Background
Parkinson’s disease (PD) is one of the most common neurodegenerative disorders, affecting ten million people worldwide, constituting 1% of the population above the age of 65 years and 5% above the age of 85 [
1]. PD is classically diagnosed with the onset of motor symptoms like rigidity, resting tremor, bradykinesia, and postural instability [
2]. These symptoms are primarily caused by the loss of dopaminergic neurons in substantia nigra pars compacta [
3‐
5]. However, non-motor symptoms (NMS) like fatigue, pain, depression, dementia, loss of smell, and sleep fragmentation are also frequently reported among PD patients. A huge phenotypic variability exists in NMS as well as in motor symptoms, and in non-motor fluctuation (NMF) after chronic levodopa therapy [
6‐
8]. Factors such as demographics, pathological changes, and genetics only partially address the symptomatic heterogeneity in PD [
9], whereas the basis for intraindividual NMS and NMF variability remains largely enigmatic.
If and how phenotypic variability relates to extent of aggregation of the pre-synaptic protein α-syn, which constitutes a major component of the Lewy bodies/neurites found as pathological hallmarks in Parkinsonian brains [
10], is also unknown. Extent of aggregation and potential spread of pathological strains of α-syn are not well correlated to clinical phenotypes of PD [
11,
12]. None the less, α-syn and its aggregated forms are prominent targets for therapy approaches trying to reduce their impact on the brain, e.g. by active and passive immunization against this highly abundant protein [
13]. However, α-syn appears to be naturally addressed by the human immune system. A recent study has shown T-cell responses to N- and C-terminal peptides of α-syn in about 40% of the studied PD patients [
14]. A plethora of studies reported the presence of autoantibodies (AAb) against α-syn in the serum of healthy subjects and PD patients [
15,
16]. Lately, it has been shown that independent of their age and sex, almost all individuals present circulating α-syn AAbs [
17]. These naturally occurring anti-α-syn AAbs have been shown to bind to both monomers and oligomers of α-syn [
18,
19]. Additionally, several reports support the notion of a leaky blood-brain barrier (BBB) in PD patients. Increased level of albumin and immunoglobulin G (IgG) in the cerebro-spinal fluid (CSF) and a reduced P-glycoprotein function in the midbrain of PD patients has been previously reported [
20,
21]. Importantly, the presence of IgGs was demonstrated in post-mortem brain tissue on neurons specifically in the midbrain region of idiopathic and familial PD patients, but not in controls [
22]. It is currently unknown if in PD these IgGs are solely derived from periphery, or might also be produced intrathecally from infiltrated B-cells as has been suggested for certain neuroinflammatory conditions [
23‐
25].
Considering a compromised BBB in PD patients, invasion of the brain by circulating α-syn AAbs is very likely. So far, α-syn AAbs were considered to act potentially neuroprotective, as they are able to modify the aggregation propensities of α-syn, and did not show any toxicity when applied to cell lines [
18,
19,
26,
27]. In the study presented here, we address the impact of α-syn AAbs present in the serum of healthy subjects and PD patients on the survival of primary rat neurons and astrocytes. We found an NMDA receptor-dependent cell death in neurons, while astrocytes in the presence of α-syn reacted to α-syn AAbs with potentially neuroprotective cytokine secretion. These findings are of relevance not only to advance our understanding of the potential role of the immune system in PD but also to recognize prospective challenges associated with α-syn immunotherapies, be they passive [
13] or active [
28].
Discussion
Autoantibodies against α-syn, which are present in the blood of almost all individuals, were initially considered potential biomarkers for PD. Recent research has shown that α-syn AAb serum levels are indeed lower in patients with neurodegenerative diseases as compared to healthy controls, but these levels do not differ between various neurodegenerative disorders, and thus cannot distinguish between e.g., PD and AD [
29]. If lower serum levels correlate with enhanced drainage of α-syn AAb into the brain through an impaired BBB remains unknown. Furthermore, serum titres of α-syn AAbs differ substantially between various cohorts of patients and controls examined thereupon [
29]. α-syn AAbs were also considered to be potentially neuroprotective, as they are able to influence the aggregation state and seeding of α-syn [
26], and were non-toxic when applied to α-syn-expressing cell lines [
27]. In pronounced contrast to these earlier studies, we now demonstrate that α-syn AAbs can induce a robust neuropathological phenotype, characterized by NMDAR-mediated calcium influx, silencing of neuronal network activity, and subsequent neurodegeneration. This neurodegenerative effect of α-syn AAbs was detected both in primary rodent neurons and in iPSC-derived human neurons.
