Light-induced damage to DNA origami nanostructures in the 193 nm–310 nm range

DNA origami nanostructures provide precisely addressable substrates for in singulo experiments as well as for applications in nanotechnology. We report on experiments evaluating the stability of DNA origami upon irradiation with light at different wavelengths and buffer solutions. DNA is irradiated with nanosecond pulsed lasers and the damage is evaluated using UV–Vis spectroscopy and atomic force microscopy imaging. We show that the wavelength dependence of the damage follows the UV absorption spectrum of DNA. Electronic excitation of DNA is primarily responsible for DNA origami damage at present wavelengths. We also demonstrate UV–Vis absorption of tris reaction products, influencing the UV–Vis absorption evaluation in experiments studying DNA damage.


Introduction
Experimental studies of DNA model systems in the gas phase enable us to reveal intriguing details of reaction dynamics or environmental effects [1]. However, such experiments are size-limited to molecular sub-units of DNA such as nucleobases or nucleotides [2,3]. To overcome these limitations, we recently implemented DNA origami in our laboratories as a versatile tool to study precisely defined DNA segments [4]. * Authors to whom any correspondence should be addressed.
Original Content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. DNA origami nanostructures (DONs) are made by folding DNA single-stranded scaffolds to desired shapes using designed complementary staple strands [5]. Knowledge of all DNA pairs of the structure allows for precise positioning of DNA segments, nanoparticles or other molecules into advanced experimental configurations on a single DON [6]. In the present study, we focused on the stability of the nanostructure itself under illumination by UV light as a prerequisite to our radiation damage experiments. Further, the obtained data are important also for many emerging applications of DONs as discussed in the following paragraphs.
Many of the recent DON applications require their exposure to electromagnetic radiation. For example, they have been shown to be effective nanosubstrates to study UV damage in DNA strands in vacuum [7], and as possible radiometers for ultraviolet exposure [8,9]. Their viability as carriers for drugs and photosensitizers in photodynamic light therapy have also been explored [10]. Stability under UV irradiation may influence the output of DON-assisted lithography [11] or the DONreinforcement strategies such as UV cross-linking [12,13]. Enhanced cross-linking between base-pairs after UV exposure has been exploited to provide additional stability to DONs but significant strand-breaking can still occur at longer exposure requiring optimization of exposure times [14]. Finally, hybrid nanostructures based on DONs are directly illuminated in various nanophotonic applications [15]. Exposure to light at various wavelengths is apparent under these conditions and determining the stability or susceptibility to damage of DONs under UV exposure will help engineer more robust, light-responsive or light-controllable nanostructures for desired applications.
Keeping the integrity of the DNA origami structure is of importance in the above mentioned applications and strand breaks induced by light can damage the structure. UV damage to DNA strands normally involves formation of photolysis products such as cyclobutane pyrimidine dimers (CPD) and 6,4-photoproducts [16]. They result from π − π * transitions around 260 nm (4.8 eV) and produce lesions in the DNA strands. Damage of the phosphate group was also observed at wavelengths greater than 295 nm (E < 4.2 eV) through XPS measurements of UV-irradiated DNA and this can lead to strand breaks in the backbone at such low energies [17]. In the same study, damage to the H-bonded nitrogen in the bases and C-O bond in the sugar ring were observed at energies above 6.9 eV (λ > 180 nm). Above 8 eV (λ > 155 nm) where DNA bases begin to be ionized under VUV, strand break cross sections increase showing that direct photoionization is already the predominant initial mechanism leading to the observed damages [18].
These direct damages to DNA are expected to also occur in DONs exposed to light through the same mechanisms. In solution, reactions from the reactive species, particularly hydroxyl (OH) radicals from H 2 O photolysis, pose a further threat to DNA [19]. However, the highly cross-linked structure and compact nature already provide DONs with a certain extent of stability in this respect [20]. The cross-linked structure can tolerate photo-lesions up to a certain point, even contributing to some relaxation of the compact structure but could fragment abruptly at high doses when enough bonds and scaffold-staple linkages are disrupted [8].
Investigating the damages to these structures in solution can explain the various DNA-damaging mechanisms and improve protocols to develop them for applications necessitating laser exposure. Here we compare the damages observed at different wavelengths where various DNA-damage mechanisms are expected: 193 nm (6.4 eV) and 260 nm (4.8 eV). Processes that could occur upon irradiation of DNA in solution at these wavelengths are one-and two-photon ionizations at 193 nm [21,22] as well as photolysis of H 2 O to produce OH radicals from direct excitation and two-photon ionizations [23]. At 260 nm, two-photon ionizations may occur but are less probable so the effect of OH radicals may be negligible and the strong UV absorption of the nucleobases at this wavelength suggest damage initiated by electronic excitation as the most dominant contributor. Here, we use atomic force microscopy (AFM) to evaluate various types of structural damage to DONs as well as UV-VIS spectroscopy to further evaluate chemical changes to DNA origami and/or buffer components (tris).

