Introduction

The development of protocell models that display structural features and biomimetic functions of life is an important milestone in systems chemistry, soft matter bioengineering, and synthetic protobiology1,2. The protocells exhibit the features of morphological flexibility, dynamic responsiveness, and collective behaviors3,4,5, facilitating implementation of complex reconfiguration into high-order architectures6,7. Coacervate microdroplets formed by liquid-liquid phase separation (LLPS) have been recently developed as membrane-less protocell models which exhibit a wide range of life-like properties8,9, including compartmentalization10,11,12,13, enrichment of biomolecules14,15, regulation of enzymatic reactions16,17,18,19, and intercellular communication20,21,22. A less explored aspect in this field is the programmable reconfiguration of these protocell droplets that allow for transformation into higher-order structures under biologically relevant conditions23,24,25.

Recent advances have demonstrated the transformation process of membrane-less coacervates into coacervate vesicles via different mechanisms, including vacuolization26,27,28,29,30, membranization31,32,33, multiphase separation34, and interface reorganization35,36,37,38. For instance, by taking the vacuolization mechanism, vacuolized protocells not only exhibit morphological changes, but are also endowed with functional promotion like biomolecular enrichment39,40,41, acceleration of biochemical reactions42,43, and enhanced interfacial stabilization44. Moreover, coacervate vesicles also enable to act as delivery vehicles for biopharmaceuticals, capable of preserving bioactivity and augmenting delivery efficacy45,46,47. However, a direct transformation from the membrane-less droplets into multicompartmentalized cell-like entities remains challenging48,49,50.

In this work, we demonstrate a helicoid pattern-featured vacuolization for developing organelle-containing coacervate vesicles. We construct liquid crystalline (LC) polysaccharide/double-tailed surfactant coacervates, realizing complex transformation into coacervate vesicles initiated by enzyme-mediated polysaccharide hydrolysis into oligosaccharide. The distinct hydrolysis rates result in LC coacervates experiencing intensive vacuolization step and coalescence step till reaching equilibrium. The resultant electrostatic interaction changes cause dissolution of LC structure which allows for the helicoid pattern within the intermediates. Alternatively, the charge screening or protonation also enables to initiate this complex transformation behavior. Significantly, the integration of interactive biomolecules enhances molecular interactions in the LC coacervates, resulting in transformation into yolk-shell coacervate vesicles capable of supporting enzymatic reactions. Our results present a typical reconfigurable coacervate protocells for implementing transformation into hierarchical compartments via complex vacuolization process and provide application potentials in dynamic life-like systems and bioengineering field.

Results

Construction of LC coacervate microdroplets

LC coacervate microdroplets were prepared with a negatively charged carboxymethyl amylose (Cm-Am, 40 k Da, substitution degree of 1.2) and a cationic surfactant, didodecyldimethylammonium bromide (DDAB). Cm-Am can be hydrolyzed into fragments as catalyzed by amylase, involving the breakdown of the amylose polymers into low molecular weight oligosaccharides that bear ionic carboxymethyl groups, i.e., hydrolyzed Cm-Am (H-Cm-Am) (Fig. 1a, Supplementary Figs. S1 and S2). DDAB is a cationic surfactant, featuring a hydrophilic head group and two hydrophobic tails. The synergistic interactions of electrostatic attraction and the hydrophobic effect drove the complexation of Cm-Am and DDAB, ultimately resulting in the formation of LC coacervate microdroplets, which underwent complex structural transition in the presence of amylase (Fig. 1b). To better understand alterations in structures, we defined assemblies related to the transformation. Specifically, LC coacervate droplets were coacervates formed by LLPS with ordered liquid crystalline structure51,52, while helicoidal vesicles represented vacuolized LC coacervates featured by helicoidal inner patterns, coacervate vesicles were assemblies consisting of a coacervate rim and a water-filled cavity45, and yolk-shell coacervate vesicles (i.e., organelle-containing coacervate vesicles) were assemblies with a coacervate rim, a water-filled cavity, and an inner coacervate organelle. Coacervate microdroplets were isotropic coacervate droplets.

Fig. 1: Construction of enzyme-responsive LC coacervate microdroplets.
Fig. 1: Construction of enzyme-responsive LC coacervate microdroplets.
Full size image

a Structural illustration of amylase-mediated hydrolysis of Cm-Am into fragments. b Schematic illustration representing morphological changes of Cm-Am/DDAB LC coacervate microdroplets driven by enzymatic hydrolysis. Two structural transformation pathways were shown: (top) LC coacervate microdroplets without incorporated biomolecules underwent vacuolization in the presence of amylase, forming a helicoidal intermediate structure that eventually reconfigured into coacervate vesicles; (bottom) LC coacervate microdroplets containing biomolecules transformed into coacervate vesicles containing internal organelles. c Phase diagram illustrating the formation of microdroplets at varying concentrations of Cm-Am and DDAB. The coacervation region was determined by optical microscope. d Optical microscopy image of Cm-Am/DDAB LC microdroplets (total concentration of 10 mM, monomer molar ratio of 1:1). e Size distribution of the LC microdroplets, with a mean diameter of 3.5 ± 1.0 μm (n = 310 samples). f POM image of the LC microdroplets exhibiting a Maltese-cross extinction pattern. g SAXS profile of the LC microdroplets exhibiting a sharp Bragg reflection peak at 0.177 Å−1, corresponding to a periodical spacing of 3.55 nm. Fluorescence microscopy images and the corresponding fluorescence intensity line scans displaying homogenous distribution of Hoechst in Cm-Am/DDAB LC microdroplets (h), while TAMRA-dsDNA (1012 bp) was excluded to the surface (i). The term (arb. units) is abbreviated for arbitrary units. Scale bars, d, f 5 µm; h, i 2 µm. The experiments for c–i were repeated three times independently with similar results. Source data are provided as a Source data file.

