Bottom‐Up Coacervate‐Based Artificial Cells: Integrating Cellular Hallmarks into Complex Life‐Like Systems

ABSTRACT

Living cells are remarkably sophisticated entities that form the basis of life. Bottom‐up artificial cell research focuses on reconstructing their essential functions and behaviors in life‐like compartments using synthetic and natural building blocks, which can advance our understanding of fundamental biological processes and drive technological and biomedical applications. In this review, we focus our discussion on recent developments in bottom‐up artificial cell research with a particular emphasis on coacervate‐based artificial cell systems that integrate multiple cellular hallmarks. We delineate how enhanced structural mimicry through organized compartmentalization affords improved control over function. We then examine how energy supply can be coupled to metabolic processes, growth, and adaptive responses in artificial cells. We also survey emerging systems that enable artificial cells to communicate with other artificial and living cells through responsive signaling and functional interactions. Finally, we present our vision on the opportunities and directions for artificial cell research moving forward.

Current interest in artificial cell research underscores its potential to deepen our understanding of life's fundamental processes. This review highlights advances in bottom‐up coacervate‐based artificial cell engineering via combined integration of cellular hallmarks. We discuss how structural mimicry enhances functional control, how energy fuels metabolism, growth and adaptivity, and how signaling systems enable communication with artificial and living cells.

1. Introduction

Cells represent the fundamental units of life, with each living organism either consisting of multiple cells or being a single cell itself. Through evolution, living cells have been endowed with an intricate and dynamic structural organization, enabling a wide range of complex functions that distinguishes them from dead matter. Understanding how a living cell operates has until recently been the sole domain of cell and structural biology. With our increased and detailed knowledge of the processes that govern cellular behavior, however, molecular scientists have become able to contribute to this field, by not only deducting how biomolecules interact to yield complex behavior, but also by recreating fundamental features of living cells. This emerging multidisciplinary field, known as artificial cell research, can be roughly divided into a top‐down and bottom‐up approach. The former, molecular biology‐based route, centers on modifying the genome of an existing living cell to elucidate the cellular features necessary for survival of a minimal cell. By contrast, the bottom‐up strategy represents a complementary approach in which biological and synthetic building blocks are assembled from scratch to construct an integrated system with cell‐like features and behaviors, following an understanding‐by‐building methodology. The design and construction of artificial cells as mimics of living cells helps in recapitulating and studying their signature functions, behaviors and structures in a well‐controlled life‐like environment. In addition to refining our fundamental understanding of the working principles of living cells, artificial cell research can also shed light on the origin of life on early Earth, and enable the development of biomimetic platforms with potential impact in application fields such as drug delivery, tissue engineering and biosensing [1, 2, 3].

For the field of bottom‐up artificial cell research it is important to identify which elements are needed to be included in the design to achieve life‐like behavior. Although many definitions are possible, in general five criteria can be identified that are associated with living cells, namely compartmentalization, energy supply & metabolism, growth, division & replication, communication, and adaptability [4]. Combining these features into a single all‐encompassing artificial cell system is extremely complex and establishing a whole functioning ‘living’ artificial cell is therefore still elusive. Nonetheless, recent years have seen significant advancements in all of these different aspects up to a point that the current state‐of‐the‐art in artificial cell research is directed to integrating multiple of these features. For example, compartmentalization has evolved into the design of life‐like organizations in which structure and function are intimately coupled, and communication has progressed from a mere exchange of molecules to functional responses and adaptive behavior.

In this review, we aim to highlight recent developments in bottom‐up artificial cell research with a focus on examples of this integrated approach. We first will describe how attention to structural mimicry has led to artificial cells with higher level of control over function. We subsequently illustrate that energy supply is integrated with metabolism, growth, and adaptive behavior. Finally, we delineate communicative processes that enable artificial cells to adapt to their environment. We predominantly limit the scope of this review to coacervate‐based artificial cells, but will also feature select examples of other artificial cell systems, such as vesicle‐based platforms, to highlight important developments in the field that have not yet been achieved using coacervate‐based artificial cells. For more comprehensive information on these other artificial cell systems, we refer the reader to a number of excellent reviews [5, 6, 7, 8, 9, 10, 11].

2. Function by Cellular Organization

2.1. Membrane‐Bound Versus Membraneless Compartments

One of the most defining characteristics of life is organization, which includes compartmentalization mediated by the cell membrane and membrane‐bound and membraneless organelles. Such organization spatially arranges functional cellular components and enables the integration and coordination of diverse cellular processes at the right place and time. Besides playing a structural role, compartmentalization fulfills several essential functions: it separates and protects molecules from harsh environments, segregates incompatible processes or components, establishes local environments, enables up‐concentration of molecules to enhance reaction efficiency, and maintains non‐equilibrium states that are crucial for cellular homeostasis and behavior.

In the context of artificial cell research, compartmentalization can be achieved using either membrane‐bound and membraneless chassis, and considerable efforts have been directed at constructing cell‐sized compartments via both approaches. Historically, vesicles with lipid bilayer membranes, such as phospholipid‐based giant unilamellar vesicles (GUVs) have been most prevalently used owing to their inherent structural resemblance to the living cell plasma membrane [7, 10]. Key advantages of such membrane‐bound systems include the ability to integrate proteins and other macromolecules into the membrane, harbor various multi‐compartmentalized architectures, and encapsulate protein synthesis machinery and delicate supramolecular systems into their aqueous interior [12, 13, 14, 15, 16, 17]. Compartments of this class can exhibit limited permeability, albeit this can be addressed by synthetic tuning or the integration of pore‐forming molecules to facilitate molecular exchange across the membrane [18, 19]. Beyond lipids, researchers have become increasingly skilled at mimicking the biological boundaries of cells using a wide array of alternative biological and synthetic materials, such as block co‐polymers (polymersomes), colloids (colloidosomes), proteins (proteinosomes), and hybrid combinations thereof, allowing tuning of physicochemical properties such as flexibility, amphiphilicity, and permeability (Figure 2A) [6, 20, 21, 22]. For example, polymersomes feature amphiphilic block‐copolymer‐based membrane bilayers and exhibit greater rigidity and stability than lipid vesicles at the cost of reduced permeability, making them suitable to provide long‐term protection of encapsulated molecules [6]. By contrast, proteinosomes are assembled from protein‐polymer conjugates and feature permeable membranes capable of supporting encapsulated biochemical reactions while allowing diffusive exchange of small molecules, proteins and nucleic acids with the surrounding environment [23, 24, 25, 26].

FIGURE 2.

(Sub‐)compartments as artificial cells and organelles. (A) Examples of membranized artificial cells. Microscopy images show artificial cells derived from (i) lipids (DOPC), (ii) natural cell membranes (HeLa cell membranes), and (iii) polymers (nylon). Adapted and reproduced with permission [56, 57, 58]. 2013, The Royal Society of Chemistry. 2019, American Association for the Advancement of Science. 1964, Science. (B) Quaternized and carboxymethylated amyloses phase separate into cell sized dense phase droplets, which can be stabilized with a triblock copolymer to form artificial cells, as visualized with confocal microscopy. Adapted and reproduced with permission [29]. 2017, American Chemical Society. (C) Protein‐loaded polymersomes can be loaded in the amylose‐based artificial cells to mimic artificial organelles [59]. 2019, American Chemical Society. (D) Membraneless organelles can be mimicked by multiphase separation which manifests as a droplet‐in‐droplet structure. Adapted and reproduced with permission [60, 61]. 2019, American Chemical Society.

While membrane‐bound compartments effectively mimic the membranous structure of living cells, their aqueous lumen lacks the chemically enriched crowdedness of the living cell cytoplasm. Especially with the goal of emulating intracellular processes in life‐like environments, integrating a similarly molecularly dense interior in artificial cell design is important. Membraneless compartments, such as liquid‐liquid phase‐separated (LLPS) systems, offer an attractive strategy to address this limitation via their ability to mimic the viscous, dynamic, and crowded cytoplasm of living cells. Owing to the absence of a membrane, these droplets are permeable and can readily accumulate and concentrate diverse molecules, such as proteins, nucleic acids and metabolites from the surroundings arising from differential molecular affinities toward the coexisting phases [27, 28]. As a result, the up‐concentration of molecules and the chemically distinct interior of the droplets can accelerate compartmentalized reactions. Despite these advantages, the innate lack of a membrane also introduces a drawback: membraneless droplets are often prone to coalescence and exhibit limited temporal stability, which can restrict their application in artificial cell research [29]. Nevertheless, studies have addressed this limitation by demonstrating the stabilization of membraneless systems using block co‐polymers [29], phospholipids [30, 31, 32], and polyanionic clusters [33]. These strategies enable the formation of semi‐permeable compartment boundaries that facilitate the sequestration of macromolecules inside the droplet while averting fusion, thereby combining the structural stability of membrane‐bound systems with the intracellular crowdedness of membraneless compartments in a single system.

With the increase in platforms used for artificial cell research and the advances made in the introduction of life‐like behavior, the question arises what distinguishes these compartments from similar particles used in for example drug delivery or catalysis. In our opinion, an artificial cell is a compartment—with or without membrane—equipped with a responsive feedback mechanism that enables it to engage in adaptive interactions with its environment. This endows these compartments with the ability to adapt and communicate as a premise for more sophisticated life‐like behavior.

Until recently, the artificial cell was kept structurally simple, while the function was included by the presence of active components. However, the use of intracellular compartments, either membrane‐bound or membraneless, as well as structural elements such as an artificial cytoskeleton has provided the field with additional tools to regulate function in a more controlled way. In this section, we will focus on recent developments in complex organization, in particular related to coacervates.

2.2. Coacervate‐Based Compartments

Coacervates, or condensates are liquid droplets formed through the spontaneous physical process of associative LLPS, resulting in two aqueous phases: a solute‐rich dense phase that coexists in dynamic equilibrium with a dilute phase. This is in contrast to segregated LLPS, which involves two soluble molecules that form coexisting phases at equilibrium, each enriched in one of the solutes (i.e. aqueous two‐phase systems). A typical phase diagram for coacervation displays two regions that identify conditions for which phase separation occurs, separated by the binodal curve (Figure 1). In the phase‐separated region, tie lines link the two points corresponding to the coacervate and dilute phase, respectively. Conditions at which phase separation cannot occur define a critical point on the binodal curve. The formation of coacervates is driven by multiple noncovalent interactions, including electrostatic interactions, dipole–dipole interactions, hydrogen bonding, π–π stacking, cation–π interactions, and van der Waals forces [27, 34]. Importantly, the process is often governed by a significant entropy gain arising from the release of solvent molecules and bound counterions during complexation. Depending on the number of components involved in the phase separation process, coacervates can be classified into two types. When coacervation occurs from a single component due to the presence of complementary interactions within different regions of the same molecule, it is referred to as a simple coacervate. In contrast, when two or more components participate in the phase separation process, the resulting droplets are known as complex coacervates. These droplets were first proposed as protocell models by Oparin in 1930s, who hypothesized that they could provide insights into how life originated on Earth. Today, coacervates are widely used as artificial cell models, not only to gain fundamental understanding of how compartmentalization enables biochemical reactions in a crowded microenvironment but also for their potential applications in biomedical engineering [35, 36]. Accordingly, a wide range of materials has been explored for coacervate formation, including peptides, synthetic polymers, biopolymers, nucleotides, proteins, ionic surfactants, and fatty acids [27, 37]. Furthermore, due to the presence of multivalent interactions among coacervate components and their dynamic exchange with the surrounding medium, coacervate protocells are highly responsive to external stimuli such as pH, temperature, and ionic strength. In this review, we primarily focus on the emerging functional behaviors of coacervate‐based artificial cells irrespective of their material composition. Readers interested in a deeper understanding of coacervate materials and the specific molecular interactions governing their formation are referred to several excellent reviews on this topic [27, 36, 37, 38, 39].

