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

Abstract

The PAR-3-atypical protein kinase C (aPKC)-PAR-6 complex has been implicated in the development of apicobasal polarity and the formation of tight junctions (TJs) in vertebrate epithelial cells. It is recruited by junctional adhesion molecule A (JAM-A) to primordial junctions where aPKC is activated by Rho family small guanosine triphosphatases. In this paper, we show that aPKC can interact directly with JAM-A in a PAR-3-independent manner. Upon recruitment to primordial junctions, aPKC phosphorylates JAM-A at S285 to promote the maturation of immature cell-cell contacts. In fully polarized cells, S285-phosphorylated JAM-A is localized exclusively at the TJs, and S285 phosphorylation of JAM-A is required for the development of a functional epithelial barrier. Protein phosphatase 2A dephosphorylates JAM-A at S285, suggesting that it antagonizes the activity of aPKC. Expression of nonphosphorylatable JAM-A/S285A interferes with single lumen specification during cyst development in three-dimensional culture. Our data suggest that aPKC phosphorylates JAM-A at S285 to regulate cell-cell contact maturation, TJ formation, and single lumen specification.

Figure 1.

JAM-A is phosphorylated by aPKCζ at S285. (A, top) Amino acid sequence of the entire cytoplasmic domain of murine JAM-A (aa 261–300). Green letters indicate S and T residues, which were mutated to A based on evolutionary conservation and match to PKC consensus sites (Kennelly and Krebs, 1991). S262 represents a PKC phosphorylation consensus site only in combination with the adjacent GST residues. (bottom) JAM-A peptides predicted to be generated by chymotrypsin cleavage. Red vertical bars indicate predicted cleavage sites. At Y265 and F266, two chymotrypsin cleavage sites are juxtaposed. As the enzymatic hydrolysis requires two additional amino acids C terminally to the cleavage site, peptides 1 and 2 may each exist in two variants, 1a/1b and 2a/2b, respectively, depending on which residue is targeted first. (B) JAM-A is phosphorylated by aPKCζ at S285. (Bi) GST–JAM-A fusion proteins with the entire cytoplasmic domain of JAM-A (aa 261–300) either wild type (WT) or mutated (S262A and S285A) were phosphorylated with aPKCζ. (top) Phosphorylation was analyzed by SDS-PAGE and ³²P autoradiography. (bottom) Equal amounts of proteins were controlled by silver staining of 5% of the eluted proteins (input). (Bii) Phosphopeptide mapping of phosphorylated JAM-A by two-dimensional electrophoresis and thin-layer chromatography. Chymotryptic digest of aPKCζ-phosphorylated GST–JAM-A/WT results in two major phosphopeptide spots (labeled 1b and 3; top left). One of the spots (3) is lost in the S285A mutant (top right, red circle). The second major spot (1b) as well as one minor spot (1a) are lost in the S262A mutant (indicated by red circles), which is consistent with the prediction of chymotrypsin generating two peptides harboring S262 (bottom left). Because S262 was known to represent an artificial PKC consensus generated at the fusion point between GST and JAM-A (GRHYS₂₆₂RGY, with GST sequence underlined and the basic residue at position −3 of the phosphoacceptor site in bold), a newly generated GST fusion protein with three additional JAM-A–specific amino acids between GST and the JAM-A sequence (RGSWFAYS₂₆₂RGY, with the GST sequence underlined) was analyzed by in vitro phosphorylation and conventional SDS-PAGE (bottom right). Note that the strong phosphorylation of this construct by aPKC is almost completely abolished after mutating S285 to Ala, suggesting that S285 is the only major phosphorylation site for aPKC in vitro. Black lines indicate that intervening lanes have been spliced out. (C) JAM-A is serine phosphorylated in cells. (Ci, left) Autoradiography of JAM-A immunoprecipitated from [³²P]orthophosphate-labeled KLN205 cells. (right) Immunoblot analysis of lysates and of immunoprecipitated JAM-A was performed as controls. (Cii) Phosphoamino acid analysis of JAM-A. Immunoprecipitated JAM-A shown in Ci (³²P autoradiography [Autoradio.]) was isolated from the membrane, subjected to acid hydrolysis, and analyzed by two-dimensional thin-layer electrophoresis. Relative positions of pSer, pThr, and pTyr spots are indicated. JAM-A is exclusively phosphorylated at Ser residues. pSer, phosphoserine; pThr, phosphothreonine; pTyr, phosphotyrosine; Pi, orthophosphate. Molecular markers are given in kilodaltons.

