Emergence of Anisotropic Subcompartments via Coassembly of Hierarchically Ordered G‑Quadruplexes and Fluid Polylysine in Droplet-Based Compartments

Abstract

Liquid–liquid phase separation (LLPS) plays a key role in cellular organization, enabling the formation of dynamic compartments that provide spatial and temporal control over biochemical processes. Although LLPS systems are typically fluidic, recent studies have revealed that specific molecular constituents can induce an internal structure. Here, we show that droplet formation between guanine quadruplexes (G4s) and polylysine peptides triggers G4-driven internal structural ordering. Using birefringence-retardation imaging, confocal fluorescence imaging, as well as X-ray diffraction and scattering techniques, we demonstrate that anisotropic subcompartments are gradually developed at specific points of their droplet–solution interfaces. The G4s possess intrinsic molecular rigidity due to their stacked G-quartet structures, and their interaction with the flexible polylysine peptides enables the formation of a hexagonal columnar phase (a = 3.64 nm; c = 0.86 nm; nine units per turn). This highly ordered subcompartment is enriched in low-fluidity G4s, while the peptides remain dynamically diffuse throughout the entire compartment. This supramolecular platform provides insights into the cooperative roles of structural order and molecular mobility in phase-separated systems, offering a foundation for the bottom-up design of synthetic condensates inspired by biomolecular organization.

Liquid–liquid phase separation (LLPS) has been widely recognized as a key mechanism in the organization of biomolecules within cells over the past several years. − This process leads to the formation of membrane-less organelles such as P-granules, nucleoli, and stress granules, which are highly dynamic compartments with internal fluidity that support transient and reversible biological functions under fluctuating cellular environments. In parallel, biological systems also rely on a wide variety of ordered structures such as filamentous actin networks, microtubule assemblies, fibrillar protein assemblies, and lipid bilayer membranes. , These ordered architectures play an essential role in maintaining cellular integrity by constraining spatial relationships. These examples highlight that the complex functional capabilities of living systems are not based exclusively on either fluidity or ordering but instead on a sophisticated integration of both.

Recent studies have increasingly demonstrated that LLPS systems can exhibit internal structural ordering under specific molecular and environmental conditions. Stress granules, a representative cellular condensate formed via LLPS, become enriched in the protein TDP-43, which subsequently undergoes intracondensate demixing and assembles into ordered fibrils. Similar LLPS-coupled fibrillization has been reported for FUS, α-synuclein, and tau, implicating this phenomenon in the pathogenesis of neurodegenerative diseases. Protein condensation has also been suggested to facilitate microtubule nucleation. , Thus, living systems exploit LLPS as a reaction field that both concentrates components and promotes the emergence of ordered architectures. Deciphering the mechanisms by which these phenomena occur may thus reveal unrecognized driving forces in cellular events.

To elucidate the mechanisms of LLPS-driven ordering, researchers have developed model condensates in which ordered internal phases, such as liquid crystalline (LC) phases, are induced within a part of or throughout the entire LLPS droplet. In particular, associative LLPS systems composed of DNAs and peptides have attracted considerable attention. In associative LLPS, electrostatic attraction between oppositely charged macromolecules serves as the primary driving force for the formation of condensed liquid phases. In this context, Shakya et al. discovered that, with suitable tuning of parameters such as DNA sequence, ionic strength, and the presence of nucleotide triphosphates, double-stranded DNAs (dsDNA) and positively charged peptides can form LLPS droplets with internal ordering. Furthermore, Fraccia and Jia characterized a variety of LC mesophases in similar systems. A key feature underlying these systems is the interplay between mechanically rigid and flexible components at the molecular level. Specifically, the rigid short dsDNA (≤20 bp) , acts as a mesogenic unit, while the flexible peptide plays a facilitative role in promoting alignment. Taken together, these insights suggest a broader principle, i.e., the emergence of anisotropic ordering in associative LLPS droplets may occur via a latent, universal mechanism inherent to structurally rigid oligonucleotide-based architectures that is not limited to short dsDNA. ,,,

Guanine quadruplexes (G4s) are noncanonical secondary structures formed by four strands of DNA or RNA and are known to play regulatory roles in various cellular functions. , They consist of a π–π stacked unit of G-quartets, which are planar assemblies of four guanine nucleobases linked via Hoogsteen-type hydrogen bonding , and form LC phases under specific conditions. − The architecture of the G4s imparts them with exceptional rigidity, exceeding that of dsDNA. In the past few years, researchers have reported the ability of G4s to promote droplet formation and to maintain structural integrity within droplets. − However, despite their inherent ordering potential, details of the spatial arrangement of G4s inside LLPS droplets remain elusive. Developing an accurate understanding of their structures in droplets is crucial for elucidating their roles in diverse biological processes.

