One approach towards the creation of bottom-up synthetic biological systems of higher complexity relies on the subcompartmentalization of synthetic cell structures using artificially generated organelles — roughly mimicking the architecture of eukaryotic cells. Organelles create dedicated chemical environments for specific synthesis tasks — they separate incompatible processes from each other and help to create or maintain chemical gradients that drive other chemical processes. Artificial organelles have been used to compartmentalize enzyme reactions, to generate chemical fuels via photosynthesis and oxidative phosphorylation, and they have been utilized to spatially organize cell-free gene expression reactions. In this short review article, we provide an overview of recent developments in this field, which involve a wide variety of compartmentalization strategies ranging from lipid and polymer membrane systems to membraneless compartmentalization via coacervation.
One of the major transitions  during the evolution of life on our planet must have been the emergence of cellular life forms — usually imagined as membrane-encapsulated chemical reaction networks capable of self-replication and Darwinian evolution  — which later evolved into the prokaryotic life forms known today. Eukaryotic cells — cells with a nucleus — have a considerably more complex internal structure involving extended membrane systems and compartments with specialized functions — so-called organelles. Such cell-internal compartments provide a large variety of beneficial functions, which supposedly also contributed to the emergence of ever more complex multicellular life forms.
While a large number of researchers in bottom-up synthetic biology have previously attempted to create protocells modeled after the imagined first simple cells [3,4], in recent years, scientists also started to work on subcompartmentalization — compartments inside of compartments — and the division of labor between artificial organelles in order to create ‘eukaryotic’ cell mimics with possibly enhanced capabilities (Figure 1).
Subcompartmentalization of artificial cells using synthetic organelles.
The origin and purpose of organelles in biology
Some of the potential benefits of having cells with specialized subcompartments may be deduced from the role of organelles in biology. In fact, only eukaryotic cells are known to harbor membranous organelle structures, and these are assumed to have arisen after the formation of the first eukaryotic cell from an archaean cell with a bacterial endosymbiont . The bacterial endosymbiont evolved into what today are the mitochondria, while the cell nucleus is supposed to have formed as a protection of the archaean genome against invasion by Group II self-splicing introns from the bacterium . This inevitably afforded the separation of transcription and translation processes in eukaryotic gene expression.
With few exceptions, eukaryotic cells are much larger and much more complex than prokaryotic cells, and it is assumed that this complexity is facilitated by the presence of organelles and specialized subcompartments. Among others, bioenergetic arguments have been put forward that relate the size of the eukaryotic cell with the presence of energy harvesting and energy-converting organelles . As bioenergetic processes such as the generation of ATP from a proton motive force involve membrane-bound molecular machines, the availability of ATP scales with the membrane area available — having internal organelles and membrane structures effectively increases the area for such processes and thus energy supply.
These examples (mitochondria, nucleus) demonstrate two of the major beneficial roles of organelles, namely the creation of dedicated intracellular compartments that carry out specialized biochemical functions, and the separation and protection of incompatible processes from each other. Organelle compartments can have quite unique chemical environments, e.g. with specific pH values, ion content, hydrophobicity, etc. From a physical point of view, membranes divide space into separate domains, allowing the establishment of chemical gradients. They can also act as two-dimensional scaffolds, which enhance interactions of membrane-bound binding partners or simply prevent loss of compounds via diffusion.
One should mention that there are other ways to compartmentalize biochemical components than via membrane enclosure, which are of interest for bottom-up biology. Even though prokaryotes do not have a nucleus, their chromosomal DNA is still organized via plectonemic supercoiling and histone-like proteins . Bacteria also feature a wide variety of protein cages such as carboxysomes, propanediol utilization (PDU) or ethanolamine utilization (EUT) microcompartments , which serve as enzyme-packed reaction containers that carry out chemical reactions such as carbon fixation, and propanediol or ethanolamine utilization (as their name indicates), respectively. Other protein cages such as ferritin serve as iron storage compartments, while encapsulins carry cargo proteins associated with oxidative stress such as peroxidases. Protein capsules have been variously engineered for synthetic biology applications such as the spatial organization of enzyme cascades . For instance, capsids of bacteriophages were engineered as self-assembling nanocompartments for enzymatic production of hydrogen , and several groups have recently demonstrated the synthesis of bacterial encapsulins as compartments inside eukaryotic cells such as in yeast  or in HEK293 cells .
Furthermore, in recent years, it has become increasingly clear that membraneless compartmentalization via coacervation or liquid–liquid phase separation plays a major role in the creation of highly dynamic intracellular chemical environments .
