Divergent effects of pathological α-synuclein truncations and mutations on phase separation

Abstract

Phase separated condensates of α-synuclein (α-Syn) accelerate amyloid fibril formation, a process implicated in Parkinson’s disease pathogenesis. Yet, the precise effects of pathologically relevant α-Syn sequence modifications on this process remain unclear. Here, we show that sequence truncations exert the strongest influence on condensate thermodynamics, material properties, and amyloid aggregation, whereas familial point mutations impart minimal effects. Among the tested familial variants (A30P, H50Q, G51D and A53T), only G51D forms condensates that show a reduced propensity for amyloid fibril formation. Truncated variants undergo rapid gelation and form amyloid fibrils almost immediately. Extending our study to multicomponent systems where α-Syn is a client, we show that α-Syn can dissolve DNA-peptide coacervates or assemble into Pickering clusters on condensate surfaces—regulating condensate fusion and nucleic acid partitioning. These functions depend on the acidic C-terminal domain of α-Syn. Together, our results show disease-relevant modifications can modulate α-Syn phase behavior, both in pathological and physiological contexts.

This study shows how disease relevant sequence alterations affect phase separation and aggregation behavior of α-Syn. Notably, terminal truncations favor phase separation and aggregation and diminish interaction with other condensates.

Introduction

Intrinsically disordered proteins (IDPs) often assemble into higher-order structures through multivalent interactions¹. A major example of such behavior is liquid–liquid phase separation, which generates dynamic, membrane-less condensates that compartmentalize and regulate key steps in gene expression, transcription, and other essential cellular processes^2–10. Phase separated condensates can concentrate biomolecules (nucleic acids and proteins) to very high local levels, potentially enhancing reaction efficiency significantly, but can also become the origins of dysfunction. For example, disease-associated mutations or post-translational modifications of IDPs can accelerate their evolution into amyloid states, contributing to a myriad of diseases^11–14. Aberrant condensate maturation and loss of fluidity and reversibility associated with amyloid aggregation have been reported for many IDPs in vitro including FUS, hnRNPA1, TDP-43, tau, and α-synuclein (α-Syn), linked to amyotrophic lateral sclerosis, frontotemporal dementia, Alzheimer’s, and Parkinson’s disease, respectively^15–22. While in some cases, the high density of biomolecules within condensates can suppress amyloid fibril growth and proliferation of client IDPs^23,24, the surfaces of condensates have been found to be efficient nucleation sites for amyloid fibrils^24–27. This is particularly relevant for α-Syn which has been shown to be almost fully resistant to aggregation in bulk solution under quiescent conditions²⁸. However, under conditions favoring phase separation, α-Syn readily and efficiently undergoes amyloid fibril formation^21,22,29. Despite this potential pathogenic role and relevance of phase separation, few studies have quantitatively investigated the effects of pathological α-Syn modifications on this process.

Here, we combine a range of biophysical techniques including our in-house developed microfluidic phase-separation assays to quantitatively elucidate the effects of terminal truncations and familial point mutations on α-Syn phase separation and amyloid aggregation. We show that terminally truncated α-Syn variants, commonly found in Lewy bodies^30–34, possess a significantly higher propensity for phase separation, and form gel-like condensates enriched in amyloid fibrils. Although the driving forces for phase separation remain similar across most familial mutants of α-Syn, G51D condensates exhibit a significantly reduced propensity for gelation and amyloid aggregation, which has also been verified recently in cellular models³⁵.

Since cellular condensates are usually composed of multiple components³⁶, we also tested the effect of α-Syn as a ‘client’ molecule within model coacervate (RP3 peptide/ssDNA) and IDP condensate (DDX4N1) systems. The RP3/ssDNA coacervates and DDX4N1 condensates represent two major models of phase separation in cells—driven by electrostatic interactions and hydrophobic/cation-π interactions, respectively^29,37,38. We show that by either partitioning within coacervates, or adsorbing on condensate surfaces as Pickering clusters, α-Syn can regulate coacervate/condensate stability, growth, and nucleic acid partitioning within these compartments. Our experiments reveal that terminal truncations also impair these functions of α-Syn as a client, as they are modulated by the acidic C-terminal region of α-Syn.

Results

Terminal truncations alter the thermodynamics of α-Syn phase separation

We employed Taylor dispersion-induced phase separation (TDIPS³⁹), a microfluidic capillary-based technique that we developed on the FIDA1 instrument, to chart phase diagrams of α-Syn variants as a function of NaCl and protein concentration (Fig. 1a, Supplementary Fig. 1, Table 1). The wild-type (WT) and familial mutants (in our study: A30P, H50Q, G51D and A53T) did not phase separate in the absence of NaCl—suggesting Debye screening of unfavorable electrostatic interactions by NaCl is required for phase separation to occur. Interestingly, apart from a broader phase diagram, i.e., formation of condensates at lower protein and NaCl concentrations, C-terminally truncated (residues 1–125, ‘ΔC1–125’ and 1–110, ‘ΔC1–110’) and both N/C-terminally truncated (residues 30–110, ‘Core’) α-Syn showed the opposite behavior to the full length proteins. These variants phase separated more strongly in the absence of NaCl⁴⁰ (Fig. 1a). This observation was further verified using confocal microscopy (Fig. 1c). The dilute phase concentration (C_dil) is a crucial parameter to determine the thermodynamic stability of condensates. This is because:

$${\mathrm{C}}_{\mathrm{dil}}\propto e^{-\chi }$$ 1

and the Flory parameter (χ) is a measure of the interaction strength between protein molecules undergoing phase separation, relative to protein-solvent interactions⁴¹. We measured C_dil of α-Syn variants as a function of NaCl concentration using Capillary flow experiments (Capflex²⁹) (Supplementary Fig. 1, Table 2). We found a systematic and very similar decrease in C_dil with increasing NaCl concentration for WT and the familial α-Syn mutants (Fig. 1b). In line with the results from TDIPS, C_dil of truncated α-Syn variants exhibited an inverted relationship with NaCl concentration, suggesting that these variants phase separate more strongly with decreasing ionic strength. At 0 mM NaCl, C_dil scaled well with the degree of truncation, being the lowest for Core (46 µM), followed by ΔC1–110 (80 µM) and ΔC1–125 (135 µM) (Fig. 1b). The same scaling was observed at 150 mM NaCl, despite the reversed electrostatic dependence, showing that terminal truncations promote phase separation of α-Syn, even at physiological ionic strength (Fig. 1d).

Fig. 1. Ionic strength dependence of phase separation of α-Syn variants.

a Coarse phase diagram of WT and variant (ΔC1–125, Core, A30P, H50Q, A53T, G51D) α-Syn as a function of NaCl concentration. Here, n = number of condensates detected, and I_avg = average intensity of fluorescence spikes/condensates. The experiments are performed with 25% (w/v) PEG-8000, in 20 mM sodium phosphate buffer, and at 25 °C. b (Left) C_dil as a function of NaCl concentration for WT and familial (A30P, H50Q, G51D, A53T) α-Syn variants. (Right) C_dil as a function of NaCl concentration for WT and terminally truncated/altered α-Syn variants (ΔC1–125, ΔC1–110, Core, C_rev). C_t c Representative confocal microscopy images of the respective proteins with 0, 150 and 250 mM NaCl. Experiments (b, c) are performed at Ct = 200 μM with 20% (w/v) PEG-8000, in 20 mM sodium phosphate buffer, pH 7.4, and at 25 °C. d Normalized C_dil at physiological (150 mM) NaCl and KCl concentration. The datapoints (b, d) represent n = 2, or n = 3 independent experiments. For n = 3, the data are presented as mean ± S.D. For n = 2, the mean value is shown. Source data are provided as a Source Data file.

Table 1.

TDIPS parameters for α-Syn phase diagrams

Tray	Vial	Pressure (mbar)	Time (s)	Outlet	Measure	Comment
2	1	3500	360	Variable	no	1 M NaOH wash for cleaning the capillary.
2	2	3500	180	Variable	no	MQ water wash to remove NaOH.
2	3	3500	30	Variable	no	20 mM sodium phosphate buffer, pH 7.4 to equilibrate the capillary.
1	Analyte (NaCl as variable)	3500	90	Variable	no	20 mM sodium phosphate buffer, pH 7.4, 25% (w/v) PEG-8000, with appropriate NaCl/KCl concentration as the experimental condition.
1	Indicator (Protein)	900	10	Variable	no	200–300 μM α-Syn+20% (w/v) PEG-8000 in 20 mM sodium phosphate buffer, pH 7.4. Each sample was spiked with 20 nM Alexa488140C-α-Syn as fluorescent reporter molecule.
1	Analyte (NaCl as variable)	2800	600	Variable	yes	20 mM sodium phosphate buffer, pH 7.4, 25% (w/v) PEG-8000, with appropriate NaCl/KCl concentration as the experimental condition.

Table 2.

Capflex parameters for α-Syn phase separation

Tray	Vial	Pressure (mbar)	Time (s)	Outlet	Measure	Comment
2	1	3500	350	Variable	no	1 M NaOH wash for cleaning the capillary.
2	2	3500	120	Variable	no	MQ water wash to remove NaOH.
2	3	3500	60	Variable	no	20 mM sodium phosphate buffer (20 mM Tris HCl, for CaCl2) equilibration with appropriate NaCl/KCl concentration and pH as the analyte (phase separated solution).
1	Analyte (Phase separated solution)	2800	600	Variable	yes	200–300 μM α-Syn variants with 20% (w/v) PEG-8000 at different NaCl/KCl concentrations and pH. Each sample was spiked with either 20 nM Alexa488-140C-α-Syn unless otherwise mentioned.

From these observations, we hypothesized that the negatively charged (acidic) amino acids (abundant in the C-terminal region) were responsible for the inversion of salt dependence. To demonstrate this further, we measured C_dil of a recombinant α-Syn variant (C_rev), where five C-terminal acidic residues were substituted with lysine²². Akin to the truncated variants, phase separation of C_rev was also electrostatically favorable with a lowest C_dil of 145 µM at 0 mM NaCl (Fig. 1b). Subsequently, we measured C_dil of WT and C-terminal truncated/modified variants across a pH range of 5.6–7.8. Our data showed that the formal net charge of α-Syn variants (theoretically derived from independent pKa values) correlated well with the extent of phase separation at different pH values (Supplementary Fig. 2)—establishing the importance of the charged C-terminal tail in modulating α-Syn phase separation.

To explore this further, we also probed phase separation of α-Syn in the presence of KCl (K⁺ being the dominant cytosolic cation) and CaCl₂ (Ca²⁺ being the major signaling cation in neurons). We observed a very similar dependence of α-Syn phase separation in the presence of KCl compared to NaCl, i.e., phase separation of the WT increased with higher KCl concentrations, whereas the Core variant exhibited the opposite trend (Supplementary Fig. 3). These findings suggest that generic Debye screening by monovalent cations facilitates phase separation of α-Syn, and the effect of K⁺ is analogous to Na⁺. Notably, at physiological relevant NaCl and KCl concentrations, the truncated variants showed a higher propensity to phase separate compared to WT and familial mutants (Fig. 1d, Supplementary Fig. 4).

Multivalent cations such as Ca²⁺, Cu²⁺, and Fe³⁺ have been shown to interact with acidic (-ve) residues of α-Syn at physiologically relevant pH^42–44. These ions can also induce α-Syn phase separation in cells²¹. Quantification of the relative C_dil revealed a progressive depletion of soluble α-Syn across all familial and C-terminally truncated variants with increasing concentrations of Ca²⁺, with a large fraction (80%) of protein partitioned into the dense phase at 2.5 mM Ca²⁺ (Supplementary Fig. 5). In contrast, the phase separation of Core variant was insensitive to the calcium concentration, arguably due to the fact that its dilute phase concentration is already low in this regime, and/or it lacks the calcium binding sites.

