Primary microglial cells were isolated from postnatal mouse pups (P0-3) following our published protocol (29). Following isolation of microglia from mixed glial culture, cells were treated with 1 μM αSyn_Agg for 24 h.

Mouse microglial cell (MMC) line was a kind gift from Dr. Golenbock from University of Massachusetts (30). The MMC line was chosen over other microglial cell lines for proteomic studies due to its similarity with primary microglia as shown previously by our group (31). We have previously shown that MMC at basal level is in a relatively resting/quiescent state which becomes activated by LPS or αSyn_Agg treatment to an M1-like state, similar to that seen in primary microglia. Due to the greater resemblance to primary microglia, the MMC line appears to be a better choice than the more commonly used immortalized microglia cell line, BV2 (32). MMC were grown in 10% fetal bovine serum, DMEM, 1% penicillin/streptomycin and 1% glutamate. Treatments were performed in 2% FBS-containing media. Cells were treated with 1 μM αSyn_Agg for 24 h (32).

Recombinant αSyn was prepared following a previously published protocol (19, 33). Briefly, transformation with plasmid encoding human αSyn was performed in E. coli cells (BL21(DE3) strain) cells. Recombinant αSyn expression was induced by using isopropyl β-D-1-thiogalactopyranoside (IPTG) (Invitrogen). Cells were lysed and recombinant αSyn was purified as previously described (34, 35). We used FPLC system from Biorad to purify the protein and the FPLC chromatogram showed one peak suggesting the purity of the protein (Supplemental Figure 1A). Further, we performed Krypton stain (Supplemental Figure 1B) to determine the purity of the protein. For αSyn aggregation, recombinant protein solution was shaken at a speed of 1000 rpm at 37°C for 7 days (36). The level of endotoxin in αSyn preparations was quantified and <5 EU was detected. Moreover, we confirmed the conformation of the aggregates by electron microscopy (28).

All animals were housed under standard conditions of constant temperature (22 ± 1°C), humidity (relative, 30%), and a 12-h light/dark cycle. Use of the animals and protocol procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at Iowa State University (ISU), Ames, IA, USA. αSyn_Agg pre-formed fibrils (αSyn_PFF) were in injected in C57/BL mice bred in our animal facility. Mice were anesthetized as previously described and then injected with 5 of μg αSyn_PFF or vehicle. The coordinates indicating distance (mm) from bregma were: AP 0.5, ML 1.9, and DV 4 (28).

Samples were prepared essentially as described with slight modifications (37). MMCs were grown to 75% confluence, exposed to αSyn_Agg (1 μM) for 24 h, and then harvested. Each cell pellet was individually homogenized in 300 μL of urea lysis buffer (8 M urea, 100 mM NaHPO₄, pH 8.5), including 3 μL (100 × stock) HALT protease and phosphatase inhibitor cocktail (Pierce) (20, 37). After lysis for 30 min at 4°C, protein supernatants were transferred to 1.5-mL Eppendorf tubes and sonicated (Sonic Dismembrator, Fisher Scientific) three times for 5 s with 15 s intervals of rest at 30% amplitude to disrupt nucleic acids and subsequently vortexed. Protein concentration was determined by the bicinchoninic acid (BCA) method, and samples were frozen in aliquots at −80°C. Protein homogenates (100 μg) were diluted with 50 mM NH₄HCO₃ to a final concentration of <2 M urea and then treated with 1 mM dithiothreitol (DTT) at 25°C for 30 min, followed by 5 mM iodoacetimide (IAA) at 25°C for 30 min in the dark. Protein was digested with 1:100 (w/w) lysyl endopeptidase (Wako) at 25°C for 2 h and further digested overnight with 1:50 (w/w) trypsin (Promega) at 25°C. Resulting peptides were desalted with a Sep-Pak C18 column (Waters) and dried under vacuum. For LC-MS/MS analysis, derived peptides were re-suspended in 100 μL of loading buffer (0.1% formic acid, 0.03% trifluoroacetic acid, 1% acetonitrile). Peptide mixtures (2 μL) were separated on a self-packed C18 (1.9 μm, Dr. Maisch, Germany) fused silica column (25 cm × 75 μM internal diameter (ID); New Objective, Woburn, MA) by a Dionex Ultimate 3000 RSLCNano and monitored on a Fusion mass spectrometer (Thermo-Fisher Scientific, San Jose, CA). Elution was performed over a 2 h gradient at a rate of 400 nL/min with buffer B ranging from 3 to 80% (buffer A: 0.1% formic acid in water, buffer B: 0.1% formic acid in acetonitrile). The mass spectrometer cycle was programmed to collect at the top speed for 3-s cycles. The MS scans (400–1,600 m/z range; 200,000 AGC; 50 ms maximum ion time) were collected at a resolution of 120,000 at 200 m/z in profile mode, and the HCD MS/MS spectra (0.7 m/z isolation width; 30% collision energy; 10,000 AGC target; 35 ms maximum ion time) were detected in the ion trap. Dynamic exclusion was set to exclude previously sequenced precursor ions for 20 s within a 10 ppm window. Precursor ions with +1 and +8 or higher charge states were excluded from sequencing.

