Single molecule DNA origami nanoarrays with controlled protein orientation

Nanoscale engineering of biomolecular arrays can facilitate the fabrication of next-generation (bio)molecular devices and assays.¹ In this regard, the organization of functional biomolecules on solid substrates with a nanoscale spatial resolution with control at the level of individual molecules is of fundamental importance in order to develop biomimetic platforms capable of high-throughput single-molecule investigations.^2,3 Nanoscale biochips with such capabilities can surpass the current limits and would permit the monitoring of biochemical processes in real time, characterization of transient intermediates, and the measurement of the distributions of molecular properties rather than their ensemble averages.⁴ In particular, the precise placement of proteins within a nanoarray is of importance for the fabrication of biomimetic surfaces to be employed in cell adhesion and spreading^5–7 investigations, drug discovery,⁸ as well as photobiophysical⁹ and biosensing^10–12 applications.

Different techniques have been employed for fabricating biomolecular nanoarrays, including electron-beam,^3,13,14 focused-ion-beam (FIB),¹⁵ nanoimprint¹⁶ and colloidal lithography,^17,18 thermochemical scanning probe lithography,¹⁹ dip-pen,²⁰ photo-lithography,²¹ polydimethylsiloxane (PDMS) imprint,²² ink-jet microdeposition,²³ polymer brushes,^7,24 particle self-assembly,^25–27 and bottom-up DNA origami.^{5,6,8,27–38} These have opened new avenues for the development of high-throughput biosensing as well as for fundamental investigations of molecular interactions, including the protein–DNA interactions.^39,40

Among different approaches for molecular organization, the use of DNA origami has shown to be an excellent bottom-up strategy due to high yielding self-assembled product in a single pot reaction and addressable sites for facile and precise functionalization with less than 6 nm (out of plane) resolution^30,41,42 subnanometer intermolecular distances^43,44 Moreover, spatial orientation control of single dyes was achieved using DNA origami.³⁸ Consequently, DNA origami has been utilized as nano-breadboards where molecular components such as proteins can be placed with single molecule control and organized into target configurations.^{29,31,36,45–50} In particular, Marth et al. have shown positional control of proteins on a DNA origami, as well as orientational control via residue-specific incorporation of useful bioorthogonal reaction handles using a reprogrammed genetic code.²⁸ Based on the combination of both bottom-up DNA origami and top-down strategies, a photolithography strategy for producing fluorescent nanoarrays for photonics was devised.³² More recently, a nanoarray strategy entirely relying on bottom-up self-assembly of microparticles and DNA origami was demonstrated and applied for super-resolution studies.²⁷ Furthermore, our group demonstrated the fabrication, via a one-step lithographic process, of DNA origami nanoarrays³³ to investigate the role of biomimetic surfaces in cancer cell spreading.⁶ 

Here, we present the fabrication of single protein nanoarrays via selective assembly of protein–DNA origami hybrids, arranged on nanopatterned surfaces, with control over the orientation of the protein tethered to the DNA nanostructures. The lattice-like nanoarrays were patterned on a metal-coated substrate via a single lithographic step using focused-ion beam milling (FIB).⁵¹ Green fluorescent protein (GFP) was modified at a specific residue with an oligonucleotide via a strain-promoted azide-alkyne coupling (SPAAC),^28,52,53 which was then hybridized to a complementary sequence at a specific position on the DNA origami. The DNA nanostructures were subsequently size-selected and selectively immobilized at the bottom of the nanoapertures of the aforementioned nanoarrays, via a biotin-streptavidin linkage strategy on silanized glass. Our strategy presents a facile method for immobilizing single proteins in a nanoarray with orientational control and is of general applicability for the investigation, via fluorescence imaging, of biomolecular events with single-molecule resolution.

All oligonucleotides and modified oligonucleotides were purchased from integrated DNA technologies (IDT). M13mp18 viral DNA was purchased from Tilibit. Streptavidin, Dulbecco's phosphate-buffered saline (DPBS), and tris-acetate-EDTA (TAE) buffer pH 7 were purchased from ThermoFisher. MgCl₂ and NaCl were purchased from Fisher Scientific. Glass coverslips size one were purchased from Agar Scientific.

