De Novo Peptides That Induce the Liquid–Liquid Phase Separation of α‑Synuclein

Abstract

Liquid–liquid phase separation (LLPS) of proteins can form membraneless organelles in the cell and can lead to pathological aggregation associated with neurodegenerative diseases. However, progress in controlling LLPS has been limited, and there has been no emergence of engineered de novo molecules to induce and modulate LLPS of targeted proteins. Here, we report de novo peptides that efficiently induce the LLPS of α-synuclein (αSyn), a protein involved in Parkinson’s disease, discovered by the RaPID (random nonstandard peptides integrated discovery) system. These peptides primarily interact with the C-terminal region of αSyn, leading to the formation of an interaction network with αSyn and resulting in efficient droplet formation. Our study demonstrates the capacity of target-specific peptides to control LLPS and the subsequent liquid–solid phase transition (LSPT). Our novel LLPS-inducing and LSPT-modulating peptides may serve as a promising tool for fundamental investigations of LLPS and potentially for therapeutic intervention in amyloid diseases.

Introduction

Liquid–liquid phase separation (LLPS) of proteins accompanies the formation of a condensed liquid droplet phase, giving rise to a wide variety of cellular functions and regulatory mechanisms. − However, these membraneless organelles can also pose a risk for the formation of amyloid fibrils, which are potentially associated with serious neurodegenerative diseases. Condensates of intrinsically disordered proteins (IDPs), such as fused in sarcoma (FUS) and Tau, have been shown to undergo LLPS and subsequently form fibrils through a process known as the liquid-to-solid phase transition. − These IDPs, which are often highly charged and flexible, induce LLPS by forming weak, transient multivalent interactions between protein–protein and protein-RNA molecules, contributing to the dynamic nature of phase-separated compartments. ,−

Diseases associated with protein aggregation are likely developed or progress due to failures to maintain homeostasis in living systems. The aggregation of proteins from their native (or soluble) state to an amyloid state may generally proceed through an intermediate condensate that is typically metastable. , Therefore, the condensed liquid droplet state of amyloidogenic proteins can be a potential therapeutic target; i.e., controlling the physical properties of droplets and the subsequent liquid-to-solid phase transition can be a relevant strategy to modulate the disease-associated aggregation. However, the molecular-level understanding of how to manipulate droplet phase behavior remains limited. Moreover, to our knowledge, a target-specific de novo molecule that induces LLPS has not yet been reported in the literature, although some studies have demonstrated non-target-specific LLPS induction using small molecules, simplified model macromolecules, or polypeptides inspired by native proteins. − The development of a target-specific de novo LLPS-inducing molecule would provide valuable tools for a broad range of applications, such as drug partition into condensates of target proteins and the formation of artificial organelles from non-LLPS-inducing proteins, without interfering with natively existing condensates in the cellular system. These applications offer substantial promise for therapeutic interventions in challenging indications, which may allow the modification of IDPs and IDRs currently considered undruggable with small molecules. ,

α-Synuclein (αSyn) is an amyloidogenic IDP involved in Parkinson’s disease, − and recent reports suggest that it also undergoes LLPS, although this process requires certain purification methods and long incubation times. , The monomeric form of αSyn is intrinsically disordered and consists of three characteristic domains: a positively charged N-terminal region, a central hydrophobic nonamyloid-β component (NAC), and a negatively charged C-terminal region. Due to its highly dynamic nature, identifying molecular species that interact with αSyn through conventional enthalpy-driven binding mechanisms is challenging, as has been similarly observed with amyloid-β. − Recent NMR and cryo-EM studies have revealed the highly ordered structure of the αSyn fibril core, while the N-terminus (1–37 residues) and C-terminus (98–140 residues) remain disordered. − These findings suggest that the structured fibril core is surrounded by dense and disordered terminal regions.

Various strategies have been developed for the discovery of cyclic peptides targeting specific proteins. Among them, the random nonstandard peptides integrated discovery (RaPID) system (Figure A) uses a combination of mRNA display and genetic code reprogramming, facilitated by the flexible in vitro translation system, referred to as FIT, system. This system enables the ribosomal synthesis of extremely large (>10¹²) libraries of natural product-like macrocyclic peptides and allows rapid selection based on binding to protein targets of interest. Owing to their conformational rigidity, macrocyclic peptides typically incur a lower entropic penalty upon binding and can thus achieve high binding affinity and specificity. − We therefore hypothesized that this approach would be effective in identifying macrocycles capable of interacting with the disordered regions of target proteins.

1.

Selection of macrocyclic peptides that interact with the amyloid fibrils of αSyn. (A) Schematic overview of the RaPID system for macrocyclic peptides. A mRNA library containing a random sequence domain, (NNK)_6–15, was transcribed from the corresponding cDNA library and was conjugated with an oligonucleotide bearing puromycin. The resulting mRNAs were translated by the FIT system in the presence of the appropriate aminoacyl-tRNAs prepared by flexizymes. Linear peptides displayed on the individual mRNAs were spontaneously cyclized after translation, and the resulting macrocyclic peptides were displayed. After reverse transcription, the peptide libraries were subjected to αSyn fibrils immobilized on magnetic beads and binding species were isolated. The cDNAs on binding mRNA-peptide fusion were recovered and amplified by PCR. (B) Distribution of charged residues in the αSyn sequence. Negatively and positively charged residues at neutral pH are shown by red and blue bars, respectively. The fibril core region in the fibrils is signified by gray rectangles. (C) A schematic depiction of the typical structure of amyloid fibrils of αSyn. The arrangement of the two protofilaments is illustrated as a cross-section of the fibril core. (D) AFM image showing the uniform morphology of αSyn fibrils prepared by six rounds of seeding. (E) Peptide sequences were identified from the pool in round six. The dissociation constant (K _D) and molar binding ratio of each peptide to αSyn were elucidated by ITC.

Here we report the discovery of de novo peptides that induce LLPS of αSyn in vitro. Five out of seven selected peptides from the RaPID screening were able to induce LLPS of αSyn. NMR studies of αSyn revealed that one peptide, FD1, primarily interacts with the C-terminal region of αSyn, allowing us to propose a mechanism by which target-specific LLPS induction occurs. By monitoring the fluidity of αSyn droplets, we demonstrate that FD1 and its variants can modulate the physical nature of αSyn droplets and the subsequent liquid-to-solid phase transition of αSyn condensates. These LLPS-inducing and phase-transition-modulating peptides represent a promising tool for fundamental studies of αSyn LLPS and may offer valuable insights into how the LLPS-mediate fibril formation of disease-related proteins can be regulated.

Results

RaPID Selection of Macrocyclic Peptides against Fibril States of αSyn

In order to obtain peptides capable of inducing the LLPS of αSyn, we conducted selection using the RaPID system with αSyn amyloid fibrils as the displaying target (Figure A). As noted earlier, recent structural studies on αSyn have revealed that approximately 40 residues at both the N- and C-termini of αSyn remain flexible and form a dense mesh of disordered regions surrounding the fibril core (Figure B,C), commonly referred to as “fuzzy coat”. By using αSyn fibrils as the selection target, we were able to exclude the NAC region and focus on these highly charged and flexible terminal regions. Screening against the fibrillar state of αSyn, where the disordered regions are spatially aligned at narrow intervals (∼4.8 Å), may facilitate the identification of peptides capable of bridging multiple αSyn molecules through multivalent interactions, thereby inducing LLPS.

