Somatic variants of MAP3K3 are sufficient to cause cerebral and spinal cord cavernous malformations

Abstract

Cerebral cavernous malformations (CCMs) and spinal cord cavernous malformations (SCCMs) are common vascular abnormalities of the CNS that can lead to seizure, haemorrhage and other neurological deficits. Approximately 85% of patients present with sporadic (versus congenital) CCMs. Somatic mutations in MAP3K3 and PIK3CA were recently reported in patients with sporadic CCM, yet it remains unknown whether MAP3K3 mutation is sufficient to induce CCMs. Here we analysed whole-exome sequencing data for patients with CCM and found that ∼40% of them have a single, specific MAP3K3 mutation [c.1323C>G (p.Ile441Met)] but not any other known mutations in CCM-related genes. We developed a mouse model of CCM with MAP3K3I441M uniquely expressed in the endothelium of the CNS. We detected pathological phenotypes similar to those found in patients with MAP3K3I441M. The combination of in vivo imaging and genetic labelling revealed that CCMs were initiated with endothelial expansion followed by disruption of the blood-brain barrier. Experiments with our MAP3K3I441M mouse model demonstrated that CCM can be alleviated by treatment with rapamycin, the mTOR inhibitor. CCM pathogenesis has usually been attributed to acquisition of two or three distinct genetic mutations involving the genes CCM1/2/3 and/or PIK3CA. However, our results demonstrate that a single genetic hit is sufficient to cause CCMs.

Keywords: MAP3K3; cavernous malformations; cerebral; somatic mutation; spinal cord; ‌endothelial cell.

Conflict of interest statement

The authors report no competing interests.

© The Author(s) 2023. Published by Oxford University Press on behalf of the Guarantors of Brain.

Figures

Figure 1

Formation of CCMs induced by MAP3K3I441M mutations alone in mice of different ages. (A) Schematic of injection of AAV-BR1-MAP3K3I441M-P2A-EGFP (i.e. AAV-MAP3K3I441M) through the superficial temporal vein of a mouse. Regular T2 MRI was used for CCM detection after aden-associated virus (AAV) injection. (B) Experimental design of a series of brain MRI in the mice infected with MAP3K3I441M. (C) Quantification of the number of lesions in mouse brains in the first, second, fourth and sixth weeks after AAV injection. Compared with the data from the mice before AAV injection, no obvious difference in CCM numbers was identified in the first week (n = 5). In the second (n = 5) and fourth weeks (n = 4) after AAV-MAP3K3I441M injection, the number of lesions increased significantly. The number of lesions reached a plateau in the sixth week after AAV injection (n = 3). The density of lesion in different brain regions. The number of lesions in the brainstem were significantly lower than that in other brain areas. Ob = olfactory bulb; Cx = cerebral cortex; Th = striatum, thalamus and hippocampus; Bs = brainstem; Cm = cerebellum. n.s. = not significant; *P < 0.05; ***P < 0.001; one-way ANOVA. (D) Longitudinal T2 MRI of different sections of a mouse brain at first, fourth and sixth weeks after injection of AAV-BR1-MAP3K3I441M. Numerous microlesions were detected throughout the mouse brain. Quantification of the number of lesions in the MRI of different brain regions including olfactory bulb, cerebral cortex, striatum, thalamus and hippocampus, brainstem and cerebellum. MRI were performed at the first (n = 5 mice), fourth (n = 4 mice), and sixth (n = 3 mice) weeks. n.s. = not significant; **P < 0.01; ***P < 0.001; one-way ANOVA. (E) Administration of AAV-MAP3K3I441M in mice at different ages was essential to drive CCM formation in their brains and spinal cords. Mice aged P5, P35, P150 and P330 were injected with AAV-MAP3K3I441M and T2 MRI was performed 15 days later. MRI results showed that all mice of different ages developed CCMs in 2 weeks (arrowheads). (F) Example images of MAP3K3I441M induced CCM lesions (arrowheads) in mice. AAVs were injected at P35 and brains were harvested at P50. [F(i–iv)] Insets from the brain. (G) Example images of MAP3K3I441M induced spinal cord CM lesions (arrowheads). AAVs were injected at P35 and spinal cords were harvested at P50. [G(i–iii)] Insets from the spinal cord.

