Design of RNA origami
The design of the RNA tiles was guided by the rules established by Geary et al. for ssRNA origami24.
The 3H-4DT-iSpinach blueprint was generated by including the iSpinach module to the three′ finish of the 3H-4DT design from Geary et al.23. The iSpinach aptamer was linked to the core design by a 2 uracil linker.
The blueprints for the nanotube-forming tiles have been designed based mostly on the 3H-3DT tile from Geary et al.23. The primary and third helices have been prolonged in order that the helical flip of the exterior 180° kissing loops would align upon meeting. This extension course of was executed manually by way of cycles of extension and visualization utilizing the RNAbuild script from ROAD23 and ChimeraX51 to attenuate the space between kissing loops upon closure of the tube. Throughout this course of, the size of the helices have been additionally adjusted in order that the interior kissing loops would correspond to eight base pair duplexes as an alternative of the unique 9 within the 3H-3DT design, reflecting the up to date understanding of inside kissing loop geometry based mostly on cryoEM buildings52.
The blueprint of the iSpi design was then put into Revolvr23. The ultimate sequence was chosen utilizing two standards: each exterior kissing loop pairs ought to have binding power beneath −8 kcal mol−1, and the distinction in binding power between the 2 pairs must be minimal. The WT design was created by eradicating the iSpi overhang. The dsOV was launched into the tile blueprint with out additional sequence optimization. All blueprints have been used as enter for the trace_analysis script from ROAD23 to test for folding irregularities after which used for oxRNA simulations.
The blueprints of the mutated ring-forming tiles have been generated as described, and the sequence optimization was carried out in the identical method described for the opposite tiles.
The RNA origami containing biotin aptamer was designed by concatenating the WT, biotin aptamer and iSpinach aptamer within the following sequence: WT tile-AAA-biotin aptamer-AAAA-iSpinach aptamer (Supplementary Fig. 29). The sequence was then checked utilizing the RNAfold net server53 to make sure appropriate folding of every aptamer within the concatenated sequence.
Synthesis of RNA origami in bulk
DNA templates have been synthesized as double-stranded gBlocks from Built-in DNA Applied sciences (IDT). DNA gBlocks (0.5 ng μl−1) have been PCR-amplified utilizing 14–25-nt primers (Supplementary Knowledge 1), utilizing a Phusion Excessive-Constancy PCR Equipment (NEB) with annealing temperature at 62 °C. The PCR product was then purified utilizing a Qiagen PCR purification equipment. RNA was transcribed and co-transcriptionally folded in a one-pot response containing PCR purified DNA template (4 ng μl−1), Mg(OAc)2 (6 mM), NaOAc (40 mM), KCl (40 mM), Tris-OAc (50 mM, pH 7.8), rNTPs (1 mM every), dithiothreitol (DTT) (1 mM) and DFHBI-1T dye (62.5 μM) and RNase inhibitor (1 U μl−1), if not described in any other case. Reactions have been initiated by including T7 RNA polymerase (0.2 U μl−1). Transcription reactions have been carried out in 100 μl volumes at 37 °C for two–12 h relying on the experimental setup. All of the experiments have been carried out utilizing nuclease-free water.
Fluorescence assay in bulk
3H-4DT RNA origami with an iSpinach fluorophore was ready with PCR purified DNA template (4 ng μl−1), NaOAc (40 mM), KCl (40 mM), Tris-OAc (50 mM, pH 7.8), rNTPs (1 mM every), DTT (1 mM) and DFHBI-1T dye (62.5 μM) in 4 totally different situations of Mg(OAc)2 (that’s, 0 mM, 1 mM, 1–6 mM and 6 mM). Within the 1–6 mM pattern, first, 1 mM Mg(OAc)2 was added for two h after which as much as 5 mM of Mg(OAc)2 added for the subsequent 2 h. To measure the fluorescence depth, 100 μl ultimate quantity was pipetted right into a Greiner 96-well black backside plate (Sigma-Aldrich), which was lastly positioned contained in the plate reader (Spark multimode plate reader from Tecan Life Sciences) for 4 h at 37 °C. RNA manufacturing was quantified by measuring the emission of DFHBI-1T upon excitation with 488 nm (DFHBI-1T dye, λex = 482 nm, λem = 505 nm) with a achieve of 100 (manually set).