A puromycin-ligated mRNA library was constructed to encode macrocyclic peptides with N-chloroacetyl-l-tyrosine (^LY-library) or N-chloroacetyl-d-tyrosine (^DY-library) as an initiator, followed by a random peptide region consisting of 6–15 residues, a cysteine and ending with a short linker sequence. Upon translation of the mRNA library, each mRNA template was ligated to the C-terminus of the cognate linear peptide to form an mRNA-peptide fusion via the puromycin linker, and then, cDNA synthesis was performed by reverse transcription (RT). During the RT step at 42 °C, the N-terminal chloroacetyl group completely cyclized with its downstream cysteine to form a thioether-macrocyclic peptide. Each library was applied first to protein-free magnetic beads to remove background nonspecific binders and then to the αSyn fibril-immobilized magnetic beads to enrich for binders specific to the fibril state of αSyn. The morphology of the amyloid fibrils used for the RaPID campaign was characterized by AFM, revealing a height of 9.0 nm and a periodic twisting pitch of 99 nm (Figure D), consistent with previous cryo-EM studies.

After seven rounds of RaPID selection, both the ^LY-library and ^DY-library yielded successful enrichment of peptide species selectively bound to immobilized αSyn fibrils over background binding. Sequence alignment analysis of the enriched peptides from the seventh round revealed three and five convergent classes of macrocyclic peptides from the ^LY-library and ^DY-library, respectively. Since the mRNA fusion used during selection generally enhances water solubility, we appended a short solubility-enhancing tag, GKKK-NH₂, to the C-termini of the unmodified macrocyclic peptides FL2–3 and FD1–5 to facilitate downstream studies (Figure E).

To evaluate the binding affinity and binding mode of the selected peptides, they were chemically synthesized and subjected to isothermal calorimetry (ITC) measurements (Figure S1). The dissociation constants K _D for the αSyn fibrils were in the single-digit micromolar range, and the binding stoichiometries (n value, peptide/αSyn) were significantly lower than 1.0 for FL2–3, and FD1–3, suggesting that these peptides likely interact with multiple αSyn molecules (Figure E). In contrast, FD4 and FD5 exhibited similar K _D values but had n values close to or greater than 1.0, implying a different binding mode.

Selected Peptides Efficiently Induced Droplet Formation of αSyn

To investigate whether the selected peptides can induce liquid droplet formation with αSyn, 100 μM of purified αSyn (see Materials and Methods) was mixed with 2 molar equiv of peptides in 20 mM sodium phosphate buffer (NaPi) (pH 7.5) containing 10% polyethylene glycol (PEG)-8000, commonly used as a molecular crowding agent. After incubating on ice for 1 h, differential interference contrast (DIC) images of the solution showed many spherical droplets in the presence of FL2–3 and FD1–3, whereas FD4 and FD5 showed aggregation-like assemblies rather than liquid-like droplets (Figure A). In contrast, αSyn alone did not undergo LLPS, even after 7 days of incubation (Figure S2A). We further examined the pH dependence of LLPS in the presence and absence of FD1 (Figure S2B). FD1 markedly enhanced LLPS at acidic pH, while αSyn alone at 250 μM formed droplets only at pH 5.2, and not at higher pH values. For comparison, we tested BD1-GKKK, a macrocyclic peptide previously identified by RaPID screening against αSyn monomer. DIC imaging showed no droplet formation in the presence of BD1-GKKK (Figure S3).

2.

Selected peptides induce LLPS of αSyn and are colocalized with αSyn in liquid droplets. (A) DIC images of the αSyn solution in the absence and presence of seven selected peptides after 1 h incubation on ice. All solutions were prepared in 20 mM NaPi buffer (pH 7.5) containing 10% PEG8000. (B) A series of DIC images of αSyn solution incubated with varying concentrations of FL2 (left) and FD1 (right) for 30 min at 4 °C. The scale bar on the DIC images indicates 20 μm. (C) Turbidity at 600 nm and fluorescence intensity of the supernatant measured at various concentrations of FL2 and FD1 in the absence and presence of αSyn. (D) Concentrations of αSyn and FD1 remaining in the supernatant (αSyn_sup and FD1_sup, respectively) after centrifugation at 15 300g, determined by HPLC. (E) Residual concentrations of αSyn and FD1 in the supernatant following 15 min incubation with varying concentrations of FD1. (F) Adsorption of a droplet onto the glass surface observed by DIC. The scale bars indicate 10 μm. (G) Fusion event of two droplets on the glass surface observed by confocal microscopy with the rhodamine labeled αSyn (αSyn-Rhod). (H) Fluorescence images showing the colocalization of αSyn-Rhod with fluorescein-labeled FL2 (FL2C-Fluor) and FD1 (FD1C-Fluor) within the liquid droplets.

Based on the sequence enrichment during selection, indicative of binding potency and the predicted higher solubility, we selected two peptides, FL2 and FD1, for further experiments. Droplet formation, in both size and quantity, was dependent on the concentration of peptide (10–1,000 μM) as well as αSyn (20–200 μM) (Figure B). Quantitative fluorescence measurements of the dilute-phase and turbidity measurements were also demonstrated at various concentrations of the peptides with 100 μM αSyn. These analyses consistently showed a concentration-dependent increase in droplet formation and a corresponding decrease in the amount of αSyn remaining in the dilute phase (Figure C). Within the time scale of our experiments, the presence of peptides appears to alter the thermodynamics of LLPS, since αSyn without peptides does not show LLPS even after prolonged incubation. To evaluate this further, we quantified the concentrations of αSyn and FD1 in the dilute phase by centrifugation at various incubation time points. The results showed that peptide-induced LLPS reaches an apparent equilibrium within 15 min (Figure D and Figure S2C) and that increasing FD1 concentration further reduced the residual αSyn in the supernatant (Figure E and Figure S2D). These findings suggest that the system likely achieves a new thermodynamic equilibrium in the presence of peptides, whereas in peptide-free LLPS systems, the concentration of αSyn in the dilute phase remains constant (∼100 μM in 20% PEG), regardless of the total protein concentration. The generated droplets showed liquid-like properties, as evidenced by their adsorption onto the glass plate surface (Figure F and Movie 1) and fusion events of droplets (Figure G and Movie 1), as observed by DIC and time-lapse fluorescence microscopy. Temperature-dependent turbidity assays revealed that the condensates dissociated to a dispersed monomeric state as the temperature increased. Reversibility of the formation of droplets was also verified by reforming droplets after heat dissociation by raising the temperature to 20 °C (Figure S4). Such reversibility of the assembly is a hallmark of liquid droplets and is not expected in stable solid aggregates. Phase diagrams of αSyn after different incubation periods (1 h or 1 day at 4 °C) were constructed based on morphological classification in DIC images (Figure S5), categorizing samples into monomer, droplet, aggregate, and gel states.