The formation of Cm-Am/DDAB LC coacervate microdroplets was observed across a broad concentration range, as illustrated by a phase diagram (Fig. 1c and Supplementary Fig. S3), exhibiting enhanced coacervation at higher concentrations. In this work, the LC microdroplets were primarily obtained at a total concentration of 10 mM with a monomer molar ratio of 1:1 (marked by a red triangle in Fig. 1c), wherein the LC coacervates were capable of exhibiting macroscopically visible phase separation in the Eppendorf tubes after centrifugation. These LC microdroplets displayed discrete spherical morphologies with high optical contrast, as well as liquid-like property, evidenced by spontaneous fusion behavior in the absence of external stimulus or special conditions (Supplementary Fig. S4 and Movie S1). The average diameter was measured to be 3.5 ± 1.0 μm, and the zeta potential was determined to be +3.9 ± 1.0 mV (Fig. 1d, e). Additionally, the LC microdroplets exhibited birefringent Maltese cross patterns with concentric retardation lines under polarized optical microscopy (POM) (Fig. 1f). This phenomenon disappeared after heating at 60 °C for 80 min (Supplementary Fig. S5), indicative of the system undergoing a phase transition at 60 °C. The LC coacervates exhibited an onion-like multilamellar structure without central water-filled lumen as characterized by cryogenic scanning electron microscopy (Supplementary Fig. S6). The ordered internal structure of the LC coacervates was further confirmed by small-angle X-ray scattering (SAXS), which revealed a sharp Bragg reflection peak at 0.177 Å−1, corresponding to a periodic molecular spacing of 3.55 nm (Fig. 1g and Supplementary Table 1). Considering the bilayer thickness of approximately 2.45 nm53, this spacing likely represented the periodic stacking of a DDAB bilayer and Cm-Am molecules.

LC coacervate microdroplets were able to sequestrate a variety of biomolecules. Experimentally, fluorescein isothiocyanate (FITC)-labeled dextran with relatively low molecular weights (i.e., 10 k and 20 k Da) were uniformly distributed within the interior of the LC microdroplets (Supplementary Fig. S7). In contrast, FITC-dextran with a high molecular weight (i.e., 40 k and 70 k Da) were predominantly absorbed onto the surface of the LC microdroplets, while FITC-dextran of 250 k Da was mostly excluded. The molecular weight-selective sequestration behavior of LC microdroplets was further demonstrated with a range of client compounds (Fig. 1h, i and Supplementary Fig. S8).

Amylase-induced LC microdroplet-to-coacervate vesicle transition

With the addition of amylase, the LC coacervate microdroplets underwent a two-step transition process toward coacervate vesicles with no other assemblies in the water-filled cavity. Initially, the LC microdroplets experienced intense vacuolization, resulting in the formation of an intermediate structure characterized by helicoidal inner patterns (225 s), followed by the second step, i.e., expansion of the assembly (375 s) caused by weakened multivalent electrostatic complexation and increased internal osmotic pressure during the hydrolysis of Cm-Am into small fragments (Fig. 2a, Supplementary Fig. S9, and Movie S2). The LC microdroplet with doped RITC-Cm-Am vacuolized into helicoidal vesicles as an intermediate, which underwent coalescence, ultimately giving rise to a hollow vesicle determined as a coacervate vesicle (Fig. 2b and Supplementary Movie S3). As a result, the helicoidal vesicles or coacervate vesicles both possessed thicker rims (i.e., 0.4 ± 0.2 or 0.3 ± 0.1 μm) and strong enrichment capability by compared to DDAB vesicles with thinner membranes (i.e., about 0.2 μm) and poor membrane permeability54, wherein the rim and membrane thickness were both measured by fluorescence microscopy and analyzed based on full width at half maximum of gray values in linescan profiles (Supplementary Figs. S10 and S11). Except for the differences in thickness of rims and vesicle membranes, we could also observe that coacervate vesicles were always hollow, whereas the vesicles sometimes possessed multilamellar structures. The cryogenic transmission electron microscopy (cryo-TEM) further confirmed the rim thickness of coacervate vesicles (i.e., about 120 nm) was much larger than membrane thickness in vesicles (i.e., about 6 nm) (Supplementary Fig. S12). In the reconfiguration process, the enzyme (RITC-amylase) was homogenously distributed (Fig. 2c). Notably, the LC coacervate microdroplet experienced a 1.6-fold increase in diameter, from 4.2 to 6.8 µm (Fig. 2d). This change was statistically supported by a comparison of the diameter of LC microdroplets (Fig. 1e, Dcoac of 3.5 ± 1.0 µm) and the resultant coacervate vesicles (Supplementary Fig. S13, Dvesicle of 4.7 ± 2.0 µm), yielding a size ratio Dvesicle/Dcoac of 1.3 ± 0.7. After reaching equilibrium, no obvious vacuolization or expansion was observed in the coacervate vesicle, even with the addition of extra amylase (Fig. 2e). Reconstructed three-dimensional (3D) images confirmed the occurrence of vacuolization and the appearance of helicoidal structure during the LC coacervate microdroplet-to-coacervate vesicle transition, accompanied by an obvious size expansion (Fig. 2f, g). Such reconfiguration process was also achieved in the LC coacervates with higher total concentrations (e.g., 15 or 20 mM) (Supplementary Fig. S14). Upon the addition of PBS buffer (10 mM, pH 6.9, containing 1 mM CaCl2), the Cm-Am/DDAB LC coacervates exhibited negligible changes. Furthermore, the helicoid patterns observed in the intermediate structures were predominantly left-handed, occurring in approximately 80% of cases (Fig. 2h). While the origin of mesoscale chirality remains unclear, we speculated that this phenomenon was related to the chirality of amylose (Supplementary Fig. S15), which possibly induced supramolecular chirality within the LC microdroplets55.

