Polymeric Giant Unilamellar Vesicles with Integrated DNA‐Origami Nanopores: An Efficient Platform for Tuning Bioreaction Dynamics Through Controlled Molecular Diffusion

Giant unilamellar vesicles (GUVs) are microcompartments serving to confine reactions, allow signaling pathways, or design synthetic cells. Polymer GUVs are composed of copolymer membranes mimicking cell membranes, and present advantages over lipid‐based GUVs, such as higher mechanical stability and chemical versatility. Such microcompartments are essential for understanding reactions/signaling occurring in cells, which are difficult to study by in vivo approaches due to the cell's complexity. However, the lack of control over their production, stability, and membrane diffusion properties is still limiting their use for bio‐related applications. Here, polymer GUVs produced by microfluidics and permeabilized with DNA‐origami nanopores (DoNs) that present a high level of control over these essential properties are introduced. After systematic optimization of conditions, DoN‐GUVs reveal a narrow size distribution, allow for high encapsulation efficiencies, and are stable for weeks, protecting encapsulated biomolecules. The kinetics of diffusion of molecules through the GUV's membrane is tuned by insertion of DoNs with a controlled 3D‐ structure. DNA polymerase I, encapsulated as model for bioreactions, successfully produced DNA duplex strands with spatiotemporal control. DoN‐GUVs loaded with active molecules open new avenues in bioreactions, from the detection of biomolecules, over the tuning of molecular transport rates, to the investigation of cellular processes/signaling.


Introduction
Compartmentalization is a fundamental concept of cells, enabling spatial control and temporal regulation of bioreactions.By DOI: 10.1002/adfm.2023047823][4][5][6][7] Studying these processes in vivo is difficult to achieve due to the high heterogeneity of cells.[17][18] Among these assemblies, giant unilamellar vesicles (GUVs) are the artificial compartments of choice when cellular processes need to be studied.[32] However, their mechanical stability is generally affected by their surrounding environment, and their chemical functionalization is restricted to those offered by the lipid structures, limiting their use for bio-related application. [22]To overcome these limitations, amphiphilic copolymers replaced lipids to generate GUVs by self-assembly using rehydration of copolymer films, [22] electroformation [33] and very recently, microfluidics. [11]The higher mechanical stability of polymer-based membranes compared to lipid-based ones combined with the possibility to engineer their chemical structure to have adjustable membrane properties, make polymer compartments more appropriate for the investigation of bioprocesses in a controlled cell-like manner.
Despite above-mentioned advantages, polymer-GUVs generally have low permeability, [22] a property that is limiting the possibility to initiate reactions inside by preventing the diffusion of small reactant molecules or reaction products through the membrane.To tackle this challenge, ion channels, [34] pore-forming membrane proteins, [11] or peptides [35] were inserted into the polymer GUVs membrane to serve as channels for the diffusion of small molecules.Permeabilized GUVs supported reactions, such as the polymerization of actin, leading to the formation of a cytoskeleton, [36] or the glucose oxidation catalyzed by a glucose oxidase enzyme. [11]Generally, the insertion of biopores is limited both by the increased hydrophobic mismatch between the size of biopores and the thickness of the polymer membranes as well as by the lower flexibility of the later ones compared to lipidic membranes. [35,37]Therefore, the insertion of biopores and membrane proteins into synthetic membranes is complex and additional steps are often required that may modify the physicochemical properties of the pores and/or alter their ability to be inserted into membranes.Besides these issues, conventional methods for GUV formation, including film rehydration and electroformation, have drawbacks including, large GUV size distribution, low encapsulation efficiency of molecules and poor control over the number of inserted biopores. [20,22]ere, we introduce a polymer GUV-based platform allowing for a high level of spatiotemporal control over bioreactions by combining a microfluidic approach for compartment formation with the co-encapsulation of reaction compounds and insertion of DNA-origami nanopores to support molecular diffusion through their membranes.We have chosen and adapted a microfluidic-based droplet generation method to produce polymer GUVs as it enables a controlled formation and loading of GUVs with water-soluble biomolecules at an encapsulation efficiency of close to 100%. [11]We selected poly(dimethylsiloxane-)bloc-poly(-2-methyl-2-oxazoline) (PDMS-b-PMOXA) diblock amphiphilic copolymers as the building blocks of GUVs, to benefit from their ability to form stable and flexible membranes. [11]o fine tune the GUV's membrane permeability allowing for a precise time-initiation of reactions inside their cavity, we selected DNA-origami nanopores (DoNs) presenting a T-shape structure.Recently, the development of DNA-origami technology, which enable single-stranded DNA (ssDNA) molecules to be folded into a desired shape with a high level of control, offered a way to obtain well-defined, tunable nanopores [38,39] that present advantages compared to other biopores.DoNs can be engineered with a control over their shape, size, and surface chemistry.For example, T-shape, [40] O-ring, [41] and even cylindrical [42] DoN, were produced using this technology.[45][46] One recent study reported on the insertion of DoNs into the membrane of polymer GUVs, however only when DoNs were attached to cholesterol, without governing control over the number of pores inserted per GUV, and their permeability properties in terms of kinetics. [47]These limitations were due both to the different morphology of the synthetic membrane based on triblock copolymers and the method used for the GUVs production which did neither reach a sufficient control over the GUV's size nor achieve a high encapsulation efficiency.These drawbacks should be resolved because the control of the polymer GUVs properties represents a key factor in studying bioreactions and mimicking in situ biological processes.We systematically optimized the conditions for the generation of GUVs and co-encapsulation of molecules and insertion of DoNs, including the composition of microfluidics phases, flow rates, and DoNs concentration.After fine-tuning and studying the resulting permeability properties upon insertion of DoNs in the GUV's membrane by different modes (e.g., internal/external insertion), we investigated the use of this platform for controlled bioreactions.Furthermore, we evaluated the mechanical stability of GUVs overtime as a prerequisite for application.Additionally, we selected the DNA strands duplex formation as a model for bioreactions due to its importance in the processes of DNA replication occurring in living cells. [48,49]We were interested to test how our DoNs-GUV platform allows co-encapsulation of mixtures of reactants (DNA single strand (ss-DNA) template and the Klenow fragment of the polymerase I) and provides a spatiotemporally control of double-strand DNA production.Besides, we evaluated the ability of DoNs-GUVs for detection of specific ss-DNA sequences from a mixture of DNA strands.By a straightforward change of the co-encapsulated molecules, this platform opens numerous applications ranging from bioprocess control to investigation of cellular processes/signaling or DNA sequence detection.

