Microglial MT1 activation inhibits LPS‐induced neuroinflammation via regulation of metabolic reprogramming

Abstract Parkinson’s disease (PD) is one of the most common neurodegenerative diseases. Although its pathogenesis remains unclear, a number of studies indicate that microglia‐mediated neuroinflammation makes a great contribution to the pathogenesis of PD. Melatonin receptor 1 (MT1) is widely expressed in glia cells and neurons in substantia nigra (SN). Neuronal MT1 is a neuroprotective factor, but it remains largely unknown whether dysfunction of microglial MT1 is involved in the PD pathogenesis. Here, we found that MT1 was reduced in microglia of SN in 1‐methyl‐4‐phenyl‐1,2,3,6‐tetrahydropyridine (MPTP)‐induced PD mouse model. Microglial MT1 activation dramatically inhibited lipopolysaccharide (LPS)‐induced neuroinflammation, whereas loss of microglial MT1 aggravated it. Metabolic reprogramming of microglia was found to contribute to the anti‐inflammatory effects of MT1 activation. LPS‐induced excessive aerobic glycolysis and impaired oxidative phosphorylation (OXPHOS) could be reversed by microglial MT1 activation. MT1 positively regulated pyruvate dehydrogenase alpha 1 (PDHA1) expression to enhance OXPHOS and suppress aerobic glycolysis. Furthermore, in LPS‐treated microglia, MT1 activation decreased the toxicity of conditioned media to the dopaminergic (DA) cell line MES23.5. Most importantly, the anti‐inflammatory effects of MT1 activation were observed in LPS‐stimulated mouse model. In general, our study demonstrates that MT1 activation inhibits LPS‐induced microglial activation through regulating its metabolic reprogramming, which provides a mechanistic insight for microglial MT1 in anti‐inflammation.

Although the underlying mechanisms of microglia activation remains ambiguous, recent studies indicate that metabolic reprogramming have shed new light on inflammation (Kelly & O'Neill, 2015). Homeostatic macrophages generate amounts of adenosine triphosphate (ATP) through mitochondrial oxidative phosphorylation (OXPHOS) (Van den Bossche et al., 2017). Upon inflammatory insults like LPS, macrophages exhibit enhanced aerobic glycolysis (Van den Bossche et al., 2017). Such macrophages generate excessive pro-inflammatory factors (Minton, 2017). Several studies have showed that restriction of aerobic glycolysis remarkably inhibits pro-inflammatory factors production in macrophages (Nikbakht et al., 2019;O'Neill, 2014). Microglia and macrophages have similar characteristics, demonstrating that regulation of microglial metabolic reprogramming might be an effective means to mitigate neuroinflammation.
It has been reported that MT1 receptors are reduced in AD patients (Wu et al., 2007). Silence of Mtnr1a increases the amyloidogenic processing of amyloid precursor protein (Sulkava et al., 2018) and exacerbates mutant huntingtin-mediated toxicity (Wang et al., 2011), indicating that MT1 has a protective role in neurons. Interestingly, both MT1 and MT2 dramatically down-regulated in SN of PD patients (Adi et al., 2010), possibly suggesting their vital roles in the progression of PD. However, it is still largely unknown whether and how microglial MT1 is involved in neuroinflammation.
In this study, we assessed the potential protective role of MT1 activation in inflammatory factor-induced damage of DA neurons.
We found that the non-selective MT1 agonist Ramelteon remarkably inhibits LPS-induced microglial activation, while silencing of Mtnr1a exacerbates microglia-mediated neuroinflammation both in vivo and in vitro. Moreover, we found that microglial metabolic reprogramming might participate in the MT1-mediated control of neuroinflammation.