In this study, we report the hierarchical ordered organization of G4s within LLPS droplets via coassembly with fluid polylysine. Droplets formed via associative LLPS of G4-forming DNAs and polylysine peptides gradually develop anisotropic subcompartments at specific points on their droplet–solution interfaces. The anisotropy of the subcompartments arises from the hierarchical alignment of the G4s into a hexagonal columnar phase, while the polylysine components retain dynamic fluidity. This system reflects fundamental physicochemical principles underlying multiphase organization, a feature also discussed in the context of membrane-less compartments, , thereby providing a basis for deeper insights into cooperative molecular assemblies and the design of functionally complex synthetic structures.

Results and Discussion

Basic Experimental Design

For the G4s, we adopted a 22-nt parallel G4-forming DNA derived from the oncogenic c-myc promoter (dPu22; Figure A and Table S1) and used Tris–EDTA (TE) buffer conditions, both of which have previously been described in LLPS studies involving G4 structures. To minimize variables other than charge and simplify the system, polylysine peptides were used as the counterparts for associative LLPS with dPu22. Various polylysine peptides were examined, and the deca-l-lysine peptide (K10; Figure B and Table S1) was selected as the shortest chemically defined peptide capable of forming droplets with dPu22 (Figure S1).

1.

(A–B) Schematic representations of (A) dPu22 and (B) K10. (C–D) Experimental phase diagrams as a function of the K10 and dPu22 concentrations; inset: bright-field images of the indicated K10–dPu22 mixtures. The samples contained [K10] = 10.5, 21, 26.6, 42, 52.5, 84, 105, 168, 210, or 420 μM and [dPu22] = 0.5, 1, 2, 4, 6, 8, 10, 30, 40, 50, 80, 100, 200, 500, or 1000 μM in TE buffer (pH 7.4). Images were acquired (C) 4 and (D) 24 h after sample preparation. The data at 8, 12, 16, and 20 h are presented in Figure S2.

Phase Diagrams of K10–dPu22 Mixtures

Initially, we generated phase diagrams over the range of 10.5–420 μM for K10 and 0.5–1000 μM for dPu22 to investigate the conditions under which the K10–dPu22 droplets form. Mixtures of DNA, buffer, and peptide were prepared, and bright-field images were acquired every 4 h. Figure C and D shows the phase diagrams at 4 and 24 h. Herein, droplets were defined as optically detectable assemblies with spherical or quasi-spherical shapes, whereas aggregates were defined as assemblies whose shapes could not be determined (for representative images, see Figure S2). Droplets and aggregates were observed 4 h after sample preparation for K10 concentrations >21 μM within a specific range of dPu22 concentrations (Figure C). For a given K10 concentration, the formation of optically detectable assemblies was suppressed under charge-imbalanced conditions; for example, at a fixed K10 concentration of 21 μM, assemblies were observed at 10 μM dPu22 (1:1 charge ratio), while assemblies were absent at either higher or lower charge ratios. This trend is consistent with the typical reentrant phase transition observed in associative LLPS between positively charged peptides and negatively charged DNAs. , The range over which visible droplets and aggregates were observed expanded gradually over time until no further changes in the droplet-forming area in the phase diagram were observed after 16 h (Figure S2). Concurrently, the roundness of the droplets gradually decreased (Figure S3), and their shape transitioned from spherical to quasi-spherical (inset of Figure D).

Formation of Anisotropic Subcompartments within K10–dPu22 Droplets

The transformation of the droplets from spherical to quasi-spherical suggests the emergence of anisotropic structures within them. To further investigate this possibility, we analyzed birefringence retardation, which indirectly reflects the degree of molecular orientation associated with anisotropic internal structures. The time-course of the birefringence retardation was observed under droplet-forming conditions with 210 μM K10 and 100 μM dPu22, at which the charge ratio (+:−) was 1:1. Images were captured at 1-hour intervals (Figure S4 and Movie S1). Representative frames are presented in Figure , and the corresponding birefringence-retardation values are given in Figure S5. These images and videos showed only negligible birefringence retardation comparable to the background level for the first 5 h after sample preparation, indicating an isotropic nature of the initial droplet-based compartments, which are hereafter termed the mother compartment. After 5 h, distinct birefringent puncta began to appear at the droplet edges (indicated by the filled black triangles in Figure ) and gradually developed into anisotropic subcompartments, whereby the birefringence retardation reached approximately 50 nm by 9 h. The formation of puncta was observed more frequently in larger droplets (Figure S6), a trend reminiscent of nucleation-driven processes such as protein crystallization and amyloid formation in phase-separated droplets. , Over time, the shape of droplets gradually shifted from spherical to quasi-spherical, suggesting that the subcompartments exhibit a rigid crystalline-like nature rather than a fluidic liquid-like one. The droplets continued to undergo frequent coalescence even after overnight incubation (red, purple, and blue arrows in Figure ), which implies that they retained liquid-like properties.

