Downregulation of miR‐7116‐5p in microglia by MPP+ sensitizes TNF‐α production to induce dopaminergic neuron damage

Activation of microglia resulting in exacerbated inflammation expression plays an important role in degeneration of dopaminergic (DA) neurons in the pathogenesis of Parkinson's disease (PD). However, how this enhanced inflammation is induced in microglia remains largely unclear. Here, in the mouse PD model induced by 1‐methyl‐4‐phenyl‐1,2,3,6‐tetra hydropyridine (MPTP), we found that miR‐7116‐5p in microglia has a crucial role in this inflammation. 1‐methyl‐4‐phenylpyridinium (MPP+) is uptaken by microglia through organic cation transporter 3 (OCT3) to downregulate miR‐7116‐5p, an miRNA found to target tumor necrosis factor alpha (TNF‐α). Production of TNF‐α in microglia is specifically potentiated by MPP+ via downregulation of miR‐7116‐5p to elicit subsequent inflammatory responses. Furthermore, enhancement of miR‐7116‐5p expression in microglia in mice inhibits the production of TNF‐α and the activation of glia, and further prevents loss of DA neurons. Together, our studies suggest that MPP+ suppresses miR‐7116‐5p level in microglia and potentiates TNF‐α production and inflammatory responses to contribute to DA neuron damage.

inhibits the oxidative phosphorylation in mitochondria to damage DA neurons (Blum et al., 2001). Besides inhibition of the oxidative phosphorylation in mitochondria, a wealth of evidence uncovers the participation of inflammation resulting from microglial activation in loss of DA neurons during administration of MPTP (Barcia et al., 2004;Hirsch & Hunot, 2009;Kanaan, Kordower, & Collier, 2008;Langston et al., 1999;Liberatore et al., 1999). In PD patients suffering from MPTP, there is a considerable microglial activation in substantia nigra (SN) (Langston et al., 1999). Additionally, reactive microglia in SN could release proinflammatory and oxidative factors and further contribute to the damage of DA neurons in MPTP model (Liberatore et al., 1999;Main et al., 2016;Martin et al., 2016;Sriram et al., 2006;Wu et al., 2002). Therefore, it is promising to clarify the process of microglial activation in pathological condition to reduce the inflammation response and delay the degeneration of DA neurons.
Although the activation of microglia by the factors released from stressed DA neurons has been widely investigated (Kim et al., 2007;Wang et al., 2006), whether MPP 1 could directly affect microglial inflammatory responses remains unclear.
Among the proinflammatory factors released by microglia in inflammatory responses, TNF-a has a prominent role in neuroinflammation during the pathogenesis of PD. Post-mortem brains and the cerebrospinal fluid of PD patients display increased amounts of TNF-a and its receptors (Boka et al., 1994;Mogi et al., 1994b). Additionally, a rare form of early-onset idiopathic PD has been associated with a single nucleotide polymorphism (SNP) in the TNF promoter, which predictably increases TNF-a expression (Nishimura et al., 2001). The TNF-a production is also controlled by miRNAs at post-transcriptional level (Guan et al., 2015;Tili et al., 2007;Zhang, Dong, Li, Hong, & Wei, 2012). Furthermore, in mouse MPTP model, the expression of dopamine and tyrosine hydroxylase (TH) was remarkably rescued by inhibition of TNF-a or ablation of its receptors (Ferger, Leng, Mura, Hengerer, & Feldon, 2004;Sriram et al., 2002).
In this study, we aim to investigate whether MPP 1 can directly affect the expression of proinflammatory factors, especially TNF-a, in microglia in mouse MPTP model. We report now that MPP 1 could be uptaken by microglia through organic cation transporter 3 (OCT3, Slc22a3), and specifically increase the production of TNF-a by downregulating miR-7116-5p to promote the death of DA neurons.

| Animals
All animal experiments were approved by the Institutional Animal

| AAV delivery and TNF-a antibody infusions
To package the miR-7116 adeno-associated virus (AAV), the fragment containing pri-miR-7116 sequence (listed in Supporting Information, Table 1) was ligated into pAOV-CMV-bGlobin-LSL-EGFP plasmid. The titers of AAV particles were between 2 3 10 12 and 4 3 10 12 units/ml. Mice were anesthetized by one intraperitoneal injection of 0.7% pentobarbital sodium and immobilized in a stereotaxic apparatus. The AAV (10 9 , 1 ll) was slowly injected into the substantia nigra (SN) at a rate of 0.1 ll/min by a 5 ll Hamilton syringe. The coordinates of the SN are AP 23.3 mm, ML 61.3 mm, DV 24.7 mm from bregma.
For TNF-a antibody infusions, the drug-delivery tubes were inserted into the lateral ventricles and fixed with a mixture of acrylic and dental cement. After 7-day recovery period, TNF-a antibody (2 lg, 1 lg/ll) was injected intracerebroventricularly at a rate of 0.2 ll/min.

