inal

neada

CytokineNeurodegenerationInammatoryPesticide

ash murecytokine, granulocyte macrophage colony stimulating factor (GM-CSF), would

aled th

Neurobiology of Disease 43 (2011) 99112

Contents lists available at ScienceDirect

Neurobiology

.e ldisease (PD) (Carvey et al., 2006; Cicchetti et al., 2009; Liu et al., 2003).Parallel work in rodents has demonstrated that administration of thepesticides, paraquat and rotenone, provoked a loss of substantia nigrapars compacta (SNc) dopamine (DA) neurons, coupledwith histologicaland behavioral symptoms reminiscent of PD (Cannon et al., 2009; DiMonte, 2003; Litteljohnet al., 2009; Sherer et al., 2007; Somayajulu-Nituet al., 2009). The neurodegenerative process induced by paraquat, aswell as other well-established DA toxins, such as MPTP, involves theactivation of microglial cells and liberation of oxidative radicals, as well

coupled with mitochondrial dysfunction, and alterations of apoptoticregulators within the SNc (Fitzmaurice et al., 2003; Hartmann et al.,2002; Jenner, 1998; Knott et al., 2000; Nagatsu and Sawada, 2007;Tatton et al., 2003).

Given that the neurodegenerative process in PD occurs over aprolonged period of time, there may be a critical window whereintherapeutic intervention may halt or even reverse disease progres-sion. In particular, the trophic cytokines, glial derived neurotrophicfactor (GDNF) and brain derived neurotrophic factor (BDNF), haveas pro-inammatory cytokines, such as interfenecrosis factor- (TNF-) (McCoy et al., 20Indeed, the deleterious impact of paraquat w

Corresponding author at: 1125 Colonel By Drive, OttFax: +1 613 520 4052.

E-mail address: shayley@ccs.carleton.ca (S. Hayley).Available online on ScienceDirect (www.scienced

0969-9961/$ see front matter 2011 Elsevier Inc. Aldoi:10.1016/j.nbd.2011.02.011at chronic exposure to., heavy metals and evenogression of Parkinson's

neuroinammatory response (Mangano andHayley, 2009; Purisai et al.,2007). These studies are in line with post-mortem analyses of PD brain,which revealed signs of neuroinammation and oxidative damage,pesticides and other environmental toxins (e.gimmune infections) are involved in the prParaquatGM-CSFParkinson's diseaseToxinTrophic factor

Introduction

Epidemiological studies have revemodest but signicant neurodegenerative effect that was markedly augmented with LPS priming. Centralinfusion of GM-CSF into the substantia nigra pars compacta (SNc) prevented the loss of SNc dopamineneurons to a degree comparable to that of glial derived neurotrophic factor. Importantly, systemicadministration of GM-CSF also had neuroprotective consequences, suggesting that the trophic cytokine cancross the blood brain barrier to promote neuronal survival. Indeed, GM-CSF acted to inhibit the LPS andparaquat induced microglial response, while augmenting astrocyte immunoreactivity within the SNc.Moreover, GM-CSF blunted the paraquat induced reduction of brain derived neurotrophic factor within thehippocampus, as well as in culturedmesencephalic neurons. Although paraquat reducedmesencephalic levelsof the anti-apoptotic protein, Bcl-2, GM-CSF had no effect in this regard. Hence, GM-CSF appears to affectinammatory and/or neuroplastic factors within the SNc that may be linked to neurodegeneration, as well asin other brain regions (hippocampus), which could be important for co-morbid non-motor symptoms in PD.These data suggest that peripheral GM-CSF administration might hold promise as a treatment of PD.

2011 Elsevier Inc. All rights reserved.

pre-treated with the bacterial endotoxin, lipopolysaccharide (LPS), andthis effect was associated with a heightened microglial-dependentron- (IFN-) and tumor08; Mount et al., 2007).as augmented in rodents

been considered2000; Porritt et aGDNF have beenmodels of PD (E1996; Levivier eSimilarly, BDNF wcoupled with it(Taliaz et al., 20

awa, Ontario, Canada K1S 5B6.

irect.com).

l rights reserved.haride (LPS). As previously observed, paraquat provoked aKeywords:inhibit the neurodegenerative effects of the pesticide, paraquat, administered either alone or followingpriming with the bacterial endotoxin, lipopolysaccAccepted 27 February 2011Available online 4 March 2011

the hematopoietic trophicGranulocyte macrophage-colony stimulatdopaminergic cell loss in an environment

E.N. Mangano, S. Peters, D. Litteljohn, R. So, C. BethuInstitute of Neuroscience, Carleton University, 1125 Colonel By Drive, Ottawa, Ontario, Can

a b s t r a c ta r t i c l e i n f o

Article history:Received 20 October 2010Revised 26 January 2011

Parkinson's disease (PD) hinammatory agents, whicassociated with toxin expos

j ourna l homepage: wwwg factor protects against substantia nigratoxin model of Parkinson's disease

, J. Bobyn, M. Clarke, S. Hayley K1S 5B6

been linked to exposure to a variety of chemical (e.g., pesticides) anday act cumulatively over time. Finding novel means of limiting pathologywould have tremendous clinical importance. To this end, we assessed whether

of Disease

sev ie r.com/ locate /ynbd ipossible therapeutic agents for PD (Howells et al.,l., 2005; Yasuhara et al., 2007). In fact, both BDNF andshown to be neuroprotective in virtually all animalslamboli, 2005; Huang et al., 2010; Hung and Lee,t al., 1995; Shults et al., 1995; Zhang et al., 2007).as shown to impart antidepressant effects that were

s positive actions upon hippocampal neurogenesis10); an important nding given the high degree of

100 E.N. Mangano et al. / Neurobiology of Disease 43 (2011) 99112depressive co-morbidity evident in PD and the fact the we previouslyfound paraquat to induce depressive-like behaviors and enhancehippocampal monoamine activity (Litteljohn et al., 2009). Unfortu-nately, clinical trials using GDNF have failed to consistently yieldbenecial clinical outcomes, stemming perhaps from surgical com-plications and side-effects [GDNF is unable to cross the blood brainbarrier (BBB)] (Lang et al., 2006), and the limited diffusion of GDNF inhuman brain (i.e., leading to reduced bioavailability in nigrostriatalbrain regions) (Nutt et al., 2003; Salvatore et al., 2006).

Granulocyte-macrophage colony stimulating factor (GM-CSF) is ahematopoietic cytokine produced mainly by peripheral immune cells(T-lymphocytes, macrophages, and NK cells) and also by centralastrocytes (Davignon et al., 1988; Guillemin et al., 1996; Hercus et al.,2009). Its basic function is to promote the differentiation andmaturation of peripheral innate immune cells, including neutrophilsand dendritic cells. The role of GM-CSF in the brain is not fullyappreciated; however, unlike GDNF, GM-CSF is able to cross theBBB and accumulate at physiologically relevant levels within thebrain parenchyma (Franzen et al., 2004; McLay et al., 1997;Thomson and Lotze, 2003). Additionally, systemic exposure to GM-CSF can promote functional recovery following traumatic spinal cordinjury (Bouhy et al., 2006; Huang et al., 2009) and ischemia (Nakagawaet al., 2006; Schabitz et al., 2008).

It has been suggested that GM-CSF may be neuroprotective bypromoting microglia to adopt a dendritic cell-like state (Fischer andReichmann, 2001; Santambrogio et al., 2001; Schermer and Humpel,2002), facilitating the release of BDNF (Hayashi et al., 2009) andincreasing the expression of anti-apoptotic proteins such as Bcl-2(Huang et al., 2007; Kim et al., 2009). To this end, elucidating the pro-survivalmechanisms of GM-CSF against DA-targeting toxinsmay offernew therapeutic targets for PD. Thus, we sought to assess whethercentral or peripheral administration of GM-CSF would prevent theneurodegenerative consequences induced by exposure to paraquatalone or with LPS pre-treatment and whether GM-CSF's pro-survivaleffects are linked to changes in the glial microenvironment. In light ofthe potentially important role for BDNF in both PD and its co-morbidpathology (e.g. depression and cognitive disturbances), it was also ofinterest to assess whether the toxin treatments affected this trophicfactor and whether GM-CSF could inuence such changes.

Materials and methods

C57BL/6 male mice were purchased from Charles River Laboratories(LaPrairie, Quebec, Canada) at 1012 weeks of age and subsequentlysingle-housed in standard polypropylene cages (272114 cm), andmaintained on a 12-h light/dark cycle. Mice were housed in atemperature (21 C) and humidity controlled room, and providedwith an ad libitum diet of Ralston Purina mouse chow and water.Animals were acclimatized for a period of 1 week prior to thecommencement of experimental procedures. All experimental testparadigms were approved by the Carleton University Committee forAnimal Care and adhered to the guidelines outlined by the CanadianCouncil for the Use and Care of Animals in Research.

Experiment 1: assessment of the benecial effects of supra-nigraladministration of GDNF and GM-CSF following LPS-paraquat inducedpathology

Given that GM-CSF can diminish apoptotic cell death and stimulatethe release of BDNF, we sought to determine if a direct infusion of GM-CSF into the SNcwould protect DA neurons from the neurodegenerativeeffects of LPS and paraquat to a degree comparable to GDNF. To this end,mice underwent stereotaxic surgery (described in an ensuing section)and cannulae were implanted just above the SNc. Following 1 week ofconvalescence, animals were infused with a single dose of either saline

or LPS (0.1 g/2.0 l) and given a 2 day rest period. Thereafter, micereceived intraperitoneal (i.p.) injections of saline or paraquat (SigmaAldrich; 10 mg/kg), 3 times aweek for 3 consecutiveweeks (aparadigmpreviously found to reliably impact the survival of SNc DA neurons;Mangano and Hayley, 2009). At the beginning of each week (i.e.,immediately prior to the 1st, 4th and 7th paraquat or saline injection),mice (n=1620/group) received supra-SNc infusion (using the samecannulae that previously delivered LPS or saline) of saline, GDNF (R&DSystems; 1 g/2 l) or GM-CSF (R&D Systems; 10 ng/2 l) into the SNc.Finally, in order to assess SNcDAneuronal survival, 5 days following thenal paraquat or saline injection, a subset of mice (n=810/group)were perfused with 4% paraformaldehyde (PFA), brains post-xed for24 h and then cryoprotected for 3 days in a solution comprising 20%sucrose, 0.1 M PBS and 0.02% sodium azide (pH 7.4).

