Postmortem human brain material
Two hundred seventy-five autopsied brains were obtained from Kyoto University Hospital and Osaka Saiseikai Nakatsu Hospital from 1992 to 2009 through a process approved by an institutional research committee. As we described previously [1
], neuropathological diagnoses were made by thorough histopathological examination of extensively sampled brain sections (Supplementary Fig. 1). In brief, in all brains at least 20 different samples, including anteroinferior frontal region, anterior cingulate region, middle frontal gyrus, superior frontal gyrus, precentral gyrus, superior and middle temporal gyri, amygdala, hippocampus, entorhinal cortex, supramarginal sulcus, occipital lobe, basal ganglia, thalamus, cerebellum, and at least 3 levels of the brainstem, were systematically taken from the formalin-fixed brains (Supplementary Fig. 1). Among the 275 patients, 31 patients were pathologically proved to have CAA by hematoxylin and eosin (H&E) staining and then by β-amyloid (Aβ) immunostaining, all of which were included in this study. Autopsies were performed at 12.9 ± 11.9 h (mean ± SD) (range 1.5–45.5 h) after death. The average fixation time was 37 ± 57 days (range 6–330 days) (Supplementary Table 2).
Clinical and pathological diagnosis
The 31 patients consisted of 14 AD (mean ± SD, 81 ± 8-year old) and 17 non-AD patients (78 ± 8-year old). The breakdown of the 17 non-AD patients is listed in Supplementary Table 2. The premortem clinical diagnoses, causes of death, vascular risk factors, postmortem pathological diagnoses and other demographic and pathological data of the 31 patients are also shown in Supplementary Table 2.
The clinical diagnosis of dementia met the criteria of the Diagnostic and Statistical Manual of Mental Disorders IV [3
]. The neuropathological diagnoses of AD were made if the postmortem brains revealed the presence of frequent neuritic plaques in the neocortex (Consortium to Establish a Registry for Alzheimer’s Disease, CERAD) [29
], and NFT stage was no less than IV, according to the Braak and Braak neuropathological staging of Alzheimer-related changes [7
] as assessed with modified Bielschowsky staining. Two observers (T.Y. and Y.O.) assessed SP and NFT stage individually, and if required, a joint assessment was scrutinized under a two headed microscope. The diagnosis of diffuse Lewy body disease was also made by thorough histopathological examination of extensively sampled brain sections [27
]. The diagnosis of subcortical ischemic vascular dementia was made clinically [5
], and was retrospectively found to meet the pathological criteria outlined by Kalaria et al. [19
]: (1) the presence of bilateral diffuse white matter lesions, (2) the presence of lacunar infarctions in the perforator territory, and (3) the presence of arteriolosclerosis.
Grading of atherosclerosis
At autopsy, the atherosclerosis stage was consistently graded by one of the authors (T.Y.). The degree of atherosclerosis at skull base was classified into four grades: ‘normal’ (no atherosclerosis), ‘mild’ (the presence of patchy atheroma), ‘moderate’ (a severity that is intermediate between mild and severe), or ‘severe’ (the presence of atheroma along the entire length of the vessels). The above staging was re-examined and confirmed by another author (Y.O.) using the autopsy report and macroscopic images taken at autopsy.
Tinctorial and immunohistochemical staining
Tissue blocks were obtained from the frontal, temporal, parietal, and occipital lobes (Supplementary Fig. 1). The blocks were embedded in paraffin and sectioned at 12 μm thickness for Congo Red staining, and 6 μm thickness for other staining on a microtome. To minimize variability in staining intensity, tissue sections were prepared by the same technician and stained with or using freshly prepared tinctorial and buffer solutions. Routine histological assessment was carried out with Congo Red, H&E, Klüver-Barrera (KB), modified Bielschowsky, and Gallyas staining. Pearls-Stieda staining was added as it was needed. The rest of the blocks were used for immunohistochemistry, involving sequential incubation with primary antibody, appropriate biotinylated secondary antibody (diluted 1:200, Vector Laboratories, Burlingame, CA, USA), and avidin–biotin–peroxidase complex (1:200, Vector Laboratories) in 0.1 M phosphate-buffered saline (PBS, pH 7.4). The sections were visualized with 0.01% diaminobenzidine tetrahydrochloride and 0.005% H2O2 in 50 mM Tris–HCl (pH 7.6). The primary antibodies were mouse anti-Aβ8–17 (6F/3D) (1:100, Novocastra, Newcastle, UK), rabbit anti-cow glial fibrillary acidic protein (GFAP) (1:200, DAKO, Glostrup, Denmark), mouse anti-human paired helical filament-tau (AT8; 1:200, Thermo Scientific, Rockford, IL, USA), and mouse anti-cluster of differentiation 68 (CD68) (1:100, DAKO) antibodies.
