Emergent Motility of Self‐Organized Particle‐Giant Unilamellar Vesicle Assembly

TL;DR

Researchers created cell-like microrobots using lipid vesicles and particles that move and adapt under electric fields.

Contribution

A new method for creating motile, self-assembled microrobots using GUVs and silica particles under electric fields is introduced.

Findings

Asymmetric particle decoration on GUVs enables self-propulsion and reversible transformation into active structures.

Electric fields control motion and allow on-demand cargo delivery via vesicle bursting.

Dynamic phase diagrams map motion regimes based on field parameters and membrane tension.

Abstract

Giant unilamellar vesicles (GUVs), soft cell‐sized compartments formed through the self‐assembly of lipid molecules, have long been utilized as model systems and passive carriers in membrane biophysics and biomedical applications. However, their potential as dynamically responsive and motile systems remains largely untapped due to challenges in achieving controlled and sustained motion in soft, deformable structures. Here, an autonomous cell‐like microrobot through the emergent self‐assembly of GUVs (5‐10 µm) and silica microparticles (1‐3 µm) under alternating current electric fields is realized. Self‐propulsion arises from asymmetric self‐organization of the particles on the vesicle surface, enabling a reversible transformation of the assembly into an active structure. Unlike rigid colloidal systems, GUVs introduce unique features enabled by their soft lipid membranes: shape deformations, membrane tension‐dependent motility, and field‐triggered live bacteria release via vesicle bursting. Through experiments and simulations, the mechanisms underlying self‐assembly and propulsion are investigated, and a dynamic phase diagram is constructed to map the motion regime as a function of field parameters. Finally, it is shown that these self‐assembled structures are capable of reconfiguration in response to local constraints in the environment, suggesting potential applications in complex environments and advancing the potential of GUVs toward the rational design of cell‐like microrobots or artificial cell systems.

Giant unilamellar vesicles (GUVs), when combined with silica particles under alternating electric fields, spontaneously self‐assemble into motile structures. Asymmetric particle decoration induces fluid flows that propel the assemblies, enabling persistent motion and reversible control. Motion can be tuned by membrane tension and field parameters, and on‐demand cargo delivery is demonstrated, advancing GUVs toward adaptive, cell‐like microrobots and artificial cells.

Figures

Experimental setup and the schematic of the formation of GUV‐particle assembly. A) Microscopic image of a GUV and lipid membrane structure. B) Schematic illustration of the experimental setup and the application of AC electric fields across ITO‐coated glasses. C) Schematic illustration of the system components: i) Bare GUVs and ii) bare particles exhibit no motion individually, while iii) GUV‐particle assemblies spontaneously form under an electric field, resulting in persistent motion. D) Time‐sequence images showing the process of assembly formation and motion in experiments. Initially, GUVs and particles exhibit random Brownian motion (t = 0 s). Upon application of the field (EON), particles attach to the GUV surface (t = 0.80 s) and form an asymmetric decoration (t = 1.63 s), leading to GUV motion (t = 2.80 s) Red line in (D) represents the trajectory of GUV.

Simulations and experimental results on GUV‐particle assembly to study particle decoration. A) Finite‐element analysis of the spatial distribution of the normalized squared electric field gradient around the GUV. The heat map represents the magnitude of the gradient of the normalized squared electric field. Arrows indicate the direction of the gradient, with lengths scaled logarithmically (see methods for details). B) Schematic of DEP forces acting on a particle near the vesicle. C) Plot of DEP forces as a function of the distance from the GUV surface along different arcs (inset: schematic illustrating the analyzed arcs around the GUV). Representative simulation images of the depletion region generated in COMSOL: showing D) top view and E) side view. F) Experimental data showing the depletion region from the top view. G) 3D reconstructed confocal images of the GUV, showing particles attached to the bottom of the vesicle. H) Snapshots of particle dynamics before (EOFF) and after (EON) field application, showing the formation of the depletion region experimentally.

Analysis of fluid flows around a bare vesicle. A,B) flows around vesicles, showing a divergent flow in the equatorial plane of the vesicle (red arrows), and a convergent flow in the top polar plane of the vesicle (blue arrows), respectively. C,D) Radial velocity profiles of the flow fields u(r) averaged over time in the equatorial and top plane, respectively. Insets show the location of the flows (side view). Velocity dependence on E) electric field amplitude E0, and F) frequency. Insets show that velocity scales quadratically with amplitude and inversely with frequency of the electric field, consistent with the ICEO mechanism. Error bars represent standard deviations (n = 7).

Role of unbalanced flows, attachment asymmetry, and electric field parameters in generating the vesicle motion. A) Snapshots of vesicle configurations and B) Time‐dependent velocity profile of a vesicle showing the motion stops after symmetric attachment. C) Eccentricity of the depletion region over time for one moving and one non‐moving vesicle. Snapshots show corresponding vesicle configurations. D) Schematic depiction of flow patterns and particle attachment. Symmetric particle attachment generates symmetric flows, resulting in no net motion, while asymmetric particle attachment induces asymmetric flows, leading to vesicle motion. The inset shows vesicles in the top view. The speed of the vesicle E) as a function of frequency, showing a linear decrease with 1/f, and F) as a function of the electric field amplitude E2 consistent with the ICEO mechanism. Error bars represent standard deviations (n = 8). G) Mean square displacement (MSD) analysis for vesicles at varying amplitude values (40, 50, 60 Vpp), demonstrating straight‐line motion, and increased displacement with stronger fields.

Phase behavior, bursting, and tunability of the motion of GUV‐particle assembly under AC electric fields. A) Phase map of the vesicle behavior (active motion, no motion, and bursting) as a function of AC electric field frequency and amplitude (peak‐to‐peak). The green region indicates active motion, the red region corresponds to vesicle bursting, and the blue region represents no motion. B) Demonstration of cargo release via vesicle bursting and the release of encapsulated particles (top) and encapsulated bacteria (bottom). C) Time‐dependent velocity profile of the vesicle with amplitude values alternating between 20 Vpp (low‐field) and 50 Vpp (high‐field). D) Snapshots of vesicle configurations for corresponding amplitudes.

