Alpha-synuclein fine-tunes neuronal response to pro-inflammatory cytokines

Pro-inflammatory cytokines are emerging as neuroinflammatory mediators in Parkinson’s disease (PD) due to their ability to act through neuronal cytokine receptors. Critical questions persist regarding the role of cytokines in neuronal dysfunction and their contribution to PD pathology. Specifically, the potential synergy of the hallmark PD protein alpha-synuclein (α-syn) with cytokines is of interest. We therefore investigated the direct impact of pro-inflammatory cytokines on neurons and hypothesized that α-syn pathology exacerbates cytokine-induced neuronal deficits in PD. iPSC-derived cortical neurons (CNs) from healthy controls and patients with α-syn gene locus duplication (SNCA dupl) were stimulated with IL-17A, TNF-α, IFN-γ, or a combination thereof. For rescue experiments, CNs were pre-treated with α-syn anti-oligomerisation compound NPT100-18A prior to IL-17A stimulation. Cytokine receptor expression, microtubule cytoskeleton, axonal transport and neuronal activity were assessed.


Introduction
Neuroinflammation is a prominent feature of many neurodegenerative disorders including Parkinson's disease (PD) (Tansey et al., 2022).Pro-inflammatory cytokines may be especially relevant in PD pathogenesis since they can cross the intact bloodbrain barrier and penetrate the brain tissue (Yarlagadda et al., 2009).Indeed, given the reports of elevated levels of several pro-inflammatory cytokines in the cerebrospinal fluid of PD patients (Qu et al., 2023), clearer insights into their effects are warranted.
Recently, it has been acknowledged that neurons express cytokine receptors, including those for interleukin (IL)-17A, tumor necrosis factor alpha (TNF-α), and interferon gamma (IFN-γ), and that these regulate neuronal function differently from glial cytokine receptors (Gate et al., 2021;McCoy et al., 2006;Panagiotakopoulou et al., 2020;Sommer et al., 2018).In fact, neurons in PD appear hypersensitive to proinflammatory cytokines: Cytokine receptors are upregulated in PD patients' brains, as well as in vivo and in vitro PD models (Mogi et al., 2000;Mohamed et al., 2023;Sommer et al., 2018).Consequently, cytokines are toxic to neurons in PD: Blocking soluble TNF-α attenuated dopamine neuron death in a 6-hydroxydopamine PD model, and TNF receptor knockout mice are resistant to MPTP-induced dopaminergic degeneration (McCoy et al., 2006;Sriram et al., 2002).IL-17A levels are increased in the substantia nigra of PD dementia patients (Gate et al., 2021), and IL-17A induced the death of stem cell-derived PD dopaminergic neurons, which was rescued by IL-17A or IL-17RA neutralising antibodies (Sommer et al., 2018).Together, these data point to the detrimental effects of pro-inflammatory cytokines on neurons in PD.
To unravel the interplay of α-syn and inflammation, we treated human stem cell-derived neurons from patients with monoallelic α-syn gene locus duplication (SNCA dupl) with pro-inflammatory cytokines.We demonstrate cytokine receptor expression in our neurons and show increased IL-17A receptor expression and disrupted receptor homeostasis in SNCA dupl neurons.We find that the basally altered levels of cytoskeletal components and tubulin post-translational modifications in SNCA dupl neurons are further aggravated by cytokine exposure.This pathology translates into altered axonal transport and neuronal activity.Finally, we rescue IL-17A-induced functional deficits in SNCA dupl neurons by reducing α-syn oligomerisation.Our study points to the complex interaction between α-syn and neuroinflammation and identifies IL-17A as the cytokine whose pathological effects are strongly dependent on an excess of α-syn.

Cytokine treatment
For the last 24 hours (h) of CN differentiation, CNs were treated with recombinant human cytokines: IL-17A (10 ng/ml; R&D Systems #7955-IL), TNF-α (100 ng/ml; Peprotech #AF-300-01A), IFN-γ (150 ng/ml; Peprotech #AF-300-02), or a combination of the three at 10-fold lower concentration of each.The concentrations used for singlecytokine treatments were 10-fold higher than those previously measured in the supernatant of iPSC-derived midbrain dopaminergic neurons co-cultured with autologous activated T cells (Sommer et al. 2018 and unpublished) and identified as neurotoxic in previous literature (Di Filippo et al., 2021;Meyer-Arndt et al., 2023;Pavlinek et al., 2022).For rescue experiments, CNs were treated with either IL-17A (10 ng/ml) for 24 h, or α-syn anti-oligomerisation compound NPT100-18A (10 nM) starting from day 7 of differentiation and IL-17A (10 ng/ml) for the last 24 h.NPT100-18A was kindly provided by Wolfgang Wrasidlo (UCSD).Fresh differentiation medium containing the treatment compounds or a respective vehicle (phosphate buffered saline [PBS], 0.3% bovine serum albumin in PBS, or a combination thereof) was added to the cells for the last 24 h, followed by immediate processing of cells for analysis.

Gene expression analysis
Cells were lysed using TRIzol (Invitrogen) followed by phenol-chloroform phase separation.Total RNA was isolated from the aqueous phase using RNeasy Mini Kit (QIAGEN #74104) according to manufacturer's instructions.300 ng of total RNA was reversely transcribed into cDNA using the QuantiTect Reverse Transcription Kit (QIAGEN #205311) according to the manufacturer's instructions and diluted 1:1 with ultrapure water.1-2 μl of cDNA was used per qRT-PCR reaction.qRT-PCR was performed as previously described (Sommer et al., 2016).Expression was assessed using the 2 -ΔCt analysis.Primer pairs (Sigma, Eurofins) are listed in Table 1.

Cell death analysis
To determine the rate of neurons in a late stage of apoptosis, Click-iT Plus TUNEL Assay (Thermo Fisher #C10619) was applied according to manufacturer's instructions, followed by immunocytochemistry as described above.The percentage of TUNEL/TUBB3 double-positive cells was evaluated by enhancing the contrast of TUNEL channel in Fiji to remove background, followed by a custom pipeline in CellProfiler.Neuronal nuclei were identified as described above, used to mask TUNEL channel to remove non-neuronal TUNEL-positive nuclei and remaining nuclei were counted.