Should this scenario hold true for the brain of PD patients, were evidently plasma-borne IgGs are present in vulnerable brain nuclei such as SNpc and STN [
22,
43], it appears plausible that α-syn AAbs that infiltrated the CNS could contribute to disease etiology. The blood-brain-barrier (BBB) is impaired in many neurodegenerative disorders including PD [
44], and the extent of its leakiness may act as a critical denominator for the extent of α-syn AAb penetration, potentially contributing to the rather variable clinical phenotype and/or disease progression seen in PD patients [
45].
The precise location where α-syn AAbs interact with α-syn is currently unknown. Within the CNS α-syn is a highly abundant protein, mostly located at pre-synaptic sites at the membrane of synaptic vesicles. However, α-syn-containing aggregates are found within the cytoplasm and neuropil, suggesting that the protein has also non-synaptic locations, at least during intracellular transport and/or degradation processes [
46]. Furthermore, α-syn is released from neurons in an activity-depending process [
47] (and see Ref 30, Fig.
3 for a detailed evaluation in cultured neurons as used in this study), suggesting that it reaches the extracellular space starting from synaptic sites. It is likely that during synaptic vesicle fusion α-syn bound to vesicle membranes is targetable by antibodies against α-syn. Binding of the antibody might then cause steric problems with vesicle recycling and the ability to release further neurotransmitters. This mechanism may explain the impact of α-syn AAbs on neuronal activity. Alternatively, or in addition, secreted α-syn might be bound by antibodies when being associated with NMDAR. Extracellularly applied α-syn in oligomeric state has been demonstrated to interact with the GluN2A subunit of NMDAR, thereby causing synaptic dysfunctions, i.e., deficits in long-term potentiation [
48‐
50], and direct activation of extra-synaptic NMDAR by α-syn oligomers has also been described [
51]. Thus, given that in our cell culture system α-syn is released from spontaneously active neurons in significant amounts [
30], it appears plausible that this secreted protein can interact with NMDAR. Targeting this receptor-bound α-syn by an antibody may then induce a conformation change resulting in increased ion permeability of NMDAR, causing calcium influx and subsequent excitotoxicity-like cell death. If this happens at synaptic and/or extra-synaptic NMDAR, for which either neuroprotective or neurodegenerative properties are postulated [
52], remains to be evaluated. It will also be important to investigate if and how these mechanisms apply in the brain, where synaptic sites are probably more shielded from the environment as compared to our two-dimensional cultures. Importantly, antibody-mediated targeting of α-syn can also induce neurotoxicity in absence of any ectopic overexpression of the human protein in rodent neurons, since an antibody directed specifically against rodent α-syn induced the same magnitude of neurotoxicity in rat neurons expressing only the endogenous rat α-syn.
Mechanistically, our data suggest that electrical activity of neurons might be important to release α-syn from synaptic sites, to serve as an interaction partner for NMDAR and thus as a mediating element between NMDAR and α-syn AAbs. However, acute electrical activity itself is not necessary to make neurons vulnerable to α-syn AABs, since neurons that were silenced by blockade of AMPA-R with NBQX or sodium channels with TTX were still susceptible to induction of neurotoxicity by α-syn AABs. This finding further strengthens the hypothesis that α-syn AABs act rather directly on NMDAR associated α-syn than on synaptic vesicle-associated α-syn.
