Cell Physiology

VE-cadherin: adhesion at arm's length

Peter A. Vincent, Kanyan Xiao, Kathleen M. Buckley, Andrew P. Kowalczyk


VE-cadherin was first identified in the early 1990s and quickly emerged as an important endothelial cell adhesion molecule. The past decade of research has revealed key roles for VE-cadherin in vascular permeability and in the morphogenic events associated with vascular remodeling. The details of how VE-cadherin functions in adhesion became apparent with structure-function analysis of the cadherin extracellular domain and with the identification of the catenins, a series of cytoplasmic proteins that bind to the cadherin tail and mediate interactions between cadherins and the cytoskeleton. Whereas early work focused on the armadillo family proteins β-catenin and plakoglobin, more recent investigations have identified p120-catenin (p120ctn) and a related group of armadillo family members as key binding partners for the cadherin tail. Furthermore, a series of new studies indicate a key role for p120ctn in regulating cadherin membrane trafficking in mammalian cells. These recent studies place p120ctn at the hub of a cadherin-catenin regulatory mechanism that controls cadherin plasma membrane levels in cells of both epithelial and endothelial origin.

  • endothelial cell
  • cytoskeleton
  • β-catenin
  • p120ctn
  • cell adhesion
  • vascular endothelial cadherin

ve-cadherin has emerged as an adhesion molecule that plays fundamental roles in microvascular permeability and in the morphogenic and proliferative events associated with angiogenesis (29, 30, 106). Like other cadherins, VE-cadherin mediates calcium-dependent, homophilic adhesion and functions as a plasma membrane attachment site for the cytoskeleton (8, 81). However, VE-cadherin is integrated into signaling pathways and cellular systems uniquely important to the vascular endothelium. Recent advances in endothelial cell biology and physiology reveal properties of VE-cadherin that may be unique among members of the cadherin family of adhesion molecules. For these reasons, VE-cadherin represents a cadherin that is both prototypical of the cadherin family and yet unique in function and physiological relevance. A number of excellent reviews have addressed the contributions of VE-cadherin to vascular barrier function, angiogenesis, and cardiovascular physiology (29, 30, 69, 105). Because of an evolving understanding of the molecular regulation of VE-cadherin, this review will focus on studies that have examined the cytoplasmic interactions that couple VE-cadherin to cytoskeletal binding partners. In addition, emphasis will be placed on emerging information indicating that cytoplasmic binding partners for VE-cadherin regulate cadherin cell surface levels by modulating VE-cadherin endocytosis and degradation.


VE-cadherin was first identified in 1991 by Suzuki, Sano, and Tanihara (107), who cloned the cDNAs for eight novel cadherin family members, one of which (clone 5) could be found in endothelial cells. In 1992, Lampugnani et al. (62) produced a monoclonal antibody (7B4) that labeled a protein localizing to intercellular boundaries of cultured endothelial cells. The 7B4 antigen was absent in fibroblasts, smooth muscle cells, and keratinocytes. Characterization of the 7B4 antibody demonstrated localization of the antigen to intercellular boundaries of endothelial cells in blood and lymphatic vessels in a number of tissues. The 7B4 monoclonal antibody immunoprecipitated a 140-kDa protein that, upon amino-terminal sequencing, revealed identity to clone 5 (cadherin 5) identified by Suzuki et al. (107). Expression of the full-length clone of cadherin-5 in Chinese hamster ovary (CHO) cells produced a 140-kDa protein that localized to cell-cell boundaries, mediated Ca2+-dependent, homophilic adhesion (aggregation), and decreased permeability to high-molecular-weight molecules across CHO cell monolayers (15). Sequence comparisons revealed a number of structural features in cadherin-5 common to other cadherin family members, including the presence of five cadherin-like repeats in the extracellular domain, a single-pass transmembrane region, and a well-conserved cytoplasmic tail (62, 108) (Fig. 1). On the basis of these structural similarities to the cadherin family and because of its selective expression in endothelial cells, cadherin-5 was given the name vascular endothelial cadherin (VE-cadherin) (15).

Fig. 1.

Schematic of vascular endothelial (VE)-cadherin domain organization and models for cadherin-based adhesion. A: representation of type I/II cadherins depicting 5 cadherin repeats in the extracellular domain. Note the amino-terminal tryptophan (Trp-2) and the cell adhesion recognition (CAR) site within the EC1 domain and the position of Ca2+ binding sites between each cadherin repeat. The cytoplasmic domain of VE-cadherin includes the “juxtamembrane region” that binds p120-catenin (p120ctn) and the “catenin binding domain” that interacts with β-catenin and plakoglobin. B: model of cadherin interactions leading to homophilic binding and cell-cell adhesion. Cadherins form cis dimers by the lateral binding of cadherins within the plasma membrane of a single cell. Trans dimers are formed by the binding of cadherins between adjacent cells. Electron tomography and crystallography suggest that 3 recurring structures termed S, W, and λ result from the binding of the EC1 domains, whereas surface force measurements suggest that overlapping interactions may also contribute to cadherin-mediated cell-cell adhesion. See text for additional details on cadherin interactions.

