Organization of Inorganic Nanomaterials via Programmable DNA Self-Assembly and Peptide Molecular Recognition

Abstract

An interesting alternative to top-down nanofabrication is to imitate biology, where nanoscale materials frequently integrate organic molecules for self-assembly and molecular recognition with ordered, inorganic minerals to achieve mechanical, sensory, or other advantageous functions. Using biological systems as inspiration, researchers have sought to mimic the nanoscale composite materials produced in nature. Here, we describe a combination of self-assembly, molecular recognition, and templating, relying on an oligonucleotide covalently conjugated to a high-affinity gold-binding peptide. After integration of the peptide-coupled DNA into a self-assembling superstructure, the templated peptides recognize and bind gold nanoparticles. In addition to providing new ways of building functional multi-nanoparticle systems, this work provides experimental proof that a single peptide molecule is sufficient for immobilization of a nanoparticle. This molecular construction strategy, combining DNA assembly and peptide recognition, can be thought of as programmable, granular, artificial biomineralization. We also describe the important observation that the addition of 1–2% Tween 20 surfactant to the solution during gold particle binding allows the gold nanoparticles to remain soluble within the magnesium containing DNA assembly buffer under conditions that usually lead to the aggregation and precipitation of the nanoparticles.

The future of nanoscale device fabrication for diverse applications including electronics, tissue engineering, biomedical imaging, drug delivery, catalysis, and photonics will likely require 3D constructs containing inorganic nanomaterials with tunable spacings as well as integrated organic subunits.^1–5 Such fabrication tasks are difficult for existing top-down and bottom-up fabrication strategies.^{6, 7} Another approach is to imitate biology, where there are numerous examples of nanoscale materials that integrate organic molecules for self-assembly and molecular recognition with ordered, inorganic minerals to achieve mechanical, sensory, or other advantageous functions. Using biological systems as inspiration, researchers have sought to mimic the nanoscale hybrid materials produced in nature. Here, we describe a new combination of self-assembly, molecular recognition, and templating, relying on a covalent conjugate⁸ between an oligonucleotide and a high-affinity gold-binding peptide (selected from a combinatorial library).⁹ After integration of the peptide-coupled DNA into a self-assembling superstructure, the templated peptides recognize and bind gold nanoparticles. In addition to providing new ways of building functional multi-nanoparticle systems, this work provides experimental proof that a single peptide molecule is sufficient for immobilization of a nanoparticle. This molecular construction strategy, combining DNA assembly and peptide recognition, can be thought of as programmable, granular, artificial biomineralization.

RESULTS AND DISCUSSION

The basis for our “molecular erector set” is the novel combination of two developing technologies, structural DNA nanotechnology,^{10, 11} and in vitro evolution of peptides for recognition of specific inorganic minerals.⁷ The self-assembling DNA template used in these experiments is based on a 4×4 cross-tile system developed for creation of two-dimensional (2D) nanogrids displaying periodic square cavities.^{12, 13} Assembly of the original structure relies on two DNA tile types (types A and B) composed of nine oligonucleotides each, where each tile features one core oligonucleotide anchoring all four tile arms and the arms carry sticky-end overhangs that are designed to be complementary to exactly one sticky end on the other tile (see the Methods section for details and sequences). An important modification to the original design is the addition of a nick into the core DNA strand anchored in one of the tile types (tile type A). The added nick divides the original 100 nucleotide core strand into two fragments of 58 and 42 nucleotides in length, exposing a 5′ end on the shorter fragment and a 3′ end on the longer fragment. To prove the ability of individual peptides to recognize and bind gold nanoparticles, a single copy of the peptide was added to each A tile in the opening created by the nick, shown schematically in Fig. 1a.⁸ To ensure the entire peptide is free to participate in the gold recognition process, a spacer, in the form of two thymine nucleotides, was added between the peptide and the end of the normal core strand DNA sequence. The 5′ end of the TT spacer was labeled with an amino-terminated phosphoramidite, with the amine facilitating coupling of the oligonucleotide with the C-terminal end of the peptide. One noteworthy feature of this self-assembling DNA system is the corrugated design strategy. In this design, adjacent tiles flip with respect to each other thereby improving the formation of large pieces of nanogrid through elimination of the additive effects of tile curvature.^{12, 13} The corrugation feature affects the orientation of each peptide as well. The location of the nick along the core strand ultimately determines the location of the covalent bond between the peptide and the oligonucleotide. The location selected in this experiment favors peptide display on only one face of tile A. As depicted in Fig. 1b, corrugation of the tiles results in an alternating up-down display pattern for the peptides.

