Capsule-like DNA Hydrogels with Patterns Formed by Lateral Phase Separation of DNA Nanostructures.

Phase separation is a key phenomenon in artificial cell construction. Recent studies have shown that the liquid-liquid phase separation of designed-DNA nanostructures induces the formation of liquid-like condensates that eventually become hydrogels by lowering the solution temperature. As a compartmental capsule is an essential artificial cell structure, many studies have focused on the lateral phase separation of artificial lipid vesicles. However, controlling phase separation using a molecular design approach remains challenging. Here, we present the lateral liquid-liquid phase separation of DNA nanostructures that leads to the formation of phase-separated capsule-like hydrogels. We designed three types of DNA nanostructures (two orthogonal and a linker nanostructure) that were adsorbed onto an interface of water-in-oil (W/O) droplets via electrostatic interactions. The phase separation of DNA nanostructures led to the formation of hydrogels with bicontinuous, patch, and mix patterns, due to the immiscibility of liquid-like DNA during the self-assembly process. The frequency of appearance of these patterns was altered by designing DNA sequences and altering the mixing ratio of the nanostructures. We constructed a phase diagram for the capsule-like DNA hydrogels by investigating pattern formation under various conditions. The phase-separated DNA hydrogels did not only form on the W/O droplet interface but also on the inner leaflet of lipid vesicles. Notably, the capsule-like hydrogels were extracted into an aqueous solution, maintaining the patterns formed by the lateral phase separation. In addition, the extracted hydrogels were successfully combined with enzymatic reactions, which induced their degradation. Our results provide a method for the design and control of phase-separated hydrogel capsules using sequence-designed DNAs. We envision that by incorporating various DNA nanodevices into DNA hydrogel capsules, the capsules will gain molecular sensing, chemical-information processing, and mechanochemical actuating functions, allowing the construction of functional molecular systems.

Introduction

Phase separation is a physical phenomenon by which a homogeneous phase separates into two (or more) distinct phases, resulting in nonuniform matter distribution. The phase separation of water-soluble molecules in an aqueous solvent is called liquid–liquid phase separation. A typical example is the phase separation of a polyethylene glycol (PEG)/dextran (DEX) mixture, which forms two liquid phases via segregative phase separation.[1] The phase separation of a polymer mixture generally occurs under specific polymer concentrations, temperature, salt concentrations, and polymer molecular weight.[2] This phenomenon generates an aqueous two-phase system, adopted in various applications, such as biomolecule extraction[3] and patterning.[4]

Phase separation of water-soluble molecules, especially biomolecules, has also been adopted for the construction of artificial cells.[5] For example, a cell-free transcription/translation machinery was mixed with a PEG/DEX solution and encapsulated into water-in-oil (W/O) droplets.[6] In another example, artificial nucleoid-like structures containing transcriptional reaction sets were formed in lipid vesicles.[7] Importantly, phase separation plays crucial roles in living cells, such as heterochromatin formation,[8] gene expression regulation,[9] and membrane-less organelle construction.[10] In cells, proteins and/or nucleic acids exhibit phase separation, leading to the formation of liquid or hydrogel-like molecular condensates.[11,12] Given the utility and potential functions of phase separation, controlling it by designing biopolymer molecules is an issue to be addressed in artificial cell studies.

The compartmental capsule is one of the most important structures in artificial cells because it is required for the integration of multiple chemical reactions into a molecular system by separating them from the environment.[13,14] W/O droplets[15−17] or lipid vesicles[18−21] have been typically adopted for compartmentalization. The compartmental capsules of living cells, i.e., cellular membrane vesicles, have two-dimensionally segregated domain structures[22] related to biological events, such as signal transduction[23] or molecular uptake.[24] Lipid vesicles prepared using two (or more) lipid species with different phase transition temperatures or headgroup charges exhibit phase separation on lipid bilayer membranes.[25−27] Studies have reported methods to control the lateral phase separation of the lipid vesicle capsule by changing the temperature,[25−27] membrane-tension,[26] or lipid composition.[27] However, designing and regulating the phase separation of the compartmental capsule using a molecular design approach remains challenging.

