JACOP, a Novel Plaque Protein Localizing at the Apical Junctional Complex with Sequence Similarity to Cingulin*

The apical junctional complex is composed of various cell adhesion molecules and cytoplasmic plaque proteins. Using a monoclonal antibody that recognizes a chicken 155-kDa cytoplasmic antigen (p155) localizing at the apical junctional complex, we have cloned a cDNA of its mouse homologue. The full-length cDNA of mouse p155 encoded a 148-kDa polypeptide containing a coiled-coil domain with sequence similarity to cingulin, a tight junc-tion (TJ)-associated plaque protein. We designated this protein JACOP (junction-associated coiled-coil protein). Immunofluorescence staining showed that JACOP was concentrated in the junctional complex in various types of epithelial and endothelial cells. Furthermore, in the liver and kidney, JACOP was also distributed along non-junctional actin filaments. Upon immunoelectron microscopy, JACOP was found to be localized to the undercoat of TJs in the liver, but in some tissues, its distribution was not restricted to TJs but extended to the area of adherens junctions. Overexpression studies have revealed that JACOP was recruited to the junctional complex in epithelial

Tight junctions (TJs), 1 the most apical component of the junctional complex, play crucial roles in the epithelial and endothelial barrier function (1)(2)(3)(4)(5). TJs create the circumferential seal around cells and work as barriers against the free diffusion of solutes through the paracellular pathway (1)(2)(3)(4)(5). The strength and selectivity of the TJ barrier vary among cell types depending on their physiological requirement of paracellular transport (2)(3)(4)(5). Furthermore, the biogenesis and barrier function of TJs are influenced by various intracellular signaling systems (4). Recent progresses in the identification and characterization of TJ-constituting proteins have enabled the analysis of the molecular basis of the structure and the functional regulation of the TJ barrier. TJ strands, the intramembrane part of TJs observed by freeze-fracture electron microscopy, mainly consist of at least two types of integral membrane proteins, claudins and occludin, both of which have four membrane-spanning regions (6,7). JAMs (junctional adhesion molecules), members of the immunoglobulin superfamily, are also included in or localized very close to TJ strands (8,9). Among these proteins, claudins, which comprise a multi-gene family containing Ͼ20 members, have been shown to contribute directly to the structure and barrier function of TJs (5).
Similar to other intercellular junctions, TJs also contain various cytoplasmic plaque proteins that are thought to play important roles in the formation and regulation of the TJ barrier by providing a scaffold for various proteins and linking adhesion molecules to cytoskeletons (4,5). Among TJ-associated plaque proteins, ZO-1, ZO-2, and ZO-3 have been well characterized, belonging to the membrane-associated guanylate kinase family (10 -12). Each of these membrane-associated guanylate kinases has multiple domains for protein-protein interaction including three PDZ domains, one SH3 domain, one guanylate kinase-like domain, and one proline-rich region, and works as a scaffold for various molecular components of TJs (13,14). ZO-1 forms independent complexes with ZO-2 and ZO-3 (15), and they are localized immediately beneath the plasma membranes of TJs (10,16). They directly bind to integral membrane proteins, claudins, JAMs, and occludin at the PDZ1, PDZ3, and guanylate kinase domains (9,17,19), respectively. C-terminal halves of ZO-1 and ZO-2 were shown to bind to F-actin in vitro (19 -21), suggesting that these membraneassociated guanylate kinases establish the molecular link between the TJ membrane and the actin cytoskeleton. Non-membrane-associated guanylate kinase TJ-associated plaque proteins such as cingulin, symplekin, and 7H6 antigen have also been identified and characterized (22)(23)(24). Cingulin, which consists of a globular "head" domain, a coiled-coil "rod" domain, and a globular "tail" domain, is predicted to form a homoparallel dimer in the rod domains and is localized farther from the plasma membrane than ZO-1 (16,25). Cingulin forms a complex with ZO-1 and ZO-2 in vivo in its head domain, and overexpression of cingulin in cultured epithelial cells affected the normal distribution of ZO-1 at cell-cell contact sites, suggesting an intimate relationship between cingulin and ZO-1 in the TJ structure (25,26). In vitro binding studies have revealed that cingulin interacts with various components of TJs including JAM, ZO-1, ZO-2, ZO-3, myosin, and F-actin, suggesting a role for cingulin as a linker between the TJ membrane and F-actin (25)(26)(27)(28). The existence of these molecular linkage systems between the TJ membrane and actin filaments is consistent with the notion that the actin cytoskeleton is involved in the regulation of the TJ barrier (4). Furthermore, an increasing number of proteins have been found as cytoplasmic components of TJs including the Par-3⅐Par-6⅐aPKC complex, MAGI-1 and MAGI-3, Pilt, JEAP, MUPP-1, ZONAB, and GEF-H1/Lfc (29 -36), suggesting that TJs are involved in various cellular functions other than the barrier function such as cell polarization, protein transport, cell growth regulation, and so forth.
