Self‐Regeneration and Self‐Healing in DNA Origami Nanostructures

Abstract DNA nanotechnology and advances in the DNA origami technique have enabled facile design and synthesis of complex and functional nanostructures. Molecular devices are, however, prone to rapid functional and structural degradation due to the high proportion of surface atoms at the nanoscale and due to complex working environments. Besides stabilizing mechanisms, approaches for the self‐repair of functional molecular devices are desirable. Here we exploit the self‐assembly and reconfigurability of DNA origami nanostructures to induce the self‐repair of defects of photoinduced and enzymatic damage. We provide examples of repair in DNA nanostructures showing the difference between unspecific self‐regeneration and damage specific self‐healing mechanisms. Using DNA origami nanorulers studied by atomic force and superresolution DNA PAINT microscopy, quantitative preservation of fluorescence properties is demonstrated with direct potential for improving nanoscale calibration samples.


General materials
For folding, purification and storage of DNA origami nanostructures, a 1x TAE buffer consisting of 20 mM acetic acid, 1 mM EDTA, and 50 mM Tris was used. The 1x PBS buffer used for fluorescence imaging consisted of 2.7 mM KCl, 137 mM NaCl, 1.8 mM KH2PO4, and 10 mM Na2HPO4. Individual concentrations of Mg 2+ used for every DNA origami nanostructure are given in Table S1. The scaffold strands were extracted from M13mp18 bacteriophages. Unmodified staple strands were purchased from Eurofins Genomics GmbH and Integrated Device Technology Inc. Dye labeled oligonucleotides for DNA PAINT imaging or brightness labeling were purchased from Eurofins Genomics GmbH (Germany).
Specific materials used for individual experiments are described in the sections below. Figure S1. Schemes of the DNA origami nanorulers used in this study. The 6 helix bundle (6HB) and the 12 helix bundle (12HB) are used as 1D nanorulers for modern fluorescence microscopy. The new rectangular origami (NRO) can be used as a 2D breadboard for the positioning of various modifications. Corresponding exemplary AFM images obtained after folding and purification shown in the lower panel illustrate the successful self-assembly of predicted structures.

DNA Origami folding
All investigated DNA origami nanostructures (depicted in Figure S1) were synthesized using the corresponding scaffold strands and temperature programs given in Table S1. Modifications of the nanorulers were realized using caDNAno (version 2.2.0). A full list of the unmodified staple strands and sequences of the 12HB DNA origami [1] is given in Table S15. A full list of the unmodified staple strands and sequences of the NRO DNA origami [2] is given in Table S16. A full list of the unmodified staple strands and sequences of the 6HB DNA origami [3] is given in Table S17.

Table S1
Corresponding scaffold strands, folding programs, folding buffers and magnesium chloride concentrations used for the folding of the DNA origami nanorulers 12HB, 6HB, and NRO shown in Figure S1. For the folding of the DNA origami nanostructures, the scaffold strand and the staple strands were mixed as given in Table S2 in the corresponding 1x folding buffer containing MgCl2 concentration as listed in Table S1. Unmodified core staple strands, which are completely incorporated in the origami structure, were used in 10-fold excess with respect to the scaffold strand. Staple strands with protruding 5'-ends, which act as docking sites for DNA PAINT or labeling experiments, were used in 30-fold excess with respect to the scaffold strand. Biotinylated staple strands, which were incorporated for enabling surface immobilization of the DNA origami structures, were used in a 30-fold excess with respect to the scaffold strand.

Table S2
Final concentrations and relative equivalents of scaffold strand, unmodified staple strands (core staple strands) and modified staple strands (e.g. biotinylated staple strands for immobilization and DNA PAINT docking site staple strands for superresolution imaging) used within this study. Folding of the 12HB and 6HB origami was realized with a non-linear thermal annealing ramp over 16 hours (Table S3) [4] , while the NRO was folded during a linear annealing ramp over 75 min (Table S4). Folding mixes had a total volume of 100 µl with final concentrations of scaffold strand, core staple strands and modified staple strands (biotinylated, DNA PAINT docking sites) as given in Table S4.

Purification of DNA origami nanostructures
Purification of folded DNA origami nanostructures was realized either by gel electrophoresis or by filter purification.
For gel electrophoresis, a 1 w% aqueous solution of agarose in a 1x TAE buffer with 12 mM MgCl2 was homogenized in a microwave. The solution was cooled down to ca. 50°C and stained with peqGreen (VWR International GmbH, Germany) and solidified by creating the wells for sample loading. The solidified gel was placed within a gel electrophoresis chamber filled with 1x TAE buffer containing 12 mM MgCl2. The whole chamber was placed in an ice bed to prevent melting of the gel. Five parts of sample solution were mixed with one part of 6x BlueJuice loading dye (Thermo Fisher Scientific, USA) and loaded in the wells. The gel was run at 60 V for ca. 2 h. Bands of interest were identified using blue light and cut out with a scalpel. Purified DNA solution was extracted by squeezing the cut gel bands. Alternatively, sample purification was realized by filtration using Amicon Ultra filters (100 K, Merck, Germany). The filter was first centrifuged with folding buffer for 7 minutes at 6000 g. The sample solution was then loaded into the filter and centrifuged for 15 minutes at 6000 g. 500 µL of folding buffer was loaded into the filter and centrifuged for 15 minutes at 6000 g, which was repeated. After three washing steps, the filter was inverted and placed into a new collection tube. The purified sample could then be collected by centrifugation for 2 minutes at 1000 g.
Concentrations of purified sample solution were measured via UV/vis spectroscopy (NanoDrop, Fischer Scientific, USA).

