α‐synuclein suppresses microglial autophagy and promotes neurodegeneration in a mouse model of Parkinson’s disease

Abstract The cell‐to‐cell transfer of α‐synuclein (α‐Syn) greatly contributes to Parkinson's disease (PD) pathogenesis and underlies the spread of α‐Syn pathology. During this process, extracellular α‐Syn can activate microglia and neuroinflammation, which plays an important role in PD. However, the effect of extracellular α‐Syn on microglia autophagy is poorly understood. In the present study, we reported that extracellular α‐Syn inhibited the autophagy initiation, as indicated by LC3‐II reduction and p62 protein elevation in BV2 and cultured primary microglia. The in vitro findings were verified in microglia‐enriched population isolated from α‐Syn‐overexpressing mice induced by adeno‐associated virus (AAV2/9)‐encoded wildtype human α‐Syn injection into the substantia nigra (SN). Mechanistically, α‐Syn led to microglial autophagic impairment through activating toll‐like receptor 4 (Tlr4) and its downstream p38 and Akt‐mTOR signaling because Tlr4 knockout and inhibition of p38, Akt as well as mTOR prevented α‐Syn‐induced autophagy inhibition. Moreover, inhibition of Akt reversed the mTOR activation but failed to affect p38 phosphorylation triggered by α‐Syn. Functionally, the in vivo evidence showed that lysozyme 2 Cre (Lyz2 cre)‐mediated depletion of autophagy‐related gene 5 (Atg5) in microglia aggravated the neuroinflammation and dopaminergic neuron losses in the SN and exacerbated the locomotor deficit in α‐Syn‐overexpressing mice. Taken together, the results suggest that extracellular α‐Syn, via Tlr4‐dependent p38 and Akt‐mTOR signaling cascades, disrupts microglial autophagy activity which synergistically contributes to neuroinflammation and PD development.

During the past two decades, great progress has been made in understanding the role of neuronal autophagy in PD. By contrast, the study of microglial autophagy in PD is still at infancy. Indeed, autophagy is critical for homeostatic regulation in the immune system. Dysregulation in microglial autophagy is emerging as a core regulator in brain development and diseases (Kim et al., 2017;Larabi et al., 2020;Lucin et al., 2013). Previously, we and other groups demonstrated the microglial autophagy alterations in response to lipopolysaccharide (LPS) and tumor necrosis factorα (TNFα) (He et al., 2018;Jin et al., 2018;Lee et al., 2019), and the in vitro work revealed a role of autophagy in microglia polarization and inflammation. Yet, little is known about the impact of extracellular α-Syn on microglial autophagy. The in vivo evidence whether microglial autophagy affects neuroinflammation and DA neurodegeneration needs to be strengthened.
In this study, we reported an impairment of microglial autophagy caused by extracellular α-Syn via toll-like receptor 4 (Tlr4) and downstream p38 and Akt-mTOR signaling pathways and provided the evidence that conditional knockout (cKO) of microglial autophagyrelated gene 5 (Atg5) in mice via lysozyme 2 Cre (Lyz2 cre )-mediated Atg5 depletion, enhanced the neuroinflammation and DA neuron losses in the midbrain and exacerbated the locomotor deficits in a viral-based α-Syn overexpression mouse model.

| Extracellular human α-synuclein inhibited autophagy initiation in microglia
To study the effect of extracellular α-Syn on microglia autophagy, the protein levels of two classic autophagy markers LC3-II and p62 were examined. hα-Syn appeared to dose-dependently reduce LC3-II but enhance p62 protein level in BV2 cells. Treatment with 10 μg/ml hα-Syn reached significance (Figure 1a,b) and was applied in the following studies. A time-dependent change of LC3-II and p62 protein levels was observed (Figure 1c,d). Similarly, hα-Syn caused LC3-II reduction but p62 protein elevation in cultured microglia (Figure 1e,f).
Moreover, p62 mRNA level was also elevated, consistent with a recent study (Choi et al., 2020). However, hα-Syn still enhanced p62 protein level even when de novo protein synthesis was blocked by cycloheximide (CHX), supporting that post-translational increase in p62 protein level also occurred (Figure 1g,h). Next, we asked if neuronsecreted α-Syn had similar effects on microglia. DA cells that overexpress hα-Syn were reported to release α-Syn and stimulate microglial inflammation (Bae et al., 2012;George et al., 2019). As such, we transferred the conditioned media from wildtype hα-Syn-overexpressing PC12 cells (α-Syn-CM) to microglia culture and tested its effect on microglial autophagy. The conditioned media from normal PC12 cells (Con-CM) served as controls. Western blotting revealed timedependent LC3-II decreases and p62 protein increases in α-Syn-CMtreated BV2 and microglia cells (Figure 1i-l), which were abolished when α-Syn-CM was pre-incubated with hα-Syn antibody but not non-specific IgG (Figure 1m,n), implying neuron-secreted hα-Syn, in line with the recombinant protein, inhibited microglial autophagy.