Glass coverslips were cleaned intensively using our previously published protocol as follows.³⁵ Coverslips were placed in a Teflon rack, then rinsed with Milli-Q water (mQ) and carefully sonicated in piranha solution (3:1, sulfuric acid and hydrogen peroxide) inside a small beaker container surrounded by ice for 2 h. After this, coverslips were rinsed with mQ and sonicated in mQ for 10 min. Then, coverslips were sonicated in acetone for 10 min, rinsed with mQ, sonicated in ethanol for 10 min, and finally, coverslips were blown dry with argon. After this, cleaned coverslips were loaded into a metal holder for thermal evaporation in a vacuum. Metal layers of ∼1.5 nm chromium layer (the deposition rate of ∼0.1 Å/s) as an adhesion layer and ∼3.3 nm gold layer (the deposition rate of ∼0.4 Å/s) on top were deposited as measured by the crystal sensor. After this step, samples were ready for nanopatterning using the FIB equipment.

Nanoapertures were patterned on the prepared metal-coated glass substrates using the FEI^TM Quanta scanning electron microscope (SEM) and FIB system.^33,54 Apertures' shapes were circles with a designed diameter of 200 nm with a spacing distance of 2 μm. Nanopatterned arrays were drawn in the FEI^TM Quanta software and milled with a gallium ion beam of 30 kV/50 pA (or 0.1 nA) with a dwelling time tuned to reach the bottom of the nanoaperture. Several arrays of at least 256 nanoapertures were produced in a single substrate. The patterned substrates were characterized immediately using the scanning electron microscope (SEM) capabilities with 5 kV/47 pA, and samples were scanned in atomic force microscopy (AFM) afterward. Patterned substrates were activated with oxygen plasma prior to the biofunctionalization and immobilization of DNA origami.

The DNA origami used was a triangular-shaped nanostructure called the Rothemund triangle.^30,33,35 This structure was a single-layer DNA sheet with a 120 nm side length. It was synthesized by mixing and annealing more than 200 short single-stranded DNAs (called staples) and a 7249-nucleotide circular single-stranded DNA (called scaffold). For simplicity, single-stranded DNA is referred to as ssDNA. During origami preparation, some staples were extended to allow the hybridization of functional components (see the list of sequences in the supplementary material). These components were designed to protrude from opposite faces of the DNA origami by selecting DNA duplexes whose terminal ends were closer to the face. The staples (100 μM in 1× TAE buffer), the 12 modified staples for biotin anchors, and the biotin-functionalized ssDNAs were mixed in 5:1, 5:1, and 500:1 ratios relative to the scaffold in 1× TAE buffer/12.5 mM Mg²⁺ buffer (called annealing buffer). For ATTO 488 photobleaching experiments, the capturing dye staple and the dye-labeled ssDNA were added in a ratio of 10:1 and 50:1. For protein attachment, the capturing protein staples were added in a ratio of 10:1, while DNA-GFP was added as described in Sec. II F. The DNA mixture was annealed in a PCR machine (Hybrid Sprint PCR Thermal Cycler, Thermo Scientific) by first heating it to 90 °C for 5 min and cooling it down gradually at the rate of 0.2 °C per min until reaching room temperature. The assembled origami structure was then purified by using microcentrifuge filters (100 kDA MWCO; Millipore Amicon Ultra 0.5) at 10 000 rpm three times, 3 min each. The final concentration of origami was 1.2 nM in 100 μl annealing buffer. The DNA sequences used for DNA origami fabrication are included in the supplementary material.

The GFP variant containing 4-azido-l-phenylalanine (AzF) at residue 204 is based on the superfolding GFP variant and was produced as described previously.⁵³ GFP protein–DNA conjugation was made via strain-promoted alkyne-azide cycloaddition.⁵⁵ The partner to the azide in GFP is bicyclononyne on ssDNA. Briefly, BCN-functionalized DNA (BCN-DNA) was prepared from 3' terminal end amino-functionalized DNA and BCN-NHS ester [(1 R,8S,9s)-Bicyclo[6.1.0]non-4-yn-9-ylmethyl N-succinimidyl carbonate, Sigma Aldrich].⁵⁵ Azide-functionalized GFP (30 μM) was mixed with BCN-DNA (150 μM) in tris buffer (50 mM; pH 6.8) and let to react overnight in an incubating shaker at 37 °C. The reacted product, i.e., GFP-DNA conjugate, was purified using native (non-denaturing) polyacrylamide gel electrophoresis (PAGE; biorad mini-PROTEAN) as follows. A 10% polyacrylamide gel was made, and GFP-DNA was pipetted into the wells. The gel was run in tris-glycine buffer at 150 V. Using a dark light reader (Clare chemical research), bands corresponding to GFP-DNA were excised from the gel, frozen, and crushed using a spatula. The crushed bands were soaked in 1× TAE for 48 h. The gel was removed from the solution using microcentrifuge filters (Corning; 0.45 μm pore size).

