Nucleobase Rec ognition By Truncated α-Hemolysin Pores

Abstract

The α-hemolysin (αHL) protein nanopore has been investigated previously as a base detector for the strand sequencing of DNA and RNA. Recent findings have suggested that shorter pores might provide improved base discrimination. New work has also shown that truncated-barrel mutants (TBM) of αHL form functional pores in lipid bilayers. Therefore, we tested TBM pores for the ability to recognize bases in DNA strands immobilized within them. In the case of TBMΔ6, in which the barrel is shortened by ~16 A, one of the three recognition sites found in the wild-type pore, R1, was almost eliminated. With further mutagenesis (Met113→Gly), R1 was completely removed, demonstrating that TBM pores can mediate sharpened recognition. Remarkably, a second mutant of TBMΔ6 (Met113→Phe) was able to bind the positively charged β-cyclodextrin, am₇βCD, unusually tightly permitting the continuous recognition of individual nucleoside monophosphates, which would be required for exonuclease sequencing mediated by nanopore base identification.

Graphical abstract

Engineered protein pores are being developed for use in biotechnology, including applications in molecular sensing.^1–3 There has been particular interest in a new generation of nanopore DNA sequencers that would operate cheaply and quickly at the single-molecule level,^{4, 5} and recently this approach has proved successful.^6–10 The aim of the present study was to examine truncated-barrel mutants (TBM) of αHL¹¹ for their ability to identify RNA and DNA bases,^12–18 with the view to facilitate the electrical read-out of the sequences of nucleic acid molecules. The 5 nm-long transmembrane β barrel of the αHL pore comprises the base recognition region of the protein and the wild-type (WT) pore contains three broad recognition sites, R1, R2 and R3.¹⁵ Previous work has shown that individual nucleobases, presented in a fixed DNA homopolymer or heteropolymer background, can be identified at each of the three sites.^{15, 16, 19, 20} However, base identification is context dependent and the signal (I_RES%) from a given base will be shifted when neighboring bases are changed. Context-dependent signals produce additional information that is useful for sequence determination.^{6, 8, 10, 21–23} However, the signal will be uninterpretable unless the number of reading heads and their width is restricted. The Mycobacterium smegmatis MspA pore has favorable properties for reading DNA sequences on single strands, because changes in the ionic current are dominated by a single reading head that spans 3–4 bases.^{9, 10, 24} In the present work, we attempted to reduce the number of reading heads in the αHL pore by using truncated pores²⁵ and thereby demonstrate an approach that might be generally useful for improving protein pores as sequence readers.

Our recent work has demonstrated that the αHL pore is able to withstand substantial truncations in the β barrel region and still form single channels in lipid bilayers.¹¹ The β barrel contains 14 antiparallel β strands, with each protomer of the heptameric pore lipid of contributing two adjacent strands, which are connected by a turn (amino acids Gly-126 through Ile-132, Figure 1). The strands themselves largely consist of alternating hydrophilic and hydrophobic amino acids, with the side chains of the hydrophilic amino acids pointing into the lumen of the pore, and the side chains of the hydrophobic amino acids pointing into the lipid bilayer (Figure 1B). To truncate the β barrel, rings of inward and outward facing residues from each of the two strands were sequentially deleted by PCR mutagenesis (leaving the turn sequence intact) to form truncated barrel mutant (TBM) proteins. The rest of the TBM sequences were unaltered, except the charged residues at the central constriction (E111 and K147), which were mutated to neutral asparagines (NN). In TBMΔ2, amino acids V124 and T125 were deleted from the “down” strand and amino acids G133 and G134 were deleted from the “up” strand. TBMΔ4 and Δ6 were formed by deleting additional pairs of amino acids from each β strand (Figure 1B). As the mutant proteins have been demonstrated to adopt WT-like folds,¹¹ it is estimated that with each sequential truncation, the protein becomes ~5 Å shorter in length, Figure 1C). To test the integrity of the barrel in TBM mutants, cyclodextrin¹¹ (CD) binding experiments were also carried out using β-cyclodextrin (βCD), heptakis-(6-deoxy-6- amino)-β-cyclodextrin (am₇βCD) and γ-cyclodextrin (γCD). CD binding within the β barrel of the αHL pore,^26–28 is sensitive to small perturbations in the structure of the pore²⁸ or the cyclodextrin itself.²⁹ Interestingly, while the TBMΔ6 bound am₇βCD weakly, the mutation Met-113→ Phe, which strengthens βCD binding in the untruncated pore,²⁸ dramatically improved am₇βCD binding to TBMΔ6, allowing am₇βCD to remain bound to TBMΔ6/M113F for more than 1.5 h (at potentials of +60 to +140 mV).

