Electrical characterization of protein molecules by a solid-state nanopore

Abstract

The authors measured ionic current blockages caused by protein translocation through voltage-biased silicon nitride nanopores in ionic solution. By calculating the mean amplitude, time duration, and the integral of current blockages, they estimated the relative charge and size of protein molecules at a single molecule level. The authors measured the change in protein charge of bovine serum albumin (BSA) protein induced by pH variation. They also confirmed that BSA molecules indeed traverse nanopores using an improved chemiluminescent analysis. They demonstrated that a larger protein fibrinogen could be distinguished from BSA by a solid-state nanopore measurement.

Measuring the charge and structural properties of individual protein molecules as well as the distribution of these properties in their native environment is currently a great challenge. Ionic current blockages measured in voltage biased nanofabricated pores have been used to sense nanoparticles and molecules of protein and DNA,^1–7 inspired by the pioneer work on measuring single polymers in protein channels.^8,9 There has been remarkable progress in the study of polymer translocation in protein pores and a recent study shows that polymer size can be resolved at high resolution.¹⁰ However, the ionic current signature and the dynamics of a charged protein molecule moving in a solid-state nanopore need to be studied. In this letter, we report our observations of well-defined current blockage signals due to single protein molecules traversing silicon nitride nanopores. We measured the changes of bovine serum albumin (BSA) as a function of pH, and studied how the protein size and structure affect the blockage signal by comparing a larger fibrinogen protein with BSA. These studies are important milestones in the development of solid-state nanopore devices for fast protein characterization.

The main component of a nanopore sensing system [Fig. 1(a)] is a nanopore in a silicon nitride membrane that separates two chambers connected electrically by ionic solution inside the nanopore. When a voltage is applied across the membrane, a stable open pore current I₀ is observed. After the addition of negatively (or positively) charged protein molecules to the cis chamber, the molecules in the vicinity of the nanopore will be captured by the electric field, and forced to traverse the nanopore to the positively (or negatively) biased trans chamber. The interaction of protein molecules with the nanopore, either by reversible partitioning into or by translocation through, will result in transient current blockages, as shown in Fig. 1(b). Three parameters are calculated from a current blockage: the mean blockage current ΔI_b, the translocation duration t_d, and the integrated area of a blockage A_ecd.⁶ A_ecd is the integral of ΔI_b(t) with respect to time in units of kiloelectron charge. The current blockages were recorded using an Axopatch 200B system (Molecular Devices) in voltage clamp mode at V=120 mV with its low pass filter set at 100 kHz for all measurements in this work. For each set of data, about 10 000 blockage events were recorded. The nanopores used in this work were about 10 nm in thickness and slightly larger in diameter than the protein molecules being studied. The details of silicon nitride nanopore fabrication and the single molecule detection apparatus were described in our previous work.^4,6 The concentration of protein molecules placed in the cis chamber was approximately 10 nM.

FIG. 1.

(Color online) Schematic diagram (not to scale) for nanopore protein analysis experiment (a), base line and typical current blockade events for BSA at pH 7.0 (b) and 4.5 (c). Event distribution plots of ΔI_b vs t_d for BSA at pH 7.0 (d) and pH 4.5 (e).

BSA (66 430 Da, Sigma) has an isoelectric point (PI) ranging from pH 5.1 to 5.5;¹¹ thus the protein has an overall negative charge (−18e) at pH 7. Applying 120 mV voltage to an ∼16 nm diameter pore in a solution of 0.4M KCl at pH 7.0, I₀∼7.4 nA was measured. After addition of BSA to the negatively biased cis chamber, downward blockage events occurred [Fig. 1(b)] indicating that the BSA molecules were negatively charged. When the cis chamber was positively biased, no blockages were observed at the beginning of the experiment. As the magnitude of the bias potential was decreased, smaller ΔI_b and longer t_d were observed. The cumulative results are presented in an event distribution plot [Fig. 1(d)]. Every dot in Fig. 1(d) represents one blockage event and every blockage is characterized by its ΔI_b and t_d. Figure 1(d) shows that there are two clusters of BSA blockage events: cluster 1 has most probable values of ΔI_b∼50 pA and t_d∼110 μs, while cluster 2 has smaller ΔI_b∼20 pA and shorter t_d∼50 μs. We attribute cluster 2 to the events that protein molecules partially entered the pore but failed to pass through it. Cluster 2 events will not be considered further in this study except to note that they can be easily distinguished from cluster 1 events, which were identified as free translocation of protein molecules through nanopores.