Along this line, we continued to identify novel TJ constituents by raising mAbs against a junction-enriched fraction isolated from the liver (37). Here, we describe the cDNA cloning and characterization of JACOP, a novel cytoplasmic plaque protein localizing at the apical junctional complex including TJs and AJs. This novel molecule showed significant sequence similarity to cingulin.

EXPERIMENTAL PROCEDURES
Cloning of Mouse JACOP cDNA and Mouse Cingulin cDNA-A TJand AJ-enriched plasma membrane fraction was isolated from chick liver as described previously (37). Peripheral membrane proteins were extracted from this membrane fraction with 1.5 M sodium trichloroacetate for 2 h at 4°C.
After ultracentrifugation at 100,000 ϫ g for 1 h at 4°C, the supernatant was recovered and dialyzed against PBS, pH 7.4. This was followed by centrifugation at 20,000 ϫ g for 20 min at 4°C. Chicken p155/JACOP protein was precipitated from the supernatant with protein G-Sepharose 4B beads (Amersham Biosciences) with which E14 mAb was covalently coupled by the use of dimethyl pimerimide (Pierce). The precipitated proteins were eluted from the beads with the SDS-PAGE sample buffer, separated by SDS-PAGE, and then electrophoretically transferred onto a polyvinylidene difluoride membrane (Bio-Rad). After staining with Coomassie Brilliant Blue R-250, the protein band of p155/JACOP on the polyvinylidene difluoride membrane was excised and subjected to amino acid sequence analysis by the in-gel digestion method described by Rosenfeld et al. (38) in which four distinct peptide sequences (Peptides 1-4 in Fig. 1B) were determined. A homology search using the GenBank TM /EMBL/DDBJ data base identified a mouse EST clone (GenBank TM accession number AA144597) encoding a polypeptide, showing significant identity with Peptide 4 (Fig. 1B). Using two primers, 5Ј-TACTATGGCTGGAGTGT-3Ј and 5Ј-CTGCAGCTGCT-CATTCA-3Ј designed from the sequence information in this mouse EST, a 348-base DNA fragment was amplified by PCR from mouse lung cDNA prepared from mouse lung total RNA with Superscript II reverse transcriptase (Invitrogen). This fragment was used as a template to generate a digoxigenin (DIG)-labeled hybridization probe with the DIG High Prime labeling kit (Roche Applied Science), and hybridization screening was performed using a Lambda ZAP mouse lung cDNA library (Clontech). Among the positive clones, cl.9Z of 2.9 kb was the longest and included the termination codon but the 5Ј part of the cDNA containing the initiation codon was not found in this cDNA library. Using a hybridization probe designed from the 5Ј sequence of cl.9Z, a Lambda ZAP F9 mouse teratocarcinoma cDNA library was screened and we successfully obtained a cDNA clone (cl.1-3), which contained the initiation codon and overlapped with cl.9Z. Based on the sequence FIG. 1. Purification and cDNA cloning of JACOP. A, immunoprecipitation of chicken p155. Using anti-p155 mAb E14, immunoprecipitation was performed in the presence (ϩextract) or absence (Ϫextract) of the protein extract prepared from the junction-enriched plasma membrane fraction isolated from chick liver. A band of p155 was detected on SDS-PAGE followed by silver staining (arrowhead). B, partial amino acid sequences of chicken p155. Four peptide sequences (Peptides 1-4) were determined. A data base search identified a mouse EST clone (GenBank TM accession number AA144597) with significant homology to Peptide 4. The identity and homology between Peptide 4 and clone AA144597 are indicated by asterisks and dots, respectively. C, amino acid sequence of mouse JACOP, the homologue of chicken p155. The mouse JACOP cDNA encodes a 1298 amino acid polypeptide, which contained four sequences homologous to Peptides 1-4 (underlined).
information of these cDNA clones, DNA fragments that cover the fulllength protein were obtained from F9 cDNA by PCR with appropriate primers and were ligated in pBluescript SK(Ϫ) to form the full-length JACOP cDNA using standard methods of molecular biology.
Mouse cingulin cDNA was obtained by PCR based on the sequence information from a mouse EST (GenBank TM accession number BC042459). Using the primer pairs 5Ј-GAATCCGGGAGCACTGATCT-GGAC-3Ј/5Ј-AGTCTGAATTGGATCACTTGTAGG-3Ј and 5Ј-AGCAT-AGCCAGAGTCCCGATTCTG-3Ј/5Ј-AGACATCTTCTGCCTCTCAGC-CTC-3Ј, two overlapping DNA fragments whose combination covers the whole open reading frame were amplified by PCR from cDNA prepared from mouse small intestine total RNA with Superscript II reverse transcriptase. Nucleotide sequences of these two fragments (1.7 and 2.4 kb, respectively) were checked not to contain mutations responsible for the amino acid replacement. They were then linked together through the SphI site in pBluescript SK(Ϫ) to generate cDNA encoding the full-length mouse cingulin. Compared with BC042459, the cloned cingulin cDNAs lacked 24 nucleotides encoding MVSPAST (aa 338 -345 in BC042459), probably because of alternative splicing.