Figure 1.

The α-hemolysin (αHL) protein nanopore. (A) Cartoon representation of the αHL pore (pdb: 7AHL). The αHL protein forms heptameric nanopores in lipid bilayers. The pore consists of an upper cap domain which contains a roughly spherical, water-filled vestibule, and a transmembrane domain. The E111N and K147N mutations¹⁵ at the top of the β barrel are highlighted in red, which replace the charged residues (Glu and Lys) at the central constriction. (B) The transmembrane domain is a 14-stranded, antiparallel β barrel. Each of the seven protomers contributes a pair of adjacent β strands (separated by a turn sequence; amino acids 126–132) to the barrel. The amino acid sequence of the transmembrane portion of the β strands for the most part alternates between hydrophilic residues (which face inwards towards the water-filled lumen of the pore) and hydrophobic residues (which face outwards towards the hydrocarbon core of the bilayer). To truncate the β barrel, rings of inward and outward facing residues from each of the two strands were deleted by PCR mutagenesis (leaving the turn sequence intact) to form truncated barrel mutant (TBM) proteins.¹¹ The three nucleobase recognition sites within the β barrel, R₁, R₂ and R₃, are also indicated. The three nucleobase recognition sites within the β barrel, R₁, R₂ and R₃, are also indicated^{15, 16, 19} (C) Cut-through representations of the truncated mutants, TBMΔ2, Δ4 and Δ6, used in the present study. The length indicated is the distance between the Cα atoms of amino acids N17 (located at the top of the cap domain), of the 3^rd subunit, and T129 (located at the bottom of the transmembrane domain), of the 7^th subunit.¹¹

RESULTS AND DISCUSSION

Defining recognition elements within the TBM pores

The TBMΔ2, Δ4 and Δ6 pores were examined for the ability to discriminate single adenine bases, within immobilized poly(dC) oligonucleotides, in a similar manner to that previously established.^{14, 15, 30} A set of fourteen poly(dC) oligonucleotides was used, each containing a single adenine nucleobase. The adenine substitutions were in positions 7 to 20 relative to a3′ biotin tag (Figure S1, S7 and Table S1), positions that span the entire length of the β barrel in full-length αHL pores. The residual current difference, ΔI_RES% (with respect to poly(dC)), was plotted against the position of the adenine nucleobase for each of the truncated pores (Figure 2 and Table S2).

Figure 2.