2.

Time-course of the bright-field and birefringence-retardation images of the K10–dPu22 droplets. The filled black triangles at 5 h indicate the emergence of birefringent puncta. The red, purple, and blue arrows at 18–21 h represent the coalescence and fusion of multiple droplets into a single droplet. Images at 4–10 h and 18–21 h were taken from different fields; full time-course data at 1 h intervals across all fields are shown in Figure S4 and Movie S1. Samples contained [K10] = 210 μM and [dPu22] = 100 μM (1:1 charge ratio) in a TE buffer (pH 7.4).

Detectable birefringence retardation was observed in the formed assemblies, even those under charge-imbalanced conditions (charge ratios (+:−) of 1:2.5, 12.5:1, and 50:1; Figure S7), even though the mean per-pixel retardation values over the entire image were lower than those under charge-balanced conditions. This could be attributed to the smaller droplet sizes typically formed under charge imbalance and the less frequent puncta formation in smaller droplets (Figure S6). Thus, although the magnitude of retardation varies with the charge ratio between K10 and dPu22, these results demonstrate that anisotropic molecular organization within droplets is a robust phenomenon that occurs under a range of conditions.

Distribution and Fluidity of K10 and dPu22 in the Mother Compartments and the Subcompartments

In addition to the birefringence-retardation measurements, fluorescence imaging experiments were carried out to reveal the distribution and fluidity of the two components. For that purpose, we used K10 and dPu22 labeled with fluorescein (FAM) at the side chain of an N-terminal lysine and tetramethyl rhodamine (TMR) at the 3′-terminus of DNA, respectively. K10–dPu22 droplets were prepared to contain 0.5% fluorophore-labeled compounds and incubated for 24 h in order to form the anisotropic subcompartments.

Confocal fluorescence imaging revealed that the FAM-labeled K10 was predominantly localized in the mother compartment (FAM-K10 in Figure A). The FAM fluorescence intensity in the subcompartment was approximately half of that in the mother compartment and higher than that of the surrounding solution, which shows the partial distribution of K10 into the subcompartment (FAM-K10 in Figure B). In contrast, the fluorescence of TMR-labeled dPu22 was observed in both the mother compartment and the subcompartment (dPu22-TMR in Figure A) and was stronger in the subcompartment, indicating its preferential accumulation there (dPu22-TMR in Figure B). These results demonstrate that the initial mother compartment is enriched in both K10 and dPu22 compared with the surrounding solution. Over time, dPu22 becomes increasingly concentrated, while K10 is progressively excluded. Consequently, anisotropic subcompartments emerge from the isotropic mother compartments. Time-lapse images demonstrate that this change in distribution patterns emerges within ∼1 min (Figure S8 and Movie S2), which is also reminiscent of nucleation-driven processes. −

3.

Distribution of K10 and dPu22 in a K10–dPu22 droplet containing an isotropic mother compartment and an anisotropic subcompartment. (A) Confocal fluorescence images. (B) Fluorescence intensity profile along the cyan line in (A). (C) Confocal images showing the xy plane together with the corresponding xz and yz cross-sectional views. Samples contained [K10] = 105 μM (0.5% FAM-K10) and [dPu22] = 50 μM (0.5% dPu22-TMR) in a TE buffer (pH 7.4).

An analysis of z-stacked confocal imaging revealed that the anisotropic subcompartment spreads uniformly along the interior of a spherical cap, giving it a convex-lens-type shape (Figure C). This shape was also consistently observed when the subcompartment was located at the bottom of the droplet (Figure S9).

