Endothelial Cell–Pericyte Interactions in the Pathogenesis of Cerebral Cavernous Malformations (CCMs)

Abstract

Cerebral cavernous malformations (CCMs), consisting of multiple, dilated capillary channels formed by a single layer of endothelium and lacking parenchymal cells, are exclusively to the brain. Patients with inherited autosomal-dominant CCMs carry loss-of-function mutations in one of three genes: CCM1, CCM2, and CCM3. It is not known why CCM lesions are confined to brain vasculature despite the ubiquitous expression of CCM proteins in all tissues, and whether cell types other than endothelial cells (ECs) contribute to CCM lesion formation. The prevailing view is that the primary defects in CCMs in humans are EC-intrinsic, such that EC-specific deletion of any one of the three genes in mice results in similar CCM lesions. An unexpected finding is that Ccm3 deletion in pericytes (PCs) also induces CCM lesions. CCM3 deletion in ECs or PCs destabilizes PC–EC associations, highlighting the importance of these interactions in CCM formation.

Cerebral vascular malformations affect 0.1%–4% of the general population and fall into four categories: arteriovenous malformations, venous malformations, capillary telangiectasias, and cerebral cavernous malformations (CCMs). CCMs consist of clusters of enlarged endothelial channels (“caverns”) that are arranged back-to-back to form densely packed sinusoids with little or no intervening brain parenchyma (Rigamonti et al. 1988; Otten et al. 1989; Revencu and Vikkula 2006; Cavalcanti et al. 2012; Draheim et al. 2014). These lesions lack vascular smooth muscle cells (SMCs)/pericytes (PCs), elastic tissue, subendothelial support, and an intact basal lamina, leaving the vessel thin and leaky. Ultrastructural analysis has revealed a decreased number of PCs, endothelial detachment from the basal lamina, and ruptures in the luminal endothelium likely due to reduced/damaged intercellular junctions (Tanriover et al. 2013). CCMs are primarily found within the neurovasculature of the central nervous system (CNS) (i.e., brain, spinal cord, retina), where they result in increased risk for stroke, seizures, and focal neurological deficits (Rigamonti et al. 1988; Otten et al. 1989; Revencu and Vikkula 2006). Currently, the only treatment for CCMs is surgical resection.

CCMs are associated with loss-of-function (LOF) mutations in any one of the three CCM genes: CCM1 (also known as Krev/Rap1 Interacting Trapped 1 [KRIT1]) (Sahoo et al. 1999), CCM2 (also known as malcavernin or osmosensing scaffold for mitogen-activated protein kinase kinase-3-osm) (Liquori et al. 2003), and CCM3 (also known as programmed cell death 10 [PDCD10]) (Bergametti et al. 2005). Three CCM proteins can be found in the same complex within the cell, and recent studies suggest that all CCM proteins regulate the MEKK3/KLF4 signaling, which contributes to the onset and progression of CCMs (Boulday et al. 2009; Chan et al. 2011; Cunningham et al. 2011; McDonald et al. 2011; Maddaluno et al. 2013; Zhou et al. 2016b). However, CCM3 might also act independently of CCM1 and CCM2, as its mutation in humans often results in a more serious form of the disease (Denier et al. 2006; Zhu et al. 2011; Shenkar et al. 2015), and CCM3 knockout (KO) mice show severe phenotypes (Zhou et al. 2016a; Tang et al. 2019).

