Visual Abstract

Aggregation of α-synuclein (αSN) is an important histological feature of Parkinson disease. Recent studies showed that the release of misfolded αSN from human and rodent neurons is relevant to the progression and spread of αSN pathology. Little is known, however, about the mechanisms responsible for clearance of extracellular αSN. This study found that human complement receptor (CR) 4 selectively bound fibrillar αSN, but not monomeric species. αSN is an abundant protein in the CNS, which potentially could overwhelm clearance of cytotoxic αSN species. The selectivity of CR4 toward binding fibrillar αSN consequently adds an important αSN receptor function for maintenance of brain homeostasis. Based on the recently solved structures of αSN fibrils and the known ligand preference of CR4, we hypothesize that the parallel monomer stacking in fibrillar αSN creates a known danger-associated molecular pattern of stretches of anionic side chains strongly bound by CR4. Conformational change in the receptor regulated tightly clearance of fibrillar αSN by human monocytes. The induced change coupled concomitantly with phagolysosome formation. Data mining of the brain transcriptome in Parkinson disease patients supported CR4 as an active αSN clearance mechanism in this disease. Our results associate an important part of the innate immune system, namely complement receptors, with the central molecular mechanisms of CNS protein aggregation in neurodegenerative disorders.

Parkinson disease (PD), Lewy body dementia, and multiple system atrophy are among the most prevalent neurodegenerative diseases. Aggregation of the cytosolic protein α-synuclein (αSN) into cell body inclusions (i.e., Lewy bodies) (1), with αSN as the main component, is a shared histological hallmark of these diseases (2). Lewy bodies are terminal products of the complex pathway of αSN aggregate formation; they consist of fibrils of many thousands of αSN monomers. Many smaller types of αSN oligomers are also formed during aggregation. These oligomers may be the species responsible for cytotoxicity due to their high mobility and ability to perturb the cell membrane. Fibrils have been considered to be more innocuous, but study results reveal an important cell toxicity role for fibrils (3, 4). The αSN aggregates can be released to the extracellular environment and then transfer from neuron to neuron through receptor-mediated endocytosis (1). Some study findings suggested that this mechanism is a means of protective clearance (5). However, the intercellular transfer can initiate an inflammatory response in microglia and nucleate further intracellular aggregation, which ultimately exacerbates neurodegeneration and promotes disease (1, 6). Consistent with this notion, genetic PD risk variants are significantly enriched in gene sets functionally linked to the regulation of leukocyte activity (7). Microglial cells are the main resident myeloid leukocyte in the CNS. They are especially enriched in the substantia nigra, which shows the most prevalent neuronal death during PD (8). It is proposed that microglial cells are αSN scavengers (9). Secreted αSN is cytotoxic to recipient neural cells in vitro (10) and in vitro–generated oligomers of recombinant αSN are up to 17-fold more cytotoxic than monomers (11). αSN is one of the most abundant proteins in the CNS; it accounts for 0.5–1% of all cytosolic brain protein (12). To limit αSN cytotoxicity during neural cell death, receptor-mediated αSN clearance must bind αSN oligomers and fibrils while avoiding saturation with the monomeric species. A study using a murine model found that predominantly aggregated αSN is an agonist of TLR-2 and stimulates microglial activation (13). A mouse model of multiple system atrophy revealed that TLR-4 is associated with αSN clearance (14). It also associated with αSN activation of microglial cells by monomeric and fibrillar forms (15). However, although TLRs are critical for the sensing of danger-associated molecular patterns (DAMPs) (16), they are only a limited part of the physical clearance of particulates that carry DAMPs. The structure of DAMPs associated with αSN aggregation remains unknown.

Studies using solid-state nuclear magnetic resonance (17) and cryo-electron microscopy (1820) have revealed atomic-resolution structures of the central core of fibrillar αSN. However, the structure(s) of the toxic oligomeric state(s) is more elusive because of the highly dynamic nature of these species (21, 22). The αSN primary structure comprises three distinct regions: the N terminus, the nonamyloid-β-component, and the C terminus (23) (Fig. 1A). The lysine-rich αSN N terminus (residues 1–60) has an overall positive charge. In an aqueous solution, the monomeric form of αSN assumes a mostly unfolded structure. A cell membrane environment stimulates α-helical folding with a resulting structure that is remarkably similar to the neural protein, myelin basic protein (24). The αSN C terminus (residues 96–140) carries an overall negative charge and remains unfolded, even when in contact with cell membranes. In the fibrillar state, a central β-sheeted core assembles from parallel-stacked monomers (Fig. 1B). It then further assembles into a dimer of cores, still exposing some of the surface of the original β-sheeted core (1720). In the available structures (1720), the N and C termini are not visible. This characteristic may be due to fibril heterogeneity and terminus flexibility. The segments protrude away from the longitudinal axis of the fibril as highly charged, brush-like appendages to the core (17). On-pathway oligomers (21), or oligomers that coexist with the fibril form (25), are likely to have a related structure. β-Sheeted αSN forms are the most toxic (26). The parallel-type of stacking of the fibril and related oligomers creates stretches, sometimes referred to as ladders (27), of repeated side chains that are most noticeable in the β-sheet core (Fig. 1C), but they may also occur at least in patches of the unfolded regions. It remains unclear if this pattern or other structural patterns affect clearance of these toxic forms.

Complement receptor 3 (CR3; also named Mac-1, integrin αMβ2, or CD11b/CD18) and CR4 (p150,95, integrin αXβ2, or CD11c/CD18) mediate microglial phagocytosis (28). Both receptors bind proteolytic cleavage products of C3, which is part of the innate immune system. Human CR4 has a strong preference for molecules with motifs of uninterrupted negative charge, such as the proteins polyglutamate and osteopontin, and the glucosamine glycan heparin (29). By contrast, CR3 appears to interact with less homogeneous motifs containing positively charged protein side chains (24, 30, 31). These differences in ligand recognition are explainable from properties of the major ligand binding site in the α-chain of CR3 (αM) and α-chain of CR4 (αX), usually referred to as the inserted (I) domain. Both the αM and αX I domain (αMI and αXI, respectively) take the Rossmann fold with seven amphipathic α helices surrounding a hydrophobic β-sheet core (29). Opposite the connection to the main body of the receptors, the I domains chelate an Mg2+ ion in the metal ion-dependent adhesion site (MIDAS) (Fig. 2A, 2B). Despite high overall sequence and structural similarity, the human αMI and αXI differ with regard to the presentation of electrostatic charge near the MIDAS (32). The human αXI, especially, carries positive charges, which are involved in ligand recognition (33, 34), and almost no negative charge (Fig. 2C), which would further act to accommodate motifs of uninterrupted negative charge (29). The stretches of anionic charge, guided by the parallel organization of the αSN monomers in the aggregates, suggest that CR4 is a strong receptor, especially for fibrillar αSN, but, to our knowledge, CR4 has not been examined as a receptor for αSN. The roles of CR3 or CR4 in αSN phagocytosis have also not been determined. Indicating their importance in PD, the cerebral/cortical ITGB2 mRNA, which encodes the β-chain (CD18) of CR3 and CR4, was significantly upregulated in microglia from a small cohort of five PD patients and three age-matched controls (35). However, it is unknown if the α-chains of these receptors, critical for ligand binding, are differentially expressed in PD.

CR3 and CR4 ligand-binding activity is tightly regulated by conformational changes in the receptor ectodomain (29). In the ligand-binding inactive state, the α- and β-chains form bent conformations, and the α-chain ligand-binding domain (i.e., the I domain) is in proximity to the cell membrane. Alterations in contact between the integrin cytoplasmic tails and the adaptor proteins talin and kindlin-3 enable cytoskeletal rearrangements, which permits opening of the ectodomain into the ligand-binding conformation (29, 36). Despite recent study findings on CD18 integrins in neurodegenerative diseases (37), the roles of conformational activation of CD18 integrins in neuroinflammatory disease remain unknown.

In this study, we found that fibrillar αSN was a strong ligand for CR4, but the monomeric form was not. Our results were also consistent with those of a previous study of CR3 as a receptor for presumably unaggregated αSN (38). Unlike CR4, CR3 showed no major difference in the recognition of monomeric and aggregated forms of αSN. Conformational activation of CD18 integrins strongly enhanced phagocytosis concomitantly with the intracellular formation of lysosomal vesicles. Conformationally activated CD18 integrins efficiently cleared fibrillar αSN but not monomeric αSN. Guided by these results, we reanalyzed published PD brain transcriptome data and found a previously unappreciated upregulation of expression of all components for CR3 and CR4 in patients with PD compared with age-matched controls. This upregulation correlated with the gene expression of the conformation-regulating proteins kindlin-3 and talin. Our data now indicate an important role in PD of CR4 ligand selectivity and conformational activation in clearance of fibrillar αSN.

Human wild-type (Wt) αSN and αSNΔ2–11 were prepared using recombinant expression in Escherichia coli as a protein source (39).

For the cellular experiments with fibrillar αSN, monomeric αSN (346 μM) was assembled into preformed fibrils (PFF) using incubation under sterile conditions at 37°C in PBS (pH 7.4) (Life Technologies) with continuous shaking at 1050 rpm (Eppendorf ThermoTop) for 7 d. Aggregation was monitored by removing samples for thioflavin S fluorescence spectroscopy. The final incubation product was sedimented using centrifugation at 15,600 × g for 20 min to isolate the insoluble PFF from the soluble αSN. After αSN sedimentation, the PFFs were diluted to 2 mg/ml in sterile PBS (pH 7.4) (Life Technologies) and subjected to ultrasound breakage for 20 min using a Branson-Emerson 250 Analog Sonifier equipped with a water jacket cooling system to avoid sample heating. The settings were a 30% duty cycle and an output control of 3. The size-distribution profiles of the PFFs in suspension were measured using dynamic light scattering (DynaPro NanoStar instrument; Wyatt) at 25°C. The data analysis of the PFFs sample showed a homogeneous monodisperse population of 44-nm hydrodynamic-radius PFFs.

For transmission electron microscopy (TEM) experiments, lyophilized powder of human Wt αSN (40) was dissolved in PBS buffer and filtrated using 0.22-μm filters. Thioflavin T (ThT) dye was added to a final concentration of 20 μM and 6 mg/ml αSN. αSN was fibrillated in a 96-well plate (Thermo Fisher Scientific) by incubation and shaking at 37°C for 7 d. The fibrillation was done in triplicates of 150 μl with a 3-mm glass bead added in each well for agitation. ThT fluorescence was followed in a Fluostar Optima plate reader (BMG Labtech) by the emission at 480 ± 5 nm upon excitation at 450 ± 5 nm to ensure full fibrillation. For buffer change and separation from soluble αSN fractions, the fibrils were sedimented, washed, and finally resuspended in Tris buffer (150 mM NaCl, 20 mM Tris [pH 7.4]). The fibrils were subsequently sonicated with 10 s pulsed sonication using a Sonopuls mini20 (Bandelin).

Primary human monocytes were isolated from buffy coats obtained with an established collaboration with the Aarhus University Hospital Blood Bank according to ethically approved protocols (Protocol No. 77). The monocytes were isolated from the buffy coat, by initial erythrocyte depletion by density gradient centrifugation (no. 17-1440-02, Ficoll-Paque PLUS; GE Healthcare) followed by negative selection using Dynabeads Untouched Human Monocytes kit (no. 11350D; Invitrogen). Purified cells were stored at −135°C in RPMI 1640 with l-glutamine, 20% (v/v) heat-inactivated FCS (Life Technologies), and 10% (v/v) DMSO. The cells were thawed immediately before use. Cell population viability was >90% for all experiments.

Immortalized myelogenous K562 cell lines with recombinant expression of CR3 or CR4 were made and cultivated as described (41) together with the parental K562 cell line. Briefly, the cells were cultured at 37°C and 5% CO2 in RPMI 1640 with NaHCO3 and 10 mM HEPES [pH 7.2], 10% (v/v) FCS, and penicillin and streptomycin. Selection for recombinant expression was maintained by adding 4 μg/ml puromycin dihydrochloride to the CR3/K562 culture medium and 16 μg/ml hygromycin B to the CR4/K562 medium.

Cell adhesion was tested using a centrifugation-based assay as described (42). In brief, polystyrene 96-well microtiter plates with v-shaped bottoms (no. 3896; Costar) were coated for 1 h at 37°C with Wt αSN diluted in 150 mM NaCl and 20 mM Tris [pH 9.4] (coating buffer) at concentrations of 0.940, 1.88, 3.75, 7.5, 15, or 30 μg/ml or left uncoated for reference. Each concentration was prepared in triplicate. To remove oligomeric αSN species, some plates were emptied, and 100 μl 6 M guanidine hydrochloride (Gu·HCl) was added. The samples were then dissolved in coating buffer, followed by incubation for 1.5 h at room temperature. All plates were then washed in 200 μl PBS with 0.05% (v/v) Tween 20 and further blocked in this buffer for 1 h at 37°C.