One-Step Formation of Polymer GUVs by Microfluidics and Insertion of DoNs into Their Membrane
To obtain a thin GUV membrane that facilitates DoNs insertion, we chose the amphiphilic diblock copolymer PDMS 25 -b-PMOXA 10 (Figure S1, Supporting Information) as it has been reported to form GUVs with a stable membrane. [50]Diblock copolymers form bilayer structures with only slight interdigitation and entanglement of the polymer chains. [51]In comparison with triblock based membranes, which adopt a mixed conformation with fractions of stretched/elongated chains, I-shape and loops, Ushape, [52] reducing lateral mobility, membrane based on diblock copolymers will favor DoN insertion. [53]DoNs were produced by assembling a DNA scaffold in a square lattice pattern, comprising a double-layered plate (46 nm × 51 nm origami rectangular design) with a 8.4 nm × 8.4 nm square aperture in its center and a 15 nm long hollow stem extending perpendicularly from the center, designed to act as a transmembrane channel with a square inner cross-section of 4.2 nm × 4.2 nm (Figure 1A-C). [40]he double-layered DNA plate was functionalized using single strand DNA (ssDNA) moieties to provide additional functionality and/or facilitate insertion in membranes.While we selected a design of DoNs previously used for the insertion into lipid membranes, [40] all the conditions necessary to generate polymer GUVs and insert DoNs were different and required systematic optimization to cope with the huge differences in the synthetic GUVs formation and membrane properties compared with lipid GUV.Polymer GUVs were formed using double emulsion templates generated in a sixway junction microfluidic device (Figure 1D; Figure S2, Supporting Information). [11]An inner aqueous (IA) phase was pinched and enveloped by a polymer organic (PO) and an outer aqueous (OA) phase, thereby forming double-emulsion droplets.Copolymer molecules at the two liquid-liquid interfaces of the droplets self-assemble into two monolayers, which are brought closer together during the polymer solvent dewetting, which finally results in a polymer bilayer formation.The fluid phase compositions and flow rates were different from those used for lipid-GUVs production by microfluidics [54] and specifically optimized for the production of stable polymer-GUVs.In this respect, we replaced the organic phases such as octanol [27] or oleic acid [25,55] by a mixture of 3:2 vol/vol hexane:chloroform, providing a better copolymers solubilization.The composition of the fluid phases has been adapted to simultaneously have a volatile organic solvent, not completely miscible in water and providing good amphiphile dispersion.The phases were respectively composed of 5 w/v% of PEG and 200 mM of sucrose dissolved in PBS (≈600 mOsm kg −1 ); 4 mg mL −1 of PDMS 25 -b-PMOXA 10 in a 3:2 v/v hexane:chloroform mixture; and 5% w/v PVA and 100 mM of NaCl in PBS solution (≈540 mOsm kg −1 ) (more details are presented in the materials and methods section).The IA and OA phase compositions were adapted to obtain similar osmolarity values in order to avoid GUVs membrane destabilization upon the osmotic pressure effect.Typically, the flow rates of the IA, PO, and OA phases were respectively set at 1, 3, and 50 μL min −1 as they provided optimal production conditions for the doubleemulsion droplets in the channels.As the formation of these droplets depends on the pinching of the phases, it therefore depends on their relative viscosities and has been adapted for each combination of fluid-phase composition.
First, we formed polymer GUVs without the insertion of DoNs to determine the conditions for obtaining stable and uniform compartments.The resulting polymer GUVs were observed by confocal laser scanning microscopy (CLSM) to determine both their architecture and size.In that respect, we encapsulated calcein, a hydrophilic dye inside the cavity of GUVs during their formation (by addition to the inner aqueous phase IA at 10 μM), and simultaneously labeled the membrane with a hydrophobic dye, BODYPY630/650 (by addition to the outer aqueous phase OA at 1.25 μM).Single-compartments GUVs with a bilayer membrane were formed and an encapsulation of calcein with an efficiency of close to 100% was obtained, in agreement with previous encapsulation efficiencies of small molecules in polymer GUVs generated by microfluidics approach. [11]We determined a mean GUV diameter of 37±1.5 μm (standard deviation calculated from N = 105 GUVs) (Figure S3, Note S1, Supporting Information) with a very narrow size distribution (Figure 1E,F).These polymer GUVs were stored at 4 °C and observed at 25 °C after 3 weeks.They preserved their integrity as no dye molecules were found to escape (constant fluorescence intensity) from their cavity and their number was constant (Figure S4, Supporting Information).When compared with lipid GUVs, which depending on their size and lipid composition can generally maintain their integrity only over hours to a few days, [22] the significant stability improvements of synthetic GUVs supports their further use as advanced microcompartments for medical applications.