Figure 2

Characteristics and morphology of MAP3K3I441M -induced CCM lesions. (A) Representative images of H&E staining of MAP3K3I441M -induced CCM lesions (arrowhead, inseti) and haemorrhage (arrowheads, insetii) in mice. (B) Characteristics of MAP3K3I441M-induced CCM lesions and the control blood vessels in mouse brains (top), and characteristics of CCM lesions from a patient with MAP3K3I441M mutations (bottom). H&E staining (lesions, arrowheads in first panel; haemorrhage, arrowheads in second panel) and iron deposits of haemosiderin (patches, arrowheads in third panel) in MAP3K3I441M-induced CCM lesions from mice or brain tissue from human patients. Dilated vascular sinusoids were lined by endothelial cells (arrowheads in first panel). Haemorrhages and haemosiderin stains surrounded the lesions (second and third panels). The images from control groups are shown in the fourth panel. (C and D) The morphology of CMs induced by MAP3K3I441M in Cdh5-CreER::Ai14tg mice. Example images of brain sections from Cdh5-CreER::Ai14tg mice injected with AAV-BR1-MAP3K3I441M-P2A-EGFP. Representative images of MAP3K3I441M-induced CM lesions with dilated lumen (top, D) or distorted lumen (bottom, D). Tdt = tdTomato signal (red) in endothelial cells from Cdh5-CreER::Ai14tg; nuclei, DAPI (blue); GFP = weak green signal from endothelial cells (green) infected by AAV-BR1-MAP3K3I441M-P2A-EGFP. (E and F) The morphology of CMs induced by MAP3K3I441M in Pdgfrb-Cre::Ai14tg mice. Example images of brain sections stained with anti-Collagen IV in Pdgfrb-Cre::Ai14tg mice 2 weeks after infection with AAV-BR1-MAP3K3I441M. Normal blood vessels in brain region (E). Representative images of MAP3K3I441M-induced CM lesions with dilated lumen (F). Blood vessels, collagen IV (green); nuclei, DAPI (blue), and tdT signal (red) from pericytes in Pdgfrb-Cre::Ai14tg mice.

Figure 3

In vivo imaging of the development of CCMs caused by MAP3K3I441M. (A) Schematic of time-lapse live imaging of the brain vasculature in a conscious mouse. (B) Imaging of the development of a CCM lesion (arrowheads) located in a mouse brain through a chronic cranial window. Before AAV injection (baseline, left); 20 days after AAV injection (Day 20, middle). Inset: The same region within the dashed box with FITC-dextran (green) circulating in blood vessels (right). (C–E) Three-dimensions of a z-stack image taken by a two-photon laser excitation microscope. Green = FITC-dextran; red = Dsred signal in the NG2BacDsRed transgenic(tg) mouse. (F–H) CCM lesions (arrowheads) with leakage (top) and without leakage (bottom) in NG2BacDsRed tg mouse (F). Percentage of CCM lesions with leakage (n = 180 lesions in total). L(+) = leakage; L(−) = with no leakage (H). (G–I) A CCM lesion (arrowheads) with coverage of pericytes (top) and without coverage of pericytes (bottom). (I) The percentage of lesions with pericyte [P(+)] or without pericyte coverage [P(−)](n = 180 lesions in total). (J) The coverage of pericytes in individual CCMs mediated by MAP3K3I441M mutations from the brain of Pdgfrb-Cre::Ai14 tg. Blood vessels were stained with antibodies against Collagen IV (green). Nuclei were stained with DAPI (blue). (K and L) Diameter and volume of the CCM lesions and normal vessels; *P < 0.05; **P < 0.01; ***P < 0.001; one-way ANOVA. (M–O) Characterization of blood flow in individual CCMs induced by MAP3K3I441M mutations. Positions of line scanning in normal vessels (V1, V6), parent vessels (V2, V4) and cavernous vessels of CCM lesions (V3, V5). (N) The velocity of blood cells in vessels shown in (M) measured through line scanning. Images were plotted with X-T mode (the y-axis represents time and the x-axis represents distance) in the locations of V1 to V6. (O) Summarized results of red blood cell velocities in the upstream (Up, n = 7) and downstream (Down, n = 9) of the parent vessels, in normal blood vessels (n = 15), and in cavernous vessels of CCM lesions (n = 7), one-way ANOVA and *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 4

