News

Oct 19, 2024

De novo designed YK peptides forming reversible amyloid for synthetic protein condensates in mammalian cells | Nature Communications

Nature Communications volume 15, Article number: 8503 (2024) Cite this article Metrics details In mammalian cells, protein condensates underlie diverse cell functions. Intensive synthetic biological

Nature Communications volume 15, Article number: 8503 (2024) Cite this article

Metrics details

In mammalian cells, protein condensates underlie diverse cell functions. Intensive synthetic biological research has been devoted to fabricating liquid droplets using de novo peptides/proteins designed from scratch in test tubes or bacterial cells. However, the development of de novo sequences for synthetic droplets forming in eukaryotes is challenging. Here, we report YK peptides, comprising 9–15 residues of alternating repeats of tyrosine and lysine, which form reversible amyloid-like fibrils accompanied by binding with poly-anion species such as ATP. By genetically tagging the YK peptide, superfolder GFPs assemble into artificial liquid-like droplets in living cells. Rational design of the YK system allows fine-tuning of the fluidity and construction of multi-component droplets. The YK system not only facilitates intracellular reconstitution of simplified models for natural protein condensates, but it also provides a toolbox for the systematic creation of droplets with different dynamics and composition for in situ evaluation.

In eukaryotes, biomolecular condensates––membrane-less compartments that concentrate specific sets of biomolecules––spatiotemporally regulate a broad spectrum of cellular functions1,2,3. Liquid–liquid phase separation (LLPS) or biomolecular assembly via weak multivalent interaction underlies the formation of these membrane-less structures with dynamic liquid-like properties4. Based on this mechanism, synthetic protein condensates have been fabricated using engineered proteins5,6,7,8,9,10. The sequence-level design for intracellular artificial protein condensates is of substantial interest in bottom-up approach research to boost understanding at the molecular level and serve as pivotal tools for synthetic biology.

From a chemical perspective, short peptides fabricating artificial coacervate droplets have been developed by taking into account extensive interactions, including van der Waals forces and electrostatic and aromatic interactions11,12. Lysine/arginine-rich cationic peptides associate with anionic biomolecules, forming complex coacervate through electrostatic interactions13,14,15,16. Hydrophobic or amphiphilic peptides exhibiting fibrillar assembly can be modified to favor droplet formation17,18. Recently, designed coiled-coil peptides capable of forming network structures have been demonstrated for artificial protein condensates in bacterial cells19,20. However, almost all of these studies have been limited either to test tubes or within the prokaryotes. Indeed, while K10 or R10 peptides16 are reported to form droplets in vitro, the peptide-fused superfolder GFPs (sfGFPs) dispersed or localized in the nucleolus in mammalian cells (Supplementary Fig. 1). The droplet formation of Tag3-fused sfGFP in bacteria21 could not be reproduced in mammalian cells. There are pioneering examples of using de novo peptides in mammalian cells by two groups: Shu et al. leveraged coiled-coil peptides as homo-oligomer building blocks for constructing protein condensates22. Jerala et al. demonstrated droplet construction by varying the stability of coiled-coil dimers, valency, and linkers23. However, these have mainly focused on coiled-coil motifs, whereas β-sheet structures are deeply involved in natural protein condensates. The core motifs of representative droplet-forming proteins, such as FUS, are known to form reversible fibrils24,25. Additionally, some droplets mature into pathogenic amyloid-like fibrils26,27. Given the ubiquity of biomolecular condensation in mammalian cells, the development of de novo β-sheet peptides creating artificial droplets in these cellular contexts is inevitable. The lack of such a platform represents a considerable bottleneck in synthetic biology research on protein condensates.

In this article, we present a series of YK de novo peptides that promote artificial protein condensates with liquid-like properties within mammalian cells. The YK peptides consist of alternating tyrosine and cationic lysine residues, spanning lengths from 9 to 15 residues. These YK-genetically fused sfGFPs exhibit droplet formation in mammalian cells, and the length of the YK peptide can modulate their dynamic fluidity. Additionally, by leveraging the homotopic assembly of YK peptides, we design and fabricate multi-component droplets in the cells. In vitro investigations show that the cationic YK peptides respond to adenosine triphosphate (ATP), forming reversible amyloid-like fibrils––a proposed essential structure observed in natural droplet-forming proteins28. In the presence of poly-anion species, the YK-fused sfGFP assembles into colloids of approximately 20 nm in diameter, which subsequently coalesce into micron-scale droplets upon introducing a crowding reagent.

To demonstrate the versatility of YK-based methodologies, we reconstruct multi-component droplets composed of Nck1 adapter protein and neural Wiskott–Aldrich syndrome protein (N-WASP), which are implicated in actin polymerization within living cells29,30. These tests show that the accumulation and lower dynamics of N-WASP in droplets enhance the local amount of polymerized actin31. Thus, YK systems provide a valuable tool for the bottom-up construction of multi-component droplets and for investigating protein condensates in mammalian cells

To create droplet-inducing peptide tags, we designed YK peptides comprising alternating tyrosine and lysine residues, following the fundamental principles of β-sheet peptide design (Fig. 1a). Alternating repeats of hydrophobic and hydrophilic amino acids yield an amphiphilic β-strand, which polymerically assembles into supramolecular fibers32,33,34,35. In our previous study, we developed a Y15 peptide36,37,38,39 in which hydrophobic tyrosine residues were aligned on one face while lysine or glutamate residues were positioned on the opposite face. Despite the Y15-fused proteins demonstrating self-assembly within mammalian cells, they formed µm-scaled gel-like structures lacking fluidity in the case of dimeric or tetrameric proteins fusion. We attributed this stable and irreversible assembly to the interplay of complementary electrostatic intermolecular interactions between cationic lysine and anionic glutamate residues. To lower the stability of formed fibrils and introduce reversibility of the peptide assembly, we incorporated electrostatic repulsion between peptides by employing solely cationic lysine residues as hydrophilic components.

a Chemical structure of YK13 peptide and the schematic illustration of droplets formation in cells. We modeled that the formation of liquid droplets is driven by the cooperative interaction between YK peptide interactions and the dimerization of sfGFP. NES means nuclear exporting signal. b YK peptide tag sequences used in this study. c Confocal fluorescence images showing droplet formation of NES-YK-sfGFPs in a length-dependent manner. Scale bar, 20 µm. d Partition coefficients of NES-YK-sfGFPs determined by fluorescence imaging. The box plot is represented with a center line, median; box limits, Q1 and Q3; whiskers, 1.5×interquartile range; points, outliers. n = 41, 43, and 39 cells, respectively. e Fluorescence anisotropy values of NES-YK-sfGFPs expressing in living cells. Data are presented as mean values +/- SD (n = 3, independent samples). f Fusion of NES-YK13-sfGFP droplets. The two independent droplets fused at 12 s into a single larger droplet. Scale bar, 1 µm. g FRAP analysis of NES-YK-sfGFP droplets. The region of interest was bleached at 0:01; then, the fluorescence recovery in the region was monitored. Scale bar, 2 µm. The right panel shows that the recovery rate reduces as the YK peptide length increases. The line plots represent mean values with translucent error bands (SD, n = 15 cells). h Fluorescence images of NES-YK13 control peptide-fused sfGFP in cells. Underlined are residues altered from the original YK13 peptide. Scale bar, 20 µm. Source data are provided as a Source Data file for d, e, and g.

Initially, we investigated the protein droplet formation by YK-tagging in cells. In accordance with our previous report revealing that tyrosine-rich peptides tend to localize in the nucleus37, we appended a nuclear export sequence (NES), from MAPKK2 (36–55)40 to the N-terminal. By using sfGFP (239 residues) forming a weakly dimer, we tested the intracellular droplet formation of NES-YK-tagged sfGFPs by varying number of repeats of YK units (Fig. 1b and Supplementary Fig. 2). This led to longer NES-YK13- and NES-YK15-sfGFPs forming fluorescent granules in the cytosol (Fig. 1c and Supplementary Fig. 3). The majority of NES-YK11-sfGFP was distributed in cells, while a minority formed puncta. In contrast, the shortest NES-YK9-sfGFP did not show any assemblies. The partition coefficients were 3.0 ± 0.9, 6.4 ± 1.8, and 5.9 ± 1.7 for YK11, YK13, and YK15, respectively (Fig. 1d). The structures overlapped with neither G3BP1 (a stress granule marker41) nor vimentin (aggresomal marker42), suggesting that the YK-based granules were independent of these natural bodies (Supplementary Fig. 4). We also confirmed that YK13 fusion had negligible effects on sfGFP fluorescence intensity (Supplementary Fig. 5) and cell viability (Supplementary Fig. 6). Moreover, to obtain evidence of the self-association, we measured the fluorescence anisotropy (r) because Homo-FRET among proximal fluorescent proteins in complexes decreases the r43,44. The results revealed that the r values of NES-YK-sfGFPs gradually decreased with a longer YK length (Fig. 1e).

