Realization of a functional artificial cell, the so-called protocell, is a major challenge posed by synthetic biology. A subsequent goal is to use the protocellular units for the bottom-up assembly of prototissues. There is, however, a looming chasm in our knowledge between protocells and prototissues. In the present paper, we give a brief overview of the work on protocells to date, followed by a discussion on the rational design of key structural elements specific to linking two protocellular bilayers. We propose that designing synthetic parts capable of simultaneous insertion into two bilayers may be crucial in the hierarchical assembly of protocells into a functional prototissue.
Synthetic biology applies principles and ideas from a wide range of disciplines for quantitative manipulation of cells for biotechnological and engineering applications (applied goal). Another aim is to recreate life-like systems for understanding the origins of life (intellectual curiosity). The two aims are not mutually exclusive; however, synthetic biologists can be broadly classified into two groups. The first group is programming cells for specific outputs, turning them into ‘intelligent’ bioreactors (applied goal) [1–5]. The second group is majorly concerned with top-down and bottom-up synthesis of a biological cell [6–8]; this serves both the applied and intellectual aims. The top-down approach of stripping a cell of its components one-by-one is cumbersome, and may alter a cell in unpredictable ways . The bottom-up methodology is more measured. It involves de novo construction of complex biological parts and systems [10,11], and reconstitution of purified cellular components to mimic cell behaviour [12,13]. In the present paper, we focus on the bottom-up construction of the protocell and prototissue in the context of our recent work on designing a communication link between artificial cells.
To understand how life began, functions and evolves, synthetic biologists seek to build artificial cells using a minimalist approach [14,15]. The approach is based on the premise that the prebiotic cell would have comprised only a few basic components: a membrane-enclosed aqueous compartment encapsulating some form of genetic material, i.e. RNA or DNA. Mimics of primitive cells have been achieved in the laboratory in the form of nanometre/micrometre-sized lipid and fatty acid vesicles. To recapitulate the structure and function of a living cell, an obvious next step is to increase the complexity of an artificial cell by reconstituting small portions of a biological cell in protocell models [16,17]. With the unprecedented increase in our knowledge of molecular biology and biochemical systems, it is now possible to entrap RNA/DNA, reconstitute simple proteins and build cytoskeletal assemblies in model systems, with their behaviour studied and compared with the existing data from other in vitro studies [18–22]. Beyond simple encapsulation, it has also been possible to express proteins in lipid-enclosed containers and thereby participate in complex bioreactions [23–25].
A model protocell would be considered alive if it were capable of growth and division [26–29]. In the absence of complex biological machinery, it has been surmised that physicochemical properties had an instrumental role in the spontaneous growth and division of these cells [30,31]. A key conceptual advance was the successful demonstration that RNA replication in fatty acid vesicles could be coupled to their growth [32,33] and division [34,35], even while allowing the passage of sugars and nucleotides across the vesicle bilayers [36,37]. Although, owing to their basic physical and chemical nature, fatty acid vesicles have been instrumental in developing a working model of a primitive cell, a mimic of the contemporary biological cell will need to have a coating of phospholipids. Sugawara and co-workers have taken the first steps in this direction by demonstrating the amplification of DNA in self-reproducing giant lipid vesicles .
In contrast with the progress made in modelling protocells, the development of prototissues has remained an unexplored field in synthetic biology. Prototissues are expected to exhibit emergent properties arising from the agglomeration of a large number of protocells. A number of interesting questions arise in such a scenario. What are the determining physical and chemical principles for constructing a functional prototissue? Can the construction of a prototissue be based on the same principles as a protocell? What features will confer bio-functionality on a prototissue?
The simplest form of an artificial prototissue is a network of lipid vesicles (nanometre or micrometre). Recently, the Luisi laboratory devised a way to self-assemble lipid vesicles [POPC (1-palmitoyl-2-oleoyl-sn-glycero-3-phosphatidylcholine) and sodium oleate] mediated by electrostatic attraction of polypeptides (poly-L-arginine or poly-L-lysine) . The properties of these so-called vesicle colonies were different from those of the isolated vesicles; the cellular membranes in the colonies were more permeable to negatively charged small molecules (ADP) and macromolecules (tRNA), and showed an increased tendency to fuse. It remains to be seen whether such tissue mimics can be assembled in a controlled manner in varied geometries.
