Nanopore technology has revolutionized the field of single molecule detection and sequencing, enabling efficient decoding of genetic codes and detailed exploration of molecular properties at the single-molecule level. A single-molecule passes through the nanopore modulates the ionic current through the volume exclusion effect, generating current blockage with unique patterns or “fingerprints”, which enable the characterization of molecular properties such as size, shape, surface charge, and interaction information at the single-molecule level. Professor Yi-Tao Long’s group has always focused on the label-free and precise identification of DNA sequences, peptide sequences and post-translational modifications of proteins using nanopore technology (Nat. Nanotechnol. 2022, 17, 1136-1146, Nat. Chem. 2023, 15, 578-586, Nat. Nanotechnol. 2024, 19, 1693–1701, Nat Chem Biol 2024, https://doi.org/10.1038/s41589-024-01734-x, J. Am. Chem. Soc. 2019, 141 (40), 15720-15729, JACS Au 2021, 1(7), 967-976, Angew. Chem. Int. Ed. 2022, 61, e202209970, Angew. Chem. Int. Ed. 2021, 60, 24582-24587). Their studies about nanopore detecting mechanisms show that the single-molecule current signals detected in nanopore reflect not only the volume exclusion effect of the molecules being analyzed but also the instantaneous noncovalent interactions between the molecules and sensitive sites within the pore. However, current research studies mainly focus on the stable step-like current for DNA sequencing and single-molecule recognition, while the brief fluctuations or spikes are frequently dismissed as random noise in feature extraction.
Recently, Yi-Tao Long’s group utilized the dynamic features in nanopore current signals to achieve precise single-molecule identification, confirming that the dynamic “fingerprint” features in the nanopore current signals are more effective than stable feature in precise single-molecule identification. They also proposed the stochastic collision model to elucidate the generation of dynamic current features through transient interactions in biological nanopores. This research provides an important theoretical foundation for further exploring nanopore signal features using machine learning.
The authors utilized K238Q Aerolysin nanopore with two sensitive sites to detect single-stranded DNA (ssDNA), identifying current signal with a pattern where transient spikes are superimposed on the two stable transition states (Figure 1). And both the stable transition feature and dynamic spike feature exhibit significant variances across different ssDNAs. Furthermore, the authors evaluated the effectiveness of stable transition feature and dynamic spike feature for single-molecule identification using machine learning (Figure 2). After detecting 10 ssDNA samples differing by a single base by nanopore, they developed a 1D ResNet50-based neural network to automatically extract dynamic spike features from the current signals (Figure 3). The study revealed that, compared to the stable transition feature, the dynamic spike feature significantly improved molecular recognition accuracy from 44% to 93%.
Through molecular dynamics simulations and single-molecule experiments using various mutated nanopores (Figure 4), the study further confirmed that the alternating stable transition feature results from the simultaneous interactions of the ssDNA molecule with the R220 and Q238 sites in the nanopore. During the persistence of these two stable states, the chemical groups on the DNA will experience transient collisions with amino acids within the nanopore. The potential energy between the two interaction chemical groups changes and then perturb the local electric field distribution, resulting in a sudden interaction force acting on the nearby ions, which ultimately generate the transient spike signals. The authors illustrated that the ionic current signals can be interpreted as 1D imaging of ssDNA by nanopore, with each transition step representing a distinct molecular snapshot. These dynamic spike signals, captured at each snapshot through random collision mechanism, provide unique molecular-specific interactions information from different perspectives. The alternating steps resemble consecutive snapshots taken from various “imaging angles”, offering complementary and comprehensive insights into single-molecule characteristics. This new understanding provides a new sight for the de novo design of nanopores with enhanced functionality, enabling more precise molecular sensing and broadening the scope of nanopore technology in single molecule analysis, DNA sequencing and protein sequencing.
Figure 1. Investigation of ionic current pattern of ssDNA in mutant K238Q aerolysin nanopore.
Figure 2. Evaluation of the effectiveness of R-T transition features in identifying single molecule.
Figure 3. Exploring the validity and interpretability of dynamic spike features.
Figure 4. R-T transition feature generation mechanism.
Figure 5. Stochastic collision mechanism of dynamic spikes features.
The work was online published onJanuary 2, 2025 in Journal of the American Chemical Society (Doi: https://doi.org/10.1021/jacs.4c13664), with Dr. Shao-Chuang Liu, Prof. Yi-Tao Long as corresponding authors, and Ph.D. candidate Jia Wang, Dr. Shao-Chuang Liu as co-first authors. This study was supported by grants from the National Key R&D Program of China (2021YFF1200200), the National Natural Science Foundation of China (22204073, 22027806, and 22090054) and Nanjing University Integrated Research Platform of the Ministry of Education.