Oriented Soft DNA Curtains for Single Molecule Imaging

Over the past twenty years, single-molecule methods have become extremely important for biophysical studies. These methods, in combination with new nanotechnological platforms, can significantly facilitate experimental design and enable faster data acquisition. A nanotechnological platform, which utilizes flow-stretch of immobilized DNA molecules, called DNA Curtains, is one of the best examples of such combinations. Here, we employed new strategies to fabricate a flow-stretch assay of stably immobilized and oriented DNA molecules using protein template-directed assembly. In our assay a protein template patterned on a glass coverslip served for directional assembly of biotinylated DNA molecules. In these arrays, DNA molecules were oriented to one another and maintained extended either by single- or both-ends immobilization to the protein templates. For oriented both-end DNA immobilization we employed heterologous DNA labeling and protein template coverage with the anti-digoxigenin antibody. In contrast to the single-end, both-ends immobilization does not require constant buffer flow for keeping DNAs in an extended configuration, allowing us to study protein-DNA interactions at more controllable reaction conditions. Additionally, we increased immobilization stability of the biotinylated DNA molecules using protein templates fabricated from traptavidin. Finally, we demonstrated that double-tethered Soft DNA Curtains can be used in nucleic acid-interacting protein (e.g. CRISPR-Cas9) binding assay that monitors binding location and position of individual fluorescently labeled proteins on DNA.


Introduction
Dynamic protein-nucleic acids (NA) interactions play a crucial role in the regulation of many cellular processes. Currently these problems are widely investigated using advanced microscopybased methods that enable direct monitoring of NA-protein interactions at the single-molecule (SM) level in real time. Information obtained from these experiments is crucially important for building mechanistic models of diverse reactions. 1,2 Nano-or micro-scopic platforms combined with microscopy techniques become very popular and allow accessing information that is otherwise hidden. [3][4][5] However, most of the SM techniques cannot be parallelized and are often technically challenging. Therefore, new high-throughput platforms for SM imaging of protein-NA interactions are in high demand. 6 One of the best combinations of SM methods with a nanotechnological platform, was the development of the Deoxyribonucleic acid (DNA) Curtains platform. It enabled high-throughput SM imaging by employing nano-engineering, microfluidics, supporting lipid bilayers (SLB) and SM microscopy. [6][7][8] This platform utilizes an inert lipid bilayer, which passivates the otherwise sticky surface of the flowcell channel, and mechanical barriers to partition the lipids. Biotinylated DNA molecules that are anchored on the biotinylated lipids via neutravidin (nAv) can be manipulated using hydrodynamic force. Another similar recently developed platform is called DNA skybridge. 9 It utilizes a structured polydimethylsiloxane (PDMS) surface for DNA immobilization and a thin Gaussian light sheet beam parallel to the immobilized DNA for visualization of DNA and protein interaction at the SM level. The original DNA Curtains platform demonstrated great benefits for studies of many different NA-interacting proteins. However, the original DNA Curtains are less stable and more expensive fabrication-wise than the platform described in this and our previous work. 10 The skybridge platform contains stably immobilized DNA molecules, but it utilizes rather unusual phenomena for visualization of fluorescently labeled DNA and proteins. However, it is an interesting alternative to the existing DNA Curtains platform.
Recently we demonstrated that streptavidin (sAv) patterns on the modified coverslip surface can be utilized to fabricate biotinylated DNA arrays. 10 The design of the protein patterns on the modified surface ensures predefined distribution and aligning of the biotinylated DNA molecules on the narrow line-features (> 200 nm). We refer to these aligned molecules as Soft DNA Curtains. The application of hydrodynamic buffer flow allows extension of the immobilized DNA molecules along the surface of the flowcell channel. These Soft DNA Curtains permit simultaneous visualization of hundreds of individual DNA molecules that are aligned with respect to one another and offer parallel data acquisition of diverse biological systems. We showed that Soft DNA Curtains are easy to fabricate in any laboratory having an access to an atomic force microscope (AFM) and objective or prism-based total internal reflection fluorescence microscopy (TIRF).
