Introduction

Cerebral cavernous malformations (CCM) are vascular lesions of the brain affecting 0.4–0.8 % of the general population [1, 3, 52]. These lesions account for 5–13 % of all the vascular malformation in the central nervous system (CNS) [9, 53] and manifest a great spectrum of clinical symptoms such as headaches, motor and sensory deficits, epilepsy, and hemorrhagic strokes [4, 62, 87]. Histologically, CCM have dilated thin-walled capillaries with clusters of enlarged endothelial channels (caverns) lacking intermediate brain parenchyma [12, 16] and tight junctions [32, 64]. CCM can be classified as either sporadic or familial with 80 % accounting to the former and 20 % to the latter [18, 58]. Unlike the sporadic lesions which are mostly accompanied with a single lesion, familial cases of CCM are characterized by the presence of multiple lesions with autosomal dominant pattern of inheritance [58]. Such familial types of lesions are associated with loss of function mutations in the following three genes: KRIT1 (Krev interaction trapped 1) [13], CCM2/OSM/MALCAVERNIN [45], and PDCD10 (programmed cell death 10)/TFAR15 [2]. Similar to other tumor suppressor genes, CCM pathogenesis also follows the Knudsonian two-hit mechanism, in which loss of one allele due to a germline mutation in an affected cell (first hit) is accompanied with somatic mutation in the other (second hit) [58]. Research in the past few years have shown extensively how different signaling pathways are regulated by the product of these mutated genes. However, understanding how and where these signaling pathways cross talk may shed light on their function and help therapeutically to patients suffering from CCM. Phosphatase and tensin homologue deleted on chromosome ten (PTEN) has a crucial role to play in the regulation of angiogenesis in tumor tissues, but its relevance to CCM pathogenesis remains unexplored. In this review, we will discuss the recent insights and provide an update literature reviews into the role of PTEN/PI3K/Akt/VEGF pathway in CCM pathogenesis. In the first section, current understanding of the brief molecular architecture of the CCM proteins is reviewed. Next, recent updates into the PTEN protein structure and its interplay with the PI3K/Akt/VEGF signaling during angiogenesis and CCM pathogenesis are discussed. Finally, in the concluding part of this article, we will focus on how such pathways could help to better understand the biology of CCM that may favor future therapeutic treatments.

Structure of CCM proteins

This section will discuss briefly the structural domain and the molecular interaction of three CCM proteins: KRIT1, CCM2, and PDCD10.

KRIT1

Among the three CCM genes known, KRIT1 remain the first gene to be identified and its linkage mapping to chromosome 7q has been broadly studied [1, 23, 34, 61]. KRIT1 gene encodes KRIT1, a 736-amino acid protein consisting of a C-terminal FERM domain (F for 4.1 protein, ezrin, radixin, moesin), an N-terminal region consisting of three NPxY motif (NPAY195, NPLF234, NPYF253) and 3 ankyrin repeating domains [24, 60, 81] (Fig. 1). In the NPxY motif, NPAY195 interacts with ICAP1 (integrin cytoplasmic-associated protein) and regulates vascular development, cell migration, and angiogenesis [48]. The FERM domain of KRIT1 comprises of three subdomains, F1, F2, and F3, in which the F3 subdomain possess a phosphotyrosine-binding fold (PTB) [17]. A large spectrum of protein binds to these domains in order to regulate cell-cell interactions and vascular permeability [22]. The F1 subdomain utilizes an ubiquitin-like shape, the F2 subdomain folds similar to acyl-CoA homology and the F3 subdomain simulate a PTB/pleckstrin homology [57]. Recent studies have shown that KRIT1 FERM domain is able to form complexes with the N-terminus of the protein, the heart of glass (HEG1) C-terminus, and the small GTPase Rap1 [17, 20, 21, 37]. The GTPase Rap1 binds to both the F1 and F2 subdomains of KRIT1 FERM domain, while HEG1 interacts within the F1 and F3 interface. The binding of KRIT1 protein and the Rap1 is important for endothelial cell (EC) stabilization [22]. KRIT1 is primarily localized in the cytoplasm and encodes for nuclear localization sequence (NLS) and a nuclear export sequence (NES) [1]. This sequence helps the intracellular protein to shuttle between the cytoplasm and the nucleus [79]. Liu et al. showed that direct binding of Rap1 to KRIT1 FERM domain enables it to localize to the cell-cell junctions and suppress Rho kinase signaling [47]. Moreover, KRIT1 interacts with CCM2 through two of its Npxy motifs, NPLF234 and NPYF253 [82], and regulate cellular functions such as endothelial cell-cell junction stabilization and vascular permeability.