Altogether, our data show that under physiological conditions, thermodynamics of α-Syn phase separation remains mostly unaffected by familial point mutations, but is significantly promoted by C- and N-terminal truncations. It is important to note here that we used 50 nM Alexa488-α-Syn WT as a fluorescent probe for the phase separation of all α-Syn variants. Using a series of additional experiments, we verified that the Alexa488-α-Syn accurately reports on the C_dil, even in cases where the electrostatic dependence is reversed (truncated variants) (Supplementary Fig. 6, Supplementary Discussion).

Terminal truncations accelerate gelation rate and modulate material properties of α-Syn condensates

We employed fluorescence recovery after photo bleaching (FRAP⁴⁵) to examine the translational dynamics of molecules within WT and variant α-Syn condensates. Condensates formed by WT and familial mutants recovered well (mobile fraction ~0.6–0.9) and maintained high translational dynamics for several tens of minutes (Fig. 2a). Interestingly, fluorescence recovery decreased substantially for ΔC1–125, ΔC1–110 and Core (mobile fraction ~0.2–0.4 within 10 min), suggesting a sol-gel transition with fast kinetics for the truncated variants (Fig. 2a, Supplementary Fig. 7). Complementary dynamic/static light scattering (DLS/SLS) measurements further supported the FRAP data obtained from truncated variants, showing that these condensates fail to fully dissolve upon dilution and were harder to dissolve even with 1 M urea (Supplementary Fig. 8). Although familial mutants displayed comparable fluorescence recovery to the WT just after formation, we hypothesized that their gelation timescales might differ due to distinct amyloid aggregation kinetics within the dense phase. We observed that over time, fluorescence recovery progressively diminished and the mobile fractions decreased to 0.3–0.45 for most full-length variants over 1 h (Fig. 2a, Supplementary Fig. 7), however not with identical kinetics. Notably, G51D retained the highest molecular mobility and solidified with the slowest kinetics, maintaining a substantial mobile fraction (~0.7) even after 1 h of incubation (Fig. 2a, Supplementary Fig. 7).

Fig. 2. Gelation timescales of α-Syn condensates.

a The change of the mobile fraction as a function of time for condensates formed by all α-Syn variants. The lines joining the data-points and the gray shaded region are guides to the eye. C_t = 300 μM for each protein. The experiment is performed n = 2 independent times and the mean values are shown. b (Left) Distance (D) between two condensates as a function of relaxation time (t) is plotted for fusion events detected for WT and ΔC1–110 (a representative case). The curves are fitted with a mono-exponential decay (see methods) to determine τ values. (Right) τ_app values as a function of D₀ are plotted for WT and ΔC1–110. The value of the slopes (corresponding to τ_app = η/γ) obtained from linear regression are indicated in the figure. The experiments (a, b) are performed with 20% (w/v) PEG-8000, in 20 mM sodium phosphate buffer, pH 7.4, and at 25 °C. Source data are provided as a Source Data file.

FRAP alone is not sufficient to infer macroscopic material properties since monomers can diffuse freely even within a network architecture inside condensates⁴⁶. To verify whether the observed translational dynamics were in agreement with the macroscopic viscosity (η) of the condensates, we measured the relaxation kinetics of condensates during fusion events^47,48 (Fig. 2b, Supplementary Fig. 9). We restricted these experiments to WT, ΔC1–110, and Core which showed considerable differences in FRAP experiments.

To obtain the apparent inverse capillary velocity (ν_app) as a measure of condensate viscosity (η, where ν_app = η/γ and γ is the surface tension), we performed mono-exponential fitting of the end-to-end distance (D) during condensate fusion to extract the relaxation time constant (τ_app) (Fig. 2b). τ_app was then plotted against $\sqrt{{\mathrm{D}}_0}$, D₀ being the distance at the beginning of fusion, as:

$${\mathrm{\tau }}_{\mathrm{app}}={\mathrm{\upsilon }}_{\mathrm{app}}\sqrt{D_0}$$ 2

(See methods) (Fig. 2c). Tens of fusing condensate pairs that had similar sizes were chosen for both WT and ΔC1–110. Despite smaller condensates, fusion kinetics were markedly slower for ΔC1–110 (D₀ = 4 μm, τ_app = 0.6 s) compared to WT (D₀ = 4.5 μm, τ_app = 0.2 s), resulting in a roughly twofold higher ν_app (~0.93 s·μm⁻¹) relative to WT (~0.45 s·μm⁻¹) (Fig. 2b, c). Since some WT condensate pairs were in contact with the glass surface, the measured ν_app for WT is likely overestimated due to additional drag that slowed fusion and hindered full relaxation. Although ν_app does not directly translate into the viscosity ‘η’ since the surface tension (γ) of condensates can be different between the two proteins, our analysis clearly shows that the overall material properties are significantly modulated by the C-terminus, with ΔC1–110 condensates being more viscous compared to those formed by the WT, while Core condensates were essentially solid-like and showed no fusion upon contact (Fig. 2b, Supplementary Fig. 9).

Pathological variants lead to different extents of amyloid fibril formation via phase separation

While often associated, condensate solidification may not always be a consequence of amyloid fibril formation. With this in mind, we designed the subsequent experiments to quantify α-Syn amyloid fibril formation via phase separation, and compared their results across the pathological variants.

Phase separated α-Syn solutions aggregate primarily via two dominant mechanisms occurring on similar timescales²² (Fig. 3a). These include: 1) Condensate gelation due to amyloid fibril formation within the dense phase, and 2) Amyloid fibril formation of the dilute phase monomers catalyzed by existing condensates^22,29. We confirmed that all α-Syn variants form amyloid fibrils under phase separating conditions using Thioflavin T (ThT) fluorescence assays, together with transmission electron microscopy (TEM) and atomic force microscopy (AFM) (Supplementary Fig. 10, Supplementary Discussion). However, these approaches provide limited insight into the underlying mechanisms. We therefore set out to dissect these processes in detail and establish quantitative measures to compare between the variants.

Fig. 3. Amyloid fibril formation driven by α-Syn condensates.

a Schematic showing two mechanisms of amyloid fibril formation driven by α-Syn condensates. Representative confocal microscopy images of Alexa488 labeled α-Syn condensates show dense phase aggregation (left) and propagation of aggregation in the dilute phase induced by the condensates (right). b The dilute phase monomer concentration (soluble fraction) decreases exponentially over time under phase separating conditions for all α-Syn variants. The data is fitted to a mono-exponential decay function to calculate the t_1/2. c ThT fluorescence assay for serially diluted WT condensates (sonicated) containing fibril seeds showing a clear concentration dependence. The slopes of these graphs from 0 to 10 h are calculated and used as a relative quantification of the seeding potential. d The slopes (ThT fluorescence/h) are plotted against the dilution factors as a comparative measure of fibril seed abundance among the α-Syn variants. The data shows H50Q, ΔC1–125 and Core condensates possess the highest seeding potential, while G51D condensates have the lowest seeding potential when treated with our protocol of seed extraction by sonication. The gray areas highlight the seeding-competent range in terms of the dilution factor for each variant. The experiments (b–d) are performed with 300 µM protein, 20% PEG, 150 mM NaCl at physiologically relevant pH (7.4). The experiments (b–d) are performed n = 2 independent times. Mean values are shown for (b), and representative sets of data are shown for (c) and (d). Source data are provided as a Source Data file.

At equilibrium, the concentrations of the dense and dilute phases of phase separated solutions are expected to stabilize and remain constant unless altered by downstream processes. However, our measurements revealed that for all α-Syn variants, the concentration of the soluble fraction decreased approximately exponentially over a 24 h period and then plateaued for the remaining experimental time (100 h) (Fig. 3b). This behavior potentially indicated condensate-mediated conversion of soluble monomer into aggregates. In agreement with our FRAP and ThT data, we observed that the G51D variant exhibited the lowest extent of monomer depletion (soluble fraction plateaued at 150 µM after 24 h) and relatively slower kinetics of depletion (t_1/2 = 5 h), compared to other full-length variants (plateaued at 50 µM after 24 h, average t_1/2 = 2.5 h, Fig. 3b). Strikingly, the Core variant, which showed instant gelation of condensates in FRAP and had a comparable ThT kinetic profile to that of the WT (Fig. 2, Supplementary Fig. 10), had the slowest monomer depletion rate (t_1/2 = 24 h) (Fig. 3b). These observations raised two key questions:

1. Why do soluble fractions stabilize at concentrations at least an order of magnitude above the critical concentration for α-Syn amyloid fibrils (typically a few µM⁴⁹)?

2. Why do Core condensates display slower dilute phase monomer depletion despite their rapid gelation kinetics?

To address these points, we performed live imaging of WT α-Syn condensates over a 5 h period, during which most of the dilute phase was depleted. We observed progressive clustering of condensates over time (Supplementary Fig. 11, Supplementary Movie 1). The clustering could also be induced within minutes by high-speed centrifugation (>5000 × g) (Supplementary Fig. 11). We found that condensate clustering reduces their ability to seed further aggregation of monomers, even after their sonication (Supplementary Fig. 12). The condensates could only catalyze (seed) further aggregation of monomers from solution when sonicated at early time points, or prior to centrifugation (Supplementary Fig. 12, see methods). We therefore concluded that the solid-like condensates contain seeding-competent fibrils whose availability for growth becomes limited with condensate clustering, leading to a plateau in soluble α-Syn concentrations at values higher than the critical concentration for amyloid aggregation.

Considering this proposed mechanism, we devised an experimental strategy (Supplementary Fig. 12) that allowed us to quantify and compare the relative abundance of elongation-competent fibril seeds across different α-Syn variants. The half-time of dilute phase monomer depletion (t₁/₂) was used as a reference time point to compare the condensate seeding potential across variants. Full-length constructs and ΔC1–125 showed an average t₁/₂ of ~3 h, whereas Core exhibited a markedly slower t₁/₂ of 24 h (Fig. 3b). At these time points, the condensates were sonicated to release fibril seeds, and incubated for additional 1 h to allow further aggregation. The formed aggregates were then isolated using centrifugation, the pellets were serially diluted and used as seeds in a ThT elongation assay with freshly added monomer solutions (40 µM, no PEG) (Fig. 3c, d). The dilution threshold at which an increase in ThT fluorescence intensity (elongation) was no longer detectable served as a comparative measure of fibril seed abundance among the variants (Fig. 3c, d).

Consistent with our previous observations, condensates formed by G51D showed considerably lower seeding potential compared to all other variants (Fig. 3d, Supplementary Fig. 13). Interestingly, we found that H50Q, ΔC1–125 and Core condensates possessed the highest seeding potential (Fig. 3d, Supplementary Fig. 13), indicating the potential pathological relevance of phase separation by these variants.

Our data suggests that the depletion of dilute phase monomers may not directly reflect the total amount of amyloid fibrils within the dense phase. As a significant fraction of fibrils may be clustered within the condensate scaffold—their free ends being inaccessible to the dilute phase monomers for further growth, unless they are released by sonication (Supplementary Fig. 12). A slower dilute phase monomer depletion rate for the condensates formed by the Core variant (Fig. 3b) is also likely due to a high degree of clustering of fibrils within the dense phase. Importantly, while the overall seeding potential of the Core condensates appeared higher than that of the WT at their respective t_1/2, the result still likely underestimates the true amyloidogenic potential of this variant. This is because this particular variant depleted significantly fewer monomers from the dilute phase than other variants (with the exception of G51D) (Fig. 3b). Given that the measured seeding potential was higher than WT despite lower monomer consumption, this implies that the monomers in these condensates were converted into more seeding competent fibrils. This observation is also consistent with previous reports demonstrating the intrinsically higher aggregation propensity of the Core region of α-Syn^50,51.

These results establish phase separation as an important modulator of α-Syn amyloid aggregation, and reveal that pathological variants perturb this process in several distinct ways—by altering the thermodynamics of condensate formation, their subsequent gelation kinetics, and the efficiency with which monomers are converted into seed-competent fibrils within the condensates. We find that truncations modulate all of these pathways dramatically, while the effects of familial mutations are either comparable to WT (A30P, A53T), or confined to specific pathways (G51D, H50Q). While condensate gelation timescales often correlate with fibril formation, our data show that this relationship may not be universal across all variants (Supplementary Fig. 14). These findings place phase separation within the broader framework of the known mutational effects of α-Syn⁵², providing a mechanistic link between sequence variation, condensate behavior, and amyloidogenic potential.

