Chengzhen Hu

Saccharides play critical roles in many forms of cellular activities. Saccharide structures are however complicated and similar, setting a technical hurdle for direct identification. Nanopores, which are emerging single molecule tools sensitive to minor structural differences between analytes, can be engineered to identity saccharides. A hetero-octameric Mycobacterium smegmatis porin A nanopore containing a phenylboronic acid was prepared, and was able to clearly identify nine monosaccharide types, including D-fructose, D-galactose, D-mannose, D-glucose, L-sorbose, D-ribose, D-xylose, L-rhamnose and N-acetyl-D-galactosamine. Minor structural differences between saccharide epimers can also be distinguished. To assist automatic event classification, a machine learning algorithm was developed, with which a general accuracy score of 0.96 was achieved. This sensing strategy is generally suitable for other saccharide types and may bring new insights to nanopore saccharide sequencing.

Chemical reactions of single molecules, caused by rapid formation or breaking of chemical bonds, are difficult to observe even with state-of-the-art instruments. A biological nanopore can be engineered into a single molecule reactor, capable of detecting the binding of a monatomic ion or the transient appearance of chemical intermediates. Pore engineering of this type is however technically challenging, which has significantly restricted further development of this technique. We propose a versatile strategy, "programmable nano-reactors for stochastic sensing" (PNRSS), by which a variety of single molecule reactions of hydrogen peroxide, metal ions, ethylene glycol, glycerol, lactic acid, vitamins, catecholamines or nucleoside analogues can be observed directly. PNRSS presents a refined sensing resolution which can be further enhanced by an artificial intelligence algorithm. Remdesivir, a nucleoside analogue and an investigational anti-viral drug used to treat COVID-19, can be distinguished from its active triphosphate form by PNRSS, suggesting applications in pharmacokinetics or drug screening.

Enantiomers, chiral isomers with opposite chirality, typically demonstrate differences in their pharmacological activity, metabolism, and toxicity. However, direct discrimination between enantiomers is challenging due to their similar physiochemical properties. Following the strategy of programmable nanoreactors for stochastic sensing (PNRSS), introduction of phenylboronic acid (PBA) to a Mycobacterium smegmatis porin A (MspA) assists in the identification of the enantiomers of norepinephrine and epinephrine. Using a machine learning algorithm, identification of the enantiomers has been achieved with an accuracy of 98.2%. The enantiomeric excess (ee) of a mixture of enantiomeric catecholamines was measured to determine the enantiomeric purity. This sensing strategy is a faster method for the determination of ee values than liquid chromatography–mass spectrometry and is useful as a quality control in the industrial production of enantiomeric drugs.