Figure 1. Peptide-modified oligonucleotides self-assemble into a grid style lattice for organization of AuNP.

a, Schematic of the A tile original and modified designs. The modified A tile (right) shows the location of the nick and covalent addition of the gold binding peptide. b, Self assembly of the 19 unique oligonucleotides into a square lattice where alternating peptides display on opposing sides of the DNA lattice plane. c, Tapping mode AFM images under buffer of DNA lattice functionalized with peptide before (left) and after (right) 1 equivalent of 5 nm AuNP has been allowed to bind for 60 minutes. AuNP treated samples were allowed to mix in solution prior to deposition on mica and immediate AFM analysis.

The linear dodecapeptide employed (WALRRSIRRQSY) was identified by our collaborators through a FliTrx bacterial expression system to have a high affinity for gold surfaces.¹⁴ The N-terminus of the peptide was acetylated to prevent the addition of multiple copies of the peptide to the oligomer chain. The peptide-oligonucleotide conjugate was prepared by coupling the C-terminal end of the peptide to the 5′ amine-terminated end of the oligonucleotide.⁸ Successful coupling of the peptide to the oligonucleotide fragment was verified using polyacrylamide gel electrophoresis and MALDI mass spectrometry. Incorporation of the peptide into the DNA lattice was accomplished by combining one equivalent of the peptide-oligonucleotide conjugate with one equivalent each of the 18 remaining oligonucleotides and heating the mixture to 90°C then allowing it to cool to room temperature over 12 hours. As shown in the tapping mode AFM image on the left-hand side of Fig. 1c, neither the addition of the nick in the core strand nor the covalent linkage of the peptide negatively affected the annealed nanogrid product (imaged in annealing buffer). Omission of the peptide-oligonucleotide conjugate from the annealing reaction resulted in samples that showed no signs of nanogrid formation.

To demonstrate successful molecular recognition by the peptides after their organization by DNA self-assembly, 5 nm gold nanoparticles (AuNP) were added to the peptide-labeled DNA nanogrid. However, prior to their addition, the AuNP required treatment with a non-ionic surfactant to ensure their compatibility with the DNA nanostructure environment. Self-assembling DNA nanostructures require excess cations (typically magnesium) to screen the repulsion of the negatively-charged phosphate backbones in close proximity to one another.¹⁵ In sharp contrast, commercially-available citrate-coated gold AuNP easily undergo irreversible aggregation in the presence of electrolytes and cannot tolerate the 12.5 mM magnesium used for DNA nanostructure stabilization in these experiments. Typically, researchers stabilize AuNP using thiol-based monolayers that are easily altered through ligand exchange reactions.^{16, 17} However, desorption of a thiolate from gold is known to require about 45 kcal/mol¹⁸ and is thus unlikely to be displaced by the peptide (ΔG_ads~ −8.7 kcal/mol)¹⁴. Instead, protection of the nanoparticles was achieved through physisorption of a nonionic surfactant, polyoxyethylene (20) sorbitan monolaurate (Tween 20), to the gold surface.¹⁹ It is hypothesized that the interaction of the Tween 20 with the surface of the AuNP is weak compared to the attraction of the genetically-engineered peptide to gold, allowing the weakly-adsorbed surfactant to be displaced by the incoming peptide. The AFM photo on the right-hand side of Fig. 1c shows the result of adding 1 equivalent of AuNP to peptide-labeled DNA nanogrid for 60 minutes prior to deposition on mica for imaging. All samples combining DNA with AuNP were allowed to mix in solution for a period of time prior to deposition of 3 μL of the mixture on mica, followed by a 3 minute wait before addition of 60 μL buffer for tapping mode AFM analysis. The addition of imaging buffer to the mica surface significantly dilutes the excess AuNP and is expected to quench any additional binding to the peptides. In addition to Tween 20, other agents were tested for their ability to stabilize the AuNP, including polyethylene glycol (PEG), bovine serum albumin (BSA), and polyoxyethylene (60) sorbitan monostearate (Tween 60). However, none of these were as successful as Tween 20 in stabilizing the AuNP. The optimal Tween 20 concentration was identified by testing concentrations ranging between 0.1% and 5% (w/v) for their ability to inhibit nanoparticle aggregation in 12.5mM Mg²⁺. At the highest nanoparticle concentration used for this report (332 nM), no aggregation was observed after several hours in magnesium containing buffer with 2% Tween 20, and all subsequent measurements were performed at this Tween 20 concentration.

