Roles of cell-adhesion molecules nectin 1 and nectin 3 in ciliary body development

Cells in multicellular organisms form cell-cell junctions and contacts that play essential roles in various cellular processes, including morphogenesis,differentiation, proliferation and migration. Cell-cell junctions and contacts are mediated by cell adhesion molecules (CAMs). The cadherin superfamily,which consists of over 80 members, serves as key Ca²⁺-dependent CAMs in a variety of cell-cell junctions(Takeichi, 1995; Yagi and Takeichi, 2000). The immunoglobulin (Ig) superfamily also plays important roles as Ca²⁺-independent CAMs(Brummendorf and Lemmon, 2001). Recently, nectins (Pvrl – Mouse Genome Informatics) have emerged as Ig-like CAMs that play roles in a variety of cell-cell junctions and contacts(Takai et al., 2003a; Takai et al., 2003b; Takai and Nakanishi, 2003). Nectins comprise a family of four members: nectin 1, nectin 2, nectin 3 and nectin 4. Extracellular regions of all nectins form homo-cis-dimers and then promote homophilic or heterophilic trans-interactions. In epithelial cells and fibroblasts in culture, nectins initiate cell-cell adhesion and recruit cadherins to the nectin-based cell-cell adhesion sites to cooperatively form adherens junctions (AJs) (Takai et al.,2003a; Takai and Nakanishi,2003). Furthermore, nectins recruit proteins of tight junctions(TJs), first junctional adhesion molecules (JAMs) and then claudins(Tsukita et al., 1999; Tsukita et al., 2001), to the apical side of AJs in cooperation with cadherins. JAMs are Ca²⁺-independent Ig-like CAMs that recruit the cell-polarity protein complex, which consists of Par3, aPKC and Par6, by directly binding Par3 (Ohno, 2001). Nectin 1 and nectin 3, but not nectin 2, also directly bind Par3. In addition,intracellular tails of nectins are associated with the actin cytoskeleton through afadin (Mllt4 – Mouse Genome Informatics), a nectin and actin filament (F-actin)-binding protein. Furthermore, nectins induce activation of Cdc42 and Rac small G proteins, which eventually enhances the formation of cadherin-based AJs through the reorganization of the actin cytoskeleton. Cdc42 activated by nectins may also bind to Par6 and activate the polarity protein complex (Ohno, 2001; Takai et al., 2003a). Thus,these mechanisms involving nectins are essential to form stable junctional complexes and cell polarity.

Physiological roles of nectins have also been demonstrated in neurons. At the synapses between the mossy fiber terminals and the pyramidal cell dendrites in the CA3 area of hippocampus of the brain, 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 of the puncta adherentia junctions,respectively (Takai et al.,2003b). As N-cadherin symmetrically localizes at the synapse, both nectins and N-cadherin are likely to cooperate in the formation of the puncta adherentia junctions. At the contacts formed between commissural axons and the processes of floor plate cells in the neural tube, nectin 1 and nectin 3 asymmetrically localize as key CAMs at the commissural axon side and the floor-plate cell side, respectively (Okabe et al., 2004b). Nectin 1 and nectin 3 play an important role in the trajectory of the commissural axons, whereas cadherins do not serve as CAMs. Nectins have furthermore been demonstrated to play roles in the testis. Nectin 2 and nectin 3 reside specifically in Sertoli cells and spermatids,respectively (Takai and Nakanishi,2003). Nectins, but not cadherins, are at least major CAMs at the Sertoli cell-spermatid junctions. Thus, evidence is accumulating that nectins play important roles in a variety of cell-cell junctions and contacts.

During mouse development, nectins and afadin are expressed in polarized epithelia, such as neuroepithelia, epithelial somites and facial primordia,which are dynamically remodeling (Okabe et al., 2004a). To explore their roles in dynamic epithelial remodeling during development, knockout studies on nectins and afadin are essential. Thus far, afadin^–/– and nectin 2^–/– mice are available. Afadin^–/– mice have revealed crucial roles of the whole nectin-based cell-adhesion system by showing embryonic lethality with severe defects in the formation of AJs and subsequent epithelial morphogenesis(Ikeda et al., 1999). They show developmental defects at stages during and after gastrulation, including disorganization of the ectoderm, impaired migration of the mesoderm, and loss of somites and other structures derived from both the ectoderm and the mesoderm (Ikeda et al., 1999). Nectin 2^–/– mice exhibit the male-specific infertility phenotype and have defects in the later steps of sperm morphogenesis,exhibiting distorted nuclei and abnormal distribution of mitochondria(Bouchard et al., 2000; Mueller et al., 2003; Ozaki-Kuroda et al., 2002). In these mice, the structure of the Sertoli cell-spermatid junctions is severely impaired, and the localization of nectin 3 and afadin is disorganized(Ozaki-Kuroda et al., 2002). Thus, it is likely that the Sertoli cell-spermatid junctions rely largely on nectins. In addition, spermatozoa of nectin 2^–/– mice have a defect in binding to the zona pellucida and oocyte penetration(Mueller et al., 2003).