Sequence-specific DNA interactions can be used to control the phase separation of DNA nanostructures at the molecular level.[28−33] Recent studies have demonstrated that sequence-designed DNA nanostructures exhibit temperature-dependent phase separation and self-assembly into liquid-like droplets.[28−32] The DNA nanostructures formed droplets under a specific temperature, and the formed droplets became hydrogels when the temperature was lowered.[30] Furthermore, two distinct liquid DNA phases were formed on the two types of DNA nanostructures with orthogonal sequence pairs,[29,30] allowing the formation of two immiscible DNA liquid droplets and hydrogels. Designing a DNA-based phase separation system is a feasible approach to the design and control of lateral phase separation of compartmental capsules.

We herein report the formation of capsule-like DNA hydrogels with several types of patterns formed by lateral phase separation of DNA nanostructures. We used W/O droplets as the substrate to form the hydrogels. A mixture of cationic and nonionic surfactants were used for droplet preparation to enable DNA nanostructures to adsorb onto the W/O droplet interface. We designed two types of DNA nanostructures (Y-motif and orthogonal Y-motif, hereafter referred to as orthY-motif), whose sequences are orthogonal to each other, with the ability to form different phases in selective and exclusive self-assembly manners. We investigated the pattern formed on the interface by the self-assembly and phase separation of the motifs using a variety of Y- and orthY-motif mixture ratios. Then, to address the sequence design-based control of phase separation, we designed additional DNA nanostructures that acted as a linker for the Y- and orthY-motifs. We constructed a phase diagram by investigating how the addition of the linker changed the phase-separation patterns. In addition to the investigation of such patterns, we showed the formation of the phase-separated hydrogels on inner leaflets of lipid vesicles. The capsule-like hydrogels were further successfully extracted into an aqueous solution and combined with an enzymatic reaction. DNA nanotechnology provides various functional DNA devices, capable of computing,[34] sensing,[35] and actuating,[36] that can be incorporated in the hydrogel capsules. Therefore, our results will serve as the means to design and construct functional microcapsules for artificial cell studies.

Experimental Section

Materials

DNAs were purchased from Eurofins Genomics (Tokyo, Japan). Nonmodified and FAM-modified DNAs were of oligonucleotide purification cartridge grade, and Alexa 405- and Cy3-modified DNAs were of high-performance liquid chromatography purification grade. DNA was stored in ultrapure water (18 MΩ·cm in resistance) at a concentration of 100 μM at −20 °C until use. DNA concentration was measured using a microvolume spectrometer (DS-11FX, DeNovix, Wilmington, DE, USA). DNA sequences are provided in Table S1. Sorbitan monooleate (Span 80) and oleylamine were purchased from Tokyo Chemical Industry (Tokyo, Japan) and Kanto Chemical (Tokyo, Japan), respectively. Ultrapure water and Tris-HCl (pH 8.0) were purchased from Thermo Fisher Scientific (Waltham, MA, USA). NaCl powder, cholesterol, polyoxyethylene (10) octylphenyl ether (Triton-X 100), sucrose, and liquid paraffin were purchased from FUJIFILM Wako Pure Chemical (Osaka, Japan). Iodixanol (OptiPrep) was purchased from Cosmo Bio (Tokyo, Japan). Silicone-coated glasses (30 mm by 40 mm with a thickness of 0.17 mm) were purchased from Matsunami Glass (Osaka, Japan). 1,2-Dioleoyl-sn-glycero-3-phosphocholine (DOPC) was purchased from NOF Corporation (Tokyo, Japan). 1,2-Dioleoyl-3-trimethylammonium-propane (DOTAP) was purchased from Avanti Polar Lipids (Alabaster, AL, USA). Exonuclease I and III were purchased from New England Biolabs (Ipswich, MS, USA).