The granules of NES-YK-sfGFP exhibited liquid-like properties. From time-lapse imaging, we observed that two independent droplets fused into one large spherical structure (Fig. 1f). In addition, rapid fluorescence recovery was clearly observed in FRAP studies (Fig. 1g). Notably, the fluidity attenuated in a length-dependent manner, where the t1/2 were 4.5 ± 1.8, 7.5 ± 1.9, and 8.9 ± 2.5 s, and the mobile fractions were 74.4 ± 13.1%, 47.3 ± 16.9%, and 26.0 ± 15.0% for YK11, YK13, and YK15 fusion, respectively (Fig. 1g). YK-sfGFPs without NES formed droplets with high roundness in the nucleus (Supplementary Fig. 7). In the same manner as for NES-YK-sfGFPs, the fluidity in droplets gradually decreased with increasing peptide chain length (Supplementary Fig. 8). The peptide length dependency proves that the YK peptide association directly affects the dynamics of droplets.

This droplet formation is a YK sequence-specific event. NES-FK13/VK13-sfGFP, in which Phe or Val was used instead of Tyr, exhibited irregularly shaped aggregates, while an Ala substituent (NES-AK13-sfGFP) dispersed uniformly (Supplementary Fig. 9). This trend is consistent with our previous studies37 using XEXK repeats (X: Ile, Phe, Leu, Val, Tyr, or Trp). In a series of Lys substituents, Glu or Gln substituents (NES-YE13- or NES-YQ13-sfGFPs) do not show any fluorescent puncta and an Arg substituent (NES-YR13-sfGFP) formed less-circular aggregates (Supplementary Fig. 10). Notably, fluorescent puncta were barely observed in cells for the sfGFPs fused with YK13(split) with a GGS linker at the center of YK13 peptide, YK13(Y7P) with a Tyr-to-Pro point substitution destabilizing the β-sheet secondary structure, scrambled scYK13, and YK13(K6Y/Y7K) in which only one pair of Tyr and Lys positions was exchanged (Fig. 1h). Furthermore, fluorescence anisotropy was not reduced, suggesting that these are monomerically dispersed (Supplementary Fig. 11). Simple cationic K6 peptide also does not induce assembly. These indicate that the droplet formation is a phenomenon specific to YK repeats.

We propose that the cationic YK13 peptide assembles into fibril structures through the interaction with anionic species such as ATP present at the millimolar level in living cells45 (Fig. 2a). Initially, we prepared the purified YK peptides by standard Fmoc solid-phase synthesis (Supplementary Fig. 12). To assess the self-assembling propensities of YK13 peptide, we used thioflavin-T (ThT) dye or X-34 dye46, fluorogenic dyes that bind to amyloid-like structures. Prior to the addition of ATP, the level of fluorescence was comparable to that of the blank control, whereas the addition dramatically enhanced the fluorescence intensity (Fig. 2b and Supplementary Fig. 13 and 14). This ATP-responsive assembly of YK13 was also monitored as an increase in turbidity (Supplementary Fig. 15). In CD measurement, YK13 peptide solution with ATP exhibited positive Cotton effects at 211 and 238 nm, potentially attributable to alignment of the phenol group of tyrosine residues (Supplementary Fig. 16). FT-IR measurements showed broad bands at 1700–1600 cm−1 from YK13, whereas the mixture of YK13 and ATP exhibited a strong sharp peak that appeared around 1635 cm−1, indicative of a β-sheet secondary structure47 (Supplementary Fig. 17). Moreover, from negatively stained TEM, the bundled nanofibril structures were observed (Fig. 2c and Supplementary Fig. 18).

a A schematic illustration of ATP-mediated self-assembly of YK13 peptide. b Thioflavin-T fluorescence assay of YK13 peptide. To the YK13 peptide (50 µM) solution in 10 mM Tris-HCl and 150 mM NaCl buffer (pH 7.5) containing ATP at various concentrations (0–3 mM) was added thioflavin-T (25 µM). The charge ratios are indicated as positive/negative (P/N). The intensity increased in an ATP-dose-dependent manner. Data are presented as mean values +/- SD (n = 3, independent samples). c Negatively stained TEM images of YK13 nanofibers. Scale bar represents 100 nm (inset, 50 nm). The enlarged image revealed that thin fibrils (<5 nm width) were bundled. d Anion species dependence in the assembly of YK13. The final concentrations of YK13, anion species, and thioflavin-T in 10 mM Tris-HCl buffer (pH 7.5) are 50 µM, 1 mM, and 25 µM, respectively. Data are presented as mean values +/- SD (n = 3 biologically independent experiments). e Simulation snapshots at 100 ns for the YK13 and ATP complex (blue: N atom, red: O atom, purple: ATP). Source data are provided as a Source Data file for b and d.

The increase in salt concentration decreased ThT fluorescence, suggesting the significance of electrostatic interactions (Supplementary Fig. 19). Adenosine diphosphate (ADP) significantly increased ThT fluorescence, while neither adenosine monophosphate (AMP) nor phosphate showed enhancement (Fig. 2d). This trend indicates that multivalent anions may have prompted the fibril formation of YK13. Unexpectedly, the addition of pyrophosphate did not increase ThT fluorescence. Furthermore, uridine triphosphate (UTP) and cytidine triphosphate (CTP) showed no significant impact (Supplementary Fig. 20). In contrast, guanosine triphosphate (GTP), containing a purine base, exhibited strong intensity comparable to that upon ATP addition. These results underscore the contribution of purine bases to peptide self-assembly. On the basis of these considerations, MD simulations were performed to explore the impact of ATP on YK13 assembly. The simulations revealed that, in the presence of ATP, the YK13 fibril structure was well maintained (RMSD = 2.2 Å) (Fig. 2e and Supplementary Fig. 21). Notably, the 100 ns structure illustrated the base stacking of ATP on the peptides. In contrast, in the absence of ATP, the YK13 fibers rapidly collapsed (Supplementary Fig. 21). These findings strongly suggest that ATP-mediated YK peptide fiber formation relies on multivalent electrostatic interactions involving triphosphate and lysine, as well as base-stacking interactions among ATPs.

ATP forms complexes with Mg(II) within cells48. We confirmed the ATP-dependent self-assembly of YK13 peptides under 5 mM MgCl2 conditions (Supplementary Fig. 13–15, 17). Notably, negative control of YK13(K6Y/Y7K), which did not induce droplet formation inside cells, showed an increase in neither turbidity nor ThT fluorescence under Mg(II) conditions (Supplementary Fig. 13–15). This consistency indicates that, within living systems, endogenous poly-anions mediate the bridging of YK peptides through interactions with the cationic side chains of lysine, arranged along one face of the YK peptide’s β-strand, thereby stabilizing the β-sheet structure.

YK13 peptides assemble one-dimensionally into a fibril structure in the presence of ATP, but why did YK13-sfGFP adopt a droplet form with an isotropic spherical structure in its assembly? To answer this question, we examined ATP-dependent assembly in a test tube using purified His-YK13-sfGFP, which was expressed in E. coli (Fig. 3a and Supplementary Fig. 22). The results revealed that the r of His-YK13-sfGFP significantly decreased upon mixing with ATP, supporting its ATP-mediated self-assembly (Fig. 3b). Intriguingly, the mere addition of ATP to the His-YK13-sfGFP solution did not lead to droplets. DLS measurements indicated colloidal structures with a particle size of approximately 20 nm (Fig. 3c). TEM observation revealed spherical structures of 20 nm, which is consistent with the DLS results (Fig. 3d). Meanwhile, untagged sfGFP showed neither a decrease in r nor an increase in particle size (Fig. 3c and Supplementary Fig. 23). The CD spectrum of His-YK13-sfGFP altered upon addition of ATP; the difference spectrum for His-sfGFP displayed positive peaks at 211 nm and 234 nm, comparable to that of the YK13 peptide (Supplementary Fig. 24). Furthermore, we conducted ThT fluorescence studies. Since the fluorescence of sfGFP overlaps with that of ThT, we prepared His-YK13-sfYFP with a T203Y mutation from sfGFP (Supplementary Fig. 25). As a result, the mixture of ATP and His-YK13-sfYFP showed ThT fluorescence, although the intensity was weaker compared to that of YK13 peptide. Given that ThT specifically detects fibrils in the case of Aβ40, while the signal for both oligomers and protofibrils are negligible or weak49, the YK13-sfYFP oligomer might also exhibit weak ThT fluorescence. While YK13 moiety in His-YK13-sfGFP adopts a structure similar to that of YK13 peptide, the fusion forms an oligomeric assembly, probably due to frustrated growth caused by steric hindrance of sfGFP.