Immunocytochemistry
For the analysis of tubulin post-translational modifications, cells were fixed with icecold methanol at -20°C for 10 minutes (min), followed by post-fixation with 4% paraformaldehyde (PFA) at 37°C for 15 min.Otherwise, cells were fixed with 4% PFA at room temperature (RT) for 15 min.Cells were blocked and permeabilised in blocking buffer (PBS + 10% donkey serum [Pan Biotech], 0.3% Triton X-100 [Sigma], and 5 μl/10 6 cells Fc receptor blocker [Biolegend]) for 1 h at RT, followed by incubation with primary antibodies in blocking buffer overnight at 4°C.After washing, cells were incubated with fluorophore-conjugated secondary antibodies for 1.5 h at RT, counterstained with 4′,6-diamidino-2-phenylindole (DAPI) (10 μg/ml; Sigma) and mounted on glass slides (Thermo Fisher) in Aqua-Poly/Mount (Polysciences) or imaged in microscopy-compatible plates (ibidi).For immunocytochemistry of microfluidic devices, the cells were blocked for 2 h and incubated with primary antibodies for 48 h, followed by 2 h incubation with secondary antibodies, utilising a volume gradient between the soma and axon side.Cells were imaged with Zeiss Axio Observer.Z1 equipped with Apotome 2 using 20x air-or 40x oil-immersion objective, maintaining constant exposure for each antibody panel.For a list of used antibodies, see Table 2.

Image analysis
Analysis of bulk fluorescence intensity was performed using custom pipelines in CellProfiler (v. 4.2.4,Broad Institute).First, DAPI was used to identify nuclei with diameter of 3-16 μm (typical for neurons).For more stringent neuronal selection, tubulin-βIII (TUBB3) signal was binarized using adaptive Minimum Cross-Entropy strategy and only nuclei with ≥10% overlap with the tubulin mask were retained.Since the initial segmenting did not fully capture cell bodies, TUBB3 signal was further used to identify somata by enlarging the DAPI mask using "Distance -B" method with global Minimum Cross-Entropy thresholding.The soma mask and tubulin mask were merged and the mean fluorescence intensity was measured within this mask.Mean fluorescence intensity was normalised to the number of neuronal nuclei and expressed as mean intensity per neuron.For tubulin post-translational modifications, the value was further normalised to TUBB3 intensity.5-6 fields-of-view (FOVs) containing a total of >75 neurons were analysed per cell line and condition and averaged to represent n = 1 biological replicate.
For the analysis of post-translationally modified tubulin distribution along neurites, ≥30 μm of neurite starting from soma was traced using Segmented Line tool with line width = 5 px in Fiji/ImageJ (v.2.0.9;NIH).Mean grey value along the line was measured using the "Plot Profile" function.20 neurites were traced per cell line and condition.Owing to the different lengths of traced neurites, the mean grey values were averaged only as long as ≥5 neurites reached given length.Mean grey values were subsequently averaged in 2.5 μm bins.Lengths of >50 μm were defined as distal neurite.

Co-localisation analysis
To determine the co-localisation of α-syn with tau, neurites in microfluidic devices were imaged with 63x objective using Zeiss LSM 780 confocal laser scanning microscope.Co-localisation was calculated using Manders Co-localisation Coefficient (MCC) which determines the fraction of non-zero value pixels in one channel that overlap with nonzero value pixels in a second channel (Manders et al., 1993).In contrast to Pearson's coefficient, MCC is insensitive to pixel intensities.Neurites were selected as regions of interest, background set to zero and the fluorescence signal thresholded, followed by the measurement of MCC using JACoP (Just Another Co-localization Plugin) plugin for Fiji (Bolte & Cordelières, 2006).Scrambled images (block size = 10 pixels) were used to control for false positive co-localisation.

Axonal transport
For axonal transport measurements, neurons were cultured in microfluidic chambers (Xona Microfluidics #SND450) as described previously (Prots et al. 2018).Briefly, 8×10 4 NPCs and 1×10 4 human cerebellar astrocytes were seeded on the soma side, and 2×10 4 astrocytes on axonal side in differentiation medium with 0.5% foetal bovine serum (Invitrogen).Based on the preliminary experiments demonstrating that the rate of axonal transport is supported by astrocytes in the microfluidics, the optimal experimental setup for measuring axonal transport in iPSC-based CNs includes coseeding neurons with astrocytes as previously published (Prots et al., 2018).Two days post-seeding, cells were transduced with Mito-DsRed lentivirus at MOI = 1.Live cell imaging was performed at week 2 of differentiation using Zeiss Axio Observer.Z1 with 40x oil-immersion objective.For imaging, the medium was replaced with a buffer (10 mM HEPES, 144 mM NaCl, 2.5 mM KCl, 10 mM glucose, 2.5 mM CaCl 2 , 2.5 mM MgCl 2 ) with a medium-equivalent osmolarity (310 mmol/kg).Time-lapse recording parameters were: 1 frame / 5 s, total duration 10 min/FOV.6-8 FOVs were acquired per microfluidic device.
For axonal transport analysis, kymographs of individual neurites were created in Fiji using the KymographBuilder plugin.At least 11 neurites from 2 microfluidic chambers per condition were analysed.Total number of mitochondria and the number of moving versus stationary events were calculated from kymographs.The threshold for movement was displacement ≥5 μm and traceability for ≥1/3 of total imaging time.Mitochondrial transport parameters were calculated from the single tracks of moving mitochondria.Mitochondria were classified as slow, medium, or fast moving (<0.1, 0.1-0.3, and >0.3 μm/s respectively) based on previous literature (Prots et al. 2018).