Preparation of Surfactants in Oil Solution

A liquid paraffin oil solution containing 20 mM Span 80 and the oleylamine oil solution were, respectively, prepared as follows. First, Span 80 and liquid paraffin were mixed in a test tube. Similarly, oleylamine was mixed with liquid paraffin. These solutions were well vortexed and then sonicated for 1 h at 50 °C using a sonicator bath (CPX1800h-J, Branson, Danbury, CT, USA). After the sonication treatment, the 20 mM Span 80 and oleylamine solutions were mixed at a molar ratio of 1:3 to prepare a 10 mM surfactant solution. The mixture was further sonicated using the sonicator bath for 1 h at 50 °C. The mixed solution was stored under N2 gas-filling and light-shielding conditions until use at 23 °C.

Generation of Water-in-Oil Droplets Containing DNA Nanostructures and Their Annealing

DNA strands were mixed in a PCR tube with a buffer containing 20 mM Tris-HCl (pH 8.0) and 350 mM NaCl. The DNA concentrations were altered depending on each experiment. Note that the dye-modified DNA strands without the sticky-end were added at a 10% molar ratio (e.g., 2.5 μM for Y-1 and Y-3, 2.25 μM for Y-2, and 0.25 μM for the fluorescence-modified strands; Table S1). The surfactant mixtures in oil and DNA solution were incubated for 2 min at 95 °C using a thermal cycler (MiniAmp Thermal Cycler, Thermo Fisher Scientific, Waltham, MA, USA). Two microliters of DNA solution was added to 50 μL of the surfactant mixture in a PCR tube and incubated for 1 min at 95 °C. Then, the tube was well mixed by tapping it to generate micrometer-sized water-in-oil droplets. The tubes containing the droplets were again placed on the thermal cycler, and the temperature was lowered from 95 to 25 °C at a rate of −1 °C/10 s (annealing process). The generated droplet radius was 4.7 ± 1.3 μm (mean ± standard deviation) (Figure S1).

Observation

After annealing, the prepared droplets were placed onto the silicone-coated glass using a micropipet with a tip that was cut off to make the diameter larger than the droplets. The bottom surfaces or cross sections of the droplets were visualized using a confocal laser scanning microscope (FV1000, Olympus, Tokyo, Japan) with a 40× objective lens (UPLAPO, Olympus, Tokyo, Japan). Alexa 405, FAM, and Cy3 were imaged using lasers at 405, 473, and 543 nm, respectively.

Generation of Lipid Vesicles with Phase-Separated Hydrogels

DOPC, a zwitterionic lipid, and DOTAP, a cationic lipid, were used as an alternative to Span 80 and oleylamine. Using this lipid mixture (1 mM each), W/O droplets containing Y- and orthY-motifs (5 μM each) were generated. Note that 142 mM of iodixanol was additionally mixed in aqueous solution for droplet generation. A DOPC (90%) and cholesterol (10%) mixture in liquid paraffin (5 mM in total) was used to create an outer leaflet of lipid vesicles (outer lipid mixture). An outer aqueous solution containing 20 mM Tris-HCl (pH 8.0), 350 mM NaCl, and 142 mM sucrose was prepared. The outer lipid mixture (150 μL) was poured onto the outer solution (300 μL) in a 1.5 mL tube and incubated for over 15 min at 23 °C. The W/O droplet after annealing was gently poured onto the outer lipid mixture and subsequently centrifuged at 8,000 g at 4 °C for 15 min. After centrifugation, the remaining liquid paraffin was removed using a micropipette. The precipitated lipid vesicles were dispersed by pipetting and observed on BSA-coated glass. A detailed protocol is provided in Supplementary Note 1.

Extraction of Capsule-like Hydrogels and Combination with Enzymatic Reaction

In the extraction, a 1.5 mm thick silicone rubber sheet with a punch hole (5 mm in diameter) was placed on BSA-coated glass (24 mm by 36 mm, with a thickness of 0.17 mm). To extract the hydrogels, 27 μL of the lipid vesicle solution was poured in the hole. After 10 min, 2 μL of 10% (v/v) Triton-X 100 in buffer (20 mM Tris-HCl and 350 mM NaCl) was dropped into the vesicle solution in the hole to remove lipid bilayers surrounding the hydrogels. Then, extracted hydrogels and exonuclease I/III were mixed on the glass. A detailed protocol of the enzymatic reaction is provided in Supplementary Note 2.

