Division and Regrowth of Phase‐Separated Giant Unilamellar Vesicles**

Abstract Success in the bottom‐up assembly of synthetic cells will depend on strategies for the division of protocellular compartments. Here, we describe the controlled division of phase‐separated giant unilamellar lipid vesicles (GUVs). We derive an analytical model based on the vesicle geometry, which makes four quantitative predictions that we verify experimentally. We find that the osmolarity ratio required for division is 2 , independent of the GUV size, while asymmetric division happens at lower osmolarity ratios. Remarkably, we show that a suitable osmolarity change can be triggered by water evaporation, enzymatic decomposition of sucrose or light‐triggered uncaging of CMNB‐fluorescein. The latter provides full spatiotemporal control, such that a target GUV undergoes division whereas the surrounding GUVs remain unaffected. Finally, we grow phase‐separated vesicles from single‐phased vesicles by targeted fusion of the opposite lipid type with programmable DNA tags to enable subsequent division cycles.


Introduction
"Omni cellulae ec ellulae." From the point of view of modern science,R aspailsr ealization from 1825, [1] popularized by Virchow, [2] may state the obvious:E very living cell found on Earth today originates from apreexisting living cell. Bottom-up synthetic biology,h owever,i sc hallenging this paradigm with the vision to create as ynthetic cell from scratch. [3,4] Success unquestionably entails that the synthetic cells must have the capacity to produce offspring,making the implementation of synthetic cell division an exciting goal. [5][6][7][8] Over the course of evolution, living cells have developed as ophisticated machinery to divide their compartments in ah ighly regulated manner.T he reconstitution of am inimal set of cellular components seems to be ap lausible albeit challenging route towards synthetic cell division. [9][10][11] These challenges leave room for creative approaches,s eeking solutions beyond the mimicry of todaysb iological cells. One exciting strategy is to assemble ad ivision machinery de novo,b yd esigning active,n ot necessarily protein-based nanomachines.D NA origami structures have been used to shape and remodel lipid vesicles, [12][13][14] although active forcegenerating motors remain ad istant goal. As hortcut towards synthetic cell division is the non-autonomous mechanical division of liposomes, [15] which may jump-start exciting directions.T he exploitation of physicochemical mechanisms, on the other hand, could lead to autonomous division. Noteworthy theoretical work describes the shape transformations of single-phase [16][17][18] as well as phase-separated liposomes [19][20][21] depending on the surface-to-volume ratio. Tw ov esicles connected with at ight neck have been theoretically predicted [20] and can readily be observed in experiments. Ar emarkable recent report triggered shape transformations of lipid vesicles by an internal enzymatic reaction, but neck fission did not occur. [22] There are few experimental reports describing the complete dissociation of small buds from ap arent vesicle. [23,24] Division into more equally sized compartments has once been reported as an occasional observation [25] or it relied on multilamellar vesicles [26] or liquid-liquid phase separation. [27] Moreover,m ultilamellar fatty acid vesicle systems have been shown to deform and sometimes divide [28] and recently,d ivision was shown as ar esult of spontaneous curvature. [29] However,w ea re still missing aw ell-controlled division mechanism where designated vesicles divide with as uccess rate close to 100 %, combined with as uitable growth mechanism. This would be an important step for the field of bottom-up synthetic biology since it could provide the basis for the evolution of synthetic cells.
Here,weexperimentally demonstrate full spatiotemporal control over the division of phase-separated GUVs with an unprecedented success rate.T op redict the process quantitatively,w es how that it is sufficient to look at the vesicle geometry.W ed escribe the shape transformations of phaseseparated vesicles without fitting parameters,w hile previous theoretical work relies on membrane-specific parameters. [19][20][21] From these geometrical considerations,w ec an extract the precise conditions required for division and thereby provide areproducible and highly controlled division mechanism. Notably,w ed emonstrate that the division of GUVs can be regulated by am etabolic reaction or triggered locally by light. We further implement vesicle fusion via programmable DNAt ags as am echanism to regrow phaseseparated vesicles from single-phased ones to enable subsequent division cycles.While our synthetic division mechanism distinctively differs from that of nowadays living cells,o ur results prompt to ask whether similar mechanisms may have sustained cell division at the onset of life [30,31] or if remnants thereof may still play arole for the generation of intracellular vesicles or to support certain division processes of todays cells. [32][33][34] Results and Discussion

