help button home button Endocrine Society Endocrine Reviews
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Copyright Permission
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Pötter, E.
Right arrow Articles by Brabant, G.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Pötter, E.
Right arrow Articles by Brabant, G.
Endocrine Reviews 20 (2): 207-239
Copyright © 1999 by The Endocrine Society

The Cadherin-Catenin System: Implications for Growth and Differentiation of Endocrine Tissues

Eyck Pötter, Clemens Bergwitz and Georg Brabant

Abteilung Klinische Endokrinologie, Zentrum Innere Medizin, Medizinische Hochschule Hannover, D-30625 Hannover, Germany


    Abstract
 Top
 Abstract
 I. Introduction
 II. Molecular Components and...
 III. Signal Transduction,...
 IV. Mouse Mutants
 V. Implications of the...
 VI. Summary and Conclusions
 References
 

I. Introduction
II. Molecular Components and Subcellular Organization of the Cadherin-Catenin System
A. Cadherins, a large family of cell adhesion proteins
B. Structure-function relationships of classic cadherins
C. Catenins and the cadherin-catenin complex
III. Signal Transduction, Hormonal Regulation, and the Cadherin-Catenin System
A. Cadherin-catenins as targets of regulation
B. Cadherin-catenins as signal transducers
IV. Mouse Mutants
V. Implications of the Cadherin-Catenin System for Tumor Development and Prognosis of Tumors in Endocrine Tissues
VI. Summary and Conclusions


    I. Introduction
 Top
 Abstract
 I. Introduction
 II. Molecular Components and...
 III. Signal Transduction,...
 IV. Mouse Mutants
 V. Implications of the...
 VI. Summary and Conclusions
 References
 
CELL adhesion is a fundamental process influencing the life of most cells and may be divided into cellular contacts with the extracellular matrix and with neighboring cells. It is essential for tissue organization during development and for the maintenance of tissue integrity in adult organisms. Physical interactions between cells are primarily mediated by four types of higher order structures at the plasma membrane: gap junctions, tight junctions (zonulae occludentes), desmosomes (maculae adhaerentes), and adherens junctions (zonulae adhaerentes), with the latter three being members of the characteristic tripartite junctional complexes observed early in epithelia of many glands (1). Gap junctions are composed of connexins and regulate the passage of small solutes between adjacent cells, allowing communication between their cytoplasms (for reviews see Refs. 2, 3, 4). Tight junctions with occludin (5) and claudins (6) as integral membrane proteins act as important apical barriers, both regulating paracellular permeability and separating apical and basolateral membrane regions, thereby creating polarity (for reviews see Refs. 7, 8, 9). Adherens junctions and desmosomes mediate cell-cell adhesion via members of the cadherin family of transmembrane proteins and their connections to the cortical actin-based cytoskeleton and the intermediate filament system, respectively, that are established by catenins, a group of cytoplasmic proteins (for reviews see Refs. 10, 11, 12, 13, 14, 15, 16).

Cadherins were first identified about 20 yr ago in independent studies addressing either cell-cell adhesion mechanisms or early morphogenetic events during development of mouse preimplantation embryos and embryonal carcinoma cells (17, 18, 19). During compaction, a Ca2+-dependent adhesion process at the eight-cell stage, the organization of the early morula drastically changes as a prerequisite for proper formation of the blastocyst. This process involves maximization of cell-cell contacts, appearance of tight and gap junctions, changes in cytoskeletal organization, and polarization of blastomeres (20, 21, 22, 23, 24). Antibodies interfering with the compaction process were used to purify a plasma membrane glycoprotein protein termed uvomorulin (19, 24), and similar immunochemical approaches identified liver cell adhesion molecule (L-CAM) in chicken liver cells (25), Arc-1 in dog kidney cells (26, 27), and cell-CAM 120/80 in human epithelial cell lines (28). These molecules appeared to be species homologs of the classic epithelial-type cadherin, now generally referred to as E-cadherin. Soon after, neural N-cadherin and placental P-cadherin were identified to form with E-cadherin the first members of the cadherin family (29). Characteristically, these molecules display their adhesive properties and are protected against proteolytic attack in the presence of Ca2+ ions (17, 24). Currently, more than 30 human members of the cadherin family are known, some of them with unknown functions. Extensive studies of classic cadherins and their associated cytoskeletal anchor proteins, the catenins, gave important insights into structure and function of adhesive cadherin-catenin complexes (Section II) and, furthermore, revealed that cadherins and catenins participate in signal transduction either as targets of regulation by a variety of mechanisms or as signal transducers by themselves (Section III). Genetically manipulated mouse models further elucidated the dual role of the cadherin-catenin system in cell adhesion and signaling during embryogenesis and tissue morphogenesis (Section IV). Loss of E-cadherin-mediated cell adhesion contributes to tumor invasion, metastasis formation, and cancer progression in many malignancies, including those of endocrine tissues and their target organs, and documents the role of E-cadherin as both a tumor suppressor and an invasion suppressor. In addition, recent progress in this rapidly expanding field indicates that alterations in the cadherin-catenin system related to catenin signaling may be causally involved in tumorigenesis (Section V). Understanding how the cadherin-catenin system participates in the multistep process of tumorigenesis will, undoubtedly, lead to the development of new molecular diagnostic tools and therapeutic strategies.


    II. Molecular Components and Subcellular Organization of the Cadherin-Catenin System
 Top
 Abstract
 I. Introduction
 II. Molecular Components and...
 III. Signal Transduction,...
 IV. Mouse Mutants
 V. Implications of the...
 VI. Summary and Conclusions
 References
 
A. Cadherins, a large family of cell adhesion proteins
Cadherins are transmembrane glycoproteins located in the plasma membrane of cells in most, if not all, solid tissues. Most of them mediate Ca2+-dependent cell-cell adhesion predominantly through homophilic interaction of their extracellular domains. Cadherins constitute an ever-growing family of proteins, and members were cloned in various species, including man, mouse, rat, chicken, frog, zebrafish, nematodes, and the fly. Based on amino acid sequence comparisons and structural features, cadherins can be subdivided in five groups, classic cadherins type I and II, desmosomal cadherins, protocadherins, and other, more distantly related cadherins (30, 31, 32, 33). Human cadherins cloned so far and some cadherins of other species expected to exist in humans are listed in Table 1Go.


View this table:
[in this window]
[in a new window]
 
Table 1. Human cadherin family members

 
The first cadherins identified by cDNA cloning are E-, P-, and N-cadherin, which are now generally referred to as classic cadherins type I together with R-cadherin (cadherin-4) (30, 34, 35, 36, 37, 38). These type I membrane proteins are composed of an amino-terminal extracellular region including five repeated subdomains, a single-pass membrane-spanning region, and a carboxy-terminal region located in the cytoplasm (for details see Section II.B and Fig. 1Go). Mouse M-cadherin, which was first identified in skeletal muscle cells and considered to be involved in myotube formation, and its putative human homolog cadherin-15 are related to this classic cadherin subfamily (39, 40).



View larger version (19K):
[in this window]
[in a new window]
 
Figure 1. Schematic representation of E-cadherin and its functional domains. The processing site of the E-cadherin precursor, the sequence motif HAV involved in specific homophilic binding, and a serine cluster in the catenin-binding region are indicated by arrows. Ca2+-binding sites are shown at domain repeat interfaces. Numbers indicate first and last amino acid positions in the human precursor protein sequence. The scheme was drawn according to Refs. 42, 116, 142, 143, 145–147, 149, 150, and 152. EC1 to EC5, cadherin repeats 1–5 in the extracellular region; S, signal sequence; Pro, propeptide; TM, transmembrane region; CD, cytoplasmic domain.

 
Type II cadherins, e.g., cadherin-5 to -12, which were initially detected in neuronal tissue by an RT-PCR-based approach, share the basic cadherin structure, but have a low overall amino acid homology with type I cadherins and, therefore, may have diverged early in the evolution of multicellular organisms. Type II cadherins lack specific amino acids and sequence motifs such as the cell adhesion sequence motif (HAV-motif; see Section II.2) found in the N-terminal part of type I cadherins (30, 41, 42).