The unique feature of the brain vasculature is the blood–brain barrier (BBB) formed by the brain neurovascular unit consisting of endothelial cells (ECs), PCs, astrocytes, microglia, and neurons. Recent studies suggest that loss of CCM3 causes BBB disruption as the inciting event for the development of CCM2. The primary cell type involved in the formation of the CCM lesion is not known. The prevailing view holds that the primary defects in CCMs in humans are EC-intrinsic, and EC-specific deletion of any one of the three genes in mice results in similar CCM lesions. However, these models with CCM gene ablation in ECs have several shortcomings that have been addressed by recently established models. An unexpected finding is that Ccm3 deletion in vascular mural cells, SMCs, and particularly PCs, also induces CCM lesions, and mice carrying this mutation can survive to adulthood with lesions that resemble human disease spreading through whole brains. These observations support the notion that deletions of CCM3 in mural cell, in addition to ECs, are a major cause of CCM lesions. While CCM3 deletion in ECs induces exocytosis of angiopoietin-2 (Ang2) from ECs, thereby destabilizing EC junctions and PC–EC associations, CCM3 deletion in PCs attenuates PC migration toward ECs, thereby disrupting EC–PC interactions, highlighting the importance of EC–PC interactions in CCM formation. Here, we will summarize recent progress in our understanding of how mural cells contribute to CCM lesions and the underlying role of CCM3 in regulation of EC–PC interaction and BBB integrity. These studies support that loss of CCM3 in either ECs or PCs alters signaling critical for PC–EC interactions, contributing to vascular disassembly, extracellular matrix (ECM) remodeling, and CCM formation. A recent review comprehensively summarizes recent progress in human genetics of CCM disease and mouse models with EC-specific deletions (Snellings et al. 2021).

CCM PATHOGENESIS AND NOVEL MOUSE MODELS

CCMs have been thought to result from a two-hit mechanism, and biallelic somatic mutations in one of the CCM genes have been identified in EC lining cavernous vessels of patients carrying germline mutations (Gault et al. 2005; Akers et al. 2009; Pagenstecher et al. 2009), leading to the notion that the primary defects in CCMs are EC-specific. Conditional EC-specific gene disruptions in mouse models support this idea. Recent studies using multicolor fluorescent reporter mice indicate the CCM lesion is initially derived from a single clonal expansion of CCM3-mutant ECs within a CCM lesion (Detter et al. 2018). It is unclear whether lesion-forming ECs have special features located at a unique niche. The current CCM models have several shortcomings: (1) EC-specific Ccm ablation in adult mice does not induce CCM lesions. CCM lesion formation is thought to require ongoing angiogenesis and/or remodeling. (2) EC-specific Ccm ablation at neonatal stages induces CCM lesions. However, these lesions develop predominantly in the cerebellum and not the cerebrum, while human CCM lesions affect the entire brain. (3) CCM-carrier mice do not survive to adulthood, limiting studies to perinatal mice. Several new models have addressed these concerns.

The BEC-Specific Deletion Models

Global EC-specific Ccm deletions (by Cdh5CreER^T2) are common CCM models. However, the CCM lesion-carrier mice from global EC-specific Ccm ablation (Pdcd10^ECKO) do not survive to adulthood, limiting studies to perinatal mice (Boulday et al. 2009; Chan et al. 2011; Cunningham et al. 2011; McDonald et al. 2011; Maddaluno et al. 2013; Zhou et al. 2016b). We have recently discovered that Pdcd10^ECKO pups with global EC deletion of Pdcd10 (Ccm3) exhibit severe splenic hemorrhage and rupture, leading to perinatal death (Zhou et al. 2021). These mice also exhibited vascular defects in other tissues such as testis, mesentery, and lymph nodes (W Min and JH Zhou, unpubl.). We have established a brain EC (BEC)-specific inducible Pdcd10 deletion model (Pdcd10^BECKO) using an Mfsd2aCreER^T2 line that promotes CCM lesions in the brain and retina without causing vascular defects in other tissues (Zhou et al. 2021). Importantly, Pdcd10^BECKO mice survive 6–12 mo, allowing visualization of vascular lesion formation by live imaging, examination of a time course of CCM pathogenesis, and testing of therapeutics in adulthood. Pdcd10^BECKO mice exhibit CCMs ranging in size from early-stage, isolated caverns to large, multicavernous lesions. Late-stage CCM lesions in older Pdcd10^BECKO mice (6 mo) display many characteristics of human CCM lesions, including hemosiderin deposits, increased EC proliferation, and PC loss. Thus, Pdcd10^BECKO mice are a reliable model that closely resembles human CCM pathogenesis. A similar model has been established for a Ccm2 or Ccm3 deletion with the Slcoc1-CreERT2 system (Tang et al. 2019; Cardoso et al. 2020).

