Male C57BL/6 mice (20–25 g) were used for these studies. They were housed in groups of 4–6, at constant room temperature of 22 ± 1°C and relative humidity (60%), with a 12-h light-dark cycle and free access to food and water. Experiments were carried out in accordance with the Guidelines of the European Union Directive (2010/63/EU) and Spanish regulations (BOE 34/11370-421, 2013) for the use of laboratory animals; the study was approved by the Scientific Committee of the University of Seville. Experiments using MPTP were carried out following the operating instructions about the use, handling and storage of chemical agents of the Prevention Service of the University of Seville (SEPRUS¹).

Acute MPTP models have been associated to strong inflammatory response in the nigro-striatal dopaminergic system (Kavanagh et al., 2015). Main MPTP acute treatment paradigm is based on four consecutive MPTP injections (16 mg/kg every 2 h). However, the combination of LPS and this acute MPTP paradigm abruptly increased early death rate. To solve this issue, we took advantage of a single injection of MPTP at higher dose (40 mg/kg), which has been shown to cause a selective death of nigral dopaminergic neurons (Mejías et al., 2006). Using this MPTP paradigm, Lopez-Barneo and colleagues demonstrated a 65% decrease of striatal dopaminergic terminals, a 48% loss of nigral TH-positive neurons and a 24% loss of Nissl-stained cells in the SN evaluated 1 week after neurotoxin injection (Mejías et al., 2006). These findings demonstrate neurodegenerative events of the nigro-striatal dopaminergic system and make this MPTP paradigm suitable to test whether or not the presence of peripheral inflammation affects nigral dopaminergic vulnerability.

Crews and colleagues demonstrated that a single dose of 5 mg/kg (i.p.) causes neuroinflammation and delayed death of nigral dopaminergic neurons (starting 7 months after LPS injection) (Qin et al., 2007). Besides, it is well-documented that very little peripheral LPS enters to the brain due to the poor passage through the BBB (Nadeau and Rivest, 1999), Consequently, to full validate the solely contribution of peripheral inflammation in MPTP-induced neurotoxicity, we used a LPS dose 2.5-fold lower than that used by Qin et al. (2007). Consequently, animals were distributed within groups according to treatments, time points and method of analysis: The V (vehicle, control) group receiving a single i.p. injection of both vehicles; the MPTP group, receiving a single i.p. dose of MPTP (40 mg/kg in 0.9% sterile saline; Sigma-Aldrich, St. Louis, MO, United States, Catalogue # M0896); the LPS group receiving a single i.p. LPS injection (2 mg/kg in 0.9% sterile saline; Sigma-Aldrich, St. Louis, MO, United States, Catalogue # L4391), and the LPS/MPTP group receiving a single i.p. dose of MPTP and LPS. In the case of using both treatments, LPS was administered 120 min before the MPTP administration. All treatments were administrated in a volume of 100 μl per 25 g of body weight. At least three animals per group were used (Table 1).

Animals were sacrificed at different time points depending on the technique assayed. For immunohistochemistry analysis, animals were sacrificed at 12 h, 24 h, and 2 weeks. For qPCR assays, animals were sacrificed at 12 h. For HPLC and stereological analysis, animals were sacrificed at 2 weeks.

At required day and hour after the treatment, animals dedicated to the immunohistochemistry were anesthetized with ketamine (50 mg/kg, Ketamidor^®, Richter Pharma, Wels, Austria) and medetomidine (10 mg/kg, Domtor^®, Ecuphar, Oostkamp, Belgium) and perfused through the heart with 0.9% saline followed by 40 ml of 4% paraformaldehyde in phosphate buffer, pH 7.4. Brains were removed and then cryoprotected serially in sucrose in PBS, pH 7.4, first in 10% (24 h), then in 20% sucrose (24 h), and then in 30% sucrose until sunk (2–5 days). The brains were then frozen in isopentane at -40°C (10 min) and kept at -40°C. We analyzed early (12 and 24 h) and delayed response (2 weeks) after challenge.

