Live Follow-Up of Enzymatic Reactions Inside the Cavities ofSynthetic Giant Unilamellar Vesicles Equipped with MembraneProteins Mimicking Cell ArchitectureMartina Garni,† Tomaz Einfalt,† Roland Goers,†,‡ Cornelia G. Palivan,*,† and Wolfgang Meier*,†

†Department of Chemistry, University of Basel, Mattenstrasse 24a, CH-4002 Basel, Switzerland‡Department of Biosystems Science and Engineering, ETH Zurich, Mattenstrasse 26, CH-4058 Basel, Switzerland

*S Supporting Information

ABSTRACT: Compartmentalization of functional biological units, cells, andorganelles serves as an inspiration for the development of biomimeticmaterials with unprecedented properties and applications in biosensing andmedicine. Because of the complexity of cells, the design of ideal functionalmaterials remains a challenge. An elegant strategy to obtain cell-like compartmentsas novel materials with biofunctionality is the combination of syntheticmicrometer-sized giant unilamellar vesicles (GUVs) with biomolecules becauseit enables studying the behavior of biomolecules and processes within confinedcavities. Here we introduce a functional cell-mimetic compartment formed byinsertion of the model biopore bacterial membrane protein OmpF in thicksynthetic membranes of an artificial GUV compartment that enclosesas amodelthe oxidative enzyme horseradish peroxidase. In this manner, asimple and robust cell mimic is designed: the biopore serves as a gate thatallows substrates to enter cavities of the GUVs, where they are converted intoproducts by the encapsulated enzyme and then released in the environments of GUVs. Our bioequipped GUVs facilitate thecontrol of specific catalytic reactions in confined microscale spaces mimicking cell size and architecture and thus provide astraightforward approach serving to obtain deeper insights into biological processes inside cells in real time.

KEYWORDS: giant unilamellar vesicle, enzyme reaction, membrane protein, synthetic membranes

Eukaryotic cells, as complex assemblies with highly definedstructures based on organelles and a cytoplasm with a

variety of biomolecules enclosed by the cellular plasma mem-brane, serve as an inspiration for the development of newfunctional systems.1 In this respect, the bottom-up approach tomimic the hierarchical organization and functionality of variouscellular structures or even of entire eukaryotic cells2 is the focusfor producing new functional materials with a large variety ofapplications, such as therapeutics,3 catalysis,4 biosensing, andsurface technology.5 Biomimetic materials are created by com-bining synthetic assemblies with active biomolecules or com-plexes (proteins, enzymes, DNA).2 Self-assembly processes thattake place in biological systems can be used to obtain biomime-tic materials in which the biomolecule induces functionalitywhile the supramolecular assemblies (micelles, nanoparticles,tubes, liposomes, polymersomes, planar membranes,6,7 ormicrometer-sized giant unilamellar vesicles (GUVs)) serve asstable matrices.8 The development of functional biomimeticsystems requires an understanding of the various reactions thatoccur within the cellular microenvironment and at cellularinterfaces.9,10 In this respect, synthetic compartments are veryappealing supramolecular assemblies because their architectureoffers well-defined and protected reaction spaces for encap-sulated catalytic compounds (enzymes, proteins), similar to the

compartimentalization of cells or natural organelles. Rudi-mentary cell mimics in the micrometer range known as lipidGUVs are among others used to study lipid membrane per-meability,11,12 membrane protein reconstitution,13 and con-fined enzymatic reactions.14 Polymer compartments (polymer-somes with nanometer sizes and synthetic GUVs with micro-meter sizes) self-assembled from amphiphilic block copoly-mers15,16 have become an elegant solution for the developmentof functional compartments with various advantages over lipidcompartments (liposomes and lipid GUVs), such as improvedstability and the huge variety of polymers, whose chemistry canbe tailored to achieve specific properties (stimuli-responsiveness,flexibility, crystallinity).15 Polymersomes have been developedfor various applications, such as conventional drug deliverysystems,17,18 catalytic nanocompartments (or nanoreactors),19−21

biosensors,22 and even artificial organelles that act as simplemimics of cellular organelles.23 Interestingly, the study ofenzymatic reactions taking place inside polymersomes as con-fined reaction spaces with sizes of less than 300 nm indicatedthat under specific conditions it was possible to induce anincrease in the bioactivity of encapsulated enzymes,24 while the

Received: March 12, 2018Published: August 26, 2018

Research Article

pubs.acs.org/synthbioCite This: ACS Synth. Biol. 2018, 7, 2116−2125

© 2018 American Chemical Society 2116 DOI: 10.1021/acssynbio.8b00104ACS Synth. Biol. 2018, 7, 2116−2125

Dow

nloa

ded

via

AIX

-MA

RSE

ILL

E U

NIV

on

Sept

embe

r 25

, 201

8 at

11:

34:4

4 (U

TC

).

See

http

s://p

ubs.

acs.

org/

shar

ingg

uide

lines

for

opt

ions

on

how

to le

gitim

atel

y sh

are

publ

ishe

d ar

ticle

s.

presence of crowding agents (i.e., poly(ethylene glycol)) insidethe cavity induced a decrease in bioactivity.24 Despite the useof polymersomes in various applications, there are still impor-tant unanswered questions, such as the following: How do bio-molecules (enzymes, proteins, substrates) behave inside con-fined compartments with sizes that mimic those of cells? Is thereany interaction between encapsulated compounds and compart-ment membranes? Do encapsulated compounds diffuse freelyinside the cavities, or are they attached to the membrane? If so,does this possible interaction affect the activity and stability ofthe encapsulated compounds? The development of advancedfunctional compartments requires detailed responses to suchquestions in order to understand the complexity and specificityof natural biocompartments.To answer such questions, polymer GUVs can serve as a

