Full-length protein classification via cysteine fingerprinting in solid-state nanopores.

Recent advances in single-molecule technologies are transforming the field of protein analysis. Solid-state nanopores provide an effective method to linearize and thread full-length proteins in a single file. However, slowing their rapid translocation remains a challenge for accurate, time-resolved ion-current-based fingerprinting. In this work, we present a click-chemistry-based strategy for covalently attaching short oligonucleotides to cysteine residues on denatured proteins across a broad range of molecular weights. The negatively charged oligonucleotides increase the capture rate by a factor of ten compared with native proteins and induce a distinct 'stick-slip' motion that slows protein passage through the nanopore by more than 20-fold. These oligonucleotide tags also produce characteristic, time-resolved ion current pulses that serve as unique protein-specific signatures. To uncover the physical mechanism responsible for the protein translocation dynamics, we model our system using all-atom molecular dynamics and finite element simulations. By leveraging a supervised machine learning classifier, we demonstrate that a small number of translocation events is sufficient to identify individual proteins, achieving near-perfect classification accuracy. To demonstrate the robustness of the method, we successfully distinguish between VEGF-A isoforms (VEGF-165 and VEGF-121), which are relevant to cancer diagnostics, within a mixed protein sample. Our nanopore-based fingerprinting technique eliminates the need for affinity reagents, such as protein-specific antibodies, or motor proteins, offering a rapid, direct and cost-effective approach for single-molecule protein identification and classification.