ReviewMechanism of homophilic cadherin adhesion
Introduction
The transmembrane glycoproteins known as cadherins mediate cell–cell adhesion in all soft tissue 1, 2. They are also responsible for cell segregation during morphogenesis 3, 4. The homophilic binding site(s) and the tissue selectivity region reside in the extracellular segment, which folds into five structurally homologous domains numbered 1–5, starting from the outer domain 1, 2, 5, 6••, 7. Adhesion requires calcium, which both rigidifies the intersegment links and apparently activates trans interactions 6••, 8••, 9, 10, 11, 12••. In addition, a substantial body of evidence shows that cadherin dimerizes on the cell surface and that cis dimerization is necessary for adhesion 13, 14, 15, 16. Because of the involvement of cadherin in development, cancer, and other diseases, there is tremendous interest in the physical mechanisms underlying its function. The altered tissue specificity conferred by exchanging the amino-terminal domains of neural cadherin (NCAD) and epithelial cadherin (ECAD) implicated this domain in the recognition of opposing cadherins [5]. However, the domains involved and the molecular interactions responsible for adhesion are still controversial topics [6••].
Structural investigations of cadherin binding primarily focused on the two outer amino-terminal domains 8••, 10, 17, 18. However, interpreting the crystal structures of domains 1 and 2 of ECAD and NCAD in terms of cadherin adhesive function yielded conflicting results. A model for homophilic cadherin binding was proposed on the basis of the structure of the amino-terminal domain of neural cadherin (NCAD1) [17]. Cis bonds were mediated by the exchange of β-strands between adjacent protomers (strand dimer). An adhesion site was assigned to a large area of contact between the antiparallel domains (adhesive dimer). The buried surface area at the putative adhesive interface is approximately 3300 Å2, and the interfacial region involves a number of long-range interactions and bridging waters 6••, 17. It does contain the conserved HAV peptide sequence, which might be involved in adhesion [19], but the alanine residue is recessed and does not contact the opposing protein [17].
The model resulting from the structure of cadherin adhesion (the zipper model) posited that the formation of trans adhesive contacts between the amino-terminal domains (NCAD1) of cis dimers generates an alternating ribbon of antiparallel dimers, which form the intercellular junction (Figure 1). Consistent with this model, fusions of the E-cadherin ectodomain with the pentamerization domain of cartilage oligomeric matrix protein (COMP) assembled into star-shaped aggregates (Figure 2). These aggregates associated via the amino-terminal ends of the cadherins to form ring-like structures (Figure 2) 8••, 12••. Shear-flow detachment studies with ECAD1 also indicated weak interactions with high dissociation rates [20]. The ∼20–25 nm intercellular gap distances observed in electron micrographs of adherens junctions and desmosomes was used to justify the hypothesis that amino-terminal domains mediate adhesion [21]. At the carboxy-terminal region of the NCAD1 structure there was a slightly disordered partial calcium-binding region between domains 1 and 2 [17], and the calculated intermembrane gap distance estimated on the basis of the 2.9 nm NCAD1 length was consistent with the electron microscopy (EM) data [21]. However, structures solved later of domains 1 and 2 of NCAD (NCAD12) and domains 1 and 2 of epithelial cadherin (ECAD12), which contained structured calcium-binding linker regions, revealed that the domain lengths are closer to 4.5 nm 8••, 10, 18. The end-to-end cadherin distance would therefore be much larger at ∼45 nm 8••, 10, 18.
Subsequent structural data are also inconsistent with direct adhesion between the amino-terminal domains. Neither the putative adhesive interface nor the strand dimer interactions were observed in the structures of NCAD12, ECAD1 or ECAD12 8••, 10, 18. A second structure of ECAD12 8••, 10 further suggested that cis dimers might form by bridging between the calcium-binding regions, rather than by β-strand exchange. Furthermore, other findings suggest the involvement of additional domains. The DECMA-1 antibody against the membrane proximal region of ECAD1–5 inhibits adhesion [22]. The reversible loss of adhesive function by DTT treatment was attributed to changes in domain 5, although the effects on the protein structure were not determined [22]. One antibody to domain 5 activates cadherin, whereas yet another antibody to domain 3 inhibits it (B Gumbiner, personal communication). Additionally, a cluster of mutations on domain 3 is associated with gastric cancers [23].
Section snippets
Testing the model of the homophilic cadherin-binding mechanism by surface force measurements
A direct way to test the cadherin-binding models is to quantify both the functional activity (adhesion) of the full length ectodomains and the relative distance(s) at which the proteins bind. For example, with the length of the fully folded amino terminus, the model of Shapiro et al. [17] would predict that adhesion between cadherin extracellular domains takes place at an intermembrane separation of 40–45 nm. This type of information is obtained routinely with the surface force apparatus (SFA),
Direct force measurements reveal three distinct antiparallel adhesive alignments
The SFA was recently used to test current models of cadherin adhesion, with surprising results [33••]. Direct force–distance measurements between full-length, oriented ectodomains of C-cadherin from Xenopus revealed that cadherin adhesion involves more than just the amino-terminal domain. Instead, the protein binds in three distinct antiparallel alignments, and the strongest bond does not involve direct interactions between the amino-terminal domains.
The first direct measurements of the
Remaining questions
The normalized force–distance data alone unfortunately do not reveal which domains of cadherin mediate adhesion. First, the profiles only give the relative cadherin alignments. Second, the 1 nm uncertainties in the positions of the outer two minima are due to errors in determining the point at which the bonds fail and the proteins begin to separate. At high cadherin densities, the slow initial protein detachment makes a more precise localization of the minima difficult [33••].
The major
Conclusions
X-ray structures have provided valuable but contradictory data regarding cadherin function. Force measurements directly tested cadherin-binding models by quantifying the adhesion and the relative protein alignments that mediate binding. The force profiles demonstrated that homophilic C-cadherin binding involves multiple, distinct binding interactions, which involve more than just the amino-terminal domain. The current structural and functional information is still insufficient to elucidate the
References and recommended reading
Papers of particular interest, published within the annual period of review, have been highlighted as:
• of special interest
•• of outstanding interest
References (44)
- et al.
Cell-to-cell contact and extracellular matrix
Curr Opin Cell Biol
(1995) Cell adhesion: the molecular basis of tissue architecture and morphogenesis
Cell
(1996)Cadherins in cancer: implications for invasion and metastasis
Curr Opin Cell Biol
(1993)- et al.
Localization of specificity determining sites in cadherin cell adhesion molecules
Cell
(1990) - et al.
Homophilic adhesion by cadherins
Curr Opin Struct Biol
(1999) - et al.
Single amino acid sustitutions in one Ca2+ binding site of ovomorulin abolish the adhesive function
Cell
(1990) - et al.
Lateral clustering of the adhesive ectodomain: a fundamental determinant of cadherin function
Curr Biol
(1997) - et al.
Structure-function analysis of cell adhesion by neural (N-) cadherin
Neuron
(1998) - et al.
Identification of a cadherin cell adhesion recognition sequence
Dev Biol
(1990) Thin film studies using multiple-beam interferometry
J Coll Int Sci
(1973)