Optimization of the Inverted Emulsion Method for High‐Yield Production of Biomimetic Giant Unilamellar Vesicles

Abstract In the field of bottom‐up synthetic biology, lipid vesicles provide an important role in the construction of artificial cells. Giant unilamellar vesicles (GUVs), due to their membrane's similarity to natural biomembranes, have been widely used as cellular mimics. So far, several methods exist for the production of GUVs with the possibility to encapsulate biological macromolecules. The inverted emulsion‐based method is one such technique, which has great potential for rapid production of GUVs with high encapsulation efficiencies for large biomolecules. However, the lack of understanding of various parameters that affect production yields has resulted in sparse adaptation within the membrane and bottom‐up synthetic biology research communities. Here, we optimize various parameters of the inverted emulsion‐based method to maximize the production of GUVs. We demonstrate that the density difference between the emulsion droplets, oil phase, and the outer aqueous phase plays a crucial role in vesicle formation. We also investigated the impact that centrifugation speed/time, lipid concentration, pH, temperature, and emulsion droplet volume has on vesicle yield and size. Compared to conventional electroformation, our preparation method was not found to significantly alter the membrane mechanical properties. Finally, we optimize the parameters to minimize the time from workbench to microscope and in this way open up the possibility of time‐sensitive experiments. In conclusion, our findings will promote the usage of the inverted emulsion method for basic membrane biophysics studies as well as the development of GUVs for use as future artificial cells.


Introduction
Minimal cell research has gained considerable interesti n recent years. [1] Successful realizationo fw hich could open up the possibilitieso fn ot only ad eeper understanding of biological cell complexitiesb ut also for engineering new artificial cells designed with specific tasks in mind. The "top-down" approach aims to reach this goal through the modification of pre-existing organisms. The "bottom-up" approach, on the other hand, hopes to achievet his goal by assembling an artificial cell from individual non-living components. [1] The latter method,a lthoughi tm ay take longert or each the goal, gives us the opportunity for complete control of the system, whichm ay lead to abroader range of future applications.
Theoretically,t he construction of am inimal cell from the bottom-up should be possible by mimickinga nd redesigning what we see in nature using individual components, such as sugars,l ipids, proteins, and genetic material. [2] Beforet hese steps can take place, as uitable compartment system to con-tain these materials shouldb ee stablished. Among all the choices, lipid vesicles remain the most likelyc andidates to succeed. These vesiculars tructures are composed of phospholipids and come in variouss izes, such as small unilamellar vesicles (SUVs) below 100 nm diameter,l arge unilamellar vesicles (LUVs) from 100-1000 mm, and giant unilamellar vesicles (GUVs) between1a nd approximately 100 mm. [3][4][5] GUVs are considered the gold standard of cellular mimics owing to their similarity in size to that of am ammalian cell. Moreover,w ith advances in technologies, such as microfluidics for their production [6] and handling, [7] GUVs are looking like an increasingly more attractive option.
The challenge lies in not only encapsulatingl arge biomolecules inside giant vesicles (something which not all GUV preparation techniques can achieve), but also in establishing ar eliable method for their production. Havingb iomolecules, such as actin networks or proteins in general on the inside of lipid vesicles is an essential requirement for the bottom-up construction of artificial cells. High protein encapsulation is also of interestf or biophysical studies when their interaction with inner leafletp lasma membranes lipids is under investigation. [8][9][10][11] Moreover,t echniques, which allow encapsulation of low samples volumes are often desirable. Unfortunately,c onventional techniques do not readily yield the encapsulation of large biomolecules and do not offer control over size or yield. [12][13][14] Recently,m icrofluidic systems have been demonstrateda se xcellent platforms for the preparation of monodisperse liposomes with high precision lossless encapsulation of biomolecules. [15][16][17] The mostc ommon methods being double In the field of bottom-up synthetic biology,l ipid vesicles provide an important role in the construction of artificial cells. Giant unilamellar vesicles (GUVs),due to their membrane's similarity to natural biomembranes, have been widely used as cellular mimics. So far,s everal methods exist for the production of GUVs with the possibility to encapsulate biological macromolecules. The inverted emulsion-based method is one such technique, which has great