Thaw-mounted 25-μm coronal sections were cut on a cryostat at -15°C and mounted in gelatin-coated slides. Primary antibodies used were rabbit-derived anti-tyrosine hydroxylase (anti-TH, Sigma, 1:300), mouse-derived anti-glial fibrillary acidic protein (anti-GFAP; Sigma, 1:400) and rabbit-derived anti-Iba-1 (Wako, 1:500). Incubations and washes for all the antibodies were in Tris-buffered saline (TBS), pH 7.4. All work was done at room temperature. Sections were washed and then treated with 0.3% hydrogen peroxide in methanol for 20 min, washed again, and incubated in a solution containing TBS and 1% goat serum (Vector) for 60 min in a humid chamber. Slides were drained and further incubated with the primary antibody (Table 2) in TBS containing 1% goat serum and 0.25% Triton-X-100 for 24 h. Sections were then incubated for 2 h with biotinylated goat anti-rabbit IgG (Vector, 1:200). The secondary antibody was diluted in TBS containing 0.25% Triton-X-100, and its addition was preceded by three 10-min rinses in TBS. Sections were then incubated with ExtrAvidin^®-Peroxidase solution (Sigma, 1:100). The peroxidase was visualized with a standard diaminobenzidine/hydrogen peroxide reaction for 6 min (do not ingest and breathe dust).

Breakdown of the BBB was assessed by employing a one-step immunohistochemical detection of IgG with some modifications (Tomas-Camardiel et al., 2004). Briefly, sections were incubated for 2 h in biotinylated horse anti-mouse IgG (Vector; 1:200; Table 2) in TBS containing 1% bovine serum albumin (BSA) and 0.2% (Triton-X-100). Visualization of IgG immunoreactivity was identical to that described above for immunohistochemistry.

Animals were perfused and sections were prepared as described above. Incubations and washes for all the antibodies were in PBS, pH 7.4. All work was done at room temperature, unless otherwise noted. Sections were blocked with PBS containing 5% BSA for 2 h. The slides were then incubated overnight at 4°C with the primary antibody (Table 2): goat-derived anti-occludin (Santa Cruz Biotechnology Inc.; 1:50), mouse-derived anti-GFAP (Sigma; 1:400), rat-derived anti-C3 (Santa Cruz Biotechnology Inc.; 1:50), sheep-derived anti-TH [NOVUS (Biomol); 1:1000], rabbit-derived anti-caspase 3 (Cell Signaling; 1:250), rabbit-derived anti-Iba1 (Wako; 1:1000) and goat-derived anti-galectin-3 (R&D Systems; 1:250). Primary antibodies were diluted in PBS containing 1% BSA and 1% Triton X-100. After three washes in PBS, sections were incubated with (Table 2) secondary antibodies conjugated to fluorescein (Vector; 1:300), Alexa Fluor^® 488 and Alexa Fluor^® 647 (Invitrogen; 1:200) for 2 h at room temperature in the dark. Fluorescence images were acquired using a confocal laser scanning microscope (Zeiss LSM 7 DUO) and processed using the associated software package (ZEN, 2010).

For GFAP immunoreactivity, the AnalySIS imaging software (Soft Imaging System GmbH, Münster, Germany) coupled to a Polaroid DMC camera (Polaroid, Cambridge, MA, United States) attached to a Leika light microscope (Leica Mikroskopie, Wetzlar, Germany) was used. Quantification of Iba-1 and TH positive cells in the SN was estimated according to a modified stereological approach. In each animal, every seventh section was used with random starting points and systematically distributed through the anteroposterior axis of the analyzed region. For counting cells showing Iba-1 immunoreactivity, a systematic sampling of the area occupied by the Iba-1 positive cells in each section was made from a random starting point with a grid adjusted to count five fields per section. An unbiased counting frame of known area (40 μm × 25 μm = 1000 μm²) was superimposed on the tissue section image under a 100× oil immersion objective. The different types of Iba-1-positive cells (displaying different shapes depending on their activation state) were counted as a whole and expressed as cells per mm². The number of TH-positive neurons in the SN was estimated using a fractionator sampling design (Gundersen et al., 1988). Counts were made at regular predetermined intervals (x = 150 μm and y = 200 μm) within each section. An unbiased counting frame of known area (15686,7 μm²) was superimposed on the tissue section image under a 40× objective. Therefore, the area sampling fraction was 15686,7 μm²/63380,6 μm² = 0.248. In all animals, 25-μm sections, each 175 μm apart, were analyzed; thus, the fraction of sections sampled was 25/175 = 0.143. The total number of neurons in the SN was estimated by multiplying the number of neurons counted within the sample regions by the reciprocals of the area sampling fraction and the fraction of section sampled.

Relative amount of IgG extravasation was measured using Image-J software (downloaded as a free software package from the public domain²). Areas of IgG occupancy were counted using three randomly distributed fields per section and three sections per animal. The system allows defining a threshold to discern between IgG immunoreactivity and background and the percentage of IgG occupancy measured.