Generation of Expression Constructs-Using PCR and standard methods of molecular biology, an HA epitope tag (YPYDVPDYA) with a three-glycine linker was bound to the N or C terminus of full-length JACOP. The C-terminal deletion constructs encoding aa 1-367 and 1-581 of JACOP with the N-terminal HA tag (j1-367 and j1-581, respectively) were produced by PCR using the N-terminal HA-tagged full-length JACOP construct as a template. An N-terminal deletion construct encoding aa 582-1298 of JACOP with a C-terminal HA tag (j582-1298) was produced by PCR using the C-terminal HA-tagged full-length JACOP construct as a template. DNA fragments for HAtagged full-length JACOP and deletion constructs were subcloned into a mammalian expression vector, pCAGGSneodelEcoRI (39).
Northern Blotting-Total RNAs of various mouse tissues were prepared with TRIzol reagents (Invitrogen) based on the manufacturer's instructions, separated by 1.0% agarose-formaldehyde gel electrophoresis, and blotted onto the nylon membrane. A ϳ800-bp DIG-labeled DNA probe was generated with the DIG-PCR amplification kit (Roche Applied Science) using two primers, 5Ј-TTTGGCGAATACCAA-CACGTA-3Ј and 5Ј-TTTCTTGGTTTCAGAATAGGC-3Ј, and full-length cDNA of JACOP as a template. Northern hybridization was performed in buffer containing 5ϫ SSC, 50% formamide, 0.1% N-lauroyl sarcosine, 0.02% SDS, and 2% blocking reagent (Roche Applied Science) at 68°C for 12 h. The membrane was washed with 2ϫ SSC containing 0.1% SDS at room temperature for 30 min and with 0.1ϫ SSC containing 0.1% SDS at 68°C for 45 min and incubated with alkaline phosphataseconjugated anti-DIG antibody (Roche Applied Science). For the detection of hybridized probe, the membrane was soaked in CSPD chemiluminescence substrate (Roche Applied Science) and exposed to x-ray film, which was processed for imaging. As a loading control, the membrane was used for re-hybridization with a DIG-labeled probe of mouse glyceraldehyde-3-phosphate dehydrogenase (GAPDH).
Antibodies-Bacterial expression constructs of the GST fusion protein containing aa 493-633 of JACOP and the maltose-binding protein fusion protein containing the identical region of JACOP were generated by subcloning the corresponding DNA fragment amplified by PCR into pGEX4T-1 (Amersham Biosciences) and pMAL-c2 (New England Biolabs), respectively. The GST-JACOP fusion protein then was produced in Escherichia coli. DH5-␣ was purified with glutathione-Sepharose 4B beads and subcutaneously injected into rabbits to raise anti-JACOP pAb. Anti-JACOP pAb was used for experiments after affinity purification. The obtained rabbit serum was incubated with nitrocellulose membranes blotted with the maltose-binding protein-JACOP fusion protein produced in E. coli, and the bound antibodies were eluted with 0.2 M glycine buffer, pH 2.8, followed by neutralization to pH 7.5 with 2 M Tris-HCl, pH 9.5. To produce rat anti-mouse cingulin mAb, an expression construct of the GST fusion protein containing aa 216 -419 of mouse cingulin was generated by subcloning the corresponding DNA fragment amplified by PCR into pGEX4T-1. The GST-cingulin fusion protein was produced in E. coli, purified, and injected into rats. mAb production was performed as described previously (6). Mouse antiafadin monoclonal antibody (40) was kindly provided by Dr. Yoshimi Takai (Osaka University Graduate School of Medicine, Osaka, Japan). Mouse anti-ZO-1 monoclonal antibody (T8 -754) was generated and characterized as reported previously (41). Mouse anti-HA tag mAb and rat anti-HA tag mAb were purchased from Covance and Roche Applied Science, respectively. Actin filaments were labeled with Alexa 488phalloidin (Molecular Probes).
Cell Culture and Transfection-Eph4 cells (a gift from Dr. Reichmann, Institute Suisse de Recherches, Lausanne, Switzerland), L cells, and NIH3T3 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. DNA transfection was performed using LipofectAMINE with Plus reagent (Invitrogen) according to the manufacturer's instructions. After 48 h in culture, cells were analyzed by immunoblotting or immunostaining. To obtain stable transfectants of Eph4 cells expressing j582-1298, cells were replated after 48 h after transfection and cultured in the presence of 500 g/ml G418 for 2 weeks.
