CRISPR/CasRx suppresses KRAS-induced brain arteriovenous malformation developed in postnatal brain endothelial cells in mice

Abstract

Brain arteriovenous malformations (bAVMs) are anomalies forming vascular tangles connecting the arteries and veins, which cause hemorrhagic stroke in young adults. Current surgical approaches are highly invasive, and alternative therapeutic methods are warranted. Recent genetic studies identified KRAS mutations in endothelial cells of bAVMs; however, the underlying process leading to malformation in the postnatal stage remains unknown. Here we established a mouse model of bAVM developing during the early postnatal stage. Among 4 methods tested, mutant KRAS specifically introduced in brain endothelial cells by brain endothelial cell-directed adeno-associated virus (AAV) and endothelial cell-specific Cdh5-CreERT2 mice successfully induced bAVMs in the postnatal period. Mutant KRAS led to the development of multiple vascular tangles and hemorrhage in the brain with increased MAPK/ERK signaling and growth in endothelial cells. Three-dimensional analyses in cleared tissue revealed dilated vascular networks connecting arteries and veins, similar to human bAVMs. Single-cell RNA-Seq revealed dysregulated gene expressions in endothelial cells and multiple cell types involved in the pathological process. Finally, we employed CRISPR/CasRx to knock down mutant KRAS expression, which efficiently suppressed bAVM development. The present model reveals pathological processes that lead to postnatal bAVMs and demonstrates the efficacy of therapeutic strategies with CRISPR/CasRx.

Keywords: Molecular pathology; Mouse models; Neurological disorders; Neuroscience.

Figure 1. Development of mouse bAVM models in the postnatal stage.

(A–C) A model of Cdh5-CreERT2;lsl-Kras^G12D mice with tamoxifen injection at P5. (B) Survival curve (n = 10). (C) A representative image of the cerebral cortex with CD31 staining (red). (D–H) A model of AAV-PHP.V1-Ple261-iCre–injected lsl-Kras^G12D;lcl-EGFP mice. (E) Survival curve (6 × 10¹⁰ GC, n = 9). (F) Representative serial brain section images of EGFP⁺ vascular tangle formation (green, P84). (G and H) Ectopic EGFP expressions in NeuN⁺ neurons (arrow) and GFAP⁺ astrocytes (arrowhead). (I–N) A model of AAV-PHP.V1-Ple261-iCre– and AAV-PHP.V1-CAG-DIO-HA-KRAS^G12D–injected lsl-tdTomato mice. (J) The brain showing multiple hemorrhagic lesions (6 × 10¹⁰ GC, P15). (K) Survival curve (n = 10). (L) Brain section images of tdTomato⁺ vascular tangles (red, P15). (M and N) Ectopic tdTomato expressions in NeuN⁺ neurons (arrow) and GFAP⁺ astrocytes (arrowhead). (O–R) A model of AAV-PHP.V1-CAG-DIO-HA-KRAS^G12D– and tamoxifen-injected Cdh5-CreERT2;lsl-tdTomato mice. (P) Brain images showing multiple hemorrhagic lesions (6 × 10¹⁰ GC, P18). (Q) Survival curve (n = 5). (R) Serial brain section images of tdTomato⁺ vascular tangles (red, P18). Scale bars: 250 μm (C), 500 μm (F and R), 50 μm (G, H, M, and N), 2 mm (J and P), 100 μm (L).

Figure 2. Vascular malformations in KRAS^G12D-induced Cdh5-CreERT2;lsl-tdTomato mice.

(A) Survival curves of Cdh5-CreERT2;lsl-tdTomato mice injected with different amounts of AAV-CAG-DIO-KRAS^G12D (1 × 10⁹, 5 × 10⁸, 1 × 10⁸, 5 × 10⁷, 3 × 10⁷, and 1 × 10⁷ GC). (B–D) Representative images of the brain (B), spinal cord (C), and organs (D) (1 × 10⁹ GC, P21). (E–H) Brain section images at days 5, 10, and 15 after AAV-PHP.V1-CAG-DIO-HA-KRAS^G12D (E–G) or control AAV-PHP.V1-CAG-DIO-MCS injection (H) in Cdh5-CreERT2;lsl-tdTomato mice (1 × 10⁹ GC; 3 to 4 images were combined in each panel). Arrowheads indicate vascular tangles. Dotted areas are magnified in the neighboring panels. (I) The size of lesions (1 × 10⁹ GC, total 380 vascular tangles, n = 3 brains, mean ± SD). N.D., not detected, in control AAV (MCS). (J) Inner vessel diameter in control AAV and of the lesions in KRAS-induced group. ****P < 0.0001. n = 357 and 862 vessels in 2 animals; data are shown as mean ± SD. Mann-Whitney U test was used. (K) The ratio of lesion numbers in brain regions (n = 3 brains). (L–N) HA and RAS^G12D expressions in the vascular tangles (P16 and P21). Scale bars: 2 mm (B), 1 mm (C), 5 mm (D), 500 μm (E–H), 100 μm (L–N, insets in F, G).

