Inflammation induces mitochondrial dysfunction and dopaminergic neurodegeneration in the nigrostriatal system


Address correspondence and reprint requests to Guoying Bing, Department of Anatomy and Neurobiology, 310 Davis Mills Building, University of Kentucky, Chandler Medical Center, 800 Rose Street, Lexington, KY 40536-0298, USA. E-mail:


Evidence suggests that chronic inflammation, mitochondrial dysfunction, and oxidative stress play significant and perhaps synergistic roles in Parkinson’s disease (PD), where the primary pathology is significant loss of the dopaminergic neurons in the substantia nigra. The use of anti-inflammatory drugs for PD treatment has been proposed, and inhibition of cyclo-oxygenase-2 (COX-2) or activation of peroxisome proliferator-activated receptor gamma (PPAR-γ) yields neuroprotection in MPTP-induced PD. Lipopolysaccharide (LPS) induces inflammation-driven dopaminergic neurodegeneration. We tested the hypothesis that celecoxib (Celebrex, COX-2 inhibitor) or pioglitazone (Actos, PPAR-γ agonist) will reduce the LPS-induced inflammatory response, spare mitochondrial bioenergetics, and improve nigral dopaminergic neuronal survival. Rats were treated with vehicle, celecoxib, or pioglitazone and were intrastriatally injected with LPS. Inflammation, mitochondrial dysfunction, oxidative stress, decreased dopamine, and nigral dopaminergic neuronal loss were observed post-LPS. Celecoxib and pioglitazone provided neuroprotective properties by decreasing inflammation and restoring mitochondrial function. Pioglitazone also attenuated oxidative stress and partially restored striatal dopamine as well as demonstrated dopaminergic neuroprotection and reduced nigral microglial activation. In summary, intrastriatal LPS served as a model for inflammation-induced dopaminergic neurodegeneration, anti-inflammatory drugs provided protective properties, and pioglitazone or celecoxib may have therapeutic potential for the treatment of neuro-inflammation and PD.

Abbreviations used

bovine serum albumin




dopamine and cAMP-regulated phosphoprotein


3,4-dihydroxyphenylacetic acid




homovanillic acid


inducible nitric oxide synthase


insulin receptor beta






phosphate-buffered saline


Parkinson’s disease


peroxisome proliferator-activated receptor gamma


respiratory control ratio


substantia nigra


tyrosine hydroxylase


ventral tegmental area

Parkinson’s disease (PD) is a neurological disorder that affects approximately 1–3% of the population (Bennett et al. 1996; Lang and Lozano 1998). The primary pathology is significant loss of the dopaminergic neurons in the substantia nigra (SN), which leads to a significant loss of striatal dopamine. Other pathological hallmarks include chronic inflammation, oxidative stress, mitochondrial or proteasomal dysfunction, and the presence of Lewy bodies (McGeer et al. 1988; Forno 1996; Hirsch et al. 1997; Jenner 1998; Greenamyre et al. 1999; Olanow and Tatton 1999; Vila et al. 2001).

Various PD models demonstrate inflammation in neurodegeneration, and anti-inflammatory drugs attenuate toxin-induced PD (He et al. 2001; Gao et al. 2002; Vijitruth et al. 2006). A high correlation of post-encephalitis with PD (Casals et al. 1998; Berger 2003) and a report of Parkinsonism induced by accidental exposure to lipopolysaccharide (LPS) from Salmonella minnesota (Niehaus 2004) also support the role of inflammation in the etiology of PD. Intranigral and intrapallidal LPS induce microglial activation and dopaminergic neuronal death (Bing et al. 1998; Castano et al. 1998; Liu et al. 2000; Lu et al. 2000; Iravani et al. 2002; Arimoto and Bing 2003; Zhang et al. 2005; Arimoto et al. 2006). Microglial activation initiates or perpetuates neuronal loss by increasing cytotoxic molecules like superoxide, NO, various pro-inflammatory cytokines, and prostaglandins (Banati et al. 1993; Minghetti and Levi 1998; Arimoto and Bing 2003; Kim et al. 2004). To improve on the current models of LPS-induced dopaminergic neurodegeneration and to elucidate potential mechanisms involved in this degeneration, we injected LPS into the striatum. This was to avoid local mechanical trauma to the SN and because degeneration maybe initiated at the terminals leading to retrograde damage (Greenamyre et al. 1999). We tested the effect of LPS on nigrostriatal mitochondrial function, because to our knowledge mitochondrial dysfunction had not been demonstrated in vivo with LPS. Mitochondrial damage implies a direct cause and effect relationship between inflammation and dopaminergic neuronal death, and is important because mitochondria are the cell’s ‘power source.’ LPS-induced mitochondrial dysfunction has been demonstrated in vitro (Xie et al. 2004), where mitochondrial dysfunction precedes cell death (Welty-Wolf et al. 1996; Crouser et al. 2002). We hypothesized that LPS would induce inflammation-mediated mitochondrial dysfunction with subsequent dopaminergic neuronal cell death. We also tested the hypothesis that celecoxib, an exogenous inhibitor of cyclo-oxygenase-2 (COX-2) activity, or pioglitazone, an agonist of peroxisome proliferator-activated receptor gamma (PPAR-γ), will reduce the LPS-induced microglial activation, inhibit the excessive production and release of cytotoxic molecules, improve mitochondrial function, and increase dopaminergic neuronal survival. This is because epidemiological evidence shows that anti-inflammatory drugs reduce the risk of PD (Chen et al. 2003), and studies using COX-2 knockout mice, COX-2 inhibitors, or agonists of PPAR-γ demonstrate partial neuroprotection in the MPTP model of PD (Teismann and Ferger 2001; Breidert et al. 2002; Feng et al. 2002; Dehmer et al. 2004; Vijitruth et al. 2006).

