Intercalating Electron Dyes for TEM Visualization of DNA at the Single‐Molecule Level

Abstract Staining compounds containing heavy elements (electron dyes) can facilitate the visualization of DNA and related biomolecules by using TEM. However, research into the synthesis and utilization of alternative electron dyes has been limited. Here, we report the synthesis of a novel DNA intercalator molecule, bis‐acridine uranyl (BAU). NMR spectroscopy and MS confirmed the validity of the synthetic strategy and gel electrophoresis verified the binding of BAU to DNA. For TEM imaging of DNA, two‐dimensional DNA origami nanostructures were used as a robust microscopy test object. By using scanning transmission electron microscopy (STEM) imaging, which is favored over conventional wide‐field TEM for improved contrast, and therefore, quantitative image analysis, it is found that the synthesized BAU intercalator can render DNA visible, even at the single‐molecule scale. For comparison, other staining compounds with a purported affinity towards DNA, such as dichloroplatinum, cisplatin, osmium tetroxide, and uranyl acetate, have been evaluated. The STEM contrast is discussed in terms of the DNA–dye association constants, number of dye molecules bound per base pair, and the electron‐scattering capacity of the metal‐containing ligands. These findings pave the way for the future development of electron dyes with specific DNA‐binding motifs for high‐resolution TEM imaging.


Introduction
UnstainedD NA and related biomolecules contain mostlyl ight elements, such as carbon, nitrogen, and phosphorous, that all have low electron-scattering strengths. As ar esult,h igh-resolution TEM imaging of unstained single DNA molecules supported on commercial carbon supports has been unsuccessful because, as the electronw ave passes through DNA,o nly negligible changes occur to its amplitude or phase, whichc onsequently leaves the unstained DNA invisible under TEM. The use of electron dyes is therefore as tringent requirement for in-creasinge lectron scattering, andh ence,t he TEM visualization of DNA.
Surprisingly,d espite the pressing need for DNA-staining compounds, there is ah istorical gap in the systematic investigation of effective and accessible electron dyes. Uranyl acetate (UA) and its analogues have been the only DNA stain since 1961. [1] UA binds covalently to the negatively charged backbone of DNA, [1] and with the heavy atomicn umber of 92 for the uranium element, it provides excellent electrons cattering in TEM. Practically,t he use of UA is not ideal because it is extremely toxic and its access requires governmental permissions due to tight restrictions on nuclearf uel materials.I ti sw ell knownt hat the DNA-binding mode, and consequently,c ellular responset oc ompounds changes with even slight modifications in the coordination chemistry of the transition metals. [2] Hence, from ab iochemistry point of view,i tr emains an interesting question whether it would be possible to engineer a synthetic compound with ac ompatible uraniumc ore element for providing excellent TEM contrast, but with an intercalation bindingm ode, rather than covalent attachment to the DNA phosphate backbone.
