1 Introduction

The plasma membrane is a rather complex organelle, formed from a bilayer composed mostly of a mixture of phospholipids, sphingolipids, cholesterol, and a variety of peripheral and integral proteins [1, 2]. Long believed to be merely a passive barrier between the cell and its surrounding environment, the plasma membrane plays a fundamental role in a number of biological processes, such as cell signaling and signal transduction, motility, and cellular division. While research of such phenomena is of high interest both on a fundamental and technological level, their complexity often enough requires the use of model systems that allow their interrogation in a controlled environment [3].

Supported lipid bilayers (SLBs) are one particular example of model lipid membranes that are of great interest, as they can not only recapitulate the heterogeneous composition and lateral dynamics of plasma membranes, but can also be patterned in 2D and 3D. While they are traditionally formed on continuous planar surfaces, over the last two decades SLBs have been combined with nanostructured surfaces (Fig. 1a) [4,5,6,7,8,9,10], patterned substrates [11,12,13], microfabricated compartments [14, 15], and 3D printed complex architectures [16].

Fig. 1
4 illustrations with photos and micrographs. A. membrane topology and a grid that highlights 2 shades. B. molecular distribution and a micrograph. D P P C and D O P C are labeled. C. protein self-organization between A T P and P subscript i. 2 micrographs with thread-like patterns. D. patterned D N A origami nanostructures and 2 micrographs with hexagonal patterns.

Molecular patterning strategies, with emphasis on model lipid membranes as mimics of natural membranes. a Epifluorescence photographs of a corral array (20 × 20 μm) of a supported lipid bilayer partitioned by a microfabricated grid of chrome lines. Corrals in the center were photobleached (dark) with a circular spot highlighting how the barriers (dark lines) prevent diffusive mixing of lipids between corrals. b Cholesteryl-TEG modified three-point star DNA tiles preferentially bind and self-assemble on the liquid-ordered phase of phase-separated DOPC/DPPC supported lipid bilayers. c Myosin II action on F-actin bound to supported lipid bilayers induces distinct patterns on the membrane such as filament bundles (left) and linked, polar asters (right). d Model and AFM image of patterned DNA origami triangles combined into a hexagon shape. Scale bars, b, 200 nm and inset 25 nm, c, 10 µm, d, 100 nm. a, adapted with permission from [5]. Copyright 1998 American Chemical Society, b, adapted with permission from [62]. Copyright 2017 American Chemical Society, c, adapted with permission from [66], d, reprinted by permission from Springer Nature: Nature, [36], Copyright 2006

Full size image

Beyond controlling membrane topology and topography, several methods can be applied to control molecular distribution and localization in and on SLBs. Lipid phase separation [17, 18] in membranes composed of lipids of different melting temperatures is itself an example of molecular demixing [19,20,21] (Fig. 1b) and can be harnessed to change the distribution of membrane-coupled proteins [22,23,24]. On the other hand, external forces can be applied to control molecular distribution: Electric fields can be used to generate gradients of charged lipids [25, 26], while surface acoustic waves have been shown to result in lipid demixing and even protein transport and accumulation [27,28,29]. Light responsive molecules can also be exploited to control membrane properties—the incorporation of azobenzene groups in a lipidic hydrophobic moiety results in control over lipid phase separation [30,31,32,33], while control over protein binding and localization can be achieved by the fusion of light responsive membrane-binding domains to proteins of interest [33,34,35]. While most of these methods offer reversible control over membrane properties and molecular localization, their applicability can be restricted due to membrane diffusion, introduced perturbations, spatio- and/or temporal resolution or scalability.