Recently, Nollet et al. (81) divided the cadherins into five subfamilies based on domain structure, genomic organization, and phylogenetic analysis. The four major groups include the classical/type I cadherins, the atypical/type II cadherins, and the desmogleins and desmocollins, which form the desmosomal cadherin subfamilies. The fifth group contains only one member, BS-cadherin. The type I and type II subfamilies contain the cadherin family members frequently found in adherens junctions. VE-cadherin can be grouped with the type II cadherins on the basis of its genomic structure (81). For example, the precursor region and the cytoplasmic tail of the atypical/type II cadherins are encoded by a single exon, whereas these regions in other cadherins contain at least one or more introns. The type II cadherins, including VE-cadherin, also contain two extra introns in the sequence coding for the extracellular region that are not found in other cadherin families (81). Sequence similarities within the first cadherin repeat (EC1) have also been used to classify cadherins. Although VE-cadherin is classified as a type II cadherin, it only contains 58% homology with the EC1 domain of cadherin 11, the prototypical type II cadherin (81). Type I cadherins all share a triple amino acid sequence, HAV, in the EC1 repeat, termed the cell adhesion recognition (CAR) sequence, that is required for cell-cell adhesion (11, 12) (Fig. 1). The HAV sequence has been postulated to interact with an amino-terminal tryptophan (Trp-2) residue also found in the EC1 domain (11, 65). The amino acid sequence surrounding the Trp-2 residue is highly conserved within the type I cadherin subfamily. The type II cadherins and the desmosomal cadherins also have a postulated CAR site that contains a central alanine, but the surrounding residues are different for each subfamily. Interestingly, VE-cadherin is one of only two cadherins lacking a central alanine in the region corresponding to the CAR site of other cadherins, although VE-cadherin does possess a tryptophan in position 2 of the EC1 domain. Even though VE-cadherin has been classified as an atypical/type II cadherin, phylogenetic analysis further suggests that it is a distant relative to members of this subfamily (81), underscoring the unique nature of this endothelium-specific adhesion molecule.

The extracellular domains of the cadherins mediate homophilic binding and adhesion between adjacent cells. The VE-cadherin extracellular domain consists of five cadherin-like repeats that form a rigid, rodlike structure that is stabilized by the binding of calcium ions to the intervening sequences located at the base of each domain (Fig. 1). Cadherins are thought to associate via both cis and trans interactions. Lateral cis dimerization is the binding of two adjacent cadherins within the plane of the plasma membrane of one cell. Although the mechanism of cis interaction of cadherins is controversial, crystallography, mutagenesis, and in vitro binding studies all support cis dimerization as being required for cadherin-mediated cell-cell adhesion (58, 65). Trans dimerization is the interaction between cadherins on adjacent cells. Early models of cadherin-based adhesive interactions described the formation of a zipperlike structure formed through the trans interdigitation of cis dimers on the membranes of opposing cells (100). More recent studies have proposed three models for how the interaction of the EC1 domain contributes to the formation of trans dimers between cadherins on adjacent cells (Fig. 1). Three trans interactions, termed S, W, and λ, were predicted from electron tomography of desmosomes (45) by modeling the desmosomal cadherins to the crystal structure of C-cadherin (13). Other studies using electron microscopy and surface force measurements have suggested a model in which there is greater overlap between cadherins on opposing cells, such as that illustrated in Fig. 1 where the EC1 domain interacts with the EC4 or EC5 domain (22, 65, 102, 103). Recent structural studies suggest that VE-cadherin homophilic binding occurs in a manner similar to that of classical cadherins, although the precise mechanism has not been defined (3, 9). Nonetheless, the importance of the VE-cadherin extracellular domain in mediating endothelial cell function has been demonstrated by using antibodies to the VE-cadherin EC1 domain. Antibodies directed against EC1 cause a decrease in the barrier function of endothelial cell monolayers in vitro and of blood vessels in vivo. In addition, antibodies to the EC1 domain also disrupt angiogenesis (24, 25). However, antibodies to the fourth cadherin repeat are able to prevent angiogenesis without causing changes in vascular permeability (25). These findings highlight the importance of understanding the structural basis for VE-cadherin-mediated cell adhesion as a foundation for the development of therapeutic interventions designed to selectively control vascular permeability and angiogenesis.