Local Combination Deletion Model in Adult Mice

Recent studies have identified gain-of-function (GOF) mutations in phosphatidylinositol-4,5-bisphosphate3-kinase catalytic subunit α (PI3KCA) and CCM LOF somatic mutations in the same cells in a majority of human CCMs (Ren et al. 2021). Moreover, a genetic gain of PIK3CA function accomplished by expressing PIK3CA H1047R exclusively in BECs with the Slco1c1(BAC)-CreERT2 transgene augments CCM formation in Krit1 LOF neonatal mice. Similarly, endothelial loss of PI3K signaling inhibitor PTEN also augments CCM formation of Krit1 LOF in a dose-dependent manner in neonatal animals. Because of leakage of the Slcoc1-CreERT2 system, however, it has not been feasible to test whether PI3KCA and Krit1 LOF at the adult stage induce CCM lesions. To overcome this limitation, AAV viruses encoding Cre recombinase are directly injected into the brains through cranial windows to achieve PI3KCA and Krit1 LOF in mice 2 mo of age. Interestingly, neither CCM LOF nor PIK3CA GOF alone is sufficient for lesion formation in the mature brain, but the combination of the two results in the rapid formation of CCM lesions (Ren et al. 2021). It worth mentioning that the Cre activity is detected in both endothelial and neuronal cells following direct viral injection. Therefore, it cannot be excluded that other cell types may contribute to the disease progression in this system.

PC-SPECIFIC CCM3 DELETION AS A NEW CCM MOUSE MODEL

It has been proposed that CCM lesions may arise from abnormalities in surrounding cells (Boulday et al. 2009). In addition to the endothelium, the CCM3 gene is expressed in PCs that function to stabilize microvessels of the neurovascular unit. Recently, we have established mural cell-specific CCM3 KO mice using SM22a-Cre, which drives a specific deletion of Ccm3 in mural cells, including PCs and SMCs (Pdcd10^SMKO) (Wang et al. 2020). Pdcd10^SMKO mice develop CCM lesions in the brain with onset at neonatal stages. In contrast to the Pdcd10^ECKO pups that do not survive beyond P15, one-third of Pdcd10^SMKO mice survive up to 6 wk of age, exhibiting seizures and severe brain hemorrhage. FITC-dextran (fluorescein isothiocyanate-dextran) perfusion assays indicate that leakage into surrounding tissues in the cerebellum and cerebrum of Pdcd10^SMKO. This is consistent with the CCM lesions in humans, which are formed by enlarged and irregular blood vessels that often result in cerebral hemorrhage. Compared to Pdcd10^ECKO, lesions in Pdcd10^SMKO are more dispersed and detected throughout cerebellum and in areas between the midbrain/cerebellum (Fig. 1), closely resembling human lesions, which are distributed throughout the brain (66% in the cerebral hemispheres, 18% brainstem, 8% in basal ganglia or thalamus, and 6% in cerebellum) (Gross et al. 2011). The large lesions are characterized by dilated endothelium with reduced associated PCs visualized by co-immunostaining with an EC marker CD31 and a PC marker NG2, closely resembling CCM lesions in Ccm3iecKO mice and human lesions.

Figure 1.

Comparison of cerebral cavernous malformation (CCM) lesions between Pdcd10^ECKO and Pdcd10^SMKO mice. (A) CCM3^lox/lox (wild-type [WT]), Pdcd10^ECKO, and Pdcd10^SMKO mice were harvested at P10. Representative fresh brain images are shown. (B) A diagram for the distributions of CCM lesions in Pdcd10^ECKO (red) and Pdcd10^SMKO (green) mice. The locations were based on H&E staining of the sagittal plane and coronal plane of brain sections.

PC PHENOTYPE DURING CCM LESION FORMATION

Unlike previous models, Pdcd10^BECKO and Pdcd10^SMKO mice develop CCM lesions not only in the cerebellum but also in the cerebrum (Wang et al. 2020; Zhou et al. 2021). Similarly, direct injection of virus-expressing Cre can induce local lesions in the cerebrum (Ren et al. 2021). These allow investigators to visualize vascular lesion formation and progression at various stages using transcranial two-photon microscopy with live imaging in these mice under the background of mT/mG reporter mice (Pdcd10^BECKO:mT/mG). Pdcd10^BECKO:mT/mG mice reveal lesions that have significantly greater capillary branches and vessel diameters with turbulent blood flow compared to control mT/mG mice. Capillaries of Pdcd10^BECKO:mT/mG mice consist of clusters of enlarged endothelial channels (“caverns”) that are arranged back-to-back to form densely packed sinusoids as described for human lesions (Rigamonti et al. 1988; Otten et al. 1989; Revencu and Vikkula 2006).

