aPKC phosphorylates JAM-A at Ser285 to promote cell contact maturation and tight junction formation

In multicellular organisms, epithelial cells cover organs and body cavities to generate a selective barrier between distinct compartments. Epithelial cells develop apicobasal polarity reflected by a defined organization of intercellular junctional complexes, the existence of distinct plasma membrane domains, and the asymmetric distribution of molecules. The epithelium is sealed by tight junctions (TJs), which form at the most apical part of cell–cell contacts (Tsukita et al., 2001). TJs are crucial for the barrier function of epithelial cells because they restrict the diffusion of ions and macromolecules along the intercellular cleft (paracellular diffusion barrier; Van Itallie and Anderson, 2004). In addition, TJs prevent the free diffusion of proteins and lipids between the apical and the basolateral membrane domain (intramembrane diffusion barrier; van Meer and Simons, 1986), implicating them in the regulation of apicobasal membrane polarity.

TJs are composed of various integral membrane proteins, cytoplasmic scaffolding proteins, and adaptor proteins as well as regulatory proteins, including kinases and phosphatases, small GTPases, and guanine nucleotide exchange factors (Matter and Balda, 2003b; Ebnet, 2008). Two major cytoplasmic scaffolding protein complexes are the PAR-3–atypical PKC (aPKC)–PAR-6 complex and the Pals1–PATJ complex (Macara, 2004). Both complexes are required for TJ formation as inferred from knockdown studies and from ectopic expression of dominant-negative mutant proteins (Shin et al., 2006; Suzuki and Ohno, 2006). PAR-3 and PAR-6 serve as scaffolding proteins to regulate the localization and Cdc42/Rac1-mediated activation of aPKC, respectively. The Pals1–PATJ complex consists of the two scaffolding proteins Pals1 and PATJ, which have no catalytic activity. However, this complex can be physically linked to the PAR–aPKC–PAR-6 complex (Hurd et al., 2003). In addition, it is linked to the Cdc42-specific Rho GTPase-activating protein Rich1, through which it may indirectly influence the activity of the PAR–aPKC complex (Wells et al., 2006). Together, these observations place aPKC at the center of a protein network that regulates the formation and integrity of TJs in epithelial cells.

During cell–cell contact formation, aPKC interacts with different scaffolding proteins and phosphorylates various target proteins. At early phases of cell–cell contact formation, it forms a ternary complex with PAR-6 and Lethal giant larvae (Lgl; Yamanaka et al., 2003). The association of Lgl with aPKC–PAR-6 prevents the interaction of aPKC–PAR-6 with PAR-3. aPKC activation leads to Lgl phosphorylation and its segregation from the aPKC–PAR-6 complex (Yamanaka et al., 2003), allowing aPKC–PAR-6 to associate with cell–cell contact-associated PAR-3 and to form an active PAR-3–aPKC–PAR-6 complex at those sites. Active aPKC then phosphorylates a defined set of target proteins, such as PAR-1 or Numb, leading to their exclusion from the aPKC-containing membrane domain (Hurov et al., 2004; Suzuki et al., 2004; Smith et al., 2007; Morais-de-Sá et al., 2010). In turn, PAR-1 phosphorylates PAR-3, which prevents PAR-3 oligomerization and its stable localization at the membrane (Benton and St Johnston, 2003a,b; Mizuno et al., 2003). These mutual phosphorylations regulate the formation of distinct membrane domains. Once TJs are formed, the activity of aPKC at TJs is most likely continuously required to maintain their functional integrity. Furthermore, it is likely that aPKC activity is not only used to exclude basolateral membrane markers, such as Lgl or PAR-1, from the apical contact region but also to regulate the function or activity of other components within the TJs (Aono and Hirai, 2008).

A putative candidate protein subject to phosphorylation by aPKC is the Ig superfamily member junctional adhesion molecule A (JAM-A). In polarized epithelial cells, JAM-A localizes to lateral cell–cell contacts and is enriched at TJs (Martìn-Padura et al., 1998; Liu et al., 2000). During cell–cell contact formation, JAM-A localizes to the earliest sites of cell–cell interactions, the primordial, spotlike adherens junctions (AJs; pAJs), in which it serves to recruit PAR-3 and to assemble an active PAR-3–aPKC–PAR-6 complex (Ebnet et al., 2001; Itoh et al., 2001). Cells expressing JAM-A mutants that do not localize at cell–cell contacts fail to develop functional TJs and apicobasal membrane polarity (Rehder et al., 2006). These observations prompted us to analyze whether phosphorylation of JAM-A by aPKC is part of the molecular mechanism by which aPKC regulates the formation of TJs and the development of apicobasal polarity.

We analyzed JAM-A phosphorylation by incubating GST–JAM-A fusion proteins containing the entire cytoplasmic domain of JAM-A (Y₂₆₁–V₃₀₀; Fig. 1 A) with aPKCζ in the presence of radiolabeled ATP followed by either conventional SDS-PAGE or by chymotryptic digest and two-dimensional analysis of the resulting peptides. These experiments indicated that aPKCζ phosphorylates JAM-A and that S285 is a major aPKCζ phosphorylation site (Fig. S1). The relatively strong autoradiography signals remaining in all single, double, or triple mutants prompted us to also mutate S262, which reflects an artificial PKC consensus site resulting from the fusion of GST sequences with JAM-A sequences. Conventional SDS-PAGE indicated that JAM-A phosphorylation was reduced after mutating either S262 or S285 (Fig. 1 Bi). Chymotrypsin treatment and phosphopeptide analysis revealed two major phosphopeptide spots (Fig. 1 Bii, top left, 1b and 3). Mutating S285 to Ala abolished spot 3 (Fig. 1 Bii, red circles). Mutating S262 abolished spot 1b as well as one of the minor spots labeled 1a (Fig. 1 Bii, red circles). To verify that S262 is phosphorylated because of the presence of an artificial PKC consensus site, we generated new fusion proteins in which three additional JAM-A–specific amino acids were inserted between the GST sequence and S262 of JAM-A. This construct showed almost no ³²P incorporation when S285 was mutated to Ala (Fig. 1 Bii, bottom right). Thus, these observations indicate that aPKCζ phosphorylates JAM-A strongly and exclusively at S285 in vitro.