Catenins and their binding partners. Early studies of E-cadherin revealed that the cytoplasmic tail of the classical cadherins interacts with proteins that couple the cadherins to the actin cytoskeleton (85, 86). These proteins were originally termed α-, β- and γ-catenin. The importance of these cadherin binding partners in regulating cadherin function became apparent with the realization that the catenins act as key regulators of cadherin-mediated adhesion. In a variety of cellular contexts, cadherin deletion mutants lacking the ability to bind to catenins exhibited compromised ability to mediate tight cell-cell adhesion (77, 79, 87). Cloning, sequencing, and subsequent structural analysis revealed that α-catenin was closely related to vinculin and played a key role in coupling the cadherin tail to the actin cytoskeleton (46, 78). β-Catenin was found to function as a bridge between the cadherin tail and α-catenin (2), whereas γ-catenin was found to be identical to the desmosomal protein plakoglobin and to have some overlapping functions with β-catenin (57, 71, 97). Like β-catenin, plakoglobin is closely related to the Drosophila protein armadillo, and these junctional components were among the founding members of a family of proteins defined by a 42-amino acid repeated motif termed an armadillo domain (9193) (Fig. 2). Interestingly, plakoglobin not only assembles into desmosomes but also binds to classical cadherins and assembles into adherens junctions (97). Because of this dual targeting of plakoglobin to both actin and intermediate filament-based junctions in epithelial cells, plakoglobin is thought to play a key role in the cross talk between adherens junctions and desmosomes and likely participates in the coordinated assembly of these discrete adhesive complexes in a variety of epithelial tissues (26, 27, 66).

Fig. 2.

Armadillo family proteins. Two major subfamilies of armadillo proteins assemble into intercellular junctions in endothelial cells and other cell types. Plakoglobin and β-catenin represent one subfamily, along with the Drosophila protein armadillo (not shown). The p120ctn subfamily includes p120ctn-related molecules as well as the plakophilins, which localize predominantly to desmosomes. p0071 is found in both actin-associated adherens junctions and desmosomes and is also referred to as plakophilin-4. Each box represents an armadillo (arm) domain. See text for additional details. δ-cat, δ-Catenin.

The role of both β-catenin and plakoglobin in the assembly of intercellular junctions hinges on their ability to bind to both the cadherin tail and to downstream linking proteins such as α-catenin, which couples the cadherin-catenin complex to the actin cytoskeleton. Structural studies of β-catenin revealed that the arm repeats are tightly packed and form a superhelix of helices, resulting in a compact structure that forms a binding interface for several interacting partners, including cadherins (48). β-Catenin and plakoglobin both bind to the distal region of the cadherin cytoplasmic tail in a highly conserved region termed the catenin binding domain (Fig. 1). Analysis of the cadherin-β-catenin complex revealed that the cadherin tail is unstructured in the absence of β-catenin and is sensitive to proteolytic degradation (49, 50). Whereas the arm domain of β-catenin binds to the cadherin tail, the amino-terminal region of β-catenin binds to α-catenin (2). It is thought that plakoglobin interacts with the cadherin tail and with α-catenin in a manner similar to β-catenin (97). By interacting with both the cadherin tail and actin binding proteins such as α-catenin, β-catenin and plakoglobin play key roles in linking the cadherin to the actin cytoskeleton.

In addition to β-catenin and plakoglobin, a number of other armadillo family proteins have been localized to intercellular junctions in both epithelial and endothelial cells. Whereas plakoglobin and β-catenin are structurally homologous, a second subfamily of arm proteins became apparent with the identification of p120-catenin (p120ctn) and a series of related molecules (7, 42) (Fig. 2). p120ctn was originally identified as a Src substrate and was subsequently found to interact with the juxtamembrane domain (Fig. 1) of the classical cadherins (96, 109, 117). Unlike β-catenin and plakoglobin, p120ctn does not appear to directly link cadherins to the cytoskeleton but, rather, appears to play a regulatory role in cadherin function. p120ctn and the juxtamembrane domain of the cadherin tail are likely to regulate cadherin clustering and the local organization of actin during the assembly of cadherin-based adhesive contacts (109, 117). In addition to p120ctn, this subfamily of armadillo proteins includes ARVCF, δ-catenin, and p0071 (also known as plakophilin-4) (42). Very little is known about the function of these p120/plakophilin family proteins in vascular endothelial cells. p0071 is expressed in endothelial cells and binds to both VE-cadherin and desmoplakin (18), although the function of p0071 in endothelial barrier function and angiogenesis has not yet been explored. A related group of proteins termed the plakophilins assemble into desmosomes (40), but the expression patterns of the plakophilins in vascular cells have not been determined.

Role of catenins in junction assembly and vascular organization. Endothelial intercellular junctions appear to be unique in that VE-cadherin associates with both actin and intermediate filament binding proteins (Fig. 3). In epithelial cells, E-cadherin associates with actin and assembles into adherens junctions, whereas the desmosomal cadherins interact with intermediate filament binding proteins and nucleate the assembly of desmosomes. Franke and colleagues (98, 99) defined a new group of adhesive intercellular junctions found in endothelial cells that were termed “complexus adhaerentes.” These junctions were found to contain the desmosomal protein desmoplakin but do not appear to contain desmosomal cadherins. In a series of subsequent studies, desmoplakin emerged as a key component of vascular endothelial junctions. Valiron et al. (112) demonstrated that desmoplakin was expressed in cultured endothelial cells, and subsequent studies indicated that plakoglobin plays a key role in recruiting desmoplakin to VE-cadherin-based adhesive contacts (61). More recently, we have found that p0071, an armadillo family protein related to p120ctn and to the plakophilins, also links VE-cadherin to desmoplakin (18). Inactivation of the desmoplakin gene in mice resulted in early embryonic lethality (36). However, the importance of VE-cadherin linkages to the vimentin intermediate filament cytoskeleton became apparent when Gallicano et al. (35) utilized chimeric embryos to reveal desmoplakin functions in nonepithelial tissues. These elegant studies demonstrated that desmoplakin null endothelial cells failed to organize properly, resulting in decreased capillary density. More recently, the Gallicano laboratory found that depletion of desmoplakin in endothelial cells with the use of a small interfering RNA (siRNA) approach inhibited endothelial tubule formation in cells cultured on Matrigel (Gallicano GI, personal communication). These results underscore the need to understand the detailed linkages between VE-cadherin and the vimentin intermediate filament cytoskeletal network.

