DNA Origami Signaling Units Transduce Chemical and Mechanical Signals in Synthetic Cells

Transmembrane proteins transmit chemical signals as well as mechanical cues. The latter is often achieved by coupling to the cytoskeleton. The incorporation of fully engineerable membrane‐spanning structures for the transduction of chemical and, in particular, mechanical signals is therefore a critical aim for bottom‐up synthetic biology. Here, a membrane‐spanning DNA origami signaling units (DOSUs) is designed and mechanically coupled to DNA cytoskeletons encapsulated within giant unilamellar vesicles (GUVs). The incorporation of the DOSUs into the GUV membranes is verified and clustering upon external stimulation is achieved. Dye‐influx assays reveal that clustering increases the insertion efficiency. The transmembrane‐spanning DOSUs act as pores to allow for the transport of single‐stranded DNA into the GUVs. This is employed to trigger the reconfiguration of DNA cytoskeletons within GUVs. In addition to chemical signaling, mechanical coupling of the DOSUs to the internal DNA cytoskeletons is induced. With chemical cues from the environment, clustering of the DOSUs is induced, which triggers a symmetry break in the organization of the DNA cytoskeleton inside of the GUV. DNA‐based transmembrane structures are engineered that transduce signals without transporting the signaling molecule itself—providing a route toward signal processing and adaptive synthetic cells.


Introduction
DNA nanotechnology emerged as a powerful tool for bottom-up synthetic biology. [1]In particular, versatile DNA DOI: 10.1002/adfm.202301176nanostructures have been engineered to function as transmembrane pores which allow for the transport of ions, [2][3][4] fluorophores, [5,6] macromolecules, [7,8] DNA, [9,10] proteins, [11] or lipids. [12]oncomitantly, DNA nanostructures have been engineered as structural and functional mimics of cytoskeletal elements.In particular, DNA nanotubes have been reconstituted in the confinement of cell-sized water-in-oil droplets [13,14] and recently also in GUVs. [15,16][20] Very recently, Li et al. demonstrated a transmembrane-spanning version of this seed, [8] which connects to DNA nanotubes on the GUV's exterior.This and all previous examples of DNA nanopores share a common function: They transport molecules across the membrane.There is no example of synthetic signal transduction where information can be transduced without transporting the signaling molecule itself.In living cells, on the other hand, this mode of signal transduction is crucial.Mechanical cues from the cell's enviroment are transmitted by a mechanical link between the transmembrane protein and the cytoskeleton. [21]Engineering signal transduction without molecular transport in synthetic cells is appealing from a conceptual point of view and it would enable information processing without the risk of dissipating gradients and "losing" encapsulated components due to a porous membrane.
Here, we equip a transmembrane-spanning DNA origami with the capability to transmit chemical and mechanical signals across the membrane of a GUV, causing remodeling of its internal DNA-based cytoskeleton upon external stimulation.We implement both of the described mechanisms for transmembrane signaling: First, the signaling molecule itself enters the GUV to induce cytoskeletal remodeling.Second, the signaling molecules serve as an environmental cue that transmits information for cytoskeletal remodeling by mechanical coupling with our engineered transmembrane receptors.We thus achieve information transduction across the GUV membrane without transporting the signaling molecule itself.

Results and Discussion
First, we set out to design a transmembrane-spanning DNA origami signaling unit (DOSU) with the following functional features: (i) The DOSUs should have the ability to act as nanopores for the translocation of small molecules.(ii) It should be possible to cluster the DOSUs to mimic receptor clustering as a transmembrane signaling mechanism.(iii) They should enable chemical signaling across the GUV membrane.(iv) The transmembrane-spanning DOSUs should be mechanically coupled to a cytoskeleton within the GUV to structurally mimic focal adhesions.Taken together, we thus aim for a minimal system enabling chemical as well as mechanical signal transduction across the GUV membrane.Figure 1a illustrates the basic design of our transmembrane DOSU.We adapt a DNA origami design from Mohammed et al., [20] which was initially used for nucleating algorithmic self-assembly, [22] then as a binding site for DNA nanotubes [18][19][20][23][24][25][26] and most recently as a transmembrane pore. [8] We odified this DNA origami [20] with 12 overhangs that can each bind to a cholesterol-tagged DNA and with 6 biotin modifications (Figure S1, Supporting Information).The first allows the attachment and insertion of the DOSU into lipid membranes by hydrophobic interactions, the latter enables the stimuli-induced clustering upon addition of streptavidin.First, we verify the successful assembly of the DOSU with confocal microscopy (Figures S2 and S4, Supporting Information) and agarose gel electrophoresis (Figure S3, Supporting Information).