To address whether JAM-A is phosphorylated in cells, we metabolically labeled confluent KLN205 cells using [³²P]orthophosphate, immunoprecipitated endogenous JAM-A, and performed phosphoamino acid analyses. SDS-PAGE and autoradiography indicated a phosphorylated band that was identified as JAM-A by Western blotting (Fig. 1 Ci). This band was excised from the membrane and subjected to hydrolysis followed by two-dimensional separation on thin-layer cellulose plates. Autoradiography revealed a prominent signal for phosphorylated serine but not threonine or tyrosine residues (Fig. 1 Cii), indicating that in the cell, JAM-A is phosphorylated exclusively on serine residues.

To analyze whether aPKCζ and JAM-A form a complex, we performed coimmunoprecipitation and pull-down experiments. JAM-A was detectable in aPKCζ immunoprecipitates obtained from transiently transfected HEK293T cells (Fig. 2 A) as well as in immunoprecipitates obtained from Caco-2 epithelial cells, a human colonic epithelial cell line that adopts a polarized morphology with well-developed TJs (Fig. 2 B). In vitro translated aPKCζ bound to recombinant JAM-A as strong as to its direct binding partner PAR-6 (Joberty et al., 2000), indicating that the interaction between aPKCζ and JAM-A can be direct and does not require PAR-3 as a scaffold (Fig. 2 C). In line with this observation, JAM-A deletion mutants lacking the PDZ domain binding motif required for the interaction with PAR-3 (Ebnet et al., 2001) were readily detectable in aPKCζ immunoprecipitates (Fig. 2 D). Together, these findings indicate that aPKCζ forms a complex with JAM-A in epithelial cells and that this interaction can be direct and does not require PAR-3.

To analyze the relevance of JAM-A phosphorylation at S285 in cells, we generated phospho-S285 (S285-P)–specific pAbs (Fig. S2). Using these antibodies, we analyzed JAM-A S285 phosphorylation in MTD-1A epithelial cells, a polarizing epithelial cell line with well-developed TJs (Suzuki et al., 2002). MTD-1A cells grown to confluence showed a strong S285-P signal at cell–cell contacts (Fig. 3 A). When the cells were grown under low Ca²⁺ to disrupt cell–cell contacts (Gumbiner and Simons, 1986), JAM-A appeared at the periphery of rounded cells. Interestingly, the S285-P signal was completely absent in these contact-naive cells. When cells were replenished with Ca²⁺ (Ca²⁺ switch [CS]) for 3 or 8 h to induce new cell–cell contact formation, the S285-P signal colocalized with total JAM-A at maturing cell–cell contacts (Fig. 3 A). These findings indicate that JAM-A is phosphorylated at S285 in confluent cells, is dephosphorylated during disruption of cell–cell contacts, and is rephosphorylated during CS-induced new cell–cell contact formation.

During cell–cell contact formation E-cadherin, ZO-1, and JAM-A are the first proteins that appear at so-called pAJs (also called puncta), whereas PAR-3, aPKC, and PAR-6 appear later (Yonemura et al., 1995; Suzuki et al., 2002). The sequential appearance of JAM-A and aPKCζ suggests that JAM-A phosphorylation at S285 occurs only after aPKCζ has been recruited. To address this question, MTD-1A cells were scratch wounded to induce new junction formation (Fig. 3 B). Total JAM-A was detected in the center of the wounds in a spotlike pattern typical for proteins present at pAJs (Fig. 3 B, top). S285 phosphorylation of JAM-A was not detected at pAJs but was detected only when pAJs had matured into more linear cell–cell contact structures. These observations indicate that JAM-A phosphorylation at S285 occurs with a significant delay after its localization at pAJs. As observed previously (Suzuki et al., 2002), aPKCζ was absent at pAJs. It appeared at more mature cell–cell contact sites with a delay similar to JAM-A S285 phosphorylation (Fig. 3 B, bottom). The similarity in the kinetics of appearance of aPKCζ and of S285 phosphorylation of JAM-A suggested that aPKCζ phosphorylates JAM-A after its recruitment to pAJs. To address this question, we inhibited aPKCζ activity using the aPKC-specific inhibitor PSζ (pseudosubstrate ζ). Incubation of wounded cells with PSζ almost completely abolished the S285 phosphorylation signal and resulted in a diffuse intracellular staining (Fig. 3 C). Together, these observations suggest the following hierarchy of events: JAM-A localizes to pAJs in the absence of S285 phosphorylation, in which it serves to recruit PAR-3. Subsequently, after assembly of the PAR-3–aPKC–PAR-6 complex at pAJs and activation of aPKC by Rac1/Cdc42 small GTPases, aPKC phosphorylates JAM-A at S285. The direct and PAR-3–independent interaction of aPKCζ with JAM-A allows a continuous S285 phosphorylation of JAM-A despite the dissociation of aPKCζ from PAR-3 as a result of aPKC-mediated S827 phosphorylation of PAR-3 (Nagai-Tamai et al., 2002; Morais-de-Sá et al., 2010).