potential for rapid productiono f GUVs with high encapsulation efficiencies for large biomolecules. However,t he lack of understanding of various parameters that affect production yields has resultedi ns parse adaptation within the membrane and bottom-up syntheticb iology research communities.H ere, we optimize variousp arameters of the inverted emulsion-based method to maximize the pro-ductiono fG UVs. We demonstrate that the density difference between the emulsion droplets, oil phase, and the outer aqueous phase plays ac rucial role in vesicle formation. We also investigated the impact that centrifugations peed/time, lipid concentration, pH, temperature, and emulsion droplet volume has on vesicle yield and size. Compared to conventionale lectroformation, our preparation methodwas not found to significantly alter the membrane mechanical properties. Finally,w e optimize the parameters to minimize the time from workbench to microscope and in this way open up the possibility of timesensitivee xperiments. In conclusion, our findings will promote the usage of the invertede mulsion method for basic membrane biophysics studies as well as the development of GUVs for use as future artificialc ells. emulsion templating by co-axially alignedg lass capillaries. [18][19][20] Alternative microfluidic techniques include cDICE, [21] jetting, [22] and pico-injection. [23] For am ore detailed review of these methods please refer to references [17] and [24].A lthough microfluidic devices do yield GUVs, which are more monodisperse in size and with higher encapsulatione fficiencies, their use involves more complex instrumental or time-consuming setups, which are not always available to researchers. [6,[25][26][27] On the contrary,t he bulk inverted emulsion-based method has shown great promise in encapsulatingm acromolecules, such as polymers, [28] DNA, [29] enzymes, [30] cells, [31] and even micron-sized particles. [32] The method can also be used to produce GUVs with complex multicomponent lipid mixtures allowing them to be used as biomimetic membrane models. [33,34] Despite this, the method suffers from af ew drawbacks resulting in poor yields or insufficiently large GUVs. For this reason,v ery few groups have adopted the method and often prefer more established techniques, such as electroformation or gentle hydration, [4,5,14] even if these choices limit the range of experimental possibilities. This papera ims to address this issue by optimizing each of the required steps in the inverted emulsion method to maximize the yield and reliability.N ormally,i ti sp erformedi nE ppendorf tubes resulting in one preparation of GUVs. Not only is this low-throughput, but in the absence of proper optimization,r esults in very low vesicle yields. Recently,w ep resented work in which we demonstrated that microtiter plates are ideal for performing this method as it allows multiple parallel experiments. [29] We take advantage of this setup within this work andp erform repeatable parallel experiments to allow optimization.
The typical procedure ( Figure 1a)s tarts with the initial dissolution of as pecific concentration of ac hosen lipid mixture in oil (such as mineral oil or mixture of multiple oils) that is later added on top of an aqueous solutiont of orm an oil-water interface. The lower aqueous solutionw ill eventually become the outer environment of the GUVs (typically ag lucose solution). At the same stage,aspecific volumeo fd enser aqueous solution is added to the lipid-oil phase (in as eparatev ial) to produce aw ater-in-oil emulsion. This aqueous phase will become the inner compartment of the GUV and will therefore contain the solutes to be encapsulated. Owing to their amphipathic nature, lipids present in the oil phase self-assemble at the interfaces( both the emulsion and the lower interfacial layer,s ee Figure 1a)t of orml ipid monolayers.T he emulsion is then added on top of the interfacea nd ac entrifugal force is appliedtoassist the movement of denser water-in-oil emulsion droplets through the interfaciall ipid monolayer.T his process will result in the simultaneous wrapping of the second monolayer of lipids aroundt he droplets and the formation of GUVs in the lower aqueous phase. Seriousc onsideration should be made to ensure the fact that the oil used to solubilize the lipids can remain in between the lipid leaflets, which can alter the naturalp hysical properties of the membrane. [35][36][37] Any protocol, which uses an oil phase to form lipid membranes should have the membrane inspected. Thisc ouldb eo ne of the major reasonsf or lack of widespreada cceptancea nd usage of this method, even though ac ouple of studies have shownn o significant changes in the membrane mechanics of GUVs prepared from invertede mulsions. [38,39] In addition to this, the methodm ust be optimized for compatibility with ar ange of salt concentrations and pH values, as well as havingc harged/ uncharged macromolecules inside or outside the GUVs to make them as biomimetic as possible.
