Fig. 5
Significant enrichment of PML bodies in human brain areas with high plaque load. (A) Haemotoxylin eosin (HE) staining of human hippocampal sections was used to define Cornu Ammonis (CA) 1–3, Gyrus dentatus (GD), and Plexus areal. Parallel sections were used for PML immunofluorescence staining. DAPI co-staining was used for low-resolution tile-scan imaging and allocation of specific CA areas according to the initial HE staining. (B) Hippocampal CA1 or CA3 areas were used for high-resolution tile scan imaging. A single scan included 5 × 5 images with 5 z-stacks, which were subsequently combined using maximum projection. (C) A representative high-resolution (100 × objective) confocal tile-scan image demonstrates identification of PML bodies in DAPI counter-stained nuclei. (D) Tile-scan imaging was then established with Thioflavin co-staining to identify areas with high plaque load in the human hippocampus (CA1 or CA3). (E) CellProfiler™ software was used to automatically annotate and extract nuclei (DAPI channel in grey scale) from the image (upper row). Subsequently, PML body identification and quantification was done in the extracted nuclei. (F) All nuclei containing 1 to 5 PML bodies were used to determine the percentage of PML positive nuclei (y axis). Hippocampal areas with high plaque load (plaque, average = 36.2%) revealed a significant lower percentage of PML positive nuclei compared to areas without plaque (no plaque, average = 44.1%) (p < 0.05; every dot (left to each bar) indicates a single tile-scan experiment; for areas with high plaque load 31 tile-scans were analysed and quantified, 28 for no plaque, 9 for control). Tile-scans of control sections (individuals without any plaque pathology, average = 53.8%) revealed again higher percentage of PML positive nuclei compared to “no plaque” areas (p < 0.05). (G) Detailed analysis of cells containing one up to ten PML bodies per nucleus revealed significant differences. Nuclei with a single PML body were underrepresented (p < 0.05) in areas with high plaque load versus areas without plaques. Difference to control tile scans were highly significant (p < 10−5). In addition, nuclei with two PML bodies (red bars) also revealed highly significant differences to the control. Representative images for nuclei containing one, two, or three PML bodies are given