FIGURE 1.

Types of liquid‐liquid phase separation (LLPS) and coacervate formation. Multiple non‐covalent interactions can contribute to LLPS and to coacervate formation. Segregative LLPS manifests as droplets of component 1 in component 2, whereas associative LLSP manifests as dense phase droplets in a dilute phase environment. For complex coacervation the dense phase contains both components 1 and 2. For simple coacervation, component 1 phase separates by itself. These types of behavior can be described by phase diagrams, of which two schematic examples are shown.

Coacervates typically manifest as droplets which typically exhibit fluid dynamics (e.g. fusion, fission, and wetting behaviors). Although most coacervates lack a physical membrane, the dense phase droplet and dilute phase pose distinct chemical environments, separated by the droplet interphase with its own characteristics. The dense phase‐forming materials largely dictate characteristics like pH, hydrophobicity, viscosity, and density. Tuning material properties of coacervate‐based artificial cells can thus be obtained by appropriate choice of the used materials. Protein condensates, i.e. have been shown to sustain significant pH gradients without any external energy input [40]. The dense phase pH shifts towards conditions of minimal electrostatic repulsion, meaning that the pH of the dense phase is buffered to the isoelectric point of the protein scaffold. By using different phase separating proteins, different compartments with different pH ranges can be realized. To change the viscoelastic properties of coacervates multiple strategies exist. These include tuning the “sticker strength” in the phase separating material by adjusting the binding strength of the stickers, or using more connected sticker units to increase multivalency. For example, in DNA nanostars the binding strength of the sticky ends of the DNA was tuned by making longer complementary ends [41, 42]. In protein‐RNA mixtures, varying the stoichiometry of RNA sequences can influence the material properties of the resulting protein‐RNA coacervates [43, 44, 45]. For synthetic polymers, different functional groups can be introduced to incorporate multiple intermolecular interaction types. In our group, coacervates have been made by phase‐separating polymers, of which one is an amino‐functional polymer. Changing the type of amine and charge density on this polymer changed coacervate formation, stability, protein partitioning, and enzyme function [46]. In other work by our group, methacrylate moieties were introduced to the phase separating polymers. The photo‐crosslinkable methacrylate groups were locally irradiated with UV light in the presence of a photoinitiator, to crosslink the polymers and thus change the viscoelastic properties of the coacervates. By varying the amount of methacrylate groups and irradiation time, the degree of crosslinking was tuned, while in the same time local irradiation provided spatial control over the crosslinking [47, 48].

Next to the distinct environments in the dense and dilute phase, also the coacervate interface has its own characteristics. Molecules experience interactive forces at interfaces that are different from forces in the bulk phase, which can lead to differences in adsorption (enrichment of certain molecules at the interface compared to the bulk phase composition), molecular orientation, molecular movement, and electric potential [49, 50, 51]. This change in orientation and movement of molecules at interfaces also affects the liquid‐to‐solid transition of certain proteins. For example, both hnRNPA1, a protein involved in amyotrophic lateral sclerosis, and FUS, an RNA‐binding protein, showed fibril formation originating at interfaces [52, 53, 54]. These effects can be ascribed to larger fluctuations in local protein density at the interface and a different protein orientation/conformation inside and outside the coacervate. Even though the dilute‐dense‐phase interface can be considered as a distinct phase and environment, it does not always provide the same physical enclosure as a membrane. For instance, membraneless condensates have a high tendency to fuse and coalesce. To overcome this limitation, interfaces are often stabilized or coated with membrane‐forming molecules like (amphiphilic) proteins, synthetic polymers, small molecules such as fatty acids, phospholipids, and surfactants, particles (e.g. polystyrene or gold coated nanoparticles), and condensate‐amphiphilic block polymers [55]. Depending on the type of material, the membranes are still permeable, but often allow for the encapsulation of functional macromolecules while preventing coalescence and content mixing. In our group, a triblock copolymer (terpolymer) was designed for the interfacial stabilization of cell‐sized coacervate microdroplets (Figure 2B) [29]. The asymmetric molecule consists of hydrophilic, hydrophobic, and polyanionic components (poly(ethylene glycol), poly(caprolactone‐gradient‐trimethylene carbonate), and poly(glutamic acid), respectively) capable of direct coacervate membranization via electrostatic surface anchoring and chain self‐association. In this way, discrete artificial cell populations were stabilized against coalescence.

2.3. Organelles in Artificial Cells

Besides the cell being a compartment itself, a broad range of different subcellular compartments contribute to the cell's organization and thus functioning. When these structures within a cell perform a specific function, they are considered organelles. Organelles in living cells can again be divided in membranebound and membraneless organelles. These different types of organelles have different properties and thus are used in different cellular processes. Membraneless organelles are often more dynamic in nature and can quickly form and dissolve in response to stress. Since they do not have a membrane, partitioning of molecules in these organelles depends mostly on interactions with the condensate scaffold. The membranized organelles on the other hand are of a more permanent nature, and can form a more or less permeable membrane to separate molecules from the inside and outside of the membrane. Since both types of organelles are present in living cells, they are also both considered to be of interest to artificial cells.

Membrane‐bound organelles are generally made from the same types of materials as the membrane‐bound artificial cells. The difference is only in the size of the different compartments and the more intricate preparation method, as the smaller compartments should be encapsulated in the other. Either the same type of compartments can be combined like liposomes‐in‐liposome, or different compartment types can be mixed. For example, protein‐loaded polymersomes were encapsulated in coacervate droplets, mimicking proto‐organelles (Figure 2C) [59]. By this extra compartmentalization, the spatial organization of enzymes can be tuned, leading to an enhancement of functionality. Next to this spatial control, temporal control can be achieved, making the system even more lifelike. This has been demonstrated with liposome‐based compartments. Phospholipid membranes allow the introduction of (triggerable) channels, and their permeability can be further tuned by using different lipids with varying melting temperature and hence fluidity. Because of these properties, systems have been designed where triggers can be temperature‐ or small molecule‐based [62, 63, 64].

Examples in living cells of membraneless organelles, or biomolecular condensates are the nucleolus, stress granules, P granules, Cajal bodies, and nuclear bodies, which often concentrate specific proteins and RNA to a macro scale level to regulate various biochemical processes efficiently. Membraneless organelles in artificial cells can be any subcellular compartment without a designed membrane‐like structure. Even though the terminology for membraneless organelles in artificial cells and living cells slightly differ, the way of formation, appearance, and broad functions are the same. Next to this, synthetic coacervates are often considered as models for biomolecular condensates. Membraneless organelles in artificial cells are either made by multiphase separation in coacervate systems, or by phase separation of macromolecules in liposomes or in aqueous droplets in oil. Multiphase separation of polyelectrolytes is a generic phenomenon in mixtures of at least three different charged macromolecules (Figure 2D) [60, 61]. The non‐covalent multivalent interactions between the different macromolecules leads to competition between the polyelectrolytes, forming an inner dense phase and outer dense phase in a dilute environment. One example of this is shown in Figure 2D where multiphase coacervates are formed by poly(L‐glutamic acid), protamine sulfate, and poly(allylamine hydrochloride) or the short peptide RRASLRRASL. Another example is the formation of a DNA‐rich dense phase in our amylose‐based coacervates, where the negatively charged DNA competes with the negatively charged amylose for interacting with the positively charged amylose. Since the DNA has a higher charge density, this molecule interacts more with the positively charged amylose and forms the inner dense phase, while the negatively charged amylose forms the outer dense phase with the rest of the positively charged amylose [65].

Membraneless organelles formed by phase separation of macromolecules in liposomes or in aqueous droplets in oil were for example achieved by the use of premade components, such as intrinsically disordered proteins or RNA, which are combined with synthetic elements to induce condensation [66]. Another approach was demonstrated by the expression of phase separating proteins in liposomes serving as artificial cells, producing biomolecular condensates in situ in artificial cells [12]. The engineered condensates, expressed by cell‐free transcription and translation, sequestered RNA in the liposomes and thereby reduced protein expression, demonstrating the intricate relationship between structure and function found in condensate systems.

2.4. Cellular Architectures

Next to having (sub)cellular compartments, also their organization (where and when they exist) plays an important role. To describe this overarching organization, we use the term ‘architecture’, since it concerns the design, function, and dynamic construction of the cell's physical structures in a specific environment. For example, the nucleolus of eukaryotic cells contains core‐shell or nested microstructures where one phase is contained within another. The bulk of the nucleolus is the granular component (GC), while ribosomal genes and RNA are transcribed in small phases of the nucleolus called the fibrillar centre (FC) and dense fibrillar component (DFC) [67]. Multiphase structures in which boundaries are in contact with each other are also known, such as the interaction between P‐bodies and stress granules in the cytoplasm of cells which can include docking and fusion [68]. Mimicking these cellular architectures in artificial cells however, is challenging especially when preserving complex and dynamic structures over longer timescales [69, 70, 71, 72].