To the best of our knowledge,a ne xtensive systematic study of the variousp arameters that might affect the productiono f GUVs using this method has not been attempted previously.I n this paper,w ea nalyze the resulting vesicles with respect to their size and yield by thoroughly optimizing each preparation step. We investigate the significance of the applied force, centrifugation duration, the inner solution density/composition, the monolayer formation time, oil phase, lipid concentration, pH, droplet volume, and temperature. Importantly,w econfirm the applicability of the GUVs by characterizingt heir membrane functionality and propertyb yp erforming am embrane proteinbased permeation assay as well as bending rigidity measurements. Finally,t he method can be performedw ithin microtiter well plates for high-throughput production and analysis.

Results and Discussion
The multiple steps involved in this invertede mulsion method can lead to unseene rrors (Figure 1b and c) which may lead to low yields. So here, we optimize each step systematically to identify the most crucial ones and the effects they have on the morphology as wella st he yield of the GUVs.P erforming the method within microtiter plates allows us to run multiple parallel conditions for fast optimization. Here, we focus on the effectso fd ensity gradientso ft he aqueous solutions, the time and speed of centrifugation, the concentration of the lipids,

Sugar-baseddensity gradients
Sugars,s uch as glucose and sucrose, have been extensively used in the preparation of giant vesiclest om aintain isoosmotic conditions across the membrane and also for their relative inertness towards the lipid bilayer.I nt he case of the inverted emulsion method, the inclusion of ad ensity gradientb etween the lipid oil and the inner aqueous solution is essential for the formationo fG UVs. The water-in-oil dropleth as to be transferred from the oil phase to the aqueous phase. Most natural oils are of low density compared to pure water and mineral oil, used primarily in this study,i sn oe xception.F urthermore, there is another density gradient to be taken into consideration-the density differenceb etween the inner aqueous solution and the outer aqueous solution. This allows the GUVs to settle at the bottomo ft he well not only for instant visualization but to avoid aggregation at the interface-made possible if the outer aqueous solution is less dense compared to the inner aqueous solution present inside the droplets (and eventually inside the GUVs). Somer esearchers have used gravity to provide the force to drive the droplets through the interface. [40] This method of production not only requires as ignificant density gradient, but also needs al onger time for all the droplets to crosst hrough the lipid monolayer to obtain ag ood yield. Alternatively,t his process can be expedited by applying ac entrifugalforce, as performed herein.
From Figure 2a,i ti se vident that there is as ignificant increase (at least~fivefold) in the number of GUVs produced when the inner sucrose solution concentrationsa re equal to 300 mm or above compared to 50 mm at any given applied centrifugal force. Note that glucose at an isoosmolar concentration is used as the outer solutiont hroughout unless otherwise stated. Furthermore, increasing the centrifugal force appliedf or as pecific time (here it was fixed at 3min) is also effective in improvingt he number of GUVs produced until it plateausat400 g. However,its hould be noted that high centrifugation speeds (above3 00 g)r esult in clustering of GUVs at the cover glass and formation of lipid clumps due to bursting ( Figure 2c and FigureS1i nt he Supporting Information). The average sizes of the giant vesicles produced tend to remain approximately the same for all the sugar concentrations and appliedc entrifugal forces tested. The range was between 20-35 mmi nd iameter,e xcept for a5 0mm sugar concentration where some were less than 20 mm ( Figure 2b). Note that vesicles smaller than 10 mma re excluded from our analysis. We conclude that ad enser inner solution promotes better yields due to al arger force experienced by each droplet. We speculate that the size variation for low sugars mayb eb ecause larger GUVs do not survive or that the droplets did not make it through the interface. For the next experiments, we chose 600 mm sucrosea st he optimal sugar concentration, that is, high yield, large GUVs, and minimum bursting. Although this ideal sugar concentration is higherthan physiological osmolari-ties (~260 to > 400 msOm depending on the cell), more relevant concentrationso f3 00 mm also produce good yields.

Centrifugal duration
From the above data it is evident that the invertede mulsion methodi sh ighly dependento nt he density of the solutions used. Whereas ah igher centrifugal speed for as hort span of time might be enough to force all the droplets to pass through the lipid interface, our data suggest that it will also result in bursting and aggregation of the GUVs along the walls of the microtiter plate (see 1200 g centrifugal speed for 10 min in Figure S2). Therefore, we aimed at achieving good yields at lower speeds but with longer time durations as shown in Figure 3.
