-, or GAD-IR). For each experiment, 2–4 wells/treatment condition were used for cell counting and results from three independent experiments were obtained. The overall dendrite length for individual TH-IR neurons in the cultures was measured as previously described (Liu et al. 2001a). For each well of cell cultures, 25–50 TH-IR neurons were selected for measurement and three wells for each condition were used. Measurements from three separate experiments were obtained.

Uptake assays for dopamine (DA), GABA or seretonin (5-HT)

Uptake assays were performed as described (Gao et al. 2002). Briefly, after rinsing with warm Krebs–Ringer buffer (KRB, 16 mm sodium phosphate, 119 mm NaCl, 4.7 mm KCl, 1.8 mm CaCl2, 1.2 mm MgSO4, 1.3 mm EDTA, and 5.6 mm glucose; pH 7.4), cultures were incubated for 15 min at 37°C with 1 µm[3H]DA (30 Ci/mmol, NEN, Boston, MA, USA), 5 µm[3H]GABA (90 Ci/mmol, NEN), or 0.1 µm[3H]5-HT (10 Ci/mmol, Amersham Pharmacia, Piscataway, NJ, USA) in KRB. After washing (3 ×) with ice-cold KRB, cells were solublized in 1 m NaOH and radioactivity was counted with a liquid scintillation counter. Non-specific uptake for DA, GABA or 5-HT, observed in the presence of mazindol (10 µm), NO-711 (10 µm), or fluoxetine (10 µm) were subtracted.

Nitrite and TNF-α assays

The production of NO was determined by measuring the accumulated levels of nitrite in the supernatant with the Griess reagent (detection limit: 0.5 µm) and release of TNF-α was measured with a rat TNF-α enzyme-linked immunosorbent assay kit (detection limit: 5 pg/mL) from R & D Systems (Minneapolis, MN, USA) as described (Gao et al. 2002).

Superoxide assay

The production of superoxide was determined by measuring the SOD-inhibitable reduction of cytochrome c as described (Gao et al. 2002). Mesencephalic neuron-glia or microglia-enriched cultures grown in 96-well plates were treated with LPS or vehicle in 120 µL of phenol red-free MEM or Hank's balanced salt solution (HBSS), containing 2% FBS, 2% HS, and with or without 600 U/mL superoxide dismutase (SOD). To each well, 80 µL of ferricytochrome c (100 µm) in treatment medium was added. The cultures were then incubated for 60 min at 37°C. Afterwards, the absorbance was read at 550 nm with a SpectraMax Plus microplate spectrophotometer (Molecular Devices, Sunnyvale, CA, USA). The amount of SOD-inhibitable superoxide was calculated using a molar extinction coefficient of 2.11 × 104m/cm for cytochrome c at 550 nm (Massy 1959) and was expressed as nmol/106 cells.

Statistical analysis

Data are expressed as the mean ± SEM. Statistical significance was assessed with an analysis of variance (anova) followed by Bonferroni's t-test using the StatView program (Abacus Concepts, Inc., Berkeley, CA, USA). A value of p < 0.05 was considered statistically significant.

Results

LPS infusion induces a delayed, progressive, and preferential loss of rat nigral dopaminergic neurons

To establish the temporal relationship between microglial activation and dopaminergic neurodegeneration, an inflammagen, LPS, was unilaterally infused, for 2 weeks at 5 ng/h, into an area right above the SN pars compact of the rat brain. Afterwards, brains were harvested at 2, 4, 6, 8 or 10 weeks, coronal sections were cut through the nigral complex, and the extent of neuronal loss was determined. As shown in Fig. 1, infusion of LPS induced a delayed yet progressive loss of nigral TH-IR neurons. Comparison of the number of nigral TH-IR neurons between the LPS-infused and the non-infused control side indicated that no significant loss of TH-IR neurons was detected at 2 or 4 weeks after the initiation of LPS infusion (Fig. 1). Significant loss of TH-IR neurons started to manifest between 4 and 6 weeks after the start of LPS infusion and at the 6-week time point, a 39% loss was observed. The LPS-induced loss of nigral TH-IR neurons further progressed over time and a 58% and 69% loss was observed at the 8- and 10-week time point (Fig. 1). Quantification of the number of Neu-N-IR neurons in the SN revealed a 14% and 29% decrease in Neu-N-IR neurons at the 8- and 10-week time points, respectively (Fig. 1).