Materials and methods


Three-month-old male Sprague-Dawley rats were obtained from Harlan (Indianapolis, IN, USA) and were housed under a 12-h light–dark cycle with free access to food and water in the Division of Lab Animal Resources at the University of Kentucky. Experimental protocols involving the animals were in strict accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee at the University of Kentucky.

Treatments and injections

Rats were randomly selected and grouped for a study to determine an optimal dose of celecoxib and an effective dose of S. minnesota LPS (Sigma-Aldrich, St Louis, MO, USA). Briefly, rats were anesthetized with sodium pentobarbital (Abbott Laboratories, North Chicago, IL, USA) and were positioned in a stereotaxic apparatus. Eight small openings were created in the skull using a dental trephine. The stereotaxic coordinates, measured in millimeters, from Bregma were anterior/posterior +1.0, medial/lateral ±2.0 and ±3.5, and dorsal/ventral −5.5 and −6.0 as well as anterior/posterior −0.5, medial/lateral ±2.5 and ±4.0, and dorsal/ventral −5.0 and −6.5 (Paxinos and Watson 1998). Next, either saline or one of three different LPS solutions was injected into each site using a 30 ga 10 μL Hamilton (Reno, NV, USA) syringe (totaling 16, 32 or 60 μg LPS). The rate of injection was 0.4 μL per minute and the needle was kept in place for 5 min post-injection before slow withdrawal. Following surgery, the rats were kept on a heating pad and 10 mL/kg of subcutaneous sterile saline was given to aid in post-operative recovery. Saline injections continued until the rats became hydrated and ambulatory. A curved syringe was used to inject Nutri-Cal (EVSCO, Buena, NJ, USA) into the corner of their mouth to make sure that the animals did not suffer from nutritional withdraw. The rats were cleaned twice a day for the first few days following injections. For the main study, rats were randomly selected, grouped, and pre-treated for 4 days via oral gavage, with 0.3 mL of vehicle, (pure olive oil; Kroger, Lexington, KY, USA; twice daily), the COX-2-specific inhibitor celecoxib in vehicle suspension (Celebrex; Pfizer Inc., New York, NY, USA; 10 mg/kg twice daily as clinically dosed), or the PPAR-γ agonist pioglitazone in vehicle suspension (Actos; Takeda Chemical Industries, Osaka, Japan, 20 mg/kg once daily as clinically dosed). On the fifth day, the rats were injected, as previously mentioned, with 2 μL/site of 0.9% sterile saline or 2 μL/site of LPS solution made from 2 μg LPS per 1 μL sterile saline (totaling 32 μg or 48 000 endotoxin units of LPS). Following surgery, the rats resumed their respective treatments and were assessed for weight loss and changes in body condition until sacrificing 3 days later. The animals used for cell counts were treated under the same regimen with the exception that they resumed their post-treatments for 7 days before sacrificing.

Tissue collection (immunocytochemistry, western blot, and HPLC)

Animals were anesthetized with CO2, killed by decapitation, and the brains were rapidly removed. Using a rat brain matrix, the lower half of the brain, containing the SN was collected and immediately fixed in 4% paraformaldehyde, was stored at 4°C for 3 days, and was transferred to a 30% sucrose 1× phosphate-buffered saline (PBS) solution stored at 4°C until the tissue was used for sectioning and immunocytochemistry. Whole left-side striatum was dissected out, snap frozen, placed into a pre-weighed Eppendorf tube, weighed, and stored at −80°C until assayed by HPLC. Right-side striatum was dissected, snap frozen, placed into an Eppendorf tube, and was stored at −80°C until assayed by western blot analysis.