To test whether the early phosphorylation of JAM-A/S285 is required for cell–cell contact formation, we analyzed the recruitment of ZO-1 as early marker for nascent cell–cell contacts during CS-induced new contact formation (Ando-Akatsuka et al., 1999). MDCK II cells stably expressing JAM-A/wild type (WT) or JAM-A/S285A under a doxycycline-regulated promoter (MDCK II Tet-Off cells; Rehder et al., 2006) were subjected to CS to induce new contact formation. 2 h after readdition of Ca²⁺, ZO-1 immunoreactivity significantly increased in cells overexpressing JAM-A/WT (Fig. 4 A), indicating that JAM-A promotes junctional maturation. In contrast, overexpression of JAM-A/S285A significantly decreased ZO-1 immunoreactivity (Fig. 4 A). The delay in ZO-1 recruitment in JAM-A/S285A–expressing cells was less pronounced 4 h after CS, suggesting that JAM-A/S285 phosphorylation is important during the early steps of junctional maturation (Fig. 4 B). A similar pattern was observed for β-catenin and occludin (Fig. 4 B). Together, these observations indicate that aPKC-mediated S285 phosphorylation of JAM-A promotes cell–cell contact maturation at early time points of contact formation.

Because JAM-A localizes along the entire cell–cell contact of epithelial cells (Liu et al., 2000), we analyzed JAM-A/S285 phosphorylation in highly polarized MTD-1A cells. In contrast to total JAM-A, S285-P JAM-A showed a restricted distribution at the apical cell–cell contact area (Fig. 5 A). Stainings for the two TJ-specific proteins occludin (Fig. 5 A) or ZO-1 (Fig. S3 A) showed a complete overlap of S285-P JAM-A with both proteins, indicating that S285-P JAM-A localizes exclusively to the TJs. In addition, S285-P JAM-A completely overlaps with aPKCζ and PAR-3 (Fig. S3 B). Treatment of polarized MTD-1A cells for 2 h with PSζ inhibitor (Fig. 5 B) or knockdown of aPKCζ using siRNA (Fig. 5 C) abolished the pAb S285-P immunoreactivity, indicating that S285 phosphorylation of TJ-associated JAM-A is mediated by aPKCζ. To test whether S285 phosphorylation regulates the targeting of JAM-A to the TJs, we ectopically expressed a phosphodeletion mutant (S285A) and a phosphomimicking mutant (S285D) of JAM-A in MDCK II cells. Both mutants colocalized with ZO-1 at the apical region of the cell–cell contacts (Fig. S3 C), indicating that S285 phosphorylation does not regulate the localization of JAM-A to TJs. In support of this, blocking JAM-A S285 phosphorylation with PSζ or by aPKCζ knockdown did not change the localization of JAM-A at TJs (Fig. 5, B and C). Together, these findings suggest a TJ-specific function of aPKC-mediated JAM-A S285 phosphorylation.

As JAM-A is S285 phosphorylated exclusively at TJs, we investigated whether aPKCζ-mediated phosphorylation of JAM-A is antagonized by PP2A, a protein phosphatase, which localizes at TJs and regulates TJ formation (Nunbhakdi-Craig et al., 2002). JAM-A was detectable in PP2A immunoprecipitates obtained from transiently transfected HEK293T cells as well as from nontransfected Caco-2 cells (Fig. 6, A and B), indicating that JAM-A associates with PP2A in polarized epithelial cells. Incubation of aPKCζ-phosphorylated GST–JAM-A with recombinant PP2A almost completely abolished JAM-A S285 phosphorylation (Fig. 6 C). Deleting either six or nine C-terminal amino acid residues of JAM-A did not abolish the association with PP2A (Fig. 6 D), indicating that PP2A binding is not mediated through a PDZ domain protein associated with JAM-A. Together, these observations indicate that PP2A interacts with JAM-A and dephosphorylates JAM-A at S285. The findings suggest that PP2A antagonizes aPKCζ-mediated phosphorylation of JAM-A at TJs.

The development of trans-epithelial electrical resistance (TER) is a hallmark of barrier-forming epithelial cells (Matter and Balda, 2003a). To study the role of S285 phosphorylation of JAM-A for TER development, we used MDCK II Tet-Off cells stably expressing JAM-A/WT or JAM-A/S285A. Induced expression of WT JAM-A did not significantly affect the development of TER (Fig. 7 A) as observed previously (Rehder et al., 2006). Expression of JAM-A/S285A resulted in a strong decrease in TER, indicating that the barrier function of TJs is severely impaired when JAM-A cannot be phosphorylated at S285. Expression of JAM-A/S285A also significantly increased the permeability for a 4.3-kD TRITC-dextran when compared with JAM-A/WT expression (Fig. 7 B), indicating that the barrier for small molecular mass hydrophilic molecules is also impaired in the absence of JAM-A S285 phosphorylation. The defect in the barrier function cannot be explained by an inability to form TJ strands, as expression of JAM-A/S285A did not affect the formation of TJ strands (Fig. 7 C). In addition, the defect in TJ function is unlikely to be caused by a mislocalization of PAR-3 because recruitment of PAR-3 to cell–cell contacts and to TJs is not affected by the S285A mutation (Fig. S4, A and B). Together, these observations suggest that aPKCζ-mediated phosphorylation of JAM-A at S285 is critical for the formation of a functional barrier for ions and small hydrophilic molecules in epithelial cells.