| Alterations of autophagy-related proteins in the midbrain of hα-Syn-overexpressing mice
The in vitro observations were validated in hα-Syn-overexpressing mice in which the SN pars compacta (SNpc) were bilaterally injected with adeno-associated virus 2/9 (AAV2/9) carrying wildtype hα-Syn (AAV2/9-hα-Syn) driven by synapsin. The mice injected with the virus carrying enhanced green fluorescent protein (eGFP) served as paralleled controls. Immunostaining displayed a predominant colocalization of eGFP or anti-hα-Syn (LB509) immunoreactivity to the cells labeled with neuron-specific marker NeuN, but not astrocyte marker GFAP or microglia marker Iba1, implying neurons were mainly infected ( Figure   S1a, b). We also observed an obvious colocalization of eGFP and LB509 fluorescence to tyrosine hydroxylase (TH)-positive neurons in the SNpc. Western blot using the antibody specifically against hα-Syn (MJFR1 clone) detected hα-Syn expression in the SN at 2 weeks, which gradually increased at 4 and 8 weeks after AAV2/9-hα-Syn injection (Figure S1c). A significant loss of DA neurons was found in the SNpc at 8 weeks (Figure S1d, e), indicating a successful establishment of a PDlike mouse model. A lower LC3-II level, with a mild p62 protein increase, was detected in the SN from AAV2/9-hα-Syn-injected mice compared with AAV2/9-eGFP group (Figure 2a, b). Double immunostaining identified a dramatic increase in p62 immunoreactive signal in Iba1 + microglia, which displayed an amoeboid morphology with enlarged soma, in AAV2/9-hα-Syn-injected mice compared with AAV2/9-eGFP group (Figure 2c), although the majority of enhanced p62 signal seemed to derive from non-microglial cells ( Figure S2). To confirm this, we acutely isolated the adult microglia and examined the autophagic markers by western blot. A remarkable increase in p62 protein, along with LC3-II decrease, was found in the microglia population from AAV2/9-hα-Syn-injected mice compared with AAV2/9-eGFP controls (Figure 2d,e; Figure S3). These supported a down-regulated autophagy activity in the microglia of hα-Syn-overexpressing mice, in line with in vitro data.

| Tlr4-dependent p38 and Akt-mTOR activation was involved in autophagy inhibition by hα-Syn
We then examined the signaling for α-Syn-induced autophagy inhibition in microglia. First, we focused on Akt-mTOR signaling because of its critical role in autophagy regulation. A rapid increase in Akt phosphorylation at Ser473 was found in BV2 cells following α-Syn-CM treatment within 1 h (Figure 3a, b). Phosphorylations of mTOR (Ser2488) and its downstream p70S6 kinase (Thr389) also enhanced.
Second, we studied MAPK pathway due to its importance in α-Syninduced microglial inflammation (Klegeris et al., 2008). Remarkably, α-Syn-CM caused a rapid increase in p38 MAPK phosphorylation early at 15 min after treatment, while Erk1/2 or Jnk phosphorylation remained unaltered. Similarly, we observed time-dependent rises in the phosphorylation of Akt, mTOR, as well as p38 in hα-Syn-treated microglia (Figure 3c, d). Moreover, inhibition of Akt with wortmannin (Wort) prevented the mTOR activation and LC3-II decrease caused by hα-Syn. As a recent study reported that Forkhead Box O3 (FOXO3) was a downstream target of Akt for microglial autophagy inhibition by LPS (Lee et al., 2019), we examined the FOXO3 levels with western blot. However, phosphorylation of FOXO3 at Ser253 did not change up to 6 h after hα-Syn treatment ( Figure S4), implying a less likely involvement of FOXO3 in hα-Syn-suppressed autophagy here. Yet, mTOR inhibitor Rapa reversed hα-Syn-caused LC3-II reduction but did not affect Akt phosphorylation, implying that mTOR was downstream to Akt activation. In addition, p38 MAPK inhibitor SB202190 Several studies showed that extracellular α-Syn interacted with plasma membrane receptor Tlr4 or Tlr2 and triggered neuroinflammation in microglia (Choi et al., 2020;Fellner et al., 2013;C. Kim et al., 2013). Tlr4 may play a major role in this process because its expression is highest within human SN and upregulated in the brains of patients with PD (Drouin-Ouellet et al., 2014;Hughes et al., 2019). To study if hα-Syn-caused microglial autophagy inhibition was mediated by Tlr4, microglia from wildtype and Tlr4 −/− pups were cultured and stimulated with hα-Syn. However, hα-Syn did not enhance the phosphorylation of p38 or Akt, nor reduce the LC3-II level in Tlr4 −/− microglia compared with wildtype microglia (Figure 3g), indicating that hα-Syn-caused microglial autophagy inhibition was Tlr4 dependent.
Tlr4 deficiency was verified by reverse transcription PCR ( Figure 3h).