Can intercalation be utilized for TEM imaging of DNA?T he designo ra pplicationo fi ntercalating molecules containing heavy elements as electron dyes has been very limited so far. For example, although many DNA-binding compounds that contain heavy elements (Pt, Ag, Au, etc.) are conceivable, only platinum has gaineda ttentiond ue to its rich libraryo fc oordination chemistry.I ndeed, since the first successful application of cisplatin as an anticancer agent,t he field of synthetic platinum compounds has evolvede xponentially. [2,3] Cisplatin is a Stainingc ompounds containing heavy elements (electron dyes) can facilitate the visualization of DNA and related biomolecules by using TEM. However,r esearch into the synthesis and utilization of alternative electron dyes has been limited. Here, we reportt he synthesis of an ovel DNA intercalator molecule, bis-acridine uranyl (BAU). NMR spectroscopy and MS confirmedt he validity of the synthetic strategy and gel electrophoresis verifiedt he binding of BAU to DNA. For TEM imaging of DNA, two-dimensional DNA origami nanostructures were used as ar obust microscopy test object.B yu sing scanning transmission electron microscopy (STEM) imaging, which is favored over conventional wide-field TEM for improved contrast, and therefore, quantitative image analysis, it is found that the synthesized BAU intercalator can render DNA visible, even at the single-molecule scale. For comparison, other stainingc ompoundswith apurported affinity towards DNA, such as dichloroplatinum,c isplatin, osmium tetroxide, and uranyl acetate, have been evaluated. The STEM contrast is discussed in terms of the DNA-dye association constants, number of dye molecules boundp er base pair,a nd the electron-scattering capacity of the metal-containing ligands. These findings pave the way for the future development of electron dyes with specific DNAbindingm otifs for high-resolution TEM imaging. covalentD NA binder,w hereas other platinum complexes,e specially those containing planar aromatic terpyridine moieties, are intensively studied DNA intercalators, owing to their vast therapeutic applicationsi nc hemotherapy and cancert reatment. [3][4][5][6] In addition, it has recently been shown that some derivativeso fp latinum complexes can penetrate into the cell nucleusa nd show cell viability. [4,6,7] This grants them ac lear advantage for TEM imaging applicationso fb iological samples. However,d espite these promising intercalation properties, no TEM visualization of single DNA molecules hasb een reported, so far,t hrough intercalation binding of platinum or any other heavy elements, such as uranium.
Twom ajor drawbacks have impeded progress towards the systematic investigation of contrast agentsf or TEM visualization of DNA:1 )demanding and sometimesa mbiguous sample preparations, and 2) nonoptimized choicesf or the TEM imaging mode. With regard to the first point, most reports in the literatureu se plastic-embedded tissue sections or viruses as the model system for microscopic investigations. [8,9] Apart from tediouss ample preparation, artifacts often affect the results due to the crowded environment (presence of proteins, lipids, DNAs, etc.) [10] The second drawback is associated with inherently noisy TEM images, even for stained DNA. Due to the faint signal of DNA, contaminants or the substrate signal often interfere with imaging and reduce the obtainedD NA contrast.A ccordingly, reliable and quantitative analysis of DNA images has proventob edifficult and sometimes impossible.
Here, we present an ovel intercalating agent that is synthesized by covalently tethering ab is-acridine moiety to as alophen ligand,that is, aSchiff base, which functions as achelator for au ranium atom (Scheme 1). We fully characterized our synthesized intercalator by meanso fN MR spectroscopy,M S, and gel electrophoresis. We then present our TEM investigations. To overcome sample preparation problems, we utilized aw elldefined DNA origami nanostructure as am icroscopic test model system for TEM imaging (Figure 1). DNA origami,t hat is, DNA folded into au ser-defined shape, [11] provides af acile and straightforward approacht oe valuate stained DNA and strongly reduces the above-mentioned sample preparationc hallenges because the characteristic shapeoft he origami designf acilitates the detection and quantitative analysis of the DNA contrast. Furthermore, we significantly improveT EM imaging by using the scanning transmission electron microscopy (STEM) technique. STEM is advantageous over conventionalw ide-field TEM because it providesahighers ignal-to-noise ratio (SNR) for single-image acquisitions. Indeed, STEM enabled us to quanti-Scheme1.Synthetic scheme for the preparationo fB AU. The complete protocols for the synthesis are given in the Supporting Information.

Results
Synthesis, characterization, and binding properties of BAU to DNA Intercalators are widely applied in analytical and medical chemistry, [12] and they offer the potentialf or tailoring the surface properties of DNA nanostructures. [13] Inspiredb yt he wide range of useful properties of acridine derivatives, andi nc ontinuation of efforts to expand its applications as staining agents, we synthesized an ovel bidentate intercalator compound, BAU( Scheme 1), that contains one uranyl cation confined inside the chelating salophen moiety in the center of its structure, which acts as aT EM contrast enhancer.T his unit is tethered to twoa cridine heterocycles (bis-acridine) that serve as DNA intercalatingl igands. The linker strategy for tethering the bis-acridine moiety to the uranium metal centerw as adopted from salophen-typec oordination chemistry. [14] Af ull description of the experimental procedures for the synthesis of BAU is given in the Experimental Sectiona nd the Supporting Information.