Organelles for powering artificial cells
As mentioned, one of the major roles played by eukaryotic organelles is to provide energy for biochemical processes in cells. Similar machinery for energy supply (or conversion) will also be needed for the realization of autonomously operating artificial cellular structures.
There have been many attempts to reconstitute photosynthetic complexes in liposomes and also polymer-based compartments (for a dedicated review, see, e.g. ). Most of the work focused on the use of bacteriorhodopsin (BR) or proteorhodopsin (PR), and of photosystems (PS I and II) extracted from plants, algae or cyanobacteria as light-driven proton pumps. In order to produce ATP from the resulting transmembrane proton gradients, ATP synthase has to be incorporated into the membrane of the same compartment in the right orientation (Figure 2a).
Functions of synthetic organelles.
One major step towards photosynthetic organelles for synthetic cells was recently taken by Lee and co-workers . They reconstituted recombinantly expressed PR (from Gamma proteobacterium) and PS II (from Spinacia oleracea) as well as ATP synthase (directly purified from Bacillus pseudofirmus) into small phospholipid vesicles and used them as ‘energy modules’ inside of giant unilamellar vesicle protocells. Among others, they demonstrated that illumination of the protocells with light results in ATP production, which in turn can be used to drive actin polymerization.
As the reconstitution or de novo creation of energy harvesting and ATP generating processes is a cumbersome and challenging endeavor, other synthetic biologists have directly introduced organelles of biological origin for that purpose — essentially resulting in ‘hybrid’ artificial cells. For instance, Kumar and co-workers extracted chloroplasts from spinach and utilized the fact that due to their negative surface potential they were readily taken up by positively charged coacervate droplets made from poly(diallyldimethylammonium chloride) and carboxymethyl-dextran . Upon light irradiation, the chloroplasts were shown to drive a reduction reaction in these membraneless protocells (see also below). Göpfrich and co-workers recently encapsulated mitochondria isolated from living cells into liposomes, where they could potentially also be used as ‘power supplies’ .
In this context, it is interesting to note that inverted membrane vesicles (IMVs) produced during the generation of cell extracts have been found to improve the yield of cell-free protein synthesis. IMVs contain functional ATP synthases , which continue to produce ATP [19,20]. For this reason, the IMVs contained in (some) cell extracts could also be interpreted as ATP-generating organelles, which could play a role when encapsulating these extracts into liposomes.
Enzyme reactions in synthetic organelles
There have been various attempts to create organelle-like compartments within artificial cells to achieve or improve functions other than energy harvesting or ATP generation, mainly with the goal to spatially organize enzymatic or gene expression reactions . In this context, liposome-in-liposome [22,23], polymersome-in-polymersome  or colloidosomal  structures were utilized, often in combination with gels, coacervates , nanoparticles and other means that provide compartmentalization or spatial localization (cf. Figure 1).
Van Hest and co-workers have created a variety of polymer-based compartments for the encapsulation of enzymes in order to improve the performance of multienzyme reactions (Figure 2b), and in this context, separated incompatible enzymes and reaction environments from each other . For instance, they created cell mimics by encapsulating artificial polymer-based organelles into larger polymersomes, which were also filled with ‘cytosolic’ enzymes (phenyl acetone monooxygenase, PAMO) and reagents. One type of organelles contained the enzyme alcohol dehydrogenase (ADH), while the other contained lipase from Candida antarctica (CalB) or alcalase. These enzymes form a catalytic reaction cascade, in which a non-fluorescent precursor molecule is first converted by PAMO into a substrate for CalB/alcalase, which is further processed by ADH into an intermediate compound which spontaneously converts into the fluorescent product resorufin. Compartmentalization of the enzymes forces the reaction intermediates to cross the organelle boundaries and thus creates dedicated reaction environments for each step. Importantly, the enzyme alcalase is a protease, which can degrade the other members of the reaction cascade — correspondingly, the overall reaction was found to proceed more effectively when alcalase was contained in organelles.
A different route towards multicompartmentalized systems based on lipid bilayer membrane vesicles was taken by other groups. Already more than a decade ago, Stamou and Vogel demonstrated the encapsulation of small unilamellar vesicles (SUVs) inside of large unilamellar vesicles (LUVs), where the SUVs contained substrates for enzymes residing in the ‘cytosol’ of the LUVs . Using temperature-induced phase transitions, the substrates could be released from the SUVs and then further processed by the enzymes.