As expected, the yield increased with increasing duration of centrifugal force applied( Figure 3a). This is to be expected as al ongera pplied force will increase the chances of the droplets crossingo vert he interface. Figure 3a also shows that the yield of GUVs remains highest with am inimum of 3min and above 200 g. This can be advantageous if speed of encapsulationi s crucial.F urthermore, as the yield of the GUVs remained the same for higher forces it suggests that most of the droplets have passed through the interface( for this specific volume of emulsion droplets and solution density gradient). However,t he plateau in yield could also be ar esult of 1) bursting of large vesicles at the glass well bottom due to higherf orces and/or 2) the interfacial monolayer not having enought imet or e-seal before the next dropleta rrives. We note that although the yield and size of the GUVs are superior from 600 g and above for all durations( see Figure S2 fori mages),f ewer lipid aggregationso ccur at 200 g for 3min (Figure 3c). Considering this

Volume of inner solution
It is easy to predict that the GUV yield depends on the number of water-in-oil droplets. Therefore, we varied the volume of the inner aqueous solution used to produce the initial emulsion. Note that all other volumes were kept constant and the same total volumeo f5 0mLo fe mulsion was added. We used inner solution volumes from 1t o2 5mLa nd as Figure 4s hows, the yield increased with volumes between 5a nd 10 mL. The size of the vesicles, however,p eakeda t2 5mLi nner solution volume. It is reasonable to assumethat the yield decreases with smaller volumes of the inner solution supplied for creating the emulsion, but it was unexpected that it decreases again at greater volumes. This phenomenonc an be attributed to the fact that ah igher volume of inner solution produces more droplets with increasing proximities to each other.T his could result in a higherp robability of spontaneous dropletf usion leadingt o larger vesicles but lower overall numbers. Indeed, this waso bserved in the case of 25 mLw hose averaged iameter is 1.5-fold more than the other conditions (Figure 4b and c).
Consequentially,t hese results suggest an opportunity to tune the size of the GUVs to an extent by changing the volumeo ft he inner solution to make the emulsion. Nevertheless, for the next experiments, we selected 7 mLf or the inner solutionv olume as it produces the highest number of GUVs with the least variation in GUV size.

Lipid concentration and incubation time
The concentration of solubilized lipids in the oil phase directly affectst he time it takes to form the monolayers at the oilwater interfaces. This, in turn, puts al imit on the speed of the overall preparation time, which can be ad isadvantage without optimization. Moreover,a ni ncomplete lipid monolayer at the interface can directly affect the yield of the GUVs produced. For example, the aqueousi nterior of the droplets can make direct contact with the aqueous outer solution insteado ft he lipid monolayer at the interface resultingi nt he unwanted release of internal contents and reduction in the overall yield. Therefore, it is important to have asufficiently assembled interfacial lipid monolayer before addition of the emulsion. In     (< 200 mm), it is not possible to increaset he overall yield with al ongeri ncubationt ime (Figure 5a). Presumably due to the unavailability of lipids neededt oc over the interfacial area of the well completely and to replenish it after dropletc rossing. To confirmt his assumption with ah igh degree of certainty,o ne would have to perform complementary simulations to investigate the minimum time required for the lipids to re-assemble at the interface after droplet crossing. For higherl ipid concentrations (200, 400, and 800 mm), the yield is similara nd reached ap lateau with am inimum of 30 min incubation time.
Considering this finding, for the next optimization we have fixed the interfacial lipid monolayer incubation time at 30 min and the lipid concentration at 200 mm.I na ddition, the range of sizes is similart ot hat of the experimentsp erformed earlier in this work (~20-30 mmi nd iameter), which demonstrates the robustnessoft he method in general (Figure 5b).

