1. In situ stepwise modification of 1,3,5-3NAM-AeL
The K238C AeL contains seven inward-facing cysteine residues at position 238. As covalent bonding of NAM to non-adjacent cysteine residues is favored In situ stepwise modification results in1,3,5-3NAM-AeL. To control the advancement a single K238C nanopore was inserted into a lipid bilayer, giving an open-pore current (Figure S1) in 3.0 M KCl and 10.0 mM Tris solution. Then, NAM solution was introduced into the reaction chamber with a final concentration of 10.0 µM. An irreversible current drop occurred as a single NAM molecule reacted with the thiol group at the K238C site to form a sulfur-carbon bond (S-C) (Fig. 1a-b). The step-by-step current drop can be followed until seven cysteine residues were all covalent bonded to seven NAM molecules (Figure S2). The addition of NAM into WT AeL or K238G AeL barely generates any blockades (Figure S3), demonstrating that the irreversible current drops were produced after addition by a single-molecule electrophilic reaction between cysteine residues (nucleophilic regent) and NAM molecules (electrophilic conjugated system). For following single-molecule sensing, the NAM molecule is introduced from the trans side (bottom opening of the pore) under a negative applied potential (Figure S4). The real-time ionic current recording reports the stepwise in situ modification, until the production of 1,3,5-3NAM-AeL (current drop in range from < 15%). Additionally, the side product of 3NAM-AeL isomers (current drop of > 15%) would be discriminated by current drop and those products are removed by breaking the lipid bilayer with a high voltage pulse. After building a 1,3,5-3NAM-AeL, the potential is switched into + 80 mV to change the direction of the driven force and to prevent further reactions. This action would not only prohibit the addition of the fourth NAM molecule but also drive the negatively charged peptides to translocate from the cis (mouth opening of the pore) to the trans side (Scheme 1 and Figure S4).
In our experiments, 83% of current traces only contain three successive current drops. This result suggests that the optimal stoichiometric ratio between the reactant NAM and the K238C AeL monomer is 3:4. That is, three of seven monomers were modified with NAM molecules (Fig. 1a-b). We further explore the current drop distributions to understand the stoichiometric selectivity and regioselectivity during the modification. As noted in Fig. 1b, the first three current drops of I1, I2, and I3 can be assigned to the modification of one, two, and three NAM molecules, respectively. The corresponding amplitude of current drops is defined as △I1, △I2, and △I3, respectively (Fig. 1b). In details, the addition of a single NAM molecule to K238C AeL yields 1NAM-AeL, which produced △I1/I0 = 3.63 ± 0.60% (n = 34, Fig. 1c). The second NAM molecule was possibly added to 2, 3, or 4 positions due to the steric preference, generating the three isomers of 1,4-2NAM-AeL, 1,3-2NAM-AeL, and 1,2-2NAM-AeL, respectively (Fig. 1a). This current distribution is 2-fold wider than 1NAM-AeL (Fig. 1c). After clustering by kernel density evaluation (Fig. 1c), the △I2/I0 is classified into two distributions of D2-1 (center at ~ 4.2%) and D2-2 (center at ~ 6.5%). The corresponding ratio of those two kinds of events is 85% and 15%, respectively. As noted by previous work37, the different relative positions of modification inside nanopores may cause multiple current distributions. Similarly, the different relative positions of the two NAM molecules could induce various orientations inside the nanopore, which might be responsible for the broad ΔI2/I0 distribution. To verify our hypothesis, we conducted MD simulations for three isomers of 2NAM-AeL (Fig. 1d-f). The spatial probability densities of the modified two phenyl groups are evaluated by projecting 238 sites at XY plane (Z = 0). As a result, both two phenyl groups of 1,4-2NAM-AeL extend away from the inner wall, which is similar to the phenyl group inside 1NAM-AeL (Figure S5-S6, Supplementary Video). Z’ is defined as the distances along the Z axis between the centroid of benzene and Cα of bonded Cysteine (Figure S5). Figure 1d shows the Z’ of the phenyl group at 1 and 4 positions illustrates a flexible conformation of NAM due to the rotatable sp3 C-C and C-S bond. As to 1,3-2NAM-AeL, a closer distance between the two NAM molecules inside a nanopore induces one phenyl to stick to the inner wall of the nanopore, and the other phenyl extends away from the inner wall (Fig. 1e). The Z’ of 1,3-2NAM-AeL shows a narrow distribution of phenyl group at