Core α-Syn condensates give rise to cytotoxic amyloid fibrils among the pathological variants

Finally, to probe the pathological relevance of our findings, we tested how fibrils derived from phase separated condensates differ from conventionally prepared pre-formed fibrils (PFFs) when introduced into living neuronal cells. While studies have established that α-Syn phase separation can trigger cytotoxicity^53,54, the situation in vivo is highly complex and direct evidence for homotypic condensate formation by α-Syn alone remains limited. In this context, we sought to probe the consequences of applying exogenously generated homotypic α-Syn fibrils derived from condensates—focusing on the propagation⁵⁵ of fibrillar species. Human neuroblastoma SH-SY5Y cells were exposed to 5 μM of WT PFFs or 5 μM of WT condensate-derived fibrils for 96 h (Fig. 4a). We observed that both PFFs and condensate derived α-Syn fibrils readily associate with cell membranes and can penetrate into the cytoplasm, where they form punctate structures (Fig. 4a). In the MTT assay, the Core α-Syn fibrils (regardless of origin) resulted in the most deleterious effects in terms of mitochondrial/metabolic impairment (Fig. 4b). Notably, even the Core α-Syn monomer showed more pronounced metabolic damage compared to other variants, which could in part be caused by its rapid aggregation. PFFs and condensate-derived fibrils showed similar effect in the MTT assay across all variants (Fig. 4b). While the MTT assay is a metabolic readout of living cells, the Propidium Iodide (PI) assay is a direct marker of cell death. Using both of those orthogonal assays provides a more complete picture of cellular damages induced by fibrils. Strikingly, in our PI assay, we observed lower cytotoxicity of condensate derived fibrils compared to PFFs for the WT, G51D, and A53T variants. This could be due to pathway-dependent formation of distinct fibril polymorphs⁵⁴ and the differences in their toxic effects (Fig. 4c). In contrast, fibrils formed by A30P, H50Q, and the truncated variants exhibited comparable cytotoxic effects regardless of the aggregation pathway, with the Core variant being the only construct displaying significantly higher cytotoxicity (Fig. 4c, Supplementary Fig. 15)—similar to that observed in the MTT assay. Our data suggests that homotypic phase separation may contribute to the toxicity-relevant diversification of α-Syn fibrils in a sequence-dependent manner. Such a diversification can essentially arise in two ways: through the formation of distinct fibril strains, as well as through differences in size distribution of the fibrils caused by a different balance of nucleation and growth rates in the presence and absence of condensates. Which of these two factors contributes to differences in toxicity in any particular case is, however, not straightforward to disentangle.

Fig. 4. Terminal truncations confer cytotoxicity to condensate-derived fibrils.

a Cells were fixed and stained for DAPI (nucleus, cyan) and Sodium Potassium ATPase (membrane, magenta). Alexa488-labeled fibrils are shown in yellow. Confocal images from the z-stack, presenting the middle of the cells are shown. b Cells were exposed to 5 μM monomers, 5 μM of PFFs or 5 μM of condensate-derived fibrils made from the different variants of α-Syn, for 96 h. 570 nm absorbance indicative of mitochondrial activity (MTT assay) is plotted for monomeric α-syn variants, PFFs and condensate-derived fibrils for all α-syn variants, relative to the buffer control. The data shows the mean ±S.D for n = 3. Statistics: one-way ANOVA with Dunnett’s multiple comparisons against PBS control (F = 30.53): PBS vs Core monomers: P < 0.0001; PBS vs all 7 PFF types: P < 0.0001; PBS vs all 7 condensate-derived fibril types: P < 0.0001, (****P < 0.0001). Additionally one-way ANOVA with Dunnett’s multiple comparisons within each group against the WT version of that group (F within monomers = 5.511; F within PFFs = 6.157; F within condensate-derived fibrils = 15.94): WT monomers vs Core monomers: P = 0.0026; WT PFFs vs Core PFFs: P = 0.0074; WT condensate-derived fibrils vs G51D condensate-derived fibrils: P = 0.0029; WT condensate-derived fibrils vs A53T condensate-derived fibrils: P = 0.0177; WT condensate-derived fibrils vs 1–125 condensate-derived fibrils: P = 0.0025; WT condensate-derived fibrils vs Core condensate-derived fibrils: P < 0.0001, (#P < 0.05, ##P < 0.01, #### P < 0.0001). c Cells were exposed to either 18 μM WT, A30P, H50Q, G51D, A53T, 1–125 or core α-syn monomers, 5 μM of PFFs or 5 μM of condensate-derived fibrils made from the different variants of α-Syn for 96 h. The percentage of PI+ cells is measured on all individual cells, which were gated previously to exclude debris and doublets. Data represent the mean ±S.D of n = 3 independent experiments performed in triplicates for each condition. Statistics: one-way ANOVA with Dunnett’s multiple comparisons against PBS control (F = 5.429): PBS vs WT PFFs: P = 0.0052; PBS vs G51D PFFs: P = 0.0278; PBS vs A53T PFFs: P = 0.0004; PBS vs Core PFFs: P = 0.0002; PBS vs Core condensate-derived fibrils: P = 0.0008, (*P < 0.05, **P < 0.01, ***P < 0.001). Additionally one-way ANOVA with Dunnett’s multiple comparisons within each group against the WT version of that group (F within monomers = 0.872; F within PFFs = 2.163; F within condensate-derived fibrils = 7.013): WT condensate-derived fibrils vs Core condensate-derived fibrils: P = 0.0063, (##P < 0.01). Source data are provided as a Source Data file.

The C-terminus of α-Syn is a primary modulator of effects imparted on other condensates when α-Syn acts as a client

Next, we set out to provide a more generic view of α-Syn phase behavior in the context of other, distinct condensate systems, bridging pathological and physiological scenarios. We selected the RP3 peptide/ssDNA system and DDX4N1 as two well-established, minimal models that individually and separately recapitulate key principles of intracellular condensate formation—electrostatic coacervation and cation–π-driven condensation, respectively^29,37,38. We then probed how α-Syn, acting as a client IDP, behaves in these distinct environments, and how pathological mutations alter this behavior.

In a series of TDIPS experiments, we placed an RP3 plug between two ssDNA plugs (with and without α-Syn, see methods, Table 3) (Fig. 5a, left) to induce RP3/ssDNA coacervation within a capillary. In the absence of α-Syn, the expected Gaussian peak of ssDNA split into two peaks with detectable spikes—indicating mass transfer from the main peak due to phase separation³⁹ (Fig. 5a, right). Interestingly, as the α-Syn concentration within the ssDNA plugs increased, these two distinct peaks (and spikes) gradually disappeared, and a single peak (indicative of the absence of phase separation) emerged at equimolar α-Syn to RP3 ratios (Fig. 5a, right). This suggested that α-Syn could alter the electrostatic balance and dissolve RP3/ssDNA coacervates or prevent their formation. We quantified the coacervate inhibition potential of WT and pathological α-Syn variants by plotting the ratio between the areas under the two peaks, as a function of α-Syn concentration (Fig. 5b, Supplementary Fig. 16). We found that all full-length variants dissolved RP3/ssDNA coacervates at near equimolar ratios (0.8–1) of α-Syn to RP3 (Supplementary Fig. 16). Intriguingly, ΔC1–125 could not dissolve the RP3/ssDNA coacervates—stressing the importance of the entire negatively charged C-terminal region for this effect (Fig. 5b, Supplementary Fig. 16). In order to establish that the C-terminal region acted as a proxy for ssDNA by binding to RP3, we performed parallel confocal microscopy experiments with Alexa488-ssDNA and N-terminally labeled Alexa488-α-Syn (both WT and ΔC1–125). WT α-Syn partitioned homogeneously within RP3/ssDNA coacervates and dissolved them in a concentration-dependent manner, while ΔC1–125  displayed the complete absence of partitioning and therefore was also unable to dissolve them (Fig. 5c, d). Moreover, we found that ΔC1–125 remained excluded from the coacervates, which stands in agreement with previous observations²⁴. Taken together, the microscopy data fully agreed with the TDIPS experiments, and clearly showed that electrostatic regulation of coacervates was dictated by the interactions between RP3 and the negatively charged C-terminal tail of α-Syn.

Table 3.

TDIPS parameters for RP₃/ssDNA coacervation

Tray	Vial	Pressure (mbar)	Time (s)	Outlet	Measure	Comment
2	1	3500	45	Variable	no	1 M NaOH wash for cleaning the capillary.
2	2	3500	45	Variable	no	MQ water wash to remove NaOH.
2	3	3500	45	Variable	no	10 mM Tris-HCl buffer, pH 8.0, 150 mM NaCl to equilibrate the capillary.
1	Analyte	50	15	Variable	no	10 μM ssDNA spiked with 240 nM Alexa488-ssDNA in 10 mM Tris-HCl buffer, pH 8.0, 150 mM NaCl. This plug had different concentrations of α-Syn variants as titrants.
1	Indicator	50	15	Variable	no	1.5 mM RP3 in 10 mM Tris-HCl buffer, pH 8.0, 150 mM NaCl.
1	Analyte	50	15	Variable	no	10 μM ssDNA spiked with 240 nM Alexa488-ssDNA in 10 mM Tris-HCl buffer, pH 8.0, 150 mM NaCl. This plug had different concentrations of α-Syn variants as titrants.
2	3	1000	90	Variable	yes	10 mM Tris-HCl buffer, pH 8.0, 150 mM NaCl to mobilize the triple-plug to the detector.

Fig. 5. α-Syn as client for electrostatic coacervates.

a (Left) A triple-plug TDIPS strategy to investigate the effects of α-Syn on RP3/ssDNA coacervate system. (Right) In the absence of α-Syn, RP3 and ssDNA form coacervates, which is detected as peak splitting (mass transfer) of an otherwise Gaussian profile. Increasing α-Syn concentrations in the ssDNA plugs dissolve coacervates in a dose-dependent manner. b The ratio of the area under the two peaks (A₁/A₂) when coacervates form are plotted as a function of WT and variant α-Syn concentrations (shown as a stoichiometric ratio relative to RP3). Datapoints represent mean of n = 2 independent experiments. Representative confocal microscopy images of RP3/ssDNA coacervates as a function of (c) WT α-Syn and (d) ΔC1–125 α-Syn. Experiments (c, d) are performed with 200 µM RP3, 20 µM ssDNA, in 10 mM Tris-HCl, pH 8.0, 150 mM NaCl, and at 20 °C. Source data are provided as a Source Data file.

We next investigated the role of α-Syn as a client in DDX4N1 condensates. Capflex measurements showed that α-Syn did not substantially affect the thermodynamics of DDX4N1 phase separation, as quantified through determination of the dilute phase concentration (C_dil = 53 μM, C_t = 70 μM; Supplementary Fig. 17, Table 4). However, even at sub-stoichiometric levels (as low as 1 μM α-Syn relative to 70 μM DDX4N1), condensates appeared significantly smaller (Fig. 6a). Strikingly, α-Syn formed distinct submicron-scale clusters on DDX4N1 condensate surfaces (Fig. 6b). These surface clusters behaved as Pickering agents, sterically hindering fusion events to a significant extent (Supplementary Fig. 17), and thereby limiting condensate growth. Except for the Core variant, all α-Syn variants, including ΔC1–125, reduced condensate size, with average diameters decreasing from ~7 µm (0.5 µM α-Syn) to ~5 µm (5 µM α-Syn). The inability of Core, but not ΔC1–125, to mediate this effect highlighted the importance of the N-terminal region in Pickering cluster formation (Supplementary Fig. 18).

Table 4.