Experimental evidence suggested an inverse correlation between the number of equivalents of added AuNP and the binding time required to reach near-saturating conditions. Fig. 2a shows the result when 4 equivalents of AuNP are added to peptide-labeled nanogrid and allowed to mix in solution for 3 minutes. The resulting density of templated AuNP in Fig. 2a is similar to the sample pictured in Fig. 1c. However, similar density was achieved for the sample with reduced AuNP concentration only after allowing mixing for much longer time. Fig. 2b demonstrates this effect with even greater contrast. For this sample, 20 equivalents of AuNP were added to the peptide-labeled nanogrid and allowed to mix in solution for 30 minutes. The resulting sample was so densely populated with AuNP that characterization was limited to low magnification images (as in Fig. 2b). High magnification scans of samples with dense AuNP coverage, such as the sample shown in Fig. 2b, resulted in poor image quality due to degradation of the DNA template. The dearth of unbound AuNP observed on the mica substrate is also noteworthy. The sharp contrast between the number of AuNP bound to DNA lattice versus the number of AuNP distributed randomly on bare mica (as shown in Fig. 1c and 2b) was not observed for control lattice samples assembled without peptide.

Figure 2. AuNP bound to peptide-labeled DNA lattice.

a, Rows of AuNP on peptide-labeled DNA lattice after 4 equivalents of nanoparticles are allowed to bind to the peptide for 3 minutes before deposition into the surface. The box-outlined region depicts the magnified area shown in c. b, Dense coverage of AuNP on DNA nanogrid is achieved after 20 equivalents AuNP are allowed to bind for 30 minutes before imaging. Very few unbound AuNP are detectable on the exposed mica surface. c, Magnified AFM image of the box-outlined region in (a). Comparison of the AFM image with the map (d) shows AuNP binding favors one peptide orientation (red) over the other (blue). The map depicts peptides displayed on opposing sides of the gray-colored 2D DNA lattice as red and blue circles. The predicted distances between rows of AuNP shown in (d) corresponds well with the measured distances in (c).

The corrugation strategy employed in this self-assembling DNA nanostructure results in lines of peptides alternating between each face of the 2D DNA lattice (depicted in Fig. 1b and as red circles on one face and blue circles on the opposite face in Fig. 2c). Thus, when mixed in solution, as with the samples in these experiments, AuNP should bind to both sides of the plane formed by the DNA lattice. However, comparison of the DNA lattice populated with AuNP depicted in Fig. 2c with the corresponding schematic map in Fig. 2d showing all possible binding sites (red and blue circles) in a similarly-sized region of DNA lattice, clearly demonstrates binding of the AuNP favors peptide rows spaced 52 nm apart. This distance represents predominately one of the two possible peptide configurations. The map shown in Fig. 2d shows peptide rows displayed on the same side of the DNA lattice are 52 nm apart, while rows formed by peptides on opposing faces of the DNA lattice are 26 nm apart. The bias for AuNP binding to peptides on the same face of the DNA lattice shown in Fig. 2c could be a result of the 3 minute wait after deposition of the DNA lattice on the mica before dilution with imaging buffer. AuNP access to the peptides during these 3 minutes may be limited to the face of the DNA lattice not obscured by the mica surface. This possibility was tested by altering the sample preparation procedure. The DNA lattice was deposited on mica for 3 minutes before gold nanoparticles were added and allowed to bind for up to two hours. However, this change resulted in a significantly reduced yield of AuNP binding to either configuration of peptide and did not result in a measureable bias. The observation that very little binding takes place when the DNA lattice mobility is restricted on the 2D mica surface suggests that other mechanisms might be affecting behavior of the system. In a published study of ssDNA-labeled AuNP binding to ssDNA on a DNA lattice, the authors observed a tendency for the particles to preferentially occupy non-adjacent binding sites and attributed the observation to electrostatic repulsion between negatively charged AuNP.²⁰ The extent to which electrostatic repulsion versus other mechanisms is operating in the current system remains an open question.