The effect of particle size and membrane tension on the motion of GUV‐particle assembly under AC field. A, B) Experimental images of big vesicles motion with 3 µm particles. The red line represents the GUV trajectory. C) GUV speed as a function of membrane tension controlled by osmotic imbalance between inner (sucrose) and outer (glucose) solutions. Tensed membrane: 600 mM sucrose inside and 300 mM glucose outside. Relaxed: 600 mM sucrose inside and 600 mM glucose outside. Deflated: 600 mM sucrose inside and 1000 mM glucose outside. Schematics below each regime represent the corresponding membrane states. D) Fluorescence microscopy images showing the decoration of particles under an electric field. i, ii) represents tensed membrane state, iii, iv) represents relaxed state, and v, vi) represents deflated state. Lipid tubules are seen in the case of a deflated membrane state.

Adaptability and reconfiguration of the particle‐GUV assembly. A) Time‐lapse images of a moving vesicle while rearranging itself according to the particles on the substrate. B) The red line represents the trajectory of the vesicle. C) Time‐lapse images of a deformed GUV, navigating through a narrow gap between two particles acting as obstacles and marked with blue circles.

Full text

Introduction

1

Giant unilamellar vesicles (GUVs) are cell‐sized vesicles made from a thin bilayer of self‐assembled lipid molecules (Figure 1A).^[^ 1 ^]^ Their similarity to the structure of cell membranes makes them suitable model systems for membrane biophysics and artificial cell research, and their high biocompatibility and biodegradability make them useful for drug delivery and biomedical applications.^[^ 2, 3, 4 ^]^ They can encapsulate several types of therapeutic agents, including hydrophilic, hydrophobic, and lipophilic drugs, providing a protective effect from the metabolic processes.^[^ 5 ^]^ On the other hand, GUVs have emerged as promising candidates for developing cell‐sized microrobots because they do not possess the limitations in architectural designs and programmability that are faced by current synthetic and rigid‐body microrobots.^[^ 6, 7 ^]^ Owing to their soft and deformable nature, GUVs can undergo shape‐changing deformations like cells, enabling them to adapt to varying environments and complex 3D structures.^[^ 8 ^]^ To transform the GUVs into functional microrobots, it is crucial to design motile and active GUVs that can have dynamic interactions with their surroundings. Efforts to introduce motility into GUVs have led to a variety of theoretical and experimental studies that focused on integrating GUVs with active matter^[^ 9, 10, 11, 12 ^]^ and adopting the mechanisms used for synthetic microrobots, such as the use of electric fields,^[^ 13, 14 ^]^ light,^[^ 15, 16 ^]^ chemically driven,^[^ 17 ^]^ adhesion‐based,^[^ 18 ^]^ or ion exchange^[^ 19 ^]^ methods. These approaches often require complex modifications of GUVs that limit their functionality as microrobots.^[^ 7 ^]^ Despite many attempts, the persistent motion of GUVs at high speeds remains challenging due to the lack of an inherent propulsion mechanism and their sensitivity to the surrounding environment parameters.^[^ 20, 21, 22 ^]^

Experimental setup and the schematic of the formation of GUV‐particle assembly. A) Microscopic image of a GUV and lipid membrane structure. B) Schematic illustration of the experimental setup and the application of AC electric fields across ITO‐coated glasses. C) Schematic illustration of the system components: i) Bare GUVs and ii) bare particles exhibit no motion individually, while iii) GUV‐particle assemblies spontaneously form under an electric field, resulting in persistent motion. D) Time‐sequence images showing the process of assembly formation and motion in experiments. Initially, GUVs and particles exhibit random Brownian motion (t = 0 s). Upon application of the field (EON), particles attach to the GUV surface (t = 0.80 s) and form an asymmetric decoration (t = 1.63 s), leading to GUV motion (t = 2.80 s) Red line in (D) represents the trajectory of GUV.

A possible solution to the need for complex modifications of GUVs would be the use of non‐invasive external methods, such as the application of electric fields. Electric fields have long been established as versatile tools for cell manipulation,^[^ 23 ^]^ electroporation,^[^ 24 ^]^ tissue ablation,^[^ 25 ^]^ and cancer treatment.^[^ 26 ^]^ Similar to biological cells, GUVs also exhibit a rich variety of behaviors under electric fields, such as electroporation, fusion, and shape change.^[^ 27 ^]^ These behaviors suggest that electric fields could serve as a tunable and noninvasive mechanism for actuating soft vesicle systems. While field‐driven propulsion has been studied in rigid colloidal systems,^[^ 28, 29, 30 ^]^ the interplay between rigid and soft, membrane‐bound compartments such as GUVs under electric fields remains largely unexplored. In particular, how membrane deformability and surface fluidity influence flow generation, particle interactions, and motility in such systems is still poorly understood.

Herein, we introduce a self‐propelled cell‐like microrobot design that contains two passive structures, GUVs and SiO_2_ particles, and becomes active by forming a spontaneous assembly under an alternating current (AC) electric field. The resulting asymmetric GUV‐particle self‐assemblies exhibit a persistent self‐propelled motion (Figure 1). By combining experiments and numerical simulations, we investigate the mechanism responsible for the observed self‐assembly and self‐propulsion. We further demonstrate that the membrane deformability and tension of GUVs modulate propulsion efficiency, particle rearrangement, and dynamic adaptability of the assemblies. In addition, we show that the vesicles can encapsulate and release live bacteria via field‐induced bursting, providing a functional demonstration of triggered cargo delivery. Overall, this approach highlights a controllable, switchable propulsion mechanism for designing self‐propelled GUVs, paving the way toward realizing cell‐like microrobots or artificial cells, with potential applications in cargo transport and as an experimental tool for biophysiological studies on cell‐particle interactions.

Results and Discussions

2

Results

2.1

We study dynamic self‐assembly and motion of GUV and silica particle assemblies. A mixture of GUVs (5–10 µm diameter, see Figure S1, Supporting Information for more details) and spherical silica particles (1 µm diameter) was placed in a 200 µm high‐chamber, and an electric field ranging between 5–60 V_pp_ and 1–40 kHz was applied (Figure 1B). In the absence of an electric field, both the GUVs and the silica particles exhibit Brownian motion due to their passive nature. We did not observe any persistent motion in the system when only GUVs or only silica particles were present in the experimental chamber and the electric field was applied (Figure 1C).