| Mtnr1a deficiency aggravates LPS-induced production of pro-inflammatory factors in microglia
It is well established that LPS causes microglia to markedly increase their production and secretion of pro-inflammatory factors, such as inducible nitric oxide synthase (iNOS), cyclooxygenase-2 (COX-2), TNFα, and IL-6 (Colonna & Butovsky, 2017). We found MT1 was expressed in primary microglia and the murine microglial cell line BV2 ( Figure S1a,b). We knocked down Mtnr1a, which encodes MT1, in BV2 cells ( Figure S2), to assess whether Mtnr1a deficiency could potentially affect LPS-induced microglial activation. We found that LPS significantly promoted the production of iNOS and COX-2, whereas the transfection of cells with Mtnr1a small interfering RNAs (siRNAs) increased the production without affecting the BV2 cell viability Primary microglia in which Mtnr1a was knocked down, exhibited amoeboid-like characteristics with an expanded cell body, implying its activation (Figure 1g).
Since we have showed that microglial MT1 deficiency could enhance pro-inflammatory factors production in protein levels, we carried out an analysis of gene expression. LPS obviously increased the mRNA levels of iNOS, COX-2, IL-6, and TNFα, with an even greater increase in Mtnr1a-knockdown BV2 cells (Figure 1b-e). These data indicated that Mtnr1a deficiency in microglia aggravates the production of pro-inflammatory factors response to LPS.

| MT agonist Ramelteon inhibits LPS-induced production of pro-inflammatory factors in microglia
Since we found that loss of MT1 in microglia promotes LPS-induced neuroinflammation, we accessed whether MT1 activation could inhibit LPS-induced microglial activation. Because no MT1-specific agonist exists, we used the non-selective MT agonist Ramelteon, which has been used clinically to treat insomnia (Kuriyama et al., 2014). Ramelteon's affinity for MT1 is 10-fold that of MT2 (Zlotos et al., 2014). We found that the abundance of MT1 on microglia was ~3-fold that of MT2 ( Figure S3b), suggesting that Ramelteon indeed acts as an agonist of microglial MT1.
Pretreatment of LPS-stimulated BV2 cells or primary microglia with Ramelteon significantly suppressed the increase of pro-inflammatory proteins like iNOS and COX-2 in a dose-dependent manner without affecting cell viability (Figure 1h,m, and Figure S3c).
Concurrently, the mRNA levels of these pro-inflammatory factors as well as those of IL-6 and TNFα were also downregulated (Figure 1i-l).
These data indicated that MT1 activation could suppress LPS-induced microglial activation.

| The anti-inflammatory effect of Ramelteon is specifically dependent on MT1
To exclude any receptor-independent effects on Ramelteon's antiinflammatory effects, we treated BV2 cells with the MT antagonist Luzindole before Ramelteon treatment. Luzindole inhibited the antiinflammatory effects of Ramelteon in a dose-dependent manner Because Ramelteon is a non-selective MT agonist that activates both MT1 and MT2 (Roth et al., 2005), we utilized MT2 specific agonist (IIK7) and antagonist (4P-PDOT) along with Ramelteon to test whether MT2 is involved in the anti-inflammatory effects of Ramelteon. In LPS-treated BV2 cells, neither IIK7 nor

| The anti-inflammatory effects of MT1 activation are independent of classical GPCR signaling
Since we have showed that MT1 has a significant role in microglial activation, we assessed how microglial MT1 regulates LPS-induced inflammation. It is well-known that MT1 is a classical GPCR, which could activate Gαi or Gαq upon ligand binding (Cecon et al., 2018).
When Gαi is activated, the activity of adenylate cyclase (AC) is restrained, thereby inhibiting the production of second messenger cyclic adenylate (cAMP) (Cecon et al., 2018). Treatment of BV2 cells with Ramelteon decreased cellular cAMP ( Figure S4a), suggesting that this decrease might be involved in the anti-inflammatory effect of MT1 activation. To test this, we used the compound SQ22536, which downregulates the cellular level of cAMP ( Figure S4b), however, LPS-induced microglial activation was not affected by the pretreatment with SQ22536 ( Figure 2g). Thus, the anti-inflammatory effect of MT1 activation is independent on Gαi signaling.
Next, another MT1-related G protein Gαq was also tested. The direct downstream signal produced by Gαq is the enhancement of phospholipase C (PLCβ) activity (Cecon et al., 2018), so we examined whether PLCβ activation participates in the anti-inflammatory effect of MT1 activation. Pretreatment of BV2 cells with the PLCβ inhibitor U73122 prior to administration of Ramelteon and LPS did not affect the anti-inflammatory effects of Ramelteon ( Figure 2h). Thus, Gαq signaling was not involved in the anti-inflammatory effects of MT1 activation.