Second, we used two different approaches to incorporate DoNs into GUVs membranes: i) "external insertion", i.e., addition of DoNs to the GUVs membranes from the external aqueous phase, and ii) "internal insertion" where DoNs are added to the IA phase during the double-emulsion production.Our aim was to distinguish whether there are differences in DoNs insertion between these approaches to optimize the insertion efficiency whilst not affecting the GUV's integrity.To evaluate the integration of DoNs, they were hybridized with the hydrophilic ssDNA-Alexa 488 and cholesterol-BODIPY 630/650 ssDNA prior to the insertion (Figure 1C).CLSM was used to visualize the GUVs after the insertion process (Figure 2).
In the first approach, a solution of DoNs dispersed in the OA phase composition was added 20 min after the double emulsion production and collection, to ensure complete partition of the organic phase, favoring the formation of GUVs. [11]A homogeneous distribution of fluorescently labelled DoNs in the OA phase sur-rounding the GUVs was observed immediately after their addition (Figure 2A).After 60 s, the fluorescent signals of both dyes were localized and formed a fluorescent ring associated with the GUVs membrane.This indicates both a complete covering of the GUVs membrane for a concentration of DoNs of 9 nM and that this DoNs concentration does neither affect the GUVs formation nor its integrity (Figure 2B).This extremely rapid insertion of the DoNs demonstrates their high affinity for the polymer membrane of GUVs.Based on the intrinsic architecture of the DoNs, their expected orientation into the membrane is with the stem and cholesterol domains inserted in the hydrophobic region of the membrane while the hydrophilic top side of the plate is facing toward the exterior of the GUVs membrane (inset of Figure 2B).
In the second approach ("internal insertion"), the insertion of the DoNs in the polymer membrane was directly observed by CLSM after the w/o/w double emulsion droplet production.Immediately after droplet collection, DoNs were found to be homogeneously distributed in the GUVs lumen where a homogeneous fluorescence intensity associated with both dyes was observed (Figure 2C).60 s after droplet collection, the fluorescent signal associated with both dyes was present only as rings associated with the GUVs membrane (Figure 2D), indicating again the high affinity of the DoNs with the polymer organic phase.Indeed, at this point, the double-emulsion dewetting is not complete and an organic layer is still present in the microstructures formed.Nevertheless, the DoNs are already present at the droplet's interface.
Compared to the external insertion approach, the internal insertion of the DoNs provides a way to lower the amount of nanopore solution used for GUV's permeabilization (from few hundred microliters up to tens).Indeed, in the internal insertion approach, DoNs are directly in contact with the GUVs membrane at the targeted concentration by encapsulating with microfluidics, and not by adding in a diluted bulk.However, no difference in the rapidity of the DoNs insertion was observed between the two approaches.
To establish the role of the cholesterol attached on the DoNs in the nanopore insertion, DoNs without cholesterol were compared to cholesterol-functionalized DoNs in terms of the rapidity of the insertion process.No significant difference in DoNs membrane insertion was observed: both nanopores with and without cholesterol were localized at the membrane interface within a minute (Figure S5, Supporting Information).We consider that the nature of the diblock copolymer membranes and their unique physical characteristics supports DoNs insertion without the help of cholesterol. [37]Indeed, it has been reported that self-assembled PDMS-b-PMOXA diblock copolymer membranes have packed bilayer structure with only slight interdigitation and entanglement of the polymer chains, [53] which makes them flexible and facilitates the insertion of biomolecules into the membrane.Nevertheless, the functionalization with cholesterol of DoNs can be used to support the insertion of the DoNs in more rigid polymer membranes such as those formed with triblock copolymers. [47]For the sake of simplicity, only pristine DoNs (without cholesterol and Alexa 488) were further used for GUVs permeabilization.

Evidence of DoNs Insertion by Molecule Diffusion Assays
While the localization of the DoNs at the GUVs interface was observed by CLSM, the observed colocalization could also be due to the adsorption of the pores onto the membrane surface without perforation as the spatial resolution of the CLSM does not allow details at the membrane's inner or outer surface to be distinguished.Therefore, we were interested to understand whether the insertion of the DoNs into the GUVs membrane was functional and induced its permeabilization.In this respect, two diffusion assays were performed using the green-fluorescent dye calcein as "reporting" molecule.First, to investigate the external insertion of DoNs, we encapsulated inside GUVs calcein and added DoNs from the outside (in the OA phase) of the GUVs after dewetting.The GUVs fluorescence intensity associated with calcein was monitored over time.A decrease of the fluorescence intensity in the lumen of the GUVs was observed when adding DoNs in the OA phase (Figure 3A), thus indicating a calcein outflux from the GUVs.Contrary, no release of calcein was observed when DoNs were absent in the GUVs (Figure 4A).Therefore, DoNs were found to be inserted into the GUVs membrane from the external phase, and supported the diffusion of calcein molecules through the GUVs membrane.