| Immunofluorescence and image analysis
Anesthetized mice were perfused with 0.9% saline and 4% paraformaldehyde in 1 mM phosphate buffer (PB). The brains were incubated in 4% paraformaldehyde over night and then dehydrated in 30% sucrose for 48 hr. Frozen sections (30 lm) of the entire midbrain were prepared with a freezing microtome and then washed with phosphatebuffered saline (PBS) followed by incubation with 10% BSA and 0.5% Triton X-100 in PBS for 1 hr at room temperature. Primary antibodies (anti-tyrosine hydroxylase (TH) 1:1,000, anti-ionized calcium-binding adapter molecule 1 (Iba1) 1:500, anti-glial fibrillary acidic protein (GFAP) 1:1,000, anti-green fluorescent protein (GFP) 1:1,000) were then added, and incubated at 48C overnight. Next day, the brain sections were washed in PBS followed by incubation with secondary antibodies (1:1,000) conjugated with fluorophore at room temperature for 2-3 hr. Every fourth section of the entire slices was chosen for determination of the number of TH-positive cells in SN and analysis of the fluorescence intensity of Iba1 and GFAP. The TH-positive cells in SN were defined as reported previously (Sauer, Rosenblad, & Bjorklund, 1995), and the cell number was counted by Image Pro Plus.
The fluorescence expression of Iba1 and GFAP was defined as the average intensity of glial scar area in MPTP-injected mice or the SN area in saline-injected mice, which was quantified by ImageJ.

| Isolation of the adult microglia and astrocytes in ventral midbrain
The isolated mouse ventral midbrains were grounded to pass to a 70 lm nylon cell strainer and then digested in a mixture of collagenase D (0.05%, Sigma, C5138) and DNase I (0.025 U/ml, Sigma) for 1 hr at room temperature. After washed with Hank's Balanced Salt Solution (HBSS, Invitrogen) to remove digestion mixture, cells were resuspended and stained with allophycocyanin (APC)-labeled CD45/Alexa Fluor 488-labeled CD11b, APC-labeled GLAST antibodies, or Alexa Fluor 488/APC-labeled isotype at 48C for 1 hr. Flow cytometry was then performed for sorting CD11b 1 /CD45 l 8 w , or GLAST 1 population.
2.6 | RNA isolation, reverse transcription, and qRT-PCR Briefly, cells or tissues were harvested in 1 ml TRIzol to extract the total RNA. The cDNA was synthesized by reverse transcription (Promega) with random primers for mRNA or stem loop for miRNAs. The levels of miRNA and mRNA were quantified by quantitative real-time PCR (qRT-PCR) using SYBR Green Realtime PCR Master Mix (Toyobo).

| ELISA
The expression of TNF-a and IL-6 was measured by ELISA kit (eBioscience) in the cell supernatant following the vender protocol. Standard curve was made to quantify the concentration of TNF-a and IL-6 from different samples.

| Chromatin immunoprecipitation
Chromatin immunoprecipitation (ChIP) was performed as per the manufacturer's instructions (Millipore,. Briefly, 1 3 10 7 BV2 cells were cross-linked by 1% formaldehyde for 10 min at room temperature and then terminated with 0.125 M glycine. Cells were resuspended with a lysis buffer (1% sodium dodecyl sulfate, 10 mM EDTA, 50 mM Tris-HCl, pH 8.1) and incubated for 30 min on ice after washed twice with cold PBS and centrifuged at 3,000 rpm for 2 min at 48C. Chromatin was sheared into 200 to 1,000 bp fragments by sonication. Cell debris was removed by centrifugation at 13,200 rpm for 10 min at 48C and sheared chromatin was incubated with anti-RNA pol II antibody (Santa Cruz Biotech) over night at 48C and then with protein G-Sepharose for 2 hr at room temperature. DNA was eluted in the elution buffer (1% sodium dodecyl sulfate, 0.1 M NaHCO 3 ) and then reverse cross-linked at 658C for 6 hr followed by proteinase K treatment at 558C for 2 hr. Last, DNA was extracted with phenolchloroform for PCR analysis.