Microelectrode array (MEA) recording
For neuronal activity measurements, 5×10 4 NPCs were seeded onto 48-well CytoView MEA plate (Axion Biosystems #M768-tMEA-48B) coated with poly-D-lysine (1 mg/ml; Sigma) and laminin (25 μg/ml; Sigma).Cells were grown in medium consisting of 1:1 Neurobasal : DMEM/F-12 medium (both Gibco) and factors as described above.Half medium change was performed every three days due to the high cell density and to prevent detachment.Recordings were acquired at 37°C, 95% O 2 and 5% CO 2 using the Maestro MEA System and AxIS software (both Axion Biosystems).The signal was sampled at a rate of 12.5 kHz with a hardware frequency bandwidth of 200-5000 Hz and subsequently digitally filtered using a single-order Butterworth band-pass filter (200-5000 Hz).The adaptive threshold for spike detection was set to 5.5 standard deviations above background for each electrode.The criterium for an active electrode was ≥5 spikes/min.Activity parameters were compiled using AxIS Metrix Plotting Tool (Axion Biosystems).

Statistical analysis
Differences between two groups were analysed by two-tailed unpaired Student's t test (gene expression), Fisher's exact test or Mann-Whitney test (axonal transport), or Kolmogorov-Smirnov test (immunofluorescence).For more than two groups, data were analysed by two-way ANOVA followed by Sidak's or Dunnett's post hoc test or Fisher's LSD test (immunofluorescence, gene expression), or Kruskal-Wallis test with post-hoc Dunn's test (immunofluorescence).MEA data was analysed using repeated-measures ANOVA with Dunnett's post hoc test.p ≤ 0.05 was considered significant.Statistical analysis was performed in GraphPad Prism 8 (GraphPad Software).Error bars represent standard error of the mean (SEM).

Cytokine receptors are expressed and regulated in human cortical neurons
We first tested how human iPSC-derived cortical neurons can respond to proinflammatory cytokines.Therefore, we first probed the expression of several PDrelevant cytokine receptors in control neurons.Immunofluorescence of a representative untreated control cell line indicated that all investigated receptor subunits were detectable in TUBB3-positive neurons on a protein level (Fig. 1B).qPCR analysis validated the expression of all examined receptor subunits (il17ra, tnfr1, ifngr2, il2ra, and il2rg) in cortical neurons from healthy controls on a transcript level (Fig. 1C, vehicle).
To determine whether the cytokine receptor expression is regulated in neurons upon an inflammatory challenge, we treated the cells with three PD-related cytokines -IL-17A, TNF-α and IFN-γ (Fig. 1A).The transcript levels of the cytokine receptors were generally downregulated following a 24-h cytokine treatment compared to vehicletreated cells (Fig. 1C).TNF-α application reduced ifngr2 and il17ra levels (0.30-and 0.16-fold of vehicle-treated control; Fig. 1C), while ifngr2 levels were additionally significantly decreased upon IFN-γ treatment (0.22-fold of vehicle; Fig. 1C).Notably, TNF-α and IFN-γ stimulation resulted in a strong upregulation of β2-microglobulin (b2m), a subunit of major histocompatibility class I (MHC-I) complex classically used as a housekeeping gene (Supplementary Fig. 1A).The response to cytokines was not primarily driven by astrocytes in culture, as human cerebellar astrocytes showed a different pattern of receptor regulation following cytokine treatment -most notably, IL-17A resulted in upregulation of il17ra, tnfr1 and ifngr2 expression (Supplementary Fig. 1C).
In summary, the expression of cytokine receptors in human CNs and its homeostatic regulation by cytokines demonstrates the ability of neurons to cell-autonomously respond to inflammation.

Cytokine receptor expression and homeostasis is dysregulated in neurons with SNCA dupl
We next compared the basal levels of cytokine receptor transcripts in healthy iPSCderived CNs and CNs from patients carrying SNCA dupl.The expression of il17ra was significantly higher in CNs harbouring SNCA dupl compared to controls (7.19-fold of control, p=0.0089;Fig. 1D), mirroring the findings from iPSC-derived dopaminergic neurons from patients with sporadic PD (Sommer et al., 2018).Levels of other receptor transcripts were not significantly altered.We further compared the regulation of the receptor expression by cytokines between the two genotypes.IL-17A treatment had αsyn increase-dependent effects on ifngr2 and il2ra expression (Fig. 1E): While ignfr2 was downregulated by IL-17A in control neurons, its levels were increased in SNCA dupl neurons (0.57 vs 1.66-fold of vehicle, p=0.0451).In contrast, IL-17A-induced increase in il2ra in control was suppressed in SNCA dupl neurons (4.72 vs 1.55-fold of vehicle, p=0.0001).TNF-α-induced downregulation of il17ra in control was also nonsignificantly suppressed in SNCA dupl.Similarly to healthy neurons, TNF-α and IFN-γ application resulted in a pronounced b2m upregulation in SNCA dupl CNs (Supplementary Fig. 1B).
Thus, neurons with SNCA dupl display alterations in cytokine receptor expression as well as in the homeostatic regulation of these receptors upon cytokine exposure, indicating an altered response to inflammatory stimuli, particularly to IL-17A.

Short exposure to pro-inflammatory cytokines does not induce neuronal death
We determined the impact of a 24 h exposure to cytokines on neuronal survival by the terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL) assay.Since recent findings indicate that cytokines can have synergistic pathological effects on neurons (Meyer-Arndt et al., 2023), we included a cytokine mix consisting of IL-17A, TNF-α and IFN-γ at concentrations reported in a PD iPSC-based cellular model, better reflecting the inflammatory milieu in the disease (Sommer et al., 2018).Immunofluorescence analysis showed no change in the percentage of TUBB3 + /TUNEL + late apoptotic neurons either between control and SNCA dupl, or upon cytokine challenge (p>0.504;Fig. 2A, B).We hence deem any observed phenotypes to be early pathologies.

Overall tubulin pathology in SNCA dupl neurons is unaffected by cytokines
We tested whether pro-inflammatory cytokines induce neuronal cytoskeletal pathology by evaluating the levels of TUBB3 and tubulin post-translational modifications (PTMs): α-tubulin K40 acetylation (acTub) and α-and β-tubulin C-terminal polyglutamylation (pgTub) which are linked to neurite stabilisation and neurite severing respectively (Bodakuntla et al., 2021;Cappelletti et al., 2021).Analysis of mean fluorescence intensity (MFI) in neurons revealed a disease-specific increases in TUBB3, acTub and pgTub levels in SNCA dupl (p≤0.0001;Fig. 2D-H).However, neither of the individual cytokines had either a significant effect on the MFI in either genotype, or a differential effect between genotypes (Fig. 2D, 2F, 2H).The only exception was a trend (p=0.09)towards a differential effect of cytokine mix on TUBB3, with increased MFI in SNCA dupl and no effect in control neurons (Fig. 2D, right).Thus, while SNCA dupl neurons per se display cytoskeletal pathology with altered tubulin and tubulin PTM levels, pro-inflammatory cytokines did not further alter these parameters on a structural level.