One of the drawbacks of our previous work was that the double-tethered Soft DNA Curtains had no defined orientation of both-end biotinylated DNA molecules. Such DNA molecules could bind to the sAv line-feature in any direction. Random orientation of DNAs would not create a huge problem because one end of the DNA molecule could be fluorescently labeled and this labeling would allow us to post-orient DNA molecules during data analysis. However, this procedure introduces an extra complication of the experiment.
Here we fabricated the uniformly oriented double-tethered DNA Curtains using heterologous labeling of the DNA molecules by biotin and digoxigenin (dig). We confirmed the defined orientation of DNA molecules using a fluorescent tag introduced asymmetrically to the DNA molecule. These improvements allowed us to demonstrate that double-tethered Soft DNA Curtains can be used in NA-interacting protein binding assay that monitors binding location and position of fluorescently labeled CRISPR-Cas9 proteins on DNA. The well-controlled fabrication procedure of high-quality protein templates was achieved using a portable printing device (PPD) developed especially for this purpose. We increased stability of the immobilized DNA molecules using a more stable alternative to sAv called traptavidin (tAv) 11,12 as an ink for the fabrication of protein templates.

Materials and methods
Chemicals and Materials. Silicone elastomer Sylgard 184 (Dow Corning, Midland, MI, USA) was used for lift-off microcontact printing (µCP) stamp production. For Si master structure production, the gold coated silicon wafers (a 20 nm-thick Au film and a 2 nm Ti adhesion layer, Ssens BV, The Netherlands) were used. Before use, substrates were cleaned in SC-1 solution: ultrapure water, 30% hydrogen peroxide (Carl Roth GmbH, Germany), 25% ammonia solution (Carl Roth GmbH, Germany) at 5:1: Production and purification of proteins. His-tagged tAv was produced and purified according to the published protocol. 11,12 E. coli BL21(DE3) cells were transformed with pET21a tAv plasmid, plated onto Luria-Broth (LB)-Carbenicillin agar plates and incubated at 37 °C overnight. An overnight culture in LB-Ampicillin was grown out of a single colony with shaking 220 r.p.m. and 37 °C. The overnight culture was diluted 100-fold into LB-Ampicillin medium, grown at 37 °C until OD 600 0.9, and protein expression was induced with 0.5 mM isopropyl-β-Dthiogalactopyranoside for 4 h at 37 °C. Cells were collected by centrifugation at 5000g and 4 °C for 10 min. The cell pellet was resuspended in a lysis buffer (300 mM NaCl, 50 mM Tris, 5 mM EDTA, 0.8 mg/mL lysozyme, 1% Triton X-100 (pH=7.8, at 25 °C)) and put on a rocker at 80 r.p.m. at room temperature for 20 min. Pulsed sonication of the cell pellet on ice at 30% amplitude was performed afterwards for 10 min. Centrifugation at 27000 g and 4 °C for 15 min followed by washing of the inclusion body pellet in a wash buffer (100 mM NaCl, 50 mM Tris, 0.5% Triton X-100 (pH=7.8, at 25 °C)) was repeated three times. Isolated inclusion bodies were dissolved in 6 M guanidinium hydrochloride (pH=1.5, at 25 °C) and then spun at 17700g and 4 °C for 20 min. Protein precipitation using solid ammonium sulfate was then carried out in order to precipitate tAv from their refolds. The precipitate was resuspended in a minimal volume of PBS at room temperature, centrifuged at 14000g and 4 °C for 5 min and the excess of ammonium sulfate was removed by running the supernatant through a NAP-25 column (GE Healthcare). The tAv was purified using HiTrap chelating column (GE Healthcare) charged with Ni 2+ equilibrated with a equilibration buffer (300 mM NaCl, 50 mM Tris-hydrochloride (pH=7.8, at 25 °C)). Protein was eluted with elution buffer (300 mM NaCl, 50 mM Tris, 0.5 M Imidazole (pH=7.8, at 25 °C)). The fractions containing tAv were dialyzed into PBS at 4 °C and concentrated by ultrafiltration using a 9 kDa MWCO centrifugal concentrator. The final yield of purification was 3 mg of tAv per 1 liter of initial culture.