platform that allows deeper insight into the molecular pro-cesses (reactions, interactions) that take place inside confinedcavities, in their membranes, and at the interface, by real timevisualization using microscopic techniques. In comparison toGUVs, nano-sized polymersomes can give only indirect insightinto membrane protein reconstitution and localization ofenzyme reactions. GUVs can be measured directly by establi-shed microscopic techniques, as their sizes match cell dimen-sions, and reactions or interactions can thus be visualized.Therefore, they represent a simple model system for studyingthe above-mentioned processes and interactions in a cell-likemanner. In this respect, polymer GUVs are of particular inter-est because their higher mechanical stability, than lipid-basedGUVs, allows these processes to be studied for longer periodsof time. Unlike lipid-based GUVs, there are currently onlya few examples of polymer GUVs being used as func-tional platforms.25 For example, synthetic GUVs based onpoly(butadiene)-block-poly(ethylene oxide) (PBut2.5-b-PEO1.3)loaded with hydrophilic dyes, liposomes (DPPC), andpolymersomes (PBut1.2-b-PEO0.6) allowed fast, selectivelytriggered release due to a light-induced increase in the osmoticpressure, which resulted in rupture of the GUVs.26 Despite thepotential application of this system for directed delivery andcontrolled local dosing of active compounds, applications ofthis approach are limited by the fast, irreversible decomposi-tion of the GUVs, because of the membrane to volume ratiocompared with nano-sized vesicles.PB-b-PEO polymer GUVs can mimic structural and func-

tional eukaryotic cells by encapsulating enzyme-filled intrinsicallyporous PS-b-PIAT polymer nanoreactors with free enzymes andsubstrates to fulfill a three-enzyme cascade reaction inside themulticompartmentalized structures.27 However, that workprovided no information about the localization of the enzymes,and only the fluorescence product was observed by spinning diskconfocal microscopy. Furthermore, transmembrane exchange ofsubstrates and products occurred because of the intrinsicpermeability of the membrane and, unlike in cell membranes,was not mediated by membrane proteins.Proton transport through polymer poly(ethylene oxide)

polybutadiene (PEO−PBD) GUV membranes has beenachieved by incorporating two different synthetic pores self-assembled from either a dendritic dipeptide or a dendritic ester.However, assembly of dendrimers into stable helical pores wasnot possible in the case of poly(2-methyloxazoline)-block-poly-(dimethylsiloxane)-block-poly(2-methyloxazoline) (PMOXA−PDMS−PMOXA)-based GUVs, either because of the signifi-cant solubility difference between PBD and PDMS or thedifferences between the di- and triblock architectures.28

A more challenging approach for the development of cellmimics based on synthetic GUVs is to insert biopores andmembrane proteins into their membranes to induce enhancedpermeability to specific ions or molecules or to support in situreactions while preserving the three-dimensional architecturesof the GUVs. This approach provides the possibility of under-standing the biological function of the biopores or membraneproteins and visualizing transmembrane processes in real time.An example is GUVs based on the rendering of a PMOXA−PDMS−PMOXA triblock copolymer permeable for Ca2+ ionsby insertion of an unselective ionophore, lasalocid A, and ahighly selective ionophore, N,N-dicyclohexyl-N′,N′-dioctadec-yl-3-oxapentane-1,5-diamide,29 while insertion of the channel-forming peptide alamethicin allowed unselective transport ofcations and anions.29 The insertion and functionality of thesmall pore-forming peptide gramicidin (gA) in syntheticPMOXA−PDMS−PMOXA GUV membranes were visualizedin real time by the fluorescence change of the encapsulated dyeinside the inner cavities.8 Interestingly, functional insertion ofgA could be achieved in thick membranes up to 5 times greaterthan the size of gA.8 The ability of synthetic membranes to allowinsertion of biopores and membrane proteins (KcsA, OmpF,and AqpZ) that are several times smaller than the membranethickness is the result of their intrinsic fluidity, which over-comes the significant hydrophobic mismatch if the copolymersare appropriately selected chemically, as in the case of diblockand triblock PMOXA−PDMS−PMOXA copolymers.25,30,31

However, to the best of our knowledge, the proof of a functionalreconstitution of membrane proteins in synthetic GUVs withcell-like membranes has not yet been established.Here we present the functional reconstitution of a mem-

brane protein in thick and stable synthetic membranes of GUVsin order to support cell-like functionality: an enzyme reactionthat takes place within a GUV microsized cavity (Figure 1).Membrane protein functionality inside GUV membranes rep-resents the key factor enabling diffusion of the substrates/prod-ucts of the reaction and therefore supporting the in situ reac-tion in a cell-like manner.We selected a cysteine double mutant of the bacterial outer

membrane protein F (OmpF) as the model membrane proteinbecause it has various advantages such as well-known stabilityand a pore size that allows diffusion of molecules up to 600 Da.Controlled labeling of the cysteine residues within the OmpFbackbone by fluorophores allows the reconstitution of OmpFto be monitored.32 In addition, OmpF has already beeninserted into PMOXA−PDMS−PMOXA polymersome mem-branes with sizes of around 200 nm either as the wild type32 oras a mutant33 and demonstrated to allow molecular diffusion.Therefore, we chose this type of amphiphilic copolymer togenerate GUVs. The crucial step in our approach was tosimultaneously encapsulate an enzyme inside GUVs and insertthe OmpF double mutant to provide the transmembraneexchange of substrates and products for the in situ enzymaticreaction. The molecular cutoff of OmpF pores plays a dualrole: (i) it prevents the model enzyme horseradish peroxidase(HRP) from escaping from the GUV cavity and (ii) it allowsdiffusion of the substrates (H2O2, Amplex UltraRed (AR)) andthe fluorescent product of the enzymatic reaction (resorufin-like product). The reaction has been characterized on wholeGUV populations by flow cytometry, while conversion into thefluorescent product inside individual GUVs has been detectedby confocal laser scanning miscroscopy (CLSM) in real timeby time-lapse recording. Real time visualization provides deep