Desmogleins and desmocollins, adhesive transmembrane glycoproteins of desmosomes, constitute the subgroup of desmosomal cadherins (43, 44, 45, 46, 47, 48). They are connected to the intermediate filament network of the cytoskeleton via associated cytoplasmic proteins such as the common plaque protein plakoglobin (49, 50) and several other cell type-specific proteins including desmoplakins and plakophilins (11). Three desmogleins (Dsg 1, 2, and 3) and three desmocollins (Dsc 1, 2, and 3) have been identified in humans; their genes locate to a single cluster on chromosome 18q21 (51, 52, 53, 54, 55, 56, 57, 58). Dsg 2 and Dsc 2 appear to be ubiquitously expressed, while type 1 and 3 desmosomal cadherins are restricted mainly to stratified squamous epithelia (59, 60, 61, 62, 63, 64). The extracellular regions of desmosomal cadherins contain cadherin-like repeats, but the composition of their cytoplasmic tails differs considerably from that of classic cadherins (11, 65). In addition, a variable number of a unique sequence repeat are present in the cytoplasmic domain of desmogleins (52, 66, 67). Differential pre-mRNA splicing gives rise to two isoforms of each desmocollin, whereas splice variants are not common for classic cadherins (65, 68). Desmosomes were best studied in the epidermis, where they exert crucial functions for differentiation and the maintenance of tissue integrity. In some acquired blistering skin diseases such as pemphigus foliaceus, pemphigus vulgaris, and IgA pemphigus, Dsg 1, Dsg 3, and Dsc 1, respectively, have been identified as targets of circulating autoantibodies causing desmosomal plaque disruption (52, 69, 70, 71, 72, 73, 74).

The protocadherin subfamily, initially identified by a PCR-based screen, includes Pcdh1, Pcdh2, and Pcdh7 (BH-cadherin) in human, Pcdh3 in rat, and NF-cadherin in Xenopus (32, 75, 76, 77, 78). While similar in their overall structure, they are missing the propeptide sequences found in classic cadherin, contain one to two additional extracellular subdomains including putative Ca2+-binding sites, and completely lack homology to classic cadherins in their intracellular tails (32, 76, 77). In contrast to classic cadherins, they mediate comparatively weak homophilic, Ca2+-dependent cell-cell aggregation that appears to be independent of interaction with catenins (32, 75, 77). Partial cDNA sequences have been obtained for other protocadherins including the fibroblast proteins FIB1 to FIB3 and ME1 to ME6 of melanoma cell lines and melanocytes in man and Pcdh4 to Pcdh6 in rat (77, 79, 80). Due to their long extracellular regions and absence of propeptide sequences, the Drosophila and human FAT proteins dFAT and hFAT, respectively, and the product of the dachsous gene (81) in Drosophila may also be placed into the protocadherin subfamily. Both FAT proteins contain an exceptionally large extracellular region with 34 tandemly arranged cadherin repeats and, in addition, 4 to 5 epidermal growth factor (EGF)-like repeats and a laminin A-G repeat (82, 83). In Drosophila the fat locus encodes a tumor suppressor gene, and recessive (loss of function) mutations lead to excessive proliferation of imaginal disc cells in the wing bud (84).

Several other cadherin-related proteins have been identified, which illustrates the structural and functional diversity of cadherins. H-cadherin is strongly expressed in brain and heart tissues and has been suggested to play a role in breast cancer (42, 85, 86). H-cadherin (cadherin-13) and its chicken homolog, T-cadherin, lack the entire cytoplasmatic tail and are integrated into the plasma membrane via a glycosyl phosphatidylinositol anchor (87, 88, 89). Similarly, the kidney-specific Ksp-cadherin, which appears to be associated with a renal Na+/HCO3--transporter (90, 91), rat LI-cadherin, expressed predominantly in liver and intestine (92, 93), and its putative human homolog, HPT-1 cadherin, reported to be associated with an intestinal peptide transporter (93, 94), lack most of the intracellular tail.

Interestingly, the transmembrane receptor tyrosine kinase RET encoded by the ret protooncogene also contains a cadherin-like sequence in its extracellular domain and displays Ca2+-dependent protection against proteolysis (95, 96). However, cell adhesive activity could not be demonstrated, and the organization of the ret gene differs considerably from those of classic cadherins, which show remarkable conservation of exon-intron boundary locations (97, 98). This indicates that RET is only distantly related. RET has recently been identified as part of multicomponent receptor complexes activated by transforming growth factor-ß (TGF-ß) family members, either glial cell line-derived neurotrophic factor (99, 100) or neurturin (101, 102). Mutations and translocations leading to constitutive activation of RET are implicated in a number of inherited and sporadic cancer syndromes such as multiple endocrine neoplasia type 2A and 2B, familial medullary thyroid carcinoma, and a subset of papillary thyroid carcinomas, while both activating and inactivating mutations were found in familial Hirschsprungs disease (for reviews see Refs. 103, 104, 105). In this context, it was speculated that interaction of RET with cadherins may contribute to tumorigenesis.

Evolution, classification, and structure-function analysis of cadherins are the subject of several excellent recent reviews (16, 29, 106, 107, 108, 109, 110, 111, 112). In the following subsection, therefore, we will briefly describe the structural and functional characteristics referring mainly to E-cadherin as perhaps the best studied cadherin.

B. Structure-function relationships of classic cadherins
Classic cadherins consist of an extracellular domain of approximately 550 residues, followed by a transmembrane region spanning the plasma membrane once and a highly conserved cytoplasmic domain of approximately 150 residues. They are synthesized as precursor proteins including a signal sequence and a propeptide of approximately 155 residues. Subsequently, the precursor (Mr ~135 kDa) undergoes posttranslational modifications (glycosylation, phosphorylation, and proteolytic cleavage) resulting in the mature protein (Mr ~120 kDa) (142, 143, 144, 145) (Fig. 1Go). The extracellular region is composed of five homologous subdomains (cadherin-repeats EC1–EC5), each consisting of approximately 110 residues (37). Single amino acids and cadherin-specific amino acid motifs such as LDRE, DXD, DXNDN (single letter code for amino acids; X refers to any amino acid residue), highly conserved in sequence and spacing, that are required for cadherin dimerization and Ca2+-binding have been identified in these repeats. EC1 mediates cell adhesion and in this domain the HAV-motif, together with variable residues in its immediate vicinity, are implicated in selective cadherin binding. Four conserved cysteine residues in EC5 are likely important for intramolecular disulfide bonds (142, 146, 147, 148, 149).

High-resolution structures of the first extracellular repeat EC1 of N-cadherin (150), E-cadherin (151), and the first two repeats EC1–2 of E-cadherin (152) have recently provided important insights into the three-dimensional structures of these molecules. According to these data, the cadherin repeats correspond roughly to structural folding units with, although unrelated in sequence, similarity to the Ig fold (151, 153). The EC1–2 structure showed that three Ca2+ ions are bound in a contiguous array at the interface between EC1 and EC2. Twelve amino acid residues contributed by both repeats and the linker region participate in Ca2+-binding. They include residues in previously predicted Ca2+-binding motifs DXNDN and DXD (142, 152, 154). One Ca2+ ion is bound primarily by residues of EC1, another primarily by EC2, and the third is coordinated by residues of both domains. In additon, Ca2+ appears to account for a number of noncovalent interactions between EC1 and EC2. Considering that residues involved in Ca2+ binding are conserved at most domain interfaces, the Ca2+ requirement for cadherin-dependent adhesion may result from a Ca2+-induced orientation of the subdomains and rigidification of the whole extracellular cadherin region (152, 155). Consistent with this model are electron microscopy studies, which indicate that Ca2+-binding induces a structural change of the extracellular E-cadherin domain from a more globular to an elongated rod-like shape (156).