Immunohistochemical procedures: SNc DA neuronal survival andmicroglial reactivity

Frozen brains were cryostat-sectioned and free-oating coronalsections (20 m) of the SNc were mounted onto gelatin-coated slides.Anti-tyrosine hydroxylase (TH) antibody along with 1% cresyl violet(Sigma-Aldrich) as a counter-stain was used to detect SNc DAneurons. The primary antibody (TH) was diluted in a solutioncontaining 1% BSA, 0.5% Triton X-100, 0.05% Tween 20, and 0.05%sodium azide in 0.01 M PBS, pH 7.3; and sections were incubatedovernight at room temperature (1:3000, ImmunoStar, Hudson, WI,USA). The primary antibodies were visualized using a biotin rabbitanti-mouse (1:500, Cedarlane) secondary and horseradish peroxi-dase-conjugated streptavidin tertiary (1:500, Cedarlane). Sectionswere then further incubated with diaminobenzidine (DAB; Sigma-Aldrich) for 10 min, counterstained with cresyl violet (Sigma-Aldrich)and dehydrated with serial alcohol washes before cover-slipping withclearene (Surgipath).

Survival of SNc DA-producing cells was determined by way ofserial section analysis of the number of TH+ cells within the SNc, atbregma levels ranging from 3.08 to 3.40. Using a double blindprocedure, 810 animals were analyzed per group and the number ofTH+ cells was counted across multiple bregma levels. Midbrainsections from each bregma level were compared across treatmentgroups to assess the extent of DA neuronal loss induced by paraquatalone or primed with LPS, and to determine if GM-CSF or GDNFprevented such effects. In order to ensure that a genuine loss of DAneurons occurred rather than simply a phenotypic suppression, thetotal number of surviving SNc neurons (TH+ and TH) was countedusing cresyl violet staining.

Striatal sections were stained using TH+ antibody (1:1000) andphotomicrographs were obtained for each animal using the sameexposure time. Image J software was used to determine thebackground threshold for each striatal section and the total numberof white (background) and black (TH+) pixels. All images wereconverted into an 8-bit format, where the grayscale varied from 0 to255. The area of interest was selected and the upper and lowerthreshold values were used across all images to separate the featuresof interest from the background. The upper and lower thresholdswere determined using an automatic thresholding option, a modiedversion of the IsoData method. This algorithm divides the image intothe object of interest and background by taking an initial threshold(histogram-derived), followed by the averages of the pixels at orbelow the threshold and pixels above are computed. The averages ofthose two values are computed, the threshold is incremented and theprocess is repeated until the threshold is larger than the compositeaverage. The data were then presented as a histogram with thenumber of black (object of interest) and white (background) pixelspresent.

Morphological changes of microglia are generally taken to indicatesome degree of change in their activation states (Mangano and

Hayley, 2009). Indeed, differing degrees of compacted soma,

101E.N. Mangano et al. / Neurobiology of Disease 43 (2011) 99112thickening of processes, and enhanced proliferation are normallytaken to reect a state of activation, wherein microglia are capable ofsubstantial release of inammatory and oxidative factors (Nimmer-jahn et al., 2005). Hence, the morphological state of microglia wasassessed histologically using CD11b (complement receptor markerpresent on microglia) antiserum (1:1000; AbD Serotec, Raleigh, NC,USA; overnight incubation at 4 C). The primary antibody was dilutedin 1% BSA, 0.03% Triton X-100, 0.05% Tween 20 and 0.05% sodiumazide in 0.01 M PBS (pH 7.3). Thereafter, sections were washed with0.01 M PBS and incubated with biotin rabbit anti-rat secondary(1:500, Cedarlane) and horseradish peroxidase-conjugated strepta-vidin tertiary (1:500, Cedarlane) antibodies. Sections were once againvisualized by incubation with DAB (Sigma-Aldrich) for 10 min anddehydrated with serial alcohol washes before cover-slipping withclearene (Surgipath).

A semi-quantitative rating scale previously published by our lab(Mangano and Hayley, 2009) was used to verify the degree ofmicroglial reactivity. The morphological change in microglia wasanalyzed using a 03 rating scale. Specially, a rating of 0 was given ifthe SNc microglia appeared quiescent (highly ramied with manylong thin primary and secondary branches). A rating of 1 reected anintermediate reactivity state in which less than 10 cells within the SNccould be considered moderately activated (thickened, short processeswith a compact soma). A rating of 2 was given if the majority of themicroglial cells appeared to be intermediately activated withoccasional highly activated cells (amoeboid and macrophage-likeappearance, circular in shape lacking processes) and a rating of 3 wasgiven only when the majority of cells displayed the most highlyactivated amoeboid shape.

Western blot analyses of hippocampal BDNF levels

It has become clear that, besides the frank nigrostriatal neuronaldamage, PD is associated with pathology involving brain regionsimportant for neuroplasticity (Bruck et al., 2004; Jokinen et al., 2009).Hence, it was of interest to assess whether paraquat would affecthippocampal BDNF levels. Moreover, it is of clinical importance toevaluate whether GM-CSF administration can reverse such effects. Tothis end, a subset of the mice from Exp. 1 (n=810/group) wereeuthanized by rapid decapitation (again 5 days after the nalinjection) followed by micropunch dissection of the hippocampus.Tissue samples were stored at 80 C until Western blot analyses(Hayley et al., 1999; Hayley et al., 2004).

All chemicals were obtained from Sigma-Aldrich (St. Louis, MO)and all antibodies were obtained from Santa Cruz Biotechnology(Santa Cruz, CA) unless otherwise indicated. Samples were dilutedwith lysis and protease inhibitor buffer up to the desired proteinconcentration, yielding whole cell lysate (50 g) in 20 l and 20 lloading buffer (5% glycerol, 5% -mercaptoethanol, 3% SDS and 0.05%bromophenol blue). To denature the proteins, the 40 l sample washeated in boiling water for 5 min. Sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE), the separating buffer[370 mM Tris-base (pH 8.8), 3.5 mM SDS], and the stacking buffer[124 mM Tris-base (pH 6.8), 3.5 mM SDS], were placed in runningbuffer (25 mM Tris-base, 190 mM glycine, 3.5 mM SDS); and samples,alongwith the Precision Plus Protein Standards Dual Color (Bio-Rad,Hercules, CA, Cat #161-0374), were loaded into the Acrylamide gel(12.5%) for molecular weight determination at 120 V.

After electrophoresis, proteins were transferred overnight, at 4 Cand 180 mA, in transfer buffer (25 mM Tris-base, 192 mM Glycine,20% methanol) onto PVDF (Bio-Rad, Cat #162-177). Thereafter,membranes were blocked for 1 h with gentle shaking in a solutionof non-fat dry milk (5% w/v) dissolved in TBS-T buffer (10 mM Tris-base (pH 8.0), 150 mM sodium chloride, 0.5% Tween-20). Themembranes were then incubated with a rabbit anti-BDNF primary

antibody (1:500) diluted in blocking solution at room temperature for1 h. Any unbound antibody was removed using three washes of 15 mlTBS-T at room temperature. Membranes were incubated on a shakerfor 1 h at room temperature with HRP (horseradish peroxidase)conjugated anti-rabbit (1:2000) secondary antibody and washedagain with TBS-T. Finally, BDNF was visualized with a chemilumines-cent substrate (Perkin Elmer, Waltham, MA, cat #NEL102001EA; for1 min) and briey exposed on a Kodak lm.

Experiment 2: assessment of the benecial effects of systemicadministration of GM-CSF following LPS-paraquat induced pathology

Mice underwent stereotaxic surgery identical to Experiment 1 andreceived a single supra-nigral infusion of either saline or LPS (same doseused in Experiment 1), followed 2 days later with i.p. saline or paraquattreatment (10 mg/kg; 3 times per week for 3 weeks). Paralleling theintra-SNc infusions delivered in the previous experiment, mice receivedi.p. injections of GM-CSF (R&D Systems; 2 g) once per week for theduration of the 3 week paraquat regimen. All mice were sacriced5 days after the nal paraquat or saline injection and brains wereperfused for immunohistochemical analyses.

Immunohistochemical analyses of SNc neuronal loss

Stereological analysis of SNc sections (60 m) was used todetermine DA degeneration for Experiment 2. Briey, the SNc wasoutlined under 2.5 magnication and TH+ neurons counted using a60 oil immersion objective. The SNc was sampled in a systematicrandom fashion according to the optical fractionator method outlinedby MicroBrightField Inc. Cells were quantied in 3-dimensionalcounting frames using a counting grid size of 9090 m and acounting frame size of 6060 m with a 15 m dissector height and3 m upper and lower guard zones. Only the portion of the SNcipsilateral to the infusion site was quantied. All analyses wereconducted with the counter blind to the treatment conditions. StriatalDA terminal staining was conducted in a manner identical to that ofExperiment 1.