Raw data files were analyzed using MaxQuant v1.6.3.4 with Thermo Foundation for RAW file reading capability, as previously published (20). The search engine Andromeda was used to build and search a concatenated target-decoy IPI/Uniprot mouse reference (downloaded Aug 14, 2015, with human alpha synuclein sequence added, Uniprot ID P37840). Protein methionine oxidation (+15.9949 Da) and protein N-terminal acetylation (+42.0106 Da) were variable modifications (up to five allowed per peptide); cysteine was assigned a fixed carbamidomethyl modification (+57.0215 Da). Only fully tryptic peptides were considered with up to two miscleavages in the database search. A precursor mass tolerance of ±20 ppm was applied prior to mass accuracy calibration and ±4.5 ppm after internal MaxQuant calibration. Other search settings included a maximum peptide mass of 6000 Da, a minimum peptide length of six residues, and 0.05 Da Tolerance for orbitrap (FTMS) HCD MS/MS scans. Co-fragmented peptide search was enabled to deconvolute multiplex spectra. The false discovery rate (FDR) for peptide spectral matches, proteins, and site decoy fraction were all set to 1%. Quantification settings were as follows: re-quantify with a second peak finding attempt after protein identification has completed; match full MS1 peaks between runs; a 0.7-min retention time match window was used after an alignment function was found with a 20 min RT search space. The label-free quantitation (LFQ) algorithm in MaxQuant (21, 22) was used for protein quantitation. Data are available via ProteomeXchange with identifier PXD013691.

RNA isolation from primary microglial cells was performed as described previously (38, 39). Total RNA concentration was measured, and 1 μg RNA was converted to cDNA using the Affinity Script qPCR cDNA synthesis system (Agilent Technologies). Real-time PCR was performed with the RT2 SYBR Green master mix (Thermo-Fisher #K0172). The housekeeping gene 18s rRNA (Qiagen #PPM57735E) was used as the reference for all qRT-PCR experiments. The ΔΔCt method was used, implementing the threshold cycle (Ct) value for the housekeeping gene and for the respective gene of interest in each sample (18, 39). The primers were generated using primer bank (40). The primers were synthesized at Iowa State DNA facility (see Table 1 for primer list).

To determine if any protein products of PD GWAS targets were enriched in a particular module, we used the single nucleotide polymorphism (SNP) summary statistics from http://www.pdgene.org/ (15) to calculate the gene level association value using MAGMA (15). MAGMA calculates the gene level association value by taking the mean of all the transformed (Z statistic) SNP P-values associated with a particular gene and uses a known approximation of the distribution to get the gene association value. MAGMA accounts for linkage disequilibrium (LD) using reference data with similar ancestry. These gene lists were further filtered to select for genes that have a MAGMA defined gene association value > 1.3 (-logP-value). For each module in the protein network, the mean GWAS significance value (-log P) was calculated as the enrichment score for the module. Random sampling (10,000 times) of the MAGMA gene list was used to assess the significance of the module enrichment score. The enrichment scores were then scaled by subtracting the mean and dividing by the standard deviation of the random samplings. The P-value was calculated as the proportion of samplings that have a scaled enrichment score greater than or equal to the module enrichment score. The psychiatric genomics consortium provides links to various data sets.

Seahorse metabolic stress test was performed as described previously using a Xfe24 Seahorse (20, 41). Briefly, primary microglial cells (100,000 per well) were plated in PDL-plated Seahorse 24-well plate. Cells were treated with αSyn_Agg (1 μM) for 24 h. For MitoStress test, 0.75 μM oligomycin, 0.75 μM FCCP and 0.5 μM rotenone/antimycin were used. Wave 2.6.0 was used to analyze the data.