DNA origami preparation
DNA origami nanotriangles were assembled following Rothemund's protocol [5]. An M13mp18 viral DNA scaffold strand (5 nM, Tilibit Nanosystems) was added with excess concentration of 208 oligonucleotide staples (Metabion International AG) in the folding buffer (FOB) composed of 1xTAE buffer with 12.5 mM MgCl 2 [24]. The solution was heated to 90 • C and steadily cooled down to 10 • C at a rate of −0.7 • C min −1 in a temperature-regulated heating block. Purification was done by spin filtration through 100 kDa MWCO Amicon filters for 4 min at 8500 rpm for three cycles.

UV irradiation
An ArF excimer laser (Excistar XS500, Coherent) was used to generate a UV beam of 193 nm (6.4 eV) with ∼4 ns pulse width. Using a deflector, the beam was directed to irradiate the center of the sample with an iris in between the deflector and the beam to regulate the spot size. The beam intensity between the iris and the sample was measured using an energy meter. The set-up was tuned to deliver 1-5 mJ pulse −1 with a frequency of 100 Hz over a spot diameter of 0.75-1 cm measured in the area where the energy meter is positioned. This results in an intensity of 1-6 × 10 9 W m −2 just above the sample. A 0.7 × 0.7 cm 2 plasma cleaned silicon chip was placed on a sample stage in the middle of the irradiated spot and the DNA solution was dropped in the middle. Irradiation at different energy fluences was performed at room temperature with the required number of pulses controlled through the Star PC-Control V2.81 Software. For instance, to reach 16 mJ mm −2 with 6 × 10 9 W m −2 laser intensity requires 667 pulses or about 6.7 s.
To generate the 225, 260, and 310 nm UV beam, an optical parametric oscillator (OPO) laser (EKSPLA NT-230) with ∼5 ns pulse width was used. The beam was then directed perpendicular to the sample by a prism. The energy measured just above the sample was between 0.9-1.5 mJ pulse −1 over a spot size of 0.8-1.1 cm giving an intensity of about 2-5 × 10 9 W m −2 with a fixed frequency of 20 Hz. The beam reaches maximum output only after ∼20 s; hence, it was first manually chopped and then released at maximum output to ensure uniform intensity over the desired irradiation times. The irradiation times are varied depending on the energy flux measured. For example, at 260 nm with 5 × 10 9 W m −2 intensity, it takes about 36 s (∼700 pulses) to reach 16 mJ mm −2 . The irradiation parameters are summarized in table 1. The irradiation was done on droplets of DNA solution (∼1 nM, ∼6 µL) on Si chips. For ex-situ UV-Vis analysis, the irradiation was done on a Si wafer that was sonicated in ethanol, rinsed with water, and then blown with N 2 to dry. Prior plasma cleaning was not performed for this substrate so that less DNA will stick to the surface and more DNA can be recovered from the droplets. After irradiation, the droplet was pipetted out from the surface. The volume of DNA solution recovered using this procedure was about 75% of the volume introduced and was consistent throughout all measurements. For the UV-Vis absorption spectroscopy, a NanoDrop TM (Thermo Scientific TM , OneC Microvolume UV-Vis Spectrophotometer) was used.

Deposition
Six microliters of DNA solution (∼1 nM) with varying TAE concentrations were dropped onto a plasma cleaned Si substrate facing the laser beam. The irradiated droplet is then added with 10 µL of 10xTAE-Mg buffer and left for 1 h over an ethanol bath for incubation. The samples were then rinsed with 1:1 ethanol-H 2 O solution and blown with N 2 to dry completely.

