Neurobiology of Disease 71 (2014) 280–291

Contents lists available at ScienceDirect

Neurobiology of Disease

j ourna l homepage: www.e lsev ie r .com/ locate /ynbd i

Dynamic changes in pro- and anti-inflammatory cytokines in microgliaafter PPAR-γ agonist neuroprotective treatment in the MPTPp mousemodel of progressive Parkinson's disease

Augusta Pisanu a, Daniela Lecca b, Giovanna Mulas b, Jadwiga Wardas c, Gabriella Simbula b,Saturnino Spiga d, Anna R. Carta b,⁎a National Research Council, Institute of Neuroscience, via Ospedale 72, 09124 Cagliari, Italyb Department of Biomedical Sciences, University of Cagliari, via Ospedale 72, 09124 Cagliari, Italyc Department of Neuropsychopharmacology, Institute of Pharmacology, Polish Academy of Sciences, 12 Smetna St., 31-343 Krakow, Polandd Department of Life and Environmental Sciences, Via Fiorelli 1, 09126 Cagliari, Italy

⁎ Corresponding author.E-mail addresses: augustapisanu@tiscali.it (A. Pisanu),

giovannamulas2@gmail.com (G. Mulas), wardas@if-pan.ksspiga@unica.it (S. Spiga), acarta@unica.it (A.R. Carta).

Available online on ScienceDirect (www.sciencedir

http://dx.doi.org/10.1016/j.nbd.2014.08.0110969-9961/© 2014 Elsevier Inc. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 27 April 2014Revised 31 July 2014Accepted 6 August 2014Available online 15 August 2014

Keywords:PPAR-γCytokinesMPTPMicrogliaNeuroprotectionParkinson's disease

Neuroinflammatory changes play a pivotal role in the progression of Parkinson's disease (PD) pathogenesis.Recent findings have suggested that activated microglia may polarize similarly to peripheral macrophagesin the central nervous system (CNS), assuming a pro-inflammatory M1 phenotype or the alternative anti-inflammatory M2 phenotype via cytokine production. A skewed M1 activation over M2 has been relatedto disease progression in Alzheimer disease, and modulation of microglia polarization may be a therapeutictarget for neuroprotection. By using the 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine–probenecid(MPTPp) mouse model of progressive PD, we investigated dynamic changes in the production of pro-inflammatory cytokines, such as tumor necrosis factor (TNF)-α and interleukin (IL)-1β, and anti-inflammatory cytokines, such as transforming growth factor (TGF)-β and IL-10, within Iba-1-positivecells in the substantia nigra compacta (SNc). In addition, to further characterize changes in the M2 pheno-type, we measured CD206 in microglia. Moreover, in order to target microglia polarization, we evaluatedthe effect of the peroxisome-proliferator-activated receptor (PPAR)-γ agonist rosiglitazone, which hasbeen shown to exert neuroprotective effects on nigral dopaminergic neurons in PD models, and acts as amodulator of cytokine production and phenotype in peripheral macrophages.Chronic treatment with MPTPp induced a progressive degeneration of SNc neurons. The neurotoxin treatmentwas associated with a gradual increase in both TNF-α and IL-1β colocalization with Iba-1-positive cells, suggest-ing an increase in pro-inflammatory microglia. In contrast, TGF-β colocalization was reduced by the neurotoxintreatment, while IL-10was mostly unchanged. Administration of rosiglitazone during the full duration ofMPTPptreatment reverted both TNF-α and IL-1β colocalization with Iba-1 to control levels. Moreover, rosiglitazoneinduced an increase in TGF-β and IL-10 colocalization comparedwith theMPTPp treatment. CD206was graduallyreduced by the chronic MPTPp treatment, while rosiglitazone restored control levels, suggesting that M2 anti-inflammatory microglia were stimulated and inflammatory microglia were inhibited by the neuroprotectivetreatment. The results show that the dopaminergic degeneration was associated with a gradual microgliapolarization to the inflammatory over the anti-inflammatory phenotype in a chronic mouse model of PD.Neuroprotective treatment with rosiglitazone modulated microglia polarization, boosting the M2 overthe pro-inflammatory phenotype. PPAR-γ agonists may offer a novel approach to neuroprotection, actingas disease-modifying drugs through an immunomodulatory action in the CNS.

© 2014 Elsevier Inc. All rights reserved.

Neuroinflammatory changes, including chronic microgliosis arekey pathological components of Parkinson's disease (PD). Activated

d-lecca@yahoo.it (D. Lecca),rakow.pl (J. Wardas),

ect.com).

microglia and recruited peripheral macrophages (which are antigen-ically not distinguishable from microglia) may promote neurotoxic-ity and play a cardinal role in the progression of neurodegenerationvia the release of neurotoxic species such as pro-inflammatory cyto-kines (Barcia et al., 2004; Hirsch et al., 1998; McGeer et al., 2003; Orret al., 2002; Sawada et al., 2006; Whitton, 2007). Hence, activatedmicroglia/macrophages undergo phenotypic polarization to produceneurotoxic factors, such as inflammatory cytokines, reactive oxygen

281A. Pisanu et al. / Neurobiology of Disease 71 (2014) 280–291

species (ROS), nitric oxide synthase (NOS), and glutamate, namelyM1 pro-inflammatory phenotype, but also a variety of neuroprotec-tive molecules, such as anti-inflammatory cytokines, neurotrophins,and membrane receptors defined as M2 phenotype (Varnum andIkezu, 2012). A prevalence of M1 over M2 microglia/macrophageshas been reported in neurodegenerative pathologies, such asAlzheimer's disease (AD), and in the elderly brain, and microgliapolarization has been suggested to be a target for neuroprotectivetherapies (Hoozemans et al., 2006; Lund et al., 2006; Sanchez-Guajardo et al., 2013). In the PD brain, proliferation of microglia isobserved early in the disease process, remaining relatively staticand unrelated to the extent of striatal degeneration or clinical sever-ity (McGeer et al., 1988; Gerhard et al., 2006). In contrast, microgliapolarization into pro-inflammatory or anti-inflammatory M2 pheno-types has been poorly investigated in PD, although post-mortemstudies measuring cytokine levels have suggested that both pro-and anti-inflammatory microglia may coexist in the Parkinsonianbrain (Rojo et al., 2010). The advanced dopaminergic degenerationin symptomatic PD has been associated with an overproduction ofcytotoxic cytokines, such as tumor necrosis factor (TNF)-α, interleu-kin (IL)-1β, IL-2, IL-4, and IL-6, in the CNS, which is reflected byincreased levels in the serum and ventricular cerebrospinal fluid(CSF), suggesting that microglia may be mainly polarized toward apro-inflammatory phenotype in advanced disease stages (Mogiet al., 1994; Brodacki et al., 2008). Nevertheless, anti-inflammatorycytokines, such as transforming growth factors (TGF)-α and -β, andIL-10, have also been reported in the CNS or serum of PD patients,which may indicate the presence of anti-inflammatory microglia aswell (Mogi et al., 1994; Brodacki et al., 2008; Mogi et al., 1995,1996; Rentzos et al., 2009). Therefore, multiple phenotypical subsetsof activated microglia may coexist in the Parkinsonian brain, drivingthe disease progression via a dynamic alteration of the balancebetween pro- and anti-inflammatory phenotypes and related cyto-kine release. The inflammatory cytokines TNF-α and IL-1β aremajor components of the neuroinflammatory response in PD patho-genesis (Mogi et al., 1994; Sriram et al., 2006; De Lella Ezcurra et al.,2010; Pott Godoy et al., 2008; McCoy and Tansey, 2008; Watsonet al., 2012; Harms et al., 2012). In contrast, cytokines belonging tothe TGF-β family as well as IL-10 are involved in the differentiationand survival of neurons, and have been suggested to exert beneficialand neuroprotective actions against 1-methyl-4-phenylpyridinium(MPP+) toxicity in vitro and in PD experimental models in vivo(Gonzalez-Aparicio et al., 2010; Schober et al., 2007; Krieglsteinet al., 1995; Arimoto et al., 2007; Johnston et al., 2008). Besidescytokines, CD206 is a membrane receptor recognizing carbohydrateligands, that reflects the M2 microglia/macrophages polarizationand is associated with neurorestoration and endocytotic processes(Perego et al., 2011). Given the double-edge nature of activatedmicroglia, defining their polarization along with nigral degener-ation is a pivotal issue in PD, and targeting the impaired pro-inflammatory/anti-inflammatory microglia balance may represent acontributing mechanism to neuroprotection in a disease-modifyingstrategy.

