Growth of Giant Peptide Vesicles Driven by Compartmentalized Transcription–Translation Activity

Abstract Compartmentalization and spatial organization of biochemical reactions are essential for the establishment of complex metabolic pathways inside synthetic cells. Phospholipid and fatty acid membranes are the most natural candidates for this purpose, but also polymers have shown great potential as enclosures of artificial cell mimics. Herein, we report on the formation of giant vesicles in a size range of 1 μm–100 μm using amphiphilic elastin‐like polypeptides. The peptide vesicles can accommodate cell‐free gene expression reactions, which is demonstrated by the transcription of a fluorescent RNA aptamer and the production of a fluorescent protein. Importantly, gene expression inside the vesicles leads to a strong growth of their size—up to an order of magnitude in volume in several cases—which is driven by changes in osmotic pressure, resulting in fusion events and uptake of membrane peptides from the environment.

ThomasF rank,Kilian Vogele,Aurore Dupin,F riedrichC.S immel,*and To bias Pirzer* [a] Abstract: Compartmentalizationa nd spatial organization of biochemical reactionsa re essential for the establishment of complex metabolic pathways inside synthetic cells. Phospholipid and fatty acid membranes are the most natural candidates for this purpose, but also polymers have shown great potential as enclosureso fa rtificial cell mimics.H erein, we reporto nt he formation of giant vesiclesi nasize range of 1 mm-100 mmu sing amphiphilic elastin-like polypeptides. The peptide vesicles can accommodatec ell-freeg ene expression reactions, which is demonstrated by the transcription of af luorescent RNA aptamer and the production of af luorescent protein. Importantly,g ene expression inside the vesicles leads to a strong growth of their size-upto an order of magnitude in volumei ns everal cases-which is driven by changes in osmotic pressure, resulting in fusion eventsa nd uptake of membrane peptides from the environment.
One of the most prominent goals of bottom-up synthetic biology is the creation of syntheticc ells. [1] Such systems are envisioned to displayaset of properties andc apabilities that are associatedw ith extant living cells, namely (i)compartmentalization, (ii)growth,s elf-maintenance and self-replication,( iii) signaling, communicationa nd sensing, and (iv) the potential for evolution through replication and transfer of genetici nformation.
Compartmentalization is an essential prerequisite for the remaining properties,a nd has been achieved using aw ide variety of different approaches. [2] Cell-scaler eactionc ontainers have been created from phospholipid membranes, [3] using fatty acids [4] or polymers, [5] emulsion droplets, [6] coacervates, [7] or even using microfluidics-based DNA chips. [8] Severals ynthetic cell modelsa lready displayeda tl east some of the desired properties listed above.F or instance, Szostak and co-workers reported on fatty acid vesicles-serving as primordial cell models-,w hich were able to grow and divide upon external feeding with fatty acids. [9] In further work, enzyme-free copying of nucleic acid templates was shown inside fatty acid vesicles. [10] Kurihara et al. succeededt os how DNA amplification inside lipid-based giant vesicles,w hich were able to grow when membrane precursors were added to the outside solution. [11] Growth was also observedf or phospholipid vesicles externally fed with fatty acids, whichc ontained ac ellfree protein synthesis reaction. [12] Other research groupsf ocused on in situ phospholipidb iosynthesis inside of liposomes. For instance, Hardy et al. catalytically synthesized phospholipids from simpler precursors,w hich also resulted in membrane growth. [13] More closely mimicking lipid synthesis in natural cells, Scott et al. established parts of the complex phospholipid synthesis pathway inside of liposomes. To this end, all the required enzymesw ere encoded on DNA templates and produced inside of the liposomes via cell-free gene expression. [3a] Due to their similarity to biological cell membranes,p hospholipid membranesa ppear to be the most natural candidates for compartmentalization of synthetic cell-mimicking systems. From atechnical point of view,h owever,p hospholipids have several drawbacks.F or instance, phospholipids form membranes with ar elatively high resistance to ac hange in membrane area, and even minor stretching causesm embrane rupture. [1b] Membranesc omposed of lipid mixtures can have more favorable mechanical properties, but in vesiculo production of mixed membranes would be even more challengingt han for homogeneous membranes.