| Age-dependent changes of microglia phenotype in microglial autophagy-deficient mice
To understand the relevance of microglial autophagy impairment, we generated the microglial autophagy-deficient mice by crossing Lyz2 Cre mice with Atg5 f/f mice followed by genotyping ( Figure S5a With this strategy, the autophagy gene Atg5 was deleted in myeloid lineage-specific lysozyme M-expressing cells including microglia (Kim et al., 2017). Autophagy deficiency was verified by assessing Atg5, LC3-II, Atg7, and Beclin1 protein levels in the brain homogenates. Atg5 and LC3-II levels reduced in both heterogeneous and homogeneous Lyz2 Cre ; Atg5 f/f (Atg5 cKO ) mice compared with Atg5 f/f littermates, with more reductions in homogeneous Atg5 cKO mice, which were then used in the following study. Atg7 and Beclin1 levels were unaffected (Figure 4a We then examined whether autophagy deficiency may affect microglial morphology and activation. For this purpose, brains of both young and aged Atg5 f/f and Atg5 cKO mice were studied by anti-Iba1 immunostaining. Overall, a higher Iba1 intensity was found in different brain regions (SN and cortex) of Atg5 cKO mice compared with Atg5 f/f littermates (Figure 4d,e). Moreover, Iba1 intensities increased with ages, which was more obvious in aged Atg5 cKO mice. A significant interaction effect was found between genotype and age (p < 0.05, twoway ANOVA). We then focused on the SN, a PD-affected region, and found the microglia in Atg5 cKO mice, compared with those in control littermates, exhibited with enlarged soma, shortened processes, and lessened branch points. The morphological alterations also occurred with age, more obviously in aged Atg5 cKO mice (Figure 4f-h), consistent with the Iba1 intensity results. These implicated that Atg5 depletion may prompt microglia into a primed state and enhance their susceptibility to the second challenge, particularly with aging.