Astrocytes are absolutely essential for proper brain function by establishing a functional BBB and by protecting and nourishing neurons [
32]. Here, native astrocytes, either in neuron-astrocyte co-culture, or in astrocyte-enriched cultures (> 99% astrocytes), did not react to α-syn AAbs or purified, commercial α-syn antibodies. However, in the presence of α-syn, either as secreted protein in neuron-astrocyte co-culture or if expressed within astrocytes, astrocytes reacted by specifically secreting the chemokine RANTES, while other cyto- and chemokines (IFNγ, TNFα, CCC2, CXCL1, IL12) were not elevated in cell culture supernatants. Previous studies have shown a similar elevation of RANTES in the serum and mid-brain of PD patients and MPTP-treated mice [
53,
54]. It remains enigmatic for the time being where and how α-syn AAbs might interact with α-syn in astrocytes and how this interaction would initiate RANTES secretion. Given that capturing of RANTES or blocking its receptors moderately enhanced the neurotoxicity of α-syn AAbs, it appears that the astrocytic reaction is rather a neuroprotective attempt. However, if and how α-syn AAbs could induce a complex interplay of neurotoxic mechanisms through NMDAR activation on neurons and neuroprotective signaling of astrocytes in human brain remains to be determined. None the less, our data suggest that the individual extent of α-syn secretion, BBB leakiness and α-syn AAbs plasma levels may significantly influence the neurodegenerative events in patients affected by synucleinopathies, adding evidence that these parameters might be important co-factors for the wide spectrum of disease manifestations seen in clinical practice.
Our work is also of importance for another field, that has gained considerable interest recently, i.e., immunotherapies targeting α-syn aggregation and propagation [
13]. Our data strongly suggest that antibodies directed against α-syn may cause remarkable side-effects if they reach the CNS via a leaky BBB, a view supported by recent work showing that mice actively immunized against a C-terminal peptide of α-syn demonstrated neurodegeneration at least under conditions of α-syn overexpression in the target tissue [
55].
Materials and methods
Serum samples
As described previously [
29], serum samples from idiopathic PD patients and age- and gender-matched healthy subjects were obtained from the Department of Neurology, University Medical Center, Göttingen. All patients underwent detailed neurological and neuropsychological examinations by experienced movement-disorder specialists. PD patients were diagnosed according to the criteria of the International Parkinson and Movement Disorder Society [
56]. All participants provided informed consent and the study protocols (Nr. 13/11/12 and 4/5/21) were pre-reviewed by the Institutional ethics committee in concordance with the Declaration of Helsinki.
Enzyme linked immuno-sorbant assay (ELISA)
The concentration of α-synuclein (α-syn) autoantibodies (AAb) in the serum was measured using ELISA as described previously [
29]. Briefly, high-binding 96-well plates coated with 20 ng/ml of α-syn were blocked with 5% bovine serum albumin (BSA) for 1 h at room temperature (RT). Serial dilutions of the pan-synuclein antibody (ab6176) were used as a standard. Standards and 1:500 diluted samples were incubated for 2 h at RT, followed by incubation with HRP-conjugated antibody and developing with TMB. The reaction was stopped with 1 M sulfuric acid and read at 450 nm. The detection range of our assay was 7-500 ng/ml with 4.5% and 12% intra- and inter-assay coefficients of variation respectively. Uncoated wells were used as negative controls.
Depletion of α-synuclein autoantibodies
Recombinant α-synuclein (α-syn) protein with 6X His-tag was synthesized as described previously [
57]. Subsequently, the α-syn protein was bound to nickel-charged nitrilotriacetic acid (Ni-NTA) magnetic beads (catalog no. 88832) in a buffer containing 10 mM Imidazole, 50 mM NaH
2PO
4 and 300 mM NaCl, pH 8.35 for 2 h with continuous rotation. Beads were then washed two times to remove unbound α-syn protein. Subsequently, α-syn AAb containing serum was incubated with α-syn-bound magnetic beads for 4–5 days with continuous rotation. Every two days, old beads were replaced with beads freshly bound to α-syn protein to allow maximum depletion of α-syn AAbs from the serum. Lastly, a strong magnetic field was applied to separate beads from the serum. This was repeated thrice with 5–8 min of incubation each time to ensure the complete removal of beads from the serum (Fig.
2A). Serum was maintained sterile and cold (4–8 °C) at all steps. As confirmed by ELISA, this method allowed up to 80–90% depletion of α-syn AAbs from the serum.