Detailed analysis of the experimental data shown in Fig. 2c as well as additional similar results, leads to another interesting and important conclusion. The instances of AuNP binding to adjacent binding sites on peptide-labeled DNA tiles, together with clear evidence that the DNA nanogrid is lying flat (as opposed to folding or warping that would allow multiple peptides to cluster together), lead us to conclude that a single peptide molecule is sufficient to immobilize an AuNP. The AFM image in Fig. 2c clearly supports this conclusion. This is the strongest experimental evidence to date supporting the contention that a single copy of the peptide binds to and immobilizes a single nanoparticle target.

We used polyacrylamide gel electrophoresis to characterize the binding of individual peptide-oligonucleotide conjugates to AuNP in measurements similar to those performed on AuNP mixed with thiolated oligonucleotides.²¹ However, no change in mobility was observed between peptide-oligonucleotide conjugates with and without gold (data not shown). This result suggests that the interaction between gold and peptide, unlike Au-S, must be too weak to survive the forces exerted during electrophoresis. Furthermore, we tested the selectivity of the peptide-oligonucleotide conjugate by adding silver nanoparticles to the peptide-labeled DNA lattice. No immobilized silver nanoparticles were observed in the course of this measurement, demonstrating the high selectivity of the peptide interaction. In addition, we tested the binding of a variety of different gold particles including 1, 3, 5, 15, and 20 nm sizes, however the best initial results were obtained with 5 nm gold particles therefore none of the other sizes were pursued in later studies.

CONCLUSION

Many imaginative future applications of nanotechnology have been envisioned. One obstacle to realizing these visions is that we are still unable to organize many of the components needed for these applications. The high-yield and high-fidelity organization of gold nanoparticles has been achieved with the simple marriage of gold and thiol chemistry with single-stranded DNA complementary to strands incorporated into DNA lattices.^20,22,23 This technique can be extended to organize additional inorganic materials using other well-known chemistries, however, the simultaneous organization of multiple inorganic species for parallel device construction favors an approach with greater selectivity, such as those exhibited by proteins in biological chemistry. This study represents the first combination of DNA-based self-assembly and peptide molecular recognition to demonstrate patterned synthetic biomineralization. With continued development, the prototype described here can be used to construct simple nanoelectronic devices, such as single electron transistors²⁴. For more sophisticated assemblies, polyvalency can be included using multiple copies of the same peptide on the 5′ and 3′ oligonucleotide ends available in the nick site of the core strand. Further complexity may be achieved with the inclusion of multiple peptides capable of independently targeting different, specific materials, such as quantum dots, or semiconducting carbon nanotubes. To further develop this construction kit as an artificial, programmable biomineralization system, perhaps peptides with metal reducing capabilities may be incorporated for assembly of devices from the in situ nucleation and growth of nanoparticles in mild conditions from dilute ionic solutions. Some efforts in that direction have already yielded promising results, although not with the level of template complexity and programmability demonstrated here.^25,26

METHODS

Purification of Oligonucleotides

Synthetic oligonucleotide were purchased from Integrated DNA Technologies (Coralville, IA) and separated from truncation products by denaturing polyacrylamide gel electrophoresis (PAGE). Denaturing PAGE was carried out on a 160 × 180 × 1.5 mm gel containing 10% acrylamide and 8.3 M urea in TBE buffer (90 mM Tris, 90 mM boric acid, and 2 mM ethylenediaminetetraacetate (EDTA), pH 8). The gel loading buffer contained 90% formamide and 0.1% bromophenol blue. After electrophoresis (300V for 45 minutes), the portion of the gel containing the desired DNA was excised, diced, and shaken overnight at 4°C in 500 μL 0.5 M ammonium acetate, 10 mM magnesium acetate, and 2 mM EDTA. The supernatant was removed and added to a centrifuge tube with 1 mL of 200 proof ethanol and stored at −20°C overnight. The mixture was centrifuged for 30 minutes at 4°C and 16,000 g and the supernatant was discarded. The purified pellet of DNA was dissolved in pure water and the concentration was determined by ultraviolet absorption at 260 nm wavelength.