Western blotting

Western blot analysis as previously described (Arimoto and Bing 2003) with a few modifications was used to analyze the striatal tissue levels of COX-1, COX-2, the inducible nitric oxide synthase (iNOS), and the insulin receptor (IRβ) beta subunit in vehicle/saline (n = 3), vehicle/LPS (n = 4), celecoxib/LPS (n = 5), and pioglitazone/LPS (n = 4) treated/injected rats. All samples were aliquoted and normalized to the desired concentration and the diluted protein solution was added to a new tube along with 3× loading buffer and 5% BME whenever necessary. The samples were boiled, cooled, loaded into an appropriate gel and run at 80 V. The Kaleidoscope pre-stained marker from BioRad (Hercules, CA, USA) was used as the molecular weight marker. Gels were transferred to polyvinylidene difluoride or nitrocellulose at 60 V for 2 h or 20 V overnight. The membrane was washed before incubation in blocking buffer. Then the membrane was washed in 1× wash buffer before incubation in the primary antibody diluted in blocking solution (anti-β-actin – 1 : 4000, Sigma; anti-COX-1 – 1 : 500, Calbiochem (San Diego, CA, USA); anti-COX-2 – 1 : 1000, Cell Signaling (Danvers, MA, USA); anti-iNOS – 1 : 500, Upstate (Charlottesville, VA, USA); and anti-insulin receptor-β (C-19) – 1 : 500, Santa Cruz (Santa Cruz, CA, USA)). Next, the membrane was washed and incubated in the appropriate secondary antibody solution (1 : 2000 in 5% milk blocking buffer) for 1 h. After incubation, the membrane was washed before using the chemiluminescent substrate solution ECL Plus kit and Hyperfilm (Amersham Biosciences, Little Chalfont Buckinghamshire, UK) was used for the chemiluminescent detection. Signal specificity was insured by omitting each primary antibody in a separate blot, and loading errors were corrected by measuring β-actin immunoreactive bands in the same membrane. The density measurement of each band was performed by a densitometer system (Image Quant version 5.1, Molecular Dynamics Inc., Emeryville, CA, USA). Background samples from an equivalent area near each lane were subtracted from each band to calculate mean band density.

Mitochondria isolation and respiration

This method has been previously described with some minor modifications (Sullivan et al. 2000, 2003). All procedures were performed on ice throughout the protocol. Briefly, the brains were rapidly dissected out and the striatum and nigral regions were isolated quickly and carefully using a rat brain matrix. Three whole rat nigra had to be pooled to make one nigral mitochondrial sample while a single whole striatum was used for one striatal sample. All the tissues were immediately homogenized using glass dounce homogenizer with isolation buffer containing 215 mmol/L mannitol, 75 mmol/L sucrose, 0.1% bovine serum albumin (BSA), 1 mmol/L EGTA and 20 mmol/L HEPES). The pH was adjusted to 7.2 (both the stock HEPES and the buffer itself with KOH). Next, the mitochondria were isolated using differential centrifugation. The homogenate was spun in an Eppendorf microcentrifuge at 4°C for 5 min at 1300 g and the supernatant was transferred into a new tube. The loose pellet was resuspended in an isolation buffer with EGTA and was spun again at 1300 g for 5 min. The resulting supernatant was transferred to a new microcentrifuge tube, topped off with the isolation buffer containing EGTA, and was spun at 13 000 g for 10 min. The supernatant was discarded and the pellet was resuspended in 500 μL of isolation buffer with EGTA. A nitrogen cell disruption bomb (model 4639; Parr Instrument Company Moline, IL, USA) cooled to 4°C was used to burst the synaptoneurosomes as the resuspended mitochondrial samples were placed inside at 1000 psi for 5 min (Brown et al. 2004). Using isolation buffer with EGTA, the samples obtained post-cell disruption were brought to a final volume of 2 mL and were centrifuged at 13 000 g for 10 min. The pellet was resuspended in isolation buffer without EGTA (75 mmol/L sucrose, 215 mmol/L mannitol, 0.1% BSA, and 20 mmol/L HEPES with the pH adjusted to 7.2 using KOH) and was centrifuged at 10 000 g for 10 min. The final mitochondrial pellet was resuspended in isolation buffer without EGTA to yield a final concentration of approximately 10 mg/mL, which was immediately stored on ice. To normalize the results, the protein concentrations were determined with all the samples on the same plate using the BCA protein assay kit and measuring absorbance at 560 nm with a Biotek Synergy HT plate reader (Winooski, VT, USA). Aliquots from each sample were maintained at −80°C.

Respiration activity of the isolated mitochondria was measured using a Clark-type oxygen electrode (Hansatech Instruments, Norfolk, UK), and approximately 100–150 μg of isolated striatal or nigral mitochondrial protein were suspended and constantly stirred in a sealed and thermostatically controlled chamber at 37°C in respiration buffer (215 mmol/L mannitol, 75 mmol/L sucrose, 0.1% BSA, 20 mmol/L HEPES, 2 mmol/L MgCl, 2.5 mmol/L KH2PO4 at pH 7.2). The rate of oxygen consumption was determined based on the response slope of the isolated mitochondria to oxidative substrates as previously described (Sullivan et al. 2003; Brown et al. 2004). State II respiration was initiated by the addition of 5 mmol/L pyruvate and 2.5 mmol/L malate. State III respiration was initiated by the addition of 150 μmol/L ADP followed by the addition of oligomycin (1 µg/mL) to inhibit the ATP synthase and induce state IV respiration. The mitochondrial uncoupler carbonyl cyanide 4-trifluoromethoxy phenylhydrazone (1 μmol/L) was added to the chamber to induce maximum NADH-linked state V respiration (complex I driven). Rotenone (1 μmol/L) was added to the chamber to inhibit complex I of the electron transport system. Then, FADH maximum respiration (complex II driven) was assessed by the addition of succinate (10 mmol/L) to the chamber. Three runs were performed for each sample and the average was used for statistical analysis. Respiratory control ratios (RCR) were calculated for NADH-linked substrates as the ratio of state III/state IV oxygen consumption. All rates are expressed as nmoles of O2 consumed per minute per milligram of protein (O2/min/mg). For the statistical analysis, the number of samples used per group was as follows: vehicle/saline (striatum n = 4 and nigra n = 3), vehicle/LPS (striatum n = 4 and nigra n = 4), celecoxib/LPS (striatum n = 4 and nigra n = 4), and pioglitazone/LPS (striatum n = 5 and nigra n = 4) treated/injected rats. This was because, to get a nigral mitochondrial sample of n = 4 we pooled 12 rats (3/sample). For the striatal mitochondrial sample to get n = 5, we used 5 separate rat striata (1/sample).

