Nectin and afadin: novel organizers of intercellular junctions

The cadherin superfamily plays key roles in intercellular adhesion. An emerging intercellular adhesion system, consisting of nectin and afadin, also has roles in organization of a variety of intercellular junctions either in cooperation with, or independently of, cadherin. Nectin is a Ca2+-independent immunoglobulin-like intercellular adhesion molecule, and afadin is a nectin- and actin-filament-binding protein that connects nectin to the actin cytoskeleton. This novel intercellular adhesion system has roles in the organization of E-cadherin-based adherens junctions and claudin-based tight junctions in epithelial cells. The adhesion system is furthermore involved in the formation of synapses in neurons and the organization of heterotypic junctions between Sertoli cells and spermatids in the testis.

Synapses are a specialized form of intercellular junction, and their specificity and plasticity provide neurons with the structural and functional basis for the formation of networks (Fig. 1B). At synapses, there are two types of junctions: synaptic junctions and puncta adherentia junctions (Peters et al., 1976). Synaptic junctions include presynaptic active zones, where synaptic vesicles and Ca 2+ channels localize, and post synaptic densities (PSDs), where neurotransmitter receptors localize. Puncta adherentia junctions by contrast are ultrastructurally similar to the AJs of epithelial cells. The homotypic intercellular junctions are formed by homophilic interactions between cell adhesion molecules (CAMs). This is in contrast to cell-matrix AJs, which form through heterophilic interactions between CAMs and matrix proteins.
In addition to the homotypic intercellular junctions described above, heterotypic intercellular junctions also exist. These include those formed between differentiating germ cells and their supporter Sertoli cells in the seminiferous epithelium in the testis and between specialized sensory cells and supporting cells in sensory epithelia. The seminiferous epithelium of the testis contains two types of intercellular junction: Sertoli-cell-Sertoli-cell junctions and Sertolicell-spermatid junctions (for a review, see Russell and Griswold, 1993) (Fig. 1C). Sertoli cells constitute the singlelayered epithelium and embrace and cultivate spermatogenic cells throughout their development (spermatogenesis). During the latter half of spermiogenesis, spermatids form prominent heterotypic intercellular junctions with Sertoli cells (Sertolicell-spermatid junctions), which are downregulated when spermatids are released as spermatozoa. Sertoli-cell-Sertoli cell junctions, by contrast, are homotypic and similar to those in epithelial cells.
We describe here an emerging intercellular adhesion system consisting of nectin, a Ca 2+ -independent immunoglobulin (Ig)like CAM, and afadin, a nectin-and actin filament (F-actin)-binding protein that connects nectin to the actin cytoskeleton. This novel adhesion system plays roles in the organization of all of these homotypic, interneuronal and heterotypic junctions.
Afadin has two splice variants: l-afadin and s-afadin (Mandai et al., 1997). l-Afadin, the larger splice variant, is a nectin-and F-actin-binding protein that has two Rasassociation (RA) domains, a forkhead-associated (FHA) domain, a DIL domain, a PDZ domain, three proline-rich (PR) domains and an F-actin-binding domain (Mandai et al., 1997;Takahashi et al., 1999). l-Afadin binds along the side of F-actin but not to the ends of F-actin, although it does not have crosslinking activity. s-Afadin, the smaller splice variant, lacks the F-actin-binding domain and the third proline-rich domain. Human s-afadin is identical to the gene product of AF-6, a gene that has been identified as an ALL-1 fusion partner involved in acute myeloid leukemias (Prasad et al., 1993). AF-6/s-afadin has been reported to bind directly to RYK, a receptor tyrosine kinase, and a subset of Eph receptor tyrosine kinases (Hock et al., 1998;Buchert et al., 1999;Halford et al., 2000), but Trivier et al. have recently shown that AF-6/s-afadin does not in fact bind to RYK (Trivier and Ganesan, 2002). AF-6/s-afadin has also been reported to interact with a deubiquitinating enzyme, Fam (Taya et al., 1998), but Chen et al. have found no genetic interaction between the Drosophila homologs of mammalian l-afadin and Fam during its eye development (Chen et al., 2000). Unless otherwise specified, afadin refers to l-afadin in this article.