Results and Discussion

Phase-Separated Hydrogel Formation

Span 80 and oleylamine were used to prepare W/O droplets (Figure a). Span 80 and oleylamine have the same alkyl chain (C18:1, 9-cis) but different head groups. Span 80 is a nonionic surfactant that was used for stabilizing the generated droplet. Oleylamine is a cationic surfactant that was used to accumulate the DNA nanostructures onto the droplet interface via electrostatic interactions. As the pKa value of oleylamine was around 10.0–10.7,[37,38] the amine part in oleylamine was protonated (positively charged) in our buffer condition (pH 8.0). Therefore, the DNA nanostructures with negatively charged-phosphate groups, were adsorbed to the droplet interface. By optimizing the mixing ratio of Span 80/oleylamine for DNA adsorption, we determined that DNA nanostructures were well adsorbed at the Span 80/oleylamine ratio of 1/3 (Figure S2).

Figure 1

Schematic illustrations of the experimental system. (a) Structural formulas of Span 80 and oleylamine. Oleylamine acts as a cationic surfactant due to the protonation of the amine group. (b) DNA nanostructures. The Y-motif and orthogonal Y-motif (orthY-motif) self-assemble into two types of network structures, respectively, as the sticky-ends in the two motifs are orthogonal sequences. (c) Formation of phase-separated DNA hydrogels on a water-in-oil (W/O) droplet interface. The motifs were adsorbed on the interface via electrostatic interaction and self-assembled on the interface.

Two types of DNA nanostructures, Y- and orthY-motif, were designed. The motifs were respectively composed of three different single-stranded DNAs (ssDNAs) (Figure b) of equal length (nucleotide (nt) number). Both motifs have three sticky-ends with 8 nt palindrome sequences, however, the sticky-end sequences between the two motifs were not complementary (orthogonal sequences). Thus, each motif could interact with other motifs of the same type. Owing to this design, each motif could selectively and exclusively self-assemble into different hydrogels by forming the network structures of the motifs (Figure b). The sequences of both motifs were designed to have approximately the same thermodynamic parameters (melting temperature: Tm) of hybridization in the formation of the motifs (Tm: ∼64.3 °C) and in the sticky-ends (Tm: ∼43.5 °C) (Table S2).[39] Phase-separated gel networks were formed on the interface by adding the two motifs, the buffer, and salts in an aqueous phase for droplet formation, after the annealing process (Figure c). These patterns were stable for at least a day.

We investigated hydrogel formation on the droplet interface using an equimolar motif concentration (2.5 μM each) (Figure a). For imaging, Y- and orthY-motifs were labeled with FAM and Alexa405, respectively (fluorescence-modified strands were added at a 10% molar ratio). Microscopic observation revealed that, after the annealing process, the DNA nanostructures exhibited lateral phase separation on the droplet interfaces and the droplets were covered with phase-separated DNA hydrogels that showed various patterns (Figure b). We classified the patterns into three types: Bicontinuous, Y-motif patch (Y-patch), and orthY-motif patch (orthY-patch) (Figure c). In the bicontinuous pattern, both the Y- and orthY-motifs formed continuous hydrogels. In the Y-patch pattern, Y-motifs formed smaller hydrogels in a continuous orthY-motif hydrogel, and vice versa in a orthY-patch pattern. Only W/O droplets without bicontinuous patterns on the surface were classified as patch patterns. The appearance frequency of the patterns was analyzed by observing the droplet surfaces. The results showed that bicontinuous pattern was over 70% under the equimolar condition.

Figure 2

DNA hydrogel patterns on the droplet interface formed by phase separation of two DNA motifs. (a) Schematic representation of the experimental condition. (b) Microscopy images of the droplet surfaces (left) and droplet cross section (right). The green and blue channels represent the FAM signal for the Y-motif and Alexa405 signal for the orthY-motif, respectively. Scale bar: 30 μm. (c) Classification of the observed patterns. Bicontinuous pattern, Y-motif patch pattern (Y-patch), and orthY-motif patch pattern (orthY-patch). Scale bar: 10 μm. (d) Pattern frequency. The number of analyzed droplets was 491 in the different experiments.