Monocytes were thawed and added to PBS supplemented with 20% (v/v) FCS, collected using centrifugation, and resuspended in RPMI 1640 supplemented with 2% (v/v) FCS. The monocytes were fluorescently labeled using incubation with 2,7-bis(2-carboxyethyl)-5(6)-carboxyfluorescein acetoxymethyl ester (no. 14562; Sigma-Aldrich) at 37°C and 5% (v/v) CO2 for 15 min, washed twice, and resuspended in 150 mM NaCl, 5 mM KCl, 1 mM MgCl2, 1.8 mM CaCl2, 10 mM HEPES [pH 7.4], with 5 mM glucose and 2.5% (v/v) FCS (binding buffer) to a final cell concentration of 6–10 × 105 cells/ml. The cells were centrifuged for 5 min at 230 × g, washed twice, and resuspended in binding buffer with 5 μg/ml CD18 integrin-activating Ab KIM127 (CRL-2838). To test the contribution of CR4 to adhesion, 1 μg/ml predialyzed function-blocking Ab was added to the αX chain (clone “3.9”, MA1-46052; Thermo Fisher Scientific). Mouse IgG1 (M7894; Dako) was added to obtain an isotypic control. The microtiter plates were emptied and kept at 37°C in a heating block, a 100-μl cell suspension was then transferred to each well using an automatic multichannel pipette, and the plates were incubated at 37°C with CO2 for 10 min. The plates were then centrifuged at 50 × g for 5 min, and the fluorescence count was read in the nadir of the wells using a Victor3 1420 multilabel counter (485-nm excitation wavelength, 535-nm emission wavelength; Wallac).

The K562 cell lines were treated as for the human monocytes, except that integrin activation was achieved by addition of 1 mM MnCl2 rather than Ab. Appropriate centrifugation force was achieved using centrifugation at 10 × g for 5 min and again at 50 × g for 5 min. The fluorescence count was recorded at each step.

The surface plasma resonance (SPR) assays were performed in CM-4 chips and run in the BIAcore 3000 instrument (GE Health Care). The chip surfaces were coupled with Wt αSN or αSN with an N-terminal deletion missing residues 2–11 (αSNΔ2–11) using the amine-coupling chemistry method as described (43). A reference surface was prepared by coupling the surfaces with ethanolamine rather than protein.

The αMI and αXI were stabilized in the activation conformation by mutation of a C-terminal Ile residue to glycine and were prepared as described earlier (31). The I domains were diluted in 150 mM NaCl, 1 mM MgCl2, 5.0 mM, HEPES [pH 7.4] (running buffer) to concentrations of 156, 325, 625, 1250, and 2500 nM. The prepared solutions were then injected over the surfaces; the contact time (tc) was 245 s and the following dissociation phase time was 255 s. The surfaces were then regenerated in 50 mM EDTA, 1.5 M NaCl, and 0.1 M HEPES [pH 7.4]. The data collection rate was one data point per 0.4 s.

The sensorgrams were manually aligned using BIAevaluation software (GE Healthcare), and the signals from the reference surfaces were subtracted from the signals from the ligand-coated surfaces. The resulting αMI sensorgrams were analyzed using an algorithm for combined affinity and rate constant distributions of ligand populations from experimental surface binding kinetics and equilibria using the fitting tool “EVILFIT” (44) implemented in MATLAB 2012a (Mathworks). The injection start was 0 s and the injection end was 240 s; the dissociation start was 250 s and the dissociation end was 450 s. The operator-set boundaries for the distributions were uniformly set to limit the KD values in the 10−10–10−2 M interval and the dissociation kd values in the 10−4–10−1 s−1 interval. In the case of the αXI, a more limited analysis was made using the BIAevaluation software (GE Healthcare, Norwalk, CT). kd was calculated from the sensorgram by 1:1 Langmuir binding isotherm for the dissociation phase defined by the kd = 1/S × dS/dt, where S is the SPR signal and with local fitting applied.

For preparation of EM grids, the sonicated αSN fibrils and the αXI were mixed in a 1:1 M ratio (12 μM, based on the αSN monomer concentrations) with added 2 mM MgCl2 or 2 mM EDTA, respectively. The mixtures were then incubated for 30 min at room temperature. Five-microliter samples were then allowed to absorb onto glow-discharged Formvar/carbon-coated 200 mesh Cu grids (Electron Microscopy Sciences) for 1 min before blotting and washing with 5 μl MQ H2O in a 1 min incubation. The absorbed material was negatively stained using a 1 min incubation with 5 μl 2% (w/v) uranyl formate. The TEM was performed using a CM100 TWIN Transmission Electron Microscope (Philips).

For establishing assays to characterize cellular uptake and size selectivity in phagocytosis of αSN, monocytes from two donors were purified and used separately in six independent experiments as described in Fig. 3. The monocytes were retrieved from −135°C storage on the day of experiment and kept on dry ice until use (Fig. 3A). Biotinylated Wt and fibrillar αSN samples were preincubated with 12.5 μl of a 40-nM streptavidin/quantum dot (Q-dot) solution (no. Q10123MP; Molecular Probes) for 30 min at 37°C; the final concentrations of proteins were 10 or 20 μg/ml. The cells were thawed and resuspended in 1 ml RPMI to a concentration of 40 × 106 cells/ml. Twenty-five microliters of cell suspension was added to the αSN/Q-dot solutions or Q-dots without protein (control) together with 5 μg/ml CD18 integrin-activating Ab KIM127 (CRL-2838; ATCC), followed by incubation for 30 min at 37°C with 5% (v/v) CO2. All samples were then washed twice with PBS. After centrifugation at 230 × g for 5 min, the supernatant was saved for nanoparticle tracking analysis (NTA) (described below) (Fig. 3B). Five microliters CD14 Ab conjugated with Brilliant Violet 421 (no. 301829; BioLegend) was then added to each sample, and the samples were incubated for 25 min at 37°C with 5% (v/v) CO2. Ten microliters of a 1:200 dilution in PBS of 1 mM Lyso Tracker stain Green DND 26 (no. L7526; Thermo Fisher Scientific) was then added to each sample, followed by an additional 5 min incubation. The samples were then washed twice in PBS and resuspended in 50 μl PBS. Five microliters of 100 μM DNA nucleus stain CyTRAK Orange (CO50050; Biostatus) was then added, and all tubes were incubated for 15 min. All samples were then kept on ice until imaging flow cytometry analysis. Flow cytometry was performed using an Amnis ImageStreamX MKII (Amnis, Seattle, WA); the sensitivity was set to high with a 60× image magnification. Images from 40,000–60,000 cells were recorded for all samples. Membrane and intracellular masks were determined from the locations of the CD14 staining. The data were analyzed using the IDEAS software package (Amnis) (Fig. 3C).

The supernatants obtained in the experiments described above (Fig. 3B) were diluted in PBS (1:1000) to obtain a particle concentration suitable for analysis. The particles present in the samples were analyzed using a NanoSight LM10 system (Malvern Instruments, Malvern, United Kingdom) (Fig. 3C). The system was configured with a 405-nm laser and a high-sensitivity scientific complementary metal–oxide–semiconductor camera (OrcaFlash2.8, Hamamatsu C11440; Malvern Instruments). The sample chamber was washed twice with PBS before each measurement. All samples were thoroughly mixed before measurement and were then injected into the sample chamber using 1-ml syringes. The measurements were initialized within 10 s of injection into the chamber. Approximately 20–70 particles were in the field of view, corresponding to 2 × 108–1.3 × 109 particles/ml. The videos were collected and analyzed using NTA software (version 2.3, build 0025). The automatic settings were used for the minimal expected particle size, minimum track length, and blur setting. To enable recording of the movements of small particles, the camera sensitivity was set to maximum (level 16) and the detection threshold was set close to minimum (level 3). A 650-nm long-pass filter was used for all recordings. The temperature (range, 23–25°C) was recorded manually. Three 60-s-duration videos were recorded for each sample (i.e., three replicates for each measurement).

To analyze the biological variation in the size-selective phagocytosis, a further seven donors were subjected to the NTA-based analysis following the procedures described in Fig. 3A–C, however, without technical replicates and only using the highest concentration of Wt or fibrillar αSN (10.0 μg/ml; or blank as control).

The RNA sequencing (RNA-seq) data from prefrontal cortex Brodmann area 9 from 29 patients with PD and 44 controls (45) were downloaded from the Gene Expression Omnibus data repository (https://www.ncbi.nlm.nih.gov/geo/; accession code no. GSE68719). The reads were quality filtered and mapped to the Ensembl Homo Sapiens. GRCh38.94 human reference genome using HISAT2 (version 2.1.0) (46). Gene expression quantification was performed using StringTie version 1.3.4 (47), the Ensemble Homo Sapiens (GRCh38.94 annotation gtf file), and the alignment BAM files. Gene abundances were reported as transcripts per million (TPM).

For the heatmaps, the rows were centered, and unit variance scaling was applied to the rows. The rows and columns were clustered using correlation distance and average linkage. Lists of glia specifically expressed genes were obtained from https://web.stanford.edu/group/barres_lab/brain_rnaseq.html with a low count threshold of 1 fragment per kilobase per million. Only genes with TPM >1 in the StringTie count files were considered for inclusion.

To assess expression of integrin-related genes in monocytes in PD patients and controls, TPM values based on RNA-seq data reported in Schlachetzki et al. (48) were obtained from Gene Expression Omnibus data repository (https://www.ncbi.nlm.nih.gov/geo/; accession code no. GSE88888). For comparison of transcriptomic profiles between microglia and circulating monocytes, normalized gene expression matrices were obtained from RNA-seq data [https://science.sciencemag.org/content/356/6344/eaal3222.long (49)].

All statistical analyses were performed using Prism software (GraphPad Software, San Diego, CA). A p value <0.05 was considered to be statistically significant. All information on experimental replicates, number of donors tested, and applied statistical tests are stated in figure legends.

Primary human monocytes coexpress CR3 and CR4 similarly to human microglial cells (50). We used these cells as a robust model system for investigating the ability of human CD18 integrins to bind in cell adhesion experiments with titrations of the αSN coating concentration (Figs. 13). The cell adhesion in each titration point was determined from independent experiments with monocytes from three donors. The titration experiments were analyzed in a one-way ANOVA involving all titration points. When the CD18 integrin-activating Ab KIM127 was added, monocytes adhered robustly to microtiter wells coated with recombinant Wt αSN and were significantly stronger than for conditions without KIM127 (Fig. 4A). When adding a function-blocking Ab to the αXI of CR4, the adhesion of the monocytes was significantly reduced compared with conditions in which an isotypic control Ab was added. The peak in difference between the titration curves for the function-blocking and control Abs was at a αSN coating concentration of 15 μg/ml. To test if this amounted to a significant difference in a single titration point, we increased the number of independent experiments for this condition with monocytes from a total of eight donors permitting the use of a nonparametric (paired) Wilcoxon test. For all donors, the addition of function-blocking Ab reduced the adhesion. The variation between donors in inhibition relative to control Ab-treated cells ranged from 60 to 0.4%, with a mean value at 16% (Fig. 4B); the difference was highly significant in testing of the absolute adhesion for function-blocking versus control Ab-treated monocytes with p < 0.0078. This result clearly indicated that CR4 participated in the adhesion to αSN as expressed in the monocyte cell membrane.

FIGURE 1.

Structural models of monomeric and filament αSN. Monomer organization of human Wt (Wt) αSN. (A) An illustrative model of human Wt αSN, generated using an ensemble optimization method based on SAXS data (77), where this model is one of many plausible conformations. The N-terminal domain (residues 1–60) is orange, the nonamyloid-β-component (NAC) region (residues 61–95) is green, and the C-terminal (residues 96–140) is teal. The positive-charge lysine residues are blue (there are no arginine residues in the sequence), and the negative-charge residues (glutamate and aspartate) are red. (B) Organization of human single protofilament/fibrillar αSN based on Protein Data Bank entry 2N0A (17). The β-sheet core, which resembles a “Greek key,” is shown with charged residues colored as for the monomeric αSN. (C) Zoom-in on part of the filament αSN in the β-sheet region with anionic stretches generated by the aligned Glu residues. The measurements also indicate the distances between the Cαs of Glu46 in the strands.

FIGURE 1.

Structural models of monomeric and filament αSN. Monomer organization of human Wt (Wt) αSN. (A) An illustrative model of human Wt αSN, generated using an ensemble optimization method based on SAXS data (77), where this model is one of many plausible conformations. The N-terminal domain (residues 1–60) is orange, the nonamyloid-β-component (NAC) region (residues 61–95) is green, and the C-terminal (residues 96–140) is teal. The positive-charge lysine residues are blue (there are no arginine residues in the sequence), and the negative-charge residues (glutamate and aspartate) are red. (B) Organization of human single protofilament/fibrillar αSN based on Protein Data Bank entry 2N0A (17). The β-sheet core, which resembles a “Greek key,” is shown with charged residues colored as for the monomeric αSN. (C) Zoom-in on part of the filament αSN in the β-sheet region with anionic stretches generated by the aligned Glu residues. The measurements also indicate the distances between the Cαs of Glu46 in the strands.

Close modal
FIGURE 2.

Homology modeling and calculation of electrostatic surface potentials for human and rodent αXI and αMI. (A and B) Models of the mouse and rat αXI and αMI were created using the SWISS-MODEL server (58), with the primary structures for Mus musculus and Rattus norvegicus taken from Bajic et al. (31) and based on the Protein Data Bank entries for open-conformation human αXI 4NEN (78) and αMI 1IDO (79) as templates. (A) Structural alignment of human (purple), mouse (teal), and rat (orange) αXI. The location of the Mg2+ ion in the MIDAS is shown as a green sphere. (B) Same analysis and coloring with the human, mouse, and rat αMI. (C) Electrostatic surface potential representation from −5 kT/e (red) to 5 kT/e (blue) for human and rodent αXIs and αMIs. All surfaces were scaled and oriented according to the structures in (A) and (B). The electrostatic surface potential was calculated using Adaptive Poisson-Boltzmann Solver using default parameters (80). Because of the difficulties in calculating the electrostatic potential of coordinated divalent metal ions (81), structures were modeled without such, which makes the electrostatic charge of the unoccupied MIDAS negative (red).

FIGURE 2.

