Engineered MspA Nanopores for Real-Time Detection of Insulin, Its Pharmaceutical Analogues, and Degradation Products

Abstract Nanopore sensors have demonstrated single-molecule resolution for nucleic acids and small molecules, yet the discrimination and detection of folded proteins in complex biological matrices remains a challenge. Insulin is a demanding test case: it circulates in blood at picomolar concentrations, differs from clinically used pharmaceutical analogues by as little as a single amino acid, and its continuous monitoring would close the remaining open loop in artificial pancreas systems. Here, we show targeted mutations near MspA nanopore’s constriction-enabled discrimination of insulin from the pharmaceutical analogues insulin aspart─differing by one amino acids, and insulin glargine. Furthermore, we reported on insulin degradation by interstrand disulfide reduction and deamidation, a ubiquitous degradation pathway in protein pharmaceuticals that compromises both potency and safety. A machine-learning classifier built on 16 single-event features separated insulin from the blood background with a per-pore limit of detection of 16 nM in 25% v/v blood, suggesting that an array of ∼1000 pores could reach physiologically relevant concentrations. A single-molecule electrical readout for insulin is a requirement for wearable sensors for making an artificial pancreas and could be used for preadministration quality control of insulin pharmaceuticals.