PAK4 [p21 protein (Cdc42/Rac)-activated kinase 4] and partitioning defective (Par) 6B are required for apical junction assembly in epithelial cells. PAK4 phosphorylates Par6B at Ser143 blocking its interaction with Cdc42 (cell division cycle 42). This could provide dynamic turnover of Par6B at junctions and/or promote Par6B interactions with alternative binding partners.
Epithelial cells form the interface between the host and its external environment and, in addition to providing a barrier function, they exhibit specialized characteristics unique to their distinctive niches. The establishment of apical–basal polarity, the formation of intimate cell–cell adhesion contacts and the reorganization of the actin cytoskeleton are three key features of epithelial morphogenesis [1,2]. Genetic screens in Caenorhabditis elegans and Drosophila have identified numerous proteins that participate in epithelial morphogenesis, but three highly conserved groups are critical: partitioning defective (Par) 3–Par6–aPKC (atypical protein kinase C) and CRB3 (crumbs3)–PALS1–PATJ (PalsALS1-associated tight junction protein) located apically and Scribble–Dlg (discs large)–Lgl (lethal giant larvae) located basolaterally [3–11]. In addition, the Par6–aPKC complex is regulated by Cdc42 (cell division cycle 42), a Rho family GTPase . There is a complex interplay between these proteins involving positive and negative feedback loops [13–20]. The subsequent assembly of adherens junctions, tight junctions and an organized actomyosin cytoskeleton at the apical region of the lateral membrane provides a fence function, preventing the diffusion of proteins and lipids between the apical and basolateral membranes, and a barrier function, protecting the host from the external environment, yet allowing controlled paracellular permeability [21,22].
Given the importance of apical junctions at both the tissue and the cellular level, their formation and maintenance are subject to tight control and are influenced by numerous signalling pathways. Rho GTPases play key roles in epithelial morphogenesis, in part through their ability to spatially organize and modify the actin cytoskeleton [23–25]. Rho, Rac and Cdc42, the three best characterized members of this family, are activated by guanine-nucleotide-exchange factors (GEFs), inactivated by GTPase-activating proteins (GAPs) and, in their active GTP-bound state, interact with target proteins to generate downstream responses [23,26,27]. Around 82 GEFs, 67 GAPs and 100 targets have been identified in mammalian cells. In work aimed at identifying the contribution of Rho GTPase pathways to the assembly of apical junctions, we previously identified three target proteins in the human bronchial epithelial cell line 16HBE: the serine/threonine kinase PRK2 (protein-kinase C-related kinase 2; a Rho target), the serine/threonine kinase PAK4 [p21 protein (Cdc42/Rac)-activated kinase 4; a Cdc42 target] and the serine/threonine kinase complex Par6B–aPKC (a Cdc42 target) [28,29]. The observation that two distinct Cdc42 targets, PAK4 and Par6B, are required for apical junction assembly raises the possibility that they provide co-ordinated contributions.
PAK4 belongs to the mammalian PAK family of kinases. Group I PAKs (PAK1, 2 and 3) are activated upon binding Cdc42 or Rac by the displacement of an auto-inhibitory switch domain (AID) [30,31]. Group II PAK (PAK4, 5 and 6) activation is less clear. It was thought that Cdc42 binding (through an N-terminal CRIB motif) is required only for localizing the kinases, but recent work has provided evidence that they too have an AID . Given that Cdc42 activity is spatially controlled in cells, it is likely that in 16HBE cells, Cdc42 binding promotes localized activation of PAK4.
Par6B is a scaffold-like protein generally found in a complex with one of the two aPKC isoforms (aPKCγ/ζ), but also additional proteins, notably Par3 and CRB3. The Par proteins (Par1–6) were originally identified in C. elegans as key players in the establishment of cell polarity during early development, but have since been shown to control cell polarity in a wide range of contexts, including epithelial morphogenesis [3,12,19,29,33]. The interaction of Par6 with aPKC and Par3 is mediated by an N-terminal PB1 domain and a neighbouring PDZ domain respectively [34,35]. Par6 also contains a CRIB motif, which interacts with the GTP-bound form of Cdc42 [12,36]. Cdc42 binding has been suggested to generate a conformational change in Par6 leading to activation of aPKC kinase activity [36,37]. However, recent work has called this into question: using purified proteins, Par6 was found to activate aPKC in the absence of Cdc42, raising the possibility that the role of Cdc42 is to localize the Par6–aPKC complex .