PCs are an integral component of the neurovascular unit and play fundamental roles in the development and maintenance of the brain vascular network (Winkler et al. 2011; Vanlandewijck et al. 2018). Previous studies from human samples (Rigamonti et al. 1988; Otten et al. 1989; Revencu and Vikkula 2006) and CCM2-deficient mouse models have suggested that CCM lesions affect the venous bed, but not the arterial compartment (Boulday et al. 2011); these studies were based on histology, immunostaining, electron microscopy (EM), and micro-computed tomography [CT] analyses, providing important information regarding vessel structural changes and EC–PC interactions (Schulz et al. 2015; Zhou et al. 2016a). However, it has not been possible to visualize dynamic vascular changes during CCM development. We have specifically labeled brain PCs using the fluorescent Nissl dye NeuroTrace 500/525 and used it to perform high-resolution optical imaging in live animals (Daneman et al. 2010). Topical application of this dye through a cranial window labeled a distinct population of cells that line cerebral blood vessels at cortical depths up to 400 µm. Dye labeling is very bright and concentrated both in cell soma and throughout the processes where it displays a punctate pattern. Colabeling of perfused vessels with Texas Red IV dye shows NeuroTrace-labeled cells lining the smallest cerebral vessels and exhibiting the morphology of capillary PCs (Zhou et al. 2021). Specifically, these cells have multiple slender processes extending longitudinally and spanning several vessel branches. Interestingly, we have observed that PCs shorten their processes and gradually dissociate from dilated vessels within CCM lesions, resulting in a reduced number of PCs surrounding the single layer of endothelium in the late CCM lesions (Fig. 2). These changes are confirmed by immunofluorescence staining of brain sections in moderate CCM lesions that exhibit loose association of PC with EC and lower rate of vessel coverage compared to normal brain microvessels (Zhou et al. 2021).

Figure 2.

Characterization of cerebral cavernous malformation (CCM) lesions and endothelial cell (EC)–pericyte (PC) interactions by two-photon microscopy. (A) P19 Ctrl Mfsd2aCreERT2:mT/mG and Pdcd10^BECKO mice were subjected to two-photon microscopy. Note that brain ECs (BECs)-Cre (Mfsd-CreER^T2) function in capillary (C), postcapillary venule, and vein (V), but not artery (A). (B) Ctrl and Pdcd10^BECKO (no mT/mG) were perfused with IV dye for vessels (red) and NeuroTrace dye for PCs. *Indicates CCM lesion. Scale bars, 10 µm (A); 50 µm (B).

It has been previously reported that ECs in the CCM lesion undergo endothelial–mesenchymal transition (EndMT) with gain of α-smooth muscle actin (SMA) expression (Maddaluno et al. 2013). A subsequent study by Zhou et al. did not detect EndMT in CCM lesions (Zhou et al. 2016b). To address this discrepancy, we use superresolution STED microscopy of brain sections and confocal imaging of whole-mount retina staining, and we detect α-SMA expression in some PCs but not in ECs during early CCM formation in both brain and retinas of Pdcd10^BECKO mice (Zhou et al. 2021) with similar results observed in the Pdcd10^SMKO mice (Wang et al. 2020; Zhou et al. 2021). Brain and retinal PCs basally express α-SMA (Alarcon-Martinez et al. 2018), which can be up-regulated by TGF-β (Verbeek et al. 1994), a cytokine that has been shown to be increased in CCM lesions (Maddaluno et al. 2013). The contribution of the SMA expression in PCs and potentially in other cells to CCM pathogenesis needs further investigation.