DA neurons were sampled throughout the rostralcaudal axis ofthe SNc in every 2nd section, labeled with TH and counterstained with1% cresyl violet. Free oating sections were blocked in 0.3% hydrogenperoxide solution for 30 min, washed in 10 mM PBS and thenincubated overnight at 4 C in mouse anti-TH primary antibody(1:500 in 10 mM PBSwith 2% BSA, 1% sodium azide and 0.5% Triton-X,Immunostar Inc.). This was followed by incubation in secondary biotinSP-conjugated Afnipure goat anti-mouse IgG (1:200 in 10 mM PBSwith 2% BSA, 1% sodium azide and 0.5% Triton-X, Jackson Immunor-esearch Laboratories) for 2 h at room temperature and, lastly,incubation in tertiary peroxidase-conjugated streptavidin (1:200 in10 mM PBS with 2% BSA and 0.3% Triton-X, Jackson ImmunoresearchLaboratories).

Following antibody incubations, sections were rinsed 3 times for5 min in 10 mM PBS. To visualize the antibody complex, sections wereincubated for 5 min in 1 ml (per well) of DAB (0.2 mg/ml) in 50 mMTris HCl solution. Sections were incubated for an additional 15 minfollowing addition of 50 l of 6% hydrogen peroxide (in distilledwater) per well. Sections were mounted on glass slides, air-driedovernight, counterstained with 1% cresyl violet, and dehydrated withserial alcohol washes. Slides were then cover-slipped and stored atroom temperature. TH+ cells were quantied as outlined above andrepresentative photomicrographs depicting TH+ neurons in the SNcwere taken using an Olympus BX52 microscope under magnicationof the 10 objective for qualitative analysis. The same free-oatingsections used to determine TH loss were also used to conrm thatneurons were actually lost and TH expression not simply reduced.Hence, the total number of SNc neurons was determined using the

cresyl violet counterstain.

102 E.N. Mangano et al. / Neurobiology of Disease 43 (2011) 99112Assessment of glial changes

Free-oating sections were incubated overnight at 4 C with eithermouse anti-GFAP (1:1000; Chemicon) to visualize astrocytes or ratanti-mouse CD11b (1:1000, Serotec) to visualize microglia. Allantibodies were diluted in a solution containing 10 mM PBS with 1%sodium azide and 0.3% Triton-X. Sections were then washed with10 mM PBS 3 times prior to incubating with their respectivesecondary biotinylated antibodies, namely goat anti-mouse and goatanti-rat (1:200, Jackson Immunoresearch Laboratories), for 2 h atroom temperature. Following 3 washes with 10 mM of PBS, thesections were then incubated in tertiary peroxidase-conjugatedstreptavidin (1:200, Jackson Immunoresearch Laboratories) for 2 hat room temperature. The antibody complexes were visualized byincubating with DAB for 10 min. Sections were mounted on glassslides, air-dried overnight, cover-slipped and stored at room temper-ature. Both astrocytic and microglial reactivity were qualitativelyassessed using a double blind procedure. Representative photomicro-graphs were taken on an Olympus BX52 microscope using a 40objective.

General stereotaxic surgical procedures for Experiments 1 and 2

All mice underwent stereotaxic surgery wherein indwelling cannu-lae (26-gauge stainless steel) were implanted just above the SNc(bregma: anterior-posterior 3.16 mm, lateral1.2 mm, ventral4.0 mm). Central administration of LPS or saline was achieved using aHarvard Apparatus Pico Plus syringe pump. A total of 2 l of uid wasdelivered to the SNc over a period of 5 min from polyethylene tubingconnected to a Hamilton microliter syringe. Infusions and injectionswere administered between 0830 h and 1400 h to minimize thepotentially confounding effects of circadian variation. It is importantto note that animals were unrestrained during the infusion procedure,allowing the mice to move freely about their home cage in order tominimize stress. Also,mice had a 1 week convalescence period betweencannulae implantation and administration of any treatment. This is animportant point to address given that inammatory processes aremarkedly affected by wounds resulting from surgical procedures andthe stress associated with restraint.

Experiment 3: in vitro assessment of the effects of paraquat and GM-CSF

Primary mesencephalic neuronglia cultures were prepared frombrains of embryonic day 18 (E18) C57Bl/6 mice. Whole brains wereextracted aseptically and the mesencephalon was isolated. Afterremoving the blood vessels andmeninges, pooledmidbrain tissuewasdissociated using a papain dissociation kit (Worthington BiochemicalCorp.). Briey, 500 l of papain solution was added to pooled tissueand mechanically homogenized and placed in an incubator for 1 h at37 C. Cultures were maintained at 37 C in a humidied atmosphereof 5% CO2, 95% air. Following dissociation, the cells were triturated 5times and centrifuged at 300g for 5 min. The cells were then re-suspended in a papain/digestion inhibitor solution and counted usinga hemacytometer (BrightLine; 394485). Cells were diluted insupplemented Neurobasal A media containing 2% B27, 1% N2,416 M L-glutamine, 41.6 U/ml penicillin and 41.6 g/ml streptomy-cin. Thereafter, cells were seeded in either 6-well plates (25.0104/ml) or 4-well slides (5000 cells/ml) pre-coated for 1 h with poly-D-lysine (20 g/ml; Sigma) and bronectin (20 g/l; Sigma).

Two days following seeding, the media was replaced with freshsupplemented Neurobasal A media alone or containing GM-CSF(250 ng/ml), paraquat (30 M) or paraquat (30 M)+GM-CSF(250 ng/ml), and incubated for 6 h. Importantly, the optimal paraquatconcentration and exposure time were empirically determined byperforming dose (0.3300 M) and time (124 h) course analyses. A

+dose and exposure time that reliably induced 3040% loss of THneuronswas chosen as this degree of cell death paralleled that observedunder in vivo conditions in the aforementioned experiments. Immedi-ately following the drug exposure, all treated cells were replaced withfresh supplementedmedia and those that had been incubatedwithGM-CSF received fresh GM-CSF for a further 24 h. Thereafter, mesencephaliccultures were scraped (for western blot analysis) or xed with 4% PFAfor 24 h for quantication of cell death. Hence, mesencephalic neuronswere exposed to paraquat for 6 h and GM-CSF for 30 h in total.

Immunocytochemistry

Cells were xed with 4% PFA for 1 h at room temperature in sterilePBS. Non-specic staining was blocked using 10% normal serumdiluent (containing 0.5% Triton X-100, 0.03% sodium azide in PBS) for1 h. The total neurons were visualized using NeuN (1:800; Chemicon)and DA neurons using TH (1:800; Chemicon) antibodies. The cultureswere incubatedwith these primary antibodies (diluted in the blockingsolution) overnight at 4 C. The following day the cultures werewashed 3 times with PBS for 10 min and incubated for 1-h withsecondary antibodies (1:200; Alexa 488 and Alexa 555). All imageswere analyzed using an up-right Nikon microscope.

Western blot

Western blot procedures were identical to those already describedfor Experiment 1, with two exceptions: 1) mesencephalic culturesinstead of hippocampal tissue were processed, and 2) in addition towhole cell lysate fractions for BDNF, sub-cellar fractions wereobtained for Bax and Bcl-2 proteins. Given the evidence that suggestsGM-CSF can affect apoptotic mitochondrial proteins, it was of interestto determine any changes in pro- and anti-apoptotic factors withinthe cytosolic and mitochondrial fractions. Specically, the pro-apoptotic factor, Bax, is translocated from the cytosol to themitochondrial membrane following activation, whereas the anti-apoptotic Bcl-2 is permanently associated with the mitochondrialmembrane. To this end, subcellular fractionationwas performed usinga kit purchased from Calbiochem (QIA88), which separated culturesinto cytosol and mitochondrial sub-fractions in order to determineBcl-2 and Bax protein concentrations. Briey, neuronal cell pelletswere suspended in 200 l of 1 cytosol extraction buffer containingprotease inhibitor cocktail and DTT. The cells were homogenizedusing a tissue grinder, centrifuged (700g for 10 min and 10,000g for30 min) to obtain the cytosolic fraction. The pelletswere re-suspendedin 100 l of mitochondrial extraction buffer containing proteaseinhibitors and sonicated in order to obtain the mitochondrial fraction.Thereafter, sampleswere dilutedwith lysis andprotease inhibitor buffer(as described earlier) up to the desired protein concentration: cytosolicfractions (50 g) and mitochondrial fraction (50 g) in 20 l and 20 lloading buffer (5% glycerol, 5% -mercaptoethanol, 3% SDS and 0.05%bromophenol blue). Finally, all sampleswere loadedonto an acrylamidegel, transferred to PDVF and visualized using chemiluminescenceprocedures that were once again identical to those of the earlier BDNFexperiment. Table 1 provides a list of the antibody dilutions and gelspecics. All antibodies were obtained from Santa Cruz Biotechnology(Santa Cruz, CA) except for anti-rabbit-HRP, whichwas purchased fromSigma Aldrich.

Statistical data analysis

All data were analyzed by ANOVA followed by Fisher's plannedcomparisons where appropriate. Data were evaluated using a Stat-View (version 6.0) statistical software package available from the SASInstitute, Inc. For the in vitro study, protein expression was quantiedby densitometry (AlphaEase FC v.3.1.2, Alpha Innotech, Co., SanLeandro, CA) and, unless otherwise indicated, all results are expressed

as meansS.E.M of at least three independent experiments.