NMR spectroscopy and MS analyses confirmed the structure of BAU, and validated our synthetic scheme. The 1 Ha nd 13 C chemicals hifts of BAU were assigned upon standard 1D and 2D NMR spectroscopy characterization (Figures 2a nd 3). Specifically,t he 2D NMRs pectroscopym easurements ( Figure 3) allowed the spin systems to be assigned to the following structural patterns: the salophen-UO 2 complex, the acridine moiety, and the connecting hexyl spacer. In particular, the signal of the magnetic resonancesa ttributable to the azomethine groups of the salophen-UO 2 complex was observed at d = 9.60 ppm, whereas significant magnetic resonances for the three phenyl rings bridgedb yt hese azomethine groups appeared in the range of d = 7.22-6.56 ppm. The signals corresponding to the aromatic pattern of the acridine moiety were assigned in the range of d = 8.45-7.30 ppm. The presence of the hexyl chain spacer was confirmed by the magnetic resonances of the CH 2 groups attached to the phenylg roup of the salophen-UO 2 complex and the amino group at the 9-position of acridine, which were underpinned at d = 4.15 and 3.93 ppm, respectively.F igures S1-S5 in the Supporting Information allow for further 1 Ha nd 13 CNMR characterizationo ft he intermediate synthesized compounds (i.e.,c ompounds 3-7). The complementary characterization of the exact mass by means of HRMS (ESI) confirmed the formation of target BAU ( Figure 2B). Based on the thorough structural confirmation by meanso fN MR spectroscopy and ESI-MS,w et hus explicitly showedt he feasibility of synthesizing am etallointercalator molecule associating a very heavy element, such as uranium.
As ap rerequisite to use BAU as aD NA electron dye,i t should bind strongly to DNA.W ep erformed gel electrophoresis to evaluate the bulk binding properties of BAU to DNA origami nanostructures ( Figure 4). DNA origami nanostructures, containing fluorescent Cy5-labeled staple strands, werei ncubated with variable molar ratios of BAU to DNA base pairs, startingf rom an excess value of 5BAU bp À1 to lower ratios down to 0.05 BAU bp À1 ,a sc alculated from the scaffold length of the DNA origami. Visualization of the origami nanostructures was achieved by fluorescent imaging of the Cy5 fluorophores,a sw ell as by stainingw ith SYBR Gold. Figure 4s hows the data and elucidates two important points:1 )A close look at the immobile band indicated inside the dashed white rectangles reveals an increasing intensityi nt he SYBR Gold channel with lower BAU concentration, but au niform intensity in the Cy5 channel. This suggests that BAU competitively inhibits the binding of SYBR Gold to DNA nanoplates,m ost likely through intercalation. Our observation is consistent with the general consensus thatb is-acridine moieties are indeede xcellent intercalators. [15,16] 2) BAU binding alters the electrophoretic mobility of the DNA origami,s ince we observet hat the origami structures become completely immobilized inside the gel pockets (lanes 1t o4 ), and can only acquire partial mobility below stoichiometric ratios of 0.5 BAU bp À1 (lanes 5a nd 6). This is attributed to both the positive charge of BAU and interorigami crosslinking that can occur if the two acridine intercalating units of BAU bind to two different origami plates. Indeed, TEM images taken from these complexes confirmed such BAU-induced origami interactions ( Figure S6). Collectively, the results of electrophoretic analysisi ndicate that BAU binds tightlyt oD NA, thereby suggesting its potential to be used as aD NA electron dye.