UNIQUE CEREBRAL EC–PC INTERACTIONS CONTRIBUTE TO THE CCM LESION DEVELOPMENT

Why CCMs are confined to the CNS in humans is not fully understood but may be attributed to the unique structure of the CNS neurovascular unit (Lee et al. 2009). PC coverage varies among different types of vessels. The PC⁄EC ratio ranges from 1:100 in skeletal muscle to 1:1 in the brain and retina (Bell et al. 2010; Daneman et al. 2010; Winkler et al. 2011). In general, vessels in the CNS exhibit the highest PC coverage, highlighting the importance of PCs in the formation and maintenance of the CNS vasculature (Bell et al. 2010; Daneman et al. 2010; Winkler et al. 2011). PCs are vascular contractile cells related to the vascular SMC lineage. However, PC exhibit some distinct cellular makers (e.g., NG2) and special functions. It is reported that regional blood flow in the normal and ischemic brain is controlled by arteriolar SMC contractility and not by capillary PCs (Hill et al. 2015). However, PC recruitment is critical for capillary assembly and stabilization as shown in mice and in cocultures (Lindahl et al. 1997; Stratman et al. 2009). Importantly, PCs and ECs share a basement membrane, enabling them to communicate directly. PC protrusions (pegs) insert into EC invaginations (sockets) at occasional interruptions in the basement membrane, providing structural support as well as direct heterotypic cell–cell communications through tight, gap, and adherence junctions (Fig. 3). The extent of EC–PC interactions in the neurovascular structure may explain why CCM lesions are confined to brain vasculature. This concept is supported by the observation that CCM3 deletion in vascular ECs or PCs induces CCM lesions with 100% penetration (Zhou et al. 2016a, 2021; Wang et al. 2020). Deletion of CCM3 in astrocytes/neuronal cells (Louvi et al. 2011), but not in myeloid cells (Zhang et al. 2013), induce disruption of the neurovascular unit and CCM lesions in mice. These results suggest that CCM3 in both vascular ECs and PCs is critical for neurovascular integrity; loss of CCM3 alters EC–PC interactions, contributing to vascular disassembly and capillary dilation within the neurovascular unit seen in CCM lesions. It is possible that CCM3-deficient astrocytes weakly induce PC–EC dissociations. A recent report supports this idea. Specifically, proliferative astrocytes are shown to play a critical role in CCM pathogenesis by serving as a major source of VEGF, which is known to promote EC–PC disruption and angiogenesis (Lopez-Ramirez et al. 2021). While Ccm ablation leads to the activation of an MEKK3-MEK5-ERK5-MEF2 signaling axis that induces a strong increase in KLF4 in ECs (Maddaluno et al. 2013; Zhou et al. 2016b), augmented KLF4 in turn activates endothelial nitric oxide (NO), driving the increase of VEGF in astrocytes, forming a positive feedback loop between CCM endothelium and astrocytes during vascular lesion development. Taken together, the high PC coverage is unique to the neurovasculature, and loss of CCM3 causes BBB disruption as the inciting event for the development of CCM.

Figure 3.

Model for CCM3 loss-induced endothelial cell (EC)–pericyte (PC) dissociation and cerebral cavernous malformation (CCM) disease. In normal condition, ECs secrete PDGF-BB that binds to PDGFRβ to recruit PCs. On the other hand, Ang1 secreted from PCs binds to the tyrosine kinase receptor Tie2 in ECs, and therefore tightens junction formation and EC–PC interaction to help orchestrate vessel maturation. However, in the condition of CCM3 mutation, Ang2 secreted from ECs antagonizes Ang1 and binds to Tie2, disrupting EC–PC interaction. Ang2 can also bind to integrins on PCs to induce PC apoptosis, thus leading to EC–PC dissociation and CCM formation.

