Introduction
Deposition of granular osmiophilic material (GOM) is the vascular pathological hallmark of CADASIL, which is the most prevalent hereditary small vessel disease [
1] and is caused by missense mutations in the
NOTCH3 gene [
2,
3]. GOM have been shown to contain NOTCH3 ectodomain (NOTCH3
ECD) and extracellular matrix proteins [
4‐
6], and can be visualized ultrastructurally in the tunica media of small arteries and capillaries. These electron dense GOM deposits are located in the basement membrane of mural cells, i.e. vascular smooth muscle cells and pericytes [
7‐
11]. In both manifest and pre-manifest CADASIL patients, GOM deposits are present not only in brain vessels, but also in vessels of other organs, such as the skin [
11‐
13]. Other CADASIL-associated vascular pathology includes mural cell degeneration, smooth muscle actin (SMA)-positive (neo)intima formation, fibrosis and vessel wall thickening [
14‐
19]. These vascular alterations are associated with compromised cerebrovascular reactivity (CVR) [
20,
21] and reduced cerebral blood flow (CBF), and eventually lead to mid-adult onset of recurrent strokes, vascular cognitive impairment and ultimately dementia [
1]. Brain MRI reveals progressive symmetrical white matter hyperintensities, lacunes, microbleeds and brain atrophy [
1].
We have previously described that our humanized CADASIL transgenic
NOTCH3Arg182Cys mouse model, which overexpresses human mutant NOTCH3 protein from a genomic construct, shows granular NOTCH3
ECD immunostaining as early as 4 weeks of age, while GOM deposits first appear around 6 months of age [
22]. However, little is known about how GOM deposits evolve over time and what their relation is to other CADASIL-associated vascular pathology and vascular dysfunction. Here, we performed a longitudinal study of GOM pathology in the transgenic
NOTCH3Arg182Cys mice. In addition, we assessed cerebrovascular, motor and cognitive function in these mice.
Methods
Mice
Transgenic mice were used that harbour the human full-length
NOTCH3 gene (located on a 143 kb BAC construct) in either the wild-type or the mutant (c.544C>T, p.Arg182Cys) form, generated on a C57BL/6J background [
22]. Mice were bred at the animal facility of the Leiden University Medical Center and housed individually under standard conditions, i.e. a 12-h light/dark cycle with food and water available ad libitum.
Three different mouse strains were used, with various human
NOTCH3 expression levels: 100% for wild-type mice (tgN3
WT100), and 100% and 350% for mutant mice (tgN3
MUT100 and tgN3
MUT350, respectively) [
22]. Non-transgenic littermates were used as additional controls. A prospective study with 6–8 mice per group was performed to study body weight and motor function at various time points (1.5, 3, 6, 12, 16 and 20 months), and cerebral hemodynamics, cognition and immunohistochemical staining was studied at 20 months. Three mice had to be sacrificed before the end of the study; one due to an eye infection (tgN3
WT100, at 15 months), one due to having a wound on its back (tgN3
MUT350, at 19 months) and one due to low body weight (tgN3
MUT100, at 20 months). In addition to the prospective study, tgN3
MUT350 mice were sacrificed at the age of 1.5 (
n = 1), 3 (
n = 1), 6 (
n = 2), 12 (
n = 2) and 20 (
n = 2) months for electron microscopy (EM) studies.
Electron Microscopy
In addition to mouse brain, post-mortem brain tissue was obtained from three CADASIL patients (deceased at age 59, 66 and 69 years) for comparison with human pathology. Mouse (frontal lobe grey matter) and human (frontal lobe grey matter) brain tissue was fixed overnight at 4 °C in 1.5% glutaraldehyde and 1% paraformaldehyde (pH = 7.4). Tissue blocks of ≤ 1 mm3 were post-fixated for 90 min in 2% osmium tetroxide and 2% potassium ferrocyanide after filtrating the post-fixative through a 0.2-μm filter. After post-fixation, the tissue was washed for 30 min in MilliQ and dehydrated in a series of ethanol (70%, 80% and 90%) for 30 min each and twice for 1 h in 100% ethanol. Blocks were incubated for 10 min in propylene oxide, 2 h in propylene oxide and Epon LX-112 (1:1) and finally for 2 h in propylene oxide and epon LX-112 (1:2). Subsequently, the epon was polymerized for 48 h at 70 °C.
One-micrometre-thick sections were checked for the presence of blood vessels by light microscopy. Then, areas with high blood vessel density were selected for further analysis with EM, i.e. 80-nm sections were collected on a one hole grid and subsequently stained with uranyl acetate and lead citrate. Images were acquired with a digital camera (One View, Gatan Inc., Pleasanton, CA) mounted on a 120 kV transmission electron microscope (Tecnai T12 with a twin objective lens, Fei Inc., Hillsburough, OR). Overviews of relatively large regions on the specimen that contained abundant numbers of cross-sections of vessels were collected by stitching many individual images together (40,000–60,000 nm
2) using software described earlier [
23]. Stitched images were examined using Aperio ImageScope (version 10.0.35).