Figure 3. Three-dimensional analyses reveal a nidus structure in cleared brains of KRAS^G12D-induced Cdh5-CreERT2;lsl-tdTomato mice.

(A–C) Three representative cleared brains of latex-infused KRAS^G12D-induced Cdh5-CreERT2;lsl-tdTomato mice (3 × 10⁷ GC, P42) with tangled vasculature. Bottom, magnified views of the dotted areas. Arrowheads, feeding arteries (white) and draining veins (black). (D–G) A representative brain image of KRAS^G12D-induced Cdh5-CreERT2;lsl-tdTomato mice (D), fluorescence image in stereomicroscopy (E; bottom, magnified view of the lesion [arrowheads]), cleared brain by CUBIC (F), and light-sheet microscopic imaging (G). (H and I) Sagittal sectioning view by IMARIS (H) and 3D magnified view (I) of the lesion in the cerebral cortex (arrow). Green, vascular tangle; red, feeding arteries (Rea, retrosplenial artery; Pif, posterior internal frontal artery); blue, draining veins (Cauds, caudal branch of the superior sagittal sinus; Vehiv, ventral hippocampal vein). A, anterior; P, posterior; R, right; L, left. (J and K) Horizontal (J) and coronal (K) sectioning views of I. White arrowheads, vascular pillars in dilated vessels; arrow, dilated vein; asterisk, necrotic area; black arrowhead, hemorrhagic area. (L–P) Another sagittal view of G showing a vascular tangle in the septum (L, arrow) and 3D magnified sagittal (M and N) and horizontal images (O and P). Nidus structures (arrowheads, green/yellow/purple) connected with feeding arteries (red; Rsa, rostral septal artery) and draining veins (blue; Dsv, dorsal septal vein). (Q–S) Magnified views of arteriovenous junctions (arrowheads in N and P), showing dilated vessels connected with arteries and veins (Q) and aneurysms (R and S). Yellow circles, arteriovenous junctions. (T–X) Sectioned views of dilated vessels in Q and aneurysm in R (T and V) with 3D images of dilated vessels (yellow, green), arteries (red), and veins (blue) (U and X). Arrowheads, vein; black arrowheads, artery. Scale bars: 2 mm (D), 1 mm (H, L), 200 μm (J, K, and T–X).

Figure 4. Histopathological changes in the lesions of KRAS^G12D-induced Cdh5-CreERT2;lsl-tdTomato mice.

(A and B) Representative images of pERK expression in tdTomato⁺ endothelial cells in the lesion of AAV-DIO-KRAS^G12D and tamoxifen-injected Cdh5-CreERT2;lsl-tdTomato mice (1 × 10⁹ GC, P16, P21). (C) Quantified pERK expression in tdTomato⁺ endothelial cells in the lesion and adjacent control area. n = 9 lesions in 3 animals, unpaired t test. (D–F) Ki67 expression in the lesion (arrowheads, P16, P21). (G and H) Quantification of Ki67⁺ cells in tdTomato⁺ endothelial cells (%) and Ki67⁺ tdTomato^– cells in the lesion and adjacent control area. n = 4 animals, Mann-Whitney U test (G) and unpaired t test (H). (I and J) Cell and nuclear area of tdTomato⁺ endothelial cells in the lesion and adjacent control area. n = 9 lesions in 3 animals, unpaired t test (I) and Mann-Whitney U test (J). (K) H&E staining showing hemorrhage and hemosiderin-laden macrophages/microglia (arrowheads) in the lesion (P16, P21). (L) Iron staining with Prussian blue in the lesion (blue, arrowheads). (M and N) Activated Iba1⁺ microglia/macrophages and GFAP⁺ astrocytes surrounding the lesion (P16, 21). (O and P) Quantification of Iba1⁺ (O) and GFAP⁺ cell areas (P) in the lesion and adjacent control area. n = 9 lesions in 3 animals, unpaired t test. (Q) Ly6G⁺ neutrophils, B220⁺ B cells, CD3⁺ T cells, and MPO⁺ neutrophils in the lesion (arrowheads). (R and S) Quantification of Ly6G⁺ (R) and B220⁺ cells (S) in the lesion and adjacent control area. n = 3 animals, unpaired t test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Scale bars: 100 μm (A, B, D–F, and K–N), 20 μm (Q).