After DNA origami and GFP-DNA were assembled and purified in solution, the origami structure (10 nM) was mixed with GFP-DNA (1 μM) in 1× TAE buffer with 12.5 mM MgCl₂. The solution was incubated on a shaker at 20 °C for 48 h. Excess GFP-DNA was removed by using microcentrifuge filters (100 kDA MWCO; Millipore Amicon Ultra 0.5) at 10 000 rpm for 2 min.

DNA origami-GFP conjugates were characterized using AFM in the fluid. A Bruker dimension icon AFM was used in PeakForce QNM mode with ScanAsyst fluid probes. Origami—either before or after centrifugal filtering—was deposited on freshly cleaved mica. Imaging conditions were critical to unambiguously distinguish proteins attached to the origamis. The imaging was carried out in water with 50 mM Mg²⁺. In a typical experiment, the origamis were localized at a scan rate of 2 Hz with 256 samples/line. When a region of interest was found, the scan rate was reduced to 0.5 Hz and samples/line was increased to 512. The setpoint was minimized to avoid the denaturation of the DNA origami-GFP conjugate. Images were processed in Bruker Nanoscope Analysis software. DNA origami-GFP and patterns were also characterized with AFM in air using the same AFM equipment but with ScanAsyst air tips (tip radius 12 nm) in tapping mode with 512 samples per line and a scan rate of 0.5 Hz. In the case of origamis, the sample solution was deposited onto a piece of freshly cleaved mica and rinsed with water and then immersed in ethanol and dried before AFM measurement.

The substrate was plasma cleaned for 15 min, and subsequently, they were allowed to cool to room temperature. The carbodiimide coupling was performed as follows. 1% CTES (carboxyethylsilanetriol di-sodium salt; Gelest) in 10 mM tris pH 8.3 was cast on the substrate for 1 h in a shaker. Then, the substrate was rinsed with mQ and blow dry with argon. The substrate was baked at 90 °C for 1.5 h. Then, a solution of 100 mM EDC [1-ethyl-3-(3-dimethylaminopropyl) carbodiimide; from Sigma-Aldrich] and 200 mM sulfo-NHS (from ThermoFisher) in mQ was cast on the substrate for 1 h in a shaker and rinsed with mQ leaving a volume of about 100 μl on top of the pattern. A solution containing an amine-functionalized ssDNA (1 μM) and its complementary Cy5-labeled ssDNA (1 μM) in 40 μM MgCl₂ was cast on top of the pattern for 1 h at room temperature. Substrates were rinsed with 1× DBPS and stored overnight. Before total internal reflection fluorescence (TIRF) microscopy, substrates were blow dried with argon.

The substrate was plasma cleaned for 20 min, and subsequently, they were allowed to cool to room temperature. Cloning cylinders were glued (ethyl 2-cyano acrylate; loctite) to the cleaned substrates in order to be used as reaction chambers or wells. 0.1 mg/μl PEG (biotin-PEG-silane, MW 3400, 500 mg; Laysan Bio) in 95% ethanol was added and incubated for 1.5 h. We then rinsed with mQ water. 0.1 mg/mL streptavidin (ThermoFisher) was added and incubated for 0.5 h; we rinsed it with mQ water and exchanged it to 1× TAE buffer/12.5 mM Mg²⁺. 2.4 nM functionalized, biotinylated origami was added and incubated for 0.5 h. Subsequently, the samples were rinsed with 1× DPBS, which we believe help preventing nonspecific adsorption of DNA origamis to the gold nanoarray.¹⁸ The buffer was exchanged back to 1× TAE/12.5 mM Mg²⁺.

