|Home | About | Journals | Submit | Contact Us | Français|
The broad role of the transforming growth factor beta (TGFβ) signaling pathway in vascular development, homeostasis, and repair is well appreciated. Endoglin is emerging as a novel, complex, and poorly understood regulatory component of the TGFβ receptor complex, whose importance is underscored by its recognition as the site of mutations causing hereditary hemorrhagic telangiectasia (HHT) [McAllister et al., 1994]. Extensive analyses of endoglin function in normal developmental mouse models [Bourdeau et al., 1999; Li et al., 1999; Arthur et al., 2000] and in HHT animal models [Bourdeau et al., 2000; Torsney et al., 2003] exemplify the importance of understanding endoglin’s biochemical functions. However, novel mechanisms underlying the regulation of these pathways continue to emerge. These mechanisms include modification of TGFβ receptor signaling at the ligand and receptor activation level, direct effects of endoglin on cell adhesion and migration, and emerging roles for endoglin in the determination of stem cell fate and tissue patterning. The purpose of this review is to highlight the cellular and molecular studies that underscore the central role of endoglin in vascular development and disease.
The canonical transforming growth factor beta (TGFβ) signaling pathway comprises seven type I and five type II TGFβ receptors [Manning etal.,2002]. The TGFβ type I receptors are serine and threonine kinases, which include activin-like kinase 1 (ALK1) and TβRI, also known as ALK5. ALK1 and ALK5 associate with, and are activated via ligand-dependent phosphorylation [Vivien and Wrana, 1995] by the type II TGFβ receptor, TβRII [Wrana et al., 1992]. The activated type I receptor propagates canonical or Smad-dependent signals by phosphorylating Smad proteins [Shi and Massague, 2003; Feng and Derynck, 2005]. Another component of the TGFβ system is endoglin. Endoglin is a transmembrane protein [Gougos and Letarte, 1990] that acts as an auxiliary receptor for TGFβ [Cheifetz et al., 1992; Barbara et al., 1999]. Of note, mutations in endoglin (ENG) [McAllister et al., 1994] and ALK1 [Johnson et al., 1996] genes cause the vascular dysplasia hereditary hemorrhagic telangiectasia (HHT), termed HHT1 and HHT2, respectively.
Endoglin is expressed in vascular endothelial and smooth muscle cells and plays an important role in the homeostasis of the vessel wall. Evidence to support this view includes: (1) human endoglin mutations result in the vascular disorder, HHT1; (2) murine endoglin is necessary for the process of angiogenesis and vascular smooth muscle development [Li et al., 1999]; (3) endoglin is up-regulated in the endothelia of neovascularized tissues such as tumors [Burrows et al., 1995; Kumar et al., 1996, 1999; Bodey et al., 1998; Fonsatti et al., 2003], in the thyroid disorders, Grave’s disease and Hashimoto’s thyroiditis [Marazuela et al., 1995], in psoriasis [van de Kerkhof et al., 1998], scleroderma [Rulo et al., 1995; Leask et al., 2002], and in ischemic stroke [Kumar et al., 1996]; and (4) endoglin is up-regulated in the smooth muscle cells of human atherosclerotic plaques [Conley et al., 2000], and in smooth muscle cells that respond to vascular injury [Ma et al., 2000]. Vascular injury also results in increased endoglin expression in endothelial cells [Botella etal., 2002]. Transcriptional activation of endoglin and TGFβ signaling components by cooperative interaction between Sp1 and KLF6 suggests that these factors play a role in the response to vascular injury [Botella et al., 2002]. These data support the view that understanding endoglin’s role in development and disease will provide considerable insight into the processes of angiogenesis, smooth muscle cell regulation, and vascular homeostasis.
HHT is a genetic vascular disorder that affects about one in 10,000 people [Lux and Marchuk, 2001], although recent studies suggest that this prevalence may be 1/5,000 or higher [Guttmacher et al., 1995; Kjeldsen et al., 1999; Dakeishi et al., 2002; Westermann et al., 2003]. HHT shows a significant age-dependent onset of symptoms. Adults positive for a mutant HHT endoglin allele have significantly greater risk of cerebral arteriovenous malformation and epistaxis (nose bleeding), which increases with age [Aassar et al., 1991; Shovlin et al., 1995]. Up to 1/3 of HHT patients have multiple organ involvement, which can be disabling and life threatening. The detection and treatment of HHT are now the focus of at least 24 HHT Centers worldwide, including 8 in the United States.