Pro-inflammatory cytokines aggravate the altered distribution of tubulin PTMs in SNCA dupl neurons
The distribution of tubulin PTMs such as acTub and pgTub along the neuronal cytoskeleton is non-uniform and defined by the function of the given neuronal segment (Cappelletti et al., 2021;Okada et al., 2004).Given the importance of PTM localisation, we asked whether pro-inflammatory cytokines alter acTub and pgTub distribution along the neurites.To this purpose, we traced neurites starting from soma and measured the MFI / mean grey value along the trace (Fig. 2I).
The distribution of both acTub and pgTub was significantly altered in vehicle treated SNCA dupl compared to control CNs (Fig. 2J, 2L).The acTub mean grey value was elevated in the proximal neurite (≤50 μm) in SNCA dupl CNs, while the distribution in the distal neurite was comparable (Fig. 2J).The intensity of pgTub was significantly reduced along the entire neurite in SNCA dupl compared to control, with stronger effects in the distal segment (Fig. 2L).
Intriguingly, individual cytokines and cytokine mix altered the distribution of tubulin PTMs along the neurites (Fig. 2K, 2M).Notably, the cytokine effects on acTub were more prominent in SNCA dupl CNs: In control cells, the total and proximal acTub distribution was not altered by any cytokine apart from a trend for TNF-α (p=0.062).In SNCA dupl, TNF-α, IFN-γ, and cytokine mix all significantly (p≤0.029)increased acTub intensity compared to vehicle, with a trend for IL-17A (p=0.0504;Fig. 2K).In the distal neurites, which were impacted by cytokines in both genotypes, the neurons exhibited differential sensitivity based on SNCA genotype.Specifically, IL-17A and TNF-α had no significant effect in control cells but increased acTub intensity in SNCA dupl (p≤0.0031;Fig. 2K).Cytokine mix reduced acTub intensity in control distal neurites only.IFN-γ significantly altered acTub distribution in both control and SNCA dupl neurites.Strikingly, cytokines had a genotype-dependent directionality of effect: a reduction of acTub in control but an upregulation in SNCA dupl (Fig. 2K).Thus, cytokines further aggravated the already basally elevated acTub levels in SNCA dupl neurons.Of note is the lack of effect of IL-17A in control CNs.
For pgTub, we observed a similar pattern of differential sensitivity to cytokines based on α-syn dosage (Fig. 2M).TNF-α significantly disrupted pgTub intensity in both genotypes along the entire and proximal neurite (Fig. 2M).However, TNF-α effects on the distal neurite pgTub were stronger in SNCA dupl compared to control (p=0.064vs p<0.0001).IFN-γ and cytokine mix had α-syn increase-dependent effects: while cytokine mix strongly reduced pgTub intensity in all parts of the control neurites only, IFN-γ significantly increased pgTub levels in SNCA dupl only.Similarly to acTub, the directionality of cytokine effects was entirely opposite depending on α-syn levels.
In summary, pro-inflammatory cytokines and their combination alter the distribution of tubulin PTMs in neurites, with both the magnitude and the directionality of the effects being modulated by α-syn.Furthermore, IL-17A emerges as the only cytokine with entirely SNCA dupl-specific effects on acTub distribution.

Tau pathology in SNCA dupl neurons is aggravated by pro-inflammatory cytokine mix
Aside from tubulin PTMs, microtubule-associated proteins such as tau play a major role in regulating cytoskeletal stability (Barbier et al., 2019).We therefore asked whether cytokine treatment can induce tau pathology in SNCA dupl CNs.
Immunofluorescence analysis showed a disease effect for tau, specifically an increase in normalised MFI in SNCA dupl neurons (Fig. 3A, 3B left).Interestingly, while individual cytokines did not alter the tau levels, cytokine mix tendentially (p=0.051)increased tau MFI in SNCA dupl only (control: 0.939-fold of vehicle; SNCA dupl: 1.637fold of vehicle) (Fig. 3B, right).We further noticed tau clumps ("speckles") of circa 1-2 μm in diameter in SNCA dupl CNs which were specifically localised to neurites as evident when only considering tau signal within TUBB3 signal (Fig. 3A insets).There was a disease effect for tau speckles, with a significant increase in SNCA dupl (Fig. 3C) which was not further altered by cytokines.
We next wondered whether tau co-localises with α-syn in neurites and whether proinflammatory cytokines may affect such localisation.Given the SNCA dupl-specific effect of IL-17A on acTub, we focused on this cytokine.We traced neurites grown in microfluidic devices and calculated Manders Colocalization Coefficient (MCC), the fraction of signal in one channel overlapping with signal in a second channel (Fig. 3D).While the basal tau and α-syn co-localisation was not significantly different between the genotypes, IL-17A strikingly increased the fraction of tau co-localising with α-syn in SNCA dupl CNs only (p=0.0263;Fig. 3E).For the fraction of α-syn co-localising with tau, there was a trend (p=0.09) for the increased co-localisation in SNCA dupl CNs following IL-17A treatment.
Therefore, pro-inflammatory cytokine mix and IL-17A can induce or aggravate tau pathology in SNCA dupl neurons, including its co-pathology with α-syn.