Fig. 2: Structural transformation of LC coacervate microdroplets into coacervate vesicles.
Fig. 2: Structural transformation of LC coacervate microdroplets into coacervate vesicles.
Full size image

a Time-dependent optical microscopy images of an Cm-Am/DDAB LC microdroplet vacuolizing into an intermediate structure with helicoid inner pattern, followed by reconfiguring into a coacervate vesicle in the presence of amylase (1.5 mg/mL). Time series of fluorescence microscopy images displaying helicoid pattern coalesced into outer rim within the vacuolized microdroplets (doped RITC-Cm-Am) (b), whilst RITC-amylase was uniformly accommodated by the assemblies emerged in the reconfiguration (c). d The corresponding diameter changes during the reconfiguration process in (a) exhibiting the size increased from 4.2 μm of droplet to 6.8 μm of coacervate vesicle. Data were extracted from Supplementary Movie S2, while points 1, 2, 3 corresponded to optical images in (a). e Time-dependent fluorescence microscopy images showing the coacervate vesicle remained unchanged with adding a new batch of amylase when the hydrolysis of Cm-Am reached equilibrium. Reconstructed 3D fluorescence microscopy images for reconfiguration process of a Cm-Am/DDAB LC coacervate doped with NBD-PE (f) and corresponding fluorescence intensity diagram according to linescan profiles along the white dashed arrows (g) showcased increase in diameters after transforming into coacervate vesicle. Grid width = 1 µm. h Statistic analysis for the directions of helicoid patterns within vacuolized coacervates indicating 80% left-handed and 20% right-handed chirality, as shown in the fluorescence microscopy images of intermediates with doped RITC-Cm-Am (n = 220 samples). i Statistics for diameters of LC microdroplets before (Dcoac) and after adding amylase (Dvesicle). Dvesicle was measured after coacervate vesicle remained stable or at the time when it ruptured. A fitted line with a slope of around 1.6 was obtained from the majority data of intact and ruptured coacervate vesicle events (n = 56 samples). j Time-dependent alterations of microstructure populations during the reconfiguration process (n = 40 samples). Scale bars, a 2 μm; b, c, e, h, 5 μm. The experiments for a–j were repeated three times independently with similar results. Source data are provided as a Source data file.

Additionally, the vacuolization process was able to induce the rupture of coacervate vesicles, leading to their division (Supplementary Fig. S16). We statistically analyzed the diameters of the initial LC microdroplets (Dcoac) and the coacervate vesicles (Dvesicle), finding a linear correlation between them (Fig. 2i). A slope of 1.6 corresponding to the Dvesicle/Dcoac ratio was obtained, consistent with the diameter ratio (Fig. 2d). The formation of coacervate vesicles was likely maintained by a balance between the osmotic pressure and Laplace pressure within coacervate vesicles30.

We then examined the changes in populations of LC coacervate microdroplets, helicoidal structured vesicles, and coacervate vesicles after the addition of amylase (Fig. 2j). At 340 s, 90% of the LC coacervate microdroplet population (blue region 1) transformed into helicoidal structured vesicles (light blue region 2, 52% of the total populations), which was further reconfigured into coacervate vesicles (pink region 3, representing 38% of the total population). At 760 s, the proportion of coacervate vesicles increased to 83%, then remained stable thereafter. Meanwhile, the fractions of the initial LC microdroplets and helicoidal structured vesicles ultimately decreased to 7% and 10%, respectively, possibly due to the insufficient enzyme activity.

To gain further insights to the transition process, we investigated the impact of amylase concentration on the final microstructures of the Cm-Am/DDAB LC coacervate microdroplets. The 1.50 mg/mL amylase concentration, primarily used in this study, led to coacervate vesicles as the predominant final microstructure, accounting for 44% of the total population (Supplementary Fig. S17). In comparison, lower amylase concentrations (e.g., 0.15 and 0.75 mg/mL) primarily resulted in intact LC coacervate microdroplets (89% and 35% of the total population, respectively), while higher amylase concentrations (e.g., 2.25 and 3.00 mg/mL) predominantly led to multicompartmentalized droplets (51% and 47% of the total population, respectively), due to either insufficient or overly rapid hydrolysis of Cm-Am.

Possible mechanism of structural transition

To demonstrate the enzyme-mediated Cm-Am hydrolysis, we applied gel permeation chromatography (GPC) to monitor changes in the molecular weight of hydrolysis fragments (Fig. 3a). Upon the addition of amylase, the molecular weight decreased sharply from 40 to 5 k Da within 300 s, consistent with transformation of LC coacervates into helicoid-structured vesicles at vacuolization stage (225 s) (Fig. 2a) and the diameter of resulted assemblies reached a plateau at 300 s (Fig. 2d). Ultimately, the Cm-Am hydrolysis reached equilibrium accompanied with the generation of isotropic coacervate vesicles at coalescence stage (Supplementary Fig. S18). To further understand the effect of Cm-Am hydrolysis, we examined the internal ordering changes of assemblies by POM. The ordered LC droplets transformed into LC helicoid-structured vesicles at vacuolization stage (225 s), which was roughly consistent with rapid hydrolysis of Cm-Am within 300 s and eventually isotropic coacervate vesicles at coalescence stage (575 s) that was almost aligned with equilibrium hydrolysis of Cm-Am after 300 s (Fig. 3b and Supplementary Movie S4). We proposed the degree of Cm-Am hydrolysis determined the structural ordering of the assemblies: low-degree hydrolysis of Cm-Am electrostatically complexed with DDAB to form LC microdroplets, while high-degree hydrolysis led to the formation of isotropic coacervate vesicles (Fig. 3c). As proof of concept, we compared the structural ordering of coacervate microdroplets by using Cm-Am with varied degree of hydrolysis. H-Cm-Am 1 with 2 min hydrolysis (9 min elution) complexed with DDAB led to LC coacervate microdroplets (Fig. 3d, e). Whereas H-Cm-Am 2 with 12 h hydrolysis (10 min elution) gave rise to isotropic coacervate vesicles with DDAB (Fig. 3f). This difference was probably because long-chain Cm-Am hindered the mobility and rotation of DDAB, readily leading to LC structure, while short-chain fragments were unfavorable to be orderly organized with DDAB56,57.