A way to change the artificial cellular architecture is by dynamically induced phase separation. This is a kinetic process that results from a change in coacervate composition (while maintaining the same materials) and results in nucleation of dilute and/or dense phase droplets, of which the formation is dependent on the rate of the composition change, the diffusion rates within the coacervates, and the coacervate size. By changing the interaction strengths of the phase separating polymers, e.g. by changing salt concentration, pH, or temperature, the phase diagram changes, leading to a different coacervate composition. When the conditions change abruptly, droplets of the dilute phase can be nucleated in the dense coacervate phase, due to limited diffusion rates [73]. When using multiphase coacervates, also the different dense phases can be nucleated in each other (Figure 3A) [65]. Depending on the coacervate size and rate of environmental change, the number of newly nucleated phases can be controlled. The bigger the condensates and the faster the change, the more droplets form with the same distance between them and the bulk. By this kinetic process, multiple small droplets are formed in a larger coacervate, which leads to more interfacial area, which can lead to the enhanced uptake of cargo [65]. These newly formed architectures however are thermodynamically unfavorable because of the increase in surface area, and thus energy. This leads to short life‐times of the transient architectures which revert back to the thermodynamically stable state. To preserve the transient coacervate architectures, the polymer building blocks can be modified with crosslinkable methacrylate groups, which upon crosslinking limit molecular diffusion and create a more robust artificial cell with custom architectures [47]. Another way to create multiple of the same compartments in coacervate based systems is by in situ photopolymerization of a third phase separating polymer [74]. As polymerization proceeds, coacervate droplets mature into multi‐compartmental coacervates, forming droplets and vesicles, which is driven by electrostatic interactions and osmotic pressure differences.

FIGURE 3.

Varying cellular architectures. (A) (i) By changing the interaction strengths of the phase separating polymers, e.g. by changing salt concentration, the phase diagram changes, leading to a different coacervate composition. Droplets can be nucleated when the conditions change abruptly, for which the maximum volume fraction can be predicted by overlapping the tie‐planes. (ii) Dynamically induced phase separation leads to nucleation of dilute and/or dense phase droplets in the dense phase, as seen by confocal microscopy. Adapted and reproduced with permission [65]. 2024, Springer Nature. (B) PDA fibrils can bundle together inside amylose‐based coacervates to form an artificial cytoskeleton. Depending on the hydrophobicity of the PDA the location of the bundles can be tuned. Hydrophobic PDA bundles are associated to the membrane and hydrophilic PDA bundles are distributed in the lumen. Adapted and reproduced with permission [75]. 2025, Springer Nature.

2.5. Cytoskeleton

Another structural component for cellular organization is the cytoskeleton, best known for providing mechanical support and regulating the cell's physical properties [76]. Moreover, the dynamic and reversible formation of the cytoskeleton governs cellular morphological changes, controls intracellular transport, membrane dynamics, signal transduction, and modulates internal physical properties, thereby orchestrating essential processes such as cell division and migration [77, 78]. Therefore, reconstituting the cytoskeleton within artificial cells is crucial for understanding the mechanical, structural, and dynamic complexity of living systems and for achieving life‐like functions with sophisticated regulatory control.

Several studies examined the effects of natural cytoskeletal components, such as actin filaments and microtubules, inside artificial cell compartments (e.g. GUVs, water‐in‐oil and coacervate droplets) through the reconstitution of cytoskeletal proteins [79]. These studies reported various shape transformations, including deformation, protrusion, and blebbing of GUVs driven by force generation or strong interactions between the filaments and the membrane [79, 80, 81, 82]. Furthermore, by coupling cytoskeletal filaments with motor and other proteins, researchers have achieved spatial organization of filaments, formation of contractile rings, and hierarchical structures, enabling artificial cells to undergo division, motility, and dynamic deformation [83, 84, 85]. More recently, actin polymerization inside polypeptide coacervates was studied, showing preferential partitioning of actin monomers and an enhanced polymerization rate, leading to the formation of ring‐shaped filamentous structures around the coacervate interface [86, 87, 88, 89, 90]. In another example, GTP‐fueled self‐assembly of the bacterial tubulin homolog FtsZ inside coacervates induced mechanical deformation and eventually led to coacervate division [83, 84, 85].

However, challenges such as limited stability and complex purification of native proteins have prompted the design of artificial cytoskeletons from synthetic building blocks to impart structural, mechanical, and dynamic functionalities reminiscent of living cells. Recent examples include cytoskeletal networks constructed from DNA, peptides, or DNA–peptide conjugates within artificial cell, which endow artificial cells with mechanical resilience against hypertonic stress, enhanced stiffness, and regulated cargo diffusion [16, 91, 92, 93, 94, 95, 96]. These networks also influence the internal artificial organelle dynamics by increasing internal viscosity and can serve as tracks for vesicular transport. Notably, a recent study demonstrated the formation of a contractile DNA tube ring inside a GUV, offering valuable insight into the design and understanding of minimal division machinery in artificial cells. Our group designed a DNA‐functionalized supramolecular polymer that acted as an artificial cytoskeleton inside coacervate‐based artificial cells and served as a signal‐localization hub for transduction and communication, capable of reversibly recruiting, and releasing signals on demand through a strand‐displacement process (vide infra) [97].

Moreover, Chen et al. demonstrated the construction of a DNA‐based artificial cytoskeleton inside a viscoelastic confinement using all‐DNA coacervate‐based artificial cell, to modulate internal mechanical properties [98]. This platform revealed distinct DNA nanotube growth mechanisms compared to solution conditions where the nanotubes remained tethered to the viscoelastic matrix during and after growth, resulting in a two‐order‐of‐magnitude increase in the artificial cell's elastic modulus. This mechanical reinforcement enabled the artificial cells to maintain stable contact with mammalian HeLa cells without leaking their internal contents.

In another example, a model artificial cytoskeleton formed from self‐assembled polydiacetylene (PDA) fibers was developed within a membranized coacervate‐based artificial cell (Figure 3B) [75]. The spatial reorganization of this artificial cytoskeleton within the coacervate closely mimicked the structural, mechanical, and organizational functions of the natural cytoskeleton in living cells. Via the introduction of hydrophobic moieties, the fibers were associated with the membrane which enabled the regulation of membrane dynamics, such as the lateral diffusivity of membrane components. Without hydrophobic modification, the fibers formed a bundled and densely entangled network within the lumen, which governed the mechanical properties of the artificial cell, exhibiting a Young's modulus which closely matched that of the natural human leukemia cell line, thereby effectively mimicking the mechanical behavior of a living cell. Furthermore, the fiber bundling facilitated protein reconstitution and colocalization of split NanoLuc luciferase on the fiber surface, restoring, and enhancing its metabolic activity as a result of the scaffolding effect. Together, these examples closely mimic the roles of the cytoskeleton in living cells, mediating the spatial organization and colocalization of biomolecules and proteins, and dynamically regulating cellular function.

3. Energy‐Dependent Metabolism and Growth

3.1. Energy Production

Irrespective of whether the cell is natural or artificial, energy is indispensable for cellular functioning, as all cellular activities rely on energy‐dissipative chemical reactions. Therefore, the continuous supply of energy in artificial cells is crucial to drive metabolic function, enable self‐maintenance, and achieve life‐like behavior [99, 100, 101].

Living cells primarily use adenosine triphosphate (ATP) as the energy currency, with its generation and consumption tightly coupled to various biological reactions. In contrast to natural systems, where ATP is self‐regenerated through metabolic pathways, artificial cells typically rely on either externally supplied or internally produced energy to execute various functions [100, 101]. Energy can be provided directly in the form of ATP or generated in situ through substrate‐level phosphorylation using high‐energy substrates such as creatine phosphate, acetyl phosphate, or phosphoenolpyruvate which have been shown to support protein synthesis inside liposomes [102, 103, 104, 105]. However, to construct truly self‐sustaining artificial cells, continuous self‐generation of energy, similar to natural cells is essential. In a recent example, this was achieved by reconstituting the transmembrane protein F₀F₁‐type ATP synthase in the membrane of GUVs to produce ATP [106]. Furthermore, by coupling ATP synthase or plant‐based photosystem II (PSII) with the light‐driven proton pump bacteriorhodopsin (bR) on the liposome membrane, researchers built a self‐sustaining ATP‐generating system capable of producing ATP using light as additional energy source [107, 108]. The self‐produced ATP was subsequently utilized to power ATP‐dependent processes such as cell‐free transcription and translation, actin polymerization, vesicle morphogenesis, and CO₂ fixation within artificial cells [108, 109, 110].

In the context of coacervate‐based artificial cells, ATP is extensively used as one of the key structural components for coacervate formation and is introduced into the system either by direct addition or through in situ enzymatic production from energy‐rich substrates [111, 112, 113, 114, 115, 116]. However, it is scarcely utilized as a power source for biosynthetic reactions. Recently our group demonstrated an example in which ATP was consumed to recruit and organize proteins within compartmentalized complex coacervates [117]. Specifically, the energy‐consuming phosphorylation step, crucial for the recruitment of client protein Green Fluorescent Protein (GFP)‐cRaf to the scaffold protein 14‐3‐3, was facilitated by protein kinase A (PKA) at the expense of ATP (Figure 4A). Moreover, dephosphorylation of the scaffold protein by phosphatase dissipated the stored chemical potential energy, resulting in the efflux of the client protein from the coacervate due to the loss of affinity for the scaffold. Furthermore, this platform was utilized to reconstitute the two parts of split‐NanoLuc luciferase on the 14‐3‐3 scaffold, serving as a docking station to restore its functionality and catalyze the metabolic conversion of its substrate, furimazine. This example closely correlates with various energy‐consuming, dynamic binding–unbinding events of protein partners that regulate diverse cellular signaling pathways.

FIGURE 4.

Energy production and metabolism of artificial cells. (A) (i) Schematic of NanoLuc reconstitution (SmBiT and LgBiT) and restoration of enzymatic activity inside a coacervate via a protein complementation assay involving the 14‐3‐3 and c‐Raf domains. This process is mediated by an energy‐intensive phosphorylation step of c‐Raf domain catalyzed by protein kinase A (PKA) using ATP as the substrate. Upon reconstitution, split NanoLuc catalyzes the conversion of furimazine to furimamide, producing bioluminescence. His denotes the histidine tag used for protein recruitment into the coacervate. (ii) Corresponding bioluminescence microscopy image with and without PKA acquired with a DAPI filter without external excitation (scale bar: 25 µm) and (iii) quantified bioluminescence intensity at 458 ± 12.5 nm (N = 3) for the coacervate samples. Adapted and reproduced with permission [117]. 2023, Wiley‐VCH. (B) (i) Schematic of living‐synthetic hybrid protocell morphogenesis through proto‐cytoskeletal actin filament growth, driven by E. coli‐mediated sustained ATP production over time. (ii) Confocal microscopy image showing protocell morphology transformation into an amoeba‐like structure. The protocell contains multiple cellular architectures, including a DNA‐histone coacervate subcompartment (blue, DAPI), surrogate mitochondria (E. coli expressing eGFP, green), an outer membrane derived from lipids of P. aeruginosa (PAO1 strain, red), and F‐actin microfilaments (red, RhB). Scale bar: 10 µm. Adapted and reproduced with permission [118]. 2022, Springer Nature. (C) (i) Schematic of the metabolic conversion of L‐tryptophan to indigo via a multi‐enzymatic cascade involving tryptophanase (TnaA) and a flavin‐containing monooxygenase (FMO), exhibiting enhanced reaction rates within a coacervate environment. TnaA catalyzes the conversion of L‐tryptophan to indole using pyridoxal‐5’‐phosphate (PLP) as a cofactor. The indole is oxidized to indoxyl by FMO in the presence of NADPH, after which indoxyl dimerizes to form indigo as the final product. (ii) Time‐dependent NADPH depletion during the reaction, monitored at 340 nm, showing a faster consumption rate in coacervates compared to the bulk solution. Adapted and reproduced with permission [119]. 2020, Springer Nature.