Intercellular adhesion activity of nectins
E-Cadherin functions as a Ca 2+ -dependent CAM at AJs in epithelial cells (Takeichi, 1991;Gumbiner, 1996;Vlemincks and Kemler, 1999;Tepass et al., 2000;Angst et al., 2000;Yagi and Takeichi, 2000). It is a member of the cadherin superfamily, which consists of >80 members. E-Cadherin has an extracellular region containing five tandemly repeated domains (EC1-EC5), a single transmembrane region and a cytoplasmic region (Fig. 3A). In the current model for its intercellular adhesion activity, which is based on crystallographic analysis of the EC1 and EC2 domains of Ecadherin and N-cadherin and of the whole domains (EC1-EC5) of C-cadherin (Shapiro et al., 1995;Nagar et al., 1996;Tamura et al., 1998;Pertz et al., 1999;Boggon et al., 2002), cadherin first forms a cis-dimer, in which the monomers are aligned in a parallel orientation, and then a trans-dimer, in which cisdimers from opposing cell surfaces interact in an anti-parallel orientation. The formation of the cis-dimer appears to be essential for the formation of the trans-dimer (Tamura et al., 1998;Pertz et al., 1999).
A mutant of nectin lacking the second Ig-like domain does not form a cis-dimer, indicating that the second Ig-like domain Journal of Cell Science 116 (1) (A) AJs and TJs. At AJs, E-cadherin serves as an essential CAM. The cytoplasmic region binds β-catenin, which in turn binds α-catenin. α-Catenin is associated with the circumferential F-actin bundles directly and indirectly through vinculin and α-actinin. Nectin also functions as a CAM at AJs, but is more highly concentrated at AJs than E-cadherin. The cytoplasmic region binds afadin that is directly associated with the F-actin bundles. Afadin and α-catenin are associated with each other presumably through an unidentified molecule 'X'. At TJs, claudin and JAM function as CAMs. Occludin is another transmembrane protein at TJs. The cytoplasmic regions of claudin, JAM and occludin bind ZO-1, ZO-2 and ZO-3. ZO-1 and ZO-2 are directly associated with F-actin and form a dimer with ZO-3. Afadin and ZO-1 may be associated with each other through an unidentified molecule 'Y'. F-Actin and peripheral membrane proteins shown in the right-side cell are omitted in the left-side cell. (B) Synapses. At the synapses between mossy fiber terminals and pyramidal cell dendrites in the CA3 area of hippocampus, both synaptic junctions and puncta adherentia junctions are highly developed. At the puncta adherentia junctions, nectin-1 and nectin-3 localize asymmetrically at the presynaptic and postsynaptic sides, respectively. Afadin, N-cadherin and catenins localize symmetrically at the both sides. Synaptic junctions are associated with presynaptic active zones where synaptic vesicles, Ca 2+ channels and many other components, such as bassoon, localize and with PSDs where neurotransmitter receptors localize. F-Actin shown in the postsynaptic side is omitted in the presynaptic side. (C) Sertoli-cell-spermatid junctions. Nectin-2 in Sertoli cells and nectin-3 at spermatids form a hetero-trans-dimer at Sertoli-cell-spermatid junctions. The F-actin bundles surround the spermatid heads like parallel rings (3D view). The nectin-based adhesive membrane microdomains show one-to-one linkage with each F-actin bundle. ES contains not only F-actin bundles but also a flattened cistern that is connected to microtubules.

F-Actin
is necessary for the formation of the cis-dimer . Nevertheless, a fragment of the first Ig-like domain can form a cis-dimer (Krummenacher et al., 2002). This domain is thus also likely to be involved in the formation of the cisdimer (Fabre et al., 2002;Krummenacher et al., 2002). In addition, the first Ig-like domain is necessary for the formation of the trans-dimer Miyahara et al., 2000;Sakisaka et al., 2001;Reymond et al., 2001;Momose et al., 2002;Krummenacher et al., 2002). The first Ig-like domains of nectin-3 and nectin-4 bind to the same region of the first Iglike domain of nectin-1 to form the respective hetero-transdimers, although nectin-3 shows higher affinity than nectin-4 (Fabre et al., 2002). The function of the third Ig-like domain is currently unknown. Note that the interaction between nectin and afadin is not essential for the formation of the cis-dimer or the trans-dimer .

