Diffusion of DNA on Atomically Flat 2D Material Surfaces

Abstract

Accurate localization and delivery of biomolecules are pivotal for building tools to understand biology. The interactions of biomolecules with atomically flat 2D surfaces offer a means to realize both the localization and delivery, yet experimental utilization of such interactions has remained elusive. By combining single-molecule detection methods with computational approaches, we comprehensively characterize the interactions of individual DNA molecules with hexagonal boron nitride (hBN) surfaces. Our experiments directly show that, upon binding to a hBN surface, a DNA molecule retains its ability to diffuse along the surface. Further, we show that the magnitude and direction of such diffusion can be controlled by the DNA length, the surface topography, and atomic defects. We observe that the diffusion speed of the biomolecules is significantly lower than indicated by molecular dynamic simulations. Through computational analysis, we present the model based on temporary trapping by atomic defects that accounts for those observations. By fabricating a narrow hBN ribbon structure, we achieve pseudo-1D confinement, demonstrating its potential for nanofluidic guiding of biomolecules.

Keywords: DNA; hexagonal boron nitride; nanofluidics; surface diffusion; van der Waals materials.

1

Single-molecule observation of ssDNA on the hBN surface. (a) A schematic representation of the sample chamber and measurement setup. Fluorescently labeled ssDNA molecules (green) adsorbed onto the surface of an hBN flake (purple) are imaged using a wide-field epifluorescence microscopy setup. M, mirror; L, lens; A, aperture; DM, dichroic mirror; EmF, emission filter. (b) Fluorescence microscopy images of Cy3-labeled ssDNA molecules adsorbed on the hBN surface after injection of 0, 1, and 10 pM DNA concentrations (from left to right) at indicated times. ssDNA molecules in the top panels are marked by a green circle. The bottom panels show the magnified view of the areas within the white boxes in the top panel. The scale bars represent 20 μm (top) and 5 μm (bottom), respectively. (c) Temporal evolution of fluorescence emission spots counted on the flake area shown in the top panels of Figure 1b. A 5 μL of droplet containing no DNA, 5 μL of 1 pM DNA and 10 μL of 10 pM DNA were added at −60 s, 340 and 740 s, respectively (blue arrows). The time scale corresponds to that shown above panels in 1b.

2

2D diffusion of ssDNA on hBN surfaces. (a) Schematic illustration of the single-molecule tracking analysis methodology. Identification of fluorescent spots in the fluorescence microscopy images were performed using differences of Gaussian (DoG) algorithm combined with a quadratic fitting scheme with subpixel resolution. The localized spot positions were then tracked across consecutive frames to establish molecule trajectories. (b) Trajectories of individual ssDNA molecules (7 nt) recorded over 400 s. Blue and red colors indicate mobile and stationary phases determined by temporal apparent diffusion coefficient (D _A,temp) values (see Figure d), respectively (see Methods for details). The lines in the top left corner denote the angles of the hBN lattice determined by the flake edges. The scale bar is 20 μm. (c) A zoom-in image from a square area in (b). The black trajectories are representative trajectories showing the confined movement in domain boundaries defined by step edges (potential location indicated by dashed lines). The scale bar is 1 μm. (d) Time traces of the apparent diffusion coefficient (D _A) of four representative molecules (left panels) with their 2D trajectories (right panels), with blue and red indicating the mobile and stationary states, respectively. The scale bars in the right panels are 1 μm. (e) Jump distance distributions at various lag times τ = nΔt, with n = 1, 2, 4, 8, 16 and the frame time Δt = 0.1 s. The distributions show two distinct components, mobile (blue) and stationary (red). (f) Mean squared displacement ⟨r ²(τ)⟩ for ssDNA molecules of different lengths plotted as a function of lag time τ. Data measured from different hBN surfaces are marked with different colors. (g) The apparent diffusion coefficient D _A and the diffusion exponent α in ⟨r ²(τ)⟩ = 4D _Aτ^α are determined through linear fitting of the data in (f), as depicted by the red lines. The fill colors differentiate the data sets, which correspond to those in (f).

3

MD simulation of ssDNA diffusion on hBN surfaces. Three distinct types of hBN surfaces were investigated: a perfectly flat surface (a), a surface with a single step edge (b), and a surface with atomic defects (c). The displacement maps on the last column show the displacement in x and y of each nucleotide as a function of time.

4

Monte Carlo (MC) simulations of ssDNA diffusion on hBN surfaces. (a) Conceptual outline of the MC simulation model, which employs particles that diffuse, with diffusion coefficient D ₀ determined from MD, in simulation steps of 1 ns with probabilities P _tr per step of encountering trapping sites. Trapped particles remain stationary for a random trap time τ_t, before resuming movement. This behavior is simulated using the probability distribution function ψ­(τ_t), derived from single-molecule fluorescence microscopy data (Figure S6). If a particle crosses the domain boundary L _d ² based on its calculated displacement, its location is adjusted to be mirrored off the boundary, ensuring it remains within the domain boundary. By fitting the resulting MSDs, the effect of the domain size L _d and the trapping probability P _tr on the apparent diffusion coefficient D _A and the diffusion exponent α are determined. (b) Simulated variations in D _A and α for various P _tr and L _d are plotted as functions of frame time Δt when D ₀ = 749 μm²/s. The dashed line indicates Δt used in our fluorescence microscopy measurements. (c) Influence of P _tr on D _A and α at set conditions of Δt = 0.1 s and L _d = 2.0 μm, with blue and red lines representing experimental data for 7 nt and 15 nt ssDNA molecules. The corresponding P _tr values are 0.12% and 0.04% for 7 nt and 15 nt ssDNA molecules, respectively.

5

Surface diffusion of ssDNA molecules on a hBN ribbon. (a) A time-based maximum projection of 3500 epifluorescence microscopy images, showing 16 nt ssDNA molecules on a hBN surface. The hBN flake consists of three parts, a large reservoir (top), a narrow channel (bottom), and a y-junction branch (middle). The scale bar is 10 μm. (b) The mean square displacement (MSD) as a function of the lag time, averaged over 292 trajectories. (c–f) Snapshots of ssDNA injection from the reservoir to the channel (c), linear motion along the channel (d), straight (e) and bending (f) movements at the y-junction. In the last column, the images show the standard deviation projection of the image stack over time, where the pixels with large intensity changes across the image stack are brighter. The presence of bright patches indicates the presence of mobile DNA molecules throughout the respective time periods (3.1, 2.4, 3.7, and 2.4 s for (c–f), respectively). The scale bars in (c–f) are 5 μm.