position 1(or 3) and a wide distribution of phenyl group at position 3(or 1). Due to the steric hindrance, the two neighboring phenyl groups of 1,2-2NAM-AeL exhibit the narrow distribution of Z’, which sticks to the inner wall of the pore (Fig. 1f). Compared to K238C AeL, two phenyl groups of 2NAM-AeL produce a narrower diameter of around K238C site (Figure S7). Consequently, the statistic results of confined ion numbers (K+ and Cl− ions) around the K238C site of 2NAM-AeL are less than K238C AeL, showing an order of K238C > 1,4-2NAM-AeL ≈ 1,3-2NAM-AeL > 1,2-2NAM-AeL (Figure S8). The lower current drop distribution of D2-1 could be assigned into 1,4-2NAM-AeL or 1,3-2NAM-AeL, while the higher current distribution drop of D2-2 is attributed to 1,2-2NAM-AeL. This result further confirmed that the conformation of phenyl groups inside K238C AeL largely affects the confined ion number, arousing the difference in the current drop. As a control experiment, the reactant of N-ethyl-maleimide (NEM) molecules without the phenyl group was added in the cis chamber (Figure S9). The results showed that NEM did not display heterogeneous current distribution at the second current drop (△I2/I0), which further supports that the current drop difference is related to the spatial orientation of phenyl groups from the modified NAM.
As a third NAM molecule covalent bonded with the 238 site, similar to I2/I0, the current distribution of △I3/I0 is divided into two regions, D3-1 and D3-2 (Fig. 2c). The resulting site-selectivity of 3NAM-AeL depends on the relative position of 2NAM-AeL. According to the reaction pathway shown in Fig. 1a, both 1,3-2NAM-AeL and 1,4-2NAM-AeL can produce 1,3,5-3NAM-AeL where three NAM molecules occupy three non-adjacent positions. Among all four 3NAM-AeL isomers, a complete reaction network analysis predicts that 1,3,5-3NAM-AeL owns the highest theoretical yield of 55% (Figure S10a and Supplementary Note 1). During the In-situ production, > 60% of events drop into the (I0-I3)/I0 range from < 15.0% resulting from the production of 1,3,5-3NAM-AeL (Figure S10b-e, and Supplementary Note 2). The nanopore current proves that 80% of 1,3-2NAM-AeL or 1,4-2NAM-AeL transfer into 1,3,5-3NAM-AeL (Table S1). Due to the steric hindrance, 1,2-2NAM-AeL with two neighboring phenyl groups cannot transfer into this current state.
We further explored the structure of 1,3,5-3NAM-AeL through MD simulations and single-molecule experiments (Fig. 2a-d). Compared with the unmodified K238C AeL nanopore, this hetero-nanopore of 1,3,5-3NAM-AeL provides a lateral hydrophilic-hydrophobic gradient. The diameter at the 238 sites is further reduced to ~ 6 Å (Fig. 2b, left). The electrostatic potential rises higher at the 238 sites, revealing a stronger electrostatic field under the applied potential (Fig. 2b, right). In the top view of the pore (Fig. 2c), the phenyl group at position 3 extends towards the center of the pore. It is stabilized by the other two phenyl groups at positions 1 and 5 through π-π interaction (Fig. 2d-e). These three phenyl groups formed a stable hydrophobic “clamp” facing the inward hydrophilic residues of β-barrel. In summary, the aforementioned features allow the enhanced interaction between the single analyte and the nanopore to improve sensing accuracy (Figure S11).
2 Single-molecule sensing ability of hetero-nanopores
To achieve the desired single-molecule sensing, the applied potential is switched from negative to positive to drive the negatively charged analyte from the cis chamber to the nanopore. This voltage shifting could prohibit the following addition of NAMs and reduce background signals from the remaining NAM in the trans solution (Figure S12). After the In situ production of 3NAM-AeL, we tested the conductance pattern and the incomplete isomers ((I0-I3)/I0 > 15%) were removed by destroying the membrane with a voltage pulse.
The model molecules of polydeoxyadenines poly(dA)4 were pulled into a NAM-AeL to study its single-molecule sensing ability (Fig. 2a). The results show that the statistical residence time of poly(dA)4 with the 1,3,5-3NAM-AeL is nearly two times longer than that with K238C AeL because of the enhanced electrostatic barrier and narrow pore size at the 238 sites (Fig. 2e). Moreover, the durations from 1,3,5-3NAM-AeL sensing are centered with a slight shift (n = 4, Figure S13 and Table S2), illustrating the stable configuration of three phenyl groups inside hetero-nanopore leads to the reproducible interaction between analyte the pore. In contrast, 1NAM-AeL and 2NAM-AeL exhibited shorter durations for sensing the single poly(dA)4.