Fig. 1
figure 1

Structural assembly of CCM proteins. KRIT1 consists of a C-terminal FERM domain (F for 4.1 protein, ezrin, radixin, moesin), an N-terminal region consisting of three NPxY motif (NPAY195, NPLF234, NPYF253) and 3 ANK domain. The NPAY195 domain binds to ICAP1, while the NPLF234 and NPYF253 domain interacts with the CCM2 PTB domain. The F3 subunit of FERM domain has a PTB-like domain and binds to multiple proteins. Additionally, KRIT1 has an N-terminal NLS and a C-terminal NES region to help shuttle from cytoplasm to nucleus. CCM2 consist a PTB domain and a C-terminal HHD. The PTB domain shares similarity to the ICAP1 and binds to KRIT1, PDCD10, and TrkA. The PDCD10 has a much simpler structure and consists of an N-terminal dimerization domain and a C-terminus FAT-H domain. PDCD10 binds to a variety of proteins including CCM2, germinal center kinases III subfamily, GCKIII (STK24, STK25, and MST4), paxillin, and VEGFR. ANK ankyrin repeat domain; FERM, F for 4.1 protein, ezrin, radixin, moesin; NLS nuclear localization signal; NES nuclear export signal; PTB phosphotyrosine-binding domain; HHD harmonin-homology domain; TrkA tyrosine kinase A; FAT-H focal adhesion targeting homology domain; VEGFR vascular endothelial growth factor receptor

Full size image

CCM2

CCM2 is important for the regulation of cell-cell interactions and endothelial cell lumen formation. It is a 444 amino acid protein and is known to interact with KRIT1 through its putative PTB domain (Fig. 1). CCM2 is thought to mediate its binding to PDCD10 through its putative C-terminus located slightly before the PTB domain, although the detailed mechanism of interaction is not well understood [18]. The KRIT1-CCM2-PDCD10 complex formation, therefore, is achievable only in the presence of CCM2 which acts as the hub bringing together KRIT1 and PDCD10. Zhu et al. also demonstrated that silencing CCM2 gene in endothelial cells activate the Akt signaling proteins [89], indicating that CCM2 regulates PI3K/Akt signaling. Notwithstanding that the C-terminal region of the CCM2 domain is approximately 200-amino acid region of unknown function, a recent study have shown that CCM2 C-terminus is able to induce cell death through TrkA receptor tyrosine kinase [27]. A similar study recently identified the molecular and crystal structure of the CCM2 C-terminus, suggesting that there exists a folded region of the CCM2 C-terminus with structural similarity to scaffolding protein harmonin [18]. The mechanism by which the CCM2 harmonin domain mediates signaling to induce cell death in the endothelial cells still remains elusive. Recently, a putative CCM2-Like (CCM2L) protein was identified in the endothelial cells of mice and zebrafish during cardiovascular development [59, 84]. This protein is able to compete with its paralog CCM2 and binds to HEG-KRIT1 complex rather interacting with PDCD10. Moreover, Rosen et al. [59] predicted that the human homologue of CCM2L, C20ORF160 may be involved in the pathogenesis of CCM.