According to both crystal structures, it was proposed that cadherins form stable lateral homodimers projecting outward from the same membrane plane with contacts mediated by parallel EC1 domains (150, 152). From the x-ray data of N-cadherin EC1, a second antiparallel interaction of EC1 domains was deduced. Extending this mode of adhesive interaction of EC1 domains to that between adjacent cells led to the proposal of a linear zipper-like mode of cadherin interaction (150, 157). This attractive model, however, is not entirely supported by the structural data with the EC1–2 dimer of E-cadherin (152); nor does it seem to be consistent with recent electron microscopy data obtained for the whole extracellular region of E-cadherin that was experimentally fused to an assembly domain (158). Regardless of the exact mode of interaction between adjacent cells, it appeared from several studies that a Ca2+-induced rigidification of cadherins’ extracellular region and Ca2+-dependent lateral homodimerization of cadherins may be a prerequisite for establishing an initial cell-cell contact, which is subsequently consolidated by intracellular associations involving catenins and by local cadherin clustering to generate stable intercellular junctions (152, 156, 158, 159, 160, 161, 162, 163, 164, 165, 166). In this context it should also be mentionend that cadherin-mediated cellular interaction is not exclusively homotypic and homophilic, since interactions of different cadherin types and of E-cadherin with other molecules, such as {alpha}Eß7-integrin of T cells, are well documented (167, 168, 169, 170, 171, 172, 173).

The cytoplasmic domain of classic cadherins is highly conserved (37, 108). The catenin-binding site has been mapped to the terminal 72 amino acids of E-cadherins’ carboxyl terminus by deletion mutagenesis experiments and construction of chimeric molecules (145, 174, 175, 176). In that location, a short core region of 30 amino acids essential for catenin binding was identified, which contains a cluster of 8 well conserved serine residues that are highly phosphorylated in the wild-type protein (145). Replacement of the whole serine cluster rendered E-cadherin unphosphorylated and led to parallel loss of catenin binding and of cell adhesion properties. Based on its recognition sequence and its ubiquitous expression, casein kinase II may be the natural kinase that phosphorylates E-cadherin (145). A different intracellular part of E-cadherin after the transmembrane region, termed the juxtamembrane domain, has been suggested to function independently in the suppression of cell motility (177). Similarly, studies in Xenopus have demonstrated that the cytoplasmic juxtamembrane domain of N-cadherin influences cell adhesion (178). The intracellular tail of E-cadherin, therefore, not only mediates its interaction with cytoplasmic proteins, but also regulates cadherin functions.

C. Catenins and the cadherin-catenin complex
Catenins, initially identified by coimmunoprecipitation with E- and N-cadherin, consist of three classic members, termed {alpha}-catenin, ß-catenin, and {gamma}-catenin, according to decreasing apparent molecular masses of 102 kDa, 88 kDa, and 80 kDa, respectively (174, 179, 180, 181). Biochemical evidence indicated that {gamma}-catenin is most likely identical with the common desmosomal plaque protein plakoglobin (182, 183, 184). In contrast to the diversity of cadherins, only two {alpha}-catenins (185, 186, 187, 188, 189, 190, 191), one ß-catenin (192, 193, 194, 195), and one plakoglobin (50) have been identified. Recently, the src-substrate p120cas has also been found to be associated with classic cadherin-catenin complexes in several cell types and represents a new catenin member, now termed p120ctn (196, 197, 198).

Previous biochemical characterizations indicated equimolar stochiometry of the complex consisting of E-cadherin, ß-catenin, and {alpha}-catenin, and differences in extractability of cadherin-bound catenins suggested a peripheral localization of {alpha}-catenin in such complexes (160, 181, 199). Several studies demonstrated that components in classic cadherin-catenin complexes are organized in a hierarchical order, e.g., E-cadherin contains a common binding site to accomodate either ß-catenin or plakoglobin in a mutually exclusive fashion, which, in turn, binds to {alpha}-catenin (Fig. 2Go). Therefore, at least two independent cadherin-catenin complexes may exist in the same cell, one containing ß-catenin and one containing plakoglobin. Both catenins appeared to associate early with E-cadherin at the endoplasmic reticulum during its synthesis, whereas {alpha}-catenin becomes integrated into the complex later at the stage of plasma membrane insertion. Upon cell-cell contact, {alpha}-catenin seems to establish a firm anchorage of the microfilament system to the complex, leading to a characteristic decrease in its detergent solubility, which may also reflect recruitment of further proteins (10, 159, 160, 181, 184, 194, 199, 200, 201, 202, 203).



View larger version (72K):
[in this window]
[in a new window]
 
Figure 2. Schematic representation of cell-cell and cell-matrix contacts of an epithelial cell. The inset depicts an adherens junction with the putative molecular interactions of E-cadherin dimers (150 152 ) and associated intracellular proteins (15 110 ).

 
The two known {alpha}-catenins coded for by separate genes are neural {alpha}N-catenin and epithelial {alpha}E-catenin, which are homologous proteins. Transfections and expression studies indicated that they can at least partially substitute for each other in supporting cadherins’ adhesive function in cell culture experiments (187), but probably not in vivo during trophoblast formation (204). The human {alpha}N-catenin (945 amino acids, Mr 104 kDa) has been mapped on chromosome 2p11.1-p12 (205). Alternative splicing generates two isoforms of {alpha}N-catenin, differing by a 48-amino acid insertion at the carboxy terminus, and both variants were found to be expressed in mouse neural tissues (206). The human {alpha}E-catenin gene containing 16 exons localizes to chromosome 5q31 and encodes a protein of 906 amino acids (Mr 102 kDa). A pseudogene with 90% similarity maps to chromosome 5q22 (188, 207, 208, 209). Three mRNA variants of {alpha}E-catenin were described, which either yield protein isoforms differing in their carboxy-terminal regions by insertion of 24 amino acids encoded by an alternatively used exon or appear as a mRNA transcript with an extended 3'-untranslated region probably affecting mRNA stability (190, 191).

Three regions of {alpha}E-catenin (Fig. 3Go) display homology to vinculin (185, 186), an important submembraneous component involved in actin bundling and F-actin attachment to plasma membrane protein complexes of adhesion sites such as cell-matrix contacts (210, 211, 212, 213, 214). Likewise it was suggested that {alpha}E-catenin may link cadherins to the actin-cytoskeleton (185, 186). Recent studies confirmed such a role of {alpha}E-catenin: it may connect the cadherin-catenin complex to actin either directly or via {alpha}-actinin (200, 215, 216, 217, 218). Furthermore, the reported binding sites in {alpha}E-catenin for these proteins are distinct, probably allowing simultaneous interaction of {alpha}-catenin with actin and {alpha}-actinin. (217). Interaction studies demonstrated, that the binding sites for plakoglobin and ß-catenin reside in the amino-terminal region comprising the first 228 amino acids of {alpha}-catenin (217, 219, 220, 221). Therein, a core of 27 residues appears to be necessary and sufficient for binding either one of both proteins by hydrophobic interaction (220) (Fig. 3Go).



View larger version (38K):
[in this window]
[in a new window]
 
Figure 3. Schematic representation of catenins and their functional domains. Numbers indicate the positions of the first and last amino acid residue in the proteins. Horizontal lines below each protein indicate binding for interacting proteins or regions mediating homodimerization. Top, Human {alpha}E-catenin. Alternative splicing generates an isoform containing 24 additional amino acids near the carboxy terminus (arrow). Shaded areas indicate three regions of homology to vinculin. The scheme was drawn with data according to Refs. 188, 190, 207, 215, 217, 220, and 221. Middle, Human ß-catenin. Shaded areas mark a repeated sequence motif, the arm repeat. The position of a consensus phosphorylation motif for glycogen synthase kinase 3ß (GSK3ß) and a putative transactivation domain is shown above the protein. APC, The tumor suppressor protein adenomatous polyposis coli; TCF, transcription factor of TCF/LEF-1 family; EGF-R, receptor for EGF (see also Section III). The scheme was drawn with data according to Refs. 194, 195, 230, 288, 411, 419, and 425. Bottom, Human plakoglobin ({gamma}-catenin). Shaded areas mark the arm repeats. Classic cadherins, E-, N-, and P-cadherin; Desmosomal cadherins, desmogleins and desmocollins. The scheme was drawn with data according to Refs. 194, 251, 258, 260, 263, and 264.