Figure 1.

Delayed and time-dependent loss of nigral dopaminergic neurons following LPS infusion. LPS was infused, with an osmotic minipump, for the first 2 weeks at a rate of 5 ng/ h (0.5 µL/h) into a region right above the SN pars compacta of one side of the rat brain. Afterwards, rats were killed at indicated time points and brain sections were cut through the nigral complex. After immunostaining, the number of TH- or Neu-N-IR neurons in the SN (covering a range of 4.5–5.4 mm posterior to bregma, rostral to cordial) was counted as described in Materials and methods. The number of animals used for the 2-, 4-, 6-, 8-, and 10-week time point is 4, 4, 6, 10, and 5, respectively. Results are expressed as percentage of the non-infused side contralateral to LPS infusion. **p < 0.005 compared with the controls.

The LPS-induced degeneration of nigral dopaminergic neurons was further characterized by confocal double-label (TH and Neu-N) immunofluorescent analysis. At 10 weeks after the start of LPS infusion, in the pars compact of the SN ipsilateral to LPS infusion, the number of neurons labeled for both cytoplasmic TH (red) and nuclear Neu-N (green) was much fewer than that in the corresponding region of the non-infused side (Figs 2a and b). Furthermore, compared with the control side, the intactness of the intricate network of the TH-IR fibers on the LPS-infused side, especially that in the pars compacta, appeared to be significantly affected (Fig. 2b). In addition, the LPS-induced neurodegeneration appeared to be preferential to TH-IR neurons since non-dopaminergic neurons (green nuclear staining only) in the SN pars compacta and areas above and below were not significantly affected (Figs 2a and b). The LPS-induced destruction of nigral neurons was confirmed by staining for the Nissl substance that helps highlight the dense band of cells in the pars compact of SN. As shown in Fig. 2(c), compared with the control side, a significant loss of the Nissl-positive neurons was observed in the pars compact of SN ipsilateral to LPS-infusion (10 week). As a control, little damage to nigral neurons was observed in the SN infused with vehicle PBS alone when compared with the non-infused side (data not shown). On the other hand, the LPS-induced destruction of nigral TH-IR neurons was not paralleled by significant damage to TH-IR neurons in the adjacent ventral tegmental area (VTA) either when LPS was infused into an area right above the middle of pars compact (Fig. 2a) or at the junction between SN and VTA (Fig. 2d). These results demonstrated that LPS-infusion resulted in a preferential destruction of nigral dopaminergic neurons.

Figure 2.

Characterization of the LPS-induced loss of rat nigral dopaminergic neurons. (a) Representative confocal double label immunofluorescent analysis of a coronal brain section from a rat 10 weeks after the start of LPS infusion. Neurons in general were labeled with an antibody against a nuclear protein (Neu-N) that is expressed in the nuclei of neurons and a green fluorescence-tagged secondary antibody. Dopaminergic neurons were labeled with an antibody against a specific cytoplasmic marker (TH) and a red fluorescence-tagged secondary antibody. In the pars compact region of the SN ipsilateral to LPS infusion, the number of neurons with both the cytoplasmic TH immunoreactivity (red) and nuclear Neu-N immunoreactivity (green) were significantly fewer than that in the non-infused side. The bright green glow is a non-specific staining seen near the edge of the needle track. (b) Magnified photomicrographs of confocal images from the boxed areas (defined by solid white lined rectangles, a and b) in Fig. 1 (a) . Arrows indicate the double-labeled neurons. Again note a substantial decrease in the number of double-labeled neurons in the LPS-infused side. (c) Representative Nissl staining of nigral neurons of a brain section from a rat 10 weeks after LPS infusion. Note the significant disappearance of Nissl body-containing neurons in the pars compact (defined by dashed ovals) on the LPS-infused side of the brain. (d) Representative immunohistochemical analysis for TH-IR neurons in the SN and VTA from a rat 10 weeks after the infusion of LPS to the junction between SN and VTA. SNPc, SN pars compacta. VTA, ventral tagmental area. Arrow, needle tract. Bars in (a), (c) and (d), 200 µm; and in (b), 50 µm. LPS to the junction between SN and VTA. SNPc, SN pars compacta. VTA, ventral tagmental area. Arrow, needle tract. Bars in (a), (c) and (d), 200 µm; and in (b), 50 µm.