Peroxisome-proliferator-activated receptor (PPAR)-γ agonists havebeen shown to exert a neuroprotective effect in experimental modelsof PDby either preventing or arresting the loss of dopaminergic neuronsin the substantia nigra compacta (SNc) (Carta et al., 2011; Carta andPisanu, 2013; Dehmer et al., 2004; Schintu et al., 2009a). PPAR-γis high-ly expressed in the central as well as peripheral immune cells (Carta,2013; Bernardo et al., 2000; Tontonoz et al., 1998; Varga et al., 1812).Studies on peripheral macrophages have shown that PPAR-γmodulatesthe production of both pro- and anti-inflammatory cytokines by thesecells (Odegaard et al., 2007; Pascual et al., 2007; Satoh et al., 2010;Straus and Glass, 2007). In addition, we have previously shown thatstimulation of PPAR-γ by rosiglitazone inhibits microglial release ofTNF-α in a PD model, and others have reported concordant results in

different neurodegenerative pathologies (Carta et al., 2011; Diab et al.,2004). Moreover, targeting of additional deregulated mechanismsmay contribute to the rosiglitazone-mediated neuroprotective ef-fects, such as reduction of oxidative stress, mitochondrial damageand an anti-apoptotic activity (Carta, 2013). Therefore, we hypothe-sized that PPAR-γ agonists may act as disease-modifying compoundsin the Parkinsonian brain and that modulation of theneuroinflammatory response largely contributes to the neuroprotec-tive effect.

To test this hypothesis, we used the 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine–probenecid (MPTPp) mouse model of progressivePD. This model, consisting of the chronic administration of the neuro-toxin in associationwith the clearance inhibitor probenecid for a periodof 5 weeks, offers several advantages over the classic acute or subacuteMPTPmodels. Firstly, the MPTPpmodel induces a progressive degener-ation of SNc neurons, which is reflected by the gradual appearance ofmotor impairment (Petroske et al., 2001; Schintu et al., 2009b). More-over, dopaminergic degeneration is associated with a gradual increasein microglial reactivity in the SNc, which accompanies the well knownmechanisms of cell loss in this model, such as mitochondrial dysfunc-tion, oxidative stress and apoptotic cell death. Therefore, this model al-lows testing of neuroprotective drugs alongside the neurodegenerativeprocess and at different stages of degeneration. In the present study, wecharacterized the evolution of microglia/macrophages polarization as amechanism contributing to neuronal damage alongside the neurode-generative process induced by MPTPp treatment, and when MPTPpwas co-administered with the PPAR-γ agonist rosiglitazone. To thisaim, we used double staining immunofluorescence associated withReal Time quantitative PCR (RTPCR) to assess changes in the productionof TNF-α and IL-1β asmarkers of the pro-inflammatory phenotype, andTGF-β, IL-10 and CD206 as markers of the anti-inflammatory M2phenotype.

Experimental procedures

Drugs

MPTP-HCl (Sigma, Italy) was dissolved in saline; probenecid(Sigma, Italy) was dissolved in 5% NaHCO3. Rosiglitazone (SantaCruz Biotechnology) was suspended in 0.5% methylcellulose.

Treatment

Three-month-old male C57BL/6J mice (Charles River, Italy) weredivided into seven groups (n = 5–10 for each group). The controlgroup received saline as a vehicle, groups M1, M3, M7, and M10received 1, 3, 7, and 10 doses of MPTP (25 mg/kg i.p.) respectively,plus probenecid (100 mg/kg i.p.) (MPTPp). MPTPp was injected twicea week up to 5 weeks (Schintu et al., 2009b). Group MR7 received 10doses of MPTPp over 5 weeks and the PPAR-γ agonist rosiglitazone(10 mg/kg i.p.) starting on the 2ndday after the seventhMPTPp admin-istration (Carta et al., 2011). Group MRchr received 10 doses of MPTPpand chronic rosiglitazone (10 mg/kg i.p.) starting 1 day before thefirst neurotoxin injection. For both the MR7 and MRchr groups,rosiglitazone was administered daily, 1 h before MPTPp and until sacri-fice. Twenty-four hours after discontinuation of MPTPp treatment, partof mice were anesthetized with chloral hydrate (400 mg/kg i.p.) andtranscardially perfused with 4% paraformaldehyde (PFA)/0.1 M phos-phate buffer (PBS). Brains were removed, postfixed in PFA for 2 h andstored in 0.1% sodium azide PBS until immunohistochemical processing.Remainingmice belonging to groups V, M10 andMRchr were sacrificedwith CO2 24 h after completion of drug treatment, brains were rapidlyremoved, liquid nitrogen-frozen and stored at −80° for RTPCR. Allanimal experimentations have been conducted in accordance with theguidelines for care and use of experimental animals of the EuropeanCommunities Council Directive 86/609/EEC.

282 A. Pisanu et al. / Neurobiology of Disease 71 (2014) 280–291

Immunohistochemistry

Coronal sections from the SNc (40-μm thick) were cut on avibratome and immunoreacted with an antibody against one of the fol-lowing cytokines for 40 h (TNF-α polyclonal rabbit anti-TNF-α 1:800,Abbiotec, USA; IL-1β polyclonal rabbit anti-IL-1β 1:200, Abbiotec,USA; TGF-β polyclonal rabbit anti TGF-1β 1:200, Abbiotec, USA; IL-10polyclonal rabbit anti-IL-10 1:200, Abbiotec, USA) or against CD11band CD206 for 12 h (CD11b monoclonal rat anti mouse 1:000,Serotec-Oxford; CD206 polyclonal goat anti CD206 1:500, Santa CruzBiotechnology, CA). For cytokines immunolabeling, after washing inPBS sections were incubated with the cytokines secondary antibodiesfor 2 h, and then with the anti Iba-1 antibody for 40 h (polyclonal goatanti-Iba-1 1:1000, Novus Biological, UK), followed by a 4-hourincubation with Iba-1 secondary antibody. For CD11b and CD206immunolabeling, after washing in PBS/0.2% Triton X-100 sectionswere incubated with the secondary antibodies for 1 h. For fluorescentdouble labeling, AlexaFluor® 594-conjugated immunoglobulin G (IgG,1:200, Jackson ImmunoResearch Europe, UK), was used to detect Iba-1and CD11b immunoreactivity (IR), while three-step detection was usedto increase the signal of the cytokines and CD206 by combining biotin-conjugated IgG (1:200, Invitrogen, US) and streptavidin–fluorescein(1:500, Vector, UK). To allow visualization of cell nuclei, sections wereincubated for 5 min in 5 μM Hoechst 33258 (H 33258) solution (Sigma,Italy). For diaminobenzidine (DAB) visualization of tyrosinehydroxylase (TH), sections were incubated with an anti-TH antibody(IgG polyclonal rabbit anti-TH, 1:1000, Biomol, UK) and the classicavidin–peroxidase complex (ABC, Vector, UK) protocol was applied,using 3,3′-diaminobenzidine (Sigma) as a chromogen.

Gene expression analysis

The midbrain was dissected and RNA was isolated using RNeasyPlus Mini Kit (Qiagen, Ambion Inc. Austin, Texas, USA) according tothe manufacturer's instructions. RNA (1 μg) was reverse transcribedusing High Capacity cDNA Reverse Transcription Kit with RNase Inhibi-tor (cat.n.4374966, Life Technologies) (Invitrogen S.r.l. Milan, Italy)according to the manufacturer's instructions. Real Time PCR wasperformed in a 7300 Real-Time PCR System (Life Technologies). Theamplification mixture contained 20 ng of cDNA, 10 μl 2× TaqManGene Expression PCRMastermix (Life Technologies) and 1 μl of specific20× TaqMan Gene Expression Assay (Mm00485148-m1 for Mrc1/CD206, Mm00443258-m1/Tnf and Mm01178820_m1 Tgfb1, LifeTechnologies). All samples were analyzed in triplicate and β-actingene expression (Mm006607939-s1cat. N.4331182 Life Technologies)was used as housekeeping control for possible differences in cDNAamount. Each experimentwas repeated twice. The relative gene expres-sion was calculated according to the 2-ΔΔCTmethod using the vehicle-treated tissue as control.

Data analysis and statistics

Stereological counting of TH-immunoreactive neuronsTH-immunoreactive neurons were counted on both hemispheres

(Carta et al., 2011). All stereological counting was performed using aLeica microscope (DMLB; Leica, Denmark) equipped with a camera(Basler Vision Technologies, Germany) and amicroscope stage connect-ed to an xyz stepper (PRIOR ProScan) and the newCAST Visiopharm(Denmark) software. Systemic uniform, random sampling was used tochoose the sections for the analysis. The first sampling item was takenat random from the frontal part of the SNc and all the followingsampling items were taken at a fixed distance from the previous one(280 μm intervals). The SNc region was outlined at low magnification(×5) for area estimation. The number of labeled neuronswas calculatedunder ×63 magnification using randomized meander sampling andoptical dissector methods. The thickness of cut sections was 40 μm

and the optical dissector heightwas 10 μm. The top (10 μm) and bottom(20 μm) layers that had shrunk during the staining procedure werediscarded. The sampling area covered 20% of the region of interest.The counting frame (8127.1 μm2) applied the exclusion and inclusionlines, and unbiased counting was performed by an experimenterblinded to the treatment. In each sampled area the dissector positionwas adjusted. TH-ir neuronsor Nissl-stained cells were counted onlywhen present completely or partially inside the frame and when theydid not touch any of the red exclusion lines. Neurons which touchedgreen inclusion lines were counted.