Effects of pH and temperature
pH and temperature play ac rucial role in the functionality of many biological processes. [41] As model membranes for biological cells, giant vesicles have been shown to be robust at various physiologically relevant pH andt emperature ranges. [42] Considering that this emulsion-based method has the advantage of encapsulating large biomolecules such as enzymes, [30] it is therefore possible to study the enzymatic activity of biologically relevant chemical reactions encapsulatedw ithin. To determine if it is possible to produce GUVs at such conditions, we prepared GUVs at various pH (inside and outside) andt emperature conditions. Figure 6s hows the change in yield and size with varying temperatures and pH conditions. Yields moderately increased from pH 4t o9and decreaseds ignificantly at pH 12 (Figure 6a). This can be explained due to the fact that at low and high pH values, the zwitterionic POPC lipid headgroup becomes more positivelyo rn egativelyc harged, respectively. [43] It is knownt hat charged lipids in monolayers repel each other, which may affect the formation and stability of the final GUVs. Moreover,a tp H12, the vesicles were larger with clustering and aggregation (Figure 6b and c). This is most likely due to the hydrolyzing effect induced by hydroxide ions andp ossible degradation of the lipids at such high pH environments. [44] In a furthere xperiment to ascertain the greater applicability of the methodf or producing biomimetic compartments we used the buffering agents 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), Tris, and phosphate-buffereds aline (PBS) within inner solution only.F igure S4 shows that the methodc an produce GUVs containing these widely used buffers-something which is not alwaysf easible with hydration-based methods especially in the case of salt containing PBS solutions.
Unlike pH, temperature does not seem to have ad rastic impact on the overall yield of the GUVs (Figure 6d). The yield appearst oi ncrease marginally with temperature, at least for the one component POPCl ipid composition implemented here. Although the maximum temperatureu sed in this study (37 8C) yielded ah ighern umber of GUVs, the averages ize of the GUVs ((23 AE 4) mm) was smaller than the average size of the GUVs ((30 AE 4) mm) produced at rest of the temperature conditions ( Figure 6e). Visually,t he GUVs produced at all the temperatures tested were without defects or aggregations except at 37 8Cw here some small lipid clumpsw ere seen (observed as bright red spots in Figure 6f). From the resultso btained, the inverted emulsion methodcan successfully produce biomimetic GUVs at different pH and temperature conditions-including physiological ones.

Polymers as alternative density gradient agents
As already shown, the density of the inner solution is vital for producing GUVs using the inverted emulsion-basedm ethod. A denser inner solutionc an easily pass through the oil phase and across the oil-water interface. Furthermore,adenser inner solution will helpt he GUVs settle down to the bottom of the well plate for better visualization and long-term experiments. The above experiments encapsulated sugar-based aqueous solutionst on ot only achieve this density difference, buta lso to osmotically balance the medium. The usage of sugar solutions might not be feasible if one has to encapsulate enzymes that can metabolize sugars or molecules that are sensitive to these carbohydrates. Moreover, encapsulatingc ells within GUVs for single-cell analysis, which hasg ained some interest in  [31] andh igh concentrations of sugars can affect cell viability or function.
In such scenarios, alternative compounds that will not adverselyaffect the lipid bilayer have to be implemented. Considering this, we have turned to polymers, such as polyethylene glycol (PEG) andp oly(vinyl alcohol) (PVA). PEG is aw ell-known compound with high hydrophilicity and anti-biofouling properties. [45] Elsewhere, PVAi su sed as ac oating materialt og row GUVs using the gentle hydration method. [46] Figure 7d epicts the yield and size of the GUVs produced at variousp olymer concentrations used as an inner solution (osmotically matched glucoseo utside) at different spin speeds. As expected, the GUVs yield increased with higherc oncentration of the polymers in the inner solution as the densities increased (Figure 7a  and d).N ote that for concentrationso f0 .5 to 5% the osmolarities increased from 1t o4 5mOsm and 4t o6 0mOsm for PEG and PVA, respectively.A t4 00 g,t he yield reached ap lateau for both the polymers whereas the average size of the GUVs remained the same for all the concentrations at that spin speed. We also observed that the vesicles encapsulatingP EG exhibit inward tubes, consistent with recent findings for stabilization of high spontaneousc urvature by this polymer. [47,48] Surprisingly,f rom 300 to 500 g,t he average size of the GUVs (~25 mm) is approximatelyt he same at all the polymer concentrations for both the polymers (Figure 7b and e). When comparing data from both polymers, the overall yield is higher with PEG for all conditions tested (almost doublef or the highest polymer concentration at 5 w/v %). This could be due to density differences between the two polymers and/ort hat PVAc an interfere with the lipid monolayero rb ilayer by incorporating itself acrosst he membrane. If the concentrations of 2.5-5 %P EG or PVAa re not suitablef or ap articular application, lower concentrations also yield GUVs. From the confocal images, it is observable that the GUVs produced with PVA ( Figures 7c and S5)a nd PEG (Figures 7f and S6) are morphologically similar to the GUVs produced by using sugar solutions ( Figure 2c)e xcept that the overall yield is lower.O verall,t he use of polymer solutions instead of sugar solutions to produce GUVs is possible using the inverted emulsion methoda nd PEG is am ore promising candidate.