Figure 2.

JAM-A directly interacts with aPKCζ. (A) JAM-A interacts with aPKCζ in HEK293T cells. aPKCζ immunoprecipitates (IP) obtained from HEK293T cells transiently transfected with aPKCζ and Flag–JAM-A were immunoblotted (IB) with Flag tag antibodies. In lanes labeled - (Lys), 0.75% of the input was loaded. (B) JAM-A associates with aPKCζ in Caco-2 epithelial cells. Immunoprecipitates obtained with antibodies against aPKCζ were analyzed for the presence of JAM-A (top) or aPKCζ (bottom). Lanes labeled - (Lys) contain 1.1% of the input. (C) The interaction of aPKCζ with JAM-A is direct. (top) GST–JAM-A was incubated with recombinant, [³⁵S]methionine-labeled aPKCζ. GST–PAR-6C served as a positive control (ctrl). (bottom) Equal loading of GST fusion proteins was analyzed by Coomassie brilliant blue staining. The lysate lane contains 10% of the input. Note that aPKCζ interacts with JAM-A as strongly as with PAR-6. (D) The association of JAM-A with aPKCζ does not require a PDZ domain protein. HEK293T cells were transfected with Flag–JAM-A constructs, either full-length JAM-A (JAM-A/f.l.), C-terminal deletion mutants lacking three (JAM-A/Δ3) or nine (JAM-A/Δ9) amino acids, which include the PDZ domain binding motif, or the entire cytoplasmic domain (JAM-A/Δ43). (top) Immunoprecipitates obtained with aPKCζ antibodies (first through fourth and sixth lanes) or with isotype-matched control antibodies (fifth lane) were analyzed for the presence of Flag-tagged constructs. (middle) The specificity of the aPKCζ immunoprecipitation was analyzed by immunoblotting 10% of the precipitated material with antibodies against aPKCζ. (bottom) Equal expression of the transfected constructs was verified by immunoblotting cell lysates with Flag tag antibodies. Asterisks indicate Ig heavy chains. Note that the absence of the PDZ domain binding motif in JAM-A does not abolish aPKCζ binding. - (Lys), lysate (no immunoprecipitation, with postnuclear supernatant loaded); Alk.Phos., alkaline phosphatase. Molecular markers are given in kilodaltons.

Figure 3.

JAM-A is phosphorylated at S285 by aPKCζ during cell–cell contact formation. (A) JAM-A is S285 phosphorylated in polarized MTD-1A cells. MTD-1A cells grown to confluence were either kept in normal medium (normal Ca²⁺), grown under low Ca²⁺ conditions for 18 h (low Ca²⁺), or grown under low Ca²⁺ followed by Ca²⁺ replenishment for 3 or 8 h to induce new contact formation (CS). Cells were stained with antibodies against total JAM-A and S285-P JAM-A. JAM-A is S285 phosphorylated at mature cell–cell contacts, is dephosphorylated upon cell–cell contact disruption, and is rephosphorylated during cell–cell contact formation. (B) JAM-A phosphorylation at S285 correlates with the appearance of aPKCζ at early sites of cell–cell contact. Wound-scratched MTD-1A cells were stained with antibodies against total JAM-A and S285-P JAM-A (top) or with antibodies against total JAM-A and aPKCζ (bottom). S285 phosphorylation of JAM-A correlates with the appearance of aPKCζ and is only observed after pAJs have matured into more linear cell–cell contacts. Arrows indicate the direction of the woundings. Insets show confluent areas of the monolayers. Boxes in the merge pictures mark the area shown in the zoomed in pictures. (C) Inhibition of aPKCζ activity blocks JAM-A S285 phosphorylation during cell–cell contact formation. Confluent MTD-1A cells were scratch wounded, grown in the presence of 25 µM PSζ, and then stained with antibodies against total JAM-A and S285-P JAM-A. JAM-A phosphorylation at S285 is abolished in the presence of PSζ. Bars: (A) 5 µm; (B and C) 10 µm.