ACS Synthetic Biology Research Article

DOI: 10.1021/acssynbio.8b00104ACS Synth. Biol. 2018, 7, 2116−2125

2117

insight regarding the membrane protein and enzyme localiza-tion, synthetic membrane permeability, and calculation of theapparent rate constant. Membrane protein functionality in thesynthetic membranes of GUVs together with the in situ enzy-matic reaction represents a necessary step toward the produc-tion of more sophisticated reactions inside functional compart-ments. Such synthetic GUVs equipped with membraneproteins represent robust functional assemblies that benefitfrom the stability of the synthetic membrane and the activity ofthe biomolecules. Thus, they open the opportunity to providedetailed insight into the design of multifunctional systems thatserve as simple models of artificial cells.Insertion of the Channel Protein OmpF. An essential

step toward the production of synthetic GUVs with controlledmembrane permeability and a confined space for enzymaticreactions is the reconstitution of OmpF-M, the double mutantof the channel porin OmpF. In order to visualize reconstitutionof OmpF-M in the PMOXA7−PDMS49−PMOXA7 polymermembrane of GUVs by CLSM, 300 mM sucrose was used as adefault rehydration solution, since it allows GUVs to settleonce they are dispersed in phosphate-buffered saline (PBS)(150 mM, pH 7.4) (see the Supporting Information (SI)).In addition, the borosilicate glass surface was plasma-treated toallow GUVs to settle even if the sucrose could diffuse out ofthe GUV cavities because of the reconstituted OmpF. The cys-teine residues replacing the native amino acids K89 and R270in OmpF-M were conjugated with a fluorophore (Atto 488maleimide),33 and nonreacted Atto 488 maleimide was removedby filtration and dialysis. Atto 488 modification of OmpF-M hasbeen evaluated by fluorescence correlation spectroscopy (FCS)and gel electrophoresis. FCS allows the determination of thediffusion time for fluorescent molecules through the confocal

volume of the confocal laser scanning microscope and thus theirbinding to compounds/assemblies with significantly highermolecular weights by measuring the respective diffusiontimes.34,35 Investigating the efficient chemical coupling reac-tion and successful purification of OmpF is of importancebecause free dye molecules could unspecificially stain the poly-mer giant membranes. By dividing the molecular brightness(counts per molecule (CPM), in kHz) of Atto 488 maleimidebound to OmpF-M in 1% n-octyl-β-D-glucopyranoside (OG)in PBS (4.5 ± 0.3 kHz) by Atto 488 in PBS (3.1 ± 1.23 kHz),an average labeling efficiency of 1.5 Atto 488 molecules perOmpF-M monomer was obtained. No residual fluorescencewas observed after purification when Atto 488 was added towild-type OmpF in 3% OG or to 3% OG alone, thus provingthat the maleimide fluorophore binds specifically to the cys-teine residues of OmpF-M (Figures 2A and S1).To determine whether OmpF was inserted into GUV mem-

branes, CLSM measurements were carried out using GUVsprepared in the presence of Atto 488-labeled OmpF-M (Atto488−OmpF-M) and compared with the results for GUVs inthe presence of unlabeled OmpF-M (Figure 2B). Membranesof the GUVs with unlabeled OmpF-M remained dark, whilethose of GUVs with Atto 488−OmpF-M presented a distinctfluorescent membrane (shown in green in Figure 2B), whichclearly indicates successful insertion of Atto 488−OmpF-M.The use of the film rehydration technique for GUV prepara-tion yielded a polydisperse GUV population (size, lamellarity),as observed by CLSM and FACS (Figure 2B,C), which helpedus investigate influence of size on the enzyme apparent rateconstant. In future studies, specific size ranges could be selectedand polydispersities improved by the use of droplet transferandmicrofluidic approaches,36,37 although these might not be

Figure 1. Design of (left) a natural cell and (right) a synthetic giant unilamellar vesicle (GUV) with encapsulated enzymes engineered for selectivepermeability to substrates and products based on reconstitution of outer membrane protein F (OmpF) to visualize a model enzymatic reactioninside the cavity.

ACS Synthetic Biology Research Article

DOI: 10.1021/acssynbio.8b00104ACS Synth. Biol. 2018, 7, 2116−2125

2118

compatible with the biomolecules because of the use of organicsolvents. In addition to CLSM measurements, insertion of Atto488−OmpF-M was evaluated by flow cytometry, which haspreviously been used for analysis of lipid GUVs.38 The sizes ofthe prepared GUVs were calibrated using the forwardscattering (FSC-A) of polystyrene latex beads in the sizerange of 2−10 μm (Figure 2C). Taking into account the GUVpopulation between 3−10 μm, a significant shift in fluores-cence intensity was observed for the GUV population withreconstituted Atto 488−OmpF-M compared with the GUVpopulation without OmpF (Figures 2D and S2). The popula-tion of GUVs smaller than 3 μm was discarded on the basis ofthe CLSM measurements, where GUVs in this size range weretoo small for precise measurements inside their cavities.Enzyme Encapsulation Inside the Cavities of GUVs.

We were interested in encapsulating enzymes within the cavi-ties of GUVs and analyzing how they behave inside a cell-likeconfined space lacking the crowded environment of the cyto-plasm. In addition, the enzymatic reaction was used to verifydiffusion of the substrates through the OmpF-M pores.In order to follow the encapsulation of the model enzyme

horseradish peroxidase in GUVs, lysine residues of HRP wereconjugated with Atto 488 succinimide ester, and the unreactedfluorophore was removed by dialysis and filtration. Theconjugation and purity of the Atto 488−HRP conjugate wasestablished with FCS (Figure 3A) and SDS-PAGE (Figure S3).Atto 488 in solution had a typical diffusion time (τD) of 44.3 ±17 μs and molecular brightness (CPM) of 4.4 ± 0.3 kHz,

whereas Atto 488−HRP presented a significantly longer τD of252 ± 43 μs with a molecular brightness (CPM) of 3.4 ±0.5 kHz. The free fraction of Atto 488 in the Atto 488−HRPsolution was determined with a two-component fit of the auto-correlation curve to be 15 ± 11% (Figure 3A).After encapsulation of Atto 488−HRP in 300 mM sucrose