a A schematic illustration of ATP-dependent self-assembly of His-YK13-sfGFP and subsequent droplet formation. b Anion species dependence of the self-assembly. The charge ratios of the positive charge of YK peptide moiety and the negative charge of anion species are shown as P/N. The fluorescence anisotropy of His-YK13-sfGFP (5 µM) exhibited sigmoidal decreases in the cases of poly-valent anion [ATP, pyrophosphate, ADP, poly-A (2,100–10,000 nucleotides), and poly-Glu (>12 kDa)] addition. In contrast, anions with less multivalency (phosphate and AMP) did not impact the assembly. Data are presented as mean values +/- SD (n = 3, independent samples). c DLS measurements of 5 µM His-YK13-sfGFP (red) or His-sfGFP (gray) in the presence (straight line) or absence (dashed line) of 1 mM ATP. d A negatively stained TEM image of His-YK13-sfGFP with ATP. Scale bar, 50 nm. e Droplet formation of His-YK13-sfGFP (5 µM) in the presence of 1 mM ATP and PEG20k. The solutions were incubated at 37 °C for 1 day before observation. Scale bar, 20 µm. f FRAP analysis of NES-YK13-sfGFP or His-YK13-sfGFP droplets. Scale bar, 2 µm. The right panel shows time course of normalized fluorescence intensities of bleached regions. The data points are presented as mean value ± SD as translucent error bands (n = 5). Source data are provided as a Source Data file for b, c, and f.

The r increased with increasing salt strength (Supplementary Fig. 26), and His-YK13-sfGFP particle size decreased with increasing temperature (Supplementary Fig. 27), suggesting that the electrostatic interaction contributed to assembly. Similar to ATP, pyrophosphate and GTP significantly reduced the r value (Fig. 3b and Supplementary Fig. 28). ADP displayed an increased concentration threshold for a response at concentrations exceeding 3 mM. In contrast, phosphate and AMP, with low anion valence, showed no changes even at 10 mM. CTP, UTP, poly-A (700–3500 kDa), and poly-glutamate (>12 kDa) also exhibited a decrease in r, albeit less pronounced compared to the decrease observed with the addition of ATP or GTP. The complex with poly-A formed smaller colloids, presumably due to a significantly negative zeta potential, suppressing fusion through charge repulsion (Supplementary Fig. 29).

Subsequently, we investigated the molecular crowding effects on droplet formation of His-YK13-sfGFP, considering that various recombinant IDRs phase-separated in vitro under crowded conditions26,27. The results revealed that a solution containing 5 µM His-YK13-sfGFP and 1 mM ATP exhibited highly spherical structures (roundness of 0.92 ± 0.08) under 3% PEG20k conditions, gradually growing in size over time (Fig. 3e and Supplementary Fig. 30). Time-lapse images at the hour scale show the fusion process of the droplets (Supplementary Fig. 31). When varying the applied concentration, the partition coefficient remained almost constant, supporting the assertion that His-YK13-sfGFP exhibits characteristics of droplets (Supplementary Fig. 32). The phase diagram showed that the phase boundary existed above 4 µM His-YK13-sfGFP and above 400 µM ATP (Supplementary Fig. 33). We observed similar droplets upon the addition of ADP, pyrophosphate, and other NTPs, whereas smaller aggregates formed upon the addition of poly-A and poly-glutamate, carrying high valence anions (Supplementary Fig. 34). Taking these findings together, His-YK13-sfGFP rapidly forms oligomers through interactions with poly-anion species driven by electrostatic interactions, followed by droplet formation in a crowded environment. During these experiments, we also noticed that at high concentrations (50 µM), His-YK13-sfGFP formed aggregates in the absence of ATP regardless of PEG20k addition (Supplementary Fig. 35). These aggregates were distinct in morphology from the spherical droplets formed in the presence of ATP and PEG20k. We assume that in the absence of ATP, the positively charged YK13 (+6) directly interacts with the negatively charged sfGFP (−7), leading to aggregation.

YK peptide chain length-dependent oligomer and droplet formation was also observed: His-YK11-, His-YK13-, and His-YK15-sfGFP showed particle sizes of 13.8 ± 3.1, 23.3 ± 0.4, and 28.9 ± 1.9 nm upon the addition of ATP, while His-YK9-sfGFP remained in a monomeric state even in the presence of ATP (Supplementary Fig. 36). Upon the addition of 5% PEG20k, droplets were observed at >5 µM for YK11-sfGFP and at >1 µM for YK13- and YK15-sfGFP, whereas no droplet formation was observed for His-YK9-sfGFP (Supplementary Fig. 37), which is consistent with the YK length dependence observed in living cells. Even in the presence of Mg(II), His-YK13-sfGFP exhibited responsiveness to ATP, forming droplets under crowded conditions (Supplementary Fig. 38).

The observed droplets in vitro, however, displayed little fluorescence recovery in FRAP experiments, indicating less dynamic within this timeframe (Fig. 3f and Supplementary Fig. 39). We attributed the discrepancy in dynamics to the differences in the N-terminal tags and the experimental conditions. We prepared NES-YK13-sfGFP without a His-tag and investigated the conditions for droplet formation in vitro. Compared to His-YK13-sfGFP, higher concentrations (>10 µM) of NES-YK13-sfGFP were required for droplet formation under 1 mM ATP conditions (Supplementary Fig. 40). In Mg(II)-containing solutions, the threshold shifts to higher concentrations, required at >30 µM NES-YK13-sfGFP and >3 mM ATP. Under conditions close to the phase boundary, NES-YK13-sfGFP droplets were more dynamic, showing a mobile fraction of 33% (Fig. 3f and Supplementary Fig. 39). The net charge of NES-YK13-sfGFP is -4 (NES: -1, YK13: +6, sfGFP: -7, and the sequence from the restriction enzyme site: -2). In contrast, the net charge of His-YK13-sfGFP is -1 (His-thrombin recognition site: +1, YK13: +6, sfGFP: -7, and the sequence from the restriction enzyme site: -1). The increased electrostatic repulsion in NES-YK13-sfGFP may contribute to its reduced self-assembly propensity and more dynamic behavior. Considering that NES-YK13-sfGFP within cells existed at high concentrations (>100 µM) (Supplementary Fig. 41), the above experimental conditions were closer to the intracellular conditions. Additionally, the crowded environment in vitro must differ from the natural cellular environment. Indeed, COS-7 cells expressing NES-YK13-sfGFP were treated with a lysis buffer containing 5% PEG20k, resulting in dynamically arrested protein condensates that showed little recovery in FRAP experiments. Conversely, in the lysis buffer lacking PEG20k, the droplets were dissolved upon dilution (Supplementary Fig. 42).

The reversible fibril formation is proposed to be responsible for the dynamic association properties of low-complexity (LC) domains in natural droplet-forming proteins24,25,28,50. As a representative example, the self-assembling motifs within the FUS LC domain adopt reversible fibers that dissolve in response to heating and phosphorylation25. This characteristic is fundamentally distinct from the thermostable fibrils observed in pathologic amyloid. We expect that, similar to these natural reversible amyloid cores, the YK13 peptide forms a reversible fibril. To explore this, we sequentially introduced ATP and ATPase into a solution of YK13 peptide. Thioflavin-T fluorescence was enhanced upon the addition of ATP and returned to background levels upon subsequent ATPase addition (Fig. 4a). Upon further ATP addition, the fluorescence was temporally enhanced, followed by a decline. This transient response to ATP was repeatedly observed. Notably, the disassembly rate correlated with ATPase concentration, with rapid dissociation on the order of seconds (t1/2 < 30 s) at the highest concentrations of ATPase (Supplementary Fig. 43).

a Reversible amyloid formation of YK13 peptide in vitro. The time course of thioflavin-T fluorescence, where black arrowheads show the addition of ATP and the blue arrow shows the addition of ATPase. b Reversible droplet formation of His-YK13-sfGFP in the presence of 5% PEG20k. In the presence of ATPase (0.66 U/mL potato apyrase, red line), the turbidity was temporarily enhanced. Arrowheads represent the addition of ATP (1.0 mM). Data are presented as mean values +/- SD (n = 3, independent samples). c Intracellular droplet dissolution triggered by the addition of 2-DOG. The COS−7 cells expressing NES-YK13-sfGFP in D-MEM (without glucose) were treated with 2-DOG (10 mM), followed by incubation for 18 min. Scale bar, 10 µm. Source data are provided as a Source Data file for a and b.