GOM Analysis
The number of GOM deposits was counted per vessel and expressed as counts per 100 μm vessel circumference. Vessel circumference was approximated using the formula of oval circumference (Ramanujan’s approximation for ellipse circumference = π[3(a + b) − √((3a + b) × (a + 3b))]) where a and b were defined as the major diameter (a) and minor diameter (b) for each vessel between endothelial basement membranes. GOM deposits were studied in vessels with a minor diameter < 8 μm, referred to as microvessels, as these were the most abundant in the stitched images. Twenty-three to 84 microvessels were studied per time point (1.5, 3, 6, 12 and 20 months). Width of the basement membrane was determined averaging 40 measurements in two ntg mice and two tgN3MUT350 mice each. In the human brain sample, 21 microvessels were analysed. Also, the GOM deposit area was measured using ImageJ after manually drawn region-of-interests around GOM deposits.
Immunohistochemistry
Two 5-μm coronal frontal brain sections per mouse (approximately at the height of the infundibulum) and human brain sections of frontal white matter were analysed with immunohistochemistry. NOTCH3
ECD staining was performed as described before [
22]. Smooth muscle actin (SMA) staining was performed after pre-treatment with trypsin for 30 min at 37 °C and washed three times for 5 min with PBS. The primary antibody (Alpha-Smooth Muscle Actin, 1:4000, goat polyclonal, NB300-978, Novus Biologicals) was incubated overnight at room temperature. The secondary antibody (Rabbit Anti-Goat IgG, biotinylated, 1:400, Jackson Immunoresearch Lab. Inc.) was incubated for 1 h at room temperature and developed with the Vectastain Elite ABC HRP Kit (PK-6100, Vectorlabs) for 30 min at room temperature. Finally, slices were stained with 0.05% 3,3′-diaminobenzidine (DAB, Sigma) supplemented with 0.0045% H
2O
2 for 10 min and stained with Harris’ haematoxylin solution (diluted 1:3, Merck) for 5 s. Sections were Verhoeff-Van Gieson and Periodic Schiff acid stained using the Artisan Link Pro staining machine (DAKO, Agilent), and Van Gieson stained as described previously [
24]. Sections were stained with Klüver-Barrera luxol fast blue to quantify white matter vacuolization (Supplementary Methods
4).
Microscopy imaging was performed with the Keyence BZ-X710 (Keyence). Using 20 times magnification, full colour images were taken of the complete section at 1-ms capture time. Images were stitched by Keyence BZ-X Analyzer software version 1.3.0.3 to obtain one high resolution full brain image per section. The SMA positive area was determined using a Colour Threshold (Hue 0–50; Saturation 0–255; Brightness 0–175) in ImageJ, and expressed as percentage of total brain surface of the section.
Neuroimaging, Cerebrovascular Reactivity, Cognition and Motor Function
At the age of 20 months, neuroimaging (T2W, FLAIR, SWI, cerebral hemodynamics) was performed under medetomidine anaesthesia using a 7 Tesla MRI (Bruker PharmaScan), see also Supplementary Methods
1. In short, absolute cerebral blood flow (CBF) was measured using arterial spin labelling (ASL)-MRI during a 7-min baseline and a 7-min CO
2 challenge, using the measured signal difference between the labelled and control images in three brain slices. Absolute CBF at baseline and absolute CBF at challenge were quantified during the last 140 s of baseline and challenge, respectively. Absolute CBF increase was calculated as well as the cerebrovascular reactivity (CVR), which was defined as the relative cerebral blood flow (CBF) increase. One non-transgenic mouse was excluded from neuroimaging analysis due to a poor hemodynamic response to anaesthesia.
A Morris Water Maze protocol was used to assess cognitive function in 20-month-old mice, starting 2 weeks before neuroimaging assessment (Supplementary Methods
2). In short, during the training phase, mice were trained to find a hidden platform in the north-west quadrant of a circular swimming pool, while at the reversal training phase, mice were retrained to find a hidden platform in the south-east quadrant. Path length of the mice was determined between release in the pool and finding the hidden platform. Motor function was determined by analysing speed on a rotarod, on a beam and during swimming in the Morris water maze (Supplementary Methods
3). Motor and cognitive mouse experiments were performed by the same experienced researcher.