Second, we evaluated the internal insertion of DoNs into the GUVs membrane by a calcein influx assay.GUVs encapsulating DoNs into their lumen were formed prior to adding calcein in the OA phase.An increase of calcein fluorescence intensity in the lumen of the GUVs was observed over time (Figure 3B), indicating the influx of calcein molecules across the GUVs membrane.Without DoNs in the IA phase, no calcein diffusion through the GUVs membrane (Figure S6, Supporting Information) was observed, indicating that the GUVs membrane is impermeable without DoNs.

Control of The GUVs Permeability by DoNs
To monitor the calcein flux, the fluorescence intensity inside GUVs was measured by analysis of the CLSM micrographs (Figure S3 and Note S1, Supporting Information).In the absence of DoNs, no change in the fluorescence intensity of the GUVs in time was observed (Figure 4A).After the addition of DoNs in the GUV dispersion, a significant decrease in GUVs fluorescence intensity in time was observed (Figure 4B).The DoNs calcein diffusion was fitted using an exponential decay model (Note S2, Supporting Information).This decrease of fluorescence intensity is due to the diffusion of calcein and was observed to be DoNs concentration dependent.For example, 240 s after the addition of DoNs at 9 nM in the OA phase, the fluorescence intensity of the GUVs decreased by ≈70%, while it decreased by ≈50% when DoNs were added at 5 nM, and ≈25% at 1 nM (Figure 4B).The different diffusion kinetics could be due to an increase in the number of inserted pores with an increase in DoNs concentration.As the concentration increases, more pores are available to allow for the diffusion of calcein, leading to a higher rate of leakage.
Note that since no difference in exchange kinetics was observed for both calcein influx and leakage (example Figure S7, Supporting Information), therefore, we will further focus only on the leakage scenario.In addition, we observed that the kinetics of diffusion was not modified by the functionalization of the DoN with cholesterol (Figure S8, Supporting Information).
The number of DoNs per GUV (Np) that supported the calcein diffusion through the membrane of GUVs was estimated by analyzing the fluorescence intensity decay (Note S2, Supporting Information).Fick's law was used to model molecular diffusion across the DoNs.This type of exponential decay analysis has already been reported in the framework of Fick's law, [25,56] which is governing how molecules passively diffuse in a gradient of concentration across a membrane.Basically, molecules diffuse from high-concentration areas to low-concentration areas to equilibrate the concentration on either side of the membrane, which is thermodynamically favorable.N p increases with DoNs concentration (Figure 5A).For DNA concentrations varying from 1 to 20 nM, N p values ranging from ≈10 2 to ≈10 5 were found, in agreement with values reported for DNA nanopores inserted into lipidic GUVs membranes. [56]At DoNs concentrations above   20 nM, bursting of GUVs was observed.This rupture is probably due to an increase of pore density leading to GUVs membrane destabilization.To understand this, the number of pores acting as channels for diffusion was compared to the theoretical maximum number of DoNs per GUV, N p,max , and the effective pore fraction was calculated (Note S2, Supporting Information).For DoNs concentrations increasing from 1 to 20 nM, the effective pores fraction first increases appearing to reach a plateau at concentrations above 3 nM, at a value of ≈20-40% active pores (Figure 5A).This indicates that at low nanopore concentrations, only a small amount of the DoNs is inserted into the membrane and is participating in the calcein diffusion.The increase of the pore concentration first enhances the capability of the pores to insert into the membrane and to serve for diffusion.We hypothesize that it may be due to the increased membrane fluidity induced by additional DoN insertions.We believe that at high DoN concentrations, the pores tend to form clusters in the membrane, which could inhibit the participation of the pores in the calcein diffusion, as it has been suggested for pores clusters inserted into lipidic membranes. [57]This effect of clusters formation would become predominant with the increase of the concentration compared to the increase of the membrane fluidity, and explains the plateau.At higher concentrations, DoNs may tend to create defects in the membrane that eventually lead to the bursting of the GUVs (observed for concentrations higher than 20 nM).
To determine whether the DoNs maintain their inner pore sizes of ≈4.2 nm × 4.2 nm once inserted in a polymer membrane, and to investigate the molar mass cutoff for diffusion, the leakage of molecules of variable sizes and molecular masses out of GUVs was studied.A set of fluorescein isothiocyanate (FITC)-labeled dextrans in the molecular weight range of 7-150 kDa was used to study diffusion from the GUVs. [58]By adding DoNs from the outside of the GUVs, the polymeric membranes were permeabilized and the FITC fluorescence associated with the initially encapsulated dextran molecules was monitored over time.