Fluorescence recovery after photobleaching (FRAP), which is used to assess molecular fluidity in biomolecular condensates and supramolecular soft materials, , was performed to evaluate the dynamics in both the isotropic mother compartments and anisotropic subcompartments (Figure A as well as Movies S3 and S4). Simultaneous photobleaching of the TMR and FAM photolabels was applied to a specific area (Figure B), and their normalized fluorescence intensities were quantified. As shown in the recovery curves in Figure C and the values listed in Figure D, TMR-labeled dPu22 exhibited a recovery time of 103 ± 20 s in the mother compartment, indicating limited but measurable fluidity. In the subcompartment, the recovery was markedly slower and could not be quantified, suggesting severely restricted dynamics. In contrast, the recovery of the fluorescence of FAM-labeled K10 was rapid, with no significant difference being observed between that in the mother compartment (6.6 ± 3.8 s) and the subcompartment (6.6 ± 3.1 s), indicating high fluidity throughout the droplet. The FRAP results indicate that the states of the mother compartment and the subcompartment of K10–dPu22 are nearly liquid and crystalline, respectively, which is highly consistent with the results of the time-course birefringence-retardation imaging experiments (Figure ). Notably, in the subcompartments, the two components exhibit different supramolecular behavior, i.e., K10 exhibits rapid, second-scale fluidity, whereas dPu22 shows ordering and remains immobile.

4.

Mobility of K10 and dPu22 was determined by FRAP experiments. (A) Confocal fluorescence images of the K10–dPu22 droplet before and after photobleaching and during fluorescence recovery. White arrows indicate the bleached area. For the corresponding videos, see Movies S3–S4. (B) Merged fluorescence image of the K10–dPu22 droplet immediately after photobleaching. (C) Fluorescence recovery curve. Fluorescence intensities in the photobleached area were normalized to those in the unbleached area and the background. (D) Fluorescence recovery times. Values were obtained by single-exponential fitting. Data are presented as the mean value ± the standard deviation (n = 4). Samples contained [K10] = 210 μM (0.5% FAM-K10) and [dPu22] = 100 μM (0.5% dPu22-TMR) in TE buffer (pH 7.4).

Structural Analyses of the Anisotropic K10–dPu22 Subcompartment

To investigate the molecular orientation in the subcompartments, the K10–dPu22 condensate was isolated by centrifugation and subjected to wide- and small-angle X-ray structural analyses (Figure A–B). Additionally, a lyophilized powder of the K10–dPu22 condensate was analyzed to provide additional lattice information, based on which the condensate was assumed to adopt a hexagonal lattice (for details, see Section 2 in the Supporting Information; Figure S10), and the peaks were indexed accordingly. In accordance with previous reports, the peak observed at 0.342 nm was assigned to the π–π-stacking distance between adjacent G-quartets. , The pattern of the condensate exhibited two peaks with d-spacings of 3.142 and 0.863 nm, corresponding to the (100) and (001) planes of a hexagonal lattice with lattice parameters a = 3.64 and c = 0.86 nm. The feature at 0.44 nm was attributed to a halo peak, suggesting the presence of an amorphous structure. The observation of a peak at 7.81 nm, which is approximately nine times the c-axis length (7.81 nm/0.86 nm ≈ 9), suggests the formation of a hexagonal columnar (Col_ho) phase with a helical pitch of nine.

5.

Structural analyses of the anisotropic subcompartments. (A) Wide-angle and (B) small-angle X-ray diffraction and scattering patterns of the K10–dPu22 condensate. The patterns were background-corrected by subtracting the signal of the buffer solution. (C–F) Schematic illustration of the proposed structures: (C) a single structural unit; (D) top and (E) side views of (F) the hexagonal columnar assemblies. (G) Bright-field and birefringence-axis images of the K10–dPu22 droplets. The color represents the slow axis of birefringence, which corresponds to the direction of the higher refractive index. Samples contained [K10] = 210 μM and [dPu22] = 100 μM in TE buffer (pH 7.4); images were acquired 9 h after sample preparation. (H) Plausible schematic representation of a K10–dPu22 droplet composed of an isotropic mother compartment and an anisotropic subcompartment.

According to a previously reported solution structure of dPu22, the stacking height of three G-quartet planes is approximately 0.65 nm, and the molecular width, including the phosphodiester linkage, is about 3.58 nm (Figure S11). Additionally, depending on its surrounding environment, the reported G4 sequence can exhibit intermolecular stacking. Drawing upon these previous findings, we assumed that the G4s within the subcompartment might also have a propensity to stack up. Combining these structural insights and the FRAP results, we inferred that the minimal structural unit was primarily organized around a well-defined Pu22 molecule with a K10 molecule contributing to the structure via flexible interactions (Figure C). These units are arranged hexagonally along the a-axis (a = 3.64 nm; Figure D) and stacked along the c-axis (c = 0.86 nm; Figure E) in a helical manner with nine units per turn (Figure F). The lattice parameters a and c are expanded in the condensate compared to the lyophilized condensate powder (Figure S10 and Section S2 in the Supporting Information), likely due to the existence of water and the fluidic K10 molecules.