LPS infusion-induced microglial activation in SN precedes nigral dopaminergic neurodegeneration

Activation of microglia is characterized by dramatic changes in morphology: from the resting ramified, to the partially activated rod-like, and to the fully activated amoeboid microglia (Kreutzberg 1996). In addition, microglial activation leads to an increased expression of surface molecules such as the complement CR3 receptor and immunostaining of this receptor has been widely used as a marker for microglial activation. To determine the time course of microglial activation following LPS infusion, brain sections obtained from rats at 3 days, 1, 2, 4, or 8 weeks after LPS infusion, were immunostained with the OX-42 antibody that recognizes rat CR3 receptor. As shown in Fig. 3, LPS-infusion caused the nigral OX-42-IR microglia to undergo dramatic morphological changes, indicative of their status of activation. In the SN contralateral to LPS infusion, the OX-42-IR microglia exhibited the typical ramified morphology of resting microglia. However, 3 days (the earliest time point examined) after the start of LPS infusion (5 ng/h), microglia in the SN ipsilateral to LPS-infusion had significantly enlarged cell bodies and increased OX-42-immunoreactivity with a sizable portion metamorphosed into the partially activated rod-like microglia. At one, especially 2 weeks after the start of LPS infusion, nearly all OX-42-IR microglia were converted to the amoeboid form, indicative of a maximal degree of activation (Fig. 3). At 4 and 8 weeks after LPS infusion, microglia appeared to remain at the stage of fully activated, although their morphology, especially of those at the 8-week time point, seemed to be different from that at the 2-week time point (Fig. 3). As a control, in the SN infused with vehicle PBS, no significant activation of OX-42-IR microglia was observed, except for the very limited area along the needle tracks. Since significant degeneration of nigral dopaminergic neurons was not detected at least 4–6 weeks after the start of LPS infusion (Fig. 1) whereas a full degree of microglial activation was achieved within 1–2 weeks (Fig. 3), LPS-induced microglial activation preceded dopaminergic neurodegeneration.

Figure 3.

LPS infusion-induced activation of microglia in the rat SN. Following the start of LPS infusion (5 ng/h), rats were killed at the indicated time points. Brain sections were obtained and immunostained with the antibody OX-42 to determine the activation status of microglia as described in the Materials and methods. In the non-LPS infused SN (control), microglia exhibited the typical ramified morphology of resting microglia. Upon stimulation with LPS, microglia progressed from being partially activated (rod-like shape, 3 day) to being fully activated (amoeboid form, 2 weeks). At longer time points (4–8 week), microglia remained to be highly activated. Bar: 25 µm.

LPS treatment induces a selective and time-dependent degeneration of dopaminergic neurons in mesencephalic neuron-glia cultures

The underlying mechanism of action responsible for the microglial activation-mediated degeneration of nigral dopaminergic neurons was further studied using the ventral mesencephalic mixed neuron-glia cultures. This ‘long-term’ (up to 12 days of treatment) in vitro culture system, coupled with stimulation with low concentrations of LPS (0.1–10 ng/mL), allowed for the study of progressive dopaminergic neurodegeneration (Gao et al. 2002) that could not be achieved with our previous ‘short-term’ cultures (2–3 days of treatment) treated with concentrations of LPS as high as 1000 ng/mL (Kim et al. 2000; Liu et al. 2000b). Neuron-glia cultures were treated for 2–10 days with vehicle or 100 pg/mL to 10 ng/mL LPS and the extent and selectivity of the degeneration of dopaminergic neurons were assessed by multiple parameters. Uptake assays for [3H]DA demonstrated that LPS treatment resulted in a concentration- and time-dependent decrease in DA uptake (Fig. 4a). Time course study revealed that the lower the concentrations of LPS, the longer it took to induce a significant reduction in DA uptake: from 1 ng/mL at day 6, to 0.3 ng/mL at day 8, and to 0.1 ng/mL at day 10 (Fig. 4a). Comparison of the neuron-glia cultures for DA, GABA, or 5-HT uptake following treatment for 10 days with vehicle or 0.1–10 ng/mL LPS showed that only at 3 and 10 ng/mL LPS, a modest decrease (15–20%) in GABA and 5-HT uptake was observed, compared with a dose-dependent and significantly more dramatic drop in DA uptake (Fig. 4b).