When the pH of the chamber solution was lowered to acidic conditions (pH<5), current blockages disappeared if the trans chamber remained positively biased. However, when the trans chamber was switched to negatively biased, current blockages appeared again, as shown in Fig. 1(c), indicating that the net charge of BSA protein had changed to positive at pH<5. This measurement is consistent with the fact that BSA is positively charged when the pH is lower than its PI (pH<5).¹² We studied the translocation of BSA through the same nanopore at three different acidic pH values (4.5, 4.1, and 2.4). The BSA molecules proved to be positively charged at all these pH values. When the pH was 4.5, near the PI of BSA, the most probable values are ΔI_b∼71 pA and t_d∼269 μs, as shown in Fig. 1(e). The same measurement was performed at a higher pH value of 10.4. The most probable values of ΔI_b and A_ecd increased as solution pH decreased [Fig. 2(a)]. The open pore current I₀ was approximately constant at all the pH values measured.

FIG. 2.

(Color online) Plot of peak values of ΔI_b and A_ecd vs pH (a) and plot of Q/Q_pH7 estimated from Eq. (3) vs pH (b).

If the dimensions of a protein molecule are close to that of a nanopore, d_m∼D_p and l_m∼H_eff, where d_m and l_m are the diameter and length of the protein molecule and D_p and H_eff are the diameter and the effective membrane thickness of the pore; theoretical description for the resistance change of the pore caused by the protein molecule translocation is complicated. However, we neglect the interaction between a protein molecule and the nanopore (free translocation); based on Ohm's law and the work by DeBlois and Bean,¹³ the parameters of ΔI_b, t_d, and A_ecd can be approximately described below,

$$\Delta I_b=\sigma V{d}_m^2{l}_m[1+f(d_m,D_p,L_m,H_{\mathrm{eff}})]∕H_{\mathrm{eff}}^2\text{when}{l}_m<H_{\mathrm{eff}},$$ (1)

$$t_d\sim M∕QV,$$ (2)

$$A_{\mathrm{ecd}}={\int }_{\text{event}}\Delta I\left(t\right)\sim M\Delta I_b∕Q.$$ (3)

Here M is the molecular weight, Q is the effective charge of a protein molecule, f(d_m, D_p, l_m, H_eff) is a correction factor that depends on the relative values associated with the dimensions of the molecule and the nanopore, and σ is the conductivity of the solution. Assuming the shape of BSA molecules is globular, as shown in Fig. 1(a), then l_m∼14 nm is smaller than H_eff∼20 nm (estimated from I₀). According to Eq. (1), an increase of ΔI_b suggests that the product d²_ml_m of BSA molecules increased. The increase in d²_ml_m at lower pH values implies a change in conformation or dimensions of BSA molecules. This is consistent with earlier reports that BSA molecules form dimers at low pH values¹⁴ or that the volume of BSA molecules was expanded at acidic pH values.¹¹ From Eq. (2), a change in the charge of BSA (Q) can be measured by the change in t_d. Since A_ecd can be measured more accurately than t_d for a current pulse, the relative charges Q/Q_pH7 were estimated from Eq. (3). The estimated values of Q/Q_pH7 are shown in Fig. 2(b).

If we assume that a BSA molecule in a nanopore is a rigid particle diffusing in one dimension under the influence of an electric force F=QV/H_eff with an average drift velocity ν_BSA=H_eff/t_d using Einstein relation, the effective diffusion constant was estimated to be D_eff=(k_BT) ν_BSA/F∼10⁻¹⁰ cm²/s. This value is three orders of magnitude smaller than it is in bulk solution (D∼10⁻⁷ cm²/s).¹⁵ The estimated characteristic time for a BSA molecule to diffuse a length of H_eff would be τ_eff=(H_eff)²/D_eff∼10 ms in a nanopore. Comparing to t_d∼10² μs for BSA measured at pH 7, this analysis suggests that there is a strong confinement for a BSA molecule in a nanopore.