Fig. 3.

Schematic representations of protein interactions within endothelial intercellular junctions. VE-cadherin has been shown to engage in protein interactions leading to both actin and vimentin associations. p120 and p0071 both associate with the VE-cadherin juxtamembrane domain, whereas β-catenin and plakoglobin associate with the distal catenin binding domain. β-Catenin and p120 are thought to mediate assembly of actinbased adhesive structures, whereas plakoglobin and p0071 both associate with the intermediate filament binding protein desmoplakin. Note that the schematic depicts known interactions but is not intended to exclude other combinations of binding events that may occur with endothelial intercellular junctions.

A recent study by Cattelino et al. (20) demonstrated that the loss of β-catenin expression alters the balance between the recruitment of α-catenin and desmoplakin to endothelial intercellular junctions. In these studies, conditional β-catenin null mice were generated in which the β-catenin gene was ablated in endothelial cells by using a conditional Cre/LoxP system. Interestingly, endothelial cells lacking β-catenin exhibited a dramatic increase in desmoplakin expression. Whereas this result likely reflects a compensation mechanism responding to the absence of β-catenin, these findings also suggest that the balance between actin and vimentin associations with VE-cadherin in endothelial cells may be a dynamically regulated process. Consistent with this idea, Shasby et al. (101) reported that VE-cadherin associations with vimentin are lost upon exposure of endothelial cells to histamine. Collectively, these studies reveal an intricate meshwork of associations among VE-cadherin, actin, and vimentin cytoskeletal networks. Ongoing work promises to resolve how endothelial cells use these cytoskeletal linkages to differentially regulate responses to angiogenic and inflammatory signaling pathways.

Catenin function and signaling are integrated with growth factor signaling pathways. In the mid-1990s, a confluence of discoveries in both flies and vertebrate model systems revealed a remarkable signaling role for β-catenin in axis specification during development and in an inherited form of colon carcinogenesis in humans (16, 21, 39, 95). In addition to the cadherin bound form of β-catenin, the protein also translocates to the nucleus and modulates the expression of genes involved in tissue patterning and growth control (74). In the nucleus, β-catenin associates with the LEF/TCF family of transcriptional regulators (104). The ability of β-catenin to accumulate in the cytosol and to translocate to the nucleus is tightly regulated by a complex of proteins that controls β-catenin metabolic stability. Glycogen synthase kinase (GSK)-3β phosphorylates the amino-terminal domain of β-catenin (118) and targets the protein for turnover via the ubiquitin proteasome pathway (1). Under normal conditions, the non-cadherin-bound pool of β-catenin is rapidly degraded, but several signaling pathways modulate β-catenin turnover rates and thereby regulate β-catenin signaling. In the canonical pathway, Wnt family growth factors bind to cell surface receptors of the frizzled family, leading to the inactivation of GSK-3, a decrease in β-catenin turnover, and enhanced β-catenin signaling.

It remains unclear whether the Wnt-β-catenin signaling pathway plays a significant role in regulating endothelial cell proliferation or morphogenic events associated with vascular remodeling. However, several observations and a growing number of new studies strongly suggest that β-catenin signaling is integrated with control mechanisms that regulate endothelial organization and function. Transfection of endothelial cells with Wnt-1 causes increased endothelial cell proliferation (114), and a growing number of studies indicate that frizzled family proteins are expressed in vascular cells (31, 37). Importantly, Frzd5 was shown to be critical for placental and yolk sac angiogenesis (52). Separate studies demonstrated accumulation of cytoplasmic β-catenin in endothelial cells during neovascularization after myocardial infarction (10). Similarly, we found that expression of a stabilized form of β-catenin lacking the GSK-3 phosphorylation sites caused enhanced endothelial growth rates, suggesting a role for the Wnt/β-catenin signaling pathway in regulating endothelial cell growth and morphogenesis (113). Further evidence that the Wnt/β-catenin signaling pathway may be utilized as part of the angiogenic response comes from studies showing that LEF/TCF-dependent transcription is activated in endothelial cells exposed to FGF-2 (47). Additional work in this area is needed to understand how β-catenin signaling is integrated with other pathways that modulate angiogenic responses. Interestingly, the VE-cadherin cytoplasmic domain and β-catenin have been shown to play a critical role in the regulation of VEGF-induced proliferation in endothelial cells (19, 64). Collectively, these studies indicate the existence of a signaling axis centered around VE-cadherin and β-catenin that modulates endothelial proliferation and organization during angiogenesis (17).