Physiological roles of other members of nectins in dynamic epithelial remodeling during development, however, have not been established. Specifically, knockout studies on nectin 1 and nectin 3 are likely to provide some insights into mouse development because the most potent heterophilic trans-interaction is detected between nectin 1 and nectin 3(Takai et al., 2003a; Takai and Nakanishi, 2003). Here, we report an unexpected ocular phenotype of nectin 1^–/– and nectin 3^–/– mice,demonstrating that the heterophilic trans-interaction between nectin 1 and nectin 3 mediates the apex-apex adhesion between the pigment and non-pigment cell layers of the ciliary epithelia, and that this nectin-mediated adhesion is essential for the morphogenesis of the ciliary body.

C57BL/6 and BALB/cA mice were purchased from CLEA Japan (Tokyo, Japan). The animals and procedures used in this study were in accordance with the guidelines and approval of Osaka University Medical School Animal Care and Use Committee. Targeting constructs of nectin 1 and nectin 3 were made to replace the coding exon 2 (amino acids 26-143 of the nectin 1 protein) and the coding exon 1 (amino acids 1-58 of the nectin 3 protein), respectively, with a neo-resistance gene cassette (Fig. 1A, part a; Fig. 1B, part a). The neo-resistance gene cassette contained the exogenous thymidine kinase promoter. RW4 embryonic stem (ES) cells were transfected with these targeting vectors and selected as described previously(Ikeda et al., 1999). Homologous recombinants were verified by Southern blotting using 5′ and 3′ external probes and the neo-resistance gene probe. Nectin 1^+/– and nectin 3^+/– ES cells were microinjected into embryonic day 3.5 (E3.5) C57BL/6 blastocysts and transferred to MCH pseudo-pregnant foster mothers to generate chimeras that were mated with BDF1 mice for germline transmission. Genotyping was performed by Southern blotting and PCR. For Southern blotting, the following fragments were used: for the nectin 1 gene, a 1.2 kb EcoRI-HindIII fragment (5′-probe) and a 0.4 kb KpnI fragment(3′-probe); and for the nectin 3 gene, a 1.1 kb HindIII fragment (5′-probe) and a 1.2 kb BamHI-EcoRV fragment(3′-probe). For PCR, the following primer sets were used: for the nectin 1 gene, 5′-CCGTAAAGGTCAAGGGCAGAG-3′ and 5′-GTGCCTGTCCCTTGTCCA-3′; for the nectin 3 gene,5′-CTGCTGCTGCTGCTTATTCCC-3′ and 5′-AACCTCAGCCTAGAAGTCCGC-3′; and for the neo gene,5′-CTGTTGTGCCCAGTCATAGCC-3′ and 5′-CACTGAAGCGGGAAGGGACTG 3′. Nectin 1^–/– and nectin 3^–/– albino mice were generated by a cross of nectin 1^–/– and nectin 3^–/– mice with BALB/cA mice twice.

Rabbit anti-nectin 1 and nectin 3 polyclonal antibodies, which recognize the intracellular regions of nectin 1 and nectin 3, respectively, were prepared as described (Takahashi et al.,1999; Satoh-Horikawa et al.,2000). Rat anti-nectin 1, nectin 2 and nectin 3 monoclonal antibodies, which recognize the extracellular regions of nectin 1 (#48-12),nectin 2 (#502-57) and nectin 3 (#103-A1), respectively, were prepared as described (Mizoguchi et al.,2002; Satoh-Horikawa et al.,2000; Takahashi et al.,1999). A rabbit anti-afadin polyclonal antibody was prepared as described (Mandai et al.,1997). A mouse anti-ZO-1 monoclonal antibody (AB01003; Sanko Junyaku), a rat anti-P-cadherin monoclonal antibody (PCD-1; Takara Bio), a mouse anti-actin monoclonal antibody (MAB1501; Chemicon), a mouse anti-cadherin 11 monoclonal antibody (32-1700; Zymed) and a mouse anti-collagen IX monoclonal antibody (#MS-344-P0; Lab Vision) were purchased from commercial sources.

