EPFL Lipid Brain Atlas Divides Mouse Brain Into 539 Distinct Lipizones

The study used laser-based mass spectrometry imaging on thin brain sections to generate roughly 7 million lipid measurements, then applied custom machine-learning tools to reconstruct a whole-brain model. Many lipizones matched known anatomical regions while others connected distant cell bodies to their projection targets, patterns not captured by gene or protein maps alone. White matter, previously viewed as uniform, resolved into chemically distinct territories, and the choroid plexus showed sharp lipid boundaries. Pregnancy induced larger lipid shifts than sex differences, including elevated galactosyl ceramide in myelin regions and cortical remodeling.

These findings establish lipids as spatial postal codes that coordinate distant but functionally related brain areas. The atlas supplies a healthy reference for investigating lipid dysregulation in depression, Alzheimer's disease, and neurodevelopmental disorders, where membrane composition changes are implicated but poorly localized. Because the map derives from only 11 mice and one species, it cannot yet address human variability or disease states directly.

Next steps require applying the same MALDI-MSI pipeline to human postmortem tissue and longitudinal rodent models of neurodegeneration to test whether lipizone boundaries shift before clinical symptoms appear. Integration with single-cell transcriptomics will clarify how lipid signatures arise from specific cell types.

The open atlas will serve as a baseline for testing whether targeted lipid interventions can restore regional chemistry in disease models.