Clinically, HHT sufferers present with vascular dysplasia characterized by arteriovenous malformation resulting from muscularization of postcapillary venules without obvious endothelial cell defects. Microscopically, vascular lesions originate as focal dilatations of postcapillary venules followed by thickening of the vessel wall with mononuclear cell infiltration (primarily lymphocytes) and proliferation of smooth muscle cells [Braverman et al., 1990;Aassar etal., 1991]. Pulmonary arteriovenous malformations occur in ~30% of patients and are associated with serious complications that include stroke and brain abscess.
HHT1 is a dominantly inherited disorder. More than 155 distinct mutations in ENG are linked to HHT1 [Prigoda et al., 2006]. These mutations tend to cluster as premature termination codons in exons that encode the extracellular domain of the protein, and lead to truncated forms of endoglin that are not readily detectable by immunological methods [McAllister et al., 1995; Berg et al., 1996; Shovlin et al., 1997; Yamaguchi et al., 1997; Gallione et al., 1998]. These observations strongly suggest that HHT results from reduced dosage or haploinsufficiency of endoglin protein [Abdalla and Letarte, 2006].
Endoglin was originally described as a type I integral membrane protein with an extracellular domain of 561 amino acids, a hydrophobic transmembrane domain, and a 47-residue cytosolic domain [Gougos and Letarte, 1990]. Comparative analysis of the primary structure reveals that endoglin belongs to the zona pellucida (ZP) family of extracellular proteins that share a ZP domain consisting of 260 amino acids with 8 conserved cysteine residues close to the transmembrane region [Bork and Sander, 1992; Jovine et al., 2005]. This consensus ZP domain is divided in two ZP subdomains that are potentially involved in endoglin receptor oligomerization [Jovine et al., 2005; Llorca et al., 2007].
In humans, endoglin contains an RGD tripeptide located in the ZP domain of the extracellular region [Gougos and Letarte, 1990]. Although this motif led to the hypothesis that endoglin binds to integrins or other RGD-binding receptors [Gougos et al., 1992; Lastres et al., 1992], the function of the RGD sequence in human endoglin may reflect a recent adaptation because this motif is absent from mouse [Ge and Butcher, 1994], porcine [Yamashita et al., 1994], rat, and canine [Llorca et al., 2007] endoglin proteins.
The primary structure of endoglin suggests that there are four N-linked glycosylation sites in the N-terminal domain and a probable O-glycan domain, which is rich in Ser and Thr residues proximal to the membrane-spanning domain [Gougos and Letarte, 1990]. Experimental studies using specific glycosidases confirmed that endoglin is glycosylated [Gougos and Letarte, 1988]. This post-translational modification occurs in multiple stages when endoglin is overexpressed in COS cells, giving rise to partially and fully glycosylated species that are present at the cell surface [Lux et al., 2000].
The 47-residue cytosolic domain of the predominant L-isoform of endoglin constitutes the region of the protein with the highest degree of conservation among endoglins from different mammalian species, as well as with the homologous protein betaglycan [Lopez-Casillas et al., 1991]. A splicing isoform of human endoglin results in the expression of a short S-endoglin species with a distinct cytosolic domain of 14 residues [Bellon et al., 1993]. Both cytosolic domains can be phosphorylated by serine and threonine kinases [Lastres et al., 1994], including the TGFβ type I and II receptors [Guerrero-Esteo et al., 2002; Koleva et al., 2006]. Recently, a short endoglin isoform was characterized in mice [Perez-Gomez et al., 2005]. Although the L-endoglin isoform and betaglycan contain a consensus PDZ-binding motif (SerSerMetAla) present at the carboxyl terminus, the S-endoglin isoform lacks this motif. As will be discussed below, the L-form of endoglin is linked to the regulation of the adhesive properties of endoglin, and thus isoform switching of the cytosolic domain of endoglin may have potential regulatory significance to the function of endoglin.