For histological analysis, mouse embryos and tissues were dissected and fixed with 10% formaldehyde at 4°C overnight, washed with 0.1 M phosphate-buffered saline, pH 7.4 (PBS), dehydrated in graded alcohols,embedded in paraffin wax, sectioned at 4 μm, and stained with Hematoxylin and Eosin. For immunofluorescence microscopy, mouse embryos and tissues were dissected and fixed with 2% paraformaldehyde in PBS at 4°C for 4 hours,followed by being washed with PBS. After being placed into 10% sucrose solution overnight and 25% sucrose solution for 4 hours, they were embedded in OCT compound (Sakura Finetechnical), frozen in liquid nitrogen, and then sectioned using a cryostat at 10 μm. The sections were mounted on glass slides, air-dried and washed three times with PBS containing 0.05% saponin. After being blocked in PBS containing 20% Block Ace (Dainippon Pharmaceutical)and 0.05% saponin for 30 minutes, the samples were incubated at 4°C overnight in PBS containing 5% Block Ace, 0.05% saponin and the primary antibodies. The samples were washed with PBS containing 0.05% saponin three times, and incubated at room temperature for 1 hour with the secondary antibodies in PBS containing 5% Block Ace and 0.05% saponin. After being washed for 5 minutes three times, the samples were mounted in 50% glycerol and viewed with a Radiance 2100 confocal laser-scanning microscope (BioRad).

Immunoelectron microscopy was performed using the silver-enhancement technique as described (Kinoshita et al.,1998). Adult mice were perfused with 4% paraformaldehyde in PBS(pH 7.4). Tissue blocks were cut to 50 μm on a vibratome. The sections were washed with 0.1 M phosphate buffer, pH 7.3 (PB) and then frozen and thawed in isopentane cooled with liquid nitrogen and PB containing 25% sucrose, 10%glycerol and 0.02% NaN₂ at room temperature. The sections were then incubated with 5 μg/ml of the primary antibodies in 50 mM Tris buffered saline, pH 7.4 (TBS) containing 2% normal goat serum. After being washed with TBS, the sections were incubated with the secondary antibodies coupled with 1.4-nm gold particles (Nanoprobes) and then reacted with HQSilver kit(Nanoprobes) for 8 minutes. After osmification, the immunostained sections were block stained with uranyl acetate, dehydrated and flat-embedded in Epon. The ultrathin sections were then prepared and viewed with an electron microscope (H-7500, HITACHI).

In situ hybridization was performed as described(Takemoto et al., 2002; Okabe et al., 2004a) with some modifications. In brief, mRNA probes were labeled with digoxigenin-UTP by RNA in vitro transcription according to the manufacturer's protocol (Roche Diagnostics). Cryosections were prepared as described above. The sections were air-dried, washed with water for 3 minutes and treated with 0.2 M HCl for 3 minutes. After being washed with water, the sections were treated with 0.1 M triethanolamine (pH 8.0) and acetylated with 0.25% dehydrated acetic acid in 0.1 M triethanolamine for 10 minutes. The sections were washed with PBS for 3 minutes twice and incubated with a prehybridization solution [50% deionized formamide, 5×SSPE (pH 7.5), 5% SDS and 1 mg/ml of yeast tRNA]. Each mRNA probe was mixed with the prehybridization solution, boiled at 80°C for 5 minutes, and added to the sections. After the sections were incubated at 50°C overnight, the sections were washed with 5×SSC and then with 2×SSC containing 50% formamide at 50°C for 30 minutes three times. After being blocked with 1.5% Blocking Reagent (Roche Diagnostics) in TNT buffer [100 mM Tris-HCl (pH 7.5), 150 mM NaCl, and 0.1% Tween-20], the sections were incubated with the anti-digoxigenin antibody conjugated to alkaline phosphatase at room temperature for 2 hours. The sections were washed with TNT buffer at room temperature for 30 minutes five times. Hybridized probes were visualized by HNPP Fluorescent Detection Set (Roche Diagnostics)and viewed with a Radiance 2100 confocal laser-scanning microscope.

Protein concentrations were determined with bovine serum albumin as a reference protein (Bradford,1976). SDS-PAGE was carried out as described by Laemmli(Laemmli, 1970).