Results are presented as the mean number of TH-immunoreactiveneurons or Nissl-stained cells per mm3 (density) ± SEM, calculatedusing the following formula:

D ¼ sum of Q=V sampling ðD ¼ density of labeled cells per mm3;

V sampling ¼ area of the region of interest � dissector height;Q ¼ total count of labeled neuronsÞ:

In addition, we extrapolated the numbers of dopaminergic neuronsin whole SNc using the following formula:

N ¼ G � V G ¼ density of cells per mm3; V ¼ volume of SNc

� �:

The volume of SNc was reconstructed by calculating the distanceamong slices, multiplied by the outlined area (Carta et al., 2011).

Double labeling analysisIba-1-positive microglia were identified at ×100/1.25 oil magnifica-

tion by comparison of Iba-1 immunoreactivity and H 33258 nuclearcounterstaining. For each animal, an average of 50 Iba-1-positivecells was captured from both the left and right SNc and analyzed. Foreach field, three images were captured using three different filtersets in order to visualize the red signal of Iba-1-IR, the green signal ofcytokine-IR and the blue signal of cell nuclei counterstaining.

By using NIH software ImageJ, a stack of the three images wasobtained for each acquired field in order to identify Iba-1-positivemicroglial cells. For eachmicroglial cell, the body and primary processeswere outlined, and the area occupied by each microglial cell and thepercentage of Iba-1 colocalization with cytokine-IR were measured.Colocalization values obtained from cells analyzed in each animal, andcollected data from each experimental group (n = 5) were plotted asa frequency distribution displaying the percentage of colocalization be-tween Iba-1 and the selected cytokine signal in the outlined microglialcells. Statistical analysis was performed to detect significant differencesamong frequency distributions for each cytokine. Moreover, histograminspection suggested that statistical differences were imputable todrug effect in a cell subpopulation, whereas a large population ofcells maintained colocalization values similar to the vehicle (Fig. 3).For that reason, a deconvolution analysis was applied to the histogramsin order to unmask subpopulations of cells affected by MPTPp and/or rosiglitazone treatment (see Additional file 1 for details of thedeconvolution analysis procedure). Based on the deconvolution results,an appropriate threshold (50)was set in order to separate the identifiedcell populations; mean values above it were calculated for each experi-mental group and statistically compared (Carta et al., 2005).

Confocal microscopy analysisQualitative and quantitative analyses for CD11b and CD206 were

performed using a Leica 4D confocal laser scanning microscope,equipped with an argon–krypton laser. Images were digitized 24 hafter the immunofluorescence procedure. Surface rendering, maximumintensity, colocalization, and simulated fluorescence process algorithmswere used (ImageJ 1.48q and Imaris 7.0). Volume of colocalized ele-ments was determined as follows: for each dataset (40–60 images), acolocalization channel was automatically composed by Imaris 7.3. In

283A. Pisanu et al. / Neurobiology of Disease 71 (2014) 280–291

the resulting stacks, four regions of interest (x=40 μm; y=40 μm; z=10 μm) were randomly chosen and volume of the elements of interestwas calculated, summed and expressed as volume/μm3 (n = 200).

StatisticsThe results from the stereological analysis of nigral cells as well as

CD11b/CD206 colocalization and RTPCR were statistically analyzedwith a one-way analysis of variance (ANOVA) followed by Tukey'spost-hoc test. Level of significance was set at p b 0.05.

Frequency distributions of cytokines were statistically comparedwith the non-parametric Kolmogorov–Smirnov test, with significancelevels set at p b 0.05. The histograms were further inspected with adeconvolution analysis in order to unmask subpopulations of cellsunder the curve. The effect of drug treatments on the identified subpop-ulation was determined by one-way ANOVA followed by the LSD post-hoc test for comparison between individual groups (Fig. 4). Level ofsignificance was set at p b 0.05. In order to evaluate changes to theM1/M2 ratio, the percentage of cells expressing different cytokineswas statistically compared within each experimental group with aone-way ANOVA followed by Tukey's post-hoc test (Fig. 5).

Results

Evaluation of MPTP-induced nigral degeneration

MPTPp chronic treatment reduced both the density and the numberof TH-positive neurons as well as Nissl-stained cells in the SNc by about40% (Fig. 1). Neurodegeneration was gradual along the chronic neuro-toxin treatment, with a slight reduction in the number of TH-positiveneurons after three MPTP injections (M3), and a larger significant neu-ronal loss after subsequent injections, in agreement with the previous

0

5000

10000

15000

20000

*

TH-p

osit

ive

neur

ons

(cou

nts)

treatment n

mean density (X 103)

TH cells Nissl

V 5 25.51 ± 0.8 28.70

M1 5 23.49 ± 1.2 28.32 ±

M3 5 21.33 ± 0.82* 25.67 ±

M7 5 19.50 ± 0.69* 24.16 ±

M10 5 17.28 ± 0.64* 20.86 ±

MR7 5 20.58 ± 0.98* 24.83 ±

MRchr 5 22.00 ± 0.90^

26.39 ±

Fig. 1. Extrapolated total number and mean density of TH-positive and Nissl-stained cells in th^p b 0.05 versus chronic MPTPp (M10), by Tukey's post-hoc test (n = 5 per group).

characterization of this MPTPp protocol (Schintu et al., 2009b). Thedaily administration of rosiglitazone, from the seventh MPTPp injectionto the end of neurotoxin treatment (MR7), arrested the dopaminergicdegeneration, while rosiglitazone administration for the full length ofneurotoxin treatment (MRchr) prevented dopaminergic degenerationin the SNc, confirming our previous findings (TH: one-way ANOVA fortreatment effect: F(6,28) = 5.02 p b 0.01; Nissl: one-way ANOVA fortreatment effect: F(6,28) = 4.37 p b 0.01) (Arimoto et al., 2007; Cartaet al., 2011) (Fig. 1).

Changes in cytokine production by microglia

As expected, we found a large variability in the level of cytokineexpression in microglia in vehicle-treated animals (Figs. 2 and 3).Thus, while most Iba-1 cells displayed about 30–40% of colocalizationwith all cytokines, cells expressing undetectable colocalization aswell as cells expressing a high level of colocalization were found, sothat the extrapolated frequency distribution was close to normality.Moreover, statistical analysis revealed a significant difference betweenthe M10 histogram and vehicle for all the four cytokines analyzed(non-parametric Kolmogorov–Smirnov test, p b 0.05), except for IL-10.Chronic treatment with rosiglitazone (MRchr) significantly affectedcolocalization levels of all cytokines comparedwithM10 (non-parametricKolmogorov–Smirnov test, p b 0.05) (Fig. 4). Histogram inspection anddeconvolution analysis indicated the presence of two cell subpopulations,where the larger subpopulation maintained similar mean values beforeand after pharmacological treatments, while a smaller subpopulationwas affected by pharmacological treatments, displaying changes ofcolocalization levels. Mean values of cells falling within this populationwere statistically compared (Fig. 5).

VehicleM1M3M7M10MR7MRchr

*^

*

cells

mean N (X 103)

TH cells Nissl cells

± 1.20 15.61 ± 0.60 17.66 ± 1.75

2.12^

13.98 ± 1.06 16.20 ± 1.68

1.14 12.37 ± 0.84 13.93 ± 1.43

0.90* 12.14 ± 0.53* 13.05 ± 1.08*

0.41* 10.25 ±0.44* 11.65 ± 0.87*

1.65 12.52 ±0.68 14.07 ± 0.53

0.68^

14.18 ± 0.88^

15.41 ± 0.69

e whole SNc, as measured by unbiased stereology. *p b 0.05 versus vehicle-treated mice;

Fig. 2. TNF-α (upper panel) and IL-1β (lower panel) immunoreactivity in microglia in SNc. Representative images show Iba-1-IR cells (red), TNF-α-IR or IL-1β-IR cells (green) andcolocalized elements (yellow). Nuclei are visualized by Hoechst (blue). Scale bar: 10 μm.