Iba-1/galectin-3 co-localizing cells were counted with Image-J software using the plugin cell counter. At least three animals were used for each condition and number of Iba-1/galectin-3 cells were measure in a minimum of three eyespot images acquired using a confocal laser scanning microscope (Zeiss LSM 7 DUO).

Analysis of striatal dopamine was performed by HPLC with electrochemical detection. A Merck L-6200A intelligent pump was used in conjunction with a glassy carbon electrode set at -550 mV (DECADE II, ANTEC, Netherlands). A Merck Lichrocart cartridge (125 mm × 4 mm) column filled with Lichrospher reverse-phase C₁₈ 5 μm material was used. The mobile phase consisted of a mixture of 0.05 M of sodium acetate, 0.4 mM of 1-octanesulfonic acid, 0.3 mM of Na₂EDTA and 70 ml methanol/l, adjusted to pH 4.1 with acetic acid. All reactive agents and water were of HPLC grade. The flow rate was 1.0 ml/min. Measurement of DA in fresh tissue was performed according to the method previously described (Ismaiel et al., 2016). Concentrations of striatal DA samples were calculated with the aid of the eDAQ PowerChrom 280 software.

The results on microglial activation in terms of Iba-1 immunoreactivity showed that the highest activation takes places after 12 h of treatment. Therefore, animals designated to RT-PCR were perfused with saline sacrificed by decapitation 12 h after the treatment. Brains were then removed and striatum and SN were dissected, snap frozen in liquid nitrogen and stored at -80°C. Real Time RT-PCR was performed essentially as described previously (Sanchez-Hidalgo et al., 2016). Total RNA was extracted from the SN using RNeasy^® Mini kit (Qiagen). cDNA was synthesized from 1 μg of total RNA using RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific) in 10 μl reaction volume as described by the manufacturer. Real-time PCR was performed with SensiFAST^TM SYBR No-ROX Kit (Bioline), 0.4 μM primers and 4.2 μl cDNA. Controls were carried out without cDNA. Amplification was run in a LightCycler 480 (Roche Molecular Systems) thermal cycler at 95°C for 2 min followed by 40 cycles of 95°C for 5 s and 60°C for 15 s. Following amplification, a melting curve analysis was performed by heating the reactions from 65 to 95°C in 0.1°C/s while monitoring fluorescence. Analysis confirmed a single PCR product at the melting temperature. β-actin served as reference gene and was used for samples normalization. Primer sequences for β-actin, tumor necrosis factor (TNF)-α, interferon (INF)-β, interleukin (IL)-1β and IL-6 are shown in Table 3. The cycle at which each sample crossed a fluorescence threshold (Ct) was determined, and the triplicate values for each cDNA were averaged. Analyses of real-time PCR were done using a 2^Ct relative quantification method.

Results are expressed as mean ± SD. Means were compared by one-way ANOVA followed by the Fisher’s LSD test for post hoc multiple range comparisons. An alpha level of 0.05 was used. The Statgraphics Plus 3.0 statistical package was used for the analyses.

Iba-1 is a protein associated with major histocompatibility complex II, which is expressed in all microglial phenotypes but overexpressed in reactive microglia. Hence, we first used Iba-1 immunohistochemistry to evaluate the number and morphology of microglial cells. In striatum, control animals showed a faint Iba1 immunostaining characterized by a small number of highly-ramified microglial cells (170.2 cell/mm² ± 45.6; Figures 1A,B). Treatment with LPS alone induced a transient activation of microglia that peaked after 12 h (224.5 ± 10.3 cells/mm²; p < 0.001; Figures 1B,C,F,I). In this brain structure, microglial cells were bigger and exhibited thicken processes in response to LPS treatment. The treatment with MPTP alone showed a time-course characterized by a significant increase after 24 h (161.5 ± 19.6 and 224.2 ± 9.7 cell/mm², respectively, p < 0.001; Figures 1B,D,G,J). This MPTP-induced effect over microglial cells was significantly enhanced when LPS was administered 12 h prior to MPTP (273.8 ± 56,9 cell/mm²; p < 0.001; Figures 1B,E,H,K). The results demonstrate how peripheral inflammation modulates microglia brain response during neurodegeneration. Two weeks after treatment, number and morphology of most of microglial cells were similar to those found in controls. Still, some microglial cells in the striatum from LPS/MPTP- and MPTP-treated animals showed a round, phagocytic morphology without processes, features of chronically activated microglia (Figure 1K).