p120ctn and the plakophilins. The role of the catenins in linking cadherins to the cytoskeleton is established as a vital structural role for cadherin binding proteins. This function in linking cadherins to the cytoskeleton has been established for β-catenin and plakoglobin, but the role of p120ctn and other members of the p120ctn subfamily of arm proteins has remained somewhat elusive. The identification of several adherens junction (ARVCF, δ-catenin, p0071) and desmosomal proteins (plakophilins) with structural homology to p120ctn (7, 42) opened up a new line of investigation into the role of these arm family proteins in junction assembly and regulation. At present, very little is known about the role of most of these proteins in cell adhesion, with the exception of p120ctn and plakophilin-1. The plakophilin-1 gene was found to be mutated in a human skin disorder characterized by ectodermal dysplasia and fragility (72). Patients with this disorder exhibit severe epidermal fragility and blistering. Subsequent molecular analysis revealed that plakophilin-1 participates in the recruitment and clustering of desmoplakin at desmosomes, thereby strengthening adhesion between keratinocytes (44, 60). At the present time, virtually nothing is known about the role of plakophilins in endothelial cell biology. δ-Catenin has been implicated in regulating cell migration (68), whereas p0071 may play a role in the balance between adherens junction and desmosome assembly in epithelial cells (43). Furthermore, p0071 may play a role in the recruitment of desmoplakin to endothelial intercellular junctions (18), underscoring the need to explore the functions of these proteins in vascular biology.

Although very little is known about the function of the plakophilins and p0071 in endothelial cells, several studies over the past few years have indicated that p120ctn plays an important regulatory role in cadherin-based adhesive contacts (7). p120ctn binds to a highly conserved sequence in the juxtamembrane domain of the classical cadherins (109). This domain is also highly conserved in VE-cadherin, and mutation of this domain of VE-cadherin eliminates both p120ctn and p0071 binding (Ref. 18; Calkins CC and Kowalczyk AP, unpublished observations). Furthermore, p120ctn and p0071 appear to compete for binding to this domain, much in the same manner as β-catenin and plakoglobin compete for binding to the catenin binding domain of the cadherin tail. An important and unresolved question is how do cells use this rather extensive repertoire of p120ctn/plakophilin family members to modulate the assembly state of cadherin-based intercellular junctions? Resolving this issue will require a more comprehensive characterization of plakophilin expression patterns and an understanding of whether the members of this family of proteins function cooperatively or antagonistically in junction assembly.

Previous studies indicated that the E-cadherin juxtamembrane domain regulates cadherin clustering (117) and plays a key role in locally modulating actin dynamics at the site of cadherin-based adhesive interactions (109). The VE-cadherin cytoplasmic domain was recently shown to activate Cdc42 (59) and Rac (63). Similarly, E-cadherin is known to activate phosphatidylinositol 3-kinase and Rac activity, and part of this signaling function of the cadherin tail requires p120ctn association with the cadherin juxtamembrane domain (38). Likewise, numerous studies have shown that cadherins are regulated by Rho family GTPases (14, 67), indicating a reciprocal regulation of cadherin-based adhesion and local actin dynamics at sites of intercellular junction formation. An important break-through was made with the discovery that p120ctn regulates the activity of Rho family GTPases. Nearly simultaneously, three laboratories published data indicating that p120ctn regulates the activity of Rho A, Rac, and Cdc42 (6, 41, 82). These findings revealed that the cadherin-p120 complex regulates Rho family GTPases, which in turn regulate junction assembly. However, the details of how this adhesion and cytoskeletal regulatory system is coordinated remain unclear. In particular, important questions remain with respect to how cadherin-based adhesion is integrated with the regulation of Rho family GTPase activity during cell migration and how these systems differ between endothelial and epithelial cell types.


Endocytosis and processing of VE-cadherin. The identification of the catenins and our growing understanding of cadherin-cytoskeletal linkages have provided important insights into the basic mechanisms of cell-cell adhesion. However, in the past decade, much less progress has been made in understanding the mechanisms driving intercellular junction disassembly, which is a hallmark feature of inflammatory, angiogenic, and tumorigenic responses. Recent studies indicate that endocytosis is an important cellular mechanism for the regulation of cadherin expression and presentation of cadherins at the plasma membrane. Growing evidence suggests that cadherin endocytosis occurs during development (73), in response to growth factors that trigger cell migration (54, 75), and in response to transforming oncogenes such as Src (34). In endothelial cells, VE-cadherin is downregulated in response to advanced glycation end products (84), during endothelial transdifferentiation (32), and in response to shear stress (83). Importantly, VE-cadherin was shown to be internalized in response to H2O2, which causes barrier dysfunction, and that inhibitors of protein kinase C modulate VE-cadherin internalization (5, 55). The small GTPases Rac1 (4) and ARF6 (89) also have been implicated in destabilizing epithelial adherens junctions by modulating endocytosis of adherens junction components. These observations suggest that endocytosis of cadherins may be used by a variety of cell types in response to signaling cues that modulate the adhesive potential of the cell surface.