Both PAK4 and Par6B were identified as downstream targets of Cdc42 required for apical junction assembly in 16HBE cells . RNAi depletion of either inhibits the assembly of mature junctions in these cells indicating that the two proteins do not act redundantly. In experiments designed to determine their function, we have now identified a direct interaction between PAK4 and Par6B that supports a co-ordinated function for these two targets during apical junction formation.
Primary antibodies were used against the following: ZO-1 (zonula occludens protein 1; rabbit polyclonal, Invitrogen); Cdc42 (rabbit polyclonal, Cell Signaling Technology; mouse, BD Transduction); Par6B (rabbit polyclonal H-64, Santa Cruz Biotechnology); aPKC (rabbit polyclonal C-20, Santa Cruz Biotechnology); Par3 (rabbit polyclonal, Millipore); β-actin (clone AC-74, Sigma–Aldrich); PAK4 (rabbit polyclonal, Cell Signaling Technology); haemagglutinin (HA; mouse, Covance); GFP (rabbit polyclonal, Invitrogen; mouse clone 3E11, Cancer Research UK); myc (clone 9E10, Cancer Research UK); Flag (M2, Sigma–Aldrich). Secondary antibodies used: Alexa Flour 488- and 568-conjugated secondary antibodies (Invitrogen); horseradish peroxidase (HRP)-conjugated secondary antibodies (Dako).
DNA constructs and siRNA sequences
Human PAK4 and PAK4 (K350M) were a gift from Dr Audrey Minden (Rutgers University, Piscataway, NJ, U.S.A.).
The HA-tag was inserted C-terminally to Par6B, since N-terminally tagged versions were not able to rescue Par6B depletion in 16HBE cells. Mouse Par6B mutants, S143A, S143E and ∆P136, were generated by single-step PCR mutagenesis using the following primers:
S143A forward: acttccggcccgtggcatccatcatcgac; reverse: gtcgatgatggatgccacgggccggaagt
S143E forward: ggacttccggcccgtggagtccatcatcgacgtgg; reverse: ccacgtcgatgatggactccacgggccggaagtcc
S144A forward: cttccggcccgtgtcagccatcatcgac; reverse: gtcgatgatggctgacacgggccggaag
∆P136 forward: acagcgacgaggacgatatcattattgaagacagtggcg; reverse: cgccactgtcttcaataatgatatcgtcctcgtcgctgt.
Primers were from Sigma. Inserts were subcloned into: (i) pQCXIP vectors, using PCR amplification to introduce BamHI (5′) and EcoRI (3′) restriction sites using the following primers: forward: cgcgcggatccgccaccatgaaccgcggccaccgg; reverse: gcgcggaattctcaagcgtaatctggaacatcgtatgggtataatgtgatgatggtaccg; or (ii) pRK5myc vector using PCR amplification to introduce BamHI (5′) and EcoR1 (3′) restriction sites using the following primers: forward: cgcgcggatccgccaccatgaaccgcggccaccgg; reverse: gcgcggaattctcataatgtgatgatggtaccg. siRNA was from Dharmacon/GE Healthcare, siGENOME individual duplex. Par6B duplex3 (D-010681-03). Sequence: CGAAGAAGATGACATTATC
16HBE14o- cells (a gift from Dr Dieter Gruenert, California Pacific Medical Center, San Francisco, CA, U.S.A.) were cultured in MEM (minimal essential media) + GlutaMAX (Invitrogen) supplemented with 10% FBS (BenchMark, Gemini Bio-Products) and penicillin (100 units/ml)/streptomycin (100 μg/ml; Invitrogen) at 37°C in 5% CO2. For siRNA transfections in 16HBE cells, cells were seeded at 1.2×105 cells/well in six-well plates (10–20% confluent). siRNAs were transfected in medium without antibiotics, using 100 pmol of siRNA (final concentration, 50 nM) and 5 μl of Lipofectamine LTX (Invitrogen) per well. The transfection reagent was removed after 12 h of incubation and replaced with fresh medium containing antibiotics. After 48 h, cells were trypsinized and seeded in 24-well plates with or without coverslips. Five to six days after transfection, coverslips were fixed for immunofluorescence (IF) staining and cells were lysed for Western blot (WB) analysis. For retroviral infection, 1.2×105 cells were seeded per well in six-well plates and infected with 1.5 ml of viral suspension supplemented with 8 μg/ml polybrene (hexadimethrine bromide) by centrifugation at 1100 g for 30 min. Two days after infection, stable pools were selected using 1.5 μg/ml puromycin (Invitrogen). To avoid clonal effects, all experiments were performed with stable pools.
