UCL researchers developed a hydrogel-based axon model using tunable micropillar arrays to better mimic the geometry and softness of native axons in multiple sclerosis. Published in Nature Methods, the system is designed to improve early-stage evaluation of remyelination drugs by reducing mismatch errors created by overly rigid lab models. The study reports that when drug testing occurs under more physiologically realistic mechanical conditions, performance dropped for candidate therapies, suggesting prior screens may have yielded misleading “hits.” The platform uses polyacrylamide hydrogels tuned to match native softness and seeded oligodendrocytes to form multilayered compact myelin. For translation-focused teams, the key takeaway is a methodological shift: drug candidates may need reassessment in human-relevant physical environments before entering expensive clinical programs. The tool also reinforces the broader push for mechanobiology-informed preclinical screening.