Figure 5. Single-cell RNA-Seq analyses of brain endothelial cells in KRAS^G12D-induced mice.

(A) t-SNE plot of the total extracted brain cells of AAV-CAG-DIO-MCS–injected (control) and AAV-CAG-DIO-KRAS^G12D–injected(KRAS-injected) Cdh5-CreERT2;lsl-tdTomato mice. (B) t-SNE plot of endothelial cells in control and KRAS-group. (C) Representative gene markers, the number of up- and downregulated DEGs, cell numbers (middle), and ratio (right) of each endothelial cell type in control and KRAS-group. (D) tdTomato and endothelial cell type marker expressions in t-SNE plot. (E) Representative GO terms in up- and downregulated DEGs (KRAS versus control). (F) Pathway analyses of genes expressed in endothelial cells (KRAS versus control). (G) Violin plots of representative DEGs expressions in endothelial cells of control and KRAS groups, which were commonly observed in human bAVMs (32). **P < 0.01, ***P < 0.001.

Figure 6. Single-cell RNA-Seq analyses of perivascular, glial, and immune cells in KRAS^G12D-induced mice.

(A) t-SNE plot of perivascular cells of AAV-CAG-DIO-MCS– (control) and AAV-CAG-DIO-KRAS^G12D–treated (KRAS-treated) Cdh5-CreERT2;lsl-tdTomato mice. (B) Representative gene markers, the number of up- and downregulated DEGs, cell numbers (middle), and ratio (right) of each perivascular cell type in control and KRAS-group. (C) Violin plots of marker gene expressions of proliferative fibroblast-like cells (FB). (D) t-SNE plot of microglia/macrophage in control and KRAS group. (E) Representative gene markers, the number of up- and downregulated DEGs, cell numbers (middle), and ratio (right) of each microglia/macrophage cluster in control and KRAS group. (F) Violin plots of representative DEGs in microglia/macrophage clusters in control and KRAS group. (G) t-SNE plot of astrocytes in control and KRAS group. (H) Representative gene markers, the number of up- and downregulated DEGs, cell numbers (middle), and ratio (right) of each cluster of astrocytes in control and KRAS group. (I) Violin plots of representative DEGs in astrocyte clusters in control and KRAS group. (J) t-SNE plot of leukocytes in control and KRAS group. (K) Cell numbers (left) and ratio (right) of leukocytes in control and KRAS group.

Figure 7. KRAS knockdown by AAV delivery of CRISPR/CasRx suppressed bAVM development.

(A) A design of CRISPR/CasRx to silence KRAS^G12D mRNA expression. DR, direct repeat. The right bottom image was modified from refs. , . (B) KRAS mRNA expressions in HEK293T cells transfected with control, KRAS^G12D, and KRAS^G12D-targeting CRISPR/CasRx plasmids in real-time PCR. n = 3, 1-way ANOVA followed by Tukey’s test. (C) Survival curves of KRAS^G12D-induced (1 × 10⁹ GC) and KRAS^G12D-CasRx–treated (1 × 10¹⁰ GC) mice. n = 13 and 10, log-rank test. (D) Brain images of KRAS^G12D-induced (1 × 10⁹ GC) and CasRx-treated mice at P21. (E and F) Representative images of vascular tangle formation (arrowheads) in KRAS^G12D-induced (1 × 10⁹ GC, E) and CasRx-treated (F) mice. (G and H) Lesion numbers (G) and size (H) in KRAS^G12D-induced (1 × 10⁹ GC) and CasRx-treated mice. n = 4, unpaired t test (G), 2-way repeated ANOVA followed by Bonferroni’s test (H). (I) Percentage of mice forming lesions in KRAS^G12D-induced (1 × 10⁹ GC, n = 13; 3 × 10⁷ GC, n = 10) and CasRx-treated mice (n = 10, 8, 7). (J–O) P42 brains of KRAS^G12D-induced (3 × 10⁷ GC, J and K) and CasRx-treated mice (L and M, treated at P5; N and O, treated at P19). Bottom panels (K, M, and O) show histological images of vascular tangles in the upper panels (arrowheads, J and N) or corresponding areas (M). (P and Q) Lesion numbers (P) and size (Q) in KRAS^G12D-induced (3 × 10⁷ GC) and CasRx-treated mice. n = 10, 8, 7, Kruskal-Wallis test followed by Dunn’s multiple-comparison test (P); n = 5, 4, Mann Whitney U test (Q). ***P < 0.001, **P < 0.01, *P < 0.05. Scale bars: 2 mm (D, J, L, and N), 500 μm (E and F), 200 μm (K, M, and O).