solution and settling of the GUVs to the imaging chambersurface (see the SI), most of the GUVs measured by CLSMshowed fluorescence inside their cavities (Figure 3B), indi-cating successful HRP encapsulation and free diffusion of theenzyme. Not all of the GUVs had the same amount of HRPencapsulated inside their cavities when observed by CLSM.Because of the nature of the film rehydration method used for thepreparation of GUVs, different localizations of Atto 488−HRPare expected. In addition, GUV membranes presented only lowfluorescence intensity because of the reflection of fluorescentlight between the GUV membrane−cavity interface and a minorenzyme−membrane interaction. No fluorescence was observedoutside the GUVs, which demonstrates successful purification ofthe GUVs after enzyme encapsulation. Self-assembly of theGUVs with encapsulated enzyme is a statistical process, meaningthat not all GUVs have the same amount of Atto 488−HRPpresent inside their cavities.To visualize in more detail the dynamics of enzyme

encapsulation in the GUVs, we performed FCS measurementsinside the cavity of Atto 488−HRP-loaded GUVs (Figure 3C).The encapsulated HRP molecules have diffusion times (273 ±144 μs) similar to those of free HRP in sucrose (252 ± 43 μs),

Figure 2. Labeling and reconstitution of OmpF-M. (A) FCS autocorrelation curves of Atto 488 in PBS (blue) and OmpF-Atto 488 in 3% OG(green). The dashed lines are the experimental autocorrelation curves, and the solid lines are fits. The curves have been normalized to 1 to facilitatecomparison. (B) CLSM micrographs of GUVs: (a) without and (b) with labeled OmpF-M (Atto 488) reconstituted into the synthetic membrane.(C) Flow cytometry analysis of HRP-loaded GUVs containing OmpF-M (green), HRP-loaded GUVs without OmpF-M (dark gray), andpolystyrene latex beads with diameters of 2, 3, 5, 7, and 10 μm (visualized in increasing size from light purple to dark purple). (D) Flow cytometryanalysis of PMOXA−PDMS−PMOXA GUVs with (green) or without (gray) Atto 488−OmpF-M.

ACS Synthetic Biology Research Article

DOI: 10.1021/acssynbio.8b00104ACS Synth. Biol. 2018, 7, 2116−2125

2119

resulting in an overlap of their curves (Figure 3C, green andblack curves). This confirmed the successful enzyme encap-sulation (at least 10 GUVs were measured by FCS) and nochange in the dynamics of the enzyme inside the confinedspace of the GUVs (Figure 3C).Enzyme Reaction Inside GUV Cavities. To address the

question of enzyme functionality within GUV cavities, the rateof oxidation of Amplex UltraRed by HRP in the presence ofH2O2 was first studied in solution by fluorescence spectros-copy, and the enzyme activity for HRP in PBS was compared tothat for HRP in 300 mM sucrose, which was used for GUVformation (Figure S4). There was no evident difference betweenthe two media. Furthermore, without H2O2 as a cosubstrate,there was barely any conversion of Amplex UltraRed into thefluorescent resorufin-like product. Similarly, there was no sub-strate conversion and therefore no fluorescence signal when noAmplex UltraRed substrate or no enzyme was present insolution (Figure S4).This model enzyme reaction has been recently studied in

synthetic compartments with sizes in the nanometer range,namely, in polymersomes equipped with OmpF-M.33 As wemodified this method for the formation of OmpF-M-equippedpolymersomes by using sucrose solution instead of PBS for thepreparation of GUVs and the addition of OmpF-M in bidi-stilled water instead of PBS (since the presence of salts dis-turbed the GUV formation), we were interested in determiningwhether these modifications affected the OmpF-M recon-stitution and further enzymatic reactions. Therefore, we com-pared the enzymatic reaction inside polymersomes under theabove-mentioned conditions used for GUVs with those previouslyreported for polymersomes by fluorescence spectroscopy.32,33

For this, OmpF-equipped polymersomes loaded with HRPwere prepared and characterized (Figures S5 and S6).The in situ enzymatic reaction was not affected, which indi-cates that the sucrose solution together with the addition of theOmpF-M in bidistilled water used for the GUVs experiments(see below) did not affect the functional reconstitution ofOmpF or the activity of the encapsulated enzyme withinpolymersomes (Figure S7). This information tells us that thereconstitution of OmpF-M and the enzyme reaction should notbe hindered by the conditions used for preparing the GUVs.Monitoring the enzymatic reaction by CLSM for up to

12 min demonstrated that the membrane protein OmpF isfully functional in the synthetic membrane of GUVs and that itrenders the membrane permeable to the substrates and produ-cts. We produced videos with an image recording time of15.7 s/image and 1 s delay. To establish a starting point (t = 0),the first image was recorded before addition of the substrates.Then the enzymatic reaction was initiated by adding the sub-strates (Amplex UltraRed and H2O2) to the external solution,and the increase in the fluorescent signal associated with theproduct of the reaction was visualized by CLSM (Figure 4 topand Video S1). The same procedure was carried out for HRP-loaded GUVs without inserted OmpF-M (Figure 4 bottom andVideo S2).The increase in the fluorescence intensity within GUVs

equipped with OmpF-M results from the formation of thefluorescent resorufin-like product and proves the functionalinsertion of OmpF-M that permits molecular transport throughthe synthetic membrane of the GUVs. On the contrary, therewas no increase in fluorescence intensity inside GUVs withoutinserted OmpF-M, since the substrates, especially Amplex

Figure 3. (A) FCS autocorrelation curves for Atto 488 in sucrose (blue) and Atto 488−HRP in sucrose (green). The dashed lines are experimentalautocorrelation curves, and the solid lines are fits. The curves have been normalized to 1 to facilitate comparison. (B) CLSM images of polymerGUVs with Atto 488-labeled HRP encapsulated. In each panel, the left image is the fluorescence channel, and the right image is the transmissionchannel. (C) FCS autocorrelation curves for Atto 488 in sucrose (blue), Atto 488−HRP in sucrose (black), and Atto 488−HRP in GUV (green).The dashed lines are experimental autocorrelation curves, and the solid lines are fits. The curves have been normalized to 1 to facilitate comparison.