Membranecomposition and functionality
We aimed to assess the membrane composition for any possible presence of oil in the lipid bilayer of the GUVs.T his is important for the widespread acceptability and usage of this technique. Previously,researchers have performed alpha-hemolysin-based assays on GUVs produced by using microfluidic method to confirmo il-free membranes, unilamellarity and thus, functionality. [49] We have performed similar experiments with calcein dye-filled GUVs produced using the inverted emulsion method. The plot presented in Figure S7 indicates that over ap eriod of 30 min, there is ad ecrease in the calceinf luorescence intensity inside the GUVs in the presence of alpha-hemolysin. This suggestst hat the lipid bilayer is oil-free enough for the membrane protein to incorporate and assemble. This clearlys uggests that the GUV membrane is unilamellar and functional for the integration of membrane proteins. Furthermore,t he fluorescence inside the GUVs remained the same in the absence of alpha-hemolysin, an observation that also confirms the stability (i.e.,nob ursting or leakage) of the GUVs.
An ancillary measurement to understand the composition and purity of lipid membranes is to measure their bending rigidity. [50] Here, we employed the method of fluctuation spectroscopy as described in Gracia et al. [51] However,f or systems with high sugar content as used here, one should be aware of two issues:1 )sugarsd ecrease the bending rigidity of GUV membranes [50,52,53] and direct comparison with literatured ata should taket his into account;a nd 2) at high sucrose/glucose contrasts, the density differencea cross the membraned istorts the vesicle shape and can affect the membrane fluctuation spectrum leadingt oe rrors in the assessed bending rigidity. [54] Vesicles containing6 00 mOsm sucrose in their interior and suspended in equimolar glucose solutionsa re affectedb yb otho f these effects. To avoid the error associatedw ith gravity, which often appearst ob en eglected when comparing inverted emulsion with electroformed GUVs (see, e.g.,r ef. [38]), we decreased the sucrose/glucose density gradient. For this, the externalG UV solutionw as adjustedt o5 75 mm sucrose and 25 mm glucose. This conditione nsures no gravity-associated error as assessed from the gravitational parameter introduced in Henriken andI psen. [54] The data presented in Figure 8s how that there is no significant difference between the bending rigiditieso ft he GUVs produced using electroformation (1.70 AE 0.71) 10 À19 Ja nd the inverted emulsion (1.42 AE 0.47) 10 À19 J methods. For comparison, we also plot the data obtained without the gravity correction showing an erroneous apparent increase in the bending rigidity.T he bending rigidity of invertedemulsion-based vesicles appears slightly lower than that of electroformed vesicles but the difference is not significant (confirmed with at -test). These experimentsa scertain that the lipid membranes produced by the inverted emulsion method are comparable to those of GUVs produced from conventional techniques.

Conclusion
The inverted emulsion-based methodi sr elatively an ew technique for producing giant vesicles when compared to electroformationo rg entle hydration.T his particular method of producing giant vesicles,l eaflet-by-leaflet, hass eries of advantages. Although the major benefiti st he possibility to encapsulate large (bio)molecules inside the GUVs under physiological conditions, the methodc an be optimized to produce GUVs with minimum preparation times and for high-throughput analyses.
In this work, we have optimized various parameters to reduce the time requiredt op roduce the GUVs. This was made easy due to the use of 96-well microtiter plates for parallel preparations/experiments. Our results demonstrate that density gradients by using sugars are vitally important in manipulating the overall yield of the GUVs. Alternatively,p olymers, such as PEG or PVA, can be used to replace inner sugar solutions, if necessary.W eh ave also explored the possibility of using different oils for solubilizing the lipids because having ah omogenous lipid solution in the carriero il free from aggregation is important for the formationo ft he interfacial lipid monolayer. From our findings it is evident that mineral oil is the most suitable amongst all the other oils (see Ta ble S1 for more details). Another significant finding of this work is that, with the exception of pH 12, there is little effect of pH and temperature on the yield and size of the GUVs at least fort he lipid composition tried here. This is beneficial for encapsulatingc omponents that are active only at specific pH and temperature conditions. For example, it is possible to encapsulate enzymesi nside the GUVs at low temperatures when they are inactive and activate them later under the microscope to study the kinetics of conversion and the effect of the resultantm olecules on the lipid bilayer. We also note the robustnesso fu sing the 96-well plate to perform the invertede mulsion procedure. This is evidenced by the relatively small error barsf or each experiment and when 600 mOsm sucrose is used the GUV yields and sizes are comparable across experiments.