Single-molecule monitoring was performed using an LSM710 ELYRA PS.1 in the TIRF mode with a 642 (150 mW) or 488 nm (100 mW) laser excitation. Cy5 nanoarrays were observed with 642 nm at 0.05% laser power and 100 ms camera acquisition time. Single dye nanoarrays were monitored with 5% laser power and 500 ms acquisition time in a 1× DBPS buffer containing 450 mM NaCl (final concentration), 4.15 mM MgCl₂, 2% Tween 20 (from Sigma-Aldrich), Trolox (from Sigma-Aldrich), and oxygen scavenger (1 mg/mL glucose oxidase and 0.4% v/v catalase, from Sigma-Aldrich). For photobleaching experiments, the laser power was increased to 100%. Single protein nanoarrays were monitored with 488 nm at 2%–5% laser power, BP495–550 filter, and 50–100 ms camera acquisition time.

Using ImageJ, we analyzed a TIRF microscopy time lapse (see the supplementary material for representative frames). Initially, a time lapse was loaded into ImageJ as a Stack. Then, bright spots were visually identified. A 2 × 2 pixels region was selected on the bright spot and shifted around the spot in such a way that the maximum intensity was obtained using the ImageJ's Plot Z-axis Profile tool. We judged the presence of fluorescence by comparing the intensity in the expected position of the nanoaperture with the background intensity upon bleaching or the background intensity of the surroundings using the image analysis software ImageJ.

In this work, nanoarrays were initially patterned on a gold/chromium metal-coated glass substrate [Fig. 1(a)] with a single focused ion beam (FIB) step, as described previously.^6,33 In this FIB step, a pattern consisting of nanoapertures with a diameter ø and with lattice periodicity d was milled [Fig. 1(b)]. This strategy allowed the exposure of the glass surface beneath the metal-coating for further passivation and biofunctionalization using DNA origami tile structures. To control the process with single-molecule resolution, DNA origami nanostructures were designed to each display a single fluorescent molecule on one face via a ssDNA anchor (the addressable element) and were then cast and immobilized onto the nanoapertures via molecular anchors on the opposite face using biotin with streptavidin as a linker between the origami structure and the biotin-silanized glass surface [Fig. 1(c)].

A representative AFM image of the nanoaperture array (ø = 200 nm, with d = 2 μm) is shown in Fig. 2(a) with the height profile indicating the successful formation of the nanoapertures. Additional pattern characterization using a scanning electron microscope (SEM) is shown in Fig. SM1 in the supplementary material. We confirmed whether the nanoapertures were available for functionalization by attaching fluorescent dye molecules. Cy5-modified DNA duplexes were immobilized on the nanoapertures' exposed glass via EDC/Sulfo-NHS carbodiimide coupling (see Sec. II). TIRF microscopy confirmed the formation of a Cy5 nanoarray with ∼100% yield [Fig. 2(b)].

To probe whether our nanoarray fabrication strategy was suitable for single molecule monitoring, we immobilized triangular DNA origami nanostructures designed to present a single dye (see Sec. II). An ATTO 488 dye was selected as its excitation and emission spectra are similar to green fluorescent proteins (GFPs). The actual DNA origami structure is shown in Fig. 3(a). The dimension of the triangular nanostructure is 120 nm per side, by design, so only one structure can, in principle, physically fit per well. Furthermore, the face of the structure presented was defined through biotin anchors [Fig. 1(c)]. In contrast to previous origami immobilization methods,^32,33,56 we chose to use biotin anchors on one origami's face to ensure the dye, and ultimately the protein, would be protruding from the other face and exposed to the buffer solution. The biotin anchors were achieved by hybridizing a unique biotin-functionalized ssDNA with 12 ssDNA anchors on the origami's face. A single ATTO 488-functionalized oligonucleotide was immobilized via DNA hybridization to the single available complementary ssDNA on the triangular DNA origami tile [schematic in Fig. 3(b)].^57,58 Next, the biotinylated DNA origami construct was immobilized onto the nanoarrays by casting them onto nanoapertures, which were previously functionalized with PEG-biotin and streptavidin for passivation (see Sec. II).