 
Plakoglobin ({gamma}-catenin) was recognized first by biochemical characterization as a major component of desmosomal plaque proteins (49, 222, 223). A protein of 745 amino acids (Mr 82 kDa) is predicted from its cDNA sequence (50); its human gene localizes to chromosome 17q21 (224), but despite being the first catenin cloned its gene structure has not yet been reported.

The cDNA for ß-catenin was first cloned from a Xenopus library (192). The gene encoding the human homolog resides on chromosome 3p21-p22 and covers a 23-kb region (195, 225, 226, 227). The 15 coding exons and the cDNA predict a protein of 781 amino acids (Mr 85 kDa) that is highly conserved among the species. Two splice variants have been reported affecting the 3'-untranslated region of the mRNA (193, 194, 195).

Due to an imperfect sequence motif of ~42 amino acids (arm-repeat) repeated 12 or 13 times in their central parts, plakoglobin, ß-catenin, and their homolog armadillo in Drosophila, constitute the armadillo family (228, 229, 230). p120ctn, Which was recently found associated intracellularly with classic cadherins, also belongs to the growing armadillo family (196, 197). Other armadillo proteins identified so far are the tumor suppressor adenomatous polyposis coli (APC), suspected to be causally involved in colon carcinogenesis due to inactivating mutations (231, 232, 233); the desmosomal plaque proteins plakophilin 1 and 2 (234, 235), p0071 (236), and the neural plakophilin-related armadillo protein NPRAP (237); the plant protein ARC1 (arm repeat containing) interacting with the S-locus receptor kinase (238); the guanine nucleotide exchange factor smgGDS (239, 240) and the related SMAP (smgGDS-associated protein) (241); SpKAP115, an accessory nonmotor component of sea urchine kinesin II (242); PF16, found at the microtubules of the flagella in the algae Chlamydomonas rheinhardtii (243); Srp1, an essential yeast gene of the importin-{alpha} family mediating nuclear import through the nuclear pore complex (244, 245) and the Srp1-related pendulin, cloned in Drosophila (246, 247); ARVCF (armadillo repeat gene deleted in velo-cardio-facial syndrome (248); and the yeast vacuolar protein VacP8 involved in protein targeting (249). These proteins display their several activities in diverse processes such as cell adhesion, nuclear transport, and growth control, but they are all characterized by the presence of a variable number of arm repeats. The arm-repeat domain in this heterogenous protein family appears to mediate primarily heterophilic protein-protein interaction. The crystal structure of the entire arm-repeat domain of murine ß-catenin revealed an elongated right-handed superhelical structure, in which each arm repeat folds into three {alpha}-helices (250). A band of positive surface potential, which resides in a groove formed by the superhelical twist of the ß-catenin structure, was proposed to facilitate the binding of cadherins or other interacting proteins such as APC and TCFs (T-cell factor; Section III) by means of charge complementation (250).

Extensive characterization of functional regions of ß-catenin and plakoglobin contributes much to a current understanding of composition and sorting of cadherin-catenin complexes (Fig. 3Go). According to these studies, the binding sites in ß-catenin may map to a broad region that includes the central-most arm repeats for E-cadherin association and to a stretch of 32 amino acids in front of or overlapping with the first arm repeat for interaction with {alpha}-catenin (194, 201, 217). Quite similar in size and location, the {alpha}-catenin binding site in plakoglobin comprises 29 amino acids close to the start of its arm region (251, 252). Several mapping studies indicate that plakoglobin harbors partially overlapping binding sites that probably enable mutually exclusive binding of classic and desmosomal cadherins via plakoglobin’s central arm repeats and arm repeats on both flanks, respectively. The amino-terminal binding sites of plakoglobin for {alpha}-catenin and desmosomal cadherins appear to partially overlap, which may explain {alpha}-catenins exclusion from desmosomes (251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262, 263). In addition to binding to desmosomal cadherins — with some preference for Dsg over Dsc — plakoglobin may support interaction of desmoplakin with these desmosomal cadherins involving plakoglobin’s carboxy-terminal half (261, 263, 264, 265). Desmoplakins are the major candidates that anchor desmosomes to the intermediate filament proteins. In addition, plakoglobin may directly contribute to cytokeratin binding since in vitro studies demonstrated a weak specific association of these proteins (265, 266, 267, 268, 269). Thus, plakoglobin may play a role in different processes important in desmosomal assembly and function, including heterodimerization of desmosomal cadherins and their anchorage to the intermediate filament.

The new catenin p120ctn, formerly termed p120cas, was initially identified as a substrate for the oncogenic tyrosine kinase src and for growth factor receptor tyrosine kinases (RTK) including receptors for EGF, platelet-derived growth factor (PDGF), and colony stimulating factor 1 (CSF-1) (270, 271, 272). p120ctn cDNA was first cloned in mouse, and the corresponding human gene mapped to chromosome 11q11 (196, 273, 274). In the mouse, four major p120ctn protein isoforms (Mr 90–120 kDa) are expressed simultaneously in several tissues (196, 197, 275). It has been inferred from the cloning of the human p120ctn gene, which comprises 21 exons, that in man the number of p120ctn isoforms is apparently much greater due to alternative splicing and alternative usage of start codons (198). The functional consequence is unknown as yet. p120ctn Is reported to directly associate with E-, N-, P- and VE-cadherins, but not with {alpha}-/ß-catenin and plakoglobin (196, 197, 276, 277, 278, 279, 280). Protein interaction studies indicate that the p120ctn-binding site resides in the carboxy-terminal tail of cadherins in the juxtamembrane region close to the ß-catenin/plakoglobin binding region and that the arm repeats of p120ctn are required for binding (277, 279, 280, 281, 282). The available data including the binding constellations are in favor of a modulatory role of p120ctn in the cadherin-catenin complex rather than a structural role.

Several studies documented complex dynamics during synthesis and initial and late assembly stages of cadherin-based intercellular junctions; biochemical characterizations demonstrated that more than 10 proteins may localize constitutively or transiently to adherens junctions, either forming additional structural components or acting as modulators (159, 160, 162, 163, 199, 283, 284, 285). Important aspects of signal transduction will be addressed in the following section.


    III. Signal Transduction, Hormonal Regulation, and the Cadherin-Catenin System
 Top
 Abstract
 I. Introduction
 II. Molecular Components and...
 III. Signal Transduction,...
 IV. Mouse Mutants
 V. Implications of the...
 VI. Summary and Conclusions
 References
 
A significant body of evidence indicates that the best studied E-cadherin-catenin complex is the target of many growth factor- and hormone-dependent signaling pathways to regulate its function and expression. On the other hand, the cadherin-catenin complex itself may modulate or initiate signaling events implicated in differentiation and growth control. Specifically, the "free" pool of cadherin-independent catenins, which may be present in the cytoplasm as monomers, homodimers, or heterodimers, may participate in signal transduction and gene transcription (160, 199, 221, 286, 287, 288, 289, 290, 291). Both aspects are the subject of several excellent recent reviews (292, 293, 294, 295, 296, 297, 298). The following two subsections will discuss relevant novel findings.

A. Cadherin-catenins as targets of regulation
The adhesive function of E-cadherin-catenin complexes and the expression of its components may be controlled by a variety of growth factors and hormones. These modulators may act at the protein level directly or indirectly, resulting in cadherin-catenin modifications by means of reversible phosphorylations, or may affect the amounts of the components resulting from effects on transcription or protein turnover.

Rapid changes in the phosphorylation patterns of cadherins and catenins in adherens junctions mediated by protein tyrosine kinases and phosphatases may represent the predominant mechanism at the posttranslational level. Also serine/threonine kinases, such as those of the protein kinase A and C families (PKA, PKC), act on cadherin-catenin complexes. Furthermore, phosphorylation of conserved serine residues in the cytoplasmic tail of E-cadherin, probably mediated by casein kinase II, appear to be necessary for catenin binding and cadherin-mediated cell adhesion (145). In addition, effectors of the organization of the cytoskeleton, including guanosine triphosphatases (GTPases) of the Rho subfamily, appear to regulate E-cadherin-dependent adhesion. The following four subsections including Fig. 4Go and Table 2Go will briefly review modulations by effectors with respect to the putative underlying mode of action.