Figure 4.

LPS-induced progressive and selective reduction in DA uptake in mesencephalic neuron-glia cultures. (a) Time course and LPS-concentration dependence for the effect of LPS on DA uptake. Cultures were treated with vehicle or 0.1–10 ng/mL LPS for 2–10 days and then assayed for [3H]DA uptake. The results are expressed as percentage of control and are the mean ± SEM of 3–6 experiments performed in triplicate. (b) Effect of LPS on DA, GABA, or 5-HT uptake. Neuron-glia cultures were treated for 10 days with vehicle alone or 0.1–10 ng/mL LPS and then assayed for [3H]DA, [3H]GABA, or [3H]5-HT uptake. The results are expressed as percentage of control and are the mean ± SEM of four experiments performed in triplicate. The amount of DA, GABA and 5-HT uptake by control cultures at 2 days post-treatment was 0.74 ± 0.03, 43.00 ± 3.00, and 0.40 ± 0.10 and was 0.55 ± 0.07, 39.33 ± 4.06 and 0.35 ± 0.07 pmol/min/well at 10 days post-treatment, respectively (n = 3). *p < 0.05, **p < 0.005 compared with the control cultures.

Immunocytochemical analysis demonstrated that LPS (1 or 10 ng/mL; 10 days)-induced degeneration of dopaminergic neurons involved a significant loss of TH-IR perikaya and destruction of TH-IR dendrites, while no significant damage was observed for MAP-2-IR neurons (Fig. 5a). Even the survived TH-IR neurons in the LPS-treated (10 ng/mL) cultures exhibited much shrunken cell body, rough perimeter, and very short dendrites, in sharp contrast to the appearance of healthy TH-IR neurons in the vehicle-treated control cultures (Fig. 5a). Quantitation of the number of immunostained neurons indicated that LPS treatment (0.3–10 ng/mL; 10 days) did not result in any significant loss of GAD-IR neurons, a marker for GABAergic neurons, or Neu-N-IR neurons, a marker for neurons in general (Fig. 5b). In contrast, the number of TH-IR neurons which represent only a small fraction (∼3%) of the Neu-N-IR neurons decreased by 20, 42, and 51% following treatment with 1, 3, and 10 ng/mL LPS, respectively (Fig. 5b). More dramatically, the average dendrite length of the TH-IR neurons in cultures treated for 10 days with 1 or 10 ng/mL LPS reduced by 50 and 75%, respectively (Fig. 5b).

Figure 5.

Immunocytochemical analysis and quantitation of LPS-induced neurotoxicity in neuron-glia cultures. (a) Immunostaining. Neuron-glia cultures treated for 10 days with vehicle control, 1, or 10 ng/mL LPS were double labeled for cytoplasmic TH (brown) and for nuclear Neu-N (blue). Note the disappearance of TH- and Neu-N double stained neurons (arrows) and the shortening of TH-IR dendrite in LPS-treated cultures over the background of non-dopaminergic neurons stained positive for Neu-N only (blue). Alternatively, cultures were single stained with the anti-MAP-2 antibody to detect both the cell body and dendrites. No significant damage to MAP-2-IR neurons was observed in LPS-treated cultures. Bar, 20 µm. (b) Quantitative analysis. The number of TH-IR, Neu-N-IR, or GAD-IR neurons in the control or LPS-treated cultures was counted as described in the Materials and methods. The overall length of the dendrites of individual TH-IR neurons was also measured as described in the Materials and methods. Results are mean ± SEM of three experiments performed in triplicate. *p < 0.05, **p < 0.005 compared with the control cultures.