Capflex parameters for DDX4N1 phase separation

Tray	Vial	Pressure (mbar)	Time (s)	Outlet	Measure	Comment
2	1	3500	200	Variable	no	1 M NaOH wash for cleaning the capillary.
2	2	3500	60	Variable	no	MQ water wash to remove NaOH.
2	3	3500	40	Variable	no	0.1% (v/v) Tween-20 coating to prevent condensate sticking on the capillary wall.
2	4	3500	40	Variable	no	10 mM sodium phosphate buffer, pH 6.5, 50 mM NaCl to equilibrate the capillary.
1	Analyte (Phase separated solution)	500	250	Variable	yes	70 μM DDX4N1 phase separated solution with different concentrations of WT α-Syn. Each sample was spiked with 600 nM DDX4N1-YFP.

Fig. 6. α-Syn as client for DDX4N1 condensates.

a (Top) Representative bright field (BF) microscopy images showing DDX4N1 condensate size decrease as a function of α-Syn concentration. (Bottom) Distribution of DDX4N1 condensate diameters as a function of α-Syn concentration (Supplementary Table 1). The datapoints are from n = 2 independent experiments. b Representative confocal microscopy images with Alexa488 labeled α-Syn showing Pickering clusters on the surface of DDX4N1 condensates. c Representative time-lapse images showing progressive sequestration of 50 nM Alexa488-ssDNA within DDX4N1 condensates in the absence and presence of 1 μM α-Syn. d ssDNA sequestration rates for WT, ΔC1–125, and ΔC1–110 with increasing α-Syn concentrations. Experiments are performed with 70 μM DDX4N1, in 10 mM sodium phosphate buffer, pH 6.5, 50 mM NaCl, and at 25 °C. The datapoints represent mean of n = 2 independent experiments. Source data are provided as a Source Data file.

We further hypothesized that negatively charged α-Syn clusters could electrostatically repel nucleic acids and thus reduce their sequestration into DDX4N1 condensates. Confocal imaging confirmed a dose-dependent reduction in ssDNA uptake, with sequestration rates dropping from 2.0 to 0.1 RFU/s as α-Syn concentration increased (Fig. 6c). FRAP experiments showed that α-Syn did not alter the mobility of DDX4N1 molecules inside condensates (Supplementary Fig. 19), indicating that the effect arose solely from changes on the condensate surfaces. Akin to the RP3/ssDNA system, truncated α-Syn variants (ΔC1–125 and ΔC1–110) exhibited attenuated inhibition of ssDNA sequestration, reducing rates only from 2.0 to ~1.0 RFU/s (Fig. 6d, Supplementary Fig. 20), again demonstrating that the C-terminal tail could be a very important modulator when α-Syn acts as a client in other nucleic acid-enriched condensates.

Discussion

Familial mutations have been shown to alter lipid membrane binding and amyloid aggregation of α-Syn by altering the charge, hydrophobicity, or flexibility of the N-terminal helical region⁵². They also influence cross-elongation of amyloid fibrils where the structure of the template fibril (e.g., formed by a mutant) shapes the structure of the resulting fibril, leading WT monomers to adopt features and toxicity of the mutant fibril^52,56,57. Despite these reported effects on several individual steps of the amyloid formation pathway of α-Syn, we do not find a significant influence of these single point mutations on the thermodynamics of phase separation (Fig. 1). This is because in contrast to fibril formation, where well-defined molecular contacts are key drivers and stabilizers of the fibrils, a constantly re-arranging network of multiple weak interactions drives protein phase separation^58,59. Single point mutations do not substantially modify the overall driving forces for condensate formation—the free energy differences being small in most cases. Moreover, the effects of such mutations can be compensated by new interactions enabled by the mutation itself. They could only impart significant effects if they disrupt critical interaction motifs—such as cation–π or electrostatic interactions, as shown for other IDPs undergoing condensation⁶⁰. This appears not to be the case for α-Syn. A lower extent of phase separation of G51D has been recently reported in human cells and in budding yeast models³⁵. However, our findings indicate that this mutation primarily affects the kinetics of condensate gelation and amyloid fibril formation, rather than the thermodynamics of condensate formation (Figs. 1–3).

In contrast, terminal truncations that lead to strong alterations of the net charge of α-Syn show a significant impact on the thermodynamics of phase separation, leading to an inverted electrostatic dependence compared to full-length α-Syn⁴⁰), as well as stronger phase separation propensity near physiological conditions (Fig. 1). We also highlight that our experimental results diverge from prediction algorithms to decipher phase separation propensities of IDPs based on their amino acid sequence^61,62. In most prediction algorithms, the C-terminal region of α-Syn is identified as major driver of phase separation. In contrast to this generally proposed role, we find that the C-terminal region has a mostly modulatory function, and truncating this region results in more widespread homotypic condensation behavior. Lower saturation concentrations close to physiological ranges (tens of µM) across varying ionic strengths suggest a higher likelihood for the truncated variants to undergo condensation within cellular crowded milieus. In a budding yeast model, masking C-terminal charges with polyethylenimine (PEI) has been shown to induce more aggressive phase separation of α-Syn, supporting our in vitro data⁶³.

Intriguingly, in stark contrast to the effects of monovalent cations discussed above, we find that physiologically relevant divalent cations (i.e., Ca²⁺) can dominate the charge screening effects that otherwise limit condensation of the truncated variants (especially ΔC1–125). This is because unlike NaCl or KCl, which primarily contribute to Debye screening, Ca²⁺ directly binds to α-Syn⁶⁴. The ΔC1–125 variant retains a sufficient number of acidic residues to enable efficient Ca²⁺ binding which can then act as a “glue” between the negatively charged regions of two or more protein molecules. Such opposite modulatory trends of divalent vs. monovalent ions have been reported previously in DDX4 systems, where Ca²⁺ and Mg²⁺ enhance phase separation in contrast to NaCl²⁹.

These findings suggest that the interplay between sequence changes (mutations and truncations) and the ionic milieu of neurons collectively determines the extent and likelihood of α-Syn phase separation in vivo. This is in contrast to phase separation- independent amyloid fibril formation pathways where the aggregation rates increase in the presence of monovalent as well as divalent cations⁶⁴, therefore highlighting the differences between condensate formation and amyloid fibril formation.

Phase-separated α-Syn solutions are dynamic and represent an intermediate step on the pathway to amyloid aggregation⁶⁵, as it has not yet been demonstrated that the condensate state can be stably maintained without converting into a gel or amyloid fibrils. This highlights the importance of also considering kinetic factors when comparing pathological α-Syn variants. In terms of the kinetics of the sol–gel transition, most familial mutants behave similarly to the WT, with one notable exception: G51D, which shows substantially delayed gelation timescales (Fig. 2). In contrast, condensates formed by truncated variants with higher thermodynamic stability also show considerably faster gelation timescales compared to full-length variants (Fig. 2). A natural question that arises from these findings is how the gelation timescales and varying material properties relate to the formation of amyloid fibrils. Our data point to two central features of α-Syn condensates: 1) condensate gelation, irrespective of its timescale, consistently gives rise to elongation-competent amyloid fibrils, albeit to different extents depending on the variant (Fig. 3). Particularly striking is the almost instantaneous sol–gel transition of truncated α-Syn variants, rapidly producing amyloid fibrils. This finding is of potential pathological relevance given the enrichment of C-terminally truncated α-Syn in Lewy bodies^30–34. And 2) comparable gelation kinetics does not necessarily translate to comparable amounts of seeding-competent amyloid fibrils, as highlighted by H50Q condensates, which possess substantially more seeding potential than WT despite having similar gelation timescales (Figs. 2 and 3). It is important to note that these conclusions rest on the assumption that the dense-phase concentrations of the different variants are at least of the same order of magnitude (generally in the mM range^22,66).

Although α-Syn condensates gave rise to amyloid fibrils, our experiments show that most of those resulting fibrils, except those of the Core variant, did not lead to cell death in neuronal cells, as measured using the PI assay (Fig. 4). The enhanced cytotoxicity of the Core variant may reflect its higher fibril yield within the dense phase. However, this explanation is insufficient, since other variants such as H50Q and ΔC1–125 also produce abundant fibrils, but do not display comparable toxicity. A second, more likely explanation is that among the different variants studied here, only the Core condensates form fibrils of a morphology and with surface properties^54,67,68 that exert significant effects on neuronal physiology. The complete absence of a “fuzzy coat⁶⁹” probably allows the hydrophobic β-sheet core to interact directly with cellular targets and infer toxicity. In addition, due to the lack of the N/C-terminal regions, the Core variant appears constrained to limited pathogenic fibril polymorphs with properties close to conventionally prepared PFFs. However, we cannot rule out the possibility that the reduced cytotoxicity of most of the condensate-derived fibrils could be due to a reduced rate of fibril internalization compared to PFFs. Similar to previous reports, we find that condensate-derived fibrils induce mitochondrial dysfunction at levels comparable to PFFs, as measured by MTT assays^53,54. This apparent discrepancy between MTT and PI assays underscores the importance of employing multiple, orthogonal assays when assessing fibril-induced cellular damages. Mitochondrial impairment alone does not necessarily translate into acute cytotoxicity, and reliance on a single assay could lead to over- or underestimation of pathogenic potential.

Pathological manifestations due to α-Syn mutations/modifications may also indirectly arise from disruptions of other condensates/coacervates essential for neurons that sequester α-Syn as a client molecule. Emerging evidence suggest that neuronal condensates formed by tau, synapsin and VAMP2 can efficiently sequester α-Syn^70,71. However, the effects of α-Syn, particularly on nucleic acid enriched condensates/coacervates, remain underexplored. We find that for both RP3/ssDNA coacervates²⁹ and DDX4N1 model condensates^29,38, the most prominent effect of α-Syn was imparted on nucleic acids (in our case: ssDNA)—modulated by the acidic C-terminal tail of α-Syn (Figs. 5 and 6). In neurons, the dissolution of phase-separated assemblies is often equally critical compared to their formation. In fact, the ability to form and dissolve rapidly and repeatedly is an advantage of a cellular condensate over many other supramolecular assembly states. For many condensate systems, this dynamic cycling is mediated by enzymes (kinases) and ATP⁷². Here, we show another regulatory mechanism that might exist for many neuronal coacervates, where α-Syn can compete with ssDNA (or RNA) for a positively charged protein, thereby dissolving the coacervates (Fig. 5). On the other hand, α-Syn can also form Pickering clusters on the surface of DDX4N1 condensates. A similar effect was previously shown for TDP43 condensates where α-Syn emulsification resulted in TDP43 aggregation⁷³. Here, we find that these Pickering clusters control condensate size and slow down ssDNA partitioning within DDX4N1 condensates, which might be implicated in the regulation and stabilization of DDX4N1 condensates against nucleic acid-induced dissolution (Fig. 6). These insights demonstrate the multi-faceted nature of a single IDP on the condensation landscape of a living cell—both enabling and preventing dissolution of phase separated states depending on their molecular profile. While we recognize that our in vitro experiments are far from recapitulating the full complexity of the cellular environment, we believe that the methods we employed in this work as well as the insights derived from them significantly contribute to the constantly evolving understanding of the native function of α-Syn in neurons—particularly from the perspective of phase separation.

Taken together, our findings suggest that familial α-Syn mutants do not manifest their pathological properties via phase separation, whereas sequence truncations alter the phase separation behavior sufficiently to clearly differentiate their effects from that of the WT protein. Given the versatility of α-Syn, it is plausible that different sequence modifications associated with an increased risk for disease exert their deleterious effects in different ways. It should also be noted that the role of α-Syn phase separation in the generation of protein pathology is very likely to facilitate the formation of the first fibrils. Whether or not the very first fibrils that form in a patient that are highly cytotoxic may not be so relevant, because it has been shown that fibrils can amplify and generate toxic oligomers through secondary nucleation^74,75. We believe that our study can contribute to sharpen the focus on more significant sequence changes, and away from single point mutations, in the context of α-Syn condensate formation. This perspective can inform future studies of the physiological role of α-Syn condensation and its connection to the emergence and spreading of pathological forms of α-Syn.