We also confirmed that the self-assembly and droplet formation of His-YK13-sfGFP were reversible. Upon addition of ATP to a mixture of ATPase and His-YK13-sfGFP, a transient decrease in anisotropy was observed (Supplementary Fig. 44). In the negative control lacking ATPase, the decrease in anisotropy was maintained. In the presence of PEG20k, upon adding ATP, a transient increase in turbidity occurred, followed by a gradual return of turbidity to the baseline (Fig. 4b and Supplementary Fig. 44). To investigate the reversible droplet formation of NES-YK13-sfGFP within cells, we added a glycolysis inhibitor, 2-deoxyglucose (2-DOG), which decreases the intracellular ATP level51. The addition of 10 mM 2-DOG resulted in dissolution of most droplets in about 20 min (Fig. 4c and Supplementary Fig. 45). The droplets of NES-YK13-sfGFP were not stained by SYTO-17, a nucleic acids binding dye (Supplementary Fig. 46). These are supporting the assertion that ATP, rather than RNA, acts as a counter-anion of YK13 peptide in living cells.

In nature, protein condensates are multi-component systems in which certain subsets of proteins are enriched and cooperatively interact1. We leveraged the simple assembly system of YK peptide for the bottom-up construction of multicomponent protein condensates in mammalian cells. To demonstrate this, we fused the NES-YK13 peptide to three fluorescent proteins (sfGFP, mCherry, and miRFP670) and co-expressed them (Fig. 5a). As a result, they co-assembled into three-component droplets in cells (Fig. 5b).

a Design and construction of multi-component droplets consisting of sfGFP, mCherry, and miRFP670-myc by tagging NES-YK13 peptide. b Confocal images of multi-component protein droplets in COS-7 cells. Scale bar, 10 µm. A line-scan plot from A and B corresponding to a dashed line in images is shown. Source data are provided as a Source Data file for b.

We noticed that, when these NES-YK13-fused fluorescent proteins were solely expressed, only the sfGFP fusion formed droplets. In the case of fusions with mAG (monomeric Azami-Green52) and mCherry with high monomeric quality53, fluorescence was observed uniformly from the cytoplasm, and droplets were scarcely observed (Supplementary Fig. 47). Despite the dispersion, NES-YK13-mAG and NES-YK13-mCherry exhibited low fluorescence anisotropy (Supplementary Fig. 48 and 49), indicating that they both formed oligomeric assemblies that were invisible by optical microscopy. Conversely, tetrameric AG fusion, even upon short YK9 fusion, led to aggregates that never recovered fluorescence in the FRAP test (Supplementary Fig. 50). Given the impact of YK peptide length on fluidity and the fact that sfGFP dimerizes at high concentrations54, we inferred that the reversible polymerization of YK peptides and the weak protein–protein interaction cooperated to form micron-scale structures with dynamics. To test the necessity of dimer formation by sfGFP, we examined two monomeric variants, sfGFP(V206K) and sfGFP(F223D). Neither NES-YK13 fusion formed droplets within cells (Supplementary Fig. 51). Moreover, to demonstrate that dimerization is sufficient for droplet formation, we fused a coiled-coil peptide (CCDi)55 forming homodimers to the C-terminus of NES-YK-mCherry. The results showed that NES-YK-mCherry-CCDi formed droplets, albeit with low dynamics (Supplementary Fig. 49).

In our multicomponent droplet system, the NES-YK13-sfGFP acts as a scaffolding protein that maintains the droplet, and the monomeric fluorescent proteins (NES-YK13-mCherry and NES-YK13-miRFP670-myc) are recruited into the YK-based droplet as client proteins. FRAP analysis of NES-YK13-mCherry or NES-YK13-miRFP670-myc showed rapid fluorescence recovery within less than one second (Supplementary Fig. 52). This suggests that the client proteins NES-YK13-mCherry and NES-YK13-miRFP670-myc accumulate in droplets relying solely on YK13 peptide interactions, whereas NES-YK13-sfGFP scaffolds droplets through the combination of YK interactions and sfGFP homodimer interactions. This scaffold/client concept is consistent with the mechanism of protein assembly behind natural liquid condensates56,57. For instance, promyelocytic leukemia (PML) protein acts as a scaffold for the PML nucleus, while Sp100 and BLM client proteins partition into the structure58. In stress granules, a client-enriched shell is formed around a core consisting of scaffolding proteins such as G3BP159. The estimates from the reported coarse-grained MD calculations suggested the importance of the network connectivity between scaffolds in the maintenance of biomolecular condensates60.

To demonstrate the utility of the YK-based droplet system, we applied it to an intracellular reconstitution assay of Nck1/N-WASP droplets, which regulate actin cytoskeleton dynamics61. Three SH3 domains of Nck1(1–258) interact with proline-rich motifs of N-WASP, leading to droplet formation via their multivalent weak interactions31,62. This interaction triggers the allosteric activation of N-WASP, which subsequently recruits an Arp2/3 complex to initiate actin polymerization63. Additionally, condensates including Nck1 and N-WASP recruit actin filaments by binding to their basic regions64. Given these findings, we first co-expressed Nck1(1–258) fused with varying lengths of YK peptides and untagged N-WASP to investigate the influence of Nck1(1–258) protein assembly on droplets formation (Fig. 6a). NES-YK9/YK11-mAG-Nck1(1–258) was observed as small granules with 0.3 ± 0.1 µm of diameter where mCherry-N-WASP was recruited (Fig. 6b). In contrast, mAG-Nck1(1–258) did not show any droplets, and the NES-YK7-mAG-Nck1(1–258) formed droplets only in high-expressing cells. Considering that NES-YK13-mAG did not form droplets, we supposed that the interaction between Nck1(1–258) and N-WASP contributed to droplet formation62,65. By elongating the YK peptide, the Nck1(1–258) dynamics were reduced while N-WASP dynamics were almost constant (Fig. 6c). Reduction in fluorescence anisotropy of mAG supports the peptide length-dependent assembly (Supplementary Fig. 53).

a Intracellular reconstitution of YK-tagged Nck1(1–258) and N-WASP droplet. b Fluorescence images of COS-7 cells co-expressing NES-YK-mAG-Nck1(1–258) and mCherry-N-WASP. Scale bar, 5 µm. c FRAP analysis of (left) NES-YK-mAG-Nck1(1–258) and (right) mCherry-N-WASP in droplets. The line plots represent mean values with error bands (SD, n = 10 cells). d Droplets composed of YK9-tagging Nck1(1–258) and YK-tagged N-WASPs. e Fluorescence images of COS-7 cells co-expressing NES-YK9-mAG-Nck1(1–258) and NES-YK-mCherry-N-WASP. Scale bar, 5 µm. f FRAP analysis of (left) NES-YK9-mAG- Nck1(1–258) and (right) NES-YK-mCherry-N-WASP in granules. (S.D., n = 10 cells). g, h Fluorescence images of phalloidin-stained COS-7 cells co-expressing (g) NES-YK9-mAG-Nck1(1−258) and mCherry-N-WASP or (h) NES-YK9-mAG-Nck1(1-258) and NES-YK9-mCherry-N-WASP. Scale bar, 1 µm. The arrowheads indicate granule positions. i The actin intensities over granules. (n = 25, 38, 35, 37, 34, 35, and 28 cells respectively). j The N-WASP-normalized actin intensities over granules. (n = 25, 38, 35, 37, 34, 35, and 28 cells respectively). The box plots are presented as follows: central line, median; box limits, Q1 and Q3; whiskers, 1.5×interquartile range; and points, outliers. Statistical comparisons between two groups were performed using unpaired two-tailed Student’s t-tests. Source data are provided as a Source Data file for c, f, i, and j.

Next, we fused various lengths of YK peptides to N-WASP, while the YK tag on Nck1(1–258) is kept as YK9 (Fig. 6d). Notably, upon YK9 fusion, the granules clustered into larger structures of 1.8 ± 0.9 µm, and the partition coefficient of mCherry-N-WASP increased from 5.0 ± 2.3 to 11.8 ± 7.3 (Fig. 6e and Supplementary Fig. 54). FRAP studies showed slow dynamics of NES-YK9-mCherry-N-WASP, a significant difference from the immediate fluorescent recovery of untagged mCherry-N-WASP (Fig. 6f).

