Rational design of peptides for epitope imprinting of polynorepinephrine: A plasmonic and machine learning integrated approach

Molecularly Imprinted Polynorepinephrine (MIPNE) has demonstrated superior performance for mimetic receptors production, facilitating their integration into techniques like Surface Plasmon Resonance (SPR), Biomimetic Enzyme-Linked ImmunoSorbent Assay (BELISA), and Bio-Layer Interferometry (BLI). Here we developed a multiplexed Localized Surface Plasmon Resonance (LSPR) assay to face the selection of appropriate epitope sequences for protein imprinting, a critical factor in optimizing MIPNE efficiency. The plasmonic properties of gold nanoparticles formed on MIPNE were used to classify epitopes as functional ( F ), uncertain ( U ), or dysfunctional ( D ). Feature extraction and machine learning analysis identified key physico-chemical descriptors influencing imprinting efficiency. Subsequent SPR testing confirmed the correlation between epitope selection and receptor performance. This study provides the first systematic approach for epitope selection in MIPNE, paving the way for their improved design and application in bioanalytics and biosensing. • Molecularly imprinted PNE is a green, cost-effective alternative to antibodies. • MIPNE enables efficient protein detection via the epitope imprinting approach. • Selecting the right epitopes is crucial for imprinting efficiency, yet remains largely unexplored. • We report an LSPR assay combined with ML analysis for epitope classification. • SPR testing confirmed the link between epitope selection and receptor performance.