The three-dimensional structure of the extracellular region of endoglin at a resolution of 25 Å was determined using single-particle electron microscopy [Llorca et al., 2007]. The molecular reconstruction suggests that endoglin exists as a dome comprised of antiparallel-oriented monomers enclosing a cavity at one end. Using these data, a high-resolution structure of endoglin indicates that each endoglin subunit comprises three well-defined domains, including the two ZP regions and one orphan domain, which are organized into an open U-shaped monomer [Llorca et al., 2007] (Fig. 1). These studies were performed by using a soluble form of the extracellular domain of endoglin. Of note, a soluble form of endoglin was recently detected in pregnant women with preeclampsia and it appears to play a pathogenic role in this disease [Levine et al., 2006; Venkatesha et al., 2006]. The metalloprotease MMP-MT1 was suggested to play a role in soluble endoglin production [Venkatesha et al., 2006]. Interestingly, a structural analysis of the extracellular region of endoglin identified a potential protease cleavage site that is highly conserved among different mammalian species and is located between the two subdomains of the ZP consensus region of endoglin [Llorca et al., 2007]. However, whether this soluble protein is generated by protease cleavage of the membrane bound endoglin or by an alternative splicing mechanism remains to be determined.
Endoglin and betaglycan bear a significant degree of sequence similarity [Ge and Butcher, 1994] and therefore, the search for functional attributes of endoglin has drawn upon results from the study of betaglycan. Betaglycan interacts with the TGFβ type II receptor [Lin and Lodish, 1993] and plays a role in the presentation of the TGFβ ligand to TβRII [Lopez-Casillas et al., 1993]. TGFβ binds to the N-terminal endoglin-related region of betaglycan, and mutational analysis suggests that the remainder of the extracellular and the cytosolic domains are not required for betaglycan-dependent enhancement of TGFβ binding to TβRII [Lopez Casillas et al., 1994].
Examination of the primary structure of betaglycan, especially in its cytosolic domain, indicated that this component of the TGFβ receptor system was a homolog of human endoglin [Lopez-Casillas et al., 1991]. Based on this finding, it was established that endoglin binds TGFβ1 and TGFβ3 but not TGFβ2 [Cheifetz et al., 1992]. This difference in affinity of endoglin for the TGFβ isoforms distinguishes it from betaglycan because betaglycan recognizes all three isoforms. These studies provided the basis for the examination of endoglin’s functions as a component of the TGFβ receptor system.
Because endoglin differs from betaglycan in its TGFβ ligand-binding profile [Cheifetz et al., 1992], it was not surprising to learn that functional differences, as well as similarities, exist between these two proteins. For example, both L- and S-endoglin isoforms bind TGFβ1 [Bellon et al., 1993], which is consistent with an exclusive role for the extracellular domain in TGFβ ligand binding. This view is supported by studies indicating that switching of the endoglin and betaglycan cytosolic domains has no effect on endoglin ligand binding [Letamendia et al., 1998]. However, in contrast to betaglycan, the binding of ligand to endoglin requires the presence of TβRII [Letamendia et al., 1998], suggesting that endoglin participates in ligand binding only within the TGFβ receptor complex. This result explains the observation that only a small fraction of the total cell surface endoglin binds ligand [Cheifetz et al., 1992].