We first generated nectin 1^+/– and nectin 3^+/– ES cells by homologous recombination using targeting vectors designed to delete the exon 2 of the nectin 1 gene and the exon 1 of the nectin 3 gene, respectively (Fig. 1A,part a; Fig. 1B, part a). When these targeting constructs were introduced into ES cells by an electroporation, six G418-resistant colonies heterozygous for the nectin 1 gene and six G418-resistant colonies heterozygous for the nectin 3 gene were obtained. The genotypes of the G418-resistant colonies were confirmed by Southern blotting (Fig. 1A, part b; Fig. 1B, part b). These ES clones were used to generate chimeric mice and successfully contributed to germ-line transmission. Nectin 1^+/– and nectin 3^+/– mice appeared normal compared with the wild-type littermates (data not shown). Each nectin 1^+/– or nectin 3^+/– mice was intercrossed and genotypes of the progeny were determined by PCR and Southern blotting using the tail DNAs (Fig. 1A, part c; Fig. 1B, part c;data not shown). Nectin 1^–/– and nectin 3^–/– mice did not express the nectin 1 and nectin 3 proteins, respectively, as neither the nectin 1 nor the nectin 3 protein was detected by the anti-nectin 1 or nectin 3 antibodies, which recognize the intracellular regions of the nectin 1 or nectin 3 proteins, respectively(Fig. 1A, part d; Fig. 1B, part d). Therefore, we referred these targeted alleles as null alleles. Nectin 1^–/– and nectin 3^–/– mice were viable and apparently healthy under specific-pathogen-free conditions, except that male nectin 3^–/– mice were infertile (data not shown), suggesting a role for nectin 3 in spermatogenesis. The analysis on this role of nectin 3 in spermatogenesis will be described elsewhere.

Microphthalmia was the most striking feature observed in common between nectin 1^–/– and nectin 3^–/– mice(Fig. 2A,B). This finding was unexpected because nectin 2^–/– mice did not show such microphthalmia (Bouchard et al.,2000; Mueller et al.,2003; Ozaki-Kuroda et al., 2000). Wild-type mice are born with both the eyes closed. However, about 30% of nectin 1^–/–pups (n=40), but not nectin 3^–/– pups(n=30), were born with one or both of the eyes open (data not shown). Nectin 1^–/– mice born with the eyes open tended to show more severe defect in their eyes than nectin 1^–/–littermates with the eyes closed, although all mice born with the eyes closed also showed microphthalmia. These results suggest that microphthalmia may be caused by the same underlying mechanism involving nectin 1 and nectin 3, but not nectin 2.

To investigate the etiology of microphthalmia in nectin 1^–/– and nectin 3^–/– mice, we histologically analyzed the eyes of adult wild-type, nectin 1^–/– and nectin 3^–/– mice. The vitreous body was absent, the retinal layers were undulating, and the lenses were somewhat deformed in the eyes of nectin 1^–/– and nectin 3^–/– mice(Fig. 3B, part a; Fig. 3C, part a). In addition,the ciliary processes were absent, and the ciliary epithelia adhered to the lens in the eyes of nectin 1^–/– and nectin 3^–/– mice (Fig. 3B,part b; Fig. 3C, part b). Because the ciliary body secretes aqueous humor and glycoproteins of the vitreous body(Bertazolli Filho et al., 1996; Francis and Alvarado, 1997; Haddad et al., 1990; Zimmerman and Fine, 1964), the impairment of the ciliary body seemed to cause microphthalmia, the absence of the vitreous body, the undulating retinal layers and the deformed lens in the eyes of nectin 1^–/– and nectin 3^–/– mice. These results indicate that the ocular phenotype of nectin 1^–/– mice is histologically indistinguishable from that of nectin 3^–/– mice.

The eye development of nectin 1^–/– and nectin 3^–/– mice was prospectively examined during embryogenesis. In mice, eyelid formation begins at embryonic day 13 (E13), and from E14 to 16 the eyelids grow, flatten across the eye, and fuse tightly(Harris and McLeod, 1982). At E15.5, the eyelids of wild-type, nectin 1^–/–, and nectin 3^–/– mice did not fuse and there was no obvious difference between their eyes (Fig. 4A,parts a, b; data not shown). At E16.5, the eyelids of wild-type and nectin 3^–/– mice completely fused(Fig. 4B, part a; data not shown). By contrast, the eyelids of nectin 1^–/– mice did not fuse at this stage (Fig. 4B, part b). Furthermore, both nectin 1^–/– and nectin 3^–/– mice showed a separation of the apex-apex contact between the pigment and non-pigment cell layers of the ciliary epithelia (Fig. 5B, parts b, c; Fig. 5C, parts b, c). The ciliary epithelia consist of two layers, the pigment and non-pigment epithelia, and these two layers make the ciliary processes(Raviola and Raviola, 1978). In normal eyes, the apices of the pigment and non-pigment epithelia were apposed and contacted each other (Fig. 5A,parts a-c). It has been reported that the apices of the pigment and non-pigment epithelia adhere with each other by puncta adherentia junctions, desmosomes and gap junctions(Raviola and Raviola, 1978). However, in the eyes of nectin 1^–/– and nectin 3^–/– pups, the apices of the pigment and non-pigment epithelia were separated at E16.5 and postnatal day 0 (P0)(Fig. 5B, parts b, c; Fig. 5C, parts b, c, see also Fig. 8). The apices of the pigment and non-pigment epithelia appeared to still contact at E15.5 in the eyes of nectin 1^–/– and nectin 3^–/– pups, as well as in wild-type pups(Fig. 5A, part a; Fig. 5B, part a; Fig. 5C, part a). There was no obvious difference in the vitreous body, the retinal layers or the lens between wild-type, nectin 1^–/– and nectin 3^–/– mice at E16.5 and P0(Fig. 4B, parts a, b; Fig. 4C, parts a, b; Fig. 4D, parts a, b). These results suggest that a separation of the apex-apex contact between the pigment and non-pigment cell layers of the ciliary epithelia is a primary defect in both nectin 1^–/– and nectin 3^–/–mice.