Total internal reflection fluorescence (TIRF) microscopy showed that 88% of the nanoapertures had observable fluorescence, hereinafter called bright spots, indicative of nanoapertures occupied by DNA origami nanostructures [Fig. 3(c)]. We observed that bright spots occur in the background too, but the majority were present in the nanoaperture. Photobleaching experiments indicated that 85.9% of bright spots had a single photobleaching step (high to low intensity transition as seen in the representative trace in Fig. 3(c). See also additional traces in Fig. SM2). The histogram of the step bleaching (Fig. SM3) shows that the majority of dyes can be monitored for a time window of at least 20 s. One-step bleaching events strongly suggested that single dye molecules were present in each nanoaperture; thus, single DNA nanostructures were immobilized in each nanoaperture; 3.1% of the bright spots exhibited two-step photobleaching, suggesting the presence of two dyes, and hence, double DNA nanostructure occupancy. 10.9% of the bright spots had diverse single-step transition events, which still suggests single occupancy (see reference note⁵⁹ and representative traces in Fig. SM4). All in all, 97% of nanoapertures with bright spots had single DNA origami structures immobilized. While, in the context of the entire nanoarray, 85% of the nanoapertures had single occupancy of DNA origami structures.

To fabricate single protein nanoarrays, we attached individual green fluorescent protein (GFP) molecules to the DNA origami nanostructure. We have previously engineered GFP to contain azide chemistry close to the chromophore [residue 204; see Fig. 4(a)] using a reprogrammed genetic approach.^28,52,53,55 The incorporation of azide chemistry via the non-canonical amino acid 4-azido-l-phenyalalanine (AzF) provides a means to link the addressing ssDNA to the protein in a single designed site via strain promoted alkyne-azide cycloaddition (SPAAC). This, in turn, generates a homogenous protein–DNA origami structure system, which is essential for monitoring single molecule events; due to the high prevalence of lysine residues on a protein's surface, primary amine attachment processes (i.e., lysine residues) would generate a highly heterogeneous system with each well on the array representing different protein–DNA origami structure configurations and potentially functional effects. GFP with azide at residue 204 (here on called GFP^204AzF) was linked to ssDNA using SPAAC; addressing ssDNA was modified with BCN at its 3′ end allowing SPAAC to occur simply on mixing the two molecules [Fig. 4(a)]. The DNA-GFP product was then purified (see Sec. II). Subsequently, the DNA-GFP conjugate was incubated in solution with the DNA origami, which had a protruding complementary ssDNA that allowed tethering of the DNA-GFP to the DNA origami structure. Previous work has shown that the attachment of BCN ssDNA to GFP^204AzF has little effect on the bulk spectral properties of GFP.²⁸ 

After incubation, GFP^AzF204-functionalized origami nanostructures (GFP-origami hereinafter) were purified. We tested two positions on the nanostructure for the GFP tethering: on the origami edge and the (top) face [schemes in Figs. 4(b) and 4(c), respectively]. AFM images of the GFP-origami structures obtained in liquid and air confirmed the successful functionalization of GFP on DNA origami in a 1:1 ratio; representative AFM images are shown in Figs. 4(b) and 4(c) (see Figs. SM5 and SM6 for additional AFM images). We found that 54.7% of DNA origami had GFP conjugated to the structures from a sample size of 53 across several preparations. Imaging conditions were critical to unambiguously distinguish proteins in liquid attached to the origamis requiring higher ionic concentrations in the imaging buffer (see also Fig. SM7) and selection of appropriate AFM tips (see Sec. II). Moreover, the fluid-mode AFM process is dynamic, where proteins may be displaced in the course of imaging.

We then immobilized the GFP-origami structures into the nanoarrays via the aforementioned biotin–streptavidin method. TIRF microscopy showed that 54% of nanoapertures exhibited fluorescent spots [Fig. 5(a)] in agreement with the yield observed via AFM. A representative single nanoaperture's intensity over time depicting the GFP photo-blinking behavior is shown in Fig. 5(b)—this on/off blinking behavior has been attributed to the charged state of GFP's chromophore, but the exact cause of the intermittent blinking is unknown.^60,61 The intensity traces were in line with a previous single molecule analysis of GFP^AzF204 (see Fig. SM8 for additional intensity traces).⁶² Overall, these results demonstrate that single proteins can be spatially arranged over large areas using our combined fabrication strategy. The lower yield of protein occupancy in the nanoapertures obtained with our DNA origami nanoarrays compared to the aforementioned DNA origami-dye arrays (Fig. 3) is likely ascribable to increased steric hindrance and electrostatic interactions between the protein and DNA origami.