Endoglin bound to ligand is isolated as a complex with the TGFβ type I receptor and the type II receptor, TβRII [Yamashita et al., 1994]. The TGFβ type I receptors include: ALK1, the bone morphogenetic protein (BMP) receptors ALK2, 3, and 6, as well as ALK5 and the activin receptors, ALK2 and ALK4. In addition to TβRII, the various type I receptors can interact with the activin (ActRII) or BMP type II receptors [Shi and Massague, 2003; Feng and Derynck, 2005]. In vitro co-immunoprecipitation studies of the interaction of endoglin with type I and type II receptors indicates that endoglin interacts with the ligands activin-A, BMP-7, and BMP-2 [Barbara et al., 1999]. These results are supported, at least for BMP-7, by functional experiments demonstrating that endoglin overexpression enhances the BMP-7/Smad1/Smad5 pathway, while inhibiting the TGF-β1-induced ALK-5/Smad3 signaling in myoblasts [Scherner et al., 2007]. As discussed above, these interactions require coexpression of the respective ligand-binding kinase receptor [Letamendia et al., 1998; Barbara et al., 1999]. Thus, endoglin binds TGFβ1 and β3 by associating with TβRII, and interacts with activin-A and BMP-7 in association with the ActRII receptors ActRIIA and ActRIIB. In addition, endoglin binds BMP-2 by interacting with the BMP ligand-binding receptors ALK3 and ALK6 [Barbara et al., 1999]. Interestingly, BMP-9 binds with high affinity to endoglin without the TGF-β signaling receptors [Scharpfenecker et al., 2007]. In agreement with this finding, overexpression of endoglin increases the BMP-9 response, whereas silencing of both BMPRII and ActRIIA expressions completely abolishes it [David et al., 2007]. These studies indicate that endoglin complexes with most ligand-type I/II receptor complexes, potentially reflecting a role for endoglin in the dynamics of type I/II receptor interactions and their downstream signaling pathways, or a regulatory role for phosphorylated endoglin occurring because of receptor activation, or both.
Studies of the interaction of endoglin with ALK5 and TβRII indicate that both ALK5 and TβRII interact with the extracellular and cytosolic domains of endoglin. However, ALK5 interacts with the endoglin cytosolic domain only when the kinase domain is inactive. Upon association, ALK5 and TβRII phosphorylate the endoglin cytosolic domain; then ALK5, but not TβRII, dissociates from the complex [Guerrero-Esteo et al., 2002]. These data suggest the hypothesis that endoglin’s extracellular and cytosolic domains play distinct roles in receptor signaling that are downstream of ligand binding and receptor activation.
Endoglin modulates TGFβ-dependent cellular responses. In human monocytic U-937 cells, TGFβ1, but not TGFβ2 responses are abrogated in both L-and S-endoglin transfectants [Lastres et al., 1996]. In a variety of cell types, including myoblasts, the TGFβ1-dependent responses opposed by endoglin include inhibition of cellular proliferation, cellular adhesion, platelet/endothelial cell adhesion molecule 1 phosphorylation, homotypic cell aggregation, and the increased expression of extracellular matrix components, including collagen and fibronectin [Lastres et al., 1996; Letamendia et al., 1998; Guerrero-Esteo et al., 1999; Diez-Marques et al., 2002; Obreo et al., 2004], and the secreted extracellular matrix-associated protein lumican [Botella et al., 2004]. Interestingly, no changes in total ligand binding were observed in L-endoglin transfectants [Lastres et al., 1996], suggesting that endoglin’s effects occur downstream of ligand binding. As with TGFβ receptor signaling in general, endoglin-dependent regulatory effects are likely to be cell type specific, subject to conditions that include the specific TGFβ type I receptors that are present and the relative levels of endoglin isoform expression.
Although TGFβ is a potent inhibitor of cell proliferation, endoglin expression counteracts this inhibitory effect in several cell types, including endothelial cells [Lastres et al., 1996; Li et al., 2000]. The positive correlation between endoglin expression and endothelial cell proliferation was confirmed in several experimental models. Thus, endoglin is markedly up-regulated in the proliferating endothelium of tissues undergoing angiogenesis [Burrows et al., 1995; Kumar et al., 1996, 1999; Bodey et al., 1998; Fonsatti et al., 2003], and in vitro inhibition of its expression on endothelial cells impairs this process [Li et al., 2000]. In addition, suppression of endoglin not only increases the TGFβ1-dependent inhibition of endothelial cell proliferation, but also endothelial cell apoptosis induced by hypoxia and TGFβ1 [Li et al., 2003]. Furthermore, using mice bearing targeted endoglin (eng) alleles, studies of derived eng−/− and eng+/− embryonic endothelial cells indicate that endoglin promotes endothelial cell proliferation via a TGFβ/ALK1 pathway [Lebrin et al., 2004]. An exception to this widely reported correlation between endoglin and endothelial cell proliferation is the finding that an endothelial cell line established from null eng−/− 8.5-day-old embryos are responsive to TGFβ and can proliferate faster than control mouse eng+/− endothelial cells [Pece-Barbara et al., 2005]. Future studies should clarify the detailed mechanism of endoglin-dependent effects on endothelial cell proliferation.