Results

Experiment 1: Supra-nigral injection of GM-CSF or GDNF attenuatedLPS-paraquat induced degeneration of SNc dopamine neurons andstriatal terminals

The ANOVAs revealed that the number of surviving TH+ neuronsat several levels of the SNc varied as a function of the LPS, paraquat,and growth factor (GM-CSF or GDNF) treatments {F7, 52=6.727,F7, 33=3.072, F7, 46=3.278 and F7, 35=4.548 psb0.001; bregmalevels 3.08, 3.16, 3.28 and 3.40, respectively}. As depicted in

Fig. 1, paraquat treatment alone caused approximately 30% loss ofTH+ neurons across SNc levels. Consistent with our previous report(Mangano and Hayley, 2009), priming the SNc with LPS 2-days priorto commencing the pesticide regimen resulted in a greater loss of TH-expressing neurons (~4050% yet, this was not signicantly differentfrom the 30% loss induced by paraquat alone). Moreover, centraladministration of either GM-CSF or GDNF completely prevented theneuronal loss induced by paraquat alone or following LPS priming.Indeed, the follow up statistical comparisons conrmed that paraquatand LPS+paraquat administration signicantly reduced the survivalof TH+ neurons across all four bregma levels and infusion of eitherGM-CSF or GDNF prevented these reductions (pb0.01).

To determine whether the loss of TH+ neurons reected a genuineneurodegenerative effect or simply a phenotypic suppression, all SNcsections were counterstained with cresyl violet and the total numberof surviving neurons was quantied across bregma levels for each ofthe treatment groups. Not surprisingly, the ANOVA revealed asignicant difference in the total number of cresyl violet stainedneurons as a function of the treatments {F7, 52=5.009, F7, 33=3.102 ,F7, 46=3.036 and F7, 35=4.380 psb0.001; bregma levels3.08, 3.16,

Table 1List of antibody dilutions and gel specics used for Western blotting.

Primary antibody Catalogue # % SDS-PAGE gel Dilutions

anti-Bcl-2 HRP sc-492 HRP 15 1:250anti- Bax HRP sc-493 HRP 15 1:100anti-BDNF sc-546 12.5 1:500anti-Actin HRP sc-47778 HRP 1:5000anti-rabbit-HRP A6154 1:2000

103E.N. Mangano et al. / Neurobiology of Disease 43 (2011) 99112Fig. 1. Central administration of GM-CSF and GDNF protected SNc neurons. Dopaminergic neSNc. The top representative photomicrographs illustrate the degree of dopaminergic neuronexposure alone (10 mg/kg; 3 per week for 3 weeks) or following LPS priming (0.01 g/2 ltreatment (* pb0.05). However, a single supra-nigral injection of GDNF (1 g/2 l) or GM-CSFinduced by paraquat alone and with the addition of LPS. Data are expressed as meanSEMuron loss was assessed using TH (1:1000) antibody staining across several levels of theloss induced by LPS and paraquat, as well as the effects of GDNF and GM-CSF. Paraquat) of the SNc 2 days earlier induced a signicant loss of TH+ neurons, relative to saline(10 ng/2 l) once per week during the paraquat regimen prevented the SNc neuron loss

; n=810.

3.28 and 3.40, respectively}. Specically, in a manner identical tothat observed for TH+ neurons, paraquat reduced the total number ofSNc neurons and once again, this effect was somewhat augmented(albeit not signicantly) by LPS priming and was absent in mice thatreceived the GM-CSF or GDNF treatments (pb0.01; see Supplemen-tary data Figure). However, this effect appeared to be restricted to theTH+ neurons, as the number of TH- cresyl violet stained neurons werenot signicantly affected by any of the treatments.

It is important to underscore that cell countswere conducted usingaccepted manual procedures across multiple bregma levels of the SNc(as in Hayley et al., 2004). However in the second experiment of thismanuscript (to be presented shortly) we used more recentlyestablished stereological assessments to quantify cell loss. Thisdifference stems from the fact that we only had access to a properstereology set-up at the time when the second study was conducted.Nonetheless, we believe the current manual cell counts to be fullyvalid and accurate. In fact, we found the two methods to yield verysimilar results with regards to the basic neurodegenerative effects ofparaquat. Yet, we fully acknowledge the limitations of the manualsystem used in the experiment.

As was the case for the SNc, the LPS, paraquat and growth factortreatments altered TH+ immunoreactivity within the striatum{F7, 33=32.195 pb0.001}. In this regard, mice exposed to paraquatalone or primed with LPS displayed reduced striatal TH+ staining(pb0.05); and, once again, infusion of either GM-CSF of GDNFprevented this effect, see Fig. 2.

Supra-nigral injection of GM-CSF inuenced LPS and paraquat induced

Specically, semi-quantitative ratings conrmed that paraquat inducedamodest elevation of CD11b+ staining in comparisonwithmice treatedonly with saline, and prior LPS infusion further enhanced this effect{F7, 32=4.937 pb0.001}. The majority of "activated" microglia receiveda score of 1, as they displayed an intermediate level of reactivitycharacterized by shortened, thick dendritic processes. A few cells,particularly in response to LPS and paraquat treatment, appeared to beclumped together with more compact soma characteristic of a reactivestate, and thus received a score of 2 (Fig. 3). Surprisingly, a single supra-nigral injection of GDNF at the beginning of each week of the paraquatregimen did not appreciably inuence the impact of paraquat or LPSupon CD11b+ immunoreactivity. However, the weekly GM-CSF infu-sionsdid cause a change in the state ofmicroglial activation,wherein thetrophic cytokine greatly attenuated the morphological changes ofmicroglia that were provoked by paraquat and LPS, causing themicroglia to be more ramied and received lower ratings (see Fig. 3).

In contrast to the augmentedmicroglial response (at least in terms ofmorphology), astrocytic expression within the SNc (as indicated byGFAP+ immunoreactivity) was modestly reduced in mice exposed toLPS+paraquat, see Fig. 4. Although a somewhat surprising nding, wedid previously nd a similar attenuated GFAP response in LPS primedmice later exposed to paraquat (Mangano and Hayley, 2009). Impor-tantly, a single infusion of either of the growth factors, GDNF or GM-CSF,once per week throughout the paraquat regimen appeared to beassociatedwith a reverse of the LPS+paraquat inducedGFAP reduction.

Supra-nigral infusion of GM-CSF modied the paraquat inducedreduction of mature hippocampal BDNF

paricrne oGM

104 E.N. Mangano et al. / Neurobiology of Disease 43 (2011) 99112Fig. 2. Central administration of GM-CSF and GDNF attenuated the impact of LPS andanalyzing pixel intensity of sections stained with TH (1:1000). The representative photomber staining. Indeed, mice exposed to paraquat (10 mg/kg; 3 per week for 3 weeks) aloloss of striatal terminals (* pb0.05). A single supra-nigral injection of GDNF (1 g/2 l) orglial changes in the SNc

LPS and paraquat induced changes in microglial morphology thatparalleled the loss of TH+ neurons within the SNc. As previouslyobserved (Mangano and Hayley, 2009), paraquat administered alone orin LPS primed mice enhanced microglial reactivity (as indicated bymorphological changes and increased CD11b density) within the SNc.of TH+ber immunoreactivity. Data are expressed as meanSEM; n=810.aquat upon striatal dopaminergic terminals. Striatal DA ber loss was determined byographs illustrate that paraquat with or without LPS priming induced a reduction of TH+

r following priming 2-days earlier with LPS (0.01 g/2 l) displayed approximately 35%-CSF (10 ng/2 l) once per week during the paraquat regimen totally prevented this lossAs already alluded to, disturbances of neurotrophin-mediatedplasticitymightbe importantnot only for the cognitive decits observedin PD, but also the other frequent co-morbid psychiatric symptoms, suchas anxiety and depression (Litteljohn et al., 2009). To this end, it isimportant that the paraquat and GM-CSF treatments provoked altera-tions of hippocampal BDNF {F2, 10=8.597 pb0.01}, see Fig. 5.

105E.N. Mangano et al. / Neurobiology of Disease 43 (2011) 99112Specically, the mature 1314 kDa form of BDNF was reduced withinthe hippocampus following paraquat administration, relative to vehicletreatment (pb0.05); however, GM-CSF infusion attenuated this effect,such that the growth factor did not differ from controls. Owing to tissuelimitations, the Western blot analyses were restricted to only thevehicle, paraquat and GM-CSF+paraquat treatment groups.

Experiment 2: systemic administration of GM-CSF had neuroprotectiveconsequences

As in Experiments 1, the ANOVA revealed a signicant differencebetween the treatment groups in terms of the number of survivingTH+ neurons within the SNc {F3, 12=79.701 pb0.0001}. The follow-up comparisons indicated that chronic paraquat exposure signi-cantly reduced the number of TH+ SNc neurons (~30% reduction)relative to vehicle-treated mice (pb0.01), and this effect was furthermodestly but non-signicantly increased when animals were pre-treated with LPS before exposure to the pesticide (N40% reduction).Paralleling the results of supra-nigral infusion in Experiment 1, i.p.GM-CSF injection appeared to attenuate the impact of LPS andparaquat, such that the number of surviving SNc TH+ neurons didnot signicantly differ from those that received vehicle alone, seeFig. 6. Once again, the impact of paraquat appeared to be restrictedto DA neurons; indeed, although the total number of cresyl violetstained neurons was reduced, no such reduction was observed forspecic TH- cresyl violet neurons (data not shown).