Surface-Immobilization of DNA origami nanorulers
For optical microscopy experiments, the DNA origami sample was immobilized on Nunc™ LabTek™ II chambers (Thermo Fisher, USA). The chambers were first cleaned with 400 µL of 1 M KOH solution and washed three times with 1x PBS buffer. Then the surfaces were passivated with 100 µL BSA-biotin (0.5 mg/mL in PBS, Sigma Aldrich, USA) for 30 minutes and washed three times with 1x PBS buffer. The passivated surfaces were incubated with 100 µL neutravidin (0.25 mg mL-1 in PBS, Sigma Aldrich, USA) for 15 minutes and washed three times with 1x PBS buffer. The sample solution with DNA origami featuring several staple strands with biotin modifications on the base was diluted to approximately 50 pM in 1x PBS buffer containing 12.5 mM MgCl2 and incubated in the chambers for 5 to 15 minutes. Sufficient surface density was probed with a TIRF microscope.

Photostabilization of fluorescent labels
Optical measurements with ATTO542 or Cy5 as imager fluorophores were carried out under photostabilizing conditions. [5] A 2.5x TAE buffer with 1 % (wt/v) D-(+)-glucose (Sigma Aldrich, USA), 165 units/mL glucose oxidase (G2133, Sigma Aldrich, USA), 2170 units/mL catalase (C3155, Sigma Aldrich, USA), 1 mM Trolox and 2 M NaCl was used. [6] The sample chamber with surface immobilized origami sample was completely filled with imaging buffer and sealed to prevent oxygen solvation. The first measurements were carried out at least 20 minutes after introducing the oxygen removal system to allow the equilibration of the oxygen concentration in the sample solution.

Fluorescence imaging of DNA origami nanorulers used as brightness standards
Automated long-term experiments with brightness nanorulers (12HB with self-regenerating label and 6HB with Nb.BtsI cleaved labels) were carried out on a commercial Nanoimager S (ONI Ltd., UK). Red excitation at 638 nm was realized with a 1100 mW laser, green excitation at 532 nm with a 1000 mW laser, respectively. The microscope was set to TIRF illumination. In order to not corrupt the first frames of the acquired intensity transients by the photobleaching of single nanorulers, the objective was first focused into the sample plane on a random section of the glass surface and the auto focus was activated. Subsequently the imaging lasers were shut off. Before starting time lapse measurements, the sample slide was moved to a new region of interest while still being kept in focus by the auto focus. The data acquisition was initialized by activating the lasers and taking frames of 100 ms over a user defined acquisition protocol (e.g. a frame of 100 ms taken every 10 min).
Throughout this study, fluorescence brightness imaging was realized with different imager strands but same imager concentration of 5 nM in 1x PBS buffer containing 12.5 mM MgCl2. Brightness data processing including background subtraction and data analysis were performed with ImageJ 1.52n (version 1.8.0_172). For drift correction the linear stack alignment with SIFT plugin in ImageJ was used. Spot detection was realized using a custom written algorithm in ImageJ.

DNA PAINT imaging
Super-resolution measurements using the DNA PAINT technique were carried out on a custom-built total internal reflection fluorescence (TIRF) microscope, based on an inverted microscope (IX71, Olympus). Red excitation at 644 nm was realized with a 150 mW laser (iBeam smart, Toptica Photonics) spectrally filtered with a clean-up filter (Brightline HC 650/13,Semrock). For yellow excitation, an additional 560 nm/1 W fiber laser (MPB Communications) also filtered with a clean-up filter (Brightline HC 561/4, Semrock) was used. The red and the yellow beams are combined with a dichroic mirror (T612lpxr, Chroma). To expand the beam profile, the laser passed through lenses (Bi-convex f50, Thorlabs; AC f120, Linos). The laser beam was coupled into the microscope with a triplecolor beam splitter (Chroma z476-488/568/647, AHF Analysentechnik) and focused on the backfocal plane of an oil-immersion objective (100 ×, NA = 1.4, UPlanSApo, Olympus) aligned for TIRF illumination. To avoid drift the objective was mounted on a nosepiece (IX-2NPS, Olympus). The fluorescence light is guided through an additional 1.6× optical magnification lens, an emission filter (ET 700/75, Chroma for red excitation or ET 605/70m, Chroma for yellow excitation) and finally focused on an scientific Complementary metaloxide-semiconductor (sCMOS) camera (pco.panda 4.2, 2048x2048 px, PCO AG) for detection. The calibrated pixel size was 42 nm/pixel. For data acquisition, a pixel binning of 2 was used resulting in an acquisition pixel size of 84 nm. Data acquisition was controlled with the software Micro-Manager 1.4. [7][8] The two used DNA PAINT imager strands (8 nt in length) with their corresponding sequences and fluorescent labels on the 3´end are given in Table S5. In general, DNA PAINT imaging was realized with the ATTO655 imager strand and imager concentrations of 1 to 5 nM in 1x PBS buffer containing 12.5 mM MgCl2. Typically, a frame time of 50 ms over an experiment time of 20 min was used with 640 nm laser excitation at 30 mW. For two color incorporation studies on the repair of kinked 12 HB nanorulers ( Figure S7), DNA PAINT images were obtained with Cy3B imager strands and excitation with the 560 nm laser.
Stabilization studies of 12HB DNA PAINT nanorulers after 2 hours incubation in 10% fetal bovine serum (FBS) solution were carried out on a commercial Nanoimager S (ONI Ltd., UK) using a 10 nM solution of the 6 nt ATTO655 imager strand (Table S5) in an 1x PBS buffer containing 12.5 mM MgCl2. A frame time of 25 ms over an experiment time of 10 min was used with excitation at 638 nm set to 160 mW output power. Table S5 DNA PAINT imager strands and fluorescent labels used for DNA PAINT measurements. DNA PAINT imaging in red channel was realized with ATTO655 labelled imager strands, while imaging in green channel with Cy3B labelled imager strands, respectively.