| Luciferase activity assays
For TNF promoter activity determination, BV2 cells were transfected with TNF promoter (listed in Supporting Information, Table 1) cloned into a luciferase expression system for 24 hr, and then treated with MPP 1 and/or LPS as described above. To verify the regulation of TNFa by miR-7116-5p, the luciferase reporter constructs containing the 3 0 UTR of mouse TNF-a (listed in Supporting Information, Table 1) and its seed sequence mutants were co-introduced with miR-7116-5p mimics (100 nM) in HEK293 cells for 24 hr. The luciferase activity was determined by Dual-Luciferase ® Reporter Assay System (Promega, E1910).

| Small RNA sequencing
Total RNA was extracted from BV2 cells by TRIzol reagent (Invitrogen) and the quality of RNA was assessed by Agilent 2200. The enriched small RNA library was then sequenced on the Illumina HiSeq 2500 performed by NovelBio Bio-Pharm Technology Company. The raw data were analyzed by EB-Sequencing algorithms to achieve the differentially expressed miRNAs (Dif_miRNAs). The Dif_miRNAs were selected according to the fold changes of two groups more than 1.5 and the false discovery rate (FDR) <0.05. The count number of small RNA library in each group was listed in Supporting Information, min. To stop the reaction, the medium was rapidly replaced with cold EC buffer and followed by two washes, and 1 M NaOH was added to cells. The radioactivity was detected using a liquid scintillation counter.

| Northern blot analysis
The RNA extracted from BV2 cells was loaded to a 15% denaturing polyacrylamide-urea gel and transferred to a Nytran nylon transfer membrane (Ambion, Austin, TX). The membranes were cross-linked and prehybridized for at least 1 hr with Ultra-Hybsolution (Ambion), followed by hybridization to the locked nucleic acid (LNA) digoxin  The miR-7116-5p mimics (0.2-1 nM in BV2 cells and 100 nM for luciferase assays in HEK293 cells) and siRNAs (100 nM) were introduced into the cells by using siIMPORTER (Millipore, 64-101) according to the manufacturer's instructions. All plasmids were introduced into the cells using Lipofectamine 2000 (Invitrogen) following manufacturer's protocol.

| Statistical analysis
All data were presented as mean 6 SEM. Comparisons between groups for statistical significance were performed with student's t test or analysis of variance (ANOVA) with post hoc test in multiple groups by using Graph-Pad Prism 5.0. Results were considered significant difference at *p < 0.05; **p < 0.01; ***p < 0.001; #p < 0.05; ##p < 0.01; ###p < 0.001.