Roles of nectin and afadin in organization of AJs in fibroblasts
In cultured fibroblasts that lack TJs, such as L cells that stably express exogenous E-cadherin (EL cells), nectin-2 and afadin colocalize with E-cadherin at intercellular AJs (Mandai et al., 1997;Takahashi et al., 1999). Neither nectin-2 nor afadin is found at cell-matrix AJs (Mandai et al., 1997;. Several lines of evidence indicate that the nectinafadin and E-cadherin-catenin systems are physically and functionally associated and that the systems play cooperative roles in the organization of AJs (Fig. 4A). Studies using L cells that express exogenous full-length or mutant nectin-1 or nectin-2 and E-cadherin have revealed that nectin recruits βcatenin through α-catenin and E-cadherin through the αcatenin and β-catenin complex to nectin-based junctions . Moreover, a recombinant extracellular fragment of nectin-3 fused to the Fc domain of IgG (Nef-3) forms a trans-dimer with nectin-1 and thereby inhibits the formation of the nectin-1-based intercellular junctions in EL cells (Honda et al., 2003). This inhibits the formation of the E-cadherin-based intercellular junctions. Conversely, Nef-3 coated on microbeads first recruits the nectin-1-afadin complex and then the E-cadherin-catenin complex to bead-cell contact sites in EL cells that stably express exogenous nectin-1 (nectin-1-EL cells) (Honda et al., 2003). An extracellular fragment of E-cadherin fused to the Fc domain of IgG (Cef) coated on microbeads recruits not only the E-cadherin-catenin complex but also the nectin-1-afadin complex to the bead-cell contact sites in nectin-1-EL cells (Honda et al., 2003). Kinetically, the nectin-based junctions form first and their formation is followed by formation of the E-cadherin-based junctions.
How the nectin-afadin system is physically associated with the E-cadherin-catenin system is not known, but both afadin and α-catenin have been shown to be essential for the association of nectin and E-cadherin. The cytoplasmic tail of E-cadherin and β-catenin might be involved in, but is not essential for, this association. Afadin directly binds α-catenin in vitro, but this binding is not strong Pokutta et al., 2002). The direct binding of these proteins may occur in vivo, but it is more likely that a post-translational modification(s) of either or both proteins and/or an unidentified molecule(s) are required for the binding of α-catenin to afadin Pokutta et al., 2002). α-Catenin also binds to ZO-1 -an AJ component in fibroblasts (Itoh et al., 1991;Itoh et al., 1997;Imamura et al., 1999;Gumbiner, 2000;Nagafuchi, 2001). Furthermore, ZO-1 binds to afadin (Yamamoto et al., 1997). ZO-1 might thus be involved in the association of nectin and E-cadherin, but this possibility is unlikely given that the direct binding of afadin to ZO-1 has not been reproduced Yokoyama et al., 2001) and nectin recruits ZO-1 to the nectinbased junctions through afadin in an α-catenin-independent manner . Ponsin, an afadin-and vinculin-binding protein , and vinculin colocalize with nectin and afadin at intercellular AJs and furthermore localize to cell-matrix AJs . Vinculin directly binds to α-catenin (Tsukita et al., 1992;Aberle et al., 1996;Gumbiner, 2000;Nagafuchi, 2001). Vinculin and ponsin might thus also be involved in the association of nectin and E-cadherin. However, ponsin forms a binary complex with either afadin or vinculin and does not form a ternary complex -at least in a cell-free assay system . Ponsin and vinculin are not essential for the association of nectin and E-cadherin .
Several lines of evidence indicate that, as in fibroblasts, the nectin-afadin and E-cadherin-catenin systems are physically and functionally associated in epithelial cells. Firstly, during the formation of the junctional complex involving AJs and TJs in Madin Darby canine kidney (MDCK) cells that stably express exogenous nectin-1 (nectin-1-MDCK cells), Ecadherin is recruited to the nectin-1-based junctions where afadin colocalizes (Honda et al., 2003). Moreover, Nef-3 inhibits the formation of nectin-1-based intercellular junctions and thereby the recruitment of E-cadherin, and this inhibits formation of E-cadherin-based AJs in wild-type and nectin-1-MDCK cells (Honda et al., 2003). In the latter, Nef-3 coated on microbeads first recruits the nectin-1-afadin complex and then the E-cadherin-catenin complex to the bead-cell contact sites (Honda et al., 2003).