Human embryonic kidney (HEK) 293T cells (A.T.C.C.) were cultured in DME HG (Dulbecco's modified Eagle, high glucose)+sodium pyruvate (Sloan–Kettering Medium Preparation Facility) supplemented with 10% FBS and penicillin (100 units/ml)/streptomycin (100 μg/ml; Invitrogen) at 37°C in 5% CO2. For DNA transfection, cells were seeded at 4×105 cells/well in six-well plate, with 5 μg of Lipofectamine 2000 (Invitrogen) and 1 μg of total plasmid DNA used per well. For retroviral production, cells were grown to 80–90% confluency in 10 cm plates and were co-transfected with 5 μg of VSV-G, 5 μg of GagPol and 5 μg of retroviral plasmids of interest, using 25 μl of Lipofectamine 2000. Six hours after transfection, the medium was replaced with of fresh growth medium. Viruses were collected every day for three consecutive days.
Immunofluoresence and microscopy
16HBE cells were seeded on glass coverslips and fixed in 3.7% (v/v) formaldehyde (Sigma–Aldrich) for 15 min and permeabilized in 0.5% (v/v) Triton X-100 (Sigma–Aldrich) for 5 min at room temperature. For staining with anti-PAK4 antibody, cells were fixed in ice-cold methanol at −20°C for 5 min. Primary and secondary antibody incubations were carried out for 1 h at room temperature. Coverslips were mounted with fluorescent mounting medium (Dako) and visualized using a Zeiss AxioImager.A1 fluorescence microscope with 40× NA 0.75 objective and AxioVision software (Zeiss). For confocal images, 16HBE cells were seeded at confluence on glass coverslips and incubated for 10 days and the medium refreshed every 2 days. Confocal z-stacks were acquired using a Zeiss LSM 780 confocal microscope, equipped with a 40× objective.
Immunoprecipitation and blotting
16HBE cells were lysed in 1× SDS sample buffer (2% SDS, 100 mM DTT, 50 mM Tris/HCl, pH 6.8, 10% glycerol and 0.1% Bromophenol Blue) and boiled for 5 min at 100°C. For immunoprecipitation, transfected HEK293T cells were lysed by scraping in lysis buffer (0.5% NP-40, 50 mM Tris/HCl, pH 8.0 and 150 mM NaCl). If immunoprecipitation was used to detect phosphorylated proteins, then lysis buffer also contained 12.5 mM β-glycerophosphate, 0.5 mM NaF, 0.5 mM Na3VO4 and 10 nM calyculin-A) supplemented with 2 mM PMSF and a complete protease inhibitor tablet (Roche). Cell lysates were centrifuged for 15 min at 18000 g at 4°C to pellet cell debris. The supernatant was incubated with primary antibody for 2 h, followed by Protein G-Sepharose beads (Sigma–Aldrich) for 1 h, at 4°C with tumbling. Beads were washed with lysis buffer and boiled in sample buffer. Proteins were resolved by SDS/PAGE, transferred on to a PVDF membrane (Millipore), blocked with 5% milk or BSA and incubated with the appropriate primary antibodies overnight at 4°C. Proteins were visualized using HRP-conjugated secondary antibodies (Dako) and ECL detection reagents (GE Healthcare).