Recent studies carried out by our laboratories demonstrated that VE-cadherin and the cadherin binding proteins plakoglobin and β-catenin regulate endothelial monolayer permeability and endothelial growth rates (113). In the course of these studies, we utilized dominant negative mutants of VE-cadherin to disrupt endothelial intercellular junctions. Dominant negative mutants of cadherins have been used extensively to investigate cadherin function, predominantly in epithelial cell model systems (33, 56, 119). Previous studies indicated that dominant negative cadherin mutants cause the downregulation of endogenous cadherins (80, 111), but the mechanism driving cadherin turnover was not known. We hypothesized that mutant cadherins might trigger endocytosis and degradation of endogenous cadherins. Consistent with previous studies in epithelial cells, we found that a cadherin mutant comprising the interleukin 2 receptor extracellular domain and the VE-cadherin cytoplasmic tail (IL-2R-VE-cadcyto) caused a dramatic decrease in endogenous cadherin levels (116). Immunofluorescence analysis and other experiments demonstrated that the mechanism by which cadherin mutants cause the downregulation of endogenous cadherins is by triggering endocytosis and lysosomal degradation of endogenous VE-cadherin (115, 116). Interestingly, chloroquine treatment, which inhibits lysosome acidification, resulted in the formation of a VE-cadherin fragment lacking the β-catenin binding domain. The subcellular compartment where this cleavage takes place is currently unknown, but the cadherin appears to be cleaved after internalization. One possibility outlined in Fig. 4 is that the removal of the β-catenin binding domain is a sorting signal that fates the cadherin for lysosomal degradation. Defining how VE-cadherin is recruited into the endocytic pathway and then sorted for recycling or degradation will be important for understanding how signaling pathways modulate endothelial cell-cell contact during angiogenesis and inflammation.

Fig. 4.

Processing of VE-cadherin by the endocytic pathway. VE-cadherin at the plasma membrane is thought to be associated with both β-catenin and p120ctn. The relative levels of VE-cadherin present in the cell are dictated by p120ctn levels. Disruption of p120ctn interactions with the VE-cadherin tail appears to trigger cadherin internalization and also may lead to disruption of β-catenin interactions with the cadherin tail during endocytosis. The loss of p120ctn binding to the cadherin tail appears to direct the cadherin to a degradative pathway involving lysosomes and perhaps also the proteasome (not shown). At some point after internalization, the VE-cadherin tail is cleaved, removing the β-catenin binding domain. This cleavage event may fate the cadherin for degradation. It is not known whether VE-cadherin enters a recycling pathway, but this possibility may involve p120ctn-kinesin interactions to deliver the cadherin back to the plasma membrane. MVB, multivesicular bodies.

p120ctn: a set point mechanism for cadherin expression levels. Growing evidence indicates that the catenins not only couple the cadherins to the cytoskeleton but also regulate cadherin-based adhesive interactions. Nonetheless, precisely how cadherins are regulated by their catenin binding partners remains somewhat of a mystery, and the function of p120ctn has been particularly elusive. As discussed below, recent evidence indicates that the core function of p120ctn is to regulate cadherin expression levels by controlling cadherin membrane trafficking.

The first study demonstrating a central role for p120ctn in regulating cadherin expression levels came from Ireton et al. (51) with the identification of an epithelial tumor cell line with mutations in the p120ctn gene. In this model system, the loss of p120ctn due to gene mutation resulted in the posttranslational loss of E-cadherin expression. These investigators demonstrated that p120ctn played a key role in maintaining E-cadherin metabolic stability. In the absence of p120ctn, E-cadherin exhibited rapid turnover, and E-cadherin metabolic stability and expression levels could be restored by replacing the mutated p120ctn gene with exogenously expressed p120ctn. Interestingly, the central armadillo domain of p120ctn was found to be necessary and sufficient to rescue E-cadherin, suggesting that p120ctn interactions with the cadherin juxtamembrane domain regulate cadherin expression levels by stabilizing cadherin turnover rates. These studies held important clues for understanding how cadherin expression levels might be altered in human tumors in which E-cadherin expression was lost even in the absence of mutations in the E-cadherin gene. Indeed, emerging evidence suggests that the loss of cadherin expression in some human tumors may well be a consequence of mutations in the p120ctn gene, resulting in increased E-cadherin turnover and decreased E-cadherin expression (110).