The formation of the phase-separated gel patterns was attributed to the immiscibility of the Y- and orthY-motifs, which was due to their orthogonal sequences. Our earlier study[30] revealed that, in a specific temperature range, the Y-motifs self-assembled into liquid-like microstructures (named DNA droplets) and two types of DNA droplets composed of Y- or orthY-motifs exhibited selective/exclusive fusion behavior. In this study, under a specific annealing temperature range between approximately 63 and 35 °C,[30] the motifs self-assembled into liquid-like structures on the W/O droplet interface. The two types of “DNA liquids” on the interface were immiscible, resulting in their phase separation. Because the DNA liquids become hydrogels following a temperature decrease,[30] phase-separated hydrogel patterns can be formed on the interface. Although DNA nanostructures without sticky-ends were adsorbed onto the interfaces, the patterns were not observed (Figure S3), clearly demonstrating that the formation of phase-separated hydrogel patterns depended on the orthogonality of the sticky-end sequences, not due to the fluorescence molecules.

Changes of the Formed Patterns by Adjusting the Nanostructure Mixing Ratio

To explore the possibility of controlling the patterns on the interface, we focused on the mixing ratio of the two motifs. We prepared W/O droplets containing the Y- and orthY-motif mixture at a concentration ratio of 1.5/3.5 or 3.5/1.5 (μM/μM). The results showed apparent differences in the trends of the pattern formed between the two conditions (Figure ). At the Y/orthY ratio of 1.5/3.5, approximately all droplet interfaces were covered with the Y-patch pattern (Figure a). In contrast, most of the droplets were covered with the orthY-patch pattern at the Y/orthY ratio of 3.5/1.5 (Figure b). The frequency analysis of the formed patterns showed that the majority were patch patterns; the Y-patch pattern or orthY-patch pattern frequency was over 90 or 75% at Y/orthY = 1.5/3.5 or 3.5/1.5, respectively (Figure c). These results showed that the type of patterns formed can be altered by adjusting the mixing ratio of the two immiscible motifs.

Figure 3

DNA hydrogel patterns on the droplet interface under unbalanced motif concentrations. (a and b) Representative microscopy images of the droplet surfaces containing 1.5 and 3.5 μM (a) or 3.5 and 1.5 μM (b) of the Y- and orthY-motifs. Scale bar: 20 μm. (c) Frequency of patterns using 1.5/3.5 or 3.5/1.5 (μM/μM) Y-/orthY-motifs. The number of analyzed droplets in each condition were 493 for 1.5/3.5 and 343 for 3.5/1.5, respectively.

The formation of patch patterns is expected to be determined during the self-assembly process of the motifs. Under an unbalanced motif concentration, a large number of motifs can easily self-assemble into large hydrogels that can almost entirely cover the droplet interface. In contrast, a smaller number of motifs is difficult to assemble or grow to large hydrogels. Notably, although the thermodynamic parameters of sticky-end hybridization were approximately the same in both motifs, the bicontinuous pattern frequency at Y/orthY = 3.5/1.5 was higher than that at Y/orthY = 1.5/3.5 (Figure c). In addition, the frequency of the orthY-patch pattern was higher than that of the Y-patch pattern (Figure d), which may imply that the sticky-end sequence in Y-motifs favors the formation of the continuous hydrogel pattern rather than the formation of patch patterns. Concentration imbalance is able to cause the pattern frequency difference. However, the DNA solution was prepared from stock solutions whose concentrations were measured using the spectrometer. Nonuniform partitioning of the DNA strands into the droplets would also change the concentration, but it will equally affect both motifs and will not explain this pattern frequency difference. Although it is challenging to reveal the detailed mechanisms causing this imbalance, which is out of the scope of the present study, clarifying it would deepen our understanding of the design of DNA nanostructures capable of phase separation.