Recombinant protein purification and in vitro kinase assay
PAK4 wild-type (WT), PAK4 kinase-dead, Par6B WT, Par6B N-terminal fragment, Par6B C-terminal fragment, Par6B S143A and Par6B S144A cDNAs were fused with an N-terminal histidine-tag and expressed in Escherichia coli BL21-CodonPlus (DE3)-RIPL competent cells (Strategen), followed by one-step affinity purification with Ni-NTA (Ni2+-nitrilotriacetic acid)-agarose beads (Qiagen), according to the manufacturer's instructions. Histidine-tags were not cleaved after purification. Imidazole (contained in the elution buffer) was removed by dilution and concentration (Amicon Ultra centrifugal filters, Millipore). In vitro kinase assays were performed in reaction buffer (12.5 mM MOPS, pH 7.5, 7.5 mM MgCl2, 0.5 mM EGTA, 12.5 mM β-glycerophosphate, 0.5 mM NaF, 0.5 mM Na3VO4 and 10 nM calyculin-A). A 2 μg amount of Par6B were incubated with 50 ng of recombinant PAK4 kinase. Reactions were initiated by the addition of 50 μM ATP containing 3 μCi of [γ-32P]ATP. Mixtures were incubated for 30 min or 1 h and stopped by adding 1× SDS sample buffer, followed by boiling at 100°C for 5 min. Phosphorylation of Par6B proteins was detected by autoradiography. In vitro kinase assay results were quantified using ImageJ (http://rsb.info.nih.gov/ij/). In brief, the intensity of each phosphorylation band was first normalized to its protein level on WB and then the adjusted phosphorylation intensities were normalized to WT sample .
In vitro kinase assays were performed as described above, except that only non-radiolabelled ATP was added to the reaction. Proteins were resolved by SDS/PAGE and visualized with SimplyBlue SafeStain (Invitrogen) according to the manufacturer's instructions.
In-gel protein digestion and MS were carried out at the Sloan–Kettering Microchemistry and Mass Spectrometry Core Facility. SDS/PAGE-resolved WT and mutant PAR6B were subjected to in-gel trypsin followed by endoproteinase AspN digestion. Resulting peptides were purified and analysed by nano-LC–MS/MS as described in . Peptides and proteins were subsequently identified with Mascot search engine (Matrix Science, version 2.3.02; www.matrixscience.com), with the mouse segment of the NCBInr database (315729 entries) and statistical analyses were performed with Scaffold 4 software (Proteome Software Inc). Search parameters were as follows: two missed cleavage sites were allowed, precursor mass tolerance was set at 10 ppm, fragment ion mass tolerance was set at 0.8 Da. Variable peptide modifications were allowed for serine, threonine and tyrosine phosphorylation, methionine oxidation, cysteine acrylamide derivatization and deamidation of asparagines and glutamines. MudPit scoring was applied using significance threshold score P<0.01. Any resulting phosphorylated peptide identification was visually inspected for ‘b’ and ‘y’ ions to confirm the location of the phosphate residue.
Preparation of phospho-specific antibodies
Rabbit polyclonal antisera recognizing pSer143 in Par6B were raised using the following phosphopeptide as antigen: C-P-Q-D-F-R-P-V-pS-S-I-I-D (where pS is phosphoserine). The corresponding non-phosphorylated peptides were also synthesized. The phosphopeptide was coupled to KLH (keyhole limpet haemocyanin) and two rabbits were immunized (Harlan Laboratories). The final bleeds were tested by ELISA and purified by sodium sulfate precipitation and affinity purification using an Affigel 15 matrix (Bio-Rad Laboratories), according to the manufacturer's instructions. Purified antibodies were tested by Western blotting, in the presence or absence of non-phospho- or phospho-peptides.