View larger version (44K):
[in this window]
[in a new window]
 
Figure 4. Participation of cadherins and catenins in signal transduction. This figure represents a simplified scheme of signaling pathways probably interacting with cadherin and catenin function. Various receptor tyrosine kinases (RTKs), nonreceptor tyrosine kinases (NRTKs), and protein tyrosine phosphatases (PTPs) may modulate the interaction of "bound" catenins with cadherins affecting their phosphorylation. This may allow redistribution of catenins between the bound and "free" (i.e., cadherin-independent cytoplasmic) pool. The free catenin pool may generally be targeted for degradation or may be available for participation in signal transduction cascades such as the wnt-signaling pathway. Through heterodimerization with transcription factors of the TCF/LEF-1 family, ß-catenin may alter transcription of target genes. Refer to Sections III.A and III.B in the text. CS, Cytoskeleton; IF, intermediate filament system; Dsh, dishevelled; GSK3ß, glycogen synthase kinase 3ß; ILK, integrin-linked kinase; CK 2, casein kinase II; WRE, wnt responsive element; casp3, caspase3.

 

View this table:
[in this window]
[in a new window]
 
Table 2. Hormonal influence on cadherin/catenin expression and function

 
1. Tyrosine kinase-coupled mechanisms. The complex architecture of adherens junctions (Section II) implies that regulatory mechanisms are required for their organization. In breast cells transformed with activated Ras, an increase of tyrosine phosphorylation of ß-catenin and p120ctn is accompanied by changes of E-cadherin-mediated adhesion and disturbed interaction of cadherin-catenin complexes with the cytoskeleton (299). Increased tyrosine phosphorylation of both ß-catenin and cadherin mediated by v-src correlates with decreased cell-cell adhesion and increased invasiveness (300, 301, 302). Similarly, inhibition of protein tyrosine phosphatases (PTPs) in MDCK dog kidney cells rapidly elevated phosphotyrosine containing proteins at intercellular contacts with subsequent disruption of adherens junctions (303) and protooncogenes, such as the tyrosine kinases c-src, c-yes, and c-lyn, have been identified in adherens junctions preparations (285). Transformation of E-cadherin-expressing L-cells with v-src led to tyrosine phosphorylation of E-cadherin, ß-catenin, and p120ctn, but not of {alpha}-catenin (290). Whereas the exact relationship between v-src-mediated tyrosine phosphorylation of the cadherin-catenin complex and loss of cell adhesion is unknown, the latter study showed that neither the stochiometry nor the amounts of the E-cadherin-catenin complex were changed by v-src action. Tyrosine phosphorylation of ß-catenin may affect the ability of {alpha}-catenin to anchor the complex to the cytoskeleton, whereas the cadherin-catenin complex itself is not disrupted (290, 304). Alternatively, as shown by inhibition of PTPs in breast epithelial cells and E-cadherin-transfected leukemia cells, tyrosine hyperphosphorylation of ß-catenin or {alpha}-catenin may trigger dissociation of {alpha}-catenin from E-cadherin-catenin complexes (218, 305). Furthermore, recent transfection studies with a chimeric E-cadherin-{alpha}-catenin fusion protein lacking the ß-catenin binding site suggested that v-src-mediated weakening of cell adhesion might be independent of ß-catenin but might result from phosphorylation of other v-src substrates (306). Recently, the association of the EGF receptor with the E-cadherin-catenin complex provided additional evidence for tyrosine phosphorylation-dependent regulation of the cadherin-catenin complex by RTKs (288, 304). EGF stimulation of carcinoma cell lines resulted in a rapid increase of tyrosine-phosphorylated ß-catenin and plakoglobin. In vitro data indicated that the direct interaction of the EGF receptor with the E-cadherin-catenin complex is mediated by the central core region of ß-catenin (288). The concept of a functional modulation of E-cadherin via tyrosine phosphorylation is supported by studies in other cell lines indicating that direct phosphorylation of E-cadherin-bound ß-catenin and plakoglobin is induced by EGF- and by hepatocyte growth factor (HGF)-dependent RTKs in parallel with scattering of the respective cells (307, 308). Likewise, v-src and RTKs for growth factors such as EGF, PDGF, and CSF-1 either directly or via growth factor-associated cytoplasmic tyrosine kinases, such as FER and Tyk2, induce tyrosine phosphorylation of the catenin p120ctn (270, 271, 272, 309, 310). While the exact role of p120ctn for the function of cadherins is currently unknown, it was suggested that massive tyrosine phosphorylation may induce its dissociation from cadherin-catenin complexes and, in addition, may lead to subcellular redistribution of uncomplexed p120ctn, enabling association with other effector proteins (290). In contrast, in ras-transformed mammary cells, elevated tyrosine phosphorylation of cadherin-bound ß-catenin and of unbound p120ctn may induce displacement of ß-catenin by p120ctn in parallel with a decreased association of the complex to the cytoskeleton (299).

Hirohashi and associates (311, 312, 313) showed, in gastric carcinoma cell lines, that the RTK c-ErbB2, a protooncogene related to the EGF receptor, associates with and phosphorylates plakoglobin and ß-catenin bound in E-cadherin-catenin complexes. Mapping studies indicated that the c-ErbB2 binding site localizes to the carboxy-terminal arm-repeat region of ß-catenin. Transfection of the gastric cancer cell line TMK-1 with an N-terminally truncated ß-catenin lacking the binding site for E-cadherin resulted in inhibition of c-ErbB2-mediated phosphorylation of endogenous ß-catenin. Furthermore, the transfected cells displayed reduced invasive capacity in vitro and when injected into nude mice produced fewer and smaller metastatic nodules compared with host cells (313). Additionally, c-ErbB2 signaling might act at the transcriptional level leading to E-cadherin gene silencing as shown with transfected human mammary cell lines (314), and a comparable mechanism for EGF receptor action was demonstrated in other cell types such as thyrocytes (315).

Functional regulation at the protein level by RTK action is further exemplified in studies of the insulin-like growth factor-I (IGF-I) receptor-positive mammary carcinoma cell line MCF-7/6. Treatment with physiological concentrations of IGF-I rapidly restored the antiinvasive properties of the E-cadherin-catenin complex involving an RTK-dependent mechanism. Specific tyrosine kinase inhibitors block E-cadherin-mediated aggregation of these cells, a process independent of de novo protein synthesis (316, 317). The mechanism appears receptor specific (IGF-IR) and E-cadherin specific, since it can be inhibited by monoclonal antibodies specific for IGF-IR and E-cadherin and not be mimicked by IGF-II. Insulin exerts similar effects, albeit at higher molar concentrations, and is not blocked by IGF-IR antibodies. A truncated form of IGF-I, (des1–3)IGF-I, which has lost most of its IGF-IR binding activity, is more than 100-fold more potent in stimulating E-cadherin-mediated fast aggregation of MCF-7/6 cells, possibly via an autocrine mechanism by an as yet unidentified factor (316, 317). In contrast to the rapid IGF-I-induced IGF-IR autophosphorylation, the phosphorylation status of {alpha}-catenin, ß-catenin, and E-cadherin appeared to be unaltered (293). Direct association of IGF-IR with the E-cadherin-catenin complex forming a multielement complex has recently been suggested in immunofluorescence studies where specific antibodies revealed colocalization of the IGF-IR and E-cadherin but also of ß- and {alpha}-catenin at the points of cell-cell contact sites, and IGF-IR as well as its substrates, the insulin receptor substrate 1 and SHC, were contained within the E-cadherin complexes (318).