Microglial activation and release of neurotoxic factors precede dopaminergic neurodegeneration in LPS-treated neuron-glia cultures

Consistent with the rapid activation of microglia observed in the SN following LPS infusion, as early as 1 day after LPS stimulation, microglia in the neuron-glia cultures exhibited significant morphological changes and increased surface expression of CR3 receptor, indicative of LPS-induced activation. As shown in Fig. 6, significant activation of OX-42-IR microglia was observed in cultures stimulated with the lowest concentration of LPS used (0.1 ng/mL). In contrast, significant neurodegeneration, as judged by reduction in DA uptake, was not detected until 10 days after treatment with 0.1 ng/mL LPS (Fig. 4a). The degree of microglial activation increased with increasing concentrations of LPS (Fig. 6).

Figure 6.

LPS-concentration-dependent activation of microglia in the neuron-glia cultures. Cultures were treated for 1 day with the indicated concentrations of LPS and immunostained with the OX-42 antibody. In control cultures, a large portion of the OX-42-IR microglia was round and small. Following stimulation with LPS (0.1–10 ng/mL), microglia became activated with the characteristics of larger in size, irregular in shape and increased immunoreactivity. Images presented are representative of at least three independent experiments. Bar: 50 µm.

More importantly, analysis of the time courses for the release of neurotoxic factors from activated microglia demonstrated that significant production of those factors was detected in the first 24 h after LPS stimulation. For instance, significant production of NO was observed as early as 1 day after stimulation with 1–10 ng/mL LPS and sustained accumulation was evident 6 days later (Fig. 7a). The release of TNF-α was even faster than that for NO. Six hours after stimulation with 3–10 ng/mL LPS, maximal release of TNF-α was detected (Fig. 7b). The levels of nitrite and TNF-α in cultures treated with 0.1–0.3 ng/mL and 0.1–1 ng/mL LPS, respectively, were not significantly different from that of the vehicle-treated control cultures (Figs 7a and b). However, robust production of superoxide, measured as SOD-inhibitable reduction of cytochrome c, was detected in cultures treated with the entire range of concentrations of LPS used in this study (0.1–10 ng/mL). Stimulation of cultures for 60 min with 0.1, 0.3, and 1 ng/mL LPS increased superoxide production to 189%, 200%, and 232% of the control level, respectively. Significantly elevated levels of superoxide production were detected at later time points. For example, the levels of superoxide production in cultures treated with 0.1, 1.0, and 10 ng/mL LPS were 1.4-, 1.8-, and 2.0- and 1.5-, 1.9-, and 2.1-fold over vehicle-treated control cultures at 1 and 6 days after stimulation with LPS.

Figure 7.

LPS-stimulated production of TNF-α, NO and superoxide in rat mesencephalic neuron-glia cultures. (a) and (b) Cultures were treated with vehicle alone or 0.1–10 ng/mL LPS. At indicated time point, levels of nitrite (an indicator of NO production) and TNF-α in the culture medium were measured. (c) Superoxide production, measured as SOD-inhibitable cytochrome c reduction, was determined in cultures treated for 60 min with LPS. The results are the mean ± SEM of four experiments performed in triplicate. *p < 0.05, **p < 0.005 compared with the control cultures.

NADPH oxidase-mediated generation of superoxide by activated microglia is a major contributor to LPS-induced neurotoxicity

The unique pattern for the LPS concentration-dependent production of neurotoxic factors prompted us to speculate that ROS play a major role in neurodegeneration. In cultures treated for 10 days with 1–10 ng/mL LPS which stimulated the production of both NO and superoxide (Figs 7a and c), significant neuroprotection was achieved by either inhibition of NO production by 1 mm NG-nitro-l-arginine methyl ester (l-NAME), or neutralization of the reactivity of superoxide by inclusion of 100 U/mL each of SOD/catalase or 2.5 mm glutathione (GSH) (Fig. 8a). In contrast, in cultures treated for 12 days with 0.1–0.3 ng/mL LPS which did not induce significant production of NO but did result in a robust generation of superoxide (Figs 7a and c), SOD/catalase and GSH, but not l-NAME, afforded significant neuroprotection (Fig. 8b).