Senile plaque and neurofibrillary tangle burden
The burden of neuritic plaques was classified into ‘none’, ‘sparse’, ‘moderate’, and ‘frequent’ categories in the cortical sections stained with the modified Bielschowsky staining according to CERAD protocol [29
]. The stage of NFTs was assessed according to the Braak and Braak neuropathological staging of Alzheimer-related changes [7
]. In this study, we used 6 μm thick paraffin sections using the modified Bielschowsky method as we have previously reported [52
]; because, the modified Bielschowsky method is far more effective over other silver staining methods in detecting NFTs.
Staging of CAA
For staging of CAA, 12 μm thick sections were stained with Congo Red and viewed with polarized light [13
]. CAA was classified into CAA Type 1 (affected capillaries with or without larger cerebral vessel involvement) and CAA Type 2 (affected leptomeningeal arteries, cortical arteries/arterioles, or rarely veins), as proposed by Attems et al. [4
]. CAA Type 2 was further analyzed, and was divided into three grades proposed by Vonsattel et al. [44
]: those with ‘mild’ (focal Aβ deposits in the smooth muscle layer of the vessel walls), ‘moderate’ (circumferential Aβ deposits in the smooth muscle layer of the vessel walls), and ‘severe’ (extensive Aβ deposition with morphological changes such as microaneurysms, fibrinoid necrosis, double barreling, inflammation, thrombus, or hemorrhage) (Fig. ). When several grades were observed in one case, the dominant grade represented the case.
Representative photomicrographs of various grades of CAA. Congo Red staining showing mild CAA (a), moderate CAA (b), and severe CAA associated with double barreling (c). Bars indicate 100 μm in a and c, and 50 μm in b
Definition and quantitative analysis of CMI
CMIs were analyzed in the same sections that were used for pathological confirmation of CAA. CMIs were defined as cerebral cortex lesions visible only microscopically [19
] and usually accompanied by reactive glial proliferation. Regions of interest with evidence of expanded Virchow-Robin space or microabscess as well as those accompanied by hemorrhagic changes or cortical laminar necrosis on H&E staining were excluded from analysis. Following further confirmation of CMIs with immunohistochemistry for GFAP and CD68, we determined the density of CMIs using a method reported previously [33
]. In brief, the number of CMIs in each lobe was counted in sections stained with H&E. Images of the H&E stained slides were scanned (GT-X770 EPSON, Nagano, Japan). The cerebral cortices were outlined on each slide and the areas were measured using the ImageJ software package (image processing and analysis in JAVA, ImageJ bundled with JAVA 1.43, NIH, USA). The number of CMIs per cm2
of the cortex was calculated as a measure of CMI density. The CMI density in the frontal cortex was the mean of the values obtained from the five areas (anteroinferior frontal region, anterior cingulate region, middle frontal gyrus, superior frontal gyrus, precentral gyrus), and that in the temporal cortex was the mean of the values obtained from the two areas (lateral and medial temporal). The CMI density in the parietal cortex was obtained from the parietal supramarginal gyrus, and that in the occipital cortex was obtained from the occipital calcarine cortex (Supplementary Fig. 1).
Assessment of white matter changes
Using H&E- and KB-stained slides cut coronally at the level of mid-hippocampus, parietal, and occipital lobes, we classified white matter lesions into four grades as reported previously [9
]: those with ‘normal’ (normal white matter), ‘mild’ (no appreciable reduction in axonal meshwork density, and a slightly increased number of reactive astrocytes), ‘moderate’ (a slight reduction of axonal meshwork density, a reduction of oligodendroglial cell nuclei, and a further increased number of reactive astrocytes), and ‘severe’ (a marked reduction of myelin, axons and oligodendroglial cell nuclei with relatively marked astrocytic reaction, loosely scattered macrophages but no complete cerebral infarction).
We used transgenic mice, C57BL/6-Tg(Thy1-APPSwDutIowa)BWevn/J [11
] (Jackson Laboratory, Bar Harbor, ME, USA), which overexpress the neuronally derived human APP gene, encoding the Swedish p.K670N/M671L, Dutch p.E693Q and Iowa p.D694N mutations, under the control of the mouse thymus cell antigen 1 (Thy1) promoter. Generally, Thy1-driven exogenous gene expression is not altered by hypoxic/ischemic condition [26
]. The mice were screened for transgene expression by polymerase chain reaction, and heterozygous mice were mated with non-transgenic C57BL/6J mice (Japan SLC, Hamamatsu, Japan). All mice were given free access to food and water.
