Cerebral Cavernous Malformations (CCMs) refer to abnormal blood vessels with thin walls that are tightly packed in the brain [1, 2]. The appearance of CCMs is similar to that of a small mulberry, as a result of abnormal cell divisions in the vascular endothelium [3]. Approximately 500 people out of every 100,000 are affected by this disease (0.5%) [4], with approximately 20% of cases being familial, resulting from the loss of function of one of the CCM1/KRIT1 [5], CCM2 [6], or CCM3/PDCD10 [7] genes. In the remaining 80% of patients, the disease occurs sporadic, and the genes responsible for it have not been well identified. Moreover, CCM disease is associated with brain aging and is prevalent in the elderly [8]. This type of vascular malformation is characterized by hemorrhages within brain tissue, suggesting that damage to the blood-brain barrier may play a significant role in the development of these conditions.

In light of the inherited nature of CCM and the discovery of disease-related genes, there has been a growing interest in investigating their functional roles. Each of the CCM1, CCM2, and CCM3 genes encodes a protein that is structurally distinct and does not share any sequence homology [9]. The gene products of each gene are multi-domain adaptor proteins, which participate in numerous signaling pathways and interact with numerous binding molecules [10]. The CCM1, CCM2, and CCM3 proteins form a heterotrimeric complex [11], the CCM signaling complex, which plays an important role in the development and pathogenesis of CCM. Recent studies suggest that CCM proteins play an important role in many cellular events, including cell polarity [12], cytoskeletal reorganization [13], cell proliferation [14], cellular adhesion [15], and migration, impacting angiogenesis [16], cell–cell junction integrity [17], vascular permeability [18], and apoptosis, whether they form part of the ternary complex or not.

Interestingly, the ankyrin repeat and sterile alpha motif domain-containing protein 1B (ANKS1B) was identified as a novel binding partner of CCM1 [19], a major causative gene for CCM disease. The CCM1 protein is characterized by three NPxY/F motifs and three ankyrin repeats, and it is abundantly expressed in vascular endothelial cells. Silencing of ANKS1B, however, did not have any significant effect on the proliferation, migration, or sprouting of angiogenesis in primary human endothelial cells. While silencing of ANKS1B expression disrupted endothelial barrier functions, resulting in increased permeability, the role of ANKS1B in CCM1-mediated vascular pathogenesis remains unclear [19].

ANKS1 family of proteins consists of two members, ANKS1A and ANKS1B, and they contain six ankyrin repeats at the N-terminus, two SAM domains, as well as a PTB domain at the C-terminus [20]. In this study, we show that ANKS1A plays a role in stabilizing and regulating the permeability of vascular endothelial cells in the aged brain. The findings of this study may provide a new paradigm for understanding the pathogenesis of CCM and other vascular diseases of the brain.

Mice

ANKS1A^+/-(LacZ) gene trap mice have been described previously [21, 22]. ES cells with a gene trap vector insertion (pGT0lxr) in the ANKS1A gene locus (cell line ID CF0537) were obtained from the Mutant Mouse Regional Resource Center. The gene trap insertion site was determined through sequence analysis. These ES cells were microinjected into C57BL/6 blastocysts and subsequently transferred to foster mother (ICR) females to create chimeric animals. Chimeric males were bred with 129/SvJ females, and the presence of the transgene in F1 agouti pups was assessed using PCR analysis of tail genomic DNA with specific primers designed to detect mutant and wild-type ANKS1A: 5’-TGAAGGCACATGACCCTGAG-3’ (F1), 5’-ATGTCATAGCTGTTTCCTGT-3’(F2), and 5’-ACAGCGTTTGCATCTTGCTG-3’(R1). The experiments on mice were conducted according to the institutional research guidelines of the Sookmyung Women’s University and the protocol approved by the Institutional Animal Care and Use Committee (IACUC) of the university (protocol approval: SMWU-IACUC-2104-002-1). All mice had ab libitum access to standard laboratory diet and water.

X-gal staining and immunostaining analysis

For X-gal staining, the brain sections were fixed with 1% paraformaldehyde (PFA), 0.2% glutaraldehyde, 1% deoxycholate (Sigma Aldrich, St. Louis, U.S.A), and 10% NP-40 in phosphate-buffered saline (PBS) for 10 min; the sections were then washed 3 times for 10 min and then incubated with the staining buffer (K4Fe(CN)6, K3Fe(CN)6, MgCl2, X-gal) at 37°C. After 15 h, the reaction was stopped by washing each slide three times with PBS.