Peptide Synthesis and Purification

The peptide (Ac-WALRRSIRRQSY-OH) was synthesized on Wang resin pre-loaded with Fmoc-L-Tyrosine (Novabiochem) at 0.1 mmole scale using a Protein Technologies PS3 automated peptide synthesizer. The coupling of standard Fmoc (9-fluorenylmethoxy-carbonyl)-protected amino acids (Chem-Impex) was achieved with HBTU (O-benzotriazole-N,N,N′,N′-tetramethyluronium hexafluorophosphate; Novabiochem) in the presence of N-methylmorpholine (NMM) in N,N′-dimethylformamide (DMF) for 20 minute cycles. Fmoc deprotection was achieved using 20% piperidine in DMF (2 X 5 minutes). The N-terminus of the peptide was acetylated with acetic anhydride and NMM. Side-chain deprotection and peptide cleavage from the resin were achieved by treating the resin-bound peptide with 5 mL of 100% trifluoroacetic acid (TFA) for 2 hours under N2. After evaporation of TFA under N2, the peptide was washed three times with cold diethyl ether, air-dried, and purified by semi-preparative reverse-phase HPLC on a YMC C18 column with a linear 40-min gradient from 3 to 70% acetonitrile in water with 0.1% TFA. The mass of the peptide was confirmed using an Agilent ESI-MS.

General Procedure for DMTMM Coupling Reactions in Water

DMTMM (4-(4,6-dimethoxy[1,3,5]triazin-2-yl)-4-methyl-morpholinium chloride) coupling reactions were run in 200mM MOPS buffer pH 7.0 (200 mM (3[N-morpholino]propanesulfonic acid, 20 mM sodium acetate, 10 mM EDTA). 450 pmoles of DNA was combined with peptide (270 nmoles, 600 eq.) and DMTMM (4.5 μmoles, 10,000 eq.).

Workup for Coupling Reactions

The peptide-oligonucleotide conjugates were isolated from the completed solution-phase coupling reactions via ethanol precipitation. In this case, 1mL of 200-proof ethanol and 50 μL of 3 M sodium acetate (150 μmoles) were added and allowed to sit overnight at −20°C. The precipitated conjugate was centrifuged for 30 minutes at 4°C and 16,000 g and the supernatant was discarded. The pellet was dried in a vacuum centrifuge for 2 hours before resuspension in 30 μL water.

Purification of Peptide-Oligonucleotide Conjugate

Peptide-oligonucleotide conjugate (WALRRSIRRQSY - TTGTG AAGTT TTTCG ATCCT AGCAC CTCTG GAGTT TTTCT TGCC) was separated from unreacted DNA by denaturing polyacrylamide gel electrophoresis. The 30 μL fraction of resuspended reaction product was combined with 30 μL of gel-loading buffer and heated to 90°C for 5 minutes before loading on to the gel. After electrophoresis (300V for 45 minutes), the portion of the gel containing the peptide-oligonucleotide conjugate was excised, diced, and shaken overnight at 4°C in 500 μL 0.5 M ammonium acetate, 10 mM magnesium acetate, and 2 mM EDTA. The supernatant was removed and added to a centrifuge tube with 1 mL of 200 proof ethanol and stored at −20°C overnight. The mixture was centrifuged for 30 minutes at 4°C and 16,000 g and the supernatant was discarded. The purified pellet of peptide-oligonucleotide conjugate was dissolved in pure water to a concentration of 30 μM. The final concentration was confirmed by ultraviolet absorption at 260 nm wavelength.

MALDI Analysis of Peptide-Oligonucleotide Conjugates

MALDI-TOF mass spectrometry analysis was used to characterize the oligonucleotide starting material and peptide-oligonucleotide reaction product (see Supplementary Information for spectra). The analysis was performed using an Applied Biosystems DE-Pro Maldi-MS in the Mass Spectrometry Facility in the Chemistry Department at Duke University. To prepare the samples, the products were recovered from the polyacrylamide gels and dissolved in pure water to a concentration of 30 μM. 10 μL volumes of the recovered gel products were then stripped of cations using Ziptips (SCX, Millipore) and added to a mixture of 9 μL of 50 mg/mL 3-hydroxypicolinic acid and 1 μL 50 mg/mL diammonium citrate. The mass spectrometer was run in negative-ion mode and spectra were collected through the summing of 50 laser pulses.