Fig. 3: Mechanism involved in the polysaccharide hydrolysis-induced reconfiguration of LC coacervate microdroplets into isotropic coacervate vesicles.
Fig. 3: Mechanism involved in the polysaccharide hydrolysis-induced reconfiguration of LC coacervate microdroplets into isotropic coacervate vesicles.
Full size image

a Molecular weight of Cm-Am fragments in the course of amylase-induced hydrolysis inside Cm-Am/DDAB LC microdroplets. b Time series of POM images showing the structural transition from LC coacervates to isotropic coacervate vesicles initiated by amylase (1.5 mg/mL). c Scheme representing the hydrolysis degree of Cm-Am determined structural ordering of the resultant assemblies, namely, low-degree-hydrolysis Cm-Am formed LC coacervates with DDAB, while high-degree-hydrolysis Cm-Am led to isotropic coacervate vesicles with DDAB. d GPC elution profiles showing H-Cm-Am 1 were the 2-min dominant species of Cm-Am hydrolysis, whilst H-Cm-Am 2 became the 12-h predominant products. The reverse peaks were attributed to solvent effects. Fluorescence microscopy and POM images displaying H-Cm-Am 1 and DDAB formed ordered LC coacervate microdroplets (e), whereas H-Cm-Am 2 and DDAB generated isotropic coacervate vesicles (f). Time-lapse fluorescence microscopy images (g) and corresponding kinetics (h) for FRAP of RITC-dextran (10 k Da)-doped Cm-Am/DDAB LC microdroplet and H-Cm-Am/DDAB coacervate vesicle, respectively, exhibiting around 8-fold increase of molecular mobility in the coacervate vesicle by compared to that in the initial LC droplets. The data represented mean ± s.d. (n = 3 independent experiments). Scale bars, b, e–g 5 μm. The experiments for a, b, d–h were repeated three times independently with similar results. Source data are provided as a Source data file.

To investigate the structural differences between the LC microdroplets and coacervate vesicles, we performed fluorescence recovery after photobleaching (FRAP) by using RITC-dextran (10 k Da) as a fluorophore. After photobleaching by 569 nm laser, a rapid fluorescence recovery was observed, with recovery rate of 58% in LC microdroplets and 80% in coacervate vesicles within 500 s (Fig. 3g, h). Although the adjacent area available for replenishing the bleached region was smaller in coacervate vesicles, the recovery rate was faster compared to the LC microdroplets, indicating higher molecular diffusivity in the vesicular system. Specifically, the diffusion coefficient in coacervate vesicles (1.1 × 10−14 m²/s) was approximately eight times higher than in LC coacervate droplets (1.4 × 10−15 m²/s). This difference was attributed to the less ordered microenvironment within the coacervate vesicles, which facilitated faster molecular migration, whereas the liquid crystalline matrix of the microdroplets imposed restrictions on passive diffusion58.

We further investigated the transition from LC microdroplets to coacervate vesicles in mixed systems with varying Cm-Am/DDAB ratios. An LC phase was observed when the monomer molar ratio ranged from 8:1 to 1:4, with helicoidal vesicles noted during the transition (Supplementary Fig. S19). In contrast, when the monomer molar ratio exceeded 8:1 (e.g., 9:1) or fell below 1:4 (e.g., 1:5), isotropic structures were formed, and no helicoid patterns appeared in the intermediate structures. Therefore, we hypothesized that the highly ordered LC structure of the initial coacervate microdroplets was associated with the formation of helicoidal vesicles.

We next examined the structural transition in a variety of protocell microdroplets by varying the coacervate components. First, we replaced DDAB with dimethyldimyristylammonium bromide (DMAB) and observed the formation of ordered LC microdroplets (Fig. 4a) with a periodic spacing of 4.19 nm (Fig. 4b, c)59. The transition from LC microdroplet to coacervate vesicle was observed, which was accompanied by the formation of helicoidal vesicles as intermediates (Fig. 4d and Supplementary Movie S5). Likewise, carboxymethylated dextrin (Cm-dextrin, substitution degree of around 0.9) was used as a short-chain analog of Cm-Am (Fig. 4e and Supplementary Fig. S20), which upon interacting with DDAB, gave rise to LC microdroplets with a periodical spacing of 3.52 nm (Fig. 4f, g)60. The Cm-dextrin/DDAB LC coacervates transitioned into helicoidal structures and eventually coacervate vesicles (Fig. 4h and Supplementary Movie S6). By contrast, no helicoidal vesicles emerged in the structural transition of isotropic coacervates formed by Cm-Am and dodecyltrimethylammonium bromide (DTAB), myristyltrimethylammonium bromide (MTAB), or poly(allylamine hydrochloride) (PAH), wherein the coacervate vesicles were also ultimately generated (Fig. 4i–l and Supplementary Fig. S21)61. Therefore, the appearance of helicoid pattern was possibly attributed to Cm-Am-hydrolysis-induced frustration of lamellar structure in LC droplets, leading to the fusion and connections between disrupted phase62,63.