Bayley and colleagues have developed a versatile droplet-based system, which is capable of forming bilayers in an oil phase [40,41]. Aqueous droplets in an oil/lipid mixture assemble a lipid monolayer; two such lipid-coated droplets form a bilayer at their connecting interface [40,41]. The bilayer thus formed is capable of hosting membrane pores and eukaryotic ion channels . The aqueous droplets are capable of sustaining transcription/translation processes, enabling the synthesis of membrane proteins and their continued insertion into the bilayer . Aqueous droplets in oil thus offer an attractive protocell model for it is likely that the initial cells were aqueous compartments floating in an oily mixture. This new system may offer unforeseen insights into the emergent behaviour of model cells in an assembly . Recently, it was shown that a 3D network of aqueous droplets in oil is capable of changing shapes if an osmotic gradient was established between the droplet layers . Contained within an oil droplet containing lipids, such droplet networks can also function in an aqueous environment . Without any protein or genetic matter, this is not a conventional tissue model; however, the work is an interesting attempt to mimic tissue behaviour. In contrast with lipid vesicles which are aqueous systems, and the aforementioned water-in-oil system, it has also been proposed that the first cells could have been oil droplets in an aqueous environment. A specific mixture of oils is capable of random motion, waste excretion and even fission–fusion cycle [47–50].
Replication, metabolism and hence sustained growth and evolution of a cell or tissue are largely dependent on the cell or tissue's capacity to effectively communicate with its surroundings. Sensing and communication (chemical, electrical or mechanical) confer functionality on cells and tissues. Membrane protein pores allow selective transport mechanism across phospholipid membranes. In the context of a minimal cell, spontaneously forming membrane pores have been shown to facilitate mass transfer across membrane boundaries from their surrounding bulk. Noireaux and Libchaber  expressed the staphylococcal pore toxin αHL (α-haemolysin) in vesicles that made pores in the vesicle bilayer and allowed a continuous and/or selective uptake of nutrients, an essential criterion for the sustenance of a minimal cell. It was found that bioreactions within liposomes could be sustained for several days upon making the protocells porous to nutrients. Transmembrane protein pores also allow electrical and chemical communication across bilayers formed of aqueous droplets in oil [41,46].
Similarly, a functionally robust artificial tissue based on such systems will require electrical and chemical communicative links between its protocells and the external milieu (Figure 1). In Nature, cells are often connected to neighbouring cells by proteins that span the cell membranes of adjacent cells (Figure 1A). Hence one of the necessary requirements for an assembly of protocells to function as a prototissue is a protein that can span two bilayers simultaneously such that a network of communication channels (connecting cell interiors) is established across the tissue (Figure 1B).
Inspired design of a prototissue
Constructing a gap junction mimic
The most notable example of proteins connecting two bilayers occurring in Nature is the gap junction channels . These channels are formed of two hemi-channels or connexons, each in the apposing bilayers of two adjacent cells (Figure 1A), forming a continuous conduit between the cytoplasm of the two cells . Several intermolecular hydrogen bonds between the extracellular loops of the connexons and a large contact area between the connexons create a tight seal and separate the pore of the gap junction channel from the extracellular region . Inspired by the gap junction design, we recently reported a gap junction mimic engineered using αHL .
αHL is a well-characterized membrane pore-forming toxin. The structure of αHL has been studied in atomic detail , it is easy to express, and native and denaturing purification procedures are well established . The αHL monomer reconstitutes as mushroom-shaped heptameric pores (α7) in cell membranes and artificial lipid bilayers [58,59]. The α7 structure has a flat cap surface and a transmembrane β-barrel (Figures 2A and 2B). It was thus surmised that two α7 units could be linked at their caps such that their transmembrane β-barrels pointed in axially opposite directions and the resulting dimer pore could span adjacent lipid bilayers . The dimer protein, (α7)2, would thus be analogous to the gap junction channel (Figure 2C).