| Microglial metabolic reprogramming is involved in the anti-inflammatory effect of MT1 activation
A growing body of evidence suggests that the pro-inflammatory factor-mediated activation of microglia requires metabolic reprogramming (in favor of aerobic glycolysis over oxidative phosphorylation) (Gimeno-Bayon et al., 2014;Li et al., 2018;Nair et al., 2019;Orihuela et al., 2016). It has been reported that MT1 is an important modulator on insulin secretion in pancreatic β-cells (Peschke & Muhlbauer, 2010), implying a role of pancreatic β cell-derived MT1 in regulation of glucose metabolism. We used kits to assay both glucose and lactate to measure glucose consumption and the production of extracellular lactic acid in microglia under different treatments. Consistent with previous studies, LPS promoted glucose consumption and lactic acid production in the cell culture medium

| The pyruvate dehydrogenase subunit PDHA1 is involved in MT1-mediated microglial metabolic reprogramming
As we found that MT1 activation inhibits LPS-induced microglial activation through regulating metabolic reprogramming, we wonder the potential mechanisms underlying it. We isolated and sequenced total RNA from Mtnr1a-deficient and normal microglia. A total of 186 genes were upregulated and 384 genes were downregulated in Mtnr1a-deficient microglia compared with the normal group (Figure 4a,b). Based on the Kyoto Encyclopedia of Genes and Genomes (KEGG), many of these genes are associated with certain metabolic pathways, including pyruvate metabolism, the citric acid cycle, and fatty acid metabolism (Figure 4a,c). A heatmap revealed genes with significant changes in expression (i.e., >2-fold) in certain metabolic pathways, and we were amazed to find that Pdha1 expression was decreased substantially in Mtnr1a-deficient microglia ( Figure 4d). Pdha1 encodes the E1 alpha 1 subunit of pyruvate dehydrogenase (PDH) complex which has a critical role in the citric acid cycle (Tan et al., 2015).
To verify RNA sequencing results, we knocked down Mtnr1a or used Ramelteon to treat BV2 cells. At the transcriptional level, Pdha1 expression

F I G U R E 1
Mtnr1a negatively regulates LPS-induced production of pro-inflammatory factors in microglia. (a) BV2 cells were transfected with control siRNA or Mtnr1a siRNA for 48 hr, then exposed to LPS (100 ng/ml) for 12 hr. After treatments, the protein levels of iNOS, COX-2, MT1 and GAPDH were determined using immunoblot analyses. The values were presented as the means ± SEM from three independent experiments. **p < 0.01, two-way ANOVA followed by Sidak's post-hoc test. (b-e) BV2 cells were transfected with control siRNA or Mtnr1a siRNA for 48 hr, then treated with LPS (100 ng/ml) for 6 hr. After treatments, total RNA were collected to detect the mRNA levels of iNOS, IL-6, COX-2 and TNFα by qRT-PCR assays. The values were presented as the means ± SEM from three independent experiments. *p < 0.05, **p < 0.01, two-way ANOVA followed by Sidak's post-hoc test. (f) Primary microglia were treated as (a), the protein levels of iNOS, COX-2, MT1 and GAPDH were determined using immunoblot analyses. The values were presented as the means ± SEM from three independent experiments. **p < 0.01, two-way ANOVA followed by Sidak's post-hoc test. The morphology of primary microglia were showed in (g).
(h) BV2 cells were pretreated with Ramelteon (10 μM, 50 μM, 100 μM) for 12 hr and then exposed to LPS (100 ng/ml) for 12 hr. After treatments, the protein levels of iNOS, COX-2 and GAPDH were measured using immunoblot analyses. The values were presented as the means ± SEM from three independent experiments. **p < 0.01 vs. the group treated with LPS alone, two-way ANOVA followed by Sidak's post-hoc test. (i-l) BV2 cells were pretreated with Ramelteon (100 μM) for 12 hr and then exposed to LPS (100 ng/ml) for 6 hr. After treatments, total RNA were collected to measure the mRNA levels of iNOS, IL-6, COX-2 and TNFα by qRT-PCR assays. The values were presented as the means ± SEM from three independent experiments. **p < 0.01 vs. the group treated with LPS alone, two-way ANOVA followed by Sidak's post-hoc test. (m) Primary microglia were treated as (h), the protein levels of iNOS, COX-2, MT1 and GAPDH were determined using immunoblot analyses. The values were presented as the means ± SEM from three independent experiments. **p < 0.01, two-way ANOVA followed by Sidak's post-hoc test. The morphology of primary microglia were showed in (n)