SDS-PAGE and Western Blotting-SDS-PAGE (10%) was performed according to the method of Laemmli (42), and the gels were stained with a silver-staining kit (Wako Pure Chemicals). For Western blotting, proteins resolved by SDS-PAGE were electrophoretically transferred from gels to nitrocellulose filters. Filters were incubated with a 1:1000 dilution of anti-JACOP rabbit serum or culture supernatant of anticingulin monoclonal antibody followed by biotinylated secondary antibodies and streptavidin-conjugated alkaline phosphatase. Nitro blue tetrazolium and bromochloroindoryl phosphate were used as substrates for the detection of alkaline phosphatase.
Immunofluorescence Microscopy-To analyze the distribution of JACOP, various mouse tissues were cut in pieces and embedded in O.C.T. compound using liquid nitrogen. Frozen sections (ϳ5-m thick) were cut in a cryostat, mounted on glass slides, air-dried, and fixed in 95% ethanol at 4°C for 30 min followed by 100% acetone at room temperature for 1 min. They were then washed with PBS, incubated with 1% bovine serum albumin/PBS for 10 min, and finally incubated with primary antibodies at room temperature for 30 min. After being washed with PBS, samples were incubated for 30 min with secondary antibodies. For the immunostaining of cultured cells, cells plated on glass coverslips were fixed with 1% formaldehyde in PBS for 10 min at room temperature. They were then permeabilized with 0.2% Triton X-100 in PBS for 5 min and washed three times with PBS. After blocking with 1% bovine serum albumin/PBS for 10 min, the samples were treated with primary antibodies, washed with PBS, and finally incubated with secondary antibodies. As secondary antibodies, Alexa Fluor 488 goat anti-rabbit IgG, Alexa Fluor 647 goat anti-mouse IgG, Alexa Fluor 488 goat anti-mouse IgG (Molecular Probes), Cy3-conjugated donkey anti-rat IgG, and Cy3-conjugated donkey anti-mouse IgG (Jackson ImmunoResearch laboratories) were used. After being washed with PBS, samples were mounted in 90% glycerol-PBS containing 0.1% para-phenylenediamine and 1% n-propylgalate. Specimens were observed using a Zeiss Axio-phot photomicroscope or Zeiss LSM510 confocal laser-scanning microscope (Carl Zeiss).
Immunoelectron Microscopy-Immunoelectron microscopy using the silver-enhancement immunogold method was performed as described previously with some modifications (43). Frozen sections (ϳ10-m thick) of tissues fixed with 2% paraformaldehyde were incubated with anti-JACOP pAb followed by a second Ab coupled with 1.4-nm gold particles (Nanoprobes Inc.). The sample-bound gold particles were silver-enhanced with the HQ-silver enhancement kit (Nanoprobes Inc.) at 25°C for 8 min. After washes with distilled water, samples were postfixed with 1% osmium oxide in 100 mM phosphate buffer, pH 7.3. They were dehydrated through a graded series of ethanol (50, 60, 70, 80, 90, 95, and 100%) and propylene oxide and embedded in epoxy resin. From these samples, ultrathin sections were cut, stained with uranyl acetate and lead citrate, and then observed with a Hitachi H7500 electron microscope.

RESULTS
Purification and cDNA Cloning of JACOP-We previously reported the identification of a novel 155-kDa protein (p155) localizing at epithelial and endothelial cell-cell junctions using the monoclonal antibody E14, which was produced in a rat immunized with a cell-cell junction-enriched plasma membrane fraction isolated from chick liver (37). To further characterize p155, we attempted to clone its cDNA. Using protein G beads covalently coupled with E14 monoclonal antibody, the chicken p155 protein was purified from the peripheral membrane proteins extracted from the cell-cell junction-enriched plasma membrane fraction (Fig. 1A). From the protease-digested chicken p155 protein, amino acid sequences of four peptides were determined, one of which corresponded to a peptide sequence encoded by a mouse EST in the data base with high homology (Fig. 1B, Peptide 4). The full-length cDNA of this EST was cloned by hybridization screening of cDNA libraries of mouse lung and mouse F9 teratocarcinoma cells. The entire open reading frame of the obtained cDNA clone encodes 1,298 amino acid residues with a calculated molecular mass of ϳ150 kDa, which is comparable with the molecular mass of p155 observed on SDS-PAGE (Fig. 1C). Furthermore, the amino acid sequence predicted from the obtained cDNA contains sequences homologous to the remaining three chicken peptides, indicating that we certainly cloned the mouse homologue of p155 (Fig. 1, B and C). Coils, Paircoil, and Multicoil programs predicted an extended dimeric coiled-coil structure in the C-terminal half ( Fig. 2A). The N-terminal half was not predicted to contain a coiled-coil and was likely to assume a globular structure. Downstream of the coiled-coil domain, there was another small domain. We referred to these three domains as the head, rod, and tail from the N terminus (Fig. 2B) and designated this novel protein JACOP (junction-associated coiled-coil protein).