The maintenance of apicobasal membrane polarity requires a diffusion barrier that prevents the intermixing of apical and basolateral membrane components. This intramembrane diffusion barrier colocalizes with the paracellular diffusion barrier at TJs but is regulated by a different molecular mechanism (Umeda et al., 2006). To analyze whether JAM-A S285 phosphorylation is required for the development of apicobasal polarity, we expressed JAM-A/S285A in MDCK II epithelial cells and analyzed lumen formation in three-dimensional cyst assays (O’Brien et al., 2002). In the absence of JAM-A/S285A expression, MDCK cells developed normal cysts with a single lumen surrounded by a monolayer of epithelial cells (Fig. 8). Expression of JAM-A/S285A resulted in the formation of cysts, which, instead of a single lumen, contained multiple lumens. Interestingly, apicobasal polarity of individual cells was not disturbed, as indicated by the apical localization of F-actin in cells surrounding the lumens (Fig. 8). These observations suggest that despite the role of JAM-A S285 phosphorylation in the regulation of the epithelial barrier function, S285 phosphorylation of JAM-A is not involved in the regulation of the intramembrane diffusion barrier.

During cell division, individual mitotic cells within the epithelial monolayer round up, but retain extensive contacts to the surrounding interphase cells (Baker and Garrod, 1993; Théry and Bornens, 2008). Apicobasal membrane polarity, as well as TJs, is maintained during all stages of mitosis (Reinsch and Karsenti, 1994; Kim and Raphael, 2007), and the barrier function of epithelial cells may even increase during mitosis (Soler et al., 1993). When stained for S285 phosphorylation, mitotic cells show a much stronger signal than the surrounding nondividing cells (Fig. 9). The increased phosphorylation of JAM-A persists to telophase and cytokinesis (Fig. 9, middle), suggesting that it is maintained throughout mitosis. The increase in signal intensity in mitotic cells was not observed with a polyclonal JAM-A antibody directed against the nonphosphorylated peptide (Fig. 9, bottom), indicating that it reflects increased S285 phosphorylation but not altered JAM-A recruitment to cell–cell contacts. Our findings thus indicate that during mitosis, JAM-A is hyperphosphorylated at S285, which may contribute to a stabilization of cell–cell contacts and barrier function in mitotic cells.

The formation of an apical junction complex, including TJs and AJs, in vertebrate epithelial cells is regulated by highly conserved cell polarity proteins, including aPKC. Once recruited to pAJs through its association with JAM-A–bound PAR-3 and activated through Cdc42/Rac1 small GTPases, aPKC promotes the maturation of primordial junctions into apical cell junctions, with TJs being separated from AJs (Suzuki et al., 2002). In this study, we find that shortly after the formation of pAJs, aPKCζ phosphorylates JAM-A at S285 to promote junctional maturation. In fully polarized epithelial cells, aPKCζ-phosphorylated JAM-A localizes specifically at the TJs, and this activity is most likely antagonized by PP2A. JAM-A S285 phosphorylation is required for the development of a functional epithelial barrier and for normal epithelial morphogenesis. We also find that JAM-A is hyperphosphorylated at S285 in mitotic cells. Our findings identify S285 of JAM-A as a novel mediator of aPKC activity during cell–cell contact maturation as well as during barrier function development in epithelial cells.

Cell–cell contact formation is a stepwise process that is characterized by the sequential localization of integral membrane and cytoplasmic proteins to sites of cell–cell adhesion. JAM-A is among the first cell adhesion molecules that appear at pAJs, and PAR-3 and aPKC are recruited later (Ebnet et al., 2001; Suzuki et al., 2002). Consistent with the predicted role for cell adhesion molecules to act as positional cues for the correct localization of cell polarity proteins (Nelson, 2003), the early localization of JAM-A most likely serves to recruit PAR-3 and promote the formation of an active PAR-3–aPKC–PAR-6 complex at these sites. Interestingly, we find that JAM-A is present at pAJs in a non-S285–phosphorylated state. S285 phosphorylation of JAM-A occurs after aPKC has been recruited to pAJs. These findings thus suggest that nonphosphorylated JAM-A present at pAJs recruits PAR-3 and promotes the assembly of the PAR-3–aPKC–PAR-6 complex. Through the binding of Rho family small GTPases to PAR-6, aPKC is activated to phosphorylate JAM-A at S285. JAM-A phosphorylation at S285 promotes the maturation of cell–cell contacts. These findings thus identify JAM-A as a novel target through which aPKC regulates junctional maturation after the establishment of pAJs.

Ectopic expression of a kinase-negative mutant of aPKC (aPKCkn) results in a more drastic phenotype than expression of JAM-A/S285A. As opposed to JAM-A/S285A–expressing cells, aPKCkn-expressing cells fail to develop the pAJs into mature cell–cell contacts, with TJs segregated from AJs (Suzuki et al., 2002). This suggests that contact maturation requires additional phosphorylation of proteins other than JAM-A. aPKC has been shown to phosphorylate PAR-1 as well as Numb, and in both cases, the phosphorylation leads to the exclusion of the proteins from the aPKC-containing membrane domain, which is part of the mechanism of how aPKC regulates the development of membrane asymmetry (Benton and St Johnston, 2003b; Hurov et al., 2004; Suzuki et al., 2004; Smith et al., 2007). In the case of JAM-A, however, phosphorylation by aPKC does not lead to its exclusion from the aPKC-containing membrane domain, indicating that JAM-A S285 phosphorylation serves a different purpose. In line with the role of JAM-A as a positional cue, we speculate that aPKC-mediated S285 phosphorylation of JAM-A generates new binding sites for the recruitment of proteins that regulate cell–cell contact maturation after pAJs have been formed. Pull-down experiments using phosphorylated JAM-A peptides combined with mass spectrometry will help to identify these proteins.