| MT1 activation inhibits microglia-mediated DA neuronal death
It is well accepted that activated microglia could release a number of pro-inflammatory factors that damage neighboring neurons (Wolf et al., 2017). As we showed that MT1 activation has remarkable antiinflammatory effects, we speculated that MT1 may mediate the neurotoxicity caused by activated microglia. To verify our speculation, we 2.9 | Ramelteon inhibits DA neuronal loss and microglial activation in MPTP-induced PD mouse model.
To further explore the protective effects of MT1 activation on DA neuronal loss in vivo, we performed the MPTP-induced PD model.
A seven consecutive days Ramelteon treatment were conducted F I G U R E 2 The anti-inflammatory effect of Ramelteon is specially dependent on MT1. (a) BV2 cells were pretreated with Luzindole (200 μM, 400 μM) for 2 hr, followed by the treatment of Ramelteon for 12 hr (100 μM), then were exposed to LPS (100 ng/ml) for 12 hr. After treatments, the protein levels of iNOS, COX-2 and GAPDH were measured using immunoblot analyses. The values were presented as the means ± SEM from three independent experiments. *p < 0.05, **p < 0.01 vs. the group treated with LPS +Ramelteon, two-way ANOVA followed by Sidak's post-hoc test. (b) BV2 cells were pretreated with IIK7 (100 μM) for 12 hr, followed by the administration of LPS for 12 hr. Then, the protein levels of iNOS, COX-2 and GAPDH were measured using immunoblot analyses. The values were presented as the means ± SEM from three independent experiments, ns, no significance, two-way ANOVA followed by Sidak's post-hoc test. (c) BV2 cells were pretreated with 4P-PDOT (200 μM) for 2 hr, followed by the treatment of Ramelteon for 12 hr (100 μM), then were exposed to LPS (100 ng/ml) for 12 hr. After treatments, the protein levels of iNOS, COX-2 and GAPDH were measured using immunoblot analyses. The values were presented as the means ± SEM from three independent experiments, ns, no significance, two-way ANOVA followed by Sidak's post-hoc test. (d) BV2 cells were transfected with control siRNA or Mtnr1a siRNA for 48 hr, followed by the administration of Ramelteon for 12 hr and then were exposed to LPS (100 ng/ml) for 12 hr. After treatments, the protein levels of iNOS, COX-2 and GAPDH were measured using immunoblot analyses. The values were presented as the means ± SEM from three independent experiments. *p < 0.05, **p < 0.01, ns, no significance, two-way ANOVA followed by Sidak's post-hoc test. (e, f) BV2 cells were transfected with control siRNA or Mtnr1a siRNA for 48 hr, followed by the administration of Ramelteon for 12 hr and then were exposed to LPS (100 ng/ml) for 6 hr. After treatments, total RNA were collected to detect the mRNA levels of TNFα and IL-1β by qRT-PCR assays. The values were presented as the means ± SEM from three independent experiments. *p < 0.05, **p < 0.01, ns, no significance, two-way ANOVA followed by Sidak's post-hoc test. (g) BV2 cells were pretreated with SQ22536 (10 μM, 50 μM, 100 μM) for 12 hr and then exposed to LPS (100 ng/ml) for 12 hr. After treatments, the protein levels of iNOS, COX-2 and GAPDH were measured using immunoblot analyses. The values were presented as the means ± SEM from three independent experiments, ns, no significance, two-way ANOVA followed by Sidak's post-hoc test. (h) BV2 cells were pretreated with U73122 for 2 hr, followed by the administration of Ramelteon for 12 hr, then were exposed to LPS (100 ng/ml) for 12 hr. Thereafter, the protein levels of iNOS, COX-2 and GAPDH were measured using immunoblot analyses. The values were presented as the means ± SEM from three independent experiments, ns, no significance, two-way ANOVA followed by Sidak's post-hoc test   (Tan et al., 2015). Many studies have shown that PDHA1 regulates metabolic reprogramming in macrophages (Semba et al., 2016;Tan et al., 2015). Interestingly, PDHA1 has been shown to be substantially downregulated in the SN of PD patients (Miki et al., 2017).