Division of Phase-Separated GUVsT riggered by Metabolic Decomposition
GUVs-that is,micron-sized vesicles enclosed by asingle lipid bilayer-are the most commonly used compartment type for the assembly of synthetic cells. [4] To realize acontrollable and efficient mechanism for their division, we propose as trategy that is based on three steps:S tep 1) Define the plane of division;S tep 2) Increase the surface-to-volume ratio,a nd Step 3) Enable neck fission to allow for the formation of two smaller second-generation compartments from asingle large compartment. To realize Step 1, we choose lipid phase separation to define the plane of division as the interface of the liquid-disordered (ld, orange) and the liquidordered (lo,g reen) phase as illustrated in Figure 1a.H ence, an increase in the surface-to-volume ratio (Step 2) requires ar eduction of the GUVsi nner volume.T ot his end, we exploit osmosis.A ni ncrease of the osmolarity outside the GUVs,t hat is,ahigher concentration of solutes in the outer aqueous solution, causes water efflux through the GUV membrane [35] as illustrated in Figure 1b.N ote that the number of lipids in the membrane,t hat is,t he surface area of the GUV,remains constant during this process ( Figure S1). There is no lipid addition. As described in previous theoretical work, [20] the GUV deforms to minimize the energy associated with the line tension at the phase boundary until ab ud is connected to the first-generation vesicle by at ight neck. Ac ommon assumption is that the energy barrier for neck scission (i.e.the final pinching of the second-generation vesicle) is too large to enable vesicle fission without coat proteins.H owever,w hile pinching of the lipid constriction comes with an energy cost for opening up the bilayer structure,italso removes the phase boundary. [21,36] Therefore, we postulate that complete division could be favorable if the line tension is high enough (Step 3). To implement the proposed division mechanism experimentally,w ef irst need ac ontrolled mechanism to increase the outer osmolarity of the solution. Metabolic processes,t hat is,t he decomposition of molecules through enzymes,i nevitably lead to an osmolarity increase.W et hus set out to metabolize the sugar Figure 1. Division of phase-separated GUVs. Schematic illustration of the division mechanism relying on a) phase separation of the GUVs and b) osmosis. C 0 ,C 1 ,a nd C 2 denote the osmolarityo utside of the GUVsand V 1 , V 2 ,a nd V 3 describe their volume at different time points. c) Chemicalr eaction pathway of sucrosed egradationc atalyzed by the enzyme invertase. d) Osmolarity ratio C/C 0 over time for GUV-containing solutions composed of 300 mm sucrose, 10 mm HEPES (pH 7.4) and 44 mg L À1 (blue) or 22 mg L À1 invertase( gray). Error bars are too small to be visible. The data was fitted with limited growth fits (solid lines). The dotted black line indicates the time point at which division occurs (see f).e)Overlay of brightfield and confocal image of ap haseseparated GUV with equally large hemispheres (Lipid Mix 1, Table S2, ld phase labeled with LissRhod PE (orange), l ex = 561 nm). f) Confocal fluorescence time series depicting the division process in the presence of 44 mg L À1 invertase. The vesicles are fully separateda nd quickly diffuse apart after division (see 45 min time point). Scale bars:10mm. solution in which GUVs are often immersed. Forthis purpose, we make use of the enzyme invertase.Extracellular invertase is secreted by yeast as af orm of cell-cell cooperation to decompose sucrose into fructose and glucose (Figure 1c). [37] We performed osmometer measurements to test if extracellular invertase in as olution of phase-separated GUVs can produce an increase of the osmolarity ratio C/C 0 as required for division. Indeed we find that the osmolarity of the initially 300 mm sucrose solution increases significantly over time (see Figure 1d). Ther ate of increase depends on the enzyme concentration. In the presence of 44 mg L À1 invertase,t he initial osmolarity almost doubles over the course of 150 minutes.
Note that we did not optimize the conditions for invertase activity but chose conditions compatible with the proposed mechanism for GUV division. Phase-separated GUVs with two distinct hemispheres (Figure 1e), were successfully electroformed using al ipid mixture consisting of DOPC, cholesterol, DPPC,C La nd LissRhod PE (DOPC (18:1 1,2dioleoyl-sn-glycero-3-phosphocholine), DPPC (16:0 1,2-dipalmitoyl-sn-glycero-3-phosphocholine), CL (Cardiolipin (Heart, Bovine)), LissRhod PE (18:1 1,2-dipalmitoyl-snglycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl));T ables S1 and S2, Mix 1). [38] LissRhod PE labels the ld phase (orange). To test the proposed mechanism for division, we add 44 mg L À1 invertase to the GUV-containing sucrose solution. Figure 1fshows atime series taken over the course of 45 minutes (see Figure S2 for an overview image with multiple dividing vesicles). We observe the formation of ac onstriction at the interface of the two phases,e ventually leading to complete division. As visible in the final timestep, the second-generation vesicles diffuse apart as soon as the division is completed, proving that complete neck scission occurred. Control experiments confirm that neither phase separation, nor osmosis alone are sufficient to promote GUV division ( Figure S3). To appreciate the continuous deformation process leading to division, Video S1 is recommended. To probe the versatility,w et ested twelve additional lipid mixtures and obtain GUVs with two distinct hemispheres from mixtures containing positively,n eutral, and negatively charged lipids.I nterestingly,w ef ind that the choice of fluorophore attached to the lipid affects the phase separation behaviour ( Figure S4, Tables S2-S4). Division was also obtained for GUVs composed from adistinctively different lipid mixture (Table S2, Mix 2, Video S2). We have thus achieved the division of phase-separated GUVs by increasing the outer osmolarity with an enzymatic reaction. It is interesting to consider that phase separation may have come into play when phospholipids emerged. [30,31] By regulating the transcription of ametabolic enzyme like invertase,primitive cells could, in principle,m aintain ah igh level of control over their division without asophisticated division machinery.