For immunostaining analysis of the mouse brain cryosections, the sections were washed with PBS three times for 10 min at RT. Then, they were incubated in PBS containing 0.3% Triton X-100 for 30 min and treated with primary antibody diluted in PBS containing 0.3% Triton X-100 and 3% BSA for overnight at 4°C. After the overnight incubation, the brain sections were washed, and treated with secondary antibodies for 2 hours at RT. The sections were then three times washed with PBS containing 0.3% Triton X-100 and mounted with Vectorshiled mounting medium. Primary antibodies were as follows: Collagen IV (goat, 1:100, 1340-01, Southern Biotech), Hemoglobin (rabbit, 1:100, ABIN1078132, Antibodies online), CD68 (rabbit, 1:100, ab125212, Abcam), Aquaporin 4 (rabbit, 1:100, AQP-004, Alomone labs), CD45 (rabbit, 1:100, ab10558, Abcam), Fibronectin (rabbit, 1:100, ab23750, Abcam), Sca-1 (rat, 1:100, ab51317, Abcam). Secondary antibodies were as follows: Donkey anti-rabbit IgG Alexa488 (1:500, A21296, Thermo Fisher Scientific, Waltham, MA USA), Donkey anti-rat IgG Alexa488 (1:500, A21208, Thermo Fisher Scientific), Donkey anti-goat IgG Alexa488 (1:500, A11055, Thermo Fisher Scientific), Donkey anti-goat IgG Alexa568 (1:500, A11057, Thermo Fisher Scientific).

Image analysis

Image data were collected using an LSM700 (Carl Zeiss Microscopy) with a Plan-Apochromat ×20/0.8 M27 objective lens of an Axio Observer camera (Zeiss). The images were taken at 0.5~1 μm z-stack intervals over a 5~10 μm thickness. The fluorophore excitations were with 488, 555, and 639 nm laser wavelengths. All images were processed by the ZEN Black software.

Cerebral blood vessel network was analyzed using the AngioTool 0.6a software (NIH) [23]. For producing the vessel skeletonized images for Col IV staining, various parameters (e.g., vessel diameter) were optimized such that only the true vessels were labelled. The skeletonized images were further used to calculate length, total area, branching points and lacunarity of microvessels in each microscopic field.

Evans blue extravasation assay

2% Evans blue (Sigma Aldrich, St. Louis, U.S.A) dissolved in saline was intraperitoneally injected into each mouse (200 μl) and they were sacrificed after 24 hours [24]. For qualitative assessment of Evans blue extravasation, the brains were fixed in 4% paraformaladehyde and cut 40 μm-thick cryosection. Brain sections were washed with PBS three times for 10 min at RT. Then, sections were mounted with Vectorshield mounting medium and observed them under ZEISS Axio Zoom.V16 fluorescence microscopy (Zeiss, Jena, Germany).

Primary mouse brain endothelial cell culture

Primary brain endothelial cells were obtained from 2 months old mice using published protocol [25, 26]. The removal of the meninges by rolling the brains on filter papers. Subsequently, the brains were dissociated using a tissue grinder (with 10 brains processed in one tissue grinder). The resulting homogenates were then subjected to centrifugation (2580 g for 7 minutes at 4°C), and the resulting cell pellets were re-suspended in a 20% BSA solution, followed by thorough vortexing. The precipitates were digested for 60 minutes at 37°C in DMEM containing Collagenase/Dispase (100 mg/ml, Roche), DNase I (4 ug/ml, Roche), and TLCK (Nα-Tosyl-L-lysine chloromethyl ketone hydrochloride) (0.147 ug/ml, Sigma). After the digestion, the precipitates were again centrifuged (2580 g for 7 minutes at 4°C) and washed with PBS. The isolated mouse brain endothelial cells were then seeded onto 24-well plates coated with type IV collagen. These cells were cultured in high-glucose DMEM supplemented with 20% PDS (Plasma-derived bovine serum, FirstLink), penicillin/streptomycin, heparin (750 U), endothelial cell growth supplement (Sciencell), and 8 μg/ml puromycin (Sigma). After seeding, the cells underwent a two-day puromycin treatment, and subculturing was performed for 7 days.

Quantification

Immunofluorescence analysis was performed on images obtained from confocal microscopy. By using the Zen Blue software (Zeiss, Jena, Germany), the staining for neuroinflammation and Endothelial-mesenchymal transition (Endo-MT) markers was analyzed and quantified. Briefly, we used the mean signal intensity per field. Quantification was conducted on three animals per each group.

Statistics

Statistical analysis was performed on data from three or four independent experimental replicates using GraphPad Prism 9.0 (San Diego, CA, USA). Error bars in the graphs represent the mean±SD of the data. Statistical significance test was via the unpaired student’s t test for two samples in all the figures with the following p-value designations in the plots: *, p<0.05; **, p<0.01; ***, p<0.001; ****, p<0.0001.