However, when we mix GUVs and particles and switch on the electric field, we observe two key effects: i) the GUVs and the particles spontaneously form assemblies, and ii) the asymmetric assemblies of GUV and silica particles propel with a persistent straight‐line motion under the electric field. The formation process of an assembly can be seen in Figure 1D as a time evolution. After the field is switched on, the particles are attracted by the vesicle and stay attached to its surface (t = 0.8 s in Figure 1D). Consequently, the GUV attracts multiple particles, forming an asymmetric decoration of particles around it, and the GUV‐particle assembly propels with the undecorated part of the GUV facing the front (Movie S1, Supporting Information). The assembly formation persists even in a crowded environment with higher particle concentrations, where some particles are pushed away, whereas others attach to the vesicle surface and cause a net motion. When the electric field is switched off again, the motion ceases, and the silica particles slowly diffuse away from the GUVs due to thermal fluctuations.

We further investigate the underlying mechanisms of GUV‐particle self‐assembly and the fundamental principles driving their emergent propulsion. We hypothesize that the decoration of the vesicle surface by particles and the subsequent formation of the GUV‐particle assembly are driven by dielectrophoretic (DEP) interaction between GUVs and particles. When polarizable dielectric objects, in our case GUVs, are subjected to a uniform electric field, they distort the surrounding electric field, leading to the emergence of a nonuniform electric field in the vicinity of the GUVs.^[^ 31 ^]^ As a result, a low electric field region forms at the poles of the vesicles while high electric field regions develop around the equator (Figure S3A, Supporting Information).^[^ 32, 33 ^]^ We quantify and evaluate this effect using finite element analysis simulations (via COMSOL Multiphysics software) (see Figure S2, Supporting Information). First, we simulate an isolated GUV in the presence of a uniform electric field (frequency = 10 kHz) to assess the modulation of the electric field around the GUV. Because of the spatial gradient and modulation of the electric field around the vesicle, a nearby dielectric particle experiences the DEP force. The DEP force on the particles is attractive toward the low‐field region at the poles and repulsive in the high‐field region at the equator of the vesicle (Figure 2A,B). The DEP force on a particle can be expressed as F_DEP_ = 2πα^3^ɛ_0_ɛ_m_Re [CM]∇|E^2^| where α is the dielectric particle radius, ɛ_0_ is the vacuum permittivity, ɛ_m_ is the relative permittivity of the medium, Re[CM] is the real part of the Clausius‐Mossotti factor, ∇ is the gradient operator, and E is the electric field localized at the particle's location.^[^ 34 ^]^ This force decays as a function of distance from the vesicle and increases with the increasing external electric field strength (Figure S3B,C, Supporting Information).

Simulations and experimental results on GUV‐particle assembly to study particle decoration. A) Finite‐element analysis of the spatial distribution of the normalized squared electric field gradient around the GUV. The heat map represents the magnitude of the gradient of the normalized squared electric field. Arrows indicate the direction of the gradient, with lengths scaled logarithmically (see methods for details). B) Schematic of DEP forces acting on a particle near the vesicle. C) Plot of DEP forces as a function of the distance from the GUV surface along different arcs (inset: schematic illustrating the analyzed arcs around the GUV). Representative simulation images of the depletion region generated in COMSOL: showing D) top view and E) side view. F) Experimental data showing the depletion region from the top view. G) 3D reconstructed confocal images of the GUV, showing particles attached to the bottom of the vesicle. H) Snapshots of particle dynamics before (EOFF) and after (EON) field application, showing the formation of the depletion region experimentally.

To further study how silica particles behave near the vesicle, we perform experiments and simulations where a GUV is placed in a concentrated solution of silica particles exhibiting Brownian motion. Both in experiments and the simulations, the silica particles spontaneously rearrange around the vesicle under the electric field, forming an empty region in the immediate vicinity of the vesicle (henceforth referred to as the depletion region) while some particles get accumulated at the vesicle's periphery (Figure 2D–H; Figure S4, and Movies S2 and S3, Supporting Information). Importantly, both the simulations and the confocal microscope images reveal that the silica particles settle to the bottom of the chamber (due to their higher mass density) and the formation of the depletion region and the decoration of the vesicles both occur on a ≈5 µm‐thick latitudinal plane passing through the bottom pole of the vesicle. This agreement between simulations and experiment confirms the validity of our description of the DEP‐driven attachment mechanism. By varying the size of the GUV and the particles in the simulations (see Methods for details), we further observe that the radius of the depletion region increases with both GUV size and particle size (Figure S5A–D, Supporting Information).

We study the decoration of the particles close to the vesicles by performing experiments where we vary the concentration of the silica particles while the GUV size is almost the same (Figure S5E, Supporting Information). At low particle concentrations (≈0.1 mg mL^−1^), the particles and GUV form an asymmetric assembly, and at higher concentrations (0.3 mg mL^−1^) the particles fully decorate the vesicle, forming a symmetric assembly. At even higher concentrations (0.5 mg mL^−1^) the particles form multiple layers around the GUV, with the outermost particles bouncing on the vesicle.

While DEP forces drive the attachment of the silica particles around the vesicle, the motion of the assemblies cannot be explained solely on the basis of DEP forces because a net force on the whole assembly is required for its motion. In addition to DEP, the application of an electric field leads to the generation of fluid flows around GUVs.^[^ 27 ^]^ The applied field induces polarization of GUVs, where the positive charge cloud in the solution accumulates on one side of the vesicle and the negative charge cloud moves to the opposite side (see Figure S6, Supporting Information). The tangential component of the electric field exerts a force on the induced charges in the solution, resulting in an electro‐osmotic slip flow from the pole of the vesicles toward the equator. This mechanism is termed as induced‐charge electro‐osmosis (ICEO)^[^ 35 ^]^ and the ICEO flows vary with the square of electric field amplitude (v_ICEO_ ∝ E_0_ ^2^). Unlike solid particles, GUV is composed of a lipid membrane that behaves as a 2D incompressible fluid. Under mechanical stress, the membrane tension resists change to its surface area, but this balance is disrupted even in weakly inhomogeneous electric fields (caused by the nearby glass substrate), causing lipid flow in the membrane (due to a gradient in membrane tension) to restore equilibrium.^[^ 1, 36 ^]^ These membrane flows, organized into quadrants, are coupled with the internal and external fluids. Given these interactions, we consider and explore the role of these fluidic flows in generating and driving the motion of particle‐vesicle assemblies.