Further evidence is provided by the gastric cancer cell line, HSC-39. These cells express E-cadherin but do not form intercellular junctions (Yanagihara et al., 1991). Overexpression of nectin-2 or nectin-3 leads to intercellular adhesion (Peng et al., 2002). In keratinocytes an N-cadherin mutant that lacks the extracellular region acts as a dominant negative mutant and reduces the E-cadherin-based adhesion activity (Fujimori and Takeichi et al., 1993). Overexpression of nectin-3 reverses this inhibitory action of the N-cadherin mutant (Y. Tanaka, H. Nakanishi, S. Kakunaga et al., unpublished). Overexpressed nectin-3 recruits the N-cadherin mutant to the nectin-3-based junctions. The N-cadherin mutant probably associates with endogenous nectin and therefore prevents it from associating with endogenous Ecadherin.
Afadin-deficient mice show developmental defects at stages around gastrulation, including disorganization of the ectoderm, impaired migration of the mesoderm, and loss of somites and other structures that are derived from both the ectoderm and the mesoderm (Ikeda et al., 1999;Zhadanov et al., 1999). Furthermore, cystic embryoid bodies derived from afadindeficient embryonic stem cells show disorganization of the ectoderm (Ikeda et al., 1999). In the ectoderm of afadindeficient mice and embryoid bodies, the organization of AJs is highly impaired. Again, these results strongly implicate the nectin-afadin system in the organization of intercellular junctions in these tissues. Single-molecule image analysis of nectin-2 and E-cadherin in mouse mammary tumor (MTD-1A) cells indicates that nectin-2 moves more rapidly than Ecadherin on the free surface of the plasma membrane (Katsuno et al., manuscript in preparation). These results suggest that the nectin-afadin and E-cadherin-catenin systems cooperatively organize AJs. The molecular mechanism by which the nectinafadin and E-cadherin-catenin systems associate in epithelial cells thus appears to be similar to that in fibroblasts but has not fully been elucidated.
At TJs, three CAMs have thus far been identified (Fig. 4A). Claudin functions as a Ca 2+ -independent CAM Tsukita et al., 2001) and constitutes a superfamily consisting of over 20 members. Occludin is another transmembrane protein at TJs, but its function has not yet been established. Junctional adhesion molecule (JAM) represents a Ca 2+ -independent Ig-like CAM family of at least three members at TJs (Martin-Padura et al., 1998;Ozaki et al., 1999;Palmeri et al., 2000;Liang et al., 2002). Claudin and occludin have four transmembrane regions, whereas JAM has an extracellular region that has two Ig-like domains, a single transmembrane region and a cytoplasmic region. Claudin, occludin and JAM are linked to the actin cytoskeleton through peripheral membrane proteins, including ZO-1, ZO-2 and ZO-3 Tsukita et al., 2001;Bazzoni et al., 2000;Ebnet et al., 2000). ZO-1 and ZO-2 are F-actin-binding proteins and form a dimer with ZO-3 Tsukita et al., 2001). The association of these CAMs with the actin cytoskeleton has been proposed to strengthen the intercellular adhesion of TJs Tsukita et al., 2001). Although the formation of TJs is dependent on the formation of E-cadherin-based AJs (Takeichi, 1991;Gumbiner, 1996;Vlemincks and Kemler, 1999), accumulating evidence suggests that the nectin-afadin system plays a role in the organization of TJs. During the formation of the junctional complex of AJs and TJs in nectin-1-MDCK cells, claudin, occludin and JAM are recruited to the apical side of the nectin-1-based junctions, where both afadin and ZO-1 colocalize (Fukuhara et al., 2002a;Fukuhara et al., 2002b). After they are recruited, ZO-1 translocates from the nectin-1-based junctions to their apical side. Recruitment of claudin, occludin and JAM is inhibited by Nef-3 (Fukuhara et al., 2002a;Fukuhara et al., 2002b). Nef-3 coated on microbeads recruits not only the nectin-1-afadin complex but also ZO-1 and JAM to the bead-cell contact sites in nectin-1-MDCK cells (Fukuhara et al., 2002a). Nectin-2 recruits both α-catenin and ZO-1 at the same time to the nectin-2-based intercellular adhesion sites in L cells that stably express exogenous nectin-2 (nectin-2-L cells) . This recruitment requires afadin but not α-catenin, ponsin or vinculin. In the ectoderm of afadin-deficient mice and embryoid bodies, organization of not only AJs but also TJs is highly impaired (Ikeda et al., 1999). Furthermore, mutations in the nectin-1 gene are responsible for cleft lip/palate ectodermal dysplasia -Margarita island ectodermal dysplasia and Zlotogora-Ogür syndrome -which is characterized by cleft lip/palate, syndactyly and ectodermal dysplasia (Suzuki et al., 2000;Sozen et al., 2001). These results suggest that the nectinafadin system plays a role in the organization of TJs. It is unknown how the nectin-afadin system regulates the organization of TJs, but ZO-1, ZO-2 and ZO-3 associated with the nectin-afadin system may play a role in recruiting JAM, claudin and occludin. It also remains unknown whether JAM is involved in the localization of claudin and occludin for organization of TJs.