Slot blot for markers of oxidative stress

Mitochondrial samples were diluted with 0.1 mol/L PBS. For each sample, 10 μL of sample, 10 μL of 12% sodium dodecyl sulfate, and 20 μL of 1× dinitrophenyl hydrazone was added into a new tube. Samples were incubated at 22–24°C, and then neutralizing solution and 0.1 mol/L PBS were added and mixed. The samples were loaded into each well and 0.1 mol/L PBS was loaded into the empty wells. Next, the samples were sucked through onto the nitrocellulose followed by 0.1 mol/L PBS. The membrane was placed into wash blot buffer with 1% casein (Sigma-Aldrich) and was agitated for 2 h at 22–24°C. After rinsing with blot buffer, the blot was agitated overnight at 4°C in primary antibody solution containing 1% casein, wash blot buffer, and the appropriate concentration of primary antibody [Anti-hydroxynonenal (HNE) – 1 : 10 000, Calbiochem; Anti-3-nitrotyrosine (3-NT) – 1 : 2000, Upstate; or Oxyblot Kit – 1 : 100, Chemicon, Temecula, CA, USA). Next, the blot was rinsed with wash blot buffer before addition of the appropriate secondary antibody. The blot was agitated in the secondary antibody for 2 h at 22–24°C. The blot was rinsed with wash blot buffer and was developed in Super Signal 1 : 1 for 3 min and a picture was taken for 600 s using the BioRad set-up. Blots were performed in triplicates and the averages were used for statistical analysis.

HPLC analysis of striatal tissue levels of dopamine and metabolites

Tissue levels of dopamine and its primary metabolites 3,4-dihydroxyphenylacetic acid (DOPAC) and homovanillic acid (HVA) were measured using HPLC as previously described (Cass et al. 2003). Retention times of standards were used to identify peaks, and peak heights were used to calculate recovery of the internal standard (dihydroxybenzylamine) and the amount of dopamine and metabolites. Results are expressed as ng/g wet weight of tissue from vehicle/saline (n = 8), vehicle/LPS (n = 5), celecoxib/LPS (n = 5), and pioglitazone/LPS (n = 4) treated/injected rats.


Every sixth section throughout the entire nigra was processed for immunocytochemical detection (n = 4–5 per group) of the dopaminergic neuronal marker tyrosine hydroxylase (TH; 1 : 2000, Pel Freez, Rogers, AR, USA) or the activated microglia marker OX-6 (1 : 1000, Serotec, Raleigh, NC, USA) using a sensitive avidin–biotin complex (ABC) peroxidase method previously described (Lu et al. 2000). Representative SN and striatal sections were also immunostained with a marker for GABAergic neurons using the dopamine and cAMP-regulated phosphoprotein (DARP-32; 1 : 1000, Chemicon). Briefly, free-floating brain sections were washed in 1× PBS and blocked for 1 h. After blocking, they were incubated overnight at 4°C with the primary antibody solution. Next, they were washed in wash buffer and incubated with the appropriate biotinylated secondary antibody (1 : 1000) for 1 h. Sections were rinsed in 1× PBS and incubated for 1 h with the ABC kit (Vector Laboratories Inc., Burlingame, CA, USA). Peroxidase activity in the tissue sections was visualized using 0.05% diaminobenzidine (Sigma-Aldrich) as a substrate. All sections were mounted on glass slides, dehydrated and coverslipped using Permount (Fisher Scientific, Fair Lawn, NJ, USA). The stained sections were used to determine the extent of dopaminergic neuronal cell damage or microglial activation. Using the optical fractionator method of the Bioquant System, an estimate of the total number of SN dopaminergic neurons or activated microglia was calculated by visually counting the number of nigral TH-positive neuronal cells with clearly visualized cell bodies or OX-6-positive microglial cells. The outlines of the SN were determined in the TH-stained sections by the distribution of the dopaminergic neurons and by referencing well-established landmarks (Hagg and Varon 1993). The number of neurons in sterile saline-injected rats was used to calculate the percentage of surviving neurons in the LPS-injected rats. The number of OX-6-positive microglia determined the amount of activated microglia. Nissl staining of adjacent SN sections was used to determine actual cell loss. These sections were mounted on gelatinized slides, stained with cresyl violet, dehydrated in ascending alcohol concentrations, and coverslipped using Permount.

Statistical analysis

Data are expressed as means ± SEM. Tests of variance homogeneity, normality, and distribution were performed to ensure that the assumptions required for standard parametric analysis of variance were satisfied. The Systat 10 software (SPSS Inc., Chicago, IL, USA) was used to perform statistical analyses using anova followed by a protected least significant differences post hoc test with statistical significance set at p < 0.05.