| Autophagy modulated microglia-mediated neuroinflammation
We then explored the effect of autophagy modulation on neuroinflammation induced by hα-Syn, which mimicking the pathogenesis of PD in vitro. The autophagy activator Rapa attenuated the increase in TNFα and IL-1β mRNA levels and decrease in CD206 mRNA level induced by hα-Syn-CM in BV2 cells. The ELISA data showed that Rapa also attenuated the rises of IL-6 and IL-1β in the culture supernatant ( Figure S6a-e). These results were verified in cultured microglia from F I G U R E 2 Alterations of autophagy-related proteins in the midbrain of hα-Syn-overexpressing mice. (a, b) LC3-II and p62 protein levels in the SN from C57BL/6 mice at 4 weeks following AAV2/9-hα-Syn or AAV2/9-eGFP injection. Student t test, n = 5. (c) Midbrain sections from AAV2/9-hα-Syn-or AAV2/9-eGFP-injected mice were double stained with anti-Iba1 and anti-p62. White arrows indicate typical colocalization of Iba1 and p62 signal under confocal scanning. Scale bar, 20 μm. (d, e) p62 and LC3-II protein levels in the microglia isolated from AAV2/9hα-Syn-or AAV2/9-eGFP-injected mice at 4 weeks. Student t test, n = 3. *p < 0.05, **p < 0.01 | 7 of 17 TU eT al.
The in vivo consequence of α-Syn-induced microglial autophagy inhibition was investigated further. Immunostaining revealed an elevation of Iba1 intensity, with remarkable morphological changes in these Iba1 + cells, in the SN reticular (SNr) of both genotypes after AAV2/9-hα-Syn injection. Specifically, microglia displayed an amoeboid morphology, with enlarged soma and thick processes, which were different from the ramified microglia with smaller soma and numerous longer branches in control (AAV2/9-eGFP) group (Figure 5f-j). These changes were more obvious in AAV2/9-hα-Syn-injected Atg5 cKO mice compared with Atg5 f/f controls. Moreover, TNFα and IL-1β mRNA levels were elevated, along with a lowered CD206 mRNA level, in the SN of Atg5 cKO mice compared with Atg5 f/f mice following AAV2/9hα-Syn injection. A mild change in inflammation-associated molecules (TNFα, IL-1β, and CD206) mRNA level was found in the SN from AAV2/9-eGFP-injected Atg5 cKO and Atg5 f/f mice, probably due to the majority of non-microglia cells in the tissue (Figure 5k-m). However, an approximate sevenfold increase in TNFα and IL-1β mRNA levels was found in the microglia population acutely isolated from Atg5 cKO mice (2 months old) relative to Atg5 f/f controls ( Figure 5n). Altogether, the data implicated hα-Syn may cause inflammation in Atg5 cKO mice, independent of autophagy inhibition. Autophagy-dependent and autophagy-independent machinery may synergistically contribute to hα-Syn-caused neuroinflammation in PD.