PDCD10

PDCD10 or alternatively named as TF-1 cell apoptosis-related protein (TFAR15) is a 212-amino acid protein consisting of a N-terminus dimerization domain and a C-terminus focal adhesion targeting homology domain (FAT-H) [12, 25, 42] (Fig. 1). It has been recently identified and known to be involved in apoptosis. Although studies have documented that PDCD10 exists in complex with KRIT1 and CCM2 [38, 67], there are lines of evidences suggesting PDCD10 might have roles distinct from that of KRIT1 and CCM2. For example, Chan et al. demonstrated that PDCD10 has a non-endothelial role in different developments from KRIT1 and CCM2 [7]. Besides, there exist separate binding partners for the three CCM proteins. Li et al. showed that the interaction between PDCD10 with CCM2 PTB domain and paxillin LD motifs is mediated through its FAT-H domain and CCM pathogenesis is related to point mutations in the PDCD10 FAT-H domain [42]. Moreover, substantial evidences implicate the critical role of germinal center kinases III subfamily, GCKIII (STK24, STK25, and MST4), in PDCD10 function and its contribution to CCM pathogenesis [31, 69, 75, 76]. Fidalgo et al. showed that PDCD10 localizes in the Golgi body and forms a complex with GCKIII kinases and GM130, a Golgi protein that is important for Golgi assembly and orientation [15]. Other partners known to interact with PDCD10 are protocadherin ƴ [44] important for neuronal survival and VEGFR2, crucial for vascular development and angiogenesis [28].

Any other additional genes?

Besides the three known CCM genes, previous studies speculate the existence of an additional CCM locus.

Bergametti et al. observed low frequency mutation of PDCD10 in 8 of 20 families that were negative for KRIT1 and CCM2 [2]. In another study, Liquori et al. reported that the sequence analysis of PDCD10 in 29 probands lacking KRIT1 and CCM2 mutations revealed only three mutations [46]. The probable explanations for such lower number of mutation in CCM affected families might be the following: (a) the existence of a fourth unidentified CCM gene which could have gone undetected by the exon sequencing, (b) hypermethylation of CpG sites in the promoter regions of the KRIT1 and PDCD10 genes that carry CpG islands, and (c) mutations in regulatory elements of the KRIT1, CCM2, and PDCD10 genes like enhancers distant to the gene locus. Further linkage and high throughput mutational screening might be interesting to shed light into the structure and function of this hypothesized CCM locus.

PTEN structure

The successful identification of homozygous deletion in human chromosome 10q23 in diverse human cancer types led to the isolation and characterization of the tumor suppressor gene PTEN/MMAC1 (phosphatase and tensin homologue deleted on chromosome ten) [41, 68]. PTEN, a 403-amino acid protein consists of a short N-terminal phosphatidylinositol-4-5-bisphosphate (Ptdlns (4,5) P2)-interacting motif (PBD), a catalytic signature motif, HCXXGXXR located in the active sites of protein tyrosine phosphatases (PTPs) [40, 66]. Located beyond the PTPs domain, is a C2 domain, a carboxy-terminal tail, and a PDZ-binding motif [56, 66]. The carboxy-terminal tail is known to stabilize PTEN, and mutations in this domain induce tumor growth and proliferation [19]. A host of other mutations in the phosphatase and C2 domain also cause protein misfolding and results in uncontrolled cellular process. An essential function of PTEN is to oppose the activity of phosphatidylinositol 3-kinases (PI3K)/Akt pathway by hydrolyzing phosphatidylinositol-3,4,5-triphosphate (PIP3) to phosphatidylinositol −4,5-biphosphate (PIP2) [49].

