Protocadherin family: diversity, structure, and function Hirofumi Morishita1 and Takeshi Yagi2 Protocadherins are predominantly expressed in the nervous in the cadherin superfamily [3]. The protocadherin sub- system, and constitute the largest subgroup within the cadherin group has only been identified and characterized in superfamily. The recent structural elucidation of the amino- studies from the past decade. These genetic and func- terminal cadherin domain in an archetypal protocadherin tional studies have revealed a divergent cytoplasmic revealed unique and remarkable features: the lack of an domain, as well as six or seven extracellular cadherin interface for homophilic adhesiveness found in classical (EC) domains with low sequence similarities to the EC cadherins, and the presence of loop structures specific to the domains of the classical cadherin subgroup. protocadherin family. The unique features of protocadherins extend to their genomic organization. Recent findings have Knowledge of protocadherin family at the molecular level revealed unexpected allelic and combinatorial gene regulation has increased profoundly in the past few years from for clustered protocadherins, a major subgroup in the publications of the first detailed structural and functional protocadherin family. The unique structural repertoire and findings, as well as identification of unusual gene regu- unusual gene regulation of the protocadherin family may lation for this family. In the following, we shall contex- provide the molecular basis for the extraordinary diversity of the tualize the emergent insights into the structure and nervous system. function of the protocadherin family that have accumu- Addresses lated in the findings from the past ten years. 1 Division of Neuroscience, Children’s Hospital Boston, Harvard Medical School, 320 Longwood Ave., Boston, MA 02115, USA Diversity and classification 2 KOKORO-Biology Group, Laboratories for Integrated Biology, Graduate School of Frontier Biosciences, Osaka University, 1-3 The term ‘protocadherin’, originally introduced about a Yamadaoka, Suita, Osaka 565-0871, Japan decade ago, is now used in various ways after the recent expansion of the molecular family [4–7]. Our subsequent Corresponding authors: Morishita, Hirofumi use of the term shall be based on phylogenic study of ([email protected]) and Yagi, Takeshi protein sequences consisting each cadherin domains. ([email protected]) More specifically, our references to protocadherins will exclude the Flamingo/CELSER subfamily, which con- Current Opinion in Cell Biology 2007, 19:584–592 tains seven-pass transmembrane proteins, and FAT sub- This review comes from a themed issue on family, whose elements consist of far more than seven EC Cell to cell contact and extracellular matrix domains [3,8]. Currently, within the protocadherin Edited by Lawrence Shapiro and Barry Honig family, more than 70 different protocadherin (Pcdh) genes have been identified, The protocadherin family can be largely divided into two groups based on their genomic 0955-0674/$ – see front matter structure: clustered protocadherins, and nonclustered # 2007 Elsevier Ltd. All rights reserved. protocadherins [8](Figure 1). Clustered protocadherins DOI 10.1016/j.ceb.2007.09.006 are consist of the Pcdha, b,andg family, each of which has a specific genomic organization clustered in a small gen- ome locus [9]. With over 50 members, the clustered Introduction protocadherin constitutes the largest subgroup within The recent explosion in genome sequencing has revealed the protocadherin family. The Pcdha was originally dis- the impressive diversity of the cadherin superfamily. To covered in the mouse brain, and has been referred to as date, more than 100 different cadherins have been ident- the cadherin-related neuronal receptor (CNR) [5]. As ified. Originally characterized as calcium-dependent cell other protocadherins do not have a specific clustered adhesion molecules, cadherin molecules are now known genome locus, here we shall collectively refer to them to be involved in many biological processes, including cell as nonclustered protocadherins. Nonclustered protocad- recognition, cell signaling during embryogenesis, and the herins can be divided into two subgroups: Pcdhd, and formation of neural circuits [1,2]. Cadherins are identified solitary protocadherins in the phylogenic tree (Pcdh-12, by the presence of cadherin sequence repeats of about 15, 20, 21)[8]. The Pcdhd family comprises at least nine 110 amino acids. Several subgroups of cadherins can be protocadherins, all of which contain highly conserved defined based on shared properties and sequence sim- motifs (CM1, CM2) in their cytoplasmic domains. Mem- ilarity: the classical (type I) and closely related type