Homology modeling and calculation of electrostatic surface potentials for human and rodent αXI and αMI. (A and B) Models of the mouse and rat αXI and αMI were created using the SWISS-MODEL server (58), with the primary structures for Mus musculus and Rattus norvegicus taken from Bajic et al. (31) and based on the Protein Data Bank entries for open-conformation human αXI 4NEN (78) and αMI 1IDO (79) as templates. (A) Structural alignment of human (purple), mouse (teal), and rat (orange) αXI. The location of the Mg2+ ion in the MIDAS is shown as a green sphere. (B) Same analysis and coloring with the human, mouse, and rat αMI. (C) Electrostatic surface potential representation from −5 kT/e (red) to 5 kT/e (blue) for human and rodent αXIs and αMIs. All surfaces were scaled and oriented according to the structures in (A) and (B). The electrostatic surface potential was calculated using Adaptive Poisson-Boltzmann Solver using default parameters (80). Because of the difficulties in calculating the electrostatic potential of coordinated divalent metal ions (81), structures were modeled without such, which makes the electrostatic charge of the unoccupied MIDAS negative (red).

Close modal
FIGURE 3.

Schematic representation of interdonor and interassay variation testing and total set of experiments for measuring Q-dot internalization using ImageStream flow cytometry and NTA. (A) Buffy coats from two donors were used for purification of monocytes. For each donor, approximately ≥9 × 106 cells were distributed into each of three vials. (B) The vials were treated with KIM127 Ab to activate CD18 integrins or were untreated. As indicated with colored lines, the KIM127 treatments were distributed to enable an indication of the interdonor and interassay variation, both with and without KIM127 treatment. Samples from each of the six experiments were further mixed with 2.5–10 μg/ml Wt or fibrillar (Fibril) αSN with Q-dots, or treated as controls. (C) Following incubation, the supernatants were collected and stored at −20°C for later NTA profiling. The cell fraction was kept on ice for immediate imaging flow cytometry analysis.

FIGURE 3.

Schematic representation of interdonor and interassay variation testing and total set of experiments for measuring Q-dot internalization using ImageStream flow cytometry and NTA. (A) Buffy coats from two donors were used for purification of monocytes. For each donor, approximately ≥9 × 106 cells were distributed into each of three vials. (B) The vials were treated with KIM127 Ab to activate CD18 integrins or were untreated. As indicated with colored lines, the KIM127 treatments were distributed to enable an indication of the interdonor and interassay variation, both with and without KIM127 treatment. Samples from each of the six experiments were further mixed with 2.5–10 μg/ml Wt or fibrillar (Fibril) αSN with Q-dots, or treated as controls. (C) Following incubation, the supernatants were collected and stored at −20°C for later NTA profiling. The cell fraction was kept on ice for immediate imaging flow cytometry analysis.

Close modal
FIGURE 4.

Cell adhesion to αSN-coated surfaces. (A) Monocyte adhesion was studied in four conditions using a centrifugation-based assay in the presence or absence of CD18 integrin-activating Ab KIM127. The contribution of CR4 was tested using murine function-blocking Ab 3.9 to CD11c with murine isotypic IgG1 Ab as a control. For each condition, the mean and SEM are indicated from independent experiments using monocytes from three donors applied to each condition. The statistical analyses were performed using two-way ANOVAs and Bonferroni correction for multiple comparisons. (B) From the peak in difference between conditions with and without function-blocking Ab at a coating concentration of 15 μg/ml Wt αSN shown in (A), the adhesion at this coating concentration was analyzed using monocytes from eight donors. Adhesion was made with function-blocking or isotypic control Ab added. Paired results for each donor, indicated with color label, are shown with connecting, hatched lines. The mean values with and without function-blocking Ab are indicated with the ends of a curly bracket and the ± SEM with connected black bars. The relative inhibition by the 3.9 Ab, shown next to the bracket, was calculated from normalization to the adhesion in the presence of the control Ab and shown as the mean value ± SEM. The statistical comparison was made with a (paired) Wilcoxon test. (C) Adhesion of K562 cells with a recombinant expression of CR3 and CR4 and parental K562 cells as a control repeated in three independent experiments. Integrins were activated by the addition of MnCl2. The effect of fibrillar αSN on CR3 and CR4 binding was tested using preincubation of the coated surfaces with Gu·HCl. For each condition, the error bars indicate the mean and SEM values from the independent experiments. The statistical analyses were performed using two-way ANOVAs and Bonferroni correction for multiple comparisons as in (A).

FIGURE 4.

Cell adhesion to αSN-coated surfaces. (A) Monocyte adhesion was studied in four conditions using a centrifugation-based assay in the presence or absence of CD18 integrin-activating Ab KIM127. The contribution of CR4 was tested using murine function-blocking Ab 3.9 to CD11c with murine isotypic IgG1 Ab as a control. For each condition, the mean and SEM are indicated from independent experiments using monocytes from three donors applied to each condition. The statistical analyses were performed using two-way ANOVAs and Bonferroni correction for multiple comparisons. (B) From the peak in difference between conditions with and without function-blocking Ab at a coating concentration of 15 μg/ml Wt αSN shown in (A), the adhesion at this coating concentration was analyzed using monocytes from eight donors. Adhesion was made with function-blocking or isotypic control Ab added. Paired results for each donor, indicated with color label, are shown with connecting, hatched lines. The mean values with and without function-blocking Ab are indicated with the ends of a curly bracket and the ± SEM with connected black bars. The relative inhibition by the 3.9 Ab, shown next to the bracket, was calculated from normalization to the adhesion in the presence of the control Ab and shown as the mean value ± SEM. The statistical comparison was made with a (paired) Wilcoxon test. (C) Adhesion of K562 cells with a recombinant expression of CR3 and CR4 and parental K562 cells as a control repeated in three independent experiments. Integrins were activated by the addition of MnCl2. The effect of fibrillar αSN on CR3 and CR4 binding was tested using preincubation of the coated surfaces with Gu·HCl. For each condition, the error bars indicate the mean and SEM values from the independent experiments. The statistical analyses were performed using two-way ANOVAs and Bonferroni correction for multiple comparisons as in (A).

Close modal

These findings were further supported by the use of K562 cells with recombinant expression of CR4 and CR3 (Fig. 4C). Mn2+ activation of the integrins again induced strong adhesion to αSN in CR4/K562 and, albeit more attenuated, also in CR3/K562 cells compared with conditions without integrin activation. As noted earlier (39), the applied Wt αSN contains a small amount of aggregated αSN (Supplemental Fig. 1). To test the effect of aggregates on cell adhesion, surfaces were treated with Gu·HCl. This exposure caused a significant reduction in the adhesion in CR4/K562 cells, whereas the adhesion in CR3/K562 cells was unchanged.

To better understand on a quantitative basis how CR3 and CR4 recognize αSN, we followed a strategy used in previous studies to design a SPR assay (43, 51). αMI and αXI were injected over surfaces covalently coupled with Wt αSN. The well-established heterogeneous αMI interactions with ligands (29) were analyzed by resolving the combined set of association and dissociation phases (Fig. 5A, 5C, 5E, 5G) into an ensemble of 1:1 interactions; each was typified by their ka and kd rates (44, 52). From the simple relationship KD = kd/ka, the distribution in ligand binding kinetics for the ensemble was shown in three-dimensional plots with axes of kd, the equilibrium constant KD, and the SPR signal, R, in arbitrary resonance units (RU) with contours to indicate the volume of each type of 1:1 interaction (Fig. 5B, 5D, 5F, 5H). For αMI, the model was clearly consistent with the experimental data as shown by the residuals in panels below the sensorgrams and small root-mean square deviations not exceeding 2% of the maximum SPR signal in either of the experiments (Fig. 5A, 5C, 5E, 5G). The ensemble binding to native Wt αSN (Fig. 5B) was easily divided by eye into four bins (I–IV); for each bin, the weighted mean KD and kd were calculated with results listed in Table I. For native Wt αSN, the ensembles comprised interactions with KD values in the orders of ∼10−4 and 10−6 M (bins I–III). A population that was minor (bin IV), yet distinct and with a good total signal, had a KD of 10−8 M (Fig. 5B). Treatment of the surface with Gu·HCl reduced the SPR signals (Fig. 5E) roughly proportional to the loss of dry-mass level (38%) of immobilized αSN (Fig. 5A, 5E). Although some features were missing because of the total lower signal, the ensemble binding was largely similar to the untreated surfaces (Fig. 5B, 5F). We also performed experiments with an N-terminal truncated αSN that contained a deletion of αSN residues Asp2 to Ala11 (αSNΔ2–11) (Fig. 5C). We used this construct to represent the minimal engineered intervention sufficient to reduce formation of αSN aggregates (39). As indicated from the SPR signal, the αSNΔ2–11 was a somewhat poorer ligand for αMI compared with Wt αSN (Fig. 5A, 5C). Shown by the contour plots, αSNΔ2–11 differed because it had almost 10-fold less (∼5 RU) of the high affinity interactions in bin IV (Fig. 5B, 5D) compared with the Wt construct (∼45 RU), whereas interactions in the other bins were only reduced ∼2-fold. As in the case of Wt αSN, Gu·HCl removed noncovalently bound material (Fig. 5C, 5G), which caused a drop in total signal but did not change the KD distribution.

FIGURE 5.

Interactions of the CR3 (αMI) and CR4 (αXI) ligand-binding domains to αSN-coupled surfaces. (AL) SPR surfaces were coupled with Wt αSN (A, E, and I) or truncated αSN (αSNΔ2-11) (C, G, and J), applied to the experiments in their native states (A, C, I, and J) or following treatment with Gu·HCl (E, G, K, and L). Sensorgrams for a titration of I domain concentration are shown with indications of the start of the injection (ti) and the end of the tc and with the residuals between the experimental data (in colored lines) and the model (in broad gray lines) shown in panels below the sensorgram. The root-mean square deviation (RMSD) between the experimental data and the model applied was also calculated and stated for each set of sensorgrams (A, C, E, G, and I). For the data in (J)–(L), no modeling was made. The amount of immobilized protein for each of the surfaces is stated in picomoles per square millimeter. SPR signals were analyzed as ensembles of 1:1 interactions, each interaction typified by its equilibrium dissociation constant (KD) and dissociation rate (kd) (29, 43, 52, 82). Results are presented on two-dimensional grids with log10 (KD) on the abscissa and log10 (kd) on the ordinate axes, and contours (in RU, with 10-RU stepping) indicating the amount of the interaction. (A–D) Binding of the αMI to native Wt αSN and αSNΔ2-11. In (B), (D), (F), and (H) the four bins, numbered I–IV, are shown for quantifying the KD and kd values listed in Table I. (E–H) Same analyses as in (A)–(D) for surfaces treated with Gu⋅HCl with 2-RU stepping between contours in (F) and (H). (I–L) Binding of the αXI to either native (I and J) or Gu⋅HCl-treated (K and L) Wt αSN and αSNΔ2-11. SPR experiments were repeated twice on the same chip with decreased signal but overall similar binding kinetics in the second run of the same chip and again twice with a fresh chip and a similar decrease in signal for the second run. Shown sensograms are representative of the runs from the fresh chip. (MO) TEM imaging of the αXI binding to sonicated αSN fibrils. Images were made either with sonicated fibrils alone (M), with sonicated fibrils and αXI in Mg2+-containing buffer permitting binding of the I domain (N), or, as a control, with sonicated fibrils and αXI in EDTA-containing buffer not permitting such binding (O). In (N) and (O), αXI-like features in proximity of the fibrils are highlighted using a yellow circle next to the feature. The diameter of the circle was equivalent to ∼4.5 nm; this result was consistent with a previous study of negative-stained TEMs of αXI using class averaging (83). A scale bar (200 nm) is below the micrographs.

FIGURE 5.

Interactions of the CR3 (αMI) and CR4 (αXI) ligand-binding domains to αSN-coupled surfaces. (AL) SPR surfaces were coupled with Wt αSN (A, E, and I) or truncated αSN (αSNΔ2-11) (C, G, and J), applied to the experiments in their native states (A, C, I, and J) or following treatment with Gu·HCl (E, G, K, and L). Sensorgrams for a titration of I domain concentration are shown with indications of the start of the injection (ti) and the end of the tc and with the residuals between the experimental data (in colored lines) and the model (in broad gray lines) shown in panels below the sensorgram. The root-mean square deviation (RMSD) between the experimental data and the model applied was also calculated and stated for each set of sensorgrams (A, C, E, G, and I). For the data in (J)–(L), no modeling was made. The amount of immobilized protein for each of the surfaces is stated in picomoles per square millimeter. SPR signals were analyzed as ensembles of 1:1 interactions, each interaction typified by its equilibrium dissociation constant (KD) and dissociation rate (kd) (29, 43, 52, 82). Results are presented on two-dimensional grids with log10 (KD) on the abscissa and log10 (kd) on the ordinate axes, and contours (in RU, with 10-RU stepping) indicating the amount of the interaction. (A–D) Binding of the αMI to native Wt αSN and αSNΔ2-11. In (B), (D), (F), and (H) the four bins, numbered I–IV, are shown for quantifying the KD and kd values listed in Table I. (E–H) Same analyses as in (A)–(D) for surfaces treated with Gu⋅HCl with 2-RU stepping between contours in (F) and (H). (I–L) Binding of the αXI to either native (I and J) or Gu⋅HCl-treated (K and L) Wt αSN and αSNΔ2-11. SPR experiments were repeated twice on the same chip with decreased signal but overall similar binding kinetics in the second run of the same chip and again twice with a fresh chip and a similar decrease in signal for the second run. Shown sensograms are representative of the runs from the fresh chip. (MO) TEM imaging of the αXI binding to sonicated αSN fibrils. Images were made either with sonicated fibrils alone (M), with sonicated fibrils and αXI in Mg2+-containing buffer permitting binding of the I domain (N), or, as a control, with sonicated fibrils and αXI in EDTA-containing buffer not permitting such binding (O). In (N) and (O), αXI-like features in proximity of the fibrils are highlighted using a yellow circle next to the feature. The diameter of the circle was equivalent to ∼4.5 nm; this result was consistent with a previous study of negative-stained TEMs of αXI using class averaging (83). A scale bar (200 nm) is below the micrographs.

