Background
Traditionally, Parkinson’s disease (PD) is considered a movement disorder with progressive dopaminergic neurodegeneration and Lewy body formation in the substantia nigra [
1]. Current evidence suggests that the pathological processes of PD extend beyond the nigrostratal system and degeneration of neuronal populations in other brain regions has also been noted [
2]. Furthermore, patients with PD display not only motor deficits but also a lot of non-motor symptoms [
3]. In the clinic, cognitive impairments are frequently observed in PD patients and are often inadequately treated, which gradually becomes an important determinant of life quality of patients [
4,
5]. Currently, the mechanisms of cognitive deficits in PD remain unclear.
Microglia-mediated neuroinflammation has been reported to contribute to the pathogenesis of PD [
6]. In addition to nigrostriatal regions, activated microglia and accumulation of inflammatory factors are also noted in the hippocampus and cortex, two critical brain regions that may contribute to cognitive decline in PD [
7]. Negative correlations between microglial activation and hippocampal volume or cerebral glucose metabolic rate within hippocampus have been observed in PD patients with dementia [
8]. Furthermore, Menza et al. found higher levels of proinflammatory factors, such as interleukin-1β (IL-1β) and tumor necrosis factor α (TNFα) in the plasma of PD patients than that of individuals without PD [
9]. A subsequent study demonstrated that increased TNFα and TNF receptor 1 contents in the plasma of PD patients are associated with poor cognitive test scores [
9]. Similarly, in Alzheimer’s disease (AD) patients, a positive correlation between microglial activation and the clinical dementia rating score has also been observed [
10]. Additionally, activated microglia have been shown to induce neurotoxic reactive astroglia [
11]. Neurotoxic astroglial activation is also considered to contribute to neuronal damage and subsequent cognitive impairment in animal models of neurodegeneration [
12]. Consistently, blocking microglia-mediated conversion of astroglia to a neurotoxic phenotype by IL-10 or fluorocitrate alleviated LPS-induced depressive-like behavior and cognitive dysfunction in mice [
13]. These results suggest that neuroinflammation mediated by microglia may contribute to cognitive decline in neurodegenerative disease.
To provide direct experimental evidence linking neuroinflammation and cognitive impairments in PD, a mouse PD model induced by rotenone was administered PLX3397 or minocycline to deplete brain microglia or inhibit microglial activation, respectively. Cognitive performance as well as neuronal loss, synaptic degeneration, and α-synuclein Ser129-phosphorylation were determined. Then, the underlying mechanisms of how microglial activation contributes to cognitive impairments were further explored. Our results suggested that microglial activation damaged cognitive performance in rotenone-induced mouse PD model by exacerbating neuroinflammation, oxidative stress, and neuronal apoptosis.
Methods
Reagents
Rotenone was purchased from Sigma-Aldrich, Inc. (R8875, St. Louis, MO, USA). PLX3397 (S7818) and minocycline (S4226) were purchased from Selleck (Shanghai, China). The antibody against Neu-N was purchased from EMD Millipore (MAB377, Temecula, CA, USA). The anti-ionized calcium binding adaptor molecule-1 (Iba-1) and anti-glial fibrillary acidic protein (GFAP) antibodies were purchased from Wako Chemicals (019-19741, Richmond, VA, USA) and Dako (Z0334, Santa Clara, CA, USA), respectively. The antibodies against postsynaptic density protein 95 (PSD-95, ab13552), Ser129 phosphorylated (ab51253), and total α-synuclein (ab6162) were purchased from Abcam (Cambridge, MA). In Situ Cell Death Detection Kit was purchased from KeyGEN BioTECH Corp (KGA7073, Jiangsu, China). The commercial assay kits for glutathione (GSH, S0052) and malondialdehyde (MDA, S1031S) were purchased from Beyotime Biotechnology (Shanghai, China). All other chemicals were of the highest grade commercially available.
Animal treatment
Eight-week old male C57BL/6 J mice (SPF Animal Laboratory of Dalian Medical University) were randomly assigned to control and two rotenone groups (
n = 12 in each group). Rotenone was freshly prepared every day in a solution of 0.1% DMSO (diluted with PBS) and was administered daily (i.p., 0.75 or 1.5 mg/kg bw/day) to mice for consecutive 3 weeks as described previously [
14,
15]. For the time course study, mice were treated with 1.5 mg/kg bw/day rotenone. After 1, 2, and 3 weeks of initial rotenone injection, mice (
n = 4 in each time point) were sacrificed and the brains were dissected. Mice in control group were given equal amount of 0.1% DMSO (diluted with PBS). All mice were housed under standard laboratory conditions (a 12-h light/dark cycle with an average room temperature of 25 °C). All animal procedures and their care were performed in strict accordance with the National Institutes of Health guidelines and were approved by the Institutional Animal Care and Use Committee of Dalian Medical University.