Materials

WT and variant (A30P, H50Q, G51D, A53T, ΔC1–125, 140C) human α-Syn genetic sequences were encoded on a pT7-7 bacterial plasmid under isopropyl β-D-1 thiogalactopyranoside (IPTG) inducible promoter sequence (Genscript, USA). Purified ΔC1–110 and Core α-Syn variants were kind gifts from Prof. Samir K. Maji (IIT-Bombay, India). The C_rev gene construct (D115, D121, E126, E131, and E137, all substituted to K) was purchased from Twist Bioscience (USA) on a pET-29b (+) plasmid with a stop codon placed before the His-tag at the end of the protein sequence. DDX4N1 CtoA gene was encoded as a GST fusion protein on a pET29a (+) bacterial plasmid (Genscript, USA). DDX4N1-YFP was encoded on pET30M-2 plasmid and was a kind gift from Dr. Tim Nott (Oxford University, UK). HPLC-purified RP₃ peptide (solid-phase synthesis) was procured from Bachem (Switzerland), and Schafer-N (Denmark). HPLC purified unlabeled, and Alexa488 labeled ssDNA was procured from TAG Copenhagen (Denmark). LB broth, ampicillin, kanamycin, and IPTG for protein production was procured from VWR (Denmark). All other relevant reagents, salts and buffer components for experiments were purchased from Sigma (USA) and VWR (Denmark), unless otherwise specified. Formvar coated electron microscopy grids were purchased from Sigma (USA). Protein, peptide and nucleic acid sequences are provided at the end of the method section.

Methods

Protein production

Expression and purification of α-Syn variants

WT, A30P, H50Q, G51D, A53T, and ΔC1–125 α-Syn were produced and purified following identical protocols²². 100 mg/L ampicillin (kanamycin for familial point mutants) was used as a bacterial selection marker for protein production. After transformation with the desired plasmid, a primary E. coli BL21 (DE3) culture was grown for 10–12 h at 37 °C. The primary culture (10 ml) was then inoculated into 1 L LB and grown under a shaking speed of 160 rpm, till the OD₆₀₀ reached 0.8. Protein expression was induced by adding 1 mM IPTG and the culture was grown for 4 more hours. The cells were harvested by centrifugation (7000 × g, 20 min, 4 °C) and the pellet was stored at −20 °C until further use. To purify the protein, the cell pellet was dissolved in 20 ml, 10 mM Tris-HCl, 1 mM ethylenediamine tetra acetic acid (EDTA), pH 8.0 with 1 mM phenylmethylsulfonyl fluoride (PMSF) and sonicated on ice (10 s on, 30 s off, 12 cycles at 40% amplitude). Next, 1 µL Benzonase was added to the cell lysate to precipitate DNA followed by centrifugation at 20,000 × g for 30 min at 4 °C. The supernatant was collected and heated at 80 °C on a water-bath for 20 min. After cooling down the solution back to room temperature, a second centrifugation step (20,000 × g for 20 min at 4 °C) was performed to precipitate heat-sensitive proteins, while α-Syn remained in the supernatant. The supernatant was collected, and 4 ml saturated (NH₄)₂SO₄ was added for 1 ml supernatant to salt out α-Syn. Salted out α-Syn was obtained in the pellet after centrifuging the solution at 20,000 × g for 20 min at 4 °C. The pellet was dissolved in 7 ml of 25 mM Tris-HCl, pH 7.7, and 1 mM dithiothreitol (DTT) was added to the solution. Subsequently, the solution was dialyzed against the same buffer for 18 h at 4 °C to remove (NH₄)₂SO₄. The tank was replenished with fresh buffer after 12 h. The α-Syn solution was then subjected to anion exchange column (AEC) (HiTrap Q Hp 5 ml, GE healthcare, USA) followed by size exclusion chromatography (SEC) (HiLoad 16/600 Superdex 200 pg. column). The purified protein was eluted in 10 mM of sodium phosphate (NaH₂PO₄.H₂0 + Na₂HPO₄.2H₂0) buffer (pH 7.4). α-Syn concentrations were measured using a NanoDrop Lite (Thermo Scientific, USA), and a UV spectrophotometer, Labbot (Labbot, Sweden) by measuring the absorption at 280 nm. The theoretical molar extinction coefficient of full-length (5960 M⁻¹cm⁻¹) and ΔC1–125 α-Syn (2980 M⁻¹cm⁻¹) was predicted by ProtParam (Expasy, Switzerland). The protein purity was checked using SDS-PAGE. To note, 140C-α-Syn was expressed and purified using an identical protocol—the only difference being addition of 1 mM DTT and 1 mM EDTA in all buffers to prevent intermolecular cysteine-disulfide linkages²⁹.

Expression and purification of C-terminal charge reversed (C_rev) α-Syn

C_rev-α-Syn was purified in a small-scale setup, as previously established in our laboratory²². Briefly, the cell pellet obtained from 50 ml of culture was resuspended in 4 ml Tris buffer (10 mM, 1 mM EDTA, pH 8) and 2 mM PMSF. The cells were sonicated on ice (10 s on, 30 s off, 12 cycles at 40% amplitude) and centrifuged at 20,000 × g for 20 min, 4 °C. The supernatant was collected and heated to 85 °C for 20 min, followed by centrifugation (20,000 × g for 20 min, 4 °C). Next, DNA was precipitated by adding 5% v/v glacial acetic acid and 1 mg/ml streptomycin sulfate, followed by centrifugation (20,000 × g for 20 min, 4 °C). The supernatant was transferred into another tube and C_rev α-Syn was salted out with saturated (NH₄)₂SO₄ (final concertation of 2.5 M). The pellet was gently washed 2 times and dissolved in 10 ml Tris buffer (10 mM, pH 8.0). The protein was loaded on a gravity column with cation exchange resin (NuviaTM S, BIO-RAD, USA). The column was washed with 4 ml of 150 mM NaCl and the protein was eluted with 3 ml of 250 mM NaCl (10 mM Tris pH 8.0). The buffer was exchanged with an illustra NAP-5 column (GE Healthcare, USA) in 10 mM of sodium phosphate (NaH₂PO₄.H₂0 + Na₂HPO₄.2H₂0) buffer at the desired pH, before use. Protein concentration was measured by NanoDrop Lite (Thermo Scientific, USA) using 5960 M⁻¹.cm⁻¹ as the molar extinction coefficient for C_rev α-Syn. The protein purity was checked using SDS-PAGE.

DDX4N1 expression and purification

A variant of DDX4N1 was used in our experiments where all cysteine residues were mutated to alanines (CtoA)²². This avoids disulfide bridge formation and simplifies the protein purification protocol, having little or no effect on LLPS. We refer to this variant as only ‘DDX4N1’ for better readability. DDX4N1 was tagged with a His_tag fused with a thioredoxin-tag (eMM9) at the C-terminal that facilitates protein expression and thermodynamic stability. The His_tag-eMM9 was followed by codon-optimized DDX4N1 CtoA linked by a 10-residue GS-linker and a TEV-cleavage site (ENLYFQ/G). For DDX4N1-YFP, the protein was produced as His_tag-GST-TEV_site-DDX4-YFP. Successful clones were selected from a transformation plate (kanamycin as selection marker) and both the proteins were produced and purified following similar, previously described established protocols^22,29. Briefly, for DDX4N1, 10 ml overnight grown culture was (50 mg/l kanamycin) was used to inoculate 1 l AB-LB. The AB-LB-medium contained 5 g/l NaCl, 40 mM sodium/potassium phosphate (K/NaPi) buffer, pH 7.0, 15 mM (NH₄)₂SO₄, 50 mM NaCl, 2 mM MgCl₂, 0.1 mM CaCl₂, 3 nM FeCl₃, and 50 mg/l kanamycin. The culture was grown at 37 °C until OD₆₀₀ reached 0.8 and protein production was induced with 1 mM IPTG, before incubating overnight at 20 °C. Cells were harvested by centrifugation at 7000 × g for 20 min at 4 °C. 30 ml lysis buffer (50 mM sodium phosphate buffer, pH 6.5, 500 mM NaCl) was used to resuspend the cell pellet. 8 µl Benzonase was added to the solution and the resuspended cells were lysed on ice with the help of a probe sonicator (40% amplitude, 30 s on, 30 s off, total time of 30 min). The cell lysate was centrifuged at 20,000 × g for 20 min at 4 °C and the supernatant was collected and heated at 80 °C for 15 min. The solution was centrifuged at 4500 × g for 30 min at 4 °C to remove heat-induced protein aggregates and the supernatant was passed through a 0.22 µm filter and collected. 10 mM imidazole was added and the solution was loaded onto 5 ml pre-equilibrated Ni-NTA resin (Thermo Scientific) on a gravity column. The column was washed with 10 column volumes of 50 mM sodium phosphate, pH 6.5, 500 mM NaCl, 20 mM imidazole, followed by 10 column volumes denaturing buffer (50 mM sodium phosphate, pH 6.5, 500 mM NaCl, 20 mM imidazole, and 3 M GdnHCl). Next, GdnHCl was removed by washing the column with 10 column volumes of the non-denaturing buffer. A protease solution (9.5 ml lysis buffer, 2 mM TCEP, 500 µl His_tag-TEV protease (Genscript) was added to the column to release DDX4N1 overnight at room temperature with gentle mixing. TEV protease remained bound, and the eluted DDX4N1 was concentrated using a 10 kDa centrifugal filter (Amicon ultra, Merck, Germany) to 2 ml. The concentrated DDX4N1 was SEC purified on a HiLoad Superdex 75 16/600 column. Pooled fractions were flash frozen in liquid nitrogen before storing at −80 °C. The protein purity was checked at different stages of purification using SDS-PAGE. The DDX4N1 concentrations were measured using by NanoDrop Lite (Thermo Scientific, USA) using molar extinction coefficients of 29350 M⁻¹.cm⁻¹ determined from ProtParam (Expasy, Switzerland).

Fluorophore labeling of proteins

In this work, we have used Alexa488-140C-α-Syn, N-terminally Alexa488 labeled WT and ΔC1–125 α-Syn, Alexa488-ssDNA, and DDX4N1-YFP as fluorescent reporter molecules. Alexa488-ssDNA was purchased from a commercial supplier as mentioned in the materials section. DDX4N1-YFP was recombinantly produced in E.coli as a fusion protein²⁹. For Alexa488-140C-α-Syn production, 13.7 mg/ml 140C-α-Syn was injected into a Superdex 200 increase, 10/300 GL column to remove DTT and EDTA. The fraction containing 140C-α-Syn was pooled at a concentration of 5.8 mg/ml. Alexa488 C-5 maleimide was dissolved in DMSO (1 mg in 200 μl) which corresponds to a concentration of ~7 mM. The labeling reaction was performed at 25 °C for 1 h, with 360 μM 140C-α-Syn and 10 fold excess (3.6 mM) dye concentration. The total reaction volume was 500 μl. After conjugation, the free dye was removed using Superdex 200 increase, 10/300 GL column and the concentration of the labeled protein was calculated according to absorbance values at 275 nm (140C-α-Syn) and 488 nm (Alexa488 C5 maleimide) with the help of NanoDrop Lite (Thermo Scientific, USA), and using 5960 M⁻¹.cm⁻¹ as the molar extinction coefficient of the protein. N-terminal Alexa488-WT/ΔC1–125 α-Syn were produced using labeling kits provided by the dye supplier: FIDAbiosystems ApS (Denmark). Here, the Alexa488-NHS labeling reaction was performed in sodium phosphate buffer (pH 7) instead of bicarbonate buffer (pH ~9, used to label the lysine side chain amines) to specifically target the primary amine group at the N-terminal of the protein.