Analysis of DNA PAINT data
Acquired DNA PAINT raw data were analyzed using the Picasso software package. [9] The obtained tiff-movies were first analyzed with the "localize" software from Picasso. Centroid position information of single imager strand binding events was localized with a minimal net gradient of 10000 and a box size of 9 for data acquired with the custom-built TIRF setup and with a minimal net gradient of 2500 and a box size of 5 for data obtained on the Nanoimager S, respectively. The fitted localizations were further analyzed with the "render" software from Picasso. X-y-drift correction of the localizations was corrected with the RCC drift correction. DNA origami nanorulers were picked with the Render software and corresponding mean off-times and number of localizations per picked nanorulers were extracted for further analysis.
For quantitative distance analysis, the localization events of the picked nanorulers were exported from Render as csv. files for further examination with the software GATTAnalysis from GATTAquant GmbH, Germany.

AFM imaging with JPK Nanowizard
For probing correct folding of the origami structures and observing structural properties, AFM images were taken. AFM scans in aqueous solution (AFM buffer = 40 mM Tris, 2 mM EDTA, 12.5 mM Mg(OAc)2·4 H2O) were realized on a NanoWizard® 3 ultra AFM (JPK Instruments AG). For sample immobilization, a freshly cleaved mica surface (Quality V1, Plano GmbH) was incubated with 10 mM solution of NiCl2 for 3 minutes or alternatively with 0.01 % (wt/v) Poly-L-ornithine solution. The mica was washed three times with ultra-pure water to get rid of unbound Ni 2+ ions or Poly-L-ornithine and blow-dried with air. The dried mica surface was incubated with 1 nM sample solution for 3 minutes and washed with AFM buffer three times. Measurements were performed in AC mode on a scan area of 3 x 3 µm with a micro cantilever (νres = 110 kHZ, kspring = 9 N/m, Olympus Corp.).

Self-Regenerating brightness label on 12HB brightness ruler
For establishing a self-regenerating brightness label on a 12HB nanoruler, we designed 5x20 docking sites for external labeling into the 12HB structure ( Figure 1 main text). The five labeling spots were equally distributed along the 200 nm axis of the 12HB with 40 nm inter-spot distances. Therefore, we exchanged 5x20 unmodified core staple strands of the 12HB in caDNAno with the docking site staple strands given in Table S6. The docking site staple strands exhibit an over 20 nt long overhang at the 3'-ends. The used imager strands, which are complementary to the sequence of the docking sites, are listed in Table S7. The hybridization of 20 nt imager strands creates at room temperature thermodynamically stable and permanent labels, which are prone to photobleaching. Using shorter imager strands of only 13 nt length leads to a transient dynamic label, which can recover after photobleaching events. All used imager strands were labelled on the 3'-end, including the oxazine dye ATTO655 and the rhodamine dye ATTO542.
Self-regenerating of brightness labels on the 12HB brightness ruler (Figure 1 main text) was investigated with respect to the ability to recover brightness labels after photodamage, i.e. photobleaching. After initial brightness measurements (100 ms with 3 mW excitation at 532 or 640 nm), the immobilized nanorulers were photobleached (3 min, 3 mW without photostabilization, 20 mW with photostabilization) to a complete breakdown of the brightness function. The brightness recovery of the bleached nanorulers was measured over time via time-lapse imaging (100 ms every min with 3 mW at 532 or 640 nm).

Repair of kinked 12HB nanoruler
To emulate a structural damage in a 12HB nanoruler, we folded the DNA origami while leaving out 9 staple strands (Table S9) in the central region of the linear nanostructure resulting in a single stranded scaffold strand across all 12 helices (Figure 2 main text). To investigate the resulting 12HB nanoruler via DNA PAINT, we exchanged 60 staple strands with DNA PAINT staple strands given in Table S8. The DNA PAINT staple strands that were used exhibit the docking site sequence complementary to the 8 nt DNA PAINT imagers in Table S5. The docking sites were equally distributed on two sides along the 200 nm axis of the 12HB in ca. 7 nm distances to visualize the overall contour shape of the 12HB nanorulers. In a first folding process using the folding Program 1 in Table S3, the 12HB was folded while leaving out the nine staple strands given in Table S9. After purification, one part of the sample solution was examined with AFM and DNA PAINT. The other part of the sample solution was folded in a second step with a mix of the 9 missing staple strands in 300x excess using the folding program 1, but starting from T=50°C to accelerate the incorporation of the missing staples but also to not degrade the already folded 12HB nanorulers. After purification the repaired sample solution could be analysed using AFM and DNA PAINT. Additionally, incorporation studies were performed. Therefore, three of the nine later added staple strands were labelled with Cy5. Via widefield imaging of the same field of view in red channel (incorporated Cy5 staple strands) and subsequent DNA PAINT imaging in yellow channel (Cy3B imager strand) quantitative incorporation of the missing staple strands could be probed. For DNA PAINT experiments, 1 nM solution of the 8 nt ATTO655 or Cy3B imager strands was used.