JAM-A is enriched at TJs, but a substantial amount of JAM-A molecules is localized along the lateral membrane (Liu et al., 2000; Rehder et al., 2006). We observed that in polarized epithelial cells, S285 phosphorylation of JAM-A is detected exclusively at TJs. Because a phosphodeficient JAM-A mutant (JAM-A/S285A) was not excluded from TJs, S285 phosphorylation does not regulate the localization of JAM-A at TJs. Therefore, it is more likely that the TJ-associated subpopulation of JAM-A, which is characterized by S285 phosphorylation, is either phosphorylated exclusively within TJs by aPKCζ or, alternatively, is phosphorylated by aPKCζ within a different subcellular compartment and is then transported to the TJs. This JAM-A population is continuously phosphorylated by aPKCζ as indicated by the loss of the S285 phosphorylation when fully polarized epithelial cells are incubated for 2 h with the PSζ inhibitor. Phosphorylation of JAM-A at S285 does probably not contribute to the overall maintenance of TJs as indicated by the apical restriction of occludin in PSζ-treated cells and by the apical localization of PAR-3 and the presence of TJ strands in JAM-A/S285A–expressing cells. JAM-A S285 phosphorylation by aPKC rather regulates specifically the barrier function of TJs. The mechanism by which JAM-A phosphorylation regulates the barrier function is presently unclear. We can rule out the possibility that it regulates the barrier by changing the expression levels of proteins involved in the formation of TJ strands because we observed no difference between JAM-A/WT– and JAM-A/S285A–expressing cells with regard to their influence on the expression of claudin-1, claudin-2, claudin-10, claudin-15, occludin, or ZO-1 (Fig. S5). It is possible that S285 phosphorylation serves to assemble a signaling module, which operates at the level of TJs to maintain their functional integrity. The presence of a Cdc42-dependent signaling module that regulates the maintenance of TJs has recently been described (Wells et al., 2006).

We have previously shown that ectopic expression of JAM-A cytoplasmic domain deletion mutants in MDCK II renal epithelial cells disrupts the epithelial barrier (Rehder et al., 2006). JAM-A knockout mice have a barrier and a morphological defect in the intestine and in the cornea but not in the kidney (Kang et al., 2007; Laukoetter et al., 2007). Most likely, tissue-specific factors contribute to the tissue-specific requirements in the barrier function. In support of this, we observed that JAM-A regulates the levels of claudin-2, but not claudin-10 and claudin-15, in MDCK II cells (Fig. S5), whereas the opposite has been observed in intestinal SK-CO15 cells (Laukoetter et al., 2007). These findings suggest that JAM-A contributes to the barrier function in a tissue-specific manner, via a S285 phosphorylation-independent mechanism involving the selective regulation of claudins.

PP2A antagonizes aPKC-mediated S285 phosphorylation in vitro. PP2A has been described to be localized at TJs, and a study using pharmacological inhibition or reduced expression of the catalytic subunit of PP2A indicated that PP2A negatively affects TJ formation and barrier function (Nunbhakdi-Craig et al., 2002). This activity of PP2A probably resides in its ability to dephosphorylate ZO-1 and occludin (Nunbhakdi-Craig et al., 2002; Seth et al., 2007). Besides these structural components of TJs, PP2A interacts with and dephosphorylates aPKCζ at T410 (Nunbhakdi-Craig et al., 2002), resulting in down-regulation of aPKC activity (Hirai and Chida, 2003). Thus, by targeting T410 of aPKCζ, PP2A can also indirectly contribute to the reduced phosphorylation of structural components of TJs, such as occludin, ZO-1, and claudin-4 (Nunbhakdi-Craig et al., 2002; Seth et al., 2007; Aono and Hirai, 2008). We identified PP2A in a complex with JAM-A in cells, and we found that PP2A dephosphorylates JAM-A at S285 in vitro. These findings point to PP2A as a negative regulator of JAM-A S285 phosphorylation. Although it is possible that other protein phosphatases, such as PP1, are involved in regulating JAM-A phosphorylation (Seth et al., 2007; Traweger et al., 2008), and although we cannot exclude the possibility that protein kinases, such as PKCα, regulate JAM-A phosphorylation at S285 under certain conditions in epithelial cells (Ozaki et al., 2000), the loss of the S285 phosphorylation in aPKCζ knockdown and PSζ-treated cells, the identification of PP2A and JAM-A in the same protein complex, and the ability of PP2A to regulate the activity of aPKCζ and to dephosphorylate JAM-A in vitro suggest that aPKCζ and PP2A cooperate in regulating the TJ-specific phosphorylation of JAM-A.