Although in living cells crowding plays a significant role in gene expression, emulating cell‐free protein expression fueled by self‐generated ATP in coacervates remains a major challenge. In this respect, the recent example of using a living‐synthetic biohybrid system by Xu et al. stands out as a single report to date because of its level of organizational and functional complexity (Figure 4B) [118]. Following a multistep living‐material assembly process, the authors constructed a membranized bacteriogenic artificial cell through the co‐encapsulation and on‐site lysis of two spatially segregated live bacterial colonies (E. coli and P. aeruginosa) within a complex coacervate microdroplet. These artificial cells exhibited diverse cytomimetic functions, including the operation of proto‐metabolic pathways such as glycolysis (a 10 enzyme cascade) and the execution of in vitro transcription and translation. Most importantly, by implanting live E. coli cells as surrogate mitochondria, continuous metabolic activity was maintained through the sustained production and secretion of ATP. This ATP supply enabled the control of the internal architecture of the artificial cell by promoting the formation of a three‐dimensional network of actin filaments, driving a morphological transformation from a spherical to an amoeboid shape.

3.2. Metabolism

Metabolism is a vital process for sustaining life and is fundamentally defined as the set of chemical transformations of metabolites (substrates) through cascades of enzymatic reactions within cellular compartments. These reactions generate energy or cellular building blocks required for proliferation and the maintenance of cellular structure [120]. Coacervates have been utilized not only for executing natural biological reactions but also for the synthesis of non‐natural molecules through chemical transformations [27, 37, 114, 121]. These reactions can occur either at the sub‐compartmental level, resembling natural organelles, or within cytomimetic crowded environments, where the coacervate serves as a hub for biomolecular reactions [122]. The micro‐environment can influence enzymatic reaction kinetics in two opposing ways: it may accelerate reaction rates by increasing the local concentrations of enzymes and substrates, or inhibit reactions through undesired molecular interactions or spatial restrictions imposed on the enzymes—topics that have been extensively reviewed elsewhere [27, 37, 114, 121].

As an example, our group demonstrated glucose oxidase (GOx) and horseradish peroxidase (HRP)‐mediated cascade reactions within organized, membranized coacervates [29]. This concept was further extended to introduce hierarchical internal complexity by encapsulating enzymes within polymersomes, which were subsequently incorporated into coacervates to function as artificial organelles (vide supra) [59]. In this system, enhanced enzymatic cascade activity was observed when both enzymes were co‐encapsulated within the same polymersome, compared to their separate encapsulation, once the entire assembly was localized inside the coacervate. In a related example, the efficiency of an enzymatic cascade converting L‐tryptophan (L‐Trp) to indigo was significantly enhanced through the sequestration and spatial organization of tryptophan anhydrase (TnaA) and flavin‐containing monooxygenase (FMO) within the coacervate (Figure 4C) [119]. The confined and highly crowded microenvironment promotes close enzyme proximity while enhancing substrate partitioning and diffusional control relative to bulk solution, collectively leading to increased reaction rates.

In living cells, metabolism maintains a non‐equilibrium state through a continuous supply of energy, typically in the form of ATP or other high‐energy molecules, via biosynthetic and energy‐dissipative chemical reactions [123]. In the context of artificial cells, one such example is cell‐free protein expression which has been performed in water‐in‐oil droplets and in GUVs by encapsulating cell lysate inside these compartments and which could be sustained for example using light [103, 105, 109]. Towards integration of transcription‐translation (TX‐TL) with coacervate‐based artificial cells, studies have shown mixed success. Early work demonstrated GFP and mCherry expression in low micromolar and nanomolar levels, respectively, though it remained unclear whether synthesis occurred inside the coacervates or in the surrounding solution followed by partitioning of the fluorescent product in the artificial cells [116, 124]. Schoenmakers et al. identified an Elastin‐like polypeptide‐single‐stranded DNA (ELP‐ssDNA) coacervate system that supported micromolar‐level GFP expression when investigating the TX‐TL compatibility of different coacervation systems [125]. More recently, Tomohara et al. combined coacervates of intrinsically disordered proteins (IDP) with aqueous two‐phase systems (ATPS) as droplet‐in‐droplet structures to demonstrate spatially separated transcription in the IDP droplet akin to a nucleus, and translation in the outer droplet as mimic of the cytosol [126]. As far as we are aware, incorporation of TX‐TL machinery into coacervate droplets has to date been more challenging than their equivalent incorporation into aqueous droplets or vesicles, perhaps due to droplet destabilization by TX‐TL reaction mixtures, or because of interaction strength mismatches between droplet components and TX‐TL reagents.

3.3. Growth and Proliferation

Artificial cell growth refers to the increase of size or volume of the compartment. In liposomes, growth means an extension of the lipid bilayer, which is for example facilitated by either external addition of membrane components, by fusion of two liposomes driven by membrane tension or by internal membrane component production through enzymatic reactions or chemical catalysts embedded into the vesicle membrane [17, 127, 128, 129, 130, 131, 132, 133]. Coacervate‐ based artificial cells without a membrane can grow passively, through coalescence or Ostwald ripening, which happen spontaneously to minimize the surface energy, or actively through generation of coacervate material at the expense external energy input [112, 134, 135, 136, 137]. For example, Nakashima et al. reported the formation of active coacervate droplets by the complexation of enzymatically produced ATP from ADP and pyruvate kinase, with lysine‐rich protein K₂; the recruitment of substrate and enzyme into the droplet increased the local concentration of ATP inside which subsequently accelerated protein sequestration in the droplet, providing positive feedback into the system leading to growth [112]. Interestingly, growth of the droplets happened without compromising the concentration of the constituents, which is important if the process of growth is to be further developed into replication and division. In DNA‐based membrane‐enclosed coacervates, growth occurred through rupture of the membrane (Figure 5A) [138]. The catalytic production of a DNA intercalator led to the weakening of the DNA‐hydrogel membrane. This induced morphological changes in the cell, resulting in coacervate fusion through membrane rupture. These results are clear examples of integrated systems in which abiotic metabolism regulates the functional properties of an artificial cell.

FIGURE 5.

Growth and division of artificial cells. (A) (i) Schematic of the growth of an all‐DNA artificial cell via intercellular fusion, induced by swelling and weakening of the DNA shell through intercalation of the metabolite umbelliferone generated by abiotic metabolism by a Ru‐containing artificial metalloenzyme inside the coacervate. Redrawn from the Ref. [138]. (B) (i) Schematic representation of coacervate division driven by FtsZ filament growth in the presence of GTP fuel. (ii–v) Corresponding confocal images showing fibril splitting at different time points (Scale bar: 10 µm). Adapted and reproduced with permission [85]. 2018, Springer Nature.

For growth to lead to proliferation, the entire content of the artificial cell should be replicated, including the functional machinery and information carriers present. Until now, the replication process in artificial cells has been limited to DNA and RNA. Living cells accomplish replication efficiently by storing genetic information in DNA, which is passed on to subsequent generations through the coordinated action of protein circuits. For self‐replication to occur, three fundamental steps must be followed: i) template recognition by the building blocks; ii) polymerization; and iii) dissociation of the duplex for information transfer [5, 139]. Executing replication within crowded coacervate environments is significantly more challenging than in vesicle‐based or water‐in‐oil droplet systems, where all the necessary components can be encapsulated more easily [140, 141, 142]. As with TX‐TL, strong interactions between coacervate materials and replication ingredients often hinder the process. Despite reports of enhanced ribozyme activity inside coacervates, the replication process still faces major challenges, such as the selective partitioning of nucleobases (NTPs), DNA, RNA, and essential enzymes. This bias in partitioning has important consequences for the overall efficiency of replication [38, 143, 144].

Regarding division, ideally a mother cell divides into two identical daughter cells that can subsequently grow, proliferate, and evolve over time. In a simple theoretical model proposed by Zwicker, the cell cycle is described as a process in which a droplet grows through the accumulation of internal material, leading to shape changes [134]. Once the droplet reaches a stationary size, it splits into two new droplets, and the cycle repeats. However, natural cell division is far more intricate, being regulated by complex internal machinery. Reported studies show that artificial cells, such as liposomes or coacervates can undergo division through external perturbations, including the application of shear forces, changes in osmotic pressure, or stimuli such as light and temperature [17, 38, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154]. Among these, coacervate droplets tend to be more sensitive to variations in external factors like osmolarity, pH, and temperature, which often lead to dissolution rather than division. In contrast, autonomous division driven by internal processes brings artificial cells closer to living systems. This has been demonstrated in liposomes where enzymes modified the surface charge of membrane components or where embedded membrane proteins induced curvature, generating mechanical strain that ultimately led to division [148]. The Huck group reported coacervate division driven by FtsZ protein filaments formed in the presence of GTP as a chemical fuel (Figure 5B) [85]. The growth of these protein filaments elongated the coacervate, and filament bundling at the tips produced a narrow, fragile middle region that eventually led to division. Remarkably, this process could be repeated multiple times. Similarly, coacervate droplets composed of crystalline actin have been shown to divide via the formation of an actomyosin ring, where active myosin proteins constrict the droplet's midplane, splitting it into two daughter droplets [152, 153, 154]. Moreover, asymmetric division was induced in multiphase coacervate droplets by tuning the interfacial tension between the two phases, leading to budding followed by fission [153].

Despite these advances, truly autonomous proliferation within a single coacervate‐based artificial cell remains unachieved to date. Although Kurihara and co‐workers demonstrated self‐reproduction of DNA coupled with liposome growth and division, the dilution of chemical catalysts and genetic material after each cycle limited sustained autonomous proliferation [128]. Therefore, the integration of hybrid artificial cells featuring a coacervate core and a lipid membrane, combined with continuous energy supply and existing proliferation mechanisms, may pave the way toward constructing artificial cells that more closely mimic living systems.

3.4. Energy‐Dependent Out‐of‐Equilibrium and Adaptive Behavior

All living organisms are out‐of‐equilibrium systems that are brought out of thermodynamic equilibrium through the continuous conversion of chemical energy into functional work. This constant energy input creates transient structures, modulates biochemical reaction cycles, and drives feedback loops that enable cells to sense and adapt to their external environment by reorganizing internal structures with high spatiotemporal precision, ultimately giving rise to essential cellular functions such as motility and division [155].