Intrastriatal LPS induces dopaminergic neurodegeneration in a dose-dependent manner

A celecoxib dose–response study using 1, 10 and 30 mg/kg of celecoxib was performed in rats receiving 60 μg total LPS. We showed that LPS induces 61% TH-positive cell loss and that 10 mg/kg of celecoxib, given b.i.d., significantly attenuates this cell loss (Fig. 1a). A LPS dose study was performed because of the poor condition and high mortality of these animals. Various concentrations of LPS (16, 32, and 60 μg total) induced significant dopaminergic cell loss (13%, 21%, and 61%, respectively) 7 days post-LPS injections (Fig. 1b). As the 32 μg LPS dose did not have a high mortality rate, we used it and the 10 mg/kg celecoxib dose for the next set of experiments.

Figure 1.

 Stereological cell counts for TH-positive cells, 7 days post-LPS, where n = 3–5 per group, except for the LPS/celecoxib 1 mg/kg group. (a) A dose response for celecoxib reveals that 10 mg/kg twice daily attenuates the LPS-induced cell loss. (b) Various concentrations of intrastriatal LPS (16, 32 and 60 μg) induce a significant decrease in the number of TH-positive cells in the SN (∼13%, 21%, and 61%). Data are expressed as means ± SEM (* and **p ≤ 0.05; *** and ****p ≤ 0.01), where * or *** represents LPS versus saline injection, ** represents vehicle versus 10 mg/kg celecoxib treatment, and **** represents the LPS dose response.

Changes in pro-inflammatory proteins and insulin receptor expression following LPS were partially restored with celecoxib or pioglitazone

Lipopolysaccharide-induced microglial activation is demonstrated by the increase in OX-6-positive cells, in the striatum, 3 days after LPS (Fig. 2a). Western blot analysis, of the striatal tissue, was used to assess the intrastriatal LPS-induced inflammatory response by measuring the levels of COX-1, COX-2, and iNOS protein expression (Fig. 2). LPS significantly increased the levels of COX-2 (Fig. 2b) and iNOS (Fig. 2c) while it had no significant effect on COX-1 (Fig. 2b). Celecoxib or pioglitazone reduced the increase in COX-2 (Fig. 2b). Celecoxib also reduced the increase in iNOS and pioglitazone showed a definite trend (p = 0.058) (Fig. 2c). As LPS is commonly used in the periphery to induce insulin desensitization (Ueki et al. 2004) and as pioglitazone is an insulin sensitizer used to treat type II diabetes mellitus (Smith 2001), we measured IRβ to determine whether insulin signaling was altered by LPS injection. LPS significantly decreased striatal IRβ, and treatment with celecoxib or pioglitazone partially restored the LPS-induced loss of IRβ (Fig. 2d).

Figure 2.

 Striatal OX-6-positive immunocytochemistry and western blot analysis, 3 days post-LPS, where n = 3–5 per group. (a) Intrastriatal LPS induces striatal microglial activation, characterized by OX-6-positive activated microglia. (b) Intrastriatal LPS leads to a significant increase in striatal COX-2 and (c) iNOS expression, and co-treatment with celecoxib or pioglitazone inhibits this increased expression. (d) LPS significantly decreased IRβ expression and treatment with celecoxib or pioglitazone partially inhibited this decreased expression. (e) Representative western blots for COX-2, COX-1, iNOS, and IRβ. Data are expressed as means ± SEM (**p ≤ 0.01), where ** represents vehicle/LPS versus vehicle/saline, celecoxib/LPS, or pioglitazone/LPS.

Detrimental changes in mitochondrial respiration induced by LPS were partially restored with celecoxib or pioglitazone

Mitochondria from the striatum and nigra were assessed for respiratory function 3 days after intrastriatal LPS injection. Respiration experiments utilizing striatal mitochondria demonstrate the negative effects of LPS on mitochondrial respiration, where LPS caused a significant decrease in state III and state V respirations, when using different oxidative substrates for complex-I- or complex-II-driven respiration. Celecoxib or pioglitazone partially restored the LPS-induced decreases in striatal mitochondrial respiration (Fig. 3a). The nigral mitochondrial respiration experiments demonstrated the negative effects of LPS on mitochondrial complex I and complex II function; however, they only showed a trend (p = 0.06) for decreased state III respiration. Treatment with celecoxib or pioglitazone partially restored the decreases in nigral mitochondrial respiration (Fig. 3b). The RCR, where a score of 5 or greater equals a ‘healthy’ well-coupled mitochondria, suggest that inflammation has negative effects on nigrostriatal mitochondrial function and that treatment with celecoxib or pioglitazone partially rescues the mitochondria from the LPS-mediated damage (Fig. 3c).

Figure 3.

 Assessment of mitochondrial function, 3 days post-LPS, where n = 3–5 per group. (a) Intrastriatal LPS significantly decreases states III and V respiration in the striatum driven by mitochondria complex I and II substrates. Treatment with celecoxib or pioglitazone partially restores mitochondrial function. (b) In nigral mitochondria, LPS significantly decreases state V respiration driven by both complex I and II substrates, and treatment with celecoxib or pioglitazone partially restores mitochondrial function. (c) Striatal and nigral respiratory control ratios, where a score of ≥5 represents healthy well-coupled mitochondria, demonstrates LPS-induced mitochondrial damage, which is restored by celecoxib or pioglitazone treatment. Data are expressed as means ± SEM (*p ≤ 0.05 and **p ≤ 0.01), where * or ** represents vehicle/LPS versus vehicle/saline, celcoxib/LPS or pioglitazone/LPS.