Fig. 3. Central GDNF and GM-CSF modulated the impact of LPS and paraquat upon SNcimmunostaining within the SNc of mice that received supra-nigral saline or LPS infusion togGM-CSF (10 ng/2 l). The above photomicrographs taken using a 40 objective demonstratethe SNc. Interestingly, although GM-CSF infusion blunted the impact of LPS and paraquat (marker. This was conrmed using a semi-quantitative rating scale (graph below); paraquat aappear to alter this effect, GM-CSF treated mice were comparable from saline treated contrAgain paralleling the earlier ndings, LPS, paraquat and GM-CSFtreatments affected TH+ immunoreactivity within the striatum{F3, 14=31.592 pb0.0001}. As depicted in Fig. 6, mice exposed toparaquat alone or primed with LPS displayed reduced striatal TH+

immunostaining (pb0.05).As was the case for central infusion of GM-CSF, systemic adminis-

tration of the trophic cytokine appeared to blunt the CD11b+changesinduced by LPS and paraquat (Fig. 7). Indeed, semi-quantitative ratingrevealed that CD11b+cell morphological signs of activation varied asa function of the treatments {F3, 11=14.65 pb0.01}. As can be seen inFig. 7 and conrmed by the planned comparisons, paraquat exposurealone or in the context of LPS pre-treatment increased ratings ofmicroglia reactivity (relative to saline treatment; pb0.01), whereas GM-CSF treated animals were indistinguishable from saline controls. Onceagain, as was observedwith central GM-CSF infusion, systemic exposureto the cytokine augmented GFAP immunoreactivity, whereas LPS andparaquat appeared tomoderately suppress the astrocyticmarker (Fig. 8).

Experiment 3: GM-CSF neuroprotection was associated with normalizedBDNF but not Bcl-2 or Bax levels

To evaluate the mechanism utilized by GM-CSF to protect SNc DAneurons from paraquat toxicity, primary mesencephalic neurongliamixed cultures were assessed following paraquat exposure. A doseresponse curve and time dependent evaluation of paraquat toxicityrevealed the dose that provoked a degree of cell death similar to that

microglia. The representative photomicrographs depict microglia (CD11b+; 1:1000)ether with systemic paraquat treatment, in presence or absence of GDNF (1 g/2 l) orthat paraquat and LPS+parquat treatments elevated CD11b+ immunostaining withinin terms of CD11b+staining intensity), GDNF had no such inuence on the microglialnd LPS+paraquat hadmarked effects onmicroglia morphology. Although GDNF did notols.

106 E.N. Mangano et al. / Neurobiology of Disease 43 (2011) 99112observed in vivo, see Fig. 9. Specically, the dose (0.350 M) and time(312 h) course analyses revealed that exposure to 30 M ofparaquat for 6 h provided optimal conditions, reliably inducing 3040% loss of TH+ neurons. Indeed, further experiments conductedusing these parameters revealed signicant differences in the numberof surviving TH+ neurons in mesencephalic cultures as a function ofpesticide exposure {F3, 16=13.048 pb0.0001}. As shown in Fig. 9,paraquat provoked a signicant loss of cultured mesencephalic TH+

neurons (pb0.01); however, again paralleling the in vivo ndings, thiseffect was completely prevented by co-treatment with GM-CSF(250 ng/ml).

Given that GM-CSF has been reported to impart protectiveconsequences through its effects upon trophic and apoptotic path-ways (Huang et al., 2007), the trophic factor, BDNF, and the pro- andanti-apoptotic factors, Bax and Bcl-2, respectively, were assessedusing Western blot. To this end, the ANOVAs revealed signicantdifferences between the treatment groups for BDNF and Bcl-2 (but notBax) protein levels within the mesencephalic cultures {F2, 4=6.412,F2, 9=3.881, respectively, pb0.05}. As depicted in Fig. 9 andconrmed by the follow up comparisons, paraquat signicantlyreduced levels of BDNF and Bcl-2 within the cultured neurons

Fig. 4. Central GDNF and GM-CSF modulated the impact of LPS and paraquat upon SNcimmunostaining within the SNc of mice that received supra-nigral saline or LPS infusion togGM-CSF (10 ng/2 l). The above photomicrographs (A) show that GFAP+ immunoreactivitparaquat, relative to saline treatment. This can be better observed with the higher magnicaCSF infusions appeared to reverse the effects of LPS and paraquat and in fact, GM-CSF treat(pb0.05). Co-treatment with GM-CSF restored BDNF protein expres-sion within the mesencephalic cultures; however, the growth factordid not appear to affect Bcl-2 protein expression.

Discussion

Inammatory and oxidative processes, together with reducedtrophic support, are widely considered to be essential players in thepathological processes of PD (Bossers et al., 2009; Chauhan et al.,2001; Guerini et al., 2009; Howells et al., 2000; Masaki et al., 2003).Indeed, accumulating evidence suggests that an augmentedmicroglialresponse, possibly driven by pro-inammatory cytokines such as IFN- and TNF- contributes to the progression of PD (Hirsch and Hunot,2009; Whitton, 2007). In this regard, the release of oxidative species,along with enhanced inammatory enzyme activity (e.g., COX-2) andthe induction of various MAP kinase pathways likely representproximal mediators of DA neuronal pathology (Gao et al., 2003;Hunot et al., 2004). In contrast to the more pro-inammatory role ofmicroglia, astrocytes have more frequently been associated with aprotective, buffering capacity and the release of trophic factors (Chenet al., 2006). At the same time, several reports detected reductions of

astrocytes. The representative photomicrographs depict astrocyte (GFAP+; 1:1000)ether with systemic paraquat treatment, in presence or absence of GDNF (1 g/2 l) ory was moderately diminished by LPS priming followed 2 days later with exposure totion (40) images in the below photomicrographs (B). Once again, the GDNF and GM-ment appeared to increase GFAP staining slightly above that of saline treated controls.

107E.N. Mangano et al. / Neurobiology of Disease 43 (2011) 99112anti-oxidants, particularly glutathione, as well as the trophic factor,BDNF, occur in the PD brain (Fitzmaurice et al., 2003; Murer et al.,2001). Whatever the case, increasing attention has been devoted toutilizing novel means of interfering with the pro-death process inanimal models of PD. To this end, we report that both central andperipheral administration of the trophic cytokine, GM-CSF, protectedSNc DA neurons from LPS and paraquat toxicity, andmodulated BDNF,microglia and astrocyte responses to these toxins.

Neurotrophic factors are essential for the maintenance and survivalof neuronal populations. Consequently, both BDNF andGDNF have beenextensively studied as potential therapeutic agents for PD patients(Peterson and Nutt, 2008). BDNF is normally present in relatively high

Fig. 5. Central administration of GM-CSF inuenced the effects of paraquat onhippocampal BDNF expression. The Western blot revealed that paraquat promoted amarked reduction of hippocampal BDNF protein. Four representative bands are shownfrom the three treatment groups. Clearly, paraquat caused a reduction of BDNF in eachof the animals exposed, whereas the addition of GM-CSF (10 ng/2 l) infusion generallyattenuated this effect (at least in 3 of the 4 animals). The bar graph on the bottomconrmed that these effects were statistically signicant (*pb0.05, relative to salinetreated). All integrated density scores were normalized against b-actin to control forany variations in loading. Data are expressed as meanSEM; n=4.concentrations within SNc DA neurons and is responsible for supplyingnutritive support, as well as promoting plastic and pro-survivalprocesses that have the potential to inhibit or possibly remediateneuronal pathology (Baquet et al., 2005;Murer et al., 2001). The trophicfactor has well known pro-survival and differentiation effects onmesencephalic DA neurons, and was reported to protect against MPTPand 6-OHDA insults (Hung and Lee, 1996; Shults et al., 1995). Inagreement with the nding that BDNF mRNA levels are reduced withinthe SNc of post-mortem PD brain (Mogi et al., 1999; Salehi andMashayekhi, 2009)we presently report that paraquat diminished BDNFlevels in mesencephalic cultures. Moreover, in vitro application of GM-CSF blunted this effect andprevented the loss of culturedmesencephalicTH+ neurons induced by paraquat. These ndings are consistent withreports showing that GM-CSF can augment BDNF to combat spinal cordinjury (Bouhy et al., 2006; Hayashi et al., 2009); and, taken together,suggest that BDNF could be contributing to some of the neuroprotectiveactions of GM-CSF in response to diverse injury-inducing stimuli.

In addition to BDNF, GDNF acts on dopaminergic neurons toenhance their morphological differentiation and survival (Lin et al.,1993), which has led to much excitement regarding the potential ofGDNF as a therapeutic agent for PD. Indeed, numerous rodent andprimate studies reported that GDNF attenuated the neurodegenera-tive effects of MPTP and further boosted DA functioning of the existingneurons (Gash et al., 1996; Kearns and Gash, 1995; Tomac et al.,1995). Importantly, the GDNF-induced restoration of DA functioningwas apparent even months after trophic support was terminated(Grondin et al., 2002; Kirik et al., 2004; Maswood et al., 2002). Despitethe overwhelming evidence supporting the benets of GDNF, it haslargely failed in human clinical trials due to complications arising fromthe fact that GDNF is unable to cross the BBB. Indeed, intracerebro-ventricular injection of GDNF was reported to cause problems relatedto (limited) diffusion into the brain parenchyma and the manifesta-tion of debilitating side-effects (Kordower et al., 1999; Nutt et al.,2003).

To circumvent the fact that trophic factors generally do not cross theBBB, new techniques are currently being developed to facilitate centralpenetration (Juillerat-Jeanneret and Schmitt, 2007; Pardridge, 2002);these include encapsulation or breaking GDNF into small bioactivefragments (Peleshok and Saragovi, 2006). However, nding novelneurotrophic factors that easily cross the BBB (and thus avoid suchcomplications entirely) may provide a new class of therapeutic agentsfor PD. In this regard, GM-CSF is one of three cytokines belonging to afamily of colony stimulating factors that have trophic effects and arecapable of crossing the BBB and accumulating in the brain parenchymaat reasonable levels (Franzen et al., 2004; McLay et al., 1997; Thomsonand Lotze, 2003). The exact mechanism by which GM-CSF permeatesthe BBB is presently unclear; however, based on experiments whichtagged GM-CSF prior to systemic administration to rodents, it is clearthat this cytokine can, indeed, gain entry into the brain (McLay et al.,1997).

