Single-Molecule Multivalent Interactions Revealed by Plasmon-Enhanced Fluorescence

Abstract

Multivalency as an interaction principle is widely utilized in nature. It enables specific and strong binding by multiple weak interactions through enhanced avidity and is a core process in immune recognition and cellular signaling, which is also a current concept in drug design. Here, we use the high signals from plasmon-enhanced fluorescence of nanoparticles to extract binding kinetics and dynamics of multivalent interactions on the single-molecule level and in real time. We study mono-, bi-, and trivalent binding interactions using a DNA Holliday Junction as a model construct with programmable valency and introduce a step-binding model for binding kinetics relevant for structured macromolecules by including an experimentally extractable binding restriction term ω to quantify the effects from conformation, steric effects, and rigidity. We used this approach to explore how length and flexibility of the DNA ligands affect binding restriction and binding strength, where the overall binding strength decreased with spacer length. For trivalent systems, increasing spacer length additionally activated binding in the trivalent state, giving insight into the design of multivalent drug or targeting moieties. By systematically changing the receptor density, we explored the binding super selectivity of the multivalent HJ at the single-molecule level. We find a polynomial behavior of the trivalent binding strength that clearly shows receptor-density-dependent selective binding. Interestingly, we could exploit the rapidly decaying near fields of the plasmon that induce a strong dependence of the signal on the position of the dye to observe binding dynamics during single multivalent binding events.

Keywords: DNA nanotechnology; fluorescence enhancement; multivalency; plasmonic nanoparticles; single-molecule fluorescence.

Figure 1

Measurement principle. (a) Schematics of the setup. We use a DNA-based framework of DNA-coated gold nanorods and 4-way junctions (HJ) as a multivalent ligand–receptor system. The DNA-coated nanorods are placed on a cover glass mounted in a flow cell on a TIRF microscope. When fluorescently labeled HJs bind to the particle through the DNA–DNA ligand–receptor interaction, the signal from a particle transiently increases. (b) Example of a field of view showing the dark background and the diffraction-limited bright spots from the gold nanorods. The blue square highlights a background region with nonspecific HJ binding, while the red square highlights a gold nanorod that HJs bind to. (c) Nanorod highlighted in the red square in (b) upon binding of a dye-labeled HJ (top), referred to as the bound-state (on), and upon absence of HJ binding (bottom), referred to as the off-state. The sketch shows the drastic reduction of the fluorescence enhancement factor (EF) with increased distance between the nanorod and dye-labeled HJ. (d) Time trace from the highlighted particle in the background region in (b) with respective color-coding (the particle highlighted in the red square is displayed in red, and the area in the blue square is displayed in blue). (e) BEM simulation of a 40 by 82 nm gold nanorod with LSPR at 650 nm showing the fluorescence enhancement factor of an ATTO 655 dye with an intrinsic quantum yield of 0.3. The inset shows the fluorescence enhancement factor along the dotted line that highlights the distance-dependent enhancement. (f) Experimental enhancement factors from four different ligands; two single-stranded ligands with dye placed at the 3′ or 5′ end and two 4-way Holliday junctions with dye and ligand placed as sketched. Black lines show the mean value of the measured enhancement factors. (g) Zoom-in region of the time trace in (d) that shows the bound-state lifetime τ_on (on-time/bright time) and off-time τ_off (dark-time) of the HJ binding to the nanorod.

Figure 2

Holliday Junction with a programmable valency. (a) Sketch of the Holliday junction scaffold with dye location and added ligand sequences that are complementary to the receptor DNA on the gold nanorod surface. (b) oxDNA structure of the Holliday junction with 3 binding ligands. (c) Examples of 100 s time traces from a monovalent (top), bivalent (middle), and trivalent (bottom) HJ with respective color-coding (blue, red, and yellow for mono-, bi-, and trivalent HJ, respectively). (d) Cumulative distribution function (CDF) of collected bound-state lifetimes from multiple time traces with thousands of events and a corresponding stretched exponential fit with same color-coding as in (c). Inset shows the CDF on a shorter time axis to better show monovalent CDF. (e) oxDNA structure of the HJ ligand showing the ligand sequence in red and the T-spacer in black. (f) Measured bound-state lifetimes of HJs with a single ligand without any T-spacer, with 1, 2, and 4 T-spacers and the bound-state lifetime of the single-stranded free ligand (ligand only).

Figure 3

Step-binding model. (a) Scheme of step binding. The first step is the binding of HJ with one intermolecular interaction. From here, the HJ can either dissociate and leave the particle or bind with the second ligand as an intramolecular interaction. The rate of the second binding step is influenced by the effective concentration C_eff and the binding restriction ω. (b) Schematics of the probing volume determining C_eff. Green shows the probing volume, and orange shows the available surface area under the probing volume. The blue spheres (and half spheres) represent the receptor DNA on the particle surface. The first sketch (top left) shows the case in which the ligand is considered a freely diffusing object within the probing volume. The second and third sketches show the unstructured model from the side and from the top. Here, the ligand is tethered to the surface by the first ligand interaction but can diffuse freely within the probing volume defined by the length of the tether. (c) Sketch of a structured model that takes into account the structure of the HJ that limits the accessible volume to be far from the first binding site. (d) Experimental and modeled values of the bound-state lifetimes for the bivalent HJ.

Figure 4

Restriction factor. (a) Schematic Illustration of the influence of the restriction factor on the probability of the position of the second ligand during a binding event. The bright areas represent areas where the ligand is more likely to be, whereas the dark areas represent a volume that the ligand is less probable to occupy. The restriction factor makes the space close to the surface less likely, thus restricting the binding of the second ligand. (b) Shows the step-binding model of the trivalent HJ including separate restriction factors, ω, and effective concentrations, C_eff, for the monovalent to bivalent (2) and trivalent step (3). (c) Table of bound-state lifetimes of the trivalent HJ and restriction factor from the two binding steps.

Figure 5

Addition of spacers. (a) Spacer sequence addition between the HJ arm and ligand sequence that increases length and flexibility of the ligand. Spacer sequences consist of multiple T nucleotides ranging from 2 to 12. (b) Histogram of the ligand extensions with added T-spacers obtained from oxDNA simulations. (c) Plot of the bound-state lifetimes of the bivalent (circle) and trivalent (square) HJ on the left axis with added T-spacers, and the ratio of bound-state lifetimes between trivalent and bivalent HJs in orange on the right axis. (d) Plot of the restriction effect, 1/ω, for the bivalent (red) and trivalent (black) HJs with added T-spacers.

Figure 6

(a) Sketch of the systematic dilution of the receptor strand with a dilution strand. (b) Plot of the bound-state lifetimes of the bivalent (orange) and trivalent (yellow) HJ with different receptor strand concentrations. Dotted lines show linear and polynomial lines as a guide to the eye.

Figure 7

Binding dynamics. (a) Time traces of a 12 nt monovalent ligand showing a “box-like signal” with fluctuations around a single fluorescence intensity level within an event. On the right plots, a histogram of the signal intensities for each frame along with the relative standard deviation (RSD) of the signal values. (b) Time traces of a trivalent HJ with 8 nt ligands that show different signal fluctuations compared to the 12 nt free ligand. Here, the signal fluctuates around multiple fluorescence intensity levels. On the right is plotted a histogram of the signal intensities for each frame along with the relative standard deviation (RSD) of the signal values. Traces were acquired with a 100 fps framerate.