Close modal
Table I.
Binding kinetic parameters for the CR3 αMI
Wt αSN
αSNΔ2–11
KD (M)kd (s−1)KD (M)kd (s−1)
Native structured proteina     
 Bin I (4.6 ± 0.8) × 10−5 (3.4 ± 1.7) × 10−3 (5.3 ± 0.3) × 10−5 (5.4 ± 0.1) × 10−3 
 Bin II (9.6 ± 8.5) × 10−5 (1.2 ± 2.2) × 10−2 (3.5 ± 2.3) × 10−5 (1.2 ± 5.0) × 10−3 
 Bin III (5.5 ± 0.4) × 10−7 (1.9 ± 0.6) × 10−2 (7.9 ± 5.1) × 10−7 (2.2 ± 0.4) × 10−2 
 Bin IV (0.8 ± 0.8) × 10−8 (3.2 ± 3.2) × 10−3 (0.8 ± 0.8) × 10−8 (3.3 ± 3.3) × 10−3 
Gu·HCl-treated (denatured) proteina  
 Bin Ib N/D N/D N/D N/D 
 Bin II (8.5 ± 6.6) × 10−5 (1.1 ± 0.6) × 10−1 (2.4 ± 2.2) × 10−5 (1.4 ± 0.7) × 10−1 
 Bin III (6.5 ± 1.9) × 10−7 (5.1 ± 1.5) × 10−2 (1.0 ± 0.4) × 10−6 (6.3 ± 1.5) × 10−2 
 Bin IV (1.5 ± 1.4) × 10−9 (4.9 ± 3.2) × 10−3 (1.4 ± 1.1) × 10−9 (4.4 ± 1.7) × 10−3 
Wt αSN
αSNΔ2–11
KD (M)kd (s−1)KD (M)kd (s−1)
Native structured proteina     
 Bin I (4.6 ± 0.8) × 10−5 (3.4 ± 1.7) × 10−3 (5.3 ± 0.3) × 10−5 (5.4 ± 0.1) × 10−3 
 Bin II (9.6 ± 8.5) × 10−5 (1.2 ± 2.2) × 10−2 (3.5 ± 2.3) × 10−5 (1.2 ± 5.0) × 10−3 
 Bin III (5.5 ± 0.4) × 10−7 (1.9 ± 0.6) × 10−2 (7.9 ± 5.1) × 10−7 (2.2 ± 0.4) × 10−2 
 Bin IV (0.8 ± 0.8) × 10−8 (3.2 ± 3.2) × 10−3 (0.8 ± 0.8) × 10−8 (3.3 ± 3.3) × 10−3 
Gu·HCl-treated (denatured) proteina  
 Bin Ib N/D N/D N/D N/D 
 Bin II (8.5 ± 6.6) × 10−5 (1.1 ± 0.6) × 10−1 (2.4 ± 2.2) × 10−5 (1.4 ± 0.7) × 10−1 
 Bin III (6.5 ± 1.9) × 10−7 (5.1 ± 1.5) × 10−2 (1.0 ± 0.4) × 10−6 (6.3 ± 1.5) × 10−2 
 Bin IV (1.5 ± 1.4) × 10−9 (4.9 ± 3.2) × 10−3 (1.4 ± 1.1) × 10−9 (4.4 ± 1.7) × 10−3 
a

Binding kinetic parameters extracted from the analysis made in Fig. 5B, 5D, 5F, and 5H. For each bin, the value was calculated as a mean weighted by the volume of the interactions in the bin. All tabulated values are the mean of two experiments ± SD.

b

Bin 1 collects weak interactions, which were N/D for the surface with denatured protein at the applied αMI concentrations.

N/D, not detected.

The binding of αXI to Wt αSN produced a more multifaceted result than for αMI, especially because the sensorgrams were not analyzable as ensembles of 1:1 interactions. At the start of the association phase (ti), a sharp peak appeared, followed by a slow raise in the signal toward the end of the tc. The dissociation phase had a remarkably slow progression (Fig. 5I). Altogether, the αXI sensorgrams shared features similar to those published earlier for the binding to heparin (53). As in the case of the binding to heparin, analysis by a simple 1:1 model, or even as an ensemble of 1:1 interactions used for αMI above (Fig. 5B, 5D, 5F, 5H), failed to fit the experimental data. Instead, to obtain at least some quantitative information on the αXI binding of αSN, exclusively the dissociation phases for the injections of 156–1250 nM αXI were analyzed as a 1:1 interaction. Generally, the experimental data matched this approach with only small residuals (Fig. 5I). On average, based on measurements for two experiments with four αXI concentration, the kd (mean ± SD) was approximately equal to (8.66 ± 1.39) × 10−4 s−1. This almost equaled the slowest dissociation rates determined for the αMI binding of Wt αSN (Table I). When surfaces were treated with Gu·HCl, the αXI SPR signal vanished (Fig. 5K), unlike what was found for αMI. Similar to the cell adhesion experiments with the CR4/K562 cells (Fig. 4C), these findings indicated a critical role of aggregated αSN, which was further supported by the αSNΔ2–11 experiments. Compared with Wt αSN, αXI had much more limited interaction with αSNΔ2–11 (Fig. 5J). This result occurred even though the surfaces were coupled with the same amounts of protein. Gu·HCl treatment again ablated the signal (Fig. 5K, 5L). Taken together, these findings indicated that CR4 αXI strongly discriminated between monomeric and aggregated αSN, whereas CR3 αMI did not.

The interaction between αXI and sonicated fibrillar αSN was investigated using negative-stain TEM. Similar to the results of other EM studies (19, 54), the slender fibrils appeared with well-defined perimeters, often with dark rims (Fig. 5M, Supplemental Fig. 2A). When αXI was added in the presence of Mg2+, the perimeters appeared more ruffled, which reduced the rim contrast (Supplemental Fig. 2B). In some locations, there were also high-contrast αXI-like features associated with the fibrils (Fig. 5N). The dark rims were restored when the EDTA-containing buffer that prevented αXI ligand interactions was used (29) (Supplemental Fig. 2C), and the associated αXI-like features appeared in only a few positions (Fig. 5O).

The results of the experiments above suggested that aggregated αSN is a ligand for CR3 and CR4 and that CR4 has a specific role in recognition of the aggregated forms. We designed an assay for direct quantification of the CD18 integrin-mediated phagocytosis of aggregated αSN analyzing the need for CD18 integrin activation found in the experiments described above. Initially, monocytes from two donors were analyzed also with replicates (Fig. 3B) to overall assess the technical stability with regard to intra-assay variation. The experiments focused on microscopically evaluating the phagocytic process by image stream flow cytometry and quantification of clearance with regard to the size of the phagocytosed particles by use of NTA.

The preparations of Wt or fibrillar αSN were biotinylated and conjugated with streptavidin-coupled Q-dots (Fig. 6A, 6B). The cells were stained for CD14 expression to locate the membrane, DNA to locate the nucleus, and for the intracellular phagolysosomes. Quantification of uptake was performed based on images of the entire cell. Image masks were made to further distinguish fully internalized Q-dots from those associated with the membrane (Fig. 6C). The analyses were performed with cells incubated with the CD18 integrin-activating Ab KIM127 (distributions indicated with a black line) or without such activation (gray lines). The results for the differences in median Q-dot fluorescence (ΔIM) and for the Kolmogorov–Smirnov variable D (measures the maximum difference between the normalized curves) (inserts, Fig. 6D–F) indicated that CD18-integrin activation increased the uptake of fibrillar αSN only (Fig. 6E). This finding was also clear when comparing the variation between donors and in the technical repeats (Supplemental Fig. 3A–F).

FIGURE 6.

Image flow cytometry of CD18 integrin-mediated αSN phagocytosis. (A and B) In preparation for the phagocytosis experiment, either Wt (A) or fibrillar αSN (B) were biotinylated and mixed with PEG and streptavidin-coated 20-nm ZnS/CdSe Q-dots. The particles and the αSN species are drawn to scale. (C) Two representative events from the population of Q-dot-positive monocytes (events 882 and 1111) are presented with masks placed to collect Q-dot fluorescence in the entire cell or in the intracellular or membrane compartments. The masks were based on staining of the monocyte cell membrane through binding of CD14 Ab, indicated in purple. The masks are indicated in dark gray. The Q-dots are red, further highlighted with white arrowheads in the mask covering the entire Q-dot positive cell. (DF) Image-based calculations on Q-dot uptake. The distributions of Q-dot fluorescence intensities in the entire cell and in the intracellular and membrane compartments, as defined in (C), were calculated from three independent experiments with purified monocytes from two donors. A combined total of 22,000 image events for all three experiments without the addition of KIM127 Ab, and 60,000 events for all three experiments with KIM127 Ab addition, were analyzed. The distributions were compared using the differences in median levels (ΔIM), and using the Kolmogorov–Smirnov statistic D, the maximum difference between the normalized cumulative distributions (shown in inserts) of the Q-dot intensity for the KIM127 untreated and treated cells. Analyses were performed with naked Q-dots (D), or 10 μg/ml fibrillar αSN and Q-dots (E), or 10 μg/ml Wt αSN and Q-dots (F). Except for the comparison of Wt αSN-coupled Q-dots in the membrane compartment (F), all other comparisons were statistically significant (Kolmogorov–Smirnov test, p < 0.002). (GJ) Stain for lysosomes. Four representative events were selected, including two events with low (167 and 700) and two events with high (411 and 869) lysosomal staining (G). For all events described above, the lysosomal stain in the presence or absence of KIM127 Ab was compared for naked Q-dots (H), or 10 μg/ml fibrillar αSN and Q-dots (I), or 10 μg/ml Wt αSN and Q-dots (J).

FIGURE 6.

Image flow cytometry of CD18 integrin-mediated αSN phagocytosis. (A and B) In preparation for the phagocytosis experiment, either Wt (A) or fibrillar αSN (B) were biotinylated and mixed with PEG and streptavidin-coated 20-nm ZnS/CdSe Q-dots. The particles and the αSN species are drawn to scale. (C) Two representative events from the population of Q-dot-positive monocytes (events 882 and 1111) are presented with masks placed to collect Q-dot fluorescence in the entire cell or in the intracellular or membrane compartments. The masks were based on staining of the monocyte cell membrane through binding of CD14 Ab, indicated in purple. The masks are indicated in dark gray. The Q-dots are red, further highlighted with white arrowheads in the mask covering the entire Q-dot positive cell. (DF) Image-based calculations on Q-dot uptake. The distributions of Q-dot fluorescence intensities in the entire cell and in the intracellular and membrane compartments, as defined in (C), were calculated from three independent experiments with purified monocytes from two donors. A combined total of 22,000 image events for all three experiments without the addition of KIM127 Ab, and 60,000 events for all three experiments with KIM127 Ab addition, were analyzed. The distributions were compared using the differences in median levels (ΔIM), and using the Kolmogorov–Smirnov statistic D, the maximum difference between the normalized cumulative distributions (shown in inserts) of the Q-dot intensity for the KIM127 untreated and treated cells. Analyses were performed with naked Q-dots (D), or 10 μg/ml fibrillar αSN and Q-dots (E), or 10 μg/ml Wt αSN and Q-dots (F). Except for the comparison of Wt αSN-coupled Q-dots in the membrane compartment (F), all other comparisons were statistically significant (Kolmogorov–Smirnov test, p < 0.002). (GJ) Stain for lysosomes. Four representative events were selected, including two events with low (167 and 700) and two events with high (411 and 869) lysosomal staining (G). For all events described above, the lysosomal stain in the presence or absence of KIM127 Ab was compared for naked Q-dots (H), or 10 μg/ml fibrillar αSN and Q-dots (I), or 10 μg/ml Wt αSN and Q-dots (J).

Close modal

The conformation-changing KIM127 Ab strongly induced the stain for phagolysosomes independent of the provided substrates (Fig. 6G–J). All experiments found an associated increased phagolysosomal response with KIM127 Ab-induced conformational change in the CD18 integrins. An experiment using high preactivation had the same result (Fig. 6I, Supplemental Fig. 3G–L).

The experiments described above served to identify the size-selective phagocytosis of αSN via CD18-integrin. A related, but not identical, question involves the ability of these processes to clear αSN from the extracellular environment to limit inflammatory responses to this material (55). Supernatants from the phagocytosis experiments above were analyzed using laser-equipped NTA equipment to track and size-determine Q-dot/αSN conjugates based on hydrodynamic-radius measurement in the complex medium, as permitted by the nonbleaching Q-dots. The size distributions obtained from phagocytosis samples were analyzed using a particle diameter cut-off <100 nm to exclude all unconjugated Q-dot particles. With the presence of 5–10 streptavidin molecules per Q-dot, the αSN conjugation capacity enabled good size separation from the unconjugated 20-nm Q-dots (Fig. 7A, 7B). Results from the NTA analysis were presented as the raw file with the particle concentration versus size and scattering intensity (insert) and as plots with particle concentration versus size for the experiments with 10 μg/ml αSN (Fig. 7A, 7B). Similar to the analysis using image stream flow cytometry, all interdonor and interassay replicates (Fig. 3B) were used to calculate a single median distribution (Fig. 7A, 7B). Addition of KIM127 reduced the median size of the Q-dot/fibrillar αSN from 417 to 108 nm (Fig. 7A, 7B). In contrast, there was essentially no change in the size distribution of Q-dot/Wt αSN conjugate (Fig. 7A, 7B). The median values were 132 nm without KIM127 addition and 160 nm with KIM127 addition. Both values were close to median sizes of the Q-dot/fibrillar αSN left behind in the experiment with KIM127. This result also supported the finding that naked Q-dots had only a minor change in size distribution and that the change was opposite from the Q-dots/fibrillar αSN result. The same results were also found when a lower concentration (5 μg/ml) of αSN was used (Supplemental Fig. 4). Taken together, the experiments (Fig. 7A, 7B) established a separation at ∼500 nm between the easily phagocytozed αSN-conjugated particles and particles smaller than this value, which were essentially not phagocytozed with the involvement of CD18 integrin activation.