Then, birefringence-axis imaging was performed to determine the relationship between the molecular orientation determined in the aforementioned experiments and the orientation of the anisotropic subcompartments observed by optical microscopy. − As shown in Figure G, the broad surface of the convex lens-shaped subcompartment corresponds to the slow axis, with a higher refractive index in that direction. According to previous reports, the nucleobase-pair planes in dsDNA show a higher refractive index. , By analogy, it can be inferred that the G-quartet planes coincide with the high-refractive-index direction; i.e., the planes of the G-quartets are mostly oriented parallel to the broad convex surface of the anisotropic subcompartment, as depicted in Figure H.

Notable Structural Characteristics

Owing to their high self-associative capability, there is a long history of studies reporting the formation of supramolecular assemblies based on G-quartets. Recent examples include guanosine–phenylboronic acid, guanosine–phenylboronic acid–surfactant, and guanosine-5′-monophosphate–guanosine assemblies in the presence of metal ions. However, to the best of our knowledge, this study is the first to report the organization of G4s within associative LLPS droplets, where the droplet itself serves as the organizing environment to encourage the formation of anisotropic subcompartments.

The formation of the subcompartments observed in this study is likely related to the liquid–liquid crystal phase separation previously reported for dsDNA. , The anisotropy of the subcompartments mainly arises from the G4-forming DNAs, which exhibit no detectable fluidity in FRAP assays (Figure ) and no peaks corresponding to 3D long-range order in the X-ray analyses (Figure A and B). Together, these results indicate that the hierarchically organized G4s can be regarded as LC molecular assemblies that form a static configuration. Notably, the subcompartments also contain polylysine peptides as a highly diffusive component distributed throughout the droplet. The subcompartments are thus binary supramolecular assemblies that integrate both static and dynamic components. In this sense, the anisotropic subcompartments can be regarded as hierarchically organized structures stabilized by the local environment in the mother compartments. The fact that the anisotropic subcompartments remain intact despite the rapid diffusion of the dynamic component suggests that this diffusive component contributes to the structural organization of the system, rather than merely occupying space within it.

Chemical Structural Constraints and Permissiveness in Anisotropic Subcompartment Formation

Associative LLPS and subsequent anisotropic subcompartment formation by dPu22 were observed not only with K10 but also with K50 and K100. The K50/K100–dPu22 systems exhibited clearly detectable birefringence retardation, albeit to a lower extent than K10–dPu22 (for details, see Section S3 in the Supporting Information; Figure S12). These results indicated that the formation of subcompartments is not specific to K10 but instead is a general phenomenon among polylysine peptides of varying lengths. However, 10-mer peptides containing both lysine and another amino acid residue (i.e., glycine, serine, leucine, or phenylalanine; Table S1) failed to form anisotropic subcompartments with dPu22 using the same charge-ratio conditions as for K10 (for details, see Section S3 in the Supporting Information; Figure S13A). Supporting the lysine-selective nature of this phenomenon, histone H1 (Table S1), a positively charged protein whose droplet formation is known to be promoted by G4s, and R50 also failed to induce anisotropy (Figure S13B and C). Considering the molecular orientation in the anisotropic subcompartments (Figure C–F), the formation of the anisotropic subcompartments likely requires that the peptides or proteins prevent interference with intermolecular G4 stacking and support the lateral assembly of the G4 stacks; nonlysine residues and longer polypeptide chains are presumed to disrupt one or both of these factors, at least under the conditions applied in this study.

Finally, to assess whether anisotropic subcompartments could also form in droplets containing K10 and other G4s, we examined two additional oligonucleotides, i.e., d22AG, a 22-nt antiparallel G4-forming DNA derived from the human telomeric sequence, and rPu22, the RNA analog of parallel G4-forming dPu22 (Figure A and Table S1). The conditions of [K10] = 210 μM and [G4] = 100 μM with a charge ratio of 1:1 were reproduced, and the systems were analyzed after 24 h of incubation. As shown in Figure B, the birefringence-retardation images of rPu22 and d22AG show anisotropic compartments within the droplets. Although a direct comparison of the magnitude of birefringence retardation across different oligonucleotides is difficult due to variations in droplet size, we noted that rPu22 exhibits higher birefringence retardation than dPu22 despite producing smaller droplets (Figure C). This enhancement may reflect the inherently stronger stacking propensity of RNA, which could more effectively promote the vertical interactions essential for the formation of the subcompartment. It is striking that the emergence of anisotropic subcompartments is not limited to parallel G4 DNA but also occurs in systems containing G4 RNA and antiparallel G4 DNA. The present results suggest that the ability to form a rigid G-quadruplex scaffold is important for anisotropic ordering. However, it remains unclear how tolerant this behavior is to base substitutions or sequence modifications and whether the observed hexagonal columnar organization reflects a general feature of G4 rigidity or depends on specific stacking geometries. Systematic sequence variation will be required to define these structural limits.