Apical junction quantification
Twelve random non-overlapping images were taken per sample at 40× magnification (about 400 cells) and the percentage of cells with normal tight junction was quantified by counting manually, using the Metamorph image analysis software (Universal Imaging). Cells with a continuous ring of ZO-1 at cell–cell contacts were defined as having intact tight junctions. Cells with punctate or discontinuous ZO-1 at cell–cell contacts were defined as having a tight junction defect. Three independent experiments were quantified and statistical analysis was carried out using Prism (GraphPad Software). Error bars are S.E.M. P-values have been calculated using a two-tailed unpaired Student's t test. ****P<0.0001.
Par6 protein sequences used in alignments were retrieved from the NCBI Entrez-Protein Database. Homo sapiens (Hs): Par6A isoform 1, NP_058644; Par6A isoform 2, NP_001032358; Par6B, NP_115910; Par6G, NP_115899. Mus musculus (Mm): Par6B, NP_067384. Drosophila melanogaster (Dm): Par-6 isoform a, NP_573238; Par-6 isoform b, NP_728094. Caenorhabditis elegans (Ce): Par-6 isoform a, NP_001040687; Par-6 isoform b, NP_001040688.
PAK4 interacts directly with Par6B
PAK4 and Par6B are required for apical junction maturation in 16HBE cells and their localization to junctions is each mediated by binding to Cdc42 [29,33]. To explore whether these two targets interact with each other, epitope-tagged proteins were expressed in HEK293T cells and the two proteins were immunoprecipitated (IP) and analysed on WBs. Under these conditions, PAK4 can be readily co-precipitated with Par6B (Figure 1A) and Par6B can be co-precipitated with PAK4 (Figure 1B). As a control, Par6B was also co-expressed with the kinase PRK2, a Rho target that is also recruited to junctions in 16HBE cells, but these two proteins could not be co-precipitated (Figure 1C; Supplementary Figure S1). To map the regions of Par6B required for interaction, N-terminal (containing the PB1 domain) and C-terminal (containing the PDZ domain and CRIB motif) fragments (Figure 1D) were co-expressed with PAK4. Only the C-terminal fragment could be co-precipitated (Figure 1E). To map the regions of PAK4 required for interaction, N-terminal (containing the CRIB motif) and C-terminal (containing the kinase domain) fragments (Figure 1D) were co-expressed with Par6B. Surprisingly, the kinase domain, but not the N-terminal regulatory region, could be co-precipitated (Figure 1F). We conclude that the kinase domain of PAK4 interacts with the PDZ–CRIB region of Par6B.
The kinase domain of PAK4 interacts with Par6B
Par6B is a PAK4 substrate
Although it is not typical for kinases to interact stably with their substrates, many substrates harbour a docking site for kinases . We wondered therefore whether Par6B could be a substrate for PAK4. Full-length WT and kinase-dead PAK4 (K350M), full-length Par6B and N- and C-terminal fragments of Par6B were each purified as His (histidine)-tag fusion proteins from E. coli . PAK4 was incubated with full-length Par6B or the Par6B fragments in the presence of radiolabelled ATP at 30°C for 30 min. The reaction products were resolved on SDS/PAGE gels and ATP incorporation was detected by autoradiography. The in vitro kinase assay results reveal that PAK4, but not kinase-dead PAK4, is able to phosphorylate Par6B, predominantly in the C-terminal region (residues 126–371; Figure 2A).
PAK4 phosphorylates Par6B at residue Ser143
To identify the phosphorylation site(s) in Par6B, we conducted in vitro kinase reactions using WT or kinase-dead PAK4 and full-length Par6B. Samples were resolved on an SDS/PAGE gel and analysed by MS to identify phosphorylation site(s) in Par6B. Ser143 or Ser144 was found to be specifically phosphorylated by WT, but not kinase-dead PAK4 (Supplementary Figure S2). To analyse this in more detail, the serine residues at 143 and 144 were replaced by alanine individually or together, and in vitro kinase assays revealed that Ser143 is the major PAK4 phosphorylation site (Figure 2B). The phosphorylation signal on Par6B is not completely absent from the Par6B S143A mutant indicating additional phosphorylation site(s) that were not detected by MS.