The effect of β barrel truncations on adenine recognition along the length of the β barrel. (A) Schematic representation of a homopolymeric DNA oligonucleotide (blue circles), immobilized inside the TBMΔ6 αHL pore (grey, cross-section) through the use of a 3′ biotin-TEG (yellow)•streptavidin (red) complex. The nucleobase recognition sites (R₁, R₂ and R₃) within the β barrel of the untruncated pore are shown alongside.^{15, 16, 19} The differences in residual current (ΔI_RES%) between blockades caused by poly(dC) oligonucleotides containing a single adenine base (A_x) and poly(dC) (Table S2) for the αHL pores NN (pink) and (B) TBMΔ2 (blue), (C) TBMΔ4 (red) and (D) TBMΔ6 (green) are plotted. I_RES% values are mean values derived from Gaussian fits to event histograms. I_RES% = (I_RES/I_O) x 100. ΔI_RES% is defined as the difference in residual current between an A_x oligonucleotide and poly(dC) (I_RES% ^{Ax oligo} - I_RES% ^pC) from an individual experiment. The means of the individual ΔI_RES% values are plotted with s.d. values as error bars. A cross-section of the β barrel domain of the truncated αHL pores, filled with DNA indicating the positions of the immobilized bases, is shown in each case. (E) The differences in residual current (ΔI_RES%) between blockades caused by poly(dC) oligonucleotides containing a single adenine base and poly(dC) for the αHL pores TBMΔ6 (green) and TBMΔ6/3G (purple) (Table S3). A cross-section of the β barrel domain of the truncated αHL pore is shown (indicating the position of the mutation M113G) filled with DNA indicating the position of the immobilized bases. The data were collected by using a voltage protocol as described in Experimental Methods. Briefly, +160 mV was applied for 900 ms to drive the negatively charged, DNA into the pore, followed by −140 mV for 50 ms, to eject the immobilized DNA and a final step to 0 mV for 50 ms.

With each sequential truncation, the recognition region of the protein is reduced. The last nucleobase recognized by the full-length NN pore is at position 19 (relative to the 3′ biotin-tag) and after this position it is assumed that the immobilized DNA chain protrudes from the β barrel into the trans compartment.¹⁵ However, in the truncated mutants, TBMΔ2, Δ4 and Δ6, the last nucleobase positions recognized are 18, 17 and 15 respectively. This suggests that the DNA protrudes from the β barrel sooner in the truncated mutants, and as expected the length of the recognition region has been reduced. However, adenine recognition by the truncated mutants is remarkable; the progressive changes in the patterns suggest that recognition site R₁ (near the central constriction) has been weakened, with R₂ and R₃ remaining, despite the removal of amino acids from the trans entrance of the pore. This suggests that recognition at sites R₂ and R₃ is not solely due to the interaction of nucleobases with specific amino acid side chains located at the bottom of the β barrel in the full-length pore. In the case of R₃, the DNA conformation upon exit from the pore or interaction with lipid head groups may affect the ionic current.

The main goal of our experiments was to reduce the complexity of nucleobase recognition as it was felt that pores with more than two recognition sites would elicit current signals that would be too complex to be reliably interpreted.

From our initial findings, TBMΔ6 appeared to have very weak nucleobase recognition at the central constriction (R₁). Previous work has suggested that amino acid substitutions at position 113 (which, along with amino acids 111 and 147, comprises the central constriction) influence nucleobase recognition at R₁ with the M113G mutant providing the weakest nucleobase recognition.¹⁹ Given this, the combined TBMΔ6/M113G mutant (Figure S1) was made in an attempt to remove the residual R₁ recognition site from the TBMΔ6 mutant and create a pore with just two recognition sites, TBMΔ6/M113G. It was indeed observed that the R₁ recognition site was removed by this mutation, while R₂ and R₃ remain largely unchanged by comparison with the TBMΔ6 pore (Figure 2E and Table S3).

Probing recognition site R₃ of αHL TBM pores for four-base discrimination

As well as examining the β barrel of each of the TBM mutant pores for the ability to discriminate single adenine bases, the lowermost recognition site, R₃, was also tested for the capability to discriminate all four bases. Sets of four poly(dC) oligonucleotides were used, which contained a single G, T, A or C nucleobase substitution at a position designed to to interact with recognition site R₃. The position of the nucleobase substitution for each set differed, and was placed at a peak of recognition site R₃, which differs for each truncated pore (Figure 2). The set used to probe TBMΔ2 and Δ4, had the substitutions at position 16, and the set used to probe TBMΔ6 and TBMΔ6/M113G, had the substitutions at position 13. Although each of the TBM pores contained an R₃ recognition site that provided strong discrimination of adenine versus cytosine, only TBMΔ6 (and the Δ6/M113G derivative) retained an R₃ site capable of distinguishing the other nucleobases. Furthermore, the dispersion of the four current levels differs in the full-length pore and the truncated pores (Figure 3, Table 1 and Table S4). This implies that the amino acid deletions towards the trans entrance may have an effect on recognition site R₃.