Wild-type Streptococcus pyogenes (Sp) Cas9 was expressed and purified as published previously. cooled to room temperature. Subsequently, T4 DNA ligase (ThermoFisher Scientific) was added, and the reaction mixture was incubated at room temperature for 2 h. After the reaction was complete, the DNA ligase was inactivated by heating to 70 °C for 10 min, the excess oligonucleotide was removed using a CHROMA SPIN TE-1000 column (Clontech, USA), and the purified DNA was stored at -20 °C.
For the insertion of an ATTO647N-labeled oligonucleotide complementary to the position 14711 bp from the biotinylated end of the λ DNA we employed the previously described strategy 14 and followed the more recently described procedure. 15  Biotinylated 5 kb long DNA was synthesized by PCR using ΦX174 RF1 DNA (ThermoFisher Scientific) as a template and oligonucleotides 5' biotin-CGAAGTGGACTGCTGGCGG-3' and 5'-CGTAAACAAGCAGTAGTAATTCCTGCTTTATCAAG-3'as primers. The product was purified using the GeneJET PCR Purification Kit (ThermoFisher Scientific).
Fabrication and characterization of a silicon master. Si masters were fabricated and characterized according to the previously published procedure. 10 Fabrication of the Si master involves formation of a self-assembled monolayer (SAM) from 1-eicosanethiol (HS-C 20 , AlfaAesar), 16 surface patterning by the nanoshaving lithography technique using an AFM (NanoWizard3, JPK Instruments AG, Germany) and wet chemical etching. [17][18][19] Characterization of the Si master was performed using an upright optical microscope BX51 (Olympus, Japan) and the AFM, operating in AC mode.

Characterization of printed protein features.
The width and morphology of the printed protein features on the glass surface were analyzed with an AFM in buffer C. For that, the glass sample was mounted into the ECCel (JPK Instruments AG, Berlin, Germany) and imaged using the QI-Advanced mode. Before each measurement the probe sensitivity and spring constant were calibrated using the contact-free calibration routine (based on the thermal spectrum of cantilever oscillations) built into the AFM software. The setpoint for measurements was set to 1.5-2 nN tip pushing force.
Protein nanopatterning by lift-off µCP and the portable printing device. Flat PDMS elastomer stamps for protein lift-off µCP were fabricated according to the previously published procedure. 10 Briefly, the prepolymer and curing agent (10:1 ratio w/w, Sylgard 184 kit) were thoroughly mixed, degassed in a vacuum desiccator (30 min), poured into a plastic Petri dish and cured in an oven (65 °C for 14 h). The thickness of the cast PDMS elastomer was ~2 mm. The PDMS surface that was in contact with the Petri dish was treated as the flat one.
The lift-off µCP was performed similarly to the published procedure. 10,20 The PDMS elastomer (5 x 5 mm 2 dimensions) was immersed in isopropanol for 10 min, held by tweezers and dried for 15 s, placed on a plastic Petri dish and dried for another 10 min. The Si master was immersed in isopropanol (20 min), held by tweezers and dried, and cleaned by air plasma (5 min, ~500 mTorr, high-power mode, PDC-002, Harrick, USA). To homogeneously cover the PDMS surface with a film of the protein ink the PDMS stamp was placed in a clean plastic Petri dish with its flat side facing up. Then a 60 µL drop of specified protein solution (in buffer A) was placed on the PDMS surface, mixed with the tip of pipette, and kept for 10 min. After incubation the protein ink was removed from the PDMS stamp by sucking it out with the pipette tip. Then, it was held with tweezers and washed with 5 mL of buffer A using a 1 mL pipette, ~50 mL ultrapure water using a wash bottle, and dried under N 2 gas stream.