ACS Synthetic Biology Research Article

DOI: 10.1021/acssynbio.8b00104ACS Synth. Biol. 2018, 7, 2116−2125

2120

UltraRed, were unable to pass through the synthetic mem-brane. Over time, the fluorescence inside the cavity of theOmpF-equipped GUVs decreased slightly (see the changefrom 3 to 6 min in Figure 4 top) because the resorufin-likeproduct of the enzyme reaction diffused out of the GUVsthrough the OmpF-M pores. As expected, after a certain timethe fluorescence inside the cavity of OmpF-equipped GUVsbecame equal to the fluorescence intensity of the environmentof the GUVs, which was not the case for GUVs withoutOmpF-M, where there was almost no fluorescence signal insidethe GUVs (the core of the GUV in Figure 4 bottom is dark).Both GUVs with OmpF-M inserted and those without OmpFshowed a fluorescent membrane over time as a result of theinteraction of small amounts of the resorufin-like product withthe GUV membrane. In addition, a second factor that causedthe membrane to appear fluorescent was the scattering andreflection of fluorescent light between the GUV membrane−cavity interface. It should be noted that in the following dataassessment, the fluorescence intensity of the membrane of theGUVs was not taken into account. The passive accumulationof the fluorescent resorufin-like product in the background wasused as a criterion to determine whether the enzyme reactiontook place within the cavity of the GUVs. The reaction wasjudged to have occurred only if the fluorescence intensity washigher inside the GUVs than in the background. Both phenom-ena were considered in the data assessment, and consequently,the membrane fluorescence was disregarded and the fluores-cence intensity of the background was subtracted.For each image of the sequences, the mean fluorescence

intensity of a region of interest (ROI), depending on the indi-vidual size of the GUVs, both inside the cavity of each GUV(Figure S8) and outside the GUVs (background) was calcu-lated using the program ImageJ. The mean fluorescence inten-sity values of the background over time were subtracted fromthe values inside the GUVs cavities (Figures 5A,C, S8, and S9).Around 70% of the GUVs with reconstituted OmpF-M showedfluorescence higher than the background inside their cavities,indicating that the enzyme reaction had occurred, whereas thefluorescence intensity inside the cavities of ∼70% of the GUVswithout OmpF showed less fluorescence intensity than thebackground (Table S1). Reasons why some GUVs did notwork as expected could be inhomogeneous encapsulation of theenzyme, membrane defects, and different amounts of reconsti-tuted OmpF resulting from the film rehydration technique.

However, despite the limitations of the film rehydration tech-nique, we observed that if the OmpF was reconstituted the cavi-ties of these GUVs, they showed a higher fluorescence activitycompared with the GUVs without OmpF.In order to estimate the apparent rate constants, kapp, for the

GUVs with and without OmpF, a linear regression of thefluorescence increase within the first 2 min was performed onall the curves, where each curve corresponded to a single GUVwith background fluorescence intensity subtracted. Only GUVswith fluorescence higher than the background in case of theOmpF-M-equipped GUVs (70%) were taken into account, andvice versa for the GUVs containing no OmpF.An apparent rate constant of kapp ≈ 0.011 s−1 was obtained

for GUVs with reconstituted OmpF-M (Figure 5B). In con-trast, for GUVs without OmpF, the negative apparent rateconstant was an order of magnitude lower (kapp ≈ −0.0036 s−1;Figure 5D). The increase in the relative fluorescence intensityfor OmpF-equipped GUVs clearly indicates successful in situbioactivity of HRP (Figure 5A,B), while the decrease towardnegative values of the curve for the GUVs without OmpF isdue to an increase in the fluorescence intensity of the back-ground that results from the HRP activity at the exterior of theGUVs (Figure 5C,D).To gain further information on the membrane permeabiliza-

tion by insertion of the OmpF-M pore, the Amplex UltraRedenzymatic assay was applied in combination with flow cyto-metry analysis. Oxidation of Amplex UltraRed was observedinside the cavities of HRP-loaded GUVs equipped with OmpF-M, in agreement with the results obtained by CLSM, and thesignificant shift of the fluorescence for HRP-loaded GUVsequipped with OmpF-M compared with HRP loaded GUVswithout OmpF-M clearly proved that the porins allow sub-strates to efficiently diffuse into the cavities of the GUVs(Figures 6 and S10).Intensities measured by FACS and CLSM are not directly

comparable. FACS represents the average fluorescence inten-sity of all measured GUVs, whereas CLSM images representonly a part of the whole sample, where our criteria were uni-lamellar GUVs above 5 μm in diameter. Not all of the GUVsobserved by CLSM worked as expected, as described above.FACS represents an average intensity of the whole sample popu-lation, and hence, the difference in signal intensity betweenFACS and CLSM is expected. However, the trend observed bythe two techniques is the same.

Figure 4. CLSM fluorescence micrographs of single GUVs obtained over time by recording images at certain time points after initial addition of thesubstrates H2O2 and Amplex UltraRed: (top) GUV with reconstituted OmpF-M, (bottom) GUV without OmpF-M.

ACS Synthetic Biology Research Article

DOI: 10.1021/acssynbio.8b00104ACS Synth. Biol. 2018, 7, 2116−2125

2121

In regard to monitoring of the enzymatic reaction by CLSM,videos over time are recorded, and the fluorescence activityover time is measured for single GUVs directly after theaddition of the substrates. In contrast, FACS measurements arebased on the flow-through principle. This means that thewhole GUV population with the added substrates is measured,and the fluorescence intensities of each GUV at different timepoints are averaged during the measurement. Because of thediffusion of the substrates and product into and out of theGUVs, it is challenging to measure the point where the fluo-rescence of each GUV is highest. However, despite the difference

in measuring methods, the 2-fold increase in the relativefluorescence intensity in CLSM is comparable to the 2-foldshift observed in FACS. Thus, these CLSM and flow cytometryresults confirm the necessity of functional membrane proteinreconstitution inside polymer GUV membranes for productionof the fluorescent product by the enzyme confined inside theGUV cavities.