Figure 3.

Single nucleobase discrimination in truncated αHL pores. Histograms of the residual current levels for NN, TBMΔ2, Δ4, Δ6 and TBMΔ6/M113G pores are shown. Three sets of four poly(dC) oligonucleotides were used, with each set containing either a single A, T, G or C nucleobase at a specific position. The position of the substituted base, N (purple), was designed to probe recognition site R₃ of each of the proteins. All experiments were conducted at least three times (Table S4), and the results displayed in the figure are from a typical experiment. (A) NN pores were interrogated with four oligonucleotides with the sequences 5′-CCCCCCCCCCCCCCCCCCCCCCNCCCCCCCCCCCCCCCCCBtn–3′. TBMΔ2 (panel B) and TBMΔ4 (panel C) pores were interrogated with 5′– CCCCCCCCCCCCCCCCCCCCCCCCNCCCCCCCCCCCCCCCBtn–3′. TBMΔ6 (panel D) and Δ6/M113G (panel E) pores were interrogated with 5′– CCCCCCCCCCCCCCCCCCCCCCCCCCNCCCCCCCCCCCCCBtn–3′. The data were collected by using a voltage protocol as described in Experimental Methods. Briefly, +160 mV was applied for 900 ms to drive the negatively charged, DNA into the pore, followed by −140 mV for 50 ms, to eject the immobilized DNA and a final step to 0 mV for 50 ms.

Table 1.

In order to calculate the overlap of the residual current distribution, each individual nucleobase peak seen in the I_RES% histograms (Figure 3 and 5) was fitted to a single Gaussian. The Gaussians were then normalized so that the probability of detecting a single nucleobase was equal. The overlap between each of the neighboring bases was calculated from the area of overlap of the normalized Gaussians. Overlaps ranged between 0 (no overlap) and 1.0 (identical distributions).

NN	G	A	T	C
G	1.00	0.75	0.42	0.00
A	0.75	1.00	0.67	0.25
T	0.42	0.67	1.00	0.58
C	0.00	0.25	0.58	1.00

TBMΔ2	G	A	T	C
G	1.00	0.47	0.47	0.53
A	0.47	1.00	1.00	0.00
T	0.47	1.00	1.00	0.00
C	0.53	0.00	0.00	1.00

TBMΔ4	G	A	T	C
G	1.00	1.00	1.00	0.00
A	1.00	1.00	1.00	0.04
T	1.00	1.00	1.00	0.04
C	0.04	0.04	0.00	1.00

TBMΔ6	G	A	T	C
G	1.00	0.46	0.00	0.54
A	0.46	1.00	0.54	0.00
T	0.00	0.54	1.00	0.00
C	0.54	0.00	0.00	1.00

TBMΔ6M113G	G	A	T	C
G	1.00	0.50	0.28	0.72
A	0.50	1.00	0.78	0.22
T	0.28	0.78	1.00	0.00
C	0.72	0.22	0.00	1.00

TBMΔ6/M113F	dGMP	dAMP	dCMP	dTMP
dGMP	1.00	0.57	0.00	0.73
dAMP	0.57	1.00	0.36	0.81
dCMP	0.00	0.36	1.00	0.81
dTMP	0.73	0.81	0.81	1.00
TBMΔ6/M113F	rGMP	rAMP	rCMP	rUMP
rGMP	1.00	0.60	0.45	0.00
rAMP	0.60	1.00	0.85	0.40
rCMP	0.45	0.85	1.00	0.55
rUMP	0.00	0.40	0.55	1.00

