Structure-guided computational design of DNA tweezers for predicted recognition of the primary glioblastoma biomarkers S100A4 and midkine

Abstract

Glioblastoma multiforme (GBM) is among the most aggressive malignant brain tumors and is characterized by severe infiltration into surrounding brain tissue, rendering early detection exceedingly difficult with current diagnostic imaging methods. Chosen biomarkers S100A4, with a role in cell motility and tumor metastasis, and midkine (MDK), connected to tumor microenvironment remodeling and expansion, are both linked to GBM's infiltrative proliferation. This study investigates a computational framework for the structure-guided design of DNA tweezer nanostructures, evaluated for predicted interactions with S100A4 and MDK. AlphaFold 3 modeling predicted the three-dimensional structures of eight three-stranded, hinge-scaffolded DNA tweezer candidates. DNA structures and target proteins were paired together in molecular docking simulations conducted with HDOCK, from which docking analyses predicted favorable binding arrangements of the complexes identified. Among the constructs investigated, DT3_8 exhibited the most favorable predicted interaction profile across the integrated computational analyses. Comparative docking against additional S100 family proteins with similar electrostatic characteristics further suggested that the predicted interaction profile of DT3_8 may not be explained solely by nonspecific electrostatic complementarity. Collectively, these findings provide a computational proof of concept for the design of dual-target DNA tweezer nanostructures and establish a foundation for future development and experimental evaluation of DNA-based biosensing platforms for glioblastoma-associated biomarkers.