Other membrane-forming molecules such as amphiphilic block co-polymers or polypeptides representa ni nteresting alternative to phospholipids. [1b, 14] Such membranes are mechanically quite robust and even capable of storinge lastic energy. [15] Furthermore, membrane forming peptides can be easily producedi nside of vesicles [16] using cell-free transcription-translation systems. [17] Ap articularly interesting class of polypeptides that can be used for membrane formation are elastin-like polypeptides (ELPs). [1b, 18] The commonly used sequence motif (VPGXG) n (shorthand notation: X n )i sd erivedf rom tropoelastin, whereXis any natural amino acid except prolinea nd ni st he number of pentapeptide repeats. Depending on the amino acid used for Xt he peptided isplays different hydrophobicity. [19] We have recently shown that ELPs can form % 200 nm sized vesiculars tructures for the compartmentalization of biochemical reactions such as transcription of RNA aptamers and the ex-pression of fluorescent proteins. [16a] In the present work we demonstrate the fabrication of much larger,c ell-sized polymersomes using as olvent evaporation method. [20] In contrastt o the glass beads method previously used by Vogele et al. [16a] the necessaryp eptide film was formed on the inner glass surface of ar ound-bottom flask, which resulted in vesicle sizes in the mms cale ( Figure 1). Importantly,E LP polymersomes encapsulating transcription (TX) or transcription-translation (TX-TL) reactions displayed as trongi ncreasei ns ize when they were externallys upplied with additional membrane peptides. In several cases, fusion eventsb etween adjacent vesicles wereo bserved as well.
Transcription of RNA aptamers inside the vesicles resulted in mixed growth behaviors, in which some vesicles started to shrink at one point,w hileo thers continued to grow.B yc ontrast, when expressing the fluorescent protein YPet, 97 %o f the vesicles continually increased in size;f or the remaining two vesicles no clear shrinkage or growth could be observed. Our experiments hence demonstrate biochemically driven growth of peptide-based synthetic cellular structures, which could set the stage for competition and selection dynamics emerging among such compartments.
We used an ELP with the sequence (R 5 Q 5 ) 2 -F 20 (in shorthand notation)a st he amphiphilic membrane component ( Figure 1, Supporting Information section 3). This sequence design ensures that the peptidesc ontain aw ell-structured hydrophobic tail and ar andom coil polar head group under our standard experimental conditions. Peptide expression and purification were performed as described previously. [21] Controlled formation of giant peptidev esicles was carried out throughs olvent evaporation based on aprotocol originally introduced by Marsden et al. [16b, 20] Initially,t he amphiphilic ELPs were lyophilized in ar ound-bottom flask to completely remove water from the peptides ( Figure 1b). The peptides were then re-dissolved by the addition of tetrahydrofuran (THF) ands onication. For encapsulation, the internal solution (IS), which containedadefined amount of sucrose in purified water,w as added to the THF/ELP mixture, followed by agitation and incubation at room temperature. Because of the amphiphilic nature of the ELPs used, droplets of the inner solution are formed, which are covered and stabilized by an ELP layer. Subsequentf ormation of ap eptided ouble-layer at the droplet interface, and hence the formation of vesicles, was accomplished by the removal of the THF solventt hrough evaporation. Nonetheless,r esidual solventw ithin the ELP double layer cannotber uled out entirely.
After formation the vesicles were mixed with an outer solution (OS) to diluter esidual IS. The OS contained an isotonic amount of glucose as well as al ow percentage of Triton X-100 (0.01 %). Due to the higherd ensity of the IS, the vesicles sedimenteda tt he bottomo famicroscopy sample chamber and could thus be easily observed for several hours. If an osmotic shock is applied the vesicles vanish ( Figures S5, S6).