GM-CSF has well documented clinical efcacy in the periphery,including its use as an immune restorative agent in certain cancertreatments (owing to its ability to stimulate leukocyte production);however, minimal attention has been devoted to elucidating thetrophic cytokine's potential as a neurotrophic agent. Albeit, a fewreports have indicated that GM-CSF had neurorestorative functions inthe spinal cord and even assisted in cerebral ischemic recovery (Ha etal., 2005; Hayashi et al., 2009; Nakagawa et al., 2006; Schabitz et al.,2008). There is also some evidence to suggest that GM-CSF canpromote the release of BDNF from microglia (Bouhy et al., 2006;Hayashi et al., 2009). In addition, GM-CSF has been shown to down-regulate the IFN- mediated induction of MHC-II expression onmicroglia (Hayashi et al., 1993). This is important since IFN- wasrecently shown to play a primary role in the degeneration of DAneurons following paraquat or MPTP exposure (Mount et al., 2007).These ndings, in combination with the fact that GM-CSF mightpromote a phenotypic shift in microglial functioning towards a moresupportive and less inammatory role (Bouhy et al., 2006; Re et al.,2002), prompted us to hypothesize that GM-CSFmight protect againstDA acting toxins by modulating the inammatory balance betweenmicroglial and astrocytic cells.

Consistent with reports showing that GM-CSF might affectneuronal survival by inducing microglia to adopt a dendritic cell-like morphology (Reddy et al., 2009), exposure to the growth factor inthe current study appreciably diminished the impact of LPS andparaquat upon microglia (as indicated by staining intensity).However, the well established neurotrophin, GDNF, did not appre-ciably affect SNc microglia staining (at least according to CD11b+ cellmorphology and proliferation), despite the fact that the growth factorprevented DA neuron loss to a degree comparable to that of GM-CSF.These ndings raise the possibility that the two trophic factors mightbe exerting some differential effects within the SNc.

In contrast to the heightened microglia response induced by LPSand paraquat, these insults seemed to reduce the SNc astrocyticresponse (as indicated by GFAP+ immunostaining). Interestingly,both GM-CSF and GDNF reversed the astrocytic reduction induced bythe toxins. This might be particularly important given that anastrocyte reduction would likely result in diminished availability oftrophic support (e.g., BDNF and GDNF release) and reduced glutamatebuffering capabilities (Malipiero et al., 1990). Although GM-CSFreceptors are certainly expressed on astrocytes (Guillemin et al.,1996; Malipiero et al., 1990), they have also been found on neurons

throughout the brain, including DA neurons of the SNc (Kim et al.,

108 E.N. Mangano et al. / Neurobiology of Disease 43 (2011) 991122009). Hence, while GM-CSF might conceivably be promotingneuronal survival through direct modulatory effects upon microgliaand astrocytes, it is also possible that these glial changes are simplysecondary to the direct impact of the trophic cytokine upon DAneurons (or other yet to be identied processes involved in theneurodegenerative response).

The capacity of GM-CSF to promote anti-apoptotic signaling(Schabitz et al., 2008) could be important in its ability to attenuatetoxin induced DA neuronal loss. Indeed, a recent study found that GM-CSF promoted Bcl-2 expression following MPTP exposure (Kim et al.,2009). However, in the present investigation, the growth factor didnot inuence Bcl-2 expression in primary mesencephalic cultures.This discrepancy may stem from the fact that the previous studyassessed Bcl-2 within PC12 cells (Kim et al., 2009), whereas wepresently assessed primarymesencephalic levels of the anti-apoptoticfactor. It also might be the case that the nature of toxin used (MPTP vs.paraquat) is relevant in this regard.

Fig. 6. Systemic GM-CSF treatment protected nigrostriatal neurons (both SNc soma and stdopamine (DA) neurons was determined using stereological cell counts of TH+ (1:1000)degree of TH+ cell loss induced by LPS and paraquat, as well as the protective effects of GM2 days prior to commencing the paraquat regimen (10 mg/kg; 3 per week for 3 weeks)mod25% (paraquat alone) to around 40% (*pb0.05). Moreover, as shown in the top right photombres (*pb0.05), however, LPS priming did not further augment this effect whatsoever. The bapparent protective effects of systemic (i.p.) administration of GM-CSF (2 g). Finally, lowmof the location of TH+terminal loss within the region. Data are expressed as meanSEM;In light of the fact that depression and other co-morbid symptomstypically occur together with the motor disturbances in PD (Farabaughet al., 2009), it was particularly interesting that paraquat reducedhippocampal BDNF expression. Indeed, substantial evidence hasindicated that reduced BDNF levels can contribute to depressive-likebehavioral effects, aswell as decits in hippocampal neuroplasticity andcognitive functioning (Anisman et al., 2008; Heldt et al., 2007; Schmidtand Duman, 2007). Similarly, other recent studies showed that BDNFadministration in mice promoted a range of anti-depressant and anti-anxiety-like effect (e.g., as assessed in forced swim, anhedoniaparadigms), as well as promoting hippocampal neurogenesis (Gourleyet al., 2008; Schmidt and Duman, 2010).

While the precise mechanisms and neural substrates underlyingthe co-morbid non-motor symptoms in PD have yet to be fullyestablished, it is possible that inammatory and/or neuroplasticprocesses in stressor-sensitive cognitive and emotional brain regions(e.g., hippocampus, prefrontal cortex, locus coeruleus) are important.

riatal terminals) from LPS and paraquat induced dopaminergic cell death. The loss ofneurons through the SNc. The top left representative photomicrographs illustrate the-CSF. Once again, priming the SNc with a single supra-nigral injection LPS (0.01 g/2 l)estly (albeit not signicantly) enhanced the degree of neuronal loss from approximatelyicrographs, mice exposed to the paraquat regimen displayed a signicant loss of striatalottom bar graphs show the quantication of SNc soma and striatal terminal loss and theagnication (2.5) images of the striatum are shown on the bottom to give a better idean=5.

Indeed, histological pathology, including accumulation of -synu-clein, occurredwithin the hippocampus of PD patients (Bertrand et al.,2003; Galvin et al., 1999). Moreover, paraquat altered hippocampal

monoamine activity, and promoted depressive- and- anxiety-likeresponses in mice (Litteljohn et al., 2009). The present investigationprovides evidence that paraquat, and possibly other environmental

Fig. 7. Systemic GM-CSF administration modulated the effects of LPS and paraquat upon SNc microglia. A. Paraquat and LPS+paraquat robustly induced microglia (CD11b+)immunostaining within the SNc. Higher magnication (40 Objective; B) better illustrates the morphological differences between the different groups. Furthermore differences inmicroglial morphology were conrmed using a semi-quantitative rating scale (C; see methods for details on the microglial rating procedures).

109E.N. Mangano et al. / Neurobiology of Disease 43 (2011) 99112Fig. 8. Systemic GM-CSF administration modulated the effects of LPS and paraquat upon SNceffects of LPS and paraquat on microglial cells, LPS primed mice that received paraquat displaParalleling the aforementioned effects apparent with central GM-CSF administration, systempopulation. Panel B reveals higher magnication (40) images to give a better idea of morastrocytes. A. The representative photomicrographs (20) reveal that in contrast to theyed a reduction of astrocyte (GFAP+) staining, whereas paraquat alone had little effect.ic injection of the trophic cytokine blunted the impact of LPS and paraquat upon the glialphology of the GFAP+cells.

110 E.N. Mangano et al. / Neurobiology of Disease 43 (2011) 99112toxins, could potentially affect emotional and cognitive processing bycausing trophic factor reductions in hippocampal pathways acting inparallel with the nigrostriatal system. Importantly, GM-CSF appearedto reverse the paraquat-induced hippocampal BDNF reduction,suggesting that its protective effects were not specic for nigrostriatalfunctioning but also translated across brain regions.

Although the current ndings do not fully elucidate a mechanismof action, it is clear that the neuroprotective effects of both systemicand central GM-CSF administration are accompanied by a modulationof the neuroinammatory glial responses provoked by LPS andparaquat. This is important in light of the accumulating evidencesuggesting an important role for glial-driven inammation in thedeath of DA neurons. However, disturbances of mitochondrialfunctioning, generation of oxidative radicals and promotion ofapoptotic factors are probably the most proximal events in theneurodegenerative process. In this regard, our data did not support arole for the classical Bax and Bcl-2 apoptotic pathways in the presentLPS-paraquat model or the protective effects of GM-CSF. Yet, it isparticularly notable that paraquat reduced BDNF within both thehippocampus, as well as cultured mesencephalic neurons, and thatGM-CSF moderately reversed both of these effects. In summary,paraquat (particularly in the context of inammatory priming withLPS) could conceivably contribute to motor and non-motor (e.g.,depression, cognitive disturbance) PD symptoms by enhancinginammatory processes and altering neuroplasticity, and GM-CSFmight have important mitigating effects in this regard.

Fig. 9. GM-CSF restored hippocampal BDNF expression inmidbrain cultures exposed to paraq(30 M) exposure for 6 h induced a signicant reduction in BDNF and Bcl-2 levels in primaryml) partially prevented the reduction of BDNF but had no effect on Bcl-2 levels within the(varying from3 to 12 h)were conducted in order to determinewhich dose and time of paraquhours reliably caused a 40% reduction in midbrain DA neurons. (C) Paralleling the aforemenmidbrain cultures (pb0.05) and once again, GM-CSF reversed this effect (bottom right bar gSupplementarymaterials related to this article can be found onlineat doi: 10.1016/j.nbd.2011.02.011.