GUV Size Distribution. Next, we were interested inassessing whether the size of the GUVs affects the enzymaticreaction that takes place within their cavities. To determine thecorresponding sizes of the GUVs for a comparison between thedata sets from GUVs with reconstituted OmpF-M and GUVswithout OmpF-M, it was necessary to investigate whetherthese two populations showed size differences. First, the dia-meters of GUVs (both OmpF-equipped GUVs and GUVswithout OmpF) were measured using ImageJ (Figure S11); aWelch two-sample t test showed no significant differencebetween the OmpF-equipped GUVs and GUVs without OmpF(p = 0.27, α = 0.05; Figure S12).To carry out a more detailed comparison, the fluorescence

intensity of each individual vesicle was taken 260 s after thereaction was initiated to ensure a stable signal. A significant dif-ference in fluorescence intensity between the averages ofOmpF-equipped GUVs and GUVs without OmpF wasobserved (p ≈ 0.024, α = 0.05; Figure S13A).Further details of the in situ enzymatic reaction were

revealed by dividing the GUVs into four groups on the basis ofsize, namely, from 6 to 9 μm (A), 9 to 11 μm (B), 11 to 13 μm(C), and 13 to 16 μm (D) (Figure S13B). The difference in

Figure 5. (top) GUVs with reconstituted OmpF-M and (bottom) GUVs without reconstituted OmpF-M. (A, C) Temporal variation in the relativefluorescence intensities of the resorufin-like product inside the cavities of single GUVs after addition of the reagents H2O2 and Amplex UltraRedafter background subtraction. Each curve represents one GUV. (B, D) Linear regressions of the change of the fluorescence intensity over time.Shown are the mean values with standard error of the mean (SEM) for n = 18−25.

Figure 6. Flow cytometry analysis of Amplex UltraRed enzymaticassay of OmpF-equipped (red) and OmpF-unequipped (blue)PMOXA7−PDMS49−PMOXA7 GUVs.

ACS Synthetic Biology Research Article

DOI: 10.1021/acssynbio.8b00104ACS Synth. Biol. 2018, 7, 2116−2125

2122

intensity of the two GUV populations was less pronounced inthe range of 6 to 9 μm, but the median intensities clearly dif-fered between OmpF-equipped GUVs and non-OmpF-equippedGUVs within the size ranges of 9 to 11, 11 to 13, and 13 to16 μm and was most pronounced with the higher sizes. In addi-tion to CLSM analysis, increasing enzymatic activity with respectto increasing GUV size was also confirmed by flow cytometry(Figure S14). This is in good agreement with the expectationthat larger GUVs should contain more enzyme moleculesinside their cavities, leading to higher substrate conversion andthus higher overall intensity. Even though a slight decrease wasobserved in going from group C to group D, it was not statis-tically significant (p ≈ 0.344).The apparent rate constant, kapp, was also calculated for each

individual GUV. Again, there were significant differencesbetween GUVs with and without inserted OmpF (Welch two-sample t test, p ≈ 0.025, α = 0.05; Figure 7A). Classification ofGUVs into the same size clusters as above shows behaviorsimilar to that described in the previous paragraph (Figure 7B).Overall, the difference in median kapp is marginal for group A,but there is an obvious difference in group B, and this increasedin groups C and D. Since there was a constant enzyme concen-tration in the rehydration solution, the number of enzymespresent in a GUV’s cavity should depend only on its size andtherefore volume. Thus, with increasing GUV size, the volumeand the number of enzymes per cavity should increase, whichexplains the lower kapp that was observed in group A. Further-more, larger sizes have more membrane surface, which allows ahigher number of OmpF pores to be present. This then resultsin increased substrate diffusion rates, and a steady stream ofsubstrates and product can be achieved. It appears that thecloser the physical characteristics (e.g., size and volume) are tothose of a natural cell, the higher the performance of the system.This is the first time that a membrane protein (OmpF) has

been reconstituted into synthetic polymer membranes ofGUVs with preserved functionality and is instrumental ininitiating an in situ enzymatic reaction by allowing diffusion ofsubstrates and products. The model enzyme reaction inside theGUV cavities was monitored in real time and gave insight intoboth the substrate apparent rate constant by the encapsulatedenzyme and the reconstitution of the functional membraneprotein (OmpF) in synthetic GUV membranes. The large cavitiesof GUVs allowed free diffusion of the encapsulated enzyme, asproven by FCS. Furthermore, there appears to be a size-dependent trend in enzyme apparent rate constant with respect

to GUV diameter. Also, whereas direct observations of internalreactions in established nanocompartment platforms, such asliposomes and polymersomes, are limited by the resolution bar-rier, GUVs provide a unique opportunity for observing thebehavior of biomolecules within defined confined spaces.The combination of biomolecules and synthetic materials foundin membrane-protein-equipped polymer GUVs not only offersthe advantages of increased stability and lower membrane perme-ability than lipid GUVs but, together with an encapsulatedenzyme, also makes the microscale compartment platform relevantfor various applications, for example, as a research tool to studybiomolecule behavior within these cell-like compartments or asbiosensors.39

Our simultaneous encapsulation and insertion of biomole-cules in GUVs allows a thorough understanding of biologicalprocesses in real time and represents a useful platform to studyreactions inside confined spaces or interactions of biomole-cules with membranes by a straightforward change of enzymetype. As GUVs have sizes similar to those of cells, they allowthe relevant study of bioprocesses in vitro in a simple manner.Thinking even further ahead, molecular factories and advancedcell mimics could be developed by encapsulating in cell-sizedGUVs various polymersomes loaded with different types ofenzymes inside their cavities.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acssynbio.8b00104.

