Astrocytes & Amyloid-β Mouse Models of Alzheimer's Disease 

Last Updated Date: December 13, 2024

Authors: Laurent Potvin-Trottier, Ph.D., Robin Guay-Lord, M.Sc., Lionel Breuiland, Ph.D., Elodie Brison, Ph.D., Simone P. Zehntner, Ph.D., Barry J. Bedell, M.D., Ph.D.

Key Takeaways

• A fully automated pipeline using deep- and machine-learning techniques can detect and classify astrocytes as normal or hypertrophic, and generate a continuous hypertrophy score.
• Applying this approach to an APP/PS1 Alzheimer's disease mouse model revealed distinct, region-specific shifts in astrocyte morphology and plaque-associated microenvironments over time.
• Astrocyte morphology provides a sensitive metric of disease state, more sensitive than GFAP density alone.
• The ratio of microglia-to-astrocyte in the plaque proximity is a complementary metric distinguishing between later disease stages.

Astrocytes and microglia are thought to play a key role in many neurodegenerative diseases, including Alzheimer's disease and Parkinson’s disease. Understanding the specific subtypes, roles, and interactions of astrocytes and microglia is important to understand disease mechanisms and to identify and assess therapeutic targets. This work aims to evaluate the spatiotemporal dynamics of astrogliosis and microgliosis in the amyloid-β plaque microenvironment in an APP/PS1 transgenic mouse model of Alzheimer’s disease.

We previously developed an automated method to analyze the plaque microenvironment in multiplex immunofluorescence tissue sections, quantifying stain density and characterizing microglia morphology. We have extended this work and implemented a deep learning-based approach to identify, count, and localize astrocytes. Astrocyte morphology was assessed using an explainable machine-learning model to distinguish cells with hypertrophic morphology, indicative of reactivity. We leveraged these methods to measure spatiotemporal cellular changes in the plaque microenvironment in tissue sections stained for Aβ, Iba-1, GFAP, and DAPI.

Our model classified hypertrophic astrocytes based on distinctive morphological features, such as thicker, more branched processes, and generated a continuous morphological score. This morphological score provided a sensitive metric of the disease state, yielding more statistically significant changes than the GFAP stain density alone. Hypertrophic astrocytes localized to plaques, but more distantly than activated microglia. To study the spatiotemporal pattern, we quantified the density of reactive in the microenvironment as a function of plaque size. We found that microglia progressively accumulate as a function of plaque size in the immediate vicinity of the plaque. In contrast, the astrocytic response was more global, with the astrocytes being progressively excluded from the larger plaques. Leveraging this differential response, we developed a new metric, the ratio of microglia-to-astrocyte in the proximity of plaques, which was sensitive to the later disease stages.

Assessment of morphological characteristics can provide additional information about the astrocytic phenotype. These features may provide sensitive measures for preclinical assessment of putative disease-modifying therapeutic agents in rodent models of neurodegenerative diseases. 

Presentation Highlights