Next, we test the first functional feature of our DOSU, namely its ability to act as a transmembrane pore (Feature (i)).By incubating DOSUs with cholesterol-tagged DNA and subsequently mixing them with SUVs with a diameter of 100 nm, we observe the binding of the DOSUs to the SUV membrane via cryo-electron microscopy (cryo-EM, Figure 1b).The optimization of our system for electron microscopy is described in Note S1 and Figure S5 (Supporting Information).If the DOSU penetrates the membrane as designed, it should be bound at a perpendicular angle to the SUV surface.By analysing the angle of bound (or nonfree) DOSUs to the SUV membrane, we find that 87% indeed attach to the SUV membrane in the desired orientation (75-105°, Figure 1c; Figures S6a and S7, Supporting Information) and only 31% of 268 analysed DOSUs are free in solution (Figure S6b, Supporting Information).
In addition, the successful incorporation of DOSUs into SUV membranes is confirmed with dithionite-mediated quenching experiments (Figure S8, Supporting Information).To further prove their insertion into the lipid membrane of giant unilamellar vesicles (GUVs), we conduct dye influx assays.To enable vizualization with confocal microscopy, we bind the DO-SUs to GUVs (Figure 1d) containing a low fraction of fluorescently labeled lipids (0.5% Atto488-PE) and add a water-soluble membrane impermeable Alexa647-NHS ester fluorescent analyte from the outside.With the DNA origami concentration measured with UV-vis spectroscopy, we approximate the number of DOSUs on the GUV membrane to be 10700 ± 1200, which corresponds to a density of (16 ± 1) DOSUs μm −2 (Note S2 and Figure S9, Supporting Information).Following the addition of the fluorescent analyte, we observe individual GUVs over time in presence and absence of the DOSUs (Figure 1e).To quantify the amount of dye influx, we analyse the intensity ratio inside and outside (I in /I out ) of the GUVs over time.The amount of fully-permeabilized GUVs increases over time for GUVs containing the DOSUs (Figure 1f).On the other hand, no dye influx can be observed in their absence or in absence of cholesterol tags (Figure S10, Supporting Information).After 330 min, the median value of the ratio I in /I out is 0.99 in presence of DO-SUs and 0.08 in their absence, respectively.This confirms the successful engineering of transmembrane DOSUs that can serve as pores for small molecules like fluorophores (Feature (i)).Importantly, the median value of the intensity ratio, and thereby the fraction of fully-permeabilized GUVs over unpermeabilized GUVs, can be tuned by changing the concentration and density of DOSUs on the membrane (Figure S11, Supporting Information).Note that we can exclude GUV permeabilization due to osmotic stress, as we add the same amount of iso-osmotic solutions for all conditions.Further, we can assume that the DOSUmediated membrane permeation is due to a non-transient insertion (Figure S12, Supporting Information).Nonetheless, we want to highlight that a toroidal lipidic pore likely forms around the DOSU channel, [3] giving rise to an alternative conductance pathway.Due to the large inner channel diameter of the DOSU the main permeation pathway leads nonetheless through the central channel lumen. [8]In addition to the dye influx assay, we also observe single GUVs over time and perform dye efflux assays.Fluorophore transport happens within minutes after signaling unit insertion (Figure S13 and Video S1, Supporting Information).