LPS-induced mitochondrial oxidative stress was partially decreased with pioglitazone

As a result of the LPS-induced microglial activation and mitochondrial dysfunction, we expect an increase in oxidative stress markers. A slot blot was used to detect the formation of protein carbonyls (Oxyblot), lipid peroxidation using 4-HNE, and tyrosine nitrosylations with 3-nitrotyrosine (3-NT) on the mitochondria (Fig. 4). LPS-induced a significant increase in striatal mitochondria Oxyblot, 4-HNE, and 3-NT levels 3 days after injections. Pioglitazone decreased the increased Oxyblot and 3-NT levels while celecoxib had no effect (Fig. 4a). No significant differences in the level of oxidative stress markers were observed in the nigral mitochondria (Fig. 4b).

Figure 4.

 Slot blot for markers of oxidative stress 3 days post-LPS. (a) The striatal slot blot shows that LPS significantly increases protein carbonyls (Oxyblot), lipid peroxidation (4-HNE), and tyrosine nitrosylations (3-NT) relative to control. Pioglitazone reduces the levels of 3-NT and Oxyblot. (b) No significant difference in the level of oxidative stress markers was observed in the nigral mitochondria. (c) Representative slot blots for striatal and nigral mitochondria: VS represents vehicle/saline; VL represents vehicle/LPS; CL represents celecoxib/LPS; and PL represents pioglitazone/LPS-treated rats. Data are expressed as means ± SEM (* or **p ≤ 0.01), where * represents treatment/LPS versus vehicle/saline injections and ** represents pioglitazone/LPS versus vehicle/LPS or celecoxib/LPS.

Decreased striatal dopamine, induced by LPS, was partially restored by pioglitazone

Dopamine loss in the striatum is an essential pathological feature in PD and it correlates with the loss of SN dopaminergic neurons. HPLC analysis demonstrates that intrastriatal LPS leads to a significant decrease in striatal dopamine (Fig. 5a) and an increase in the turnover ratios of DOPAC or HVA to dopamine (Fig. 5b). No significant difference was observed in the DOPAC or HVA metabolite levels in the vehicle/LPS rats relative to the vehicle/saline control. Pioglitazone partially reduced the LPS-induced striatal dopamine loss, but did not restore the turnover ratios (Fig. 5). Celecoxib had no effect on restoring dopamine loss nor did it restore the turnover ratios (Fig. 5). No significant changes were observed in serotonin or its primary metabolite levels (data not shown).

Figure 5.

 HPLC for dopamine and its metabolites 3 days post-LPS. (a) There was a significant decrease in striatal dopamine following LPS injection and pioglitazone treatment partially restored the striatal dopamine level. (b) LPS also induced a significant increase in the turnover ratio of HVA and DOPAC to dopamine. Data are expressed as means ± SEM (* or **p ≤ 0.01), where * represents treatment/LPS versus vehicle/saline injections and ** represents pioglitazone/LPS versus vehicle/LPS or celecoxib/LPS.

Pioglitazone protects dopaminergic neurons against LPS-neurotoxicity and inhibits microglial activation

Significant SN dopaminergic cell loss underlies as the primary pathology of PD; therefore, we tested the effects of pioglitazone on LPS-induced dopaminergic cell loss. Immunocytochemistry and stereological cell counts of the TH-positive neurons in the SN revealed that intrastriatal LPS leads to a significant decrease in the number of TH-positive cells (∼21%) 7 days post-injection, and that pioglitazone attenuated the TH-positive cell loss (Fig. 6a). Nissl staining of adjacent sections revealed actual cell loss and not a decrease in TH expression, and staining with DARP-32 showed no apparent damage to the GABAergic neurons in the striatum or SN (data not shown).

Figure 6.

 Immunocytochemistry and stereological cell counts for TH and OX-6-positive cells 7 days post-injections. (a) LPS induces a significant decrease in the number of TH-positive cells in the SN (∼21%) and treatment with pioglitazone attenuates the TH-positive cell loss. (b) LPS induces a significant increase in the number of OX-6-positive cells and treatment with pioglitazone attenuates the increase in OX-6 positive cells. The small figures below the OX-6 stain is a 40X view of microglial activation. Data are expressed as means ± SEM (* and **p ≤ 0.01), where * represents vehicle/LPS versus vehicle/saline or pioglitazone/LPS, and ** represents vehicle/saline versus pioglitazone/LPS.

Perpetuation of the inflammatory response and the vicious cycle, which ultimately leads to cell death, implies that microglial activation will be exacerbated in the SN and may play a role in driving the progression of SN dopaminergic neuronal cell loss. Therefore, adjacent SN sections were immunostained with OX-6, a marker of microglial activation, and the number of OX-6-positive activated microglia was determined. The activated microglial cell counts revealed that LPS significantly increased SN microglial activation (∼522%) 7 days post-injection, and that pioglitazone attenuated SN microglial activation (Fig. 6b).