Immunocytochemical analysis was performed per previously published protocols (42). Briefly, primary microglia were isolated and plated on poly-D-lysine coated coverslips and treated. Following treatment, cells were fixed with 4% paraformaldehyde, and blocked with 2% BSA, 0.5% TritonX and 0.05% Tween. The cells were then incubated with primary antibody overnight, washed with PBS, incubated in secondary antibody. The following primary antibodies were used: STAT3 (Cell Signaling Technologies) and pSTAT3 (Y705) (Cell Signaling Technologies).

Differential expression analyses of proteomic data were performed using pairwise t-test applied to log2-transformed expression data and adjusted for multiple comparisons using the Benjamini-Hochberg method. For comparisons across more than 2 groups, one-way ANOVA with Tukey post analysis was used. For 2 groups students t-test were performed using graphpad prism 5.0. Volcano plots were plotted with the ggplot2 package in R. Proteins with missing data were filtered for minimum criteria as described in results, and missing LFQ abundances were imputed according to an in-house implementation of the Perseus algorithm (43, 44) in R. Gene Ontology (GO) enrichment analyses were performed using GO-Elite software as previously described using input lists of differentially expressed proteins that were either increased or decreased following αSyn_Agg exposure. Pathway analyses were also performed (Metacore, Thompson Reuters) as previously described (37).

We first performed qRT-PCR experiments showing that αSyn_Agg treatment induced the expression of M1-like pro-inflammatory markers by mouse microglia cell (MMC) line including Nos2, IL-6, TNF, and IL-1β without affecting or decreasing M2 markers such as IRF-4, IGF-1 and MRC1 (Supplemental Figure 2) suggesting that αSyn_Agg induces a pro-inflammatory M1-like state in-vitro. Moreover, the increased expression of M1-like pro-inflammatory genes were seen only αSyn_Agg and not with αSyn monomers. Treatment with monomeric αSyn lead to no significant changes in expression of M1-like markers including Nos2, IL-6, TNF, and IL-1β (Supplemental Figure 3).

To identify proteomic changes in microglia in response to αSyn_Agg, whole cell lysates of MMCs that had been exposed to αSyn_Agg were used for label-free mass spectrometry studies (6 biological replicates per group). MMC lysates in 8M urea buffer were enzymatically digested by trypsin and lysyl endopeptidase C, followed by LC-MS/MS, peptide identification and quantification. In total, we identified 35,725 total peptides (33,957 unique and razor assigned to proteomic database entries) that mapped to 3,816 unique mouse protein IDs and 3,738 unique mouse gene symbols. Of these, 3,345 proteins met inclusion criteria for further analysis (at least 3 non-missing values in either group or at least 2 non-missing values in one group if completely missing in the second group). Missing values in these included proteins were imputed using a R-based script designed to recapitulate the columnwise missing-value imputation algorithm of Perseus (44, 46, 47) commonly applied to MaxQuant LFQ abundances (see Supplemental Table 1).

Differential expression analysis comparing αSyn_Agg-treated with untreated MMCs identified 501 up-regulated and 749 down-regulated proteins (T-test unadjusted p-value < 0.05 and at least 1.25-fold change in either direction, Figure 1A). Even after adjusting for multi-pairwise comparisons (BH FDR <5%), 1,578 proteins were differentially expressed of which 109 proteins demonstrated at least 4-fold change in either direction (38 up-regulated and 71 down-regulated, see Supplemental Table 1). The top αSyn_Agg up-regulated proteins included several pro-inflammatory proteins such as Irg1, Ifit1, and Pyhin1 while αSyn_Agg-downregulated proteins included Sod1, Ahnak2, Cd93, and Thumpd1.

GO enrichment analyses of αSyn_Agg-upregulated and downregulated proteins were then performed. Nuclear and nucleolar proteins involved in RNA binding and ribosomal biogenesis, RNA splicing and anti-viral defense responses were highly represented within the αSyn_Agg-upregulated proteins (Figure 1B). KEGG pathways highly represented within these proteins included ribosome biogenesis, spliceosome, and fatty acid biosynthesis. Conversely, cytosolic proteins involved in oxidation-reduction and catabolic processes, proteasome core complex function and calcium binding were highly enriched within αSyn_Agg-downregulated proteins (Figure 1B). KEGG pathways enriched in this list included several small molecule metabolic pathways, as well as proteasome and lysosomal pathways. These results suggest that αSyn_Agg strongly induces RNA synthesis and splicing while suppressing homeostatic metabolic, mitochondrial, proteasomal, and lysosomal activities.