FIGURE 7.

Concentrations of Q-dots and in monocyte culture supernatants following incubation without and with KIM127 Ab. (A and B) NTA analysis of the sizes and concentrations of αSN-coupled or naked Q-dots in supernatants from monocyte cultures, either treated with KIM127 (B) or left as untreated (A). Interdonor and intra-assay replicates are described in Fig. 3B. Because of large numbers of unbound Q-dots, the curves display the particle distributions with a lower cut-off at 100 nm. From the raw files (inserts) containing information on particle concentration, size, and scatter intensity, two-dimensional plots of particle size versus concentration were made for each condition. For each particle size, the median concentration calculated from the three experiments (Fig. 3B) is indicated with a solid black line, and the SEM result in gray. For each condition (Q-dots with no αSN, Wt αSN, or fibrillar αSN), the median value (M) was calculated. Black, hatched lines separates particle sizes below and above 500 nm. (C and D) Analysis of biological variation (interdonor variation) of phagocytic clearance of αSN-coupled or naked Q-dots. Monocytes from nine donors were incubated with Q-dots with no αSN, Wt αSN, or fibrillar αSN either in the presence of KIM127 (eight donors) or absence of KIM127 (nine donors). For each donor, a two-dimensional plot of particle size versus concentration was established as in Fig. 7A, 7B for the size intervals 100–500 nm and 500–1000 nm and further used to calculate the cumulative distribution for each interval. The median cumulative curve (read in % on left axis) was produced for conditions with (black color) and without (gray color) addition of KIM127 from the eight and nine experiments, respectively. The two-dimensional plots were also used to calculate the total number of particles with GraphPad Prism’s area-under-curve function. For each type of particle and application of KIM127, the mean particle concentration (shown as bars in inserts, read on the right axis in 108 particles ml, and with color coding as for the curves) and confidence interval (error bars showing upper-half interval) was calculated. Based on the mean value, confidence interval and number of donors tested, the statistical significance were calculated in an unequal variance t test (Welch test). For all calculations made, the p values are stated in the inserts.

FIGURE 7.

Concentrations of Q-dots and in monocyte culture supernatants following incubation without and with KIM127 Ab. (A and B) NTA analysis of the sizes and concentrations of αSN-coupled or naked Q-dots in supernatants from monocyte cultures, either treated with KIM127 (B) or left as untreated (A). Interdonor and intra-assay replicates are described in Fig. 3B. Because of large numbers of unbound Q-dots, the curves display the particle distributions with a lower cut-off at 100 nm. From the raw files (inserts) containing information on particle concentration, size, and scatter intensity, two-dimensional plots of particle size versus concentration were made for each condition. For each particle size, the median concentration calculated from the three experiments (Fig. 3B) is indicated with a solid black line, and the SEM result in gray. For each condition (Q-dots with no αSN, Wt αSN, or fibrillar αSN), the median value (M) was calculated. Black, hatched lines separates particle sizes below and above 500 nm. (C and D) Analysis of biological variation (interdonor variation) of phagocytic clearance of αSN-coupled or naked Q-dots. Monocytes from nine donors were incubated with Q-dots with no αSN, Wt αSN, or fibrillar αSN either in the presence of KIM127 (eight donors) or absence of KIM127 (nine donors). For each donor, a two-dimensional plot of particle size versus concentration was established as in Fig. 7A, 7B for the size intervals 100–500 nm and 500–1000 nm and further used to calculate the cumulative distribution for each interval. The median cumulative curve (read in % on left axis) was produced for conditions with (black color) and without (gray color) addition of KIM127 from the eight and nine experiments, respectively. The two-dimensional plots were also used to calculate the total number of particles with GraphPad Prism’s area-under-curve function. For each type of particle and application of KIM127, the mean particle concentration (shown as bars in inserts, read on the right axis in 108 particles ml, and with color coding as for the curves) and confidence interval (error bars showing upper-half interval) was calculated. Based on the mean value, confidence interval and number of donors tested, the statistical significance were calculated in an unequal variance t test (Welch test). For all calculations made, the p values are stated in the inserts.

Close modal

To further test the biological validity of our findings concerning size-selective αSN phagocytosis, we used the NTA for investigating the response with monocytes from nine different donors (Fig. 7C, 7D). The size distribution of particles conjugated and incubated as for the results shown in Fig. 7A and 7B were analyzed in two bins: one ranging from 100 to 500 nm and another from 500 to 1000 nm. In each bin, the normalized cumulative size distribution was calculated for each of the nine donors, either in the presence or absence of KIM127. From these curves, median size distributions were calculated and compared for each bin and for each of the three types of particle. To make a statistical evaluation, the total number of particles in each bin and in the presence or absence of KIM127 was also calculated and compared in inserts in the panels (Fig. 7C, 7D). As already noted above, there was a clear removal from the supernatants of particles with fibrillar αSN in the size interval 500–1000 nm when KIM127 was added (Fig. 7D), but almost never in the size interval 100–500 nm (Fig. 7C). A closer inspection of the cumulative distributions revealed that the most pronounced induction of clearance by KIM127 occurred for particles with a size of ∼700 nm (Fig. 7D). The total number of particles in the 500–1000 nm bin showed a highly significant (p = 0.0024) 38% lowering when KIM127 was added to cells incubated with fibrillar αSN conjugates (Fig. 7D). As expected, for Wt αSN conjugates, the number of particles in the 500–1000 nm bin was low, only permitting a description of the full cumulative distribution in the absence of KIM127 (Fig. 7D), but again confirming the efficient clearance of the larger particles. The number of smaller particles in the 100- to 500-nm bin were not affected by KIM127 addition (Fig. 7C). Unconjugated particles behaved opposite the αSN conjugate with no evidence of efficient phagocytosis (Fig. 7C, 7D).

PD has a strong genetic component (56); 44 risk loci have been found (57). Pathway-based analysis of associated variants has provided support for immune-related genetic susceptibility to PD with, in particular, enrichment of genes implicated with leukocyte function (7, 48). Findings using PD cortical tissue indicate that microglial genes are generally upregulated, including the CD18 encoding gene, ITGB2 (35). Guided by the in vitro results, we assessed the expression of an integrin-related panel of genes using a dataset from a recent, large RNA-seq study involving cortical samples from 29 patients with PD and 44 aged-matched controls (45) (Fig. 8A). The hierarchical clustering result generally stratified samples based on disease state. Most (60%) of the PD cases were in a distinct cluster of 94% PD cases; there were also two clusters consisting of 86 and 70% controls (Fig. 8A). The PD-enriched cluster was characterized by a marked upregulation of genes in a cluster containing microglia-specific genes implicated with leukocyte-specific function in particular (Fig. 8A). These CD18 integrin-related genes included ITGB2 and ITGAL, ITGAM, and ITGAX, which encode LFA-1, CR3, and CR4, respectively. TLN1 and FERMT3, which encode talin and kindlin-3, respectively, important for CD18 integrin conformation regulation, were also included. The results were similar for C3, which upon complement activation and proteolytic regulation is converted to iC3b, a major ligand for CR3 and CR4 (58). Among the CD18 integrin α-chain genes, ITGAM and ITGAX mRNA were notably more abundant than ITGAL mRNA (Fig. 8B). To assess whether the increased mRNA level of microglial integrin-related genes in cortical tissue from PD patients was due to a general upregulation of microglial genes reflecting multiplying microglial cells (59) or monocytic invasion (60), we examined the expression of microglia-specific genes in the dataset (Fig. 9A). Although the majority of microglial genes were upregulated in PD, as reported by others (35), we found that approximately one fourth were downregulated (Fig. 9A). In the cortical tissue, there was a notable correlation of expression of genes for CR4 and for kindlin-3 (Fig. 9B). We also examined the similarity in gene expression between human peripheral monocytes and microglia isolated from healthy tissue ex vivo from 19 individual with no diagnosis of PD (49). Although the expression pattern contains several differences overall, the expression of the CD18 integrin chains and their conformational regulators was highly correlated, and, in consequence, had a ranking in expression largely similar for the two cell types, again showing a higher expression of ITGAM and, more moderately, ITGAX than ITGAL. The C3 expression was not following this pattern with a notably higher expression in microglia than in monocytes (Fig. 9C). To investigate if changes in the CD18 integrin-related expression also could be found in peripheral monocytes from PD patients, we used the recent RNA-seq data (48) on peripheral monocytes from a smaller cohort of PD patients and aged-matched controls (Fig. 8C). In this case, no striking differences in expression were observed between patients and controls.

FIGURE 8.

Expression of integrins and related molecules in cortical tissue and monocytes. (A and B) Expression in the prefrontal cortex (Brodmann area 9) in postmortem samples obtained from patients with PD (n = 29) and healthy controls (n = 44) reported in (45). (A) Heat map of row normalized mRNA expression values (TPM) for expression of genes encoding all core components of integrins and integrin inside out signaling, including all integrin α-chains (ITGAs), all integrin β-chains (ITGBs), select integrin ligands (C3, ICAM1, MADCAM-1, VCAM1, and ITGB1BP1), and intracellular regulators of CD18 integrin conformation through inside-out signaling (TLN1 and FERMT3). The genes were clustered based on their correlation among patients (horizontal clades) and cell expression (vertical clades). Tight correlation between integrin α-chains and conformational regulators is indicated using braces. The clade collecting the most patients with PD is indicated with an asterisk. (B) Normalized mRNA expression values (TPM) for the CD18 integrin genes, regulators of their conformation, and C3. (C) Normalized mRNA expression in TPM for the same gene as in (B) from peripheral monocytes obtained from early-stage PD patients (n = 10) and age-matched controls (n = 10) (48).

FIGURE 8.

Expression of integrins and related molecules in cortical tissue and monocytes. (A and B) Expression in the prefrontal cortex (Brodmann area 9) in postmortem samples obtained from patients with PD (n = 29) and healthy controls (n = 44) reported in (45). (A) Heat map of row normalized mRNA expression values (TPM) for expression of genes encoding all core components of integrins and integrin inside out signaling, including all integrin α-chains (ITGAs), all integrin β-chains (ITGBs), select integrin ligands (C3, ICAM1, MADCAM-1, VCAM1, and ITGB1BP1), and intracellular regulators of CD18 integrin conformation through inside-out signaling (TLN1 and FERMT3). The genes were clustered based on their correlation among patients (horizontal clades) and cell expression (vertical clades). Tight correlation between integrin α-chains and conformational regulators is indicated using braces. The clade collecting the most patients with PD is indicated with an asterisk. (B) Normalized mRNA expression values (TPM) for the CD18 integrin genes, regulators of their conformation, and C3. (C) Normalized mRNA expression in TPM for the same gene as in (B) from peripheral monocytes obtained from early-stage PD patients (n = 10) and age-matched controls (n = 10) (48).

Close modal
FIGURE 9.

Gene expression of microglial-specific genes in cortical tissue and correlation of expression in microglia and monocytes. (A and B) Gene expression and correlation of expression of microglia-specific genes in cortical tissuefrom data (45). (A) Altered microglial gene expression in patients with PD versus controls. The 500 most-expressed genes in microglial cells were analyzed for changes in patients with PD versus controls, as indicated by colors. (B) Correlation coefficients for gene expression of CD18 integrin loci and function-related genes. (C) Correlation between gene expression in microglial cells and peripheral monocytes from data reported in Gosselin et al. (49). Microglia were isolated from brain tissue resected for treatment of epilepsy, brain tumors, or acute ischemia in 19 individuals. Leukocyte-specific gene expressions in the microglia and monocytes were correlated with CD18 integrin genes, regulators of their conformation, and C3 indicated in red in the plot.

FIGURE 9.

Gene expression of microglial-specific genes in cortical tissue and correlation of expression in microglia and monocytes. (A and B) Gene expression and correlation of expression of microglia-specific genes in cortical tissuefrom data (45). (A) Altered microglial gene expression in patients with PD versus controls. The 500 most-expressed genes in microglial cells were analyzed for changes in patients with PD versus controls, as indicated by colors. (B) Correlation coefficients for gene expression of CD18 integrin loci and function-related genes. (C) Correlation between gene expression in microglial cells and peripheral monocytes from data reported in Gosselin et al. (49). Microglia were isolated from brain tissue resected for treatment of epilepsy, brain tumors, or acute ischemia in 19 individuals. Leukocyte-specific gene expressions in the microglia and monocytes were correlated with CD18 integrin genes, regulators of their conformation, and C3 indicated in red in the plot.

Close modal

Taken together, the results of the analysis revealed the strong linkage between CD18 integrin-related gene expression and PD. Also, the similarity in CD18 integrin-related gene expression between microglial cell and monocytes supports the use of the latter cell type as a model for CD18 integrin function in microglial cells (Fig. 10).

FIGURE 10.

Schematic overview of the relationship between CR3 and CR4 conformational activation and αSN clearance. (A) The endosomal transport in neurons of αSN (in a mixture of oligomeric or fibrillar forms) to the extracellular environment enables microglial phagocytosis of the material. The conformational regulation of CR3 and CR4 involves the adaptor proteins talin and kindling-3 under physiologic conditions. Part of the conformational regulation can be mimicked by the KIM127 Ab by stabilizing CR3 and CR4 in their active conformations. (B) If CR4 is appropriately activated, the oligomeric or fibrillar forms of αSN are phagocytosed. Once inside the microglial cells they are loaded into more phagolysosomes, which enables degradation. (C) If CR4 is not activated, the oligomeric or fibrillar forms of αSN remain in the extracellular environment.