Staple strand exchange from dual spot to triple spot in NRO nanoruler
To probe potential exchange of staple strands within a DNA origami with staple strands from solution, we designed an NRO DNA PAINT nanoruler with initially two DNA PAINT labelling spots, each consisting of 3 docking sites, with a distance of 40 nm. Therefore, we exchanged the 6 unmodified staple strands with the DNA PAINT staple strands given in Table S10. After folding and purification, the dual spot NRO could be imaged with the DNA PAINT imagers from Table S5.
For probing the exchange of staple strands within the DNA origami with staples from solution, we incubated the dual spot NRO with a mix of three DNA PAINT staple strands, which are selected to form a third label spot for DNA PAINT in 40 and 70 nm distance to the initial two spots, resulting in a triangle-shaped triple spot. To accelerate the incorporation, we used an 300x excess of the invasive DNA PAINT staple strands with respect to the dual spot NRO in the corresponding folding buffer (Table S1) and put the sample solution into the NRO folding program (Table S4) starting from just T= 50°C to prevent melting of the already folded dual spot NRO. In order to also emulate partially damaged staple strands, we folded the initial dual spot NRO with shorter unmodified staple strands, which are to be displaced by the invasive DNA PAINT staple strands. We used 3 initial staple strands, which are in one case 4 nt too short, in a second case 8 nt too short, resulting in a 4 nt and 8 nt toehold on the scaffold strand. For all three cases of initial staple strands (0, 4, 8 nt scaffold toeholds), the initial dual spot NROs were incubated with the invasive DNA PAINT staple strands as mentioned above.
To examine the invasion of the DNA PAINT staple strands, DNA PAINT experiments with 5 nM solution of the 8 nt ATTO655 imager strands were conducted. Table S10 DNA PAINT staples for a double spot NRO nanoruler (2x3 docking sites) and invasive staple strands forming a third labelling spot. The three initial staple strands designed to be displaced by invasive third label spot staple strands exhibited 0 to 8 nt shortened sequences to establish toeholds in the scaffold strand (left out sequences highlighted in green and blue). All DNA PAINT staples exhibit at their 3'-ends the 8 nt docking site sequence for the 8 nt imager strands in Table S5 (highlighted in red). The numbers for the 5'-end 3'-end of the staples represent the helix number in the corresponding caDNAno file. Number in brackets represents the starting and ending position of the staple in the corresponding helix. For immobilization, the 3'-biotinylated staple strands are used.

Self-Healing of 12HB nanoruler in degrading conditions
For probing self-healing processes of a DNA origami nanoruler in degrading conditions, we designed a triple-spot 12HB nanoruler suitable for DNA PAINT imaging using the 6 or 8 nt imager strands shown in Table S5. We designed three labeling spots on the 12HB origami with 107 and 70 nm interspot distances, by exchanging 3x10 staple strands by the corresponding DNA PAINT staple strands in Table S11. The DNA PAINT staple strands exhibit a 10 nt long docking site for DNA PAINT experiments, which show complementary sequence to the used imager strands.
Self-Healing studies in degrading conditions were performed by incubation of the immobilized DNA PAINT nanorulers in the folding buffer of 12HB (Table S1) containing additionally 0.2% (vol) or 10% (vol) fetal bovine serum (FBS) from ThermoFisher Scientific. For testing the stabilizing effect of random additional oligonucleotide sequences, we added a mix of unmodified staples strands from the 6HB origami (Table S17)   1.14. Self-Regeneration and Self-Healing of an enzymatically cleavable brightness label on 6HB brightness nanoruler For establishing a self-regenerating brightness label on a 6HB nanoruler by enzymatic cleavage, we designed 2x10 docking sites for the external labeling of the 6HB structure. The two labeling spots were distributed along the 400 nm axis of the 16HB with a 290 nm inter-spot distance. Therefore, we exchanged the 2x10 unmodified core staple strands of the 6HB in caDNAno with the docking site staple strands given in Table S12. The 20 nt ATTO655 imager strand sequence (Table S14) was designed to exhibit the specific Nb.BtsI binding sequence CACTGC, so that the enzyme could bind to a labeled imager strand and cut it into two 10 nt fragments.
6HB brightness rulers were immobilized and externally labeled in 1x PBS with 12.5 mM MgCl2 and 5 nM ATTO655 imager strand (Table  S14) over 1 hour. Excessive imager strands were washed away. Enzymatic cleavage of imager strands bound to the docking sites on the 6HB was realized in 1x CutSmart® buffer with 12.5 mM MgCl2 and 100 units/ml Nb.BtsI. For probing the activity of the restriction enzyme, internally labeled 6HB brightness rulers were immobilized, externally labeled and imaged. Then the enzyme was added and the sample was imaged after one night of incubation. After washing the enzyme away, the immobilized brightness rulers were again externally labeled and imaged. Brightness values were extracted for individual DNA origami, averaged and normalized to initial brightness. For a self-healing label by enzymatic cleavage, immobilized 6HB brightness rulers were incubated with 1x CutSmart® buffer with 12.5 mM MgCl2,100 units/ml Nb.BtsI and 5 nM of imager strands simultaneously. After waiting for steady-state conditions for 30 minutes, the self-healing label could be imaged.

Sequence (5` to 3`) Fluorophore label on 3´ Docking length (nt)
ATGCTAAGAT/CACTGCTAGTTT ATTO655 20 Figure S2. (A) Exemplary TIRF images of gradual bleaching and recovery (initial, bleached, recovered after 180 min) of orthogonal permanent imager strand, permanent label and self-regenerating label with ATTO655 (without photostabilization). Samples were bleached over 3 min with 75 W/cm 2 excitation at 640 nm. Scale bars represent 2 µm. (B) Corresponding extracted averaged and normalized single DNA origami intensity transients after photobleaching. Self-regenerating labels (blue) show a recovery of around 40%. The permanent label exhibits a small recovery due to post labelling (15%), while an orthogonal imager strand reference shows no significant recovery of brightness. Data represent average of three experiments, highlighted areas represent the standard deviation.