| Microglia autophagy deficiency exacerbated DA neurodegeneration and locomotor deficits in hα-Syn-overexpressing mice
Next, we tested whether microglia Atg5 deficiency had any effect on DA neuron losses induced by hα-Syn overexpression.
Immunostaining with anti-TH (a DA neuron marker) showed great losses of DA neurons and terminals in the SNpc and striatum in both Atg5 f/f and Atg5 cKO mice at 8 weeks after AAV2/9-hα-Syn injection (Figure 6a, b), with more significant losses in Atg5 cKO mice.
These were verified by western blotting which showed a lower TH protein level in the SN of Atg5 cKO mice compared with Atg5 f/f littermates that received AAV2/9-hα-Syn injection (Figure 6c). We also assessed the density of DA transporter (DAT), which is located on striatal DA terminals and expressed independently of TH expression or activity. AAV2/9-hα-Syn injection caused a dramatic reduction of striatal DAT density, regardless of its genotype, implicating DA neurodegeneration in the nigrostriatal pathway. Compared with that in AAV2/9-hα-Syn-injected Atg5 f/f controls, the DAT density in AAV2/9-hα-Syn-injected Atg5 cKO mice was further lowered, which was almost undetectable (Figure 6d). Atg5 cKO mice did not exhibit obvious DA neurodegeneration, only with a subtle change of TH and DAT density compared with Atg5 f/f mice following AAV2/9-eGFP injection. These data suggested that microglial Atg5 cKO exacerbated the vulnerability of DA neurons to hα-Syn overexpression.
α-Syn phosphorylation at serine 129 (S129) is associated with α-Syn pathology and reported to be elevated in PD patients (Samuel et al., 2016). Therefore, we examined the phosphorylated level of human α-Syn (pS129α-Syn). A higher pS129α-Syn intensity was found in the SN of Atg5 cKO mice compared with Atg5 f/f littermates after AAV2/9-hα-Syn injection, while total α-Syn level was comparable between two groups (Figure 6e). Double immunostaining found abundant hα-Syn pathology in the remaining SNpc TH + neurons after AAV2/9-hα-Syn injection. High magnification image showed a colocalization of pS129α-Syn to TH + neurons and displayed a reduced TH immunoreactivity in these co-localized neurons (pS129α-Syn + TH + ) compared with unaffected DA neurons (pS129α-Syn − TH + ).
The result was confirmed by western blot, revealing a higher ratio of pS129α-Syn over total α-Syn in AAV2/9-hα-Syn-injected Atg5 cKO mice relative to Atg5 f/f controls (Figure 6f). We did not detect any band of human pS129α-Syn or α-Syn in AAV2/9-eGFP-injected mice, implying the specificity of the antibodies used.
Finally, we evaluated the animal behaviors at 8 weeks after AAV2/9-hα-Syn or AAV2/9-eGFP injection. Atg5 cKO mice did not show any sign of locomotor or motor coordination impairment compared with Atg5 f/f mice even after AAV2/9-eGFP injection Extracellular α-Syn activates microglia and neuroinflammatory responses, which contribute to PD pathogenesis. Our in vitro study showed that neuron-released α-Syn suppressed microglial autophagy initiation. This was verified by the in vivo data showing the accumulation of p62 protein, concomitant with LC3-II reduction, in microglia isolated from hα-Syn-overexpressing mice. This provided the evidence that extracellular α-Syn suppressed microglial autophagy activity. Additionally, we found an increased p62 mRNA level in α-Syn-treated microglia. This was consistent with Choi et al. that hα-Syn induced p62 transcription, which may help to sequester the internalized α-Syn for degradation via a process termed synucleinphagy (Choi et al., 2020). But, α-Syn was still able to enhance p62 protein level when de novo protein synthesis was blocked, supporting that non-transcriptional increase in p62 protein, probably due to autophagy impairment, also existed. In other words, α-Syn can cause microglial p62 protein increases through both transcriptional upregulation and autophagy inhibition.
Seemingly, our data are inconsistent with the study by Choi et al. that 250 nM hα-Syn treatment up to 9 h had little effect on total LC3-II levels. But we consistently demonstrated that microglial autophagy impairment further promoted neurodegeneration in a synucleinopathy model. In fact, 10 μg/ml (approximately equal to 500 nM) hα-Syn was applied in this study. We detected slight but insignificant LC3-II change when microglial cells were exposed to hα-Syn <10 μg/ ml for shorter periods. We speculate that α-Syn affects microglial autophagy in a concentration-dependent manner. At lower levels, α-Syn induces transcriptional upregulation of p62 through NF-κB signaling and promotes p62-dependent α-Syn degradation via selective autophagy; however, it also suppresses microglial autophagy and synergistically aggravates neuroinflammation at higher concentrations when α-Syn pathology continues. Admittedly, the α-Syn concentration applied in this study was relatively high compared with its reported level in the cerebrospinal and interstitial fluids (Stuendl et al., 2016). But, α-Syn level may be spatiotemporally different and affected by several factors such as disease progression, neuronal activity, and circadian rhythm (Elfarrash et al., 2019;Wu et al., 2020).
Our in vivo and in vitro studies consistently revealed the microglial autophagy inhibition by α-Syn. Therefore, the α-Syn concentration used here may still be of pathological relevance. The impact of mutant α-Syn on microglia autophagy deserves to be studied further.
Early in 2006, two pioneer studies in nature demonstrated an essential role of basal autophagy in neuronal survival (Hara et al., 2006;Komatsu et al., 2006). Conditional deletion of Atg7 in the SNpc DA neurons recapitulated several pathologic features of PD (Ahmed et al., 2012). The role of glial autophagy gained attention in the field of PD recently. Our study showed that conditional deletion of microglial Atg5 by crossing Atg5 f/f with Lyz2 Cre lines did not induce obvious neurodegeneration or motor coordination defects in mice up to 5 months old, but promoted the neurodegeneration in different PD mouse models, consistent with previous studies (Choi et al., 2020;H. J. Kim et al., 2017;Qin et al., 2021). However, there is a discrepancy with a recent report that Itgam/CD11b-Cre-mediated microglial Atg5 depletion caused early PD-like neurodegeneration, which displayed motor coordination defects at 3 months old and cognitive decline at 8-month-old mice (Cheng et al., 2020). Despite this discrepancy, the microglial activation and enhanced inflammatory levels were still consistent among these studies. The discrepancy may be attributable to the expression and specificity of various Cre recombinases in the myeloid lineage, deserving to be clarified further. Given that Lyz2 is extensively expressed in myeloid cells, the impact of autophagy defects in other myeloid cells from Lyz2 Cre ; Atg5 f/f mice cannot be completely ignored. The newly generated mouse line such as Cx3cr1 CreER which express Cre in a tamoxifeninducible manner may serve as a better tool to corroborate the role of microglial autophagy.
