DNA stretching and optimization of nucleobase recognition in enzymatic nanopore sequencing

We analyzed biotinylated ssDNA molecules in complex with streptavidin, where the DNA is allowed to thread through the nanopore and the biotin:streptavidin complex prevents the full translocation of the nucleic acid. This configuration mimics the pauses between base additions in enzymatic nanopore sequencing, during which the DNA strand is held within the pore lumen and stretched in the applied potential. DNA threading is manifested by the reduction of the open pore current (I_O) to a blocked pore current (I_B), and we quote the residual current (I_RES) as a percentage of the open pore current. The ssDNA molecules are captured by the nanopore at +200 mV and then the bias is reduced to +40 mV in 20 mV steps by using an automated voltage protocol. At potential lower than +100 mV DNA occasionally retreated out of the nanopore, as observed by stepwise increases of the ionic current.

A DNA molecule with a charge density Q translocating through the nanopore experiences an electrophoretic driving force, F_EF = V_biasQ, due to the electric field in the nanopore. For dsDNA inside solid-state nanopores the amplitude of the effective driving force (F_eff) on DNA has been measured directly by optical tweezers or indirectly, and has been found to vary between 240 and 100 pN V⁻¹ depending on the nanopore size, geometry and the solution ionic strength [22–25]. DNA also experiences forces that oppose DNA translocation, being a stochastic thermal (Brownian) force from random molecular collisions that is strongly influenced by the electroosmotic-induced flow field [26]. MD simulations showed that the electroosmotic flow that develops near the DNA surface and is driven by the motion of counterions can reverse the effective electrophoretic force on DNA [27]. Interactions between the nanopore surface and the DNA are usually ignored. Hence, the effective force on the DNA inside the pore (F_ef) should include the electrical force on DNA's bare charge (F_EF) and the drag force due to electro-osmotic flow (EOM) inside the nanopore (F_EOM): F_eff = F_EF − F_EOM [23, 24]. Therefore, the threaded DNA strands in our experiments will escape the nanopore when F_EF ∼ F_EOM and F_eff approaches DNA thermal fluctuations, which for the αHL wild type (αHL-WT) nanopore is at potentials lower than 80−100 mV [28–30].

The WT-αHL β-barrel contains three nucleobase recognition points, named R1, R2, and R3 (figure 1(A)), which are capable of discriminating between individual bases in immobilized ssDNA molecules. Recognition point R1 is located at the central constriction. At +160 mV and with immobilized DNA strands R1 in the WT pore recognizes bases at positions 8–11 with a peak at position 9 (figure 1(B)). R2 is located near the middle of the β-barrel and recognizes bases 13–15 (peak at position 14 at +160 mV), and R3 is located near the trans entrance of the pore and recognizes bases 16–20 (peak at position 17 at +160 mV, figure 1(B)) [31]. While WT-αHL can discriminate differences between A, T, C and G only at position R2, the 2N-αHL (E111N/K147N) mutant increased the residual current during DNA blockade thus improved nucleobase recognition [8]. 2N-αHL pores could discriminate differences between all four DNA bases at R2 and R3 but not at R1 [9]. Among the αHL mutants that are able to recognize the four DNA bases at R1, 2N/M113Y (E111N/M113Y/K147N) showed the best characteristics [9].