Fluorescence images confirmed the colocalization of both αSyn and peptides within the droplet phase (Figure H and Figure S6A, B).

To assess the sequence specificity of FL2 and FD1 toward α-synuclein, we tested sequence-scrambled peptides (FL2 _scr and FD1 _scr ) (Figure S7A). Both FL2 _scr and FD1 _scr showed significantly reduced LLPS-inducing activity, indicating that the specific amino acid sequences of FL2 and FD1 are required for interaction with αSyn and for triggering LLPS. We further evaluated the LLPS-inducing capacity of FL2 and FD1 with other intrinsically disordered proteins that undergo LLPS, as well as in the presence of a cellular extract (Figure S7B,C). The results showed that FL2 strongly promoted LLPS but lacked target specificity (Figure S7B). In contrast, FD1 selectively promoted LLPS with α-synuclein, suggesting a higher degree of target specificity. To investigate the contribution of the cyclic scaffold to LLPS induction, we compared the cyclic and linear forms of FL2 and FD1 (FL2_linear and FD1_linear, respectively; Figure S8A). The linear peptides did not exhibit significant changes in their LLPS-inducing ability in the absence and presence of DTT (Figure S8B). To probe the internal dynamics and permeability of the peptide-induced droplets, we examined whether the droplets allowed the colocalization of protein-sized molecules fused to the linear peptides.

Using sfGFP as a model cargo, fluorescence microscopy revealed that both sfGFP-labeled FL2 _Linear (FL2_Linear-sfGFP) and sfGFP-labeled FD1_Linear (FD1_Linear-sfGFP) colocalized homogeneously with αSyn within the droplet phase (Figure S8C–E).

Interactions Responsible for LLPS of αSyn and Peptides

To identify the interacting region(s) in αSyn responsible for LLPS induction by FL2 and FD1, we performed ¹H–¹⁵N heteronuclear single quantum coherence (HSQC) experiments at 5 °C with 100 μM ¹⁵N-labeled αSyn in the presence and absence of FL2 and FD1. The presence of FL2 did not cause a major chemical shift perturbation (Δδ) in the HSQC spectra, suggesting the absence of strong site-specific interactions (Figure S9A–C). In contrast, the HSQC spectra in the presence of FD1 showed significant chemical shift perturbation (Δδ) at the C-terminal region of αSyn (Figure A,B) as well as their peak intensities (I _+FD1/I _–FD1) being slightly decreased (Figure B). We further performed paramagnetic relaxation enhancement (PRE) measurements to detect a transient long-range interaction using FD1 labeled with a paramagnetic spin-label, (1-oxyl-2,2,5,5-tetramethylpyrrolidin-3-yl)­methylmethane­thiosulfonate (MTSL), referred to as FD1-C21R1. The result showed a dramatic decrease in the cross-peak intensity at the C-terminal region of αSyn (Figure C). To determine whether the observed chemical shift perturbations originated from αSyn within or outside the droplets, we performed additional measurements using centrifuged samples (Figure S10A,B). The chemical shift perturbations observed in the supernatant indicate that the spectral changes in the mixture primarily result from interactions between FD1 and αSyn in the dilute phase (Figure S10C). Although droplet formation was accompanied by peak broadening, which hindered the detailed mechanistic interpretation, the interactions observed in the dilute phase may contribute to the initial stages of condensation. Together, these results suggest that two-component LLPS can be induced through weak, nonsite-specific interactions, as exemplified by FL2; however, site-specific interactions are likely required for achieving target-specific LLPS induction.

3.

Peptides primarily interact with the C-terminal region of αSyn via electrostatic interactions. (A) ¹H–¹⁵N HSQC spectrum of 100 μM ¹⁵N-labeled αSyn monomers in the absence (red) and presence (blue) of 200 μM FD1. (B) Chemical shift differences (Δδ, top) and peak intensity ratios (I _+FD1 /I _–FD1, bottom) of αSyn in the presence of FD1 suggest the interaction of FD1 with the C-terminal region of monomeric αSyn. The positions of negatively and positively charged residues are shown by red and blue bars, respectively. (C) PRE-NMR normalized intensity ratio (I _+FD1‑C21R1 /I _–FD1) of αSyn generated using MTSL labeled FD1-C21R1 showed significant decrease at both N- and C-terminus of αSyn. Error bar represents inverse of the s/n ratio. (D) DIC images and turbidity at 600 nm show a salt concentration dependence of the formation of droplets in the presence of 100 μM FL2 and FD1. (E) Comparison of LLPS efficiency of the full-length αSyn, the charge-deleted mutants (αSyn_1–103 and αSyn_104–140), and the NAC-deleted mutants (αSyn_NPC) with 400 μM FL2 and FD1 observed by DIC images. The turbidity of solution with FD1 is shown on the right.

To investigate the importance of electrostatic interactions via highly charged N- and C-terminal regions of αSyn for LLPS, we further tested the effect of salt on LLPS. The DIC images and turbidity measurements clearly showed a decrease in the total amount of droplets as the NaCl concentration was increased (0–200 mM) (Figure D). We next evaluated the LLPS-inducing efficiency of FL2 lacking the C-terminal GKKK solubility tag (designated as FL2 _NoTag ). The deletion of solubility-tag did not reduce the efficiency of LLPS induction at 0.2–1.0 mol equivalent (Figure S11A), suggesting that the residues within the peptide core sequence identified during the selection campaign contribute to the observed LLPS of αSyn. To confirm whether the N- and/or C-terminal regions of αSyn are required for droplet formation, we prepared truncated αSyn constructs at the C-terminal region (αSyn_1–103), N-terminal region (αSyn_104–140), and NAC region (αSyn_NPC), which N- and C-terminal regions are connected by a PEG linker in place of the NAC region. LLPS assays with FL2 and FD1 (Figure E) clearly showed that neither truncated αSyn efficiently yields LLPS compared to full-length αSyn. Strikingly, αSyn_1–103 with FL2 formed aggregates rather than liquid droplets, whereas that with FD1 completely lost the ability to form neither liquid droplets nor aggregates. Notably, co-incubation of αSyn_1–103 and αSyn_104–140 in the presence of peptides did not restore LLPS, suggesting that heterotypic interaction with the N- and C-terminal regions within the same αSyn molecule may be required for such efficient LLPS events. Consistent with this, αSyn_NPC showed an improved LLPS efficiency, indicating that the NAC region is not essential for LLPS.