| DISCUSS ION
PDHA1 is a mitochondrial localization factor for nuclearencoded proteins, but it remains unclear how MT1 regulates Pdha1 expression. One study reported that the MT1 also exists on the mitochondrial membrane in neurons (Suofu et al., 2017), suggesting MT1 may directly combined with PDHA1 in microglial mitochondria and regulated its expression. In addition we found that MT1 regulates PDHA1 abundance at both the translational and transcriptional F I G U R E 3 Microglial metabolic reprogramming is involved in the anti-inflammatory effect of MT1 receptor activation. (a) BV2 cells were transfected with control siRNA or Mtnr1a siRNA for 48 hr, then exposed to LPS (100 ng/ml) for 12 hr. After treatments, the culture media were collected and were used to detect the glucose consumption by the QuantiChrom™ Glucose Assay kits. The values were presented as the means ± SEM from three independent experiments. **p < 0.01, two-way ANOVA followed by Sidak's post-hoc test. (b) BV2 cells were treated as described in (a) and then the culture media were collected to determine the lactate production using Lactate Assay kits. The values were presented as the means ± SEM from three independent experiments. *p < 0.05, **p < 0.01, two-way ANOVA followed by Sidak's post-hoc test. (c) BV2 cells were pretreated with Ramelteon (10 μM, 50 μM, 100 μM) for 12 hr and then exposed to LPS (100 ng/ml) for 12 hr. After treatments, the culture media were collected and were used to detect the glucose consumption by the QuantiChrom™ Glucose Assay kits. The values were presented as the means ± SEM from three independent experiments. *p < 0.05, **p < 0.01, two-way ANOVA followed by Sidak's post-hoc test. (d) BV2 cells were treated as described in (c) and then the culture media were collected to determine the lactate production using Lactate Assay kits. The values were presented as the means ± SEM from three independent experiments. *p < 0.05, **p < 0.01, two-way ANOVA followed by Sidak's post-hoc test. (e, f) Primary microglia were treated as (a) and (b), after treatments, the culture media were collected and were used to detect the glucose consumption and lactate production by the QuantiChrom™ Glucose Assay kits and Lactate Assay kits. (g-h) Primary microglia were treated as (c) and (d), after treatments, the culture media were collected and were used to detect the glucose consumption and lactate production by the QuantiChrom™ Glucose Assay kits and Lactate Assay kits. (i) BV2 cells were pretreated with 2-DG for 2 hr, followed by the knocking down of Mtnr1a and a sequential LPS stimulation. Next, the protein levels of iNOS, COX-2 and GAPDH were measured using immunoblot analyses. Quantitative analyses of panel (i) were shown in panel below. The values were presented as the means ± SEM from three independent experiments. *p < 0.05, **p < 0.01, ns, no significance, two-way ANOVA followed by Sidak's post-hoc test
However, our study shows that MT1 has no significant effect on The values were presented as the means ± SEM from three independent experiments. **p < 0.01 vs. si-Control group, two-way ANOVA followed by Sidak's post-hoc test. (f) BV2 cells were treated with Ramelteon (100 μM) in a time-dependent manner, then the mRNA levels of Pdha1 and HK2 were determined by qRT-PCR assays. The values were presented as the means ± SEM from three independent experiments. **p < 0.01 vs. the group treated with Ramelteon for 0 hr, two-way ANOVA followed by Sidak's post-hoc test. (g) BV2 cells were treated as described in (e), then, immunoblot analyses were used to detect the protein levels of PDHA1 and HK2. The values were presented as the means ± SEM from three independent experiments. **p < 0.01 vs. si-Control group, two-way ANOVA followed by Sidak's post-hoc test. (h) BV2 cells were treated as described in (f), then, immunoblot analyses were used to detect the protein levels of PDHA1 and HK2. *p < 0.05, **p < 0.01 vs. the group treated with Ramelteon for 0 hr, two-way ANOVA followed by Sidak's post-hoc test. (i) Primary microglia were treated as (h), immunoblot analyses were used to detect the protein levels of PDHA1. (j) Primary microglia were treated as (g), immunoblot analyses were used to detect the protein levels of PDHA1. (k) BV2 cells were transfected with control siRNA or Pdha1 siRNA for 48 hr, then with a 12-h Ramelteon treatment, followed by LPS stimulation for 12 hr. The protein levels of iNOS and COX-2 were detected with immunoblot analyses. (l, m) BV2 cells were transfected with control siRNA or Pdha1 siRNA for 48 hr, then with a 12-hr Ramelteon treatment, followed by LPS stimulation for 6 hr. After treatments, total RNA were collected to detect the mRNA levels of TNFα and IL-1β by qRT-PCR assays. The values were presented as the means ± SEM from three independent experiments. *p < 0.05, **p < 0.01, ns, no significance, twoway ANOVA followed by Sidak's post-hoc test 4 | E XPERIMENTAL PROCEDURE S