Adeno-associated viral (AAV) vectors
All vectors were generated using the AAV-6 serotype. This serotype can transduce both neurons and astrocytes. Strictly neuron- or astrocyte-specific transgene expression was achieved by using either the synapsin1 or the full length GFAP promoter [
33,
58]. AAVs were generated in transiently transfected HEK293 cells and purified by iodixanol gradient centrifugation, and heparin affinity chromatography. Following purification, FPLC eluates were dialyzed against PBS, aliquoted, and frozen at -80
oC. The titer of vector genomes (vg) was determined by qPCR. The number of transducing units (tu) was calculated based on the experimentally determined 1:30 (tu:vg) ratio.
Cell culture
Rat primary neuron-astrocyte mixed cultures were prepared from embryonic day 18 (E18) rat brain cortices as described previously [
58]. Briefly, the cortices were excised, enzymatically treated, and triturated. Cells were plated at a density of 250,000 cells per well in 24-well plates coated with 0.1 mg/ml poly-L-ornithine and 1 µg/ml laminin. Cultures were maintained in the supplemented neurobasal medium at 37 °C, 5% CO
2, and 95% humidity. The medium was replaced once during the first week of culture while transducing with α-synuclein-expressing AAV vector. The medium was not changed further to allow the accumulation of released synuclein. Cells were transduced with AAVs expressing Bcl-XL to avoid toxicity due to overexpressed α-synuclein.
The astrocyte-enriched cultures were generated as described previously [
59]. Briefly, astrocytes from the triturated cortices the of E18 rat brain were targeted with a biotinylated-anti-GLAST antibody and captured with an anti-biotin antibody bound to magnetic beads. The cell suspension mixed with antibodies was passed through a column attached to a strong magnetic field. This allowed the selection of antibody-bound astrocytes while other cell types passed in the flow-through. Astrocytes were collected by detaching the column from the magnetic field. Cells were maintained in a serum-free medium containing DMEM-F12, Penstrep, Glutamax, 1X B27 without retinoic acid, and 0.5X N2 supplement. Enriched astrocytes were used for a maximum of 5 passages.
iPSC-derived glutamatergic neurons were generated as described previously [
60]. Briefly, CT-01 hiPSC line from a healthy subject was plated on matrigel coated wells and cultured for 12 days. The neural induction medium consisted of knockout DMEM medium, 15% knockout serum replacement and 2 mM L-Glutamine that was gradually changed to Neurobasal, 1% B27 and 2 mM L-Glutamine. For neural induction, dual SMAD inhibition was applied using LDN193189 (1 µM) and SB432545 (10 µM) from day 1 to 7. FGF-2 (10 ng/ml; Peprotech; 100-18B) was added to the medium from day 3 to 12 to expand the neural progenitors. At day 13, the cells were dissociated using accutase and re-plated with rat astrocytes in wells pre-coated with poly-L-ornithine and laminin. The cells were maintained in the maturation medium until analysis. The maturation medium consisted of Neurobasal, 1% B27, 2 mM L-Glutamine, 20 ng/ml BDNF, 20 ng/ml GDNF and 10 µM DAPT.
Transductions and treatments
Rat neuron-glia co-cultures were transduced with AAV vectors expressing low levels of Bcl-XL, to prevent any neurotoxicity caused by α-synuclein expression [
30,
31], and the genetically encoded calcium sensor GCaMP6f [
30] to record neuronal activity, on day in vitro (DIV) 2 (3 × 10
7tu/well). Both transgenes were expressed from the strictly neuron-specific synapsin1 promoter. On DIV4, cells were transduced with AAV vectors expressing either the fluorophore nuclear-targeted mCherry (NmC) alone or with bi-cistronic vectors expressing human α-synuclein + NmC (2 × 10
8tu/well) (Fig.
1A). These transgenes were also expressed under the synapsin1 promoter to allow neuron-specific expression. On DIV14, cells were treated with serum containing α-synuclein autoantibodies or with commercial antibodies. For experiments with human iPSC-derived glutamatergic neurons, 1 day after plating, the cells were transduced with AAV-GCaMP6f (1 × 10
8 tu/well). Two days later, full medium change was performed and cells were transduced with neuron-specific AAV-α-synuclein + NmC (2 × 10
8tu/well). After 30 days of culture, human neurons were treated with serum containing α-synuclein autoantibodies or with commercial antibodies. Human neurons were cultured for 30 days before being used for experiments as they acquired a similar spontaneous activity at a later time-point as compared to the rat neurons.