GUV preparation
GUVs have been ready by the polyvinyl alcohol (PVA) gel-assisted swelling technique27,29. Intimately, a PVA answer was ready by mixing 5% (w/v) PVA (MW 145,000 Da) in nuclease-free water with sucrose (100 mM) for twenty-four h, at 400 rpm, at 90 °C. The PVA answer (50 μl) was dried as a skinny movie on a glass slide (60 mm × 24 mm) at 50 °C for 30 min. Then, 5 μl of a lipid combination in chloroform containing 10 mol% DOPG (10 μg μl−1) and 1 mol% DiD dye in 10 μg μl−1 DOPC was unfold onto the PVA layer and dried for 1 h at 37 °C. The DiD dye inventory answer (10 μg μl−1) was ready in chloroform. Utilizing a Teflon chamber (roughly 40 mm × 24 mm) as a spacer and a second glass slide, a chamber was assembled on high of the slide with the lipid-coated PVA. Then, the lipids have been hydrated with 1 ml of 100 mM sucrose containing all of the transcription parts to be loaded into the GUV relying on the experiment for 1 h at room temperature, permitting for GUV formation. After that, the chamber was inverted for five min and gently tapped twice utilizing a pipette tip, and the GUVs have been collected right into a 1.5 ml Eppendorf tube and left to accept 30 min. For the washing, 350 μl of GUVs was taken from the underside and added to 1 ml of 150 mM glucose buffer. The GUVs have been allowed to settle in a single day at 4 °C. The subsequent day morning, the highest 1 ml of the buffer was eliminated with out disturbing the underside layer. The GUVs have been washed second time with 0.5 ml of 150 mM glucose buffer, and the GUVs have been solely allowed to accept 2–3 h at 4 °C. For subsequent experiments, the GUV answer was sourced from the underside, the place there was the next chance of acquiring GUVs with successfully encapsulated DNA templates and RNA polymerase, attributable to their higher density. All experiments have been carried out utilizing nuclease-free water. For imaging, an 18-well Ibidi glass backside chamber was pre-incubated with 3% BSA answer and washed with nuclease-free water twice earlier than GUV addition.
Expression of RNA origami triggered by Mg2+ in GUVs
GUVs containing a combination of the DNA template (4 ng μl−1), RNA polymerase (0.2 U μl−1), every rNTP (1 mM), Mg(OAc)2 (1 mM), NaOAc (40 mM), KCl (40 mM), Tris-OAc (50 mM, pH 7.8), DTT (1 mM), DFHBI-1T dye (62.5 μM) and sucrose (150 mM) have been fashioned as described. Subsequently, a particular washing protocol was used, whereby 350 μl of GUVs was rinsed as soon as with 1 ml of 570 mM glucose buffer, maintained at room temperature for two–3 h. On the identical day, 100 μl of the washed GUV answer was transferred to an 18-well imaging chamber and supplemented with 10 µM magnesium-ionophore I (Merck). As soon as the pattern was positioned on the confocal microscope with an incubation chamber held at 37 °C for two h. Following this incubation, a further 5 mM of Mg(OAc)2 was launched into the GUV answer. The GUVs have been constantly monitored for as much as 4 h.