When implementing the inverted emulsion method in microtiter plates, the following conditions can be used for optimal GUV yields:2 0mLo fm ineral oil with 200 mm lipids to form the interface over 600 mOsm outer glucose solution for a period of 30 min. This is followedb y7mLi nner sucrose solution at 600 mOsm mixed with 250 mLl ipid mineral oil to form the emulsion. Then 50 mLe mulsion is added on top of the interface and subjected to centrifugation at 200 g for ap eriod of 3min. The total time taken for optimal production of GUVs is 35 min. We also calculated the minimum time needed to produce GUVs, which was 18 min by reducingt he incubation period from 30 to 15 min. Note that in this case as uboptimal number of vesiclesi so btained but it is better suited for timesensitivee xperiments.
The mostl ikely hurdlef or widespread acceptance of this technique could be the possibility of oil present in the lipid bilayer that could potentially alter the biophysical properties of the membrane. To this end, we have addressed this concern (at least for our setup) by incorporating af unctionally active membrane protein. Alpha-hemolysin in itsm onomeric form in solutionc an assemble as ah eptamer into the lipid bilayer to form apore. [55,56] Not only have we showed that this is possible within the lipid bilayer of the GUVs produced by using an inverted emulsion, but also that the time and concentration requiredf or total loss of fluorescence inside the GUVs is comparable to that of electroformed GUVs. [26] Along with this assay, our membrane bending rigidity analysiso ft hese GUVs supports the conclusiont hat the GUVs are oil-free (or,a tl east, that mineral oil does not alter their mechanicalp roperties) and are therefore biomimetic. Furthermore, performing the . The vesicle interior contains 600 mOsm sucrose. Data are shownfor preparations where the gravity effects were overcome by replacing the ex-ternal600 mOsm glucosesolution (black symbols;hatched area) with asolution of 25 mOsm glucose and 575 mOsm sucrose( dark cyan). Mean data are shown with error bars taken from the standard deviation(n ! 10). Snapshots above the graph show phasec ontrastimagesofG UVs under the respective conditions. Notethat for 600 mOsm glucose in the outer solution, the contrast is enhanced due to an increased refractivei ndex difference. Scale bars: 50 mm.
ChemBioChem 2019, 20,2674 -2682 www.chembiochem.org 2019 The Authors. Publishedb yWiley-VCH Verlag GmbH &Co. KGaA, Weinheim methodi n9 6-well plates provides the advantage of scalability for greater statistically significant results and high-throughput experiments. Overall,w eb elieve that our findings will aid in the widespread adaptation of this method for membrane studies where high encapsulatione fficiencies, physiological conditions and biomimetic GUVs are required.
Lipids in oil preparation:C hloroform-solubilized POPC (99.9 %) and DilC18 (0.01 %) were used to make lipid-oil solution. Briefly, the preparation of lipid-mineral oil solution starts with forming a thin film of dried lipids in ac lean glass vial by evaporating the chloroform under argon flow.Ad esiccator was used to dry the lipids for ap eriod of 1hand remove any leftover traces of chloroform. Finally,m ineral oil was added to the glass vial under low humidity (< 10 %) conditions using an AtmosBag (Sigma-Aldrich) filled with nitrogen gas and monitored with ah umidity gauge (Klimalogg Pro, TFAD ostmann). This was followed by as onication step for better solubilization of lipids in the oil. At this stage, the total concentration of the lipids in the mineral oil can be from 50 to 600 mm,d epending on the amount of mineral oil added. The lipid-mineral oil solution was then incubated overnight at room temperature in the dark to ensure that the lipids are completely dissolved. These lipid-oil solutions can be stored for up to one week at 4 8Cand brought to room temperature before usage.