Fig. 4: Structural transition of LC microdroplets under varied conditions.
Fig. 4: Structural transition of LC microdroplets under varied conditions.
Full size image

Schematic representations (a, e, i), POM images (b, f, j), SAXS profiles (c, g, k), and time-lapse optical (d, h) or fluorescence microscopy images (l) depicting ordered Cm-Am/DMAB (a–d) or Cm-dextrin/DDAB LC microdroplets (e–h) transformed into coacervate vesicles via helicoid pattern-featured vacuolization in the presence of amylase (0.9 or 0.3 mg/mL), while disordered Cm-Am/DTAB isotropic microdroplets (doped with RITC-Cm-Am) (i–l) reconfigured into coacervate vesicles without helicoid inner patterns after adding amylase (0.9 mg/mL). m Schematic illustration represented the addition of sodium chloride or hydrochloric acid initiated structural changes of Cm-Am/DDAB LC microdroplets into coacervate vesicles by charge screening of LC microdroplets or protonation of Cm-Am. n Time-dependent optical microscopy images showed the reconfiguration process of the LC microdroplets induced by 80 mM of sodium chloride. o Diagram of the LC microdroplets after adding varied concentrations of sodium chloride, displaying three regions divided based on the final states, including LC coacervates, coacervate vesicles, and solutions. Time series of optical microscopy images displaying the LC microdroplets remained unchanged in the presence of 15 mM of sodium chloride (p), while the LC droplets collapsed with adding 50 mM of sodium chloride (q). Scale bars, 5 μm. The experiments for b–d, f–h, j–l, n–q were repeated three times independently with similar results. Source data are provided as a Source data file.

Finally, we examined the role of electrostatic interactions by adding sodium chloride to screen the charge attraction and hydrochloride acid to protonate carboxymethyl group of Cm-Am (Fig. 4m). In both cases, LC microdroplets transformed into helicoidal vesicles and eventually coacervate vesicles (Fig. 4n, Supplementary Fig. S22 and Movies S7, S8). A lower dose of sodium chloride (15 mM) or hydrochloric acid (1 mM) caused negligible changes in the LC microdroplets (Fig. 4o–q). In contrast, high concentrations of perturbation reagents (e.g., 150 mM of sodium chloride or 50 mM of hydrochloric acid) led to the direct dissociation of the LC microdroplets due to strong charge screening or protonation. This was also observed by the addition of both amylase and sodium chloride or hydrochloric acid (Supplementary Fig. S23).

Simultaneous formation of artificial organelles within protocell compartments

Interestingly, the incorporation of biomolecules into ordered LC coacervate microdroplets led to the formation of organelle-containing protocells, which were also referred to as yolk-shell coacervate vesicles (Fig. 5a). Compared to the coacervate vesicles obtained after the reconfiguration of LC coacervates without biomolecules, the yolk-shell coacervate vesicles featured a coacervate organelle in the water-filled cavity. To further interrogate the complex process, we separately loaded proteins and nucleic acids into the LC microdroplets, such as fluorescence-labeled bovine serum albumin (RITC-BSA), horseradish peroxidase (RITC-HRP), glucose oxidase (FITC-GOx), carboxytetramethylrhodamine-labeled single-stranded oligonucleotide (TAMRA-ssDNA, 99 nt), and TAMRA-dsDNA (25 bp) (Fig. 5b, c, Supplementary Fig. S24 and Movies S9, S10). The resulting coacervates maintained liquid crystalline structures and predominantly transformed into coacervate vesicles with an internal coacervate organelle (i.e., yolk-shell coacervate vesicles) in the presence of amylase (Supplementary Figs. S25 and S26). The thickness of rims in yolk-shell coacervate vesicles was measured by fluorescence microscopy and analyzed based on full width at half maximum of gray values in linescan profiles, exhibiting thicker rims (i.e., 0.4 ± 0.1  μm) by compared to DDAB vesicle membranes (i.e., about 0.2 μm) (Supplementary Fig. S27). The yolk-shell coacervate vesicles also possessed a strong enrichment ability, in contrast to controllable permeability of vesicles. During the structural transition, the coacervate components were homogenously distributed throughout the assemblies (Fig. 5d and Supplementary Movie S11).

Fig. 5: LC coacervate microdroplets with integrated biomolecules reconfigured into isotropic yolk-shell coacervate vesicles.
Fig. 5: LC coacervate microdroplets with integrated biomolecules reconfigured into isotropic yolk-shell coacervate vesicles.
Full size image

a Scheme showed the transformation of LC coacervate microdroplets with integrated proteins or nucleic acids into organelle-containing coacervate vesicles induced by the hydrolysis of Cm-Am. Time-lapse fluorescence microscopy images displaying LC microdroplets with the incorporation of RITC-BSA (0.3 mg/mL) (b) or TAMRA-ssDNA (99 nt, 1.0 μM) (c) reconfigured into yolk-shell coacervate vesicles initiated by amylase (0.9 mg/mL). Time sequence of fluorescence microscopy images exhibited homogenous distribution of RITC-Cm-Am during the reconfiguration of HRP-integrated LC microdroplets (d), while the LC microdroplets with integrated FITC-dextran (10 k Da, 0.5 mg/mL) transformed into coacervate vesicles in the presence of amylase (0.9 mg/mL) (e). f Time-dependent FRAP kinetics as evaluated by fluorescent probe of FITC-PEI for LC microdroplets with integrated HRP, ssDNA (99 nt), or dextran (10 k Da), LC microdroplets without biomolecules, and rim or inner organelle within yolk-shell coacervate vesicles. The data represented mean ± s.d. (n = 3 independent experiments). g Reconstructed 3D merged fluorescence microscopy image displaying an artificial organelle-containing coacervate vesicle was formed after reconfiguration of LC droplets with the simultaneous integration of Py-HRP, TAMRA-ssDNA (99 nt), and FITC-dextran (10 k Da). It exhibited nearly white fluorescence, indicating the homogenous distributions of these biomolecules. Grid width = 1 µm. Schematic illustration of HH-min-mediated cleavage of a FRET-ssRNA substrate to liberate the fluorophore from the quencher strand (h) and the corresponding ribozyme reaction was supported within artificial organelle-containing coacervate vesicles (i). Time-dependent fluorescence microscopy images of a yolk-shell coacervate vesicle (j), Cm-Am/DDAB LC microdroplet (k), and the corresponding fluorescence intensity plots (l) exhibited faster reaction rates within rim and inner organelle than that in the LC coacervate microdroplet. The data represented mean ± s.d. (n = 3 independent experiments). Scale bars, b–e, 5 μm; j, k, 2 µm. The experiments for b–g, j–k were repeated three times independently with similar results. Source data are provided as a Source data file.