TEM imaging
We find that it is not possible to image the stainedD NA nanostructures with high contrasti nw ide-field TEM, even with a state-of-the-art direct detection device (DDD). In recent years, the fieldo fT EM has gone through ad ramatic improvement in instrumentation, especiallyd ue to the emergenceo fD DD technology.B ye liminating the scintillator layer,t he DDDs becamea dvantageous over conventional charge-coupled device( CCD) cameras because they yield ah igherd ynamic range,h igher detective quantum efficiency for all spatialf requencies up to the Nyquist limit, sample-motion correction, and al ower shot noise. [17] Because improvements in the detector would be beneficial to boost the contrasti nw ide-field TEM, ar ecent state-of-the-artD DD (Model:D E16, Direct Electron, California, USA) was employed in our aberration-corrected microscope.F igure 5A and Bs hows the best micrographs that we coulda cquire under near-focus and strongly defocused ob-   Figure 5B). Only the origami main rectangle and occasionally the side arms (4 nm wide) are distinguishable, but the dsDNA loop at the bottomi sn ot recognized at all. Note that the contrasti nF igure 5A and Bi s very faint because we do not perform negative staining, with which the contrasti sg eneratedb yas hadow image of the DNA in au niform stain background. Rather,t oi nvestigate the "selectivity" of electron dyes, the contrasti sg enerated by direct interaction of the compounds with DNA (that is, positive staining). To investigate the visibility at the single-molecule level, our origami contains only one layer of DNA, unlike multilayer 3D DNA origami designs, [18] and hence, it yields very low electron scattering.
The poor visibility of the stained DNA origami structures in wide-fieldT EM imaging is explained by electron-optical reasons. For materials science specimens (mostly metals and ceramics),w ide-field TEM at af ocus close to zero is extensively used to resolve microscopic features with atomicr esolution (even below 1 resolution). However,i nt he case of biological samples, TEM imaging at near-focus conditions fails to provide enough contrastf or visualization. [19] This is illustrated in Figure 5C and D: we calculated the CTF of our aberration-corrected Titan microscope at near-focus and at strongly defocused imaging conditions, respectively.T he CTF is am athematical functiont hat takes into account variousi maging parameters, such as the objective-lens defocus,a cceleration voltage, and aberration coefficients, and it plots the phase-to-amplitude conversion efficiency as af unction of spatialf requencies of the electron wave (i.e.,t he k vectors). For example, in the case of near-focus imaging ( Figure 5C), we observe no information transfer for the 2nmf ringe, which is the periodicity of the stacked DNA bundles within the origami rectangle, although the microscope can stillr esolve spatial features up to 2 (the right side of the curve with k = 0.5 À1 ). All spatialf requencies in the green-highlighteda rea in Figure 5C will be absent in the final image. If the objective lens is strongly defocused up to 3000 nm ( Figure 5D), however,t he 2nmp hase component is converted to amplitude with better efficiency,a lthough damped by the total spatial and temporal envelope. At this strong-defocus illumination, the CTF oscillates rapidly,w ith its first zero shifting to lower spatial frequencies, which consequently resultsi naloss of resolution (well below 5 ,t hat is, more than a5 0% drop in resolution relative to Figure 5C). We therefore conclude that wide-field TEM is not ag ood approach to visualize DNA nanostructures with high contrast.
Unlike wide-field TEM, we find that the STEM technique is suited to probe the DNA-dye interactions. STEM is an imaging technique with av ery different image-formation mechanism, [20] which does not suffer from the CTF constraints. One main advantage of STEM overT EM for our purpose hinges on the fact that the obtained contrast is almostq uadratically proportional to the atomic number (Z-contrast imaging), [20] and dyes containing elements with atomic numbers of 92 (UA, BAU) or 78 (DP) thus generate ac risp contrasto ver the substrate with a   Figure 6A-C summarizes our STEM resultsf or electron dyes that rendered origami nanoplates visible. All images were taken under the same STEM acquisitionp arameters in the microscope and were treatedw ith as imilar despecklen oise reduction and contrast adjustment for better visibility.