284 A. Pisanu et al. / Neurobiology of Disease 71 (2014) 280–291

TNF-α immunoreactivityAn increase in Iba-1/TNF-α colocalizationwas induced by 1, 3, 7, and

10MPTPp injections (Figs. 2A and 5A).Moreover, the chronic treatmentwith rosiglitazone, when either administered starting at the first(MRchr) or the seventh (MR7) neurotoxin injection, reduced theamount of TNF-α colocalization. MRchr fully prevented the MPTPp-induced increase in TNF-α (Figs. 2A and 5A) (one-wayANOVA for treat-ment effect: F(6,28) = 2.67; p b 0.05).

IL-1β immunoreactivityIba-1/IL-1β colocalization was unchanged in the first part of

the MPTPp treatment, displaying increasing levels only after 7 and10 MPTPp injections (Figs. 2B and 5B). Such increases were fully

counteracted by treatments with the PPAR-γ agonist (Figs. 2B and 5B)(one-way ANOVA for treatment effect: F(6,28) = 5.81; p b 0.001).

TGF-β immunoreactivityIba-1/TGF-β levels of colocalization were mostly unchanged along

the MPTPp treatment, except for MPTP10, which induced a significantdecrease in cytokine levels (Figs. 3A and 5C).Moreover, both treatmentswith rosiglitazone reverted theMPTPp effect, restoring control levels ofTGF-β (Figs. 3A and 5C) (one-way ANOVA for treatment effect:F(6,28) = 3.33; p b 0.01).

IL-10 immunoreactivityIba-1/IL-10 colocalizationwas unchanged byMPTPp administration,

although a tendency to decrease was observed after the full MPTPp

Fig. 3. TGF-β (upper panel) and IL-10 (lower panel) immunoreactivity in microglia in SNc. Representative images show Iba-1-IR cells (red), TGF-β-IR or IL-10-IR cells (green) andcolocalized elements (yellow). Nuclei are visualized by Hoechst (blue). Scale bar: 10 μm.

285A. Pisanu et al. / Neurobiology of Disease 71 (2014) 280–291

treatment (Figs. 3B and 5D). Rosiglitazone treatments increased IL-10levels comparedwithM10 (Figs.3B and 5D) (one-way ANOVA for treat-ment effect: F(6,28) = 2,63; p b 0.01).

Changes in CD206 immunoreactivity

Confocal microscopy analysis showed that CD206 positive microglia/macrophages were present in vehicle treated mice, and graduallydecreased along the MPTPp treatment. Remarkably rosiglitazone giveneither during the full neurotoxin treatment or in the last part onlyreverted the MPTPp effect, restoring control levels of CD206 (Figs. 6Aand B). One way ANOVA showed a treatment effect for the colocalizedvolume of CD206/CD11b (F(4,195) = 45.81; P b 0.0001). Tukey's posthoc analysis reveled that microglia displayed a significantly decrease of

CD206/CD11b colocalization which was reduced by 22.38% in M3, andby 55.17% in M10 treated mice, and was restored to control levels inMR7 and MRchr treated mice.

Anti-inflammatory/pro-inflammatory microglia ratioIn order to evaluate the ratio between microglia producing anti-

inflammatory cytokines (putatively M2) and pro-inflammatory cyto-kines (putatively M1), we statistically compared colocalization levelsof all cytokines within each experimental group (Figs. 7A–G). For de-scriptive purposes, the ratio of anti- against pro-inflammatory cytokineswas plotted (Fig. 7H). In vehicle-treated animals, anti-inflammatorymi-croglia expressing TGF-β and IL-10 prevailed over pro-inflammatorymicroglia expressing TNF-α and IL-1β, the ratio being 2.46. During thechronic treatment with MPTPp, anti-inflammatory microglia remained

0

10

20

30

40

50

60

0 20 40 60 80 100

V

M10

MRch

0

10

20

30

40

50

60

0 20 40 60 80 100

V

M10

MRch

0

10

20

30

40

50

60

0 20 40 60 80 100

V

M10

MRch

0

10

20

30

40

50

60

0 20 40 60 80 100

V

M10

MRch

TNF-α

TGF-β

IL-1β

IL-10

% o

f mic

rogl

ia

% o

f mic

rogl

ia

% o

f mic

rogl

ia

% o

f mic

rogl

ia

% of colocaliza�on % of colocaliza�on

% of colocaliza�on % of colocaliza�on

Fig. 4. Frequency distribution of TNF-α, IL-1β, TGF-β, and IL-10 colocalization within Iba-1-IR cells. For each cytokine, values from vehicle, chronic MPTPp (M10), and chronic MPTPpplus rosiglitazone (MRchr) are shown. Histograms express the percentage of Iba-1-IR cells displaying various levels of cytokines colocalization, ranging from 0 to 100. Histograms werestatistically compared with Kolmogorov–Smirnov test, with significance levels set at p b 0.05.

286 A. Pisanu et al. / Neurobiology of Disease 71 (2014) 280–291

stable until the seventh MPTPp injection, while pro-inflammatory mi-croglia gradually increased, bringing the ratio close to the unit (0.93,1.16, and 1.061 for M1, M3, and M7, respectively). Anti-inflammatory

A

C

Fig. 5. Mean values ± S.E. of cytokine colocalization with Iba-1-IR cells. Only cells falling withrecognized as being responsive to pharmacological treatments, were considered for statistical

microglia were significantly reduced after the full MPTPp chronic treat-ment (M10), where the ratio was reversed to a negative value com-pared with vehicle (0.31). Treatment with rosiglitazone administered

B

D

in the cell subpopulation identified by deconvolution analysis of related histograms, andcomparison. *p b 0.05 versus vehicle; ^p b 0.05 versus M10, by LSD post-hoc test.

vehicle M10 MRchrCD206/CD11b

A

B

Fig. 6. A) CD206 expression in CD11b-positive cells. Representative confocal images showing colocalized elements (yellow) in CD11b (red) positive cells in the SNc. B) Histogram showslevels of colocalization expressed as volume/μm3. *p b 0.001 versus vehicle; ^p b 0.001 versus M3 and M10 by Tukey's post-hoc test.

287A. Pisanu et al. / Neurobiology of Disease 71 (2014) 280–291

in the late part of neurotoxin treatment (MR7) increased anti-inflammatory microglia while decreasing pro-inflammatory microglia,partially restoring the ratio to 1.08. Instead, chronic rosiglitazone ad-ministration for the full length of the neurotoxin treatment (MRchr)prevented any increase in pro-inflammatory microglia, maintainingthe ratio at 2.43, similar to vehicle levels.

Changes in mRNA expression

Real Time PCR of RNA from midbrain did not produce significantdifferences in gene expression among experimental groups (Fig. 8).However, the trend of gene expression changes induced by the chronicMPTPp treatment was similar to changes observed in protein levels,displaying a tendency to increased TNF-α expression and decreasedTGF-β and CD206 expression. Rosiglitazone co-administration induceda slight not significant increase of TGF-β and CD206 expression, in linewith the changes observed in the protein levels, but not in TNF-αexpression (Fig. 8).

Discussion

We report that the progressive degeneration of nigral neurons andtheir rescue by rosiglitazone in the chronic MPTPp treatment wereassociated with dynamic changes in cytokine production by microglia/macrophages in the SNc. Although fluorescent immunohistochemistryonly provides a semi-quantitative measurement of protein levels, itallows to assess the extent of proteins colocalization, and to use thisvalue for comparison among experimental groups. Moreover this tech-nique, by evaluating protein level within each single cell, allows dis-criminating protein modulation in different cell types. Based oncytokine evaluation, microglia were activated early in this area andremained reactive for the whole length of the MPTPp treatment.Reactive microglia were also found in the SNc of mice receivingMPTPp in association with rosiglitazone. However, a variable balance

was found in the content of pro- and anti-inflammatory cytokinesas well as CD206, depending on the pharmacological treatment andstage of degeneration, suggesting dynamic changes in the anti-inflammatory/pro-inflammatory microglia ratio. Hence, during thechronic treatment with MPTPp pro-inflammatory microglia increased,while anti-inflammatory microglia decreased. Remarkably, treatmentwith rosiglitazone, which limited or prevented the damage to the SNc,induced a switch in the MPTP-activated microglia, boosting anti-inflammatory microglia, while reverting to vehicle levels the pro-inflammatory microglia. Analysis of mRNA expression by RTPCRshowed only slight but not significant changes in cytokines and CD206expression. It must be pointed out thatmRNA expressionwasmeasuredin the whole midbrain tissue, i.e. in glial as well in neuronal cells.Analysis of cytokines by immunofluorescence indicated a specificrosiglitazone effect in microglial cells, which might be largely lost bywhole tissue analysis. Nevertheless, changes in mRNA expressiondisplayed a trend similar to the changes at the protein level, with theexception of TNF-α expression,whichwas not affected by rosiglitazone.A cell-specific effect of rosiglitazone has been previously reported forPPAR-γ immunoreactivity, which was reduced in microglia but not inTH-positive neurons (Carta et al., 2011). Moreover, cytokines displaya peak-trend of expression following CNS insults. We analyzed mRNAexpression 24 h after completion of pharmacological treatment, whichmay not correspond to the highest mRNA expression time point. Alltogether results suggest an induction of pro-inflammatory over anti-inflammatory microglia by the neurotoxin treatment, and restorationof the physiological ratio by rosiglitazone.