Cytokine mix impairs mitochondrial motility in SNCA dupl neurons
Given the cytokine-induced alterations in the neuronal cytoskeleton, we next investigated whether this pathology results in impairments to neuronal functionality.We focused on mitochondrial axonal transport, a process critically important for neuronal health.
To address this question, iPSC-derived control and SNCA dupl neurons were transduced with Mito-DsRed, a fluorophore with a mitochondrial targeting sequence, and differentiated in microfluidic chambers which allow the unidirectional growth of neurites and thus the evaluation of retrograde and anterograde transport (Fig. 4A, 4B).Individual cytokines (IL-17A, TNF-α, and IFN-γ) did not significantly alter the proportion of motile mitochondria in either control or SNCA dupl neurons (Fig. 4C, left).We therefore speculated that a combination of the three cytokines is necessary to produce synergistic effects.Indeed, the cytokine mix significantly reduced the proportion of moving mitochondria in SNCA dupl (18.4% to 14.5%, p=0.0164), but not control CNs (Fig. 4B, 4C right).Moreover, the basal mitochondrial movement was significantly reduced in SNCA dupl compared to control (18.4% vs 23.3%, p=0.039) (Fig. 4B, 4C right).Thus, pro-inflammatory cytokines can synergise to impair neuronal physiology beyond the actions of single cytokines, particularly in the presence of increased α-syn dosage.

Cytokines differentially slow down axonal transport based on SNCA genotype
We then investigated further parameters of mitochondrial axonal transport.IL-17A had a differential effect on the direction of mitochondrial movement based on α-syn levels, increasing the proportion of retrogradely moving mitochondria in the control and decreasing it in SNCA dupl CNs (Fig. 4D, left), while no difference for other cytokines or cytokine mix was present (Fig. 4D).The average and maximum retrograde mitochondrial velocity per neurite was comparable between control and SNCA dupl neurons, and not altered by cytokines (Fig. 4E, F).However, TNF-α tendentially and significantly reduced the average and maximum anterograde velocity, respectively, in SNCA neurons only (by 0.07 [average] and 0.18 [maximum] μm/s) (Fig, 4E, F).Furthermore, there was a trend towards reduced maximum anterograde mitochondrial velocity with IL-17A application only in SNCA dupl axons (-0.14 μm/s) (Fig. 4E).
Next, mitochondria were classified into slow-(<0.1μm/s), medium-(0.1-0.3 μm/s), and fast-moving (>0.3 μm/s) as previously described (Prots et al., 2018).The retrograde velocity distribution was comparable between control and SNCA dupl neurons, but the proportion of medium-velocity anterogradely moving mitochondria was significantly lower in SNCA dupl (Fig. 4G).Intriguingly, the differential response to cytokines based on α-syn dosage was prominent.In general, cytokines slowed down mitochondrial movement.TNF-α increased the proportion of slow-moving mitochondria in both directions only in control neurons, while IL-17A had the same effect only in SNCA dupl (p=0.0387 and p=0.085;Fig. 4G).IFN-γ increased the percentage of slow retrogradely moving mitochondria in both genotypes, but of slow anterogradely moving mitochondria in control neurons only.Finally, the cytokine mix increased the fraction of slow retrogradely moving mitochondria in SNCA dupl neurons only (p=0.0387).
In conclusion, cytokines slow down mitochondrial axonal transport in human CNs, with IL-17A, TNF-α, and cytokine mix exhibiting α-syn increase-dependent effects.As observed previously with cytoskeleton, specifically IL-17A impaired mitochondrial axonal transport only in SNCA dupl neurons.

Reducing α-syn pathology rescues IL-17A-mediated deficits in neuronal functionality in SNCA dupl neurons
In our study, IL-17A consistently emerged as the only pro-inflammatory cytokine which specifically acted on SNCA dupl CNs.We therefore speculated that a detrimental synergy exists between IL-17A and α-syn pathology and that reducing α-syn oligomerisation can rescue IL-17A-induced functional deficits in SNCA dupl neurons.To this end, we treated the neurons with NPT100-18A (NPT), a de novo compound targeting the C-terminal of α-syn which disrupts its dimerization and further oligomerisation (Wrasidlo et al., 2016).NPT was applied from day 7 of differentiation to avoid treating neural precursor cells, thus yielding better translatability.IL-17A was added for the last 24 h as described above.
We first determined neuronal function using mitochondrial axonal transport as described (Fig. 5A).The application of NPT100-18A had no influence on the incidence of movement, demonstrating that the compound does not affect the basal neuronal functionality (Fig. 5B).When categorising mitochondria based on velocity, we observed no effect of IL-17A or IL-17A + NPT on control neurons.The proportion of middlevelocity retrograde mitochondria was reduced by IL-17A in SNCA dupl neurons (p=0.0437;Fig. 5C), while the pre-treatment with NPT rescued the IL-17A-induced mitochondrial slowing in these cells, with no significant difference between vehicle-treated and IL-17A+NPT-treated neurons (p=0.3931;Fig. 5C left).The nonsignificant increase in the proportion of middle-velocity anterograde mitochondria in SNCA dupl induced by IL-17A was likewise rescued by NPT treatment (Fig. 5C right).
To probe neuronal functionality in depth, we utilised the microelectrode array (MEA) system to measure neuronal activity at baseline and at multiple timepoints following IL-17A treatment (Fig. 5D).The weighted firing rate was not different between control and SNCA dupl neurons (Supplementary Fig. 2A) and IL-17A with or without NPT did not alter weighted firing rate in either genotype (Supplementary Fig. 2B).However, the number of active electrodes, indicating whether neurons are electrically active, was tendentially lower in SNCA dupl than in control CNs.Intriguingly, NPT significantly increased neuronal activity of SNCA dupl neurons while not affecting the control neuron activity (Fig. 5E).IL-17A resulted in a temporary dip in control neuron activity at 12 h post treatment, which could not be rescued by NPT (p≤0.0379;Fig. 5F top).While IL-17A did not cause a further reduction in active electrodes in SNCA dupl neurons at any time point compared to baseline (likely due to the basally lower activity), astonishingly, the pre-treatment with NPT preserved the baseline neuronal activity in these cells (baseline vs 12 h: p=0.1747;Fig. 5F bottom).This effect could be observed by comparing the % change in active electrodes in each well treated with IL-17A in the presence of NPT to its own baseline -at 12 h, there was a -50% median change in active electrodes in control, but only -13.4% in SNCA dupl (Fig. 5G).
In conclusion, reducing α-syn pathology rescues the functional deficits induced by the pro-inflammatory cytokine IL-17A in neurons with SNCA dupl.