Self-Regenerating brightness label on 12HB brightness nanoruler
Exemplary TIRF images with initial, bleached and recovered brightness after 180 min for permanent and dynamic brightness labels (ATTO655) are given in Figure S2A. To probe, if a potential recovery is due to unspecific binding of imager strands to the immobilized DNA origami, we also measured the recovery of an orthogonal imager strand, i.e. an oligonucleotide labeled with ATTO655 but with a 20 nt sequence, which is not complementary to the used docking sites. Corresponding extracted and averaged recovery intensity transients per single nanoruler are shown in in Figure S2B. While the orthogonal imager strand exhibited only a very slow, insignificant recovery due to unspecific binding of imager strands to the nanorulers, the permanent label revealed a slightly higher recovery of around 15%. This low recovery could be explained by post-labeling of initially inaccessible docking sites. Accessibility studies of externally labeled DNA origami reveal usually accessibilities in the range of 60 to 90%. [10] After bleaching of permanent labels and subsequent ROS induced damage to the docking sites, initially inaccessible docking sites might become more accessible for intact permanent imager strands from solution. The self-regenerating label though exhibited a significantly improved recovery of around 40% under identical conditions. Figure S3. (A) Exemplary TIRF images of gradual bleaching and recovery (initial, bleached, recovered after 180 min) of self-regenerating labels without photostabilization (ATTO655 and ATTO542) and self-regenerating label with photostabilization (ATTO542). Samples were bleached over 3 min with 0.5 kW/cm 2 excitation power. Scale bars represent 2 µm. (B) Corresponding extracted averaged and normalized single DNA origami intensity transients after photobleaching. Self-regenerating labels without photostabilization (red and grey) show a limited recovery of around 50 to 60%. The photostabilized (GODCAT, Trolox) ATTO542 label exhibits complete recovery of up to 100% of its initial brightness. Data represent average of three experiments, highlighted areas represent the standard deviation.
To overcome limitations by photoinduced damage, we used an imager strand modified with the rhodamine dye ATTO542, which can be photostabilized by an enzymatic oxygen scavenging system and ROXS [5,11] . For oxygen removal, a 2.5x TAE buffer with glucose, glucose oxidase and catalase was used. To deplete triplet states of the ATTO542 dyes Trolox/Trolox quinone mixture was used as reducing and oxidizing system (ROXS). Gradual bleaching and recovery of the brightness of self-regenerating labels with and without photostabilization is given in Figure S3. In order to bleach the photostabilized labels completely, higher bleaching laser powers at 0.5 kW/cm 2 over 3 min were applied. While the self-regenerating ATTO655 and ATTO542 labels without photostabilization showed again limited recovery of only up to 60 %, the photostabilized ATTO542 label revealed a complete recovery of up to 100% of initial brightness. Under the used time lapse imaging conditions and the applied photostabilization, complete repair of the photoinduced damage to the brightness functionality could be realized by self-regenerating labels. To probe self-regeneration of the labels over multiple damaging events, we bleached the same field of view multiple times and measured the time lapse recovery of the brightness after every bleaching event. Exemplary TIRF images after every bleaching and recovery cycle of the self-regenerating label without photostabilization ( Figure S4A) and corresponding extracted average single nanoruler intensity transients in ( Figure S4B) revealed that even after 4 bleaching events the self-regenerating labels were able to recover back to over 20% of initial brightness. The photoinduced damage to docking sites by ROS is still clearly visible, since the recovery decreases over every bleaching event from initial ca. 60% to ca. 20%. Analogous multiple bleaching and recovery of the ATTO542 self-regenerating brightness label with photostabilization ( Figure S4C) revealed a strongly increased recovery over multiple bleaching events. Under the used imaging parameters (120 min for recovery), the photostabilized self-regenerating label recovered to over 80% for the first three recovery cycles. Only after the fourth bleaching, a decreased recovery of around 70% and 60% aft er fifth bleaching and hence an increasing damage to the docking sites was visible. Results from Figure S3 indicate a full recovery of the photostabilized self-regenerating label after 180 min. To minimize the chance of defocusing or too much sample drift during data acquisition, the recovery for multiple bleaching events was investigated over 120 min (i.e. before full exchanged had occurred) until next bleaching cycle was initialized. Figure S5. A) Exemplary AFM images of 12HB DNA origami folded leaving out 9 staple strands in the central region using folding program 1, repaired 12HB DNA origami with incorporated missing staples after a second folding with folding program 1 for T≤50°C and an intact reference 12HB folded with complete set of staple strands, respectively. Scale bars represent 500 nm. B) Corresponding angular distribution histograms obtained by manual angle measurement of AFM images over N picked molecules. Lines indicate cumulative distributions.

