High-throughput 3D whole-brain quantitative histopathology in rodents.

Imaging pathophysiological markers in animal models is essential to understand brain disorders and to develop new therapies. To date, microscopic imaging of immunohistochemically-stained tissue section is the state-of-the art technique to reveal specific biomarkers in the brain. With the rising use of automated slide staining systems and whole slide imaging microscopes, the capabilities for production of histopathological material as well as image digitization have dramatically increased1. Histopathology is entering the era of big data but quantitative analysis remains a bottleneck. Adapted tools are still lacking to mine brain histopathological data and extract valuable biological information. Indeed, current standards for the analysis of brain histopathological markers heavily rely on manual intervention to delineate regions of interest (ROIs) and quantify the staining. Hence, data collection is usually restricted to a few tissue sections and a few manually defined brain ROIs. While this can be sufficient for many studies, it drastically limits the scope of the analysis in studies with low prior knowledge about biomarkers of interest, if any. It also heightens the risk of inter-subject and intra-subject variability. Moreover, as brain histological processing relies on tissue sectioning, direct spatial correlation of histopathological two-dimensional (2D) information with in vivo three-dimensional (3D) imaging modalities such as magnetic resonance imaging (MRI) and positron emission tomography (PET) is arduous2. Several approaches have been proposed for brain ex vivo 3D microscopy imaging. Notably, serial electron microscopy have been used for 3D reconstruction of small tissue volumes at the nanoscopic scale3,4 and allowed the fine analysis of neuronal networks5,6,7,8. At the microscopic scale, serial optical microscopy enabled the visualization of neuronal tracers and blood vessels throughout the entire brain9,10,11,12. Recently, clearing techniques such as ultramicroscopy13, CLARITY14, SeeDB15, CUBIC16 and 3DISCO17, have drawn considerable attention. Clearing protocols have been proposed as a preprocessing step to optically image thick tissues in 3D. Thick tissue immunostaining is possible with some of the aforementioned techniques but it requires long incubations to allow for antibody diffusion. Thus, for practical use, the brain often needs to be cut into thick sections before immunostaining18,19. Although, 3D reconstruction of brain histological section images has already been in use20,21,22,23,24, there is no generic method for quantifying cellular and pathological markers at the whole-brain level. Also, a pending question is how to make sense of this large amount of data to provide new biological insight. To address these issues, we propose an integrated approach that relies on block-face photography25,26, serial histology and 3D-HAPi (3D Histology Analysis Pipeline), an innovative image analysis pipeline freely available for non-commercial use (http://brainvisa.info). The method is scalable so that whole-brain analysis of multiple markers can be performed in dozens of animals in a short time. The proposed approach relies on two strategies in order to make sense of whole-brain histopathological data: (1) an ontology-based analysis which is used to extract information about the amount of biomarkers of interest at many anatomical levels of the brain and (2) a multimodal analysis which allows to spatially correlate several histopathological biomarkers across the whole-brain and to characterize signal from in vivo imaging with 3D histopathology. As a proof of concept of our method usability, we studied well-known mouse models of cerebral Aβ peptide deposition which is a neuropathological hallmark of Alzheimer’s disease27. In these mice, we thoroughly characterized Aβ peptide deposition, we evaluated the brain-wide effect of a new anti-amyloid immunotherapy, we spatially correlated neuroinflammatory cell markers with Aβ peptide deposition across the whole brain and we registered 3D histopathological data with in vivo contrast-enhanced MRI.