We hypothesized that the formation of organelle-containing coacervate vesicles was attributed to the enhanced association between interactive biomolecules and DDAB through strong hydrophobic and electrostatic interactions64,65. The multivalent complexation between biomolecules (e.g., proteins, nucleic acids), Cm-Am, and DDAB resulted in LC coacervate microphase that restricted the amylase-catalyzed digestion of Cm-Am, leaving the unreacted inner LC coacervates. In a control experiment, we introduced FITC-dextran (10 k Da) into the LC coacervates and observed the generation of coacervate vesicles without internal organelles (Fig. 5e and Supplementary Movie S12). This is due to the weak interactions between FITC-dextran (10 k Da) and Cm-Am/DDAB LC coacervates. To evaluate the binding affinities between DDAB and biomolecules, we utilized isothermal titration calorimetry (ITC) to determine the dissociation constants (Kd), showing 1.57 ± 0.35 × 10−7, 1.00 ± 0.00 × 10−9, or 1.00 ± 0.00 × 10−3 M for the interactions of DDAB with BSA, ssDNA (99 nt), or dextran (10 k Da), respectively (Supplementary Fig. S28). It indicated significantly stronger affinities of DDAB with BSA (i.e., 7-fold) or ssDNA (i.e., 1060-fold) by compared to that with Cm-Am (i.e., 1.06 ± 0.28 × 10−6 M), while the interactions between dextran and DDAB could be neglected. To further support this, we evaluated the interaction strength in the initial LC coacervates, with and without the incorporation of biomolecules, using the critical salt concentration required to dissolve the droplets66. The resulting critical salt concentrations were 353, 429, and 188 mM for the LC droplets with incorporated HRP, ssDNA (99 nt), and dextran (10 k Da), corresponding to 3.6-, 4.4-, or 1.9-fold increases compared to 97 mM for LC droplets without biomolecules (Supplementary Fig. S29). To assess the interaction strength, we measured the diffusion coefficients of FITC-labeled poly(ethylenimine) (PEI) within these droplet systems67,68, which was determined to be 2.2 × 10−15, 1.8 × 10−15, and 3.5 × 10−15 m2/s in the LC microdroplets with incorporated HRP, ssDNA (99 nt), and dextran (10 k Da), respectively, corresponding to 1.9-, 2.3-, and 1.1-fold decrease in comparison to that of LC droplets without biomolecules (4.1 × 10−15 m2/s) (Fig. 5f and Supplementary Fig. S30). We therefore believed that molecular associations between guest biomolecules and Cm-Am/DDAB LC coacervates resulted into the formation of inner organelle-containing coacervate vesicles (Fig. 5g).

In the yolk-shell coacervate vesicles, after reaching equilibrium, we observed that the rim and inner organelle remained persistently attached without coalescing, even after the addition of a new batch of amylase, when the hydrolysis of Cm-Am reached equilibrium (Supplementary Figs. S31 and S32). We hypothesized that the composition in the rim and internalized organelle was different, i.e., the degree of hydrolysis of Cm-Am was higher in the rim than in the internal organelles. Specifically, the critical salt concentration in the rim and the organelles was determined to be 9.7 and 50 mM, respectively (Supplementary Fig. S33). Additionally, the diffusion coefficient of FITC-PEI within the rim was 1.4 × 10−14 m2/s, 1.8-fold higher than that in the organelle (i.e., 7.6 × 10−15 m2/s) (Fig. 5f), suggesting a molecularly more crowded environment inside the organelle.

The ability confining biochemical reactions within organelle-containing coacervate vesicles was crucial for the functionality of artificial cells. To investigate this, we performed a hammerhead ribozyme (HH-min)-mediated RNA catalysis reaction within the organelle-embedded coacervate vesicles69. In this system, the cleavage of a fluorescence resonance energy transfer (FRET)-substrate strand catalyzed by HH-min increased the distance between the fluorophore (6-carboxyfluorescein) and Black Hole quencher 1 (BHQ1), resulting in an increase of green fluorescence intensity (Fig. 5h, i). Significantly, the LC coacervates with homogenously integrated cyanine 5 labeled-HH-min (Cy5-HH-min) and carboxyfluorescein-labeled substrate (FAM-substrate) were capable of transforming to yolk-shell coacervate vesicles initiated by amylase (Supplementary Fig. S34), while a progressive increase of green fluorescence intensity from HH-min-mediated reaction was observed in the rim and inner artificial organelle (Fig. 5j and Supplementary Movie S13), confirming the support of enzymatic reactions by the system. Moreover, the RNA catalysis rates in yolk-shell coacervate vesicles were faster than that in the initial LC microdroplets (Fig. 5k, l), attributing to the higher molecular diffusivity in the isotropic yolk-shell coacervate vesicles, which were confirmed by the disappearance of birefringence phenomenon in the POM image and Bragg reflection peak in the SAXS profile (Supplementary Figs. S35 and S36). Meanwhile, the reaction rate in the rim was slightly faster than that in the organelle.