| Microglia uptakes MPP 1 by OCT3
We initially investigated whether microglia can uptake MPP 1 and found that both the primary cultured microglia (Figure 1a,b) and a  5 4). CTRL, DMSO. ns, no significance. All data were presented as mean 6 SEM. All comparisons between groups for statistical significance were performed with one-way analysis of variance (ANOVA) with Tukey's post hoc test. ***p < 0.001 versus NC or CTRL mouse microglia cell line, BV2 cells, (Supporting Information, Figure 1a, b), uptake [ 3 H]-MPP 1 in a time-and dose-dependent manner. It has been reported that DAT and OCT3 can transport MPP 1 into DA neurons and astrocytes (Bezard et al., 1999;Cui et al., 2009;Gainetdinov et al., 1997). When microglia ( Figure 1C  (b-e) After treatment with MPP 1 or/and LPS, qRT-PCR examination of mRNA levels of TNF-a, IL-6, IL-1b, iNOS, and COX-2 in BV2 cells (b, TNF-a, IL-6, and IL-1b, n 5 5, iNOS and COX-2, n 5 4), primary cultured microglia (c, TNF-a, n 5 5, IL-6, n 5 6), astrocytes (d, n 5 4) and cortical neurons (e, n 5 4). (f) ELISA examination of protein levels of TNF-a and IL-6 from the supernatant in primary cultured microglia (TNF-a, n 5 4, IL-6, n 5 7) treated with MPP 1 or/and LPS. (g) Effects of knocking down OCT3 by si-1050 on mRNA levels of TNF-a after MPP 1 or/and LPS in primary cultured microglia (n 5 4), si-NC as a negative control (NC). CTRL, vehicle. ns, no significance. All data were presented as mean 6 SEM. Comparisons between groups for statistical significance were performed with one-way ANOVA with Tukey's post hoc test (b-f) or two-way ANOVA with Bonferroni post hoc test (g). *p < 0.05, **p < 0.01, ***p < 0.001 versus NC or CTRL. #p < 0.05, ##p < 0.01, ###p < 0.001 versus LPS group HE ET AL. | 1255 3.2 | The TNF-a production by microglia is potentiated by MPP 1 We next explored the possible role of MPP 1 uptaken by microglia and examined the expression of proinflammatory factors in these cells. The mRNA levels of the inflammatory factors, including TNF-a, interleukin-1b (IL-1b), interleukin-6 (IL-6), and interferon-g (IFN-g), all were not changed when the cells were incubated with MPP 1 for 12 hr, indicating that MPP 1 alone did not induce the production of the proinflammatory factors in these cells (Figure 2a-c). It has been reported that microglia is activated by the factors released from stressed or dead DA neurons in the MPTP model (Kim et al., 2007). We then investigated whether MPP 1 could regulate activation of microglia in inflammation. Treatment of microglia and BV2 cells with lipopolysaccharides (LPS) to mimic the factors released from stressed or dead DA neurons led to an increase in mRNA level of TNF-a ( Figure 2a) and this increase was greatly potentiated by MPP 1 pretreatment (Figure 2b,c). In contrast, there were no differences in mRNA levels of IL-1b, IL-6, or IFN-g between MPP 1 with LPS group and LPS group (Figure 2b,c). Additionally, mRNA levels of cyclooxygenase (COX-2) and inducible nitric oxide synthase (iNOS), enzymes associated with inflammation (Simon, 1999;Vane et al., 1994), were not different between these two groups (Figure 2b,c). As astrocytes and neurons are also responsible for production of TNF-a and other cytokines (Forloni, Mangiarotti, Angeretti, Lucca, & De Simoni, 1997;Liu et al., 1994), we assessed the MPP 1 potentiation of TNF-a production in these cells. However, mRNA level of TNF-a was not potentiated by MPP 1 in astrocytes or neurons (Figure 2d,e). Moreover, 6-OHDA is also a selective catecholaminergic neurotoxin that induces DA neuron damage (Ungerstedt, 1968). We thus treated microglia with 6-OHDA to explore the specificity of potentiation of TNF-a expression by MPP 1 . As shown in Supporting Information, Figure 2c, 6-OHDA pretreatment did not affect the TNF-a expression induced by LPS. We next measured TNF-a protein level by enzyme-linked immunosorbent assay (ELISA) (Figure 2a). Consistently, protein level of TNF-a was also notably potentiated by MPP 1 in microglia and BV2 cells. In contrast, IL-6 protein level was not changed between MPP 1 with LPS group and LPS group (Figure 2f and Supporting Information, Figure 2a). Therefore, MPP 1 specifically potentiated TNF-a production in microglia.
To study whether OCT3 was involved in the MPP 1 potentiation of TNF-a production, we introduced si-1050 into microglia and BV2 cells to knock down OCT3. The potentiation of TNF-a production by MPP 1 was not observed when OCT3 was knocked down in microglia and BV2 cells (Figure 2g and Supporting Information, Figure 2b). Collectively, these results point to a possibility that MPP 1 was uptaken via OCT3 in microglia to potentiate the subsequent TNF-a production induced by LPS.

| Transcription of TNF-a is not affected by MPP 1
In the next attempt to study whether the MPP 1 potentiation of TNF-a production was due to the enhancement of its transcription activity, chromosome immunoprecipitation (ChIP) was conducted to evaluate the transcriptional activity of TNF-a promoter. By quantitative realtime PCR (qRT-PCR) analysis, we found that there was no difference in the level of TNF-a promoter pulled down by Polymerase II (Pol II) between MPP 1 with LPS group and LPS group (Figure 3a). Furthermore, we constructed a luciferase reporter to monitor TNF-a promoter activity by inserting the mouse TNF-a promoter into the start of the gene of luciferase. In accordance with the result of Pol II immunoprecipitation, the expression level of luciferase was not changed after MPP 1 with LPS, compared to LPS alone (Figure 3b). Together, these results are consistent with an explanation that increased level of TNFa mRNA by MPP 1 was not due to the enhancement in its transcriptional activity.