II bers of Pcdhd subfamily can be further divided in two cadherins, desmosomal cadherins, and protocadherins subgroups, Pcdhd1 and d2, on the basis of overall (Figure 1). Protocadherins are predominantly expressed homology, number of EC repeats (seven versus six), in the nervous system, and constitute the largest subgroup and conservation of specific amino acid motifs in Current Opinion in Cell Biology 2007, 19:584–592 www.sciencedirect.com Protocadherin family Morishita and Yagi 585 Figure 1 Classification of cadherin superfamily and protocadherin family. (a) Classification and schematic diagram of cadherin superfamily. Classical type I cadherins have a conserved tryptophan (W2) in EC1 domain, and a hydrophobic pocket to accommodate W2 of the other EC1, which are crucial for homophilic adhesiveness. The prodomain will be removed to mediate functional adhesion. Highly related type II cadherins are slightly different from type I cadherins in that they have two conserved tryptophan residues (W2 and W4) and the hydrophobic pocket are correspondingly extensive. The cytoplasmic regions of classical cadherins have catenin binding site which links to the actin cytoskeleton. Desmosomal cadherins are similar to type I cadherins in that they have a conserved W2 and five EC domains, but have distinctive cytoplasmic regions. The protocadherin family is strikingly different from other cadherins in that they neither have W2 nor hydrophobic pocket but they have characteristic disulfide-bonded loop in EC1 domain. Their cytoplasmic regions do not have catenin binding site. The protocadherin family can be divided into two subgroups; clustered, and nonclustered protocadherin based on the genomic organization. Clustered protocadherins have six EC domains. Nonclustered protocadherins have various number of EC domains. Proteins that contain an identifiable cadherin-like domain have been loosely referred to others. (b) Classification of mouse protocadherin family. Protocadherin family can be divided into two groups based on their genomic structures: the clustered, and nonclustered protocadherins. Clustered Protocadherins are consisted of Pcdha, b and g families, which are clustered in a small genome locus. Nonclustered protocadherins can be divided into two subgroups, Pcdhd family, and other solitary protocadherins. All Pcdhd contains highly conserved motifs (CM1, CM2) in their cytoplasmic domains. They can be further divided in two subgroups, d1 and d2. Number of members for each group is also mentioned in the figure. cytoplasmic domains. Nine protocadherins have been interfaces of classical cadherin has been shown to be classified from existing research: Pcdh-1, 7, 9, and 11(X/ localized primarily within the amino-terminal EC1 domain Y) in the d1 subgroup, and Pcdh-8, 10, 17, 18, and 19 in the [11,12]. The new report of the solution protein-structure d2 subgroup. of EC1 domain of protocadherin (CNR/Pcdha4), which was determined by NMR, assisted the elucidation of Protein structure and adhesive property the character of protocadherin proteins compared with In the case of classical cadherins, structural biology has classical cadherins at an atomic level [13,14]. Despite provided fruitful insights to reveal the molecular basis for low sequence similarities between the EC1 domains of cell adhesion [10]. The homophilic adhesive binding CNR/Pcdha4 and of classical cadherins (30% at the www.sciencedirect.com Current Opinion in Cell Biology 2007, 19:584–592 586 Cell to cell contact and extracellular matrix maximum), the overall topology was similar to classical side chain of W2. The lack of a hydrophobic pocket cadherins: a b-sandwich-like structure composed of two suggests that the homophilic adhesion interface that is packed b-sheets (Figure 2). By contrast, the interconnect- important for classical cadherins does not exist in proto- ing loops of protocadherins were quite divergent from the cadherins [14]. classical cadherins, both in length and in biochemical structure. The structural characterization of the protocadherin EC1 domain also revealed crucial variations mainly in the loop A remarkable finding in the structural elucidation was regions, including the protocadherin-specific disulfide that the CNR/Pcdha4 EC1 domain does not have trypto- bonded Cys-(X)5-Cys motif, and the RGD (Arg-Gly- phan (W2) or a hydrophobic pocket, which is essential for Asp) motif [14 ](Figure 2). The Cys-(X)5-Cys sequence the adhesiveness of the classical cadherins [11,14] is well conserved among the clustered and non clustered (Figures 2 and 3). Among the classical, type II, and protocadherins, but not in classical cadherins. Structural desmosomal cadherins, the amino acid residues consti- analysis of the amino acid sequence revealed that the tuting the