6.

Formation of anisotropic subcompartments in K10–G4 droplets with different G4 sequences. (A) Schematic illustration of rPu22 and d22AG. The drawn G4 topologies were determined based on circular dichroism spectra (Figure S14). (B) Bright-field and birefringence-retardation images of K10–dPu22, K10–rPu22, and K10–d22AG droplets. (C) Quantified birefringence-retardation values of K10–dPu22, K10–rPu22, and K10–d22AG. Samples contained [K10] = 210 μM and [G4] = 100 μM in TE buffer (pH 7.4), with a charge ratio of 1:1. Images were acquired 24 h after sample preparation.

Conclusions

In conclusion, this study demonstrates that droplet-based compartments formed by G4-forming DNAs and polylysine peptides gradually develop anisotropic subcompartments at specific points on their droplet–solution interfaces. Birefringence-retardation imaging, confocal fluorescence microscopy, and X-ray analyses revealed that these convex-lens-shaped subcompartments originate from hierarchically aligned G4s assembled into hexagonal columnar arrays via intermolecular stacking and lateral interactions, and that the dynamic fluidity of the polylysine is maintained. Moreover, although there are sequence constraints on the positively charged peptides that can induce subcompartment formation, diverse G4 structures are applicable, indicating greater structural tolerance in terms of the oligonucleotide moiety.

Beyond these structural insights, the anisotropic subcompartments observed in this study are of interest not only as a fundamental experimental observation but also as a simplified model to explore how ordered and dynamic components can coexist within a single phase-separated compartment. In this context, our observations suggest that molecular order and fluidity are not necessarily mutually exclusive but may instead cooperate, allowing a dynamic component to coexist with and support ordered structures within the same compartment. Furthermore, the localized emergence of anisotropic substructures at specific regions of the droplet–solution interface suggests that interfacial heterogeneity relative to the droplet interior may play an active role in initiating internal ordering and mesoscale domain formation. − The multiphase organization here shares conceptual similarities with internal structuring discussed for membraneless compartments such as nucleoli and paraspeckles, , while remaining a minimal physicochemical system composed of only oligonucleotides and peptides. This simplicity highlights the utility of such minimal coassemblies as a platform for the investigation of general physicochemical principles underlying hierarchical organization, rather than directly replicating complex biological systems. Building on these findings, the system reported in this study may facilitate bottom-up approaches toward understanding the cooperation of ordered and dynamic components in molecular assemblies and toward the rational design of artificial assemblies with integrated structural and dynamic features.

Methods

Materials

Bovine serum albumin (BSA) was procured from Sigma-Aldrich Co. LLC (St. Louis, MO, USA). Tris­(hydroxymethyl)­aminomethane–ethylenediaminetetraacetic acid (Tris–EDTA) powder was purchased from Takara Bio Inc. (Shiga, Japan) and used to prepare TE buffer at concentrations of 100 mM Tris-HCl (pH 7.4) and 10 mM EDTA. K5, K10, K7G3, K7S3, K7F3, and K7L3 peptides were synthesized and purified (HPLC purity >95%) by GenScript Biotech Corp. (Piscataway, NJ, USA). Poly-l-lysine hydrochloride peptides (K50 and K100) and poly-l-arginine hydrochloride peptide (R50) were purchased from Alamanda Polymers Inc. (Huntsville, AL, USA). All peptide stock solutions were prepared based on the net weight determined through elemental nitrogen analysis. K10 labeled with fluorescein (FAM) at the ε-amino group of the N-terminal lysine was synthesized and purified (HPLC purity >95%) by Biologica Co. (Nagoya, Japan). Histone H1 from bovine thymus was procured from SignalChem Biotech Inc. (Richmond, Canada). DNAs and RNAs were synthesized and purified (salt-free grade) by Eurofins Genomics K.K. (Tokyo, Japan). DNAs and RNAs labeled with tetramethylrhodamine (TMR) at the 3′-terminus were synthesized and purified (HPLC grade) by Eurofins Genomics K.K. (Tokyo, Japan). The concentrations of DNA, RNA, and protein were determined via UV absorbance using a NanoDrop One spectrophotometer (Thermo Fisher Scientific Inc., Waltham, MA, USA). The extinction coefficients for unlabeled DNA and RNA at 260 nm and proteins at 280 nm were calculated based on previously published literature. − For the fluorophore-labeled compounds, extinction coefficients of 80,000 M^–1cm^–1 for FAM and 65,000 M^–1cm^–1 for TMR were used.