For a similar treatment of astrocyte-enriched cultures, cultures were transduced with AAV vectors expressing enhanced green fluorescent protein (AAV-GFAP-eGFP, at 3 × 10
7tu/well) either alone or in addition with human α-synuclein, under the control of glial fibrillary acidic protein (GFAP) promoter (AAV-GFAP- α-syn, 2 × 10
8tu/well) (Supplementary Fig.
5A). Treatments with depleted serum and commercial antibodies were performed as described above. All commercial antibodies (GAPDH Ab-G8795, GFP Ab-1181446001, Synapsin1 Ab-106011, and α-syn Abs-ab138501, 32-8100 Invitrogen, and sc7011) were applied at a concentration of 1 µg/ml. The expression of NmC in neurons and eGFP in astrocytes was used to determine cell survival after treatments, in live imaging microscopy.
100 ng/ml of anti-RANTES antibody (MAB678) was applied to sequester the secreted RANTES. BIX513 hydrochloride, SB328437, and DAPTA (all at 1 µM) were applied to block RANTES receptors, CCR1, CCR3, and CCR5 respectively. Recombinant RANTES (RnD Systems) was used at a concentration of 10 µg/ml. To block NMDA receptors (NMDAR), cells were pretreated for 30 min with 100 µM AP5 or 30 µM memantine followed by serum application. 100 nM Dantrolene and 50 µM 2-APB were used to block ryanodine and IP3 receptors, thereby inhibiting the release of calcium ions from the endoplasmic reticulum to the cytoplasm. AMPA receptor (AMPA-R) and sodium channels were blocked by 30 min of pretreatment with 10 µM NBQX and 1 µM tetrodotoxin (TTX), respectively.
Measurement of cell survival and spontaneous network activity
To investigate the number of NmC-expressing cells and spontaneous network activity, imaging was performed as described previously [
30]. Briefly, cells were imaged at 37
oC and 5% CO
2 with a Zeiss 5x Fluor objective (0.25 aperture) using a Zeiss Observer Z1 microscope. The same field of view was used to acquire NmC images and to record changes in GCaMP6f fluorescence for 1 min. For analysis, nuclear mCherry (NmC) was used to locate the nuclei to mark the region of interest (ROI). The number and location of cells were obtained using a custom-made Image J macro to segment the nuclear mCherry images. Calcium influx events were identified by analyzing changes in the signal of the calcium sensor using the FluoroSNNAP software. A minimum threshold of 10% of NmC-expressing neurons undergoing a calcium influx was used to identify network bursts and the percentage of active cells. This final step was performed either manually using Microsoft Excel or using a python script. To analyze the percent of calcium-filled cells (i.e. cells that showed a non-reversible increase in GCaMP6f fluorescence), a time frame without spontaneous network activity was selected from GCaMP6f recording. ROIs from segmented NmC images were used to locate the cells. The mean intensity of each ROI was determined to account for calcium-filled cells.
The number of surviving astrocytes under different conditions was determined by counting GFAP-eGFP-positive cells. 5–6 images were taken per well using a Zeiss 10X Plan-Apochromat objective (0.45 aperture) using a Zeiss Observer Z1 microscope.
Immunostaining
Cells were washed with 1X PBS and fixed with 4% PFA for 15 min. After permeabilization with 0.25% Triton X-100 for 10 min, cells were blocked with 2% BSA and 10% NGS for 1 h to reduce the non-specific binding of antibodies. The cells were incubated overnight at 4 °C with MAP2 (ab5622) antibody. After three washes with 1X PBS, cells were incubated with an anti-rabbit secondary antibody (1:500) for 1 h. Cells were imaged with a Zeiss 20X LD Plan NeoFluar objective (0.4 aperture) using a Zeiss Observer Z1 microscope.
Immunoblot
20 µg of cell lysates were resolved on 10% SDS-Polyacrylamide gel and transferred to the nitrocellulose membrane. After blocking with 5% milk, the membranes were incubated overnight with anti-GFAP (Z0334) and anti-vinculin (V9131) antibodies. After washing, the blots were incubated with horseradish peroxidase (HRP)-coupled secondary antibodies and developed using ECL. The intensity of the bands was determined using BioRad Image Lab software. Relative GFAP expression was calculated after normalization to vinculin intensity.