A homology search of the databases revealed that the Cterminal half of the amino acid sequence shared similarity with the ␣-helical coiled-coil domain of various proteins, among which cingulin, a cytoplasmic component of TJs, was most related to JACOP (Fig. 2B). The coiled-coil domains of JACOP and mouse cingulin were 39% identical at the amino acid level (Fig. 2B). In addition, JACOP and cingulin had two highly homologous regions in their head domains, confirming the structural similarity between these two proteins (Fig. 2C). The JACOP gene appeared to be conserved among mammalian species. The human and rat (GenBank TM accession numbers NM_032866 and XM_236385, respectively) homologues of JA-COP were 85 and 94% identical to mouse JACOP at the amino acid level, respectively (data not shown).
To examine the tissue distribution of JACOP, Northern blot analysis was performed using mRNAs prepared from various mouse tissues. As shown in Fig. 3, a single band of ϳ8 kb was detected in most tissues including non-epithelial ones. JACOP mRNA was detected in large amounts in the kidney and lung, whereas only a trace amount of the transcript was observed in the small intestine, a typical epithelial tissue.
Production and Characterization of Specific Antibodies for Mouse JACOP and Mouse Cingulin-To analyze the subcellu- FIG. 3. Northern blot analyses of JACOP. Total RNAs isolated from various mouse tissues were blotted on a membrane and hybridized with a DIG-labeled JACOP cDNA fragment. JACOP mRNA was detected as an ϳ8-kb band. As a loading control, the same membrane was reprobed with a DIG-labeled GAPDH cDNA fragment (bottom). lar localization of JACOP in mouse tissues, we raised rabbit pAb against a recombinant GST fusion protein, which contained aa 493-633 in the head domain of mouse JACOP. Furthermore, to compare the expression and distribution of JA-COP with that of cingulin, we also produced a monoclonal antibody to the head domain of mouse cingulin and checked the specificity of these antibodies. Initially, mouse L fibroblasts were transiently transfected with the mammalian expression vector for mouse JACOP or mouse cingulin and the lysates of these cells were immunoblotted with each antibody. As shown in Fig. 4A, anti-JACOP pAb detected an ϳ160 kDa band in JACOP-expressing L cells but not in cingulin-expressing L cells. Conversely, anti-cingulin mAb recognized an ϳ140 kDa protein band in cingulin-expressing L cells but not in JACOPexpressing L cells. In the mouse lung lysates, anti-JACOP pAb and anti-cingulin mAb detected ϳ160 and ϳ140 kDa protein bands, respectively, confirming not only the specificity of these Abs but also the correct cloning of full-length mouse JACOP and cingulin cDNAs. The specificity of these antibodies then was also examined by immunostaining L cells transiently expressing HA-tagged JACOP or HA-tagged cingulin. As shown in Fig. 4B, anti-JACOP pAb and anti-cingulin mAb specifically recognized each antigen and no cross-reaction was detected. Thus, we concluded that these Abs can be used to compare the expression and distribution of JACOP and cingulin. During the course of antibody production, we noticed that some anti-JACOP Abs cross-reacted with cingulin. For example, a rabbit pAb raised against a GST fusion protein with aa 5-271 of JACOP cross-reacted to cingulin both in immunofluorescence staining and in immunoblotting (data not shown).