For the printing procedure we build a semi-automated printing machine, which allowed us to apply different printing pressure (PP, see SI file for the detailed description). First, the cleaned Si master was placed on the silicon rubber on the bottom of the printing machine (SI Fig. 2A). The surface of the glass coverslips was modified in the same way as described previously, 21 using the biotin-PEG:methoxy-PEG (bt-PEG:m-PEG) ratio 1:10 (w/w). In order to minimize non-specific protein adsorption to the surface, we performed a second round of PEGylation with the short NHSester PEG molecules (333 Da) according to the published procedure. 22 Next, the glass slide with the PDMS stamp was removed from the glass coverslip using the tweezers and discarded. The patterned glass coverslip was assembled into the flowcell, which was prepared as described earlier. 10 The Si masters were reused for the lift-off µCP multiple times and in between the experiments they were stored in 100% isopropanol solution.
TIRF microscopy. The employed home-build TIRF microscopy setup was described previously. 10

Results and discussion
To utilize the DNA Curtains platform for complex protein-NA interaction studies, it is required to obtain double-tethered DNA molecule arrays with defined orientation. Therefore, in this work, we upgraded the existing Soft DNA curtains platform 10 and further optimized its fabrication method by introducing several new steps that made the platform more stable and more controllable.
Optimization of DNA arrays fabrication. Here, we expanded our previous work 10   line-features. Such protein array templates can be considered as a soft functional element and therefore we term our platform the Soft DNA Curtains.
First, to achieve protein array templates allowing desired distribution of biotinylated DNA molecules, we fabricated Si masters using the previously described procedure 10 with line-widths ranging from ~200 nm to 1 µm and line-spacing corresponding to ~75% (~12 µm) of the mean extension of λ DNA. 6 The dimensions of Si masters' patterned area were from 0.5 × 0.9 mm 2 to 2.5 × 1.2 mm 2 . Typical line-depth of the Si masters used in this work is ~200 nm. SI Figure 1 shows the Si masters' overall optical images, line-width and -depth measurements using AFM. SI Table 1 summarizes the characteristics of Si masters that were measured by AFM.
To improve the patterning reproducibility and to control the PP in the lift-off µCP, we built the PPD, which is similar to the previously published device. 25 However, our PPD was assembled from commonly used parts in an optics laboratory and does not require sticking of the PDMS stamp to the moving piston (SI Fig. 2). In addition to that, our PPD employs PDMS stamp attachment to the glass slide surface, which helps to keep the stamp flat ( Fig. 1A and SI Fig. 2B). It is worth noting that a similar effect (PP vs. protein array quality) could be achieved by changing the lift-off µCP printing time, but that would tremendously increase its duration. In our experiments the pressure applied by this easy to use and relatively simple device ranged from ~7 N/cm 2 to ~13.5 N/cm 2 . To keep it simple, instead of N/cm 2 we chose to report the PP in terms of the position of the syringe 3 piston (SI Fig. 2C). We calibrated this value and the results are given in SI Figure 3. To test the quality of sAv line-features printed using PPD on the m-PEG/bt-PEG (10:1 w/w) modified glass coverslip surface, we immobilized 5 kb long biotinylated double-stranded DNA (dsDNA) (Fig. 2A).
TIRF images showed DNA molecules mainly immobilized on the sAv line-features, but their density was dependent on the applied PP ( Fig. 2B and SI Fig. 4). Quantitatively the best results were obtained with a PP of 0.6 mL (SI Table 2). We noticed that at 0.45 mL and especially at Results of these measurements showed no evidence of either line breaks or line-width change (Fig.   2B). Therefore, we concluded that too high PP (starting at 0.45 mL) inactivates a fraction of sAv on the surface, which results in reduced binding of biotinylated DNA.
Another parameter that we assessed in order to optimize the immobilization of biotinylated DNA molecules was the sAv concentration during PDMS elastomer inking under constant PP of 0.6 mL.
In these experiments, the sAv ink concentration ranged from 0.013 to 0.027 mg/mL. We performed lift-off µCP with sAv ink, assembled the flowcell, and immobilized the biotinylated 5 kb long DNA molecules. Acquired TIRF images showed similar results as previously observed 10 -the highest quality protein templates were fabricated at the moderate sAv concentration of 0.017 mg/mL (SI Fig. 5 and SI Table 2). The optimal range of sAv concentration was rather narrow, since concentrations 52% higher or lower than optimal concentration immediately gave worse results.
