ChemScan: A computational platform for drug selection in platelet-mediated delivery

Platelet-mediated drug delivery is a promising strategy for targeted therapy, particularly in oncology and cardiovascular medicine. A major obstacle to its clinical translation is the lack of tools for identifying drugs that can be efficiently loaded into platelets and released upon platelet activation. In this study, we hypothesized that the key determinant of suitability for platelet-mediated delivery is a drug's ability to accumulate in platelet secretory granules, which serve as natural intracellular reservoirs for the storage and stimulus-triggered release of small molecules. This accumulation is driven by protonation within the acidic environment of the granules, which reduces membrane permeability, prevents back-diffusion, and promotes retention within these compartments. We experimentally validated this concept using fluorescent dyes with defined acid-base properties. Based on this mechanism, we developed ChemScan, an interactive web platform for screening approved drugs for their suitability for platelet-mediated delivery. ChemScan integrates a database of 2063 pharmaceutical compounds annotated with key physicochemical properties, including basic pKa, acidic pKa, and logP, and supports multilevel filtering based on the World Health Organization Anatomical Therapeutic Chemical (ATC) classification system. The platform also provides a fluorescence-based metric, the DyeLeS score, which estimates the likelihood that a compound can be visualized experimentally through intracellular accumulation. Using these criteria, ChemScan identifies candidate drugs within seconds, offering a practical decision-support tool for researchers designing platelet-mediated delivery systems and facilitating experimental validation through its integrated fluorescence assessment. Beyond platelet-mediated drug delivery, the platform may support drug repurposing and experimental design workflows in which lysosomotropic behavior nd fluorescence-related properties are relevant.