Tissue Distribution and Subcellular Localization of JACOP Protein-Frozen sections of mouse tissues were immunofluorescently stained with anti-JACOP pAb and anti-cingulin mAb. Consistent with our previous studies performed in chicken tissues, the junctional complex regions of most epithelial cells were labeled with anti-JACOP pAb in mouse tissues. JACOP and cingulin showed identical staining patterns at the junctional complex of epithelial cells in the lung (Fig. 5, a and b). In the kidney, JACOP was detected at the junctional complex of renal epithelial cells together with cingulin although the intensity of the staining signal differed depending on the type of renal tubule (Fig. 5, c and d). In contrast, JACOP was not detected at the junctional complex of epithelial cells in intestinal villi where cingulin was concentrated in large amounts (Fig. 5, e and f). Interestingly, in some tissues, non-junctional staining of JACOP was detected. In the liver, JACOP was concentrated at the junctional complex along the bile canaliculus colocalized with cingulin but JACOP signal was also detected along the lumen of canaliculi (Fig. 5, g and h). This luminal staining of JACOP overlapped with the phalloidin staining, suggesting that JACOP is distributed along actin bundles underlying the luminal plasma membranes of bile canaliculi in addition to the junctional complex (Fig. 6, a-c). Similarly, JACOP was localized at basal phalloidin-positive regions of renal epithelial cells, suggesting that it is colocalized with actin filaments in these regions (Fig. 6, d-f). JACOP was also concentrated at the cell-cell junctions of endothelial cells in various tissues in which cingulin expression was not detected (Fig. 5, e and f). To evaluate the specificity of the staining with anti-JACOP pAb, we next performed competition experiments. Sections of the kidney were stained with anti-JACOP pAb in the presence of the GST-JACOP fusion protein containing aa 493-633 of mouse JACOP or the GST-cingulin fusion protein containing aa 216 -419 of mouse cingulin as a control. As expected, the staining was competed away very effectively by pretreatment of anti-JACOP pAb with the GST-JACOP fusion protein, whereas pretreatment with the GST-cingulin fusion protein did not affect the staining ability (Fig, 5, i-l). The same results were obtained in other tissues (data not shown). Furthermore, in immunoblotting of whole lysates of mouse lung, kidney, and liver, a single band of ϳ160 kDa was predominantly detected with anti-JACOP pAb with no treatment (data not shown) or pretreated with the GST-cingulin fusion protein (Fig. 4C). This band was undetectable with anti-JACOP pAb pretreated with the GST-JACOP fusion protein (Fig. 4C). These results confirmed the specificity of our anti-JACOP pAb for tissue staining. In immunoblot analyses, we could not detect the band of ϳ160 kDa in the lysate of small intestine (data not shown), probably because of the very low expression level of JACOP as indicated in Fig. 3.
The distribution of JACOP in mouse cell lines then was examined by immunofluorescence staining. In Eph4 mammary epithelial cells, an intense signal for JACOP was detected at cell-cell junctions (Fig. 5, i and j). Interestingly, even in fibroblasts such as NIH3T3 cells, which did not have TJs, JACOP was localized to cadherin-based spotlike AJs where ZO-1 was concentrated (Fig. 5, k and l).
We further evaluated the precise localization of JACOP in the junctional complex by immunofluorescence microscopy. We compared the distribution of JACOP with that of cingulin or afadin (a marker for TJs or AJs, respectively) in the isolated bile canaliculi. In isolated bile canaliculi, the beltlike junctional complex occurs in two parallel lines and in individual complexes, TJs and AJs are aligned in this order from the lumen side. On double staining with anti-JACOP and anti-cingulin antibodies, both fluorescent signals overlapped with each other at the junctional complex (Fig. 7A, a-d). JACOP was also well colocalized with ZO-1 (data not shown). In contrast, the JACOP signal was clearly detected from a more luminal side than the afadin signal (Fig. 7A, e-h). These observations indicated that JACOP is localized at TJs in the liver.
Finally, the precise localization of JACOP was examined by immunoelectron microscopy. In the liver, consistent with the results of immunofluorescence staining of bile canaliculi, immunolabeling for JACOP was detected in the cytoplasm underlying TJs in the area of the junctional complex (Fig. 7B, a and  b). However, in some other tissues, the distribution of JACOP did not appear to be restricted to the undercoat of TJs. For example, in the kidney, JACOP appeared to be localized to the undercoat of both TJs and AJs (Fig. 7B, c and d). This is consistent with our previous immunoelectron microscopic observation that JACOP was expressed at fibrous structures (actin filaments) associated with AJs and TJs in the homogenized plasma membrane fraction of chicken renal epithelial cells (37).
The Domains of JACOP Required for the Recruitment to Cell-Cell Junctions-JACOP consists of head, rod, and tail FIG. 5. Immunofluorescent localization of JACOP in mouse tissues and cultured cells. Frozen sections of mouse lung, kidney, small intestine, and liver were double-stained with anti-JACOP pAb and anti-cingulin mAb. In the lung (a and b), JACOP and cingulin were colocalized at the cell-cell junctions of epithelial cells in bronchioles. In the kidney (c and d), JACOP and cingulin were colocalized at cell-cell junctions of renal epithelial cells (arrows), and in addition, JACOP, but not cingulin was detected faintly at basal plasma membranes of renal tubules (arrowheads). In the small intestine (e and f), JACOP was not detected at the TJs (arrowheads) where cingulin was concentrated in large amounts. Instead, JACOP was concentrated at cell-cell junctions of endothelial cells where cingulin was undetectable (arrows). In the liver (g and h), JACOP was concentrated not only at cingulin-positive cell-cell junctions along bile canaliculi but also on the canalicular luminal surfaces (insets). The staining signal was undetectable with anti-JACOP pAb pretreated with the GST-JACOP fusion protein (k). Pretreatment of anti-JACOP pAb with GST-cingulin fusion protein did not affect the staining ability (i). j and l are the corresponding phase-contrast images for i and k, respectively. Mouse Eph4 mammary epithelial cells (m and n) and NIH3T3 fibroblasts (o and p) were doublestained with anti-JACOP pAb and anti-ZO-1 mAb. JACOP was colocalized with ZO-1 at apical cell-cell junctions in Eph4 cells. In NIH3T3 cells, JACOP was colocalized with ZO-1 at cell-cell contact sites (arrowheads). Scale bars, 20 (a-l); 5 (insets); and 10 m (m-p).
