Inflammatory dysregulation of blood monocytes in Parkinson’s disease patients

All human experiments were performed in accordance with the declaration of Helsinki and approved by the Ethics Committee of Ulm University. All study participants gave informed written consent to participate in the study.

PD patients as well as healthy controls were recruited at the University and rehab hospital Ulm (RKU). Healthy controls were chosen to match the patient cohort’s characteristics. A detailed description of the patient cohort for each experiment can be found as Supplementary Tables 1–5. In general, PD patients with confounding factors affecting the immune system were excluded for all experiments. PD patients with other co-morbidities like hypertension or depression were included in the study (see Supplementary Tables 1–5). All PD patients included in the study were diagnosed using the UK PD Society Brain Bank clinical diagnostic criteria at the RKU which is a specialized center for PD. Using a standard procedure peripheral venous blood in non-fasting state was collected with a Monovette™ blood drawing system (Sarstedt, Germany) and processed within 3 h post-collection. For isolation of peripheral blood leukocytes, EDTA was used as anticoagulant. Li-Heparin monovettes (Sarstedt, Germany) were used for the collection of whole blood for flow cytometry. For determination of peripheral blood levels of cytokines, serum collection tubes (Sarstedt, Germany) were used. Patients and healthy controls were age and sex matched. For age matching, both groups were designed to have no significant difference in age when comparing average, median and range. For sex matching, differences between investigated groups were no more than 10 % in advantage of either gender.

To assess the progression of PD in patients, the modified Hoehn & Yahr staging system was used [29]. A score of 0 indicates no detectable signs of disease and was assigned to all healthy controls. Disease progression is broken down to seven stages (including 1.5 and 2.5 in the modified version of the system) ranging from mild unilateral involvement (1.0) to confinement to wheelchair or bed (5.0).

Peripheral blood mononuclear cells were isolated from venous whole blood by density centrifugation on Histopaque^® density gradient. After two subsequent washes to get rid of contaminating platelets, cells were labeled with magnetic beads coupled to monoclonal antibodies against CD14 (Miltenyi Biotec, Germany) and CD14-positive cells separated in a column placed in a strong magnetic field (Miltenyi Biotec, Germany). Subsequently, the column was removed from the magnetic field and the retracted cells eluted with MACS™ buffer (Miltenyi Biotec, Germany). For untouched monocytes isolation, non-monocytes were labeled with biotin-conjugated antibodies against CD3, CD7, CD16, CD19, CD56, CD123 and Glycophorin A, and Anti-Biotin MicroBeads (Miltenyi Biotec, Germany). The flow-through fraction represents the unlabeled monocytes. Cells were then centrifuged at 300 g for 10 min at RT and resuspended in cell culture medium (RPMI 1640, 10 % FCS (PAA, Austria), 1 % penicillin/streptomycin (PAA, Austria) or RLT-lysis buffer (Qiagen, The Netherlands) for further applications. The purity of the isolated cell population was controlled by flow cytometry analysis and immunocytochemistry.

To functionally characterize monocyte activation, monocytes were plated on uncoated culture plates (24-well, BD Biosciences, USA) at a density of 10⁶ cells/ml (5.10⁵ cells per 24-well). Monocytes were allowed to rest for 24 h prior to the LPS stimulation (1 ng/ml, [15]) to eradicate the influence of mechanical/chemical/temperature stress during the isolation procedure on experimental results. Monocytes were cultured in sterile RPMI 1640 medium supplemented with 10 % v/v fetal calf serum (FCS, PAA, Austria) and 1 % v/v penicillin–streptomycin (P/S; 1,000 U/ml, PAA, Austria). Sterile culture medium was used as a negative control for monocyte stimulation. LPS stimulation was performed in duplicates. After 24 h stimulation, the supernatant was collected and the cells were lysed with extraction lysis buffer (1 M NaCl, 0.6 % w/v sodium deoxycholate, 0.6 % v/v Igepal, 5 % v/v 1 M Tris pH 8.0, protease inhibitor cocktail (Roche, Switzerland)).