PLX3397 and minocycline treatment
PLX3397 was dissolved in 0.1% DMSO (diluted with PBS) [
16]. PLX3397 (40 mg/kg/day) was administered to mice by gavage 7 days prior to rotenone injection (1.5 mg/kg/day, i.p., daily). Based on previous reports and our pilot study, the number of microglia in the PLX3397-treated mice was reduced to approximately 70% of that in control mice at this time point [
16,
17]. After 7 days of initial PLX3397 administration, mice (
n = 12 in each group) received PLX3397 30 mins prior to rotenone by every other day until the end of the experiment. Minocycline (dissolved in PBS, 50 mg/kg/day) was administered (i.p.) to mice 2 days before rotenone (
n = 11). After 2 days of treatment, minocycline was injected 30 min prior to rotenone by consecutive 3 weeks. The chosen of concentrations of PLX3397 and minocycline was based on previous reports [
16,
18]. Mice in control group were given equal amounts of 0.1% DMSO (diluted with PBS).
Morris water maze test
Morris water maze (MWM) test was performed as described previously [
19,
20]. The test included the navigation test and the spatial probe test. In the spatial navigation test, a circular platform was placed in the middle of one quadrant (1 cm below the water surface). In one trial, mice were randomly placed in one of the quadrants and were allowed to swim until they found the platform or for a maximum 90 s. Mice were gently guided onto the platform if they could not find the platform within 90 s, and their latencies were recorded as 90 s. All mice were given a break of 5 min on the platform between trials. There were four trials per day for each mouse, and the mice were tested for 4 days. A video camera was used to record the swimming paths of mice. Different parameters for evaluating learning performance, such as latency for escape and traveled distance were analyzed by using the tracking software (NoldusEtho Vision system, version 5, Everett, WA, USA).
On the fifth day, the spatial probe test for spatial memory function was performed. Briefly, the platform was removed and mice were permitted to navigate in the pool freely for 60 s. Then, different parameters, including the time of mice spent in different quadrants, the latency to initially cross the position where the platform was located and the total number of platform crossings were recorded and analyzed by using the tracking software (NoldusEtho Vision system, version 5).
Novel objective recognition
Novel objective recognition (NOR) test was conducted based on a previous protocol [
21]. In brief, mice were trained over 3 days (3 times per day) to discriminate a novel object from a familiar one. On the first day, mice were put in an empty chamber and permitted to move freely. On the following day, mice were allowed to explore two the same objects placed in opposite corners of the chamber for 5 min. On the last day, mice were put back in the chamber and were permitted to explore a familiar object (the same object as day 2) and a new object with different colors and shapes from the familiar object. After each test, 70% ethanol was used to thoroughly clean the chamber and objects. The time the mice spent exploring familiar and novel objects was recorded, and the recognition index was calculated by using the following formula.
$$ \mathrm{Recognition}\ \mathrm{index}=\frac{\mathrm{Time}\ \mathrm{spent}\ \mathrm{exploring}\ \mathrm{novel}\ \mathrm{object}}{\mathrm{Time}\ \mathrm{spent}\ \mathrm{exploring}\ \mathrm{both}\ \mathrm{novel}\ \mathrm{and}\ \mathrm{familiar}\ \mathrm{object}\mathrm{s}}\times 100\% $$
Passive avoidance test
The passive avoidance test was done based on previous protocol [
22]. In brief, mice were put in a chamber that was separated into light and dark compartments (the same size) by a guillotine door. On the first day, the guillotine door between the light and dark compartments was opened and mice were permitted to move freely for 5 min in both compartments of the chamber. The next day, mice were put in the light compartment. After 60 s, the guillotine door was opened, and mice were permitted to move to the dark compartment freely. Once the mice entered, the guillotine door was closed and an electric shock was given (0.3 mA, 5 min). This test was repeated with 10-min intervals until the latency to enter the dark compartment reached 120 s. On the last day, mice were placed in the light compartment. Sixty seconds later, the guillotine door was opened and the latency of mice to enter the dark compartment (step-through latency) and the number of total entrances (step-through number) were recorded.