Continuous four-base mononucleotide discrimination using a cyclodextrin adapter

Successful identification of nucleoside monophosphates has been obtained with engineered αHL pores carrying cyclodextrin adapters, which can be non-covalently bound within the pore¹² or covalently attached for continuous base identification.^{13, 18} This approach has been further proposed as an exosequencing platform, where bases are cleaved from a DNA strand by a processive exonuclease and identified as individual nucleotides by the nanopore. To test the TBMs for this purpose, two further mutants of TBMΔ6, TBM (Met113→Phe) Δ6/M113F and (Met113→Asn) Δ6/M113N were prepared to test the ability to detect individual DNA and RNA mononucleotides (NMPs) (Figures 4A, S2). These mutations in the full-length αHL pore have been shown to bind cyclodextrin (CD) adapters strongly (~10⁴ times longer than the wild-type (WT) αHL pore).^{31, 32} Cyclodextrins in turn bind dNMPs and rNMPs, allowing their identification by current recording for potential exosequencing.^{12, 13, 18} In the present study, the cyclodextrin heptakis-(6-deoxy-6-amino)-β-cyclodextrin (am₇βCD) was added to the trans compartment, and the positively charged amino groups promoted an extended residence time for the CD, at positive applied potentials.^{12, 13, 18} The TBMΔ6 and Δ6/M113N pores released am₇βCD quickly (TBM Δ6: k_off = 4600 ± 300 s⁻¹; K_D= 143 ± 8 mM; TBMΔ6/M113N: k_off = 2200 ± 200 s⁻¹; K_D= 104 ± 4 mM) (Figure S3 and Table S5). Upon the addition of all four dNMPs (dGMP, dAMP, dTMP, dCMP) or rNMPs (rGMP, rAMP, rUMP, rCMP) to the cis compartment, additional current blockades were observed originating from the CD blockade level, which represented the binding of NMPs to am₇βCD. The difference in residual current between the two most widely dispersed current peaks were: TBMΔ6, ΔI_RES% ^OVERALL = 2.6 ±0.4% and Δ6/M113N, ΔI_RES% ^OVERALL = 2.8 ±0.6% for dNMPs (Table S6). By contrast, the TBMΔ6/M113F pore showed the remarkable ability to bind am₇βCD for over 1.5 h, thereby overcoming the difficult issue of using covalent chemistry to attach CDs (Figure 4B).^{12, 13, 18, 31, 33, 34} In the absence of am₇βCD, the Δ6/M113F pore remained open and passed a current, I_O ^TBMΔ6/M113F = 113.0 ± 6.0 pA (n = 9, independent experiments) at +120 mV, in 1 M KCl, 25 mM Tris.HCl, at pH 6.0. The addition of 40 μM am₇βCD to the trans compartment produced a blocked level (lasting for >1.5 h) with a residual current level I_RES-am7βCD = 65.0 ± 4.0 pA (n = 9, independent experiments) (Figure S3, typical open-states and CD binding traces for TBMΔ6, Δ6/M113N and Δ6/M113F pores).

Figure 4.

Continuous nucleobase discrimination in a truncated αHL pore with a cyclodextrin adapter. (A) Left: Schematic representation of individual DNA mononucleotides (blue circles), binding inside the TBMΔ6 pore (grey, cross-section) equipped with a cyclodextrin adapter (am₇βCD, orange). Right: Cartoon schematics of the TBMΔ6 β-barrel domain, showing the interaction of am₇βCD (orange) with the mutants M113F (red) and M113N (purple) as determined for the untruncated pore mutants.³² (B) Representative single-channel trace of the TBMΔ6/M113F pore, in the presence of 40 μM am₇βCD. The recording was made in 1 M KCl, 25 mM Tris.HCl, pH 7.5, at +120 mV. (C) Single-channel recording from the TBMΔ6/M113F•am₇βCD pore showing continuous deoxyribonucleoside monophosphate (dNMP) detection (cis: 10 μM dGMP, 10 μM dAMP, 10 μM dCMP and 10 μM dTMP). Data were acquired at +120 mV. The amplified signal was low-pass filtered at 5 kHz and sampled at 25 kHz with a computer equipped with a Digidata 1440A digitizer (Molecular Devices).

