Nanopore Analysis of DNA Molecules

Our laboratory has developed a technology that can measure DNA structure and dynamics with angstrom-level precision. It is well suited to analysis of duplex termini, and it is amenable to high throughput experiments. The approach is based on a nanoscale pore formed by the alpha-hemolysin channel, which is inserted in a lipid bilayer. Using X-ray diffraction to analyze the assembled protein reveals the shape of the heptameric pore: a 2.6-nm-wide aperture leading into a slightly wider vestibule, which then abruptly narrows to 1.5 nm in the transmembrane region (Figure 1a,b). The 2.6-nm-wide aperture is formed by a threonine ring (T9), and the 1.5-nm-wide transmembrane region is formed by a ring of glutamic acid (E113) and lysine residues (K147).

Figure 1. Details of the alpha-hemolysin pore. a) Crystal structure of the assembled heptamer at 1.9 Å resolution (from Song et al, 1996). b) Dimensions of the heptameric pore. The limiting 1.5-nm-diameter aperture near the center is formed by a ring of glutamic acid (E113) and lysine residues (K147). The 2.6 nm-diameter opening at the entrance to the pore vestibule is formed by a threonine ring (T9). c) A scale model of ssDNA superimposed on the pore. The space filling residues are K147. Note that the ssDNA molecule is sufficiently narrow (~1.3 nm diameter) to fit through this limiting aperture. d) A B-form dsDNA molecule superimposed to scale on the pore. In this case, the space filling residues are threonine 9. The duplex can fit into the pore vestibule but it is too large (~2.2 nm diameter) to fit through the narrow aperture at K147.

Single-stranded DNA (ssDNA) is narrow enough to fit through the limiting aperture (Figure 1c), but double-stranded DNA (dsDNA) is too wide, and it is limited to entry into the pore vestibule (Figure 1d). In a solution of 1.0M KCl (pH 8.0), a 120 mV applied potential produces a steady current (Io) of 120 ± 5 pA at 23 ºC through the open channel. Translocation of single-stranded linear DNA (~1.3 nm diameter) reduces this current to ~14 pA (I/Io=12%). Each monomer within ssDNA traverses the length of the 10-nm pore in 1-3 microseconds at ambient temperature.

• View a movie

Double-stranded DNA causes a very different current blockade, as revealed by studies using synthetic oligonucleotide hairpins. We chose DNA and RNA hairpins as model duplexes, because they can be formed from short, highly pure oligonucleotides that can be designed to adopt one base-paired secondary structure in 1.0M salt at room temperature. The initial experiments involved a well-characterized DNA hairpin with a six-base-pair stem and a four-deoxythymidine loop. When captured within an alpha-hemolysin nanopore, molecules such as this can cause a partial current blockade (or ‘shoulder’) lasting hundreds of milliseconds (“B” in lower part of Figure 2) followed by a rapid downward spike (“C” in lower part of Figure 2). This “shoulder-spike” signature is consistent with two sequential steps: 1) capture of a hairpin stem in the vestibule (“B” in upper part of Figure 2), where the molecule rattles in place because the duplex stem cannot fit through the 1.5-nm diameter-limiting aperture of the pore; and 2) simultaneous dissociation of the six base pairs in the hairpin stem, thus allowing the extended single-strand to traverse the channel (“C” in upper part of Figure 2).

Recently we used the nanopore device to detect single-nucleotide substitutions in dsDNA using hairpin molecules with longer duplex stems [1]. Unlike current signatures caused by short hairpins (≤7 bp), individual blockades caused by 9- and 10-bp DNA hairpins gated between several discrete conductance states. We used a combination of Hidden Markov Models and Support Vector Machines to analyze these gating patterns [2], which allowed us to detect the identity and orientation of Watson-Crick base pairs at the termini of individual DNA hairpin molecules. The mechanism underlying these discrete current transitions involves binding between the protein and nucleotides of the hairpin loop or terminus and dynamic properties of the DNA molecule itself. The latter component is the focus of this study. Preliminary results are discussed in the following section.

Figure 2. Impedance of ionic current through the alpha-hemolysin pore by a 6 bp DNA hairpin. The current trace is shown in the lower panel. Each letter corresponds to a diagram in the top panel. A) Open channel current of ~120 pA. B) Capture of the hairpin molecule in the vestibule reduces the current to ~55 pA. Our data indicate that the hairpin loop is perched at the mouth of the vestibule and the stem is inside the vestibule. Note that the current resides at this amplitude for about 100 ms, which is more than 3 orders of magnitude longer than for a linear strand of similar length. C) When the duplex stem dissociates, the applied electric field pulls the resulting ssDNA through the limiting aperture, causing a transient spike to about 15 pA residual current.