Constructing pores that can link adjacent bilayers
As αHL does not contain any native cysteine residues, (α7)2 was constructed by linking the caps of two α7 units by mutating a surface-accessible cap residue, Lys237, to cysteine in each monomer (Figure 2B). TEM (transmission electron microscopy) micrographs showed side and top views of (α7)2 as elongated (average length ~19.6 nm and diameter ~8.4 nm) and ring-shaped particles (average diameter ~7.8 nm) respectively (Figures 2D and 2E). Upon reconstitution into lipid vesicles, single (α7)2 particles were seen as protrusions of average length ~17.0 nm (Figure 2F). (α7)2 particles spanning the bilayers of two closely placed liposomes (distance ~12 nm) were also observed (Figure 2G). The measured dimensions of individual (α7)2 particles were in agreement with the molecular model of the cap-to-cap dimer (Figure 2C).
The electrical properties of a protein pore, such as open pore conductance, ion selectivity and rectification, are indicative of its structure and charges of amino acids lining the pore . The unitary conductance of (α7)2 pores (570 pS) was ~66% of the conductance value of α7 (857 pS), implying that the (α7)2 pore was longer than α7. Simultaneous insertion of (α7)2 into two bilayers could also be detected electrically. Insertion of an attolitre liposome into a β-barrel of a single (α7)2 pore whose other β-barrel was already inserted in a planar lipid bilayer led to a consistent current blockade of ~23%.
The present design of (α7)2, held together by disulfide bridges, does not form a leak-proof structure as evidenced by the residual electric conductance of (α7)2 inserted in a planar lipid bilayer and a liposome. In future, the design and expression of (α7)2 need to be optimized to increase the yield of perfectly aligned cap-to-cap (α7)2 structures. We propose that linking the α7 caps with a larger number of disulfide linkages and introducing additional electrostatic interactions between the α7 caps may reduce the leak. The current (α7)2 structure could, however, be useful for, for example, electrical signalling and transfer of large molecules such as phosphate ions from one protocell to another within an artificial tissue.
Engineering novel double bilayertraversing pores
The α7 pore has been engineered to have several interesting properties such as blocker binding, rectification, size exclusion and gating. Previous studies have shown that αHL can withstand extensive mutagenesis and chemical modifications , and is robust in a broad range of pH  or temperatures , and in the presence of proteases  and chemical denaturants [65,66]. With α7 as its scaffold, (α7)2 could also be engineered for specific responses. Elements responsive to biological (fusion proteins or peptides responsive to analytes or proteolysis ), chemical (chemical moieties, metal ions ) or physical (light , temperature , osmotic pressure) cues could be incorporated in the dimer pore for triggering, for example, gating and pore insertion [71,72]. Hetero-(α7)2 pores can also be made of two α7 units engineered to have different characteristics. This can yield (α7)2 with unique pore properties, such as the ability to allow unidirectional flow of analytes. Such hetero-(α7)2 pores will be similar to the heterogenous gap junctions found in Nature .
Other membrane pores and strategies could also be explored to engineer pores similar to (α7)2. Dimers of the transmembrane proteins ClyA (cytolysin A), and an engineered version of MspA (Mycobacterium smegmatis porin A) with a truncated membrane-inserting region have been observed under certain conditions [74,75]. ClyA and the truncated mutant of MspA pores self-associated at the ends of their transmembrane region to form homodimers; however, these were incapable of inserting into bilayers. In an analogous way to (α7)2, ClyA, MspA and other pore-forming toxins could be linked cap-to-cap to be able to insert into two bilayers simultaneously. Polymer (chemical or biological) filters could be introduced at the cap regions of dimer nanopores for selective transfer of analytes or small molecules . Alternatively, to increase the functionality of dimer pores, hybrid pores could be constructed by linking two different transmembrane pores. It will be interesting to note whether the inter-cap domain leakage would change depending on the protein pair used.