A remarkable feature of transmembrane proteins is the ability to cluster in response to a molecular trigger to enable a certain function, for example, the formation of focal adhesions. [27]o engineer the clustering of our DOSUs (Feature (ii)), we modified 6 DNA oligos with biotin around the DNA origami circumference (Figure 2a; Figure S1b, Supporting Information).Streptavidin serves as a stimulus to induce DOSU clustering in bulk (Figure 2b; Figure S14, Supporting Information) and on the GUV membrane (Figure 2c) as we confirmed with confocal microscopy and FRAP measurements (Figures S15 and S16, Supporting Information).In addition, the 71% higher diffusion coefficient of unclustered transmembrane DOSUs indicates successful 2D-clustering of the biotinylated DOSUs on GUV membranes (Figure S16, Supporting Information).The degree of clustering increases for increasing streptavidin concentrations within the range from 0 to 4 μM for a given DOSU concentration (100 ng μL −1 of DNA).Interestingly, dye influx experiments indicate that the membrane permeabilization is enhanced for clustered DOSUs compared to unclustered DOSUs (Figure 2d; Figure S17, Supporting Information).The median value of I in /I out 30 and 60 min after addition of streptavidin is 0.24, 0.54 for clustered and 0.13, 0.19 for unclustered DOSUs, respectively.This underlines that the increasing line tension for the formation of a larger pore can be overcompensated by the energy gain for the insertion of additional cholesterol moieties.
A second major function of natural pores is the transport of essential molecules across the membrane for chemical transmembrane signaling.To employ the engineered DOSUs for chemical signaling (Feature (iii)), we interfaced our DNA pores with DNAbased mimics of cytoskeletons.We used DNA double-crossover tiles [28] which self-assemble into hollow DNA filaments and encapsulated them into GUVs. [15]The DNA filaments were modified with single-stranded overhangs so that they can be disassembled via toehold-mediated strand displacement. [29]The presence of an invader DNA strand thereby leads to filament disassembly by binding the overhang of a single DNA tile (Video S2, Supporting Information).The specific disassembly function of that invader sequence compared to a random sequence has been proven by agarose gel electrophoresis (Figure S18, Supporting Information).By transporting the invader strand across the GUV membrane via the clustered DOSUs, we thus aim to create a chemical signaling pathway, whereby an external stimulus leads to filament disassembly in the interior of the GUV (Figure 3a).First, we verify that the DOSUs can transport fluorescently-labeled single-stranded DNA (ssDNA) across the GUV membrane.Over the course of 330 min, we observe that the median value of I in /I out reaches 0.75 (Figure 3b,c).The smaller amount of fully-permeabilized GUVs at this time point is in agreement with the bigger size and slower diffusion of ssDNA in comparison to the fluorescent analyte.Notably, ssDNA influx was also achieved with the unclustered signaling unit (Figure S19, Supporting Information), which is consistent with the outer diameter of the pore (12 nm).We can thus translocate the invader strand into the GUV lumen, which is filled with DNA filaments.In presence of clustered DOSUs, we observe the filament disassembly inside GUVs over time (Figure 3d).As a measure for the degree of filament disassembly, we quantify the porosity of the GUV lumen containing DNA filaments from confocal images (see Experimental Section and Figure S20, Supporting Information).After 5 h of incubation with streptavidin, the normalized porosity decreases from 1 to 0.53 ± 0.18 in presence of the DOSUs (Figure 3e).As we know from the influx experiments with fluorescently labeled DNA (Figure S19, Supporting Information), the GUVs are fully permeabilized at this time point.Moreover, our results indicate that filament disassembly for clustered DO-SUs (Figure 3e) is enhanced compared to unclustered DOSUs (Figure S21, Supporting Information).This shows the successful filament disassembly using chemical signaling across the GUV membrane.Ultimately, we set out to also mechanically couple the externally induced DOSU clustering to DNA filament remodeling inside the GUVs (Feature (iv)).For this purpose, we make use of the cylindrical shape of the DOSU which was designed for the templated binding to DNA filaments. [20]We explicitly adapt one of both sides, namely the cholesterol-modified side which is pointing into the GUV lumen (Figure 4a).First, we verify the successful functionalization of the DOSU with confocal microscopy (Figure 4b).Second, we quantify the DOSU-filament attachment for pre-annelaed filaments incubated with the pre-annealed DO-SUs and for the complete set of single stranded DNA sequences for DNA tile and filament assembly with the pre-annealed DO-SUs.After 2 h already 39% of filaments bind to the DOSU for both pre-annealed DNA filaments and also un-annealed ssDNA at the correct ratio of 1:1 (1 DOSU per filament).Successful binding increases to 85% after 22 h (Figure 4c).Note that the remaining 15% of filaments correspond to free filaments without DOSU.Binding of multipe DOSUs to one filament has not been observed.