Figure 8.

Effect of l-NAME, GSH or SOD/catalase on LPS-induced degeneration of dopaminergic neurons. Neuron-glia cultures were pretreated for 30 min with l-NAME (1 mm) or cotreated with GSH (2.5 mm) or SOD/catalase (100 U/mL each) followed with the vehicle or 1–10 ng/mL LPS for 10 days (a) or 0.1–1 ng/mL LPS for 12 days (b). Cultures were then assayed for DA uptake. The results are the mean ± SEM of four experiments performed in triplicate. *p < 0.05 compared with the LPS-treated cultures.

One of the major sources of superoxide generation in activated microglia, as well as macrophages and neutrophils, is through the membrane-associated NADPH oxidase (Babior 1999). To examine the involvement of microglial NADPH oxidase in LPS-induced neurotoxicity, the LPS-stimulated superoxide generation was subjected to inhibition by two functionally dissimilar inhibitors of NADPH oxidase, apocynin and diphenylene iodonium (DPI) with the former preventing the assembly of the enzyme complex and the latter directly inhibiting its catalytic activity (Stolk et al. 1994; Babior 1999). Pretreatment of microglia-enriched cultures with 5 µm DPI or 0.25 mm apocynin prior to stimulation with LPS (0.1–10 ng/mL) significantly reduced the LPS-induced superoxide generation (Fig. 9). Furthermore, inhibition of microglial NADPH oxidase afforded neuroprotection in neuron-glia cultures. The LPS-induced decrease in DA uptake, loss of TH-IR neurons, and shortening of TH-IR dendrites were significantly attenuated by preincubation of cultures for 30 min with 0.25 mm apocynin prior to treatment for 10 days with 0.1–10 ng/mL LPS (Fig. 10). Interestingly, treatment of cultures for 10 days with DPI (0.1–5 µm) alone was toxic to the cells as previously reported (Gao et al. 2002).

Figure 9.

Effect of NADPH oxidase inhibitors on the LPS-induced release of superoxide in microglia. Microglia-enriched cultures were pretreated for 30 min with vehicle control, DPI (5 µm), or apocynin (0.25 mm) prior to stimulation with indicated concentrations of LPS and the production of superoxide was determined. Results are expressed as percentage of control and are mean ± SEM of three experiments performed in triplicate. +p < 0.05, ++p < 0.005 compared with the control cultures, *p < 0.05, **p < 0.005 compared with the correspondingly LPS-treated cultures.

Figure 10.

Protective effect of apocynin against LPS-induced neurotoxicity. Neuron-glia cultures were pretreated for 30 min with the vehicle control or apocynin (0.25 mm) followed with the vehicle or 1–10 ng/mL LPS for 10 days. Neurotoxicity was determined by assaying for DA uptake (a) or counting of the number of TH-IR neurons (b). The results are the mean ± SEM of three experiments performed in triplicate. *p < 0.05, **p < 0.005 compared with the LPS-treated cultures. (c) Representative images of TH-IR neurons in the neuron-glia cultures. Note that apocynin significantly attenuated LPS-induced loss of TH-IR neurons and shortening of TH-IR dendrite. Bar: 50 µm.