MECHANISM BY WHICH CCM3 REGULATES EC–PC INTERACTIONS IN CCM

The mechanistic basis of several key signaling pathways involved in EC–PC interactions has been described. These pathways include PDGF-BB/PDGFRβ-dependent PC recruitment, integrin-mediated matrix interactions and cell migration, TGF-β-signaling regulation of PC differentiation, Ang-Tie2 receptor-dependent PC maturation, and EC–PC adherent junction-dependent vascular stability (Gaengel et al. 2009). The fact that both EC and PC deletion of CCM3 genes generate similar CCM lesions prompted us to examine whether CCM3 mediates these pathways involved in EC–PC interactions. We found that loss of CCM3 in ECs increases not only the ligand Ang2 release from ECs, but also its cognate receptor Tie2 on ECs. Importantly, through genetic and pharmacological approaches, we have demonstrated that Ang2 secretion and Tie2 activation play critical roles in CCM lesion formation (Zhou et al. 2016a, 2021). Several other signaling pathways in ECs have been implicated in CCM including RhoA-dependent stress fiber formation, TGF-β/Smad/BMP-mediated EndMT signaling, and MEKK3-ERK5-KLF4-mediated matrix remodeling (Boulday et al. 2009; Chan et al. 2011; Cunningham et al. 2011; McDonald et al. 2011; Maddaluno et al. 2013; Zhou et al. 2016b). Our recent study shows that conditional CCM3 deletion in PCs strongly induces disruption of the neurovascular unit and CCM lesion formation in mice (Wang et al. 2020), suggesting that CCM3 functions in both vascular ECs and PCs to maintain neurovascular integrity, likely by regulating molecules involved in EC–PC interactions.

It is of interest then to understand how PC-specific CCM3 deletion induces a similar lesion as Pdcd10^BECKO. Ang1 is known to lead to tighter EC junctions and mature vessels (Gaengel et al. 2009). However, CCM3 deletion in PC does not alter Ang1 expression. Interestingly, CCM3 loss in PC drastically augments cell surface expression and activity of β-integrins concomitant with enhanced PC adhesion but reduced migration. Moreover, reduced PC migration leads to disrupted EC–PC associations, a common event in the CCM lesion development.

This leads to the question of whether CCM gene mutations occur in PCs of human CCM lesions. Laser capture microdissection and immunohistochemical staining of the CCM proteins in the lesion endothelium have established that these mutations occur in ECs (Gault et al. 2005; Akers et al. 2009; Pagenstecher et al. 2009). However, one important caveat of human CCM specimen studies is that genetic and immunostaining analysis of lesions is limited to those that have been surgically resected, as asymptomatic lesions are rarely removed. Therefore, selected lesions tend to be very late stages, a time when PCs had dissociated from endothelium or were not present at all. It is critical to analyze early or asymptomatic lesions to determine whether CCM genes are mutated in PCs. Interestingly, it has been reported that Angt2 can engage α3β1-integrin on PCs to induce PC apoptosis in diabetic retinopathy (Felcht et al. 2012; Park et al. 2019). Thus, it is plausible that CCM ECs and PCs also form a positive feedback loop in promoting vascular lesion development. Specifically, CCM3 loss in PCs may increase surface expression of β-integrins, thereby reducing PC migration toward ECs and attenuating vascular stability; on the other hand, CCM3 loss in ECs increases Ang2 secretion, which induces apoptosis/senescence of integrin-high PCs, leading to disrupted EC–PC association and vascular malformations (Fig. 3).

CCM3-MEDIATED VESICLE TRAFFICKING AND Ang2/Tie2 IN ECs

It has been proposed that the MEKK3/KLF4 signaling could be a central hub that is regulated by all CCM proteins contributing to the onset and progression of CCM (Boulday et al. 2009; Chan et al. 2011; Cunningham et al. 2011; McDonald et al. 2011; Maddaluno et al. 2013; Zhou et al. 2016b). However, it is unclear how MEKK3/KLF4 regulates EC–PC associations. Moreover, CCM3 has been shown to be signaling distinct from CCM1 and CCM2, and CCM3 deficiency in various species and tissues induces specific phenotypes. It appears that CCM3 has a specific cellular function in controlling vesicle trafficking, a concept supported by a number of studies: (1) CCM3 suppresses VEGFR2 endocytosis to stabilize VEGFR2 on the endothelial membrane (He et al. 2010); (2) CCM3 stabilizes GCKIII proteins to promote Golgi assembly (Fidalgo et al. 2010); (3) LOF mutants of fly CCM3 or GCKIII kinase ortholog have dilated tracheal tubes, which resembles dilated blood vessels found in CCM patients (Song et al. 2013). This tube dilation phenotype can be suppressed by a reduction in the expression of N-ethylmaleimide-sensitive factor 2 (NSF2), a protein involved in SNARE recycling and the secondary phase of exocytosis (Zhao et al. 2012). (4) Mice with deficiency of CCM3, but not of Krit1 or CCM2, in the kidney tubules develop polyuria and display increased water consumption, resulting from the reduced aquaporin 2 protein in the apical membrane of tubular epithelial cells. The loss of aquaporin 2 is associated with increased expression and membrane targeting of ezrin, radixin, moesin (ERM) proteins and impaired intracellular vesicle trafficking in Ccm3-deficient mice. Moreover, treatment with erlotinib, a tyrosine kinase inhibitor promoting exocytosis and inhibiting endocytosis, normalized the expression level and membrane abundance of aquaporin 2 protein, and partially rescued the water reabsorption defect observed in the Ccm3-deficient mice (Wang et al. 2021).