To investigate the actin polymerization, we conducted phalloidin staining after cell fixation (Supplementary Fig. 55 and 56). When Nck1(1–258) was solely tagged with YK9, the actin comet tails29,66 extending from the artificial droplets were observed (Fig. 6g). In contrast, in conditions where N-WASP dynamics were lost, the actin intensity over granules remarkably increased (Fig. 6h, i). Although the N-WASP-normalized actin signal was also elevated by tagging longer YK on N-WASP, the marked increase in polymerized actin is pronounced particularly due to the enrichment of N-WASP (Fig. 6j and Supplementary Fig. 54). This suggests that the binding capacity for actin could be enforced in the condensates as well. Taken together, by intentionally manipulating the dynamics and accumulation of Nck1(1–258) and N-WASP through YK fusion, we obtained substantial evidence of their contribution to actin polymerization in living cells.

Synthetic biological research on protein condensates in eukaryotes provides insights into droplet formation at the molecular level and engineering tools in living systems. Although de novo peptides forming droplets in test tubes or within bacterial cells were previously developed, the use of designed sequences in mammalian cells has been limited, with reported cases primarily involving the use of coiled-coil motifs as building blocks22,23. Despite the strong association between β-sheets and protein condensates, to our knowledge, this is the first report investigating de novo β-sheet peptides specifically tailored for droplet formation within mammalian cells. The substantial disparities in environments between bacterial and mammalian cells67 might present an obstacle to the design and construction of artificial droplets. Meanwhile, we have developed amphiphilic peptides that self-assemble into amyloid-like fibrils in cells; sfGFP fusion with Yn peptides consisting of YEYK repeats resulted in dynamically arrested gel-like aggregates36. In this study, by changing the sequence from YEYK repeats to YKYK repeats to weaken the peptide interaction, we succeeded in fabricating droplets intracellularly. The essence of YK peptide-based droplet formation should lie in the reversible fibril formation of the peptide. The cationic YK peptides lose their self-assembling propensity due to electrostatic repulsion, whereas they readily self-assemble with poly-anions like ATP. Notably, the YK sequence is entirely designed from scratch without any consideration of trends in sequences found within IDRs.

Recent protein engineering methods using IDR or continuous folded domains have successfully induced artificial droplet formation in mammalian cells6,7,8,9,23,68. In comparison, our peptide-based approach exhibits several advantages. First, our peptide enables the fine-tuning of droplet dynamic properties by simply adjusting peptide length. Second, its small size, namely 9–15 residues, significantly streamlines experimental procedures at the laboratory scale69. To drive protein condensates, both a dimeric motif and the YK peptide are required, whereas for recruiting client proteins, simply fusing the YK tag is sufficient. Additionally, for proteins with the potential to form droplets, such as Nck1 and N-WASP, fusing only the YK tag is adequate, and there is no need to attach artificial oligomer motifs. Thanks to the shortness of the YK tag, the YK gene can be incorporated at desired positions by inserting annealed oligo DNAs or via inverse PCR using primers containing the peptide gene. Third, owing to its homotypic interactions, our approach enables the flexible design and construction of multi-component droplets. Methods using a hetero-interacting pair often require the co-expression of two scaffolding proteins to construct droplets, necessitating careful adjustment of expression ratios. In contrast, the homotypic and polymeric interactions of the YK tag allow for droplet formation with one scaffolding protein (YK-sfGFP). Clients can be easily recruited using the same YK tags. Fourth, YK peptides exhibit selectivity for specific anions such as ATP and GTP. The YK tag has the potential to enable the creation of molecular systems that utilize ATP as a fuel to drive droplet formation. Consequently, this study emphasizes the universality of de novo self-assembling peptides within cells.

The YK peptide also offers a platform for the bottom-up reconstitution of protein condensates in living cells. By building artificial protein condensates with different compositions or dynamic properties, we can evaluate the impacts on functions in living cells. Furthermore, by creating multi-component droplets with a biotin ligase, we expect to identify the endogenous proteome recruited into the designed droplets, thereby elucidating the interaction network39. This could contribute to the identification of factors essential for the biological functions involving protein condensates.

For the expression of YK-proteins in mammalian cells, the genes were cloned into a pCI-neo vector (Promega, E1841) by standard restriction digestion and ligation70. For the construction of NES-peptide-FP and YK-sfGFP, the oligo DNAs coding NES sequence, peptide tags were ordered (FASMAC), phosphorylated by T4 polynucleotide kinase (PNK, Takara), incubated at 70 °C for 20 min to inactivate PNK, 90 °C for 5 min to denature DNA, and cooled at −1 °C/min for DNA annealing. The annealed DNA fragments were inserted into the pCI-FP vector between NheI-EcoRI sites (NES) and EcoRI-SalI sites (YK peptides) or NheI-SalI sites (YK peptides). For the sfGFP (V206K), sfGFP (F223D), and sfYFP genes, point mutations were introduced by PCR using PrimeSTAR® Mutagenesis Basal Kit (Takara, R045A). As for NES-YK-mAG-Nck1(1-258), NES-YK-mCherry-N-WASP constructs, the DNA fragments of NES-YK were inserted into the pCI-mAG or pCI-mCherry vector between NheI-EcoRI sites. The genes of the Nck1(1-258) and N-WASP were cloned by nested-PCR from Human Brain, whole MarathonⓇ-Ready cDNA (Clontech, 639300). For all plasmids constructed in this study, we read the open reading frame by Sanger sequencing, and the analysis was supported by the Biomaterials Analysis Division, Tokyo Institute of Technology.

BL21-CodonPlus(DE3)-RIPL competent cells (Agilent) were transformed with a pET28a vector coding these YK9, YK11, YK13 or YK15-sfGFP or sfGFP. The expression was induced with 1 mM IPTG (isopropyl-β-D-thiogalactoside) and incubated at 37 °C for 5 hours. The cells were harvested by centrifugation (6000 g, 15 min at 4 °C), suspended in lysis buffer (20 mM Tris-HCl (pH 7.5), 300 mM NaCl, one mM PMSF (phenylmethanesulfonyl fluoride) and sonicated for 15 min (30-second interval) on ice. We purified YK13 and YK15-sfGFP from the inclusion body. The pellets obtained after lysate centrifugation (10,000 g, 30 min at 4 °C) were dissolved in equilibrium buffer (8 M urea, 10 mM Tris-HCl (pH 7.5), 300 mM NaCl buffer). The supernatants were mixed with equilibrated TALON resin for 40 min at 4 °C. After rinsed with the same buffer, the YK13 and YK15-sfGFP were collected by the addition of elution buffer (8 M urea, 10 mM Tris-HCl (pH 7.5), 270 mM NaCl, 150 mM imidazole). The solvent was exchanged by dialysis using Spectra/Por Dialysis membrane (MWCO; 12–14 kDa) into 10 mM Tris-HCl (pH 7.5). After removing insoluble fraction by centrifugation (10,000 g, 5 min, 4 °C), the YK13 and YK15-sfGFP solutions were afforded. The YK9 and YK11-sfGFP and untagged sfGFP were expressed and purified using our previously reported protocol. The concentration of protein was determined by 485 nm absorption.

Ac-YK13-NH2 and Ac-YK13(K6Y/Y7K)-NH2 peptides were synthesized by standard Fmoc solid phase synthesis. We used TentaGel® S-RAM as a resin, and the peptide chains were elongated by the mixture of Fmoc-amino acids (3 eq.), N,N-diisopropylethylamine (DIPEA, 6 eq.), HBTU (3 eq.) and HOBt (3 eq.). Fmoc deprotection was performed using 20% piperidine in NMP. Peptide cleavage from the resin and side chain deprotection was performed with trifluoroacetic acid (TFA)/triisopropylsilane/water (95:2.5:2.5 v/v). The peptide was precipitated with cold diethyl ether and purified by semi-preparative reverse phase HPLC using COSMOSIL 5C18-AR-300 or 5C18-AR-II packed column (10 × 250 mm). Eluents of 0.1% TFA ultra-pure water and 0.08% TFA acetonitrile flowed at 3.0 mL/min with a linear gradient. Peptides were detected by absorption at 220 nm, and the fractions were collected. The purity of the isolated peptides was confirmed by RP-HPLC and analyzed by MALDI-TOF MS (CHCA matrix).

For molecular modeling of YK13 octamer, we used ColabFold71. For the complex model with ATP, we manually placed ATP molecules (24) into the bilayer structure of the YK13 octamer obtained by ColabFold. We carried out the energy minimization for 300 steps with restraints on heavy atoms. Then, the systems were equilibrated for 500 ps under an NVT condition (300 K) and for 500 ps under an NPT condition (300 K and 1 bar) with the same restraints above. Finally, 100 ns production runs were conducted without restraints. We performed the MD simulations using the Amber ff14SB force field, TIP3P water model, and Carlson for ATP72,73. The box size is around 100 × 90 × 100 Å3. The temperature and pressure were regulated by the Berendsen thermostat and barostat, respectively. The time step was set to 2 fs, and the trajectories were recorded every 0.1 ns. PyMOL was used for visualization of the structures.