Sequence-Specific Channel Gating Mechanisms: DNA Hairpins of 9 and 10 Base Pairs

Capture of a single 9- or 10-bp DNA hairpin results in a current signature with discrete steps (Figure 3, top panel) that are not observed for shorter hairpins. These current levels include four phenomena: an intermediate level (IL) that initiates all 9- or-10 bp hairpin events; an upper conductance level (UL); a lower conductance level (LL) that occurs only after an upper level (UL) trace; and spikes (S) down from the lower level that indicate close proximity of the terminal base pair to the pore limiting aperture. Before the ssDNA extends into the limiting aperture, another step called the “frayed state” (F) occurs, but this is too fast for capture on the current signature.

We have developed a working model to explain the current transitions caused by hairpin capture (Figure 3, bottom panel). In our model, each 9-bp hairpin is captured so that the terminal base pair can interact with amino acids in the vestibule wall near the limiting aperture formed by lysine K147 and glutamic acid E113 of alpha-hemolysin. Circular dichroism assays indicate that the 9-bp hairpin stem is predominantly a B form duplex in bulk phase. The length per base pair of B form DNA is 3.4 angstroms, therefore the total stem length would be 30.6 angstroms. The distance between the narrowest part of the vestibule mouth at the T9 threonine ring and the pore-limiting aperture at lysine K147/ glutamic acid

Figure 3. Current signature caused by capture of a 9bp hairpin in the alpha-hemolysin pore vestibule. The upper panel shows the current trace acquired at 10 kHz bandwidth. The four most prominent levels are UL (upper level), IL (intermediate level), LL (lower level), and S (spike level). The lower panel (a-e) shows the orientation and dynamics of the captured DNA hairpin that we propose to account for the current transitions. The F or ‘frayed’ state is an essential step prior to ssDNA extension into the limiting aperture, however it is too fast to observe using our current electronics. A detailed discussion of the blockade mechanism is given in the text.

E113 is 33 angstroms. Therefore, if the hairpin loop is perched at the ring formed by the threonine ring, the 9-bp stem would reach to ~3 angstroms of the limiting aperture. Given our uncertainty about the exact position of the hairpin loop and the finite precision of the alpha-hemolysin X-ray crystal structure (1.9 angstroms), the estimated position of the hairpin terminus is accurate within ±1 bp or ±3.4 angstroms.

What do the different conductance states represent? Our data indicate that the intermediate conductance state (IL) is caused by orientation and immobilization of the hairpin terminus when it interacts electrostatically with the 3´ terminal nucleotide and residues in the vestibule wall (Figure 3a). The dwell time for this intermediate conductance state is independent of the 5´ nucleotide. The IL state invariably transitions to the upper conductance state, UL (Figure 3b). In the model, this state corresponds to desorption of the terminal base pair from the protein wall and thermal motion of the hairpin stem in orientations that allow greater ion current flow through the limiting aperture. These orientations may be angular displacement of the hairpin terminus away from the channel axis or axial orientation of the molecule allowing ion current to flow along the major groove of the duplex stem.

From the UL conductance state, the hairpin may return to the IL state or it may transition into a third conductance state, LL (Figure 3c). Residence time in the LL state depends strongly on the identity of the terminal 5' base pair. The 3´ nucleotide alone has no effect on LL dwell time; however, it does appear to augment binding when it is base paired with a 5´ nucleotide. When the duplex end frays from this bound state (Figure 3d), the 3´ strand may extend and penetrate the limiting aperture, resulting in a transient spike (Figure 3e). Based on this model, it is clear that kinetics in several of the conductance states are strongly influenced by protein-DNA interactions that are unique to this system. For the UL conductance state, however, the model suggests that the end of the duplex stem is not bound to the protein. Recent experiments support this conclusion and highlight the sensitivity of the UL current to very subtle changes in stem sequence identity.

References

1. Nanopores and nucleic acids: prospects for ultrarapid sequencing. Deamer DW, Akeson M. Trends Biotechnol 2000 Apr; 18(4):147-51.
2. Water transport by the bacterial channel alpha-hemolysin. Paula S, Akeson M, Deamer D. Biochim BiophysActa. 1999 Apr 14; 1418(1): 117-26.