To examine whether differentiation of the pigment and non-pigment epithelia of the ciliary body is impaired in nectin 1^–/– and nectin 3^–/– mice, we next examined the expression of the marker genes for the presumptive iris and ciliary epithelia in the ciliary body of wild-type, nectin 1^–/– and nectin 3^–/– mice at P0(Fig. 6). It has been reported that collagen IX is expressed in the cells at optic cap margin, where the presumptive iris and ciliary epithelia are located, between E14.5 and P2, and that cadherin 11 is expressed there between E18.5 and P2(Thut et al., 2001). Both collagen IX and cadherin 11 were similarly expressed in the ciliary body of wild-type, nectin 1^–/– and nectin 3^–/– mice at P0. Thus, regardless of a separation of the apex-apex contact between the pigment and non-pigment epithelia, the development of the ciliary epithelial cells appears restored in nectin 1^–/– and nectin 3^–/– mice as observed at least by the expression of these marker genes.

We next examined the localization of nectin 1 and nectin 3 in the ciliary body of adult wild-type mice. Because melanin in the pigment epithelia inhibits transmission of laser, we used albino mice. The immunofluorescence signals for P-cadherin and ZO-1 were concentrated at the contact sites between the pigment and non-pigment cell layers of the ciliary epithelia(Fig. 7A, parts a-d; Fig. 7E, parts a-d), consistent with the previous observations(Tserentsoodol et al., 1998; Wu et al., 2000). The signal for nectin 1 was also concentrated at the contact sites between the pigment and non-pigment cell layers of the ciliary epithelia and colocalized with that for ZO-1 (Fig. 7A, parts a-d). The signals for nectin 3 and afadin were also concentrated and colocalized with those of nectin 1, ZO-1 and P-cadherin(Fig. 7B, parts a-d; Fig. 7D, parts a-d; Fig. 7E, parts a-d; data not shown). Although the signal for nectin 2 was also observed(Fig. 7C, part a), it appeared as punctate spots and was clearly distinguished from those for nectin 1,nectin 3, afadin, ZO-1 and P-cadherin, which appeared as lines(Fig. 7A, parts a, b; Fig. 7C, parts a, b; Fig. 7D, part a; Fig. 7E, part a; data not shown). This punctate signal for nectin 2 indicates that nectin 2 localizes only at the apicolateral junctions, i.e. the most apical side of the lateral membrane domain, but not at the entire apex-apex junctions. Thus, nectin 1 and nectin 3, but not nectin 2, are likely to participate in adhesion at the apex-apex junctions between the pigment and non-pigment cell layers of the ciliary epithelia.

We next examined the localization of nectin 1, nectin 2, nectin 3 and afadin in the ciliary body of wild-type pups at E15.5, E16.5, and P0. The signals for nectin 1 and nectin 3, and afadin were concentrated as lines at the contact sites between the pigment and non-pigment cell layers of the ciliary epithelia at E15.5, E16.5, and P0(Fig. 8). Although the signal for nectin 2 was also observed there, it appeared as punctate spots and was clearly different from those for nectin 1, nectin 3 and afadin, which appeared as lines (Fig. 8). Taken together with the observations that the apices of the pigment and non-pigment epithelia became to be separated at E16.5 in the eyes of nectin 1^–/– and nectin 3^–/– pups(Fig. 5B, parts b, c; Fig. 5C, parts a, b) and that such a separation of the apex-apex contact between the pigment and non-pigment cell layers was not observed in the eyes of wild-type pups, these results suggest that nectin 1, nectin 3 and afadin is likely to participate in adhesion at the apex-apex junctions between the pigment and non-pigment cell layers of the ciliary epithelia from the embryonic stage before E16.5. However, it still remains to be defined whether the apex-apex junctions between the pigment and non-pigment cell layers of the ciliary epithelia are formed from the embryonic stage before E16.5, as apposition of two cell layers does not immediately indicate cell adhesion. It also remains unknown whether the residual apposed regions between the pigment and non-pigment cell layers in nectin 1^–/– and nectin 3^–/–pups adhere or just contact with each other.