How endoglin regulates these TGFβ-dependent responses is unknown. A potential mechanism of action is via endoglin-dependent effects on TGFβ receptor phosphorylation. TβRII is thought to be a constitutively active (ca) receptor that activates the type I receptor via phosphorylation upon ligand-induced association. Betaglycan functions by selectively binding the phosphorylated TβRII via its cytosolic domain to promote TGFβ2 signaling [Blobe et al., 2001]. Interestingly, endoglin association with TβRII results in an altered phosphorylation state of TβRII and loss of ALK5 from the complex [Guerrero-Esteo et al., 2002], either of which could explain the inhibitory effects of endoglin on ALK5 signaling, which requires phosphorylation by the TβRII kinase after its association with TGFβ1. Additionally, studies in primary human umbilical vein endothelial cells suggest that endoglin phosphorylation opposes the activated ALK1-dependent inhibition of cell adhesion [Koleva et al., 2006]. These results suggest that by interacting through its extracellular and cytosolic domains with the signaling receptors, endoglin might affect TGFβ responses.
As endoglin directly interacts with a variety of TGFβ type I receptors [Barbara et al., 1999; Guerrero-Esteo et al., 2002; Blanco et al., 2005], this raises the possibility for additive or opposing effects of endoglin on TGFβ receptor signaling. Thus, although endoglin shows an inhibitory effect on TGFβ/ALK5/Smad3 cellular responses [Letamendia et al., 1998; Guo et al., 2004; Lebrin et al., 2004; Blanco et al., 2005; Scherner et al., 2007], it enhances ALK5/Smad2 signaling [Guerrero-Esteo et al., 2002; Carvalho et al., 2004; Santibanez et al., 2007]. In addition, endoglin may be required for TGFβ1/ALK1 signaling in some cell types, especially endothelial cells. This balance between ALK5 and ALK1 may play a role in the regulation of cell growth and differentiation in cells that express endoglin, as well as ALK1 and ALK5 [Lebrin et al., 2004]. The mechanism by which endoglin potentiates TGFβ/ALK1 signaling appears to involve direct association of ALK1 with the cytosolic and extracellular domains of endoglin, with the extracellular domain mediating the enhancement of ALK1 signaling [Blanco et al., 2005]. These studies suggest that the functional association of endoglin with ALK1 is critical for endothelial cell responses to TGFβ.
Recent studies indicate that endoglin regulates the levels of expression and the activities of proteins that mediate vascular tone. The vasoregulatory protein endothelial nitric oxide synthase (eNOS) is decreased in endoglin-deficient cells, whereas it is increased in endoglin-overexpressing cells [Jerkic et al., 2004; Toporsian et al., 2005]. At least in part, the endoglin-dependent increase of eNOS levels is mediated by increased stabilization of eNOS protein in caveolae, via a post-transcriptional mechanism that involves direct association of endoglin with caveolar proteins and potentially heat shock protein 90 [Toporsian et al., 2005]. In addition, endoglin stabilizes the Smad2 protein, potentially via reduction in the levels of the Smad ubiquitination response factor 2, Smurf2 [Santibanez et al., 2007]. Thus, in the presence of endoglin Smad2 protein levels are increased, leading to TGFβ receptor-dependent induction of eNOS mRNA, and enhancement of Smad-dependent signaling. Because of the endoglin-dependent regulation of eNOS, changes in nitric oxide levels lead to altered COX-2 expression, which is suggestive of a Smad-independent mechanism underlying endoglin function [Jerkic et al., 2006].