Increased IL-1β and TNFα mRNA levels were identified in microglia acutely isolated from adult Atg5 cKO mice compared with F I G U R E 6 Microglia autophagy deficiency exacerbated DA neurodegeneration and locomotor deficits in hα-Syn-overexpressing mice. (a, b) Coronal SNpc (a) and striatum (b) sections at 8 weeks post injection and quantification for TH + neuron number (n = 6 brains) and terminal density (n = 4 brains), respectively. (c) TH protein level in the SN. Two-way ANOVA with Tukey's post hoc, n = 5. (d) Striatal sections and group data of DAT density (n = 4 brains). Two-way ANOVA with Tukey's post hoc. (e, f) Total and phosphorylated (S129) hα-Syn in the SN of AAV2/9-hα-Syn-or AAV2/9-eGFP-injected Atg5 f/f and Atg5 cKO mice at 8 weeks, assessed by immunostaining (e) and western blot (f). Scale bar, 200 μm and 10 μm. Student t test, n = 4. (g-i) Open field test for AAV2/9-hα-Syn (n = 8 for each genotype) or AAV2/9-eGFP (n = 10 for each genotype) -injected mice. Mean speed of locomotion (g) and entries into the central zone (h) were quantified. (i) Representative trajectory pmap. Two-way ANOVA with Tukey's post hoc, *p < 0.05, **p < 0.01, and ***p < 0.001 Atg5 f/f littermates, supporting the role of Atg5-mediated autophagy in regulating neuroinflammation. This was in line with the observations that microglia activation was more obvious in Atg5 cKO mice compared with Atg5 f/f under basal conditions, regardless of its age (Figure 4). Unexpectedly, we did not detect any change in TNFα mRNA or protein level in cultured Atg5 −/− microglia.
Probably, other compensatory mechanism that regulates TNFα production occurred during in vitro culture of microglia. But, the increase in IL-1β generation was consistently found. Besides the increased transcription, we detected a small amount of IL-1β protein in the culture supernatant of Atg5 cKO -derived microglia, while that from Atg5 f/f microglia was undetectable with ELISA using the same procedure. This implicated a critical role of autophagy in controlling inflammasome activation and IL-1β generation, consistent with previous reports (Deretic & Levine, 2018;Qin et al., 2021).
These data support the perception that basal autophagy balances inflammation in microglia (Deretic & Levine, 2018), and autophagy defects prime microglia to be susceptible to the second challenge.
Autophagy is a homeostatic process with multiple effects on microglia. Besides inflammation, autophagy affects other processes including phagocytosis (Lucin et al., 2013;Nash et al., 2017), exosome secretion , and metabolic homeostasis (Xu et al., 2021). Microglial autophagy also plays a role in brain development. Deficient autophagy in microglia impaired synaptic pruning and caused social behavioral defects (Kim et al., 2017). The impact of autophagy on microglial phagocytosis was not explored here. However, we found more microglia activation, concomitant with morphological changes, in the brains of Atg5 cKO mice compared with Atg5 f/f controls under normal conditions.
Whether this reflected an alteration in microglial phagocytosis needs to be investigated further.
Interestingly, we found significant changes in microglia morphology in the brains of aged mice, and these were more obvious in Atg5 cKO compared with Atg5 f/f mice, implicating Atg5 depletion produced an additional effect on aging-related microglia morphology alterations. This is of great importance for age-related neurodegenerative disorders. Microglia display diminished capacity for migration and phagocytosis and shift into a pro-inflammatory state with ages. Although microglia-specific changes in the expression of autophagy genes during aging still await to be clarified in future by single-cell RNA sequencing and other novel technologies, mounting evidence suggests a critical role for autophagy in regulating age-related changes in microglia. For example, Atg7 depletion caused lipid droplets accumulation in microglia, which represented a dysfunctional and pro-inflammatory state in the aging brain (Marschallinger et al., 2020;Xu et al., 2021). Moreover, defects in mitophagy, a selective autophagy for dysfunctional mitochondria, resulted in metabolic imbalance and inflammation in microglia (Fang et al., 2019;Rose et al., 2017). Therefore, α-Syn-caused autophagy inhibition in microglia may affect their susceptibility to aging, accelerating the progression of PD.
Autophagy and innate immunity are closely related. Autophagy modulates immune responses by regulating macrophage/microglia polarization and cytokines production. Conversely, inflammation differentially affects autophagy in macrophage and microglia, although they show a lot of similarities in origin and immune defense.
For instance, the classic Tlr4 agonist LPS enhanced autophagy activity in macrophage but reduced it in microglia (He et al., 2018;Xu et al., 2007), with unknown reason for this discrepancy. β amyloid and α-Syn, as we and others reported, disrupted the microglial autophagy processes (Cho et al., 2014;Lee et al., 2019). Different mechanisms may be involved. He et al. reported a critical role of p38 MAPK-mediated phosphorylation of Unc-51-like autophagy activating kinase 1 (ULK1) in LPS-inhibited microglial autophagy (He et al., 2018). Another study showed that Tlr4 activation by LPS suppressed microglial autophagy through PI3K-FOXO3 pathway, independent of mTOR or MAPK1/3 (Lee et al., 2019). In this study, we found the involvement of Tlr4-dependent p38 activation and Akt-mTOR signaling, but not FOXO3 inhibition, in α-Syn-suppressed microglia autophagy. Surely, we cannot exclude the possibility that the cytokines released from α-Syn-activated microglia may also indirectly interfere the autophagic processes at later stages since TNFα and other cytokines were reported to inhibit microglial autophagy flux (Jin et al., 2018). α-Syn was well demonstrated to interact with Tlr4 or Tlr2 in glia, which may be concentration-and conformation-sensitive (Kim et al., 2013).
Recent studies identified the highest expression of Tlr4 in the human SN and its upregulation in the brains of patients with PD and revealed a sensitivity of glial Tlr4 signaling to prolonged exposure to low levels of α-Syn oligomers (Drouin-Ouellet et al., 2014;Hughes et al., 2019). These highlighted the relevance of Tlr4 in α-Syn-trigged responses in microglia, in line with our data.
In sum, our findings demonstrate that α-Syn disrupts microglial autophagy initiation via Tlr4-dependent p38 and Akt-mTOR signaling and reveal that microglial autophagy impairment contributes to neuroinflammation and other PD pathogenesis. Therefore, the pharmacologic and genetic strategies that aim to modulate autophagy activity in the brain may become a potential venue for PD therapy.
The inhibitor for p38 (SB202190) and PI3K (Wort) was purchased from Selleck, BafA1 from Abcam, and CHX from Med Chem Express.