Upon the addition of dNMPs or rNMPs to the cis compartment, additional current blockades were observed originating from the CD-blockade level corresponding to the binding of NMPs (Figure 4C). Both dNMPs and rNMPs could be discriminated in I_RES% histograms (Figure 5A–B and Table 1) with ΔI_RES% ^OVERALL = 6.0 ± 1.4% for dNMPs and ΔI_RES% ^OVERALL = 5.5 ± 1.2% for rNMPs (Table S6). The products of the sequential differences (δ) between each of the four residual current levels in the histograms were used to measure the ability of the different mutant pores to discriminate between the four DNA and RNA nucleotides.¹⁹ A pore that is unable to discriminate between all NMPs has δ = 0 (i.e., the current levels of two or more NMPs overlap). For the experiments shown in Figure 5A and B, δ^dNMP = 5.4 ± 0.4 and δ^rNMP = 4.4 ± 0.6 were obtained. These results gave a significantly improved dispersion of the four standard nucleotides compared to previous studies using the full-length αHL pore.^{13, 18}

Figure 5.

Residual current histogram of nucleotide binding events within the TBMΔ6/M113F•am₇βCD pore. Data were acquired in 1 M KCl, 25 mM Tris.HCl, pH 7.5, at +120 mV in the presence of 40 μM am₇βCD (trans). (A) dNMP results corresponding to the experiment shown in Figure 4B. (B) rNMP results from a typical experiment in the presence of 10 μM rGMP, 10 μM rAMP, 10 μM rCMP and 10 μM rUMP (all cis). (C) k_off values for each NMP detected with the TBMΔ6•am₇βCD, TBMΔ6/M113F•am₇βCD and TBMΔ6 M113N•am₇βCD pores. Values of k_off were determined by using k_off = 1/τ_off, where τ_off is the mean dwell time for each rNMP in the pore. Each experiment was conducted at least three times.

The NMPs also showed variations in mean dwell time (τ_off) within the CD adapter, which were used to derive the dissociation rate constants k_off (1/τ_off) (Figure 5C).^{13, 18, 31, 34, 35} At higher potentials, the binding of the charged nucleotides to the cyclodextrin adapter was promoted, resulting in a general decrease in k_off (Figures S5–S6 and Tables S7–S8), which suggests that an optimal potential would be required to obtain k_off values suitable for continuous nucleotide identification and accurate base calling to accommodate the rate of cleavage by the exonuclease.

CONCLUSION

Truncated αHL pores were examined for their ability to identify RNA and DNA bases with the view to facilitate the cheap electrical read-out of the sequences of nucleic acid molecules. First, nucleobase discrimination in ssDNA was examined by using 3′-biotinylated oligonucleotides bound to streptavidin.¹⁵ Truncations in the β barrel of the αHL pore reduced the number of bases showing measurable interactions with the pore. The TBMΔ6 protein showed a weakened R₁ recognition site and it was possible to weaken this recognition site still further by additional site-directed mutagenesis to generate the TBMΔ6/M113G mutant pore, which displayed just two recognition sites: R₂ and R₃. Subsequent analysis of the TBMΔ6/M113G pore showed that recognition site R₃ was still capable of four nucleobase discrimination in ssDNA, although the order of the current levels differed from that found with the full-length αHL pore. Such changes in the order of the current levels have been noted previously.^{15, 16, 19}

The ability of a truncated pore to detect mononucleotides was also examined. Such detection is a requisite for single-molecule nanopore exo-sequencing,^{1, 4, 12, 13, 18} where bases are cleaved from a DNA or RNA strand by a processive exonuclease and identified as individual nucleotides by the nanopore.^1–3 Full identification of mononucleotides has been obtained with engineered αHL pores carrying cyclodextrin adapters, which can be non-covalently bound within the pore^{12, 18} or covalently attached for continuous base identification.^{13, 18} Remarkably, in the present work, the cyclodextrin adapter am₇βCD was found to bind essentially irreversibly to the mutant truncated pore TBMΔ6/M113F, allowing all four dNMPs and all four to be distinguished without breaks in current recording. This is as effective as working with a covalently attached cyclodextrin, an approach that requires difficult chemistry.^{13, 18, 33}