A schematic model of the modulatory role of endoglin in the TGFβ signaling pathways is depicted in Figure 2. Endoglin physically interacts and functionally modulates ALK1 and ALK5 signaling leading to the potentiation of Smad1 and Smad2 and inhibition of Smad3, which, in turn, regulates expression of Id1, eNOS, and plasminogen activator inhibitor-1 (PAI-1) genes, respectively. In the future, a complete identification of all the downstream genes affected by endoglin expression will be of interest, especially in HHT, in which endoglin haploinsufficiency is supposed to trigger the vascular lesions. In a step toward this goal, the gene expression fingerprinting of HHT endothelial cells revealed 277 down-regulated and 63 up-regulated genes that are potentially involved in biological processes relevant to the HHT pathology, including genes involved in angiogenesis, the cytoskeleton, cell migration, proliferation, and nitric oxide synthesis [Fernandez-L et al., 2007].
As noted, endoglin possesses properties of an adhesion molecule. This view was extended by studies indicating that endoglin expression results in the inhibition of cell migration in a variety of in vitro [Guerrero-Esteo et al., 1999; Liu et al., 2002; Conley et al., 2004] and in vivo [Ma et al., 2000] models. Efforts to address potential mechanisms underlying these properties of endoglin were based on the high degree of sequence conservation within the endoglin cytosolic domain and the lack of HHT-causing mutations in this domain. Yeast two-hybrid and cell biological approaches identified zyxin and zyxin-related protein 1 (ZRP-1) as the first examples of cytosolic proteins that interact with endoglin’s cytosolic domain [Conley et al., 2004; Sanz-Rodriguez et al., 2004]. Because these interactions are localized within endoglin’s cytosolic domain, which contains the sites of serine and threonine phosphorylation [Koleva et al., 2006], these data suggest that the endoglin cytosolic domain is a site of protein–protein interactions that are regulated by phosphorylation.
Several studies have illustrated how endoglin–zyxin interactions influence cell migration. Expression of endoglin is associated with the inhibition of cell migration and redistribution of zyxin from sites of focal adhesion (FA). Expression of endoglin caused reduction in zyxin associated with an integrin-rich FA-associated protein fraction obtained using RGD-tagged magnetic microspheres [Conley et al., 2004]. This reduction was correlated with: (1) inhibition of cell migration, (2) reduction of FA-associated p130(Cas)/Crk protein levels, and (3) that FA-associated endoglin levels were strongly mediated by endoglin’s cytosolic domain. It is noteworthy that the p130(Cas)/Crk interaction is required for the induction of cell migration [Klemke et al., 1998] and was implicated in vessel wall assembly [Foo et al., 2006].
Independently, it was discovered that endoglin also interacts with ZRP-1 [Sanz-Rodriguez et al., 2004]. Although zyxin and ZRP-1 share significant sequence homology, especially in the LIM3 domain, which contributes to endoglin binding [Conley et al., 2004], the amino terminal regions of zyxin and ZRP-1 are distinct. This distinction may underlie the different responses observed because of the interaction of endoglin with ZRP-1, which include the redistribution of ZRP-1 from sites of FA to F-actin stress fibers in endothelial cells, and dynamic rearrangement of F-actin fibers [Sanz-Rodriguez et al., 2004].
The interaction between endoglin and the ZRPs is exclusive because the interaction was not observed with betaglycan [Conley et al., 2004; Sanz-Rodriguez et al., 2004], even though their cytosolic domains are 70% identical. However, in addition to endoglin-specific protein–protein interactions, endoglin associates with proteins that also interact with beta-glycan. For example, beta-arrestin2 interacts with the conserved distal end of the betaglycan cytosolic domain and regulates betaglycan internalization [Chen et al., 2003]. This interaction also occurs with the endoglin cytosolic domain and results in endoglin internalization with beta-arrestin2 in endocytic vesicles [Lee and Blobe, 2007]. Endoglin’s cytosolic domain also interacts with a member of the Tctex1/2 family of cytosolic dynein light chains, Tctex2b, linking endoglin to the microtubule-based transport machinery [Meng et al., 2006]. Interestingly, Tctex1 is phosphorylated by the BMP type RII receptor, BMPRII [Machado et al., 2003] further supporting a functional linkage between Tctex proteins, endoglin, and TGFβ receptor complexes. Together, these studies point to a critical role for diverse protein–protein interactions involving the endoglin cytosolic domain in endoglin function (Fig. 3).