We also demonstrate that JAM-A is hyperphosphorylated at S285 during mitosis. Previous studies indicate that cell–cell contacts, including TJs, between mitotic and interphase cells are maintained during mitosis and cytokinesis (Baker and Garrod, 1993; Soler et al., 1993). The increased S285 phosphorylation of JAM-A in mitotic cells could be required to stabilize and maintain the functional integrity of TJs between mitotic and interphase cells to avoid a loss of the barrier function during mitosis. On the other hand, our observation that ectopic expression of JAM-AS/S285A in epithelial cells leads to a multiluminal phenotype when grown in a three-dimensional collagen matrix points to a role for JAM-A phosphorylation that is distinct from its role in regulating the barrier function. Recent evidence correlates mitotic spindle orientation with single lumen specification during cystogenesis, and several proteins associated with JAM-A, including aPKC, PAR-3, and PAR-6, have been implicated in this process (Jaffe et al., 2008; Mitsushima et al., 2009; Hao et al., 2010; Durgan et al., 2011), which opens the possibility that JAM-A is involved in spindle orientation. JAM-A has also been described to regulate epithelial proliferation (Laukoetter et al., 2007; Nava et al., 2011) and to be phosphorylated during mitosis at several serine residues, including the S285-corresponding S284 residue in HeLa cells (Dephoure et al., 2008; Gauci et al., 2009). Together, these observations point to an important role of JAM-A S285 phosphorylation during cell division, which may be independent of its function in regulating the cell–cell contact and epithelial barrier formation.

Caco-2 cells (American Type Culture Collection), HEK293T cells, KLN205 cells (Kaneko et al., 1980), MDCK II Tet-Off cells (BD), and MTD-1A cells (provided by M. Takeichi, Kyoto University, Japan) were grown in DME with 10% FCS, 2 mM glutamine, and 100 U/ml penicillin/streptomycin. MDCK II Tet-Off cells expressing Flag-tagged JAM-A/S285A were generated by transfecting MDCK II Tet-Off cells with pTRE2hyg expression vectors (BD). Transfected cells were selected in DME medium supplemented with 100 µg/ml G418, 1 µg/ml puromycin, 150 µg/ml hygromycin, and 50 ng/ml doxycycline. Flag–JAM-A/S285A expression was induced by growing cells in medium lacking doxycycline. CHO cells were maintained in α-MEM supplemented with 10% FCS, 2 mM glutamine, and 100 U/ml penicillin/streptomycin. Transient transfections of plasmids using either GeneJammer (Agilent Technologies) or FuGENE 6 (Roche) transfection reagents were performed according to the manufacturer’s instructions.

The following antibodies were used in this study: rat pAb anti–ZO-1, mouse mAb anti-PP2A (Millipore), mouse mAb antioccludin (Invitrogen), anti-Flag tag mouse mAb M2 and rabbit pAb (Sigma-Aldrich), rabbit pAb anti-aPKCζ, rabbit anti–human JAM-A (Santa Cruz Biotechnology, Inc.), and mouse anti–human JAM-A mAb (BD). mAbs and pAbs directed against murine JAM-A (H2O2-106-7-4 and pAb JAM-A extracellular domain [Affi828], respectively) have been previously described (Malergue et al., 1998; Rehder et al., 2006). A pAb against Ser285-phosphorylated JAM-A was generated by immunizing rabbits with a synthetic peptide comprising aa 282–294 (SQP[pSer]TRSEGEFKQ) of murine JAM-A coupled to ovalbumin. For affinity purification, specific antibodies were adsorbed against the phosphorylated peptide coupled to BSA (pAb S285-P). Antibodies against the remaining part of the peptide were removed by adsorption against the nonphosphorylated peptide, resulting in a pAb recognizing the same peptide in a nonphosphorylated form (pAb S285-N). The following reagents were used in this study: PKCζ PS peptide (PSζ; Sigma-Aldrich), recombinant aPKCζ, and PP2A (Millipore).

For transient expression of Flag-tagged JAM-A constructs, murine JAM-A cDNA (aa 25–300) or human JAM-A cDNA (aa 26–299) and C-terminal deletion constructs (hJAM-A/Δ3 [aa 26–296], hJAM-A/Δ6 [aa 26–293], hJAM-A/Δ9 [aa 26–290], and hJAM-A/Δ43 [aa 26–256, lacks entire cytoplasmic domain]) were cloned into the pFlag-CMV1 vector (Ebnet et al., 2001). For inducible expression in MDCK II cells, a pTRE2hyg-based vector encoding Flag-tagged JAM-A fusion proteins was used (Rehder et al., 2006). aPKCζ was cloned in pcDNA4-V5His (Invitrogen). The PP2A cDNA was provided by E. Sontag (University of Newcastle, Callaghan, Australia). For recombinant protein expression in Escherichia coli, the pGEX-4T-1 vector containing the entire cytoplasmic tail of murine JAM-A (aa 261–300 and aa 258–300; Ebnet et al., 2000) was used. GST–PAR-6C was generated by cloning full-length mouse PAR-6C in pGEX-6P-2. All point mutations were generated by a PCR-based approach using mismatch primer pairs.

To deplete JAM-A and aPKCζ in epithelial cells, mixtures of four siRNA duplexes (ON-TARGETplus SMARTpool siRNAs; Thermo Fisher Scientific) were used. 2 × 10⁶ cells were transfected with 100 pmol (JAM-A) or 20 pmol (aPKCζ) siRNAs by electroporation (Lonza) according to the manufacturer’s instructions. To obtain high knockdown efficiencies, the cells were harvested after 24 h and subjected to a second electroporation using the same siRNAs. After 48 h, the cells were fixed for immunofluorescence analysis. MDCK cells stably expressing JAM-A small hairpin RNAs (shRNAs) under a tetracycline-regulated promoter have been previously described (Rehder et al., 2006).

