Altered astrocyte morphology in pathogenic HEPACAM variants: a multipoint analysis and new machine learning framework for 3D Sholl analysis
Astrocytes are morphologically complex glial cells that play critical roles in brain development and function. Altered astrocyte morphology is associated with altered astrocyte function and is a common feature of many neurological disorders. Astrocytes express numerous membrane proteins that are important for their morphogenesis, including hepaCAM, an astrocyte-enriched cell adhesion molecule that regulates astrocyte branching organization, tiling, and coupling. Pathogenic variants of HEPACAM that impair homophilic protein interaction cause megalencephalic leukoencephalopathy with subcortical cysts (MLC), a rare and early-onset leukodystrophy characterized by white matter edema, seizures, and cognitive and motor decline. Pathogenic variants show altered subcellular localization and impaired interaction with key binding partners, but the impact on astrocyte morphology remains unexplored. Here we expressed three different dominant pathogenic HEPACAM variants in astrocytes of the developing mouse cortex and performed a comprehensive multipoint analysis of astrocyte morphology. Using established analysis workflows and a new machine learning model for efficient 3D Sholl analysis, we found small, but significant increases in morphological complexity for the G89S pathogenic variant, which impairs homophilic cis interaction of hepaCAM, and the Q56P pathogenic variant, which impairs homophilic trans interaction. This phenotype is distinct from the morphological changes we previously observed in Hepacam knockout astrocytes, suggesting that dominant variants may impact astrocyte morphology through a gain-of-function, rather than a loss-of-function, mechanism. Our study also provides a new machine learning workflow for streamlined 3D analysis of astrocyte branching complexity along with a framework for performing multivariate analysis of astrocyte morphology metrics across multiple conditions.