The obtained optimal sAv ink concentration under 0.6 mL PP is similar to the optimal sAv concentration without applied PP. 10 However, in contrast to the manual lift-off µCP performed by hand without application of the pressure, the PPD device allows production of a consistent and high-quality protein template across the entire patterned area (SI Fig. 6). This is the main advantage of lift-off µCP using PPD in comparison to the manual procedure.
In our previous work, we showed that the number of single-end tethered biotinylated λ DNA molecules decreased slowly over time, with a half-life of > 2 h, 10 in a good agreement with the expectations for a high-affinity biotin-sAv interaction. However, a recently developed super-stable variant of sAv -called traptavidin (tAv) 11 -should allow us to observe immobilized biotinylated DNA molecules for an even longer period of time. We verified our tAv functionality (see SI file and SI Fig. 7) and then tested whether tAv is suitable for fabrication of the fixed DNA molecule arrays.
For these experiments we used the lift-off µCP with variable tAv concentration, which ranged from 0.015 mg/mL to 0.06 mg/mL, at constant PP of 0.6 mL. Once line-features were formed and the flowcell was assembled, we immobilized biotinylated 5 kb long DNA molecules. TIRF images showed that the optimal tAv concentration was ~0.03 -0.02 mg/mL (SI Fig. 8C-D). Both at lower and higher concentration than the optimal we observed more DNA bound in the interline areas or lower DNA density on line-features (SI Fig. 8A-B and E). This visual inspection was also well reflected by the quantitative QF-based characterization (SI Table 2). However, the absolute density of DNA molecules immobilized on the line-features seems to be lower than that with sAv, but this could be rationalized by the different activity of tAv, which requires a higher concentration of DNA molecules to achieve similar densities. Here we decided to use the same concentration of sAv and tAv for the sake of consistency.
To test whether tAv indeed allows observing the immobilized DNA molecules for a longer period of time than sAv, we fabricated single-tethered Soft λ DNA Curtains on sAv (0.017 mg/mL) and tAv (0.03 mg/mL) line-features at a constant PP of 0.6 mL. Next, we performed the stability test of bound DNA molecules by performing the acquisition cycles of image series every 20 minutes.
During these cycles 20 frames were acquired with the 1 mL/min buffer flow and 20 frames without the flow. Between the acquisitions there was no buffer flow applied. Acquired TIRF images showed that the number of full-length single-tethered DNA molecules anchored to the surface decreased slowly over time, with the half-life > 2 h for both sAv and tAv (SI Fig. 9). DNA molecules were dissociating slower from tAv than from sAv, and that is in good agreement with the expectation for the lower biotin dissociation constant of tAv. 11 Namely, after 2 h of observation time only ~20% of DNA molecules were dissociated from tAv line-features, while ~40% of them were dissociated from sAv (Fig. 2C). These results proved tAv to be better suited for the long-lasting experiments, which require observation of the same DNA molecules, and can be more beneficial for all types of DNA Curtains.
Assembly and characterization of double-tethered Soft DNA Curtains. The patterns of our platform utilize tAv functional elements, and an overview of the general design is presented in Figure 1C. Line-feature width of the tAv patterns was assessed in order to optimize assembly of the doubletethered Soft DNA Curtains. Three different patterns of tAv were fabricated on the separate coverslips using Si masters #4, #5 or #6 at PP of 0.6 mL. The double-tethered Soft DNA Curtains were assembled on the patterned coverslips as described above (Fig. 3). The anchoring efficiency was tested for tAv patterns made with variable line-widths, but constant line-spacing (i.e. ~12 µm).
As expected, the wider lines allowed us to achieve more efficient anchoring. Percentages and densities (average number of both-end anchored DNA molecules per line-feature) of both-end anchored DNA molecules are presented in SI Once line-feature width was optimized, line-separation distance of the tAv patterns was assessed.
Four different patterns of tAv were fabricated on the separate coverslips using Si masters #5, #7, #8 and #9 at PP of 0.6 mL. The double-tethered Soft DNA Curtains were assembled on the patterned coverslips as described above (Fig. 3). This time, the anchoring efficiency was tested for tAv patterns made with variable line-separation distances, but constant line-widths (i.e. ~0.5 µm). As expected, the most efficient anchoring occurred with the 12 and 13 µm line-spacing distances.