Damage quantification by AFM
AFM was done in air on DNA-origami-deposited substrates using the Dimension Icon AFM (Bruker) operating with PeakForce Tapping Technology and ScanAsyst probes (40 kHz, 0.4 N m −1 ). Areas of 3 × 3 µm in dimension were scanned for each sample with 512 × 512 pixel resolution. Following the procedure by Fang et al we quantified the damage of the origami from the area of the intact origami and fragments [8]. This was done using the ImageJ software through threshold adjustments isolating the background from the particles and the built-in particle analysis function that readily calculates the area of the detected particles. In certain images, where the surface roughness of the Si surface interferes with the thresholding, background subtraction was performed using the Mosaic plug-in in the same software [25]. The particle detection is limited to about 3%-150% of the average triangle area in the control sample. Particles less than 3% of the average triangle area are excluded as they cannot be differentiated from minute artefacts in the AFM images while those above 150% are mostly from clusters. The area distribution is then plotted with a resolution set to 5% of the average triangle area. The distributions are typically characterized by two peaks, one associated with larger fragments and intact triangles (> 50% of the average triangle area) and the smaller fragments (< 50% of the triangle area). A sample area distribution histogram is shown in figure S2 of the supporting material. Asymmetric Gaussian functions were fit inside the distribution graph and the areas (I) under the peaks are assumed to be proportional to the concentration of large and small particles, respectively. Equation (1) describes the percentage of intact structures left (Percent Intact) with I Intact as the intensity associated with intact nanostructures and A Total = I Intact + I Fragments ,

Wavelength dependence
To reveal the wavelength dependence of the damage, we performed experiments using two lasers. In the 220-310 nm range, explored using UV OPO system, we focused on inflection points of the UV absorption curve. The results can be observed in figure   process. This suggests that the damage could be initiated by the electronic excitation of the DNA components associated with direct effects of the radiation on the DNA. This is understandable, since the photolysis of water requires higher energies [26].
Using the ArF laser, we measured also the damage at 193 nm. Even though we kept the same flux of photons per pulse and per area, the results at lower and higher energies cannot be directly compared due to different parameters of the light pulse as well as the higher energy flux used in the 193 nm experiment. The 193 nm experiment, however, allows us to see qualitatively a different behavior due to the occurrence of reactive H 2 O photolysis products primarily in the form of OH radicals. We can see in panel (A) of figure 2 that in pure water, the damage is practically immediate. Quantum yields of damage to plasmid DNA in solution have been profiled by Görner, and in all of the observed products, damage at 193 nm laser irradiation is more pronounced than at 254 nm except for the generation of pyrimidine dimers (CPD) [27]. The latter, however, cannot be evaluated simply from AFM as these crosslinks cannot be resolved.
Our present results can be compared to the previous detailed study of light induced damage to DONs by Fang and coworkers [8]. Low intensity light sources were used in the study focused on the damages by broadband UV-C and UV-A radiation centered at 254 and 365 nm respectively. Similar to our wavelength-resolved results, the authors report no damage at UV-A wavelengths. Under UV-C irradiation conditions, the energy fluence dependence of the damage is less steep than in the present laser irradiation case, which can be attributed to the different brightness of the used irradiation sources. Considering the fact that the difference in the photon flux for the previously used UV-C lamp and the ns pulsed laser used in the present study is several orders of magnitude, the difference in the energy fluence dependencies is not so high indicating that the laser intensity dependence is rather weak. We made a short tests within the limited intensity variation of our 193 nm laser source confirming this fact (see figure S3 of the supporting material). In the UVB range (310 nm), cross-linking through CPD formation was performed in small DNA origami objects with additional thymine bases in defined cross-linking sites improving stabilization [14,28]. Similarly, we do not see pronounced damage within the energy fluence ranges we have accessed at 310 nm, which could be due to these cross-links, in addition to the lower absorption cross-section of DNA at this wavelength.
There exist already many studies concerning the damage induced to DNA origami by various environmental [24,29,30] or energetic stimuli [8,20]. The damage types range from effects conserving the shape of the origami such as scaffold strand breaks [20] or corner lift offs [29] to complete distortion of the origami shape and structure. The damage observed in the present case, figures 1 panel (C) and 2 panel (B), is consistent with the light-induced damage observed in the work of Fang and co-workers [8]. At low energy fluences, the shape remains intact, with possible evidence for tiny increase in size of the origami structure, which is however difficult to ascertain using AFM, because lateral resolution can highly depend on the tip quality and sample preparation (drying). At higher energy fluences, the structure is abruptly decomposed into fragments that freely migrate on the substrate. This is indicating that the electronic excitation of DNA is actually very efficient in breaking DNA as well as DNA to substrate bonds. This is in contrast to thermal excitation that results in the evaporation of the DNA components in a way that shape of the origami on the surface remains conserved [29].