The PI3K/Akt signaling

The PI3K/Akt signaling pathway is well established and targeted by PTEN through its lipid phosphatase activity [78]. The identification of this pathway dates back to 1980s with the characterization of insulin receptor signaling that subsequently led to the recognition of the components of PI3K/Akt pathway [5, 29]. The PI3K/Akt pathway is highly conserved and is regulated by a multistep signaling process. Based on their structural and functional attributes, the PI3K are divided into three separate classes: class I, II, and III [33, 66], of which class I is the best characterized and is involved in regulating cell survival and growth [6]. The class IA PI3Ks consists of p110α, p110β, and p110δ catalytic subunits and are activated by receptor tyrosine kinases (RTKs), whereas the class IB PI3Ks (PI3Kƴ) are activated by G protein-coupled receptors (GPCRs) [14]. The class II PI3Ks (PIK3C2α, PIK3C2β, and PIK3C2ƴ) lack regulatory subunits and comprises a common C-terminus C2 motif [33]. Additionally, the class III PI3Ks uses phosphatidylinositol (Ptdlns) as their substrate and bears a catalytic as well as an adaptor subunit [74]. The protein-serine-threonine kinase AKT (protein kinase B) is a crucial downstream regulator of the PI3Ks pathway and consists of three isoforms: AKT1, AKT2, and AKT3 [63]. These isoforms have a common pleckstrin homology (PH) domain, a catalytic, and a C-terminal hydrophobic motif which when phosphorylated regulates downstream signaling process such as cell survival, growth, and metabolism. A major role of PTEN is to negatively regulate the PI3K/Akt signaling. Following functional loss of PTEN, PIP3 accumulation in the cell activates the downstream Akt signaling, resulting in excessive cell growth, decreased apoptosis, and angiogenesis.

Role of PTEN/PI3K/VEGF signaling in angiogenesis

Here, we will briefly discuss the interplay of PTEN, PI3K, VEGF, and hypoxia in angiogenesis and then proceed to define their role in CCM pathogenesis. Previous studies have identified the central role of PTEN mutation in angiogenesis [72, 77]. However, Kini et al. demonstrated the potential role of PTEN in inducing angiogenesis in endothelial cells [36]. Although the exact mechanism is not clear, further studies are required to delineate the role of PTEN activation in endothelial cells. Angiogenesis can be triggered by growth factors, by PI3K pathway alterations, or by mutations in the PTEN gene. The vascular endothelial growth factor (VEGF), a common proangiogenic factor, is crucial in regulating vasculogenesis and angiogenesis via the PI3K pathway. Mammalian VEGF are classified into five related growth factors: VEGF-A, VEGF-B, VEGF-C, and VEGF-D and placental growth factor (PLGF) [33]. Among them, VEGF-A is crucial in regulating angiogenesis in the endothelial cells [33]. VEGF-B role in angiogenesis is not clear while VEGF-C and -D are predicted to stimulate lymphangiogenesis. Studies showed that PTEN loss promotes VEGF-mediated angiogenesis during normal vascular development and tumor formation both in vitro and in vivo [8, 26, 30]. Apart from the role of VEGF in triggering angiogenesis, angiopoietins (ANG) also possess angiogenic property and mediate signaling through the endothelial membrane receptor (TIE2). ANG-1 and −2 are solely expressed in the endothelial cells and may have antagonistic properties [10, 50]. ANG-1 can activate the PI3K pathway by interacting with the p85 subunit of PI3K in a phosphotyrosine-dependent manner [35], whereas ANG-2 can be downregulated following the activation of PI3K/Akt pathway [73]. Tsigkos et al. also demonstrated that PTEN acts as an agonist for ANG-2 release in the endothelial cells. The potent role of PTEN, PI3K, VEGF, and ANG under hypoxic conditions [85, 90] has also been well documented. Hypoxia, a condition of decreased oxygen concentration in the cells, is reported to be a key regulator of angiogenesis. Hypoxia increases the VEGF expression by binding to the hypoxia-responsive element (HRE) and results in the formation of hypoxia-inducible factor 1 (HIF-1), a heterodimer consisting of HIF-1α and HIF-1β subunits [65]. HIF-1 is predicted to regulate several genes involved in cell survival, proliferation, migration, and angiogenesis [65]. The role of p53 tumor suppressor protein, PTEN, and HIF-1 in tumor angiogenesis has been previously reported [55]. The oncoprotein, MDM2 (mammalian double minute 2), is phosphorylated upstream by the Akt molecule, and its activation results in the inhibition of p53 protein post transcriptionally (Fig. 2).