The importance of endoglin’s cytosolic domain in cell adhesion was corroborated by Muenzner and colleagues, who showed that endoglin expression mediated an increase in cell adhesion that was dependent on an intact cytosolic domain as well as the expression of integrin β1 [Muenzner et al., 2005]. These results further implicate endoglin in the regulation of integrin-mediated cell adhesion and detachment.
An interesting observation suggesting conservation of the endoglin–LIM domain interaction comes from the study of the Drosophila protein, piopio (Pio). Pio is an apically secreted extracellular matrix protein that has an important role in the regulation of tracheal tube growth. As with mammalian endoglin, Pio possesses an extracellular ZP domain, and a C-terminal sequence whose closest mammalian homolog is endoglin [Jazwinska et al., 2003]. Interestingly, other genes that mimic Pio-mutant phenotypes in Drosophila include steamer duck (stdk) [Prout et al., 1997]. Stdk, whose mammalian homolog is Pinch, is an evolutionarily conserved LIM-domain protein that is postulated to act as part of an integrin-dependent signaling complex that colocalizes to sites of actin filament anchorage in both muscle and wing epithelial cells. Thus, interactions involving Pio and Stdk may be functionally analogous to endoglin and Lim-domain proteins. Future studies are needed to clarify the evolutionary origins of endoglin and betaglycan and their overlapping and distinct networks of interactions.
Endoglin phosphorylation is a potential Smad-independent mechanism of endoglin function that regulates Smad-independent effects on endothelial cell growth and adhesion [Koleva et al., 2006]. Endoglin phosphorylation influences its subcellular localization [Koleva et al., 2006], potentially by modulating endoglin’s interactions with adhesive proteins such as zyxin and ZRP-1, and thus modifying the adhesive properties of endoglin-expressing cells.
The regulation and pattern of endoglin phosphorylation by the TGFβ receptors is complex. Koleva et al.  conducted a detailed study of endoglin phosphorylation by ca forms of the TGFβ receptors caALK1, caALK5, and wild type TβRII. Site-directed mutagenesis of endoglin suggests that caALK5 and TβRII phosphorylate the 634SerSer635 motif within endoglin’s cytosolic domain. In contrast to serine phosphorylation, ALK1 phosphorylates wild type endoglin preferentially on threonine residues. Interestingly, mutation of the 634SerSer635 residues to 634AlaAla635 strongly reduces threonine phosphorylation of endoglin, suggesting that phosphorylation of 634SerSer635 is a prerequisite for subsequent endoglin threonine phosphorylation. This hypothesis was verified by replacement of one mutated alanine with a phospho-mimicking aspartate residue (634Asp Ala635), which restores threonine phosphorylation by caALK1.
Studies of additional endoglin site-specific mutations are also informative. For example, removal of endoglin’s putative C-terminal PDZ-binding motif results in endoglin hyper-phosphorylation of distal threonine residues [Koleva et al., 2006]. These data reveal that receptor-mediated phosphorylation of endoglin is a complex process involving negative regulation by the PDZ-binding motif and an unexpected sequential mechanism of serine and threonine phosphorylation. Future studies will be needed to gain a comprehensive understanding of the full range of functions that are mediated by endoglin phosphorylation.
Involvement of endoglin in alternative Smad-independent TGFβ signaling pathways is further supported by the phenotypic similarities between the eng−/− and TGFβ-activated kinase-1 (TAK1)−/− developing mouse embryos [Jadrich et al., 2006]. TAK1 is a noncanonical Smad-independent effector of TGFβ and BMP signaling. Similar to the eng−/− mouse, smooth muscle cell development is impaired with normal endothelial cell development in the TAK1−/− mouse [Jadrich et al., 2006], thereby raising the possibility that TAK1 may mediate Smad-independent signals downstream of endoglin. Consistent with this idea, genetic data obtained combining Smad4 conditional inactivation with endoglin overexpression in cells of the embryonic neural crest suggest that endoglin operates in pathways that are separate from the canonical TGFβ receptor signaling pathways required for smooth muscle cell fate determination [Mancini et al., 2007]. The aforementioned studies suggest that endoglin modulates multiple interactions between TGFβ Smad-dependent and -independent signaling pathways.