The question remained whether the loss of p120ctn in tumor cells, and the corresponding loss of cadherin expression, applied to normal cells or to cells of vascular origin. Importantly, the loss of p120ctn expression in Drosophila or mutation of the juxtamembrane domain of Drosophila cadherin does not cause loss of cadherin expression or embryonic lethality in flies (76, 88). These findings seemed to contradict the previous studies in mammalian cells (51) and suggested that the loss of p120ctn and E-cadherin expression in human tumor cells might be part of a more complicated regulatory system that is disrupted during epithelial tumorigenesis. To address these apparent discrepancies and to understand p120ctn regulation of cadherin function in vascular cells, Iyer et al. (53) utilized a soluble peptide corresponding to the juxtamembrane domain of VE-cadherin, as well as siRNA to decrease the levels of p120ctn, in pulmonary artery endothelial cells (53). The loss of p120ctn resulted in a dramatic loss of endothelial barrier function. Analysis of VE-cadherin protein levels demonstrated that the loss of p120ctn resulted in a corresponding loss of VE-cadherin expression, indicating a key role for p120ctn in the maintenance of vascular barrier function by modulating VE-cadherin accumulation. Similar results were found in primary cultures of human microvascular endothelial cells and in normal epithelium. Remarkably, the levels of both VE-cadherin (115) and E-cadherin (28) expression could be adjusted by manipulating expression levels or availability of cytoplasmic p120ctn. These studies clearly establish that p120ctn plays a major role in regulating cadherin levels in mammalian cells, leading to the suggestion that cellular levels of p120ctn function as a rheostat for cadherin expression levels (94). Furthermore, these findings reveal a reciprocal regulatory mechanism whereby β-catenin levels are regulated by cadherin availability, and cadherin expression levels are in turn regulated by p120ctn availability. These findings place p120ctn at the hub of a cadherin-catenin expression and signaling system.

p120ctn: a plasma membrane retention signal that regulates VE-cadherin internalization. Several recent studies indicated that p120ctn regulates the presentation of cadherins on the cell surface by controlling cadherin membrane trafficking. Two separate studies found that both E-cadherin (28) and VE-cadherin (115) are rapidly internalized from the cell surface and degraded in the absence of p120ctn. Interestingly, p120ctn does not appear to be required for proper delivery of newly synthesized E-cadherin to the plasma membrane. Davis et al. (28) utilized metabolic labeling of newly synthesized E-cadherin followed by cell surface biotinylation in an elegant series of experiments to demonstrate that E-cadherin is delivered to the membrane properly in the absence of p120ctn but is then rapidly internalized and degraded (28). E-cadherin degradation could be prevented by treating the epithelial cells with chloroquine to inhibit lysosomal degradation or by treating the cells with proteasome inhibitors. Similarly, proteasome inhibitors prevent the disruption of endothelial intercellular junctions and the downregulation of VE-cadherin in endothelial cells expressing VE-cadherin mutants (116). Because these inhibitors can deplete cellular levels of ubiquitin, it is likely that the inhibition of cadherin degradation in cells treated with proteasome inhibitors such as MG132 reflects a role for ubiquitination in cadherin internalization. Alternatively, it is possible that the proteasome plays a more direct role in cadherin catabolism by degrading the cadherin directly or, perhaps more likely, by controlling the accumulation of regulatory proteins involved in cadherin internalization.

Although the role of the ubiquitin-proteasome system in cadherin internalization remains unclear, it is likely that both E-cadherin and VE-cadherin are delivered to lysosomes for degradation in the absence of p120ctn. As mentioned above, the degradation of both VE-cadherin and E-cadherin is attenuated by treating cells with chloroquine. Furthermore, using antibodies to label cell surface VE-cadherin, we found that the internalized cadherin colocalizes with CD-63, a marker for lysosomes and late endosomal compartments (115). In vascular endothelial cells, the internalized pool of VE-cadherin did not colocalize with either p120ctn or β-catenin, leading to the hypothesis that the disruption of the cadherin-catenin complex leads to cadherin internalization and degradation (Fig. 4). Consistent with this hypothesis, expression of VE-cadherin mutants that compete with the endogenous cadherin for p120ctn interactions cause VE-cadherin internalization. Conversely, overexpression of p120ctn to saturate the VE-cadherin juxtamembrane domain with excess p120ctn blocks VE-cadherin internalization and delivery to lysosomes. Interestingly, manipulation of β-catenin availability did not substantially alter VE-cadherin plasma membrane levels, indicating that p120ctn is the key cadherin regulatory protein in terms of cadherin expression levels.

How does p120ctn regulate cadherin internalization? The cadherin juxtamembrane domain appears to play a key role in regulating cadherin turnover. Hakai, an E3 ubiquitin ligase, was found to bind to the juxtamembrane domain of E-cadherin, leading to E-cadherin ubiquitination, internalization, and degradation (34). The interaction of hakai with E-cadherin is regulated by Src kinase. Interestingly, two E-cadherin-specific tyrosine residues present in the E-cadherin cytoplasmic domain are necessary for hakai to interact with the cadherin tail. Furthermore, the two predominant endothelial cadherins, N-cadherin and VE-cadherin, both lack these residues necessary for hakai binding, suggesting that other mechanisms are involved in regulating endothelial cadherin expression levels. Nonetheless, because VE-cadherin internalization can be blocked by treating cells with MG132, these studies indicate that the ubiquitin-proteosome system is likely to play a role in regulating VE-cadherin internalization and degradation. Furthermore, it is likely that p120ctn blocks the interactions between hakai or hakai-like molecules and the cadherin juxtamembrane domain. In addition to p120ctn, both ARVCF and p0071 also block cadherin internalization and rescue cadherin expression levels when p120ctn is depleted by siRNA knockdown or by expression of cadherin mutants as competition for cytoplasmic p120ctn (28, 115). Because ARVCF, p0071, and p120ctn all bind to the cadherin juxtamembrane domain (18, 90, 109), it is likely that p120ctn functions as a “cap” to prevent interactions between the cadherin juxtamembrane domain and regulatory proteins that target the cadherin for internalization.

