Diversity of the cadherin superfamily

Current reviews classify the cadherin superfamily into several subfamilies (Nollet et al., 2000): the type I, or “classical” cadherins as well as type II cadherins are both strongly connected with the actin cytoskeleton. A third subfamily contains the desmosomal cadherins, desmoglein proteins and desmocollins, which are linked to intermediate filaments (see above). Finally, protocadherins are predominately expressed in the brain and not or only weakly associated with the cell cytoskeleton.

Aside from these subfamilies there exist various “atypical cadherins”, i.e. molecules containing one or more cadherin repeats in their extracellular domain but do not exhibit more striking features of cadherins. A schematical overview of the distinct structure of several members of the cadherin family is given in Figure 2.

Attempting to regulate the cadherins nomenclature, the “HUGO Nomenclature Committee” assigned each cadherin a distinct number and gene symbol – listed in table 1 beside trivial names of the cadherins and their amount of extracellular cadherin repeats.

Figure 2 Schematic overview of the cadherin superfamily (taken from Angst et al., 2001)

table 1 Nomenclature and tissue expression of common members of the cadherin family

(taken from Nollet et al., 2000) HAV = histidine-alanine-valine motif, Rn = rattus norwegicus, Hs = homo sapiens, Mm = mus musculus, Xl = xenopus, Dm = drosophila melanogaster, Ce = Caenorhabditis elegans, Bs = Botryllus schlosseri.

Cadherins of the classical type (type I cadherins) are e.g. E-cadherin, N-cadherin, P-cadherin, R-P-cadherin, VE-cadherin and cadherin 7, all of them exhibiting exactly five extracellular cadherin repeats (EC). They all share a highly conserved cell adhesion recognition sequence in the first cadherin motif, consisting of the three amino acids histidine-alanine-valine (HAV) (Blaschuk et al., 1990). Blocking this sequence affects drastically the adhesion function of classical cadherins (Noe et al., 1999). But not only the HAV motif is important, also the flanking sequences seem to play a crucial role especially for the selectiveness of binding to the HAV motif (Renaud-Young and Gallin, 2002). And although the EC1 domain is supposed to be primarily responsible for adhesion activity, different studies with truncated C-cadherin constructs have been shown that the isolated EC1 domain is not capable to mediate cell-cell adhesion and a construct bearing EC1 and EC2 exhibits only faint adhesion activity, but an EC1-EC3-chimera is nearly fully adhesion-competent (Chappuis-Flament et al., 2001).

These findings can be explained either by the simple demand of a distinct distance between the active center of adhesion, localized on the EC1 motif and the cell surface, bearing various other proteins, which luminal domains might interact with calcium ions or adhesion sites of the cadherin. Another possibility is that discrete domains, which are not considered yet, play a crucial role in dimerization.

The close relation between classical cadherins is reflected by its amino acid sequence which is strikingly similar with 68 % to 78 % for the EC1, compared to E-cadherin. Moreover, the cytoplasmic tail is almost identical (69 % to 89 % similarity).

Type II cadherins, former denamed “atypical cadherin subfamily”, with cadherin 11 as prototype, are closely related to the classical cadherins. They too consist of five EC domains and show still sequence similarities (43 % to 50 %) to E-cadherin, but they do not bear the HAV-motif.

Cadherins present in desmosomes (see above) are arranged in two groups – desmocollins and desmogleins. They resemble E-cadherin in their amino acid sequence (50 % and 56 % similarity, respectively) as well as in structure but differ especially in the intracellular signalling and linkage to the cytoskeleton. One typical feature is the in comparison with classical cadherins prolonged cytoplasmic tail,

whereby they are linked to intracellular adaptor proteins and finally intermediate filaments of the cytoskeleton (see Huber, 2003 for review).

All other cadherins – except of protocadherins – are subsumed as cadherin related proteins with only a low sequence similarity (lee than 44 %) some examples are FAT-1, exhibiting 34 EC units, C-cadherin or the “Flamingo”-cadherins.

An extraordinary position is taken by T-cadherin (cadherin 13, H-cadherin), the only known cadherin which does not posses a transmembrane domain but is anchored to the membrane by a GPI anchor (Kuzmenko et al., 1998).

Finally the protocadherins – meanwhile the largest group of cadherin related molecules. They have been termed by one of the pioneers in cadherin research, Shintaro Suzuki, whose group discovered discrete fragments similar to cadherins by chance when he screened for more cadherins exhibiting the five EC domains by RT-PCR (Sano et al., 1993). As they were able to isolate mRNA of these proteins in a wide range of vertebrates and invertebrates, he proposed them to be a kind of ancestors for cadherins of the classical type and named them proto-cadherins (from Greek “protos” - the first). This hypothesis is derived from phylogenetic studies implying that the clustered protocadherins emerged from multiple gene duplications (Vanhalst et al., 2001).