Liquidity of Droplet Induced by FL2 and FD1

To gain more insight into how the dynamic properties of the liquid droplets of αSyn condensates induced by the peptides can be modulated, we investigated the dynamic properties of αSyn droplets formed in the presence of selected peptides. The internal mobility of the condensates was assessed with fluorescence recovery after photobleaching (FRAP) technique using the rhodamine-labeled αSyn (αSyn-Rhod) (Figure A,B). In droplets induced by 200 μM FD1, αSyn-Rhod showed approximately 94% αSyn-Rhod fluorescence recovery after 2  min of incubation, indicating high internal fluidity. However, this liquid-like property declined in an incubation-time-dependent manner, with only ∼8% recovery observed after 1 h. These findings demonstrate that the liquid droplets induced by FD1 undergo maturation into a more solid-like state. Interestingly, the phase transition was sensitive to the temperature. While 1 h incubation at room temperature significantly reduced fluorescence recovery, samples incubated on ice maintained high fluidity (Figure A), indicating that lower temperatures slow the maturation (aging) process of the droplets. To further explore modulation of droplet dynamics, we synthesized FD1 peptides with varying lengths of lysine tags (FD1 _NoTag , FD1KKK, and FD1­(KKK) ₂ ) which affect solubility and charge distribution and may consequently alter condensate properties (Figure S11B). The results showed that longer lysine tags enhanced fluidity and resistance to maturation without altering the mode of interaction (Figure C,D and Figure S11C–F). For comparison, the fluidity of droplets formed by αSyn alone is shown in Figure S11G, although it should be noted that the pH and αSyn concentrations differ significantly in these conditions. Taken together, we have shown that the liquidity and maturation behaviors of αSyn condensates can be modulated by FD1 and its variants. In particular, while αSyn alone undergoes a liquid-to-solid phase transition within 1 h, FD1­(KKK) ₂ maintains the liquid-like state of αSyn droplets for at least the same duration.

4.

Fluidity of αSyn-Rhod in liquid droplets induced by FD1 and its variants. (A) Representative fluorescence images of αSyn-FD1 droplets during FRAP measurements at various incubation time points (2, 15, 30, and 60 min) at room temperature. (B) FRAP kinetics at each incubation time point show that the liquid droplets lose their liquidity as time passes. (C) Representative fluorescence images of droplets induced by FD1 variants. (D) Comparison of fluorescence recovery rate of αSyn-Rhod after photobleaching (130 s) at various incubation times (5, 15, 30, and 60 min) reveals that FD1 peptides with varying lengths of lysine tags modulate the fluidity and maturation propensity of the droplets. The data represent the mean ± STD for n = 3 independent experiments.

Aggregation of αSyn via LLPS Modulated by FL2 and FD1

LLPS-mediated liquid-to-solid phase transitions into pathological aggregates have been reported for various IDPs. , αSyn accumulation in the brain has been observed with coaggregation with other pathological species such as tau in Lewy body , and prion protein in early cytoplasmic inclusions bodies. Multicomponent colocalization with other interacting partners may be generated by synergistic interactions in cell bodies. Indeed, αSyn can form liquid droplets with prion protein, and their interaction promotes amyloid fibril formation. Therefore, we next explored how LLPS induced by the αSyn-peptide heterotypic interactions could affect the fibril formation of αSyn (Figure ). We performed the kinetic analysis of αSyn aggregation under three characteristic conditions: (i) conditions where αSyn does not form condensates (no PEG, PBS buffer), (ii) conditions with moderate LLPS propensity (10% PEG in PBS, condensates form only at high peptide concentrations), and (iii) conditions with high LLPS propensity (10% PEG in sodium phosphate (NaPi) buffer, condensates form even at low peptide concentrations). We monitored the spontaneous fibril formation of αSyn under these conditions (i–iii) in the absence and presence of peptides (from 0.1 to 10 mol equiv) using the fluorescence of thioflavin T (ThT) as an amyloid-sensitive probe (Figure A). The lag time (t _1/2), defined as the time to reach half-maximal ThT intensity, was compared across conditions and concentrations (Figure B). Under condition i, both FL2 and FD1 showed no acceleration of the fibril formation of αSyn but rather expanded the lag time, suggesting an inhibitory effect on early stage fibril nucleation or elongation, consistent with previous findings. Under condition ii, a reduction in both ThT fluorescence intensity and lag time was observed, while condition iii led to pronounced acceleration of αSyn aggregation, accompanied by increased ThT intensity. These data collectively indicate that the enhanced LLPS propensity, modulated by peptide binding, promotes primary nucleation of amyloid fibrils. The observed decrease in the intensity of the ThT signal at high peptide concentrations may result from a reduction in ordered β-sheet structures. Additionally, we conducted seeding experiments to confirm the effect of FL2 and FD1 on the elongation reaction of αSyn fibrils under the conditions ii and iii (Figure S12). The results showed that FL2 and FD1 significantly delayed the seed-dependent elongation reaction.

5.

Aggregation of αSyn is accelerated under conditions that promote LLPS. (A) ThT fluorescence kinetics of αSyn aggregation under three conditions: PBS without PEG-8000 (top), PBS with 10% PEG-8000 (middle), and 20 mM sodium phosphate (NaPi) with 10% PEG-8000 (bottom) (represented by different colors). (B) Logarithmic comparison of the lag time of the αSyn fibril formation in the presence of peptides to the one in the absence of the peptides (log­(t _1/2,peptide+/t _{1/2, peptide‑})) shows acceleration of aggregation under conditions promoting LLPS. Lag times of αSyn fibril formation in PBS without PEG-8000 at high peptide concentrations are not available because ThT increase was not observed within the 14-day time scale. (C) Fluorescence images of the droplets formed in the presence of 400 μM FL2 show an increase in ANS fluorescence intensity in the droplets after 3 h incubation at 37 °C. (D) Simultaneous observation of ThT (closed circle) and ANS (opened circle) fluorescence during the αSyn aggregation in PBS with 10% PEG-8000 reveals two-step fibril formation modalities, with an early LLPS and maturation phase (ANS increase colored by green area) and subsequent fibril formation (ThT increase colored by red area). (E) Fluorescence images of the aggregates after 70 h incubation in the presence of the peptides. (F) Comparison of maximum intensity of ThT (black) and ANS (red) at various peptide concentrations indicates that αSyn forms more ANS positive aggregates at high peptide concentrations, whereas the ThT intensity decreases.

Liquid-to-solid phase transitions of protein condensates can be broadly classified into two categories: amyloid fibril formation and nonamyloidogenic aggregation. Nonamyloidogenic aggregation, e.g., amorphous aggregation, is formed without a lag time; therefore, it generally competes with the slower, nucleation-dependent amyloid formation. To monitor the formation of nonfibrillar aggregates of αSyn, we simultaneously measured the fluorescence of 8-anilino-1-naphthalenesulfonic acid (ANS), which is commonly used as a probe for amorphous aggregates, although it can also respond positively to amyloid fibrils. Prior to this experiment, we first confirmed that the ANS was incorporated into the liquid droplets and showed fluorescence as time passed (Figure C). ANS fluorescence profiles monitored at conditions ii and iii indicated an early appearance of hydrophobic assemblies within the initial 10 h (Figure D and Figure S13A). The ThT profiles, however, did not show any increase until at least 10 h, showing that the early assemblies do not yet have a fibrillar structure, although final product images revealed the presence of fibrils (Figure E).