In conclusion, we have demonstrated an LC microdroplet-based protocell system that can transform into hierarchical coacervate vesicles. This system was formed through the direct complexation of a negatively charged polysaccharide and a double-tailed cationic surfactant, driven by electrostatic interactions and hydrophobic forces, which resulted in the formation of multilamellar LC microdroplets. The addition of amylase catalyzed the hydrolysis of the polysaccharide, leading to the dissociation of the multilamellar structures and subsequent swelling of the assemblies. This process ultimately drove the transition from LC microdroplets to helicoid-structured vesicles and eventually coacervate vesicles. Importantly, the incorporation of biomolecules, such as proteins and nucleic acids, facilitated the morphological transformation into coacervate vesicles embedding a coacervate droplet which structurally resembled artificial organelles capable of supporting biochemical reactions. Our findings highlight a protocell model capable of transforming into cell-like complex compartments, opening new avenues for replicating cellular structure and function.

Methods

Preparation of LC or isotropic coacervate microdroplets

Typically, the pH of Cm-Am solution (10 mM) and DDAB solution (10 mM) were adjusted to 8 before mixing. Cm-Am/DDAB LC coacervates were prepared by directly mixing the aforementioned aqueous solutions of Cm-Am (25 μL, 10 mM, pH 8) and DDAB (25 μL, 10 mM, pH 8), following vortexing for 5 s. The pH was then checked and adjusted as necessary to ensure it remained at 8. Other coacervate suspensions were prepared following the similar procedure except for using other surfactants or polyelectrolytes under varied total concentrations and monomer molar ratios, such as Cm-Am/DTAB isotropic coacervate suspension (50 mM, 1:1), Cm-Am/MTAB isotropic microdroplet dispersion (50 mM, 1:1), Cm-Am/DMAB LC coacervate dispersion (5 mM, 1:1), Cm-dextrin/DDAB LC microdroplet suspension (10 mM, 1:1), and Cm-Am/PAH isotropic coacervate suspension (50 mM, 2:1).

Turbidity diagram

The turbidity diagram was measured by mixing different concentrations of Cm-Am (5, 10, 20, 40, and 50 mM, pH 8) and DDAB (10, 20, 30, and 40 mM, pH 8) solutions with different monomer molar ratios. The turbidity of these samples was recorded by a plate reader (CLARIO star plus, BMG Labtech) at an excitation wavelength of 600 nm, where no absorbance for coacervate components. The coacervation behaviors of the samples were investigated on an optical microscope after incubation for 5 h. Measurements were repeated three times.

Sequestration properties of Cm-Am/DDAB LC coacervate microdroplets

Typically, Cm-Am/DDAB LC coacervate microdroplets (100 μL) were prepared in Eppendorf tubes using the procedure described above. Then, Hoechst (1 μL, 5 μM, in water), HPTS (1 μL, 20 μM, in water), Rh6G (1 μL, 50 μM, in water), NBD-PE (1 μL, 20 μg/mL, in ethanol), FITC-dextran (10, 20, 40, 70, and 250 k Da, 1 μL, 1 mg/mL, in water), RITC-DEAE-dextran (20 k Da, 1 μL, 1 mg/mL, in water), RITC-PDDA (1 μL, 0.5 mg/mL, in water), TAMRA-dsDNA (1012 bp, 1 μL, 2 mg/mL, in 10 mM of Tris-HCl buffer with containing 1 mM of EDTA, pH 8), or RITC-Cm-Am (1 μL, 1 mg/mL, in water) was added to the Cm-Am/DDAB LC microdroplet suspension, respectively.

Enzyme-mediated reconfiguration from coacervate microdroplets to coacervate vesicles

Typically, 30 µL of Cm-Am/DDAB LC coacervate microdroplets prepared as the above procedure were added to a sample cell with a glass coverslip, settling for ca. 10 min. Then, 5.3 µL of amylase (10 mg/mL in 10 mM PBS buffer containing 1 mM CaCl2, pH 6.9) was added into the coacervate-containing sample cell. The reconfiguration process was monitored by confocal laser scanning microscopy (CLSM).

The amylase-mediated transformation of LC coacervates with higher concentrations was explored following the above procedures except using 15 mM or 20 mM of Cm-Am/DDAB LC coacervate microdroplets (30 µL, monomer molar ratio of 1:1). The amylase-mediated reconfiguration behaviors for other LC or isotropic coacervate microdroplets were investigated following similar procedures except for using different doses of amylase (10 mg/mL). Specifically, 3 µL of amylase (10 mg/mL) was added to Cm-Am/DTAB isotropic coacervate microdroplets (30 µL, total concentration of 50 mM, monomer molar ratio of 1:1), Cm-Am/MTAB isotropic microdroplets (30 µL, total concentration of 50 mM, monomer molar ratio of 1:1), or Cm-Am/DMAB LC coacervate droplets (30 µL, total concentration of 5 mM, monomer molar ratio of 1:1). One microliter of amylase (10 mg/mL) was added to Cm-dextrin/DDAB LC microdroplets (30 µL, total concentration of 10 mM, monomer molar ratio of 1:1). Two microliter of amylase (10 mg/mL) was added to Cm-Am/PAH isotropic microdroplets (30 µL, total concentration of 10 mM, monomer molar ratio of 2:1).

The effects of PBS buffer on LC coacervates were also investigated by adding 5.3 µL of PBS buffer (10 mM, pH 6.9, containing 1 mM CaCl2) to 30 µL of Cm-Am/DDAB LC coacervates (total concentration: 10 mM; monomer molar ratio: 1:1). All experiments were independently repeated at least three times, yielding consistent results.

