Development and In Vitro Verification of a Polymersome for Blood-Brain Barrier Transport Through a Novel Machine Learning Model

Overview

Across neurological drug development, researchers struggle with the high selectivity of the blood -brain barrier (BBB), where most drugs are unable to pass from the blood into the brain. As such, drug-treatment of Alzheimer’s, Parkinson’s, brain cancer, and strokes remains difficult. Recent research identified polymersomes (polymer -based vesicular shells) as an avenue for transport of otherwise non -BBB- permeable drugs across the BBB. However, the number of discovered BBB -traversing polymersomes remains low, and they are not able to carry all drugs. Recently, machine learning has emerged as a powerful tool in medicine. This research developed a machine learning model to identify likely polymer candidates for polymersomic drug-delivery across the BBB. The model was programmed in Python using TensorFlow and trained on 7,807 molecules from the B3DB -database. It achieved 93% accuracy and identified 13 encapsulation candidates. The top candidate for BBB permeability, ammonio methacrylate (AM), had never been considered for BBB permeability before. To validate the model, a polymersomic nanoparticle (AM - DOX) was developed by encapsulating doxorubicin (DOX, an anti -tumor drug) with AM, for eventual passage across an in-vitro BBB model (parallel artificial membrane permeability assay). 500µM-DOX and 500µM-AM-DOX were separately introduced to the BBB model for 24 hours at 37 oC. While DOX was predictably unable to penetrate the BBB, the AM -DOX nanoparticle successfully passed, producing an equilibrium 250µM concentration su rrounding the barrier. This was validated via UV -Vis and ATR -FTIR spectroscopies, providing compelling evidence for a new, effective BBB -encapsulation polymer, identified via machine learning, to deliver treatments for a wide array of neurological disorders.