Figure 4.

JAM-A S285 phosphorylation promotes junction maturation. MDCK II Tet-Off cells stably transfected with either empty vector (mock) or wild-type JAM-A (JAM-A/WT) or S285A JAM-A (JAM-A/S285A; two independent clones, #1 and #2) were subjected to Ca²⁺ switch (CS) for 2 or 4 h and then fixed and stained with antibodies against ZO-1, β-catenin, or occludin. (A) Representative ZO-1 immunofluorescence pictures taken 2 h after CS. Insets show expression of the Flag-tagged JAM-A constructs. (B) Statistical analysis of ZO-1 (left), β-catenin (middle), and occludin (right) immunofluorescence at cell–cell contacts at different time points after CS as indicated. The intensity of the immunofluorescence signals at cell–cell contacts was analyzed using ImageJ software. Data are given as ratios of the mean fluorescence intensities per cell after and before induction of expression of the respective constructs. Error bars denote the means ± SEM from three separate experiments. Statistical significance was evaluated using unpaired t tests; *, P < 0.05; **, P < 0.01; ***, P < 0.001. not ind., not induced; rel., relative; β-cat, β-catenin; Occl, occludin. Bars, 10 µm.

Figure 5.

JAM-A S285 phosphorylation is restricted to TJs in polarized epithelial cells. (A) Polarized MTD-1A epithelial cells were stained with antibodies against S285-P JAM-A and total JAM-A (left) or against S285-P JAM-A and occludin (right). S285 phosphorylation is detectable exclusively at the TJs. Red lines indicate the position of the xz sections. (B) Polarized MTD-1A cells were incubated for 2 h with 25 µM PSζ inhibitor and then stained as indicated. (C) MTD-1A cells were transfected with either control siRNAs or aPKCζ-specific siRNAs and then stained as indicated. PSζ incubation or aPKCζ down-regulation abolished the S285 phosphorylation of JAM-A. Bars: (A) 10 µm; (B and C) 5 µm.

Figure 6.

PP2A interacts with JAM-A, and dephosphorylates JAM-A at Ser285. (A) PP2A interacts with JAM-A in HEK293T cells. PP2A immunoprecipitates (IP) obtained from HEK293T cells transfected with HA-tagged PP2A-Cα and Flag-tagged JAM-A were immunoblotted (IB) with antibodies against the Flag tag. The arrowheads in PP2A immunoblots indicate endogenous PP2A-C and ectopically expressed HA-tagged PP2A-Cα. Lanes labeled - (Lys) contain 0.75% of the input. (B) PP2A exists in a complex with JAM-A in Caco-2 epithelial cells. Immunoprecipitates obtained with antibodies against PP2A were analyzed for the presence of JAM-A (top) or for the presence of PP2A (bottom). Isotype-matched control antibodies (IgG ctrl) were used as controls. Lanes labeled - (Lys) contain 1.1% of the input. (C) PP2A dephosphorylates JAM-A at Ser285. GST–JAM-A/WT and GST–JAM-A/S285A phosphorylated with aPKCζ in the presence of γ-[³²P]ATP were incubated with PP2A and then analyzed by SDS-PAGE and autoradiography (Autoradiog.). PP2A completely reversed aPKCζ-mediated phosphorylation. (D) The association of JAM-A with PP2A is not mediated by the PDZ domain binding motif of JAM-A. HEK293T cells were transfected with various Flag–JAM-A constructs as indicated. (top) Immunoprecipitates obtained with PP2A antibodies (first through fourth and sixth lanes) or isotype-matched control antibodies (fifth lane) were analyzed for the presence of Flag constructs. (middle) The specificity of the PP2A immunoprecipitation was verified by immunoblotting with PP2A antibodies. (bottom) Equal expression of the transfected constructs was verified by immunoblotting cell lysates with Flag antibodies. Asterisks denote Ig heavy chains. White lines indicate that intervening lanes have been spliced out. - (Lys), lysate; Alk.Phos., alkaline phosphatase; JAM-A/f.l., full-length JAM-A. Molecular markers are given in kilodaltons.