Data availability

All other data are available from the corresponding author upon reasonable request.

The specific expression of ANKS1A in the brain endothelial cells

To determine the brain region where ANKS1A is specifically expressed, we used ANKS1A^+/-(LacZ) mice in which the LacZ reporter was inserted into the ANKS1A gene locus: the targeted allele expresses the LacZ reporter instead of ANKS1A. Strikingly, whole mount X-gal staining of the dissected brains of adult mice at P45 revealed that ANKS1A was predominantly expressed in the brain vessels (Fig. 1A). Using Collagen IV, a marker specific for brain vessels, brain sections were stained to determine whether LacZ expression is restricted to brain vessels. Consequently, we observed that LacZ expression co-localized with Collagen IV staining. These results indicate that ANKS1A is expressed specifically in cerebrovascular cells (Fig. 1B).

In order to investigate whether the ANKS1A deficiency impairs the development of cerebrovascular patterning, brain sections were stained with Collagen IV and then analyzed for the indices that represent cerebrovascular patterning (Fig. 1C). Accordingly, we found that the density of vessels, the average vessel length, the degree of vessel branching, and the gap between neighboring vessels were not significantly different between wild type and KO groups (Fig. 1D). The results suggest that ANKS1A deficiency does not affect the overall development of cerebrovascular networks in the adult brain.

CCM-like vessel lesions in the aged ANKS1A KO brain

The ANKS1A KO mice display normal behavior, sexual reproduction, and healthy lifespan, similar to the WT littermates. However, we hypothesized that aged brains of ANKS1A KO mice would exhibit some defects in the cerebrovascular system. To accomplish this, the mice were allowed to grow for over 16 months and then their brains were carefully examined. A striking finding was that ANKS1A KO mice exhibited abnormal clusters of closely packed vessels whereas their WT littermates did not. This small mulberry-shaped blood vessel is known as a cerebral cavernous malformation (CCM) [3]. As an example of a symptom associated with CCM is bleeding surrounding the lesion [27]. In accordance with this hypothesis, we found a significant amount of hemoglobin in the CCM-like lesion of the neocortex (Fig. 2A) but also in the vessels near the brain ventricular region (Fig. 2B), suggesting brain hemorrhage in the aged ANKS1A KO brain. To further confirm the leakage from the CCM-like lesion, the brain sections were immunostained with mouse IgG-specific antibodies. Consequently, we observed that the IgG-positive signal was highly detectable only in the aged ANKS1A KO mice (Fig. 2C). Finally, we injected Evans Blue, the most widely used marker of brain barrier integrity [28], into the peritoneal cavity of aged mice and examined whether the dyes were abundantly detected around the CCM-like lesions. The intensity of Evans Blue was measured in the areas containing CCM-like lesions as well as focal cavitation found adjacent to the lesions (see Fig. 2D, right and top panel). As we expected, the intensity of Evans Blue in the aged ANKS1A KO brain was substantially higher than in the WT littermate brains (Fig. 2D). Together, our studies strongly indicate that the aged ANKS1A KO brain exhibits abnormal blood vessel integrity with bleeding.

Immune cells are focally increased around the CCM-like lesion

An analysis of high resolution confocal images revealed that the IgG-positive cells were detectable around the CCM-like lesion (Fig. 2C), suggesting peripheral blood immune cells have been infiltrated into the brain parenchymal tissue. To investigate whether peripheral immune cells infiltrate the aged ANKS1A KO brain, brain sections were stained with CD68, a marker for macrophages and other monocyte lineage cells [29]. As expected, CD68-positive cells were highly detectable in aged ANKS1A KO brains, but not WT brains (Fig. 3A). A similar experiment was conducted using CD45 antibody, which stains leukocytes [30], showing that CD45-positive cells were highly detectable around the CCM-like lesions (Fig. 3B). Together, our results suggest that lesions associated with CCM may cause local neuroinflammation in the aged ANKS1A KO brain.

CCM-like lesions are characterized by irregularly gathered vessels that swell to form a large cavity within the neighboring parenchyma (see Fig. 2D, right and top panel). We found that CCM-like lesions with focal cavities were detected in various sizes in the aged ANKS1A KO brain and that those more than 2,000 μm² were the most common (Fig. 3D). In addition, immunostaining using Aquaporin 4 (AQ4) antibody, a marker for astrocyte endfeet [31], revealed that astrocyte endfeet in the CCM-like lesion and cavity were much less abundant in the aged ANKS1A KO brain (Fig. 3C). However, in the aged WT brain, Aquaporin 4 staining was completely matched with Collagen IV staining (Fig. 3C). Importantly, the aged ANKS1A KO brain exhibited much lower levels of ZO-1 (Fig. 3E), a tight junctional marker for brain endothelial cells [32]. The expression of ZO-1 was only affected in CCM-like lesions and focal cavitations adjacent to the lesions. However, ZO-1 expression was not affected in blood vessels away from CCM-like lesions.