We then examined the precise localization of nectin 1 and nectin 3 at the contact sites between the pigment and non-pigment epithelia by immunoelectron microscopy (Fig. 9A, parts a,b; Fig. 9B). For this purpose, we used the anti-nectin 1 polyclonal antibody that recognized the intracellular region of nectin 1, and the anti-nectin 3 monoclonal antibody that recognized the extracellular region of nectin 3. Although the immunoparticles for nectin 1 were detected at both the cytoplasmic faces of the plasma membranes of the pigment and non-pigment epithelia, they were more frequently detected on the side of the pigment epithelia than on the side of non-pigment epithelia (Fig. 9A, parts a,b, arrows). Consistently, the nectin 1 mRNA was expressed in both the pigment and non-pigment epithelia, but it was more abundantly expressed in the pigment epithelia (Fig. 9C, parts a,b). The immunoparticles for nectin 3 were concentrated at the junctions between the plasma membranes of both the pigment and non-pigment epithelia (Fig. 9B, arrows). The nectin 3 mRNA was expressed in both the pigment and non-pigment epithelia(Fig. 9D, part a), suggesting that nectin 3 localized at both the plasma membranes of the pigment and non-pigmental epithelia. No signal for nectin 1 or nectin 3 was observed at gap junctions (Fig. 9B,arrowhead; data not shown). These results further indicate that nectin 1 and nectin 3 localize at the apex-apex junctions between the pigment and non-pigment cell layers of the ciliary epithelia.

We next examined the localization of nectin 1, nectin 2, nectin 3, afadin,ZO-1 and P-cadherin at the apicolateral junctions of both the pigment and non-pigment epithelia in the ciliary body. As the ciliary processes were absent in adult nectin 1^–/– and nectin 3^–/– mice (see Fig. 3B, part b; Fig. 3C, part b), we examined their localization in the ciliary body of nectin 1^–/– and nectin 3^–/– albino mice at P0 when their ciliary body appeared to be abnormal(Fig. 10). We first confirmed that the nectin 1^–/– and nectin 3^–/– albino mice showed a separation of the apex-apex contact between the pigment and non-pigment cell layers of the ciliary epithelia at P0 (Fig. 10A). The immunofluorescence signal for nectin 1 was concentrated at the contact sites between the pigment and non-pigment cell layers of wild-type mice,whereas that for nectin 1 was abrogated in nectin 1^–/–mice. The signal for nectin 1 remained at the contact sites between two neighboring pigment cells, but not at those between two neighboring non-pigment cells of nectin 3^–/– mice(Fig. 10A,B). The signal for nectin 2 was concentrated at the contact sites between the pigment and non-pigment cell layers of wild-type mice as a punctate spots. The signal for nectin 2 remained at the contact sites between two neighboring pigment cells and those between two neighboring non-pigment cells of both nectin 1^–/– and nectin 3^–/– mice(Fig. 10A). The signal for nectin 3 was concentrated at the contact sites between the pigment and non-pigment cell layers of wild-type mice, whereas that for nectin 3 was abrogated in nectin 3^–/– mice. The signal for nectin 3 remained at the contact sites between two neighboring pigment cells and at those between two neighboring non-pigment cells of nectin 1^–/– mice (Fig. 10A,C). The signals for afadin and ZO-1 were concentrated at the contact sites between the pigment and non-pigment cell layers of wild-type mice, and the signals remained at the contact sites between two neighboring pigment cells and at those between two neighboring non-pigment cells of nectin 1^–/– and nectin 3^–/– mice(Fig. 10A). The signal for P-cadherin was concentrated at the contact sites between the pigment and non-pigment cell layers of wild-type mice as punctate spots(Fig. 10A). As the signal for P-cadherin appeared as lines in adult mice (see Fig. 7E, part a), P-cadherin seemed to be recruited to the apex-apex junction between the pigment and non-pigment epithelia at days later than P0. The signal for P-cadherin remained at the contact sites between two neighboring pigment cells, but not at those between two neighboring non-pigment cells of nectin 1^–/– and nectin 3^–/– mice(Fig. 10). No signal for E-cadherin or N-cadherin was observed at the contact sites between two neighboring non-pigment cells (data not shown). Thus, another unknown cadherin(s) might localize at the junctions between two neighboring non-pigment cells. Taken together, nectin 1, nectin 2, nectin 3, afadin, ZO-1 and P-cadherin localized at the apicolateral junctions between the pigment epithelia, whereas nectin 2, nectin 3, afadin and ZO-1, but not nectin 1 or P-cadherin, localized at the apicolateral junctions between the non-pigment epithelia.