The solvent evaporation method resulted in two populations of vesicles of different sizes, namely small vesicles (SV) with radii far below 1 mma nd larger vesicles with sizes spanning two to three orders of magnitude, which we will collectively refer to as giant vesicles (GV). In order to characterize the size of the vesicles across these scales, we utilized transmission electron microscopy (TEM), dynamic light scattering (DLS) as well as light microscopy (LM). For the SVs, am ean radius of 0.03 mm AE 0.01 mmw as determined using TEM (Figure 2), where the estimated uncertainty is the standard deviation of the distribution. The GV fraction displayed aw ide rangeo f radii between 400 nm and severalm icrometers. Due to the size resolution limits of the DLS and LM characterization methods, we were not able to acquire aq uantitative size distribution across the whole range, but they allowed us to determine the lower and upper size limits of the GV population (Figure 2b). In LM, we occasionally also observed vesicles with radii much larger than 10 mm. Ar ationalef or the occurrence of the SV andG Vp opulations is the presence of two alternative processes of vesicle formation. SVs are presumably created through spontaneousf ormation in aqueous solutionf rom ELP monomers, whereas the GVs are generated only through the application of the solventevaporation method.
We next studied the capability of the GVs for fusion and growth. We speculated that-similarly as previously observed with fatty acid vesicles [4] -an osmotic imbalance could lead to an influx of water and thus promote vesicle growth. In our experiments the imbalance was created through biopolymerization of polyelectrolytes. We therefore transcribed the fluorogenic RNA aptamerd Broccoli inside the vesicles by encapsulating at ranscription mix containing T7R NA polymerase,r NTPs,   NA, the ligand of the aptamer,D FHBI (3,5-difluoro-4-hydroxybenzylidine imidazoline), and sucrose( Supporting Informations ection 1.2). In order to be able to control the start of the transcription reaction,w es eparately produced two types of vesicles, which were either missing the templateD NA or the T7 RNA polymerase (Figure 3a). In the experiments, these vesicles were mixed, resulting in the transcription of fluorescent aptamers only inside of vesicles,w hich were generated through fusion of the two vesicle types. We added DNase It o the OS to preventt ranscription by accidentally released transcriptionmix. In addition, the surrounding OS was supplemented with al ow percentage of Triton X-100 (0.01 %) and an isotonic amount of glucosetobalance the initial osmotic pressure in the vesicles. Control experimentss howed that no TX activity was observed in the absence of Triton X-100,w hich apparently promoted vesicle fusion. Surprisingly,T riton X-100 alone cannoti nduce peptide vesiclef usion ( Figure S4), and we suppose that aslight osmotic imbalancei snecessary for successful fusion events. It was essential to add additional ELP monomers (200 mm)t ot he OS to facilitate vesicle growth through their incorporation into the membrane. We found that in the absence of external ELPs, the observedv esicles were generally smaller and less abundant. Furthermore, the vesicles were not stable during the experimentsand tended to shrink.
The initial vesiclep opulation was highly polydisperse in size, which likely resulted in large variations in the contents of the compartments generatedv ia fusion of the two typeso fv esicles. This in turn was expected to result in ab road distribution of transcriptional activities among the vesicles. [22] As shown in Figure 3b and Video S1, vesicles containing the transcription mix strongly varied in number and size over at ime period of severalh ours. Next to as trong increase in size of the GVs, the appearance of small "satellite" vesicles aroundt he GVs was observed. These are potentially generated by spontaneous budding events, [23] membrane instabilities followed by budding due to osmotic imbalances and Triton X-100 or interactions with the microscopy glass slide. However,s imilarv esicles were observed to emerge throughout the whole micrograph, which suggestst hat they could also simply originate from growing SVs, whose size initially was below the observation limit.