Discussion

Inflammation in the brain has been increasingly associated with the pathogenesis of several neurodegenerative disorders including PD, Alzheimer's disease, multiple sclerosis, and the AIDS dementia complex (McGeer et al. 1988; Dickson et al. 1993; Raine 1994; Rogers and Shen 2000). The hallmark of neuro-inflammation is the activation of glial cells, especially that of microglia, in response to neuronal injury or immunological challenges (Kreutzberg 1996; Aloisi 1999; Hauss-Wegrzyniak et al. 1998; Hirsch 2000; Streit 2000). Post mortem analysis of the brains of PD patients indicated the activation of microglia and accumulation of proinflammatory cytokines and free radicals (Vawter et al. 1996; Cassarino et al. 1997; Banati et al. 1998; Hunot et al. 1999). In the meantime, several lines of evidence appear to lend support to the speculation that inflammation in the brain, in particular, microglial activation, plays a critical role in the earlier stages of the pathogenesis of PD. First, case-control studies have associated antecedent traumatic brain injury with the later development of PD as well as Alzheimer's disease or dementia in general (Factor et al. 1988; Williams et al. 1991; Plassman et al. 2000). Activation of glia, especially microglia, as a result of the initial neuronal death may initiate a cascade of events leading to progressive neurodegeneration (Streit 2000). Second, epidemiological studies have suggested the possibility of developing postencephalitic Parkinsonism, sometimes several decades later, in people exposed to viruses or other infectious agents (Ravenholt and Foege 1982; Casals et al. 1998). Third, intrauterine fetal brain inflammation may play a role in the later development of PD (Mattock et al. 1988). Therefore, inflammation in the brain, with microglial activation as a major component of the process, may be able to trigger the onset of the degeneration of nigral dopaminergic neurons. Hence, it has become especially necessary to examine the temporal and possibly causative relationship between microglial activation and dopaminergic neurodegeneration.

In this study, we induced microglial activation by chronically infusing an inflammagen (LPS) into the rat brain and compared the time courses for microglial activation and degeneration of dopaminergic neurons in the SN. We found that significant microglial activation occurred at a much earlier time point (1–2 weeks, Fig. 3) than the onset of dopaminergic neurodegeneration (4–6 weeks, Fig. 1). The delayed and progressive nature of the neurodegeneration and the possible sparing of dopamine neurons in the VTA share similarities with the pattern of neuronal loss observed in PD patients (Damier et al. 1999). The temporal relationship between microglial activation and neurodegeneration was further analyzed in the in vitro cell culture model (mesencephalic neuron-glia cultures) where production of superoxide, NO, and TNF-α from activated microglia (Fig. 7) preceded the selective degeneration of dopaminergic neurons (Fig. 4). More importantly, studies with enzyme inhibitors and free radical scavengers demonstrated that the NADPH oxidase-mediated production of superoxide free radical from activated microglia appeared to be a major driving force for the induction of dopaminergic neurodegeneration (Figs 9 and 10). Hence, results from both in vivo and in vitro studies demonstrated that, at least in rodents, exposure to an inflammagen which triggers a rapid activation of microglia and production of neurotoxic factors including free radicals, was sufficient to set in motion a series of events in the brain, leading to the eventual and selective destruction of nigral dopaminergic neurons.

Previously, LPS has been delivered through a single injection into the SN (Castano et al. 1998; Herrera et al. 2000; Kim et al. 2000; Liu et al. 2000a; Lu et al. 2000). However, the neurodegeneration induced by a bolus application of microgram quantities of LPS was too fast and/or not specific to dopaminergic neurons. More importantly, the rapid time course did not allow for a readily examination of the temporal and potentially causative relationship between microglial activation and neurodegeneration. In contrast, the delayed and progressive neurodegeneration induced by chronic infusion of nanogram quantities of LPS serves as a valuable tool to study the mechanism of action for the microglial activation-mediated neurodegeneration (this study). In addition, the reported effect of intrauterus exposure of developing fetuses to LPS on the dopaminergic system in adult animals offers an alternative means to study the relationship between inflammation and neurodegeneration (Ling et al. 2002).

The LPS-induced degeneration of dopaminergic neurons observed both in vivo and in vitro in this study is not due to a transient down-regulation of TH-immunoreactivity, but rather the consequence of a complete degradation of the dopaminergic neurons. This notion is supported by several lines of evidences. First, in the in vivo studies, confocal double-label-immunofluorescent analysis demonstrated the LPS-induced disappearance of both the cytoplasmic marker (TH) and nuclear marker (Neu-N, Fig. 2). Second, the loss of nigral TH- and Neu-N-doubly immunoreactive neurons (i.e. dopaminergic neurons) was corroborated by the marked disappearance of Nissl substance-containing neurons in the pars compact of the SN ipsilateral to LPS infusion (Fig. 2). Third, double label immunocytochemical analysis of neuron-glia cultures showed that LPS-induced destruction of the intricate dendrite networks and the neuronal perikaya (Fig. 5).