In addition to regulating cadherin internalization, Chen et al. (23) recently found that p120ctn also binds to kinesin, an outwardly directed motor protein involved in transport of proteins to the plasma membrane (23). Kinesin was found to associate with the non-armadillo head domain of p120ctn. Because the armadillo domain of p120ctn binds to cadherins, these data suggest that p120ctn functions as a link between the cadherin cytoplasmic domain and motor proteins involved in vesicular transport. Consistent with this possibility, the cadherin-p120ctn interaction appeared to accelerate the rate of transport of vesicular pools of cadherin from the cytoplasm to intercellular junctions. These findings raise the exciting possibility that the rates of both cadherin internalization and transport back to the plasma membrane are regulated by p120ctn. One implication of these results is that multiple points of regulation may be available for cellular control over cadherin accumulation at the membrane (94). Interestingly, the aminoterminal domain of p120ctn contains numerous sites for phosphorylation (70), raising the possibility that posttranslational modification of the p120ctn head domain might modulate the affinity of p120ctn-kinesin interactions and thereby regulate cadherin transport rates.

The findings that p120ctn regulates cadherin internalization and cadherin transport to the plasma membrane raise several important questions about how this catenin may regulate cadherin-mediated adhesion. As shown in Fig. 4, one possibility is that the major function of p120ctn is to control cadherin expression levels by regulating entry of plasma membrane-localized cadherin into the endocytic pathway. This model is based on the finding that VE-cadherin is internalized and degraded in the lysosome in the absence of p120ctn (115). A prediction of this hypothesis is that p120ctn dissociates from the cadherin, perhaps as a consequence of posttranslational changes to the cadherin tail or to p120ctn itself. The loss of p120ctn then leads to the recruitment of the cadherin into endocytic vesicles. The details of how this process takes place are currently unknown, and it is unclear whether VE-cadherin enters the endocytic pathway via clathrin-dependent or -independent mechanisms. A second possibility is that p120ctn regulates cadherin recycling back to the plasma membrane after the cadherin is internalized. This possibility arises from the interactions found between p120ctn and kinesin (23) and the enhanced transport of the cadherin to the plasma membrane after calcium switch in the presence of excess p120ctn. In this scenario, p120ctn might remain bound to the cadherin during internalization, and the decision to sort to lysosomes or recycle back to the plasma membrane would be made in early endosomal compartments. In the presence of p120ctn, the cadherin would recycle, whereas the cadherin would be routed to lysosomes in the absence of p120ctn. The studies reported to date do not distinguish between these possibilities, and in fact, these scenarios are not mutually exclusive if p120ctn can dissociate and then reassociate with the cadherin tail at various control points in the endocytic pathway. Future studies are required to address these issues and to define how this process is regulated during inflammation or angiogenesis.


During the past decade of research on cadherin-associated proteins, investigators have focused on how cadherins are coupled to the cytoskeleton. This area of investigation remains critical because the linkage of the cadherin tail to cytoskeletal components is thought to regulate cell shape, migration, and the strength of cadherin adhesive interactions. In particular, the mechanisms by which VE-cadherin and other cadherin family members interact with the intermediate filament cytoskeleton appear to be a critical area for investigation in both epithelial and endothelial cell model systems. The catenins, and particularly the p120ctn/plakophilin family of proteins, are likely to appear at the center of these investigations. Understanding how mammalian cells use their inventory of p120ctn/plakophilin proteins to assemble functionally distinct junctional complexes may be among the most important questions to address. Finally, research now emerging in the literature puts p120ctn, and perhaps other p120ctn family proteins, at the center of a sensing mechanism that controls cadherin expression levels in mammalian cells. These studies indicate that p120ctn is a critical regulator of cadherin trafficking in vascular and epithelial cell types. Understanding precisely how p120ctn regulates cadherin internalization and/or recycling will be key to understanding how cadherin expression levels might be modulated during angiogenesis, tumor formation, and epithelial-to-mesenchymal transitions during development.


We are grateful to Drs. Kathleen J. Green and Spiro Getsios for helpful comments on this review and to Drs. Al Reynolds and Alpha Yap for sharing unpublished reagents and insights.


This work was supported by National Institute of Arthritis and Musculoskeletal and Skin Diseases Grant R01-AR-048266, American Cancer Society Grant RPG CSM-100348, and American Heart Association Grant 0355293B (to A. P. Kowalczyk). K. Xiao was supported by a postdoctoral fellowship from the American Heart Association. P. Vincent was supported by National Heart, Lung, and Blood Institute Grant K02-HL-004332.


View Abstract