Early work on base identification in DNA strands used the αHL pore, but it has been shown that the MspA pore gives a wider dispersion of current levels^{9, 24} suggesting that radical protein engineering (rather than point mutation of known pores) to produce sharper reading heads might improve current peak separation still further. The present work shows that a pore in which three recognition sites have been reduced to two can be quickly developed demonstrating the potential of such an approach, which should be generally applicable to a variety of pore-forming proteins. If further reduction in the length of the αHL pore is required to facilitate base identification, then shorter barrels¹¹ (e.g. TBMΔ8) that do not form completely stable pores in lipid bilayers, may be inserted into solid-state apertures, to form protein•solid-state nanopore hybrids.³⁶ An additional advantage to the protein•solid-state hybrid system is that it is amenable to parallelization, which would make the throughput of nanopore sequencing competitive with highly parallel second generation sequencing technologies. Indeed, an array of 10⁶ nanopores, each sequencing 100 bases per second could sequence a human genome cheaply in around 10 minutes, a feat which would make genomic medicine readily available.

EXPERIMENTAL METHODS

Protein preparation

The αHL truncated barrel mutant (TBM) proteins were produced as described previously.¹¹ Aliquots of the purified protein were stored at −80°C.

Mutagenesis

The NN mutant αHL gene was prepared with a kit for site-directed mutagenesis (QuikChange II XL, Catalog no. 200522-5, Stratagene) and verified by sequencing. The αHL truncated barrel mutants (TBMΔ2, Δ4 and Δ6) were generated from the NN gene in a pT7 vector^{37, 38} by PCR mutagenesis and ligation-free in vivo recombination,³⁹ and their sequences were verified. αHL TBMs NN/M113G/Δ120-125/Δ133-138 (TBMΔ6/M113G), NN/M113N/Δ120-125/Δ133-138 (TBMΔ6/M113N) and NN/M113F/Δ120-125/Δ133-138 (TBMΔ6/M113F) were also prepared by PCR mutagenesis and ligation-free recombination by using the NN/Δ120-125/Δ133-138 gene (TBMΔ6) in pT7 vector as a template (Figure S3).

Planar bilayer recording

Electrical recordings were carried out with a planar lipid bilayer apparatus with bilayers of 1,2- diphytanoyl-sn-glycero-3-phosphocholine (DPhPC, Avanti Polar Lipids). Bilayers were formed⁴⁰ across an aperture (~100 μm in diameter) in a 25-μm thick polytetrafluoroethylene (Teflon) film (Goodfellow, Cambridge, Cat.⁴⁰ across an aperture (~100 μm in diameter) in a 25-μm thick polytetrafluoroethylene (Teflon) film (Goodfellow, Cambridge, Cat. #FP301200/10), which separated the apparatus into cis and trans compartments. The transmembrane potential is given as the potential on the trans side (i.e. the trans potential minus the cis potential, which was at ground). A positive current is one in which positive charge moves through the bilayer from the trans to cis side.

The aperture was first pre-treated with hexadecane in n-pentane (10 mg mL⁻¹). Electrolyte solution (0.5 mL: 1 M KCl, 25 mM Tris.HCl, 0.1 mM EDTA, pH 8.0) was added to both compartments. Then, DPhPC in n-pentane (5 μL, 10 mg mL⁻¹) was added to both sides. The pentane was allowed to evaporate and a bilayer was formed by lowering and raising the electrolyte level past the aperture. All current recordings were performed with a patch clamp amplifier (Axopatch 200B, Molecular Devices). αHL heptameric pores were added to the cis compartment.