Importantly,o ur new BAU dye is seen to yield good DNA contrasti nS TEM ( Figure 6B). For BAUa nd DP,w es ee contrast enhancement not only for the main origami rectangle (72 nm 50 nm), but also for individual DNA strands( the floppy loop at the bottom that is only 2nm( single DNA helix) wide, as well as side arms that are composed of 2D NA helices, i.e.,4nm wide). Such high-contrast single-shot STEM images of DNA supportedo nc ommercial carbon membranes indicate that STEM is ap articularly fit technique for imaging stained DNA nanostructures. Excitingly,u nlike staining protocols used for tissue sections, it is clear that no post-fixationp rocess is necessaryf or our DNA origami samples, which presentsa ni mportanta dvantage for artefact-free imaging. To probe whether it was possible to increaset he contrast even further,w e stainedt he DNA nanoplates with highs tain concentrations and increased the incubation times. We found that the contrast saturated atacertain concentration for each dye (Figure S7). The images in Figure 6t hus represent the highest contrast that could be achieved for each compound;l ikely representing the maximum number of dye molecules bound to the DNA strands.
We also investigated other staining agentsr eported in the literature, such as OsO 4 and cisplatin, but we found no visibility in STEM imaging of DNA origami withthese compounds. OsO 4 , with its high atomicn umber of 76 for osmium, plays au nique role in TEM imaging of biological samples, [21] and it is often reporteda safixative agent forp lastic-embedded tissue sections. Bahr reported that OsO 4 didn ot react with nucleic acids, [21] althoughn om icroscopic evidencew as presented. To attempt to provide microscopice vidence for DNA binding, we stained our origami nanoplates with OsO 4 and performedS TEM imaging. We could, however,n ot detect any discernible contrast. Cisplatin, one of the most successful antitumor drugs, can potentially also be considered as an electron dye. It is known that cisplatin binds quasi-covalently to DNA on the N 7 -positions of the bases, which consequently causes double-helix unwinding, kinking, and DNA crosslinking. [2,3,[22][23][24] Despite this purported covalenta ffinity towards DNA, we were unable to visualize cisplatin-stained DNA origami through STEM. Some studies pointed out that cisplatin required long incubation times to interact with DNA. [3,7,23] We therefore prolonged the incubation time of cisplatin with DNA origami nanoplateso nt he TEM grid up to 2days. After this long incubation period, we could observe large amounts of nanoparticles deposited on the TEM grids, but unfortunately still not any visible DNA origami ( Figure S8). Our microscopic findings thus show that cisplatin andO sO 4 are not suitable for in vitro staining studies of DNA.
We quantified the contrasto ft he stainedD NA origami nanoplates. Different origami nanoplates were selected from random acquisitions at different locations on the TEM grids, and the HAADF detector signals were extracted along the origami main body ( Figure 6D). SNR and contrastwerecomputed according to SNR = (I s ÀI n )/Std n and contrast = (I s ÀI n )/I n ,r espectively,i nw hich I s denotes the average signal along the origami main rectangle, I n is the average background noise measured along the carbon support, and Std n is the standard deviation of the background noise. Remarkably,F igure 6E and Fs hows that BAU, with the bis-acridine intercalating ligand, provides an excellent SNR and contrast, approaching the values of the UA, but now as ap ositive-stain intercalator.T he SNR andc ontrast values for BAU exceed that of DP,w ith the terpyridine intercalating moiety (more than twice as high SNR and 60 % higher contrast). This indicates that our design strategy to consider the bis-acridine ligand was indeed more effective than the intensively studied terpyridine conjugation. Notably,a lthought he backboneb inder UA provides the highest contrast among the stains, we aimed to synthesize and investigate in- tercalating electron dyes. When using the intercalating BAU and DP compounds,w ea lwayso bserved ah igher noise (carbon appears to be brighter), which we could not avoid, even after severalw ashing steps. This may be duet os ome affinity of the intercalating ligands to the surface of amorphous carbon.N evertheless, single-DNA molecules are clearly visible over the background stained with BAU and DP intercalating dyes.