To determine any potential carry-over of the α-syn protein in the depleted serum, 500 ng of recombinant α + β + γ-syn, α-syn protein before binding, flow-through after α-syn binding, α-synAAb containing serum before and after depletion were resolved on 12% SDS-Polyacrylamide gel and transferred to the PVDF membrane. Blots were fixed with 4% Paraformaldehyde (PFA) and 0.25% Glutaraldehyde, blocked with 3% BSA and probed with a pan-synuclein antibody (32-8100 Invitrogen). Species specificity of commercial, purified antibodies, MJFR (ab138501), and D37A6 (4179), used in this study was confirmed by probing them against cell lysate from rat neuron-glia co-culture, recombinant mouse α/γ-syn and recombinant human α/β/γ-syn protein. Pan synuclein antibody (ab6176) was used as control.
Bead-based multiplex assay
A capture bead-based assay was performed to simultaneously detect multiple cytokines and chemokines, namely IFN-γ (Interferon-γ), TNF-α, CCL2, CXCL1, IL-12 (Interleukin-12), and RANTES (Regulated on Activation, Normal T-cell Expressed and Secreted). Briefly, 1d post-treatment, the cell culture medium was collected and centrifuged at 2000 rpm at 4 °C to remove debris. Samples or standards were diluted in assay buffer. The mix was incubated with capture beads conjugated to cytokine- or chemokine-specific antibody for 2 h on a shaker at RT. Beads were collected by spinning at 1000 rpm for 5 min. After washing, a biotinylated detection antibody cocktail was incubated with the beads for 1 h. This forms a capture bead-analyte-detection antibody sandwich which was then incubated with streptavidin-phycoerythrin (SA-PE) for 30 min. The resulting fluorescence intensity of SA-PE is directly proportional to the amount of bound analytes. Beads were collected by spinning at 1000 rpm for 5 min, washed, and resuspended in wash buffer. Samples were read using FACS Canto, and BD Biosciences, and analyzed with FACS Diva software. Each analyte-specific population was segregated by size and internal allophycocyanin (APC) fluorescence of the capture beads. The concentration of each analyte was determined using a standard curve generated in the same assay.
Glutamate measurement
The amount of glutamate in the medium 1 min post treatment with α-syn (auto)antibodies was measured using glutamate determination kit (GLN1, Sigma Aldrich) following manufacturer’s instructions. The assay measures the reduction of NAD + to NADH. It is proportional to the amount of glutamate that is oxidized to α-Ketoglutarate and ammonium ions in the presence of Glutamic dehydrogenase (GLDH) enzyme. To perform the assay, 100 µl of Tris (0.1 M)-EDTA (0.002 M)-Hydrazine buffer (64%) was added to each well for standard and samples. Then 10 µl of 30 mM NAD and 1 µl of 100 mM ATP were added in every well. Next, the required amount of water and 1 mM L-Glutamate was added to the wells for a standard curve and 89 µl of sample was added to their respective wells. The background signal was immediately measured at 340 nm using Infinite M200Pro plate reader (Tecan). Without any delay, 2 µl of L-GLDH was added to each well. The plate was covered and incubated at 22 ºC for 50 min and the change in signal was measured again at 340 nm. All the samples were assayed in duplicates to account for pipetting variation. The amount of glutamate in the medium was determined using the standard curve. Medium spiked with glutamate was used as a positive control to establish the assay.
Statistics
Statistical analysis was performed using GraphPad Prism software. Normal distribution was assessed with D’Agostino and Pearson test. Significant differences between two groups was tested using an unpaired, two-tailed, Student’s t-test. One-way ANOVA followed by Tukey’s or Dunnet’s posthoc analysis was used to compare 3 or more groups. Correlations were tested with Pearson-rank correlation. Power analysis was performed using G*Power software (version 3.1.9.4). Achieved power was determined using a two-tailed t-test for the difference between two independent means with a posthoc analysis considering sample size for each group, effect size, and α error of probability (0.05). All bar graphs represent mean ± standard deviation. A p-value < 0.05 was regarded as significant after a power analysis with 80% stringency.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.