MPTP-induced cytokine production

We report a gradual increase in TNF-α and IL-1β production bymicroglia alongside the chronic MPTPp treatment. Changes in cytokineproduction were limited to a cell subpopulation with the MPTPpprotocol used in the present study. This was not surprising, being in

A B

C D

E F

G H

Fig. 7. The percentage of Iba-I-IR cells expressing TNF-α, IL-1β, TGF-β, and IL-10was calculated for each experimental group and statistically compared (A–G). Data are expressed asmeanvalues± S.E. *p b 0.05 versus vehicle; ^p b 0.05 versusM10. For each experimental group, cells expressing pro-inflammatory cytokines, and cells expressing anti-inflammatory cytokineswere grouped and results are summarized in a histogram (H).

288 A. Pisanu et al. / Neurobiology of Disease 71 (2014) 280–291

accordance with the partial nigral degeneration induced by this proto-col. An increase in neurotoxic microglia and the neurotoxic role ofinflammatory cytokines in PD has been widely reported, althoughmost preclinical studies were conducted in acute or subacute modelsof dopaminergic degeneration (Sawada et al., 2006; Sriram et al.,2006; De Lella Ezcurra et al., 2010; Pott Godoy et al., 2008; McCoyand Tansey, 2008; Watson et al., 2012; Depino et al., 2003; Leal et al.,2013; Walsh et al., 2011). Activated microglia and inflammatory

cytokines have been described in distinct areas of the Parkinsonianbrain and in animal models of PD (Imamura et al., 2003; Barcia et al.,2011). Elevated expression of TNF-α has been described after acuteMPTP intoxication in mice, and deletion of its receptor is neuroprotec-tive in PD models (Sriram et al., 2006; Lofrumento et al., 2011). More-over, administration of TNF-α or IL-1β exacerbated 6-OHDA-inducedcell death (De Lella Ezcurra et al., 2010; Pott Godoy et al., 2008).In line with our results recent studies have shown an increase of

Fig. 8.mRNA expression for TNF-α, TGF-β and CD206 as measured by RTPCR in midbraintissue homogenate, 24 h after completion of chronic treatment with vehicle, MPTPp(M10) or rosiglitazone plus MPTPp (MRchr).

289A. Pisanu et al. / Neurobiology of Disease 71 (2014) 280–291

inflammatory cytokines, as measured by protein or mRNA levels, inthe midbrain and striatum of mice chronically treated with MPTPp(Luchman et al., 2009; Luchtman et al., 2012). In contrast, others failedto detect any cytokine increase in theMPTPpmodel, althoughmeasure-ment was made in the striatum only (Alvarez-Fischer et al., 2013).Markers of phagocytic activity, such as CD68, were found elevated inthe late disease stages and closely associated with neuronal death inthe post-mortem human substantia nigra and in animal models of PD(Sawada et al., 2006; Sanchez-Guajardo et al., 2013; Croisier et al.,2005;Marinova-Mutafchieva et al., 2009). In the present study, increas-ing levels of inflammatory cytokines may reflect the increasing neuro-toxic activity of microglia.

Moreover, we report for the first time changes in anti-inflammatorycytokine production in the chronic MPTPmodel of PD. Although elevat-ed levels of TGF-β as well as IL-10 have been described in Parkinsonianpatients, they have been poorly investigated in preclinical models ofPD. In one study, no changes in TGF-β mRNA expression were foundin α-synuclein-overexpressing mice, albeit expression by different celltypes was not investigated in that study (Watson et al., 2012). More-over, IL-10 mRNA levels were elevated in the acute MPTP model(Ciesielska et al., 2003). Changes in TGF-β and IL-10 production maybe a cardinal event in PD neuropathology. TGF-β is involved in the dif-ferentiation and survival of neurons, exerting neuroprotective actionsin vitro and in vivo experimental models of PD (Gonzalez-Aparicio

et al., 2010; Arimoto et al., 2007; Pratt and McPherson, 1997; Roussaet al., 2009). In humans, a positive association between genetic varia-tions in the TGF-β gene and PDhas been suggested, whilemice deficientfor TGF-β displayed a reduced neuronal loss in the SNc (Andrews et al.,2006; Goris et al., 2007). Furthermore, TGF-β co-infusion synergizedwith glial cell-derived neurotrophic factor (GDNF) beneficial effectsin PD models (Gonzalez-Aparicio et al., 2010; Schober et al., 2007).Infusion of IL-10 protected against lipopolysaccharide (LPS)-inducedcell death of dopaminergic neurons (Krieglstein et al., 1995). IL-10and TGF-β-induced neuroprotective effects have been suggested torely, at least in part, on the inhibition of cytokine and toxic species pro-duction by glial cells (Chao et al., 1995; Ledeboer et al., 2000; Lodge andSriram, 1996; Sawada et al., 1999). The decrease of CD206 levels, whichspecifically discriminate M2 microglia/macrophages, further supportsa suppression of M2 polarization by the MPTPp treatment. Therefore,our study suggests that decreased levels of anti-inflammatory cytokinesandM2microglia may contribute to the degenerative process along theneurotoxin treatment.

PPAR-γ agonist-induced cytokine production

PPAR-γ agonist rosiglitazone modulated cytokine production bymicroglia in the SNc, by boosting anti-inflammatory while revertingpro-inflammatory microglia to control levels. Of note, such changes incytokine as well as CD206 levels where observed when rosiglitazonewas administered during the full length of MPTPp treatment as well aswhen it was given in the late stage only, i.e. in presence of advancedneuronal loss and neuroinflammation. Therefore, rosiglitazone wasable to either prevent the detrimental polarization of microglia, orswitch polarization of reactive microglia from the pro-inflammatoryto the anti-inflammatory phenotype. Hitherto, modulation of anti-inflammatory cytokine production by PPAR-γ agonists has not beeninvestigated in neurodegenerative conditions. Studies onperipheralmac-rophages have shown that PPAR-γ agonists increased the production ofanti-inflammatory cytokines while decreasing pro-inflammatory cyto-kines, and drove macrophage polarization toward the M2, suppressingthe M1 phenotype (Satoh et al., 2010; Fujisaka et al., 2009; Kim et al.,2005). Moreover, the selective disruption of PPAR-γ in myeloid cellsimpaired M2 macrophage activation (Odegaard et al., 2007). Evidencehas shown the involvement of PPAR-γ in survival pathways, such as thephosphoinositide 3-kinase/protein kinase B (P13K/AKT) and heat-shockprotein pathways (Bernardo and Minghetti, 2006; Park et al., 2007;Xing et al., 2008). On the other hand a wealth of data has suggestedthat PPAR-γ agonists inhibit the expression of inflammatory cytokinesby microglia in several models of neurodegeneration, included PD, bysuppressing the activity of the transcription factor, nuclear factor-κB(NF-κB) (Carta et al., 2011; Dehmer et al., 2004; Varga et al., 1812; Diabet al., 2004; Bernardo and Minghetti, 2006; Xu and Drew, 2007; Breidertet al., 2002; Kiaei et al., 2005; Luo et al., 2006; Patzer et al., 2008; Ricoteand Glass, 2007). In MPTP-treated mice and monkeys, pioglitazone androsiglitazone attenuated microglia proliferation and reduced TNF-α pro-duction (Carta et al., 2011; Dehmer et al., 2004; Schintu et al., 2009a;Swanson et al., 2011). Moreover, both compoundswere shown to inhibitLPS or β-amyloid-stimulated production of pro-inflammatory cytokines(Xing et al., 2008; Xu and Drew, 2007; Combs et al., 2000; Luna-Medinaet al., 2005). Accordingly, in mouse macrophages, PPAR-γ agonistsinhibited the inflammatory response by transrepression of NF-kB targetgenes (Pascual et al., 2007). In the PD model used in the present study,rosiglitazone drove microglia polarization through modulation of cyto-kine and CD206 levels, displaying an effect in the CNS similar to thatreported in peripheral macrophages. Results suggest that boosting ofmicroglia polarization toward an M2 anti-inflammatory phenotype maycontribute to the disease-modifying effect of PPAR-γ agonists reportedin preclinical models of PD.