However, investigations in the genomic sequences of minor developed species like Drosophila or C. elegans contradict this conclusion. For none of the 15 or 17 protocadherins expressed in fruit fly and worm a direct human analogue was identified (Hill et al., 2001).

Moreover, sequence similarities to the cytoplasmic tail of classical cadherins were revealed, leading Frank et al. to the conclusion that protocadherins evolved - contrary to the prior assumed hypothesis – rather late during evolution in chordates or early vertebrates (Frank and Kemler, 2002).

Most of the approximately 80 described members of the human protocadherin subfamily are arranged in three clusters, termed α, β, γ. The β-protocadherin comprises single exons, flanked by individual 5’-promoters, whereas α- and γ-protocadherin genes contain three additional downstream exons, coding for a

cluster-specific constant domain, represented by the cytosolic tail of the protein (for review see Frank and Kemler, 2002).

Almost all protocadherins are expressed in the nervous system (Kohmura et al., 1998) but there are also several proteins known to be expressed in other tissues – like the presented protocadherin LKC.

The structure of protocadherins is variable. Beside the amount of extracellular repeats varying from 4 (mµ-protocadherin, Goldberg et al., 2000) to 34 (FAT-1 cadherin, Mahoney et al., 1991), their cytoplasmic tail is unique. Although to a lower extent, most of the protocadherins tested exhibit a definite cell-cell adhesion activity (protocadherin 1, 12, Arcadlin, µ-protocadherin and VE-cadherin 2) which was basically calcium dependent (reviewed by Frank and Kemler, 2002). However, they do not share the conserved tryptophane residue described for classical cadherins.

Therefore, cell adhesion mediated by these proteins can not be explained by the common strand dimer model. But still discrete interfaces capable of cell aggregation need to be identified.

As mentioned before, classical cadherins exhibit a homophilic adhesion activity which was thought until recently to be established only between two molecules of the same protein. In contrast, younger studies showed that some cadherins are able to co-aggregate with other types of cadherins in heterologous complexes, e.g.

conglomerates of cells transfected with N-cadherin mixed up with R-cadherin-expressing cells (Shan et al., 2000). However, some combinations of cadherins do not exhibit this feature – E-cadherin for example will not form adhesive dimers with N-cadherin (Duguay et al., 2003). Further insights into the molecular mechanism underlying this phenomenon are discussed in 5.4 Oligomerization of members of the cadherin family.

Similar results have been obtained for class II cadherins. Some of them are also capable to interact with members of the same subfamily. Nevertheless, up to know it was not possible to prove any interaction between members of the different two subfamilies. This is an important precondition for the hypothesis that by expression of different types or clusters of cadherins, specific cell sorting and directed cell

migration can be achieved. Given this, the establishment of tissues composed by only one cell type or a specific cell targeting during embryogenesis might be explainable. For example, the separation of the neural tube from the ectoderm is supposed to be mediated by expression of N-cadherin (neuronal precursor cells) or E-cadherin (epithelial type), respectively.

Furthermore, the differentiation of skeletal muscle precursor cells (myoblasts) into mature cells (myotubes) is promoted by N-cadherin in a calcium dependent fashion in a cell culture model (George-Weinstein et al., 1997).

Particularly for the development of the brain it has been suggested that a directed expression of various protocadherins and their interaction with each other on juxtaposed cell or with the extracellular matrix plays a key role in the formation and maintaining of distinct compartments of the central nervous system. This assumption was arisen from the observation that γ-protocadherins are already present in early embryogenesis and are especially enriched in the synapses (Wang et al., 2002;Weiner et al., 2005).

This mechanism requires a very exact coordinated temporal and spatial regulation of the expression of different members of the cadherin family. Yet the underlying molecular mechanism has not been explored. One hint might be the clustering of protocadherins on three chromosomal loci (5q31, 13q21 and Xq21) – raising the possibility to be under the control of a complex network of interacting promoters.

Wang et al. (2002) generated knock out mouse missing the whole γ-protocadherin-cluster which contains 22 single genes. These animals show a decreased synapse density in the spinal cord and a significant loss of discrete subpopulations of spinal neurons by apoptosis late in embryogenesis. Interestingly, this happens exclusively to interneuronal cells and axonal growth, adhesion capacity and migration of neuronal cells of the spinal cord, hippocampus cortex and peripheral sensory neurons are not affected – although they too physiologically express high levels of γ-protocadherins.

This led the authors to the conclusion that γ-protocadherins or at least some members of this subfamily are required for the development of the nervous system by preventing apoptosis of the interneurons and not by direct affection of neuronal

cells. A different explanation could be that the missing γ-protocadherins are functionally compensated by protocadherins of the α- and β-cluster. However, this can only be accomplished to a distinct degree as the mice die in the early postnatal period.

2.4 Intracellular signalling of cadherins