Microglia play a role of immune surveillance in the normal brain and become readily activated in response to neuronal injuries and immunological challenges (Kreutzberg 1996; Streit 2000). Historically, a large body of knowledge on the characteristics of microglia has been obtained through the studies of neuronal injury-induced microglial activation. It is just beginning to explore the relationship between early occurring microglial activation and late onset neurodegeneration in relation to several neurodegenerative diseases. The demonstration of microglial activation-mediated dopaminergic neurodegeneration presented in this study does not exclude the involvement of neuronal death-induced additional microglial activation (McMillian et al. 1994). Rather, the microglial activation at the 8-week time point (Fig. 3) certainly included elements of microglial activation induced by dying/dead neurons killed in the first place by toxic factors released from LPS-activated microglia. By the same token, neuronal death-induced secondary and sustained microglial activation would exacerbate the neurodegeneration and this may actually represent the scenario occurring in the later stage of the progression of neurodegenerative diseases.

Activated microglia produce a variety of proinflammatory and neurotoxic factors, including cytokines such as TNF-α and IL-1β, fatty acids and metabolites such as eicosanoids, and the free radicals NO and superoxide (Bronstein et al. 1995; Minghetti and Levi 1998; Liu et al. 2002), which work in concert to induce neuronal death (Chao et al. 1992; Dawson et al. 1994; Jeohn et al. 1998). Among the factors produced by activated microglia, oxygen free radicals may play a prominent role in the degeneration of dopaminergic neurons. Neutralization of the reactivity of ROS with either SOD/catalase or GSH (Liu et al. 1998) afforded significant neuroprotection (Fig. 8). The NADPH-oxidase-mediated release of superoxide from activated microglia may be a main component of ROS responsible for the induction of inflammation mediated neurodegeneration (Figs 9 and 10). Oxygen free radicals are highly reactive and can react with proteins, DNA, or RNA to alter their functions, or induce lipid peroxidation, leading to eventual neuronal death (Farber 1994). In addition, more potent intermediates such as peroxynitrite can be formed from reactive nitrosative and oxidative species (Beckman et al. 1993). Dopaminergic neurons in the SN are known to be uniquely vulnerable to oxidative insults (Jenner and Olanow 1998; Greenamyre et al. 1999; Zhang et al. 2000). Because the midbrain region has the highest density of microglia in the brain (Lawson et al. 1990; Kim et al. 2000), dopaminergic neurons in the SN may encounter an excessively high level of oxidative stress during brain inflammation. The combination of these two factors may be partially responsible for the selective degeneration of dopaminergic neurons during the progression of PD.

In addition to microglia, astrocytes may play a role in neurodegeneraton. Astrocytes can secrete neurotrophic factors to help the survival of neurons (Aloisi 1999). However, in terms of the capacity to induce neurodegeneration, astrocytes may play a smaller role than microglia. First, the repertoire and quantities of the proinflammatory and neurotoxic factors produced by activated astrocytes are somewhat limited. For example, the amount of nitric oxide produced by LPS-activated astrocytes is significantly lower than that by microglia (Liu et al. 2002). Second, Kim et al. (2000) reported a positive correlation between the abundance of microglia in various brain regions and the sensitivity of neurons to LPS-induced degeneration.

PD is most likely a consequence of interactions among genetic predisposition, environmental factors and the unique properties of the SN dopaminergic neurons (Olanow and Tatton 1999; Kidd 2000). The present study is the first to present evidence showing inflammagen-triggered microglial activation results in a delayed and progressive degeneration of dopaminergic neurons in the SN. Further exploration of the relationship between early stage inflammation in the brain and later development of PD would be of tremendous importance to the understanding of the etiology as well as pathogenesis of PD.

Acknowledgements

The authors thank Dr Jerome L. Maderdrut for reading the manuscript and Jeffrey Reece for assistance with the confocal microscope.

Ancillary