domains. We then determined the domains responsible for the recruitment of JACOP to cell-cell junctions. When transiently overexpressed in Eph4 epithelial cells, the full-length JACOP with a HA epitope tag was recruited to cell-cell junctions (Fig.  8B, a and b). We then transfected Eph4 cells with the expression vector for the head or rod-tail domain with the HA tag ( Fig.  8A) and examined the localization of these transiently expressed proteins. We found that the head domain of JACOP (aa 1-581) was recruited to cell-cell junctions, although it was also detected in large amounts in the cytoplasm (Fig. 8B, c and d).
When a shorter form of the head domain (aa 1-367) was overexpressed, it was concentrated into cell-cell junctions more clearly (Fig. 8B, e and f). Interestingly, the rod-tail domain (aa 582-1298) also appeared to be localized at cell-cell junctions (Fig. 8B, g and h). These observations indicate that JACOP has at least two distinct regions to be recruited to cell-cell junctions in epithelial cells.
Recruitment of Overexpressed JACOP to Actin Filaments-JACOP was shown to be associated with the undercoat of TJs and AJs and also to be distributed along non-junctional actin filaments in some tissues (Fig. 6). These observations suggested that JACOP interacts directly or indirectly with the circumferential actin bundles at the junctional complex. Unfortunately, our in vitro cosedimentation assay of the recombinant JACOP with actin filaments did not work well because the recombinant JACOP protein easily self-aggregated even in the absence of actin filaments (data not shown). Therefore, we examined the localization of the overexpressed JACOP in fibroblasts bearing stress fibers, well developed actin filaments. As shown in Fig. 5, NIH3T3 cells express endogenous JACOP, which can be detected by immunofluorescence staining only at cadherin-based spotlike AJs. However, when JACOP was overexpressed in these cells, exogenous JACOP with an epitope tag was found at stress fibers in addition to cell-cell contact sites, suggesting direct/indirect interaction of JACOP with actin filaments (Fig. 9, a and b). The introduction of several deletion constructs of JACOP into NIH3T3 cells revealed that the head domain of JACOP (aa 1-581) and the shorter construct (aa 1-367), but not the rod-tail domain (aa 582-1298), had the ability to associate with actin filaments (Fig. 9, c-h). DISCUSSION In this study, we have cloned cDNA encoding JACOP, a novel plaque protein localizing at the apical junctional complex including TJs and AJs. Its amino acid sequence showed significant homology to that of cingulin, a previously characterized TJ plaque protein. Furthermore, in some tissues, JACOP was distributed along non-junctional actin filament bundles in addition to the junctions and its N-terminal half was shown to be responsible for both its junction-and actin filament-based distribution. Therefore, JACOP appeared to be similar to cingulin not only structurally but also functionally and would be involved in the molecular linkage between the apical junctional complex and actin-based cytoskeletons. A remarkable feature in the structure of JACOP is its coiled-coil domain in the C-terminal half, which shows 39% sequence homology with that of cingulin at the amino acid level. Since biochemical analyses showed that cingulin forms a parallel dimer through its coiled-coil rod domain (25), JACOP may also form a dimer or A, double immunofluorescence staining of isolated bile canaliculi with anti-JACOP pAb and anti-cingulin mAb (a-d) or with anti-JACOP pAb and anti-afadin mAb (e-h). d and h, the merged images in c and g were enlarged, respectively. JACOP was colocalized with cingulin, a marker of TJs (a-d) but was distributed more on the luminal side of bile canaliculi than was afadin, a marker of AJs (e-h). Scale bars: 5 (a-g) and 3 m (d and h). B, immunoelectron microscopy. Formalin-fixed mouse liver (a and b) and kidney (c and d) were immunolabeled with anti-JACOP pAb, which was detected with colloidal gold-conjugated secondary antibodies followed by silver enhancement. JACOP was detected at TJs within the junctional complex of hepatocytes (a and b). By contrast, in both proximal and distal renal tubules, JACOP was clearly detected along AJs in addition to TJs. DS, desmosomes. Scale bar, 300 nm.
