Localizing and quantifying amyloid in three dimensions furthers understanding of spatial patterns of amyloid deposition and disease progression in experimental models of cerebral amyloid angiopathy (CAA). Serial two-photon tomography (STPT) is ideal for this purpose because it produces 3D datasets of the whole mouse brain in a matter of hours. We present an optimized pipeline for fluorescent whole-brain labeling of amyloid and vasculature, STPT imaging of labeled brains, machine learning-aided classification of vascular and parenchymal amyloid, registration of images to a standard reference atlas, and automated quantification of these signals across brain regions. We used traditional histological sectioning and slide mounted imaging to compare and optimize published methods of fluorescent amyloid and vascular labeling. An intraperitoneal injection of methoxy-X04 delivered the day before euthanasia was chosen to label amyloid. Cerebral vessels are visualized by transcardial perfusion of rhodamine-B isothiocyanate in gelatin. Labeled brains are imaged by STPT and rendered in 3D to appreciate regional amyloid distribution. Volumetric image datasets are quantified using a modification of our custom image analysis pipeline. Pixels are classified as vascular or parenchymal amyloid with supervised machine learning methods. Classified signals are registered into the Allen Institute for Brain Science Common Coordinate Framework version 3.0 for quantification. This strategy produces high resolution volumetric images amenable to automatic quantification of parenchymal and vascular amyloid deposits throughout the brain, appropriate for studies involving large numbers of experimental brains in rodent models of CAA. Figure 1: A 3D rendering of prediction probability maps of a whole labelled mouse brain depicting fluorescent signals that were automatically classified as parenchymal amyloid (cyan), vascular amyloid (yellow) and brain parenchyma.
|Original language||English (US)|
|State||Published - 2021|