General Procedures

G4 structures were prepared immediately before experiments using a thermocycler (TP-600; Takara Bio Inc., Shiga, Japan) and the following annealing program: initial heating at 95 °C for 5 min followed by gradual cooling to 25 °C at a rate of 1 °C/min. Unless otherwise noted, all experiments were conducted at room temperature (25 °C). The surfaces of glass coverslips (30 × 40 mm; 0.13–0.17 mm thick; Matsunami Glass Ind. Ltd., Osaka, Japan) were coated with BSA according to previously established procedures. The coverslips were immersed in a 3.5 w/v% aqueous solution of BSA for 15 min, subsequently rinsed with Milli-Q water, and then air-dried.

Bright-Field Imaging for the Preparation of Phase Diagrams

A solution or suspension (25 μL/well) with a final concentration of [K10] = 10.5–420 μM and [dPu22] = 0.5–1000 μM in TE buffer (pH 7.4) was prepared in each well of the 384-well microplate (CELLSTAR cell-repellent μClear; Greiner Bio-One GmbH, Frickenhausen, Germany). All dispensing and mixing procedures were automated using an epMotion 5073 liquid handling system (Eppendorf, Hamburg, Germany). The microplates were sealed with microplate tape to prevent evaporation and then incubated for 0–24 h. Bright-field images were obtained using a Cytation 5 imaging multimode reader (Agilent Technologies Inc., Santa Clara, CA, USA) with a Plan Fluorite 40× objective lens (NA 0.75).

Quantification of Droplet Roundness

The roundness of each droplet was quantified from bright-field images using the Shape Descriptors function in the Fiji software (ImageJ v1.54p; NIH, Bethesda, MD, USA). Dot plots were visualized in the R software (ver. 4.4.2; RStudio Inc., Boston, MA, USA) using the geom_jitter function, overlaid with box plots and the mean value.

Birefringence-Retardation and Axis-Orientation Imaging

A portion of the G4–peptide suspension was placed on the BSA-coated glass with a double-faced adhesive spacer (1.2 mm thickness). Birefringence-retardation and axis-orientation images were captured using a BX50 microscope (Evident Corp., Tokyo, Japan) equipped with a PA-micro-S birefringence measurement system (Photonic Lattice Inc., Miyagi, Japan) and a LMPLFLN 50× or 100× objective lens (NA 0.50 and 0.80). The birefringence-retardation and axis-orientation images were visualized using the PA-View software (ver. 2.4.10.2; Photonic Lattice Inc., Miyagi, Japan), and the retardation images were subsequently colorized using Fiji software.

Quantification of Birefringence-Retardation Values

Each retardation image (5,065,984 pixels per image) was quantified per pixel using the PA-View software. Retardation values below the background signal level (7.0 nm) were excluded. Dot plots were generated by randomly selecting up to 1,000 data points from each sample and visualizing them in the R software package using the geom_jitter function, overlaid with box plots and the mean value.

Confocal Fluorescence Imaging

A portion of the G4–peptide suspension with fluorescence labeling was placed onto BSA-coated glass with a double-faced adhesive spacer (thickness: 1.2 mm) and incubated overnight in the dark. Confocal fluorescence images were acquired with an AX-R on an NSPARC confocal microscope (Nikon Solutions Co., Ltd., Tokyo, Japan) using a PLAN APO λD 100× oil immersion objective lens (NA 1.45). For fluorescence recovery after photobleaching (FRAP) measurements, 488 nm (FAM) and 561 nm (TMR) lasers were applied to the region of interest. Fluorescence recovery was monitored for 45 (FAM) and 180 s (TMR) at intervals of 1.56 s. Bleaching was performed by using identical ROIs for both channels. The fluorescence intensity of the photobleached areas was normalized relative to that of the unbleached area and the background. The time-course of the normalized fluorescence intensity was fitted to a single exponential function (eq ) to determine the recovery time τ, according to

$$f(t)=f_0+C[1-\exp (-t/\tau )]$$ 1

where f ₀ is the fluorescence intensity at the time of photobleaching, C is an arbitrary constant, and t is the time after photobleaching. The analyses were conducted using Fiji software. The Bleach Correction plugin was applied to correct for time-dependent intensity variation.