Statistical Analyses
Differences between groups were analysed using one-way ANOVA analyses with Tukey’s post-hoc correction. The increase in GOM area over time, as well as the association between baseline CBF and CVR were analysed using simple linear regression. All statistical analyses were two-sided tests with threshold for statistical significance of 0.05, using the IBM SPSS Statistics version 23.0.0.2 software.
Discussion
We investigated the development of GOM deposits over time in a transgenic mouse model of CADASIL, which overexpresses mutant human NOTCH3 protein from a large genomic construct. GOM deposits evolved with respect to size, morphology and number in microvessels of the mutant mouse brain. Here, we propose a five-stage GOM classification system to facilitate uniform analysis and description of GOM deposits and show that this staging can also be used to systematically classify GOM in vessels of CADASIL patient material.
The GOM deposits observed in aged mice (20 months) included stages I–IV, i.e. from small, circumscript deposits within the basement membrane (stage I), to large, amorphous GOM that induced bulging of the basement membrane (stage IV). As mice aged 6 months only showed stages I–III GOM, individual GOM deposits seem to increase in size and become increasingly amorphous over time, and new GOM seem to be continuously formed. This is further illustrated by the observation that GOM deposits of stage I and II are also present in post-mortem CADASIL patient brain microvessels, next to the more extensive GOM pathology, including patches of confluent GOM (stage V) [
7,
25,
26].
Although many studies have reported the presence and morphology of GOM in CADASIL patients [
9,
11‐
13,
25,
27‐
30], little is known about how GOM deposits progress over time. Brulin et al. analysed GOM in skin biopsies of CADASIL patients of different ages, and found that the number GOM deposits increase up to 50 years of age, but also found that the number of GOM seems to decrease in elderly patients [
11]. The latter may either be attributed to other end-stage vessel wall changes hampering the visualization of GOM, or because GOM seem to become confluent and disintegrate over time [
26]. Whether different GOM stages in skin biopsies of CADASIL patients are associated with disease severity and disease progression, therefore, remains to be determined. If so, the proposed GOM classification system may aid future efforts to monitor and predict disease progression at the individual patient level.
In our humanized CADASIL mouse model and in a rat Notch3 CADASIL mouse model, GOM deposits are observed several months after the first signs of NOTCH3
ECD-positive granular immunostaining [
22,
30], suggesting that NOTCH3
ECD granules may act as seeds for GOM development. Since NOTCH3
ECD aggregates attract extracellular matrix proteins that are components of GOM deposits, such as TIMP3 and clusterin [
4,
5], there seems a direct temporal relation between NOTCH3
ECD aggregates and the formation of GOM deposits. Delineating molecular differences between early- and end-stage GOM deposits may help to understand the sequence of events in CADASIL vascular pathology, and perhaps the identification of early therapeutic targets.
Whereas the CADASIL mice in our study seem to faithfully replicate early signs of disease pathology (NOTCH3
ECD accumulation and GOM deposition), we did not observe other CADASIL-associated disease features which have been observed in other CADASIL mouse models, such as vessel wall thickening, changes in SMA staining, mural cell degeneration and blood-brain barrier leakage [
7,
9,
14,
15,
17,
31]. In line with this, we previously showed that our mutant mice do not show brain parenchyma pathology at age 20 months, which we confirmed in this study using high-resolution T2W neuroimaging [
22]. Furthermore, we extended the characterization of the mice with functional tests, which did not reveal any cognitive or motor dysfunction. Our cerebral blood flow studies showed a small, not significant, reduction in CVR in tgN3
MUT350 mice, while studies in other, genetically different, CADASIL mouse models did show a reduced CVR [
8,
30,
32]. It may be relevant that we used a different method to anaesthetize mice and also to measure CVR, namely an ASL-based MRI approach. Although less sensitive in detecting small CVR changes, the advantage of ASL-MRI is that it allows for absolute perfusion quantification and detection of differences in baseline CBF. In that way, we found that the slight reduction of CVR in tgN3
MUT350 mice could at least partially be explained by a higher CBF at baseline. Future research is needed to determine whether the differences in CVR findings between this study and others [
8,
30,
32] can be explained by differences in genetics of the mouse models, by differences in baseline perfusion states or by differences in the study set-up, including the anaesthesia protocol. Although we did not find evidence for overt functional cerebrovascular deficits in our mouse model, this humanized mouse model captures early markers of CADASIL pathology, making it suitable for therapeutic studies targeting human mutant NOTCH3
ECD accumulation early in the disease course.
In summary, we show progression of GOM deposits in a humanized CADASIL mouse model. We propose a five-stage GOM classification system for uniform assessment of GOM depositions in translational research. In future pre-clinical studies of therapeutic approaches aimed at reducing or preventing NOTCH3ECD aggregation, GOM classification may serve as a valuable tool to monitor therapeutic efficiency on an ultrastructural level.
Compliance with Ethical Standards
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.