We have uncovered a critical role of CCM3 via GCK member STK24/25 in UNC13-mediated exocytosis (Zhang et al. 2013). CCM3 deletion results in reduced STK24/25, which leads to a reduction in STK24/25-mediated inhibition of UNC13 binding to lipid and an increase in granule docking/priming. In brain vascular ECs, CCM3 suppresses UNC13B- and vesicle-associated membrane protein 3 (VAMP3)-dependent exocytosis of Ang2. CCM3 deficiency in ECs increases the exocytosis and secretion of Ang2, which leads to destabilized EC junctions, enlarged lumen formation, and EC–PC dissociation. UNC13B deficiency, which blunts Angt2 secretion by ECs, or treatment with an Ang2-neutralizing antibody normalizes the defects in the brain and retina caused by EC-specific CCM3 deficiency, including the disruption of endothelial junctions, vessel dilation, and PC dissociation (Zhou et al. 2016a).

Interestingly, CCM3 also regulates Tie2, the cognate receptor of Ang2, through a distinct mechanism (Zhou et al. 2021). We have demonstrated that CCM3 deletion enhances the expression of Tie2 at the endothelial surface via caveolae-dependent but Unc13B-mediated exocytosis-independent mechanisms. On the other hand, Cav1-caveolae regulates Tie2 signaling via modulating Tie2 protein stability. Caveolae is limited in the mouse brain vasculature and increased caveolae are associated with increased vascular permeability (Ben-Zvi et al. 2014; Andreone et al. 2017). There is a dramatic increase in the number of caveolae and increased caveolae structural protein Cav1 protein in Pdcd10^BECKO BECs. Caveolae are detected in intracellular, luminal, and abluminal compartments with “rosettes” patterns. Similarly, Cav1 protein is distributed throughout the cells in CCM3-deficient ECs compared to more membrane-association in wild-type (WT) cells. These results indicate an active caveolae-mediated trafficking in Pdcd10^BECKO BECs (Parton et al. 2020). Indeed, mechanistic studies support that Cav1 mediates Tie2 surface expression and trafficking in CCM3-deficient ECs. Importantly, Cav1 and Tie2 protein and Tie2 activity are highly up-regulated in the lesions of human CCM specimens. Moreover, genetic rescue with Cav1-deficient mice supports that increased caveolae are required for CCM3 loss-induced Tie2 signaling, vascular permeability, and CCM lesion formation. It has been reported that Tie2 activation by Ang1 or Ang2 results in colocalization of Tie2 with clathrin-coated vesicles followed by Tie2 endocytosis and degradation (Bogacka et al. 2005). In contrast, colocalization of Tie2 with clathrin-positive endosome and subsequently with lysosomes is dramatically reduced and colocalization of Tie2 with Cav1 is increased in CCM3-KO cells. These studies support that Cav1/caveolae mediate Tie2 intracellular trafficking, leading to increased Tie2 protein stabilization and signaling in CCM3-deficient EC. Of note, CCM3, but not CCM1 or CCM2, specifically interacts with GCKs and the STRIPAK complex. Similarly, Ang2-Tie2 signaling is increased in CCM3-knockdown (KD), but not in CCM1- or CCM2-deficient ECs (Zhou et al. 2016a).