Changes of the Formed Patterns by Adding Sequence-Designed Linker Structures

In addition to the mixing ratio of the motifs, the possibility to design a DNA sequence to control the formed patterns was examined. We designed an X-shaped linker (X-linker) with four sticky-ends, two for the Y-motif and two for the orthY-motif (Figure a, left). The X-linkers can form homogeneous hydrogels on the droplet interface (Figure S4), similar to those of either Y- or orthY-motifs (Figure S4). The X-linker is able to cross-link the Y- and orthY-motifs, resulting in the elimination of orthogonality (Figure a, right). At a higher X-linker concentration, the Y- and orthY-motifs should be homogeneously distributed on the interface due to cross-bridging. This pattern was named Mix pattern.

Figure 4

Changes of formed patterns by the addition of sequence-designed X-shaped linker (X-linker). (a) Schematic representation of the X-linker capable of cross-linking the Y- and orthY-motifs. (b) Representative microscopy images of the droplet surfaces containing 2.5 μM Y- and orthY-motif with the designated concentration of the X-linker. Scale bar: 20 μm. (c) Frequency of patterns under various X-linker amounts.

The X-linker concentration ranged from 0 to 1 μM using a concentration of 2.5 μM of Y- and orthY-motifs. The appearance frequency of the patterns was analyzed using imaging (Figure b and c). At 0 and 0.1 μM, most of the droplet interfaces were covered with phase-separated hydrogels. At 0.25 and 0.5 μM, although the Mix pattern formed on some droplets, phase-separated hydrogels were still observed, where the boundary between the hydrogels of each motif was vague and the size of patches seemed to become smaller (Figure S5, Supplementary Note 3). At 1 μM, all droplets were covered with the Mix pattern. To further investigate the effect of the X-linker concentration on Mix pattern appearance, sigmoidal curve fitting of the X-linker concentration-dependent frequency of Mix pattern formation was performed using a Hill-type sigmoidal curve[40] (Figure S6, Supplementary Note 4). The fitting result showed that 0.43 μM of the X-linker concentration results in a 50% Mix pattern appearance.

The size decrease of the patches (Figure S5) at the X-linker concentrations of 0.25 and 0.5 μM was likely attributed to the emulsification of DNA liquid during annealing. Since the X-linker linked the Y- and orthY-motifs, it can be regarded as a surfactant between the two motifs. This surfactant-like role has been confirmed in previous studies in bulk solution.[29,30] Jeon et al. indicated that DNA liquid composed of DNA nanostars can form microemulsions in another DNA liquid using cross-linker nanostars.[29] In addition, they showed that the cross-linkers formed micelles in DNA liquids where the cross-linker nanostars were distributed in the bulk DNA liquid. In the present study, the X-linkers were almost homogeneously distributed in both hydrogels formed by the Y- and orthY-motifs on the droplet interface (Figure S7), suggesting that the X-linkers also formed micelles in DNA liquid. Due to the surfactant-like behavior, the Y-motif (or orthY-motif) DNA liquid in the orthY-motif (or Y-motif) DNA liquid on the interface can be surrounded by the X-linkers during annealing, forming DNA liquid emulsions and contributing to forming smaller patches due to the decrease of interfacial tension between the two DNA liquids.[29] The emulsified DNA liquids eventually became hydrogels with temperature lowering.

To gain further insight into regulating the patterns of the phase-separated hydrogels, we investigated the effect of X-linker concentration on the appearance frequency of the patterns using the 1.5/3.5 or 3.5/1.5 of Y-/orthY-motif ratio. The obtained data were fitted using the sigmoidal curve (Figure S6). Based on the results, we constructed a phase diagram of the gel patterns on the interface (Figures and S8). The phase diagram showed that the appearance of the Mix pattern was determined by the amount of the X-linker rather than the Y-/orthY-motif ratio. Notably, using 1 μM X-linker resulted in the formation of the Mix pattern in all tested Y- and orthY-motif concentrations. The phase diagram also suggested that the Mix pattern could appear at lower X-linker concentration under an unbalanced Y-/orthY-motif ratio. The appearance of the Mix pattern is a result of DNA liquid emulsification, as discussed above. When the concentration of the one motif is lower than that of the other, X-linker-induced emulsification occurs more easily, leading to high frequency of Mix pattern appearance. Taken together, these findings show that sequence design of the motifs and adjustment of the amount of each motif can be used to regulate the patterns formed on the droplet interface.