Fig. 2
figure 2

Predicted role of PTEN/PI3K/Akt/VEGF signaling in CCM pathogenesis. PI3K activation is induced by the binding of VEGF to its receptor VEGFR2. PTEN, a tumor suppressor protein, has a crucial role in regulating angiogenesis by hydrolyzing PIP3 to PIP2, thus negatively regulating the PI3K signaling pathway. When PTEN protein is lost, PIP3 accumulation in the cells activates the downstream Akt signaling by increasing MDM2, HIF-1α, or VEGF expression. CCM angiogenesis may be mediated by phosphorylation of MDM2 which binds to p53 and inhibits its post transcriptional regulation. Such inhibition results in increased angiogenic activity. Therefore, a major strategy to check angiogenesis in endothelial cells might be to target the PTEN/PI3K/Akt or the MDM2-p53 interface by the application of inhibitors. P phosphate, VEGF vascular endothelial growth factor, VEGFR2 vascular endothelial growth factor receptor 2, PI3K phosphatidylinositol 3-kinases, PTEN phosphatase and tensin homologue, PDK1 phosphoinositide-dependent kinase 1, Akt protein-serine-threonine kinase, HIF-1α hypoxia-inducible factor-1α, MDM2 mammalian double minute 2, p53 tumor suppressor

Full size image

PTEN/PI3K/VEGF signaling in CCM pathogenesis

Endothelial cells in the cerebral vasculature are vital in maintaining brain homeostasis. However, disruption of these cells under pathological condition results in abnormal angiogenesis, leading to CCM and other vascular malformations [39, 71, 83]. Previous studies have demonstrated the role of PTEN mutation in various harmatomas, such as Cowden syndrome, Proteus-like syndrome, and Bannayan-Riley-Ruvalcava syndrome [43, 51, 86]; however, the role of PTEN deletion in CCM have only recently been identified. The involvement of PTEN promoter methylation in CCM via PI3K/Akt-independent pathway was first reported by Zhu et al. [86]. The authors demonstrated the activation of Erk 1,2 of the MAPK/Erk (mitogen-activated protein kinase/extracellular signal-related kinases) signaling pathway but not the PI3K/Akt pathway, suggesting the regulation of PTEN promoter methylation in a tissue- and disease specific manner. Interestingly, Zhu et al. were able to show that following addition of the PI3K inhibitor, wortmannin to the endothelial cells, there was no change in the increased endothelial proliferation induced by PTEN promoter silencing. How PTEN promoter methylation resulted in Erk1, 2 activation and not PI3K/Akt needs further study. The role of VEGF-PI3K signaling in CCM pathogenesis has also been reported. VEGF acts upstream in the PI3K signaling and regulates vascular development by promoting VEGFR2 binding to the C-terminal domain of PDCD10 [28]. Through computational modeling, Dibble et al. demonstrated the involvement of PDCD10 in the PI3K signaling pathway [11]. Using site-directed mutagenesis, the authors constructed a three-dimensional model of PDCD10 and were able to co-localize the wild type PDCD10 with the membrane-bound active PI3K (p110-CAAX) in the plasma membrane. Further studies are needed to rule out the role of PDCD10 in VEGF-PI3K signaling and their contribution to CCM pathogenesis. Recent reports show that following PDCD10 gene silencing, VEGF signaling was activated in the endothelial cells, implying its importance in angiogenesis [80, 88]. Moreover, the role of hypoxia has been documented not only in cancer biology but also in noncancerous vascular disease [54, 70, 88].

Conclusion

Despite the significant advances in understanding PTEN/PI3K/Akt/VEGF signaling in angiogenesis, much remains to be explored how these pathways simultaneously regulate endothelial cell function in patients suffering from CCM. It is difficult to figure out whether PTEN insufficiency alone or other molecular factors in the PI3K/Akt/VEGF signaling contribute to CCM pathogenesis. Preferably, the current available treatment for CCM suffering patients is surgical intervention, but not all the lesions located in the eloquent areas (brainstem, pons) can be successfully removed. Therefore, generation of suitable CCM gene knock-out animal models mimicking human CCM lesion would be helpful in understanding the role of PTEN in PI3K/Akt/VEGF signaling. Moreover, in order to develop therapeutic targets for CCM treatment, extensive linkage and mutational studies in humans are required to delineate the existence of the hypothesized fourth CCM locus, its interaction with other CCM genes and their contribution to the above mentioned signaling pathways.