| The miR-7116-5p is responsible for TNF-a overproduction by MPP 1
It has been known that mRNA of TNF-a can be regulated by micro-RNAs (miRNAs) at post-transcription level (Guan et al., 2015;Tili et al., 2007;Zhang et al., 2012). Therefore, we examined the role of miRNAs in MPP 1 potentiation of TNF-a production in microglia. We then performed small RNA sequencing to search miRNA candidates that could target TNF-a and be responsible for the MPP 1 potentiation of TNF-a production. After careful analysis, we found that mmu-miR-7116-5p, whose seed sequence could perfectly target to the 3 0 untranslated regions (UTR) of mouse TNF-a, was downregulated by MPP 1 (Figure   4a). To assess the possible regulation of TNF-a by miR-7116-5p, the 3 0 UTR of mouse TNF-a and its seed sequence mutants (AT exchange mutant, GC exchange mutant, and 8 nucleotides (8 nts) mutant) were cloned into a luciferase reporter construct. We found that miR-7116-5p largely inhibited the luciferase activity in HEK293 cells transfected with wild-type 3 0 UTR of mouse TNF-a. When transfecting HEK293 cells with mutants, the inhibition of luciferase activity by miR-7116-5p was partially (AT and GC exchange mutant) or totally (8 nts mutant) stopped (Figure 4b).
To further verify the regulation of TNF-a by miR-7116-5p in response to MPP 1 and LPS, mimics of miR-7116-5p were introduced into microglia and BV2 cells, and expression of TNF-a was then analyzed by qRT-PCR and ELISA. The miR-7116-5p greatly suppressed the MPP 1 potentiation of TNF-a production, compared to the group transfected with control mimics (Figure 4c,e and Sequence alignment of miR-7116-5p and its target sites in 3 0 -UTR of mouse TNF-a. The different mutants of mouse TNF-a 3 0 -UTR were listed above the wild-type (WT) form (top panel). The luciferase activity of HEK293 cells transfected with WT or mutants and miR-7116-5p mimics for 24 hr (bottom, n 5 6). (c-f) Examination of mRNA and protein levels of TNF-a and IL-6 in microglia (c, n 5 4; d, n 5 6; e, n 5 6; f, n 5 6) transfected with miR-7116-5p mimics for 24 hr and then treated with MPP 1 (50 lM, 12 hr) or/and LPS (50 ng/ml, 4 hr for qRT-PCR, 12 hr for ELISA). NC, negative microRNA mimic control. CTRL, vehicle. ns, no significance. All data were presented as mean 6 SEM. All comparisons between groups for statistical significance were performed with two-way ANOVA with Bonferroni post hoc test. **p < 0.01, ***p < 0.001 versus NC or CTRL. #p < 0.05, ###p < 0.001 versus LPS group [Color figure can be viewed at wileyonlinelibrary.com] HE ET AL.

| The miR-7116-5p is specifically downregulated in microglia in mouse MPTP model
We next verified the MPP 1 downregulation of miR-7116-5p by northern blot and qRT-PCR. The level of miR-7116-5p was markedly suppressed in microglia and BV2 cells incubated with MPP 1 (Figure 5a,b and Supporting Information, Figure 4a), whereas its level was not affected when incubated with LPS (Figure 5b and Supporting Information, Figure 4a), consistent with the results obtained by small RNA sequencing. However, the level of miR-125b-5p, a miRNA known to target TNF-a mRNA (Guan et al., 2015;Tili et al., 2007), was not affected by MPP 1 (Figure 5c and Supporting Information, Figure 4b).
In addition, no difference was found in the level of miR-7116-5p in microglia incubated with 6-OHDA (Supporting Information, Figure 5).
Moreover, suppression of miR-7116-5p by MPP 1 was largely prevented when the OCT3 was downregulated by si-1050 in microglia and BV2 cells (Figure 5d and Supporting Information, Figure 4c).
Collectively, all these results are consistent with an explanation that enhancement of miR-7116-5p expression suppresses the overproduction of TNF-a to reduce the inflammatory activation, which then prevents loss of DA neurons in the MPTP model.