We analyze the fluid flows outside the bare vesicles using particle imaging velocimetry (PIV) analysis (see methods for details). Figure 3A,B and Movie S4 (Supporting Information) show a representative example of flows outside a large undecorated vesicle (≈30 µm in diameter) under a 7 kHz AC field at 50 V_pp_. The flows are divergent at the equator, while they converge near the pole of the vesicle. The time‐averaged radial component of flow, u(r), confirmed the divergent (u(r) >0) and convergent (u(r) <0) flow regions (Figure 3C,D). Moreover, the flow velocity varies with field strength, demonstrating a quadratic dependence (u(r) ∝ E^2^) and varies inversely with field frequency (u(r) ∝ f^−1^) consistent with the ICEO mechanism (Figure 3E,F).^[^ 35, 37, 38 ^]^ These results further support that the electric field induces charge redistribution on the vesicle's polarizable membrane, driving the observed fluid flows.

Analysis of fluid flows around a bare vesicle. A,B) flows around vesicles, showing a divergent flow in the equatorial plane of the vesicle (red arrows), and a convergent flow in the top polar plane of the vesicle (blue arrows), respectively. C,D) Radial velocity profiles of the flow fields u(r) averaged over time in the equatorial and top plane, respectively. Insets show the location of the flows (side view). Velocity dependence on E) electric field amplitude E0, and F) frequency. Insets show that velocity scales quadratically with amplitude and inversely with frequency of the electric field, consistent with the ICEO mechanism. Error bars represent standard deviations (n = 7).

The flows around a bare vesicle are radially symmetric because of the symmetric shape of the vesicle and therefore do not cause any persistent motion of the vesicle. However, when the nearby silica particles attach to the vesicles via DEP, the radial symmetry in the flows is broken, leading to unbalanced flows around the vesicle that cause a net propulsion of the GUV‐particle assembly. To investigate the role of asymmetric particle decoration in vesicle motion, we conducted experiments demonstrating that vesicles begin attracting the particles upon the application of the electric field and start moving (Movie S5, Supporting Information). Over time, the vesicle accumulates enough particles to achieve symmetric coverage, at which point the flows around the symmetric assembly become balanced and the motion ceases (Figure 4A). This behavior is evident in the velocity deviation over time (Figure 4B). The asymmetric flows around moving vesicles can be further visualized by measuring the shape (eccentricity) of the depletion region around the vesicles. We analyzed two cases: one vesicle that moves due to asymmetric particle attachment and another that remains stationary due to symmetric attachment (Figure 4C). For the moving vesicle, the formed depletion region has an oval shape with eccentricity ≈0.5, and the depletion region around the stationary vesicle is almost circular with eccentricity ≈0.1. (Movie S6, Supporting Information). This behavior is attributed to the flow patterns: asymmetric ICEO flows around an asymmetrically decorated vesicle drive its motion, while symmetric flows around a symmetrically decorated vesicle do not (Figure 4D). The oval depletion region is caused by the asymmetric flows around a moving vesicle, and this observation is also supported by the simulations, because the DEP forces due to a stationary, asymmetrically decorated GUV also produce a circular depletion region (Figure S8, Supporting Information). However, the attached particles locally distort the electric field at the vesicle surface. This distortion alters the nearby field gradients and creates an imbalance in the induced flows, providing a direct physical mechanism by which asymmetric particle decoration drives propulsion. Taken together, our observations show that the persistent motion of particle–GUV assemblies arises from the combined action of DEP forces and ICEO flows: while DEP forces keep the particles bound to the vesicle, their asymmetric attachment continuously breaks the flow symmetry and sustains vesicle motion.

Role of unbalanced flows, attachment asymmetry, and electric field parameters in generating the vesicle motion. A) Snapshots of vesicle configurations and B) Time‐dependent velocity profile of a vesicle showing the motion stops after symmetric attachment. C) Eccentricity of the depletion region over time for one moving and one non‐moving vesicle. Snapshots show corresponding vesicle configurations. D) Schematic depiction of flow patterns and particle attachment. Symmetric particle attachment generates symmetric flows, resulting in no net motion, while asymmetric particle attachment induces asymmetric flows, leading to vesicle motion. The inset shows vesicles in the top view. The speed of the vesicle E) as a function of frequency, showing a linear decrease with 1/f, and F) as a function of the electric field amplitude E2 consistent with the ICEO mechanism. Error bars represent standard deviations (n = 8). G) Mean square displacement (MSD) analysis for vesicles at varying amplitude values (40, 50, 60 Vpp), demonstrating straight‐line motion, and increased displacement with stronger fields.

To further confirm that the observed motion is driven by the ICEO mechanism, we characterized the vesicle velocities at different frequencies and amplitudes of electric field (Figure 4E,F). The vesicles exhibited high velocities, reaching ≈8 µm s^−1^. The linear dependence of velocity on f^−1^ and E^2^ confirmed that the propulsion mechanism is consistent with the ICEO mechanism, as suggested by previous results.^[^ 35 ^]^ Additionally, we analyzed the mean square displacement (MSD) of vesicles subjected to varying amplitudes (40, 50, and 60 V_pp_) as shown in Figure 4G. Straight‐line trajectories, indicative of directed motion, were observed.

We constructed a phase diagram to understand and demonstrate the vesicle behavior in different field parameters (Figure 5A). We observed three distinct regions: active motion region represented by the green area, no motion region represented by the blue area in high frequency and low amplitude values, and vesicle bursting region in high amplitude and low‐frequency represented by the red area. Active motion (green area) occurs at lower frequencies and higher amplitude values because the ICEO flows v_ICEO_ depend on the electric field parameters as v_ICEO_ ∝ E^2^/f. At higher frequencies and lower amplitudes (blue area), the flows become weaker and thus no net motion was observed.