CCM3 REGULATES INTRINSIC PC SIGNALING IN CCM

Interestingly, CCM3 regulates integrins in PCs through a similar exocytosis mechanism. As reported for ECs (Zhou et al. 2016a), KD of CCM3 in human brain microvascular PCs (HBMVPCs) reduced GCK STK25 because CCM3 stabilizes STK25. CCM3 KD significantly increases β1-integrin expression, the downstream effectors phosphorylation of focal adhesion kinase (FAK), and the binding protein paxillin (Wang et al. 2020). These CCM3KD-induced effects can be suppressed by cosilencing UNC13B or VAMP3 (W Min and JH Zhou, unpubl.), suggesting that CCM3/UNC13B/VAMP3 may regulate β1-integrin exocytosis in PCs. We have previously reported the crystal structure of CCM3 and reported that CCM3 contains two distinct domains—the amino-terminal dimerization domain (aa1-91), which are almost entirely hydrophobic, and the carboxy terminal of focal adhesion targeting (FAT) domains (aa 92-212) resembling that of Pyk2 and FAK (Li et al. 2010, 2011). While there is high structural similarity between CCM3 and the FAT domains, there is low sequence homology. We show that CCM3 binds paxillin via the FAT domain and that mutation of a highly conserved FAK-like hydrophobic pocket (HP1) abrogates CCM3–paxillin interactions. Moreover, CCM3 and paxillin are colocalized in PCs. Interestingly, KD of CCM3 in PCs enhances paxillin localization at the focal adhesion with an enlarged focal adhesion complex, suggesting CCM3 may act as a suppressor of paxillin–FAK complex assembly (Li et al. 2010, 2011). Disease-related CCM3 truncations affect the FAT domain suggesting a role for the FAT-homology domain in the etiology of CCM. Despite their augmented adhesion to the ECM, CCM3-deficient PCs exhibit reduced cell spreading, protrusion, and migration, defects that directly contribute to diminished association with ECs, a hallmark of CCM initiation and progression (Zhou et al. 2016a, 2021; Wang et al. 2020). Increased adhesion with reduced protrusion and migration likely restrain PC processes over capillaries, leading to vascular dilation and leakage in Pdcd10^SMKO mice (Fig. 4).

Figure 4.

A model for caveolae-mediated Tie2 signaling in cerebral cavernous malformation (CCM) progression. CCM3 in brain endothelial cells (BECs) normally suppresses caveolae by a yet unknown mechanism. CCM3 loss in endothelial cells (ECs) causes enhanced caveolae-mediated Tie2 surface expression and endocytosis, leading to EC junctional disruption, EC–pericyte (PC) dissociation, blood–brain barrier (BBB) dysfunction, and lumen enlargement. Tie2 inhibition, like neutralization of Ang2 ligand, ameliorates CCM lesion progression in the mouse CCM models, with normalization of EC–PC interaction and vascular integrity. Loss of CCM3 augments UNC13B/VAMP3-mediated integrin surface targeting (not depicted in the cartoon), increasing PC adhesion with reduced migration, which leads to disruption of junctions and associations of PCs with ECs.

CONCLUDING REMARKS

Nearly all current studies of inheritable CCMs have focused on the role of CCM proteins in ECs. However, this focus fails to explain why CCMs have a predilection for the vessels of the CNS. We note that CNS microvessels are unique in having an extraordinarily high PC to EC ratio, forming a high resistance BBB further supplemented by astrocyte foot processes. Importantly, we have shown that targeted disruption of CCM3 in PC (and to a lesser extent astrocytes) also disrupts BBB and phenocopies gene disruption in EC. These data indicate that CCM arises from dysregulated heterotypic cell interactions, especially between EC and PC. We have defined the EC- and PC-intrinsic pathways contributing to CCM disease, and the pivotal role of CCM3 in regulating cellular exocytosis and membrane targeting. Loss of CCM3 in EC increases EC-derived Ang2 release and the expression of its receptor Tie2, which disrupts EC–PC association and induces CCM lesion formation. Loss of CCM3 in PC induces augmented integrin expression with enlarged focal adhesion complex yet reduced PC migration, leading to vascular malformation with blunted PC recruitment. Therefore, CCM3 loss in ECs or PCs induces CCM lesions with similar characteristics—lack of PCs in a single dilated endothelium layer. Therefore, disrupted EC–PC association may present a common pathway in CCM lesion formation, and stabilization of EC–PC interactions in the neurovascular unit by anti-Ang2, Tie2 inhibitors, and/or inhibition of β1-integrins may be a potential therapeutic strategy for the treatment of CCM disease.