Consistent with a predominantly tyrosine phosphorylation-dependent regulation at the protein level, many studies report either an increase of tyrosine phosphorylation in cadherin-catenin complexes by PTP-inhibitors accompanied by loss of cadherin function (218, 280, 300, 303, 305, 310, 319) or the direct association of cadherin-catenin complexes with transmembrane or membrane-associated PTPs such as PTPµ, PTP{kappa}, LAR-PTP, and PTP1B-LP (320, 321, 322, 323, 324). Interestingly, PTP1B-LP inactivation might be triggered via extracellular proteoglycan interaction with a cell surface N-acetyl-galactosaminyl-phosphotransferase, thereby leading in concert with N-cadherin inactivation to elevated tyrosine phosphorylation of ß-catenin and dissociation of associated actin by extracellular signals (322, 325, 326). Both PTPµ and PTP{kappa} belong to the class of receptor-like PTPs located in intercellular contact regions and are implicated in mediating homophilic cell-cell adhesion by themselves (reviewed in Ref. 327). PTPµ has been demonstrated to interact with E-, N- and R-cadherin, mediating their dephosphorylation (320, 324). In contrast, the more ubiquitously expressed PTP{kappa} appeared to use ß-catenin and plakoglobin as substrates by binding to their arm repeats, whereas binding to E-cadherin or {alpha}-catenin was undetectable (321). Thus, PTPµ and PTP{kappa} are good candidates to reverse the tyrosine phosphorylation status of cadherins and catenins and to allow an additional modulation of tyrosine kinase signaling mentioned above. Apparently, a unique mechanism exists neither for tyrosine kinase- nor tyrosine phosphatase-mediated signaling and regulation, but, instead, different molecular mechanisms acting at the protein or transcriptional level, alone or in concert, are active and appear to depend on both the effectors, i.e., the respective kinases/phosphatases, and the cellular context, i.e., the expressed cadherin/catenin species and cell types involved.These cell-specific effects appear not to be restricted to RTK pathways but may also hold true when expanded to signaling generated by other growth factors, hormone receptors, and kinases, including those of the PKC and PKA family with their respective phosphatases (see Sections III.A.3 and 4).

2. Regulation by cytoskeletal effectors. Recently, several studies gave new insights into other important signaling mechanisms involving the coordinated assembly of cadherin-catenin complexes with the actin cytoskeleton and the reorganization of the microfilament system upon cadherin-mediated adhesion. The organization of the actin cytoskeleton is regulated by proteins of the ras-GTPase superfamily, the Rho subgroup of small GTPases including Rho and Rac proteins, which control actin polymerization, and a variety of other cytoskeleton-associated processes (for review see Ref. 328). Studies with normal human keratinocytes demonstrated that blocking Rac or Rho activity leads to removal of cadherins from intercellular junctions and, furthermore, actin recruitment to cadherin-based junctions requires Rac and Rho activity (329). Similar results for Rac, Rho, and another Rho family member, Cdc42, were obtained with dog MDCK cells and human lung carcinoma cells (330, 331, 332). Ras-transformed MDCK cells showed a fibroblast-like phenotype with increased invasiveness as a result of reduced E-cadherin-dependent adhesion (333, 334). Transfection with the guanine nucleotide exchange factor, Tiam1, an activator of Rac, restored E-cadherin-dependent adhesion and inhibited invasiveness in the ras-transformed cells (335). Furthermore, ectopic expression of Tiam1 or Rac inhibits HGF-dependent scattering in epithelial MDCK cells, probably due to a Rac-mediated increase in F-actin polymerization that facilitates stable E-cadherin-catenin anchoring to the actin cytoskeleton and increases cell adhesion (335). Similarly, inhibition of Rho/Rac by Rho-GDI, a regulatory protein belonging to the class of the guanine nucleotide dissociation inhibitors (328), perturbs E-cadherin-dependent cell adhesion (332). It has been suggested that Rho-like GTPases may mediate their effects on cell adhesion through a target protein of Rac and Cdc42, IQGAP1, which has been shown to be localized to cell-cell contact sites (336, 337). Even if a detailed mechanism is still lacking, in vitro and in vivo data indicate that IQGAP1 may bind simultaneously to E-cadherin and ß-catenin and may induce dissociation of {alpha}-catenin from the E-cadherin-catenin complex (336). These studies showed an additional level of cadherin-catenin regulation exerted by Rho family members, their regulators including Tiam1 and Rho-GDI, and their targets such as IQGAP1. GTPases of the Rho family, which are modulated by many hormones and growth factors leading to both direct cytoskeletal changes and transcriptional changes (328), consequently integrate cadherins and catenins into the complex network of intracellular and extracellular interactions.

3. PKC- and PKA-coupled mechanisms. The first evidence that PKC signals at the protein level to E-cadherin and catenins came from experiments in colon (338) and lung carcinoma cells (339). In human colon carcinoma cell lines lacking {alpha}-catenin, the invasive phenotype was reversed upon treatment with the PKC activator, 12-O-tetradecanoyl-phorbol-13-acetate (TPA) by a dual mechanism on desmosomes via plakoglobin and Dsg 2 and via E-cadherin-mediated cell adhesion. It is interesting to note that this effect, which is comparable to the restoration of tight intercellular adhesion by {alpha}-catenin transfection in these cells, appears to bypass {alpha}-catenin (338). Activation of a G protein-coupled PKC-linked receptor, the M3 muscarinic acetylcholine receptor, rapidly induced E-cadherin-mediated adhesion in a small cell lung carcinoma (SCLC) cell line (339). Conversely, in many other cell lines, phorbol esters and the activation of the PKC pathway down-regulate cadherin expression. In thyrocytes, TPA acutely induced a barrier dysfunction within 24 h of treatment, accompanied by dissociation of cadherin-based junctions and shedding of cells into the apical medium (340). Similarly, in other cell types down-modulation of the functional activity of E-cadherin by phorbol esters parallels an increase of cell scattering. This is accompanied by increased phosphorylation of p120ctn and by an elevated association of this protein to E-cadherin. It may be reversed by the tyrosine kinase inhibitor, herbimycin A, supporting a role of src and p120ctn in the functional control of E-cadherin (341, 342). In the highly metastatic human adenocarcinoma cell line, L-10, PKC activation apparently led to misrouting of the cadherin-catenin complex that no longer was detected in the membrane fraction, even though the localization of {alpha}-/ß-catenins or p120ctn was unaltered (343).

Activation of PKA via cAMP-dependent pathways stimulates cadherin transcription in several cell types such as gonadal cells, placenta, thyrocytes, and osteoblasts (315, 344, 345, 346, 347, 348, 349, 350), but the mechanisms have not been clarified. Studies on cAMP-dependent regulation of the E-cadherin promoter suggest that signaling is not mediated directly via classic cAMP-response elements (CREs) since the regulatory region of the mouse E-cadherin gene is composed of a promoter consisting of positive regulatory elements such as a CCAAT-box, two AP-2 binding sites, and the palindromic element E-Pal, but contains no classic CRE motif, also lacking in the human E-cadherin promoter (97, 351, 352, 353). While the underlying mechanisms are currently unknown, studies with thyrocyte cultures that were exposed to TSH increased E-cadherin steady-state mRNA and protein levels (315). This has led to speculations that cAMP-induced transcription factors, such as c-fos or c-jun, acting on an AP-2 element may mediate the stimulatory action on E-cadherin expression (315). Gonadotropins and sex steroids may modulate N-cadherin expression in gonadal cells, a process that may be synergistically regulated by effects of sex steroids on the cadherin-catenin complex (see Section III.A.4). For dispersed, but not for aggregated, rat granulosa cells, a multihormone dependence of N-cadherin mRNA expression on estradiol and FSH has been described (354). Similarly, in rat Sertoli cells, FSH-stimulated N-cadherin mRNA transcription was potentiated by estrogens whereas estrogens alone were ineffective (349). Furthermore, N-cadherin-mediated binding of round spermatids to Sertoli cells was increased only by combined treatment with FSH and testosterone while the single hormones were inactive (350). Another cadherin, cadherin-11, appears to be positively controlled by cAMP stimulation in endometrium cells, in syncytial trophoblast cells, and in extravillous cytotrophoblast cells (131). However, cAMP-mediated regulation is not stimulatory in all cell types. Exposure of BeWo chorion carcinoma cells to cAMP induced cellular fusion and syncytium formation and resulted in the simultaneous disappearance of E-cadherin from the cell surface and a reduction in E-cadherin mRNA steady-state levels (348).