Cytokine levels in culture supernatants were quantified by ELISA (IL-6, IL-10, CCL2, IFNγ, IL-1β; all antibody pairs obtained from BioLegend, USA) or with an electrochemiluminescence immunoassay (TNFα, IFN-8; MesoScale Discovery, USA). Each standard and sample were measured by means of two replicates, average value calculated and blank absorption of matrix subtracted. A threshold of 0.2 coefficient of variance (CV) was set as a limit to consider a value significant. Absorption values of known standards were plotted against concentrations and the resulting sigmoidal curve fitted using non-weighted five-parameter logistics with ReaderFit™ (Hitachi Solutions, Japan). Cytokine release was normalized to cell number by quantification of protein concentration in cell lysate, measured by BCA assay (Thermo Scientific, USA) according to the supplier’s instructions.

Four milliliters of anticoagulated whole blood (Sarstedt, Germany) was collected in Li-Heparin monovettes from healthy volunteers and PD patients after informed consent. Osmotic lysis of red blood cells was performed with a red blood cell lysis buffer (RBC-LB: 50 mM NH₄Cl, 70 mM NaHCO₃, 0.1 % EDTA, dH₂O). After lysis and washing with FACS buffer [PBS (Life Technologies) + 10 % FCS (PAA, Austria)], cells were resuspended in RPMI + 10 % FCS + 2 % P/S and treated 24 h or 48 h at 37 °C with vehicle control (PBS) or APG101 (250 µg/ml). After incubation cells were resuspended, washed with FACS buffer and stained with anti-human antibodies specific for anti-HLA-DR (1:200; Biolegend, USA, L243, Pe-Cy7), anti-CD14 (1:200; BD, USA; M5E2, PerCP-Cy5.5), anti-CD16 (1:10; BD, 3G8, PE), anti-CD2 (1:20; eBioscience, USA RPA-2.10, FITC), anti-CD335 (1:100; AbDSerotec, UK; 9E2, FITC), anti-CD15 (1:13; Miltenyi Biotec, VIMC6, FITC) and anti-CD19 (1:100; eBioscience, USA; HIB19, FITC) for 25 min at 4 °C in the dark. After washing, the stained cells were fixed 20 min in 2 % PFA, filtered (70 μm) and analyzed by flow cytometry (LSRII, BD) using appropriate color compensation to correct for spectral overlap and autofluorescence. Therefore, following isotype controls were used at the same concentration as the primary antibodies: mouse IgG_2a, κ (Biolegend, USA MOPC-173, PE-Cy7), mouse IgG₁, κ (eBioscience, USA, P3.6.2.8.1, FITC), mouse IgG_2a, κ (BD, USA; G155-178, PerCP-Cy5.5), mouse IgG₁, κ (BD, MOPC-21, PE). Monocytes are among HLA-DR+ cells that do not express B-cell (CD19−), T-cell (CD2−), NK-cell (CD335−), or granulocyte makers (CD15−). Within this gate, CD14+ CD16− and CD14+ CD16+  monocytes, CD14− CD16+ monocytes and CD14− CD16− DCs can be separated. To exclude cell doublets, cells were gated using area vs. width signal intensity. Data are presented as mean ± SEM from four independent experiments.

For whole-transcriptome analysis of expression, monocytes were isolated from the peripheral blood of healthy controls and PD patients as described above and lysed with RLT buffer (Qiagen, The Netherlands). After isolation of total RNA (miRNeasy Mini Kit, Qiagen, The Netherlands), 200 ng total RNA was used for the generation of mRNA-focused libraries (TruSeq RNA Sample Preparation Kit v2, Set B, Illumina, USA). Clusters were generated with cBot (Illumina, USA) with the TruSeq SR Cluster Kit v3-cBot-HS (Illumina, USA). Samples were sequenced on a HiSeq 2000 sequencing system (Illumina, USA) with TruSeq SBS Kit HS-v3 (50 cycles). 52 bases were sequenced with single read and 2.0–2.5 × 107 reads per sample, 7 bases IndexRead.

For the unsupervised hierarchical clustering of subjects based on gene expression, genes with average expression among all samples <5 FPKM were excluded. Cosine correlation and complete linkage were used for clustering with Genesis [39]. Heatmaps and cluster diagrams were generated with Genesis. For pathway analysis, genes were filtered by expression level (≥5 in at least one of the two groups to be compared), log ratio (log2 ≥ 1 or ≤−1) and significance of the difference (q value, significance corrected for multiple testing by Benjamini Hochberg, q ≤ 0.05) between the compared groups. Canonical pathways and upstream regulators were predicted by an Ingenuity pathway analysis (Qiagen, The Netherlands); involved molecular functions and biological processes were identified with Genomatix genome analyser pathway system (Genomatix Software GmbH, Germany). For the identification of genes which are consistently expressed within the two groups to compare, we filtered out all genes whose expression pattern was determined by less than 25 % of the expression values. For heatmap generation, z scores were built for each value.