Acknowledgments

This work was supported by funds from the Canadian Institutes ofHealth Research (CIHR) and Parkinson's Society Canada. S.H. is aCanadian Research Chair in Behavioural Neuroscience.

References

Anisman, H., et al., 2008. Neurotransmitter, peptide and cytokine processes in relationto depressive disorder: comorbidity between depression and neurodegenerativedisorders. Prog. Neurobiol. 85, 174.

Baquet, Z.C., et al., 2005. Brain-derived neurotrophic factor is required for theestablishment of the proper number of dopaminergic neurons in the substantianigra pars compacta. J. Neurosci. 25, 62516259.

Bertrand, E., et al., 2003. Degenerative axonal changes in the hippocampus andamygdala in Parkinson's disease. Folia Neuropathol. 41, 197207.

Bossers, K., et al., 2009. Analysis of gene expression in Parkinson's disease: possibleinvolvement of neurotrophic support and axon guidance in dopaminergic celldeath. Brain Pathol. 19, 91107.

Bouhy, D., et al., 2006. Delayed GM-CSF treatment stimulates axonal regeneration andfunctional recovery in paraplegic rats via an increased BDNF expression byendogenous macrophages. FASEB J. 20, 12391241.

Bruck, A., et al., 2004. Hippocampal and prefrontal atrophy in patients with early non-demented Parkinson's disease is related to cognitive impairment. J. Neurol.Neurosurg. Psychiatry 75, 14671469.

Cannon, J.R., et al., 2009. A highly reproducible rotenone model of Parkinson's disease.Neurobiol. Dis. 34, 279290.

uat. (A) As depicted by the representative blots and conrmed by densitometry, paraquatmidbrain tissue (*pb0.05). When applied concomitant with paraquat, GM-CSF (250 ng/midbrain. (B) A dose response (varying from 3 to 50 M of paraquat) and time courseat exposurewould yield approximately 40% ofmidbrain neurons. 30 Mof Paraquat for 6tioned in vivo data, paraquat exposure reduced the number of TH+ neurons within theraph). Data are expressed as meanSEM; n=23 (BDNF) and n=45 (Bcl-2).

111E.N. Mangano et al. / Neurobiology of Disease 43 (2011) 99112Carvey, P.M., et al., 2006. Progressive dopamine neuron loss in Parkinon's disease: themultiple hit hypothesis. Cell Transplant. 15, 239250.

Chauhan, N.B., et al., 2001. Depletion of glial cell line-derived neurotrophic factor insubstantia nigra neurons of Parkinson's disease brain. J. Chem. Neuroanat. 21,277288.

Chen, P.S., et al., 2006. Valproate protects dopaminergic neurons in midbrain neuron/glia cultures by stimulating the release of neurotrophic factors from astrocytes.Mol. Psychiatry 11, 11161125.

Cicchetti, F., et al., 2009. Environmental toxins and Parkinson's disease: what have welearned from pesticide-induced animal models? Trends Pharmacol. Sci. 30,475483.

Davignon, J.L., et al., 1988. Selective production of interleukin 3 (IL3) and granulocyte-macrophage colony-stimulating factor (GM-CSF) in vitro by murine L3T4+ T cells:lack of spontaneous IL3 and GM-CSF production by Ly-2-/L3T4- lpr subset. Eur. J.Immunol. 18, 13671372.

Di Monte, D.A., 2003. The environment and Parkinson's disease: is the nigrostriatalsystem preferentially targeted by neurotoxins? Lancet Neurol. 2, 531538.

Eslamboli, A., 2005. Assessment of GDNF in primate models of Parkinson's disease:comparison with human studies. Rev. Neurosci. 16, 303310.

Farabaugh, A.H., et al., 2009. Pattern of depressive symptoms in Parkinson's disease.Psychosomatics 50, 448454.

Fischer, H.G., Reichmann, G., 2001. Brain dendritic cells and macrophages/microglia incentral nervous system inammation. J. Immunol. 166, 27172726.

Fitzmaurice, P.S., et al., 2003. Nigral glutathione deciency is not specic for idiopathicParkinson's disease. Mov. Disord. 18, 969976.

Franzen, R., et al., 2004. Nervous system injury: focus on the inammatory cytokine'granulocyte-macrophage colony stimulating factor'. Neurosci. Lett. 361, 7678.

Galvin, J.E., et al., 1999. Axon pathology in Parkinson's disease and Lewy body dementiahippocampus contains alpha-, beta-, and gamma-synuclein. Proc. Natl. Acad. Sci. U.S. A. 96, 1345013455.

Gao, H.M., et al., 2003. Critical role for microglial NADPH oxidase in rotenone-induceddegeneration of dopaminergic neurons. J. Neurosci. 23, 61816187.

Gash, D.M., et al., 1996. Functional recovery in parkinsonian monkeys treated withGDNF. Nature 380, 252255.

Gourley, S.L., et al., 2008. Acute hippocampal brain-derived neurotrophic factor restoresmotivational and forced swim performance after corticosterone. Biol. Psychiatry64, 884890.

Grondin, R., et al., 2002. Chronic, controlled GDNF infusion promotes structural andfunctional recovery in advanced parkinsonian monkeys. Brain 125, 21912201.

Guerini, F.R., et al., 2009. BDNF Val66Met polymorphism is associated with cognitiveimpairment in Italian patients with Parkinson's disease. Eur. J. Neurol. 16,12401245.

Guillemin, G., et al., 1996. Granulocyte macrophage colony stimulating factorstimulates in vitro proliferation of astrocytes derived from simian mature brains.Glia 16, 7180.

Ha, Y., et al., 2005. Synthes Award for Resident Research on Spinal Cord and SpinalColumn Injury: granulocyte macrophage colony stimulating factor (GM-CSF)prevents apoptosis and improves functional outcome in experimental spinal cordcontusion injury. Clin. Neurosurg. 52, 341347.

Hartmann, A., et al., 2002. FADD: A link between TNF family receptors and caspases inParkinson's disease. Neurology 58, 308310.

Hayashi, M., et al., 1993. Granulocyte-macrophage colony stimulating factor inhibitsclass II major histocompatibility complex expression and antigen presentation bymicroglia. J. Neuroimmunol. 48, 2332.

Hayashi, K., et al., 2009. Activation of dendritic-like cells and neural stem/progenitorcells in injured spinal cord by GM-CSF. Neurosci. Res. 64, 96103.

Hayley, S., et al., 1999. Sensitization to the effects of tumor necrosis factor-alpha:neuroendocrine, central monoamine, and behavioral variations. J. Neurosci. 19,56545665.

Hayley, S., et al., 2004. Regulation of dopaminergic loss by Fas in a 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine model of Parkinson's disease. J. Neurosci. 24,20452053.

Heldt, S.A., et al., 2007. Hippocampus-specic deletion of BDNF in adult mice impairsspatial memory and extinction of aversive memories. Mol. Psychiatry 12,656670.

Hercus, T.R., et al., 2009. The granulocyte-macrophage colony-stimulating factorreceptor: linking its structure to cell signaling and its role in disease. Blood 114,12891298.

Hirsch, E.C., Hunot, S., 2009. Neuroinammation in Parkinson's disease: a target forneuroprotection? Lancet Neurol. 8, 382397.

Howells, D.W., et al., 2000. Reduced BDNF mRNA expression in the Parkinson's diseasesubstantia nigra. Exp. Neurol. 166, 127135.

Huang, X., et al., 2007. GM-CSF inhibits apoptosis of neural cells via regulating theexpression of apoptosis-related proteins. Neurosci. Res. 58, 5057.

Huang, X., et al., 2009. GM-CSF inhibits glial scar formation and shows long-termprotective effect after spinal cord injury. J. Neurol. Sci. 277, 8797.

Huang, R., et al., 2010. Gene therapy using lactoferrin-modied nanoparticles in arotenone-induced chronic Parkinson model. J. Neurol. Sci. 290, 123130.

Hung, H.C., Lee, E.H., 1996. The mesolimbic dopaminergic pathway is more resistantthan the nigrostriatal dopaminergic pathway to MPTP and MPP+toxicity: role ofBDNF gene expression. Brain Res. Mol. Brain Res. 41, 1426.

Hunot, S., et al., 2004. JNK-mediated induction of cyclooxygenase 2 is required forneurodegeneration in amouse model of Parkinson's disease. Proc. Natl. Acad. Sci. U.S. A. 101, 665670.

Jenner, P., 1998. Oxidative mechanisms in nigral cell death in Parkinson's disease. Mov.Disord. 13 (Suppl 1), 2434.Jokinen, P., et al., 2009. Impaired cognitive performance in Parkinson's disease is relatedto caudate dopaminergic hypofunction and hippocampal atrophy. ParkinsonismRelat. Disord. 15, 8893.

Juillerat-Jeanneret, L., Schmitt, F., 2007. Chemical modication of therapeutic drugs ordrug vector systems to achieve targeted therapy: looking for the grail. Med. Res.Rev. 27, 574590.

Kearns, C.M., Gash, D.M., 1995. GDNF protects nigral dopamine neurons against 6-hydroxydopamine in vivo. Brain Res. 672, 104111.

Kim, N.K., et al., 2009. Granulocyte-macrophage colony-stimulating factor promotessurvival of dopaminergic neurons in the 1-methyl-4-phenyl-1,2,3,6-tetrahydro-pyridine-induced murine Parkinson's disease model. Eur. J. Neurosci. 29, 891900.

Kirik, D., et al., 2004. Localized striatal delivery of GDNF as a treatment for Parkinsondisease. Nat. Neurosci. 7, 105110.