We can thus link DNA filaments reconstituted on the inside of GUVs to DOSUs added externally due to the transmembranespanning nature of the DOSU.After 5 h of incubation, we observe DNA filament remodeling, inducing a symmetry break in the distribution of the filaments in presence of streptavidin (Figure 4d).This showcases the mechanical coupling across the GUV membrane upon the addition of an external chemical stimulus (Video S3, Supporting Information).To analyse the amount of filament remodeling, we quantify the normalized center-ofmass displacement r/r 0 of the filament fluorescence inside GUVs (see Experimental Section and Figure 4e).In presence of streptavidin, the center-of-mass displacement r/r 0 normalized to the radius of the GUV is 0.63 ± 0.21, whereas it is only 0.20 ± 0.10 in the absence of streptavidin.We thereby show how two separate modules for synthetic cells, namely cytoskeletons and transmembrane signaling units, can be combined to realize simplified focal adhesion mimics.

Conclusion
Every natural cell interacts with its environment or neighbouring cells using chemical and mechanical signaling pathways across the compartment barrier.This is mainly achieved by transmembrane proteins that can translate external cues into intracellular instructions. [30]Due to its inherent complexity, the reconstitution of natural signaling pathways into synthetic cells is challenging.Moreover, it is especially exciting to build a completely synthetic and rationally engineerable transmembrane mimic that can guide downstream signal transduction in synthetic cells.Out of the two signal transduction mechanisms that exist in nature, only one has so far been realized in synthetic systems: While there is several examples of DNA nanopores which transport molecules across membranes, there is no example of engineered components for information transduction without transporting the signaling molecule itself.
Here, we designed minimal signal transduction pathways which can transduce chemical signals as well as information without transport of the signaling molecule itself.We would like to acknowledge that our work builds on a DNA origami design by Mohammed et al. [20] This design has been used to seed the growth of DNA nanotubes, and in parallel to our work it has also been equipped with membrane insertion capabilities. [8]Li et al. added DNA nanotubes from the outside of the GUV to elongate their nanopore and analyzed the transport, [8] but without mechanical feedback.Here, on the other hand, the DNA nanotubes are reconstituted inside of the GUV and we engineer a transmembrane spanning version of the seed that can realize a mechanical response of the internal DNA cytoskeleton to an environmental stimulus.For the purpose of our work, it was key to link the DNA nanotubes to the transmembrane pore, because otherwise mechanical coupling would not be possible.This was the main reason for us to choose a DNA-based transmembrane pore over a protein pore.A protein-based pore like alphahemolysin would not exactly match the symmetry and geometry of the DNA filaments, which would likely be an impediment to filament growth and appropriate mechanical coupling.Generally speaking, different pieces of DNA-based hardware for synthetic cells can be engineered to be inherently compatible with one another.Additionally, we show that the clustering of our artificial DNA-based receptors increases their insertion efficiency.By reconstituting DNA cytoskeletons into GUVs we implement chemical signaling that induces the filament disassembly upon transport of an invader strand.By binding the DNA filaments to the DOSUs, we also show the mechanical remodeling of the DNA cytoskeleton inside the GUV upon a chemical stimulus from the outside, which clusters the DOSUs.It is important to note that these are two completely different signal transduction mechanisms: In the first case, the signaling molecule passes across the membrane to remodel the DNA cytoskeleton.In the latter case, the signaling molecule does not need to enter the vesicle, it transmits information for cytoskeletal remodeling by mechanical coupling with our engineered transmembrane receptors-a mechanism which deserves to be explored further in the context of bottom-up synthetic biology.
Information processing for synthetic cells is an exciting direction of future research.It will be interesting to direct the cell-tocell communication in between two or more GUVs and to add more elements to the GUVs in order to fully mimic the complex and highly evolved symmetry breaking process present in natural cells.From a practical point of view, sense-and-respond systems for vesicles will be key elements of advanced drug delivery systems.All in all, this shows the great potential of DNA nanotechnology for the development of engineerable and fully-synthetic cells with applications in biomedicine, cellular biophysics and synthetic biology.