4. Sex steroid- and retinoic acid-coupled mechanisms. As delineated above for other signal transduction cascades, the effects of sex steroids on the cadherin-catenin complex mostly studied immunohistochemically and by functional tests are critically dependent on the cellular system used. Early reports focussing on N-cadherin expression in granulosa and Sertoli cells suggest a stimulatory action of estrogen or testosterone, respectively (350, 355, 356). When treated with 17ß-estradiol, the human mammary carcinoma cell line MCF-7 undergoes morphological differentiation in vitro, including rearrangement of subapical/basolateral F-actin and plakoglobin staining into a more uniform pattern and a gradual loss of vinculin from cell-matrix and cell-cell adherens junctions (357). Defective E-cadherin-catenin function in these cells is restored in fast aggregation assays by tamoxifen and, to a lesser degree, by its metabolites 4-OH-tamoxifen and N-desmethyl-tamoxifen, supporting a mechanism mediated by classic nuclear receptor pathways, as this effect was not observed in estrogen receptor-negative breast cell lines (316, 358). However, in other human breast cancer cells, tamoxifen induced an epithelial to mesenchymal transdifferentiation along with a decreased expression of E-cadherin, which may involve stimulation of TGF-ß and may be mediated by the type I TGF-ß Tsk7L receptor (359, 360).

Not all effects of sex steroids are mediated via nuclear steroid receptors as indicated by experiments using specific receptor antagonists (361). Regulation of intracellular calcium and calcium influx via voltage-dependent calcium channels appears to be involved in sex steroid action. In experiments with MCF7/6 cells (362), the calcium channel agonist, Bay K8644, completely abolished the tamoxifen-induced effect, whereas an antagonist of this channel, verapamil, reduced the effective dose of tamoxifen to restore E-cadherin function in these cells.

Similar to sex steroids, retinoic acid, like all-trans-retinoic acid, induces rapid positive effects in mammary and colon cancer cells in fast aggregation assays, which can be specifically blocked by anti-E-cadherin antibodies (363). Byers et al. (364) suggested that these effects of all-trans retinoic acid and of 9-cis-retinoic acid converge on ß-catenin and involve cytosolic calcium. There was no change of the steady-state mRNA levels of ß-catenin, but 9-cis-retinoic acid did increase ß-catenin stability, a process reversed by low calcium levels. Broad spectrum kinase inhibitors, such as staurosporine and the PKC inhibitor bisindoylmaleidimide, counteracted these effects of 9-cis-retinoic acid. The incidence of mammary carcinomas induced in intact rats with N-nitroso-N-methylurea, furthermore, was effectively reduced by 9-cis-retinoic acid treatment, whereas all-trans-retinoic acid was less potent (365). Since cotreatment with retinoic acid and tamoxifen potentiated their effects (365), however, independent mechanisms may be involved.

In hypertransformed epithelial cells transfected with E1A 12S, retinoic acid may act on the subcellular routing of E-cadherin and may induce a correct localization of immunoreactive E-cadherin (366). In contrast to up-regulation in mammary cancer cells, estrogens decrease cell aggregation and selectively reduced E-cadherin and {alpha}- and ß-catenin steady-state mRNA levels in endometrial cancer cells, whereas antiestrogenic compounds such as progesterone or danazol reversed this effect (367). This mechanism, shown for Ishikawa cells, a cell line derived from a differentiated adenocarcinoma of the endometrium, may help to explain not only the advantage of cancer cells for detachment in an estrogen-containing milieu, but furthermore for the cyclic regulation of normal endometrial cells. The relative proportion of estrogen to progesterone appears to have direct effects on E-cadherin expression, and it was speculated that the increased incidence of endometriosis in patients with corpus luteum insufficiency may be attributed to down-regulation of E-cadherin due to the lower than normal progesterone levels (368). Formal studies, however, are needed to validate this observation.

B. Cadherin-catenins as signal transducers
In addition to the impact of E-cadherin on adherens junction formation, cell-cell adhesion, and cytoskeleton organization, many studies, including those of compaction in embryonic development (Section I) and of its expression in cancer (Section V), document its role of one — but not the only—master molecule triggering formation and function of other intercellular junctions. These include tight junctions, gap junctions, and desmosomes, whereby PKC has been suggested as one downstream component activated by E-cadherin-mediated outside-in signaling (338, 344, 393, 394, 395, 396, 397). Possibly due to the low intrinsic binding affinity of the single extracellular domain of E-cadherin, direct assays of a receptor-like activity generating intracellular signals such as classic second messengers are lacking. However, a recent study with MDCK cells and human breast MCF10A cells using immobilized E-cadherin antibodies to mimick homophilic adhesion demonstrated rapid and specific increase of tyrosine-phosphorylated proteins with apparent molecular masses of 120 kDa, 130 kDa, and 180 kDa, one of which was identified as ras-GAP (398). The consequences of cadherin-mediated ras-GAP tyrosine phosphorylation are not clear. However, in light of the Rho/Rac-dependent regulation of E-cadherin (Section III.A.2), it is possible that E-cadherin-mediated outside-in signaling could alter specifity or activity of ras-GAP, which may in turn activate Rho/Rac and lead to an increase of stable adherens junctions via increased assembly of actin with E-cadherin-catenin complexes. This assumption would imply an interesting autoregulatory loop operating at early stages of intercellular junction assembly.

Additional new evidence for cadherin-mediated outside-in signaling via tyrosine phosphorylation came from N-cadherin-transfected Chinese hamster ovary cells challenged with beads coated with N-cadherin or N-cadherin antibody (399). This treatment resulted in increased levels of cadherin and ß-catenin in adherens junctions of the challenged cells, irrespective of their proximity to the site of bead interaction, and was accompanied by increased levels of tyrosine-phosphorylated proteins in cell-cell junctions. Furthermore, treatment of the cells with beads coated with the integrin ligand fibronectin increases tyrosine phosphorylation and vinculin levels at focal contact sites and suppressed cadherin and {alpha}-catenin at intercellular junctions. As judged from tyrosine kinase inhibition, both cadherin and integrin ligand-mediated effects depended on tyrosine kinases (399). Although the downstream effectors have not been identified, this study elegantly demonstrates a selective autoregulation of adherens junction formation and focal contact formation by cadherins and integrin ligands, respectively, and suggests specific reciprocal regulation mechanisms resulting in the mutually exclusive formation of either cell-cell contacts or cell-matrix contacts.

One promising candidate for a downstream effector of integrin-dependent outside-in signaling via integrin ligands suppressing adherens junctions is integrin-linked kinase (ILK), a serine/threonine kinase recently identified and further characterized by Dedhar and associates (400, 401, 402). ILK binds to the cytoplasmic domain of both ß1-and ß3-integrins and phosphorylates ß1-integrin (Ref. 400 ; for review Ref. 403). ILK overexpression in epithelial cells decreased E-cadherin levels concomitant with loss of cell-cell adhesion and induction of tumor formation in vivo (401). Furthermore, ILK overexpression induced translocation of ß-catenin to the nucleus and activated lymphoid enhancer factor-1 (LEF-1)/ß-catenin-dependent transcription indicative for oncogenic transformation and for cross-talk between wnt- and integrin-signaling pathways (402) (Fig. 4Go).

The wnt-signaling pathway(s) in vertebrates and the related wingless (wg) pathway(s) in Drosophila have been the subject of intensive research in the past years, and a direct participation of the cadherin-catenin system in this signal transduction pathway has been well documented (for recent reviews see Refs. 297, 298, 404). Nevertheless, despite considerable recent progress, which includes identification of missing links in this pathway(s) such as receptors, downstream signaling components, and the first target genes, several parts and actions of the complex signaling machinery represented by this pathway(s) are not fully understood. Since covering all aspects is far beyond the scope of this review, we will describe the wnt pathway in relation to cadherins-catenins in terms of the simplified scheme shown in Fig. 4Go.