| Cell culture
Primary microglia were prepared from P1-P3 pups as we previously reported (Yuan et al., 2018). Briefly, cortex was isolated and dissociated in 0.05% trypsin followed by filtering through a 70μm strainer.
After culture for 1-2 weeks, microglia were detached off the flasks by orbital shaking at 180 rpm for 2 h and seeded into dishes for experimentation. BV2 cells were maintained in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin mixture in a 37°C incubator. The reagents for cell culture were obtained from Gibco.

| Preparation of α-Syn-enriched conditioned media
To prepare α-Syn-enriched conditioned media (α-Syn-CM), PC12 cells were infected with pLentiVENUS-YFP-hα-Syn as previously reported (Wang et al., 2015). At 24 h post infection, media were replaced with fresh DMEM and incubated for 48 h. The cell-free culture media were harvested and centrifuged at 10,000 g using 10 kDa cut-off filters (Millipore). The resulting supernatants were referred to as α-Syn-CM for microglia treatment further. To verify if CM-contained α-Syn affected microglial autophagy, CM was pre-incubated with 2 μg/ml antiα-Syn antibody (#2642, CST) or non-specific IgG for 10 min before being transferred to microglia culture.

| Western blotting
Cells were lysed in RIPA buffer (25 mM Tris, 150 mM NaCl, 5 mM EDTA, 1% NP-40, pH 7.5) with a protease/phosphatase inhibitor tablet. Brain tissues were homogenized in the lysis buffer at 1:10 (w/v) and centrifuged at 13,000 rpm for 30 min. Western blot analysis was performed as described previously (Yuan et al., 2018). For α-Syn phosphorylation analysis, the membranes were fixed in 4% paraformaldehyde (PFA) before subjected to blocking and incubation with anti-p-S129 α-Syn.

| GFP-LC3 plasmid transfection and punctate visualization
To monitor microglial autophagy activity, BV2 cells were transiently transfected with eGFP-LC3 plasmid (Addgene) using lipofectamine 2000. At 48 h post transfection, cells were subjected to indicated treatments for 6 h. Cells were fixed in 4% PFA for 10 min, stained with DAPI, and observed under a confocal microscope (LSM700; Carl Zeiss).