Phase behavior, bursting, and tunability of the motion of GUV‐particle assembly under AC electric fields. A) Phase map of the vesicle behavior (active motion, no motion, and bursting) as a function of AC electric field frequency and amplitude (peak‐to‐peak). The green region indicates active motion, the red region corresponds to vesicle bursting, and the blue region represents no motion. B) Demonstration of cargo release via vesicle bursting and the release of encapsulated particles (top) and encapsulated bacteria (bottom). C) Time‐dependent velocity profile of the vesicle with amplitude values alternating between 20 Vpp (low‐field) and 50 Vpp (high‐field). D) Snapshots of vesicle configurations for corresponding amplitudes.

Vesicle bursting (red area) occurs even in bare vesicles in the absence of silica particles, and it is a result of the unbalanced electric stresses on the membrane. During the application of the electric field, the shear stress on the membrane increases and causes tension on the membrane and deformation (see Note S2, Supporting Information for more details). If the tension is greater than a critical value, which is reported as 10 mN m^−1^ previously, temporary or irreversible membrane rupture can be observed.^[^ 39, 40 ^]^ We observe a bursting behavior usually in the frequency values lower than 5 kHz where electric stress cannot be balanced by the membrane tension anymore, causing membrane rupture. The bursting of vesicles can be utilized for on‐demand cargo delivery applications with GUVs, where the release of encapsulated materials is triggered by an external electric field. To demonstrate the potential application, we have encapsulated passive particles (500 nm in diameter) inside the GUVs, and burst the vesicle to release them by lowering the frequency to 5 kHz (Figure 5B top, Movie S7, Supporting Information). We also demonstrate the release of encapsulated bacteria by bursting the GUV, and the bacteria remain motile after bursting the GUV and being released in the bulk aqueous media (Figure 5B bottom).

The electric field‐driven motion highlights potential future applications of GUVs in externally controlled microrobots, such as targeted drug delivery, adaptive soft robotics, electrically driven shape change, and controlled cargo uptake/release.^[^ 41 ^]^ To further explore the tunability of motion in our system, we alternated the amplitude of the AC electric field between 20 V_pp_ (insufficient DEP forces and ICEO flows to induce particle decoration and motion, respectively) and 50 V_pp_ (where vesicle and particle assembly moves), in Figure 5C,D. The vesicle velocity increased and decreased correspondingly. In the 20 V_pp_ regime, the DEP forces were too weak to hold particles tightly, causing them to oscillate around the vesicle. Movie S8 (Supporting Information) further illustrates how particle arrangement on the vesicle surface changes dynamically with amplitude variations.

After characterizing the effect of electric field parameters on the observed emergent motion of the vesicles, we now study how the GUV and silica particle size and membrane tension affect the motion of the GUVs. So far, we observed the active motion of the vesicles up to 10‐12 µm in diameter. Bigger vesicles do not move with 1 µm particles, as the particles fail to create sufficient asymmetry in the ICEO flows to generate the motion of bigger vesicles. Because the larger vesicles are much heavier compared to the smaller ones, they require stronger unbalanced flows to be able to move. One way to generate stronger asymmetry is to use larger particles. Larger dielectric particles can generate stronger asymmetry in the fluid flows due to their size and contribute to the net propulsion of GUVs as seen for 3 µm particles in Figure 6A,B and Movie S9 (Supporting Information).^[^ 42 ^]^ However, this motion is not as fast as in the case of smaller vesicles. The drag force on the vesicle scales with its size, and larger vesicles experience stronger resistance from the surrounding fluid. In turn, symmetry can be broken, but a weak motion can be obtained.

The effect of particle size and membrane tension on the motion of GUV‐particle assembly under AC field. A, B) Experimental images of big vesicles motion with 3 µm particles. The red line represents the GUV trajectory. C) GUV speed as a function of membrane tension controlled by osmotic imbalance between inner (sucrose) and outer (glucose) solutions. Tensed membrane: 600 mM sucrose inside and 300 mM glucose outside. Relaxed: 600 mM sucrose inside and 600 mM glucose outside. Deflated: 600 mM sucrose inside and 1000 mM glucose outside. Schematics below each regime represent the corresponding membrane states. D) Fluorescence microscopy images showing the decoration of particles under an electric field. i, ii) represents tensed membrane state, iii, iv) represents relaxed state, and v, vi) represents deflated state. Lipid tubules are seen in the case of a deflated membrane state.

The membrane tension of the GUVs can be influenced by varying the osmotic balance between the internal and external solutions, specifically by varying the concentration of sucrose inside the GUVs and glucose outside. This adjustment allows control over the vesicle's membrane tension from tensed (hyperosmotic inside) to relaxed (isotonic) and deflated (hyperosmotic outside). We investigated how membrane tension affects motility by comparing the propulsion speeds of GUVs across these different tension states. Our results show that the propulsion speeds of the GUVs did not differ significantly between the GUVs with a tensed and a relaxed membrane. For GUVs with deflated membranes, the speeds were reduced roughly to half of the speed of the GUVs with tensed membranes (Figure 6C,D; Movie S10, Supporting Information). Interestingly, the deflated GUVs still moved at ≈2 µm s^−1^ despite being highly deformable and containing flexible tubules within them.

Finally, we would like to highlight the advantages of using a self‐assembled and adaptive particle‐vesicle system. As shown in Figure 7A,B and Movie S11 (Supporting Information), the assembly maintains its motion even when some particles become trapped at the bottom of the experimental setup–the particles autonomously rearrange around the vesicle periphery, consistently adapting the assembly dynamics to the environmental conditions. Figure 7C highlights similar adaptation dynamics of an assembly containing a deflated GUV in a similar environment where some silica particles get trapped at the bottom of the experimental setup. The GUV deforms and squeezes through the gap between the trapped particles to adapt to its environment and continue its motion. (Movie S12, Supporting Information) This adaptability arises from the dynamic interactions between the particles and the vesicle, and the deformable nature of the vesicles, enabling the system to redistribute its components and sustain movement without significant disruption.

Adaptability and reconfiguration of the particle‐GUV assembly. A) Time‐lapse images of a moving vesicle while rearranging itself according to the particles on the substrate. B) The red line represents the trajectory of the vesicle. C) Time‐lapse images of a deformed GUV, navigating through a narrow gap between two particles acting as obstacles and marked with blue circles.