Using dynamic light scattering (DLS), the hydrodynamic diameters (D H ) of dextran macromolecules with molecular weight ranging from 7 to 150 kDa were determined to be within a range of 1 to 10 nm (Figure S9, Supporting Information).Then, we used CLSM to visualize the release of FITC-dextran molecules encapsulated inside GUVs prior to their formation.FITC-dextran molecules diffuse through DoNs up to a molar mass of 40 kDa.Furthermore, a decrease of the fluorescence intensity was observed in the lumen of the GUVs for FITC-dextran molar mass ranging from 7 to 40 kDa with a complete reduction of fluorescence intensity after 30 min.Contrary, the fluorescence intensity associated with FITC-dextran molecules remained constant over time for molar masses above 50 kDa (Figure 5B).This molar mass cutoff corresponds to a D H of ≈4-5 nm (Figure S9, Supporting Information), indicating that the DoN preserves its channel diameter of 4.2 nm upon its insertion into the GUV's membrane.

GUVs Permeabilized With DoNs as a Platform For Bioreactions
With the basic permeability functionalities demonstrated, we now aim at evidencing the versatility of the GUV-DoNs platform for performing bioreactions by model reaction, namely the DNA duplex formation reaction.This reaction is based on the 68 kDa large (Klenow) fragment of DNA polymerase I, which amplifies DNA fragments through primer extension. [59]DNA duplex formation mediated by DNA polymerase I occurs if an added primer dimerizes with a complementary sequencing template and thereby induces DNA double strand (dsDNA) extension.DNA sequences selected to test the reaction inside DoN-GUVs are presented in Table S1 (Supporting Information).SYBR Green I (SG), a fluorescent dye which increases in brightness upon interaction with dsDNA, [60] was employed to visualize in situ the DNA duplex formation (Figure 6A).The specificity of the DNA polymerase I was determined by comparison between a primer sequence and a scrambled primer sequence, which consists of a modified sequence.We first co-encapsulated the DNA polymerase I and the template within GUVs by their addition in the IA phase used for the GUV production as previously described for calcein loading.Second, the GUVs were permeabilized by adding DoNs at 9 nM in their OA phase (external insertion) 30 min after their collection out of the chip (after dewetting).After one minute, the resulting permeabilized GUVs were then exposed to a mixture of ssDNA primer, SG, and nucleotides (dNTP mix) and their fluorescence intensity was immediately monitored overtime using CLSM.The enzymatic reaction is initiated by the following steps: 1) the mixture just added in the GUVs dispersion penetrates into the core of the GUVs, 2) the DNA polymerase I starts the DNA duplex formation, leading to the formation of ds-DNA 3) where SG can intercalate resulting in a fluorescence intensity increase (Figure 6A).
In case DoNs were not added, no fluorescence intensity was observed in the GUVs containing DNA polymerase I and the template (Figure 6C), which indicate non-diffusion of the other reaction reactants (ssDNA primer, dNTPmix, and SG) through the non-permeabilized GUVs polymer membrane.Contrary, when permeabilized GUVs were exposed to a mixture of ssDNA primer, SG and nucleotides in the OA phase, an increase of SG fluorescence intensity in the lumen of the GUVs was observed within 1200 s (Figure 6B,C).This increase was not observed in absence of ssDNA primer (Figure 6C).Therefore, the successful dsDNA formation inside DoN-GUVs was possible and detected only when the ssDNA primer diffused through the DoNs pores.No increase of the SG fluorescence intensity was observed when a scrambled primer was used, featuring a nucleotide sequence incompatible with the template sequence.When a 1:1 molar mixture of scrambled and unscrambled primers was used, the SG fluorescence intensity increased comparably to that obtained when only unscrambled ssDNA primer was used (Figure 6B).Therefore, this DoN-GUVs platform can be successfully used to host and visualize reactions such as the templated-driven detection of DNA sequence when mixtures of different ssDNA are present, which represent a key aspect for further applications.
Our results are in agreement with this type of model reaction be studied in nano-sized polymersomes, [61] but here the control of the number of DoNs allows the kinetics of reaction to be increased by a factor two, most likely due to the higher pore density enhancing the kinetics of diffusion of the reactants.Indeed, the monitoring of the in situ fluorescence intensity associated with SG upon ds-DNA production allowed to observe an increase of ≈80% and a plateau reached after ≈500 s (Figure 6C) while such a plateau was observed using fluorescence spectroscopy after 1200 s in nano-sized polymersomes.
The observation of differences in fluorescence intensity provides a straightforward manner to directly monitor bioreactions by simple in situ visualization within the DoN-GUV-based platform.Note that the kinetics curves of the enzymatic reactions studied here in the GUVs are comparable to those obtained in control bulk conditions for the enzyme freely dispersed in the IA buffer, showing that the encapsulation of the enzyme in the GUVs is not affecting its activity (Figure S10, Supporting Information).Based on the previous cutoff determination (≈40-50 kDa), the ssDNA template and dsDNA product (Mw≈22-25 kDa) should diffuse through the nanopores.However, no fluorescence decrease indicating leakage was observed, possibly due to a faster enzymatic reaction kinetics compared to the diffusion rate through the DoNs pores of the GUVs.ss/dsDNA strands may take more time to diffuse compared to the calcein small molecule (Mw = 622 Da) used for the previous kinetics study, due to their higher molar mass.Indeed, the leakage kinetics of FITC-dextran molecules with a molar mass (20 kDa) comparable to those of ssDNA and dsDNA was investigated (Figure S11, Supporting Information).The kinetics of diffusion for FITC-dextran 20 kDa was slower compared to the diffusion of the calcein, while faster compared to the one of FITC-dextran with higher molar mass value (50 kDa).This validates the effect of the molar mass on the molecular diffusion through the pores.In addition, it is important to note that the shape of the molecule that diffuse through the pores may also play a role on the diffusion kinetics.A macromolecule with a more compacted, globular shape such as dextran in water is expected to diffuse easier through the pores compared to an elongated DNA strand.Also, the DNA template is reaching concentration equilibrium between the IA and OA phases on both sides of the membrane, which is thermodynamically favorable.Thus, the DNA template is still present in the GUVs after 1200 s, when the components used for DNA duplex formation continue to penetrate, explaining the absence of a drop of fluorescence intensity in the lumen of the GUVs in 1200 s.