To confirm the current blockages observed were due to BSA molecules translocating from the cis to the trans chamber via a nanopore, the trans chamber solution was collected after a BSA translocation experiment and subsequently used for an improved chemiluminescent BSA enzyme linked immunosorbent assay (supplemental materials). An 18 nm diameter pore with 200 mV bias voltage and an increased concentration of BSA protein (100 μM) in the cis chamber were used to increase the rate of blockage events. The translocation experiment lasted ∼50 h and the estimated number of blockages was ∼2×10⁷. The event trigger level was set so that cluster 2 events were not counted. An immunoenzymetric assay kit (Cygnus Technologies, Catalog No. F030) and a chemiluminescent substrate (Lumigen PS-atto, Lumigen) were used to quantify the amount of BSA in the trans sample. The luminescence signal generated from the reaction of the substrate with the horse radish peroxidase (HRP) on the anti-BSA/HRP-labeled antibody was measured. The chemiluminescent intensity analysis (Fig. 3) indicated that approximately 1.8±0.3 pg of BSA was in the trans chamber. This amount of BSA corresponds to ∼10⁷ molecules, in good agreement with the number of blockages estimated. This assay demonstrated that BSA molecules were detected in the trans chamber; hence the current blockages observed were the result of protein traversing the nanopore.

FIG. 3.

(Color online) Luminescence intensity calibration curve for BSA standards (○) and for half of the sample collected from the trans chamber (as indicated).

To characterize how the size (M) and the structure of protein molecules change current blockages, fibrinogen, a protein with charge (−16e) similar to BSA but larger in size (340 000 Da, Sigma) was studied. A schematic model of the fibrinogen structure¹⁶ describes it as an elongated molecule ∼47.5 nm long, featuring three nodule regions, as illustrated in Fig. 4(a). In this study, an ∼18 nm diameter nanopore was used. First, BSA molecules were measured in this pore, subsequently the cis chamber was washed, and the fibrinogen was added and measured. Several typical current blockages are shown in Fig. 4(a) and the event distribution plot for fibrinogen is shown in Fig. 4(b). Most blockages do not reveal the nodular structure of fibrinogen except for a few of long t_d∼1 ms events [inset of Fig. 4(b)]. When cluster 1 events of fibrinogen are compared with those of BSA, fibrinogen has larger ΔI_b (Fig. 4(c)), longer t_d [Fig. 4(d)] and greater A_ecd [Fig. 4(e)] with much broader distributions. For BSA, the peak values of these parameters are ΔI_b=52±13 pA, t_d=95±28 μs, and A_ecd=27±12 ke. For fibrinogen, these values are ΔI_b=95±29 pA, t_d=170±48 μs, and A_ecd=56±32 ke. Thus, using these parameters we could clearly distinguish fibrinogen from BSA by a nanopore experiment.

FIG. 4.

(Color online) Schematic drawing of a fibrinogen molecule and several typical current blockages for fibrinogen (a), event distribution plot for fibrinogen (b), distribution comparisons of BSA with fibrinogen for ΔI_b (c), t_d (d), and A_ecd (e). The experiment was performed in a 0.4M KCl and TE buffered solution at pH=7.0.

A larger ΔI_b for fibrinogen can be attributed to a larger d²_ml_m as indicated in Eq. (1), although when l_mH_eff, Eq. (1) is not accurate. A longer t_d was expected based on Eq. (2) because the ratio of M/Q is about five times larger for fibrinogen than BSA. Considering the charge distribution and the structure [Fig. 4(a)], a fibrinogen molecule could enter a nanopore with many possible configurations. This would result in different levels of ΔI_b and values of t_d, thus a broad distribution in the ΔI_b vs t_d plot [Fig. 4(b)] was compared to BSA [Fig. 1(d)].

In summary, by measuring the change in current caused by protein molecules traversing a nanopore, this work demonstrated that the relative charge and size of protein molecules could be estimated based on the values of ΔI_b, t_d, and A_ecd. This nanopore technique can measure properties of individual protein molecules sequentially and would allow one to determine the distribution of these properties in real time and under natural conditions. The results presented here suggest that by using a marker protein with a known charge, conformation, and size, a solid-state nanopore can be used to characterize unknown proteins. Furthermore, if the spatial and temporal resolutions of the nanopore sensing system can be improved, structural features of protein molecules could be measured in more detail.