The increase in droplet formation and subsequent maturation, rather than fibril formation, likely explains the decrease of magnitude of ThT signal and increase of ANS intensity at high concentrations of FL2 and FD1 (Figure F and Figure S13B).

Discussion

Various peptide-based LLPS systems, including peptide–peptide and peptide–nucleic acids condensates, have been reported previously. The peptides reported to date can be divided into two types: mimic peptides, which contain amino acids homologous to self-assembling proteins, and simplified polyelectrolyte peptides, such as polylysine or polyarginine, whose electrostatic interactions are their main driving force. Although such peptides have not been able to target specific biomolecules, we successfully identified de novo macrocyclic peptides that induce LLPS of αSyn with a certain level of specificity by means of the RaPID system. We considered that peptides with multiple interactions with αSyn could be screened by displaying the fibril state of αSyn. Indeed, five of seven identified peptides indicated low molar binding ratios to αSyn, and they induced LLPS of αSyn. For comparison, we demonstrated that BD1-GKKK, a macrocyclic peptide identified by RaPID screening against the monomeric state of αSyn, did not induce LLPS under the same conditions. Among the selected candidates, FL2 and FD1 efficiently induced αSyn LLPS in a concentration-dependent manner and without a lag-phase, whereas αSyn alone did not form LLPS at least within a week. Although we adopted the cyclic form of peptides during RaPID selection to enhance the entropic favorability of binding and reduce incorporation into fibrillar structures, peptide cyclization appears to have only a minor effect on their LLPS-inducing ability. Nonetheless, cyclic peptides typically exhibit improved proteolytic stability and conformational rigidity, which may offer advantages for future in vivo applications.

Despite the high prevalence of disordered proteins in human disease, − no clinically approved drug directly targets disordered proteins in their monomeric forms. Here, the condensates induced by the selected peptides showed a dynamic, liquid-like nature. We also demonstrated that small molecules and proteins, such as fluorescein and sfGFP, can be incorporated into the liquid droplets by conjugating them with FL2 and FD1, as well as their linearized form. This ability of the peptides to colocalize the conjugated clients into the dynamic condensed phase of the target suggests a potential application: drug- or enzyme-conjugation with these peptides can efficiently promote their reaction activity in the droplets, as suggested in cancer therapeutics in nuclear condensates or in enhancing enzyme activity in cells. − The specificity of interaction with αSyn was proved to a certain extent by comparison with scrambled FD1, and this was further supported by the lack of activity of FD1 against other IDPs known to undergo LLPS. In contrast, FL2 did not exhibit target specificity. Microscopy and turbidity measurements imply that FL2 itself tends to undergo self-assembly and/or LLPS at high concentrations, likely due to its high content of aromatic and cationic amino acids, which favor phase separation. NMR measurements also indicated that FL2 exhibits no strong site-specific interactions, which may account for its lack of target specificity. Taken together, these results support the conclusion that the RaPID system successfully identified LLPS-inducing peptides, including both target-specific and nonspecific variants. Further optimization of peptide affinity and specificity may be necessary to minimize interference from endogenous protein–protein and protein–RNA interactions in the cell system. Such optimization could be advantageously achieved through display-facilitated deep mutational scanning, as reported elsewhere.

The NMR analyses of the condensed αSyn revealed that FD1 mainly interacts with the C-terminal region of αSyn. Key residues in FD1, H5, K6, Q12, R14, and S15, are likely major contributors to electrostatic interactions and dipole–dipole interactions with the C-terminal region of αSyn (residues 99–140), which is enriched in 15 anionic and 5 polar residues. Notably, HSQC spectra exhibited significant chemical shift perturbation (Δδ) at Y133, and Y136 in αSyn, suggesting that π–π interactions and cation−π interactions may also contribute to LLPS. The spatial arrangement and distance between these key residues in FD1 may underlie its target specificity. These interactions with the C-terminal regions of αSyn also support the notion that the RaPID system successfully screened macrocycles against the displayed fuzzy coats of αSyn fibrils. Based on the interaction analysis, we propose a tentative mechanistic model for the induction of LLPS of αSyn by FD1 as follows (Figure ). In the monomeric state of αSyn, the long-range intramolecular interaction between charged N- and C-terminal regions has led to a compact form of αSyn and maintains its solubility. − Interaction of FL2 and FD1 with the C-terminal regions of αSyn may shield the intramolecular interactions and lead to the opened state of αSyn, which is preferred to form interaction network. Consistent with this hypothesis, our deletion mutant experiments demonstrated the essential role of both the N- and C-terminal domains in peptide-induced two-component LLPS. Furthermore, the conformational opening of αSyn monomers may also promote the nucleation phase of fibril formation. In particular, exposure of the NAC region, which possesses high amyloid propensity as predicted by Zyggregator (Figure S14), and involvement of the N-terminal segment of αSyn (residues 36–42) may facilitate this process.

6.

A proposed hypothetical model of αSyn LLPS induced by FD1. Under physiological conditions, αSyn adopts a compact conformation stabilized by a long-range intramolecular interaction between the N- and C-terminal regions. The interaction FD1 with the C-terminal regions of αSyn may disrupt this interaction, inducing a more extended conformation of αSyn that is favorable for intermolecular interactions and network formation. Moreover, these peptides may facilitate the formation of multivalent interactions by bridging multiple αSyn molecules, thereby promoting the formation of a dynamic interaction network and efficient droplet formation. The structural modification of αSyn monomers to an opened state might also accelerate the nucleation of fibril formation, where the exposed NAC region has a high amyloid propensity and an N-terminal segment of αSyn (residues 36–42) is required for aggregation.

The concentration effect of molecules via LLPS can lead proteins into a metastable supersaturated state, which often precedes the formation of thermodynamically stable amyloid fibrils. ,,, This stepwise progression of fibrillar aggregation within condensates aligns with Ostwald’s rule of stages, according to which the morphologies of crystals change over time, guided by their kinetic accessibility and thermodynamic stabilities. , Therefore, liquid droplets formed by amyloidogenic proteins may represent an initial stage of amyloid formation and thus serve as attractive therapeutic targets. In this study, we demonstrated that modifications of FD1 modulate the dynamic properties of the αSyn condensates and their liquid to solid phase transition (LSPT). Aggregation assays revealed that FL2- and FD1-induced droplets undergo a phase transition to nonamyloidogenic aggregates in the lag-phase of αSyn fibrillation. The rapid formation of competitive nonfibrillar aggregated species might result in a reduction of the mass of pathological aggregates and could offer opportunities for suppressing their toxicity. Furthermore, whereas the acceleration of the nucleation step of amyloid fibril formation was accompanied by LLPS-induction, seeding experiments showed that both FL2 and FD1 inhibit the seed-dependent elongation phase. Under the conditions i where αSyn does not undergo LLPS, FL2 and FD1 suppressed αSyn fibril formation. Thus, LLPS induction might offer opportunities for colocalizing the inhibitor and promoting the inhibitory effect on fibril formation, although competitively prompting the inhibitory effect on the aggregation was dominant in this study. Overall, these LLPS-inducing and LSPT-modulating peptides would become promising tools for fundamental investigations of LLPS and may represent novel leads for therapeutic interventions in amyloid-related diseases.