Conclusions
We present here advanced synthetic microcompartments equipped with DNA origami nanopores (DoNs) as an efficient platform for spatiotemporal control of in situ bioreactions.By using a microfluidics approach, we simultaneously generated PDMS 25 -b-PMOXA 10 -based giant unilamellar vesicles (GUVs), decorated them with tunable DoNs and co-encapsulated proteins and selected reaction components inside.The advantages of using a microfluidics approach include an almost homogeneous size of the microcompartments, a very high encapsulation efficiency (>99%) and a directed spatial location of the biomolecules depending on their intrinsic hydrophilic/hydrophobic character.By control of the conditions at molecular levels, DoNs insertion in the GUVs membrane occurs within one minute without affecting its architecture and integrity.The insertion can be achieved with similar rapidity either by encapsulation of the DoNs in the GUV's cavity or by external addition to the GUVs dispersion.Permeabilization of synthetic GUVs with DoNs provides a controlled exchange of molecules with diameters below 4-5 nm across the membrane and thus offers excellent conditions for in situ bioreactions.We tested the versatility of the DoN-GUVs by exploring an in situ protein-based reaction consisting in the template-directed polymerization of nucleotides resulting in the formation of duplex DNA, with an encapsulated single DNA template strand.Such polymerization based on the use of polymerase is an essential bioreaction occurring in living cells and involved in the process of DNA replication.
Importantly, the specificity of the reaction allowed duplicating the appropriate ssDNA in GUVs, even in presence of mixtures of primer sequences.This, opens efficient screening solutions for applications, such as the detection of mutations in DNA sequences.Besides, due to its biomimicry, this DoN-GUVs platform can serve for the study of the diffusion of molecules such as nutrients/substrates/drugs across a flexible cell-like membrane, thus providing a better understanding of their internalization and/or reactions in a simplified and controlled environment.The significant stability compared to lipid GUVs together with their versatility allowing a straightforward change of the co-encapsulated molecules make these DoN-GUVs ideal candidates for advanced applications, including screening of molecules in complex media, cascade bioreactions, signaling or biosensing.
Polymer Synthesis and Characterization: The PDMS 25 -b-PMOXA 10 diblock copolymer used in this study was synthesized as published previously. [62]The chemical structure (Figure S1, Supporting Information) and block ratio were confirmed by 1 H NMR spectroscopy.The dispersity (Ð = 1.49) and the number-average molar mass (M n = 1800 g mol −1 ) were determined by gel permeation chromatography, as recently presented elsewhere. [61]oNs Characterization and Hybridization: The DoNs structure and purity were characterized by the provider (Tilibit Nanosystem) using a combination of transmission electron microscopy (TEM) (Figure S12, Supporting Information) and gel electrophoresis (Figure S13, Supporting Information) experiments.The results indicated the absence of staple strands used for the DoNs synthesis, as well as the absence of DoNs aggregates after filtration and purification.For specific experiments (data for Figure 2), they were hybridized with cholesterol-modified single-strand DNA (ssDNA-cholesterol) stained with BODIPY630/650.ssDNA-cholesterol was added to a DNA origami nanopore dispersion at a final ssDNA-cholesterol/DNA origami ratio of 228:1 (57 hydrophobic moieties/pore; effective 4:1 ratio).Then, ultrafiltration at 10 000 g was performed at 25 °C using Amicon Ultra 0.5 mL Ultracel filters (MWCO = 3 kDa) to purify the samples of excess ssDNA-cholesterol.The resulting cholesterol-hybridized DoNs were then stained using 5 μM BODIPY 630/650 to the nanopores dispersion.To remove free BODIPY molecules, ultrafiltration at 10 000•g was performed at 25°C using Amicon Ultra 0.5 mL Ultracel filters (MWCO = 3 kDa).DoNs were also hybridized with single-strand DNA modified with Alexa 488 (ssDNA-A488).To do so, ssDNA-A488 was added to a DNA origami nanopore dispersion at a final ssDNA-A488/DNA origami ratio of 160:1 (40 hydrophilic moieties/pore; effective 4:1 ratio).Then, ultrafiltration at 10 000 g was performed at 25 °C using Amicon Ultra 0.5 mL Ultracel filters to purify the samples of excess ssDNA-cholesterol.
Composition of Fluid Phases: The inner aqueous (IA) phase composition differed depending on the experiment performed.Its basic composition comprised 5 w/v% of PEG and 200 mM of sucrose dissolved in PBS (≈600 mOsm kg −1 ).The polymer organic (PO) phase was composed of 4 mg mL −1 of PDMS 25 -b-PMOXA 10 in a 3:2 v/v hexane:chloroform mixture.The outer aqueous (OA) phase composition was also slightly different depending on the experiments performed.It was mainly composed of 5% w/v PVA and 100 mM of NaCl in PBS solution (≈540 mOsm kg −1 ).