Figure 7.

JAM-A S285 phosphorylation is required for the formation of functional TJs. (A) MDCK II cells expressing WT JAM-A (left), S285A JAM-A (middle), or JAM-A shRNAs (right) under a tetracycline-regulated promoter were left uninduced (red graphs) or were induced (green graphs). At confluence, cells were subjected to CS, and the development of TER was monitored over a period of 50 h. The gray area indicates deviation of triplicates. Experiments were performed at least three times for each clone. (insets) Immunofluorescence analysis of ectopic JAM-A constructs (left and middle) or endogenous JAM-A (right). (B) MDCK cells used in A (two independent clones for JAM-A/S285A, #1 and #2) were analyzed for 4.3-kD TRITC-dextran permeability. Data are given as a ratio of the fluorescence intensities present in the lower compartments with and without expression of the respective constructs (left) or as relative fluorescence intensities after shRNA expression (right). Error bars denote the means ± SD from three separate experiments. Statistical significance was evaluated using unpaired t tests; *, P < 0.05; **, P < 0.01, ***, P < 0.001. (C) MDCK II Tet-Off cells expressing JAM-A/S285A were analyzed by freeze fracture electron microscopy. The TJs are composed of approximately two to three strands with a few interconnections running parallel to the cell surface and are surrounded by a looser network comprising often anastomosing strands of variable orientation. No significant differences were observed between control cells (top) and JAM-A/S285A–expressing cells (bottom). (insets) Immunofluorescence analysis of JAM-A/S285A expression. not ind., not induced; EF, exoplasmic face; PF, protoplasmic face. Bars: (C, main images) 0.5 µm; (A and C, insets) 20 µm.

Figure 8.

JAM-A S285 phosphorylation is not required for the development of the apical membrane domain but for single lumen specification. MDCK II Tet-Off cells inducibly expressing wild-type Flag–JAM-A (JAM-A/WT) or Flag–S285A JAM-A (JAM-A/S285A) were grown in a three-dimensional collagen matrix for 5–7 d and then fixed and stained with antibodies against JAM-A (anti-Flag) and with rhodamine-phalloidin to detect apical F-actin. Cysts were classified as normal (spherical with single lumen, white columns), multiluminal (spherical with more than one lumen, black columns), and disorganized (loss of spherical morphology without lumen, gray columns). Expression of JAM-A/S285A significantly increased the number of multiluminal cysts in which apicobasal polarity was retained. Error bars denote the means ± SD from three separate experiments. Statistical analysis was performed with unpaired t tests. Bars, 10 µm.

Figure 9.

JAM-A is hyperphosphorylated at S285 during mitosis. Confluent monolayers of MTD-1A cells were stained with antibodies against S285-P JAM-A (top and middle rows) or total JAM-A (bottom row) and antibodies against α-tubulin, β-tubulin, or γ-tubulin as indicated. Note that JAM-A phosphorylation at S285 is strongly increased in dividing cells when compared with surrounding nondividing cells. Bars, 5 µm.