oligomer through its rod domain. Indeed, when the rod-tail fragment of JACOP was overexpressed, it was targeted to junctions, probably through its direct interaction with the rod domain of endogenous JACOP. Interestingly, when overex-pressed, the head domain fragment of JACOP was also targeted to junctions. In the head domain, there are two short regions that showed significant homology between JACOP and cingulin. Recently, Citi and colleagues (26) narrowed down the domains of cingulin required for its interaction with ZO-1 and ZO-2 based on results of in vitro binding assays using various deletion constructs. Interestingly, the ZO-1 binding domain, which was mapped to aa 34 -48 in the head domain of mouse cingulin (26), corresponds to one of the homologous regions between JACOP and cingulin (Fig. 2C). Taken together, the head domain of JACOP may be recruited to cell-cell junctions by direct binding to ZO-1, which remains to be demonstrated in future studies.
Close inspection revealed that the manner in which JACOP is distributed in the junctional complex was not simple. In the liver, it was clearly concentrated at TJs, but in the kidney, it appeared to be distributed both at TJs and at AJs. However, considering that JACOP was concentrated fairly far from the plasma membranes as compared with ZO-1/ZO-2, it would not be meaningful to further discuss this point. Instead, it is of interest to compare the localization of JACOP with that of other TJ plaque proteins, which were also reported to distribute far from plasma membranes of TJs, such as cingulin (16), 7H6 antigen (24), and Pilt (32), in various tissues in detail. JACOP was localized to cell-cell contact sites in cultured fibroblasts lacking TJs, indicating that it is a component of cadherin-based AJs in these cells. Similarly, Citi and colleagues (26) reported that overexpressed cingulin was recruited to cadherin-based cell-cell contacts in Rat-1 fibroblasts. This behavior of JACOP/cingulin is similar to that of ZO-1/ZO-2 (20,21). If JACOP binds to ZO-1/ZO-2, this behavior appears reasonable.
JACOP was detected not only on the cytoplasmic side of the apical junctional complex where the circumferential actin bundles occur but also along non-junctional actin filaments in some tissues such as the liver and kidney. Furthermore, when over- ; rod (closed box); and tail (shaded box) from the N terminus. All of the constructs were tagged with HA at their N termini (j1-581 and j1-367) or C termini (JACOP-HA and j582-1298). B, Eph4 cells transiently expressing HA-tagged full-length or truncated JACOP were double-stained with mouse anti-HA-tag mAb and rat anti-ZO-1 mAb (a-h). j1-367 (e and f) was recruited to cell-cell junctions to the same extent as JACOP-HA (a and b). j1-581 was recruited to cell-cell junctions (arrows) with concomitant intense cytoplasmic staining (c and d). Similarly, j582-1298 also appeared to be recruited to cell-cell junctions in transient transfectants in addition to its diffuse cytoplasmic distribution (arrows in g and h). This junctionassociated localization was clearly enhanced in stable transfectants (j582-1298*, i and j). Scale bar, 10 m.  d, f, and h). JACOP-HA was distributed along stress fibers (a and b). j1-581 and j1-367, the head domain constructs, were colocalized with stress fibers (c-f), whereas j582-1298, the rod/tail domain construct, was distributed diffusely in the cytoplasm (g and h). Asterisk, a non-transfected cell. Scale bar, 10 m. expressed in fibroblasts, JACOP was recruited to stress fibers in addition to cell-cell contact sites. Therefore, from the side of actin-based cytoskeletons, JACOP appeared to be always associated with actin filaments. Based on these observations, another possibility that has to be considered is that the interaction with circumferential actin bundles rather than the interaction with junction-associated molecules such as ZO-1/ ZO-2 plays an important role for the recruitment of JACOP to cell-cell junctions. It will be of interest in future studies to investigate whether the N-terminal construct of JACOP can be further divided into two domains: one recruited to cell-cell junctions in epithelial cells but not to stress fibers in fibroblasts and the other recruited to stress fibers in fibroblasts but not to epithelial cell-cell junctions.
Most of the actin filaments underlying the apical junctional complex are associated with AJs, interacting with ␣-actinin, vinculin, myosin II, and so forth (4), but some junctional actin filaments have an intimate spatial relationship with TJs (44). It has been believed that these perijunctional actin filaments are involved in the maintenance and regulation of not only the cell-cell adhesion function of AJs but also the barrier function of TJs (4,45,46). For example, treatment of cultured epithelial cells with actin-disrupting drugs like cytochalasin led to the perturbation of perijunctional actin filaments and disrupted the structure of TJs (47,48). Signaling proteins controlling the assembly and contraction of actin filaments such as Rho family small GTPases and myosin light chain kinase markedly affected the barrier function of TJs (18, 49 -51). Taken together, it is reasonable to speculate that the barrier function of TJs is regulated by cortical tension generated by actomyosin contraction and that JACOP would be involved in this regulatory mechanism.