GUVs Production By Microfluidics: Three different fluid phases (IA, PO, and OA) were injected into the microfluidic device as schematically represented in Figure 1; Figure S4 (Supporting Information) using syringe pumps (low pressure neMESYS, Cetoni).For all the experiments, the IA, PO, and OA phases flow rates were respectively set at 2, 1, and 50 μL min −1 to produce water-in-oil-in-water (w/o/w) emulsion droplets as reported by dos Santos et al. [11] The resulting droplets were collected in a vial for 10 min, resulting in ≈10 6 GUV mL −1 .Due to the presence of sucrose in the IA phase, the droplets sink to the bottom of the tube.Finally, the w/o/w droplets turned spontaneously into GUVs by dewetting over time (≈20 min) and separation of the PO phase.
Confocal Laser Scanning Microscopy: All images presented in this study were obtained with an inverted Zeiss 880 confocal microscope (Zeiss, Germany) equipped with 20x (Plan-Apochromat M27) and 40x (C-Apochromat) water immersion objectives.GUVs were visualized in eightwell chambers (Nunc Lab-Tek, Thermo Fisher Scientific, USA).To do so, a 488 nm argon laser and a 633 nm HeNe laser were used.For the visualization of the membrane of GUVs, 5 μL of 50 μM BODIPY 630/650 was added to 200 μL of GUVs dispersion and the sample was excited with an HeNe laser (633 nm).For calcein, SYBR green, and FITC-dextran visualization, the sample was excited with an argon laser (488 nm).The measurements of fluorescence intensity were performed using the image J software, using a macro presented in Note S1 (Supporting Information).The error bars in the corresponding plots were calculated from the standard deviation of the resulting measurement data.
GUVs Functionalization with DoNs Assessed By Leakage and Influx Assays: The insertion of DoNs within the GUVs membranes was performed in two different ways: 1) by addition of pores in the IA phase before GUVs formation, and 2): by addition of pores in the OA phase after GUVs formation.In these two cases, the pore formation was assessed using the green fluorescent dye calcein.To do so, calcein powder was dissolved at 10 mM in ultrapure Milli-Q water under stirring for 1 h.The resulting solution was filtered (Sartorius, 0.2 mm), aliquoted and stored in the dark at −20 °C.The aliquots were used for a week after defrosting to produce GUVs containing 10 μM of calcein in their IA phase or to add calcein in the OA phase of already formed GUVs at a final concentration of 10 μM.When calcein was added in the IA phase of the GUVs, DoNs were added to the GUVs dispersion at concentrations ranging from 2 to 20 nM to visualize a leakage of the calcein indicating the pore insertion, while when calcein was added in the OA, the pores were added at the same concentration in the IA to visualize an influx of calcein.
DoNs Molecular Weight Cut-off Determination by Leakage Assay: The molecular weight cut-off of the DoNs was determined using a leakage assay.Dextran-FITC of molar mass of 7, 10, 20, 40, 50, 70, and 150 kDA were encapsulated into the GUVs at an FITC/sucrose molar ratio of 0.003.Then, the FITC-dextran fluorescence intensity in the lumen of the GUVs was monitored over time by CLSM after the addition of DoNs to the GUVs dispersion.
The size cut-off of the DoNs was also investigated by measuring the size of dextran macromolecules with molar masses similar to those of the FITC-dextran molecules previously used for the leakage assay.To do so, dynamic light scattering (DLS) measurements were performed to determine the hydrodynamic radius (R H ) using a Zetasizer Nano ZSP at 25 °C.Dextran dispersion (5 μL) were added to 800 μL of PBS solution in the cuvette at a final concentration of 2 g L −1 .The measurement angle was 173°, and the data were analyzed by intensity distribution.
UV-visible Spectroscopy: A 3.0 μL sample of DNA Polymerase I Large (Klenow) Fragment (5000 U mL −1 ) was analyzed at OD 280 using a Nanodrop ND-1000 spectrophotometer and blanked with 10 mM Tris-HCl, 5 mM MgCl 2 .The observed average measurement of 3 replicate samples at 280 nm was converted to mg mL −1 using an extinction coefficient of 55 450 M −1 cm −1 and a molecular weight of 68 kDa for Klenow polymerase.

DNA Duplex Formation in GUVs:
The formation of the DNA duplex into GUVs encapsulating the Klenow fragments DNA polymerase I at 10 nM, and a DNA template at 20 nM was monitored at room temperature.The DNA polymerase concentration was measured by UV-vis spectroscopy.Then, 200 μL of the GUVs dispersion, 1 μL of dNTP Mix at 10 mM, 1 μL of scrambled and/or unscrambled primer at 200 nM, and 1 μL (1X concentration) of LightCycler-FastStart DNA Master SYBR Green (Roche Diagnostics GmbH) in 10 mM Tris-HCl, 5 mM MgCl 2 were added into the microscope wells.Then, the fluorescence intensity in the lumen of the GUVs was monitored by CLSM.