Materials and Methods

Recombinant Expression and Purification of αSyn

The plasmid that expresses human αSyn and the mutants were amplified as previously described. αSyn, αSyn_1–103, and αSyn_104–140 were expressed in an Escherichia coli BL21­(DE3) transformed by pET-αSyn in 2 L flasks at 37 °C with 1 L of Luria–Bertani (LB) medium. Isotopically labeled ¹⁵N-αSyn were expressed in 2 L flasks at 37 °C with 1 L of minimal M9 batch medium. Cells were suspended in purification buffer (50 mM Tris-HCl, pH 7.5, containing 1 mM EDTA and 0.1 mM dithiothreitol), disrupted using sonication, and centrifuged (10 000g, 30 min). Streptomycin sulfate (final 5%) was added to the supernatant to remove nucleic acids. After removal of the nucleic acids by centrifugation, the supernatant was heated to 80 °C for 30 min and then centrifuged. In this step, αSyn remained in the supernatant. The supernatant was precipitated by the addition of solid ammonium sulfate to 70% saturation, centrifuged, dialyzed overnight, and then applied onto a HiTrap-Q column (cytiva) with 50 mM Tris-HCl buffer, pH 7.5, containing 1 mM EDTA and 0.1 mM dithiothreitol as running buffer. Samples were eluted with a linear gradient of 0–1 M NaCl. Collected fractions were dialyzed overnight and then applied onto reversed-phase HPLC (RP-HPLC), using a Prominence HPLC system (Shimadzu) under linear gradient conditions. Mobile phase A (comprising water with 0.1% TFA) was mixed with mobile phase B (0.1% TFA in acetonitrile). Purified αSyn were lyophilized, and molecular mass was confirmed by matrix-assisted laser desorption/ionization time-of-flight mass spectroscopy (MALDI MS), using an UltrafleXtreme instrument (Bruker Daltonics).

αSyn Fibrils for Selection

Since polymorphism is a characteristic property of amyloid fibrils, it is important to prepare and apply homogeneous fibril for the RaPID screening. We first amplified specific αSyn fibrils by repeating seeding experiment. Solutions of monomeric αSyn were prepared by dissolving lyophilized αSyn in PBS buffer. Solutions were filtered using a 0.22 μm PVDF filter, and the αSyn concentration was determined by NanoDrop using ε₂₈₀ = 5120 L mol^–1 cm^–1. Seeding experiments were performed by adding 5% (v/v) preformed fibrils to a 100 μM monomeric αSyn solution. First generation of fibrils for seeding experiments were prepared by spontaneous fibril formation by monitoring ThT fluorescence. Assays were initiated by placing the 96-well plate at 37 °C with a cycle of 3 min of shaking and 27 min of quiescence in a plate reader (Flex station; Molecular Devices). Preformed fibrils were well fragmented by ultrasonication before seeding, and the seeded solution was incubated at 37 °C for 1 week. This seeding experiment was repeated six times with PBS buffer at pH 7.5. The homogeneity of the morphology of the seventh generation of amyloid fibrils was confirmed by analyzing 10 AFM images. We thus decided to use this particular sample for our RaPID campaign. In order to immobilize the fibrils on Dynabeads, human (His)₆-αSyn (Wako Pure Chemical Industries Ltd.) was attached to fibril ends in an amyloid propagation manner by adding 5% of (His)₆-αSyn to the seventh generation of fibrils.

Atomic Force Microscopy

Conventional atomic force microscopy (AFM) measurements were performed in air with the sample deposited on a cleaved bare mica substrate. To detect small assemblies in Figure , the mica surface was functionalized. The mica substrate was incubated with a 10 μL drop of 0.05% (v/v) APTES ((3-aminopropyl)­triethoxysilane, Nacalai Tesque) in Milli-Q water for 1 min at room temperature, rinsed with ultrapure water, and then dried by airflow. The preparation of the mica AFM samples was realized at room temperature by deposition of a 10 μL aliquot of a 10 μM solution of seventh generation αSyn fibrils for 10 min. Then, the sample was rinsed with ultrapure water and dried with a gentle flow of air. Imaging was performed in tapping mode on a Bruker Multimode-8A AFM with 0.9 Hz line-rate for 5 μm × 5 μm images.

Selection of Anti-α-Syn-Fibril Peptides

Thioether macrocycles targeting human αSyn were selected using the RaPID system as previously reported with slight modification. Briefly, the peptide-oligonucleotide (mRNA/cDNA) fusions were incubated with human αSyn fibrils immobilized on Dynabeads for 30 min. αSyn fibrils were masked with 2 mg/mL yeast tRNA (invitrogen) in advance of applying the library. During selection, αSyn fibrils were treated at room temperature to avoid cold denaturation. Our first selection attempt of the mRNA-macrocycle fusion library using our ordinary protocol produced too rapidly enriched binder species in only two rounds, but our control experiments without the translation step also showed an increase in the recovery rate (data not shown). This suggested that the oligonucleotides (mRNA/cDNA) could interact nonspecifically with the fibrils, disrupting the enrichment of our desired species of active macrocycle binders. Therefore, we modified the protocol of the RaPID selection by applying an excess amount of commercial yeast tRNAs to the selection process with the aim of saturating the region(s) of the αSyn fibrils that could potentially interact with oligonucleotides. Finally, the observed enrichments in the seventh round were subjected to further DNA deep sequencing using the MiSeq sequencing system (Illumina). All DNA oligos were purchased from eurofins Genomics and are listed in Table S1. Note that we decided to exclude FL1 for further studies due to its extremely low theoretical score of water solubility, even though it was most enriched with the form of its mRNA fusion.

Chemical Synthesis of Peptides

Macrocyclic peptides were synthesized by standard Fmoc solid-phase peptide synthesis (SPPS) using a Syro Wave automated peptide synthesizer (Biotage). The resulting peptide–resin (25 μmol scale) was treated with a solution of 92.5% trifluoroacetic acid (TFA), 2.5% water, 2.5% triisopropylsilane, and 2.5% ethanedithiol, to yield the free linear N-ClAc-peptide. Following diethyl ether precipitation, the pellet was dissolved in 10 mL triethylamine containing DMSO and incubated for 1 h at 25 °C, to yield the corresponding macrocycle. The peptide suspensions were then acidified by addition of TFA to quench the macrocyclization reaction. The macrocycle was purified by RP-HPLC, using a Prominence HPLC system (Shimadzu) under linear gradient conditions. Mobile phase A (comprising water with 0.1% TFA) was mixed with mobile phase B (0.1% TFA in acetonitrile). Purified peptides were lyophilized in vacuo, and molecular mass was confirmed by MALDI MS, using an Autoflex II instrument (Bruker Daltonics).