The question whether microglia activation is entirely detrimental orwhether part of the inflammatory process may hold a beneficial role to

290 A. Pisanu et al. / Neurobiology of Disease 71 (2014) 280–291

counteract neuronal insult is still a matter of debate (Sanchez-Guajardoet al., 2013). Studies in AD brains and in animal models of AD haveshown that the progression of neurodegeneration is associated witha switch of microglia to the classical activation M1 phenotype over thealternative M2 phenotype (Hoozemans et al., 2006; Jankowsky et al.,2004). Moreover, a similar pattern of activation has been observed inLPS-treated mice and in aged brain, suggesting that a skewed polariza-tion of microglia toward the M1 phenotype may be a detrimentalevent underlying and sustaining neuronal damage (Lund et al., 2006;Jimenez et al., 2008; Nolan et al., 2005). The results of the presentstudy strongly suggest that a similar pattern of microglia activationmay occur in the chronic model of PD, and that rosiglitazone switchedthe disease-induced pattern of microglia polarization, suggesting thatthe immunomodulatory properties may contribute to the disease-modifying potential of PPAR-γ agonists in PD.

Conclusions

Progression of neurodegeneration in a chronic model of PD wasassociated with a gradual increase of neurotoxic pro-inflammatory mi-croglia over the anti-inflammatory phenotype.Moreover, pharmacolog-ical treatmentwith a PPAR-γ agonist switchedmicroglia polarization toanti-inflammatory, turning microglia activation into a beneficial eventin the diseased brain. Therefore, although microgliosis is generallyregarded as a toxic event in PD pathogenesis, results suggest that mi-croglia might display different effector function at progressive stagesof the disease, and that an immunomodulatory approach, which selec-tively targets microglia polarization by promoting theM2 over M1 phe-notype, may provide a pivotal contribution for a successful disease-modifying therapy for PD.

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.nbd.2014.08.011.

Acknowledgments

We thank Prof. Giacomo Diaz, PhD, Dept. of Biomedical Sciences,Univ. of Cagliari, for assistance in statistical analysis. We thank SusanBos of Medical Edit for language revision of the manuscript.

References

Alvarez-Fischer, D., Noelker, C., Vulinović, F., Grünewald, A., Chevarin, C., Klein, C., Oertel,W.H., Hirsch, E.C., Michel, P.P., Hartmann, A., 2013. Bee venom and its componentapamin as neuroprotective agents in a Parkinson disease mouse model. PLoS One 8,e61700.

Andrews, Z.B., Zhao, H., Frugier, T., Meguro, R., Grattan, D.R., Koishi, K., McLennan, I.S.,2006. Transforming growth factor beta2 haploinsufficient mice develop age-relatednigrostriatal dopamine deficits. Neurobiol. Dis. 21, 568–575.

Arimoto, T., Choi, D.Y., Lu, X., Liu, M., Nguyen, X.V., Zheng, N., Stewart, C.A., Kim, H.C., Bing,G., 2007. Interleukin-10 protects against inflammation-mediated degeneration of do-paminergic neurons in substantia nigra. Neurobiol. Aging 6, 894–906.

Barcia, C., Sánchez Bahillo, A., Fernández-Villalba, E., Bautista, V., Poza, Y., Poza, M.,Fernández-Barreiro, A., Hirsch, E.C., Herrero, M.T., 2004. Evidence of active microgliain substantia nigra pars compacta of Parkinsonian monkeys 1 year after MPTP expo-sure. Glia 46, 402–409.

Barcia, C., Ros, C.M., Annese, V., Gómez, A., Ros-Bernal, F., Aguado-Yera, D., Martínez-Pagán, M.E., de Pablos, V., Fernandez-Villalba, E., Herrero, M.T., 2011. IFN-γ signaling,with the synergistic contribution of TNF-α, mediates cell specific microglial andastroglial activation in experimental models of Parkinson's disease. Cell Death Dis.2, e142.

Bernardo, A., Minghetti, L., 2006. PPAR-gamma agonists as regulators of microglial activa-tion and brain inflammation. Curr. Pharm. Des. 12, 93–109.

Bernardo, A., Levi, G., Minghetti, L., 2000. Role of the peroxisome proliferator-activatedreceptor-gamma (PPAR-gamma) and its natural ligand 15-deoxy-Delta12, 14-prostaglandin J2 in the regulation of microglial functions. Eur. J. Neurosci. 12,2215–2223.

Breidert, T., Callebert, J., Heneka, M.T., Landreth, G., Launay, J.M., Hirsch, E.C., 2002. Protec-tive action of the peroxisome proliferator-activated receptor-gamma agonist pioglit-azone in a mouse model of Parkinson's disease. J. Neurochem. 82, 615–624.

Brodacki, B., Staszewski, J., Toczyłowska, B., Kozłowska, E., Drela, N., Chalimoniuk, M.,Stepien, A., 2008. Serum interleukin (IL-2, IL-10, IL-6, IL-4), TNFalpha, and INFgammaconcentrations are elevated in patients with atypical and idiopathic parkinsonism.Neurosci. Lett. 2, 158–162.

Carta, A.R., 2013. PPAR-γ: therapeutic prospects in Parkinson's disease. Curr. Drug Targets14, 743–751.

Carta, A.R., Pisanu, A., 2013. Modulating microglia activity with PPAR-γ agonists: a prom-ising therapy for Parkinson's disease? Neurotox. Res. 23, 112–123.

Carta, A.R., Tronci, E., Pinna, A., Morelli, M., 2005. Different responsiveness of striatonigraland striatopallidal neurons to L-DOPA after a subchronic intermittent L-DOPA treat-ment. Eur. J. Neurosci. 21, 1196–1204.

Carta, A.R., Frau, L., Pisanu, A., Wardas, J., Spiga, S., Carboni, E., 2011. Rosiglitazone de-creases peroxisome proliferator receptor-γ levels in microglia and inhibits TNF-αproduction: new evidences on neuroprotection in a progressive Parkinson's diseasemodel. Neuroscience 194, 250–261.

Chao, C.C., Hu, S., Peterson, P.K., 1995. Glia, cytokines, and neurotoxicity. Neurobiology2–3, 189–205.

Ciesielska, A., Joniec, I., Przybyłkowski, A., Gromadzka, G., Kurkowska-Jastrzebska, I.,Członkowska, A., Członkowski, A., 2003. Dynamics of expression of the mRNA forcytokines and inducible nitric synthase in a murine model of the Parkinson's disease.Acta Neurobiol. Exp. (Wars) 2, 117–126.

Combs, C.K., Johnson, D.E., Karlo, J.C., Cannady, S.B., Landreth, G.E., 2000. Inflammatorymechanisms in Alzheimer's disease: inhibition of beta-amyloid-stimulated proin-flammatory responses and neurotoxicity by PPARgamma agonists. J. Neurosci. 20,558–567.

Croisier, E., Moran, L.B., Dexter, D.T., Pearce, R.K., Graeber, M.B., 2005. Microglial inflam-mation in the Parkinsonian substantia nigra: relationship to alpha-synuclein deposi-tion. J. Neuroinflammation 2, 14.

De Lella Ezcurra, A.L., Chertoff, M., Ferrari, C., Graciarena, M., Pitossi, F., 2010. Chronicexpression of low levels of tumor necrosis factor-alpha in the substantia nigraelicits progressive neurodegeneration, delayed motor symptoms and microglia/macrophage activation. Neurobiol. Dis. 37, 630–640.

Dehmer, T., Heneka, M.T., Sastre, M., Dichgans, J., Schulz, J.B., 2004. Protection by pioglit-azone in the MPTP model of Parkinson's disease correlates with I kappa B alphainduction and block of NF kappa B and iNOS activation. J. Neurochem. 88, 494–501.

Depino, A.M., Earl, C., Kaczmarczyk, E., Ferrari, C., Besedovsky, H., del Rey, A., Pitossi, F.J.,Oertel, W.H., 2003. Microglial activation with atypical proinflammatory cytokineexpression in a rat model of Parkinson's disease. Eur. J. Neurosci. 18, 2731–2742.

Diab, A., Hussain, R.Z., Lovett-Racke, A.E., Chavis, J.A., Drew, P.D., Racke, M.K., 2004.Ligands for the peroxisome proliferator-activated receptor-gamma and the retinoidX receptor exert additive anti-inflammatory effects on experimental autoimmuneencephalomyelitis. J. Neuroimmunol. 148, 116–126.

Fujisaka, S., Usui, I., Bukhari, A., Ikutani, M., Oya, T., Kanatani, Y., Tsuneyama, K., Nagai, Y.,Takatsu, K., Urakaze, M., Kobayashi, M., Tobe, K., 2009. Regulatory mechanisms foradipose tissue M1 and M2 macrophages in diet-induced obese mice. Diabetes 58,2574–2582.