In vitro phase separation assays

TDIPS experiments with α-Syn

Taylor dispersion induced phase separation (TDIPS) is an in-house developed, high-throughput capable phase separation screening platform based on the FIDA1 instrument (FIDAbiosystems ApS, Denmark)³⁹. This method was used to chart the orthogonal phase diagrams (as a function of NaCl and protein concentration) of α-Syn variants. A small plug (tens of nl) of α-Syn (different concentrations, with 25% (w/v) PEG-8000, in 20 mM sodium phosphate buffer, pH 7.4) was injected into the FIDA1 capillary (75 μm diameter, 100 cm length). This plug was sandwiched between the same buffer (20 mM sodium phosphate buffer, pH 7.4, 25% (w/v) PEG-8000) containing different concentrations (0–250 mM) of NaCl/KCl to induce phase separation. The experimental parameters are described in the following table. The experiments were conducted at room temperature (25 °C).

In the TDIPS traces, condensates (as fluorescent spikes) populated the trace at a time window from 3 to 4 min (Supplementary Fig. 1). The extent of phase separation was calculated by multiplying the number of detected spikes (n) to the average intensity of the spikes (I_avg, reflecting the average size/volume of condensates). A threshold of 15% above the baseline signal was selected to find spikes in the TDIPS traces. The spike number and average spike intensity was calculated using a bult-in peak analyzer tool in Origin Pro software (Origin Labs, USA).

Capflex experiments with α-Syn

Capillary flow experiments (Capflex²⁹) is another in-house developed phase separation platform based on the FIDA1 instrument (FIDAbiosystems ApS, Denmark) where we can measure dilute phase protein concentrations (C_dil) of a phase separating system within a capillary, without physically separating the dense and the dilute phases. Here, we quantify the relative decrease of the baseline fluorescence signal upon phase separation, compared to a non-phase separated sample of identical concentration of protein (Supplementary Fig. 1). Presence of condensates can also be confirmed by emergence of fluorescence spikes—each one corresponding to an individual condensate passing the detector, and the spike intensity scales with the size/volume of the condensate. A calibration curve is generated from the baseline fluorescence values of known protein concentrations (without phase separation), which is used to determine the C_dil of a phase separating solution. Using Capflex, we measured the C_dil of different α-Syn variants under a range of solution conditions. We induced phase separation in the presence of 20% (w/v) PEG-8000, 20 mM sodium phosphate buffer (20 mM Tris HCl, for CaCl₂), at different NaCl/KCl concentrations and pH. We quantified C_dil using 20 nM Alexa488-140C-α-Syn as a fluorescent reporter. Unlike TDIPS where phase separation occurs within the capillary, in Capflex, we induced phase separation by pre-mixing the necessary components in 1 ml glass vials (FIDAbiosystems ApS, Denmark) and/or 96 well plates, which are compatible with the FIDA1 instrument and immediately injected the samples within the capillary for detection. The parameters used for these experiments are provided in the following table. All experiments were performed at 25 °C.

We analyzed all datasets using Origin Pro software (Origin Labs, USA).

Technical note

Phase separated α-Syn samples, due to formation of clumped condensates and aggregates, often clogged the capillary in FIDA1. To reliably and reproducibly estimate the C_dil with Capflex experiments, we advise to spin down the condensates at 10,000 × g, 23 min, at 25 °C, and measure the fluorescence baseline of the supernatant. This is also the reason why the appearance of spikes (corresponding to condensates) are sometimes less apparent in our Capflex traces (Supplementary Figs. 4, 5).

TDIPS experiments with RP₃/ssDNA

The RP₃/ssDNA TDIPS experiments were performed using a triple-plug strategy, where a small plug of RP₃ was placed between two identical ssDNA containing plugs (same volume as the RP₃ plug). This triple-plug was sandwiched between a buffer solution of identical ionic strength and pH of the plugs. We performed all experiments with 10 mM Tris-HCl buffer, pH 8.0, in presence of 150 mM NaCl. α-Syn variants were titrated in the ssDNA plugs. The experimental parameters are described in the following table. The experiments were conducted at 20 °C.

In the absence of and at low stoichiometric ratios of α-Syn, RP₃ and ssDNA underwent electrostatic coacervation, giving rise to a splitting of an otherwise Gaussian profile of the detected ssDNA signal. The splitting of the Gaussian profile stems from mass transfer due to phase separation because condensates that form in front and at the back of the plug have essentially zero diffusivity compared to the monomeric peptide and DNA molecules³⁹. The maximum peak intensity without phase separation was selected as a reference point and the area underneath the peak(s) on either side (A₁ and A₂, Supplementary Fig. 16) were integrated using a bult-in peak analyzer tool in Origin Pro software (Origin Labs, USA). The ratio of the areas under the leading and the trailing peaks (A₁/A₂) was used as an indication of phase separation as a function of α-Syn concentrations relative to RP₃.

Capflex experiments with DDX4N1

For DDX4N1 phase separation, the capillary was coated with 0.1% (v/v) Tween-20 before the beginning of every run to prevent the condensates from sticking to the walls. The experiments were carried out at room temperature (25 °C), in 10 mM sodium phosphate buffer, pH 6.5, and 50 mM NaCl. The total DDX4N1 concentration was 70 µM, with 600 nM DDX4N1-YFP as a fluorescent reporter. The C_dil of DDX4N1 was measured as a function of α-Syn concentrations (Supplementary Fig. 17). The experimental parameters are provided in the following table.

Dilute phase concentration measurements using centrifugation

Phase separation of α-Syn variants were induced at 200 μM total protein concentration, in presence of 20% (w/v) PEG-8000, in 20 mM sodium phosphate buffer, at different NaCl concentrations, and at 25 °C. α-Syn condensates are particularly difficult to spin down using centrifugation due to the presence of high PEG concentrations in the system. The centrifugal acceleration required to spin down spherical protein condensates by a distance ‘h’ is given by the following equation:

$$n.g=\frac{6\pi \eta r.h}{t.\left({\rho }_d-{\rho }_s\right)V.}$$ 3

Here, the product of ‘n’ and ‘g (acceleration due to gravity)’ denotes the centrifugal RCF value; ‘ρ_d’ is the density of protein inside the condensate; ‘ρ_s’ is the density of the solvent; ‘V’ is the volume of the condensate; ‘r’ is the radius of the condensate, ‘t’ is time, and ‘η’ is the viscosity of the solvent²². We assumed that the density of condensates to be: ρ_d ~ 1.1 g/cm³ (at 25 °C) and ρ_s = 1.01 g/cm³ (at 25 °C). Therefore, to move even the smaller α-Syn condensates having an arbitrary radius of 0.3 μm by 1 cm in a 20% (w/v) PEG-8000 solution (η = 20 mPa.s) within 2 h, required an RCF value of ~15,500 × g. We used 16,000 × g RCF for 2 h to ensure most α-Syn condensates are separated in pellets, and measured the C_dil using a NanoDrop Lite (Thermo Scientific, USA). Because of the larger volume fraction and size of condensates (visible turbidity), the experiments performed with 300 µM α-Syn were centrifuged at a lower speed: 10,000 × g RCF for 23 min, at 25 °C. The complete isolation of the dense phase was confirmed by measuring the C_dil using both Capflex and UV-absorption. These centrifugation parameters were used for experiments described in Fig. 3, Supplementary Figs. 6, 11, 12.

Dynamic/static light scattering and fluorescence spectroscopy

DLS/SLS and spectroscopy experiments (Supplementary Figs. 8, 12, respectively) were carried out using a capillary based DLS instrument (PANTA, Nanotemper, Germany) and a thermally controlled, multichannel spectrophotometer (Labbot, Sweden). For DLS/SLS measurements, 200–300 μM α-Syn variants were phase separated in the presence of 20% (w/v) PEG-8000, in 20 mM sodium phosphate buffer, pH 7.4, at different NaCl concentrations (optimal for phase separation for each variant). For DLS, the phase separated samples were centrifuged, and the dense phase was resuspended in an identical buffer without PEG—to dissolve liquid-like condensates by shifting thermodynamic equilibria. The samples were then loaded into 10 µl high sensitivity microcapillaries (Nanotemper, Germany) for DLS measurements. The DLS laser (405 nm) was set to 100% power and size analysis measurements were performed at 25 °C. Ten iterations were performed for one sample per experiment. The autocorrelation functions from the samples were used to obtain the size distribution profile by an in-built algorithm.

For SLS, the phase separated samples were dispensed in a 3 × 3 mm light path quartz glass cuvette (Hellma Analytics, Germany) and measured for 40 min at a constant temperature of 25 °C. Immediately after loading the samples, 1 M Urea (prepared in 20 mM sodium phosphate buffer, pH 7.4) was added from an 8 M stock solution to induce condensate dissolution. The dilution of the protein and PEG concentrations by the addition of the urea solution does not place the composition of the sample outside of the two-phase region of the phase diagram. We measured SLS at 90° with a 636 nm laser with a 20 s equilibration time and 30 s interval time between each measurement, over 40 min. The sample was mixed every few minutes using a micropipette to prevent sedimentation of the condensates.

For the experiments described in Supplementary Fig. 12 (panel c), 200 µM monomeric α-Syn solution (no PEG) in 20 mM sodium phosphate buffer, pH 7.4, with 200 µM ThT was sonicated in 1.5 µl microcentrifuge tubes (Eppendorf) using a no-contact ultrasound probe at 80% amplitude, 1 s on/3 s off, for a total duration of 10 min (on ice, 4 °C). 60 µl of this sonicated sample was dispensed in a 3 × 3 mm light path quartz glass cuvette (Hellma Analytics, Germany) and the emission spectrum (440–600 nm) was measured with an excitation wavelength of 405 nm. Subsequently, 1 µM of pre-formed amyloid fibrils were added to the same sample and the emission spectrum was measured again. This was to confirm fibril binding ability of ThT was retained after sonication.

Thioflavin-T aggregation assays

Ensemble ThT aggregation kinetics of α-Syn condensate solutions

ThT kinetics experiments described in Supplementary Fig. 10 were performed with 300 μM α-Syn variants, 20% (w/v) PEG-8000, in 20 mM sodium phosphate buffer, pH 7.4, and at 150 mM NaCl for full-length, and at 0 mM NaCl for truncated variants. The samples were dispensed in low-binding, clear bottom 96 well plates (Thermo Scientific, USA) and measurements were conducted under quiescent conditions, at 25 °C, with the help of a fluorescence microplate reader: FLUOstar Omega (BMG Labtech, Germany). The surrounding wells were filled with MQ water to prevent evaporation. Each sample contained monomer equivalent concentration of ThT (i.e., 300 μM). ThT fluorescence over time was monitored every 15 minutes for ~7 days, by exciting the samples at 440 nm and recording the emission at 482 nm. Notably, 0.05% (w/v) sodium azide was added to all samples to prevent bacterial/fungal contamination over long incubation periods.

Estimation of templating/elongation competent fibril in α-Syn condensate solutions

Phase separated solutions (120 µl) were prepared with 300 μM α-Syn variants, 20% (w/v) PEG-8000, in 20 mM sodium phosphate buffer, pH 7.4, and at different NaCl concentrations (optimal for their respective phase separation; i.e., 150 mM for all variants and 0 mM for truncated variants only). From the soluble fraction measurements (Fig. 3b), we identified the time points for each variant where the dilute phase monomers are half-depleted (t_1/2). This was used as a standardized reference point to compare the seeding potential across variants. At this time point, the condensates were sonicated (100% amplitude, 1 sec ON, 1 sec OFF, total of 4 min, 2 times) using a no-contact ultrasound probe (VialTweeter). The condensates were then incubated for 60 min at 25 °C and pelleted by centrifugation (10,000 × g, 23 min, at 25 °C). A fixed concentration of α-Syn monomer (40 µM, no PEG) was added to each pellet (respective monomers for each variant). The pellets were dissolved and the solutions were sonicated again to release and homogenize amyloid species capable of seeding the monomers. The samples were then serially diluted with solutions containing identical concentrations of the monomers (40 µM, no PEG), and ThT fluorescence assay was used to monitor fibril elongation using FLUOstar Omega (BMG Labtech, Germany). We would like to emphasize that, through careful optimization, we identified a critical experimental consideration necessary for reliable detection of seeding-competent fibrils from α-Syn condensates. Reproducible detection of seeding was only achieved when condensates were first sonicated prior to centrifugation—releasing the seeds into the bulk solution (Supplementary Fig. 12). The resulting mixture was then centrifuged to pellet all aggregated species, followed by a second sonication step after the addition of monomers to ensure homogeneous seed dispersal. This two-step sonication protocol (Supplementary Fig. 12) proved essential for quantifying seeding potential across α-Syn variants.