Theoretical Prediction of the Division Process
To gain control over the process,weset out to predict the osmolarity ratio required to achieve division of ap haseseparated GUV.F or this purpose,w ed evelop an analytical model describing the geometrical GUV shape throughout the deformation process as two spherical caps with abase radius s 0 for the initially spherical GUV and s < s 0 for the deformed GUV.One of them represents the ld phase with asurface area A ld and the other one the lo phase with as urface area A lo , respectively.T he relevant geometrical properties ( Figure 2a) can be extracted from confocal images.T his representation provides ag ood approximation of our experimentally observed GUV shapes including ak ink at the phase boundary compared to the dumbbell shape expected for single-phased GUVs.W ea ssume that the total area A tot remains constant

Angewandte Chemie
Forschungsartikel throughout the division process.I ft he outer osmolarity increases (C > C 0 ), the volume of the GUV will decrease due to water efflux. This process is fast compared to the time scale of the division process [39] and therefore assumed to be instantaneous in our model. Thee quilibrated inner volume is then given by V = C/C 0 V 0 .T he resulting excess membrane area allows for deformation of the initially spherical GUV. Deformation minimizes the phase boundary (s < s 0 )toreduce the energy associated with the line tension. [20] To quantify the progression of the division process,w ed efine ad ivision parameter d: d is 0f or the initial spherical GUV and 1f or adivided GUV. Based on these geometrical considerations,t he osmolarity ratio C/C 0 needed to achieve ac ertain deformation d for asymmetric GUV (A ld = A lo )can be calculated as Them odel thus postulates that the osmolarity ratio required for complete division (d = 1) is C/C 0 = ffiffi ffi 2 p % 1.41 (Prediction 1). Since Equation (2) does not depend on the initial radius r 0 of the GUV,the osmolarity ratio required for division is independent of the size of the GUV (Prediction 2). While living cells normally undergo symmetric division, where both second-generation compartments are of similar size,s ome processes like oocyte maturation rely on asymmetric division. [40] To extend our model for asymmetric GUVs with A ld ¼ 6 A lo we define al ipid ratio parameter l ¼ A ld =A tot ¼ 1 À A lo =A tot and hence obtain See Note S1 in the Supporting Information for adetailed derivation of the equations.
It follows that GUVs with higher asymmetry should require lower osmolarity ratios for complete division and should hence divide faster (Prediction 3). Figure 2bshows the predicted division parameter dasafunction of the osmolarity ratio C/C 0 for different lipid ratios l. AG UV with l = 0.8 divides already at an osmolarity ratio of approximately 1.22 (compared to 1.41 for symmetric GUVs with l = 0.5). For clarity,F igure 2c displays the predicted shapes of the GUVs corresponding to specific points of the phase space spanned by the division parameter and the osmolarity ratio as indicated in Figure 2b.F inally,a ny process that provides as ufficient change in the osmolarity ratio should lead to division of phase-separated vesicles,i ndependent of the chemical nature of the process (Prediction 4). Compared to previous models describing the shape transformations of lipid vesicles, [19][20][21] our model merely considers geometric proper-ties without fitting parameters.Nevertheless,the model yields four predictions,w hich we will now test experimentally.