Single base identification with the streptavidin-immobilization technique

ssDNA molecules, with a biotinyl group covalently attached to the 3′ end through a linker (Figure S1 and Table S1), were obtained from Sigma-Aldrich (UK). Solutions of the biotinylated ssDNAs, 100 μM in 1 M KCl, 10 mM Tris.HCl, pH 8.0, 0.1 mM EDTA, were pre-incubated with equal volumes of 25 μM streptavidin (Sigma-Aldrich) in the same buffer for at least 5 min. Oligonucleotides were added to the cis compartment to a final concentration of 400 nM. A voltage protocol was then initiated. First, +160 mV was applied for 900 ms to drive the negatively charged, DNA into the pore. The capture of a ssDNA molecule by an αHL pore is observed as a stepwise decrease in the open pore current level (I_O) to a lower, stable, current level (I_B). A potential of −140 mV was then applied for 50 ms to eject the immobilized DNA from the pore. The applied potential was then stepped to 0 mV for 50 ms. This one-second sequence was repeated for at least 100 cycles for each ssDNA sequence added. The amplified signal arising from the ionic current was low-pass filtered at 1 kHz and sampled at 5 kHz with a computer equipped with a Digidata 1440A digitizer (Molecular Devices).

Data were analyzed and prepared for presentation with pClamp software (version 10.1, Molecular Devices). Event searches were performed to obtain the average residual current level for each ssDNA blockade (I_RES). The mean I_B value for each oligonucleotide (I_RES) was determined by a performing a Gaussian fit to a histogram of the I_B values. The percent residual current blockade was I_RES% = (I_RES/I_O) x 100. In general, when comparing several oligonucleotide species, a single oligonucleotide species was first added to the cis chamber and the current traces required for the determination of I_RES recorded. Subsequently, a second, third and then fourth oligonucleotide was added, and additional currents recorded. When such experiments were repeated, the oligonucleotides were added to the chamber in a different order.

Nucleoside monophosphate base identification with a cyclodextrin adapter

The mononucleotides, 2-deoxyguanosine 5-monophosphate sodium salt (dGMP, >99%, Acros); 2-deoxycytosine 5-monophosphate disodium salt (dCMP, >95%, Fluka); 2- deoxythymidine 5-monophosphate disodium salt (dTMP, >97%, Fluka); 2-deoxyadenosine 5- monophosphate disodium salt (dAMP, >95%, Fluka); uridine 5-monophosphate disodium salt (rUMP, 99%, Fluka); cytosine 5-monophosphate (rCMP, >98%, Fluka); adenosine 5- monophosphate (rAMP, 99%, Acros); guanosine 5-monophosphate disodium salt (rGMP, 97%, Acros) and heptakis(6-deoxy-6-amino)-β-cyclodextrin.7HCl (am₇βCD, >99%, Cyclolab) were used without further purification.

TBMΔ6/M113F pores and dNMPs or rNMPs (10 μM) were added to the cis compartment, which was connected to ground. Once a single channel was obtained, am₇βCD (40 μM) was added to the trans compartment. Both compartments contained 0.5 mL of buffer: 1 M KCl, 25 mM Tris.HCl, pH 6.0. After the addition of am₇βCD to the trans compartment, a permanent drop in the current was observed (~70%). Data were typically acquired for measurement, at +120 mV. The amplified signal (arising from the ionic current passing through the pore) was low-pass filtered at 5 kHz and sampled at 25 kHz with a computer equipped with a Digidata 1440A digitizer (Molecular Devices). The data were analyzed and presented by using pClamp software (version 10.1, Molecular Devices). Events were detected by using the ‘Event Detection’ feature. The mean residual current value (I_RES) for each NMP interacting with am₇βCD was determined by performing a Gaussian fit to a histogram of the current values for individual blockades (I_B). The current blockade for each NMP was further described by I_RES%, where I_RES% = (I_RES/I_CD) x 100, where I_CD is the current in the presence of the cyclodextrin. The mean dwell time (τ_off) for each NMP was determined by fitting an exponential function to a dwell time histogram.