Materials and Methods and Figures S1−S14 (PDF)Movie showing a GUV containing OmpF (AVI)Movie showing a control GUV without OmpF (AVI)

■ AUTHOR INFORMATIONCorresponding Authors*E-mail: Cornelia.Palivan@unibas.ch.*E-mail: wolfgang.meier@unibas.ch.ORCIDMartina Garni: 0000-0001-7576-7503Cornelia G. Palivan: 0000-0001-7777-5355Wolfgang Meier: 0000-0002-7551-8272Author ContributionsM.G. contributed to OmpF modification and characterization,GUV production and characterization, enzymatic assays,

Figure 7. (A) Apparent rate constants, kapp, of two populations of GUVs with and without reconstituted OmpF-M. Each GUV was fittedindependently. (B) Variation of kapp with GUV diameter. Black lines indicate medians, white squares indicate average values, and white circlesindicate data points.

ACS Synthetic Biology Research Article

DOI: 10.1021/acssynbio.8b00104ACS Synth. Biol. 2018, 7, 2116−2125

2123

CLSM, flow cytometry, FCS, and writing of the manuscript;T.E. contributed to OmpF and HRP characterization andmodification, FCS, and flow cytometry; R.G. contributed toenzymatic data evaluation; W.M. and C.G.P contributed theenzymatic GUV concept and to writing of the manuscript.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

We gratefully acknowledge the financial support provided bythe Swiss Nanoscience Institute, the Swiss National ScienceFoundation, and the National Centre of Competence inResearch - Molecular Systems Engineering. The authors thankS. Lorcher (University of Basel) for synthesizing the polymer.M.G. thanks G. Persy (University of Basel) for TEMmeasurements. Dr. B. A. Goodman is thanked for editing themanuscript.

■ REFERENCES(1) Rafelski, S. M., and Marshall, W. F. (2008) Building the Cell:Design Principles of Cellular Architecture. Nat. Rev. Mol. Cell Biol. 9(8), 593.(2) Salehi-Reyhani, A., Ces, O., and Elani, Y. (2017) Artificial CellMimics as Simplified Models for the Study of Cell Biology. Exp. Biol.Med. 242 (13), 1309−1317.(3) Sueyoshi, D., Anraku, Y., Komatsu, T., Urano, Y., and Kataoka,K. (2017) Enzyme-Loaded Polyion Complex Vesicles as in VivoNanoreactors Working Sustainably under the Blood Circulation:Characterization and Functional Evaluation. Biomacromolecules 18(4), 1189−1196.(4) Dinu, M. V., Spulber, M., Renggli, K., Wu, D., Monnier, C. A.,Petri-Fink, A., and Bruns, N. (2015) Filling Polymersomes withPolymers by Peroxidase-Catalyzed Atom Transfer Radical Polymer-ization. Macromol. Rapid Commun. 36 (6), 507−514.(5) Zhang, X., Lomora, M., Einfalt, T., Meier, W., Klein, N.,Schneider, D., and Palivan, C. G. (2016) Active Surfaces Engineeredby Immobilizing Protein-Polymer Nanoreactors for SelectivelyDetecting Sugar Alcohols. Biomaterials 89, 79−88.(6) Kowal, J., Zhang, X., Dinu, I. A., Palivan, C. G., and Meier, W.(2014) Planar Biomimetic Membranes Based on Amphiphilic BlockCopolymers. ACS Macro Lett. 3 (1), 59−63.(7) Letchford, K., and Burt, H. (2007) A Review of the Formationand Classification of Amphiphilic Block Copolymer NanoparticulateStructures: Micelles, Nanospheres, Nanocapsules and Polymersomes.Eur. J. Pharm. Biopharm. 65 (3), 259−269.(8) Lomora, M., Garni, M., Itel, F., Tanner, P., Spulber, M., andPalivan, C. G. (2015) Polymersomes with Engineered Ion SelectivePermeability as Stimuli-Responsive Nanocompartments with Pre-served Architecture. Biomaterials 53, 406−414.(9) Kuchler, A., Yoshimoto, M., Luginbuhl, S., Mavelli, F., andWalde, P. (2016) Enzymatic Reactions in Confined Environments.Nat. Nanotechnol. 11 (5), 409.(10) Garni, M., Einfalt, T., Lomora, M., Car, A., Meier, W., andPalivan, C. G. (2016) Artificial Organelles: Reactions inside Protein−Polymer Supramolecular Assemblies. Chimia 70 (6), 424−427.(11) Yoshitani, T., and Yamazaki, M. (2008) Water Permeability ofLipid Membranes of GUVs and Its Dependence on ActinCytoskeletons inside the GUVs. 2008 International Symposium onMicro-NanoMechatronics and Human Science, 130−134.(12) Li, S., Hu, P. C., and Malmstadt, N. (2011) Imaging MolecularTransport across Lipid Bilayers. Biophys. J. 101 (3), 700−708.(13) Jørgensen, I. L., Kemmer, G. C., and Pomorski, T. G. (2017)Membrane Protein Reconstitution into Giant Unilamellar Vesicles: AReview on Current Techniques. Eur. Biophys. J. 46 (2), 103−119.(14) Trantidou, T., Friddin, M., Elani, Y., Brooks, N. J., Law, R. V.,Seddon, J. M., and Ces, O. (2017) Engineering Compartmentalized