To emphasize the importance of the hydrophobic "clamp" in single-molecule sensing within 1,3,5-3NAM-AeL, a control experiment was performed using NEM without phenyl groups. Since the short ethyl group cannot extend into the center of 3NEM-AeL, the NEM-AeL exhibits a weak interaction with the poly(dA)4. Therefore, with the three-step-wise modification of the NEM, the duration of poly(dA)4 is gradually reduced from K238C AeL to 3NEM-AeL, respectively (Fig. 2f and Table S3). Furthermore, the deviations in duration times of poly(dA)4 inside NEM-AeLs are greater than those observed inside 1,3,5-3NAM-AeL. Another control experiment is conducted with a hydrophobic homo-nanopore K238Y AeL in which the seven cysteine residues at the 238 site are replaced by Tyrosine residues. The residence time of poly(dA)4 is shorter than that in K238C AeL38. This result further proved that the rigid structure of the hydrophobic “clamp” inside 1,3,5-3NAM-AeL is favorable for a uniform interaction manner.
2.3 Identifying amino acid stereoisomers in a single peptide.
The discrimination of a single amino acid stereoisomer from a single peptide is still challenging due to their similar structures and identical molecular weight. Our group previously demonstrated that the heterogeneous distribution of charged residues of a WT OmpF provides a strong lateral electrostatic field at the constriction for identifying peptide epimer30. Here, the hydrophobic “clamp” causes a lateral hydrophilic-hydrophobic gradient inside a 1,3,5-3NAM-AeL. It could provide a laterally anisotropic interaction with the hydrophobic amino acid, which empowers the discrimination of single hydrophobic amino acid stereoisomers from a single peptide.
For the demonstration, we designed a series of peptides N’-DDFFXFFDD (X is L-Leu, D-Leu, L-Ile, and D-Ile) named P-L-Leu, P-D-Leu, P-L-Ile, and P-D-Ile, respectively (Fig. 3a). A previous study showed that the WT AeL hardly discriminates the pure Leu and Ile contained peptides due to the heavily overlapped blockage current39. Herein, we conducted multiple-channel nanopore experiments on the Orbit Mini (Nanion, Germany, Figure S14) to enable parallel sensing of peptide stereoisomers. Within a recording period of 20 minutes, we obtained over 10,000 events. The 1,3,5-3NAM-AeL significantly enhances the interaction with the peptide isomers, leading to the naked-eye distinguishable △I/I0 difference for direct discrimination of P-L-Leu and P-L-Ile from the mixture. Further statistical analysis reveals that the discrimination accuracy reaches as high as 97% (Fig. 3b), which has not been reported before. In contrast, the unmodified K238C AeL can not identify the P-L-Leu and P-L-Ile (Figure S15a), which further supports the vital role of the hydrophobic “clamp” in single-molecule sensing. More surprisingly, 1,3,5-3NAM-AeL can identify the unlabeled Leu/Ile and their enantiomers. As shown in Fig. 3c, the P-L-Leu and P-D-Leu with single amino acid chirality difference of leucine could be discriminated through the standard deviation of current blockade fluctuations (std). To the best of our knowledge, no single-molecule methods have exhibited such excellent discrimination ability for enantiomers so far. Finally, the four stereoisomers could be clearly identified from the mixture through std and I/I0 clustering (Fig. 3e), which further demonstrates the strong enantiomer recognition ability of 1,3,5-3NAM-AeL.
With a view to quantifying the peptide stereo-isomers, we combine the capture efficiency of the four peptides and the numbers of events in the mixture with different ratios (see details in Supplementary Note S3). As is shown in Fig. 3f, the presence of 10 µM P-L-Leu, P-D-Leu, P-L-Ile, or P-D-Ile in the mixture with different ratios could be identified and quantification. Peptide abundance measured by the 1,3,5-3NAM-AeL in the peptide stereo-isomers mixtures matched relatively well with different ratios. Hence, 1,3,5-3NAM-AeL could be used for the identification and quantification of peptide stereo-isomers that are not easily studied by mass spectrum.