DNA-based nanostructures take advantage of the unique and inherent properties of the DNA double helix and are the paragon of a molecular breadboard. With its nanoscale addressability and diversity of possible modifications, complex molecular patterns can be built on the surface of DNA assemblies in 2D and 3D (Fig. 1d) [36,37,38,39]. It is thus tempting to employ DNA origami as a tool for patterning lipid membranes. Indeed, a lot of attention has been given in recent years to the development and investigation of membrane-active DNA nanostructures [40]. From the exploration of a variety of membrane-binding strategies, including different hydrophobic moieties [41,42,43,44], ligand-receptor type binding [43], covalent conjugation [45], and ionic strength [46], to the development of triggerable membrane-binding nanostructures [47, 48], channels [49,50,51] and membrane shaping coats [52,53,54,55,56], we now have a good understanding of how to control membrane binding and dynamics of DNA origami nanostructures [57,58,59]. For example, we know that not only the number, but also the position and accessibility of membrane anchors, governs the efficiency of membrane binding. Different strategies to attach the membrane anchor to the DNA origami will also result in different levels of sensitivity to variations in local electrostatics (Fig. 2a, b). However, patterning of DNA origami on lipid membranes has so far been limited to the assembly of large-scale arrays or the preferential binding of DNA nanostructures to different lipid phases (Fig. 1b) [42, 60,61,62,63].

Fig. 2
35 micrographs, 6 illustrations, and 3 line graphs. Micrographs present patterns of D N A origami, protein self-organization of Min D, quasi-stationary patterns of Min D and origami, cargo 2 and 42 with Cy 3 B and Cy 5. Illustrations are, Min D and E through D N A origami. Cargo 2 and 42. Mechanism of bulk diffusion and membrane diffusion. 2 fluctuating line graphs with 2 lines each.

DNA origami as a tool to elucidate the physical mechanism underlying cargo transport by MinDE self-organization. a AFM image of the bare DNA origami nanostructure (20-helix bundle; 110 × 16 × 8 nm) deposited on mica. b Charge sensitivity of membrane-binding DNA origami structures depends on the used attachment strategy of membrane-anchoring moiety. While TEG-chol moieties inserted directly into the structure result in poor DNA origami binding to membranes containing negatively charged lipids, when compared to bare DOPC, the use of an 18 nucleotide long double stranded DNA linker contributes to an efficient membrane binding in both conditions. Representative confocal microscopy images of the equatorial plane of giant unilamellar vesicles (GUVs) incubated with a 3 nM solution of DNA nanostructures modified with two TEG-chol anchors. GUVs contained 0.005 mol% Atto655-DOPE for fluorescence imaging, while each origami structure carried three Atto488 dyes. c Examples of patterns formed by MinDE self-organization on planar supported lipid bilayers in vitro (left: 0.75 µM MinD, 2 µM His-MinE; right: 1 µM MinD, 1.5 µM MinE-His). d Schematic of the MinDE self-organization mechanism and the synthetic membrane-anchored cargo consisting of a DNA origami nanostructure and streptavidin building blocks. The DNA origami nanostructure has 7 dyes at the upper facet and 42 addressable sites for incorporation of biotinylated oligonucleotides at the lower facet which in turn bind to lipid-anchored streptavidin on the SLB. Fueled by ATP hydrolysis, MinDE attach and detach to and from the membrane in a concerted manner. e The contrast of the patterns resulting from DNA origami transport by MinDE increases with increasing number of incorporated streptavidin per cargo, as does the size of the MinD minima. Representative time series and line plots for origami nanostructures equipped with 2 or 42 streptavidin building blocks (1 μM MinD (30% EGFP-MinD), 1.5 μM MinE-His in presence of 0.1 nM origami-Cy5 with 2 or 42 biotinylated oligonucleotides, streptavidin). f MinDE self-organization induces sorting of two cargo species with distinct membrane footprint. Representative images and line plots for simultaneous transport of cargo-2 and cargo-42 by MinDE (1 μM MinD (30% EGFP-MinD), 1.5 μM MinE-His, 50 pM origami-Cy3b with two biotinylated oligonucleotides, and 50 pM origami-Cy5 with 42 biotinylated oligonucleotides, non-labeled streptavidin). g Schematic of the diffusiophoretic mechanism underlying the molecular transport by MinDE. MinDE reactions and diffusion generate MinDE patterns and density gradients. The diffusive fluxes of MinD exert a frictional force, fc, on the cargo molecules that depends on the effective size of the cargo molecules. Scale bars a, 400 nm, b, 10 μm, c–f, 50 μm; a, adapted from [58], b, adapted with permission from [59]. Copyright 2018 American Chemical Society. d–g, adapted from [89] under a CC BY 4.0 license