FIGURE 10.

Schematic overview of the relationship between CR3 and CR4 conformational activation and αSN clearance. (A) The endosomal transport in neurons of αSN (in a mixture of oligomeric or fibrillar forms) to the extracellular environment enables microglial phagocytosis of the material. The conformational regulation of CR3 and CR4 involves the adaptor proteins talin and kindling-3 under physiologic conditions. Part of the conformational regulation can be mimicked by the KIM127 Ab by stabilizing CR3 and CR4 in their active conformations. (B) If CR4 is appropriately activated, the oligomeric or fibrillar forms of αSN are phagocytosed. Once inside the microglial cells they are loaded into more phagolysosomes, which enables degradation. (C) If CR4 is not activated, the oligomeric or fibrillar forms of αSN remain in the extracellular environment.

Close modal

PD is an important disease of the CNS with an urgent, unmet need for effective therapy. However, understanding of the molecular-level mechanisms associated with PD pathogenesis remains incomplete. This deficiency impedes the progress in designing appropriate pharmacological treatments. Recently, the intercellular transmission of aggregated αSN has received attention as especially relevant for disease etiology and progression (61). Microglia clears aggregated αSN from the extracellular milieu, but the molecular mechanisms responsible for this process are not well understood (9). We found that the strong binding of microglial-expressed receptor CR4 to αSN selected the fibrillary forms over the monomeric species. αSN phagocytosis was dependent on CD18-integrin conformational regulation and also activated a phagolysosomal response. Our findings are directly connected to the recently characterized structures of fibrillar or similarly-folded αSN aggregates (1720); they define a novel role for the CR4 receptor in αSN-related diseases (Fig. 10).

Study results suggest that CR3 and CR4 are functionally different with regard to ligand recognition (29, 62). This difference is also relevant for recognition of αSN, with surprising consequences for the clearance of potentially toxic aggregates. We confirmed (38) that CR3 is a receptor for αSN. Our study now also reveals that CR4 binds αSN. Expressed in monocytes and presented to a surface coated with Wt αSN, which is mostly nonfibrillar, CR4 contributed ∼16% of adhesion as judged from the use of function-blocking Ab. In experiments with a recombinant expression of CR4 in K562 cells, the adhesion was strong but attenuated by treatment with Gu⋅HCl, which was unlike the improvement in CR4 binding when albumin and fibrinogen was treated with denaturing agents as reported earlier (51, 63).These observations prompted us to investigate if aggregated αSN played a special role as ligand for CR4. We immobilized preparations of αSN in SPR flow cells to study the binding of αXI. The binding to Wt αSN was strong, with fast association and slow dissociation rates, but also with a striking feature in the early association phase, where the SPR signal peaked initially followed by a slow increase in signal toward the end of the tc. The results resemble those of a previous study of binding to heparin (53). Both fibrillar αSN and heparin share the exposure of dense negatively charged carboxylates, known to be good ligands for αXI (64). The electrically charged groups engage in a hydrated layer, which is at least temporarily distorted by the αXI binding, producing the sharp peak in SPR signal. This similarity in SPR response is consequently pointing to the densely packed carboxylates in both materials as involved in the αXI binding. By contrast, the binding of the CR3 αMI to Wt αSN, and as previously reported to heparin (53), produced a more orderly response in keeping with inability of αMI to bind well polyanions (51). Probably in consequence of the complexity of the binding scheme, we were not able to model comprehensively the binding of αXI to Wt αSN. However, an estimate of the dissociation rate at ∼10−3 s−1 seemed to account well for most of the dissociation phase, suggesting that such slowly releasing bonds constituted most of the interaction between αXI and Wt αSN. Again, this was unlike αMI, which had several interactions with considerable faster dissociation rates at ∼10−2 s−1. Removal of aggregated αSN by Gu⋅HCl treatment ablated the binding by the CR4 αXI. Compared with the Wt αSN, binding to the αSNΔ2–11 construct was much reduced; the αSNΔ2–11 construct contains smaller amounts of aggregated αSN than Wt αSN according to ThT spectrosopy (39). This type of aggregate quantification is directly linked with the formation of side chain ladders (27).

Why is CR4 a receptor selective for aggregated αSN? CR4 binds well to uninterrupted stretches of anionic moieties, such as those found in polyglutamate, osteopontin, and heparin (29). This preference is consistent with the electrostatic charge distribution on ligand binding interface of human αXI as also noted elsewhere (29, 32, 51). Other studies found these mostly positive charges as important in αXI ligand recognition (33, 34), which agrees with our data, although we did not map the specific interactions between αXI and the αSN aggregates. Although monomeric αSN contains an acidic C-terminal domain, uninterrupted stretches of negative charge in the primary structure are not abundant, at least not compared with other CR4 ligands (29). The recent structural description of fibrillar αSN indicates how these stretches, nevertheless, are formed by this material (1720). The C-terminal acidic appendages with high flexibility offer an anionic environment similar to the glutamate-coupled matrix found to efficiently pull down αXI (51). The parallel stacking of the monomeric αSN within the fibrils is even more striking. This structure creates long stretches, or ladders, of uninterrupted, anionic side chains perpendicular to the peptide backbone, with an inter-Glu-Cα distance of 4.9 Å (Fig. 1C). This distance is within the range of the 3.7-Å distance of the peptide backbone in polyglutamate and the 5-Å distance of the anionic monosaccharide units in heparin; both are confirmed ligands of CR4 (51, 53). The highlighted stretches (Fig. 1C) would be equally accessible in the double-stranded models of fibrillar αSN (1820). Because of its flexibility, the distance between Glu residues could be even lower in the acidic C-terminal domain. The TEM imaging indicated that the αXI domains unevenly decorated the perimeter of the αSN fibrils under conditions permitting such binding. Consequently, CR4 binding motifs previously found to represent endogenous DAMPs (51) are critically linked with the parallel stacking of at least some types of αSN aggregates. The strong toxicity of these oligomers was recently revealed (26), further emphasizing the necessity for rapid and selective clearance.

The CR4 binding of αSN aggregates was repeated in the downstream process of phagocytosis, although now with direct demonstration of the CD18 integrin-mediated binding to fibrillar forms. Conformational activation of CD18 integrins (i.e., CR3 and CR4) with the Ab KIM127 promoted uptake of fibrillar αSN but not Wt αSN or unconjugated Q-dots. The results of the image stream flow cytometry analysis confirmed the presence of the particles in the intracellular compartment. KIM127 Ab stimulation also induced robust formation of phagolysosomes. The conformation-regulated CD18 integrin signaling associated with formation of lysosomes is a novel aspect of the structural biology of integrin conformation; it is, of course, a logical coupling to the phagocytic function of CR3 and CR4.

The choice of monocytes as our experimental model system was made as they are primary leukocytes with a physiologic regulation of integrin ligand binding activity and a similar phagocytic capacity of microglial cells through the shared expression of both CR3 and CR4 (65). From the data presented in our study, we conclude that CD18 integrin activation is required for αSN clearance by primary leukocytes. Notwithstanding the challenges in getting access to human brain tissue with viable microglial cells, if integrin activation studies were to be made on cells extracted from resected brain tissue as used by others for genetic analyses (49), it would be difficult to ensure that their native integrin conformation remained unperturbed. Hence, the role of integrin activation would be complex to ascertain. A recent development suggests that microglia-like cells can be derived from differentiation of human monocytes in vitro, especially permitting investigations on neurologic disease-associated alleles (66). It is not clear, however, how this model functions with regard to CD18 integrins, and more characterization is required to appropriately apply it for the studies presented in our report. The uncertainties with regard to using primary microglia or microglial-like cells are in contrast to monocytes extracted from blood by a negative selection protocol, which routinely has provided a source of leukocytes suitable for studying CD18 integrin activation and phagocytosis (24, 6769). Human microglial cell lines are available through immortalization by viral transduction with oncogenes (70). As a model for studying integrin function, concerns arrive from the cellular attenuation of integrin-mediated functions by the transduced oncogenes (71), although the specific consequences still need to be elucidated for CD18 integrins. Rodent in vivo models have recently be successfully used for understanding the contribution of CR3 to human neuropathology (72), whereas, to our knowledge, no similar progress was made in the case of CR4. From our structural comparison of the αMI and αXI (Fig. 2), it seems likely that CR3 ligand binding is conserved between rodents and human, whereas the same is not expected to be the case for CR4. This supports the relevance of the cross-species studies in the case of CR3 functions, whereas it is unlikely that a direct comparison of human and rodent CR4 would be equally helpful in understanding the role of CR4 in clearing αSN aggregates. Of course, this would be true for analyzing cellular systems of rodent origin in vitro as well. Finally, several lines of evidence now relate changes in the peripheral immune system to PD as recently demonstrated for blood monocytes (73). The experimental observation that enteric αSN pathology may spread to the brain points to tissues outside the brain as possible sites for initiation of PD (74). As shown by us, RNA-seq data (48) do not suggest changes in expression of CD18 integrins or their conformational regulators in monocytes from early-stage PD patients compared with controls. This is unlike what seems to be the case for microglia in brain tissue (45). Even with the reservations in such a comparison, which come from the differences between these studies in mode of tissue extraction and study cohorts, this clearly fits an understanding of the brain as the major site of disease-related inflammatory processes in PD, at least with regard to those involving CD18 integrins. Nevertheless, CD18 integrin-mediated αSN phagocytosis by peripheral monocytes may still be speculated to be triggered as an adjunct consequence of non–PD-related inflammatory responses, including infection or autoinflammatory responses.

Among the integrins, roles for especially CR3 and CR4 in PD are supported by transcriptional analysis of samples from the CNS of patients. Microarray analysis of human brain microglia revealed PD-associated alterations in ITGB2 expression (35). Our data mining of a larger data set provided (45) was prompted by the results of the biochemical and cellular studies; this research approach enabled a guided inquiry into the PD brain transcriptome of microglia. The analysis revealed a high expression of the genes encoding the CR3 and CR4 α-chains that was further increased in PD. In contrast, LFA-1 α-chain expression was almost 10-fold less than that of the CR4 α-chain and was not strongly changed in PD. Consistently, LFA-1 has little or no role in phagocytosis, whereas this is the major function of CR3 and CR4 (29). This observation, which to our knowledge is novel, strongly supports CR3 and CR4 as parts of an αSN clearance mechanism differentially regulated in the CNS and hence as participants in PD molecular pathogenesis. The correlation in gene expression between the adaptor proteins talin and kindlin-3 versus CR3 and CR4 represents even stronger evidence. There were pairwise correlations of TLN1 with ITGAM and FERMT3 with ITGAX. This result was not expected during the design of the genetic analysis, as these proteins also associate with other integrins included in the analysis (36, 75). Nevertheless, it seems to be consistent with the role of CR3 and CR4 conformational regulation in αSN clearance revealed by this study. The genetic analyses also lend support to our use of monocytes as models of CD18 integrin-mediated functions in microglia. From broad transcriptomic analyses, it is clear that the human microglial cells share several expression patterns relevant to integrin function with human monocytes across neurodegenerative diseases (35, 50). In recently obtained data (49), we found that the expression of CD18 integrin genes, together with genes encoding the molecules regulating integrin conformation, are highly correlated between human microglia and monocytes. Although select parts of the phagocytic capabilities of microglia (i.e., those involving CD33) may be subject to transcriptional regulations different from monocytes (66), the evidence presented in our study suggests that CD18 integrin-related functions are conserved between the two cell types.

Taken together, our study now identifies human CR4 as a prominent part of αSN clearance in its aggregated form. The structural biology responsible for the aggregate recognition is, on one hand, entirely consistent with what has been found for several other CR4 ligands, and, on the other hand, apparently the consequence of a previously unappreciated evolutionary specialization of the αXI compared both to its rodent homologs and the αMI. The ligand recognition by the rodent homologs are not extensively tested. Hence, the functional differences between human and rodent CR4 remain speculative at this stage and would seem to deserve more attention. Nevertheless, with ageing as the most important risk factor for development of pathological protein aggregation (76), the associated diseases are especially challenging to long-lived humans relative to the shorter-lived rodents. In this perspective, human CR4’s ability to convey aggregate clearance is well in accordance with a needed mechanism of maintaining tissue homeostasis.

We thank Bettina W. Grumsen and Kirsten S. Petersen for excellent technical assistance. We acknowledge the kind help of the Core Facility for Integrated Microscopy, Faculty of Health and Medical Sciences, University of Copenhagen and the FACS core facility in Department of Biomedicine, Aarhus University.

This work was supported by an Aarhus University Research Foundation “NOVA” grant (AUFF-E-2015FLS-9-6) to T.V.-J. and K.J.-M. and an IDEAS Center grant to M.R.-R. K.L.B., A.E.L., and B.V. acknowledge funding from the Lundbeck Foundation Initiative BRAINSTRUC (2015-2666).

The online version of this article contains supplemental material.

Abbreviations used in this article:

CR

complement receptor

DAMP

danger-associated molecular pattern

Gu·HCl

guanidine hydrochloride

I

inserted

αM

α-chain of CR3

αMI

αM I domain

MIDAS

metal ion-dependent adhesion site

NTA

nanoparticle tracking analysis

PD

Parkinson disease

PFF

preformed fibril

Q-dot

quantum dot

RNA-seq

RNA sequencing

RU

resonance unit

αSN

α-synuclein

SPR

surface plasma resonance

tc

contact time

TEM

transmission electron microscopy

ThT

thioflavin T

TPM

transcript per million

Wt

wild-type

αX

α-chain of CR4

αXI

αX I domain.