Expression of RNA origami triggered by rNTPs in GUVs
GUVs containing a combination of the DNA template (4 ng μl−1), RNA polymerase (0.2 U μl−1) and α-haemolysin (15 ng μl−1) have been fashioned as described. On the next day, 80 μl of purified GUVs was supplemented with a feeding answer containing sucrose (100 mM), Mg(OAc)2 (6 mM), NaOAc (40 mM), KCl (40 mM), Tris-OAc (50 mM, pH 7.8), DTT (1 mM) and DFHBI-1T dye (62.5 μM). These parts might enter the GUV lumen by way of the α-haemolysin pores. The GUVs, now containing the mandatory parts for co-transcriptional folding besides rNTPs, have been incubated for two h at 37 °C. They have been then transferred to an 18-well imaging chamber and allowed to settle to the underside of the slide for 20 min earlier than imaging commenced. As soon as the pattern was positioned on the confocal microscope with an incubation chamber held at 37 °C, rNTPs have been added externally to attain a ultimate focus of 1 mM for every nucleotide. The rNTPs additionally translocate into the GUVs by way of α-haemolysin. GUVs have been monitored over a interval of as much as 4 h. Alternatively, to picture the formation of the RNA origami cytoskeleton-like at discrete time factors (0 h, 2 h, 4 h and 6 h), GUVs have been incubated with all above-mentioned parts together with rNTPs in an Eppendorf tube inside a thermal block set to 37 °C earlier than imaging.
oxRNA simulations
Coarse-grained modelling of particular person RNA tiles and the nanotubes was carried out utilizing the oxRNA2 power discipline40,54 with the CUDA-accelerated oxDNA MD simulation engine (model 3.5.2)41,55,56. Constructions have been exported from ROAD schematics to PDB format utilizing the RNAbuild script from ROAD23. The PDB buildings have been then transformed to oxDNA simulation recordsdata utilizing an area copy of TacoxDNA up to date to work with Python 3.11 (ref. 57) (pipeline script in Code availability). Constructions have been relaxed utilizing the protocol detailed in refs. 58,59. In short, a brief Monte Carlo simulation was carried out to take away excluded quantity clashes between nucleotides, adopted by an extended (no less than 5 × 107 steps, dt = 0.003) MD leisure carried out utilizing the ‘Langevin’ thermostat and a modified spine FENE potential to keep away from numerical instabilities owing to excessive forces. Harmonic traps (stiff = 3.1) have been positioned between all paired nucleotides to keep up the construction throughout leisure. Profitable leisure was verified by observing per-nucleotide power values beneath −1.4 and visualization of the meant construction with oxView59,60.
A number of manufacturing equilibrium MD runs have been carried out on NVIDIA A100 GPUs utilizing edge-based calculation parallelization56 for 1 × 109 simulation steps at dt = 0.003. This roughly corresponds to a time step of 9.09 fs and a complete simulation time of 9.09 μs by direct unit conversion. It must be famous, nevertheless, that calculating precise time correspondence in coarse-grained simulations is just not easy. On the idea of a comparability between DNA binding charges in experiments and oxDNA simulations (see Supplementary Info of ref. 61), the time represented by these simulations could also be nearer to 30 ms. The temperature was maintained at 37 °C utilizing an Andersen-like thermostat (thermostat = ‘brownian’ or ‘john’ within the oxDNA parameter file)62. Debye–Huckel electrostatics have been utilized to the system mimicking 0.5 M NaCl54. Configurations have been saved each 5 × 105 steps, leading to trajectories with 2,000 frames every for evaluation. Because the dynamics of accurately folded buildings are of curiosity, we ran a number of simulations and chosen the primary 4 simulations the place the interior kissing loops maintained a minimal of 11 out of designed 12 bonds on common all through the simulation for additional evaluation.
The imply buildings, root-mean-square fluctuations (RMSFs) and bond occupancies have been calculated from simulation trajectories utilizing oxDNA Evaluation Instruments (OAT, model 2.3.5)60. The construction closest to the imply construction was extracted from one replicate for every construction utilizing the centroid perform in OAT, and this construction was used to construct the nanotube simulations utilizing a custom-written Python script (Code availability); we constructed 100-layer (300 tiles, ~90,000 nucleotides) simulations from every single tile. The nanotubes have been then relaxed with externally utilized harmonic traps to make sure that the kissing loop cohesion was efficiently fashioned. As soon as over 90% of goal bonds have been fashioned, the forces have been dropped and triplicate manufacturing simulations with the identical parameters as the only tiles have been carried out.