Experimental Section
DOSU Design and Assembly: The DNA sequences for the DOSU were adapted from Mohammed et al. [20] The design was shown in Figure S1 (Supporting Information).To allow binding of the DOSU to the lipid membrane, we incorporated 12 single-stranded overhangs at column 1 and designed complementary cholesterol-tagged DNA linkers (Figure S1, Supporting Information).Moreover, we added 6 biotin-modifications on overhangs at column 6 in order to induce the DOSU clustering with streptavidin (Figure S1, Supporting Information).
Typically, 50 to 100 μL of the DOSUs were assembled in 1 × tris-acetate-EDTA (TAE) buffer and 12 mM MgCl 2 by using 16 nM of p7429 scaffold (tilibit), 160 nM of staples and 80 or 160 nM of Atto647N-tagged DNA.The solution was annealed using a thermocycler (Bio-Rad) by heating the solution to 90°C and subsequently cooling it in steps of 1°C every minute until holding it for 1 h at a temperature of 45°C.Subsequently, the solution was further cooled to 32°C in steps of 0.1°C.Following the annealing of the DOSU, it was purified three times by spin filtration with an Amicon 50 kDa filter.Annealed solution (50μL) were diluted with 450 μL of folding buffer (1 × TAE, 12 mM MgCl 2 ) and centrifuged for 5 min at 10770 g.This process was repeated three times.After annealing and purification the DOSU concentration was measured with a spectrophotometer (IM-PLEN, NanoPhotometer NP80) and stored at 4°C for up to three days.The DNA strands were either purchased from Integrated DNA Technologies or Biomers (purification: standard desalting for unmodified DNA oligomers, HPLC for DNA oligomers with modifications).All DNA sequences were listed in Table S1 (Supporting Information).
Confocal Fluorescence Microscopy: A confocal laser scanning microscope LSM 900 (Carl Zeiss AG) was used for confocal microscopy.The pinhole aperture was set to one Airy Unit and the experiments were performed at room temperature.Images of DNA filaments in Figure 4 and Figures S2, S4, and S9c-e (Supporting Information) were acquired using the Airyscan mode.The images were acquired using a 20× (Plan-Apochromat 20×/0.8Air M27, Carl Zeiss AG) or 63× objective (Plan-Apochromat 63×/1.4Oil DIC M27).Images were analysed and processed with ImageJ (NIH, brightness and contrast adjusted).
Agarose Gel Electrophoresis: DOSU assembly and filament disassembly were verified using agarose gel electrophoresis.0.7% agarose gel was casted using 0.5 × TAE buffer containing 12 mM MgCl 2 to visualize DOSU assembly.For filament disassembly, 2% and 4% agarose gel was casted using 1 × TBE buffer containing 10 mM MgCl 2 .The gel casting buffer was also used as running buffer.In Figure S3 (Supporting Information), 10μL of purified origami (200 ng μL −1 ) was mixed with 6 × Blue Loading Dye (New England Biolabs, MA, USA).Samples in Figure S16 (Supporting Information) were mixed with 6 × Gel Loading Dye (Tilibit, Munich, Germany).The samples were then loaded into the gel pockets.Quick-Load 1 kb Extend DNA Ladder (Figure S3, Supporting Information), 1kb DNA Ladder (Figure S18a, Supporting Information) and TriDye Ultra Low Range DNA Ladder (Figure S18b, Supporting Information) (New England Biolabs, MA, USA) were used as reference.The gel was run on ice at constant voltage of 60 V for 3-5 h (Figure S3, Supporting Information) and 75 V for 2-3 h (Figure S18, Supporting Information).The gel was subsequently stained with GelRed (Sigma-Aldrich) then imaged using Azure 600 imager (Azure biosystems).
Image Analysis of DOSU Insertion Angle: The insertion angles to the SUVs of the DOSUs were measured from cryo-EM micrographs using the Angle Tool from ImageJ. [31]The insertion angle was formed by a tangent line to the surface of the SUV and a line parallel to the DOSU.DOSUs were considered to be inserted correctly if the insertion angle was greater than 75°and lower than 105°.All angles with x > 90°were defined as 180°− x for plotting.The angle histogram of DOSU insertion (Figure 1c) was plotted using Python [32] (code available: https://github.com/Biophysical-Engineering-Group/DOSU_angleplot.git).The histogram of DOSU insertion (Figure S7, Supporting Information) was plotted using Graph-Pad Prism.
