DNA translocation and unzipping through a nanopore: some geometrical effects

Abstract

This article explores the role of some geometrical factors on the electrophoretically driven translocations of macromolecules through nanopores. In the case of asymmetric pores, we show how the entry requirements and the direction of translocation can modify the information content of the blocked ionic current as well as the transduction of the electrophoretic drive into a mechanical force. To address these effects we studied the translocation of single-stranded DNA through an asymmetric alpha-hemolysin pore. Depending on the direction of the translocation, we measure the capacity of the pore to discriminate between both DNA orientations. By unzipping DNA hairpins from both sides of the pores we show that the presence of single strand or double strand in the pore can be discriminated based on ionic current levels. We also show that the transduction of the electrophoretic drive into a denaturing mechanical force depends on the local geometry of the pore entrance. Eventually we discuss the application of this work to the measurement of energy barriers for DNA unzipping as well as for protein binding and unfolding.

Figure 1

Schematic representations of the DNA inside the pore for both DNA orientations and both translocation directions. The tilt of the bases is purely schematic to illustrate the MD simulation results obtained in Mathé et al. (11). Note the equivalence of the static situations for DNA with their 5′ end first in the forward direction (respectively, 3′ end first in the forward direction) and DNA with their 3′ end first in the backward direction (respectively, 5′ end first in the backward direction).

Figure 2

Scatter plot of the normalized blocked current as a function of the event durations in the forward and backward directions. We discriminate two groups of events. The dark-colored events are identified as true translocations. The light-colored events are identified as blocked molecules or aborted translocations. The first group represents 90% of the 5700 events in the forward direction (A) and 62% of the 4700 events in the backward direction (B). In the forward direction, note the bimodal distributions corresponding to DNA, translocating with their 3′ and 5′ end first. In the backward direction, the bimodal distribution turns into a major cluster with a long tail to the low current and long time. We attribute the main cluster to DNA translocating with their 3′ end first and the tail to the other DNA orientation. See text for the clustering criteria.

Figure 3

Normalized current distributions for the selected events in the (A) forward and (B) backward directions. Double-Gaussian fits (continuous line) yield the following values for the mean blocked current: forward direction, i_b3′f = 10.6% ± 1.9% and i_b5′f = 14.6% ± 2.3%; backward direction, i_b5′b = 8.5% ± 3.6% and i_b3′b = 12.1% ± 2.4%. Each peak corresponds to the entry of the 3′ or 5′ end first as previously demonstrated in the forward geometry. In the case of the backward direction, the double peak does not clearly appear, and a single Gaussian poorly fits the data. Note that i_b5′f > i_b3′f and i_b3′b > i_b5′b. The 5′ end first in the forward direction and 3′ end first in the backward direction correspond to the same static configuration of the pore, with respect to the DNA strand.

Figure 4

Histograms of the event durations in the forward (A) and backward (B) directions for the 3′ (light) and 5′ (dark) end first. In the forward direction, both distributions are sharply defined with the characteristic times τ_3′f = 115 ± 5 μs and τ_5′f = 239 ± 5 μs, in agreement with the reported values in the literature. In the backward direction, the distributions are broader with the characteristic times of τ_3′b = 110 ± 10 μs and τ_5′b = 510 ± 40 μs. The 3′ end-first configuration is the faster one and translocates at the same speed for both pore orientations.

Figure 5

Possible configurations of dsDNA in the α-hemolysin vestibule and docking of dsDNA in the trans part of the stem. Note that, in the backward case, the double-stranded part can penetrate 1.4 nm into the stem. The Lysines 147 corresponding to the vestibule/stem constriction are highlighted in red. The threonines 125 are in orange. (For the sake of clarity, the 3′ single strand overhang is not represented here; however, were it to be represented, it should span the entire stem.)

Figure 6

Typical traces for translocation events of hp_polyA at 150 mV. The raw data is acquired at 10⁵ S/s and filtered at 10 kHz (red). The nonlinearly filtered data (black) and the multistep analysis (green) are presented for both translocation directions. (A) In the forward direction, the event has a well-defined blocked current level. (B) In the backward direction, the translocation events present a steplike noise with three levels: a high (I) level that occurs only at the beginning and end of the event; an intermediate level (II); and a low level (III). Histograms of these level values are shown on Fig. 7.

Figure 7

(A) Histogram of the normalized substep levels. The three-Gaussians fit (continuous line) has a major maximum at 2.3%, type III; an intermediate one at 7%, type II; and a minor maximum at 13.1%, type I. (B) Two-dimensional histogram of the sublevels versus the substeps number (up to the 20th substep). Eighty-five-percent of the first substeps have a blocked, type-I current. Eighty-five-percent of the second steps have type II values. Ninety-percent of the translocations end with a type-I sublevel. We have checked that this is not an artifact occasioned by the steep fall (respectively, rise) of the current from (respectively, to) the open pore level. This value of 13% is reminiscent of the blocked current value found for ssDNA entering with its 3′ end first on Fig. 2. Two-hundred-and-thirty translocations were analyzed. We interpret the different blocking levels as: I, the single-stranded overhang alone in the pore; II, the double-stranded part of the DNA partially blocking the trans pore entrance; and III, the double-stranded part lodged in the pore entrance cavity as depicted in panel A.

Figure 8

Scatter plot of the normalized blocked current versus translocation times for hp_polyA in the forward and backward directions and hp_blunt in the backward direction. Dark-colored points are the points selected as real translocations. In the forward direction, only those points that belonged to the main cluster were selected. In the backward direction, we applied a criterion based on the presence of a first and last substep between 10% and 18% (see text for details). Some hp_blunt events are unexpectedly identified as true translocations due to the width of the distribution of type II events and are treated as false-positive to the selection criterion.

Figure 9

(A) Distribution of the translocation times for the events identified as hairpin translocations in the forward and backward cases. Note that the most probable translocation time is five times slower in the backward geometry (50 ms 8 ± 5 ms) than in the forward geometry (10 ms ± 5 ms). The distribution in the backward direction spans also a decade more toward large unzipping times. We attribute these differences to different modes of force application on the unzipping fork and to enhanced thermal fluctuations of the unzipping force in the backward direction. (B) Schematic representation of the two possible opening modes: shear and traction.