Discussion
In our study, we took advantage of a human iPSC-derived model to investigate the damage dealt by pro-inflammatory cytokines to neurons in the context of neurodegeneration.Our data reveal the multifaceted, cytokine-specific alterations of neuronal cytoskeleton, axonal transport, and electrical activity.Strikingly, we describe for the first time that α-syn fine-tunes the neuronal response to pro-inflammatory cytokines by sensitising neurons to cytokine-induced structural and functional pathology, especially to IL-17A-mediated injury.
Multiple lines of evidence now highlight the critical role of inflammation in PD pathogenesis, as seen by the positive correlation between inflammatory bowel disease (IBD) and the risk of PD, the reduced PD risk in IBD patients using anti-TNF therapies, and the early T cell responses to α-syn peptides in PD (Kang et al., 2022;Lindestam Arlehamn et al., 2020;Peter et al., 2018).Importantly, a myriad of cell types contributes to this pro-inflammatory milieu, ranging from astrocytes to peripheral immune cells (Tansey et al., 2022).Thus, it may be more beneficial to target the ubiquitous inflammatory mediators such as cytokines rather than individual cell types, evidenced by the successful pre-clinical anti-TNF immunotherapy trials (Barnum et al., 2014;McCoy et al., 2006).We therefore investigated neuronal response to three proinflammatory cytokines, IL-17A, TNF-α and IFN-γ, all of which are produced by multiple cell types and whose increased serum levels correlate with disease severity in PD patients (Green et al., 2019;Reale et al., 2009;Williams-Gray et al., 2016).
We confirmed the expression of a range of cytokine receptors in human neurons which has been observed in previous studies (Gate et al., 2021;Meyer-Arndt et al., 2023;Sommer et al., 2016).Importantly, for the first time we demonstrate the cytokinemediated regulation of these receptors, indicating that neurons can not only directly react but also adapt to immune signals.In fact, many pro-inflammatory cytokines modulate neuronal function: IL-17A has been shown to control anxiety via IL17-RA in cortical neurons, while IFN-γ supports neuronal connectivity though neuronally expressed receptors and can promote neurite outgrowth (Alves de Lima et al., 2020;Filiano et al., 2016;Wong et al., 2004).Therefore, the cytokine-induced changes observed in our system may represent a homeostatic response, particularly in healthy neurons.However, at high concentrations, pathological effects of cytokines become apparent -the reported levels of IL-17A in the healthy brain range between 0-100 pg/ml, and 10 ng/ml IL-17A blocked long-term potentiation while 2 ng/ml did not (Alves de Lima et al., 2020;Di Filippo et al., 2021;Sommer et al., 2018;Wang et al., 2014).Similarly, physiological levels of IFN-γ (20 pg/ml) potentiated the inhibitory currents maintaining neural connectivity, while concentrations in the ng/ml range disrupted neuronal differentiation and synaptic protein expression (Filiano et al., 2016;Pavlinek et al., 2022;Warre-Cornish et al., 2020).Hence, the cytokine levels used in our study mimic a pro-inflammatory environment.
We observed an increased expression of IL-17RA in neurons harbouring SNCA locus duplication.These findings are in line with the increased IL-17RA in human-derived dopaminergic neurons from sporadic PD patients and in a rotenone PD model, indicating that a dysregulated IL-17A response may be a crucial feature of PD (Mohamed et al., 2023;Sommer et al., 2018).We also show that IL-17RA upregulation has consequences for the homeostasis of other cytokine receptors such as IFNGR2.IFNGR2 transcription may be regulated e.g. by IL-17A-inducible transcription factors p65 / C/EBPα, the binding sequences for which exist in IFNGR2 promoter / enhancer (Lim et al., 2007;Van Oevelen et al., 2015).This data suggests a generalised dysregulation of immune response in neurons with SNCA duplication.
In our study, neurons with increased α-syn dosage presented with several altered microtubule parameters.Cytoskeletal abnormalities are a notable component of PD pathology (Calogero et al., 2019;Kouroupi et al., 2017;Mazzetti et al., 2023).Here, we demonstrate for the first time that pro-inflammatory cytokines regulate the neuronal cytoskeleton by altering the distribution of tubulin PTMs and tau-α-syn co-localisation in neurites, elucidating a novel mechanism through which inflammation contributes to cytoskeletal pathology, which may trigger neurodegeneration.Two recent publications in non-neuronal cells have shown that TNF-α indirectly upregulates tubulin acetylation by activating TAK1 kinase which phosphorylates and activates α-tubulin acetyltransferase 1 (αTAT1) (Ortiz et al., 2023;Yoshimoto et al., 2021).In line with these findings, tubulin acetylation was increased in control neurites following TNF-α treatment in our study.Investigations into the effects of IL-17A and IFN-γ are lacking, although one study suggested an increase in acetylated tubulin in IFN-γ-treated macrophages, contrary to our observations in neurons (Khandani et al., 2007).Hence, the mechanisms through which cytokines regulate tubulin PTMs are likely cell typespecific and require further investigation.A plausible pathway involves the activation of kinases which modify the activity of tubulin-modifying enzymes -for instance, IL-17A, TNF-α, and IFN-γ all regulate the levels of PAK1 which has been shown to inhibit αTAT1 (Ahn et al., 2016;Ridzuan et al., 2023;Van Dijk et al., 2020;Zhou et al., 2009).
Astonishingly, the cytoskeletal response to cytokines was dependent on α-syn in two ways: first, SNCA locus duplication resulted in an increased sensitivity to cytokines; second, the direction of neuronal response was completely inversed based on SNCA genotype.Although most cytokine receptor levels were not altered in SNCA dupl neurons, the heightened response magnitude in these cells points to synergistic effects of α-syn and cytokines that may occur due to common downstream pathways.Intriguingly, α-syn signalling through Toll-like receptors upregulates TAK1, αTAT1 activator (Du et al., 2018;Quan et al., 2023;Yi et al., 2022).Notably, the direction of response in SNCA dupl neurons was such that the basal pathologies were enhanced by cytokine application, supporting the hypothesis that α-syn and cytokines act in concert to produce cytoskeletal damage.