Similar results were found when microglia were analyzed in the SN. Strikingly, LPS treatment alone showed increased density of microglial cells that peaked again 12 h after exposition (156.2 ± 9.4; p < 0.0001; Figures 2B,C,F,I) compared to controls (44.9 ± 47.2 cell/mm², Figures 2A,B). In MPTP-treated animals, an increase in the number of microglial cells was found at 24 h (173.5 ± 0.8 cells/mm²; p < 0.0001; Figures 2B,D,G,J). Remarkably, LPS/MPTP-injected animals showed a stronger microglia activation compared with animals treated with LPS or MPTP alone, this effect being especially evident at 12 h (337.9 ± 140.8 cells/mm²; p < 0.0001; Figures 2B,E,H,K). These data demonstrate again a synergistic effect between peripheral and central inflammation.

In the striatum, conversely, a significant number of microglial cells was seen in the SN of MPTP-injected animals (157 ± 25.7 cell/mm²), which further increased in the LPS + MPTP group (220.4 ± 36.0 cells/mm²; p < 0.0001; Figure 2B,I–K). Microglia exhibited typical morphological features of activated cells with large bodies and thick processes (Figure 2K). These features were even more evident than those seen in striatal tissue (Figure 1K).

Recent transcriptomic studies have characterized the molecular signature of microglia under different disease conditions and identified a common disease-associated phenotype (DAM) or microglia neurodegenerative phenotype (Keren-Shaul et al., 2017; Krasemann et al., 2017; Mathys et al., 2017). A selected group of genes were found to be strongly up-regulated in the microglial phenotype associated to neurodegeneration, including Itgax, Clec7a and Lgals3 (galectin-3), which differs from the classically M1-like pro-inflammatory phenotype (Keren-Shaul et al., 2017; Krasemann et al., 2017; Mathys et al., 2017). We have long characterized novel roles associated to galectin-3 within microglial cells under different disease conditions (Boza-Serrano et al., 2014; Burguillos et al., 2015; Yip et al., 2017). Consequently, we analyzed microglial galectin-3 expression under the different experimental conditions. In agreement with previous findings, homeostatic microglia lack galectin-3 expression (Burguillos et al., 2015). Since most significant changes in terms of microglia density and activation, showed by Iba1-labeled microglia, was seen at 12 h after LPS/MPTP, we analyzed the presence of microglial galectin-3 at this time under the different experimental conditions, in both striatum and SN. Systemic LPS or MPTP alone failed to induce microglial galectin-3 expression in both structures. In contrast, in the combinative group, a strong up-regulation of galectin-3 was found in striatum (20-fold vs. control levels; p < 0.0001; Figure 3) and SN (6-fold vs. control levels; p < 0.01; Figure 4).

Since most relevant microglial changes were found 12 h after LPS/MPTP, we wanted to know if the appearance of microglia polarization toward a neurodegenerative phenotype precedes degenerative events in nigral dopaminergic neurons. Hence, we performed dual confocal immunofluorescence of TH/cleaved caspase-3 (apoptotic marker) in the ventral mesencephalon at 12 and 24 h after the different experimental conditions. No apparent early loss of integrity of nigral dopaminergic neurons was evident in any of the experimental conditions tested, including LPS/MPTP at 12 h (not shown) and 24 h (Supplementary Figure S1). Only under combined LPS/MPTP treatment was evident the appearance of cleaved caspase-3 in the ventral mesencephalon 24 h after. We, however, failed to detect any TH-labeled neuron showing cleaved caspase-3 (Supplementary Figure S1).

Overall, we conclude that systemic inflammation enhances the switch from homeostatic to disease-associated microglia in the nigrostriatal system in response to MPTP treatment preceding the onset of nigrostriatal dopaminergic degeneration.

We quantified the expression of mRNAs encoding for classical neurotoxic pro-inflammatory markers including TNF-α, IFN-β, IL-1β, and IL-6 in striatum and SN of mice sacrificed 12 h after the different experimental conditions including: control animals, LPS-, MPTP-, and LPS/MPTP-injected animals. Systemic LPS induced a clear pro-inflammatory response in the nigrostriatal system with significant increase in mRNA levels for TNF-α and IL-6 in striatum and IFN-β and IL-6 in the ventral mesencephalon (Figure 5). In contrast, the effect of MPTP alone was negligible. Yet, significant increase of TNF-α expression was found in striatum (Figure 5A) and also IFN- β in SN (Figure 5F). Highest pro-inflammatory response was clearly achieved when LPS and MPTP were combined, with significant increase of TNF-α, IFN-β, and IL-6 in striatum and SN (Figure 5). Statistical analysis demonstrated that combined LPS/MPTP treatment acted synergistically to increase mRNA levels for TNF-α and IFN-β in both striatum and SN (Figure 5).