Immunohistochemistry and immunofluorescence staining
Mice in each group were perfused transcardially with PBS, followed by 4% paraformaldehyde and then the brains were collected. Brain samples were postfixed with 4% paraformaldehyde at 4 °C for 48 h, and then transferred to 30% sucrose in PBS before sectioning. Free-floating coronal sections (30 μm) containing hippocampal and cortical regions were used for immunohistochemistry and immunofluorescence staining as described previously [
23,
24]. Briefly, brain sections were blocked in 0.25% Triton/PBS containing 4% goat serum for 2 h and then incubated with anti-PSD-95, anti-Neu-N, anti-Iba-1, anti-GFAP or anti-Ser129-phosphorylated α-synuclein antibodies in PBS containing 0.1% Triton X-100 at 4 °C for an additional 24 h. Then, the sections were washed three times prior to incubation with an appropriate biotinylated or Alexa-594-conjugated secondary antibody. Immunohistochemical staining was visualized by using 3,3'-diaminobenzidine. Digital images were acquired (×10 and ×40 magnification for immunohistochemistry and immunofluorescence staining, respectively) under an Olympus microscope (BX51; Olympus, Tokyo, Japan) using an attached digital microscope camera (DP72; Olympus).
The densities of Iba-1, PSD-95, and GFAP immunostaining from two to three brain sections with 120-μm intervals from each mouse in each group were measured by using ImageJ software [
24‐
26]. Briefly, the image was first converted into the grayscale picture, and the background was adjusted before the quantifying area was selected for the measurement of the total pixels. The relative density of the staining was compared based on the density of the total pixels of a certain brain region (total pixels/area). The quantification of the staining was corrected for background staining by subtracting the pixels without primary antibody.
Automated counting assessment of neurodegeneration
The number of Neu-N
+ neurons in mice among the groups was quantified by the automated counting method in ImageJ software [
19,
24].
TUNEL assay
In Situ Cell Death Detection Kit was used to perform TUNEL assays in free-floating coronal sections (30 μm) containing hippocampal and cortical regions [
22,
27]. TUNEL
+ cells were observed using a fluorescence microscope (×40 magnification). The mean number of TUNEL
+ cells in each mouse was obtained by counting three coronal sections with 120 μm intervals by using ImageJ software.
Western blot analysis
For Western blot analysis, mice were perfused transcardially with PBS to remove the blood. The brains were collected and hippocampal, and the cortical brain regions were dissected immediately on ice. Tissue samples including the hippocampus and cortex were homogenized by using ice-cold RIPA buffer with a protease inhibitor mixture and then centrifuged at 10,000×g. After 10 min of centrifugation, protein concentrations were measured in the collected supernatants by using a BCA protein assay kit. Samples containing equal amounts of protein from each group were separated by 4–12% SDS-PAGE, and then, immunoblot analysis was performed using anti-PSD-95, anti-Neu-N, anti-α-synuclein, and anti-α-synuclein (phospho S129) antibodies at 4 °C for 24 h. After washing three times with PBST, the membranes were incubated with appropriate HRP-linked anti-rabbit or mouse IgG. The signal was detected by ECL reagents, and relative density of blots was quantified using ImageJ software.
Real-time PCR analysis
For RT-PCR analysis, mice in each group (
n = 6) were perfused transcardially with PBS to remove the blood. The hippocampal and cortical brain regions were dissected immediately on ice and then transferred to liquid nitrogen, followed by a − 80 °C freezer. RNA was isolated from the hippocampal and cortical samples by using TRIzol reagent. Quantitative analysis of RNA was performed by using Nanodrop spectrophotometer. A total 1-μg RNA from each sample was used for complementary DNA (cDNA) synthesis using MuLV reverse transcriptase and oligo dT primers according to previous reports [
28,
29]. The reaction conditions were set as 25 °C for 5 min, 42 °C for 60 min, and 70 °C for 15 min. SYBR Premix Ex TaqTM II and a Takara Thermal Cycler Dice™ Real Time System were subsequently used for real-time PCR detection based on the manufacturer’s protocols. The primers were designed with Vector NTI software and validated for efficacy through melting curve analyses. The sequences of the primers were the follows: GAPDH F (5′-TTCAACGGCACAGTCAAGGC-3′; 300 nM), GAPDH R (5′-GACTCCACGACATACTCAGCACC-3′; 300 nM), TNFα F (5′-GACCCTCACACTCAGATCATCTTCT-3′; 300 nM), TNFα R (5′-CCTCCACTTGGTGGTTTGCT-3′; 300 nM), IL-1β F (5′-CTGGTGTGT GACGTTCCCATTA-3′; 300 nM), IL-1β R (5′-CCGACAGCACGAGGCTTT-3′; 300 nM), iNOS F(5′-CTGCCCCCCTGCTCACTC-3′; 300 nM), and iNOS R (5′-TGGGAGGGGTCGTAATGTCC-3′; 300 nM). The following PCR conditions were used: 95 °C for 10 s, 55 °C for 30 s, and 72 °C for 30 s for 40 cycles (for a final reaction volume of 25 μl). All samples were tested in duplicate and normalized to GAPDH using the 2
-ΔΔCt method. Fold changes for each treatment were normalized and are shown as percentages of the control.