Canonical pathway analysis (Metacore) revealed that signaling via Stat3, Stat1, Oct3/4, and C/ebp transcriptional pathways are likely to be involved in αSyn_Agg-mediated regulation of protein expression (Figure 2A). In confirmatory studies in primary microglia, we further observed that αSyn_Agg robustly increased both native Stat3 (Figure 2B) and Stat3 tyrosine phosphorylation (Y705) (Figure 2C). Further experiments with monomeric αSyn showed no increase in Stat3 (Supplemental Figure 4A) or pStat3 (Supplemental Figure 4B) protein expression indicating αSyn_Agg-specificity of microglial responses. Together, we have identified probable pathways that regulate αSyn_Agg-induced microglial activation and pro-inflammatory mechanisms.

αSyn_Agg, like LPS, may induce microglial pro-inflammatory activation via TLR signaling (27, 48, 49) but in addition, may have unique effects on microglial activation via distinct mechanisms that are not completely understood. To identify αSyn_Agg-induced microglial protein changes that overlap with, or are distinct from LPS pro-inflammatory activation of microglia, we compared αSyn_Agg-induced differentially expressed proteins in this dataset with existing proteomic data from LPS-treated BV2 mouse microglia (37). 2,598 proteins quantified in our dataset were also quantified in this reference mouse microglial proteome comparing LPS-treated to untreated BV2 microglia (Supplemental Table 2) (37). Among these shared proteins, 1,472 were differentially expressed by αSyn_Agg (p < 0.05) of which 233 proteins were differentially expressed in both LPS and αSyn_Agg datasets (unadjusted p < 0.05), and overall level of concordance was low (Pearson's R = 0.18) (Figure 3A). While majority of LPS-differentially expressed proteins (67.9%) were also differentially expressed following αSyn_Agg, only 15.8% of αSyn_Agg-differentially expressed proteins were differentially expressed following LPS stimulation (Figure 3B). Among the shared proteins, the top concordant proteins included Irg1, Saa3, Sqstm1, Ehd1, Nadk, Icam1, and Marcksl1. These results indicate that while αSyn_Agg induces an LPS-like pro-inflammatory activation profile in microglia, the majority of αSyn_Agg-induced changes are distinct from LPS-induced changes.

To define the unique molecular mechanisms regulated by αSyn_Agg that are distinct from LPS-induced changes, we performed an analysis restricted to 596 proteins that were only differentially regulated by αSyn_Agg but not by LPS (proteins with ≥1.25-fold differential expression [p < 0.05] in response to αSyn_Agg, but p > 0.2 for LPS vs. control comparisons) (Figure 3C). GO analysis of 216 αSyn_Agg-upregulated (but not by LPS) proteins revealed enrichment of nuclear and nucleolar proteins involved in RNA metabolic processes, ribonuclear biogenesis, and splicing (Figures 3D,E). On the other hand, 380 αSyn_Agg-specific and downregulated proteins (Figures 3F,G) were enriched for cytosolic, extracellular and exosomal proteins involved in proteasomal function, small molecular metabolism, peptidase activity, cellular catabolic processes, mTOR signaling and proteolysis. Overall, these comparative analyses show that while some microglial responses to αSyn_Agg are similar to pro-inflammatory effects of LPS, αSyn_Agg also uniquely increases the expression of ribonucleoprotein and the RNA binding machinery while suppressing catabolic and protein degradation/proteasomal processes in microglia.

To derive a comprehensive list of known human PD risk genes identified by GWAS, we performed a meta-analysis of existing GWAS studies using MAGMA, and identified 622 genetic risk factors for PD (Supplemental Table 3) (15). We cross-referenced this list of PD risk genes with our microglia proteomic dataset and identified 28 proteins that were differentially expressed in microglia (≥2-fold change in either direction) in response to αSyn_Agg that also met GWAS-level statistical significance (MAGMA p-value <0.05) (Figure 4A, Table 2). We then performed qRT-PCR studies (Table 3, Figure 4B) in primary murine microglia after exposure to αSyn_Agg using identical experimental conditions to determine whether findings observed in the MMC microglial cell line can be replicated in primary mouse microglia. Of the 26 transcripts evaluated, congruent changes were observed for 3 Syn-upregulated (Brd2, Clk3, Siglec1) and 11 Syn-downregulated (Memo1, Arhgap18, Blnk, Fyn, Hspa1b, Isyna1, Itpa, Tpp2, Grn, Naglu, and Fam175b) proteins (Figure 4B, Table 3).