Figure 5

Phase diagram of the hydrogel patterns on the W/O droplet interfaces, representing the effects of the Y- and orthY-motif mixing ratio and the added amount of X-linker on the patterns formed. Note that the dashed lines are eye-guide sketches representing the phase boundary. Pie charts in the phase diagram show the frequency of each pattern.

Generation of Lipid Vesicles with Phase-Separated Hydrogels

Although W/O droplets have widely been used as a model for artificial cells, a lipid vesicle is a closer model because the cellular membrane is composed of a lipid bilayer. We found that the phase-separated hydrogels could also be formed on droplets surrounded by the cationic and zwitterionic lipid mixture (Figure S9). Because lipid-surrounded W/O droplets were used as precursors of lipid vesicles in a droplet transfer method,[41,42] lipid vesicles supported by the phase-separated hydrogels could be constructed (Figure a).

Figure 6

Lipid vesicle with phase-separated DNA hydrogels on its inner leaflet. (a) Schematic illustrations of the formation of the lipid vesicle. (b) Differential interference contrast (DIC), surface, and cross-sectional microscopy images of the lipid vesicle. Scale bar: 10 μm.

Using the lipid-surrounded droplets, a lipid vesicle with the phase-separated DNA hydrogel on its inner leaflet was successfully generated (Figure b). This result showed that our strategy to use positively charged interfaces for generating phase-separated hydrogel capsules was not limited to W/O droplets composed of Span80 and oleylamine but also applies to other charged materials, including lipid molecules. This result also suggests that interactions with interfaces via other approaches (e.g., hydrophobic interaction) can apply in generating phase-separated capsule-like hydrogels.

Extraction of Phase-Separated Capsule-like Hydrogels into an Aqueous Solution

To showcase the further potential of the phase-separated hydrogels in artificial cell studies, we extracted them into an aqueous solution from lipid vesicles. In the extraction, 5 μM Y- and orthY-motifs and 0.1 μM of the X-linker were adopted to connect the two types of hydrogels. Removing the bilayers by adding detergents (Triton-X 100) allowed for the extraction (Figure S10).

The extracted hydrogels successfully maintained a capsule-like shape (Figure a). Without the X-linker, only destroyed structures (fragments of gel films or gel particles) were observed (Figure S11), suggesting that connecting the two types of hydrogels by the linker is essential to maintain capsule-like shapes after the extraction. It is noteworthy that even after the extraction, the formed patterns by the phase separation remained, which was well visualized in the 3D-reconstructed image (Figure b).

Figure 7

Extraction of the phase-separated DNA hydrogel capsules in aqueous solution from W/O droplets. (a) Representative microscopy images of the extracted capsule surfaces (left) and cross section (right). The capsules were formed using 5/5/0.1 μM Y-motif/orthY-motif/X-linker. Scale bar: 30 μm. (b) Three-dimensional (3D) reconstructed images shown in (a). Scale bar: 30 μm. (c) Schematic illustrations of the enzymatic degradation of the extracted DNA hydrogel capsule. (d) Time series images of the degradation of the DNA hydrogel capsule. t = 0 min means observation start time after the addition of exonuclease. Scale bar: 20 μm.

Regarding the shape of the extracted structures, although the hydrogels formed spherical shapes on the droplet interface before the extraction, they were distorted, i.e., not complete spherical shapes, after the extraction (Figure a and b). This may be attributed to the release of hydrogels from the droplet interface. The hydrogels formed on the spherical droplet’s interface were forced to adsorb onto the interface by electrostatic interaction. After the extraction, the hydrogels were released from the adsorption force. If the actual surface area of the capsule-like hydrogels before the extraction is larger than the ideal values determined by the droplet size, the capsule would have a wavy interface and change into a nonspherical shape after the extraction because of the excess surface area. The elasticity or stiffness of the DNA hydrogel may also affect the shape after the extraction.