Discussion
We designed an ew intercalating electron dye and presented single-molecule visualizations of DNA. In our experiments, Z-contrast STEM images were generated through direct interaction of these dyes with the DNA strands, insteado fc onventional negative stainingp rotocols in which the contrasti sr endered through as hadow image of DNA.T able 1p rovides an overview of association constants, number of dye molecules bound to each DNA base pair,a nd the electron-scattering capacityo ft he heavy-metal center.V arious of these attributes are important for rendering high-contrast images,a sd iscussed below.
First, we notet hat the common denominator between all of these dyes is their high binding constants, in the range of 10 4 -10 6 m À1 .T his strong affinity is indispensable for visualization, rendering single-molecule DNA visible even after severalw ashing steps during sample preparation. Regarding BAU, this strong binding affinity was verifiedb ym eans of gel electrophoresis (Figure4). The use of the bis-acridine moiety as the intercalating ligand in the BAU design approach was intentional, with respectt oahigh bindingc onstant,b ecause it was shown that bis-acridines had as ubstantially higher affinity to DNA than that of their corresponding mono-acridine analogues. [15,16] Next to ah igh DNA affinity,a nother advantage of bis-acridine conjugation to thes alophen ligand containing uranium is that bis-acridine moieties are known to be biocompatible agentswith excellent antitumorp roperties. [15,16] Second, the dye density along the DNA helix should be maximized to obtain the highest scattering efficiency.C ertainly, UA providest he highest value of one dye bound per base pair, compared with BAU and DP that perform af actor of 4o r5 worse than UA. Beer and Moudrianakisen showed that attachment of three heavy markersp er base pair was required for one full amplitude contrast onto thin carbon membranes in wide-fieldTEM, [25] whereas staining with UA could yield am aximum of only one heavy atom of uranium.T his, together with phase-contrast limitations imposed by CTFs shown in Figure 5C and D, explainst he faint contrast that one obtains in normalw ide-field TEM mode, for which even utilization of a DDD is not helpful to boost the contrast( Figure 5A and B).
Finally,a lthough on the basis of theoretical arguments, the STEM contrast is expectedt os cale as approximately Z 2 versus atomic number-technically the current in the HAADF detector reaches ap lateau-which consequently saturates the grayscale values in the STEM images and impedes heavy elements from being distinguishable from one another.F or example, Ferrer et al. showed that the intensity differenceb etween the Au (Z = 79) and Pd (Z = 46) columns in aA u/Pd nanoparticle was very small. [26] Accordingly,w ee xpect that the obtained STEM contrastf or the dyes mentioned in Ta ble 1( with Z = 92 for Ua nd Z = 78 for Pt) is not likely to be strongly dependent on the core heavy element. In other words, we estimate that the effects of binding constant and dye loading density are much more pronounced in determining the overall contrast in STEM than the atomic number of the core heavy metal. Note that this statement is only true for heavy elements, whereas Zcontrasti maging remains strongly beneficial if heavy atoms are supported on light substrates, such as carbon ( Figure 6).