Our analysis revealed the direct functional impact of pro-inflammatory cytokines on mitochondrial axonal transport in an α-syn increase-dependent manner.In line with the mild cytoskeletal alterations in control neurons, the cytokine effects on mitochondrial velocity in these cells were limited.The most prominent phenotype was the bidirectional reduction in fast-moving mitochondria by TNF-α, corresponding to the fact that TNF-α was the only cytokine altering both tubulin acetylation and polyglutamylation in healthy neurons.TNF-α-mediated mitochondrial slowing was reported in hippocampal neurons and airway smooth muscle cells (Delmotte et al., 2017;Stagi et al., 2006).Notably in neurons, mitochondrial movement was strongly reduced after 20 minutes but recovered within 6 h (Stagi et al., 2006).We likewise did not observe changes in motility after 24 h, indicating that cytokine effects in healthy neurons are temporary, at least during a short-term inflammation.Intriguingly, the cytoskeletal alterations induced by cytokine mix did not translate into functional deficits in healthy cells, indicating their resilience to a low-grade transient inflammation.While synergistic effect of a combinatorial cytokine treatment on neurite outgrowth and integrity has been reported (Matelski et al., 2020;Meyer-Arndt et al., 2023), we did not find prominent cytokine mix effects in healthy cortical neurons using functional readouts.In contrast, SNCA dupl neurons were uniquely vulnerable to the IL-17A/TNFα/IFN-γ mix-mediated reduction in mitochondrial motility, which was already basally reduced in these cells, replicating our previous findings (Prots et al., 2018).An efficient axonal transport is intimately linked to the cytoskeletal health: excessive tubulin polyglutamylation has been shown to perturb mitochondrial axonal transport (Magiera et al., 2018), in line with the reduction in mitochondrial speed in our TNFα-treated SNCA dupl neurons where polyglutamylation was aggravated in the neurites.Tau-αsyn interplay also powerfully regulates transport, as the overexpression of either protein alters kinesin attachment, motility and expression (Dixit et al., 2008;Prots et al., 2013Prots et al., , 2018)).The basally increased tau levels in SNCA dupl neurons and enhanced α-syn co-localisation upon IL-17A exposure may therefore create axonal loci of kinesin detachment, resulting in the changes in mitochondrial speed and directionality observed in these cells.Thus, the subtle but numerous cytokine-induced cytoskeletal alterations likely contribute to the axonal transport dysfunction via multiple mechanisms.
Notably, IL-17A again emerged as the only cytokine with SNCA dupl-specific effects, in line with the cytoskeletal data.While its impact was not overt, the major IL-17A pathology was the reduction in medium velocity retrograde mitochondria.Most mitochondria in mature neurons are stationary and are released upon injury to be retrogradely transported for somatic mitophagy (Cheng et al., 2022).In contrast, more severe damage leads to their arrest and axonal mitophagy.Interestingly, IL-17A has been shown to induce mitochondrial dysfunction and mitophagy (Kim et al., 2017;Ramakrishnan et al., 2020;Yang et al., 2020).Alternatively, energetic stress in distal axons inhibits retrograde transport, locking mitochondria at the stressed site (Watters et al., 2020).Such explanation would be in line with the relative IL-17A-induced shift of transport in the anterograde direction.A-syn-mediated mitochondrial pathology in PD is well described (Thorne & Tumbarello, 2022), and may therefore cooperate with IL-17A to produce mitochondrial transport deficits.
We demonstrate that IL-17A-mediated damage is dependent on α-syn pathology by rescuing functional deficits in SNCA dupl neurons by a pre-treatment with the α-syn anti-oligomerisation compound NPT100-18A.The peptidomimetic, whose more bioavailable version NPT200-11 in its single enantiomeric form UCB0599 (minzasolmin) is due to complete Phase II trials for PD in autumn 2024, binds to 96-102 amino acid region of α-syn and prevents its dimerisation and further aggregation (Schwarz et al., 2023;Smit et al., 2022;Wrasidlo et al., 2016).Application of NPT to neurons during differentiation had no impact on neurite morphology and did not influence basal mitochondrial transport in healthy neurons (Kouroupi et al., 2017;Prots et al., 2018).We likewise did not observe NPT-induced changes in functional parameters that were comparable between control and SNCA dupl neurons, indicating that NPT100-18A does not disrupt physiological α-syn function.However, NPT rescued the IL-17A-induced slowing of retrogradely transported mitochondria and prevented the transient IL-17A-mediated reduction of neuronal activity, demonstrating that α-syn is a necessary but not sufficient component of some pathological PD features, potentiating the effects of other pathogenic factors such as inflammatory stimuli.Our findings shed the first light on the reason behind increased sensitivity to inflammation in PD neurons.The mechanistic link between α-syn and IL-17A nevertheless requires further investigation, as it was of the scope of our study to explore this relationship.The two main possibilities include a) direct or indirect enhancement of α-syn aggregation by cytokines; or b) the common downstream pathways of cytokines and α-syn.A recent study has shown increased α-syn aggregation in SH-SY5Y neuroblastoma and neural progenitor cells upon TNF-α and IL-6 application (Cheng et al., 2022).However, a deeper biochemical analysis confirming this observation is needed in human neurons in the context of PD.
Open questions remain regarding the translatability of findings from SNCA dupl cases to other forms of PD with different aetiology.A particularly interesting comparison would be with synuclein-negative PD cases such as some individuals with LRRK2 variants who do not have Lewy body pathology or α-syn amplification in seeding assays (Kalia et al., 2015;Siderowf et al., 2023).Based on our findings, the neuronal response to inflammation is predicted to differ from that of patients with synuclein pathology.Further investigation of other synucleinopathies, such as multiple system atrophy where synuclein pathology localises to oligodendroglia (Poewe et al., 2022), would reveal whether and how α-syn also regulates the sensitivity to cytokines in nonneuronal cells.
While we have harnessed a neuronal monoculture to investigate direct cytokine effects on neurons, multicell type models such as brain organoids will be necessary to answer questions regarding the spatial cellular interplay during cytokine-mediated neuroinflammation.For instance, it remains unclear whether cytokine-exposed neurons release chemokines attracting or activating neighbouring immune cells.Further, iPSCs lose the ageing signature during reprogramming, possibly masking some pathologies, an issue which may be addressed by the direct reprogramming of somatic cells into neurons (Pitrez et al., 2024).Nevertheless, this approach suffers from low efficiency, and iPSC-based systems currently remain among the most relevant models to investigate human-specific disease on an individual level.Recent data from postmortem PD brains showing the neuron-specific accumulation of IL-17A and the redistribution of neuritic acetylated tubulin towards the soma support the validity of our iPSC-based model (Gate et al., 2021;Mazzetti et al., 2024).
In conclusion, human cortical neurons directly respond to pro-inflammatory cytokines which can induce cytoskeletal and functional pathology especially in the presence of increased α-syn dosage and/or pathology.By demonstrating that the unique vulnerability of SNCA dupl neurons to IL-17A can be reversed by inhibiting α-syn oligomerisation, we thus highlight that a multi-pathway therapeutic approach in PD may be particularly promising.