Surface treatment of the microtiter well plates:T oa chieve highthroughput and parallelized experiments, the entire inverted emulsion method has been performed in 96-well microtiter plate. Surface treatment of the glass bottom of the wells is important to avoid adhesion and bursting of the GUVs. Wells within the plates are pre-coated using aqueous solutions of b-casein (2 mg mL À1 ). Typically,3 0mLo ft he coating solution was added to each well and allowed to dry in the presence of vacuum for 30 min. The well was gently washed multiple times with the outer solution before making the lipid interface. BSA at 2mgmL À1 was used as ac oating material in experiments involving high (12) or low (4) pH where the b-casein coating is not stable. [57] Inner and outer aqueous solutions:A ll aqueous solutions were made using MilliQ Millipore water.I nner solutions contained either sucrose (50, 300, 600, and 900 mm)o rP EG or PVAa t0 .5, 1, 2.5, 5 w/v %. For pH-based experiments, the solutions were buffered by using HEPES at 20 mm concentration and NaOH as well as acetic acid were used to adjust the pH. Inner solutions were also made with either HEPES (50, 100, 200, 400 mm), Tris (50, 100, 200, 400 mm), or PBS, and sucrose was added to provide enough density.F or all the above inner solutions, corresponding concentrations of glucose solutions were used as the outer solutions to maintain isoosmotic conditions across the membrane.
Inverted emulsion method:I nitially,5 0mLo fg lucose (outer) solution were added to the pre-coated wells in the microtiter plate. This was followed by the addition of 20 mLo ft he lipid-oil mixture on top of the glucose. The entire setup was allowed to incubate for ap eriod of 5, 15, 30, or 60 min to form an interfacial lipid monolayer (surface area of~192 mm 2 ). Then 250 mLo ft he same lipidoil mixture were added to a1 .5 mL Eppendorf tube. To this tube, 1, 3, 5, 7, 10, or 25 mLi nner solution were added and then agitated mechanically along as tandard Eppendorf tube rack to yield a water-in-oil emulsion. Note that four emulsion preparation techniques were compared. Figure S8 shows that mechanical agitation yields the narrowest size distribution compared to vortexing and pipetting by hand. Sonication did not provide any useable droplets. We also show that the sizes of the emulsions result in similar sizes of the final GUVs (Figure S8 b). Based on these findings and the fact that sonication and vortexing can cause protein degradation, we used mechanical agitation in this work, but for some applications vortexing may be sufficient. An aliquot of 50 mLo ft he emulsion was pipetted into the wells containing the lipid monolayer interface. Immediately after this, the microtiter plate was transferred into ac entrifuge. Based on the experiment either 50, 100, 200, 300, 400, 600, or 1200 g force was applied for periods of 30 s, 1min, 3min, or 10 min.
Membrane composition and functionality:G UVs containing 10 mm of the fluorescent dye calcein were incubated with 2.5 mgmL À1 alpha-hemolysin for ap eriod of 30 min. Confocal fluorescence images were acquired before and after the assay.F or bending rigidity measurements, GUVs containing 600 mOsm sucrose inside and 600 mOsm glucose outside were prepared by using the inverted emulsion method and then diluted with sucrose to achieve an out solution of 575 mOsm sucrose and 25 mOsm glucose as required. Electroformed GUVs were produced by using ap rocedure explained elsewhere with minor modifications:2V p-p at 10 Hz for ap eriod of 3h in 600 mOsm sucrose and diluted as above when required. [55] Measurements and subsequent analysis for bending rigidity measurements have been performed by using the methods developed by Garcia et al. [51] Acquisitions were performed by using an inverted microscope (Axiovert 135 Zeiss, Germany) equipped with 20 objective and af ast digital camera (eCMOS PCO.edge, PCO AG, Germany) with an exposure time of 200 msa t20f rames per second.
Microscopy:T he produced GUVs can be directly observed within the microtiter plate using an inverted microscope without any further sample preparation. Here, the samples were observed and images acquired using ac onfocal microscope (Leica microsystems TCS SP8, Wetzlar,D E) equipped with a6 3 /1.4 NA water immersion objective. At otal of six confocal images were randomly taken in each well and with the same field of view for comparable yield assessments. The diameter and population of individual GUVs were measured using the Vesicle-Analyser-Pro software. [58] Note that multiple z-stacks were acquired at each position to ensure all vesicles were imaged and that each vesicle was measured at the equatorial plane. GUVs with diameters below 10 mmw ere excluded. DiIC 18 fluorescence in the membrane was excited by using a 552 nm diode laser with emission collected at 562-635 nm and calcein present inside the GUVs by a488 nm diode laser with emission collected at 498-535 nm.