Electrostatic melting in a single-molecule field-effect transistor with applications in genomic identification

Abstract

The study of biomolecular interactions at the single-molecule level holds great potential for both basic science and biotechnology applications. Single-molecule studies often rely on fluorescence-based reporting, with signal levels limited by photon emission from single optical reporters. The point-functionalized carbon nanotube transistor, known as the single-molecule field-effect transistor, is a bioelectronics alternative based on intrinsic molecular charge that offers significantly higher signal levels for detection. Such devices are effective for characterizing DNA hybridization kinetics and thermodynamics and enabling emerging applications in genomic identification. In this work, we show that hybridization kinetics can be directly controlled by electrostatic bias applied between the device and the surrounding electrolyte. We perform the first single-molecule experiments demonstrating the use of electrostatics to control molecular binding. Using bias as a proxy for temperature, we demonstrate the feasibility of detecting various concentrations of 20-nt target sequences from the Ebolavirus nucleoprotein gene in a constant-temperature environment.

Figure 1. smFET device characteristics.

(a) To-scale illustration of the smFET device structure. An amine-modified oligonucleotide probe is covalently coupled to a single defect site, generated on a single-walled carbon nanotube (SWCNT) sidewall via diazonium chemistry. The CNT acts as the channel of a field-effect transistor, transducing biomolecular charge into a conductance change in the smFET. Binding and melting reactions of a complementary oligonucleotide generate random telegraph signals (RTSs, shown in inset), which correspond to hybridization and melting events. Experiments are performed at a constant source–drain bias (V_s) of 100 mV, while the electrolytic gate bias is modulated by an on-chip pseudo-reference electrode. (b) Representative transfer (I–V) characteristics of four CNT devices before and after point functionalization and DNA attachment. V_g is the bias on the platinum pseudo-reference electrode relative to the device. V_g is varied from −1.5 V to +1.5 V (in acetonitrile solution supported by 0.1 M tetrabutylamonium hexafluorophosphate). (c) The power spectral densities (PSDs) of the devices from b showing a strong flicker (1/f) noise component before (black line) and after (red line) modification. While the newly formed defect dominates transfer characteristics, the CNT sidewall becomes less sensitive to charge traps, leading to lower flicker noise.

Figure 2. Temperature-dependent DNA hybridization and corresponding data analysis.

(a) Time traces of 300-s length are recorded for complementary DNA target at a concentration of 100 nM. Overlaid raw real-time data (black) and idealized fits (red) for a temperature series from 30 to 60 °C. The rate of bi-stable activity increases with temperature. (b) A representative survival probability curve for the 50 °C trace from a, with an event count of 763. The fit quality is characterized by R²=0.999 and R²=0.998 for τ_low→high and τ_high→low, respectively. (c) Comparison of nearest-neighbour calculated and single-molecule measured melting curves for a 100 nM complementary target showing the same temperature dependence as a ‘traditional' melting curve exhibiting a lower hybridized fraction with increasing temperature and a T_m of 52.99 °C compared with a calculated T_m=49.66 °C. (d–f) Arrhenius plots for 100, 10 and 1 nM target concentrations, showing the temperature dependence of the melting and the hybridization rates. Error bars are calculated from five different 60-s intervals at each temperature.

Figure 3. The effects of bias on hybridization and melting kinetics.

(a) Comparison of temperature and bias-dependent melting. With increasing temperature, the melting rate increases and hybridization rate decreases. Equivalently, when the gate bias increases, the equilibrium favours the melted state. (b–d) Arrhenius plots showing how application of gate bias affects the kinetics of melting and hybridization for 100, 10 and 1 nM complementary target. Upon the application of 300 mV of gate bias, the entropy of activation for the melting reactions increases, while the enthalpy of activation for the hybridization reaction decreases. Error bars are calculated from five different 60-s intervals at each temperature.

Figure 4. Sequence dependence and the effect of e-melting on reaction kinetics.

(a) Overlay of raw real-time data (black) and idealized fits (red) for a bias series ranging from V_g=0 V to V_g=500 mV of a 100 nM complementary target. At 0 V, no melting events are observed and the hybridized state is dominant at 40 °C. When V_g is increased, the melting rate increases, demonstrating longer and more frequent melting events. (b) k_melts of a complementary target DNA at a temperature of 40 °C increases exponentially with increasing V_g. Behaviour of a target containing a single-base mismatch (SNP) has a noticeably smaller activation energy and higher melting rate constant at each bias point. (c) k_hyb is less sensitive to bias. At higher bias values, k_hyb decreases, indicating that base-pairing is affected under repulsive electrostatic force, while the SNP, which cannot pair its terminal base, does not show this effect. Error bars are calculated from five different 60-s intervals at each temperature.

Figure 5. Reaction thermodynamic parameters are affected by electrostatic force.

Van't Hoff plots depict the effect of bias on the free energy landscape, revealing an order of magnitude lower equilibrium constants and higher enthalpy for the 300 mV biased reaction at (a) 100 nM complementary target, where an increase in ΔH° of 6.5 kJ mol⁻¹ making hybridization less favourable, (b) at 10 nM with an enthalpy increase of 38.46 kJ mol⁻¹ and (c) at 1 nM with an enthalpy increase of 74.78 kJ mol⁻¹. Error bars are calculated from five different 60-s intervals at each temperature.

Figure 6. Sequence-dependent effects of electrostatic force.

(a) Dependence of K_eq on V_g, demonstrating higher values for the complementary compared to the SNP across all V_g values. K_eq=1 defines a melting potential of E_m=400.05 (±2.55) mV for the complementary target and E_m=326.45 (±1.33) mV for the SNP; R² values for the SNP and comp are 0.95 and 0.97, respectively. Error bars are calculated from five different 60-s intervals at each bias value. (b) An effective T_m curve showing the extrapolated temperatures for each hybridized fraction of both complementary and SNP target against their perspective nearest-neighbour models; and (c) calibration curves demonstrate the interchangeable effect of bias and temperature on reaction rates and further correlate each bias point with an effective temperature.