When line-width and -separation distance were optimized, we tested the orientation of λ DNA fragments on the double-tethered Soft DNA Curtains using a fluorescent tag (ATTO647N at the To tether the dig-labeled end of bt-λ DNA-dig, continuous slow flow of buffer containing biotinylated anti-dig (bt-anti-dig) antibody is applied and DNA molecules are dragged slightly, but do not reach the neighboring line-feature. At the next step, buffer flow rate is increased and the dig-labeled DNA ends encounter the neighboring line-feature, which is now covered with the bt-anti-dig, and the dig-labeled ends become anchored through dig -anti-dig interaction. C) Cartoon illustrates the doubletethered bt-λ DNA-dig molecule after dig-labeled end tethering. TIRF images shows that those DNA molecules that remain stretched without buffer flow (-flow) were successfully both-end tethered. D) Cartoon illustrates the DNA immobilization strategy and internal ATTO647N tag, which was located at 14711 bp from the biotinylated DNA end. TIRF images shows SG stained DNA molecules in the absence of buffer flow. Excitation wavelength and emission channel is indicated above each image. Histogram showing the distribution of ATTO647N locations that were determined by fitting the images to 2D Gaussian functions. Si master #8. position 14711 bp) introduced asymmetrically to the specific position of the λ DNA. As mentioned above, we used differential chemistries (biotin on the left end, and dig -right end) to tag two ends of the DNA (Fig. 3C). Therefore molecules within the double tethered curtains should be immobilized in a defined orientation. To confirm that the DNA was oriented correctly, we assembled double-tethered Soft DNA Curtains from the ATTO647N labeled λ DNA (bt-λ DNA ATTO647N-dig) as described above (Fig. 3C). The ATTO647N labels were present at a single location within the DNA molecules and aligned with one another. Their mean position was found to be 14.1 ± 0.02 kb (N = 50) from the biotinylated DNA end. This result coincided well with the expected location and practically no ATTO647N labels were observed at other locations.

Deployment of double-tethered Soft DNA Curtains for visualization of protein-nucleic acid
interactions. To demonstrate that double-tethered Soft DNA Curtains can be utilized to visualize protein-DNA interaction, we selected previously characterized Cas9 nuclease from the CRISPR-Cas system of S. pyogenes (Sp), that is involved in bacterial defense against foreign invading DNA and has been adopted as a genome editing tool. 26 (Fig. 4C). There were 172 DNA molecules and 902 binding events of proteins in total monitored during this experiment. As in Figure 4, the DNA-bound Cas9-ATTO647N demonstrated broad binding distribution along the full length of the λ DNA. However, in total protein spend more time at the target site (Fig. 4C). There were more binding events on the left side of the DNA, which likely is dictated by higher GC content. Binding events on target had significantly longer dwell times (some of them lasted as long as the acquisition time 150 s) than on non-target binding, which lasted < 20 s (Fig. 4D). On average target binding events lasted for ~52 s and non-target binding events ~7 s (SI Fig. 11). We note, that this experiment was conducted in the absence of Mg 2+ ions in the imaging buffer. Therefore, obtained results suggest that Mg 2+ ions are not essential for specificity of SpCas9 target recognition. This experiment provides direct evidence that the double-tethered Soft DNA curtains can be utilized to visualize the DNA and fluorescently labeled proteins' interaction at the SM level.

Conclusion
The oriented double-tethered DNA Curtains allow intuitive and simple parallel examination of hundreds of SM in a single TIRF experiment. In this work, we showed that it is possible to fabricate oriented and aligned Soft DNA Curtains using high-quality protein template-directed assembly of biotinylated DNA molecules and hydrodynamic force. Also, we showed by the Cas9 binding experiment that our oriented double-tethered Soft DNA Curtains are suitable for visualization and characterization of individual NA-interacting protein studies. We believe that each of the optimizations and improvements described in this work will be useful for SM studies and make the Soft DNA Curtains platform better characterized.