Preparation of Giant Unilamellar Vesicles Using Electroformation: Giant unilamellar vesicles without DNA filaments were prepared using the electroformation method [33] using a VesiclePrepPro device (Nanion Technologies GmbH).Lipid mix (30 μL of 5 mM containing 99 % or 99.5 % 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) and 1 % or 0.5 % 1,2dioleoyl-sn-glycero-3-phospho ethanolamine-Atto488 (Atto488-PE) or no fluorophore-modified lipid) in CHCl 3 were homogeneously spread on the conductive side of an indium tin oxide (ITO) coated glass slide (Visiontek Systems Ltd).After evaporating the chloroform for 20 min under vacuum, a rubber ring was placed on the lipid-coated ITO slide and filled with 270 μL of 270 mM sucrose solution to match the omsolarity of the phosphate buffered saline buffer.The second ITO slide was put on top and the chamber connected to the electrodes of the VesiclePrepPro (Nanion).An AC field (3 V, 5 Hz) was applied via the electrodes for 138 min while the solution was heated to 37 °C.GUVs were collected immediately after electroformation and stored at 6°C for up to 7 days.
Dye Influx Assays: GUVs were prepared using electroformation from a lipid mixture containing 99.5% DOPC and 0.5% fluorescent Atto488-DOPE lipids.The DOSU was mixed in a ratio of 0.96: 0.02: 0.02 with two types of cholesterol-tagged DNA (100 mM) in order to modify the DOSU with cholesterol.The mixture of DOSU and cholesterol-tagged DNA was incubated for 10 min at 35°C and 200 rpm.Electroformed GUVs (7.5 μL) were diluted in 34μL 1 × DPBS containing 5 mM MgCl 2 and incubated with 5 μL cholesterol-tagged DOSUs for 1 min to allow binding to the GUVs.Subsequently, 1 μL of 1: 20 diluted AlexaFluor 647 NHS-ester fluorescent analyte (10 mM) was added.At each time point, 5-7 μL were pipetted out of the PCR tube and imaged within a custom-built observation chamber that was coated with 2 mg ml −1 bovine serum albumin (Sigma-Aldrich) for 15 min to prevent fusion of the GUVs using a confocal microscope.

Fluorescence Recovery After Photobleaching (FRAP):
The GUV-DOSU system was prepared as for the dye influx assay.The diameter of the spherical bleaching region was set to 10 mm on the upper surface of the selected GUV.Bleaching was started after three initial images.The laser power for the bleaching step ( ex = 640 nm) was set to 100%.The diffusion coefficients were calculated from the intensity profiles according to a protocol published by Axelrod et al. and Soumpasis. [34,35]NA Tile Design and Filament Assembly: DNA filament sequences were adapted from Rothemund et al. [28] The individual DNA oligomers (5 per tile) were mixed to a final concentration of 5 mM in 1 × DPBS and 10 mM MgCl 2 .The solution (7.5 μL) were annealed using a thermocycler (Bio-Rad) by heating the solution to 90°C and cooling it to 25°C in steps of 1-0.5°C for 2 or 4.5 h, respectively.The assembled DNA filaments were stored at 4°C and used within a week after annealing.The DNA strands were either purchased from Integrated DNA Technologies or Biomers (purification: standard desalting for unmodified DNA oligomers, HPLC for DNA oligomers with modifications).All DNA sequences were listed in Tables S2 and S3 (Supporting Information).
Filament Disassembly by Invader Strand Influx: DNA origami signaling units (DOSUs) were assembled with adapter tiles and cholesterol-tags and used within one week.Cy3-labeled DNA filaments were assembled at a concentration of 5 mM and designed so that they do not bind to the DOSU.DNA filaments were encapsulated into GUVs at 500 nM in a 1 × DPBS inner solution with 10 mM MgCl 2 using the droplet-stabilized GUV formation method. [36]The iso-osmotic release buffer contained the same MgCl 2 concentration of 10 mM to stabilize DNA nanostructures on the outside of GUVs.After releasing the DNA filament-containing GUVs, they were reannealed in the thermocycler in steps of 1°C per 45 s from 90 to 25°C in order to ensure filament assembly within GUVs.Subsequently, 45 μL filamentencapsulating GUVs were mixed with 5 μL cholesterol-tagged DOSUs.After incubation for 1 min to allow DOSUs binding to GUVs, 2.5 mM invader strand solution was added to disassemble DNA filaments. [29]In the case of the biotinylated clustered DOSUs, an additional 2.5 μL streptavidin (5 mg ml −1 ) in DPBS was added to the DOSU-GUV mixture after 1 min.For the negative controls either the addition of the DOSUs or invader strand was omitted.At each time point 5-7 μL were pipetted out of the PCR tube and imaged within a custom-built observation chamber.