Nectin-based cell-cell adhesion plays important roles in morphogenesis of multicellular organisms (Takai et al.,2003a; Takai and Nakanishi,2003). Implications of nectins have first been demonstrated in the knockout study of mice lacking afadin, a major adaptor protein connecting nectins to the actin cytoskeleton that binds all members of the nectin family(Ikeda et al., 1999). In the absence of afadin, mouse embryos cease to develop at E8.5. However, knockout studies on individual members of the nectin family have been proven to be more informative. Although nectin 2^–/– mice show impaired spermatogenesis (Ozaki-Kuroda et al., 2000), our present study on mice lacking nectin 1 and nectin 3 show a virtually identical ocular phenotype:microphthalmia. This work has demonstrated that the potent heterophilic trans-interaction between nectin 1 and nectin 3, originally detected by an in vitro study, actually promotes the apex-apex adhesion between the pigment and non-pigment cell layers of the ciliary epithelia in mice. To our knowledge,the apex-apex junction between the epithelia is unique to the ciliary body. Impairment of the trans-interaction between nectin 1 and nectin 3 causes separation of the pigment and non-pigment cell layers and disruption of the ciliary body. The differentiation markers for the presumptive iris and ciliary epithelia are similarly expressed in the ciliary body of wild-type, nectin 1^–/– and nectin 3^–/– mice. Thus,the apex-apex adhesion mediated by nectin 1 and nectin 3 between the pigment and non-pigment cell layers is unlikely to be involved in the ciliary epithelial cell surviving and differentiating pathway, but is required for a process of the ciliary body formation along with the subsequent folding of these two cell layers. The ciliary body is known to produce both the aqueous humor and some components of the vitreous body and is the source of the zonules that support the lens (Bertazolli Filho et al., 1996; Francis and Alvarado, 1997; Haddad et al.,1990; Zimmerman and Fine,1964). Consistently, the vitreous body was absent, the retinal layers were undulating and the lenses were deformed in adult nectin 1^–/– and nectin 3^–/– mice. Thus,the apex-apex adhesion mediated by nectin 1 and nectin 3 between the pigment and non-pigment cell layer are required for eye development. To our knowledge,this phenotype of a separation between the pigment and non-pigment cell layer,which is observed in nectin 1^–/– and nectin 3^–/– mice, has not been reported before. We have shown that the primary defect of this trans-interaction is attributed to the differential localization of nectins: nectin 1 and nectin 3 localize at both the apex-apex and apicolateral junctions, whereas nectin 2 localizes at the apicolateral junctions. These results are further supported by the observation that mice lacking nectin 2 exhibit no microphthalmia. Thus, we conclude that the ciliary epithelia consisting of two layers, the pigment and non-pigment epithelia (Raviola and Raviola,1978), are apposed and adhered by puncta adherentia junctions and gap junctions in which nectin 1 and nectin 3, together with P-cadherin and ZO-1, play essential roles for CAMs (Fig. 11). It remains unknown how nectin 2 is excluded from the apical surface of pigment and non-pigment epithelia. It might be due to a higher affinity between nectin 1 and nectin 3 than between nectin 2 and nectin 3(Takai et al., 2003a). Alternatively, nectin 2 might be sorted to the basolateral membrane, but not to the apical membrane, in the pigment and non-pigment epithelia of the ciliary body. The human nectin 1 mutations are responsible for cleft lip/palate-ectodermal dysplasia, Margarita island ectodermal dysplasia and Zlotogora-Ogür syndrome, which is characterized by cleft lip/palate,syndactyly, mental retardation and ectodermal dysplasia(Sozen et al., 2001; Suzuki et al., 2000). Although ocular involvement in individuals with this syndrome has not been reported,the human nectin 1 mutations might potentially cause an ocular defect.