Since the extracted gel capsules were formed by DNA nanostructures only, these capsules are able to be degraded by nuclease enzymes. To demonstrate this process, exonuclease I and III were added to the solution containing the extracted capsules (Figure c, Supplementary Note 2). After the addition of the enzymes, the capsules were successfully degraded over time (Figure d). Such degradation was not observed before enzyme addition (Figure S12). These results suggest that our phase-separated DNA hydrogels extracted in an aqueous solution can be combined with enzymatic or other biochemical reactions, which may offer a means to construct cell-sized functional “DNA vesicles.”

Conclusion

We constructed phase-separated DNA hydrogels on W/O droplet interfaces, formed by the lateral phase separation of two orthogonal DNA nanostructure pairs (Figure ). Microscopic observation revealed that the two DNA nanostructures self-assembled into phase-separated hydrogels with bicontinuous, Y-patch, and orthY-patch patterns (Figure ). Differences in the mixing ratio of Y- and orthY-motifs changed the appearance frequency of the patterns (Figure ). Moreover, the addition of the X-linkers, which were designed to cross-link the Y- and orthY-motifs, resulted in the formation of the Mix pattern (Figure ), where both motifs were homogeneously distributed on the interface due to the elimination of orthogonality. We also demonstrated that the patterns of the hydrogels formed by phase separation were altered by adjusting the mixing ratio of Y- and orthY-motifs and the added amount of the sequence-designed X-linkers, represented as a phase diagram (Figure ). The formation of the phase-separated DNA hydrogels was also achieved on a lipid vesicle interface (Figure ), whose inner leaflet was covered with the hydrogels. We finally showed that the capsule-like DNA hydrogels can be extracted in an aqueous solution and maintain their capsule-like shape, with patterns derived from the phase separation (Figure ). These results offer an approach to design and fabricate the capsule made of phase-separated hydrogel using sequence-designed DNAs.

Further investigation of the effect of DNA sequences on phase separation may provide a way for precise pattern control. The results we showed here did not lead to the formation of targeted patterns, such as an only bicontinuous pattern, Y-patch pattern, or orthY-patch pattern. As we discussed in the results shown in Figures d and 3c, the sticky-end sequences may influence pattern formation. Thus, understanding the relations between the sequence and phase-separated hydrogel formation may lead to the formation of the desired pattern.

The formation of the phase-separated hydrogel on the inner leaflet of lipid vesicles (Figure ) would serve to increase the designability of lipid vesicles. Lipid bilayers can phase separate into a cholesterol-rich liquid ordered phase and cholesterol-less liquid-disordered phase. Because DNA can be modified with cholesterol, it can be expected that the phase separation of lipid bilayers by the phase-separated hydrogels on the inner leaflet can be controlled. Moreover, such a phase-separated hydrogel “cortex” in lipid vesicles not only can stabilize vesicles[42] but also may provide heterogeneous mechanochemical properties for lipid vesicle interfaces.

We envision the construction of functional microcapsules using our phase-separated hydrogels. As shown in Figure , the extracted gel capsules were successfully degraded by exonuclease enzymes. Apart from enzymes, also photo-[43] or molecular-stimulation[44] can alter the interaction between DNA molecules. Such stimuli-responsivity will allow the degradation of targeted patches in the hydrogels, which offers a cargo release function encapsulated in capsule-like hydrogels. In addition, because DNA hydrogels can dynamically alter their shapes via stiffness changes in response to DNA strands with specific sequence,[45] our hydrogel capsules may obtain motility function by repetitive shape changes. Moreover, DNA can be modified with various functional proteins,[46] which may enable us to construct phase-separated hydrogel capsules comparable to cellular membranes. Furthermore, DNA molecular devices capable of signal processing[47] or stimuli-responsive actuation[48,49] will be implemented into our microsized DNA-hydrogel capsule. We believe that such functional capsules composed of DNA will provide a new approach to developing capsule structures for artificial cell studies and molecular robotics.[50]