Repair of kinked 12HB nanoruler
Exemplary AFM images in Figure S5A revealed a large population of defective 12HB DNA origami when leaving out 9 staple strands in the central region of the nanoruler during the first DNA origami folding. The defective structures were mostly kinked and showed a decreased height in the region of single stranded scaffold. To accelerate incorporation of the 9 missing staples into the already folded defective 12HB nanorulers, we used a temperature ramp according to folding Program 1 in Table S3, but with T starting from 50°C, and a 300x excess of the 9 added staples with respect to the purified 12HB. The lower starting T was chosen to prevent melting of the already folded DNA origami. In a similar approach a scaffold strand was folded with a low number of staple strands in a first folding step with high starting temperature. In a second folding with lower starting temperatures, the set of missing staples could successfully fold the prescribed scaffold strand into the desired shape. [12] After addition of the missing staple strands, the population of defective 12HB was significantly decreased. Quantitative analysis was carried out by manual angular measurements between the two halves of picked nanorulers using ImageJ. To investigate nanorulers, whose structures were only influenced by the incomplete stapling of the scaffold strand, only those were analyzed which were immobilized as isolated monomers, while aggregates were dismissed. The obtained angular distributions in Figure S5B show a broad distribution from 0 to 180° for the defective structures. The 12HB sample after repair with the 9 missing staples exhibited an improved and narrowed angular distribution, which was shifted close to the angular distribution of an intact reference 12HB sample, which was folded with the complete set of staple strands. Defining all nanorulers with an angle under 160° as defective, resulted in a defective population of 63% after first folding with 9 staple strands left out and of only 32% after second folding with addition of the missing staples. The angular distributions indicate that the 9 missing staple strands were successfully incorporated into the defective 12HB origami and the emulated structural damage was partially repaired resulting in improved structural integrity of the nanorulers. Additionally, we investigated the defective and repaired 12HB nanorulers using DNA PAINT imaging and a dense docking site labeling along the whole length of the 12HB. By this labeling strategy, we were able to image and visualize the contour of the nanorulers with super resolution similar to AFM imaging ( Figure S6). DNA PAINT images of the defective 12HB nanorulers showed a large population (72%) of collapsed or kinked 12HB and a small population of linear nanorulers. The repaired 12HB nanorulers revealed a significantly increased population of linear, intact nanorulers, while the population of visibly defective nanorulers was decreased to 38%. It stands out, that the visibly defective 12HB nanorulers in the DNA PAINT images showed more collapsed structures than in corresponding AFM images. We ascribe this difference to the different immobilization strategies (Poly-L-Ornithine on mica in AFM, Biotin-NeutrAvidin immobilization on BSA passivated glass surface in DNA PAINT). In AFM imaging, the nanorulers were immobilized via ionic interactions with the positively charged surface and thus over the whole length of the 12HB. In DNA PAINT, the 12HB were immobilized via only 4 biotinylated staple strands, two on each site of the emulated damage, and should thus have a higher flexibility. The collapsed nanorulers could represent defective, flexible structures, which where immobilized via only one site of the emulated damage. Besides the conducted AFM and DNA PAINT characterization of the defective and repaired 12HB nanorulers, we carried out a colocalized widefield DNA PAINT experiment to prove incorporation of the 9 missing staple strands ( Figure S7A). Therefore, we exchanged 3 of the 9 missing staple strands with Cy5 labeled staple strands. After repair with the partially Cy5 labeled set of staple strands, we acquired diffraction limited Cy5 signals in red (640 nm) and corresponding DNA PAINT images of same regions with a Cy3B labeled 8 nt DNA PAINT imager strand. The colocalized image of diffraction limited Cy5 signals and DNA PAINT information revealed a successful incorporation of Cy5 labeled staple strands into most of the 12HB nanorulers measured with DNA PAINT, considering that only one third of the nine missing staples were labeled with Cy5. While the DNA PAINT images revealed the structure of the nanorulers, the diffraction limited single spots could be further investigated by extracting their corresponding time transients until bleaching. While most of the transients exhibited one incorporated Cy5 labeled staple, a minority also exhibited two or even three incorporated Cy5 labeled staple strand (see exemplary transients in Figure S7B). The colocalized images indicate significant incorporation of at least a subset of the nine missing staples, so that the above-mentioned changes in AFM and DNA PAINT experiments during repair indeed can be assigned to the incorporation of missing staple strands.
The results from AFM imaging and DNA PAINT experiments showed consistent results. The incomplete set of staple strands in the first folding resulted in a defective DNA origami population which could partially be repaired and improved in its structural integrity. Comparison with an intact reference structure showed that the repair could not remove the emulated damage in the whole nanoruler population but lead to a significant improvement of the structural distribution. The previous experiment with the kinked 12HB nanoruler shows that DNA strands can be incorporated into existing DNA origami nanostructures but it does not prove a self-healing mechanism as it is conceivable that staple strands would also constantly exchange in intact DNA origami structures. To this end, we designed a rectangular DNA origami with two spots (40 nm distance) consisting of docking strands for DNA PAINT measurements ( Figure S8). We then added staple strands with DNA PAINT docking strands extensions that would form a third spot on the DNA origami when incorporated. To increase the exchange kinetics, we incubated the double spot NROs with a 300x excess of the invasive docking strands in solution using the temperature range of the NRO folding program but starting at T=50°C, i.e. below the denaturing temperature (Table S4). Interestingly, only a vanishingly small number of triple-spot DNA origamis was observed indicating that staple exchange was kinetically blocked. If, however, the DNA origami was previously synthesized with shorter staple strands in the region of the third mark so that a toehold of 4 or 8 nucleotides was formed in the scaffold, the extended staple strands could invade and replace the existing staple strands more efficiently (see scheme and images in Figure  S8). After incubation with the extended staple strands, between 20 and 40% of DNA origamis exhibited the triple mark pattern as displayed in Figure S8B and Figure S8C confirming the notion that a toehold is required for efficient strand displacement reactions also within an intact DNA origami. [12][13][14] Successful incorporation of the staple strands forming the third labeling spot could also be probed by looking at the number of localizations per DNA origami nanostructure. The number of docking sites is increased during incorporation of the third labeling spot, which should also lead to an increase of localizations per DNA PAINT experiment. For the 8 nt toehold sample, the picked triple-spot nanorulers revealed an average number of localizations of around 354, compared to only 264 localizations for the picked double spots nanorulers within the same sample. For quantitative exchange and incorporation 150% of the localizations of the double spot nanorulers are expected for the triple spot population (since 6 docking sites are increased to 9 docking sites theoretically). The observed increase of localizations to 134% indicates, that on average around two out of the three docking sites are efficiently incorporated under the used conditions. A further comparison of exemplary DNA PAINT images of 0 to 8 nt toehold samples and corresponding extracted pick numbers and number of localizations are given in Figure S9. During manual picking of dual and triple-spot nanorulers in the obtained DNA PAINT images, we observed a small fraction of pseudo triple-spots in the samples even before adding the third labeling spot staple strands. Picking of the 0 nt toehold sample before addition of the invasive staple strands revealed e.g. a small fraction of around 5% of such pseudo triple-spots. We ascribe this population to NRO dimers, which accidently form triple spots by superposition of two individual double spot nanorulers. While the addition of the invasive staple strands led to no significant increase of the triple spot fraction for the 0 nt toehold sample, a significant higher fraction of around 20% could be found for 4 nt toehold and around 40% for the 8 nt toehold sample. According to toehold mediated strand displacement kinetics, the exchange is accelerated for increasing toehold lengths. To probe, if the increasing number of triple spots NROs was primarily due to the formation of more pseudo triple spots by unspecific dimerization of the DNA origami during the second folding, we made a reference sample for the second folding. An 8 nt toehold double spot NRO sample was treated with the same buffer and temperature ramp from NRO folding (starting at T= 50°C) but without addition of the invasive staple strands forming the third labelling spot. DNA PAINT images revealed no significant increase of the pseudo triplespot population ( Figure S9), indicating that the picked triple-spot populations after addition of the invasive staple strands can be attributed to successful incorporation into the existing DNA origami nanostructures. Figure S10. Exemplary DNA PAINT images of 12HB triple-spot nanorulers in 0.2% FBS solution (top), with added non-matching DNA strands (middle) and matching staple strands of the nanoruler (bottom) after immobilization, and 1 to 11 days of incubation, respectively. Triple-spot nanorulers are highlighted by green, doublespot nanorulers by yellow and single-spot nanorulers by red circles. Scale bars represent 200 nm.