Phosphorylation experiments were performed essentially as previously described (Ebnet et al., 2003). GST fusion proteins were coupled to glutathione Sepharose (GE Healthcare) using buffer B (10 mM Hepes-NaOH, pH 7.4, 100 mM KCl, 1 mM MgCl₂, and 0.1% Triton X-100) and subjected to in vitro kinase reactions with either recombinant active aPKCζ or PKCα (Enzo Life Sciences). aPKCζ incubations were performed for 30 min at 30°C with 10 ng enzyme in the presence of 5 µCi γ-[³²P]ATP in PKCζ kinase buffer (18 mM Hepes, pH 7.4, 0.3 mM EDTA, 2 mM EGTA, 1% BSA, 400 µM DTT, 400 µM sodium pervanadate, 30 mM MgCl₂, 200 µM of cold ATP, 10 mM β-glycerophosphate, 50 µg/ml phosphatidylserine, 5 µg/ml diacylglycerides, and 10 µg/ml leupeptin). Similarly, kinase assays with PKCα were performed with 10 ng recombinant enzyme and 5 µCi γ-[³²P]ATP in PKCα kinase buffer (14 mM Hepes, pH 7.4, 700 µM DTT, 700 µM sodium pervanadate, 700 µM CaCl₂, 900 µM sodium fluoride, 15 mM MgCl₂, 100 µM of cold ATP, 18 mM β-glycerophosphate, 50 µg/ml phosphatidylserine, 5 µg/ml diacylglycerides, and 12 µg/ml leupeptin) for 30 min at 30°C. For in vitro dephosphorylation assays, the phosphorylated GST fusion proteins were washed and then incubated for 30 min at 30°C with recombinant PP2A (Millipore). The GST fusion proteins were washed and subsequently eluted from the beads by boiling in SDS sample buffer. Supernatants were separated by SDS-PAGE and analyzed by autoradiography.

For phosphopeptide mapping, the bands corresponding to JAM-A were eluted from the polyacrylamide gels and prepared as previously described (Boyle et al., 1991). In brief, the protein was precipitated using trichloroacetic acid in the presence of RNase A (Sigma-Aldrich), oxidized with freshly prepared performic acid, and digested with 0.5 U chymotrypsin (Worthington Biochemical Corporation) for 3–5 h at 37°C. An additional 0.5 U enzyme was added and incubated as in the previous paragraph to ensure complete digestion. The separation of chymotryptic peptides in the first dimension was performed in pH 1.9 buffer (2.5% [vol/vol] formic acid and 7.8% [vol/vol] glacial acetic acid) using a thin-layer electrophoresis apparatus (Hunter HTLE-7000; C.B.S. Scientific Company, Inc.) followed by thin-layer chromatography along the second dimension in phosphochromatography buffer (37.5% [vol/vol] n-butanol, 25% [vol/vol] pyridine, and 30% [vol/vol] glacial acetic acid). Phosphopeptides were visualized by autoradiography after 10 d of exposure.

Confluent KLN205 cells were washed in phosphate-free DME and subsequently metabolically labeled for 20 h in phosphate-free DME supplemented with 0.5 mCi/ml [³²P]orthophosphate and 0.5% dialyzed FCS. Cells were lysed in lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.5% [vol/vol] Triton X-100, 10 µg aprotinin/ml, 1 mM Pefabloc, 10 µg leupeptin/ml, 1 µg Pepstatin A/ml, and 1 mM DTT), and endogenous JAM-A was immunoprecipitated using affinity-purified polyclonal rabbit antibodies. Phosphorylated proteins were resolved by SDS-PAGE and transferred to polyvinylidene fluoride membranes. After excision of the bands corresponding to JAM-A, amino acids were released by acid hydrolysis in constantly boiling hydrochloric acid for 1 h at 100°C and separated by two-dimensional electrophoresis on thin-layer cellulose plates using the thin-layer electrophoresis apparatus (Hunter HTLE-7000) according to the manufacturer’s protocol. Separation of phosphoamino acids was controlled by addition of phosphoserine, phosphothreonine, and phosphotyrosine standards, which were visualized by ninhydrin staining. Radioactively labeled phosphoamino acids originating from the sample were visualized by autoradiography after 8 d of exposure.

For CS experiments, adherent cells were incubated in low Ca²⁺ medium (Ca²⁺-free DME; Invitrogen; 5% dialyzed FCS, 4 mM l-glutamine, and 3 µM CaCl₂) for 16–20 h. To induce new cell contact formation, the medium was replaced by medium supplemented with 1.8 mM CaCl₂ (CS). The rate of junction assembly was determined by measuring the accumulation of ZO-1 at cell–cell contacts using ImageJ software (National Institutes of Health). In brief, the maximal intensity of the ZO-1 signal (projections from confocal stacks) at cell–cell contacts within a given field of view (between 70 and 100 cells) was measured using the NeuronJ tracing tool of the ImageJ software. The resulting mean intensity value for the given field of view was multiplied with the total length of ZO-1–positive cell contact sites to account for ZO-1–negative cell contacts and divided by the number of cells, resulting in the mean ZO-1 intensity per cell. For each clone and each experiment, at least three fields of view were analyzed. Each experiment was performed at least three times. Mean values and SDs were calculated from three independent experiments.