Zoom In Zoom Out Reset image size

In immobilized DNA strands, Stoddart et al showed that at +160 mV substitutions in about 12 DNA bases changed the ionic output signal of WT-αHL and 2N-αHL nanopores [8]; while for freely translocating short oligonucleotides Meller et al found that 12 nucleobases occupy the barrel of WT-αHL nanopores at +120 mV [22]. Let be exactly 12 DNA bases in the 5 nm αHL barrel in immobilized strands at +160 mV, than the inter-base distance between nucleobases is ∼4.2 Å, which corresponds to the average inter-base distance found in ssDNA molecule stretched by optical tweezers with a force of about 10 pN [32]. We investigated the voltage dependent stretching of ssDNA by probing the recognition of the WT-αHL pore with a set of 14 poly(dC) oligos containing a single dA at different positions relative to the biotinylated 3' end (from position 7 to position 20, figure 1). ΔI_RES (with respect from poly(dC)) from +60, to +220 mV were then plotted against the position of the adenine nucleotide (figure 2). At +220 mV, the peak of R2 recognition was shown for ssDNA containing dA at position near 14, while at +60 mV the peak of R2 recognition was shifted by almost exactly one nucleobase to the position near 15 (figure 2). At intermediate applied potentials the peak of R2 recognition lied at intermediate positions, suggesting that over 160 mV the DNA immobilized within the nanopore is stretched by the length of approximately one nucleobase. Therefore, assuming the DNA inside the lumen of the pore is uniformly stretched and exactly 12 nucleobases fill the αHL barrel at +160 mV, the occupancy of the αHL barrel by the DNA changes from ∼12.5 to ∼11.5 nucleobases from +60 to +220 mV, indicating that the inter-base distance between the nucleobases inside the β-barrel increased by about 2.2 Å V⁻¹. It should be noted, however, that it is possible that the DNA inside the pore is stretched less in the β-barrel than in the vestibule, where it is less constrained.

Zoom In Zoom Out Reset image size

In order to sequence DNA with a nanopore, the rapid translocation of DNA through the pore is controlled by enzymes [12, 21, 33, 34], the activities of which depend on the ionic concentration of the solution and the pulling force applied on the nucleic acid. We tested the voltage dependence of the recognition of individual bases at R1 by using the 2N/M113Y αHL mutant in 1 M KCl (figure 3) by using four poly dC oligos containing A, C, T and G at position 9 (figure 1(B)). The difference between the I_B of poly(dC)₄₀ and the oligonucleotides containing A and T (ΔI_RES^C−A and ΔI_RES^C−T) at position 9 increased almost linearly with the applied potential, in parallel with the increase in the ionic current through the DNA blocked pores with the applied potential. By contrast, the voltage dependence of the I_B difference of the oligonucleotides containing G (ΔI_RES^C−G) at position 9 was biphasic (figure 3): from +60 to +100 mV ΔI_RES^C−G increased in parallel with ΔI_RES^C−A and ΔI_RES^C−T, while at applied potentials higher than +120 mV the increase of ΔI_RES^C−G was less rapid (figure 3). As a consequence, the I_RES values for the four nucleobases could be separated at voltages higher than +60 mV, with the exception of +120 mV, where T9 and G9 showed the same I_RES value and at +200 mV, where G9 and A9 showed identical I_RES values. The recognition profile in αHL appears relatively diffused, with multiple contiguous bases showing similar recognition (figure 2). Therefore, as also observed with MspA nanopores [35], it is likely that each recognition sites in αHL nanopores reads three or four nucleobases simultaneously. Hence, the voltage dependence of nucleobase recognition might be explained by the base at position 9 moving away from the centre of the recognition site to a position where the ΔI_RES^C−G signal is reduced (but not that of $\Delta I_{{\rm RES}}^{{\rm C}-{\rm A}}$ and ΔI_RES^C−T) as the DNA is stretched in the applied potential.

Zoom In Zoom Out Reset image size

The effect of the ionic strength on nucleobase recognition was tested by sampling 2N-αHL mutant with the four poly(dC) oligos containing A, C, T and G at position 14 (R2) when the ionic strength of the solution was varied from 1 M KCl to 200 mM KCl in both compartments. Although the separation between the nucleobases decreased with the ionic strength of the solution, we found that 2N-αHL pores separated C, T, A, and G nucleobase at all ionic strengths tested (figure 4(A)). Both the conductance of the nanopore (G_O^sym) and the DNA blocked pore (G_B^sym) decreased with the ionic strength of the solution. However, while G_O correlated linearly with the conductivity of the cis and trans solutions (figure 4(B), blue diamonds), G_B decreased less rapidly (figure 4(B), yellow triangles). For example, an 80% reduction of the conductivity of the cis and trans solutions resulted in ∼80% reduction in G_O^sym and ∼50% reduction of G_B^sym. This effect, which has been observed with solid-state nanopores [36] and ion channels [37], has been explained in terms of an additional ionic current through DNA blocked pores that is due to counter ions (K⁺ ions this case) accumulating along the DNA polymer or near the surface charged of the nanopores [36].