Maximizing the above-mentionedp arameters provides a strategyf or biochemists to furtherd evelop highly efficient electron dyes. Most importantly,t he binding constanta nd dyeloading density significantly contribute to contrast. We have shown the feasibility of bis-acridine conjugation to the salophen ring as one strategy for incorporating uranium (in BAU). Also, the Pt II compounds containing terpyridine moieties (in DP) are shown to be suitable candidates as intercalating electron dyes. The application of many platinum(II) or osmium complexes have been demonstratedf or chromatin imaging. In fact, there are numerous papers presenting such chromatin data, [4,7,27] but virtually none on in vitro imaging of single-molecule DNA throughi ntercalation binding, which is the focus of our work. Based on our quantitative analyses in Figure 6E and F, such compounds perform more poorly than our bis-acridine design strategy for single-molecule visualizationb ecause our BAU compound has two DNA binding sites and as patially distant heavy metal, which allows at weezer-like binding to the DNA molecule. Te theringo ther intercalating ligandsc an be considered. The rich coordination chemistry of the transition metals, togetherw ith the diversec onjugation possibilities to flat aromatic ligands, are promising factors fort he future synthesis of electron dyes. We hope that our findings and methodology will spur the interestso fc hemists to furtherd evelop highly efficient intercalating electron dyes.

Conclusion
We imaged single DNA molecules with high contrast by using an ewly designed intercalating molecule, and we compared the results to those for other electron dyes. By introducing DNA origami as am icroscopy test objecta nd by selecting STEM over conventional wide-field TEM, we could characterize electron dyes that were appropriatef or single-molecule DNA visualization. Imaging artefacts in previous investigations were eliminated, as ar esult of facile sample preparation and the absence of other biological molecules, such as lipids and proteins. Our methodology is suited for investigation of other newly synthesized electron dyes in the future. Looking forward, one intriguing application of intercalating electrond yes is in multicolor electron microscopy of biological systems. The first ever multicolor TEM images of such samples was shown by Adamse tal. in 2016, who visualizedd ifferent cellular components,very similar to what multicolor fluorescence microscopy could offer,b ut with the full spatial resolution of TEM. [28] This is opening up new possibilities to push the resolution limits in life-science TEM, and thus, expanding our understanding of the molecular processes of life.

Experimental Section
Staining compounds:T he BAU dye was synthetized as shown in Scheme 1. To obtain an optimal linker length between the DNA intercalating ligand (i.e.,t he acridine moiety) and the TEM-contrastenhancer scaffold (the uranyl unit), commercially available 9-chloroacridine (1)w as reacted with 6-aminohexan-1-ol (2), as described previously. [29] The isolated 9-hexylaminoacridine compound (3)w as tosylated to afford 4,w hich was subsequently reacted with the selective allyl-protected 2-(2-propenyloxy)-3-hydroxybenzaldehyde (5)t oy ield compound 6,w ith ab enzaldehyde unit. The successful deprotection of 6 provided the free hydroxybenzaldehyde unit (7). The salophen base unit was synthesized by heating 7 at reflux with 1,2-benzenediamine (8)i nm ethanol, followed by treating the mixture with UA dihydrate to afford the uranyl-bis-acridine metallointercalator (9). For details of the complete synthesis, please see the Supporting Information. The stock concentration for staining experiments was 250 mm in DMSO. For staining compounds as reference measurements, we selected compounds with ak nown chemistry and specific DNA binding modes;commercial drug cocktails of unknown types of interactions with DNA were neglected. For UA, as olution of 2% UA in Milli-Q water was used, which was filtered through a0 .2 mmp olytetrafluoroethylene (PTFE) membrane. Among the huge family of Pt II compounds, the rationale for choosing an appropriate molecule was to have the simplest molecule with the presence of ac ommon intercalating motif, namely, the planar terpyridine subunit. Based on this, we selected DP (Sigma Aldrich;10mm in TE buffer). [30][31][32] NMR spectroscopy:N MR ( 1 Ha nd 13 C) spectroscopy measurements were performed by using either aB ruker Avance III 400 spectrometer ( 1 H, 400 MHz; 13  . Ac onstant spray voltage of 4.7 kV and ad imensionless sheath gas of 5w ere applied. The capillary temperature and the S-lens RF level were set to 320 8Ca nd 62.0 V, respectively.T he samples were dissolved to ac oncentration of 0.05 mg mL À1 in am ixture of THF/ MeOH (3:2) containing 100 mmol sodium trifluoroacetate (NaTFA) and infused with af low of 5mLmin À1 .