In artificial cells, out‐of‐equilibrium behavior refers to the creation of sustained or transient dynamic processes that rely on the continuous input and consumption of chemical energy within a compartmentalized system. Litareture until now mostly has explored stimuli‐responsive reversible supramolecular assembly triggered by pH, light or using DNA nanotechnology inside compartments [15, 94, 156]. Recently, our group reported an enzyme‐catalyzed, pH‐fueled reversible assembly of a hydrogelator (UPy‐PEG) inside giant unilamellar vesicles (HL‐GUVs), where urease‐driven urea hydrolysis increases the internal pH and triggers a reversible gel–sol transition [157]. This structural change regulates diffusion and modulates the activity of a co‐encapsulated enzyme (HRP), establishing a programmable structure–function relationship. Using the same pH control, membrane‐embedded Ni^{2 +}–NTA sites enabled reversible, pH‐dependent recruitment of His‐tagged split‐luciferase fragments (LgBiT and SmBiT), allowing spatiotemporal regulation of enzymatic signaling within artificial cells [157].

Another example reported by Krehan et al. describes artificial cells that exhibit homeostasis and self‐regulatory behavior within artificial cell consortia and prototissue‐like spheroids through a pH‐regulatory mechanism (Figure 6A) [158]. In this system, pH‐modulating enzymes encapsulated within polymeric microcapsules interact with a pH‐responsive membrane, enabling chemo‐structural feedback that maintains homeostasis. For example, Urease‐loaded artificial cells remain closed at high pH but open and consume urea upon acidification, thereby increasing the pH. In contrast, glucose oxidase (GOx)‐loaded cells generate acid at high pH by oxidizing glucose. The complementary activity of these two cell types allows the system to autonomously stabilize pH and return to a dormant state after environmental fluctuations. When organized into prototissue‐like spheroids, the artificial cells also exhibit collective self‐protection: the outer layer senses and neutralizes external acidic threats, shielding the inner core from environmental changes. Similarly, in mixed populations, urease‐containing cells counteract the internal acidification produced by neighboring GOx‐containing cells. This work provides a strategy for designing functional synthetic tissues capable of regulating their microenvironment and may inspire approaches to improve immunotherapy by counteracting the tumor‐induced acidic microenvironment.

FIGURE 6.

Adaptive behaviors of artificial cells. (A) (i) Schematic illustration of pH‐responsive homeostatic artificial cells composed of polycationic and polyanionic polymeric capsules encapsulating urease (left: Urease‐AC) and glucose oxidase (right: GOx‐AC), respectively. The capsules possess pH‐responsive membranes made of poly(2‐(diethylamino)ethyl methacrylate) (PDEAEMA) and poly(methacrylic acid‐co‐N‐butylmethacrylamide), which regulate permeability: PDEAEMA becomes permeable below ∼pH 7.0 due to protonation of the amine groups, whereas the poly(methacrylic acid‐co‐N‐butylmethacrylamide) membrane becomes permeable above pH ∼4.6 due to deprotonation of the acidic functional groups. (ii) Self‐protection in multi‐AC spheroids: cargo encapsulated inside the Urease‐AC remains retained due to membrane impermeability at pH > 7. However, in the presence of only glucose fuel and without pH feedback (absence of urea), glucose oxidation lowers the pH, rendering the membrane permeable and causing cargo diffusion from the AC (left). In contrast, in the presence of urea fuel (right), urease‐mediated pH feedback maintains membrane impermeability and protects the cargo from diffusion. Adapted and reproduced with permission [158]. 2024, Elsevier. (B) (i) Schematic of a light‐triggered motile coacervate decorated with Au nanoparticle–containing stomatocyte nanomotors on its surface. (ii) Corresponding motion trajectories under two‐photon laser irradiation and (iii) confocal microscopy images of coacervates with different surface nanomotor densities. Scale bar: 10 µm. Adapted and reproduced with permission [159]. 2025, American Chemical Society. (C) (i) Schematic of intracellular motility of a catalase‐functionalized coacervate within a GUV, powered by the catalase‐mediated decomposition of H₂O₂ into O₂ at the coacervate surface. (ii) Mean‐square displacement (MSD) profiles of catalase–coacervates confined within GUVs at different H₂O₂ concentrations, showing increased MSD at higher fuel concentrations. (iii) Confocal microscopy image with the corresponding motion trajectory of the coacervate inside a GUV at the fuel concentration yielding the highest MSD. Scale bar: 10 µm. Adapted and reproduced with permission [160]. 2022, American Chemical Society.

Although significant conceptual progress has been made in forming fuel‐driven transient membraneless coacervates, where fuel regulates both the formation of the coacervate and its temporal functions, major challenges remain for membrane‐bound artificial cells [111, 112, 137, 161]. Implementing energy‐dissipative processes inside these systems is difficult due to limited compatible chemistries, restricted membrane permeability, selective molecular partitioning, and unwanted interactions between the compartment material and the assembling unit. These factors often disrupt the kinetic balance required for controlled assembly–disassembly cycles [162, 163]. Nevertheless, overcoming these limitations is essential for developing artificial cells that more faithfully capture the structure and dynamic functions of living systems.

A special type of adaptive behavior that requires energy input is motility. Cell motility is a fundamental and essential process in living organisms, crucial not only for normal physiological functioning but also for key biological events such as immune responses, wound healing, and tissue repair [164, 165]. Defects in cellular motility often lead to pathological conditions including tumor progression, developmental abnormalities, and neurological disorders. Therefore, recreating motility in artificial cells is important both from a fundamental perspective and for its potential biotechnological and medical applications [166].

Inspired by natural systems, artificial cell motility can generally be classified into two categories: free‐solution motility and surface‐adhered motility [4, 167]. In free solution, artificial cells can move by generating propulsion forces either stochastically through internal chemical reactions or directionally under external stimuli such as electric fields, magnetic fields, light, or interfacial tension gradients (Marangoni flow) [168, 169, 170, 171, 172]. Surface‐guided motion is achieved when artificial cells adhere to a substrate and utilize either natural protein‐based cytoskeletal machinery or DNA nanotechnology to generate traction forces. In both cases, achieving motility requires asymmetry in the propulsion mechanism within the compartmentalized artificial cell [101, 173, 174, 175].

An example of stochastic movement was obtained via the construction of coacervate artificial cells which were powered by enzyme‐fueled chemical propulsion [176]. In this design, two independent enzymatic motors, urease and catalase, were conjugated to the coacervate membrane via click chemistry. These enzymes catalyze the decomposition of urea to ammonia and the conversion from H₂O₂ to O₂, respectively, generating thrust for movement. Because the membrane was fluidic, the enzymes underwent lateral diffusion, creating a stochastic and dynamically changing spatial distributions and patches. Optimal performance was observed at intermediate enzyme densities, which provided an ideal balance between stochastic redistribution and the formation of sufficient transient asymmetry to drive persistent propulsion.

Motile artificial cells can also be created by static positioning of motors on the cell surface. This was achieved by decorating membranized coacervates with bowl‐shaped polymer nanovesicles which were coated with gold nanoparticles (Figure 6B) [159]. Gold nanoparticles are responsive to NIR light, leading to a plasmonic photothermal effect. The anisotropic heating that results lead to a thermal gradient, inducing motility in the artificial cells. The motile behavior of the coacervates was tuned by varying the surface distribution of nanomotors from spotted, to patchy, to fully covered configurations or by altering the coacervate size and laser power density. Among these conditions, the highest propulsion speeds were obtained for coacervates with patchy nanomotor distributions and smaller diameters, owing to maximal structural anisotropy and reduced overall mass. Apart from light and chemical fuels, an oscillating AC electric field has also been used to move membraneless coacervates [177].

Another important form of motility frequently observed in living cells is intracellular transport, which includes cargo trafficking, organelle motion, and vesicular transport—processes essential for maintaining cellular function. These movements occur either by passive diffusion or through active, directed (ballistic) transport mediated by molecular motors moving along the cytoskeletal filament track. Zhan et al. demonstrated this type of intracellular cargo transport inside a GUV by engineering a DNA‐based artificial cytoskeleton [16]. Using a burnt‐bridge mechanism powered by RNase H, they guided vesicles and nanoparticles along DNA filaments within the compartment, closely mimicking natural cytoskeletal dynamics. A recent study by our group investigated the diffusive autonomous motion of an active coacervate microparticle inside a GUV (Figure 6C) [160]. Catalase‐functionalized coacervate microparticles were encapsulated within GUVs of varying diameters (10–50 µm) to systematically analyze their dynamics under confinement. In the presence of H₂O₂ fuel, the particles exhibited self‐propulsion driven by a self‐diffusiophoretic mechanism, in which catalytic decomposition of H₂O₂ creates local solute gradients that generate osmotic imbalances, resulting in particle motion. However, confinement markedly reduced particle motility compared to free movement in bulk solution. Moreover, the motion became more pronounced as the confinement size decreased. These results indicate that decelerating hydrodynamic effects, particularly increased viscous drag near the boundary, dominate the motion dynamics of chemically fueled microswimmers under three‐dimensional confinement.

4. Intercellular Communication

Across all kingdoms in nature, communication between cells is a fundamental feature of life and drives essential biological processes that underpin growth, organization, survival, and evolution [178]. For example, bacteria use quorum sensing via diffusible molecules to coordinate their proliferation and differentiation [179], fungi rely on pheromone‐mediated communication for mating [180], and mammalian cells leverage both diffusible chemical molecules and contact‐dependent signaling to regulate processes including immune responses [181, 182], tissue repair and differentiation [183], organ responses [184], and neurotransmission [185]. By dynamically sending and receiving molecular information, cells are able to monitor their population size and the presence of cooperating or competing species, and give rise to collective behaviors that cannot be accomplished by individual isolated cells [178, 186]. For multicellular organisms these processes are key for cells to understand their role, enabling the formation of complex functional tissues composed of specialized subpopulations of cells.

As with living cells, the ability to exchange information is crucial for the adaptation of artificial cells to dynamic environments. In recent years, artificial cell research has progressively turned toward the engineering of communication processes that enable artificial cells to interact, cooperate, and compete with one another via chemical signaling [186, 187]. Implementing such communication modes provides well‐defined platforms for studying communication principles and paves the way towards artificial cell consortia capable of performing higher‐order collective tasks. Furthermore, artificial cells capable of interfacing with living cells could be employed to modulate, perceive and respond to cellular behaviors, thereby offering potential for tissue engineering, drug delivery and smart biomaterial development applications [1, 188].

A key advantage of coacervate‐based artificial cells lies in their ability to sequester a wide range of (bio)molecules, among which small molecules, proteins, and nucleic acids [28, 97, 189, 190, 191]. This sequestration can be driven by preferential partitioning [28, 192, 193], specific noncovalent interactions [119], crosslinking [7], or direct structural integration [192, 194]. At the same time, coacervates permit selective exchange of signaling molecules with their environment via diffusion, a feature that can be further regulated by semi‐permeable membranes around these droplets or via modulation of the affinity of these molecules for the coacervate microenvironment.