Knott, C., et al., 2000. Inammatory regulators in Parkinson's disease: iNOS, lipocortin-1, and cyclooxygenases-1 and 2. Mol. Cell. Neurosci. 16, 724739.

Kordower, J.H., et al., 1999. Clinicopathological ndings following intraventricular glial-derived neurotrophic factor treatment in a patient with Parkinson's disease. Ann.Neurol. 46, 419424.

Lang, A.E., et al., 2006. Randomized controlled trial of intraputamenal glial cell line-derived neurotrophic factor infusion in Parkinson disease. Ann. Neurol. 59,459466.

Levivier, M., et al., 1995. Intrastriatal implantation of broblasts genetically engineeredto produce brain-derived neurotrophic factor prevents degeneration of dopami-nergic neurons in a rat model of Parkinson's disease. J. Neurosci. 15, 78107820.

Lin, L.F., et al., 1993. GDNF: a glial cell line-derived neurotrophic factor for midbraindopaminergic neurons. Science 260, 11301132.

Litteljohn, D., et al., 2009. Interferon-gamma deciency modies the motor and co-morbid behavioral pathology and neurochemical changes provoked by thepesticide paraquat. Neuroscience 164, 18941906.

Liu, B., et al., 2003. Parkinson's disease and exposure to infectious agents and pesticidesand the occurrence of brain injuries: role of neuroinammation. Environ. HealthPerspect. 111, 10651073.

Malipiero, U.V., et al., 1990. Production of hemopoietic colony-stimulating factors byastrocytes. J. Immunol. 144, 38163821.

Mangano, E.N., Hayley, S., 2009. Inammatory priming of the substantia nigrainuences the impact of later paraquat exposure: Neuroimmune sensitization ofneurodegeneration. Neurobiol. Aging 30, 13611378.

Masaki, T., et al., 2003. Association between a polymorphism of brain-derivedneurotrophic factor gene and sporadic Parkinson's disease. Ann. Neurol. 54,276277.

Maswood, N., et al., 2002. Effects of chronic intraputamenal infusion of glial cell line-derived neurotrophic factor (GDNF) in aged Rhesus monkeys. Neurobiol. Aging 23,881889.

McCoy, M.K., et al., 2008. Intranigral lentiviral delivery of dominant-negative TNFattenuates neurodegeneration and behavioral decits in hemiparkinsonian rats.Mol. Ther. 16, 15721579.

McLay, R.N., et al., 1997. Granulocyte-macrophage colony-stimulating factor crosses thebloodbrain and bloodspinal cord barriers. Brain 120 (Pt 11), 20832091.

Mogi, M., et al., 1999. Brain-derived growth factor and nerve growth factorconcentrations are decreased in the substantia nigra in Parkinson's disease.Neurosci. Lett. 270, 4548.

Mount, M.P., et al., 2007. Involvement of interferon-gamma in microglial-mediated lossof dopaminergic neurons. J. Neurosci. 27, 33283337.

Murer, M.G., et al., 2001. Brain-derived neurotrophic factor in the control human brain,and in Alzheimer's disease and Parkinson's disease. Prog. Neurobiol. 63, 71124.

Nagatsu, T., Sawada, M., 2007. Biochemistry of postmortem brains in Parkinson's disease:historical overview and future prospects. J. Neural Transm. 113120 (Suppl.).

Nakagawa, T., et al., 2006. Intracarotid injection of granulocyte-macrophage colony-stimulating factor induces neuroprotection in a rat transient middle cerebral arteryocclusion model. Brain Res. 1089, 179185.

Nimmerjahn, A., et al., 2005. Resting microglial cells are highly dynamic surveillants ofbrain parenchyma in vivo. Science 308, 13141318.

Nutt, J.G., et al., 2003. Randomized, double-blind trial of glial cell line-derivedneurotrophic factor (GDNF) in PD. Neurology 60, 6973.

Pardridge, W.M., 2002. Targeting neurotherapeutic agents through the blood-brainbarrier. Arch. Neurol. 59, 3540.

Peleshok, J., Saragovi, H.U., 2006. Functional mimetics of neurotrophins and theirreceptors. Biochem. Soc. Trans. 34, 612617.

Peterson, A.L., Nutt, J.G., 2008. Treatment of Parkinson's disease with trophic factors.Neurotherapeutics 5, 270280.

Porritt, M.J., et al., 2005. Inhibiting BDNF expression by antisense oligonucleotideinfusion causes loss of nigral dopaminergic neurons. Exp. Neurol. 192, 226234.

Purisai, M.G., et al., 2007. Microglial activation as a priming event leading to paraquat-induced dopaminergic cell degeneration. Neurobiol. Dis. 25, 392400.

Re, F., et al., 2002. Granulocyte-macrophage colony-stimulating factor induces anexpression program in neonatal microglia that primes them for antigenpresentation. J. Immunol. 169, 22642273.

Reddy, P.H., et al., 2009. Granulocyte-macrophage colony-stimulating factor antibodysuppresses microglial activity: implications for anti-inammatory effects inAlzheimer's disease and multiple sclerosis. J. Neurochem. 111, 15141528.

Salehi, Z., Mashayekhi, F., 2009. Brain-derived neurotrophic factor concentrations in thecerebrospinal uid of patients with Parkinson's disease. J. Clin. Neurosci. 16, 9093.

Salvatore, M.F., et al., 2006. Point source concentration of GDNF may explain failure ofphase II clinical trial. Exp. Neurol. 202, 497505.

Santambrogio, L., et al., 2001. Developmental plasticity of CNS microglia. Proc. Natl.Acad. Sci. U. S. A. 98, 62956300.

Schabitz, W.R., et al., 2008. A neuroprotective function for the hematopoietic proteingranulocyte-macrophage colony stimulating factor (GM-CSF). J. Cereb. Blood FlowMetab. 28, 2943.

Schermer, C., Humpel, C., 2002. Granulocyte macrophage-colony stimulating factoractivatesmicroglia in rat cortex organotypic brain slices. Neurosci. Lett. 328, 180184.

Schmidt, H.D., Duman, R.S., 2007. The role of neurotrophic factors in adult hippocampalneurogenesis, antidepressant treatments and animal models of depressive-likebehavior. Behav. Pharmacol. 18, 391418.

Schmidt, H.D., Duman, R.S., 2010. Peripheral BDNF Produces Antidepressant-Like Effects inCellular and Behavioral Models. Neuropsychopharmacology 35, 23782391.

Sherer, T.B., et al., 2007. Mechanism of toxicity of pesticides acting at complex I:relevance to environmental etiologies of Parkinson's disease. J. Neurochem. 100,14691479.

Shults, C.W., et al., 1995. BDNF attenuates the effects of intrastriatal injection of 6-hydroxydopamine. NeuroReport 6, 11091112.

Somayajulu-Nitu, M., et al., 2009. Paraquat induces oxidative stress, neuronal loss insubstantia nigra region and parkinsonism in adult rats: neuroprotection and

amelioration of symptoms by water-soluble formulation of coenzyme Q10. BMCNeurosci. 10, 88.

Taliaz, D., Stall, N., Dar, D.E., Zangen, A., 2010. Knockdown of brain-derivedneurotrophic factor in specic brain sites precipitates behaviors associated withdepression and reduces neurogenesis. Mol. Psychiatry 15 (1), 8092.

Tatton, W.G., et al., 2003. Apoptosis in Parkinson's disease: signals for neuronaldegradation. Ann. Neurol. 53 (Suppl. 3), S61S70 (discussion S70-2).

Thomson, A.W., Lotze, M.T. (Eds.), 2003. The Cytokine Handbook. Academic Press,London.

Tomac, A., et al., 1995. Protection and repair of the nigrostriatal dopaminergic systemby GDNF in vivo. Nature 373, 335339.

Whitton, P.S., 2007. Inammation as a causative factor in the aetiology of Parkinson'sdisease. Br. J. Pharmacol. 150, 963976.

Yasuhara, T., et al., 2007. Glial cell line-derived neurotrophic factor (GDNF) therapy forParkinson's disease. Acta Med. Okayama 61, 5156.

Zhang, H.T., et al., 2007. Immunohistochemical distribution of NGF, BDNF, NT-3, andNT-4 in adult rhesus monkey brains. J. Histochem. Cytochem. 55, 119.

112 E.N. Mangano et al. / Neurobiology of Disease 43 (2011) 99112

Granulocyte macrophage-colony stimulating factor protects against substantia nigra dopaminergic cell loss in an environment...IntroductionMaterials and methodsExperiment 1: assessment of the beneficial effects of supra-nigral administration of GDNF and GM-CSF following LPS-paraquat...Immunohistochemical procedures: SNc DA neuronal survival and microglial reactivityWestern blot analyses of hippocampal BDNF levelsExperiment 2: assessment of the beneficial effects of systemic administration of GM-CSF following LPS-paraquat induced path...Immunohistochemical analyses of SNc neuronal lossAssessment of glial changesGeneral stereotaxic surgical procedures for Experiments 1 and 2Experiment 3: in vitro assessment of the effects of paraquat and GM-CSFImmunocytochemistryWestern blotStatistical data analysis

ResultsExperiment 1: Supra-nigral injection of GM-CSF or GDNF attenuated LPS-paraquat induced degeneration of SNc dopamine neurons...Supra-nigral injection of GM-CSF influenced LPS and paraquat induced glial changes in the SNcSupra-nigral infusion of GM-CSF modified the paraquat induced reduction of mature hippocampal BDNFExperiment 2: systemic administration of GM-CSF had neuroprotective consequencesExperiment 3: GM-CSF neuroprotection was associated with normalized BDNF but not Bcl-2 or Bax levels

DiscussionAcknowledgmentsReferences