Staining of endothelial cell proliferation in tissues from MAP3K3I441M mouse models and CCM human samples. (A) Cell proliferation assays (EdU) in Cdh5-CreER::Ai14tg mice 2 weeks after infected with AAV-BR1-MAP3K3I441M. EdU positive cells (green, third row) were detected in the subventricular zone (SVZ), a neurogenic region in the brain. (B) No EdU+ endothelial cells were detected in MAP3K3I441M-mutated CCMs (0.0%, n = 0/50 lesions in total) or in normal vessels (0.0%, n = 0/100 normal vessels in total). (C) H&E staining and Ki67 staining in CCMs of human with MAP3K3I441M mutations. No positive immunoreaction for Ki67 was observed. (i) Insets from H&E staining in C. (D) Ki67 staining in the ‘control’ tissue (without CCMs) of the temporal lobe from an epilepsy patient. No positive immunoreaction for Ki67 was observed. (E) Ki67 staining in the tissue of human glioma. (F) Example images of normal capillaries and MAP3K3I441M mutated CCMs in Cdh5-CreER::Ai14 mice after injection of AAV- MAP3K3I441M. DAPI (blue) and tdTomato (tdT, red) signal from endothelial cells in Cdh5-CreER::Ai14tg. (G) Comparison of the membrane area of single endothelial cell of MAP3K3I441M induced CCMs (n = 10) with that of normal capillaries (n = 10) in Cdh5-CreER::Ai14 mice after injection of AAV- MAP3K3I441M. ***P < 0.001, unpaired two-tailed Student’s t-test. (H) Schematic of the morphology of endothelial cells with MAP3K3I441M expression and without MAP3K3I441M expression.

Figure 5

Molecular mechanism of the induction of CCMs by MAP3K3I441M. (A) Hierarchical cluster of all samples of bulk RNA-sequencing from brain endothelial cell lines, bEnd.3. Three samples from endothelial cells with lentivirus-MAP3K3I441M infection (Mut 1–3) and five control samples with lentivirus-EGFP infection (Ctrl 1–5). (B) Downregulated (blue) and upregulated (red) genes [log2 fold-change (FC) > 1.5, adjusted P-value < 0.05] in the endothelial cells with expression of MAP3K3I441M. (C) GSEA showed the activated pathway of structural constituent of cytoskeleton in the endothelial cells with MAP3K3I441M expression. (D) Normalized counts of the top eight genes encoding actins in bulk RNA-sequencing data acquired from the control endothelial cells (n = 5) and those cells with MAP3K3I441M expression (n = 3). ***P < 0.001, unpaired two-tailed Student’s t-test. (E) The heat map showing the z-scores of normalized counts for differential genes, as measured by bulk RNA-Sequencing. Some featured genes (e.g. Traf6, Actn1, etc.) were highlighted. (F) Western blot results for demonstrating MAP3K3-mTOR interactions. Co-immunoprecipitation (IP) results showed MAP3K3 interacted with mTOR.

Figure 6

Blocking mTOR activity with rapamycin prevents progression of MAP3K3I441M-mediated formation of CCMs. (A) Experimental design for our strategy of the targeted therapy using mTOR inhibitor rapamycin in mice with MAP3K3I441M CCMs. (B) T2 MRI of mouse brains after treatment using rapamycin or vehicle at the second and fourth week, and visual images of the hindbrains of mice treated by rapamycin or vehicle for 4 weeks. Scale bars = 1 mm. (C) Quantification of the number of CCM lesions in mice with and without the rapamycin treatment. Vehicle group, n = 7 mice; rapamycin, n = 9 mice; *P < 0.05, unpaired two-sided Student’s t-test. (D) Quantification of the increase of CCM lesion numbers between the second and fourth week after the treatment with vehicle or rapamycin. Vehicle, n = 7 mice; rapamycin, n = 9 mice; *P < 0.05, unpaired two-sided Student’s t-test. (E) H&E staining of mouse brains after treatment using rapamycin or vehicle at the fourth week (lesions, arrowheads). (i and ii) Insets from H&E staining in E. (F–H) Quantification of CCM: the density of lesions (F), the diameter of lesions (G), and the area of a single lesion (H) with H&E staining at the fourth week in vehicle or rapamycin-treat CCM mice. ***P < 0.001, unpaired two-tailed Student’s t-test. (I) Autophagy marker, p62 staining of the brain sections from mice treated with rapamycin or vehicle for 4 weeks. (J) The number of p62 bodies was significantly decreased in the rapamycin-treated group. *P < 0.05, unpaired two-tailed Student’s t-test. (K) Experimental design for our strategy of the targeted therapy using mTOR inhibitor rapamycin in aged mice (18–24 months) with MAP3K3I441M CCMs. (L) Quantification of the number of CCM lesions in aged mice in vehicle or rapamycin-treated groups. Vehicle group, n = 8 mice; rapamycin, n = 4 mice; **P < 0.01, unpaired two-tailed Student’s t-test.