In a different approach, Ces and colleagues realized small assemblies of lipid vesicles, in which adjacent vesicles were connected to each other via lipid bilayers containing membrane pores . Distinct enzymes of a multienzyme cascade (leading from lactose over glucose and hydrogen peroxide to resorufin) were contained in separate compartments, and thus each step of the cascade proceeded isolated in a single compartment, while intermediate compounds were exchanged via the compartment boundaries. More recently, the same group demonstrated a system of nested vesicle bioreactors, in which giant POPC vesicles hosted smaller vesicles made from a mixture of DPPC and the photo-crosslinkable lipid 1,2-bis(10,12-tricosadiynoyl)-sn-glycero-3-phosphocholine (DC89PC). Photocrosslinking of DC89PC leads to pore formation in the small vesicles, which could be used to release a fluorogenic substrate for β-galactosidase, which was also contained inside the giant vesicle .
Spatial organization of cell-free gene expression
Apart from multienzyme systems, several groups have attempted to spatially organize gene expression reactions using dedicated artificial organelles. Huck and co-workers immobilized gene length DNA molecules onto thiol-modified hyaluronic acid (HA) and further cross-linked it with poly(ethylene glycol) diacrylate (PEGDA), resulting in DNA-loaded HA-PEGDA gel particles . These porous particles were encapsulated into water-in-oil emulsion droplets also containing a cell-free gene expression system. Experiments with fluorescent mRNA sensors and labeled ribosomes suggested that both transcription and translation reactions can take place inside of the gel beads. In a similar approach, Aufinger and Simmel immobilized different DNA species in agarose gel beads , which were co-encapsulated as organelles into emulsion droplets (Figure 2c). Beads loaded with transcription templates could act as sources of RNA molecules, which could be captured by gel beads containing DNA species partially complementary to the RNA. When including DNA-switchable riboregulators  in the 5′-untranslated region of the transcribed mRNAs, the mRNAs were only activated for translation inside of the capture beads. In this way, transcription and translation processes were separated into two different types of organelles, superficially mimicking the situation in eukaryotic cells.
Using gel-based compartmentalization involving positively charged clay particles and DNA templates, another type of eukaryotic cell mimic was prepared by Devaraj and co-workers . In their work, the DNA/clay hydrogels acted as ‘nuclei’ inside of artificial cells bounded by porous polymer membranes. The cells were loaded with genes coding for T3 RNA polymerases and for the transcriptional repressor TetR, which both were able to diffuse through the polymer membrane and thus influence gene expression also in other cells. Among other functions, they were able to demonstrate a simple ‘quorum sensing’ mechanism, in which gene expression was activated depending on the density of the artificial cells.
Rather than using gel-immobilized genetic templates, Huck and co-workers also demonstrated gene expression reactions in nested liposomal structures (termed ‘vesosomes’) , more similar to the structures used for compartmentalization of multienzyme systems. In their system, they were able to show distinct and spatially separated transcription and protein expression reactions inside of the artificial nuclei and the cytosol, respectively.
Introducing artificial organelles into living cells
Several researchers have adopted a stricter interpretation of the term ‘organelle’ in the context of synthetic biology and recently started to operate synthetic organelles inside of living biological cells, mainly with the goal to supplement naturally existing enzyme cascades. Hosta-Rigau and co-workers developed ‘capsosomes’, which were created via layer-by-layer assembly of fluorescent gold nanoclusters, various polymer layers and liposomes on the surface of silicon dioxide particles (which were dissolved post-assembly) .The liposomes were loaded with glucose oxidase in an inner layer and horseradish peroxidase in an outer layer, constituting a well-known enzyme cascade that can be used to convert glucose via hydrogen peroxide into the fluorescent compound resorufin. Remarkably, the capsosomes were readily taken up by macrophages, where they were able to utilize intracellularly available glucose.
In even more advanced work, van Hest and colleagues created biodegradable polymersomal enzyme reactors, which were loaded with catalase and surface-modified with cell-penetrating peptides (CPPs) . The CPPs promoted the take-up of the polymersomes by HEK293 cells and human primary fibroblasts, where they were used to shield the cells from reactive oxygen species (ROS). Remarkably, in the case of particularly sensitive primary fibroblasts derived from patients with isolated mitochondrial complex I deficiency, the artificial organelles protected the cells when challenged with externally supplied hydrogen peroxide. Supplementing absent or deficient cellular functions with enzyme-loaded synthetic organelles thus appears to be a promising perspective for applications in cellular therapy.
Compartmentalization by phase separation
Organelles have been traditionally viewed as membranous compartments found in eukaryotic cells. However, there are other means of compartmentalization that do not require membranes, which have recently received considerable interest for the realization of plausible protocell models as well as for the understanding of dynamic liquid-like organelles found in modern cells [39–42].
Positively and negatively charged macromolecules can associate with each other either to form solid precipitates or into dense, liquid-like droplets termed ‘coacervates’. The latter have been shown to be viscous and polymer-rich and capable of enriching other solutes to high concentrations .