As P-cadherin localizes at this apex-apex junction(Wu et al., 2000), the roles of the trans-interaction between nectin 1 and nectin 3 may be challenged by the presence of P-cadherin. However, P-cadherin^–/– mice have no ocular phenotype (Radice et al.,1997). Furthermore, the signal for P-cadherin at the apex-apex junctions is observed in adult mice but not at P0, while those for nectin 1 and nectin 3 have already been observed at P0. Thus, nectin 1 and nectin 3,but not P-cadherin, are essential for the formation of the apex-apex adhesion between the pigment and non-pigment cell layers of the ciliary epithelia. The force of the heterophilic trans-interaction between nectin 1 and nectin 3 has been thought to be weaker than that of the homophilic trans-interaction of E-cadherin. For example, whereas cadherins show the strong cell-cell adhesion accompanied by compaction, maximization of adhesive contacts, the nectin 1-and nectin 3-mediated adhesion is not accompanied by compaction(Takai et al., 2003a; Takai and Nakanishi, 2003). In addition, force measurement analysis on paired cells has revealed that the force of the heterophilic-trans-interaction between nectin 1 and nectin 3 is much weaker than that of the homophilic-trans-interaction of E-cadherin(Martinez-Rico et al., 2005). However, a recent single-molecule analysis with the purified proteins of nectins and E-cadherin has remarkably shown that the force of the heterophilic-trans-interaction between nectin 1 and nectin 3 is stronger than that of the homophilic-trans-interaction of E-cadherin in a low loading rate condition (less than 100 pN/s), while the force of the heterophilic-trans-interaction between nectin 1 and nectin 3 is weaker than that of the homophilic-trans-interaction of E-cadherin in a high loading rate condition (more than 100 pN/s) (Y. Tsukazaki, K. Kitamura, K. Shimizu, A. Iwane, Y.T. and T. Yanagida, unpublished). Thus, the force of the heterophilic-trans-interaction between nectin 1 and nectin 3 itself should be strong enough to establish the apex-apex adhesion between the pigment and non-pigment epithelia of the ciliary body at P0, even when the signal for P-cadherin is not observed there.

Another important insight learned from this study is that nectin 1 and nectin 3 also play sentinel roles to form solid apex-apex junctional structures between the pigment and non-pigment epithelia of the ciliary body in adult mice. We have previously shown that nectins are initially involved in the formation of AJs in cooperation with E-cadherin, and subsequently promote the formation of TJs in epithelial cells in culture(Takai et al., 2003a; Takai and Nakanishi, 2003). These findings may be extended to the formation of the ciliary body: the heterophilic trans-interaction of nectin 1 and nectin 3 first forms the cell adhesion, then recruits P-cadherin to the nectin 1- and nectin 3-based adhesion sites, and finally ends up establishing the strong adhesion undercoated with F-actin, mediated by afadin and ZO-1, at the apex-apex junctions between the pigment and non-pigment cell layers of the ciliary epithelia. Thus, the heterophilic trans-interaction between nectin 1 and nectin 3 is required for the formation of three junctional structures in the ciliary body: gap junctions, desmosomes and puncta adherentia junctions(Raviola and Raviola,1978).

Phenotypes of mice lacking nectins are modified by functional redundancy that depends on the localization of nectins and their heterophilic trans-interaction. Afadin^–/– mice show embryonic lethality with developmental defects because all of the combinations of heterophilic trans-interactions between nectins are disrupted. By contrast,nectin 1^–/–, nectin 2^–/– and nectin 3^–/– mice are viable and show no life-threatening disorder. Neither nectin 1^–/– nor nectin 3^–/– mice apparently show impaired organization of AJs and TJs in most tissues where various types of nectins are expressed(data not shown). There is no doubt that the apex-apex adhesion between the pigment epithelia and non-pigment epithelia is one of the places with reduced functional redundancy of nectins and thus vulnerable to the gene disruption experiments. However, the apicolateral junctions between the pigment epithelia or those between the non-pigment epithelia are preserved in the absence of nectin 1 or nectin 3 because of the presence of nectin 2. The finding that male nectin 3^–/– mice exhibit infertility is also probably explained by the specific localization of nectin 2 and nectin 3 in Sertoli cells and spermatids, respectively(Ozaki-Kuroda et al., 2002; Takai and Nakanishi, 2003), as well as by the absence of nectin 1. Although about 30% of nectin 1^–/– pups are born with one or both eyes open, this phenotype is not observed in nectin 3^–/– mice,indicating an additional role specific to nectin 1. Consistently, nectin 1 has recently shown to localize at the sites of eyelid fusion(Okabe et al., 2004a). Further knockout studies would reveal additional roles of nectins in tissue morphogenesis and organogenesis during mouse development.

We thank Drs M. Takeichi (Center for Developmental Biology, RIKEN, Kobe,Japan), N. Azuma (National Center for Child Health and Development, Tokyo,Japan) and K. Takata (Gunma University, Gunma, Japan) for helpful discussions and critical readings of the manuscript. The investigation at Osaka University Medical School was supported by grants-in-aid for Scientific Research and for Cancer Research from the Ministry of Education, Science, Sports, Culture and Technology, Japan (2003, 2004).