Electronic signatures of all four DNA nucleosides in a tunneling gap

Abstract

Nucleosides diffusing through a 2 nm electron-tunneling junction generate current spikes of sub-millisecond duration with a broad distribution of peak currents. This distribution narrows 10-fold when one of the electrodes is functionalized with a reagent that traps nucleosides in a specific orientation with hydrogen bonds. Functionalizing the second electrode reduces contact resistance to the nucleosides, allowing them to be identified via their peak currents according to deoxyadenosine > deoxycytidine > deoxyguanosine > thymidine, in agreement with the order predicted by a density functional calculation.

Figure 1

Tunneling measurements with functionalized electrodes. (A) A gold probe and a gold substrate are functionalized with a monolayer of 4-mercaptobenzoic acid, and the size of the gap between two electrodes maintained under servo control at a value such that the two monolayers do not interact with one another, resulting in a tunnel current signal that is free of spikes (B). When a solution of nucleosides is introduced, current spikes appear, as shown here for(24) 0.7 μM deoxyadenine in trichlorbenzene with a baseline tunneling current of 6 pA at a bias of 0.5 V (C). Hydrogen-bonding schemes for all four nucleosides are shown in panels D−F. “S” represents the modified deoxyribose sugar (Supporting Information) and the hydrogen bonds are circled.

Figure 2

Effect of electrode functionalization on the distribution of current spikes for purines. Bare electrodes (A, dA; C, dG) give broad distributions (gap conductance 20 pS, 0.7 μM dA, 2.9 μM dG in TCB). Fits are Gaussian in the log of the current (Figures S6−8, Supporting Information). Distributions narrow 10-fold when one electrode is functionalized with 4-mercaptobenzoic acid (B, dA; D, dG) (gap conductance 12 pS, I_bl = 6 pA, V = 0.5 V). Fits are to two Gaussians in the log of the current with a peak at i₀ (“1”) and a second at 2 i₀ (“2”) (eq S3). i₀ = 5.9 pA for dA and 5.6 pA for dG. When both electrodes are functionalized (E, dA; G, dG) the peak currents are clearly different (i₀ = 9.4 pA for dG, i₀ = 16.5 pA for dA). (F) Distribution for a mixture of dA and dG. The assignment of the higher peak to dA is confirmed by the distribution measured with a reduced concentration of dA (H). The high current tail in (F) and (H) is consistent with a small number of two molecule (dA+dG) reads. Distributions of the spike widths are given in Figure S16 (Supporting Information).

Figure 3

Effect of electrode functionalization for pyrimidine reads. For bare reads (broad distributions in A and B) G_bl was increased to 40 pS to increase the count rate. The narrow distributions in A and B are taken with both electrodes functionalized and yield i₀ = 6.7 pA for dT and 13.3 pA for dC (G_bl = 12 pS, I_bl = 6 pA, V = 0.5 V). In a mixed solution, (C) the dT peak occurs at 8 pA and the dC peak occurs at 13.4 pA, an assignment verified by measuring a mixture with half the concentration of dT (Figure S9 in Supporting Information).

Figure 4

Summary of the reads. (A) The measured molecular conductance increases linearly with G_bl (black circles dT, black squares dA, error bars are ±HWHH). The number of two molecule reads (open circles, dT, open squares, dA) increases at G_bl = 20 pS, and the read rate is substantially reduced at G_bl = 4 pS (Supporting Information). (B) Peak currents measured in three independent runs for the four nucleosides (cross hatched bars). Error bars represent the HWHH of the current distributions. Reads for a functionalized surface and a bare Pt (light shaded bars) and bare Au (dark shaded bars) probe are relatively insensitive to the identity of the nucleoside, as shown quantitatively in (C) where the junction resistance is plotted vs the molecular resistance determined with two functionalized probes.