At the initial stage of formation of AJs and TJs, primordial spot-like junctions first form at the tips of the cellular protrusions that radiate from adjacent cells (Yonemura et al., 1995;Adams et al., 1998;Vasioukhin et al., 2000). Components of the E-cadherin-catenin and nectin-afadin systems, ZO-1 and JAM colocalize to these spot-like junctions, at which neither claudin or occludin is concentrated (Ando-Akatsuka et al., 1996;Asakura et al., 1999;Sakisaka et al., 1999;Ebnet et al., 2001). The spot-like junctions begin to fuse to form short line-like junctions, at which claudin and occludin do accumulate. Although the precise timing of arrival of each component at the junctional complex during its formation is unclear, we can propose the following speculative model for formation of the junctional complex.
In our model, all CAMs, including nectin, E-cadherin, JAM, claudin and occludin, are diffusely distributed on the free surface of the plasma membranes of migrating cells (Fig. 5A). When the two migrating cells contact through protrusions such as filopodia and lamellipodia, nectin and E-cadherin separately form trans-dimers that form micro-clusters at intercellular contact sites. Because, kinetically, nectin forms micro-clusters more rapidly than E-cadherin, the nectin-based micro-clusters are mainly formed at the initial stage. The nectin-based microclusters then recruit E-cadherin, which results in the formation of a mixture of nectin-and E-cadherinbased microclusters. The nectin and E-cadherin molecules in these microclusters are associated through afadin and catenins that are linked to the actin cytoskeleton. E-Cadherin-based micro-clusters that form slowly and independently of the nectin-based microclusters rapidly recruit the nectin-afadin complex to form other primordial spot-like junctions. These primordial junctions fuse with each other to form short linelike junctions, which develop into more matured AJs. During the formation of AJs, JAM is first assembled at the apical side of AJs, and this is followed by the recruitment of claudin and occludin presumably through ZO-1, ZO-2 and ZO-3, which eventually leads to the establishment of claudin-based TJs (Fig. 5B).

Roles of nectin and afadin in organization of synapses
At the synapses between mossy fiber terminals and pyramidal cell dendrites in the CA3 area of the hippocampus, both synaptic junctions and puncta adherentia junctions are highly developed and actively remodeled in an activity-dependent manner (Amaral and Dent, 1981). Nectin-1 and nectin-3 localize asymmetrically at the presynaptic and postsynaptic sides, respectively, of the plasma membranes of the puncta adherentia junctions and form hetero-trans-dimers (Mizoguchi et al., 2002) (Fig. 4B). Afadin, N-cadherin and αN-catenin localize symmetrically at both sides (Nishioka et al., 2000;Mizoguchi et al., 2002). s-Afadin, but not l-afadin, localizes at PSDs (Buchert et al., 1999;Nishioka et al., 2000). In at least some brain regions other than the CA3 area of the hippocampus, such as the optic tectum, N-cadherin localizes at both synaptic junctions and puncta adherentia junctions (Yamagata et al., 1995).