Wnt proteins are secreted glycoproteins that form a large protein family implicated in regulation of cell fate decisions during vertebrate development (296, 405). Wnt-1, the first identified member, was discovered by its contribution to mouse mammary tumorigenesis induced by ectopic Wnt-1 expression due to proviral insertion (406). Wnt proteins, such as Wnt-1, appear to interact with transmembrane receptors of the frizzled (Fz) family (407, 408). Upon binding to its ligand, Fz may activate dishevelled proteins (Dsh), which in turn inhibit glycogen synthase kinase 3ß (GSK3ß). These steps are less well understood and may require additional proteins such as casein kinase II, which has been suggested to phosphorylate Dsh (298). The inhibition of GSK3ß results in the activation of cytoplasmic uncomplexed ß-catenin by inhibiting its turnover, which leads to an increase of its cytoplasmic levels above a critical threshold (289, 290, 409). ß-Catenin, in turn, may associate with high-mobility group transcription factors of the TCF/LEF-1 family (410) such as hTCF4 and LEF-1, and the complex may translocate to the nucleus leading to altered transcription of target genes (411, 412, 413). In the absence of a Wnt-1 ligand, the levels of ß-catenin pools are kept low due to a high turnover rate by proteolytic degradation. This default route in the absence of Wnt-1 involves ß-catenin binding to the tumor suppressor APC and phosphorylation of both APC and conserved serine residues in the amino-terminal region of ß-catenin by the action of constitutively active GSK3ß (286, 287, 414, 415, 416, 417). Furthermore, two related proteins, termed axin and conductin, were found recently to be functional parts of higher order protein complexes consisting of APC, GSK3ß, ß-catenin, and either axin or conductin inhibiting ß-catenin/LEF-1-dependent transcription by means of promoting ß-catenin degradation (418, 419). Ubiquitination and proteasome-mediated proteolysis appear as final steps of ß-catenin’s turnover (420, 421). In addition, ß-catenin was found to be inactivated by proteolytic cleavage mediated by caspase 3, a killer protease activated upon induction of apoptotic programs (422). Thus, ß-catenin-dependent nuclear signaling can be switched off rapidly by different mechanisms. In the nucleus, TCF/LEF-1-ß-catenin complexes may bind to responsive elements in the promoters of TCF/LEF-1-target genes and induce local bending of DNA, thereby changing transcription probably via the transactivation domain at ß-catenin’s carboxy-terminal end (410, 411, 412, 413, 423, 424, 425). Physiological targets are homeobox genes including engrailed and ultrabithorax in Drosophila and twin and siamois in Xenopus (297, 298, 426, 427, 428, 429). Recent studies showed an increase of the gap junction protein connexin-43 (Cx43) at the protein and transcriptional level in Wnt-1-overexpressing rat neural crest-derived PC12 cells (430). Analysis of the rat Cx43 promoter revealed two putative TCF/LEF-1-binding elements, and transfection experiments indicated a Wnt-1-dependent activation of Cx43 promoter constructs via the TCF/LEF-1-ß-catenin route (430, 431). Thus, the Cx43 gene, in contrast to other connexins studied (430), represents another promising candidate to be regulated directly by the wnt pathway in a cell-specific manner. Furthermore, in search of new genes, which are regulated by APC, He et al. (432) introduced an inducible APC gene in in a colorectal cancer cell line that lacks endogenous functional APC. Screening of differentially expressed genes after wild-type APC induction led to the identification of the protooncogene c-myc as a direct target in the wnt pathway in these cells. The expression of c-myc was found to be repressed at the transcriptional level after APC induction and activated by TCF4/ß-catenin presumably acting on two TCF4-responsive elements identified in the c-myc promoter (432). Therefore, in normal colorectal epithelial cells, APC may prevent ß-catenin from interaction with TCF4 and may repress TCF4/ß-catenin-dependent c-myc expression. It has been proposed that mutational alterations, such as inactivating APC mutations or activating ß-catenin mutations, that would result in TCF4/ß-catenin activity may lead to overexpression of c-myc, which in turn may promote neoplastic growth (432). Recent genetic studies in mice, which show that TCF4 plays a crucial role for the maintenance of the proliferative compartment in intestinal crypts, are compatible with a growth control function of both APC and ß-catenin (433). The wnt pathways document an important additional role of ß-catenin in the nucleus influencing transcription, which closely relates both cell adhesion and tumorigenesis (Section V). This is not only reflected by the participation of the tumor suppressor APC, which in principle may compete with E-cadherin for ß-catenin (194), but also by a recent study in MDCK cells, which demonstrates a transcriptional activation of E-cadherin expression by c-myc as well as by the retinoblastoma gene product (RB), a classic tumor suppressor (434). Direct association of c-myc and RB with AP-2 transcription factors, which act on AP-2 elements in the E-cadherin promoter, may account for the effects on E-cadherin gene transcription, but the exact mechanism needs to be further explored (434) as well as its speculative relationship to the TCF4/ß-catenin-dependent c-myc regulation mentioned above (432). Interestingly, a region in the E-cadherin promoter also appears to bind to the LEF-1/ß-catenin complex in vitro, and it has been suggested that the LEF-1/ß-catenin complex regulates transcription of the E-cadherin gene (413). Overexpression of Wnt-1 in cell lines was found to increase Ca2+-dependent cellular adhesion with accumulation of E-cadherin/catenins in intercellular junctions. In addition, ß-catenin and plakoglobin expression was elevated at the posttranscriptional level (371, 435). Both effects might be short-term consequences of blocking the described catenin degradation pathway, which induces a transcription-independent increase and redistribution of uncomplexed catenin pools to other cellular compartments, including the nucleus and adherens junctions (289, 290, 436). Plakoglobin appears to use the same default route of degradation as ß-catenin, since it contains the same GSK3ß consensus phosphorylation site, directly binds to APC, and is found to be ubiquitinated after proteasome inhibition. In addition, overexpression or misrouting of plakoglobin may compete with ß-catenin for degradation and thereby affect ß-catenin-dependent functions in both signaling and cell adhesion, which implies that the complexed and unbound pools of catenins must be strictly controlled in normal physiology (420, 436, 437). Even if many molecular mechanisms await clarification, the wnt-pathway appears to directly integrate the cadherin-catenin system, especially ß-catenin, in growth control processes, which are relevant for normal cell physiology and cancer.


    IV. Mouse Mutants
 Top
 Abstract
 I. Introduction
 II. Molecular Components and...
 III. Signal Transduction,...
 IV. Mouse Mutants
 V. Implications of the...
 VI. Summary and Conclusions
 References
 
Several components of the cadherin-catenin system and the wnt-signaling cascade have been analyzed in genetically modified mice (see Table 3Go), which gave important insight into the role of these genes during vertebrate development. Many gene deletion mutants were embryonal lethal, which emphasizes the important function of the cadherin-catenin system during vertebrate development. This includes the knockouts for E-cadherin (438, 439), N-cadherin (440), ß-catenin (441), plakoglobin (442, 443), APC (444), and wnt3a (445). Other knockouts were less severe than expected from their abundant expression in the organism, which may be due to maternal rescue during early development and redundancy, and many of them display a recessive phenotype. The cadherin-catenin system may, therefore, be organized as a dynamic network that is frequently not dysregulated by an altered gene expression but subject to posttranslational modification.


View this table:
[in this window]
[in a new window]
 
Table 3. Cell-cell adhesion-related mouse mutants

 
Studies on E-cadherin mutations in lobular breast cancer provide convincing evidence for a role of E-cadherin as a tumor suppressor, but many other tumors lack defects in their cadherin/catenin-system (446, 447). Similarily, among the presently available mouse mutants, an increase in tissue neoplasia is only seen in a single loss-of-function mutant affecting the cadherin-catenin system. An increase in tissue neoplasia is only seen in a single loss-of-function mutant affecting the cadherin-catenin system. Mice bearing a deletion of exon 14 in their APC gene, resulting in a truncated gene product at codon 1638 (444), exhibit multiple intestinal adenomas and carcinomas in up to 80% of their heterozygous offspring. In the majority of murine deletion mutants, however, including those for cadherins, </