Full size image

Biological spatiotemporal organization is achieved by self-organizing protein systems that are capable of pattern formation and large-scale molecular transport through energy dissipation. Much emphasis has been laid on understanding and harnessing complex eukaryotic protein systems based on cytoskeletal and translational motor proteins (Fig. 1c) [64,65,66,67,68,69]. In contrast, bacterial systems have largely been neglected even though they are often simpler both, on a mechanistic and compositional level [70], making them ideal targets for nanotechnological applications. One such example comes from the bacterium Escherichia coli—the Min system. At its core, it consists of two proteins only, the ATPase MinD and the ATPase activating protein MinE, ATP as an energy source, and a lipid membrane as a reaction platform. The two proteins interact with each other and the lipid membrane to form patterns via a reaction-diffusion mechanism. This compositional simplicity paired with its rich dynamics made the system a paradigm for the study of biological pattern formation [71]. Over 30 years of research in vivo, in vitro, and in silico allowed us to obtain a rather detailed understanding of the underlying molecular mechanism [72,73,74,75,76,77,78,79]. In short, MinD binds to the membrane cooperatively upon ATP-induced dimerization, presumably involving the formation of higher order structures. MinE binds to membrane-bound MinD stimulating its ATPase activity, which leads to protein detachment from the membrane (Fig. 2d). In E. coli, MinDE performs pole-to-pole oscillations which generate a time-averaged gradient of the passenger protein MinC with a minimum at midcell. As MinC is an inhibitor of the main divisome protein FtsZ, this time-averaged protein gradient restricts the assembly of the cell division machinery to the middle of the cell. In 2008, MinDE dynamics were reconstituted in vitro: The purified proteins MinD and MinE supplied with ATP as an energy source self-organized on supported lipid bilayers in aqueous buffer forming traveling surface waves [77].

Depending on the specific reaction conditions, the system has been shown to exhibit a variety of patterns in vitro, such as traveling surface waves, dot and labyrinthine patterns, and also oscillatory behavior (Fig. 2c) [12, 77, 80, 81]. Indeed, while the reaction system is truly nonlinear, i.e., small changes in the reaction conditions can lead to drastic changes in the reaction outcome; considerable effort has been invested in determining the parameters influencing pattern formation in order to achieve control over it. Protein concentration and ratios [77, 81], ionic strength of the buffer [82], molecular crowding [7, 14, 83], membrane charge and fluidity [12, 82, 84], and in particular the reaction space geometry, including the form and size of the membrane as well as the surface to volume ratio, heavily influence the type of obtained patterns. For example, when MinDE is confined in membrane-coated PDMS microcompartments with an elongated, cell-like shape, they recapitulate the oscillatory behavior that occurs in the bacterial cells in vivo [12, 14, 78]. On planar rectangular membrane patches, in turn, MinDE traveling surface waves align along the longest axis [7]. Beyond modifications to the reaction surface, engineering of both, MinD and MinE, has revealed that modifications of key molecular motifs, such as the membrane binding and dimerization helices, can change the type of MinDE patterns, such that robust standing waves or MinDE traveling waves with wavelength on a millimeter-scale can be generated [85, 86]. These examples and many others found in the literature [71] clearly demonstrate that the Min system, although compositionally simple, is a striking pattern-forming system whose behavior can be tightly controlled using a varied set of parameters.

Here, we first review how DNA nanotechnology has been previously harnessed to elucidate physical mechanisms governing complex biological phenomena, i.e., the transport of molecular cargo by MinDE dynamics. Building on this work, we then investigate how MinDE-dependent transport of DNA origami nanostructures can be used to generate stable, biologically compatible patterns at the micron to millimeter-scale. We further speculate how combining the rich dynamics of the self-organizing MinDE system with the nanometer-precision addressability of DNA origami nanostructures opens up new possibilities in creating the next generation of hybrid materials with multiscale molecular patterning capabilities.

2 DNA Origami as a Tool to Elucidate Molecular Mechanisms

Recently, we and others have shown that the Min system can not only form mesmerizing patterns on membranes itself, but also induce patterns of other functionally unrelated proteins [87, 88]. These studies showed that MinDE-induced regulation of lipid-bound proteins that have a long membrane dwell-time results in their net transport. This suggested that the non-specific interaction of MinDE with these proteins modulates their diffusion on the membrane, an effect that should depend on the properties of the transported, membrane-bound molecules, such as the area they occupy on the membrane, i.e., the membrane footprint, or their diffusion on the membrane. In order to specifically probe distinct hypotheses regarding the underlying mechanism, it was desirable to modify specific cargo properties over a wide range, while keeping all other parameters comparable. While the properties of membrane-attached proteins cannot easily be tuned in a defined fashion, DNA origami nanostructures are fully programmable, as illustrated across this book. Importantly, binding and diffusion of DNA origami nanostructures on model membranes have been extensively characterized in detail, as described above, rendering them an ideal cargo for the detailed study of MinDE-dependent cargo transport [89].

The DNA origami nanostructure employed was a 20-helix bundle with 42 addressable positions at the bottom facet that can be modified with cholesteryl or biotinyl moieties for membrane anchoring (110 × 16 × 8 nm) [58] (Fig. 2a, d). While the origami structures with cholesteryl oligonucleotides directly bound to the membrane, the ones with biotinylated oligonucleotides could be decorated with streptavidin which in turn bound to biotinylated lipids on the SLB. The former design allowed us to vary the diffusion of the structures while maintaining similar membrane footprints; the latter was used to vary the membrane footprint and thus, the effective size of the molecule via the amount of incorporated streptavidin molecules. For visualization by fluorescence microscopy, the DNA origami nanostructures were functionalized with seven dyes on the upper facet. In order to test the redistribution of these structures in a controlled manner, we used conditions under which MinDE formed quasi-stationary patterns, i.e., after an initial chaotic self-organization phase, labyrinthine patterns emerge, whose macroscopic appearance is stable over long periods of time, but which are nevertheless maintained by continuous binding and unbinding of the individual proteins. We found that MinDE was also capable of redistributing these rather large structures (compared to the 5 nm-sized MinDE proteins [90] and the previously investigated membrane-bound proteins [87, 88]) leading to an anti-correlated pattern of the DNA origami on the membrane. Comparing the final quasi-stationary patterns for structures bearing different kinds and numbers of membrane anchors, we were able to show that the extent of the molecular transport by MinDE increased with the membrane footprint of the cargo molecule, i.e., the effective size of the cargo molecule on the membrane surface (Fig. 2e), but did not directly correlate with cargo diffusion. In contrast, in experiments with conditions under which MinDE formed traveling surface waves, we found that both, the cargo’s effective size and its diffusion coefficient, play a role in its transport efficiency. Cargo molecules that are too slow in comparison to the wave propagation will not be transported effectively. Even more intriguingly, we showed that MinDE self-organization was capable of spatially sorting DNA origami nanostructures with distinct effective sizes, i.e., one with 2 and one with 42 streptavidin molecules bound to its bottom surface (Fig. 2f).

Our experiments allowed us to rule out several possible physical mechanisms arriving at diffusiophoresis as the simplest one that could fully explain our observations (Fig. 2g). Diffusiophoresis generally describes the transport of particles in fluids along concentration gradients of small solutes and has mostly been reported in a non-biological context such as colloid transport [91, 92]. During MinDE self-organization, density gradients of the proteins on the membrane are established which lead to diffusive protein fluxes on the membrane toward low protein densities. Due to the high density of proteins on the membrane, MinDE proteins directly interact with other molecules on the membrane, generating a frictional force on these “cargo” molecules. As a result, the diffusive fluxes of MinDE and cargo couple on the membrane, leading to their transport toward, and subsequent accumulation in, areas of low MinDE density. As the frictional force increases with the effective size of the molecule, molecules with a larger membrane footprint experience a stronger redistribution than smaller ones. Although this first description of diffusiophoresis by a biological self-organizing system was achieved in vitro, diffusiophoresis could be more widespread in cellular systems, but might be masked by the stronger, specific molecular interactions within the cell.

3 Stable DNA Origami Patterns on Lipid Membranes

Harnessing the established experimental framework, we set out to explore the possibilities arising from the ability to pattern DNA origami nanostructures by MinDE self-organization. In this manuscript, we started by testing whether the resulting patterns could be stabilized after they had been established. DNA origami distribution out of equilibrium is maintained by the energy dissipation upon MinDE self-organization and is as such reversible: DNA origami patterns respond to changes in MinDE patterns induced, for example, by the addition of more MinE [89] and DNA origami patterns disappear altogether by thermal mixing when MinDE activity subsides [87, 88]. This reversibility may become inconvenient for any downstream applications, considering the dynamic nature of lipid membranes. Thus, it would be desirable to “freeze” MinDE-induced cargo patterns, e.g., through crosslinking. Previous work has described two main strategies for hierarchical self-assembly of large DNA-based structures (for a more detailed review, refer to [93]): “sticky” interactions [94,95,96], based on DNA base-pairing and “stacking” interactions [97,98,99], based on blunt-end interactions and shape complementarity of building blocks. While the first one can be accomplished by programming complementary single-stranded DNA extensions between building blocks or by addition of a single-stranded DNA oligonucleotide complementary to single-stranded portions on the DNA structures, the second one can be controlled by the concentration of Mg2+ in solution. As MinDE self-organization is sensitive to the ionic strength of the buffer [82], we here opted to modify our DNA origami nanostructures to contain 8 nucleotide long poly-A ssDNA extensions on each end. The addition of a complementary 14 nucleotide long poly-T polymerization staple would result in the crosslinking of DNA nanostructures and their polymerization in an end-to-end fashion.

First, we verified the behavior of a priori polymerized DNA origami nanostructures upon MinDE self-organization (Fig. 3a). A homogeneous, membrane-bound layer of DNA origami was crosslinked before MinDE self-organization was started with ATP. Upon ATP addition, MinDE self-organized into labyrinthian patterns, as did the excess membrane-bound streptavidin. However, polymerized DNA origami did not get patterned, remaining homogeneously distributed. This shows that the DNA origami nanostructures indeed formed large stable polymers on the membrane whose diffusional behavior was virtually frozen and as such could not be transported by MinDE, as the diffusiophoretic transport requires diffusive mobility of both the cargo and MinDE proteins.

Fig. 3
5 illustrations and 50 micrographs present origami patterns of D N A. Illustrations depict polymerization staple with add Min D E, polymerizable structure, Min D E induced D N A origami pattern, and regular structure. Micrographs present patterns of cross-linked Min D, origami, streptavidin, and actin.

DNA origami patterns induced by MinDE self-organization can be “frozen” on lipid membranes. a Pre-polymerized DNA origami cannot be reorganized by the MinDE system. Representative time series and kymographs of pre-polymerized DNA origami and MinD channels after initiating the start of MinDE self-organization by addition of ATP (1 µM MinD, 1.5 µM MinE-His, 0.1 nM cargo-2 pre-polymerized by addition of 17 µM polymerization staple, Alexa568-streptavidin). b MinDE-induced DNA origami patterns are stabilized by DNA origami crosslinking upon addition of polymerization staple, c but not if complementary overhangs are absent in the DNA nanostructure. Representative images and kymographs of DNA origami, MinD and streptavidin patterns after addition of polymerization staple and subsequent addition of MinE for a structure with extensions that can be polymerized (b, left) and a regular structure (c, right) (initial conditions: 1 µM MinD, 1.5 µM MinE-His, 0.1 nM (polymerizable) cargo-2, Alexa568-streptavidin; addition of 17 µM polymerization staple and subsequently 1.5 µM MinE-His). d MinDE induces patterns of streptavidin-decorated origami (1 µM MinD (30% EGFP-MinD), 1.5 µM MinE-His, 0.1 nM cargo-chol-15 with biotin extension on upper facet, Alexa568-streptavidin), e which can be crosslinked upon addition of biotinylated stabilized actin filaments, which preferentially localize to the regions enriched in streptavidin-decorated origami (1 µM MinD, 1.5 µM MinE-His, 0.1 nM cargo-chol-15 with biotin extension, Alexa568-streptavidin, Alexa-647-Phalloidin-labeled, biotinylated actin). Scale bars a–d, 50 µm; e, 10 µm

Full size image

In contrast, when we started MinDE self-organization in presence of these DNA origami structures prior to the addition of polymerization staples, MinDE dynamics induced anti-correlated DNA origami patterns similar to those observed for non-modified DNA origami nanostructures (Fig. 3b). After MinDE patterns entered the quasi-stationary phase, we added the polymerization staple to “freeze” the patterns in place. Indeed, when we added more MinE, only the MinDE patterns changed, whereas the DNA origami retained their original pattern. Intriguingly, we could observe the uncorrelated patterning of two cargo molecules: While the crosslinked DNA origami retained the original pattern, the excess membrane-anchored protein streptavidin, here used to bind DNA origami to the membrane, was transported by the reorganized MinDE patterns. In the control experiment, in which the polymerization staple was added to a DNA origami structure that cannot be crosslinked (i.e., without single-stranded overhangs), the addition of MinE resulted in concomitant and anti-correlated pattern reorganization of MinDE and DNA origami (Fig. 3c).

Having shown that MinDE-induced DNA origami patterns could be spatially stabilized, we next asked whether other molecules could be transported or targeted using DNA origami as a shuttle or scaffold. Over the years, many studies have shown that molecules can be targeted to DNA origami nanostructures with nanometer precision via a wide variety of different chemistries [100]. As a proof of concept, we used a DNA origami structure that contains 15 cholesteryl moieties for membrane binding and which is strongly redistributed by MinDE self-organization [89]. We replaced the fluorescent dye staples on the upper facet of these structures with biotinyl moieties which allowed the coupling of fluorescently labeled streptavidin to the upper facet of the DNA origami (Fig. 3d). Indeed, MinDE was also capable of transporting this streptavidin-loaded DNA origami, resulting in high-contrast, anti-correlated patterns (Fig. 3e). When adding biotinylated stabilized actin filaments to such pre-patterned streptavidin-bearing DNA origami structures, actin preferentially bound to the regions enriched in DNA origami, i.e., the MinDE minima. Moreover, the presence of multiple binding sites per filament resulted in pattern crosslinking.

Taken together, the new results presented herein are a proof of concept of how DNA origami, when combined with the MinDE system, can be used to create stable and spatially uncorrelated patterns of distinct molecules on a dynamic support, such as a lipid membrane.

4 Challenges and Opportunities

Since the birth of DNA nanotechnology, the biophysical community has been exploiting the well understood behavior of DNA and its versatility regarding conjugation/modification [100] to interrogate a number of fundamental biological processes [101]. From single molecule methods to cellular studies, researchers used DNA origami to investigate enzymatic cascades [102,103,104], motor proteins [105, 106], DNA binding proteins [107,108,109], and even receptor-mediated cellular responses, e.g., apoptosis, phagocytosis, B cell activation, T cell stimulation [110,111,112,113,114], to name only a few. Here, we reviewed how our detailed knowledge and fine control over membrane binding and diffusion of DNA origami nanostructures allowed us to unravel a new unspecific transport mechanism by the self-organizing MinDE system of E. coli [89]. Our recent results reiterate how the power of DNA origami can be used to interrogate complex dynamic phenomena in a controlled fashion and further expand the available in vitro toolkit.

Non-specific transport based on diffusiophoresis should in principle occur in any system that generates and/or maintains concentration gradients. We have shown that (membrane-bound) DNA origami is the ideal tool to characterize such phenomena in detail. For example, similar approaches could be used to explore diffusiopheritc transport or related effects in other self-organizing membrane systems such as small GTPases [115], membrane-bound kinase/phosphatase networks [10], or minimal actin cortices [64,65,66]. Soluble DNA origami structures might even be harnessed to study incorporation and molecular transport that occur during liquid–liquid phase separation of biomolecules, a phenomena that recently gained considerable attention in cell biology [116]. With the routine establishment of uncorrelated patterns, the complexity of such assays can be increased, allowing to explore the interplay between simultaneously occurring dynamics of distinct systems, as observed in cells.

To date, in the quest for biomolecular spatiotemporal patterning and transport, the focus has mostly been laid on using eukaryotic active matter elements on solid supports, which mediate cargo transport via specific interactions, such as the ParMRC system, actin, microtubules, and motor proteins [67,68,69]. Despite not being a “conventional” biological motor system, diffusiophoretic cargo transport by MinDE self-organization has the potential to enable complex biomolecular patterning on membranes with particularly high modularity when coupled to DNA origami, the model tool of molecular nanopatterning [117]. The variety of modification chemistries for membrane binding or molecular targeting, crosslinking and actuation strategies of DNA origami, paired with the diversity of possible patterns and various control elements for the Min system, as well as the ease of purification of the MinDE proteins (small proteins as compared to eukaryotic molecular motors) and the non-specific nature of the MinDE-mediated transport that overcomes the need for additional adapter proteins, are all factors that contribute to making this an attractive combination for a variety of applications (Fig. 4a).

Fig. 4
4 illustrations with 17 micrographs and 2 line graphs. A. toolbox for self-organizing protein systems. Micrographs of pattern types, scales, reaction space geometry, and light control. B. patterned S L B in chromium. 10 micrographs of Min D and origami. C. biotinyl CAP, P E with micrographs of m s f G F P and streptavidin with swirl patterns. D. air reaction mix with 10 micrographs for Min D and streptavidin. A line graph with 2 fluctuating lines. E, a fluctuating line graph with 3 lines.

Combination of DNA origami nanotechnology with self-organizing protein systems opens up new possibilities for molecular transport and patterning. a Schematic highlighting the variety of control elements available for self-organizing protein systems such as MinDE and DNA origami nanostructure toolboxes. b MinDE accumulates DNA origami structures along the long axis of patterned lipid membranes. Representative time series of MinDE traveling surface waves transporting DNA origami along the wave vector on chromium-patterned SLBs. c Large-scale gradients of cargo molecules established by a minimal MinDE system (indicated by white arrows). Representative images of MinE(1-31)-msfGFP and MinD self-organization in the presence of membrane-bound streptavidin (2.5 µM MinD (20% Alexa647-MinD), 200 nM MinE(1-31)-msfGFP, Alexa568-streptavidin), d MinDE oscillations in cell-shaped compartments result in preferential localization of the cargo molecules toward the center of the compartment. Representative time-lapse images and time-averaged fluorescence intensity profile of MinDE oscillations and streptavidin counter-oscillations in PDMS microcompartments (1 µM MinD, 2 µM MinE, streptavidin-Alexa647). e Schematic highlighting the potential of generating two distinct time-averaged gradients of DNA origami in cell-shaped compartments simultaneously. b was adapted from [89] and d from [87] under a CC BY 4.0 license. Scale bars b, 25 µm, c, 1 mm, d, 10 µm

Full size image

For example, by combining the ability to direct MinDE traveling waves on planar patterned bilayers [7] with DNA origami, reproducible directed molecular transport can be achieved, resulting in a gradient of the cargo of interest along the longest axis of a membrane patch [89] (Fig. 4b). On the other hand, we show here that MinDE-dependent net transport of molecules can occur over several millimeters and establish millimeter-sized gradients on the membrane (Fig. 4c), when MinE mutant peptides are employed [85]. As such, arbitrary cargo tethered to a DNA origami shuttle could be transported directionally over several millimeters and could even be controlled by light [33]. By harnessing the current technologies of microfabrication and membrane-coating of surfaces (see Introduction for more detail), exciting possibilities arise for macromolecular “writing” at a scale visible to the naked eye, resulting in a new generation of biohybrid devices. Hence, directed and controlled diffusiophoretic transport of DNA origami structures by MinDE could potentially be explored for molecular patterning, molecular delivery, biocomputation, or molecular separation based on membrane footprint.

Besides its potential application in nanotechnology, DNA origami transport by MinDE also offers new possibilities to achieve spatiotemporal organization in synthetic cells. We have shown in the past that MinDE self-organization can spatiotemporally position molecules via diffusiophoresis to the center of cell-shaped compartments [87, 88] (Fig. 4d). Using DNA origami structures with distinct membrane footprints, it should now be possible to target distinct molecules to at least three different regions of such a cell-like compartment (Fig. 4e). Furthermore, the concentration contrast between these regions can be controlled by adjusting the differences between DNA origami footprints. Combined with the recent strides that have been made to encapsulate a functional Min system into 3D compartments, such as monolayer sealed microfabricated compartments [15], water-in-oil droplets [118] or deformable giant unilamellar vesicles (GUV) [119, 120], DNA origami transport by MinDE could then be exploited for basic spatial compartmentalization of synthetic cells, e.g., to tether and segregate genetic material or compartmentalize the activity of molecular machines. Using functional DNA origami nanostructures that can be actuated by a number of mechanisms, e.g., strand displacement or insertion, pH, temperature, electric field, or light [121,122,123,124,125,126,127], even a functional divisome could potentially be generated, a long-standing goal of the field.

In conclusion, the versatility of DNA nanotechnology and the rich dynamics of self-organizing proteins such as the MinDE system can be harnessed together to transport, deliver, pattern, and sort functionalized cargo on membranes, thereby creating a new class of hybrid biomaterials for applications in nanotechnology and the bottom-up construction of synthetic cells.

5 Materials and Methods

Most of the experimental methods and materials for this chapter have been described in detail before [58, 89, 128], but are described in brief below, highlighting modifications and new methods.

MinDE plasmids and proteins—the plasmids pET28a-His-MinD_MinE and pET28a-His-MinE [77], pET28a-His-EGFP-MinD [11], pET28a-MinE-His [11, 77, 80, 81] and pET28a-MinE(1-31)-msfGFP [85] were used for purification of His-MinD, His-EGFP-MinD, His-MinE, MinE-His, and MinE(1–31)-msfGFP, respectively, as described in detail before [128].

DNA origami nanostructures—the modifications and the functionalization of the previously designed elongated DNA origami nanostructure [58] have been described in detail in [89]. For polymerizable DNA origami structures, end staples with 8A extensions are added upon DNA origami folding. Biotin-top-labeled DNA origami structures contained 7 modified oligonucleotides attached to extended staples on the upper facet. The assembly of the origami structure was performed in a one-pot reaction mix as described previously [58].

Self-organization assay on SLBs—SLBs were prepared as described before with a lipid composition of 30 mol% DOPG, 69 or 70% DOPC, and 1% CAP-Biotinyl-PEG in case of DNA origami with biotinyl moieties [128]. Binding of DNA origami to lipid membranes via streptavidin interactions or cholesteryl anchors has been previously described [89]. Self-organization assays were performed essentially as detailed in [128]. For large-scale MinDE traveling waves a MinE peptide [85], 200 nM MinE(1-31)-msfGFP, and 2.5 µM MinD (20% Alexa647-MinD) were used in the self-organization assay that was performed in sticky-Slide VI 0.4 chambers (ibidi GmbH, Gräfelfing, Germany).

Crosslinking of DNA origami by strand hybridization—to trigger origami crosslinking, polymerization staple (14 T) was added to the chamber at a final concentration of 17 µM, and the reaction mixture was mixed by pipetting. The large volume addition (40 µl) induced changes to MinDE patterns in some experiments. To further change the MinDE pattern, 1.5 µM MinE-His was added to the reaction mixture.

Attachment of actin to DNA origami—for attachment of labeled streptavidin to DNA origami, cholesteryl-modified DNA nanostructures were first bound to the membrane. Subsequently, the chamber was incubated with Alexa568-labeled streptavidin (ThermoFisher Scientific) at a final concentration of 1 µg/ml. After 5–10 min incubation, unbound streptavidin was removed by gently washing 3 times with a total volume of 600 µl reaction buffer. MinDE proteins were added to the reaction mixture, and the self-organization assay was started by addition of ATP. After pattern establishment, Alexa-647-Phalloidin-labeled biotinylated actin (produced as described in [65]) was added at a final concentration of 0.1 µM and allowed to bind to the DNA origami before image acquisition.