Studies using antibodies to N-cadherin and antagonistic peptides indicate that N-cadherin plays a role in the organization of synapses (Yamagata et al., 1995;Tang et al., 1998). Inhibition of nectin-1-and nectin-3-based adhesion by an inhibitor of nectin-1 (gD) in cultured rat hippocampal neurons results in a decrease in size and a concomitant increase in the number of synapses (Mizoguchi et al., 2002). The exact mechanism by which this occurs remains to be clarified but reflects partial inhibition of the formation of the hetero-transdimer between nectin-1 and nectin-3, which may affect Ncadherin-mediated adhesion and eventually lead to formation of smaller synapses. Why the number of synapses increases is not known either but may be due to a failure of neurons to determine the proper positions of synapses or compensation for functionally less competent smaller synapses. Thus, it is likely that the formation of the hetero-trans-dimer between nectin-1 and nectin-3 plays an important role in the determination of the position and the size of synapses -in cooperation with the Ncadherin-catenin system. This role of the nectin-afadin system is consistent with the finding that mutations in the nectin-1 gene are responsible for cleft lip/palate ectodermal dysplasia -Zlotogora-Ogür syndrome -which is characterized by mental retardation in addition to ectodermal dysplasia (Suzuki et al., 2000). Synapses are formed by the meeting of axons and dendrites during their maturation. At primitive synapses, synaptic junctions and puncta adherentia junctions are not morphologically differentiated, but during the maturation of synapses membrane domain specialization gradually occurs (Amaral and Dent, 1981). This neural membrane domain specialization appears to be similar to that found during formation of the junctional complex in epithelial cells with respect to the dynamic localization patterns of the junctional proteins. It is postulated that primordial junctions form first, followed by transport of the components of active zones on dense core vesicles and subsequent formation of active zones at the presynaptic side (for reviews, see Desbach et al., 2001;Ziv and Garner, 2001). At the postsynaptic side, the components of PSDs are assembled, and membrane receptors on vesicles are transported to this region. We imagine that, in synaptogenesis, the nectin-afadin unit first forms primordial junctions, and this is followed by the recruitment of the Ncadherin-catenin unit. The components of active zones would then be recruited to the primordial junctions to form active zones at the presynaptic side. At the postsynaptic side, the components of PSDs would be assembled, and membrane receptors would be recruited. The membrane domains, comprising synaptic junctions and puncta adherentia junctions, would then gradually become segregated, and this would be followed by maturation of synapses.

Roles of nectin and afadin in organization of Sertolicell-spermatid junctions
Sertoli cells form a unique type of F-actin-based junctional complex referred to as the ectoplasmic specialization (ES) (for reviews, see Russell and Griswold, 1993;Vogl et al., 2000). ES contains F-actin bundles, which are arranged at regular intervals beneath the plasma membrane, and a flattened cistern of the endoplasmic reticulum, which is connected to microtubules (Fig. 4C). This is supposed to function as a scaffold that stabilizes an adhesive domain in the plasma membrane of Sertoli cells. Vinculin is concentrated at Sertoli-cell-Sertoli-cell junctions and Sertolicell-spermatid junctions (Grove and Vogl, 1989;Pfeiffer and Vogl, 1991). Since the ES is formed only in Sertoli cells, Sertoli-cell-spermatid junctions are asymmetric. Moreover, unlike at typical AJs, the existence of the cadherin-catenin system has been questioned at Sertoli-cell-spermatid junctions (Byers et al., 1991;Cyr et al., 1992;Anderson et al., 1994). Thus, little is known about the intercellular CAMs responsible for Sertoli-cell-spermatid junctions. In contrast to Sertoli-cell-spermatid junctions, Sertoli-cell-Sertoli-cell junctions are equipped with TJs as well as AJs, which serve as the 'blood-testis' barrier.
Nectin-2, nectin-3 and afadin colocalize with the F-actin that underlies Sertoli-cell-spermatid junctions   (Fig. 4C). Nectin-2 and nectin-3 reside specifically in Sertoli cells and spermatids, respectively, which suggests the formation of a hetero-trans-dimer from nectin-3 in spermatids and nectin-2 in Sertoli cells. The nectin-based adhesive membrane microdomains show one-to-one linkage with each F-actin bundle at Sertoli cell -spermatid junctions. Nectin-2 and afadin localize at Sertoli-cell-Sertoli-cell junctions. Nectin-2-deficient mice exhibit male-specific infertility and have defects in the later steps of sperm morphogenesis, including distorted nuclei and an abnormal distribution of mitochondria (Bouchard et al., 2000). In these mice, the structure of Sertoli-cell-spermatid junctions is severely impaired, and the localization of nectin-3 and afadin is disorganized, whereas Sertoli-cell-Sertoli-cell junctions are apparently normal . The loosened adhesion and the lack of an F-actin scaffold due to mislocalization of afadin at Sertoli-cell-spermatid junctions may act together to render the contact site weak and convoluted and also produce the drastic condensation of spermatid nuclei.