Quantitative Comparison of Experiments and Theoretical Predictions
To test the predictions of our model in aq uantitative manner,w ef irst observe symmetric phase-separated GUVs (l = 0.5) in solutions with different well-defined osmolarity ratios C/C 0 .Itiscrucial to immerse the GUVs slowly to avoid lipid tubulation (see Figures S2 and S5). To be able to extract geometrical parameters more precisely from the confocal images,w ea dditionally label the lo phase.F or this purpose, we add cholesterol-tagged 6-FAM-labeled DNAt ot he GUVs,w hich self-assembles selectively into the lo phase (green) in aMg 2+ -containing buffer (see Figure S6). Note that Mg 2+ leads to asignificant reduction of the invertase activity in the presence of GUVs (see Figure S7), likely due to electrostatic interactions between the invertase and the GUVs mediated by divalent ions.T herefore,l abelling of the lo phase was omitted for experiments involving invertase. Similarly,w ef ind that in the presence of Mg 2+ ions,t he vesicles remain in close contact after division, again likely due to electrostatic interactions.After soft shaking,they are found in complete isolation (see Figure S8). Figure 3a shows the theoretically predicted shapes for the different osmolarity ratios.T he corresponding representative confocal fluorescence images are presented in Figure 3b.Note that the shapes are static since the osmolarity ratio is kept constant, unlike in the case of invertase activity.W ee xtract the geometrical parameters required to calculate the division parameter d from multiple images.A sp ostulated, we observe division at an osmolarity ratio of approximately ffiffi ffi 2 p (Prediction 1). We find that 90 %p ercent of the GUVs are single-phased (n = 200) at this osmolarity ratio,s uggesting ar emarkably high division rate.T ov erify the size independence of the division process (Prediction 2), we used the images of the deformed GUVs to calculate the radius r 0 of the initially spherical GUV. Thescatter plot of the division parameter dover r 0 is shown in Figure 3c.A se xpected, no significant size-dependent deviations from the theoretical value (blue line) can be observed in the size range of GUVs.F or vesicles below 1 mm, size effects and membrane-specific parameters will likely come into play. As aquantitative comparison of the experimental results with the theoretical prediction [Eq. (2)],weplot the mean division parameter as af unction of the osmolarity ratio.F igure 3d shows that the experimental data agrees well with the theoretical prediction. Deviations may occur due to the fact that GUVs are imaged in solution and can hence rotate in the confocal plane.T rapping can lead to lipid tubulation and hinder the division process (see Figure S9). Note that the quantitative understanding of the vesicle shape as afunction of the osmolarity ratio allows us to use phase-separated GUVs as precise osmolarity sensors.T his could for instance be useful for measuring extracellular osmolarity in cell culture based on conventional microscopy without any additional equipment (conventional osmometer measurements require freezing of the sample).

Forschungsartikel
To show that our geometrical description does not only predict the static GUV shapes but also the dynamic division process,w ea nalyse confocal time lapses of the division process in the presence of invertase.W ec an extract the osmolarity ratio at ag iven time point from the osmometer measurements in Figure 1d.F igure 3f confirms that the division process of symmetric GUVs with two equally large hemispheres (l = 0.5) agrees well with the prediction. Finally, asymmetric GUVs with l > 0.5 should require lower osmolarity ratios for division and hence divide faster (Prediction 3). To test this,w eo bserved GUVs with different lipid ratios. Figure 3e shows that asymmetric GUVs indeed exhibit shorter division times-approximately 27 min for l = 0.65 and 20 min for l = 0.80 compared to 40 min for l = 0.50 (see Figure 1f). Figure 3g confirms that the division parameter plotted as af unction of the osmolarity ratio follows the theoretical predictions [solid lines,E q. (3)].T he fact that asymmetric division happens at lower osmolarity ratios may explain why budding was more frequently reported in the literature [23,24] than symmetric division.