DNA Strands for Tile A

Strand Name	DNA Strand Sequences (5′ → 3′) (green text is peptide sequence)
4×4-1N	GGCGTGTGGTTGC
AFC2	GAGCGCAACCTGCCTGGCAAGACTCCAGAGGACTACTCATCCGT
3	GGATAGCGCCTGATCGGAACGCCTACGATGGACACGCCGACC
A44	TCACGACGGATGAGTAGTGGGCTCAGTCGGATGAGC
C5	TCCGACTGAGCCCTGCTAGGATCGACTTCACTGGACCGTTCTACCGA
C6	CTCGCTCGGTAGAACGGTGGAAGCCTCCGGTGCATG
AFC7	ACCGGAGGCTTCCTGTACGGCAGAACTCCGTTGGACGAACACTCC
4×4-8N	TGTTCGTGGCGCT
A9.5	AGGCACCATCGTAGGTTTTCGTTCCGATCACCAACGGAGTTTTTTCTGCCGTACACCA
Amine-labeled	/5AmMC6/TTGTGAAGTTTTTCGATCCTAGCACCTCTGGAGTTTTTCTTGCC
POC	WALRRSIRRQSY—TTGTGAAGTTTTTCGATCCTAGCACCTCTGGAGTTTTTCTTGCC

DNA Strands for Tile B

Strand Name	DNA Strand Sequences (5′ → 3′)
B1=BB1	GCGAGGGTAGCGTGGGTAATCCATGC
BFC2	GATTACCCTGTTACCGTCGAGAAGGCCGGACCGTTCTACC
BB3	GATGTACCTGTCTCACTCGCGAGCGAAGGACGCTACC
4×4-6N	GCTCGGTAGAACGGTGGAAGCCAACGGTC
BFC5	GTTGGCTTCCTGACACTATCGAGATGATAGGACTACTCATCC
4X4-4N	ATCCGGATGAGTAGTGGGCTCAGTCGGAG
BFC7	GACTGAGCCCTGGTCTCGTCAAGGTCGGCGGACTCTATC
B88	CGTGAGATAGAGTGGTACATCGCTCA
B9	TAACACCTTCGCTCGTTTTCGAGTGAGACACCGCCGACCTTTTTTGACGAGACCACCTATCATCTTTTTCGATAGTGTCACCGGCCTTCTTTTTCGACGG

DNA Nanostructure Formation

For DNA nanogrid formation, 18 individual standard DNA oligos and the peptide-oligo conjugate (POC) were mixed together stoichiometrically at 1.0 μM in 1X TAE/Mg2+ buffer (40 mM Tris-HCl (pH 8.0), 20 mM acetic acid, 2 mM EDTA, and 12.5 mM magnesium acetate) and slowly cooled from 95 to 20 °C over a period of 16 hours. For AFM imaging, 3 μL sample was spotted on freshly cleaved mica for 3 min. 30 μL 1X TAE/Mg2+ buffer was then placed onto the mica and another 30 μL of 1X TAE/Mg2+ buffer was placed onto the AFM tip (for a total of 60μL). AFM images were obtained on a Digital Instruments Nanoscope IIIa with a multimode head by tapping mode under buffer using NP-S tips (Veeco Inc.).

AuNP Preparation

5 nm gold nanoparticles (AuNP) were purchased from Ted Pella (product # 15702-20). To ensure AuNP stability in the high salt environment, the nanoparticles were pretreated with a non-ionic surfactant. For cases in which Tween 20 or Tween 60 (Sigma Aldrich product #’s P1379 and P1629, respectively) were used, 1 mL of stock AuNP solution was mixed with the appropriate amount of stabilizing agent to achieve final Tween concentrations ranging between 0.1% and 5% (w/v). After addition of the appropriate stabilizing agent, the AuNP solutions were mixed overnight at room temperature. After incubation, the AuNP were concentrated and excess stabilizing agent was removed by 6 hours of centrifugation at 16,000 g’s. The resulting AuNP pellet was then isolated using a pipette and resuspended in 10 μL of supernatant. The concentrations of the AuNP were determined using the AuNP absorbance at 520 nm (ε520 nm for 5 nm AuNP = 1×107 M-1cm-1).²⁷

AFM Imaging of AuNP labeled DNA Nanogrid

AuNP labeled DNA nanogrid was prepared by adding 0.5 μL annealed nanogrid solution to 6 μL of concentrated AuNP in 1X TAE/Mg2+ buffer. Unless noted otherwise, all samples combining DNA with AuNP were allowed to mix in solution for a specified period of time prior to deposition of 3 μL of the mixture on mica, followed by a 3 minute wait before addition of 60 μL buffer for tapping mode AFM analysis. AFM images were obtained on a Digital Instruments Nanoscope IIIa with a multimode head by tapping mode under buffer using NP-S tips (Veeco Inc.).