Microscopy

Microscopic observation of phase separated condensates

For α-Syn, condensate solutions containing 200–300 μM α-Syn variants, 20% (w/v) PEG-8000, in 20 mM sodium phosphate buffer were prepared at different pH and NaCl concentrations. 100 nM Alexa488-140C-α-Syn was used as a fluorescent reporter molecule. For RP₃/ssDNA, phase separated coacervates were prepared by mixing 200 μM RP₃ with 20 μM ssDNA, and as a function of WT and ΔC1–125 α-Syn. These experiments were performed in 10 mM Tris-HCl buffer, pH 8.0, in the presence of 150 mM NaCl. 100 nM Alexa488 ssDNA or N-terminally labeled Alexa488-WT and Alexa488-ΔC1–125 α-Syn were used as fluorescent reporters (as mentioned for different experiments). For DDX4N1, 70 µM protein was phase separated by rapidly diluting a 700 µM stock solution (at 500 mM NaCl) tenfold to 50 mM NaCl, in 10 mM sodium phosphate buffer, pH 6.5. 100 nM DDX4N1-YFP was used as a fluorescent reporter. To visualize α-Syn Pickering clusters on DDX4N1 condensates, 100 nM of Alexa488-140C-α-Syn was used as a fluorescent reporter. All experiments described above were visualized with an LMI-005-Confocal Microscope SP8 (Leica Microsystems, Germany). Images were acquired immediately after dispensing 5 μl phase separated solution to 15 μl PDMS wells on glass coverslips, with a 63X (oil immersion) objective at a resolution of either 512/512 or 1024/1024 pixels, and at 16 bit-depth. The excitation wavelength was set at 480 nm and images were captured with an emission range of 500–700 nm. The laser exposure was adjusted for individual sample so that maximum number of condensates/clusters could be detected. All experiments were performed at 25 °C.

Fluorescence recovery after photobleaching

FRAP experiments with α-Syn and DDX4N1 condensates were performed and analyzed according to previously established protocols^21,45. Briefly, a bleaching radius of 1–3 µm was chosen depending on the α-Syn/DDX4N1 condensate size. The condensates were photobleached with a 488 nm laser at 100% power for 2 s. The fluorescence recovery was recorded at a 371 ms frame rate for 20–90 s (post-bleach) and subsequently corrected for the effect of passive bleaching and the background fluorescence. The post-bleach fluorescence intensity values were normalized with respect to the pre-bleach fluorescence for individual condensates. All experiments were carried out at 25 °C. The images were processed and analyzed using ImageJ (NIH, USA). 100 nM of Alexa488-140C-α-Syn and DDX4N1-YFP were used as a fluorescent reporter, for α-Syn and DDX4N1 condensates, respectively. All data were plotted and statistical significance was tested using Origin Pro (Origin Labs, USA).

Condensate fusion analysis

Fusion events of α-Syn and DDX4N1 condensates were recorded using a wide field Zeiss Axio vert A1 microscope (Zeiss, Germany) in bright-field mode, with a 40× objective lens. A resolution of 1626/1236 pixels with a frame rate of 50 ms were chosen. Images were acquired immediately after dispensing 5 μl phase separated solution to 15 μl PDMS wells on glass coverslips. DDX4N1 fusion events were obtained from phase separated samples at a concentration of 70 µM, in 10 mM sodium phosphate buffer, pH 6.5, 50 mM NaCl (in the absence and presence of 5 µM α-Syn). Fusion events of WT and ΔC1–110 α-Syn condensates were recorded in phase separated solutions containing 200 μM α-Syn variants, 20% (w/v) PEG-8000, in 20 mM sodium phosphate buffer, pH 7.4, at optimal NaCl concentrations. We noted that fusion events were rare in phase separated solutions and could only detect tens of events over 3 independent experiments. This was likely because of the high viscosity of the solution (slower diffusion) due to presence of 20% (w/v) PEG-8000. In general, timescales of fusion events of liquid-like condensates are used to determine the ratio between the viscosity (η) and the surface tension (γ) of the condensates—known as the inverse capillary velocity (ν = η/γ)^47,48. When two spherical condensates of similar sizes are in contact (as in our case), a hypothetical ellipse is drawn which has a major axis of D₀ (D₀/2) being the diameter of each condensate just before fusion, at time=0). Let the minor axis of this ellipse be S₀. During fusion, the aspect ratio of the ellipse, i.e., the value of D₀/S₀ exponentially decays with the relaxation time, reaching a value of 1 after complete relaxation into a larger condensate. Fitting this data to a mono-exponential decay function is used to calculate the characteristic time (τ, exponential constant). For each condensate pair undergoing fusion, the geometric mean diameter, or the lengthscale (l₀) of fusion (at time=0) is defined with the following equation:

$$l_0=\sqrt{S_0\left(D_0-S_0\right)}$$ 4

Plotting l₀ as a function of τ for a series of fusion events, and fitting the data with a linear equation is then used to calculate ν, because τ/l₀ = η/γ = ν. In our case, we used a simplified, comparative analysis where we considered only the change in D₀ as a function of relaxation time—the aspect ratio being approximately proportional to D₀, and the l₀ being proportional to $\sqrt{D_0.}$ The distances are measured along the diameters of fusing condensates, i.e., from one end of a condensate to the other end of the other condensate. End-to-end distances of detected fusion events as a function of relaxation time were fitted with a mono-exponential decay function:

$$D\left(t\right)=D_0{e}^{-t/{\tau }_{app}}+C$$ 5

Here D(t) corresponds to distance at time=t, D₀ corresponds to distance at time=0, τ_app is the apparent time constant of fusion events (τ_app = kτ, where k is a proportionality constant relating D₀ and the aspect ratio). C is the long-term asymptote, i.e., the value of D(t) when t approaches infinity. We then plotted τ_app as a function of $\sqrt{D_0}$ to calculate the apparent inverse capillary velocity (ν_app) for WT and ΔC1–110 α-Syn condensates. The images were analyzed using ImageJ (NIH, USA) and Origin Pro (Origin Labs, USA).

Nucleic acid partitioning within DDX4N1 condensates

The PDMS wells on glass coverslips²² were filled with 6 µl of unlabeled 70 µM phase separated DDX4N1 solution prepared in 10 mM phosphate buffer, pH 6.5, 50 mM NaCl. Subsequently, 1 µl of 500 nM Alexa488-ssDNA was dispensed to the well to monitor partitioning of labeled DNA within unlabeled DDX4N1 condensates (in the presence and absence of 5 μM α-Syn variants: WT, ΔC1–125, ΔC1–110). The samples were visualized with an LMI-005-Confocal Microscope SP8 (Leica Microsystems, Germany) with a laser power of 5%, gain of 800, PMT gain of 250, and a framerate of 1.2 s/frame. The data were analyzed using ImageJ (Fiji, NIH, USA). The fluorescence intensity from each condensate was averaged and background corrected for N ~ 50 individual condensates at each time point, and the values were plotted as a function of time (Supplementary Fig. 20). The slopes were calculated using linear fitting of the data using Origin Pro (Origin Labs, USA).

Transmission electron microscopy (TEM)

TEM imaging of condensate derived amyloid fibrils of α-Syn variants was performed following previously established protocols²². The samples were aged for 7 days at 25 °C under optimal phase separating conditions and subsequently sonicated as previously described. Before imaging, the samples were diluted to 50 μM total protein concentration (monomer equivalent) using 20 mM sodium phosphate buffer, pH 7.4, and 10 μl of these diluted solutions were drop casted onto formvar coated TEM grids (Sigma, USA). The grids were left for drying at room temperature for 10 min. The excess solution was blotted carefully using a Whatman filter paper without touching the grid. Next, 10 μl 1% (w/v) uranyl acetate (prepared in MQ water) was drop casted on the grids and the samples were negatively stained for 30 s. Excess dye was blotted using filter paper and the grids were left for drying at room temperature for 20 min. The grids were subsequently imaged using a 200 kV Tecnai T20 G2 electron microscope (FEI, USA) at desired magnification (10,000×). The images were captured using a TVIPS XF415 CMOS 4 K camera and TVIPS EMplify v0.4.5software (50–100 ms exposure time depending on the sample).

Atomic force microscopy (AFM)

The phase separated solutions were aged for 7 days at 25 °C under optimal phase separating conditions. The samples were centrifuged at 16,000 × g RCF for 30 min at 25 °C, and the pellets were diluted by buffer to 2–5 µM monomer equivalent concentration. The samples were sonicated as previously described, and 25 µl of the solution was deposited onto freshly cleaved mica substrates. Following 2 min of incubation, the substrates were cleaned extensively with MQ water and dried under nitrogen gas flow. All fibrils were imaged in tapping mode in air using a DriveAFM (Nanosurf, Liestal, Switzerland) using PPP-NCLAuD cantilevers (Nanosensors, Neuchatel, Switzerland). The scanning speed was 1 s per line, 1000 points per line.

Cellular toxicity

MTT and propidium iodide assay on SH-SY5Y cells

Condensate solutions containing 300 μM α-Syn variants, 20% (w/v) PEG-8000, in 20 mM sodium phosphate buffer (no azide), pH 7.4, and at different NaCl concentrations (optimal for phase separation; i.e., 150 mM for full-length and 0 mM for truncated variants) were prepared, and aged in 1.5 ml microcentrifuge tubes (Eppendorf) for 24 h at 25 °C to allow sufficient protein aggregation. The condensate solutions were sonicated to allow propagation of the condensate derived amyloid fibrils primarily via elongation mechanisms—consuming the dilute phase monomers. Sonication was performed (100% amplitude, 1 sec ON, 1 sec OFF, total of 4 min, 2 times) using a no-contact ultrasound probe (VialTweeter). The condensate solutions were subsequently aged for 7 days to allow amyloid conversion of most monomeric protein originally present in the dilute phase. Finally, the samples were centrifuged (16,000 × g RCF, 1 h, at 25 °C), and the supernatant concentrations were measured with UV-absorption (NanoDrop Lite, Thermo Scientific, USA). The monomer equivalent protein concentration in the pellet fractions were normalized to 50 µM by adding appropriate volume of buffer in each case. The de novo, pre-formed fibrils (PFFs, for comparison) were prepared at identical protein concentrations and ionic strengths as the condensate solutions, but without PEG. The fibril formation was accelerated by shaking the solutions (100 µl total volume) at 700 rpm for 7 days, at 37 °C.

For MTT assay, undifferentiated SH-SY5Y cells were seeded at 25,000 cells/well in a cell culture grade 96-well plate. The fibrils made from α-Syn variants and corresponding α-Syn variant monomers were diluted to 5 μM in cell media (DMEM GlutaMAX, high glucose with 10% (v/v) Foetal bovine serum (FBS), 1% (v/v) Penicillin/Streptomycin, and 1% (v/v) non-essential amino acids (NEAA). The PFFs and condensate-derived fibrils The PFFs and condensate-derived fibrils were sonicated with the probe sonicator UP100H (Hielscher, Germany) at 20% Amplitude with 50% on/off pulse for 3 × 30 s with 30 s breaks in-between to avoid overheating. This was done to fragment amyloid fibrils and increase their cytotoxic potential just before adding to the cells. After 96 h, the medium was removed and thiazolyl blue tetrazolium bromide (MTT, Sigma, USA) was diluted to 0.5 mg/ml in cell medium and added to the cells. After three hours, MTT solvent (20% (w/v) SDS in 0.02 M HCl) was added to the wells and incubated overnight at 37 °C. The next day, absorbance at 570 nm was measured using a plate reader. The background absorbance of MTT without cells was subtracted and the values were normalized to the PBS control condition of each plate, which was set as 100%.