Zoom In Zoom Out Reset image size

DNA blocked nanopores are cation selective (${{P}_{{\rm K}+}}/{{P}_{{\rm Cl}}}\;=\;\gt 100$ when the cis and trans sides contain 300 and 500 mM KCl, respectively [38]), indicating that in DNA blocked nanopores most charge carriers are K⁺ ions that at positive applied potential translocate through the pore from the trans to the cis compartment [39]. Therefore, we tested nucleobase recognition when the ionic strength of the solution was reduced in cis compartment alone. Our reasoning was that the cis compartment is more likely to host a DNA handling enzyme [12, 21, 33, 34], the activity of which is usually sensitive to high ionic strength, while the trans compartment would provide the pool of K⁺ ions that recognize the nucleobase. As expected, we found that the four nucleobases were better separated when the ionic strength of the trans compartment was maintained at 1 M KCl (figure 4(A)). Interestingly, we found that in asymmetric ionic solutions, the G_O^asym was the average of the values measured in symmetric solutions (i.e. ${{{\rm G}}_{{\rm O}}}^{0.2\,{\rm M}\,{\rm cis}/1\,{\rm M}\,{\rm trans}}\;=\left( {{{\rm G}}_{{\rm O}}}^{1\,{\rm M}\,{\rm cis}/{\rm trans}}+{{{\rm G}}_{{\rm O}}}^{0.2\,{\rm M}\,{\rm cis}/{\rm trans}} \right)/2).$ By contrast, the I_B^asym values hardly decreased with the ionic strength of the cis chamber (${{{\rm G}}_{{\rm B}}}^{0.2\,{\rm M}\,{\rm cis}/1\,{\rm M}\,{\rm trans}}=0.88\;{{{\rm G}}_{{\rm B}}}^{1\,{\rm M}\,{\rm cis}/{\rm trans}},$ figure 4(C)), further suggesting that most of the current through the DNA blocked pore arises from the translocation of K⁺ from trans to cis. In asymmetric solution the difference between the I_B of poly(dC)₄₀ and the oligonucleotides containing A and T at position 14 did not vary significantly with the ionic strength of the cis solution, while the I_B difference of the oligonucleotides containing G at position 14 decreased to about 1/3 of the initial value as the ionic strength of the cis solution is reduced to 0.2 M KCl (figure 4(D)).

Heptameric αHL proteins were prepared in vitro as described in detail [40]. In short, proteins were produced by expression in an E. coli in vitro transcription and translation system and assembled into heptamers on rabbit blood cell membranes. The heptamers were run in a 5% SDS-polyacrylamide gel and the region of the dried gel containing αHL heptamers was cut out, rehydrated and crushed in 10 mM Tris.HCl, pH 8.0, containing 100 μM EDTA. Aliquots of the purified proteins were stored at −80 °C.

Ionic currents were measured by recording from planar bilayers formed from diphytanoyl-sn-glycero-3-phosphocholine (Avanti Polar Lipids, Alabaster, AL). Currents were measured with Ag/AgCl electrodes by using a patch-clamp amplifier (Axopatch 200B, Axon Instruments, Foster City, CA) as described in detail [41]. The solution of the cis and trans chambers were exchanged manually from buffered solutions in 1 M KCl by replacing aliquots with buffered solutions containing no salt. Amplified electrical signals were low-passed filtered at a corner frequency of 10 kHz and data acquisition was performed at a sampling frequency of 50 kHz.