CCM-like lesions in the aged ANKS1A brain exhibit increased mesenchymal markers

We further hypothesized that formation of a CCM-like lesion in ANKS1A-deficient vessels may be due to changes in the fate of brain endothelial cells with aging. To determine whether the endothelial cells in the brain are altered in the aged ANKS1A KO mice, we examined the brain tissues with various markers for mesenchymal tissue. First, we used fibronectin as a mesenchymal marker, which is known to increase in epithelial-mesenchymal transition (EMT) induced by TGF-β [33]. In the WT brain, fibronectin was barely observed. Contrary to this, fibronectin was highly detectable in the CCM-like lesions of the aged ANKS1A KO brain (Fig. 4A, 4B). We also used Sca-1, a commonly used marker for mesenchymal stem cell [14, 34], to determine whether the CCM-like lesion lost its normal endothelial properties. As a result, we found that, as opposed to vessels from WT mice, the aged ANKS1A KO brain displayed CCM-likes lesions with high levels of Sca-1 (Fig. 4C, 4D). In addition, we cultured primary brain endothelial cells from WT and ANKS1A KO brains (2 month of age) and found that fibronectin levels were higher in ANKS1A-deficient cells (Fig. 4E, 4F). More importantly, we observed that ANKS1A-deficient cells showed Ki67-positive staining, suggesting that they are more proliferative (Fig. 4G, 4H). Taken together, these results strongly indicate that during brain aging, ANKS1A-deficient brain endothelial cells gradually lose their endothelial property and become mesenchymal cells.

In this study, we found that ANKS1A is specifically expressed in the brain endothelial cells of adult mice. Even though ANKS1A deficiency did not impair the overall differentiation, growth, and patterning of the cerebrovascular system, its absence had a significant impact on the cerebrovascular system of the aged brain. First, vessel lesions that resemble cerebral cavernous malformations (CCMs) were found. Second, the CCM-like lesions also showed a local leakage of peripheral blood into the brain tissue. In addition, the CCM-like lesions revealed that immune cells had infiltrated the parenchyma. Third, the CCM-like lesions also contained much less astrocyte endfeet with reduced tight junctions, implying that the integrity of the BBB has been partially compromised. Lastly, the CCM-like lesions displayed an increased mesenchymal marker expression. In conclusion, our findings suggest that ANKS1A may play a role in maintaining a healthy blood vessel in the brain during the aging process.

In aged ANKS1A KO brains, CCM-like lesions were detected at a low frequency (3~5 per brain) whereas they were never seen in aged WT brains. Furthermore, we did not detect CCM-like lesions in the ANKS1A KO brains from 2 months of age to 12 months of age. Since ANKS1A KO mice over 16 months of age were highly susceptible to developing CCM-like lesions, brain aging is likely to play a critical role in the development of CCM-like lesions in ANKS1A KO brains. A key finding was that the brain endothelial cells in the CCM-like lesion exhibit mesenchymal characteristics. The Endo-MT transition of brain endothelial cells plays an important role in the development of dysfunctional BBB with hemorrhage [35]. Furthermore, vascular cell adhesion molecule 1 (VCAM1) is known to be upregulated in aged brain endothelial cells and facilitate vascular-immune cell interactions [36]. Together, these reports support our findings of local bleeding and immune cell infiltration in CCM-like lesions. Nevertheless, in light of the abundant expression of ANKS1A in brain endothelial cells, this focal phenotype in the brain is not well understood. One plausible hypothesis is that environmental factors combined with ANKS1A deficiency in the localized vessels trigger the development of CCM-like vessel lesions. It is therefore essential to identify the extrinsic factors that can severely disrupt the integrity of vessels together with ANKS1A deficiency. For example, animal models with a breakdown of the BBB are more susceptible to lipopolysaccharide (LPS) [37] and experimental autoimmune encephalomyelitis (EAE) [38]. It remains to be determined whether these experimental tools can disintegrate brain endothelial cells when combined with an ANKS1A deficiency.

This work was supported by grants NRF-2021R1A4A1027355, NRF-2021R1C1C2009319 from the National Research Foundation of Korea (NRF) and HU23C0017 from KHIDI and KDRC.