Persistence size evaluation
All AFM photos have been pre-processed with the open-access software program Gwyddion63 utilizing standardized Align Rows and Levelling instruments after which exported into Tagged Picture File Format (TIFF). The nanotubes have been manually masked utilizing Object Choice Software and different Choice Instruments in Adobe Photoshop 2024 model 25.12.0. The background was then eliminated and exported as a TIFF file. Owing to branching of the nanotubes, the background-removed photos have been then traced utilizing a custom-written Python script (Code availability) that skeletonized the pre-selected nanotubes and traced alongside the longest path in case of branching (see evaluation pipeline and Supplementary Fig. 30 in Supplementary Be aware 1).
The extracted coordinates have been then used as enter for a custom-written Python script (Code availability) to calculate the persistence size of the in vitro-transcribed nanotubes. To scale back detection error owing to low decision in imaging, the info have been filtered to exclude all detection beneath 200 nm. The squared end-to-end distance of every nanotube was plotted towards its contour size. The info have been then match utilizing the theoretical relation
$${langle {R}^{2}rangle }_{3D}=2sPLleft(1-frac{sP}{L}(1-{e}^{-L/sP})proper)$$
(1)
the place R2 is the squared end-to-end distance, L is the contour size, P is the persistence size and s is a floor parameter that’s set to 1 (Supplementary Be aware 1). For every nanotube, the end-to-end distance and the overall contour size of the nanotube have been extracted for becoming. The info have been masked by residuals ±1 s.d. For the simulated buildings, owing to the slender distribution of contour size (because the simulation is just one nanotube over time), one persistence size is calculated for every body. The tip-to-end distance of each layer pair and their contour lengths have been fitted utilizing the identical equation and masking as described above. The info plotted in Fig. 3g are the final 10% of the simulation frames (t ≥ 9 × 108).
Evaluation of RNA origami nanotube progress
Photos of nanotube-producing GUVs at totally different time factors (Fig. 5c) have been analysed in ImageJ2 (ImageJ2 2.14.0/1.54f (ref. 64)) utilizing custom-written ImageJ macro scripts (Code availability). In abstract, for time level t = 0 h, the GUVs have been robotically detected utilizing the Hough Circle Remodel plug-in. For additional time factors, the GUVs that produced RNA nanotubes have been manually chosen. The areas of curiosity (ROIs) within the RNA fluorescence channel have been then filtered and thresholded utilizing the Gaussian Blur, Convert to Masks and Erode features. The realm fraction was then extracted for every GUV utilizing the Measure perform. For time level 0, false detections have been manually eliminated. For different time factors, solely GUVs containing greater than 1% space fraction have been taken under consideration.
Evaluation of RNA cytoskeleton attachment to GUV membranes
Photos of single GUVs that produced RNA cytoskeleton with and with out biotin aptamer have been analysed in ImageJ2 (ImageJ2 2.14.0/1.54f) utilizing custom-written ImageJ macro scripts (Code availability). The GUVs have been robotically detected utilizing the Analyze Particles technique. In brief, the lipid fluorescence channel was processed utilizing the Auto Threshold, Fill Holes, Erode, Gaussian Blur after which Convert to Masks features earlier than particle evaluation. The detected ROIs have been then manually filtered to take away false detection of non-GUV particles. The radial profile of every chosen ROI was extracted utilizing the Radial Profile perform. The origin of the radial profile was the measured centroid of the ROI. The radius is the first axis of a fitted ellipse to the ROI.
The radial profile was then analysed and plotted utilizing a custom-written Python script (Code availability). The centre of mass of the RNA fluorescence relative to the GUV centre was calculated for every GUV utilizing the components
$${x}_{mathrm{c}}=frac{mathop{sum }nolimits_{i = 1}^{N}{x}_{i}{y}_{i}}{mathop{sum }nolimits_{i = 1}^{N}{y}_{i}}$$
(2)
the place xc is the centre of mass of the radial fluorescence, xi are the normalized distance from the centre of the GUV, yi is the normalized RNA radial fluorescence at distance xi and N is the variety of x, y pairs.