Discussion

2.2

Here, we present a novel and autonomous self‐propulsion mechanism for GUV‐particle assemblies under AC fields. Our findings demonstrate that asymmetric particle decoration via DEP forces can induce asymmetry in the ICEO flows of the GUV and enable vesicle motion. These results show that two passive structures, GUVs and silica particles, can spontaneously form self‐assemblies and transform into an active system, offering a new strategy for the rational design of cell‐like microrobots. Achieving such a persistent motion of GUVs has always been a challenge due to their soft and flexible structure, which limits their responsiveness to external stimuli. Previous studies have shown that membrane modifications can be used to induce vesicle motility. For example, incorporating nanoparticles into the GUV membrane^[^ 14 ^]^ can generate ICEO flows and drive propulsion, providing an approach on how nanoparticle‐membrane interactions can lead to motion, or using phase‐separating lipids to induce asymmetry.^[^ 13 ^]^ Our system achieves a persistent motion at high speeds (up to 12 µm s^−1^), without any chemical modification or fuel requirement. This makes our design broadly applicable to any GUV or particle type under correct field parameters that provide sufficient DEP forces and asymmetric ICEO flows. Since both particle attachment and motion arise from purely physical interactions, the system remains fully reversible and switchable, allowing for dynamic control over motion.

Despite the advantages of our system, some parts remain to be explored since this is the first proof‐of‐concept study on the motion of GUVs. For simplicity, we did not explicitly consider the role of membrane deformations in generating particle decoration and self‐propulsion. For example, particles attached to a vesicle tend to rearrange around the vesicle periphery to form an asymmetric decoration of the vesicle (Movie S1, Supporting Information). Such rearrangements could be a result of membrane‐mediated interactions between the attached particles.^[^ 43, 44 ^]^ If these deformations were better understood, a more controlled and precise system could be designed and engineered. Since it is a very dynamic and switchable strategy, the concentration of the particles around vesicles determines the formation of GUV‐particle assembly and the motion behavior.

Although the mechanism for the generation of the fluid flows around the GUVs may appear similar to that for a solid dielectric particle, a closer look at the flows, especially close to the vesicle, reveals clear differences between the two. The radial component of the flows around the vesicle decays with the square of the distance from the GUV (u_r_∼r^−2^), which is much slower than the r^−4^ decay of the flows expected for a solid dielectric particle near an electrode.^[^ 28, 29, 45 ^]^ While the flow profile far from the GUV in our experiments is similar to the theoretically predicted values for the ICEO mechanism for a spherical object,^[^ 35 ^]^ close to the GUVs, there are clear differences between the ICEO and the experimentally observed flows (Figure S7, Supporting Information). These differences likely arise from the soft and deformable structure of the GUV membrane, which can sustain additional interfacial stresses. In particular, Marangoni‐type flows generated by surface tension gradients along the membrane represent a plausible contribution.^[^ 46 ^]^ Similar ∼r^−2^ far‐field decays have been described for droplets with interfacial tension gradients, and strong field‐induced membrane flows have been observed in multicomponent GUVs, supporting the idea that vesicle membranes can exhibit Marangoni‐like stresses under electric driving.^[^ 47, 48 ^]^ Notably, Marangoni‐driven propulsion of droplets has been linked to asymmetric micelle adsorption at the interface, which is analogous to our case, where asymmetric adsorption of particles on the GUV membrane generates interfacial stresses that bias the flow field and result in motion.^[^ 47 ^]^ Based on these observations, we hypothesize that the flows in our system result from a combined action of ICEO and Marangoni effects, where ICEO provides the primary driving mechanism and Marangoni stresses arising from the membrane modulate its decay and overall characteristics.

Since Marangoni contributions are inherently linked to the mechanical state of the lipid bilayer, our experiments on GUVs with varying membrane tension revealed that tensed GUVs propel faster than deflated vesicles containing tubules. In deflated vesicles, excess membrane area is stored in undulations and nanotubules. Such a surplus area dissipates applied stresses into shape rearrangements rather than sustaining tension across the main vesicle body. This redistribution of energy weakens surface‐tension gradients, thereby diminishing the Marangoni contribution to the ICEO‐driven flows and leading to slower propulsion. It would be interesting to investigate in detail the effect of membrane tension and deformation (including tubules) of the GUV in the formation of the assemblies and their self‐propulsion dynamics. The DEP forces lead to the formation of the depletion region and the GUV‐particle assemblies via the attraction at short distances and repulsion at long distances between the silica particles and GUVs. For bigger vesicles surrounded by smaller silica particles and for higher concentrations of silica particles near smaller GUVs, we also observed the particles moving in periodic orbits around the vesicles, appearing to repeatedly bounce on the membrane of the GUVs (see Movie S13, Supporting Information). This periodic bouncing of the particles has not been seen previously and could be caused due to the deformable nature of the GUVs.

The design of GUV‐based microrobots can be further advanced with many strategies, as we only present a fundamental tool to generate vesicle motion. Currently, the silica particles in our system are passive, but additional functionalities could be integrated. Active particles or magnetic particles could replace the passive particles to enable the steering and directed motion of the vesicles. At the present stage, the propulsion relies on stochastic particle adsorption, which gives rise to variable particle arrangements and therefore variable motion directions. While this does not prevent robust switching of motion with field parameters, it indicates that achieving deterministic control over directionality remains an important next step. Importantly, however, the propulsion speed can be tuned reproducibly by adjusting the applied field strength and frequency for a given concentration of silica particles around, demonstrating that certain aspects of the motility, particularly speed and on/off behavior, are controllable in our system. A more detailed understanding of vesicle‐particle interactions could be achieved through advanced numerical simulations, as DEP forces are well‐studied for rigid structures but remain less understood for soft, deformable bodies. Additionally, GUVs naturally encapsulate and transport cargo due to their large size and hollow interior, making them promising candidates for multifunctional microrobotics systems. When combined with silica particles, this creates a versatile platform for simultaneous motion and cargo transport, offering potential applications in targeted drug delivery and microscale transport technologies.