4.1. Communication Between Artificial Cells

Diffusible signaling is one of the most universal communication mechanisms in nature, involving the secretion of a signaling molecule by sender cells that can be perceived and processed by receiver cells harboring the cognate receptor to regulate downstream functions [186]. Such information exchange can be governed by small molecules [195], cytokines [196], growth factors [197, 198], hormones [199, 200], and neuropeptides [201], to control a diversity of processes such as cell proliferation, differentiation, tissue repair, immune system regulation, and cell metabolism. Artificial cells differ from their living counterparts in their lack of intricate signal processing machinery. Instead, their communication is typically mediated by signals that directly engage their target regulatory modules after traversing from sender to receiver compartment. A common strategy to mimic diffusion‐based signaling therefore involves the selective incorporation of reaction modules into separate artificial cells. In this design, one population generates a diffusible reaction intermediate that serves as a chemical messenger to elicit a response in another population [186, 202].

Communication has primarily been demonstrated using enzyme cascades in coacervate‐based artificial cells, frequently employing GOx and HRP as model enzymes. For example, an early study from our group established unidirectional signaling between two terpolymer‐stabilized coacervate populations through separate incorporation of GOx and HRP. GOx‐loaded coacervates served as sender cells that converted glucose into H₂O₂, which subsequently diffused to HRP‐containing receiver cells to drive the HRP‐mediated oxidation of Amplex Red to resorufin [29]. This study demonstrated an early design principle for unidirectional signaling pathways, in which receiver cells elicit their output only in response to a specific signal generated by sender cells. More sophisticated signaling systems have since been realized, as exemplified by a study of Gobbo et al. in which coacervates were functionalized with a ruthenium polyoxometalate (RuPOM) membrane that decomposes H₂O₂ into O₂ (Figure 7A) [203]. By separately loading GOx and HRP, artificial cell populations with parallel catalytic capabilities were constructed. H₂O₂ produced by a GOx‐loaded sender population served as shared input for two receiver populations: HRP‐loaded coacervates and RuPOM‐bound coacervates. Consequently, the two populations competed in processing H₂O₂ via distinct catalytic pathways, displaying spatially distributed and parallel biochemical communication originating from a single sender population.

FIGURE 7.

Communication between artificial cells. (A) (i) Schematic of communication among coacervates functionalized with glucose oxidase (GOx), horseradish peroxidase (HRP), or ruthenium polyoxometalate (RuPOM) catalytic membranes. GOx‐generated H₂O₂ is employed as signal for competing reactions in HRP (peroxidase activity) and RuPOM (catalase‐like activity) populations, producing fluorescent 2,3‐diaminophenazine or O₂, respectively. (ii) Brightfield‐confocal microscopy overlay of coacervates functionalized with GOx (green), HRP (red) and RuPOM (no fluorescence). Adapted and reproduced with permission [203]. 2020, Springer Nature. (B). (i) Schematic of a hydrogel vessel containing coacervates (CVs) decorated with GOx, HRP, and catalase (CAT) in the outer, middle and inner layers, respectively. (ii) GOx‐produced H₂O₂ serves as input for HRP‐mediated nitric oxide (NO) production. Excess H₂O₂ is removed by CAT‐CVs. (iii) Schematic showing NO generation via AND‐gate processing. Only dual substrate input gives rise to a distinct NO output. (iv) Measurement of H₂O₂ and NO using colorimetric reactions (ABTS oxidation, and Greiss reagent) upon dual‐substrate input. A‐D designate the positions indicated in iii. Adapted and reproduced with permission [204]. 2022, Springer Nature. (C) (i) Schematic of DNA‐mediated protein shuttling between coacervates. Sender coacervates are loaded with His‐tagged yellow fluorescent protein (YFP) via a single‐stranded DNA (ssDNA) strand, whereas receiver coacervates are functionalized with NTA‐amylose (NTA‐am) and Ni²⁺. Addition of displacer strand triggers YFP release, resulting in its uptake into the receiver population. (ii) The confocal microscopy images show YFP transmission from sender (blue) to receiver (red) cells with the quantified fluorescence intensities (iii). Scale bar: 20 µm. Adapted and reproduced with permission [205]. 2022, Wiley‐VCH. (D) (i) Schematic of DNA‐mediated coacervate clustering. DNAzyme‐catalyzed cleavage of input palindromic loop DNA (subs‐4) results in the release of Cy5‐ssDNA, which substitutes Atto488‐ssDNA at the shell via DNA strand displacement (DSD) and changes the color from green to magenta. Clustering is mediated by exposure of the palindromic (p) sequence, inducing artificial cell crosslinking via palindromic duplex formation. (ii) Confocal microscopy images of Atto488‐ssDNA‐labeled artificial cells before and after the addition of subs‐4, resulting in cluster formation after 600 s. Scale bar: 5 µm. Adapted and reproduced with permission [206]. 2022, Springer Nature.

Living cells respond to stimuli by generating adaptive responses, with dynamic environments requiring sophisticated interpretive mechanisms to evoke appropriate outputs. The engineering of artificial cells capable of processing signals therefore represents another challenge. Liu et al. took a first step in addressing this by developing an integrated coacervate system capable of logic‐gate processing (Figure 7B) [204]. In their design, membrane‐bound coacervates were functionalized with GOx, HRP, or catalase, and embedded in concentric hydrogels to form a vessel capable of modulating nitric oxide (NO) production. Only the simultaneous exposure to glucose and hydroxyurea induced NO production via AND‐gate processing, which inhibited blood clot formation and platelet activation in plasma and blood samples. This example illustrates that the incorporation of signal‐processing is crucial toward creating more advanced artificial cell ensembles that better mimic the complex signaling networks of living cells and lay the foundation for future functional materials with information‐processing capabilities.

A limitation of diffusion‐based signaling is signal dilution, which can restrict the communication range. Also, in living systems, molecules are commonly present at low concentrations and their effects must be amplified to achieve long‐distance signaling [207]. To emulate this, our group implemented a signal amplification strategy based on allosteric enzyme activation [208]. Sender cells contained an enzyme that converted ATP to AMP, whereas a receiver cell population was engineered to enclose a glycogen phosphorylase (GP)‐initiated enzyme cascade that generated fluorescent NADH as output. The AMP produced by sender cells diffused to the receivers and allosterically activated the GP‐initiated cascade, enabling a single equivalent of AMP to generate ten equivalents of NADH. Alternatively, long‐distance signaling can be achieved by transporting signaling molecules via flow, akin to hormone transport in circulatory systems. This was mimicked by Liu et al. using coacervates immobilized in hydrogels [209]. In their system, coacervates containing TiO₂/Ag nanoparticles generated H₂O₂, which was transported by a unidirectionally applied flow to distant receiver coacervates. DNAzymes sequestered in the receiver coacervates converted the H₂O₂ into fluorescent resorufin to allow for visualization of the propagating signaling front.

Beyond enzyme cascades, the use of DNA nanotechnology has also been gaining traction as a platform for programmable artificial cell communication. Pioneering work by the De Greef group reported proteinosomes that could engage in diffusion mediated signaling using ssDNA signals via DNA strand‐displacement (DSD) circuits, formed by conjugating biotinylated ssDNA sequences to streptavidin‐loaded proteinosomes [25]. The ssDNA strands diffused through the proteinosome membrane, enabling input‐output exchange that supported multi‐step cascades, Boolean logic operations, bidirectional feedback, and RNA‐based signaling when combined with RNA transcription and CRISPR‐Cas‐based detection [26].

In the realm of coacervate‐based artificial cells, DNA nanotechnology has similarly been integrated. Our group demonstrated DSD‐mediated communication between coacervates using supramolecular nanoscaffolds [97], and more recently in conjunction with protein‐mediated signaling (Figure 7C) [205]. This is particularly relevant as the regulated uptake and secretion of proteins represents one of the most ubiquitous and crucial modes of communication among living cells in nature [210, 211]. As a first step towards mimicking such events, a His‐tagged Yellow Fluorescent Protein (YFP) was functionalized with ssDNA handles via click chemistry and sequestered into positively charged coacervates following hybridization to a longer corresponding uptake ssDNA strand. The addition of a releaser strand displaced the ssDNA‐conjugated YFP, triggering YFP release and its subsequent sequestration into a Ni²⁺‐NTA‐modified receiver population via complexation with its His‐tag. Although conceptually demonstrated using a fluorescent protein, this platform should be adaptable to other proteins or enzymes with diverse functionalities to afford more sophisticated outputs in future studies.

In a different study, the Walther group engineered DNA‐based coacervates containing DNAzymes that processed DNA inputs into ssDNA signals capable of directing communication and adaptive processes via DSD reactions (Figure 7D) [138]. Outputs included dynamic remodeling of coacervate shells via the replacement of original shell strands by enzymatically produced ones, as well as the assembly of coacervate‐prototissue clusters via multivalent DNA interactions. Furthermore, ssDNA signals could be transmitted from active sender cells to inactive receiver cells to drive the assembly of hetero‐prototissue clusters. This example shows that besides simple transmission of signaling molecules, dynamic changes in artificial cell populations, such as clustering, can also be modulated.

Together, these examples highlight that while enzyme cascades constitute the majority of studies on artificial cell communication, DNA‐based systems also show promise to engineer programmable communication. Systems involving conditional signal amplification or directed transport can both enhance the communication range and increase signal detection sensitivity. Moreover, integration of signal processing and basic computation capabilities showcases the potential of cooperative artificial cell consortia toward more emergent, life‐like functions. Despite these advances, outputs remain relatively limited by relying on fluorescent molecules to confirm signaling functionality. While conceptually invaluable, such systems do not approach natural processes, which involve sophisticated events such as protein expression and secretion, and phenotypic change. Expanding output diversity is therefore key towards integrating advanced biomimetic functionalities. This could be pursued via TX‐TL circuits, as these have been leveraged to express proteins with diverse functions in artificial cells [14, 212, 213, 214, 215]. While extending TX‐TL systems to coacervates remains a challenging frontier for future research, its successful implementation could empower sophisticated protein and nucleic acid‐mediated communication, or even enable integrated production of cytoskeletal or cell‐division machinery to afford communication‐mediated adaptivity.

4.2. Stimulus‐Responsivity and Control Over Communication

Many coacervate‐based artificial cells initiate communication upon external addition of diffusible signaling molecules, generally providing limited control over when and where communication is activated. This approach is further constrained to molecules that can cross (semi‐permeable) membranes and partition into the coacervate phase. By contrast, living cells adaptively coordinate communication across membranes via transmembrane receptors, coupled to interconnected signaling networks that provide precise control while minimizing spurious activation. Emulating such levels of regulation in artificial cells holds great promise for stimuli‐responsive and spatiotemporal control. Yet, as fully recapitulating the regulatory complexity of living cells is still elusive, alternative strategies are investigated that allow artificial cells to be activated under specific, user‐defined conditions.