Self-Healing of 12HB nanorulers in degrading conditions
Exemplary DNA PAINT images of immobilized triple-spot nanorulers (107 and 70 nm inter-mark distance) in the three different incubation conditions are given in Figure S10A. While the sample incubated in 0.2% FBS solution showed rapid degradation, i.e. loss of labeling spots and decreasing surface density, the addition of non-matching oligonucleotides led to a visible stabilization over the investigated time of 11 days. The addition of a set of matching unmodified staple strands stabilized the nanorulers significantly in the degrading environment so that even after 11 days, a majority of the nanorulers still exhibited a double or triple spot. For a more quantitative analysis of the induced structural damage, we extracted the number of localizations and off-times per picked nanoruler. For a degradation of the DNA PAINT nanorulers, a decrease of docking sites over time is expected. Lowered numbers of docking sites lead to lower numbers of binding events and thus also of localization events of bound imager dyes within a given time. Simultaneously, a decreasing number of docking sites increases the time between to binding event, i.e. the off-time. To decrease the influence of systematic fluctuations of the used widefield setup on quantitative analysis of the occurring damage of the DNA PAINT nanorulers, we conducted the degradations study for each incubation condition three times. Averaging over all three sets of experiments resulted in the curves given in Figure 3F-G in the main text. The extracted localizations and off-times correspond to the qualitative results from DNA PAINT images.
Combining the results from the previous NRO studies that only damaged staples with incomplete stapling of the scaffold strand are exchanged effectively, with the shown stabilization of 12HB nanorulers by presence of intact staple strands, we conclude that the given example fulfills our definition of self-healing. The applied self/repairing system improved the structural integrity of the nanorulers under wear significantly and could be applied to realize long lasting super resolution nanorulers. To investigate potential self-healing at higher damage rates, we carried out DNA PAINT studies of DNA PAINT nanorulers incubated in 10% FBS solution for 2 hours. To achieve DNA PAINT images faster, we used a 10 nM solution of the 6 nt ATTO655 imager strand in Table S5 in an 1x PBS buffer containing 12.5 mM MgCl2. With these parameters and a frame time of 25 ms, we were able to achieve super-resolution images of immobilized 12HB nanorulers within 10 minutes. Exemplary DNA PAINT images of an intact reference and after 2 h incubation in 10% FBS solution are given in Figure S11A. Incubation with 10% FBS led to fast degradation of immobilized 12HB nanorulers, while addition of intact non-matching or matching DNA staple strands led to significant stabilization so that a majority of the nanorulers still contained three spot pattern characteristic to an intact structure. To have a more quantitative comparison, we extracted the number of binding events per DNA origami nanoruler after incubation, averaged and normalized to the number of binding events of an intact reference structure ( Figure S11B). The commercial setup used for these experiments (ONI nanoimager S) is a closed system. During measurements with continuous excitation such as DNA PAINT imaging, the heat input by the laser illumination leads to an uncontrolled heating of the whole microscope body. On the other hand, the number of localisations during a DNA PAINT experiment is the product of binding times (in units of single frame time) and binding events. Since the binding times are highly dependent on the temperature thus are the number of localisations. The binding events, on the contrary, are more stable for small temperature variations, since they depend mostly on the concentration of the imager strand in solution. To have a temperatureindependent comparison, we extracted the number of binding events per picked nanorulers and compared the different incubation conditions ( Figure S11B). Two hours incubation with 10% FBS led to a decrease of binding events per nanoruler to under 50%. The addition of a set of non-matching staple strands and the set of matching staple strands at a total concentration of 5 µM lead to a significant stabilization resulting in a number of binding events of around 75%. The comparable results for non-matching and matching DNA staples indicated that the sacrificial degradation of the added DNA is the effective stabilization mechanism at 10% FBS. However, no self-healing effect could be observed. The high concentration of nucleases in the 10% FBS solution induce fast degradation of the DNA origami nanostructures, which cannot be compensated by self-healing as shown for 0.2% FBS incubation over days. Self-healing of DNA origami is thus limited to lower damaging rates, while the sacrificial degradation of added DNA can stabilize the nanorulers effectively even at fast degradation rates.
To further examine the stabilization of the DNA origami nanoruler in 10% FBS by sacrificial degradation of added DNA staples, we added different concentrations of DNA staples (total concentration of 5 µM, 500 nM and 50 nM) to the 10% FBS incubation solution. Exemplary distributions of number of binding events extracted from picked DNA PAINT nanorulers after two hours of incubation represented as box plots are given in Figure S12. For any of the added concentrations of DNA staples, the non-matching and matching staples resulted in comparable binding events, showing that no self-healing was stabilizing the nanorulers. Addition of DNA staple strands with a total concentration of 50 nM led to comparable damage than no addition of DNA, while the addition of staples with 500 nM led to a significant stabilization but lower binding event per nanoruler than addition of 5 µM of DNA staples. The results indicate that even with 500 nM solution of added DNA, DNA origami structures could be stabilized significantly in highly degrading conditions such as in 10% FBS via sacrificial degradation. Figure S12. Exemplary box plots of number of binding events per single nanoruler. Intact reference sample highlighted in black, sample incubated in 10% FBS in red, samples incubated with 10% FBS and non-matching DNA staple strands in blue and samples incubated with 10% FBS and matching staple strands in green, respectively. Squares indicate the average, central lines the median, box lines the 25% and 75% quantiles and whiskers the 1.5 times interquartile range, respectively. To estimate the external labeling efficiency of the brightness ruler, we compared the intensity of single immobilized ATTO655 labeled oligonucleotides and the 6HB brightness ruler ( Figure S13A). The designed brightness ruler exhibited a labeling number of 11.8 (59%). Next we probed the enzymatic activity of the used Nb.BtsI by comparing the brightness values before and after the incubation with the restriction enzyme. Therefore, we internally labeled the 6HB brightness rulers with ATTO532 (labeled staple strands in Table S13) to localize brightness rulers with complete label cleavage (B). Cleavage by Nb.BtsI overnight led to average brightness loss to below 20%.