1
Lashuel
,
H. A.
,
C. R.
Overk
,
A.
Oueslati
,
E.
Masliah
.
2013
.
The many faces of α-synuclein: from structure and toxicity to therapeutic target.
Nat. Rev. Neurosci.
14
:
38
48
.
2
Spillantini
,
M. G.
,
R. A.
Crowther
,
R.
Jakes
,
M.
Hasegawa
,
M.
Goedert
.
1998
.
alpha-Synuclein in filamentous inclusions of Lewy bodies from Parkinson’s disease and dementia with lewy bodies.
Proc. Natl. Acad. Sci. USA
95
:
6469
6473
.
3
Peelaerts
,
W.
,
L.
Bousset
,
A.
Van der Perren
,
A.
Moskalyuk
,
R.
Pulizzi
,
M.
Giugliano
,
C.
Van den Haute
,
R.
Melki
,
V.
Baekelandt
.
2015
.
α-Synuclein strains cause distinct synucleinopathies after local and systemic administration.
Nature
522
:
340
344
.
4
Peelaerts
,
W.
,
L.
Bousset
,
V.
Baekelandt
,
R.
Melki
.
2018
.
ɑ-Synuclein strains and seeding in Parkinson’s disease, incidental Lewy body disease, dementia with Lewy bodies and multiple system atrophy: similarities and differences.
Cell Tissue Res.
373
:
195
212
.
5
Lee
,
H. J.
,
J. E.
Suk
,
E. J.
Bae
,
J. H.
Lee
,
S. R.
Paik
,
S. J.
Lee
.
2008
.
Assembly-dependent endocytosis and clearance of extracellular alpha-synuclein.
Int. J. Biochem. Cell Biol.
40
:
1835
1849
.
6
Ferreira
,
S. A.
,
M.
Romero-Ramos
.
2018
.
Microglia response during Parkinson’s disease: alpha-synuclein intervention.
Front. Cell. Neurosci.
12
:
247
.
7
Holmans
,
P.
,
V.
Moskvina
,
L.
Jones
,
M.
Sharma
,
A.
Vedernikov
,
F.
Buchel
,
M.
Saad
,
J. M.
Bras
,
F.
Bettella
,
N.
Nicolaou
, et al
International Parkinson’s Disease Genomics Consortium
.
2013
.
A pathway-based analysis provides additional support for an immune-related genetic susceptibility to Parkinson’s disease. [Published erratum appears in 2014 Hum. Mol. Genet. 23: 562.]
Hum. Mol. Genet.
22
:
1039
1049
.
8
Perry
,
V. H.
2012
.
Innate inflammation in Parkinson’s disease.
Cold Spring Harb. Perspect. Med.
2
: a009373.
9
Lee
,
H. J.
,
J. E.
Suk
,
E. J.
Bae
,
S. J.
Lee
.
2008
.
Clearance and deposition of extracellular alpha-synuclein aggregates in microglia.
Biochem. Biophys. Res. Commun.
372
:
423
428
.
10
Emmanouilidou
,
E.
,
K.
Melachroinou
,
T.
Roumeliotis
,
S. D.
Garbis
,
M.
Ntzouni
,
L. H.
Margaritis
,
L.
Stefanis
,
K.
Vekrellis
.
2010
.
Cell-produced alpha-synuclein is secreted in a calcium-dependent manner by exosomes and impacts neuronal survival.
J. Neurosci.
30
:
6838
6851
.
11
Lorenzen
,
N.
,
S. B.
Nielsen
,
A. K.
Buell
,
J. D.
Kaspersen
,
P.
Arosio
,
B. S.
Vad
,
W.
Paslawski
,
G.
Christiansen
,
Z.
Valnickova-Hansen
,
M.
Andreasen
, et al
.
2014
.
The role of stable α-synuclein oligomers in the molecular events underlying amyloid formation.
J. Am. Chem. Soc.
136
:
3859
3868
.
12
Iwai
,
A.
,
E.
Masliah
,
M.
Yoshimoto
,
N.
Ge
,
L.
Flanagan
,
H. A.
de Silva
,
A.
Kittel
,
T.
Saitoh
.
1995
.
The precursor protein of non-A beta component of Alzheimer’s disease amyloid is a presynaptic protein of the central nervous system.
Neuron
14
:
467
475
.
13
Kim
,
C.
,
D.-H.
Ho
,
J.-E.
Suk
,
S.
You
,
S.
Michael
,
J.
Kang
,
S.
Joong Lee
,
E.
Masliah
,
D.
Hwang
,
H.-J.
Lee
,
S.-J.
Lee
.
2013
.
Neuron-released oligomeric α-synuclein is an endogenous agonist of TLR2 for paracrine activation of microglia.
Nat. Commun.
4
:
1562
.
14
Stefanova
,
N.
,
L.
Fellner
,
M.
Reindl
,
E.
Masliah
,
W.
Poewe
,
G. K.
Wenning
.
2011
.
Toll-like receptor 4 promotes α-synuclein clearance and survival of nigral dopaminergic neurons.
Am. J. Pathol.
179
:
954
963
.
15
Fellner
,
L.
,
R.
Irschick
,
K.
Schanda
,
M.
Reindl
,
L.
Klimaschewski
,
W.
Poewe
,
G. K.
Wenning
,
N.
Stefanova
.
2013
.
Toll-like receptor 4 is required for α-synuclein dependent activation of microglia and astroglia.
Glia
61
:
349
360
.
16
Blander
,
J. M.
,
R.
Medzhitov
.
2004
.
Regulation of phagosome maturation by signals from toll-like receptors.
Science
304
:
1014
1018
.
17
Tuttle
,
M. D.
,
G.
Comellas
,
A. J.
Nieuwkoop
,
D. J.
Covell
,
D. A.
Berthold
,
K. D.
Kloepper
,
J. M.
Courtney
,
J. K.
Kim
,
A. M.
Barclay
,
A.
Kendall
, et al
.
2016
.
Solid-state NMR structure of a pathogenic fibril of full-length human α-synuclein.
Nat. Struct. Mol. Biol.
23
:
409
415
.
18
Li
,
Y.
,
C.
Zhao
,
F.
Luo
,
Z.
Liu
,
X.
Gui
,
Z.
Luo
,
X.
Zhang
,
D.
Li
,
C.
Liu
,
X.
Li
.
2018
.
Amyloid fibril structure of α-synuclein determined by cryo-electron microscopy.
Cell Res.
28
:
897
903
.
19
Li
,
B.
,
P.
Ge
,
K. A.
Murray
,
P.
Sheth
,
M.
Zhang
,
G.
Nair
,
M. R.
Sawaya
,
W. S.
Shin
,
D. R.
Boyer
,
S.
Ye
, et al
.
2018
.
Cryo-EM of full-length α-synuclein reveals fibril polymorphs with a common structural kernel.
Nat. Commun.
9
:
3609
.
20
Guerrero-Ferreira
,
R.
,
N. M.
Taylor
,
D.
Mona
,
P.
Ringler
,
M. E.
Lauer
,
R.
Riek
,
M.
Britschgi
,
H.
Stahlberg
.
2018
.
Cryo-EM structure of alpha-synuclein fibrils.
Elife
7
: e36402.
21
Giehm
,
L.
,
D. I.
Svergun
,
D. E.
Otzen
,
B.
Vestergaard
.
2011
.
Low-resolution structure of a vesicle disrupting &alpha;-synuclein oligomer that accumulates during fibrillation.
Proc. Natl. Acad. Sci. USA
108
:
3246
3251
.
22
Pieri
,
L.
,
K.
Madiona
,
L.
Bousset
,
R.
Melki
.
2012
.
Fibrillar α-synuclein and huntingtin exon 1 assemblies are toxic to the cells.
Biophys. J.
102
:
2894
2905
.
23
Hashimoto
,
M.
,
E.
Masliah
.
1999
.
Alpha-synuclein in Lewy body disease and Alzheimer’s disease.
Brain Pathol.
9
:
707
720
.
24
Stapulionis
,
R.
,
C. L.
Oliveira
,
M. C.
Gjelstrup
,
J. S.
Pedersen
,
M. E.
Hokland
,
S. V.
Hoffmann
,
K.
Poulsen
,
C.
Jacobsen
,
T.
Vorup-Jensen
.
2008
.
Structural insight into the function of myelin basic protein as a ligand for integrin alpha M beta 2.
J. Immunol.
180
:
3946
3956
.
25
Skamris
,
T.
,
C.
Marasini
,
K. L.
Madsen
,
V.
Foderà
,
B.
Vestergaard
.
2019
.
Early stage alpha-synuclein amyloid fibrils are reservoirs of membrane-binding species.
Sci. Rep.
9
:
1733
.
26
Fusco
,
G.
,
S. W.
Chen
,
P. T. F.
Williamson
,
R.
Cascella
,
M.
Perni
,
J. A.
Jarvis
,
C.
Cecchi
,
M.
Vendruscolo
,
F.
Chiti
,
N.
Cremades
, et al
.
2017
.
Structural basis of membrane disruption and cellular toxicity by α-synuclein oligomers.
Science
358
:
1440
1443
.
27
Riek
,
R.
,
D. S.
Eisenberg
.
2016
.
The activities of amyloids from a structural perspective.
Nature
539
:
227
235
.
28
Akiyama
,
H.
,
P. L.
McGeer
.
1990
.
Brain microglia constitutively express beta-2 integrins.
J. Neuroimmunol.
30
:
81
93
.
29
Vorup-Jensen
,
T.
,
R. K.
Jensen
.
2018
.
Structural immunology of complement receptors 3 and 4.
Front. Immunol.
9
:
2716
.
30
Podolnikova
,
N. P.
,
A. V.
Podolnikov
,
T. A.
Haas
,
V. K.
Lishko
,
T. P.
Ugarova
.
2015
.
Ligand recognition specificity of leukocyte integrin αMβ2 (Mac-1, CD11b/CD18) and its functional consequences.
Biochemistry
54
:
1408
1420
.
31
Bajic
,
G.
,
L.
Yatime
,
R. B.
Sim
,
T.
Vorup-Jensen
,
G. R.
Andersen
.
2013
.
Structural insight on the recognition of surface-bound opsonins by the integrin I domain of complement receptor 3.
Proc. Natl. Acad. Sci. USA
110
:
16426
16431
.
32
Vorup-Jensen
,
T.
,
C.
Ostermeier
,
M.
Shimaoka
,
U.
Hommel
,
T. A.
Springer
.
2003
.
Structure and allosteric regulation of the alpha X beta 2 integrin I domain.
Proc. Natl. Acad. Sci. USA
100
:
1873
1878
.
33
Lee
,
J. H.
,
J.
Choi
,
S. U.
Nham
.
2007
.
Critical residues of alphaX I-domain recognizing fibrinogen central domain.
Biochem. Biophys. Res. Commun.
355
:
1058
1063
.
34
Gang
,
J.
,
J.
Choi
,
J. H.
Lee
,
S. U.
Nham
.
2007
.
Identification of critical residues for plasminogen binding by the alphaX I-domain of the beta2 integrin, alphaXbeta2.
Mol. Cells
24
:
240
246
.
35
Itoh
,
Y.
,
R. R.
Voskuhl
.
2017
.
Cell specificity dictates similarities in gene expression in multiple sclerosis, Parkinson’s disease, and Alzheimer’s disease.
PLoS One
12
: e0181349.
36
Hogg
,
N.
,
I.
Patzak
,
F.
Willenbrock
.
2011
.
The insider’s guide to leukocyte integrin signalling and function.
Nat. Rev. Immunol.
11
:
416
426
.
37
Shi
,
Q.
,
S.
Chowdhury
,
R.
Ma
,
K. X.
Le
,
S.
Hong
,
B. J.
Caldarone
,
B.
Stevens
,
C. A.
Lemere
.
2017
.
Complement C3 deficiency protects against neurodegeneration in aged plaque-rich APP/PS1 mice.
Sci. Transl. Med.
9
: eaaf6295.
38
Hou
,
L.
,
X.
Bao
,
C.
Zang
,
H.
Yang
,
F.
Sun
,
Y.
Che
,
X.
Wu
,
S.
Li
,
D.
Zhang
,
Q.
Wang
.
2018
.
Integrin CD11b mediates α-synuclein-induced activation of NADPH oxidase through a Rho-dependent pathway.
Redox Biol.
14
:
600
608
.
39
Lorenzen
,
N.
,
L.
Lemminger
,
J. N.
Pedersen
,
S. B.
Nielsen
,
D. E.
Otzen
.
2014
.
The N-terminus of α-synuclein is essential for both monomeric and oligomeric interactions with membranes.
FEBS Lett.
588
:
497
502
.
40
van Maarschalkerweerd
,
A.
,
V.
Vetri
,
A. E.
Langkilde
,
V.
Foderà
,
B.
Vestergaard
.
2014
.
Protein/lipid coaggregates are formed during α-synuclein-induced disruption of lipid bilayers.
Biomacromolecules
15
:
3643
3654
.
41
Petruzzelli
,
L.
,
J.
Luk
,
T. A.
Springer
.
1995
.
Adhesion structure subpanel 5, leukocyte integrins: CD11a, CD11b, CD11c, CD18
. In
Leucocyte Typing V: White Cell Differentiation Antigens.
S. F.
Schlossman
,
L.
Boumsell
,
W.
Gilks
,
J.
Harlan
,
T.
Kishimoto
,
T.
Morimoto
,
J.
Ritz
,
S.
Shaw
,
R.
Silverstein
,
T. A.
Springer
, eds.
Oxford University Press
,
New York
, p.
1581
.
42
Weetall
,
M.
,
R.
Hugo
,
C.
Friedman
,
S.
Maida
,
S.
West
,
S.
Wattanasin
,
R.
Bouhel
,
G.