COS-7 cells were provided by RIKEN BRC through the National BioResource Project of MEXT/AMED, Japan. These cells were cultured in D-MEM (043-30085, FUJIFILM) supplemented with 10% FBS (F7524, Sigma Aldrich) and penicillin/streptomycin (Nacalai Tesque) under a humid 5% CO2 atmosphere at 37 °C. A total of 1–2×104 cells/well were seeded on an EZVIEW 96-well culture plate (AGC) 1 day before transfection. In accordance with the manual, the COS-7 cells were transfected with 100 ng of plasmid DNA using Lipofectamine® 3000 (Thermo Fisher Scientific). Two days after transfection, the medium was exchanged with D-MEM (no phenol red, 044-32955, FUJIFILM) and the cells were assessed. For the construction of multi-component droplets, we transfected COS-7 cells using a mixture of DNA (100, 5, and 20 ng of plasmids expressing NES-YK13-sfGFP, NES-YK13-mCherry, or NES-YK13-miRFP670-myc, respectively).

The transfected cells were observed on a Zeiss LSM 780 confocal microscope equipped with a 63×/1.4NA oil immersion objective using ZEN2.3 (black edition) software (405 nm diode laser for mTagBFP2 and moxBFP; 488 nm argon laser for mAG, sfGFP, and AG; 561 nm DPSS laser for mKO2 and mCherry; 633 nm HeNe633 laser for miRFP670) or Leica TCS SP8 confocal microscope equipped with a 63×/1.4NA oil immersion objective. Partition coefficients of the granules were calculated using our previously reported protocol36 with slight modifications. The whole-cell areas were defined as 50 presential thresholds. The sfGFP-rich phase areas were defined as the pixels displaying 1.5 times greater value than the average fluorescence intensity of the entire cell. The other area in whole cells was set as the bulk phase. The fluorescence ratio of the sfGFP-rich phase/bulk phase was calculated as the partition coefficient. For FRAP analysis, the circular regions of interest were bleached with a 488 nm laser at 100% power, followed by time-lapse observation (0.5–2 s intervals). The fluorescence recovery curves were fitted in a single exponential mode using ZEN2.3 to calculate the half-time and mobile fraction. For in vitro studies, protein solutions were placed on an EZVIEW 96-well culture plate (AGC) and observed similarly.

Two days after transfection, the medium was exchanged with D-MEM (no phenol red, 044-32955, FUJIFILM). The fluorescence images for COS-7 cells co-expressing Nck1(1–258) were obtained by using Leica TCS SP8 confocal microscope equipped with a 63×/1.4NA oil immersion objective (488 nm argon laser for mAG; 561 nm DPSS laser for mCherry; 633 nm HeNe633 laser for phalloidin-iFluor633). For FRAP analysis, the circular regions of interest were bleached with a 488 nm laser at 100% power, followed by time-lapse observation (0.5 s intervals). The cells were rinsed with PBS and then fixed with a 4% paraformaldehyde phosphate buffer solution (161-20141, FUJIFILM) for 30 minutes at room temperature (~20 °C). After rinsing twice with PBS, the fixed cells were treated with 0.1% tween-20 PBS for two minutes, followed by washing with PBS. Actin fibrils were stained by phalloidin-iFluor633 (ab176758, abcam, 1/1000 dilution) for 60 min at room temperature and then rinsed three times with PBS. The actin intensity over granules was calculated using our reported protocol36. First, we determined the area of the green fluorescent granules by setting a value at twice Otsu’s threshold. Next, we subtracted the background signal, defined at the 50-percentile threshold, and calculated the fluorescence intensity within the defined granule area for all three channels. To obtain the high-resolution images, we also observed cells using a Zeiss LSM 780 confocal microscope equipped with an Airyscan detector using ZEN2.3 (black edition) software.

The fluorescence intensity and anisotropy of transfected cells or protein solutions were measured using an ARVO MX plate reader with Wallac1420 Workstation [excitation: 485/14 nm (F485 filter), emission: 535/25 nm (F535 filter)].

All solutions and solvents were filtered through a 0.45 µm PTFE membrane (Millipore) in advance. DLS measurements of the 5 µM sfGFP or YK-sfGFP solution with or without 1 mM ATP were conducted using Zetasizer Nano ZS90 (Malvern) at 25 °C. All measurements were performed in triplicate.

We mixed the peptide (50 µM), thioflavin-T (25 µM), and anion species in TBS buffer (10 mM Tris-HCl, 150 mM NaCl, pH 7.5). The thioflavin-T fluorescence intensities were measured using an ARVO MX plate reader (PerkinElmer) with a Wallac1420 Workstation [excitation: 450/6 nm (P450 filter), emission: 535/25 nm (F535 filter)]. For the time-course experiment (evaluation of reversibility by ATPase), fluorescence was measured using a HITACHI F-7000 fluorescence spectrophotometer (Ex: 420 nm, Em: 450–550 nm).

YK13 peptide or YK13-sfGFP solutions were dropped on collodion-coated copper EM grids (Nisshin EM). For 60 s incubation, the solutions were removed and the grid surface was rinsed with filtered ultra-pure water. The YK13 peptide samples were stained by a 10% EM stainer (Nishin EM). The YK13-sfGFP samples were stained by Nano-W® (twofold diluted), followed by rinsing with ultra-pure water. All images were taken using a JEOL 1400Plus electron microscope.

The sample size for biochemical studies was set at three. For cell image analysis, we obtained three to ten different images from various fields of view in a single biological experiment. We conducted three biologically independent experiments. For actin intensity measurements, we analyzed more than 25 cells per condition according to the appropriate literature31. Statistical comparisons between two groups were performed using unpaired two-tailed Student’s t-tests. The experiments shown in Figs. 1f, 3d, 3e, 4c, and 5b were independently repeated four, two, three, four, and three times, respectively, with similar results.

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

Data supporting the findings of this study are included in the article, along with supplementary information. Source data are provided with this paper.

Banani, S. F., Lee, H. O., Hyman, A. A. & Rosen, M. K. Biomolecular condensates: Organizers of cellular biochemistry. Nat. Rev. Mol. Cell Biol. 18, 285–298 (2017).

Article CAS PubMed PubMed Central Google Scholar

Shin, Y. & Brangwynne, C. P. Liquid phase condensation in cell physiology and disease. Science 357, eaaf4382 (2017).

Article PubMed Google Scholar

Hirose, T., Ninomiya, K., Nakagawa, S. & Yamazaki, T. A guide to membraneless organelles and their various roles in gene regulation. Nat. Rev. Mol. Cell Biol. 24, 288–304 (2023).

Article CAS PubMed Google Scholar

Banani, S. F. et al. Compositional Control of Phase-Separated Cellular Bodies. Cell 166, 651–663 (2016).

Article CAS PubMed PubMed Central Google Scholar

Yoshikawa, M., Yoshii, T., Ikuta, M. & Tsukiji, S. Synthetic Protein Condensates That Inducibly Recruit and Release Protein Activity in Living Cells. J. Am. Chem. Soc. 143, 6434–6446 (2021).

Article CAS PubMed Google Scholar

Nakamura, H. et al. Intracellular production of hydrogels and synthetic RNA granules by multivalent molecular interactions. Nat. Mater. 17, 79–88 (2018).

Article ADS CAS PubMed Google Scholar

Schuster, B. S. et al. Controllable protein phase separation and modular recruitment to form responsive membraneless organelles. Nat. Commun. 9, 2985 (2018).

Article ADS PubMed PubMed Central Google Scholar

Shin, Y. et al. Spatiotemporal Control of Intracellular Phase Transitions Using Light-Activated optoDroplets. Cell 168, 159–171 (2017).

Article CAS PubMed Google Scholar

Qian, Z. G., Huang, S. C. & Xia, X. X. Synthetic protein condensates for cellular and metabolic engineering. Nat. Chem. Biol. 18, 1330–1340 (2022).

Article CAS PubMed Google Scholar

Bracha, D., Walls, M. T. & Brangwynne, C. P. Probing and engineering liquid-phase organelles. Nat. Biotechnol. 37, 1435–1445 (2019).

Article CAS PubMed Google Scholar

Sementa, D. et al. Sequence-Tunable Phase Behavior and Intrinsic Fluorescence in Dynamically Interacting Peptides. Angew. Chem. - Int. Ed. 62, e202311479 (2023).

Article CAS Google Scholar

Niu, J., Qiu, C., Abbott, N. L. & Gellman, S. H. Formation of versus Recruitment to RNA-Rich Condensates: Controlling Effects Exerted by Peptide Side Chain Identity. J. Am. Chem. Soc. 144, 10386–10395 (2022).