Gerhard, A., Pavese, N., Hotton, G., Turkheimer, F., Es, M., Hammers, A., Eggert, K., Oertel,W., Banati, R.B., Brooks, D.J., 2006. In vivo imaging of microglial activation with[11C](R)-PK11195 PET in idiopathic Parkinson's disease. Neurobiol. Dis. 21, 404–412.

Gonzalez-Aparicio, R., Flores, J.A., Fernandez-Espejo, E., 2010. Antiparkinsonian trophicaction of glial cell line-derived neurotrophic factor and transforming growth factorβ1 is enhanced after co-infusion in rats. Exp. Neurol. 226, 136–147.

Goris, A., Williams-Gray, C.H., Foltynie, T., Brown, J., Maranian, M., Walton, A., Compston,D.A., Barker, R.A., Sawcer, S.J., 2007. Investigation of TGFB2 as a candidate gene inmultiple sclerosis and Parkinson's disease. J. Neurol. 254, 846–848.

Harms, A.S., Lee, J.K., Nguyen, T.A., Chang, J., Ruhn, K.M., Treviño, I., Tansey, M.G., 2012.Regulation of microglia effector functions by tumor necrosis factor signaling. Glia 2,189–202.

Hirsch, E.C., Hunot, S., Damier, P., Faucheux, B., 1998. Glial cells and inflammation inParkinson's disease: a role in neurodegeneration? Ann. Neurol. 44, S115–S120.

Hoozemans, J.J., Veerhuis, R., Rozemuller, J.M., Eikelenboom, P., 2006. Neuroinflammationand regeneration in the early stages of Alzheimer's disease pathology. Int. J. Dev.Neurosci. 24, 157–165.

Imamura, K., Hishikawa, N., Sawada, M., Nagatsu, T., Yoshida, M., Hashizume, Y., 2003.Distribution of major histocompatibility complex class II-positive microglia and cyto-kine profile of Parkinson's disease brains. Acta Neuropathol. 106, 518–526.

Jankowsky, J.L., Fadale, D.J., Anderson, J., Xu, G.M., Gonzales, V., Jenkins, N.A., Copeland, N.G.,Lee, M.K., Younkin, L.H., Wagner, S.L., Younkin, S.G., Borchelt, D.R., 2004. Mutantpresenilins specifically elevate the levels of the 42 residue beta-amyloid peptidein vivo: evidence for augmentation of a 42-specific gamma secretase. Hum. Mol.Genet. 13, 159–170.

Jimenez, S., Baglietto-Vargas, D., Caballero, C., Moreno-Gonzalez, I., Torres, M., Sanchez-Varo, R., Ruano, D., Vizuete, M., Gutierrez, A., Vitorica, J., 2008. Inflammatory responsein the hippocampus of pS1M146L/APP751SL mouse model of Alzheimer's disease:age-dependent switch in the microglial phenotype from alternative to classic.J. Neurosci. 28, 11650–11661.

Johnston, L.C., Su, X., Maguire-Zeiss, K., Horovitz, K., Ankoudinova, I., Guschin, D.,Hadaczek, P., Federoff, H.J., Bankiewicz, K., Forsayeth, J., 2008. Human interleukin-10 gene transfer is protective in a rat model of Parkinson's disease. Mol. Ther. 8,1392–1399.

Kiaei, M., Kipiani, K., Chen, J., Calingasan, N.Y., Beal, M.F., 2005. Peroxisome proliferator-activated receptor-gamma agonist extends survival in transgenic mouse model ofamyotrophic lateral sclerosis. Exp. Neurol. 191, 331–336.

Kim, S.R., Lee, K.S., Park, H.S., Park, S.J., Min, K.H., Jin, S.M., Lee, Y.C., 2005. Involvementof IL-10 in peroxisome proliferator-activated receptor gamma-mediated anti-inflammatory response in asthma. Mol. Pharmacol. 68, 1568–1575.

Krieglstein, K., Suter-Crazzolara, C., Hötten, G., Pohl, J., Unsicker, K., 1995. Trophic and pro-tective effects of growth/differentiation factor 5, a member of the transforminggrowth factor-beta superfamily, on midbrain dopaminergic neurons. J. Neurosci.Res. 42, 724–732.

291A. Pisanu et al. / Neurobiology of Disease 71 (2014) 280–291

Leal, M.C., Casabona, J.C., Puntel, M., Pitossi, F.J., 2013. Interleukin-1β and tumor necrosisfactor-α: reliable targets for protective therapies in Parkinson's disease? Front. Cell.Neurosci. 7, 53.

Ledeboer, A., Brevé, J.J., Poole, S., Tilders, F.J., Van Dam, A.M., 2000. Interleukin-10,interleukin-4, and transforming growth factor-beta differentially regulatelipopolysaccharide-induced production of pro-inflammatory cytokines and nitricoxide in co-cultures of rat astroglial and microglial cells. Glia 2, 134–142.

Lodge, P.A., Sriram, S., 1996. Regulation of microglial activation by TGF-beta, IL-10, andCSF-1. J. Leukoc. Biol. 4, 502–508.

Lofrumento, D.D., Saponaro, C., Cianciulli, A., De Nuccio, F., Mitolo, V., Nicolardi, G., Panaro,M.A., 2011. MPTP-induced neuroinflammation increases the expression of pro-inflammatory cytokines and their receptors inmouse brain. Neuroimmunomodulation18, 79–88.

Luchman, D.W., Shao, D., Song, C., 2009. Behavior, neurotransmitters and inflammation inthree regimens of the MPTPmouse model of Parkinson's disease. Physiol. Behav. 1–2,130–138.

Luchtman, D.W., Meng, Q., Song, C., 2012. Ethyl-eicosapentaenoate (E-EPA) attenuatesmotor impairments and inflammation in the MPTP-probenecid mouse model ofParkinson's disease. Behav. Brain Res. 226, 386–396.

Luna-Medina, R., Cortes-Canteli, M., Alonso, M., Santos, A., Martínez, A., Perez-Castillo, A., 2005. Regulation of inflammatory response in neural cells in vitroby thiadiazolidinones derivatives through peroxisome proliferator-activated receptorgamma activation. Biol. Chem. 280, 21453–21462.

Lund, S., Christensen, K.V., Hedtjärn, M., Mortensen, A.L., Hagberg, H., Falsig, J., Hasseldam,H., Schrattenholz, A., Pörzgen, P., Leist, M., 2006. The dynamics of the LPS triggeredinflammatory response of murine microglia under different culture and in vivoconditions. J. Neuroimmunol. 180, 71–87.

Luo, Y., Yin, W., Signore, A.P., Zhang, F., Hong, Z., Wang, S., Graham, S.H., Chen, J., 2006.Neuroprotection against focal ischemic brain injury by the peroxisome proliferator-activated receptor-gamma agonist rosiglitazone. J. Neurochem. 97, 435–448.

Marinova-Mutafchieva, L., Sadeghian,M., Broom, L., Davis, J.B.,Medhurst, A.D., Dexter, D.T.,2009. Relationship between microglial activation and dopaminergic neuronal lossin the substantia nigra: a time course study in a 6-hydroxydopamine model ofParkinson's disease. J. Neurochem. 110, 966–975.

McCoy, M.K., Tansey, M.G., 2008. TNF signaling inhibition in the CNS: implications fornormal brain function and neurodegenerative disease. J. Neuroinflammation 5, 45.

McGeer, P.L., Itagaki, S., Akiyama, H., McGeer, E.G., 1988. Rate of cell death in parkinson-ism indicates active neuropathological process. Ann. Neurol. 24, 574–576.

McGeer, P.L., Schwab, C., Parent, A., Doudet, D., 2003. Presence of reactive microglia inmonkey substantia nigra years after 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridineadministration. Ann. Neurol. 54, 599–604.

Mogi, M., Harada, M., Kondo, T., Riederer, P., Inagaki, H., Minami, M., Nagatsu, T., 1994.Interleukin-1 beta, interleukin-6, epidermal growth factor and transforming growthfactor-alpha are elevated in the brain from Parkinsonian patients. Neurosci. Lett.180, 147–150.

Mogi, M., Harada, M., Kondo, T., Narabayashi, H., Riederer, P., Nagatsu, T., 1995.Transforming growth factor-beta 1 levels are elevated in the striatum and in ventric-ular cerebrospinal fluid in Parkinson's disease. Neurosci. Lett. 193, 129–132.

Mogi, M., Harada, M., Narabayashi, H., Inagaki, H., Minami, M., Nagatsu, T., 1996. Interleu-kin (IL)-1 beta, IL-2, IL-4, IL-6 and transforminggrowth factor-alpha levels are elevatedin ventricular cerebrospinal fluid in juvenile Parkinsonism and Parkinson's disease.Neurosci. Lett. 211, 13–16.