Preparation of Giant Unilamellar Vesicles Containing DNA Filaments: Giant unilamellar vesicles (GUVs) containing DNA filaments were formed using the droplet-stabilized GUV formation method. [36]Briefly, 1.25 mM SUVs, 5 mM DNA filaments, 10 mM MgCl 2 , and phosphate buffered saline (DPBS consisting of 137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , and 1.8 mM KH 2 PO 4 ) were mixed together.The aqueous mix was layered on top of an oil-surfactant mix containing 1.4 wt% perfluoropolyetherpolyethylene glycol (PFPE-PEG) fluorosurfactants (Ran Biotechnologies) and 10.5 mM PFPE-carboxylic acid (Krytox, MW, 7000-7500 g mol −1 , DuPont) in a microtube (Eppendorf).The ratio in between aqueous and oil phase was 1: 3, generally leading to volumes of 100 μL:300 μL.Dropletstabilized GUVs were generated by shaking the microtube vigorously by hand.The water-in-oil emulsion droplets were left at room temperature for 0.5-1 h.Within this incubation period, the SUVs fused at the droplet periphery to create a spherical supported lipid bilayer, termed dropletstabilized GUV.Afterward, the oil phase was removed and 100 μL of 1 × DPBS was added on top of the emulsion droplets.The droplet-stabilized GUVs were destabilized by addition of 100 μL of perfluoro-1-octanol (PFO, Sigma-Aldrich) to release freestanding GUVs into the DPBS.GUVs were stored for up to two days at 6°C.GUVs were imaged in a custom-built observation chamber that was coated with 2 mg ml −1 bovine serum albumin (Sigma-Aldrich) for 15 min to prevent fusion of the GUVs with the glass coverslide.
Image Analysis of the State of Filament Assembly: 100 μL of GUVs containing 500 nM Cy3-labeled DNA filaments were mixed with 2 μL DNase (4 units, Sigma-Aldrich).GUVs were then incubated at 37°C for 20 min to digest DNA filaments present in the outer aqueous solution due to an imperfect release.The solution was then heated to 95°C to inactivate the DNase and slowly cooled to 25°C in steps of 1°C for 2 h.Subsequently, 2 μL of DOSU was incubated with cholesterol-tagged DNA and added to 36 μL of GUVs.After 1 min of incubation to allow DOSU attachment, 2 μL of streptavidin (Sigma-Aldrich, 5 mg ml -1 ) was added.For the negative control the addition of streptavidin was omitted.GUVs were imaged in the equatorial plane after 5 h.The center-of-mass displacement of DNA filaments was analysed by choosing the GUV center as center (0,0) of a 2D coordinate system.DNA filament fluorescence was thresholded to remove background fluorescence of unpolymerised DNA tiles and the center-ofmass (x,y) of the fluorescence intensity analysed.The absolute value of the DNA filament center-of-mass was then calculated in the reference of the GUV according to |d| = √ (x − 0) 2 + (y − 0) 2 .Statistical Analysis: All the experimental data were reported as mean ± SD from n experiments, filaments or GUVs.The respective value for n was stated in the corresponding figure captions.All experiments were repeated at least twice.In Figure 1c the number of analysed individual SUVs was n = 185.In Figure 1f, the number of analysed individual GUVs per timestep was in the range of 20 < n < 24 (n = 150 in total) for GUVs with DOSUs (upper plot) and 30 < n < 32 (n = 216 in total) for plain GUVs (lower plot).In Figure 2d, the number of analysed individual GUVs per time step was in the range of 39 < n < 44 (n = 567 in total).In Figure 3c, the number of analysed individual GUVs for each time point was n = 20 (n = 140 in total).In Figure 3e, the number of analysed individual GUVs was as follows: n = 34 (-0 h), n = 28 (−5 h), n = 31 (+0 h), n = 69 (+5 h).In Figure 4c, the numbers of analysed DNA filaments were n = 628 and n = 162 and individual tiles ("ssDNA") were n = 186 and n = 196, respectively.In Figure 4e, the number of analysed individual GUVs was n = 10 per condition (n = 20 in total).To analyse the significance of the data, a Mann-Whitney test (Figure 2d) or student's t-test with Welch's correction (Figure 4e) were performed using Prism GraphPad (Version 9.1.2) and p-values correspond to **** p ⩽ 0.0001, *** p ⩽ 0.001, ** p ⩽ 0.01, * p ⩽ 0.05, and ns: p ⩾ 0.05.