For the MTSL and fluorescein labeling of peptide-Cys, N-ClAc-peptide-Cys­(Dpm)-NH-resin was synthesized by Fmoc SPPS. The Mmt group on Cys in the peptide sequence was then deprotected using a mixture of 98% dichloromethane, 3% TFA, and 2.5% triisopropylsilane. The resulting N-ClAc-peptide-Cys­(Dpm)-NH-resin was cyclized by incubating it overnight with 5% N,N-diisopropylethylamine (DIPEA) in N-methylpyrrolidone (NMP) at room temperature. The cleavage and purification of peptide-Cys-NH₂ were performed as described above. The obtained peptides were treated with the same equivalent of (1-oxyl-2,2,5,5-tetramethyl-Δ3-pyrroline-3-methyl) methanethiosulfonate (MTSL, Toronto Research Chemicals Inc.), fluorescein isothiocyanate (FITC, invitrogen), or Alexa568-maleimide (invitrogen) in DMSO. The resulting peptides were purified by RP-HPLC and lyophilized in vacuo. For the fluorescein labeling of peptide-^βAla-Lys, Fmoc-peptide-Cys­(Mmt)-NH-resin, which has Cys­(Dpm) for cyclization and Cys­(StBu) for free Cys in the peptide sequence, was synthesized by Fmoc SPPS. The Mmt group on Cys in peptide sequence was then deprotected using 98% dichloromethane, 3% TFA and 2.5% triisopropylsilane. The resulting Fmoc-peptide-Cys-NH-resin was treated with 4 equiv of FITC in 5% DIPEA/NMP for 1 h at room temperature. The Fmoc group on the N-terminus of the peptide was deprotected using 20% piperidine and then chloroacetylated using ClAc-NHS. The cleavage and purification of N-ClAc-peptide-Cys­(Fluor)-NH₂ were performed as described above. The obtained N-ClAc-peptide-Cys­(Fluor)-NH₂ was cyclized by incubation overnight with 5% DIPEA in NMP at room temperature. The StBu group on Cys in the peptide sequence was then deprotected using tributylphosphine with 10% H₂O, and the resulting peptides were purified by RP-HPLC and lyophilized in vacuo. Theoretical scores of water solubility of linear peptides were determined using the CamSol method.

For the NAC-region-truncated αSyn (αSyn_NPC), αSyn_1–40-Cys-NH₂ and Lys­(N₃)-αSyn_100–140-NH₂ were synthesized by Fmoc SPPS. These peptides are conjugated via click chemistry using DBCO-PEG₂₄-Maleimide. After purification by HPLC, the molecular mass was confirmed by MALDI MS.

Molecular Cloning

pMGdB_sfGFP: The linear dsDNA encoding the sfGFP gene was purchased from Integrated DNA Technologies. The pMGdB vector was digested by XbaI (R0145, New England BioLabs) and XhoI (R0146, New England BioLabs). The gene was cloned into the linear vector by an In-Fusion HD Cloning Kit (639648, Takara), yielding pMGdB_sfGFP. pAC-Ptet_FL2_Linear-sfGFP: The DNA fragment encoding FL2_Linear was prepared by assembly PCR using primers oligo1, oligo2, and oligo3. The linear vector encoding sfGFP was amplified by inverse PCR from pMGdB_sfGFP using prepared by oligo4 and oligo5. The gene was cloned into the linear vector by the In-Fusion HD Cloning Kit, yielding pMGdB_FL2_Linear-sfGFP. The gene encoding FL2_Linear-sfGFP was amplified by PCR using primers oligo6 and oligo7. The linearized pAcTet vector was amplified by inverse PCR from pAC-Ptet-cpAaLS­(119-aMD4L) using oligo8 and oligo9. The gene was cloned into the linear vector by In-Fusion HD Cloning Kit, yielding pAC-Ptet_FL2_Linear-sfGFP. pMGdB_FD1_Linear-sfGFP: The DNA fragment encoding FD1_Linear was prepared by assembly PCR using primers oligo10, oligo11 and oligo12. The linear vector encoding sfGFP was amplified by inverse PCR from pMGdB_sfGFP prepared by oligo13 and oligo14. The gene was cloned into the linear vector by In-Fusion HD Cloning Kit, yielding pMGdB_FD1_Linear-sfGFP. All PCR products were purified by NucleoSpin Gel and PCR Clean-up (U0609A, MACHEREY-NAGEL). Escherichia. coli strain XL1-blue (200249, Agilent Technologies) was used as the host for plasmid preparation. All plasmids were purified using a FastGene Plasmid Mini Kit (FG-90502, NIPPON Genetics). Plasmid sequences were confirmed by Sanger sequencing (FASMAC). All DNA oligos were purchased from eurofins Genomics and are listed in Table S1.

Protein Expression and Purification of FL2_Linear-sfGFP and FD1_Linear-sfGFP

E. coli BL21-gold (DE3)-pLysS competent cells (230134, Agilent Technologies) were transformed with pMGdB_sfGFP or pMGdB_FD1_Linear-sfGFP. The cells were grown at 37 °C in Luria–Bertani (LB) medium containing ampicillin (50 μg/mL) until the OD600 reached 0.4–0.6, at which point protein production was induced by adding isopropyl β-d-1-thiogalactopyranoside (IPTG) to a final concentration of 0.25 mM. Then, the cells were cultured at 20 °C for 16 h. E. coli DH10B competent cells (EC0113, ThermoFischer) were transformed with pAC-Ptet_FL2_Linear-sfGFP. The cells were grown at 37 °C in Luria–Bertani (LB) medium containing chloramphenicol (30 μg/mL) until the OD600 reached 0.4–0.6, at which point protein production was induced by adding tetracycline to a final concentration of 1 μg/mL. Then, the cells were cultured at 20 °C for 16 h. Cells were harvested by centrifugation at 5000g and 4 °C for 10 min. Cells were resuspended in 15 mL of lysis buffer [50 mM sodium phosphate buffer (pH 7.4), 1 M NaCl and 20 mM imidazole] After lysis by sonication and clearance by centrifugation at 15 000g and 25 °C for 25 min, the supernatant was loaded onto 2 mL of Ni Separose 6 Fast Flow resin (Cytiva) in a gravity flow column. Beads were washed with lysis buffer, and protein was eluted with elution buffer [50 mM sodium phosphate buffer (pH 7.4), 200 mM NaCl and 500 mM imidazole]. The buffer was exchanged to PBS, using SnakeSkin Dialysis Tubing, 3.5K MWCO (ThermoFischer).

Isothermal Titration Calorimetry (ITC)

ITC measurements were performed to study the binding between αSyn fibrils and synthesized seven peptides using a Nano ITC instrument (TA Instruments). The peptides were dissolved in DMSO and a 100 mM peptide stock solution was prepared. A 1 mM peptide in PBS buffer (1% DMSO) was then injected into the sample cell containing approximately 190 μL of ultrasonicated αSyn fibrils (seventh generation) at 100 μM in PBS with 1% DMSO. ITC titrations were carried out at 25 °C with 2.5 μL injections for a total of 18 injections with stirring at 400 rpm. The data were fitted using an independent one-binding site model.