Synthetic DNA and chemical pores offer advantages for the rational design of a precise system, which is difficult to achieve with protein pores. Recently, Langecker et al.  reported a synthetic mimic of αHL pore based on a DNA scaffold. DNA origami structures were made with geometry similar to αHL, i.e. cap-like and barrel-like structures; specific DNA strands functionalized with cholesterol enabled the insertion of the whole structure into a planar lipid bilayer and lipid vesicles. Electrical measurements confirmed the presence of an open channel (inner diameter ~2 nm) running through the length (~47 nm) of the structure. The DNA pore was capable of transporting single strands of DNA under an applied potential. Open-pore gating characteristics similar to those of protein nanopores were observed; a conductance of 1 nS was in good agreement with that of αHL. A precise methodology based on DNA origami is an attractive possibility to fabricate synthetic channels. A symmetric DNA pore analogous to the (α7)2 pore could be made with two rings of hydrophobic moieties capable of simultaneous insertion into two bilayers (Figure 2H). A DNA origami pore could be tailored to have different lengths and pore diameters. Higher mechanical stability of DNA will afford a better structural integrity to these pores. Site-specific functionalization may be more readily achieved on a DNA pore bypassing the hassles of protein expression. Use of RNA polymers or a combination of DNA and RNA could be beneficial in designing pores (or linkers) of varying flexural rigidity.
Ion channels with properties similar to those of naturally occurring protein nanopores have been synthesized of macrocycles, π-stack and metal-organic architectures, peptide bundles and nanotubes . Such rationally designed synthetic chemical channels could also be checked for linking two adjacent membranes (lipid bilayer or polymer).
Semi-synthetic cells: linking protocells to real cells
The α7 pore is capable of inserting into the cell membrane of mammalian cells . Metal-actuated α7 pores have been used to inject cryoprotectants in living cells . Similar to α7, engineered (α7)2 pores could be employed for specialized roles in biological cells or tissues. For example, liposomes pierced with (α7)2 can be used to deliver doses of analytes or drug molecules into biological cells (Figure 3A). This will require extensive protein engineering efforts to introduce features that are capable of response under triggered physical or chemical stimuli [72,80]. Pores designed on the principles of DNA origami may also be used to connect a cellular and an artificial bilayer surpassing the advantages of a protein pore.
Potential synthetic biology and nanobiotechnology applications of (α7)2 (dimer) pores
Beyond synthetic biology: using dimer pores for high throughput single molecule studies
Protein pores behave as filters (α7 has a cut-off of 3 kDa), making them useful experimental tools. Using α7-studded liposomes to facilitate diffusion of small molecules from the exterior bulk solution, Ha and colleagues have performed elegant single-molecule enzymatic studies . Instead of using isolated liposomes, if an array of two or more liposomes connected by (α7)2 is used, the throughput of single-molecule fluorescence measurements could be increased severalfold. In a connected liposome pair, one liposome would function as the reaction chamber and the other liposome would constantly feed in analytes/reactants to it via a dimer pore (Figure 3B). An array of connected liposome pairs could thus be used to study several different reactions simultaneously. Complex multi-step reactions could also be studied using a larger network of liposomes with different reagents in each liposome. Engineered dimer pores could be used to create complex movement pathways of reagents in vesicle networks.
Sensing and communication are key feedback elements of a living system. A feedback mechanism based on these two parameters would have been a major impetus for the evolution of the first cells. For example, sensing mediated by physical and chemical forces (changes in temperature, surface energy) could have led to the assembly of simple cells into networks or primitive tissues. Furthermore, communication between cells across membrane boundaries in a network and the network with its milieu could have led to the proliferation of some tissues and consequently the end of others.
Networks of communicating lipid vesicles in an aqueous phase could be developed as a model system for a tissue mimic. Lipid-coated aqueous droplets in oil offer another exciting possibility for simulating cells and tissues. Creating synthetic elements such as the protein nanopore dimer are small, yet key, steps in designing artificial systems to understand physical and biological phenomena. The principles outlined for future studies will not only help us to realize long-term goals in synthetic biology, but also have an outreach in bio-nanotechnology, medicine and smart biosensors.
Protein Engineering: New Approaches and Applications: A joint Biochemical Society/Protein Society Focused Meeting held at the University of Chester, U.K., 10–12 April 2013. Organized and Edited by Ross Anderson (Bristol, U.K.) and Dafydd Jones (Cardiff, U.K.).
S.M. acknowledges the support of the Clarendon Scholarship (2009–2012) as a Ph.D. student in Professor Hagan Bayley's laboratory K.T.S. acknowledges support from the EU Commission in the form of a Marie Curie Intra-European fellowship for his stay at the University of Oxford, where most of these ideas were developed.