Article CAS PubMed PubMed Central Google Scholar

Aumiller, W. M. & Keating, C. D. Phosphorylation-mediated RNA/peptide complex coacervation as a model for intracellular liquid organelles. Nat. Chem. 8, 129–137 (2016).

Article CAS PubMed Google Scholar

Perry, S. L. et al. Chirality-selected phase behaviour in ionic polypeptide complexes. Nat. Commun. 6, 6052 (2015).

Article ADS CAS PubMed Google Scholar

Donau, C. et al. Active coacervate droplets as a model for membraneless organelles and protocells. Nat. Commun. 11, 5167 (2020).

Article ADS CAS PubMed PubMed Central Google Scholar

Cakmak, F. P., Choi, S., Meyer, M. C. O., Bevilacqua, P. C. & Keating, C. D. Prebiotically-relevant low polyion multivalency can improve functionality of membraneless compartments. Nat. Commun. 11, 5949 (2020).

Article ADS CAS PubMed PubMed Central Google Scholar

Abbas, M., Lipiński, W. P., Nakashima, K. K., Huck, W. T. S. & Spruijt, E. A short peptide synthon for liquid–liquid phase separation. Nat. Chem. 13, 1046–1054 (2021).

Article CAS PubMed Google Scholar

Kubota, R., Torigoe, S. & Hamachi, I. Temporal Stimulus Patterns Drive Differentiation of a Synthetic Dipeptide-Based Coacervate. J. Am. Chem. Soc. 144, 15155–15164 (2022).

Article CAS PubMed Google Scholar

Hilditch, A. T. et al. Assembling membraneless organelles from de novo designed proteins. Nat. Chem. 16, 89–97 (2023).

Article PubMed PubMed Central Google Scholar

Hilditch, A. T. et al. Maturation and Conformational Switching of a De Novo Designed Phase-Separating Polypeptide. J. Am. Chem. Soc. 146, 10240–10245 (2024).

Article CAS PubMed PubMed Central Google Scholar

Yeong, V., wen Wang, J., Horn, J. M. & Obermeyer, A. C. Intracellular phase separation of globular proteins facilitated by short cationic peptides. Nat. Commun. 13, 7882 (2022).

Article ADS CAS PubMed PubMed Central Google Scholar

Zhang, Q. et al. Visualizing Dynamics of Cell Signaling In Vivo with a Phase Separation-Based Kinase Reporter. Mol. Cell 69, 334–346.e4 (2018).

Article MathSciNet CAS PubMed PubMed Central Google Scholar

Ramšak, M. et al. Programmable de novo designed coiled coil-mediated phase separation in mammalian cells. Nat. Commun. 14, 7973 (2023).

Article ADS PubMed PubMed Central Google Scholar

Kato, M. et al. Cell-free formation of RNA granules: Low complexity sequence domains form dynamic fibers within hydrogels. Cell 149, 753–767 (2012).

Article CAS PubMed PubMed Central Google Scholar

Luo, F. et al. Atomic structures of FUS LC domain segments reveal bases for reversible amyloid fibril formation. Nat. Struct. Mol. Biol. 25, 341–346 (2018).

Article CAS PubMed Google Scholar

Ray, S. et al. α-Synuclein aggregation nucleates through liquid–liquid phase separation. Nat. Chem. 12, 705–716 (2020).

Article CAS PubMed Google Scholar

Patel, A. et al. A Liquid-to-Solid Phase Transition of the ALS Protein FUS Accelerated by Disease Mutation. Cell 162, 1066–1077 (2015).

Article CAS PubMed Google Scholar

Hughes, M. P. et al. Atomic structures of low-complexity protein segments reveal kinked b sheets that assemble networks. Science 359, 698–701 (2018).

Article ADS CAS PubMed PubMed Central Google Scholar

Rivera, G. M., Briceño, C. A., Takeshima, F., Snapper, S. B. & Mayer, B. J. Inducible Clustering of Membrane-Targeted SH3 Domains of the Adaptor Protein Nck Triggers Localized Actin Polymerization. Curr. Biol. 14, 11–22 (2004).

Article CAS PubMed Google Scholar

Rivera, G. M., Vasilescu, D., Papayannopoulos, V., Lim, W. A. & Mayer, B. J. A Reciprocal Interdependence between Nck and PI(4,5)P2 Promotes Localized N-WASp-Mediated Actin Polymerization in Living Cells. Mol. Cell 36, 525–535 (2009).

Article CAS PubMed PubMed Central Google Scholar

Case, L. B., Zhang, X., Ditlev, J. A. & Rosen, M. K. Stoichiometry controls activity of phase-separated clusters of actin signaling proteins. Science 363, 1093–1097 (2019).

Article ADS CAS PubMed PubMed Central Google Scholar

Zhang, S. Fabrication of novel biomaterials through molecular self-assembly. Nat. Biotechnol. 21, 1171–1178 (2003).

Article CAS PubMed Google Scholar

Caplan, M. R., Moore, P. N., Zhang, S., Kamm, R. D. & Lauffenburger, D. A. Self-assembly of a β-sheet protein governed by relief of electrostatic repulsion relative to van der Waals attraction. Biomacromolecules 1, 627–631 (2000).

Article CAS PubMed Google Scholar

Yokoi, H., Kinoshita, T. & Zhang, S. Dynamic reassembly of peptide RADA16 nanofiber scaffold. Proc. Natl Acad. Sci. 102, 8414–8419 (2005).

Article ADS CAS PubMed PubMed Central Google Scholar

Ozbas, B., Kretsinger, J., Rajagopal, K., Schneider, J. P. & Pochan, D. J. Salt-triggered peptide folding and consequent self-assembly into hydrogels with tunable modulus. Macromolecules 37, 7331–7337 (2004).

Article ADS CAS Google Scholar

Miki, T. et al. Intracellular artificial supramolecules based on de novo designed Y15 peptides. Nat. Commun. 12, 3412 (2021).

Article ADS CAS PubMed PubMed Central Google Scholar

Miki, T., Kajiwara, K., Nakayama, S., Hashimoto, M. & Mihara, H. Effects of Hydrophobic Residues on the Intracellular Self-Assembly of de Novo Designed Peptide Tags and Their Orthogonality. ACS Synth. Biol. 11, 2144–2153 (2022).

Article CAS PubMed Google Scholar

Miki, T., Hashimoto, M., Nakai, T. & Mihara, H. A guide-tag system controlling client enrichment into Y15 peptide-based granules for an in-cell protein recruitment assay. Chem. Commun. 57, 11338–11341 (2021).

Article CAS Google Scholar

Hashimoto, M., Miki, T., Niwa, T. & Mihara, H. Proximity labeling and identification of endogenous client proteins recruited to Y15-based artificial granules tethering a bait protein. J. Pept. Sci. 30, e3536 (2024).

Article CAS PubMed Google Scholar

Fukuda, M., Gotoh, I., Gotoh, Y. & Nishidaf, E. Cytoplasmic localization of mitogen-activated protein kinase kinase directed by its NH2-terminal, leucine-rich short amino acid sequence, which acts as a nuclear export signal. J. Biol. Chem. 271, 20024–20028 (1996).

Article CAS PubMed Google Scholar

Guillén-Boixet, J. et al. RNA-Induced Conformational Switching and Clustering of G3BP Drive Stress Granule Assembly by Condensation. Cell 181, 346–361 (2020).

Article PubMed PubMed Central Google Scholar

García-Mata, R., Bebök, Z., Sorscher, E. J. & Sztul, E. S. Characterization and dynamics of aggresome formation by a cytosolic GFP- chimera. J. Cell Biol. 146, 1239–1254 (1999).

Article PubMed PubMed Central Google Scholar

Chan, F. T. S., Kaminski, C. F. & Schierle, G. S. K. HomoFRET fluorescence anisotropy imaging as a tool to study molecular self-assembly in live cells. ChemPhysChem 12, 500–509 (2011).

Article CAS PubMed Google Scholar

Bader, A. N., Hofman, E. G., Voortman, J., Van Bergen En Henegouwen, P. M. P. & Gerritsen, H. C. Homo-FRET imaging enables quantification of protein cluster sizes with subcellular resolution. Biophys. J. 97, 2613–2622 (2009).

Article ADS CAS PubMed PubMed Central Google Scholar

Traut, T. W. Physiological concentrations of purines and pyrimidines. Mol. Cell. Biochem. 140, 1–22 (1994).

Article CAS PubMed Google Scholar

Zhang, J. et al. Detection and Imaging of Aβ1-42 and Tau Fibrils by Redesigned Fluorescent X-34 Analogues. Chem. - A Eur. J. 24, 7210–7216 (2018).