On the basis of these observations, we propose a model for the molecular organization of Sertoli-cell-spermatid junctions in which nectin-2 on the Sertoli cell membrane forms a heterotrans-dimer with nectin-3 on the spermatid membrane to form discrete adhesive membrane domains (Fig. 4C). In this model, the cytoplasmic region of nectin-2 in Sertoli cells binds to afadin to connect F-actin bundles to the membrane. The localization of afadin in spermatids remains unknown. It is likely that Sertoli-cell-spermatid junctions rely largely on the nectin-afadin system, whereas Sertoli-cell-Sertoli-cell junctions are formed through cooperation of multiple intercellular adhesion systems.

Role of nectin as a receptor for α-herpes viruses for entry and intercellular spreading
Nectin-1 was originally isolated as one of the poliovirusreceptor-related proteins and named PRR-1 . Nectin-2 was originally isolated as the murine homolog of human PVR (Morrison and Racaniello, 1992) but turned out to be another poliovirus-receptor-related protein and was named PRR-2 . Neither PRR-1 nor PRR-2 has thus far been shown to serve as a PVR. They were later shown to serve as receptors for α-herpes viruses, facilitating their entry and intercellular spreading, and renamed HveC and HveB, respectively (Cocchi et al., 1998;Cocchi et al., 2000;Geraghty et al., 1998;Warner et al., 1998;Sakisaka et al., 2001). It remains unknown whether nectin-3 and nectin-4 serve as receptors for viruses. Human nectin-1 allows entry of all αherpes viruses tested so far, including herpes simplex virus (HSV) type 1, HSV type 2 and pseudorabies virus (Spear et al., 2000;Campadelli-Fiume et al., 2000). Human nectin-2 can mediate entry of a restricted number of α-herpes viruses (Spear et al., 2000;Campadelli-Fiume et al., 2000). The usual manifestations of HSV disease are mucocutaneous lesions. HSV disease establishes latent infection of neurons in sensory ganglia and causes recurrent lesions at the sites of primary infection. In HSV disease, the intercellular spreading significantly contributes to the pathogenesis. The underlying mechanisms for entry and intercellular spreading of HSVs through nectin-1 are not fully understood, but at least four viral glycoproteins, gD, gB, and the gH-gL heterodimer, participate in the entry of HSV-1 (Spear et al., 2000;Campadelli-Fiume et al., 2000). Recombinant gD binds to nectin-1 in host cells and inhibits HSV-1 infection, which indicates that gD serves as a viral component that specifically interacts with its cellular receptor, nectin-1. The first Ig-like domain of nectin-1 is sufficient for the binding of gD and entry of HSV-1, and the second and third Ig-like domains increase the efficiency of entry (Cocchi et al., 1998). The interaction of nectin-1 with afadin increases the efficiency of intercellular spreading, but not the entry, of HSV-1 (Sakisaka et al., 2001). The Ecadherin-catenin system increases efficiency of both entry and intercellular spreading of HSV-1, which provides an additional line of evidence that nectin and E-cadherin are physically and functionally associated (Sakisaka et al., 2001).

Conclusions and perspectives
We have described here the roles of nectin and afadin in the organization of a variety of intercellular junctions in cooperation with, or independently of, cadherin. However, the detailed molecular mechanisms underlying the connection between the nectin-afadin system and other intercellular adhesion systems have not yet been fully elucidated. There may be other unidentified molecules that connect these different intercellular adhesion units, or post-translational modifications, such as protein phosphorylation, of the connectors. Elucidation of the molecular linkage between the nectin-afadin and other systems is essential for our understanding of the whole picture of the intercellular junctions. The nectin-afadin system may be involved in the dynamic formation and disruption of various types of intercellular junction and in the concentration of many biologically active molecules at intercellular junctions. In addition, the role of the nectin-afadin system at Sertolicell-spermatid junctions raises the possibility that this system is involved in many other heterotypic junctions between different types of cell. Since the nectin-based junctions form more rapidly than cadherin-based junctions, the nectin-afadin system may play a role in cell recognition. Elucidation of these unresolved issues will give us deeper insights into the molecular linkage between intercellular junctions and various cell functions, such as morphogenesis, differentiation, proliferation and migration, and also into the molecular mechanisms that underlie many human diseases arising from junctional disorders, such as cancer and vascular and mental diseases.