While most risk genes for PD regulate non-immune functions, genes/proteins that are most highly expressed in microglia are also most likely to regulate microglial functions and neuroinflammation in PD. Of 622 PD GWAS risk genes identified by MAGMA, 26 genes were most highly expressed in microglia based on a CNS cell-type-specific proteome from purified mouse microglia, astrocytes, oligodendrocytes and neurons (50). Of these microglial PD-risk genes, 5 proteins (Psmb9, Fam49b, Isyna1, Grn, and Naglu) were also differentially regulated by αSyn_Agg in our dataset (Figure 4A). Interestingly, these 5 proteins were all suppressed by αSyn_Agg by at least 2-fold, suggesting that polymorphisms in these 5 genes may partly replicate downstream immune effects of αSyn_Agg. These αSyn_Agg-regulated and human PD-risk proteins may represent immune genes with causative roles in PD.

We also compared αSyn_Agg-induced proteomic changes in microglia with an existing gene microarray dataset obtained from the post-mortem samples from the substantia nigra regions of PD and non-PD control patients in which 5,933 genes were differentially expressed (45) of which, 782 gene symbols were also identified in our microglial proteomic dataset. 339 genes/proteins of these 782 demonstrated differential expression in response to αSyn_Agg (Figure 4C) although poor concordance between direction of change was observed (rho = 0.1).

We observed that PGRN protein levels as well as Grn mRNA transcripts were concordantly decreased in mouse primary microglia in response to αSyn_Agg. Furthermore, Grn is highly expressed at the transcript and protein levels in mouse and human microglia (50, 51) in addition to being identified as a risk gene for PD (15). Therefore, we performed validation studies using an in-vivo model of αSyn-aggregate induced neuroinflammation in mice (36, 52). We analyzed brain tissues from mice that received stereotaxic injections with αSyn pre-formed fibrils (αSyn_PFF) (Supplemental Figure 5A). In this model, open-field versamax test revealed that αSyn_PFF induced motor behavioral changes in mice (Supplementary Figures 5B,C). Further, qPCR analysis of substantia nigra from the injected side validated Grn and other genes which were differentially regulated in our microglial proteomic dataset (Clk3, Golga3, Memo1, and Isyna1) (Supplemental Figure 5D). While these effects observed do not reflect microglia-specific alterations induced by αSyn, future studies will clarify the cell types responsible for these gene expression changes. Next, we determined patterns of PGRN protein expression in post-mortem brain tissues obtained from subjects with PD and age-matched non-disease controls, in SNpc and PFC regions (Figure 5). As expected, we observed the presence of Iba1+ microglia cells with ramified morphology in PFC in both PD and HC brains; whereas the PD SNpc displayed increased numbers of Iba1+ cells as compared to HC SNpc. PGRN immunoreactivity was predominantly observed in the melanized dopaminergic neurons in the SNpc, and to a lesser extent in cells with glial morphology in the SNpc. However, relative to HC SNpc, PD SNpc displayed more PGRN+ inclusions and intense labeling in cells lining vessels which did not have glial morphology. Unlike the SNpc, PGRN immunoreactivity in the PFC was seen in glial cells that predominantly had microglial morphology regardless of disease status. Double immunofluorescence studies of PGRN and microglial markers were confounded by lipofuscin-associated auto-fluorescence, limiting our ability to perform quantitative and microglia-specific analyses. Overall, since Grn is highly expressed at the transcriptomic level in microglia in mammalian brain, our in-vivo studies show a general agreement that PGRN expression is indeed observed in microglia in human PD cases, especially in the pre-frontal cortex.

Since our in vitro mouse microglia showed concordant dysregulation of mitochondrial proteins induced by αSyn_Agg, we performed Seahorse studies of mitochondrial stress in primary mouse microglia (Figures 6A,B). Though the role of mitochondrial dysfunction is well studied in neurons (53), the exact function of microglial mitochondrial dysfunction is still not well-understood. We have recently shown that mitochondrial dysfunction in microglia leads to inflammation by activation of NLRP3 inflammasome activation (31). Hence, to further validate that αSyn can induce mitochondrial damage in microglial cells, we performed seahorse mitostress test on primary mouse microglial cells, treated with 1 μM αSyn_Agg for 24 h to mirror our proteomic studies. αSyn altered mitochondrial dynamics in microglial cells as shown by changes in maximal respiration (Figure 6C), proton leak (Figure 6D), ATP production (Figure 6E) and basal respiration (Figure 6F). These confirmatory findings, together with our proteomics results, show that αSyn can cause mitochondrial dysfunction in microglial cells.

Patent pending related to this work entitled “Methods to treat neurodegeneration with granulins” to TK. AK is a shareholder of PK Biosciences Corporation (Ames, IA), which is interested in identifying novel biomarkers and potential therapeutic targets for PD. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.