For PI assay, undifferentiated SH-SY5Y cells were seeded at 25,000 cells/well in a cell culture grade 96-well plate. The next day, PFFs and the condensate-derived amyloid fibrils/aggregates made from α-Syn variants were diluted to 10 μM in cell media (DMEM GlutaMAX (Gibco, USA), high glucose with 10% (v/v) Fetal bovine serum (FBS), 1% (v/v) Penicillin/Streptomycin, and 1% (v/v) non-essential amino acids (NEAA) in 1.5 ml microcentrifuge tubes (Greiner, Austria). The PFFs and condensate-derived fibrils were sonicated with the probe sonicator UP100H (Hielscher, Germany) following identical protocol discussed before. The fibrils were added to the cells and diluted 1:1 in cell media resulting in a final concentration of 5 μM fibrils. Each treatment condition was done in triplicates. A buffer control was included, containing the same PBS that the monomer and fibrils stock were resuspended in. After 96 h, the cells were detached using Trypsin/EDTA (Sigma, USA). The cells were transferred into a 96-well V-bottom plate and washed twice with FACS buffer (PBS + 2% FCS + 2 mM EDTA). The cells were resuspended in 1 μg/ml Propidium Iodide (Bio Legend, USA) in FACS buffer and then measured on the CytoFLEX flow cytometer (Beckman Coulter, USA), with 10,000 events being recorded per well. The subsequent analysis was done using the CytExpert software (Beckman Coulter, USA). The acquired events were gated for the general cell population to exclude debris (SSC-A vs FSC-A), with a second gate for single cells to exclude doublets (SSC-A, SSC-H). Finally, the single cells were gated for Propidium Iodide-positive cells (ECD channel) and quantified as a percentage of the parent population. This was done for three separate passages of the cells, with the mean of each triplicate measurement taken for analysis.

Uptake of labeled fibrils into SH-SY5Y cells

Alexa488-140C-α-Syn labeled WT fibrils (condensate-derived and PFFs) were prepared using the same protocol as described before, but with 1 µM of labeled protein (300 µM total protein) added to each sample. Undifferentiated SH-SY5Y cells were seeded at 25,000 cells/well in an 18-well chambered glass bottom coverslip (ibidi, Germany). The next day, labeled AF488-PFFs and condensate-derived fibrils were diluted in cell media and sonicated with the probe sonicator UP100H (Hielscher, Germany) at 20% Amplitude with 50% on/off pulse for 3 × 30 s with 30 s breaks to avoid overheating. They were then added to the cells at a final concentration of 5 μM and incubated for 96 h. The cells were then washed with PBS and fixed with 4% PFA for 20 min. Then the cells were washed three times with PBS, after which they were blocked with blocking buffer (PBS + 5% Donkey serum + 0.1% Triton-X) for an hour. Sodium Potassium ATPase antibody (Santa Cruz Biotechnology, USA) was added at 1:100 and incubated overnight to detect the membrane. The cells were washed and stained with DAPI (Sigma, USA, 1:500) and the secondary antibody Alexa Fluor® 594-AffiniPure Donkey Anti-Mouse (Jackson ImmunoResearch, USA, 1:200) for 2 h. After washing three times with PBS, the cells were imaged with the CellObserver Spinning Disk microscope (Zeiss, Germany), using an Alpha Plan-Apochromat 100× objective with a numerical aperture of 1.46. A Z-stack was generated for each channel, which goes through the cells to confirm localization of fibrils inside the cells. The exposure times were 500 ms for the AF495 channel, 1 s for the AF488 and 300 ms for the DAPI channel. Subsequent image analysis and background subtraction was done using the Zeiss Zen 3.12 software (Zeiss, Germany), with arrows and lines added using Inkscape.

Amino acid (protein) and nucleotide sequences used in this study

WT α-Synuclein: MDVFMKGLSKAKEGVVAAAEKTKQGVAEAAGKTKEGVLYVGSKTKEGVVHGVATVAEKTKEQVTNVGGAVVTGVTAVAQKTVEGAGSIAAATGFVKKDQLGKNEEGAPQEGILEDMPVDPDNEAYEMPSEEGYQDYEPEA

A30P α-Synuclein: MDVFMKGLSKAKEGVVAAAEKTKQGVAEAPGKTKEGVLYVGSKTKEGVVHGVATVAEKTKEQVTNVGGAVVTGVTAVAQKTVEGAGSIAAATGFVKKDQLGKNEEGAPQEGILEDMPVDPDNEAYEMPSEEGYQDYEPEA

H50Q α-Synuclein: MDVFMKGLSKAKEGVVAAAEKTKQGVAEAAGKTKEGVLYVGSKTKEGVVQGVATVAEKTKEQVTNVGGAVVTGVTAVAQKTVEGAGSIAAATGFVKKDQLGKNEEGAPQEGILEDMPVDPDNEAYEMPSEEGYQDYEPEA

G51D α-Synuclein: MDVFMKGLSKAKEGVVAAAEKTKQGVAEAAGKTKEGVLYVGSKTKEGVVHDVATVAEKTKEQVTNVGGAVVTGVTAVAQKTVEGAGSIAAATGFVKKDQLGKNEEGAPQEGILEDMPVDPDNEAYEMPSEEGYQDYEPEA

A53T α-Synuclein: MDVFMKGLSKAKEGVVAAAEKTKQGVAEAAGKTKEGVLYVGSKTKEGVVHGVTTVAEKTKEQVTNVGGAVVTGVTAVAQKTVEGAGSIAAATGFVKKDQLGKNEEGAPQEGILEDMPVDPDNEAYEMPSEEGYQDYEPEA

ΔC1–125 α-Synuclein: MDVFMKGLSKAKEGVVAAAEKTKQGVAEAAGKTKEGVLYVGSKTKEGVVHGVATVAEKTKEQVTNVGGAVVTGVTAVAQKTVEGAGSIAAATGFVKKDQLGKNEEGAPQEGILEDMPVDPDNEAY

ΔC1–110 α-Synuclein: MDVFMKGLSKAKEGVVAAAEKTKQGVAEAAGKTKEGVLYVGSKTKEGVVHGVATVAEKTKEQVTNVGGAVVTGVTAVAQKTVEGAGSIAAATGFVKKDQLGKNEEGAPQE

Core α-Synuclein: AGKTKEGVLYVGSKTKEGVVHGVATVAEKTKEQVTNVGGAVVTGVTAVAQKTVEGAGSIAAATGFVKKDQLGKNEEGAPQE

C_rev α-Synuclein: MDVFMKGLSKAKEGVVAAAEKTKQGVAEAAGKTKEGVLYVGSKTKEGVVHGVATVAEKTKEQVTNVGGAVVTGVTAVAQKTVEGAGSIAAATGFVKKDQLGKNEEGAPQEGILEKMPVDPKNEAYKMPSEKGYQDYKPEA

A140C α-Synuclein: MDVFMKGLSKAKEGVVAAAEKTKQGVAEAAGKTKEGVLYVGSKTKEGVVHGVATVAEKTKEQVTNVGGAVVTGVTAVAQKTVEGAGSIAAATGFVKKDQLGKNEEGAPQEGILEDMPVDPDNEAYEMPSEEGYQDYEPEC

DDX4N1 CtoA: GMGDEDWEAEINPHMSSYVPIFEKDRYSGENGDNFNRTPASSSEMDDGPSRRDHFMKSGFASGRNFGNRDAGEANKRDNTSTMGGFGVGKSFGNRGFSNSRFEDGDSSGFWRESSNDAEDNPTRNRGFSKRGGYRDGNNSEASGPYRRGGRGSFRGARGGFGLGSPNNDLDPDEAMQRTGGLFGSRRPVLSGTGNGDTSQSRSGSGSERGGYKGLNEEVITGSGKNSWKSEAEGGES

DDX4N1-YFP: His_tag-GST-TEV_site-Ddx4n1-YFP: [HisGSTTEVsiteDdx4n1YFP]

HHHHHHMSPILGYWKIKGLVQPTRLLLEYLEEKYEEHLYERDEGDKWRNKKFELGLEFPNLPYYIDGDVKLTQSMAIIRYIADKHNMLGGCPKERAEISMLEGAVLDIRYGVSRIAYSKDFETLKVDFLSKLPEMLKMFEDRLCHKTYLNGDHVTHPDFMLYDALDVVLYMDPMCLDAFPKLVCFKKRIEAIPQIDKYLKSSKYIAWPLQGWQATFGGGDHPPKSDLVPRGSPGIHRDENLYFQGGAMGSNMGDEDWEAEINPHMSSYVPIFEKDRYSGENGDNFNRTPASSSEMDDGPSRRDHFMKSGFASGRNFGNRDAGECNKRDNTSTMGGFGVGKSFGNRGFSNSRFEDGDSSGFWRESSNDCEDNPTRNRGFSKRGGYRDGNNSEASGPYRRGGRGSFRGCRGGFGLGSPNNDLDPDECMQRTGGLFGSRRPVLSGTGNGDTSQSRSGSGSERGGYKGLNEEVITGSGKNSWKSEAEGGESSDTQGPKVTLQMVSKGEELFTGVVPILVELDGDVNGHKFSVSGEGEGDATYGKLTLKFICTTGKLPVPWPTLVTTFGYGLMCFARYPDHMKQHDFFKSAMPEGYVQERTIFFKDDGNYKTRAEVKFEGDTLVNRIELKGIDFKEDGNILGHKLEYNYNSHNVYIMADKQKNGIKVNFKIRHNIEDGSVQLADHYQQNTPIGDGPVLLPDNHYLSYQSKLSKDPNEKRDHMVLLEFVTAAGIT

RP₃: RRASLRRASLRRASL

ssDNA: 5’-TTT TTC CTA GAG AGT AGA GCC TGC TTC GTG G-3’

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

S.R., A.K., S.H., C.C, K.S., A.F., and K.M. designed and performed experiments and analyzed data. L.K.K. expressed and purified α-Syn proteins. A.F. expressed and purified Crev α-Syn. S.H., C.C., and S.R. purified DDX4N1. K.S., K.M., F.R.M., and C.G. conceptualized and performed all in cell experiments. A.K.B. acquired funding, conceived and supervised the study, designed experiments and analyzed data. S.R., A.K., S.H., C.C, K.S., C.G., and A.K.B. wrote the manuscript. S.R. and S.H. prepared all schematics and illustrations. All authors commented on the manuscript and approved it.

Peer review

Peer review information

Nature Communications thanks Sasanka Chakrabarti and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

Funding from Novo Nordisk Foundation (grant NNF19OC0055625) for the infrastructure “Imaging microbial language in biocontrol (IMLiB)” is acknowledged. A.K.B. would like to acknowledge funding through an ERC CoG (101088163 EMMA) for funding, as well as funding from the Novo Nordisk Foundation (grant NNFSA170028392). S.R. would like to acknowledge Horizon MSCA individual postdoctoral fellowship (Grant number 110361). S.R. and C.C. would like to acknowledge a Lundbeck foundation fellowship (Grant number 116392) for funding. A.K. would like to acknowledge Horizon MSCA individual postdoctoral fellowship (Grant number 101106115) for funding. S.H. would like to acknowledge a Lundbeck foundation postdoctoral fellowship (Grant number R449-2023-1527) for funding. C.C. would like to acknowledge an Erasmus Masters scholarship. A.K.B. and A.F. would like to acknowledge the Michael J Fox foundation for funding. C.G. would like to acknowledge an Independent Research Fund Denmark DFF-Research Project 1 (Grant number: 3103-00220B) for funding. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.

Data availability

The authors declare that unless otherwise stated, all data supporting the results of this study can be found in the article, supplementary, and source data files. All data are analyzed using published tools and packages and are duly cited wherever applicable. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-74627-3.