Liquid colloidal assemblies have been famously proposed as protocell candidates by Oparin already in the 1920s [43,44]. While most protocell research later focused on membranous compartments , the finding that coacervate droplets can also form from prebiotically plausible small molecules such as nucleoside triphosphates and oligopeptides  has renewed the interest in membraneless compartments in this field. Since then coacervates have also been created from peptides and polynucleotides , they have been shown to support transcription and translation reactions , and also RNA-based catalysis [48,49]. Remarkably, membraneless liquid organelles have also been found in modern cells in the form of RNA granules, P granules, the nucleolus or the Cajal body [41,42].
Based on these insights, liquid phase-separated compartments are also of great interest for the creation of artificial organelles. Among the advantages of coacervates are their rapid formation  and the ease of exchange of molecules with the surroundings  (which in contrast with membrane compartments does not require dedicated membrane pores or transporters). It has also been demonstrated that the formation of coacervates can be reversibly controlled, e.g. using temperature  or (de-)phosphorylation .
As it has already been shown that coacervation can be combined with other types of compartmentalization, cell mimics comprising both membranous and membraneless organelles are thus easily conceivable . For instance, the Mann group has demonstrated the assembly of fatty acid membranes on coacervate droplets , while Keating and co-workers have shown interfacial assembly of liposomes on the surface of RNA-based coacervate droplets . As an alternative, van Hest and co-workers utilized triblock copolymers containing hydrophilic, hydrophobic and polyanionic components that covered biopolymer coacervates with a stabilizing membrane .
Conclusion and outlook
Over the past years, we have witnessed the emergence of ever more complex cell-mimicking chemical systems that involve organelle-like subcompartmentalization. In contrast with simpler artificial cellular structures, these systems are inspired by the structure of eukaryotic cells, which contain a wide variety of organelles that create specialized environments for specific chemical tasks. Accordingly, artificial organelles have been used to compartmentalize enzymatic cascade reactions, as power supply units for the generation of chemical fuels and for the spatial organization of gene expression reactions.
In this context, different types of compartmentalization using a wide variety of materials are utilized — from rather conventional lipid membrane systems over polymersomes to colloidal assemblies, gels and other membraneless confinements.
Most of the systems realized so far have only provided ‘proof-of-principle’ demonstrations. In the future, it will therefore be important to demonstrate that organelles can indeed improve relevant chemical processes and thus the performance of artificial cell structures.
From a synthetic biology perspective, organelles could also be interpreted as functional ‘modules’ with defined interfaces to their environment. Based on a well-characterized (and well-behaved) library of such modules, programming of synthetic biological systems at a higher organizational level could become possible — similar as in going from machine language to a more abstract level in computer programming. An obvious next step in this context will be the creation of different artificial cell types and their combination into multicellular structures [55–57]. In such structures — as in biological multicellular life forms-, organelles might play a crucial role, e.g. as powerhouses that carry out the bioenergetic processes required to continuously drive the systems.
One of the major challenges for the creation of larger systems will be the considerable variability that is typically observed in compartmentalized biochemical systems, which is caused, e.g. by the large variations in the numbers of active components or the effect of compartmentalization on enzyme activity [58–60]. Creation of multicompartmentalized systems will also result in context-dependent performance of organelles, depending on the specific chemical environment they experience in a particular setting . In order to engineer larger systems, strategies will therefore have to be developed to insulate components sufficiently from each other and to buffer away the detrimental effects of chemical context  and cell-to-cell (or organelle-to-organelle) variability.
If this program succeeds, it will result in sophisticated hierarchically and spatially organized chemical systems that could be useful in a wide variety of applications. Even though the complexity of the systems will most likely still be a far cry from those of living organisms, they should be of considerable interest for the creation of ‘life-like’ materials, for the improvement of cell-free bioproduction processes, and for potential cell therapeutic applications.
Cell mimics of higher complexity can be created using synthetic organelle structures
Synthetic organelles create dedicated chemical environments and help establish or maintain chemical gradients
Synthetic organelles are used for energy conversion processes, enzymatic synthesis and cell-free gene expression reactions
Organelles can be created with and without membranes, using lipids, polymers and colloidal assemblies
inverted membrane vesicles
large unilamellar vesicles
phenyl acetone monooxygenase
poly(ethylene glycol) diacrylate
small unilamellar vesicles
F.C.S. wrote the paper.
This work was supported by European Research Council [grant number 694410] and the Deutsche Forschungsgemeinschaft [TRR 235/P15].
The author gratefully acknowledges financial support by the DFG through TRR 235 (Emergence of Life) and the European Research Council [project AEDNA, grant number 694410].
The Author declares that there are no competing interests associated with this manuscript.