Quantification of the affinities and kinetics of protein interactions using silicon nanowire biosensors

Abstract

Monitoring the binding affinities and kinetics of protein interactions is important in clinical diagnostics and drug development because such information is used to identify new therapeutic candidates. Surface plasmon resonance is at present the standard method used for such analysis, but this is limited by low sensitivity and low-throughput analysis. Here, we show that silicon nanowire field-effect transistors can be used as biosensors to measure protein-ligand binding affinities and kinetics with sensitivities down to femtomolar concentrations. Based on this sensing mechanism, we develop an analytical model to calibrate the sensor response and quantify the molecular binding affinities of two representative protein-ligand binding pairs. The rate constant of the association and dissociation of the protein-ligand pair is determined by monitoring the reaction kinetics, demonstrating that silicon nanowire field-effect transistors can be readily used as high-throughput biosensors to quantify protein interactions.

Figure 1. Schematic of the Si-NW FET biosensor set-up and binding cycles

a, Cross-section of the p-doped (P+) Si-NW biosensor set-up showing the source (S) and drain (D). FOX, front oxide (20 nm silicon oxide in direct contact with solutions); BOX, buried oxide (145 nm silicon oxide). A platinum solution gate is used as a reference electrode to bias the Si-NW FETs to the desired operating point, and the back-gate is used as a screening tool for device characterization before solution measurements. The sensing element is a 10-μm-long, 1-μm-wide nanoribbon with a typical active layer thickness of 45 nm (light blue). The fluid delivery system (trapezoidal) with flow inlet and outlet (PTFE tubes) is mounted on top of the chip. The Si-NW is functionalized with amino silanes followed by immobilized receptors [B] (red ‘Y’ shapes). Analytes [A] (blue ‘Y’ shapes) are delivered through the PTFE tubes. When the analytes bind to the receptors, the change in conductance in the Si-NWs is detected by the FET. Protein interactions, which are reversible, can be described as a reversed reaction (left bottom) with an association rate constant k₁ and dissociation rate constant k₋₁. b, Typical schematic binding cycle for measurements obtained using the Si-NW FET biosensor. At t = 100 s, a solution of analyte in the flowing buffer is passed over the receptor. As the analyte binds to the surface, the charges of the analytes cause a change in the current signal. Analysis of this part of the binding curve gives the apparent association rate. If the concentration of the analyte is known, then the association rate constant of the interaction (k₁) can be determined. At equilibrium, the amount of analyte associating and dissociating with the receptor is equal. The response level at equilibrium is related to the concentration of active analyte in the sample. At t = 350 s, the analyte solution is replaced by pure buffer and the receptor–analyte complex is allowed to dissociate. Analysis of these data gives the dissociation rate constant k₋₁ for the interaction. Introduction of a regeneration solution (for example, high salt, low pH) is used at t = 460 s to disrupt binding and regenerate the free receptor.

Figure 2. Measurements of HMGB1 and DNA binding using the Si-NW FET

a, I_d2V_g characteristics of HMGB1–DNA binding measured by Si-NW FETs. Black, after HMGB1 immobilization; green, after interaction with 300 nM DNA; blue, after DNA desorption and surface regeneration with 1 M NaCl. Threshold voltage V_T was obtained by linear extrapolation of the I–V curve to x = 0. b, Normalized Si-NW sensor responses (ΔI_ds/g_m) of HMGB1–DNA interactions as a function of DNA concentration. Five different sets of nanowires are used. The solid line is a fit using equation (1), and K_D is determined by least-squares fitting, giving 105±6 nM. c, Real-time sensor responses of HMGB1–DNA binding. Each curve represents the measurement of a different DNA concentration from the same device, and sensor responses are plotted using (I_ds−I₀)/g_m. Apparent association rates (k₁[A] + k₋₁) were determined by fitting with equation (3a) for each concentration, and dissociation data were fitted by equation (3b) with a best-fit rate constant of k₋₁ = (1.59±0.06) × 10⁻³ s⁻¹ Dashed lines represent the fits. d, Plot of apparent association rates (k₁[A] + k₋₁) versus DNA concentration. The linear fit of the data gives the association rate constant k₁ = (1.55±0.02) × 10⁵ M⁻¹ s⁻¹. For all measurements, V_ds = 100 mV, flow rate = 30 μl min⁻¹.

Figure 3. Sensor response of the binding of biotin and streptavidin as measured by the Si-NW FET

a, I_d2V_g characteristics of biotin–streptavidin binding measured by Si-NW FETs. Black, biotin-functionalized Si-NW; blue, after interaction with 2 nM streptavidin. b, Real-time sensor responses of biotin–streptavidin binding. Each curve represents measurement from a different device and sensor responses were normalized using (I_ds−I₀)/g_m. Data were fitted using equations (2a–c), and the rate constants were determined as k₁ = 5.50±0.08 × 10⁸ M⁻¹ s⁻¹ and k₋₁ = 8.80 ± 0.06 × 10⁻⁵ s ⁻¹. Dashed lines represent the fits. For all measurements, V_ds = 100 mV and flow rate = 300μ min₋₁.

Figure 4. Competitive dissociation processes of streptavidin from a biotin-functionalized surface

a, Schematic of the competitive dissociation processes of streptavidin from the biotin-functionalized surface. Streptavidin (blue) are attached via two biotin-binding sites. Desorption of the complex is achieved by introduction of d-biotin (green), and desorption occurs by sequential dissociation of the streptavidin–biotin bonds. b, Sensor response of streptavidin competitive dissociation with d-biotin (solid line), fitted with a bi-exponential decay function (dashed line).