GSH and MDA assay
The tissue samples of mice perfused with PBS only were homogenized in ice-cold lysis buffer containing 20 mM Tris (pH 7.5), 150 mM NaCl, 1% Triton X-100, and protease inhibitor mixture. After 10 min of centrifugation at 10,000×
g (4 °C), the contents of GSH and MDA in the collected supernatant were measured by using commercial kits based on the protocol provided by the manufacturer [
30].
Statistical analysis
Data were expressed as the mean ± SEM. Except for the MWM test, comparison of means among two or more groups was conducted using one-way (one parameter) or two-way ANOVA (two parameters). Subsequently, Tukey’s post hoc test was used for pairwise comparisons between means once ANOVA showed significant differences. The MWM test data were analyzed using repeated measures ANOVA. Fisher’s PLSD tests were used for comparing group means only when a significant F value was observed in the overall ANOVA. A p value less than 0.05 was considered statistically significant.
Discussion
In this study, we provided strong evidence to support that microglial activation contributed to cognitive decline in a mouse PD model generated by rotenone. Four salient features were observed: (1) rotenone-induced microglial activation preceded neurodegeneration in mice; (2) microglial depletion by PLX3397 and inactivation by minocycline significantly ameliorated rotenone-induced cognitive deficits and neuronal damage in mice; (3) PLX3397 and minocycline suppressed astroglial activation, production of proinflammatory factors, and oxidative stress in rotenone-treated mice; and (4) PLX3397 and minocycline abrogated rotenone-induced neuronal apoptosis in mice.
PD is traditionally recognized as a movement disorder with progressive loss of nigral dopaminergic neurons [
37]. However, accumulating evidence suggests that PD is a heterogeneous multisystem disorder and patients with PD display both motor deficits and multiple nonmotor symptoms [
19,
38]. Cognitive impairments are one of the most common nonmotor symptoms in patients with PD. Clinical studies have revealed that 25~46.8% of newly diagnosed patients exhibit mild cognitive deficits, and up to 80% exhibit dementia in the late stage of the disease [
39,
40]. A postmortem study in PD patients indicates that limbic and cortical neuron damage and Lewy body pathology correlate with cognitive decline [
41]. However, studies in widely used parkinsonian animal model induced by 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) generated mixed results, which greatly hampered the progress of mechanical studies on cognitive deficits. Although cognitive decline was reported in MPTP-induced mouse model in previous reports [
42,
43], Fifel et al. found no significant decrease in cognitive alterations in either acute or chronic MPTP-treated mice [
44]. Our preliminary data also showed that MPTP failed to damage learning and memory capacity in mice (data not shown), although significant dopaminergic neurodegeneration was reported in this model [
45,
46]. The rotenone-induced mouse PD model is also widely used. However, mixed results regarding the effects of rotenone on cognition in mice have been generated. Alabi et al. reported that daily doses of rotenone (2.5 mg/kg, i.p) for 4 weeks impaired memory in mice as shown by a significantly impaired performance in the Y-maze test compared with that of vehicle controls [
47]. In contrast, Jia and colleagues found that intragastric delivery of rotenone (5 mg/kg) for 3 months improved spatial learning and memory abilities of mice in the MWM test, although spatial memory ability was impaired at 1 month after treatment [
48]. In this study, we further investigated the effects of rotenone on cognition in mice by using a subchronic dosing regimen. Results showed that rotenone dose-dependently decreased novel objective recognition, passive avoidance, and MWM performance in mice after 3 weeks of treatment. Furthermore, elevated neuronal damage, loss of synapses, and Ser129-phosphorylation of α-synuclein were also detected in rotenone-injected mice, which was consistent with observations in PD patients.