The observed growth of the vesicles is consistent with our expectation that compartmentalized RNA polymerization is accompanied by an increasing osmotic pressure in the vesicles. [4] In the absence of bio-polymerization reactions no vesicle growth was observed, even with externally provided ELPs (Figure S4, Video S2). Figure 3c shows example time traces for growingv esicles and for growth followed by shrinkage when RNA aptamersa re transcribed inside av esicle.W hen in close proximity,t wo or more GVs can also fuse and thereby rapidly increasei ns ize. We find that in all fusion events, the final intensities, radii and volumes are slightly less than expected from the sum of the fusing vesicles (Figures 3d,S 8, Video S3), implyingt hat some of the vesicle content leaks out during fusion events. The excessm embrane resulting from fusion may either be lost tos olution,o ri ncorporated into am ultilamellar membrane structure.
In order to understand the dynamicso fv esicle growth and shrinkage, we have to consider the interplay of compartmen-talizedR NA polymerization, the incorporation of externally provided ELPs into the membrane and water influx. As shown in Figure4a, fluorescence intensities and volumes are almostl inearly correlated,w hich indicates that the concentrationo ft he transcribed RNA molecules in the vesicless tays approximately constant.T his in turn suggests that water influx into the growing vesicles is fast enough to compensate fort he excesso smotic pressure generated by the newly formed polyelectrolytes. [4,24] As can be seen in the inset of Figure 4a,t he global linear trend is occasionally interrupted by phases of alternating growth and shrinkage, where the correlation between intensity and volumebecomesn on-linear.
Our experimental observations furthers uggest that the availability of as upply of ELPs (as monomers, micelles or small vesicles)i nt heir immediate vicinity determines whether vesicles will grow or shrink. Some vesicles are found to repeatedly grow and shrink, and finally even disappear ( Figure S8, Videos S4.1 and S4.2). Figures 4b,c show two representative examplesf or vesicles surrounded by either many or by af ew smaller vesicles.I tc an be clearly seen that in both cases the smaller vesicles slowly disappear whereas the largestv esicle continually grows. We suppose that the large sizes of the central vesicles are reached by the consumption of the surrounding "prey" vesicles.Assoon as the supply of prey vesiclesi sd epleted, the central vesicle starts to shrink (Figure4b, green).
In anothers et of experiments, we encapsulated ab acterial cell extract-basedp rotein expression system (TX-TL) mixed As in the TX case the correlation between YPet fluorescence and volume is approximatelyl inear (Figure 5c), which again suggestsabalance between in vesiculo productiono fb iopolymers and osmotically driven growth. In contrast to the TX mix, the much more complex cell extract contains nearly the complete proteome of BL21 rosetta E. coli cell, in whichcase the osmotic pressure of the vesicle will be influenced by am ore complex network of biochemical reactions. Only very few cases of shrinking vesiclesw ere observed, but most vesiclese nter a plateau phaseo fc onstant volume and constant YPet fluorescence after completiono ft he TX-TL reaction, which indicates that an osmotic equilibrium has been attained between the inner and the outer solution.
In conclusion we have demonstrated the generation of cellsized peptide vesicles, which upon encapsulationo fc ell-free transcription and protein expression reactions exhibit volume changes over at least an order of magnitude in several cases. The size changes appear to be caused by ac ombination of fusion eventsa nd osmoticallyd riven growth, when fed with membrane components from the outside. Vesicle growth promoted by internal bio-polymerization reactions is thus much more pronounced than previously observed for other membrane systems, which may be related to the high permeability of the ELP membranes for water combined with their considerable mechanical stability.
It is conceivable that usage of more complex membrane compositions will facilitate the implementationo fo ther celllike behaviors such as compartmental divisiona nd reproduction. In fact, we already observed occasional budding events in our experiments (Supporting Information, Figure S10, Video S6), which may be taken as precursors for such processes. As the growth of our peptide vesicles is coupled to internal gene expression activity,o ur resultsa lso may lay the ground for ac ompetition between different peptide compartments based on the efficiency of the compartmentalized reactions.