Effect of radical scavenger concentration
In solution, the damage to DNA is mostly attributed to radicals not just from photolyzed DNA molecules but also from reactions with H 2 O photolysis products. At 193 nm, this can be a contributing factor because hydroxyl radicals are produced at this energy [23,31]. The formation of hydroxyl radicals is well confirmed from the UV-Vis spectra of pure folding buffer (FOB composed of 1XTAE and 12 mM MgCl 2 ) measured for 193 nm and 260 nm irradiation in panel (C) of figures 3 and 4, respectively. While there is no change to UV-Vis spectra of pure FOB in the case of the 260 nm irradiation, a strong peak at 268 nm appears upon irradiation at 193 nm that can be assigned to reaction products produced during the reaction of tris molecules, comprising bulk of the FOB, with OH radicals [32]. These spectra also indicate that no multiphoton effects are present at laser intensities explored in this work since no OH radical formation is observed at longer wavelengths.
The 268 nm peak then influences evaluation of the damages from the UV-Vis spectra as it overlaps with DNA absorption band around 260 nm. Still, we can evaluate some trends from the experiments with 193 nm and 260 nm presented in figures 3 and 4, respectively, showing results for irradiation in pure H 2 O (with residual FOB from the buffer exchange) in panel (B) and for irradiation in FOB in panel (A). The absorbance at 260 nm decrease in the absence of tris (figure 4 panel (B)) consistent with previous UV irradiation experiments of plasmid DNA in solution [33]. In the presence of tris ( figure 4 panel (A)), on the other hand, some absorbance at around 260 nm and 200 nm starts to appear with increasing energy fluence. This could be attributed to some reaction products of DNA radicals and tris since tris reaction products were not observed in the UV-Vis spectra of 260 nm irradiated FOB without the DNA. However, their closer identification cannot be performed in the present experiments.
The sample irradiation at 193 nm in H 2 O presented in panel (B) of figure 3 is influenced by residual tris from the buffer exchange procedure. This is demonstrated by the fact that despite the decrease of the DNA absorbance at low energy fluences, the 268 and 200 nm peaks typical for tris and OH radical reaction products appear at higher energy fluences, presumably upon saturating the residual tris scavenging capacity. The problem of residual tris was not so pronounced upon 260 nm irradiation as no OH radicals are formed at this wavelength.
Similarly, the 193 nm irradiation results in FOB are strongly influenced by tris absorbance and therefore cannot be used for making conclusions on the damage to DNA origami. The damage by OH radicals can be, however, disentangled from AFM analysis. The effect of varying hydroxyl radical scavenger concentration on the damage yields is shown in figure 2. Clearly, OH radicals are important contributors to the damage at this wavelength.

Conclusions
The present study demonstrates the correlation of damage to DNA origami nanostructures in solution to UV excitation of DNA. Electronic excitation of DNA results in efficient breaking of both inter-structure and structure to substrate bonds. Measurement at 193 nm demonstrates that OH radicals from water photodissociation contribute to the damage observed at this wavelength and presumably also at shorter wavelengths. The radical formation was confirmed also by experiments with radical scavenger tris, which is the major component of the FOB. It is then important to note that the observed dependence of the tris absorption spectrum on the energy fluence at 193 nm show that reaction products of tris with H 2 O and DNA photolysis products are strong UV absorbers, influencing the evaluation of DNA concentrations in UV irradiation experiments in tris buffers. Fragmentation or decomposition energy fluences for each wavelength can also be identified through this method which is crucial in experiments where DNA origami is exposed to light/lasers.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).
funding acquisition, supervision, formal analysis, writingoriginal draft; all authors: writing-review and editing.

Conflicts of interest
There are no conflicts to declare.