The mechanisms underlying rotenone-induced cognitive dysfunction remain unclear. Inflammation has long been found to be inversely correlated cognitive decline [
20,
49]. Increasing evidence suggested that activated microglia is a key causative factor in inflammation-mediated neurodegeneration and behavioral deficits [
50]. Dysregulated microglial activation is reported to be able to increase pathological protein aggregation and impair synaptic pruning and neuron plasticity in key brain regions subserving cognition [
51,
52]. In the present study, activated microglia were detected in the hippocampus and cortex of rotenone-induced mouse PD model. A time experiment revealed that microglial activation induced by rotenone preceded neuronal damage and α-synuclein pathology in mice. Furthermore, microglial depletion by PLX3397 or inactivation by minocycline significantly reduced neuronal damage in the hippocampus and cortex of mice treated with rotenone. Consistently, neuroprotective effects of both PLX3397 and minocycline against rotenone-induced dopaminergic neurodegeneration were also observed in mice (data not shown). More importantly, PLX3397 and minocycline-afforded neuroprotection were associated with improved cognitive performance in rotenone-treated mice, indicating an important role of activated microglia in mediating cognitive dysfunction. In agreement with our findings, Cope and colleagues recently found that microglial activation plays an active role in obesity-associated cognitive decline since blocking microglial activation by minocycline prevented loss of dendritic spines and cognitive decline in obese mice [
53]. Depletion of microglia by CSF1R inhibitors, such as PLX3397 and PLX5622, was also associated with improved cognitive performance in experimental mouse models of AD [
54], intracerebral hemorrhage [
55], and
irradiation-induced memory deficits [
56].
Microglia can become overactivated in response to certain injuries and release a variety of cytotoxic factors that cause neurotoxicity [
11]. In addition, activated microglia have recently been shown to be able to induce astrocytes into a neurotoxic A1 status by releasing TNFα, IL-1α, and C1q to amplify neuronal damage [
11]. Proinflammatory factors and reactive oxygen species (ROS) are considered to be critical to mediate neuroinflammatory damage since neutralization of proinflammatory factors or inhibition of ROS during neuroinflammation showed potent neuroprotection in a variety of neurodegeneration models [
57]. Consistent with these findings, in this study, elevated astroglial activation, production of proinflammatory cytokines, and oxidative stress were observed in rotenone-treated mice. Furthermore, we demonstrated that microglial depletion by PLX3397 and inactivation by minocycline suppressed these toxic events and was accompanied by decreased apoptosis of neurons and expression of cleaved caspase-3 and Bax as well as elevated expression of Bcl-xL. Our findings suggested that microglial activation contributed to cognitive deficits and neuronal apoptosis through neurotoxic astroglial activation, neuroinflammation, and oxidative damage. Consistent with our findings, endotoxin LPS, an inflammatory stimulator, has been reported to induce cognitive deficits in mice through neuroinflammation and neuronal apoptosis [
21]. Inhibition of inflammatory cytokines, oxidative stress, and neuronal apoptosis in the hippocampus also ameliorated cognitive deficits in mouse model of AD [
58], vascular dementia, and sepsis [
59]. Notably, a pharmacological manipulation method (PLX3397 and minocycline) was used to block microglial activation in the current study, and we could not exclude the possibility that PLX3397 and minocycline might be have direct effects on astroglial activation, oxidative stress, or neuronal apoptosis. Further study focusing on this issue should be performed in the future.
It is interesting to compare the similar protective effects of minocycline and PLX3397 in this study. Minocycline is a semisynthetic tetracycline derivative that displays potent anti-inflammatory effects. Multiple studies have revealed that minocycline can dampen proinflammatory microglial activation (M1) and might simultaneously promote microglial M2 polarization in neuropathological conditions [
60‐
62]. In contrast, PLX3397 is reported to directly deplete microglia in an apoptosis-dependent manner in the brains of mice [
17]. Although microglial elimination by PLX3397 is not accompanied by an inflammatory response in the brain, PLX3397-treated mice exhibited robust reductions in the expression of many inflammatory genes, including TNF-α and other cytokines, in response to LPS [
17]. Consistently, Liang et al. reported that PLX3397 treatment significantly reduced the number of M1 phenotype-like microglia and production of proinflammatory factors in a mouse model of high-fat-diet (HFD)-induced obesity [
63]. Our preliminary data showed reduced expression of iNOS, a marker of M1 microglia, and elevated expression of the M2 microglial marker Arg-1 in the brains of mice treated with combined PLX3397 and rotenone compared with those of mice treated with rotenone alone (Supplementary Fig.
5), suggesting that PLX3397 might prefer to target/eliminate M1-trended microglia in neuroinflammatory or pathological conditions. Thus, we believe that the equal potency of PLX3397 and minocycline in dampening M1 microglial activation by microglial depletion and inhibition, respectively, might be one of the potential reasons for similar neuroprotective effects of these two compounds.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.