The present study demonstrates that intrastriatal LPS induces an inflammatory response characterized by increased striatal COX-2 and iNOS as well as increased OX-6 3 days post-injection. Seven days post-LPS injection, OX-6 was increased within the SN, supporting the progression of the inflammatory response. Significant striatal and nigral mitochondrial complex I and complex II dysfunction along with correlating decreased RCR scores and increased oxidative stress markers were also observed 3 days post-LPS injection. A decrease in striatal dopamine was observed at 3 days, and at 7 days there was a significant decrease in the number of TH-positive neurons in the SN. We report that celecoxib or pioglitazone decreased the striatal inflammatory response and partially restored nigrostriatal mitochondrial function. Pioglitazone also reduced the LPS-induced decrease in striatal dopamine and some markers of oxidative stress. Therefore, we further tested the effects of pioglitazone, and not celecoxib, on the attenuation of the LPS-induced cell loss. Seven days post-LPS injection, pioglitazone attenuated dopaminergic neurodegeneration and SN microglial activation. We speculate that the protective properties are a result of the decreased inflammatory response, the attenuated oxidative stress, and the restored mitochondrial function.

Our findings are consistent with other previous models of LPS-induced dopaminergic cell loss where intranigral or intrapallidal injected rats display increased microglial activation, increased iNOS, COX-2, and pro-inflammatory cytokines, decreased striatal dopamine, and SN dopaminergic cell loss (Bing et al. 1998; Castano et al. 1998; Liu et al. 2000; Lu et al. 2000; Arimoto and Bing 2003; Zhang et al. 2005; Arimoto et al. 2006). It has also been shown that the increased iNOS, following LPS injection is co-localized with the activated microglia (Arimoto and Bing 2003).

Lipopolysaccharide significantly decreased striatal dopamine and not serotonin, which implies that at 3 days the dopaminergic neurons are damaged. The LPS effect on striatal dopamine was previously demonstrated in an intranigral LPS model where only dopamine and not serotonin levels were altered (Herrera et al. 2000; Hsieh et al. 2002).

We observed, for the first time, that intrastriatal LPS induces mitochondrial complex I and complex II dysfunction in the striatum and nigra. This was demonstrated by a decrease in oxygen uptake after the addition of substrates and ADP, which suggests that the striatal mitochondria could not effectively run electron transport post-LPS-induced inflammation. The RCR scores for the striatal mitochondria also support impaired mitochondrial respiration. In the nigral mitochondria, we showed decreased complex I and complex II functions as well as low RCR scores, and we only saw a trend for decreased state III respiration (p = 0.06), probably because of the low n used.

Our data clearly demonstrate that following LPS administration, mitochondrial bioenergetics is significantly impaired in both the striatum and the SN. Presently, it is unclear whether or not the changes we measured in mitochondrial bioenergetics are solely because of changes in dopaminergic neurons, because the mitochondrial samples include mitochondria from all the cell types found in the striatum and SN. This is further confounded by the fact that in the SN, the TH-positive cells account for approximately 20% of the total neurons and that following LPS treatment, we observe approximately 21% loss of the TH-positive cells. These facts and limitations beg to question whether or not the relatively large change in mitochondrial bioenergetics we measured could be due to changes in this subset of TH-positive cells and whether LPS administration is specific for dopaminergic neurons. Mitochondrial respiration reflects the total amount of oxygen consumed by a known amount of mitochondrial protein under various experimental conditions; thus, changes in a subset of the mitochondrial population can be readily assessed, but are only truly representative of the composition found in the sample of isolated mitochondria. In the current set of experiments, it is tempting to relate the loss of TH-positive cells directly to the loss of mitochondria bioenergetics; but as we do not know the exact distribution of mitochondria in the various subtypes of cells found in the striatum and SN, it is unwise to make this assumption. It is also possible that some of the mitochondrial dysfunction measured occurs ex vivo because of increased mitochondrial free-radical production in the population of mitochondria affected by LPS administration, causing oxidative damage to mitochondria not originally damaged in vivo, erroneously amplifying the effect of LPS on mitochondrial function. However, based on previously published work and several other changes demonstrated in the present study, it is reasonable that much of the mitochondrial dysfunction measured may be due to changes in dopaminergic neurons. Our DARP-32 stain revealed no apparent GABAergic neuron loss, and direct intranigral LPS has been shown to only damage dopaminergic and not GABA or serotonergic neurons (Herrera et al. 2000). We also demonstrated alterations in striatal dopamine and not serotonin after LPS, which implies dopaminergic neuronal damage. It is even possible that in many of the TH-positive stained neurons, mitochondrial damage and dysfunction is ongoing and contributes to the loss of mitochondrial bioenergetics measured in this study. Our data are supported by studies showing LPS-induced mitochondrial dysfunction (Crouser et al. 2002; Xie et al. 2004), where cell death was dependent upon mitochondrial dysfunction (Kuwabara and Imajoh-Ohmi 2004).