Association constant (K a )m easurements of BAU:Q uantification of the association constant of BAU toward dsDNA was performed by means of UV/Vis absorbance titration following two different procedures. In the first set of experiments, af reshly prepared solution of BAU (45 mm in 11.5 %D MSO and 88.5 %T E-Mg buffer (20 mm Tris, 2mm ethylenediaminetetraacetic acid (EDTA), 12.5 mm MgCl 2 ,p H8)) was transferred into aq uartz cuvette. Subsequently,s mall volumes (2-10 mL) of as olution of dsDNA of known concentration were subsequently added to the solution of ligand and incubated for 5min, followed by recording of the UV/ Viss pectra. Dilution of the ligand was taken into account during data analysis. In the second set of experiments, af reshly prepared solution of BAU (40 mm in 10 %D MSO and 90 %T E-Mg buffer) was divided into different aliquots and individually mixed with aknown amount of dsDNA and incubated for 5min, followed by recording of the UV/Vis spectra. In each sample, the volume was maintained constant (280 mL). All UV/Vis absorbance spectra were recorded by using aC ary Series UV/Vis spectrophotometer,A gilent Te chnologies. For these experiments, a5 438 bp bacterial plasmid (109Z5 [33] ) was used as dsDNA. The concentration of dsDNA stock solution (expressed as bp concentration) was determined by means of UV/ Vism easurements in TE buffer (20 mm Tris, 2mm EDTAp H8). The affinityc onstants were determined from changes in the absorbance, according to ar eported equation, [34] which was derived from the previously reported neighbor-exclusion model. [35] DNA origami as ac alibration tool for TEM imaging:W edesigned a2 DD NA origami nanoplate as at est object for the scattering contrast of dsDNA ( Figure 1). Our origami structure contained various DNA features of interests, such as side arms (4 nm wide, 27 and 43 nm long), two cavities in the middle (4 and 8nmw ide, 19 nm long), and a( flexible) individual dsDNA (2 nm wide) molecule loop at the bottom. Incorporation of these features into the design made it convenient to evaluate if various contrast agents could visualize individual DNA molecules and more extended DNA structures. For details of the origami design and its characterization, we refer to our previous work. [36] TEM sample preparation:C ommercial carbon-coated TEM grids (nominal 3-4 nm thin carbon supported by a5 -6 nm formvar layer;E lectron Microscopy Science, USA) were used to support the origami nanoplates. The origami solution (4 mL, 5nm)w as applied to freshly glow-discharged TEM grids and incubated for 2min. Grids were washed with ultraclean Milli-Q water to remove unadhered origami plates. Next, without drying the sample, the staining compound was pipetted onto the TEM grids and left to react for 1min. Finally,t he grids were thoroughly washed to remove all chemicals from the carbon surface. Removing the residual stain in the last step is ak ey point to avoid generation of contrast by the common negative staining protocols (for which the thickness variation of the stain layer across the biomolecule creates detectable mass-thickness contrast). In the current work, the observed contrast was instead the consequence of direct physiochemical interaction of the staining chemicals with DNA. Following the above TEM preparation protocol, we obtained uniform distributions of origami nanoplates with ah igh density on the TEM grids, which allowed facile investigation of the staining compounds;t his is a TEM imaging:F or S/TEM imaging, we used aF EI Titan microscope equipped with ap ostspecimen aberration corrector operating at an acceleration voltage of 300 kV.I maging at 300 kV was preferred over 80 kV for lower ionization damage to DNA. AH AADF detector (Fischione, USA) at ac amera length of 28.3 cm was used for obtaining mass-thickness-dominated contrast (Z-contrast). For quantitative comparison, all STEM images were acquired with the same imaging parameters (convergence angle, dwell time, pixel size) and postprocessing for contrast evaluation. In TEM mode, the third-order spherical aberration (C3) coefficient was set to zero in the image corrector to minimize information delocalization.