Phase Separation Assay

Solutions of monomeric αSyn were prepared by dissolving the lyophilized αSyn in a 10 mM NaOH solution to achieve a neutral pH. The solutions were then filtered using a 0.22 μm filter, and the αSyn concentration was determined by NanoDrop. The resulting monomer was diluted with water to a concentration of 400 μM and stored at −80 °C. Phase separation was induced by mixing αSyn dissolved in the desired buffer (pH7.5) with 10% PEG and the peptide dissolved in 100% DMSO at a concentration of 100 μM. The peptide solution was then diluted to a final concentration of 1% DMSO. Differential interference contrast (DIC) images were obtained at room temperature using a Leica DMI6000 B microscope with a 40× objective lens. All the images were obtained at a resolution of 696 × 520 pixels at 24-bit depth. The αSyn concentration was fixed at 100 μM, unless otherwise stated. For turbidimetry, 100 μM αSyn in the presence of 10% PEG was incubated for 30 min at 4 °C with various concentrations of peptides before measurements. The measurements were carried out using a Jasco V670 spectrometer (JASCO) with excitation and emission at 600 nm. Temperature regulation was carried out using a Peltier unit (JASCO) with a 1 mm light path cell.

For confocal microscopy, we used mixture of 1% FL2C-Fluor or FD1C-Fluor and 99% nonlabeled FL2 or FD1 as LLPS inducer peptides for αSyn (Figure H). Nonlabeled wild-type α-synuclein was mixed with αSyn-Rhod at a 99:1 molar ratio. αSyn formed aggregate-like assemblies with 100% FD1C-Fluor due to its low solubility, whereas αSyn formed spherical droplets with 100% FD1-sfGFP (Figure S8C,D). To minimize the effect of FL2C-Fluor and FD1C-Fluor on the liquidity of formed αSyn droplets, we decided to use low concentration of FL2C-Fluor and FD1C-Fluor by mixing them with nonlabeled FL2 and FD1 for confocal microscopy. Fluorescein and sfGFP without a peptide tag did not show efficient localization in αSyn droplets induced by FL2 and FD1 (Figure S7E).

Quantification of the Dilute Phase by Centrifugation

Centrifugation of αSyn and LLPS samples was performed at 15 300g for 10 min. To assess whether centrifugation influences the distribution of αSyn and to determine whether LLPS leads to the formation of αSyn nanoclusters, ultracentrifugation was additionally conducted at 180 000g for 10 min (Beckman Coulter, USA). For each sample, 50 μL was centrifuged, and 25 μL of the resulting supernatant was collected for quantification using ultrahigh-performance liquid chromatography (UHPLC; Shimadzu). The residual concentrations of αSyn and FD1 in the supernatant were determined based on their respective peak areas.

Confocal Microscopy

The fusion event of αSyn liquid droplets in vitro was visualized with a Leica TCS SP8 confocal microscope with a 63× oil objective lens at room temperature. Rhodamine-labeled αSyn, fluorescein-labeled peptide, peptide-tagged sfGFP, Thiolfavin T (ThT) (Wako Pure Chemical Industries Ltd.), and 1-anilinonaphthalene-8-sulfonic acid (ANS) (Nacalai Tesque) were observed using appropriate fluorescence channels (488 nm for fluorescein and sfGFP, 561 nm for rhodamine, 442 nm for ThT, and 405 nm for ANS). All the images were captured at a resolution of 512 × 512 pixels at 24-bit depth. Fluorescence recovery after photobleaching (FRAP) measurements were performed using a Leica TCS SP8 confocal microscope. A region of interest (ROI) with a radius of 1.0 μm was bleached using an appropriate laser, and the recovery of the bleached spots was recorded by using the software provided with the instrument. The fluorescence recovery was background-corrected, normalized, and plotted using an Igor Pro.

NMR Measurements

¹⁵N-αSyn was dissolved in 20 mM sodium phosphate buffer (pH 7.4), 2% (v/v) D₂O, 10% PEG, and 100 mM FL2 and FD1, which were dissolved in DMSO, were diluted to 100 and 200 μM, respectively (0.1% DMSO in final solution). Different concentration of FL2 and FD1 were used based on their LLPS efficiency, and the pH of the mixture was checked immediately before measurement. The spectra were measured at 4 °C using a Bruker Avance-III 950 MHz spectrometer equipped with a cryogenic probe, and 16 scans were taken for each spectrum. Signal assignments were achieved by comparing the chemical shifts to those previously published and obtained from the temperature- and PEG-concentration-titration measurements. Chemical shift perturbations (CSP) were calculated as Δδ = ((Δδ_N/5)² + (Δδ_HN)²)^1/2. For PRE measurements, nitroxide spin-labeled FD1 (FD1-C21R1) was synthesized as described in the synthesis section, and 2 molar equiv of 10% FD1-C21R1 ([FD1]:[FD1-C21R1] = 9:1) was added to a 100 μM ¹⁵N-αSyn solution. The NMR spectra were processed using TopSpin 4.0 (Bruker), and resonance assignment and intensity calculations were performed using the Sparky program.

Fluorescence Assay

The αSyn monomer was diluted to the desired concentration with 10× PBS buffer and supplemented with 20 μM ThT and 50 μM ANS from a 1 mM stock. All samples were prepared in low-binding Eppendorf tubes on ice. Each sample was then pipetted into multiple wells of a 96-well half-area, low-binding polyethylene glycol coating plate (Corning 3881) with a clear bottom at 80 μL per well. Assays were initiated by placing the 96-well plate at 37 °C with a cycle of 3 min shaking and 27 min quiescence in a plate reader (Flex station; Molecular Devices). The fluorescence of ThT and ANS was measured through the bottom of the plate with a 440 and 380 nm excitation filter, respectively, and a 480 nm emission filter, with three repeats per sample.

Glossary

Abbreviations

LLPS

liquid–liquid phase separation

IDPs

intrinsically disordered proteins

NAC

non-amyloid-β component

RaPID

random nonstandard peptides integrated discovery

RT

reverse transcription

ITC

isothermal calorimetry

PEG

polyethylene glycol

DIC

differential interference contrast

HSQC

heteronuclear single quantum coherence

PRE

paramagnetic relaxation enhancement

MTSL

(1-oxyl-2,2,5,5-tetramethylpyrrolidin-3-yl)­methylmethane­thiosulfonate

PBS

phosphate buffered saline

FRAP

fluorescence recovery after photobleaching

ANS

8-anilino-1-naphthalenesulfonic acid

ThT

thioflavin T

LSPT

liquid-to-solid phase transition

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.5c08019.

• Movie 1 of fusion events of droplets (MP4)

• Oligonucleotide sequence and protein sequence; ITC, turbidimetry, fluorometry, and NMR data and their respective analysis; images of DIC and confocal microscopy (PDF)

#.

Earth-Life Science Institute, Institute of Future Science, Institute of Science Tokyo, Tokyo 152-8550, Japan

The authors declare no competing financial interest.