Table legends
Table 1.The list of primers used for qPCR analysis.
Table 2.The list of primary and secondary antibodies used for immunofluorescence.

Fig. 2 .
Fig. 2. Cytoskeleton analysis of cytokine-treated CNs.For each statistical test: *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ns, not significant.(A) Representative immunofluorescence images of vehicle-and cytokine mix-treated SNCA dupl CNs labelled with TUNEL.Scale bar = 50 μm.(B) Quantification of TUBB3 + /TUNEL + control and SNCA dupl cytokine-treated neurons.Each data point represents the average of >75 neurons.n = 3 / genotype.Two-way ANOVA with post hoc Sidak's / Dunnett's test.(C) Representative immunofluorescence images of vehicle-and cytokine mix-treated CNs stained for TUBB3.Scale bar = 50 μm.(D) Mean fluorescence intensity (MFI) of TUBB3 per neuron normalised to mean of control vehicle (left), and the fold change in MFI versus the vehicle of each respective cell line (right).a.u, arbitrary units.n = 3 / genotype.Two-way ANOVA with Fisher's LSD test.(E) Representative immunofluorescence images of vehicle-and cytokine mix-treated CNs stained for αtubulin acetyl K40 (acTub).Scale bar = 50 μm.(F) MFI of acTub per neuron normalised to TUBB3 MFI and the mean of control vehicle (left), and the fold change in MFI versus the vehicle of each respective cell line (right).n = 3 / genotype.Two-way ANOVA with Sidak's post hoc test: ns, not significant.(G) Representative immunofluorescence images of vehicle-and IL-17A-treated CNs stained for polyglutamylated tubulin (pgTub).Scale bar = 50 μm.(H) Quantification of (MFI) of pgTub per neuron normalised to TUBB3 MFI and the mean of control vehicle (left), and the fold change in MFI versus the vehicle of each respective cell line (right).n = 3 / genotype.Two-way ANOVA with Sidak's post hoc test.(I) Workflow for the quantification of acTub and pgTub in neurites from immunofluorescence images.(J) Basal acTub distribution along neurites, expressed as mean grey value, in control and SNCA dupl CNs.Each data point represents an average of 20 neurites.n = 3 / genotype.Kolmogorov-Smirnov test.(K) acTub distribution along neurites in cytokine-treated control (left) and SNCA dupl (right) CNs.n = 3 / genotype.Kruskal-Wallis test with post hoc Dunn's test.(L) Basal pgTub distribution along neurites in control and SNCA dupl CNs.n = 3 / genotype.Kolmogorov-Smirnov test.(M) pgTub distribution along neurites in cytokine-

Fig. 3 .
Fig. 3. Analysis of tau pathology in cytokine-treated CNs.For each statistical test: *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ns, not significant.(A) Representative immunofluorescence images of vehicle-and cytokine mix-treated CNs stained for tau.Scale bar 50 μm.Insets: Close-up immunofluorescence images of tau signal only within TUBB3 signal in control (A') and SNCA dupl (A'') CNs.Pseudocolour heatmap was used to enhance contrast.Arrowheads point to tau speckles.Scale bars = 25 μm.(B) Quantification of tau MFI per neuron normalised to the mean of control vehicle (left), and the fold change in MFI versus the vehicle of each respective cell line (right) in vehicle-and cytokine-treated CNs.Each data point represents an average of >75 neurons.n = 3 / genotype.Two-way ANOVA with Sidak's post hoc test.(C) The number of tau speckles ≥ 1 μm in diameter per neuron in vehicle-and cytokine-treated CNs.n = 3 / genotype.Two-way ANOVA.(D) Workflow for the quantification of tau and α-syn co-localisation in neurites.(E) Representative thresholded images of total tau signal (magenta) and tau signal overlapping with α-syn signal (white) in neurites of vehicle and IL-17A-treated control and SNCA dupl CNs (left), and quantification of the fraction of tau overlapping with α-syn (middle) and of fraction of α-syn overlapping with α-tau (right) in neurites treated with vehicle or IL-17A.n = 19-21 neurites.Two-way ANOVA with Sidak's post hoc test.

Fig. 5 .
Fig. 5. Rescue of IL-17A-mediated functional pathologies in dupl CNs by NPT100-18A (NPT).For statistical test: *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ns, not significant.(A) Experimental paradigm for the rescue of mitochondrial axonal transport.(B) Mitochondrial movement incidence in control and SNCA dupl CNs treated with IL-17A ± NPT. n = 20 neurites.Two-way ANOVA with Sidak's Dunnet's post hoc tests.(C) Mitochondrial retrograde (left) and anterograde (right) speed classification following IL-17A ± NPT in control and SNCA dupl neurites.n = 31 neurites.Two-way ANOVA with Sidak's and Dunnet's post hoc tests.(D) Experimental paradigm for the rescue of neuronal activity.(E) The number of active electrodes in SNCA dupl CNs after 4 weeks of differentiation in the presence or absence of NPT.n = 6 wells.Two-way ANOVA with Sidak's / Dunnet's post hoc test.(F) The timeline of IL-17A-mediated effects on the number of electrodes in control (top) and SNCA dupl (bottom) CNs with (right) or without (left) NPT pre-treatment.n = 6 wells.Repeated-measures one-way ANOVA with Dunnet's post hoc test.(G) The percentage change in the number of active electrodes compared to baseline following IL-17A treatment with NPT pre-treatment.n = 6 wells.Two-way ANOVA with Fisher's LSD test.