Article ADS CAS Google Scholar

Barth, A. Infrared spectroscopy of proteins. Biochim. Biophys. Acta - Bioenerg. 1767, 1073–1101 (2007).

Article CAS Google Scholar

Gout, E., Rébeillé, F., Douce, R. & Bligny, R. Interplay of Mg2+, ADP, and ATP in the cytosol and mitochondria: Unravelling the role of Mg2+ in cell respiration. Proc. Natl Acad. Sci. USA 111, E4560–E4567 (2014).

Article ADS CAS PubMed PubMed Central Google Scholar

Younan, N. D. & Viles, J. H. A Comparison of Three Fluorophores for the Detection of Amyloid Fibers and Prefibrillar Oligomeric Assemblies. ThT (Thioflavin T); ANS (1-Anilinonaphthalene-8-sulfonic Acid); and bisANS (4,4′-Dianilino-1,1′-binaphthyl-5,5′-disulfonic Acid). Biochemistry 54, 4297–4306 (2015).

Article CAS PubMed Google Scholar

Gui, X. et al. Structural basis for reversible amyloids of hnRNPA1 elucidates their role in stress granule assembly. Nat. Commun. 10, 2006 (2019).

Article ADS PubMed PubMed Central Google Scholar

Lobas, M. A. et al. A genetically encoded single-wavelength sensor for imaging cytosolic and cell surface ATP. Nat. Commun. 10, 711 (2019).

Article ADS CAS PubMed PubMed Central Google Scholar

Karasawa, S., Araki, T., Yamamoto-Hino, M. & Miyawaki, A. A Green-emitting Fluorescent Protein from Galaxeidae Coral and Its Monomeric Version for Use in Fluorescent Labeling. J. Biol. Chem. 278, 34167–34171 (2003).

Article CAS PubMed Google Scholar

Cranfill, P. J. et al. Quantitative assessment of fluorescent proteins. Nat. Methods 13, 557–562 (2016).

Article CAS PubMed PubMed Central Google Scholar

Scott, D. J. et al. A Novel Ultra-Stable, Monomeric Green Fluorescent Protein for Direct Volumetric Imaging of Whole Organs Using CLARITY. Sci. Rep. 8, 667 (2018).

Article ADS PubMed PubMed Central Google Scholar

Fletcher, J. M. et al. A basis set of de novo coiled-Coil peptide oligomers for rational protein design and synthetic biology. ACS Synth. Biol. 1, 240–250 (2012).

Article CAS PubMed Google Scholar

Ditlev, J. A., Case, L. B. & Rosen, M. K. Who’s In and Who’s Out—Compositional Control of Biomolecular Condensates. J. Mol. Biol. 430, 4666–4684 (2018).

Article CAS PubMed PubMed Central Google Scholar

Youn, J. Y. et al. Properties of Stress Granule and P-Body Proteomes. Mol. Cell 76, 286–294 (2019).

Article CAS PubMed Google Scholar

Ishov, A. M. et al. PML is critical for ND10 formation and recruits the PML-interacting protein Daxx to this nuclear structure when modified by SUMO-1. J. Cell Biol. 147, 221–233 (1999).

Article CAS PubMed PubMed Central Google Scholar

Jain, S. et al. ATPase-Modulated Stress Granules Contain a Diverse Proteome and Substructure. Cell 164, 487–498 (2016).

Article CAS PubMed PubMed Central Google Scholar

Espinosa, J. R. et al. Liquid network connectivity regulates the stability and composition of biomolecular condensates with many components. Proc. Natl Acad. Sci. USA 117, 13238–13247 (2020).

Article ADS CAS PubMed PubMed Central Google Scholar

Yamaguchi, H. et al. Molecular mechanisms of invadopodium formation: The role of the N-WASP-Arp2/3 complex pathway and cofilin. J. Cell Biol. 168, 441–452 (2005).

Article CAS PubMed PubMed Central Google Scholar

Banjade, S. et al. Conserved interdomain linker promotes phase separation of the multivalent adaptor protein Nck. Proc. Natl Acad. Sci. USA 112, E6426–E6435 (2015).

Article CAS PubMed PubMed Central Google Scholar

Okrut, J., Prakash, S., Wu, Q., Kelly, M. J. S. & Taunton, J. Allosteric N-WASP activation by an inter-SH3 domain linker in Nck. Proc. Natl Acad. Sci. USA 112, E6436–E6445 (2015).

Article ADS CAS PubMed PubMed Central Google Scholar

Ditlev, J. A. et al. A composition-dependent molecular clutch between T cell signaling condensates and actin. Elife 8, 1–44 (2019).

Article Google Scholar

Li, P. et al. Phase transitions in the assembly of multivalent signalling proteins. Nature 483, 336–340 (2012).

Article ADS CAS PubMed PubMed Central Google Scholar

Ditlev, J. A. et al. Stoichiometry of Nck-dependent actin polymerization in living cells. J. Cell Biol. 197, 643–658 (2012).

Article CAS PubMed PubMed Central Google Scholar

Theillet, F. X. et al. Physicochemical properties of cells and their effects on intrinsically disordered proteins (IDPs). Chem. Rev. 114, 6661–6714 (2014).

Article CAS PubMed PubMed Central Google Scholar

Li, R. et al. Programming cell-surface signaling by phase-separation-controlled compartmentalization. Nat. Chem. Biol. 18, 1351–1360 (2022).

Article CAS PubMed Google Scholar

Kamiyama, D. et al. Versatile protein tagging in cells with split fluorescent protein. Nat. Commun. 7, 11046 (2016).

Article ADS CAS PubMed PubMed Central Google Scholar

Yang, Q and Miki, T Characterization of peptide-fused protein assemblies in living cells. Methods Enzymol 697, 293–319 (2024).

Mirdita, M. et al. ColabFold: making protein folding accessible to all. Nat. Methods 19, 679–682 (2022).

Article CAS PubMed PubMed Central Google Scholar

Maier, J. A. et al. ff14SB: Improving the Accuracy of Protein Side Chain and Backbone Parameters from ff99SB. J. Chem. Theory Comput. 11, 3696–3713 (2015).

Article CAS PubMed PubMed Central Google Scholar

Meagher, K. L., Redman, L. T. & Carlson, H. A. Development of polyphosphate parameters for use with the AMBER force field. J. Comput. Chem. 24, 1016–1025 (2003).

Article CAS PubMed Google Scholar

Download references

We thank Open Research Facilities for Life Science and Technology, Tokyo Institute of Technology for equipment and technical support. We thank the Biomaterials Analysis Division, Tokyo Institute of Technology, for DNA sequencing and technical assistance with TEM observation. We thank Kao Corporation for their generous support. This work was supported by the JSPS KAKENHI Grant Number 21K14739 and 23H05408, and JST FOREST Program (Grant Number JPMJFR2251, Japan). T.M. is the recipient of these grants. M. H. is supported by the JST, the establishment of university fellowships towards the creation of science technology innovation (Grant Number JPMJFS2112) and the JSPS KAKENHI Grant Number 23KJ0920.

These authors contributed equally: Takayuki Miki, Masahiro Hashimoto

School of Life Science and Technology, Tokyo Institute of Technology, 4259 Nagatsuta-cho, Midori-ku, Yokohama, Kanagawa, 226-8501, Japan

Takayuki Miki, Masahiro Hashimoto, Hiroki Takahashi, Masatoshi Shimizu, Sae Nakayama, Tadaomi Furuta & Hisakazu Mihara

Department of Chemistry and Biotechnology, School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo, 113-0033, Japan

Takayuki Miki & Masatoshi Shimizu

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

T.M., M.H., and H.M. designed the project. T.M., M.H., and M.S. synthesized peptides and performed assembly tests. T.M., M.H., H.T., and S.N. constructed the expression plasmids. M.H., H.T., and T.M. expressed and purified proteins and tested their assembling properties in a test tube. T.M., M.H., H.T., and S.N. perform cell experiments. T.F. performed MD simulations. According to discussion with all authors, the manuscript was written by M.H., H.M. and T.M.

Correspondence to Takayuki Miki.

The authors declare no competing interests.

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

Miki, T., Hashimoto, M., Takahashi, H. et al. De novo designed YK peptides forming reversible amyloid for synthetic protein condensates in mammalian cells. Nat Commun 15, 8503 (2024). https://doi.org/10.1038/s41467-024-52708-5

Download citation

Received: 01 March 2024

Accepted: 18 September 2024

Published: 18 October 2024

DOI: https://doi.org/10.1038/s41467-024-52708-5

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative