Functional DNA-based cytoskeletons for synthetic cells

The cytoskeleton is an essential component of a cell. It controls the cell shape, establishes the internal organization, and performs vital biological functions. Building synthetic cytoskeletons that mimic key features of their natural counterparts delineates a crucial step towards synthetic cells assembled from the bottom up. To this end, DNA nanotechnology represents one of the most promising routes, given the inherent sequence specificity, addressability and programmability of DNA. Here we demonstrate functional DNA-based cytoskeletons operating in microfluidic cell-sized compartments. The synthetic cytoskeletons consist of DNA tiles self-assembled into filament networks. These filaments can be rationally designed and controlled to imitate features of natural cytoskeletons, including reversible assembly and ATP-triggered polymerization, and we also explore their potential for guided vesicle transport in cell-sized confinement. Also, they possess engineerable characteristics, including assembly and disassembly powered by DNA hybridization or aptamer–target interactions and autonomous transport of gold nanoparticles. This work underpins DNA nanotechnology as a key player in building synthetic cells.

Supplementary Note S1: Estimation of the SUV velocity along the

DNA filaments
From the quantitative measurements of the porosity over time (see Fig. 4f, main text), we can provide a rough estimate for the transport velocity of the individual SUVs. For this purpose we need to first estimate 1) the total length of the DNA filaments inside a droplet and 2) the total number of the SUVs inside a droplet.
1) Each DNA tile has an approximate length of l t = 14 nm, and the tubular filaments have a circumference of approximately 6 tiles. [1] Assuming that all tiles are polymerized into filaments, we can calculate the maximum total length of the DNA filaments per liter to be with c t being the DNA tile concentration (250 nM) and N A the Avogadro number.
With an average droplet diameter of 40 µm in this experiment and hence a droplet volume of approximately 40 pl, the total filament length per droplet amounts to ≈ 14 mm.
2) We can calculate the SUV concentration from the total lipid concentration. We first compute the amount of lipids per SUV and from this the total number of SUVs per volume.
The SUV diameter has been measured to be 65 nm. This yields where the area of each lipid is assumed to be 0.7 nm 2 and the lipid bilayer is composed of two layers of lipids. Given a total lipid concentration of 10 µM as used in the experiment, this leads to This equals an SUV concentration of approximately 250 pM. With a droplet volume of 40 pl, we estimate that a total of 6400 SUVs have been encapsulated per droplet.
From Figure 4, we find that the filament fluorescence decreases with a half life of t 1/2 = 25 min for 250 pM SUVs. This means that 6400 SUVs roll on average over half of the total filament length in a droplet, hence 7 mm, during this time. A single SUV thus covers a distance of approximately 1 µm in 25 minutes, or v SUV ≈44 nm/min. Note that photobleaching only accounts for a 2 % decrease of the porosity (see Fig. S19a). This would result in a negligible increase of the calculated transport velocities, but its minor contribution has been disregarded here. This can be transformed into rotations of single SUVs via All in all, this yields transport velocities for the different conditions:   Figure S1: Strand routing diagrams for the DNA tiles Figure 1: Strand routing diagrams for the DNA tiles. a DNA tile for the toehold strand displacement reaction in Fig. 2 according to Green et al. [2] b DNA tile for the ATP splitaptameric reaction in Figs. 3a and b. The ATP aptamer sequence was taken from Huizenga et al. [3] c DNA tile for the dual-aptamer reaction in Figs. 3d and e. d DNA tile for the SUV and gold nanoparticle transport reaction in Fig. 4. Overhang sequences for the rolling were adapted from Yehl et al. [4] ferent tile concentrations  Addition of 20 µM anti-invader strand is not sufficient to trigger the filament assembly, whereas 37.5 µM can trigger the assembly. Images were taken approximately 10 minutes after droplet production. From these results, the higher concentrations have been chosen for the respective experiments.  origami seed Figure 11: Strand routing diagram for the DNA origami seed to implement a seeded growth mechanism for the DNA filaments. [6] a Layout of the DNA origami seed with the DNA overhangs for the attachment of a gold nanoparticle (black box) and the DNA tiles (red box). b Design and sequences of the tiles, which are attached to the overhangs on the DNA origami seed for the directional growth of the filament. All DNA sequences are provided in Supplementary Dataset 1, adapted from Mohammed et al. [6] Figure 12: a Mechanism for the seeded growth of the DNA filaments. [6] The M13 scaffold is hybridized with specifically designed and custom-synthesized staple strands to form a hollow seed tube.    Supplementary Figure S18: DNA filaments remain intact after transport Figure 18: TEM images of the DNA-based filaments after apparent cargo transport of a SUVs and b gold nanoparticles. The filaments have been released from the droplets after cargo transport by breaking them up with addition of perfluoro-1-octanol (PFO) destabilizing agent. [7] The filaments remain intact after transport, confirming that the decrease in porosity ((1 − Φ) · 100%) is not due to filament disassembly. The fluorophores have been cleaved from the filaments, while leaving the filaments intact. Furthermore, the cargos (SUVs or gold nanoparticles) have detached from the filaments after cargo transport. Scale bar: 200 nm.
Supplementary Figure  To demonstrate whether the SUVs roll, hop or glide along the DNA filaments, the free DNA on the SUVs is blocked by hybridization with blocking DNA strands (purple). The same experiment has been performed for gold nanoparticles. b Porosity ((1-Φ)·100%) measurements in the presence of the blocking DNA strands for gold nanoparticles (blue) and SUVs (red), respectively. The measured porosity remains nearly constant in each case. This indicates that the cargo transport along the filaments will halt, when the rolling motion is inhibited. Figure 23: Quantification of the DNA density on gold nanoparticles. a Illustration of the experimental procedure used for the quantification of the DNA density on gold nanoparticles. [8] b UV-Vis spectra of the released DNA (mean and standard deviation from n=5 independent experiments). DNA coverage of 126±10 strands per gold nanoparticle or a density of 0.10±0.01 strands per nm 2 is obtained.
Supplementary Figure S24: DNA density on SUVs Figure 24: Quantification of the DNA density on SUVs. a Quantification of the loss of lipids after SUV extrusion. The fluorescence intensity of the lipid mixture (99% DOPC, 1% Atto633-DOPE) is determined before and after extrusion. 9.07% of the lipids are lost during the extrusion process (mean and standard deviation from n=4 independent measurements). b Incorporation efficiency of single-stranded cholesterol-tagged DNA into SUVs. The concentration of DNA is determined from UV absorbance measurements. The reference measurement is taken before addition to SUVs ("before purification"). Afterwards, the cholesterol-tagged DNA is incubated in excess with SUVs. The DNA concentration in the supernatant is measured ("after purification"). It corresponds to the unbound fraction of DNA (mean and standard deviation from n=3 independent experiments). 27.8±2.2% of 2 µM cholesterol-tagged DNA binds to SUVs (10 µM lipids before extrusion). This gives rise to a DNA density on the SUVs of 0.18±0.01 DNA strands per nm 2 .

Supplementary Videos
Supplementary Video S1: Dynamics of the toehold-modified DNA filaments inside water-in-oil droplets Confocal timelapse of 500 nM Cy3-labelled toehold-modified DNA filaments (λ ex = 561 nm) in a 1× TAE buffer containing 12 mM MgCl 2 . After polymerization, the DNA filaments remain highly dynamic with constant rearrangement and remodelling. The video corresponds to the filaments shown in Fig. 2c. Scale bar: 50 µm.