Figure 1 .
Figure 1.Cholesterol-tagged DNA origami signaling units (DOSUs) insert into phospholipid bilayers.a) Schematic representation of the cylindrical DNA origami signaling unit design inserted into the GUV membrane via 12 cholesterol tags.The DOSU creates a passage for fluorophores to be transported into the GUV lumen.b) Representative cryo electron micrograph of small unilamellar vesicles (SUVs) with transmembrane DOSUs.Scale bar: 100 nm.c) Insertion angle of the DOSU in SUVs extracted from cryo-EM micrographs (n = 185).d) Confocal image of DOSUs (red, Atto647-labeled,  ex = 640 nm) inserted into a GUV membrane.Scale bar: 10 m.e) Confocal images of GUVs (green, Atto488-labeled,  ex = 488 nm) immersed in a solution containing a fluorescent analyte (cyan, Atto647,  ex = 640 nm).The first row shows GUVs with DOSUs and the second row without DOSUs at t = 0 and t=330 min, respectively.Scale bars: 10 m.f) Ratio of inner and outer intensity I in /I out of n ⩾ 20 individual GUVs over time and histogram of the median value of I in ∕I out with (top, n ⩾ 20) and without (bottom, n ⩾ 30) DOSUs.

Figure 2 .
Figure 2. Stimulus-induced DOSU clustering.a) Schematic representation of biotinylated DOSUs inserted into a GUV membrane and clustered upon addition of streptavidin.b) Confocal images of clustered DOSUs (red, Atto647-labeled,  ex = 640 nm) with increasing streptavidin concentration, forming visible clusters from 0.2 to 4 μM streptavidin.Scale bars: 10 μm.c) Confocal image of a GUV with clustered DOSUs (red, Atto647-labeled,  ex = 640 nm) inserted into the membrane.Scale bar: 10 μm.d) Ratio of inner over outer intensity I in /I out of individual GUVs (n ⩾ 40, two technical repeats) over time and histogram of the median value of I in /I out for GUVs without strepatvidin and GUVs 60 min after the addition of streptavidin.The dye influx is significantly increased after addition of streptavidin indicating better insertion and transport rates for clustered DOSUs.A Mann-Whitney test was performed, the p-values are 0.019 and 0.007 for t=90 and t = 120 min, respectively.

Figure 3 .
Figure 3.Chemical signal transduction across the GUV membrane.a) Schematic representation of the influx of single-stranded invader DNA strands through clustered DOSUs to induce disassembly of reconstituted DNA cytoskeletons.b) Confocal images of GUVs (green, Atto488-labeled,  ex =488 nm) with clustered DOSUs immersed in an ssDNA-containing solution (cyan, Atto647-labeled,  ex =640 nm) at t=0 and t = 330 min.Scale bars: 10 μm.c) Ratio of inner over outer intensity I in /I out of individual GUVs over time and histogram of the median value of I in ∕I out (n = 20).d) Confocal images of GUVs with clustered DOSUs (red, Atto647,  ex = 640 nm) and encapsulated DNA cytoskeletons (orange, Cy3-labeled,  ex = 561 nm) immersed in a DNA invader-containing (5 μM) solution.Scale bars: 10 m.e) Normalized porosity of DNA cytoskeletons within GUVs after 0 and 5 h in presence and absence of DOSUs (2 technical repeats, total n⩾28, mean±SD).