| D I SCUSSION
Here, we report a novel mechanism by which MPP 1 potentiation of TNF-a production in microglia contributes to damage of DA neurons.
Several lines of evidence support this conclusion. First, microglia uptook MPP 1 through OCT3. Second, MPP 1 alone did not induce the production of the proinflammatory factors, but specifically potentiated the production of TNF-a without affecting the expression of IL-1b, IL-6, COX-2, or iNOS in microglia. Third, MPP 1 potentiated TNF-a production via downregulation of miR-7116-5p. Last, elevating the expression of miR-7116-5p specifically in microglia prevented the overproduction of TNF-a and reduced loss of DA neurons in the mouse MPTP model. Together, our results suggest a model that in addition to its direct inhibitory effects on neuronal survival, MPP 1 could also damage the DA neurons by potentiating TNF-a production in microglia in a manner that is dependent on the inflammation induced by DA neuron damage (Figure 7). Therefore, inhibiting the reduction of miRNAs to suppress the exacerbated inflammation induced by MPP 1 or other environmental neurotoxins could be a promising strategy to prevent neurodegeneration of PD.
It has been reported that OCT3 is expressed in neurons and astrocytes (Cui et al., 2009;Shang, Uihlein, Van Asten, Kalyanaraman, & Hillard, 2003). However, MPP 1 in microglia, but not in astrocytes or neurons, potentiated the production of TNF-a. It is not clear at the present why MPP 1 specifically potentiated the TNF-a production in microglia. A possible explanation is that MPP 1 metabolized from MPTP in astrocytes is released to extracellular spaces (Di Monte, Wu, & Langston, 1992) and OCT3 in astrocytes primarily mediates the release of MPP 1 in mouse MPTP model (Cui et al., 2009), whereas OCT3 in microglia uptakes MPP 1 . Another possibility is that OCT3 in astrocytes and microglia has different roles in affecting the downstream events of MPP 1 . Clarification of the mechanism by which OCT3 exhibits different roles in transportation of MPP 1 by different cells will further expand our understanding on its physiological roles.
Previous studies have uncovered the regulation of miRNAs by MPP 1 in DA neurons or their cell lines (Asci et al., 2013;Niu, Xu, Wang, Hou, & Xie, 2016), while little is known about miRNA regulation by MPP 1 in microglia. Thus, one of the important findings of this study is that miR-7116-5p in microglia is downregulated by MPP 1 . Although a series of miRNAs is found to inhibit TNF-a production in different cell types (Guan et al., 2015;Tili et al., 2007;Zhang et al., 2012), miR-7116-5p in microglia is first identified as a miRNA targeting TNF-a in response to MPP 1 . Recently, in a clinical study of expression of miR-NAs in PD patients, hsa-miR-542-3p was found significantly downregulated in the SN of 8 PD patients, compared with four healthy donors. Additionally, hsa-miR-542-3p might target human TNF-a as predicted by several bioinformatic tools, such as miRandola and Tar-getScan (Cardo et al., 2014). In this context, it is possible that miRNA regulation of TNF-a production affects DA neuron survival in PD patients and in other animal models. It remains unclear how miR-7116-5p in microglia is downregulated. Reported studies have proposed that transcription-dependent or -independent mechanisms are involved in the regulation of miRNA levels (Winter, Jung, Keller, Gregory, & Diederichs, 2009;Zhang et al., 2015). Clarifying the mechanism underlying the downregulation of miR-7116-5p by MPP 1 in microglia will help our understanding of the miRNA regulation of TNF-a production in PD patients.
An important implication of the current findings is that a key inflammatory factor regulates consequent inflammatory responses contributing to the pathogenesis of PD. Here, the change in TNF-a expression well precedes the change in IL-1b, IL-6, iNOS, and COX-2 expression. Inhibition of TNF-a expression could almost abolish the consequent inflammatory responses and reduce neuronal damage.
Therefore, it suggests that inhibiting inflammation would be unsuccessful to protect DA neurons when exacerbated inflammation and irreversible neuron damage have occurred. Therefore, targeting the key inflammatory factor in early stage of pathogenesis of PD would probably be more effective. Indeed, the current preliminary evidence of clinical trials of neurodegenerative diseases such as Alzheimer's disease (AD) indicates that there is no beneficial effect to target inflammation in AD patients who have been with significant symptoms (McGeer & McGeer, 2007;Reines et al., 2004;Scharf, Mander, Ugoni, Vajda, & Christophidis, 1999). Therefore, new and early strategies should be proposed to prevent neuron damage resulted from the inflammatory injury.
Collectively, our results point to a previously unknown role of microglia in uptaking MPP 1 and existence of an MPP 1 -miR-7116-5p pathway in microglia that specifically regulates TNF-a production to affect DA neuron survival. The demonstration of miR-7116-5pdependent TNF-a production in microglia further expands our understanding of the diverse functions of microglia.

ACKNOWLEDGMENT
We thank J. Zhou for helpful discussion and Q. Hu for imaging. This work was supported by the National Natural Science Foundation of China (81130081).