| Animals and stereotaxic surgery
To induce α-Syn overexpression in vivo, the SN of 8-week-old male C57BL/6 mice (SLRC Laboratory) were bilaterally injected with AAV2/9-hα-Syn (1.28E+10 v.g) driven by the promoter synapsin (Obio). For control group, mice were injected with AAV2/9-eGFP. For surgery, mice were anaesthetized with 2% isoflurane and injected with 1.5 μl of AAV2/9-hα-Syn or AAV2/9-eGFP at the rate of 0.3 μl/ min, with the coordinates at AP −3.0 mm, ML ±1.25 mm, and DV +4.5 mm relative to the bregma ( Figure S1a). The needle was left in place for 5 min before being retracted.  (Hara et al., 2006). The breeding strategy and genotyping results are shown in Figure S2. Tlr4 KO mice were obtained from Nanjing University MARC. Animals were bred at the Laboratory Animal Facility in Soochow University under specific pathogen-free conditions and maintained at a constant ambient temperature, with access to tap water and food ad libitum. All animal experiments were carried out in accordance with the guidelines of the Institutional Animal Care and Use Committee of Soochow University.

| Adult microglia isolation
Adult microglia were isolated according to a reported protocol (Galatro et al., 2017) with minor modifications. Briefly, whole brains were isolated from mice after intracardial perfusion with chilled PBS and homogenized in 5 ml ice-cold serum-free DMEM. Homogenates were filtered through a 70µm cell strainer and centrifuged with 1000 ×g for 10 min at 4°C, followed by suspension with 22% Percoll and centrifuge again to remove the myelin fragments. After that, the pellet was resuspended in 4 ml 37% Percoll and slowly covered on the top of 4 ml 70% Percoll and centrifuged with 1000 x g for 30 min with no brake. The 70-37% interphase was carefully collected and resuspended with 3 times volume of Hank's balanced salt solution (HBSS) followed by centrifuge at 800 ×g for 10 min at 4°C. The pellet was collected as microglia-enriched population, and the purity was determined via flow cytometry ( Figure S3).

| Quantitative polymerase chain reaction (qPCR)
Total RNAs were reverse transcribed using the cDNA synthesis kit (Thermo Fisher Scientific). Target genes were amplified using SYBR Green PCR Master Mix (Q711, Vazyme) with specific primers in Table 1. Amplification of DNA products was performed using the 7500 Real-Time PCR system (Applied Biosystems). Relative mRNA levels were calculated according to the 2 −△△Ct method. All delta Ct values were normalized to GAPDH.

| Cytokine measurement
The contents of TNFα and IL-1β were determined by the enzymelinked immunosorbent assay (ELISA) kits from eBioscience, while IL-6 were quantified using the kit from R&D system according to the manufacturer's instructions.
The number of TH + neurons was counted using the "cell counter" plug-in in Fiji software and corrected by a researcher blinded to testing groups. Every fifth section throughout the SN was counted, and at least four sections per brain were included for quantification. For striatal TH or DAT intensity, at least four sections per brain were included for analysis using the ImageJ software. The microglial morphological analyses were performed using Fiji plug-in in AnalyzeSkeleton (2D/3D) software according to a previously published protocol (Young & Morrison, 2018).

| Behavior tests
The mice were subjected to several behavior tests at 8 weeks after stereotaxic injection, by the researcher blinded to treatment groups.

| Statistical analysis
Data are presented as mean ± SEM of at least three independent experiments unless indicated. The statistical analysis between two groups was determined using Student unpaired t test. For

ACK N OWLED G M ENTS
We thank Prof. Honghong Yao (Southeast University) for the technical support for adult microglia isolation. This work was supported by grants from National Natural Science Foundation of China

CO N FLI C T S O F I NTE R E S T
We have no potential conflict of interest to be disclosed.

AUTH O R CO NTR I B UTI O N S
LFH, HYT, BSY, and XOH conceived the study and designed the experiments. HYT, BSY, and XOH involved in acquisition of data. XJZ involved in adult microglia isolation. CSP and YTM involved in primary microglia culture. ZHQ, YF, and CFL contributed to the revision of the manuscript for important intellectual content. LFH, HYT, BSY, and XOH assembled the figures, and LFH wrote the manuscript with input from all authors. LFH and YPY involved in obtained funding.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.