Light-Triggered Local Division
Any process that achieves as ufficient increase of the osmolarity ratio should, in principle,b es uitable to trigger division of phase-separated GUVs (Prediction 4). We first demonstrate this by showing that water evaporation can be used instead of invertase activity to increase the osmolarity ratio,see Figure S3 a. This confirms that the division process is not dependent on the chemical nature of the enzymatic reaction but relies on the resulting osmolarity increase. Exploiting this versatility,w ew ant to realize am echanism with full spatiotemporal control over the division process, such that as elected vesicle divides at ac hosen time point whereas surrounding vesicles remain unaffected. We successfully achieve this aim based on the light-triggered uncaging of bis-(5-carboxymethoxy-2-nitrobenzyl)-ether (CMNB)-caged fluorescein. Upon 405 nm illumination, this initially nonfluorescent compound splits into three components-two CMNB molecules and the fluorophore fluorescein (Figure 4a). Its contribution to the overall osmolarity should thus triple.T he successful uncaging of fluorescein can be monitored with UV/Vis spectrometry (see Figure S10 a). Figure 4b illustrates our concept for the localized lighttriggered division:P hase-separated vesicles are immersed in asolution containing CMNB-fluorescein. Subsequently,atarget GUV is chosen for division. Thed ivision process is initiated by illuminating the surrounding area with a4 05 nm laser diode leading to uncaging of CMNB-fluorescein. Fluorescein release increases the osmolarity locally,h ence leading to division of the selected GUV,w hile surrounding GUVs remain unaffected. Based on theoretical considerations (Note S2) and osmometer measurements ( Figure S10b), we set the initial concentrations to achieve the required increase of ffiffi ffi 2 p in the overall osmolarity.F igure 4c shows snapshots from aconfocal fluorescence time series before (i) and during illumination (ii)o ft he selected area (Video S3). While division previously happened within tens of minutes (see Figures 1and 3), the rapid uncaging dynamics of CMNBfluorescein promote division after af ew seconds.O ther representative examples of GUVs undergoing similarly fast division are shown in Figure S11. Note that we could only record one fluorescence track to capture the fast dynamics. Thei ncrease in the background fluorescence intensity is due to bleed through from the 405 nm excitation and the release of fluorescein (for confocal images of the fluorescein channel before and after release,s ee Figure S12). Finally,F igure 4d highlights the locality of the division:A se xpected, av esicle outside the illuminated area does not undergo division. Moreover,i llumination alone,i nt he absence of CMNB-

Angewandte Chemie
Forschungsartikel fluorescein, does not lead to division of phase-separated GUVs ( Figure S13). Figure 4e plots the division parameter for the vesicle shown in ca safunction of time (for more examples see Figure S11). Thep lot clearly shows that no shape changes occur before illumination (the frame rate was reduced to avoid bleaching and the GUV was observed for in total 100 sb efore illumination). As soon as the local osmolarity change is induced by uncaging of CMNB-fluorescein at t = 0s ,the vesicle starts to deform and fully divides after 7.9 s. Theshape of the curve is likely to be aresult of the non-linear increase in osmolarity (Figure S10 b): Uncaging increases the osmolarity locally in the illuminated confocal volume,y et components freely diffuse in and out. Caged compounds have previously been used to change the osmolarity to induce compartment rupture. [41] Here,wehave shown that they offer the additional possibility to trigger the division of phase-separated GUVs locally with light, achieving arapid time response and division within seconds.

Regrowth of Phase-Separated Vesicles After Division
Crucially,s ynthetic cell division should be followed by agrowth phase in order to ultimately sustain multiple growth and division cycles.Ino ur system, this process has to restore the initial phase separation of the GUV.D ifferent methods for vesicle fusion have previously been employed to grow GUVs. [42][43][44][45] However, it is not trivial that these conventional fusion mechanisms can lead to phase-separated GUVs:T he emerging line tension adds to the energy barrier for the fusion of al o-phase vesicle to al d-phase vesicle.A saproof-ofprinciple experiment, we produced carboxyfluorescein-la-beled lo GUVs and rhodamine-labeled ld GUVs separately, mimicking the single-phase GUVs after division. With this strategy we can be absolutely sure that aGUV,which contains both fluorescent dyes,results from afusion event.
We mixed ld and lo GUVs and added Ca 2+ -ions.This leads to attractive interactions between the GUVs [46] and has been shown to mediate fusion between identical lo-phase GUVs. [47] We find that this process yields phase-separated GUVs,which unambiguously demonstrates that fusion between the lo and ld GUVs has occurred ( Figure S14, Video S4). It should be noted, however, that despite frequently observed hemifusion and attachment of GUVs to one another, full fusion is arare event and the vast majority of GUVs (over 95 %) remains single-phased. Moreover,f using GUVs again after division cannot lead to growth of the GUV population.
We ultimately need a" feeding mechanism" as illustrated in Figure 5a,w here each growth-division cycle can increase the total number of GUVs.CaCl 2 -mediated fusion can restore phase-separation upon addition of small unilamellar vesicles (SUVs) to GUVs ( Figure S15). However, this approach lacks programmability.Inorder to achieve targeted fusion of SUVs to GUVs of the,r espectively other lipid phase in am ixture, we thus make use of the sequence-programmable basepairing of DNA. As we already demonstrated, cholesteroltagged DNAs elf-assembles selectively into the liquidordered phase.W efind that in our system tocopherol-tagged DNA, on the other hand, attaches to both phases equally ( Figure S16). By designing complementary single-strands of DNA, one with a3 ' cholesterol and the other one with a5 ' tocopherol, we can thus selectively bring vesicle membranes into close proximity as illustrated in the zoom in Figure 5a. Such zipper-like DNA-based mimics of SNARE proteins