We first analyzed the expression pattern of occludin (an integral membrane protein that directly regulates the tight junction paracellular permeability) 12 h after challenge (maximal response of microglia density and activation) (Kuan et al., 2016). We analyzed both, striatum and SN and found a decrease in the expression of this protein in animals treated concomitantly with LPS and MPTP (Supplementary Figure S2). This finding suggests an alteration in the BBB permeability in the nigrostriatal system. Consequently, we next quantitatively analyzed the integrity of the BBB. To achieve this, we studied IgG extravasation into the nigrostriatal system 12 h after challenge. Treatment with LPS or MPTP alone failed to alter BBB integrity in terms of IgG extravasation (Figure 6A–C,E). However, combined LPS/MPTP treatment highly altered BBB integrity in the striatum as demonstrated by IgG extravasation (Figures 6D,E). Similar results were found in SN, where only LPS/MPTP group showed increased immunoreactivity of IgG into the analyzed brain parenchyma (Figures 6F–J). We additionally extended our analysis to other brain areas not directly related to the nigrostriatal system. Thus, we analyzed BBB integrity in cerebral cortex and midbrain (adjacent to the superior colliculus) in response to LPS, MPTP, and LPS + MPTP 12 h after challenge. Interestingly, BBB disruption was again observed in response to LPS + MPTP but not to LPS or MPTP alone in both analyzed areas (Supplementary Figure S3), further supporting that peripheral systemic inflammation and MPTP synergistically enhance/induce BBB breakdown.

Reactive astrocytes showing the A1 phenotype have been recently identified and found to be neurotoxic as opposed to A2 astrocytes, which are neuroprotective (Liddelow et al., 2017). Reactive microglia induces reactive A1 astrocyte polarization, which highly up-regulates complement component 3 (C3) expression, in sharp contrast to A2 astrocytes (Liddelow et al., 2017). Therefore, we have performed a double immunofluorescence confocal analysis in order to study the astrocytic phenotype in our experimental conditions. In control animals, astrocytes were not activated and failed to express C3, both in striatum and SN (Figures 7, 8A–C). Treatment with LPS or MPTP alone also failed to induce the A1 phenotype in striatum, at least at the post-injection time analyzed (12 h) (Figures 7D–I). However, the combination of both treatments clearly induced the A1 neurotoxic phenotype in the striatum (Figures 7J–L). In the SN, LPS alone again failed to induce the A1 astrocytic phenotype at 12 h post-injection (Figures 8D–F). However, A1 reactive astrocytes population were minority after the treatment with MPTP alone (Figures 8G–I). In contrast, this phenotype shift was much more evident when LPS and MPTP were combined (Figures 8J–L). Collectively, these data demonstrate that peripheral inflammation early triggers A1 astrocytic phenotype under conditions of dopaminergic neurodegeneration.

We used TH antibody to label dopaminergic neurons in SN of mice treated with LPS, MPTP, combined LPS/MPTP and controls. Analysis performed 2 weeks after the neurotoxin administration showed clear neurodegenerative features. This result was further evaluated by stereology. An even distribution of TH-positive neurons was seen in the SN of control animals (5803 ± 604; Figures 9A,E), showing that the primary antibody was highly selective to dopaminergic neurons. Our stereological analysis of dopaminergic neurons in the SN agrees with previously published reports (Smeyne et al., 2016; Alam et al., 2017; Ip et al., 2017). As expected, MPTP-treated mice showed a significant decrease of 60.7% in the number of dopaminergic neurons of SN (Figures 9C,E, p < 0.001). LPS-injected animals failed to induce loss of nigral dopaminergic neurons 2 weeks after the treatment (4770 ± 1924; Figures 9B,E). Strikingly, LPS and MPTP acted synergistically and dopaminergic cell death was significantly increased compared with MPTP alone, reaching 83.6% of cell loss (Figures 9D,E, p < 0.001).

We also analyzed dopaminergic terminals in the striatum 2 weeks after neurotoxin administration by using TH-immunohistochemistry and neurochemical analysis of DA and its metabolites. LPS alone failed to alter DA levels in the striatum as compared with control non-lesioned animals (Figures 10A,B,E). MPTP treatment largely decreased DA levels (about 40% control levels; Figures 10C,E). Combined LPS/MPTP further decreased DA levels (about 25% control levels; Figures 10D,E), but failed to reach statistical significance, thus suggesting compensatory mechanisms at the dopaminergic terminal.