Self-regeneration and self-healing of an enzymatically cleavable label on 6HB brightness ruler
After washing and addition of imager strands brightness recovered back to over 90% of initial brightness, indicating that the majority of the docking sites were not affected by the enzyme ( Figure S13C). To investigate and emphasize the concepts of self-regeneration and self-healing, we compared three different incubation conditions via time lapse TIRF imaging (3 mW at 640 nm every 10 min). Figure S14 shows exemplary TIRF images of immobilized 6HB brightness rulers over time incubated with a 5 nM solution of imager strands (top), with a solution of Nb.BtsI (middle) and a solution containing 5 nM imager strands and Nb.BtsI (bottom). While the addition of the restriction enzyme alone cleaved the labels and led to a rapid loss of the brightness signal after 1 hour, the incubation in a 5 nM imager strand solution could not recover the slow photobleaching during time-lapse imaging. Simultaneous addition of the restriction enzyme Nb.BtsI and of imager strands led to a dynamic stable brightness label showing no photobleaching effects under the used time-lapse imaging conditions even after 20 h.
In summary, the applied system recovers the building units of the brightness function, i.e. the imager strands, unspecifically with respect to photobleaching, as bleached and photoactive dyes are exchanged. With respect to the enzymatic damage by the restriction enzyme, only damaged units dissociate fast enough as two 10 nt fragments which can be replaced by an intact imager strand from solution. The self-repair in this example shows that self-regeneration and self-healing can occur simultaneously within one system when different sources of damage are present. Table S17. Unmodified staple strands of 6HB DNA origami. Sequences are denoted from 5'-to 3'-end. The numbers for the 5'-end 3'-end of the staples represent the helix number in the corresponding caDNAno file. Number in brackets represent the starting and ending position of the staple in the corresponding helix. Alle untersuchten DNA-Origami-Nanostrukturen (dargestellt in Abbildung S1) wurden mit den entsprechenden Scaffold-Strängen und Temperaturprogrammen in Tabelle S1 synthetisiert. Modifikationen der Nanolineale wurden mit caDNAno (Version 2.2.0) vorgenommen. Eine vollständige Liste der unmodifizierten Staple-Stränge und Sequenzen des 12HB DNA Origami [1] Tabelle S15 gegeben. Eine vollständige Liste der unmodifizierten Staple-Stränge und Sequenzen des NRO DNA Origami [2] ist in Tabelle S16 enthalten. Eine vollständige Liste der unmodifizierten Staple-Stränge und Sequenzen des 6HB DNA Origami [3] ist in Tabelle S17 gegeben.