Recent progress has witnessed the integration of synthetic receptors in vesicle‐based artificial cells. Examples include DNA‐peptide hybrid receptors that sense extracellular pH and trigger intracellular signaling via conformational changes [216], and DNA‐origami transmembrane structures that rearrange DNA‐based cytoskeletons upon ligand‐induced receptor‐clustering [217]. Beyond DNA‐based receptors, the Zelikin group integrated synthetic self‐immolative linker receptors that release messengers to activate enzymatic cascades and transcription in artificial cells [218, 219]. Together, these studies demonstrate how synthetic receptors can confer adaptive responses to external stimuli without relying on signal diffusion across the membrane. Future development of receptors that recognize complex biologically relevant cues, especially when coupled to synthesis or secretion‐based outputs, could enable advanced signal sensing and controlled intercellular communication.

An alternate strategy to control artificial cell communication exploits responsivity to physical stimuli, which can be applied with spatiotemporal resolution and readily cross compartments. To date, light has been the most dominantly employed physical stimulus in artificial cell research, owing to its advantages of bio‐orthogonality, tunability, reversibility, and remote‐controllability [220]. For example, the Wegner group employed light‐switchable protein‐protein interactions to induce reversible binding between vesicle‐based artificial cells, reducing intercellular distance and enabling communication [221]. This strategy was used to mimic predator‐prey behavior via light‐regulated Ca^{2 +} exchange, which induced membrane disruption in receiver cells harboring a Ca²⁺‐dependent phospholipase, and was later extended to light‐controlled bidirectional signaling using GOx and HRP [222]. Related approaches have adopted light as direct signaling modality [223], or as activating trigger to regulate connexin‐nanopore signaling [224], DSD reactions [225], and TX‐TL‐mediated production of enzymes and pore‐proteins to facilitate communication [226].

Light‐controlled communication has also been implemented in coacervates, particularly via incorporation of light‐switchable azobenzene moieties. Early work used light‐induced coacervate disassembly to release ssDNA signals that could be taken up by receiver populations [227], and similar designs were later integrated into light‐activatable systems capable of signal processing via Boolean logic operations [228]. Although these studies showcase the utility of light as reversible stimulus, they rely on coacervate disassembly and reassembly, which limits their cell‐like character. In contrast, Zhao et al. recently reported a light‐controlled communication system that preserves coacervate integrity by employing DNA Y‐motifs functionalized with azobenzene‐tethered ssDNA signaling strands (Figure 8A) [229]. The ssDNA signal was initially hybridized in sender coacervates, while a corresponding sequence in receiver coacervates was blocked by a hairpin. Upon UV irradiation, ssDNA signals were released and selectively internalized by the receiver coacervates via DNA hybridization, whereas subsequent exposure to visible light enabled bidirectional communication via transmission of ssDNA back to the original senders.

FIGURE 8.

Light‐controlled communication between artificial cells. (A) (i) Schematic of light‐activated bidirectional signaling between coacervates, mediated by base‐pairing of ssDNA signal and DNA sticky ends in both populations. UV and visible light reversibly modulate the accessibility for base‐pairing of the signal to sticky ends, thereby switching the direction of signaling. (ii) Confocal microscopy images and quantified fluorescence intensity of directional transport of FITC‐labeled ssDNA (green) from sender coacervates to receiver coacervates triggered by UV light (first and second panel) and directional transport of FITC‐ssDNA back to sender coacervates induced by vis light (third panel) at 30°C. Adapted and reproduced with permission [229]. 2022, Wiley‐VCH. (B) (i) Schematic of light‐activated enzymatic protein transfer between two coacervate populations. Sender coacervates, functionalized with Ni^{2 +}‐NTA amylose, are loaded with His‐tagged superfolder green fluorescent protein (sfGFP)‐cRaf cargo and light‐activatable TEV protease (LaTEV‐His). Irradiation using 365 nm light uncages the active‐site cysteine of LaTEV, triggering cleavage of the His‐tag linkage and release of the cargo. The released fusion protein is then selectively taken up by receiver coacervates containing 14‐3‐3 protein via specific cRaf‐14‐3‐3 interactions. (ii) Confocal microscopy images of sender coacervates loaded with sfGFP‐cRaf‐His (green) and LaTEV‐His, mixed with receiver coacervates containing DyLight650‐labeled T14‐3‐3‐His (cyan), shown at 0 and 12 h after 5 min of 365 nm irradiation. Cy3‐labeled quaternized amylose (Q‐Am; magenta) was added to visualize coacervates. Scale bar: 15 µm. Adapted and reproduced with permission [230]. 2025, Wiley‐VCH.

In a different study, our group recently demonstrated light‐controlled protein‐mediated communication between coacervate‐based artificial cells (Figure 8B) [230]. By incorporating Ni²⁺‐NTA‐functionalized polymers, a His‐tagged, light‐activatable tobacco etch virus protease (LaTEV), and a cargo protein bearing a cleavable His‐tag could be co‐sequestered within coacervates. Subsequent UV activation of LaTEV resulted in His‐tag cleavage of the cargo protein, followed by its release due to loss of affinity to the coacervate core. To demonstrate intercellular communication, sender cells were co‐loaded with LaTEV and a His‐tagged GFP fused to a phosphorylated cRaf domain, and combined with receiver cells containing cRaf binding partner 14‐3‐3. Upon UV‐irradiation, GFP‐cRaf was enzymatically released from sender cells and selectively internalized by receiver cells through protein‐protein interactions between 14‐3‐3 and the cRaf domain.

4.3. Communication with Living Cells

In the previous sections, communication was performed solely between artificial cells. Communication between artificial and living cells is however highly appealing, as living cells can be induced to exhibit more complex behaviors. Moreover, such interactions align with the premise that life‐like behavior is most convincingly demonstrated when living cells respond to artificial cells in a manner comparable to their responses to other living cell.

Early efforts toward communication with living cells focused on interfacing with prokaryotes, owing to their ease of cultivation and genetic tractability. Artificial cells have mostly been employed as translators in this context, converting environmental cues into readable stimuli via TX‐TL circuits involving quorum sensing. For example, seminal work by the Mansy group demonstrated vesicle‐based artificial cells that expressed a pore‐forming protein upon sensing theophylline, which acted as riboswitch. This triggered the release of IPTG, which induced GFP expression in E. coli [231]. This system was later adapted toward uni‐ and bidirectional communication via transmission of homoserine lactones, as well as via the use of UV‐activatable DNA templates for light‐activated control over communication [226, 232, 233]. Similarly, artificial cells have been employed for antimicrobial purposes. Ding et al. engineered vesicle‐based artificial cells that detected E. coli‐produced autoinducers and released an antimicrobial peptide, establishing a feedback loop that eliminated the bacteria [234]. Contact‐dependent approaches have also been demonstrated in this context. Zhao et al. developed coacervate‐based artificial cells coated with yeast cell wall fragments that engulfed and killed E. coli, thereby exhibiting phagocyte‐like behavior [235].

Artificial cells have also been engineered for communication with mammalian cells, albeit mainly unidirectionally. For example, the Städler group established a signaling system comprising two hydrogel‐based artificial cell populations enclosing metalloporphyrins. In the presence of H₂O₂, these catalysts produced resorufin and 7‐hydroxycoumarin, which could be internalized by Hep2G cells to generate dual intracellular fluorescence [236]. The same group later developed antioxidative artificial cells using a similar chassis that preserved the viability of Hep2G cells by scavenging peroxides from the extracellular environment [237]. Alternatively, intercellular communication modes that evoke cell death, rather than cell viability have also been reported. Zhang et al. reported an invasion‐defense interaction between coacervate‐based artificial cells and HepG2 cells [238]. In their system, GOx‐loaded coacervates served as invaders that impacted the viability of co‐cultured HepG2 cells by consuming glucose in the production of H₂O₂, whereas cells pre‐incubated with catalase withstood the oxidative challenge.

A particularly promising biomedical application for artificial cells lies in tissue engineering, by serving as adaptive senders of functional molecules such as growth factors and cytokines to support cell growth, function, and differentiation. Although artificial cells that dynamically respond to specific cues from surrounding cells or the extracellular matrix (ECM) are still elusive, exciting systems have demonstrated the possibility to direct cell growth and differentiation using artificial cells. To this end, Toparlak et al. designed liposomal artificial cells that influenced the differentiation behavior of murine neural stem cells (mNS) [213]. Their designs contained TX‐TL machinery and DNA templates encoding for brain‐derived neurotrophic factor (BDNF), LuxR, and pore‐forming protein perfringolysin O (PFO) that together operated as an AND‐gate, requiring both LuxR and an autoinducer for BDNF and PFO expression. Upon activation, BDNF was released, which was shown to induce mNS differentiation by promoting neurite outgrowth and synapse formation. Using a similar approach, Chen et al. recently developed implantable artificial cells in Matrigel that secreted recombinant fibroblast growth factor to promote angiogenesis in mice [239]. These examples highlight the potential of artificial cells to express and deliver functional proteins to modulate the behavior of nearby cells.

Apart from proteins, an alternative strategy is to establish communication by harnessing nucleic acids. For example, RNA exchange via vesicular and exosomal transport has been recognized as a universal regulatory communicative mechanism [240, 241]. Our group recently developed coacervate‐based artificial cells that encapsulated synthetic, siRNA‐loaded polymersomes functionalized with photocleavable groups, as exosome mimics (Figure 9A) [242]. Upon UV irradiation, charge switching induced polymersome expulsion, followed by uptake into mammalian cells where the RNA could be delivered to induce gene knockdown. The ability to spatiotemporally control RNA‐loaded vesicle‐based communication may lead to exciting tools to modulate cellular behavior and therapeutic tools to address disease.

FIGURE 9.