Immunofluorescence analyses were performed with cells grown on either fibronectin-coated chamber slides (Lab-Tek: Thermo Fisher Scientific) or on fibronectin-coated filters (0.4-µm pore size; Transwell; Corning). Cells were fixed in either 4% paraformaldehyde for 10 min followed by incubation in PBS/0.5% Triton X-100 for 15 min at RT or in ice-cold EtOH for 30 min and acetone for 3 min at RT followed by rehydration in blocking buffer (PBS/10% FCS). After blocking for 1 h, cells were incubated with primary antibodies in 10 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.01% Tween 20, and 0.1% BSA for 1 h at RT or overnight at 4°C. After washing, cells were incubated with fluorochrome-conjugated, highly cross-adsorbed secondary antibodies for 1 h at RT. As fluorochromes, either Cy2 and Cy3 conjugates (Jackson ImmunoResearch Laboratories, Inc.) or Alexa Fluor 488 and Alexa Fluor 568 conjugates (Invitrogen) were used. After washing, cells were mounted in fluorescence mounting medium (Dako) and stored at 4°C. Nuclear staining was performed with DNA dye (DRAQ5; Biostatus, Ltd.). Immunofluorescence microscopy was performed using a confocal microscope (LSM 510 Meta; Carl Zeiss) equipped with Plan-Neofluar lenses (Plan-Neofluar: air, 20× magnification, numerical aperture 0.5; Plan-Neofluar differential interference contrast: oil, 40× magnification, numerical aperture 1.3; Carl Zeiss) or Plan-Apochromat lenses (Plan-Apochromat differential interference contrast: oil, 63× magnification, numerical aperture 1.4; Carl Zeiss).

The TER of MDCK II cells was analyzed by seeding cells on fibronectin-coated polycarbonate filters (0.4-µm pore size; CoStar) in a 24-well device. After reaching confluence, cells were grown in low Ca²⁺ medium for 18 h and then subjected to a CS. TER was monitored online over a period of 60–80 h using an automated multiwell device (cellZscope; nanoAnalytics). This device was placed within the CO₂ incubator, and online recordings of the TER were taken in 15-min intervals, resulting in a smooth curve when plotted versus time over a period of 50 h. Data points are expressed as mean values of three different filters measured simultaneously.

The paracellular diffusion of TRITC-dextran was analyzed as described previously (Rehder et al., 2006). In brief, cells grown to confluence were subjected to a CS to induce new cell contact formation. 4 h after CS, 4.3-kD TRITC-dextran was added to the upper compartment of the filters at 1 mg/ml and allowed to diffuse for 2 h. The fluorescence in the lower compartment was analyzed using a fluorescence reader (excitation at 485 nm and emission at 530 nm; Lambda Fluoro 320; MWG Biotech). Each experiment was performed at least three times. Mean values and SDs were calculated from three independent experiments.

MDCK cyst assays were performed essentially as previously described (Rehder et al., 2006). In brief, MDCK cells were seeded as single-cell suspension in 0.18% rat tail type I collagen (BD). After 5–6 d, the gels were washed, treated with 100 U/ml collagenase VII (Sigma-Aldrich) for 15 min at RT, and then fixed. Cells were permeabilized by treatment with 0.25% Triton X-100/PBS (for 1 h at RT), washed with 2% goat serum/PBS (for 1 h at RT), and then incubated with primary and fluorochrome-conjugated secondary antibodies in 2% goat serum/PBS for a minimum of 12 h at 4°C each. After extensive washing, the gels were mounted on glass coverslips using Kaiser’s glycerol gelatin (Merck). Cysts were analyzed using a confocal microscope as specified in the Immunofluorescence microscopy paragraph.

MDCK cells were fixed with 2% glutaraldehyde in PBS for 2 h at RT and then gently scraped from the culture vessels. Cells incubated in 30% glycerol for 2 h were mounted on gold-nickel carriers and immediately rapidly frozen in Freon 22 cooled with liquid nitrogen. The samples were fractured using a freeze fracture unit (BA 310; Balzers, AG) at −100°C. Replicas of the fractured cells were made by electron beam evaporation of platinum-carbon and carbon at angles of 38° and 90° and to a thickness of 2 and 20 nm, respectively. The replicas were incubated overnight in household bleach at RT to remove the cells from the replicas. The replicas were washed in distilled water, mounted on grids, and examined in a transmission electron microscope (410; Philips).

Fig. S1 shows initial characterization of JAM-A phosphorylation by aPKCζ in vitro. Fig. S2 shows the characterization of a pAb specific for S285-phosphorylated JAM-A. Fig. S3 shows that S285-phosphorylated JAM-A colocalizes with ZO-1, PAR-3, and aPKCζ at TJs. Fig. S4 shows that S285 phosphorylation of JAM-A does not influence the localization of PAR-3. Fig. S5 shows that JAM-A regulates claudin-2 expression independent of S285 phosphorylation.

We thank Friedemann Kiefer (Max Planck Institute for Molecular Biomedicine, Münster, Germany) for his advice in phosphopeptide mapping and phosphoamino acid analysis experiments. We thank Estelle Sontag for the PP2A cDNAs. We thank Melissa Cudmore (University of Edinburgh, Edinburgh, Scotland, UK) for helpful comments and for proofreading of the manuscript. We thank Frauke Brinkmann and Annette Janning for expert technical help. We also thank Karin Schlattmann for technical help in electron microscopy.

This work was supported by grants from the Deutsche Forschungsgemeinschaft (EB 160/2-3 [SPP1111] and EB 160/4-1 to K. Ebnet).