Nolan, Y., Maher, F.O., Martin, D.S., Clarke, R.M., Brady, M.T., Bolton, A.E., Mills, K.H., Lynch,M.A., 2005. Role of interleukin-4 in regulation of age-related inflammatory changes inthe hippocampus. J. Biol. Chem. 280, 9354–9362.

Odegaard, J.I., Ricardo-Gonzalez, R.R., Goforth, M.H., Morel, C.R., Subramanian, V.,Mukundan, L., Red Eagle, A., Vats, D., Brombacher, F., Ferrante, A.W., Chawla, A.,2007. Macrophage-specific PPARgamma controls alternative activation and improvesinsulin resistance. Nature 447, 1116–1120.

Orr, C.F., Rowe, D.B., Halliday, G.M., 2002. An inflammatory review of Parkinson's disease.Prog. Neurobiol. 68, 325–340.

Park, S.W., Yi, J.H., Miranpuri, G., Satriotomo, I., Bowen, K., Resnick, D.K., Vemuganti, R.,2007. Thiazolidinedione class of peroxisome proliferator-activated receptor gammaagonists prevents neuronal damage, motor dysfunction, myelin loss, neuropathicpain, and inflammation after spinal cord injury in adult rats. J. Pharmacol. Exp.Ther. 320, 1002–1012.

Pascual, G., Sullivan, A.L., Ogawa, S., Gamliel, A., Perissi, V., Rosenfeld, M.G., Glass, C.K.,2007. Anti-inflammatory and antidiabetic roles of PPARgamma. Novartis Found.Symp. 286, 183–196.

Patzer, A., Zhao, Y., Stöck, I., Gohlke, P., Herdegen, T., Culman, J., 2008. Peroxisomeproliferator-activated receptorsgamma (PPARgamma) differently modulate theinterleukin-6 expression in the peri-infarct cortical tissue in the acute and delayedphases of cerebral ischaemia. Eur. J. Neurosci. 28, 1786–1794.

Perego, C., Fumagalli, S., De Simoni, M.G., 2011. Temporal pattern of expression andcolocalization of microglia/macrophage phenotype markers following brain ischemicinjury in mice. J. Neuroinflammation 8, 174.

Petroske, E., Meredith, G.E., Callen, S., Totterdell, S., Lau, Y.S., 2001. Mouse model ofParkinsonism: a comparison between subacute MPTP and chronic MPTP/probenecidtreatment. Neuroscience 106, 589–601.

Pott Godoy,M.C., Tarelli, R., Ferrari, C.C., Sarchi, M.I., Pitossi, F.J., 2008. Central and systemicIL-1 exacerbates neurodegeneration and motor symptoms in a model of Parkinson'sdisease. Brain 131, 1880–1894.

Pratt, B.M., McPherson, J.M., 1997. TGF-beta in the central nervous system: potential rolesin ischemic injury and neurodegenerative diseases. Cytokine Growth Factor Rev. 8,267–292.

Rentzos, M., Nikolaou, C., Andreadou, E., Paraskevas, G.P., Rombos, A., Zoga, M.,Tsoutsou, A., Boufidou, F., Kapaki, E., Vassilopoulos, D., 2009. Circulatinginterleukin-10 and interleukin-12 in Parkinson's disease. Acta Neurol. Scand. 5,332–337.

Ricote, M., Glass, C.K., 2007. PPARs and molecular mechanisms of transrepression.Biochim. Biophys. Acta 1771, 926–935.

Rojo, A.I., Innamorato, N.G., Martín-Moreno, A.M., De Ceballos, M.L., Yamamoto, M.,Cuadrado, A., 2010. Nrf2 regulates microglial dynamics and neuroinflammation inexperimental Parkinson's disease. Glia 58, 588–598.

Roussa, E., von Bohlen und Halbach, O., Krieglstein, K., 2009. TGF-beta in dopamineneuron development, maintenance and neuroprotection. Adv. Exp. Med. Biol. 651,81–90.

Sanchez-Guajardo, V., Barnum, C.J., Tansey, M.G., Romero-Ramos, M., 2013.Neuroimmunological processes in Parkinson's disease and their relation to α-synuclein: microglia as the referee between neuronal processes and peripheralimmunity. ASN Neuro 30, 113–139.

Satoh, N., Shimatsu, A., Himeno, A., Sasaki, Y., Yamakage, H., Yamada, K., Suganami, T.,Ogawa, Y., 2010. Unbalanced M1/M2 phenotype of peripheral blood monocytes inobese diabetic patients: effect of pioglitazone. Diabetes Care 33, e7.

Sawada, M., Suzumura, A., Hosoya, H., Marunouchi, T., Nagatsu, T., 1999. Interleukin-10inhibits both production of cytokines and expression of cytokine receptors inmicroglia.J. Neurochem. 4, 1466–1471.

Sawada, M., Imamura, K., Nagatsu, T., 2006. Role of cytokines in inflammatory process inParkinson's disease. J. Neural Transm. Suppl. 70, 373–381.

Schintu, N., Frau, L., Ibba, M., Caboni, P., Garau, A., Carboni, E., Carta, A.R., 2009a. PPAR-gamma-mediated neuroprotection in a chronic mouse model of Parkinson's disease.Eur. J. Neurosci. 29, 954–963.

Schintu, N., Frau, L., Ibba, M., Garau, A., Carboni, E., Carta, A.R., 2009b. Progressive dopami-nergic degeneration in the chronic MPTPp mouse model of Parkinson's disease.Neurotox. Res. 16, 127–139.

Schober, A., Peterziel, H., von Bartheld, C.S., Simon, H., Krieglstein, K., Unsicker, K., 2007.GDNF applied to the MPTP-lesioned nigrostriatal system requires TGF-beta for itsneuroprotective action. Neurobiol. Dis. 25, 378–391.

Sriram, K., Matheson, J.M., Benkovic, S.A., Miller, D.B., Luster, M.I., O'Callaghan, J.P., 2006.Deficiency of TNF receptors suppresses microglial activation and alters the suscepti-bility of brain regions to MPTP-induced neurotoxicity: role of TNF-alpha. FASEB J.20, 670–682.

Straus, D.S., Glass, C.K., 2007. Anti-inflammatory actions of PPAR ligands: new insights oncellular and molecular mechanisms. Trends Immunol. 28, 551–558.

Swanson, C.R., Joers, V., Bondarenko, V., Brunner, K., Simmons, H.A., Ziegler, T.E.,Kemnitz, J.W., Johnson, J.A., Emborg, M.E., 2011. The PPAR-γ agonist pioglitazonemodulates inflammation and induces neuroprotection in Parkinsonian monkeys.J. Neuroinflammation 8, 91.

Tontonoz, P., Nagy, L., Alvarez, J.G., Thomazy, V.A., Evans, R.M., 1998. PPARgamma pro-motes monocyte/macrophage differentiation and uptake of oxidized LDL. Cell 93,241–252.

Varga, T., Czimmerer, Z., Nagy, L., 1812. PPARs are a unique set of fatty acid regulated tran-scription factors controlling both lipid metabolism and inflammation. Biochim.Biophys. Acta 2011, 1007–1022.

Varnum, M.M., Ikezu, T., 2012. The classification of microglial activation phenotypes onneurodegeneration and regeneration in Alzheimer's disease brain. Arch. Immunol.Ther. Exp. (Warsz) 60, 251–266.

Walsh, S., Finn, D.P., Dowd, E., 2011. Time-course of nigrostriatal neurodegenera-tion and neuroinflammation in the 6-hydroxydopamine-induced axonal andterminal lesion models of Parkinson's disease in the rat. Neuroscience 175,251–261.

Watson, M.B., Richter, F., Lee, S.K., Gabby, L., Wu, J., Masliah, E., Effros, R.B., Chesselet, M.F.,2012. Regionally-specific microglial activation in young mice over-expressing humanwildtype alpha-synuclein. Exp. Neurol. 237, 318–334.

Whitton, P.S., 2007. Inflammation as a causative factor in the aetiology of Parkinson'sdisease. Br. J. Pharmacol. 150, 963–976.

Xing, B., Xin, T., Hunter, R.L., Bing, G., 2008. Pioglitazone inhibition of lipopolysaccharide-induced nitric oxide synthase is associated with altered activity of p38 MAP kinaseand PI3K/Akt. J. Neuroinflammation 5, 4.

Xu, J., Drew, P.D., 2007. Peroxisome proliferator-activated receptor-gamma agonistssuppress the production of IL-12 family cytokines by activated glia. J. Immunol.178, 1904–1913.