Weitz-Schmidt
,
P.
Lake
.
2001
.
A homogeneous fluorometric assay for measuring cell adhesion to immobilized ligand using V-well microtiter plates.
Anal. Biochem.
293
:
277
287
.
43
Vorup-Jensen
,
T.
2012
.
Surface plasmon resonance biosensing in studies of the binding between β2 integrin I domains and their ligands.
Methods Mol. Biol.
757
:
55
71
.
44
Gorshkova
,
I. I.
,
J.
Svitel
,
F.
Razjouyan
,
P.
Schuck
.
2008
.
Bayesian analysis of heterogeneity in the distribution of binding properties of immobilized surface sites.
Langmuir
24
:
11577
11586
.
45
Dumitriu
,
A.
,
J.
Golji
,
A. T.
Labadorf
,
B.
Gao
,
T. G.
Beach
,
R. H.
Myers
,
K. A.
Longo
,
J. C.
Latourelle
.
2016
.
Integrative analyses of proteomics and RNA transcriptomics implicate mitochondrial processes, protein folding pathways and GWAS loci in Parkinson disease.
BMC Med. Genomics
9
:
5
.
46
Kim
,
D.
,
B.
Langmead
,
S. L.
Salzberg
.
2015
.
HISAT: a fast spliced aligner with low memory requirements.
Nat. Methods
12
:
357
360
.
47
Pertea
,
M.
,
G. M.
Pertea
,
C. M.
Antonescu
,
T. C.
Chang
,
J. T.
Mendell
,
S. L.
Salzberg
.
2015
.
StringTie enables improved reconstruction of a transcriptome from RNA-seq reads.
Nat. Biotechnol.
33
:
290
295
.
48
Schlachetzki
,
J. C. M.
,
I.
Prots
,
J.
Tao
,
H. B.
Chun
,
K.
Saijo
,
D.
Gosselin
,
B.
Winner
,
C. K.
Glass
,
J.
Winkler
.
2018
.
A monocyte gene expression signature in the early clinical course of Parkinson’s disease.
Sci. Rep.
8
:
10757
.
49
Gosselin
,
D.
,
D.
Skola
,
N. G.
Coufal
,
I. R.
Holtman
,
J. C. M.
Schlachetzki
,
E.
Sajti
,
B. N.
Jaeger
,
C.
O’Connor
,
C.
Fitzpatrick
,
M. P.
Pasillas
, et al
.
2017
.
An environment-dependent transcriptional network specifies human microglia identity.
Science
356
: eaal3222.
50
Raj
,
T.
,
K.
Rothamel
,
S.
Mostafavi
,
C.
Ye
,
M. N.
Lee
,
J. M.
Replogle
,
T.
Feng
,
M.
Lee
,
N.
Asinovski
,
I.
Frohlich
, et al
.
2014
.
Polarization of the effects of autoimmune and neurodegenerative risk alleles in leukocytes.
Science
344
:
519
523
.
51
Vorup-Jensen
,
T.
,
C. V.
Carman
,
M.
Shimaoka
,
P.
Schuck
,
J.
Svitel
,
T. A.
Springer
.
2005
.
Exposure of acidic residues as a danger signal for recognition of fibrinogen and other macromolecules by integrin alphaXbeta2.
Proc. Natl. Acad. Sci. USA
102
:
1614
1619
.
52
Svitel
,
J.
,
A.
Balbo
,
R. A.
Mariuzza
,
N. R.
Gonzales
,
P.
Schuck
.
2003
.
Combined affinity and rate constant distributions of ligand populations from experimental surface binding kinetics and equilibria.
Biophys. J.
84
:
4062
4077
.
53
Vorup-Jensen
,
T.
,
L.
Chi
,
L. C.
Gjelstrup
,
U. B.
Jensen
,
C. A.
Jewett
,
C.
Xie
,
M.
Shimaoka
,
R. J.
Linhardt
,
T. A.
Springer
.
2007
.
Binding between the integrin alphaXbeta2 (CD11c/CD18) and heparin.
J. Biol. Chem.
282
:
30869
30877
.
54
Vilar
,
M.
,
H. T.
Chou
,
T.
Lührs
,
S. K.
Maji
,
D.
Riek-Loher
,
R.
Verel
,
G.
Manning
,
H.
Stahlberg
,
R.
Riek
.
2008
.
The fold of alpha-synuclein fibrils.
Proc. Natl. Acad. Sci. USA
105
:
8637
8642
.
55
Allen Reish
,
H. E.
,
D. G.
Standaert
.
2015
.
Role of α-synuclein in inducing innate and adaptive immunity in Parkinson disease.
J. Parkinsons Dis.
5
:
1
19
.
56
Thacker
,
E. L.
,
A.
Ascherio
.
2008
.
Familial aggregation of Parkinson’s disease: a meta-analysis.
Mov. Disord.
23
:
1174
1183
.
57
Chang
,
D.
,
M. A.
Nalls
,
I. B.
Hallgrímsdóttir
,
J.
Hunkapiller
,
M.
van der Brug
,
F.
Cai
,
G. A.
Kerchner
,
G.
Ayalon
,
B.
Bingol
,
M.
Sheng
, et al
International Parkinson’s Disease Genomics Consortium
; 
23andMe Research Team
.
2017
.
A meta-analysis of genome-wide association studies identifies 17 new Parkinson’s disease risk loci.
Nat. Genet.
49
:
1511
1516
.
58
Waterhouse
,
A.
,
M.
Bertoni
,
S.
Bienert
,
G.
Studer
,
G.
Tauriello
,
R.
Gumienny
,
F. T.
Heer
,
T. A. P.
de Beer
,
C.
Rempfer
,
L.
Bordoli
, et al
.
2018
.
SWISS-MODEL: homology modelling of protein structures and complexes.
Nucleic Acids Res.
46
(
W1
):
W296
W303
.
59
Perry
,
V. H.
,
C.
Holmes
.
2014
.
Microglial priming in neurodegenerative disease.
Nat. Rev. Neurol.
10
:
217
224
.
60
Harms
,
A. S.
,
D. G.
Standaert
.
2014
.
Monocytes and Parkinson’s disease: invaders from outside?
Mov. Disord.
29
:
1242
.
61
Brundin
,
P.
,
R.
Melki
.
2017
.
Prying into the prion hypothesis for Parkinson’s disease.
J. Neurosci.
37
:
9808
9818
.
62
Erdei
,
A.
,
S.
Lukácsi
,
B.
Mácsik-Valent
,
Z.
Nagy-Baló
,
I.
Kurucz
,
Z.
Bajtay
.
2019
.
Non-identical twins: different faces of CR3 and CR4 in myeloid and lymphoid cells of mice and men.
Semin. Cell Dev. Biol.
85
:
110
121
.
63
Davis
,
G. E.
1992
.
The Mac-1 and p150,95 beta 2 integrins bind denatured proteins to mediate leukocyte cell-substrate adhesion.
Exp. Cell Res.
200
:
242
252
.
64
Kläning
,
E.
,
B.
Christensen
,
G.
Bajic
,
S. V.
Hoffmann
,
N. C.
Jones
,
M. M.
Callesen
,
G. R.
Andersen
,
E. S.
Sørensen
,
T.
Vorup-Jensen
.
2015
.
Multiple low-affinity interactions support binding of human osteopontin to integrin αXβ2.
Biochim. Biophys. Acta
1854
:
930
938
.
65
Griffiths
,
M. R.
,
P.
Gasque
,
J. W.
Neal
.
2009
.
The multiple roles of the innate immune system in the regulation of apoptosis and inflammation in the brain.
J. Neuropathol. Exp. Neurol.
68
:
217
226
.
66
Ryan
,
K. J.
,
C. C.
White
,
K.
Patel
,
J.
Xu
,
M.
Olah
,
J. M.
Replogle
,
M.
Frangieh
,
M.
Cimpean
,
P.
Winn
,
A.
McHenry
, et al
.
2017
.
A human microglia-like cellular model for assessing the effects of neurodegenerative disease gene variants.
Sci. Transl. Med.
9
: eaai7635.
67
Støy
,
S.
,
T. D.
Sandahl
,
A. L.
Hansen
,
B.
Deleuran
,
T.
Vorup-Jensen
,
H.
Vilstrup
,
T. W.
Kragstrup
.
2018
.
Decreased monocyte shedding of the migration inhibitor soluble CD18 in alcoholic hepatitis. [Published erratum appears in 2018 Clin. Transl. Gastroenterol. 9: 171.]
Clin. Transl. Gastroenterol.
9
:
160
.
68
Jensen
,
M. R.
,
G.
Bajic
,
X.
Zhang
,
A. K.
Laustsen
,
H.
Koldsø
,
K. K.
Skeby
,
B.
Schiøtt
,
G. R.
Andersen
,
T.
Vorup-Jensen
.
2016
.
Structural basis for simvastatin competitive antagonism of complement receptor 3.
J. Biol. Chem.
291
:
16963
16976
.
69
Zhang
,
X.
,
G.
Bajic
,
G. R.
Andersen
,
S. H.
Christiansen
,
T.
Vorup-Jensen
.
2016
.
The cationic peptide LL-37 binds Mac-1 (CD11b/CD18) with a low dissociation rate and promotes phagocytosis.
Biochim. Biophys. Acta
1864
:
471
478
.
70
Timmerman
,
R.
,
S. M.
Burm
,
J. J.
Bajramovic
.
2018
.
An overview of in vitro methods to study microglia.
Front. Cell. Neurosci.
12
:
242
.
71
Cheng
,
A.
,
G. S.
Bal
,
B. P.
Kennedy
,
M. L.
Tremblay
.
2001
.
Attenuation of adhesion-dependent signaling and cell spreading in transformed fibroblasts lacking protein tyrosine phosphatase-1B.
J. Biol. Chem.
276
:
25848
25855
.
72
Hong
,
S.
,
V. F.
Beja-Glasser
,
B. M.
Nfonoyim
,
A.
Frouin
,
S.
Li
,
S.
Ramakrishnan
,
K. M.
Merry
,
Q.
Shi
,
A.
Rosenthal
,
B. A.
Barres
, et al
.
2016
.
Complement and microglia mediate early synapse loss in Alzheimer mouse models.
Science
352
:
712
716
.
73
Nissen
,
S. K.
,
K.
Shrivastava
,
C.
Schulte
,
D. E.
Otzen
,
D.
Goldeck
,
D.
Berg
,
H. J.
Møller
,
W.
Maetzler
,
M.
Romero-Ramos
.
2019
.
Alterations in blood monocyte functions in Parkinson’s disease.
Mov. Disord.
34
:
1711
1721
.
74
Van Den Berge
,
N.
,
N.
Ferreira
,
H.
Gram
,
T. W.
Mikkelsen
,
A. K. O.
Alstrup
,
N.
Casadei
,
P.
Tsung-Pin
,
O.
Riess
,
J. R.
Nyengaard
,
G.
Tamgüney
, et al
.
2019
.
Evidence for bidirectional and trans-synaptic parasympathetic and sympathetic propagation of alpha-synuclein in rats.
Acta Neuropathol.
138
:
535
550
.
75
Sun
,
Z.
,
M.
Costell
,
R.
Fässler
.
2019
.
Integrin activation by talin, kindlin and mechanical forces.
Nat. Cell Biol.
21
:
25
31
.
76
Groh
,
N.
,
A.
Bühler
,
C.
Huang
,
K. W.
Li
,
P.
van Nierop
,
A. B.
Smit
,
M.
Fändrich
,
F.
Baumann
,
D. C.
David
.
2017
.
Age-dependent protein aggregation initiates amyloid-β aggregation.
Front. Aging Neurosci.
9
:
138
.
77
Tria
,
G.
,
H. D.
Mertens
,
M.
Kachala
,
D. I.
Svergun
.
2015
.
Advanced ensemble modelling of flexible macromolecules using X-ray solution scattering.
IUCrJ
2
:
207
217
.
78
Sen
,
M.
,
K.
Yuki
,
T. A.
Springer
.
2013
.
An internal ligand-bound, metastable state of a leukocyte integrin, αXβ2.
J. Cell Biol.
203
:
629
642
.
79
Lee
,
J. O.
,
P.
Rieu
,
M. A.
Arnaout
,
R.
Liddington
.
1995
.
Crystal structure of the A domain from the alpha subunit of integrin CR3 (CD11b/CD18).
Cell
80
:
631
638
.
80
Baker
,
N. A.
,
D.
Sept
,
S.
Joseph
,
M. J.
Holst
,
J. A.
McCammon
.
2001
.
Electrostatics of nanosystems: application to microtubules and the ribosome.
Proc. Natl. Acad. Sci. USA
98
:
10037
10041
.
81
San Sebastian
,
E.
,
J. M.
Mercero
,
R. H.
Stote
,
A.
Dejaegere
,
F. P.
Cossío
,
X.
Lopez
.
2006
.
On the affinity regulation of the metal-ion-dependent adhesion sites in integrins.
J. Am. Chem. Soc.
128
:
3554
3563
.
82
Zhao
,
H.
,
I. I.
Gorshkova
,
G. L.
Fu
,
P.
Schuck
.
2013
.
A comparison of binding surfaces for SPR biosensing using an antibody-antigen system and affinity distribution analysis.
Methods
59
:
328
335
.
83
Xu
,
S.
,
J.
Wang
,
J. H.
Wang
,
T. A.
Springer
.
2017
.
Distinct recognition of complement iC3b by integrins αXβ2 and αMβ2.
Proc. Natl. Acad. Sci. USA
114
:
3403
3408
.

The authors have no financial conflicts of interest.

Supplementary data