Communication between artificial and living cells. (A) (i) Schematic of coacervate‐based artificial cell with light‐responsive polymersomes for exosome‐inspired communication to living cells. UV irradiation triggers exosome expulsion from the artificial cells for siRNA delivery from artificial cells to living cells. The polymersomes feature UV‐cleavable nitrobenzyl‐caged amine moieties to enable charge‐switching. (ii) Confocal microscopy images of artificial cells with HeLa cells (green) 1 h after UV irradiation, showing internalization and Cy5‐labeled RNA delivery (red) by charge‐switchable polymersomes. Scale bar: 25 µM. (iii) Lysosomal‐associated membrane protein 1 (LAMP1) knockdown in HeLa cells following siRNA delivery, irradiated with UV light between 0 and 10 min. Psome represents a control of pre‐irradiated polymersomes. Data represent individual cells (n = 50). Significance was assessed using a two‐tailed student t‐test (*p ≤ 0.05, **p ≤ 0.01, or ***p ≤ 0.001.) Adapted and reproduced with permission [242]. 2025, Wiley‐VCH. (B) (i) Schematic of stimulable artificial cell designated to regulate mammalian cells (STARM) that can be activated by ions or light to induce aptamer ligand presentation and contact‐dependent communication with living cells. (ii) Schematic of mouse muscle regeneration model. Muscle damage is induced by daily cardiotoxin administration for 3 days, followed by injection of light‐activatable artificial cells and daily 3 min near‐infrared (NIR) irradiation. STARMs feature a fibroblast growth factor 1‐targeting aptamer (Apt^FGFR1) and DNA‐functionalized gold nanorods. Upon NIR treatment, gold nanorod‐mediated photothermal heating triggers DNA duplex melting, increasing permeability and enabling Apt^FGFR1 presentation for communication with muscle stem cells. (iii) Analysis of muscle satellite cell activation and myogenesis by immunofluorescence intensity quantification of myogenic markers Pax7, Ki67, MyoD, and MyoG. Data represent mean ± s.d. (n = 5 independent measurements). A two‐tailed unpaired t‐test was used for statistical analysis. Adapted and reproduced with permission [243]. 2025, Springer Nature.

Recently, the utility of nucleic acids was further extended towards establishing contact‐dependent communication between coacervate‐based artificial cells and mammalian cells. Wang et al. employed surface‐anchored DNAzymes that could be activated upon exposure to ions (e.g. Zn²⁺) or light to cleave DNA substrates, thereby altering membrane permeability and exposing functional ligands for targeted interactions with specific mammalian cell receptors (Figure 9B) [243]. This system was engineered to feature responsivity to external stimuli to elicit inducible contact‐dependent communication with mammalian cells, which was shown to modulate both signaling pathways and tissue formation via near‐infrared light‐induced regeneration of muscle tissue in a mouse model.

Taken together, the aforementioned studies showcase recent advances in the capacity of artificial cells to communicate with living cells, both in in vitro and in vivo contexts. However, additional factors have to be considered to translate artificial cells as robust entities in living systems. Besides biocompatibility and stability under physiological conditions [244], further aspects have to be taken into account for truly adaptive integration in tissue‐like environments. These include the engineering of bidirectional interactions and feedback between artificial and mammalian cells for collaborative behavior. In addition, considerations on the extracellular environment towards integration of artificial cells in tissue‐like settings are also essential, as previously reviewed in detail [245].

5. Summary and Outlook

Living cells operate with a level of complexity that far exceeds simple mechanistic descriptions, as individual processes are intricately integrated to yield specific functionality with spatiotemporal control. Over the past few decades, significant progress has been made in understanding and mimicking individual cellular characteristics and combinations of key cellular hallmarks through bottom‐up engineering approaches in artificial cell research. Advances in soft matter science, analytical techniques, and design strategies have endowed artificial cells with increasingly life‐like levels of structural and functional complexity. In this review, we have discussed recent advances in artificial cell research in which multiple cellular functions are integrated within a system, with a particular focus on coacervate‐based systems, due to their close resemblance to the crowded, heterogeneous cytoplasmic environment of living cells. In particular, we have discussed key design strategies illustrating how progress in structural mimicry has enabled enhanced control over artificial cell functionality, followed by developments that couple energy input with metabolism, growth, and adaptive or transient behavior. Finally, we have highlighted emerging communication mechanisms that allow artificial cells to sense, respond to, and interact with their environment.

Despite these advances, artificial cells still remain at a highly primitive stage compared to living cells. On all three topics discussed much progress is to be expected. Researchers have only just begun to explore the intricate relation between (transient) structure and function as found in living cells to create complex, multifunctional systems capable of dynamic, life‐like adaptiveness. Rudimentary organelles, cytoskeletons, and membranes have successfully been developed. The challenge is to connect the achievement on this level of structural mimicry to functional output. A wide range of membrane components has been reported in literature, each offering distinct advantages and limitations in terms of substrate permeability, mechanical stability, and membrane fluidity. However, achieving high specificity and selectivity in transport and regulation remains a major challenge. Moreover, critical cellular phenomena such as membrane self‐regeneration and its coupling to internal processes including growth and division are still exceedingly rare in artificial cell systems. Substantial effort must be devoted into developing systems in which membranes and internal cellular functions engage in active cross‐talk, allowing membrane properties to be autonomously regulated in response to internal metabolic or mechanical cues. Beyond membranes, the cytoskeleton plays a critical role in maintaining cellular architecture and coordinating intracellular dynamics. Despite notable progress, a significant gap remains in the development of reversible, adaptive cytoskeletal systems that are fully integrated with membrane dynamics and internal functionality. Achieving dynamic control over cytoskeletal organization, where structure can be modified in response to functional demands, will be a key step toward the realization of intelligent, adaptive artificial cells with enhanced functional efficacy. Finally, the presence of organelles allows the execution of multiple processes in an artificial cell system which are compatibilized due to the local organelle microenvironment. By connecting the formation of these organelle structures with the functions they need to perform, self‐regulated and autonomous processes can be achieved, in line with the performance of living cells.

One of the crucial functions artificial organelles need to execute is energy supply. The future development of artificial cells necessitates a transition from simple, passive microreactors to fully autonomous, self‐sustaining active systems achieved through the integration of complex metabolic networks with active transport machinery. Central to this evolution is the incorporation of internal energy‐regeneration modules that can autonomously replenish biofuels and regulate internal biomachinery under far‐from‐thermodynamic‐equilibrium conditions. Such capabilities may be realized through the construction of artificial organelles, such as chloroplast or mitochondrion analogues, or through the integration of synthetic–living biohybrid modules, which have recently gained significant attention due to their potential for long‐term sustainability compared to passive energy input strategies. Moreover, bottom‐up design approaches render artificial cells highly customizable, enabling the implementation of advanced metabolic architectures that extend beyond natural biochemistry. Examples include non‐natural pathways such as the CETCH cycle for enhanced carbon fixation, as well as programmable waste‐management and nutrient‐recycling systems. Further integration of autonomous metabolic networks and energy‐flux modules with genetic information processing, in a coordinated manner, may ultimately enable the emergence of Darwinian evolution in artificial cells, wherein metabolic performance is directly coupled to growth, replication, and reproductive success. In addition, energy autonomy and sustained out‐of‐equilibrium internal processes could allow artificial cells to transition from static adaptiveness to dynamic adaptation through the integration of sensing, information processing, and actuation. Such capabilities would enable artificial cells to autonomously navigate, communicate with their environment and with cellular communities, and respond appropriately to external stimuli.

Recent advances in the engineering of artificial cell have been substantial in expanding their communication capabilities, progressing from simple unidirectional signaling to rudimentary signal processing, bidirectional exchange and communication with living cells. Notwithstanding, artificial cell communication remains in its infancy compared to that of living cells. Communication between artificial cells is limited by narrow output modalities, motivating efforts to expand functionality via incorporation of more advanced protein‐ or DNA‐ or TX‐TL‐based machinery. By contrast, communication with living cells is mostly limited by unidirectionality and lacks selectivity. Future efforts should focus on stimulus‐responsive communication systems that preserve artificial cell integrity and afford specific cell targeting, for example via integration of contact‐dependent signaling modules that target cognate receptors and the use of on‐demand activating stimuli such as light. These developments are also crucial for the next phase in artificial cell research, namely population behavior. As in nature, cells hardly operate as single entities, but interact with other cells, from bacterial colonies to tissues. Mimicking this behavior, either with fully artificial cell‐based systems or with living and artificial cell hybrid communities represents the next challenge for the field. Establishing such intricate communication networks opens up many perspectives. For example, toward tissue engineering, we envision artificial cells with the capacity to monitor their environment, for instance via synthetic receptors, and release functional cargos in response to biochemical cues to support cell growth and differentiation. Achieving truly adaptive behavior will require reciprocal feedback loops, which could ultimately enable coordinated, higher‐order communication processes in artificial tissues and organoid‐like systems. If we succeed in reconstructing complex life‐like behavior within synthetic compartments in an integrative approach this will not only lead to a better fundamental understanding of the biophysical and biochemical principles governing living cells but will also profoundly advance biotechnology, targeted drug delivery, medical science, and materials science.

Conflicts of Interest

The authors declare no conflicts of interest.

Biographies

Arjan Hazegh Nikroo obtained his M.Sc. in Biomedical Engineering from Eindhoven University of Technology in 2022. He specialized in chemical biology and protein engineering, completing his thesis on the engineering of phospho‐regulated 14‐3‐3 scaffold proteins under supervision of Prof. Luc Brunsveld and Prof. Maarten Merkx. Currently, he is a Ph.D. candidate in the group of Prof. Jan van Hest at Eindhoven University of Technology and the Institute for Complex Molecular Systems. His research centers on the engineering of protein‐mediated communication between coacervate‐based artificial cells and living cells.

Angshuman Das obtained his Ph.D. in 2024 from the Jawaharlal Nehru Centre for Advanced Scientific Research, under the supervision of Prof. Subi J. George. His doctoral research focused on the development of bio‐inspired strategies to modulate the structure and function of synthetic supramolecular polymers. He is currently a postdoctoral researcher at Eindhoven University of Technology in the group of Prof. Jan van Hest, where his research centers on the development of dynamic, reversible cytoskeletal systems to control artificial cell morphogenesis.

Madelief A. M. Verwiel obtained her B.Sc. in Molecular Life Sciences at Wageningen University and Research and her M.Sc. in Molecular Sciences at Radboud University Nijmegen. During her internship in Prof. Jan van Hest's Bio‐Organic Chemistry group, she explored synthetic signaling in coacervate‐based artificial cells. In December 2022, she joined the same group as a Ph.D. candidate to advance research on coacervate systems. Her interests include liquid‐liquid phase separation, membraneless organelles, compartmentalization, and cellular architectures.

Jan C. M. Van Hest obtained his Ph.D. from Eindhoven University of Technology (1996) in Macro‐Organic Chemistry with Prof E. W. Meijer. He worked as a postdoc with Prof D. A. Tirrell on protein engineering. In 2000, he was appointed full professor in Bio‐Organic Chemistry at Radboud University Nijmegen. As of September 2016, he holds the chair of Bio‐Organic Chemistry at Eindhoven University of Technology. Since May 2017, he is scientific director of the Institute for Complex Molecular Systems. The group's focus is to develop well‐defined compartments for nanomedicine and artificial cell research.

Data Availability Statement

Data sharing is not applicable to this article as no new data were created or analyzed in this study.