Angewandte Chemie
Forschungsartikel have been used to trigger fusion of SUVs of the same kind, [44] but it is not trivial that phase-separated vesicles can be formed. We hence immersed ld GUVs in af eeding bath containing DNA-functionalized lo SUVs (Figure 5b). Note that the SUVs (green) with ad iameter of around 100 nm ( Figure S17) are too small to be resolved individually.U pon addition of the complementary DNA, we observe phaseseparated GUVs with as ufficiently large lo phase to restore the initial condition. Given the area of the ld phase in Figure 5c,w ee stimate that approximately 5600 SUVs have fused to the GUV.W eh ypothesize that the line tension at phase boundary present after the first fusion event lowers the energy barrier for subsequent fusion. Lipid phase boundaries have been shown to promote other fusion events including HIV entry. [48] Thetime-resolved growth process is depicted in Figure S18. Note that the SUVs have al arger surface-tovolume ratio compared to the GUVs and thus are supplied in alower osmolarity solution in order to obtain spherical GUVs after fusion. Thes mall osmolarity mismatch is likely to be beneficial for the fusion process itself. [15] By preventing duplex formation, that is,i na bsence of tocopherol-tagged DNA, we do not observe vesicle fusion ( Figure S19). Compared to Ca 2+ mediated fusion, we did not only gain programmability,b ut also increased the efficiencyo ft he process.F igure 5d shows an overview image,w hich demonstrates that phase-separation was restored in the majority of GUVs after incubation. Al ipid ratio of l % 0.5 could be restored reproducibly.

Conclusion
Synthetic cell division is one of the most exciting albeit challenging tasks towards the bottom-up construction of cellular systems.O ur study realizes the division of GUVs, fully controllable by two physical parameters-phase separation and osmosis.P hase separation of the lipids in the GUV membrane defines the plane of division such that an increase of the surface-to-volume ratio by osmosis leads to contraction at the phase boundary and thus the formation of two secondgeneration compartments.W ederived amodel of the division process based on geometrical considerations.T he analytical model makes four predictions,w hich were all verified experimentally:F irst of all, the osmolarity ratio required for division of GUVs with equally sized phases is ffiffi ffi 2 p ;s econdly, the time-point of division is independent of vesicle size;third, asymmetric division happens faster (i.e.a tl ower osmolarity ratios) and fourth, any process,w hich leads to as ufficiently large osmolarity increase,can trigger division. We showcased the latter by demonstrating division as aresult of fundamentally distinct processes,i ncluding water evaporation, metabolic decomposition of sugars and light-triggered uncaging of CNMB-fluorescein. Using light as as timulus for division provides full spatiotemporal control, which could, in the future be exploited to perform directed evolution of avesicle population. Theconcept to exploit caged compounds for local vesicle division is new and broadly applicable.Itdoes not rely on specific environmental conditions and can directly be extended from CMNB-caged fluorescein to other caged compounds.A ny suitable division mechanism for synthetic cells should have the capacity to sustain multiple growth-anddivision cycles.I no ur case,g rowth has to restore phase separation in the second-generation compartments.W e achieve fusion of SUVs of the other phase to single-phased GUVs with programmable DNA-based SNARE protein mimics-thus restoring the initial conditions for subsequent division cycles,w hich will undoubtedly be ap rerequisite for the evolution of synthetic cellular systems.T he future integration of information storage and replication will be yet another important milestone towards the visionary transition from matter to life,o r, in other words,t owards as ynthetic cell which truly deserves its name.I nt he meantime,o ur engineering approach to synthetic cell division prompts questions about cellular life as we know it:W emay be curious to discover whether phase separation and osmosis may have sustained compartment division at the onset of life, possibly regulated by the expression of metabolic enzymes. And we may further ask how remnants thereof play arole in cell biology today-continuously nurturing the emergence of cells from cells.