Although endoglin’s expression was originally described as endothelial cell-restricted, it was later detected in the endocardium at 4 weeks of gestation and in the endocardial cushion mesenchyme by 5–8 weeks of gestation, suggesting a role in cardiac septation and valve formation [Qu et al., 1998]. Endoglin-targeted embryos die by E11.5 [Bourdeau et al., 1999; Arthur et al., 2000] due to defects in angiogenesis and cardiac morphogenesis, resulting in septation defects, thereby suggesting a loss of endocardial to mesenchymal transitions and a possible absence of vascular smooth muscle cells [Li et al., 1999]. However, it is unclear whether loss of endoglin results in a delay or loss of smooth muscle cell specification, differentiation, or both.
Endoglin is expressed on injured and atheromatous, but not in normal vascular smooth muscle [Adam et al., 1998; Conley et al., 2000; Ma et al., 2000], suggesting that endoglin plays a functional role in myofibroblast or pericyte responses to injury, and implicating a role for endoglin in vascular precursor cell physiology. This view is further supported by studies indicating that endoglin is expressed by circulating mesenchymal stem cells [Barry et al., 1999] and is a functional marker of long-term repopulating hematopoietic stem cells [Chen et al., 2002].
Evidence suggests that endoglin plays a role in myogenic specification during development. Because many of the smooth muscle cells that invest the large vessels and form the cardiac cushions are derived from the neural crest [Jiang et al., 2000], Mancini et al.  examined whether endoglin plays a role in specification from the neural crest. These studies show that endoglin is required for the maintenance of neural crest stem cell myogenic potential. Moreover, expression of endoglin in neural crest stem cells declines with age, coinciding with a reduction in both smooth muscle differentiation potential and TGFβ1 responsiveness.
Endoglin also plays a role in bone marrow mesenchymal stem cell regulation [Yamada et al., 2007], and in the regulation of the epithelial-mesenchymal transformation during cardiac valve formation [Mercado-Pimentel et al., 2007]. In addition, endoglin affects the efficiency of formation of the hemangioblast, a common embryonic progenitor of the hematopoietic and endothelial lineages [Perlingeiro, 2007]. Finally, supporting the relevance of endoglin-expressing circulating precursors, it was reported that endoglin has a crucial role in blood mononuclear cell-mediated vascular repair [van Laake et al., 2006]. Together, these studies support the hypothesis that endoglin expression is required for multiple cell precursors to begin tissue formation, respond to injury, and suggest that age-dependent loss of endoglin underlies an impaired response to vascular injury.
These reports point to novel and important emerging roles for endoglin in the differentiation and determination of the differentiation fate of vascular precursor cells. Thus, endoglin may participate in the integration of diverse TGFβ signals, and may directly mediate important cell-adhesive, proliferative, and migration processes in the developing and adult vasculature. Although a biochemical basis exists for understanding endoglin’s diverse effects at the cellular level, much work remains to better understand the role of endoglin in vascular development and disease.
The authors acknowledge the contributions of many investigators that, although relevant to this subject of this review, could not be included due to space limitations. C.P.H.V. was supported by NIH grants RR15555 from the COBRE program of the National Center for Research Resources, R01-HL083151 from the NIH National Heart, Lung and Blood Institute, and the Maine Cancer Foundation. The research work of C.B. was supported by grants from the Ministerio de Educación y Ciencia (SAF2004-01390) and the Instituto de Salud Carlos III (ISCIII-CIBER CB/06/07/0038) of Spain.
Grant sponsor: National Center for Research Resources, NIH; Grant number: RR15555; Grant sponsor: NIH National Heart, Lung and Blood Institute; Grant number: R01-HL083151; Grant sponsor: Maine Cancer Foundation; Grant sponsor: Ministerio de Educación y Ciencia; Grant number: SAF2004-01390; Grant sponsor: Instituto de Salud Carlos III; Grant number: ISCIII-CIBER CB/06/07/0038.