Surgical procedures and rearing methods
Male heterozygous mice were subjected to either sham or BCAS operation using microcoils [22
]. Body weight and rectal temperature were measured, and blood pressure was monitored from the tail artery of the sham- or BCAS-operated mice. Under anesthesia with 1.5% isoflurane, the common carotid arteries were exposed through a midline cervical incision, and a microcoil with an internal diameter of 0.18 mm was applied to the bilateral common carotid arteries. Sham-operated animals underwent bilateral exposure of the common carotid arteries without inserting the microcoil. Rectal temperature was maintained between 36.5 and 37.5°C, and body weight monitored until the animals were euthanized. After the operation, the mice were housed in cages with a 12-h light/dark cycle (lights on at 7:00 a.m.). Three animal groups were prepared in this experiment (6 groups, n
= 4–7 per group, n
= 31 in total; detailed information given in Supplementary Table 3). The timing of sham/BCAS and the duration of cerebral hypoperfusion are described in Supplementary Fig. 2. In brief, Tg-SwDI mice were subjected to sham or BCAS operation at 10, 16, and 20 weeks of age, and they were subsequently killed at 18, 24, and 32 weeks of age, respectively, to assess histology. Wild-type age-matched C57BL/6J mice (n
= 5 per group) were also subjected to BCAS operation. The number of mice used in this study was minimized for ethical reasons, and all procedures were performed in accordance with the guidelines for animal experimentation from the ethical committee of Kyoto University.
Systolic blood pressure and cerebral blood flow measurements
Mice were thermostatically controlled at 37°C on a warming pad and cerebral blood flow (CBF) and systolic blood pressure (SBP) recorded preoperatively and postoperatively at 32 weeks of age (12 weeks after BCAS or sham operation). Mean value of ten replicate measurements of CBF or SBP was determined for each mouse. The SBP was monitored in conscious mice by the tail-cuff method (MK-2000ST; Muromachi Co., Kyoto, Japan). The CBF was measured in identically sized regions of interest (900 pixels) located 1 mm posterior and 2 mm lateral from the bregma by Laser speckle blood flow imager (Omega Zone; Omegawave, Tokyo, Japan) under anesthesia with 1.5% isoflurane after the periosteum was widely removed with fine-tip forceps and calibration was carried out with a calibration reference device (Calibrator S/N 080715-5, Omegawave, Inc., Tokyo, Japan). CBF values were expressed as a percentage of the preoperative value. Since repetitive CBF measurement leads to fibrous scar tissue build up and bone opacification, CBF was measured only at two time points.
Histological investigation in mice
Mice were anesthetized with sodium pentobarbital (50 mg/kg, intraperitoneal) and transcardially perfused with 0.01 M phosphate buffer (PB) in normal saline. The removed brains were immersion-fixed in 4% paraformaldehyde in 0.1 M PB, and embedded in paraffin. The brains were then sliced into 6 μm thick sagittal sections at 1, 2, 3, and 4 mm lateral from the midline, and subjected to H&E and modified Bielschowsky staining. Immunohistochemical staining was performed according to the same protocol as human tissue. Mouse anti-Aβ1–40 (BA27) (Wako amyloid kit, Wako, Osaka, Japan), mouse anti-Aβ1–42 (BC05) (Wako amyloid kit, Wako), and mouse anti-Aβ5–10 (6E10) (1:500, Covance, Princeton, NJ, USA) antibodies were used.
Densitometric analysis of mouse brains
The Aβ-stained slides were captured with a digital camera (BZ-9000 KEYENCE, Osaka, Japan). Then, using the ImageJ software, the densitometric analysis of Aβ was performed blindly to animal groups by setting regions of interest in the cerebral cortex, the hippocampus, and the leptomeninges with the identical threshold in the Aβ (6E10)-immunostained sections. Leptomeningeal vascular, as well as pial, Aβ accumulations were jointly analyzed as ‘leptomeningeal Aβ’.
For human samples, numerical scores were computed from the data analysis as follows: age, 65- to 93-year old; disease group, AD = 1 or non-AD = 0; the grade of atherosclerosis, 0–3; the severity of CAA, 1–3; SP burden, 0–3; NFT stage, 0–6; and the grade of WML, 0–3. We first performed univariate analysis to determine whether age, disease type, or the above pathological changes were predictive of CMI formation using Fisher’s exact test. Next, multivariate analysis was performed while taking into account the effects of all variables on the parameters measured, including CMI formation.
In mice, Student’s t test was used to evaluate possible differences between the sham- and BCAS-operated mice groups at each time point, and two-way ANOVA was used to test for the effect of age and operation on Aβ deposition in the hippocampus, cerebral cortex, and leptomeninges. Differences with p < 0.05 were considered statistically significant in all analyses used.