X-ray Diffraction and Scattering Measurements

A suspension (40 mL) was prepared to final concentrations of [K10] = 210 μM and [dPu22] = 100 μM in TE buffer (pH 7.4), with a 1:1 charge ratio. After centrifugation (2,330g, 10 min), the resulting condensate was collected for further analysis. The viscous condensate was placed in a sample holder. The powder sample obtained from the lyophilization of the condensate was sealed in a glass capillary (φ = 1.5 mm). The X-ray diffraction and scattering patterns were collected using a NANO-Viewer (Rigaku Corp., Tokyo, Japan) equipped with a PILATUS 100k detector (Dectris AG, Baden, Switzerland), employing CuKα radiation (λ = 1.541 Å). Ag Behenate was used as the standard sample. The X-ray diffraction and scattering patterns obtained using the two-dimensional detector were processed to convert them into one-dimensional profiles using Rigaku 2DP software.

Circular Dichroism Measurement

A solution (400 μL) was prepared to final concentrations of [G4] = 60 μM or [K10] = 126 μM in TE buffer (pH 7.4) and transferred to a quartz cuvette with an optical path length of 0.1 cm. Spectra were acquired using a J-1100 spectropolarimeter (JASCO Corp., Tokyo, Japan).

Ethical Statement

This study did not involve human participants or animal experiments.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsnano.5c21437.

• Sequences of oligonucleotides and peptides (Table S1); bright-field images of mixtures of dPu22 with polylysine (Figure S1); experimental phase diagrams as a function of K10 and dPu22 concentrations (Figure S2); time-course of the changes in roundness of K10–dPu22 droplets (Figure S3); time-course of the changes in the birefringence-retardation images of K10–dPu22 (Figure S4); time-course of the changes in the birefringence-retardation values of K10–dPu22 (Figure S5); size distribution of K10–dPu22 droplets with and without anisotropic puncta (Figure S6); birefringence-retardation images of K10–dPu22 with different charge ratios (Figure S7); time-lapse confocal fluorescence images of K10–dPu22 droplets (Figure S8); confocal fluorescence image of a K10–dPu22 droplet with a subcompartment at its bottom (Figure S9); X-ray diffraction and scattering patterns of lyophilized K10–dPu22 condensate powder (Figure S10); reported solution structure of dPu22 (Figure S11); birefringence-retardation images of K50–dPu22 and K100–dPu22 (Figure S12); birefringence-retardation images of dPu22–K7 × 3, dPu22–histone H1, and dPu22–R50 (Figure S13); circular dichroism spectra of dPu22, d22AG, and rPu22 in dilute solution (Figure S14); discussion of the X-ray diffraction and scattering analysis of K10–dPu22 condensate powder; discussion of the effects of the peptide sequence on anisotropic subcompartment formation (PDF)

• Time-course of the changes in the birefringence-retardation images (Movie S1) (MP4)

• Confocal fluorescence images showing the emergence of anisotropic subcompartments (Movie S2) (MP4)

• FRAP experiments in FAM-K10 (Movie S3) (MP4)

• FRAP experiments in dPu22-TMR (Movie S4) (MP4)

Conceptualization: H.S. and S.T. Investigation: H.S., M.T., K.W., and T.K. Data curation and formal analysis: H.S., M.T., K.W., and T.K. Resources: H.S., S.T., and K.K. Funding acquisition: H.S. and S.T. Project administration: H.S. and K.K. Supervision, visualization, and writingoriginal draft: H.S. Writingreview and editing: S.T., M.F., T.K., and K.K. All authors have approved the final version of the manuscript.

This research was partially supported by grants from JSPS KAKENHI [JP22K14777 (to H.S.), JP23H01995 (to S.T.), and JP25K18125 (to H.S.)]; JST PRESTO JPMJPR22E5 (to H.S.); the Kao Foundation for Arts and Sciences (Kao Crescent Award to H.S.); the Toyota Physical and Chemical Research Institute (Toyota Riken Scholar to H.S.); the Grant-in-Aid for Challenging Research (Center for Fundamental Research to H.S.); and the Yoshinori Ohsumi Fund for Fundamental Research (Yoshinori Ohsumi Award to H.S.).

A preprint version of this work was previously posted: Sugai, H.; Tomita, S.; Toyoda, M.; Watanuki, K.; Fukuyama, M.; Kajitani, T.; Kinbara, K. Emergence of Anisotropic Subcompartments via Co-Assembly of Hierarchically Ordered G-Quadruplexes and Dynamic Polylysine in Droplet-Based Compartments. 2025, chemrxiv-2025-xh4zs. ChemRxiv. 10.26434/chemrxiv-2025-xh4zs (accessed March 14, 2026).

The authors declare no competing financial interest.