Total RNA was extracted using RNeasy Mini Kit (Qiagen, USA) in accordance with the manufacturer’s instructions. RNA concentration and quality were controlled using NanoDrop (ThermoScientific, Waltham, MA, USA) and 1 µg of total RNA was used to synthesize cDNA using Iscript reverse transcriptase (Bio-Rad Laboratories) containing oligo-dT primers and random primers. cDNA was then diluted ten times and PCR analysis was performed on a Bio-Rad iCycler System using iQSYBR Green Supermix. A specific standard curve was performed in parallel for each gene to assess the specificity of the products, for quantification of the respective transcripts in duplicate. PCR conditions were 3 min at 95 °C, followed by 40 cycles of 30 s at 94 °C and 10 s at 60 °C (for FASLG, 50 cycles of 20 s). The relative levels of each gene of interest were normalized to TBP, GAPDH (Fig. 5) and B2M (Fig. 2). Primers for FAS, FASLG and GAPDH were obtained from Qiagen. A list of all other primers used is provided in Supplementary Table 8.

Phagocytosis assay was performed as per supplier’s instructions with pHrodo™ Green Zymosan A BioParticles^® (Life Technologies, Carlsbad, USA) conjugate for phagocytosis. In brief, monocytes were plated on a 96-well tissue culture plate and incubated in RPMI1640 10 % FCS 1 %P/S containing fluorescently labeled zymosan particles (0.5 mg/ml) for 4 h at 37 °C, 5 % CO₂. Fluorescence was measured with Victor X3™ microplate reader (Perkin-Elmer, Waltham, USA).

Statistical analyses were carried out using GraphPad Prism version 6.02 for Windows, GraphPad Software, San Diego, CA, USA, “www.​graphpad.​com”. All values in the figures are presented as mean ± SEM unless stated otherwise. Outliers were identified by ROUT analysis with Q = 1 %. Gaussian distribution was tested with D’Agostino and Pearson omnibus normality test, Shapiro–Wilk normality test and KS normality test. Variances were compared with F test. Unpaired t test with (unequal SD) or without (equal SD) Welch’s correction was used for unpaired data and paired t test for paired data from Gaussian distribution. Mann–Whitney test (unpaired data) or Wilcoxon matched-pairs signed-rank test (paired data) was used for data from non-Gaussian distribution. All tests for significance were two tailed.

To investigate whether the composition of peripheral myeloid cells is altered in PD, we used a FACS-based analysis to characterize different monocyte subpopulations from PD patients and age-matched healthy controls. To avoid confounding factors affecting the immune system all PD patients with acute inflammatory co-morbidities were excluded from all following experiments. A detailed description of the PD patient cohort for each of the following experiments can be found as Supplementary Tables 1–5. We applied a FACS analysis strategy that was published recently [13], in which blood monocytes can be discriminated from the rest of the leukocytes based on high expression of HLA-DR and absence of B-cell (CD19−), T-cell (CD2−), NK-cell (CD335−), or granulocyte markers (CD15−). Within this cell population, classical (CD14+CD16−) and non-classical (CD14−CD16+) monocytes can be distinguished (Fig. 1a). We included the intermediate population (CD14+CD16+) to the non-classical population and refer to this combinatory population as CD16+ population according to Ziegler and Heitbrock [44]. Comparing total numbers of monocytes from PD patients and healthy controls revealed no significant difference (Fig. 1b). Interestingly, we found that in the peripheral blood of PD patients, classical monocytes (CD16−) are enriched, and concurrently, CD16+ monocytes are reduced compared to the monocyte composition from healthy donors (Fig. 1c). It has been suggested that a non-coding variant of HLA-DR A influences the expression of HLA-DR [1, 42]. To control that HLA-DR expression did not influence our FACS results, we quantified HLA-DR-positive monocytes in PD patients and healthy controls. As demonstrated in Supplementary Fig. 1, no significant difference in HLA-DR expression between the PD and control group was detected.

To investigate the transcriptome profile of peripheral monocytes in PD and to identify molecular pathways which are differentially regulated compared to healthy controls, we purified peripheral blood monocytes from PD patients (n = 8) and healthy controls (n = 9) by positive magnetic selection, isolated total RNA and subjected it to next-generation sequencing. Detected transcripts were mapped to a whole-genome library and a total of 8,862 genes were expressed over the lower threshold (average FPKM >5, Fig. 2a). Strikingly, unsupervised hierarchical clustering (complete linkage by cosine correlation) revealed similar expression patterns among PD patients and largely separated disease cases from healthy controls with only two exceptions (Fig. 2b).

A detailed analysis based on filtering genes by expression and significant difference between the disease and control groups revealed that genes which are differentially expressed in PD monocytes include transcription factors, secreted proteins and micro-RNAs (Fig. 2c, d).

Pathway (Table 1) and upstream (Table 2) analysis (IPA) demonstrated that the most significantly affected pathways are related to inflammation: granulocyte and agranulocyte adhesion, diapedesis, regulation of cytokine production, response to chemical stimulus and organic substances, response to stress and to LPS (Supplementary Table 6). Bacterial LPS, TGFβ, TNFα, PDGF and FAS were identified as most probably involved signal mediators (Table 2). In addition, we found that chemokine binding and activity, MAP-kinase signaling and cytokine signaling are dysregulated in PD monocytes (Supplementary Table 7).

In accordance with the next-generation sequencing results from the PD discovery cohort, we could confirm by qRT-PCR in a separate validation cohort of 14 PD patients and 13 controls the highly significant downregulation of six mRNAs: JUN, EGR1, DUSP1, RHOB, FOS, NFKBIZ and a tendency for downregulation for FOSB (Fig. 2e).

To explore the physiological relevance of our findings and to test the hypothesis that the function of myeloid cells is altered in PD patients, we established a human primary monocyte culture system using positive magnetic selection based on CD14 expression and subsequent culture. The purity of the monocytic fraction was higher than 95 % (Supplementary Fig. 2a–c) and all monocytic subpopulations were present in the monocytic fraction even when CD14 is expressed at low levels in non-classical monocytes (Supplementary Fig. 2d). To exclude the possibility that anti-CD14 magnetic beads impair monocyte activation or pre-activate monocytes, IL-6 release was measured after LPS stimulation using both anti-CD14 positive and negative isolation methods. No difference in monocyte activation was detected, thereby ensuring the quality of our assay system (Supplementary Fig. 2e). Monocytes from PD patients and controls were then stimulated with LPS in culture. Strikingly, we found that monocytes from PD patients behave abnormally, displaying excessive pro-inflammatory IL-1β, IL-6, IL-8 and IFNγ production compared to monocytes from age- and sex-matched healthy controls (Fig. 3a–d). Even though statistical significance was not reached (p = 0.06), a trend for increased TNFα and IL-10 secretion was also observed for PD monocytes compared to control monocytes (Fig. 3e–f). The most striking increase of cytokine release upon LPS stimulation from PD monocytes was found regarding IL-6 production (p < 0.0001) (Fig. 3g). Since most PD patients included in the study were on l-dopa treatment (see Supplementary Tables 1–5), we also tested the effect of l-dopa on IL-6 release in our monocyte cultures to rule out the possibility that l-dopa leads to an excess release of IL-6. As shown in Supplementary Fig. 3, l-dopa had no significant effect on IL-6 release of monocytes from healthy donors. Interestingly, we did not detect any excess production of cytokines by PD monocytes compared to monocytes from healthy controls without LPS stimulation.

We hypothesized that if monocyte hyperactivation upon LPS challenge is pathophysiologically relevant, the extent of monocyte hyperactivation should be reflected in clinical parameters. Indeed, we found that IL-6 secretion from LPS-stimulated monocytes derived from PD patients was strongly positively correlated with the Hoehn & Yahr (H&Y) stage of disease severity (Fig. 3h).

To further characterize the functional abnormalities of isolated PD monocytes, we assessed another main function of myeloid phagocytes—the phagocytosis of extracellular material. To this end, we applied a highly specific phagocytosis assay employing pH-sensitive-labeled yeast wall particles exhibiting a weak fluorescence at neutral pH and a dramatic increase in fluorescence at lower pH. This enables the quantification of cellular uptake and acidification of the phagocytic compartments. FACS analysis and microscopic pictures at the cellular level confirm the uptake of fluorescent particles (Supplementary Fig. 4). Using this phagocytosis assay, we found a significant reduction in phagocytic activity in the monocytes of PD patients compared to those of healthy controls (Fig. 3i).

Classical monocytes are characterized by high expression of C–C chemokine receptor type 2 (CCR2) and are recruited from the bone marrow in a chemokine-dependent manner. C–C motif ligand 2 (CCL2, also known as monocyte chemoattractant protein-1 MCP-1) is involved in monocyte recruitment from the bone marrow and spleen to the site of inflammation and is dependent on the presence of its receptor C–C chemokine receptor type 2 (CCR2) [43]. To further support the idea that PD patients exhibit altered monocyte function in the periphery and an ongoing inflammatory reaction where monocytes could be potentially recruited from the bone marrow in a CCL2-dependent manner, CCL2 measurements were performed from monocytes that were stimulated with LPS. Notably, LPS stimulation of monocytes also led to an abnormal CCL2 release by PD monocytes compared to monocytes isolated from healthy donors (Fig. 4a). Most importantly, elevated levels of CCL2 were also detected in the serum of PD patients compared to age-matched healthy controls (Fig. 4b), thereby additionally supporting the relevance of our in vitro human primary monocyte culture system. Interestingly, the gene set enrichment analysis identified CCR chemokine receptor binding as one of the top affected molecular functions as well as CCL2 as top affected signal transduction pathway (Supplementary Table 7). Matrix metalloproteinase activity, which is important for the release, movement and migration of monocytes, was also amongst the top affected pathways.

To investigate whether the observed inflammatory dysregulation is specific for PD we also included patients with other parkinsonian symptoms like patients with multiple system atrophy (MSA) and progressive supranuclear palsy (PSP) in our study. FACS-based analysis of monocytic subpopulations revealed an enrichment of classical monocytes also in PSP and MSA patients (Supplementary Fig. 5b), whereas the number of total monocytes was not significantly altered between the PD, PSP, MSA and control group (Supplementary Fig. 5a). LPS stimulation of cultured monocytes from PD patients and controls leads to an excess release of IL-6 from PD monocytes compared to controls as observed in our initial experiments. However, we also observed increased IL-6 release from PSP monocytes compared to controls. In contrast, MSA monocytes did not release excess IL-6 upon LPS stimulation (Supplementary Fig. 5c). Regarding phagocytic activity of cultured monocytes, we found only reduced phagocytic activity in PD monocytes. PSP and MSA monocytes had similar phagocytic activity than control monocytes (Supplementary Fig. 5d). We also measured CCL2 levels in the serum of PSP and MSA patients with no significant differences compared to control serum, whereas PD serum showed increased CCL2 levels (Supplementary Fig. 5e).

Another top upstream regulator identified by the gene set enrichment analysis is FAS, which has been previously linked to inflammatory processes in mouse models of inflammation [5]. In humans and in a mouse model for spinal cord lesion, upregulation of FAS ligand (FASLG) is required for the recruitment of peripheral myeloid cells to the injured tissue [31]. To determine whether the FAS/FASLG system is involved in the peripheral dysregulation of monocytes in PD, FAS and FASLG expression levels in monocytes from PD patients were measured. Strikingly, a strong upregulation of both FAS and FASLG in monocytes from PD patients was observed (Fig. 5a, b). Importantly, deletion of myeloid FASLG or neutralization of FASLG on myeloid cells was reported to improve functional recovery after CNS trauma [31]. We, therefore, asked whether neutralization of FASLG would also have therapeutic effects on the regulation of monocyte subpopulations in PD patients. To this end, leukocytes derived from erythrocyte-depleted whole blood of PD patients was treated with the FASLG antagonist APG101 [41] for 48 h in culture. Using an FACS-based approach, we found a therapeutic effect of APG101 by a 2.5-fold reduction of the ratio of classical to non-classical monocytes derived from PD patients compared to vehicle-treated cells (Fig. 5d). The same reduction of the ratio of classical to non-classical monocytes was observed for healthy controls (Supplementary Fig. 6).