Biomimetic Micro- and Nanocontainers. ACS Nano 11 (7), 6549−6565.(15) Discher, D. E., and Eisenberg, A. (2002) Polymer Vesicles.Science 297 (5583), 967−973.(16) Discher, B. M., Won, Y.-Y., Ege, D. S., Lee, J. C.-M., Bates, F. S.,Discher, D. E., and Hammer, D. A. (1999) Polymersomes: ToughVesicles Made from Diblock Copolymers. Science 284 (5417), 1143−1146.(17) Pawar, P. V., Gohil, S. V., Jain, J. P., and Kumar, N. (2013)Functionalized Polymersomes for Biomedical Applications. Polym.Chem. 4 (11), 3160−3176.(18) Onaca, O., Enea, R., Hughes, D. W., and Meier, W. (2009)Stimuli-Responsive Polymersomes as Nanocarriers for Drug and GeneDelivery. Macromol. Biosci. 9 (2), 129−139.(19) Dobrunz, D., Toma, A. C., Tanner, P., Pfohl, T., and Palivan, C.G. (2012) Polymer Nanoreactors with Dual Functionality: Simulta-neous Detoxification of Peroxynitrite and Oxygen Transport.Langmuir 28 (45), 15889−15899.(20) Graff, A., Winterhalter, M., and Meier, W. (2001) Nanoreactorsfrom Polymer-Stabilized Liposomes. Langmuir 17 (3), 919−923.(21) Nardin, C., Thoeni, S., Widmer, J., Winterhalter, M., and Meier,W. (2000) Nanoreactors Based on (Polymerized) ABA-TriblockCopolymer Vesicles. Chem. Commun. 0 (15), 1433−1434.(22) Kuiper, S. M., Nallani, M., Vriezema, D. M., Cornelissen, J. J. L.M., van Hest, J. C. M., Nolte, R. J. M., and Rowan, A. E. (2008)Enzymes Containing Porous Polymersomes as Nano Reaction Vesselsfor Cascade Reactions. Org. Biomol. Chem. 6 (23), 4315−4318.(23) Tanner, P., Balasubramanian, V., and Palivan, C. G. (2013)Aiding Nature’s Organelles: Artificial Peroxisomes Play Their Role.Nano Lett. 13 (6), 2875−2883.(24) Baumann, P., Spulber, M., Fischer, O., Car, A., and Meier, W.(2017) Investigation of Horseradish Peroxidase Kinetics in an“Organelle-like” Environment. Small 13 (17), 1603943.(25) Xiao, Q., Yadavalli, S. S., Zhang, S., Sherman, S. E., Fiorin, E.,da Silva, L., Wilson, D. A., Hammer, D. A., Andre, S., Gabius, H.-J.,et al. (2016) Bioactive Cell-like Hybrids Coassembled from(Glyco)Dendrimersomes with Bacterial Membranes. Proc. Natl.Acad. Sci. U. S. A. 113 (9), E1134−E1141.(26) Peyret, A., Ibarboure, E., Tron, A., Beaute, L., Rust, R., Sandre,O., McClenaghan, N. D., and Lecommandoux, S. (2017) Polymer-some Popping by Light-Induced Osmotic Shock under Temporal,Spatial, and Spectral Control. Angew. Chem., Int. Ed. 56 (6), 1566−1570.(27) Peters, R. J. R. W., Marguet, M., Marais, S., Fraaije, M. W., vanHest, J. C. M., and Lecommandoux, S. (2014) Cascade Reactions inMulticompartmentalized Polymersomes. Angew. Chem., Int. Ed. 53(1), 146−150.(28) Kim, A. J., Kaucher, M. S., Davis, K. P., Peterca, M., Imam, M.R., Christian, N. A., Levine, D. H., Bates, F. S., Percec, V., andHammer, D. A. (2009) Proton Transport from Dendritic Helical-Pore-Incorporated Polymersomes. Adv. Funct. Mater. 19 (18), 2930−2936.(29) Sauer, M., Haefele, T., Graff, A., Nardin, C., and Meier, W.(2001) Ion-Carrier Controlled Precipitation of Calcium Phosphate inGiant ABA Triblock Copolymer Vesicles. Chem. Commun. 0 (23),2452−2453.(30) Itel, F., Chami, M., Najer, A., Lorcher, S., Wu, D., Dinu, I. A.,and Meier, W. (2014) Molecular Organization and Dynamics inPolymersome Membranes: A Lateral Diffusion Study. Macromolecules47 (21), 7588−7596.(31) Itel, F., Najer, A., Palivan, C. G., and Meier, W. (2015)Dynamics of Membrane Proteins within Synthetic PolymerMembranes with Large Hydrophobic Mismatch. Nano Lett. 15 (6),3871−3878.(32) Einfalt, T., Goers, R., Dinu, I. A., Najer, A., Spulber, M., Onaca-Fischer, O., and Palivan, C. G. (2015) Stimuli-Triggered Activity ofNanoreactors by Biomimetic Engineering Polymer Membranes. NanoLett. 15 (11), 7596−7603.

ACS Synthetic Biology Research Article

DOI: 10.1021/acssynbio.8b00104ACS Synth. Biol. 2018, 7, 2116−2125

2124

(33) Edlinger, C., Einfalt, T., Spulber, M., Car, A., Meier, W., andPalivan, C. G. (2017) Biomimetic Strategy To Reversibly TriggerFunctionality of Catalytic Nanocompartments by the Insertion of PH-Responsive Biovalves. Nano Lett. 17 (9), 5790−5798.(34) Machan, R., and Wohland, T. (2014) Recent Applications ofFluorescence Correlation Spectroscopy in Live Systems. FEBS Lett.588 (19), 3571−3584.(35) Rigler, P., and Meier, W. (2006) Encapsulation of FluorescentMolecules by Functionalized Polymeric Nanocontainers: Investiga-tion by Confocal Fluorescence Imaging and Fluorescence CorrelationSpectroscopy. J. Am. Chem. Soc. 128 (1), 367−373.(36) Martino, C., and deMello, A. J. (2016) Droplet-BasedMicrofluidics for Artificial Cell Generation: A Brief Review. InterfaceFocus 6 (4), 20160011.(37) Nishimura, K., Matsuura, T., Nishimura, K., Sunami, T., Suzuki,H., and Yomo, T. (2012) Cell-Free Protein Synthesis inside GiantUnilamellar Vesicles Analyzed by Flow Cytometry. Langmuir 28 (22),8426−8432.(38) Qiao, Y., Li, M., Booth, R., and Mann, S. (2017) PredatoryBehaviour in Synthetic Protocell Communities. Nat. Chem. 9 (2), 110.(39) Kita-Tokarczyk, K., Grumelard, J., Haefele, T., and Meier, W.(2005) Block Copolymer VesiclesUsing Concepts from PolymerChemistry to Mimic Biomembranes. Polymer 46 (11), 3540−3563.

ACS Synthetic Biology Research Article

DOI: 10.1021/acssynbio.8b00104ACS Synth. Biol. 2018, 7, 2116−2125

2125