| Animal experiments
Male C57BL/6 mice (age 6-8 months, 20-25 g) were purchased from JIHUI Animal Ltd. The mice were housed on the conditions of 20-26°C, 50%-60% relative humidity and a 12-hr light and a 12-hr dark cycle for 2 weeks before experimentation. Mice had free access to water and as described before

| Cell viability assay
3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assays were performed to assess the cell viability. BV2 cells were pretreated with varying concentrations of Ramelteon or with siR-NAs against Mtnr1a, followed by the LPS administration. Thereafter, BV2 cells were incubated with 0.5 mg/ml MTT at 37°C for 2 hr. The reaction was then stopped with 150 μl of DMSO. The absorbance was detected at 570 nm to determine the cell viability.

| RNA interference
RNAiMAX ( Relative mRNA expression levels were calculated using the 2 ΔΔC t method.

| Glucose and lactate detection
After different treatments of BV2 cells or primary microlgia, the cul-

| Immunoblot analysis and antibodies
After treatments, BV2 cells and primary microlglia or isolated tissues anti-TH (Millipore). The secondary antibodies, sheep anti-rabbit or anti-mouse IgG-HRP, were purchased from Thermo Fisher Scientific.
The proteins were visualized using ECL detection kits (Thermo Fisher Scientific).

| Immunofluorescence Staining
BV2 cells or Primary microglia were fixed with 4% paraformaldehyde in PBS, followed by permeabilizing with 0.25% Triton X-100 in PBS for 5 min at room temperature. Then, the cells were washed with PBS for three times and blocked with 0.5% fetal bovine serum. Thereafter, the cells were incubated with anti-Iba1 (Wako Chemicals) or anti-MT1 (Santa Cruz Biotechnology) for 6 hr at room temperature, followed by an incubation with rhodamine-conjugated (red) -or FITC (green)-conjugated secondary antibody for 2 hr.
Subsequently, the cells were stained with DAPI (Sigma) for 5 min.
Finally, the cells were observed using an inverted IX71 microscope system (Olympus).