Preparation of H-Cm-Am/DDAB LC coacervate microdroplets or coacervate vesicles

Typically, 300 μL of Cm-Am solution (10 mM) was priorly hydrolyzed by amylase (100 μL, 10 mg/mL in 10 mM PBS buffer containing 1 mM CaCl2, pH 6.9) for 2 min, after which the mixture was heated under 90 °C for 40 min to denature amylase. Then, the mixture was centrifugated, discarding the precipitation of amylase. Afterwards, H-Cm-Am 1 were obtained. The H-Cm-Am 1/DDAB LC coacervate microdroplets were prepared by mixing the aforementioned H-Cm-Am 1 (10 mM) and DDAB solution (10 mM) in a monomer molar ratio of 1:1. Likewise, H-Cm-Am 2/DDAB isotropic coacervate vesicles (total concentration of 10 mM, monomer molar ratio of 1:1) were prepared following the above procedure except the used H-Cm-Am 2 (10 mM) was obtained after prior hydrolysis of Cm-Am (10 mM) for 12 h.

Transformation from LC coacervate microdroplets to isotropic coacervate vesicles induced by sodium chloride or hydrochloric acid

Typically, 30 µL of Cm-Am/DDAB LC coacervate suspension prepared as the above procedure was added to a sample cell with a glass coverslip, settling for ca. 10 min. Then, 3 µL of sodium chloride with a concentration of 800 mM was added, and the reconfiguration process was recorded by CLSM. Similarly, the dynamic behaviors of Cm-Am/DDAB LC coacervate microdroplets with adding 3 µL of 150 mM or 1500 mM of sodium chloride were performed following the above procedures.

The hydrochloric acid-mediated dynamic behaviors were achieved following the above procedures except that 3 µL of hydrochloric acid solution with a concentration of 10, 100, and 500 mM was added, respectively.

The dynamic behaviors of LC coacervates initiated by mixture of amylase and high concentrations of sodium chloride or hydrochloric acid were achieved following the above procedures except the addition of 5 µL of amylase solution (10 mg/mL) with containing sodium chloride (955 mM) or hydrochloric acid (355 mM). All experiments were repeated at least three times independently with similar results.

Diagram of LC coacervate microdroplets with the addition of sodium chloride or hydrochloric acid

Typically, 30 µL of Cm-Am/DDAB LC coacervate suspension prepared as the above procedure was added to a sample cell with a glass coverslip, settling for ca. 10 min. Then, different doses of sodium chloride (1500 mM) were added into the LC coacervates, respectively, and the changes of LC coacervates were monitored by CLSM. When the assemblies almost remained unchanged, we set a period of 30 min for salt concentration equilibration and then the ultimate structures were recorded. Similarly, the effects of different concentrations of hydrochloric acid on LC coacervates were also explored following the above procedures except using varied doses of hydrochloric acid (500 mM).

Reconfiguration of LC coacervate microdroplets integrated with biomacromolecules into organelle-containing coacervate vesicles

Typically, 5 µL of RITC-BSA (3 mg/mL, in 10 mM of PBS buffer, pH 8) was added to 50 µL of Cm-Am/DDAB LC coacervate suspension (total concentration of 10 mM, monomer molar ratio of 1:1, pH 8), followed by a 10-min incubation to establish partitioning equilibrium. The structural change of the sample was recorded by CLSM after adding 5.3 µL of amylase (10 mg/mL).

Likewise, the structural changes of Cm-Am/DDAB LC coacervate microdroplets with the incorporation of RITC-HRP (5 µL, 5 mg/mL, in 10 mM of PBS buffer, pH 8), RITC-GOx (5 µL, 5 mg/mL, in 10 mM of PBS buffer, pH 8), TAMRA-ssDNA (99 nt, 2 µL, 25 μM, in 10 mM of Tris-HCl buffer with containing 1 mM of EDTA, pH 8), TAMRA-dsDNA (25 bp, 1 µL, 50 μM, in 10 mM of Tris-HCl buffer with containing 1 mM of EDTA, pH 8), and FITC-dextran (10 k Da, 5 µL, 5 mg/mL, in water), Cy5-HH-min (0.5 µL, 50 μM, in water), and FAM-substrate (0.5 µL, 50 μM, in water) was recorded following the above procedure. All experiments were independently repeated at least three times, yielding consistent results.

The ribozyme-mediated catalysis within LC coacervate microdroplets and organelle-containing coacervate vesicle

Typically, 0.5 µL of hammerhead ribozyme derived from satellite RNA of tobacco ringspot virus (HH-min, 50 µM) was added into 30 µL of Cm-Am/DDAB LC coacervate suspension (total concentration of 10 mM, monomer molar ratio of 1:1, pH 8). Thirty microliters of HH-min-loaded Cm-Am/DDAB LC coacervate suspension was added to the sample cell, followed by settling for ca. 10 min on the coverslip before imaging. After adding amylase (5.3 µL, 10 mg/mL) for the reconfiguration process from LC coacervate microdroplets to yolk-shell coacervate vesicles, 0.5 µL of FRET-substrate (20 µM) was added to initiate the HH-min-mediated cleavage of the FRET-substrate strand. The enhancement of green fluorescence intensity was monitored by CLSM, which was associated with the increase in the distance between FAM and Black Hole quencher 1 (BHQ1). In a control experiment, the same condition was provided as mentioned above except none of amylase was added into Cm-Am/DDAB LC microdroplets. The data were obtained from three independent experiments.

Statistics and reproducibility

No statistical method was used to predetermine sample size. No data were excluded from the analysis. The experiments were not randomized. The investigators were not blinded to allocation during experiments and outcome assessment. All representative CLSM images were obtained from three independent experiments with similar results. Statistical parameters including the definitions and exact value of n (e.g., total number of measured samples) were reported in the figures and corresponding figure legends. Error bars displayed on graphs represent the mean ± standard deviation (s.d.) from at least three independent experiments, as indicated in the corresponding figure legends. CLSM images were analyzed by ImageJ software.

Reporting summary

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