Monitoring single protein molecule events requires overcoming various challenges. The first is represented by the constraint of a single protein molecule to a defined area. We achieved this here by combining the functionalization of DNA origami with individual proteins and their fixed organization in FIB nanopatterned wells for long-term monitoring. As the wells will potentially hold several protein molecules of the size of GFP, to achieve single protein molecules per well, we successfully employed DNA origami nanostructures with direction functionality: one face to bind to the well surface and the other to present a sequence for attaching a single incoming protein. The final challenge is the nature of the protein–DNA nanostructure interface. For single molecule experiments, the conjugation between the protein and the underlying supporting material (in this case DNA origami structure) should ideally be defined, designed, and homogenous so that only a single population of protein species is being observed. Without such control, multiple different orientations and configurations will be observed with the positions of attachments making data interpretation difficult and potentially affecting function (e.g., changing active site access, structure, and dynamics). We have addressed this challenge here by using a reprogrammed genetic code to engineer into a protein new chemistry not present in nature to allow residue-specific linkage to an addressable ssDNA unit via bioorthogonal click chemistry (SPAAC in this case). The result is a nanoarray system where the contents of each well are highly defined and uniform in terms of their molecular arrangement. Direct quantification of the occupancy of the single molecule into the nanoapertures was provided, demonstrating an excellent yield comparable to other nanoarray methods.^27,63

The single-molecule fabrication strategy that we presented is of general applicability for the fabrication of high-throughput and automated addressable protein biochips that can allow the investigation, with single-molecule control, of biomolecular events such as aptamer-biomarker recognition,^35,64 protein–DNA interactions (e.g., CRISPR),⁶⁵ protein–protein interactions,⁶⁶ enzyme (cascade) reactions,^67–69 and general biosensing and biomimetic assays.^48,70,71 For instance, relevant components could be introduced into the DNA nano-breadboard for the development of biosensor nanoarrays supported by current advances in protein engineering.⁷² Moreover, the nanoarray configuration can allow for the development of high-throughput DNA platforms for super-resolution standards.^73,74 Finally, the strategy that we presented can be applied to a myriad of (bio)physical studies that take advantage of the programmability of DNA origami;³⁸ these include tracking cellular forces,⁷⁵ the use of stimuli-responsive DNA-powered structures^76,77 for the construction of in vitro DNA nanodevices,⁷⁸ as well as for the development of quantum-related applications,^79,80 DNA computing circuits,^81,82 and information coding.^83,84

See the supplementary material for the additional detailed AFM, TIRF images, and the DNA sequences.

The authors gratefully acknowledge financial support from the Air Force Office of Scientific Research under Award No. FA9550-17-1-0179. R.E.A.G. was supported by the BBSRC-funded South West Biosciences Doctoral Training Partnership (training Grant Reference No. BB/M009122/1). For the purpose of open access, the author has applied a creative commons attribution (CC BY) license (where permitted by UKRI, “open government license” or “creative commons attribution no-derivatives (CC BY-ND) license” may be stated instead) to any author accepted manuscript version arising.

Disclaimer: The views expressed are those of the authors and do not reflect the official guidance or position of the United States Government, the Department of Defense or of the United States Air Force.' Approved for public release. Case Number: AFRL-2022-3527, cleared on 25 July 2022

The authors have no conflicts to disclose.

Keitel Cervantes-Salguero: Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Visualization (equal); Writing – original draft (equal); Writing – review and editing (equal). Mark Freeley: Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Visualization (equal); Writing – original draft (equal); Writing – review and editing (equal). Rebecca E. A. Gwyther: Investigation (equal); Methodology (equal). Daffyd D. Jones: Resources (equal); Writing – original draft (equal). Jorge L. Chavez: Funding acquisition (equal); Resources (equal). Matteo Palma: Conceptualization (equal); Funding acquisition (equal); Project administration (equal); Resources (equal); Writing – original draft (equal); Writing – review and editing (equal).

The data that support the findings of this study are available within the article and its supplementary material.