We also supported a role for oxidative stress in our model, by showing increased protein carbonyls, lipid peroxidation, and tyrosine nitrosylations. Pioglitazone attenuated the increase in protein carbonyls and tyrosine nitrosylations. Although we showed impaired nigral mitochondrial respiration, we failed to show any significant increase in oxidative stress markers. This may be a result of (i) a maximal inflammatory response in the striatum and not in the SN, at the 3-day time interval; therefore, there would be less nigral oxidative stress; (ii) having ventral tegmental area (VTA), neuronal, and glial mitochondria in our nigral sample; thus, we may not see significant oxidative damage, because the VTA may not be damaged in our sample, as it is not damaged in PD, and as the SN has the highest concentration of microglia in the brain (Lawson et al. 1990; Kim et al. 2000); and (iii) synaptosomal mitochondria responding differently than non-synaptosomal mitochondria (Brown et al. 2006). More detailed studies will need to be performed to determine what mitochondria are dysfunctional, a time course for dysfunction, a signal factor that could lead to this retrograde demise, or to determine a reason responsible for the differences in the striatal versus nigral mitochondria.

Consistent with our model of inflammation-mediated neurodegeneration, we also saw that celecoxib, a specific COX-2 inhibitor, had potential neuroprotective properties. Celecoxib attenuated the inflammatory response via reducing the increased COX-2 and iNOS as well as by partially restoring mitochondrial function. Celecoxib attenuated the TH-positive cell loss in our dose study, which supports the hypothesis that attenuation of the inflammatory response provides neuroprotective properties. Studies showing partial neuroprotection in MPTP-induced PD via specific pharmacological inhibition or genetic ablation of COX-2 (Teismann and Ferger 2001; Feng et al. 2002; Schiess 2003; Teismann et al. 2003; Vijitruth et al. 2006) support our data. Pioglitazone attenuated the inflammatory response, restored mitochondrial function, and reduced cell loss like celecoxib. However, it also attenuated the increase in oxidative stress markers as well as reduced the loss of striatal dopamine. Therefore, we speculated that there may be alternate pathways involved in the neuroprotective properties afforded by pioglitazone. Studies showing that pioglitazone provides neuroprotection in MPTP-induced PD (Breidert et al. 2002; Dehmer et al. 2004) support our data.

In an attempt to discover why pioglitazone appeared to offer alternate neuroprotective properties that celecoxib did not, we examined alternative mechanisms inherent within our study, such as our experimental model and the current clinical use of pioglitazone. Peripheral LPS injection is used to study insulin desensitization (Ueki et al. 2004). This is of particular interest, because insulin receptors and dopaminergic neurons are densely represented in the SN (Unger et al. 1991; Lang and Lozano 1998), there is a significant decrease in insulin receptor expression in the SN of Parkinson’s patients (Moroo et al. 1994; Takahashi et al. 1996), and epidemiological evidence shows that 7% of PD patients have type-II diabetes mellitus (Chalmanov and Vurbanova 1987). Support for examining changes related to insulin also comes from a review of various reports that link insulin or hyperinsulinaemia with Alzheimer’s disease (Qiu and Folstein 2006). Pioglitazone is also currently used to treat type-II diabetes mellitus (Smith 2001); therefore, we speculated that some of the alternate protective effects seen with pioglitazone may be a result of its ability to regulate insulin signaling. In our study, LPS induced a significant decrease in IRβ, and models that use LPS to generate insulin desensitization (Ueki et al. 2004) support our finding. However, in our study, we found that both pioglitazone and celecoxib attenuated the LPS-induced decrease in IRβ, which implies that changes in insulin sensitivity may not account for the potentially alternate protective properties afforded by pioglitazone. This also suggests that the decrease in IRβ may be a consequence of dopaminergic neuronal loss, because the insulin receptor is co-expressed with TH-positive neurons (Unger et al. 1991; Lang and Lozano 1998). To make any claims about the role of insulin in LPS-mediated dopaminergic neurodegeneration, future experiments will need to be performed to characterize the changes in insulin-mediated signal transduction.

In summary, we demonstrated a role for inflammation-mediated neurodegeneration, where LPS toxicity was associated with respiratory chain dysfunction. In mitochondria isolated from the striatum, there was evidence of oxidative damage to mitochondrial components, which suggests that mitochondria may be a target of free-radical stress initiated by activated microglia. The fact that both celecoxib and pioglitazone can prevent mitochondrial dysfunction suggests that mitochondrial impairment is secondary to inflammation. This may represent both a pathogenic event and a potential target for treatment, in neuroinflammatory processes. The combination of inflammation, oxidative stress, and mitochondrial dysfunction generates a vicious cycle that appears to lead to progressive dopaminergic neuronal cell death and an exacerbated activation of microglia. Pioglitazone also restored striatal dopamine and decreased oxidative stress markers in addition to attenuating the LPS-induced dopaminergic neuronal cell loss; therefore it may be a useful therapeutic for PD. More extensive studies need to be performed to determine an optimal dose and the mechanisms by which pioglitazone affords its neuroprotective properties.


Dr Jeff Smiley and the Department of lab animal resources, Xuan Nguyen, Rattanavijit Vijitruth, Bin Xing, Laura Peters, Rebecca Rankin, April Claggett, Dale Hunter, and the funding supported by M103KV01001306K22010310 from the Brain Research Center from the 21st Century Frontier Research Program funded by the Ministry of Science and Technology, Republic of Korea (to HCK), AG017963 (to WAC), NS046426, NS048191 (to PGS), and NS044157 (to GYB).