| Immunohistochemistry
After the treatments of Ramelteon and LPS or MPTP described above, the mice were perfused with 0.9% saline followed by 4% paraformaldehyde in 0.1 M PBS (pH 7.4). The mice brains were then removed and post-fixed in the same fixation agent overnight at 4°C, followed by a treatment with 30% sucrose at 4°C with another night. Serial 20 μM-thick mouse midbrain slices were cut with a freezing microtome.
Immunohistochemical staining was conducted with anti-MT1 antibody, anti-Iba1antibody (Wako Chemicals), anti-GFAP, and anti-TH antibodies (Millipore) against six slices per mouse (120 μm interval). After incubation with primary antibodies at room temperature for 6 hr, the slices were incubated with rhodamine (red)-or FITC (green)-conjugated secondary antibody (Invitrogen) for 2 hr. Thereafter, the slices were stained with DAPI for 5 min and observed using an inverted IX71 microscope system (Olympus). The number of TH + neurons and the fluorescence intensity of MT1, Iba1, and GFAP were counted using Image J (National Institute of Health).   (c) Immunohistochemical staining was performed via anti-Iba1 and anti-GFAP antibodies. Scale bar, 50 μm. n = 5 per group. **p < 0.01 vs. the group treated with LPS alone using two-way ANOVA followed by Sidak's post-hoc test. (d) Immunohistochemical staining was conducted using anti-TH antibodies. The quantification of TH + cell numbers was shown in below. Scale bar, 50 μm. n = 4 per group. *p < 0.05 vs. the group treated with LPS alone, two-way ANOVA followed by Sidak's post-hoc test. (e) Mice midbrains were isolated and the proteins were collected to measure the protein levels of Iba1 and TH in different groups using immunoblot analyses. Quantitative analyses of panel (h) were shown in panel below. n = 3 per group. *p < 0.05 vs. the group treated with LPS alone, two-way ANOVA followed by Sidak's post-hoc test. (f) The co-location of MT1 (green) and Iba1 (red) in microglia in mice SN. Scale bar, 10 μm, n = 4 per group, the relative fluorescence intensity of MT1 to Iba1 was shown in below. *p < 0.05 vs. the group treated with MPTP using one-way ANOVA followed by Sidak's post-hoc test. (g) Immunohistochemical staining was conducted using anti-TH antibodies. The quantification of TH + cell numbers was shown in below. Scale bar, 50 μm. n = 4 per group. **p < 0.01 vs. the group treated with MPTP alone, two-way ANOVA followed by Sidak's post-hoc test. (h) Mice SN were isolated and the proteins were collected to measure the protein levels of TH in different groups using immunoblot analyses. Quantitative analyses were shown in right. n = 3 per group. **p < 0.01 vs. the group treated with MPTP alone, two-way ANOVA followed by Sidak's post-hoc test. (i) Immunohistochemical staining was performed via anti-Iba1 and anti-GFAP antibodies and the relative fluorescence intensity of Iba-1 and GFAP were shown in right. Scale bar, 50 μm. n = 3 per group. **p < 0.01 vs. the group treated with MPTP alone using two-way ANOVA followed by Sidak's post-hoc test drugs, and chemical substances (http://en.wikip edia.org/wiki/KEGG).

| RNA sequencing and bioinformatics analyses
We used scripts in house to enrich significant differential expression gene in KEGG pathways.

| Statistical analysis
Two-way ANOVA followed by Sidak's post-hoc test was performed to analyze differences in different treatments. All analyses were performed using GraphPad Prism version 7.00 (GraphPad Software). P values <0.05 was considered as significant difference. All values are displayed as the mean ± SEM.

CO N FLI C T O F I NTE R E S T
The authors declare no competing financial interests.

AUTH O R CO NTR I B UTI O N S
C.-F.L. and C.G. designed the study. C.G. drafted the manuscript. C.G. analyzed the data. F.W. contributed critical reagents.

DATA AVA I L A B I L I T Y S TAT E M E N T
Our RNA-seq RAW data are available in: https://eur03.safel inks.prote ction.outlo ok.com/?url=https %3A%2F%2Fwww.