GUVs, with their high flexibility and adaptability, offer a unique platform for studying both fundamental physics and biomedical applications. Their ability to undergo shape deformations, dynamically rearrange surface‐bound particles, and respond to external forces makes them ideal for soft‐matter physics, membrane biophysics, and microrobotics systems. Unlike rigid synthetic microrobots, their soft and reconfigurable nature allows them to navigate in complex microenvironments, mimicking biological processes and enabling applications in drug delivery, targeted transport, and synthetic cell engineering. The ability to precisely control their motion using external electric fields further enhances their potential for biomedical technologies and lab‐on‐a‐chip systems, providing a versatile tool for future research in physics, synthetic biology, bioengineering, and medicine.

Experimental Section

3

Materials

All chemicals were used as received. Lipids 1,2‐dioleoyl‐sn‐glycero‐3‐phosphocholine (DOPC) and the fluorescent 1,2‐dioleoyl‐sn‐glycero‐3‐ phosphoethanolamine‐N‐(lissamine rhodamine B sulfonyl) (ammonium salt) (Liss Rhod PE) dissolved in chloroform were obtained from Avanti Polar Lipids (Alabaster, AL). Sucrose (≥99.5%), glucose (≥99.5%), chloroform (≥99.5%), paraffin oil, 0.5 µm 5% silica particles in water, LB Broth with agar‐Lennox, ampicillin, kanamycin, NaCl, biotin and L‐(+)‐Arabinosewere purchased from Sigma‐Aldrich. Tryptone was purchased from Gibco. 1 and 3 µm silica plain white particles, and 0.5 µm fluorescent polystyrene (PS) particles (PS‐FluoRed) were purchased from Microparticles. ITO slides with surface resistivity of 70–100 Ω sq^−1^ were purchased from Merck.

Methods—Vesicle Formation—Electroformation Method

GUVs were produced by the electroformation method using VesiclePrepPro device (Nanion Technologies, Germany). The 80 µL of 4 mg mL^−1^ lipid mixture, including 99% DOPC and 1% Liss Rhod PE in chloroform, was spread into the conductive side of an indium tin oxide (ITO) coated glass using a spin coater to ensure a homogenous distribution. The coated ITO slide was then vacuumed for at least 2 h to evaporate the remaining chloroform. After placing a rubber ring on the lipid‐coated area, it is filled with 250 µL of 600 mM sucrose solution. To prevent leakage from the ring, the outside of the ring was sealed with vacuum grease. The chamber was sandwiched by placing the second ITO on top and putting it into the device. Protocols provided by VesiclePrepPro software were used. A low‐voltage procedure was used in experiments involving 3 µm silica particles and in visualizing flow around vesicles to produce bigger vesicles. After the procedure, the produced vesicles were transferred to an Eppendorf tube and diluted eightfold with 600 mM glucose to make the vesicles sediment down to the bottom of the tube due to the higher density of sucrose within GUVs.

For the membrane deflation experiments, the external osmolarity is gradually changed by stepwise replacement of the surrounding solution. GUVs were initially prepared in 600 mM sucrose and transferred into 300, 600, and 1000 mM glucose solutions to obtain tensed, relaxed, and deflated membranes, respectively. Subsequently, the external solution was incrementally added in three steps with appropriate glucose solutions.

Vesicle Formation—Emulsion Transfer Method

Vesicles were produced by the emulsion transfer method, and the modified protocol of Vutukuri et al.^[^ 10 ^]^ was used. Lipids in oil solutions (LOS) were prepared by adding DOPC and Liss Rhod PE to a glass vial and gently evaporated using N_2_ gas flow. Then, the vials were put into a vacuum desiccator for further drying of lipids. Next, paraffin oil was added to the dried lipid vial to have a final lipid concentration of 2 mg mL^−1^. Lipids were dissolved by sonicating for 1.5 h. The temperature was set to 55 °C to ensure the good solubility of the lipids and prevent aggregates.

The 600 mM sucrose and 600 mM glucose solutions were used as inner and outer solutions, respectively, to make the vesicles settle down to the bottom with the help of the density difference. The solutions were filtered with a 0.2 µm filter and freshly prepared before the experiments to prevent any contamination. The 50 µL sucrose solution and 200 µL LOS solution were added to a 1.5 mL test tube and vortexed at a moderate speed for 10 s to ensure the two phases turned into a turbid solution and droplets were formed by water‐in‐oil emulsion. We have also tried the hand‐tapping method to mix the two phases, but we found that vortex is the most efficient way to have clean and symmetric vesicles. In another 1.5 mL test tube, 600 µL of glucose and 200 µL of LOS were added to form an interface from the lipid monolayer. Then, 90 µL of the emulsion solution was layered to the interface. The tube was centrifuged at 1500 g for 5 min to transfer the droplets through the interface. To harvest the vesicle pellet, a 21‐gauge needle was inserted into the place where the pellet was from the outside, and a hole was made while the lid of the tube was open. The lid was closed after removing the needle, and the vesicle suspension dripping from the hole was collected into another test tube. The solution was diluted four times for the motion experiments.

Bacterial Strain and Culture

A genetically modified strain of Escherichia coli MG1655 was used in this study.^[^ 49 ^]^ Bacterial cells stored at −80 °C in a 50% glycerol stock were first streaked onto the Luria‐Bertani (LB) agar plates (15 g L^−1^ agar, 5 g L^−1^ NaCl, 40 g L^−1^ tryptone and 5 g L^−1^ yeast extract) supplemented with 100 µg mL^−1^ ampicillin and 50 µg mL^−1^ kanamycin as single colonies. For preculture, a single colony was inoculated into 5 mL tryptone broth (TB) medium (10 g L^−1^ tryptone and 5 g L^−1^ NaCl at pH 7.0) supplemented with 100 µg mL^−1^ ampicillin and 50 µg mL^−1^ kanamycin and incubated overnight at 37 °C and 200 rpm. The overnight culture was subsequently diluted 1:100 into 10 mL of fresh TB medium supplemented with biotin (1 µm,) and the antibiotics, and incubated at 34 °C and 270 rpm for 2 h. The green fluorescent protein (GFP) expression was induced by the addition of 0.005% (w/v) L‐(+)‐Arabinose, and incubation continued until the culture reached an optical density at 600 nm (OD600) of 0.2, measured using a microplate reader (TECAN Infinite M Plex). The bacterial suspension was then collected and centrifuged at 2000 g for 5 min to pellet the cells. After discarding half of the supernatant, the pellet was diluted twofold using 600 mM sucrose solution, which is the same as the inner medium used in GUV formation. This step ensured both osmotic compatibility and efficient encapsulation during vesicle formation. The estimated final bacterial concentration was ≈8 × 10⁷ cells per mL.