GUV deformation
For all of the deformation characterization, GUVs have been encapsulated with DNA template (4–8 ng μl−1) with or with out biotin aptamer, RNA polymerase (0.2–1 U μl−1), α-haemolysin (15 ng μl−1) and sucrose (200 mM). The GUVs have been washed twice as talked about above utilizing a glucose answer (250 mM). On the subsequent day, 80 μl of purified GUVs was supplemented with a feeding answer, both containing a do-it-yourself buffer consisting of Mg(OAc)2 (6 mM), NaOAc (40 mM), KCl (40 mM), Tris-OAc (50 mM, pH 7.8), DTT (1 mM) or a industrial 1× transcription buffer from Thermo Scientific (catalogue quantity EP0112) and DFHBI-1T dye (62.5 μM). To provoke the transcription course of, GUVs have been incubated for six–12 h with 4 mM rNTPs in an Eppendorf tube inside a thermal block set to 37 °C earlier than imaging. For the management experiment, the identical set-up was used with out the addition of rNTPs.
Evaluation of deformation of GUVs owing to RNA nanotubes
Membrane detection and angular RNA origami nanotube depth measurements have been carried out with a custom-written ImageJ macro utilizing ImageJ2 model 2.14.0/1.54f and afterwards analysed and plotted utilizing a Jupyter Pocket book (Python 3.10.13) (Code availability). For membrane detection, the membrane channel of the confocal microscopy photos was processed utilizing Gaussian Blur with σ = 3 and auto-thresholded utilizing the Li technique. The Convert to Masks, Fill Holes, Erode and Analyze Particle features have been then utilized to acquire an ROI. The binary ROI picture was saved within the output folder and the centroid coordinates of the ROI have been calculated. The ROI photos have been then used to measure circularity utilizing the ImageJ Measure perform (set for Form descriptors) the place the circularity was outlined as
$$R=frac{4pi A}{{P}^{2}}$$
(3)
the place A is the world of the ROI and P is the perimeter of the ROI. For the GUVs the place a z-stack was acquired, solely the imply circularity is proven in Fig. 5g.
The space for every coordinate within the ROI after membrane detection to the centroid is calculated and saved in a consequence file. To detect the RNA origami nanotubes, a second, rectangular ROI was drawn from the centroid in the direction of the membrane with a size equal to 80% of the space between the centroid and the membrane. The oblong ROI was drawn for each tenth level on the membrane and the rectangle width was set to three μm. The imply depth for every rectangle was measured and saved in one other file. The Jupyter Pocket book script was then used to load the info from the 2 recordsdata and extract and plot the radii and intensities.
Statistics and reproducibility
Statistical analyses have been performed utilizing a parametric, unpaired t-test with Welch’s correction. Knowledge are introduced because the imply ± s.d. from no less than 6–78 GUVs noticed with confocal microscopy, relying on the precise experimental situations. For visualization, knowledge in Fig. 2g,h have been all plotted to 174 min. Persistence size values are introduced as fitted worth ± error of match. Persistence size evaluation was carried out solely on co-transcribed nanotubes that have been longer than 200 nm. Persistence size evaluation was executed on the center 80% layer of every simulated nanotube with overhangs excluded from the evaluation. Confocal and AFM experiments have been every carried out with greater than six and two impartial organic replicates, respectively. For plate reader assays, measurements have been performed in impartial triplicates. Simulation knowledge have been generated from three impartial runs. Solely simulations of single tiles the place the interior kissing loops include greater than 11 out of 12 base pairings have been used for additional in silico meeting and analyses. All analyses of nanotubes inside GUVs have been performed solely on GUVs containing nanotubes. Empty GUVs have been excluded from the evaluation. No statistical technique was used to predetermine pattern measurement. The experiments weren’t randomized. The investigators weren’t blinded to allocation throughout experiments and final result evaluation.
Knowledge visualization and evaluation
Plot graphing and statistical assessments have been carried out utilizing GraphPad Prism (model 5 and model 10.2.3) and Matplotlib 3.9.0.
Reporting abstract
Additional data on analysis design is out there within the Nature Portfolio Reporting Abstract linked to this text.