In Bulk Klenow Enzymatic Assay: Fluorescence spectra of samples were measured on a Spectramax id3 plate reader at room temperature.200 μL of the reaction mix consisting of 1 μL of dNTP Mix (10 mM), 1 μL of scrambled or unscrambled primer (200 nM), and 1 μL (1X concentration) of LightCycler-FastStart DNA Master SYBR Green (SG) (Roche Diagnostics GmbH) in 10 mM Tris-HCl, 5 mM MgCl 2 was added in each well of a 96 well plate.SG was excited at 485 nm and the fluorescence monitored overtime.

Figure 1 .
Figure 1.Schematic representations of the DoNs structure and GUVs production by microfluidics.A): 3D renderings of the T-shaped DoN with DNA double helices represented as grey cylinders.B): 2D projections representing the top (left) and bottom (right) of the pore, with the positions of singlestranded DNA exposed on the top marked in yellow and on the bottom in blue.C) Side view of the DoN showing hybridization of complementary ssDNA modified with hydrophilic Alexa 488 (green) or ssDNA coupled to hydrophobic cholesterol labelled with BODIPY 630/650 (red).D) Schematic representation of the microfluidic setup used for generating double-emulsion droplets via an inner aqueous (IA), a polymer organic (PO) and an outer aqueous (OA) phase.E) CLSM micrograph of GUVs produced by the microfluidics setup shown in D).GUVs membranes are stained with BODIPY 630/650 dye (red signal) and the hydrophilic cavity contains calcein (green signal).(Scale bar = 100 μm).F) Histogram revealing the size distribution of N = 105 GUVs (mean GUV diameter = 37 μm and standard deviation = 1.5 μm).

Figure 2 .
Figure 2. Modes of insertion of DoNs into GUVs polymeric membranes.Left-hand column: schematic representations (thickness of the membrane not to scale with the diameter), middle column: CLSM red channel micrographs, right-hand column: CLSM green channel micrographs.A): GUVs just after DoNs addition into the GUVs dispersion (external addition).B): GUVs 60 s after DoNs addition into the GUV dispersion (external addition).C): GUVs just after their production with DoNs added in the IA phase before GUVs formation (internal addition).D): GUVs 60 s after their production with DoNs added in the IA phase before GUV formation (internal addition).The DoNs are hybridized with cholesterol ssDNA labelled with BODIPY630/650 (red strands) and ssDNA labelled with Alexa 488 (green strands).

Figure 3 .
Figure 3. Dye diffusion through DoNs, investigated via calcein leakage and influx assays.Left-hand column: schematic representations, middle and right columns (membrane thickness not scale): CLSM micrographs of GUVs at t = 0 min and t = 30 min after start of leakage or influx.A): GUVs with membrane stained with BODIPY 630/650 (red), and encapsulating calcein (green) at 10 μM, at t = 0 and 30 min after addition of DoNs at 9 nM in the OA phase (external insertion).B): GUVs with membrane stained in red with BODIPY 630/650, and equipped with DoNs initially added in the IA phase at 9 nM (internal insertion), at t = 0 and 30 min after addition of calcein (green) at 10 μM in the OA phase.

Figure 4 .
Figure 4. Temporal fluorescence analysis of GUVs encapsulating calcein with and without DoNs.A): CLSM micrographs of a GUV with calcein (green) and polymeric membrane stained with BODIPY 630/650 (red) under control conditions (no DoN) and after adding 9 nM DoNs in the OA phase.B): Overtime fluorescence intensity measured for 19 ≤ N ≤ 25 GUVs after adding DoNs at different concentrations in the OA phase.Data fit to exponential decay models (dashed lines).For each condition, the data were divided by the initial fluorescence intensity value.

Figure 5 .
Figure 5. DoNs insertion and diffusion efficiency.A): The number of DoNs per GUV (Np) (black triangles), and the percentage of effective pores for calcein diffusion (blue squares), obtained at different DoNs concentrations in the outer phase of the GUVs.The values were calculated from the fluorescence intensity curves fitting obtained for 20 ≤ N ≤ 30 GUVs, using a Fick's law (Note S2, Supporting Information).B): CLSM micrographs of GUVs (stained with BODIPY 630/650, red signal) encapsulating FITC-dextran (green signal) of molecular weight ranging from 7 to 150 kDa before (0 min) and after (30 min) DoNs addition at 9 nM in the OA phase.The light blue backgrounds in B) highlight the size and molecular weight conditions for which dextran can diffuse through the DNA-nanopores inserted into the GUVs membrane, while the light red backgrounds highlight the conditions for the non-diffusion.

Figure 6 .
Figure 6.Specific DNA duplex formation in GUVs permeabilized with DoNs.A): Schematic representation of the DNA duplex formation within GUVs functionalized with DNA-nanopores.The Klenow fragment of DNA polymerase I (blue) in the lumen of the GUV catalyzes the templated duplex formation of the complementary DNA strand.B): CLSM micrographs of GUVs initially encapsulating Klenow fragment of DNA polymerase I, SG, and the ssDNA template, obtained at different times after the addition of dNTP mix, SG, unscrambled and/or scrambled primer in the GUVs' OA phase.C): Mean overtime fluorescence intensity measured inside 20 ≤ N ≤ 29 GUVs after the addition of scrambled and/or unscrambled primers, SG, and dNTPmix in the OA in presence and in absence of pores (dark blue pentagons) (all the data were divided by the maximum fluorescence intensity value of the "unscrambled" conditions).