Vesicles Under the AC Field

Both of the fabrication methods resulted in GUVs of various sizes, ranging from 1 to 70 µm, with the majority measuring between 5 and 20 µm (Figure S1, Supporting Information). For the experiments, clean GUVs without lipid residuals in a size range of 5–10 µm were chosen.

For experiments, 150 µL of particle solution was mixed with 50 µL of GUV solution. We have used 0.1 mg mL^−1^ 1 µm and 0.2 mg mL^−1^ 3 µm SiO2 particle solutions for velocity experiments, and 1 mg mL^−1^ 0.5 µm PS particles for flow experiments to prevent the effect of DEP forces. A custom‐made experimental chamber was made to apply the AC field to vesicles. A spacer with a thickness of 200 µm was placed onto the conductive side of an ITO‐coated glass slide and filled with GUV and particle solution (≈10 µL). A second ITO slide was placed on top. Adhesive copper tapes were attached to ITO slides to connect the electrodes. A sinusoidal AC field was applied by a function generator and amplifier in the range of 5–60 V_PP_ and 1–40 kHz. Vesicles were characterized by both fluorescent and bright‐field microscopy imaging by Zeiss (Axio ObserverA1) and Nikon (Eclipse, Ti‐E) inverted optical microscopes. In addition, confocal imaging was performed using a Nikon Ti‐E spinning disk confocal microscope.

Liquid Conductivity Measurements

The ionic conductivity of the solutions was measured using electrical impedance spectroscopy. We utilized a three‐electrode system, where the ITO utilized in all experiments was the working electrode, an Ag/AgCl electrode and a platinum electrode was the reference and counter electrode, respectively.

Numerical Simulations

Numerical simulations were performed using COMSOL Multiphysics 6.2 (COMSOL, Inc.) to estimate the electric field strength, the gradient of the normalized squared electric field, and the behavior of dielectric particles in the vicinity of a giant unilamellar vesicle (GUV) under experimental conditions. The simulation geometry comprised a rectangular block representing the experimental chamber with indium tin oxide (ITO) electrodes on the sides, and a sphere representing the GUV. To emulate the lipid bilayer, a 10 nm shell was incorporated on the surface of the sphere.

Three materials were assigned in the model: i) the external solution in the chamber, ii) the internal solution of the GUV, and iii) the dielectric lipid membrane. The material properties, including relative permittivity, electrical conductivity, dynamic viscosity, and density, are detailed in Table S1 (Supporting Information), and the overall simulation environment is illustrated in Figure S2 (Supporting Information).

The Electric Currents module was used to compute the spatial distribution of the electric field around the GUV, applying current conservation across the entire simulation domain. All external boundaries were electrically insulated, except the ITO and GUV surfaces (Figure S2, Supporting Information). These results were used to calculate the electric field gradient and the resulting dielectrophoretic (DEP) force acting on nearby particles (see Note S1, Supporting Information).

The dynamics of dielectric particles were examined using the Creeping Flow and Particle Tracing for Fluid Flow modules. The Creeping Flow module assumed incompressible flow with no‐slip boundary conditions, neglecting inertial terms. Virtual silica particles were randomly distributed in the simulated fluid and released at the initial time step of the transient study. In the Particle Tracing module, the system was treated as Newtonian (with inertial effects neglected) and subject to freeze‐wall boundary conditions. Contributions from Brownian motion, drag, gravity, and DEP forces were incorporated to fully capture particle dynamics.

For spatial discretization, extra‐fine user‐defined meshing was applied to the top‐bottom electrode boundaries as well as GUV domains and boundaries with corner refinement, while free tetrahedral meshing was used for the remaining geometry levels. Output data, including electric field strength, field gradients, DEP forces, and the temporal and spatial distributions of dielectric particles around the GUV, were extracted for subsequent quantitative analyses.

Particle Tracking

A custom Python script and openCV libraries were used to analyze experimental videos to detect the vesicles and silica particles. The detected objects were tracked over multiple video frames using the trackpy library. Speeds, trajectories, and MSD were calculated from the detected positions in each frame.

PIV Analysis

Experiments were performed using 500 nm Polystyrene particles dispersed in aqueous solutions as tracers, and videos were recorded at 10 fps. Dynamic Studio (version 6.1) software was used to perform PIV analysis. Adaptive PIV analysis was applied with minimum and maximum IA size (in pixels) being 64 × 64 and 32 × 32, respectively, and Grid step size being 16 × 16. Then, a 7 × 7 average filter was applied to the results of the adaptive PIV. The final results were overlaid with the experimental images to visualize flows around a vesicle. A custom Python script was used to further analyze the results. Radial component of the flows was calculated and averaged over angle and time for different experiments where the amplitude and the frequency of the electric field were varied. The peak of the average radial flows was plotted for different amplitudes and frequencies of the electric field.

Statistical Analysis

The number of data points for each analysis is provided in the corresponding figure captions, where the sample size is indicated with n. All data are presented as mean ± standard deviation (SD), and error bars in the figures represent the standard deviation.

Conflict of Interest

The authors declare no conflict of interest.

Author Contributions

S.K. and G.G. contributed equally to this work. S.K., G.G., and M.S. conceived and proposed the research. S.K. and G.G. designed the study, performed the fabrication of GUVs, experimental analysis, and characterizations. M.H. performed simulation experiments and conductivity measurements. S.F.B. and M.B.A. performed the bacterial culture. M.S. supervised the research. S.K. and G.G. wrote the manuscript. All authors participated in the discussions and critically reviewed the manuscript.

Supporting information

Supporting Information

Supplemental Movie 1

Supplemental Movie 2

Supplemental Movie 3

Supplemental Movie 4

Supplemental Movie 5

Supplemental Movie 6

Supplemental Movie 7

Supplemental Movie 8

Supplemental Movie 9

Supplemental Movie 10

Supplemental Movie 11

Supplemental Movie 12

Supplemental Movie 13