To begin to explore the potential cellular role of this phosphorylation event, a rabbit polyclonal antiserum was raised using a phosphorylated (pSer143) peptide. The specificity of the purified antibody was first examined in an in vitro kinase assay using recombinant Par6B. The antibody was capable of recognizing WT, but not S143A, Par6B after incubation with PAK4 and the signal with WT Par6B can be competed using the phospho- but not non-phospho-peptide (Figure 2C). To confirm that PAK4 can phosphorylate Par6B in cells, WT or kinase-dead versions of PAK4 were co-expressed in HEK293T cells with WT Par6B and cell lysates were analysed on WBs. Par6B was phosphorylated by WT, but not kinase-dead, PAK4 in HEK293T cells (Figure 2D). Despite numerous attempts, the antibody was not sufficiently sensitive to detect endogenous phospho-Par6B on WBs of 16HBE cell lysates (results not shown).
Par6B phosphorylation blocks Cdc42 binding
Residue Ser143 is evolutionarily conserved in metazoans: C. elegans, D. melanogaster, mouse and human (Figure 3A). In addition, this residue is adjacent to the semi-CRIB motif, which is critical for Cdc42 binding, suggesting that this could be an important regulatory site. To examine the effect of Ser143 phosphorylation on Cdc42 binding, we co-expressed WT, non-phosphorylatable (S143A) or phosphomimetic (S143E) versions of Par6B together with dominant negative Cdc42 (Cdc42N17) or constitutively activated Cdc42 (Cdc42V12) in HEK293T cells and performed co-immunoprecipitations. Neither the WT Par6B nor the mutants were co-precipitated with dominant negative Cdc42. The non-phosphorylatable mutant (Par6B S143A) and the WT Par6B efficiently co-precipitated constitutively activated Cdc42 as expected (Figure 3B). However, the phosphomimetic mutant (Par6B S143E) could not co-precipitate activated Cdc42 (Figure 3B). The lack of binding to the phosphomimetic mutant is consistent with the crystal structure of Par6B–Cdc42 complex, in which the hydroxy group in the side chain of Ser143 forms a hydrogen bond with the side chain of Asn39 in Cdc42 .
Par6B phosphorylation blocks its interaction with Cdc42
To determine whether this interaction can also be detected in cells expressing endogenous levels of Par6B, we generated 16HBE cell lines with stably expressed HA-tagged WT or Par6B mutants. As can be seen in Figure 3(C), HA–Par6B is expressed at levels comparable to that of the endogenous Par6B (compare first and second Par6B input lanes). Endogenous Cdc42 was co-precipitated with HA–Par6B and HA–Par6B (S143A), but not with HA–Par6B (S143E); Figure 3C; note that a non-specific background band migrates just above Cdc42 on gels). We conclude that phosphorylation of Ser143 specifically blocks Cdc42 binding to Par6B.
Par6B phosphorylation and apical junction assembly
To explore the effect of Ser143 phosphorylation on apical junction assembly, 16HBE cells were infected with retroviral vectors containing no insert [control, empty vector (EV)] or RNAi-resistant versions of HA-tagged WT, S143A or S143E Par6B mutants. A Par6B Pro136 deletion mutant (∆P136), previously shown to be incapable of binding Cdc42, was also used . The HA-tagged Par6B proteins were expressed at comparable levels to endogenous Par6B, as shown by Western blotting of whole-cell lysates (Figure 4C) and they did not induce any dominant negative effects on junction formation in monolayers (Figure 4A). Immunolocalization after fixing and staining with anti-HA antibody revealed WT and S143A Par6B localized to apical junctions, whereas the S143E and ∆P136 Par6B mutants could not be detected at apical junctions (Figure 4A). Finally, to determine the functional activity of the mutants, endogenous Par6B was depleted using siRNA (duplex3). Transfection of control 16HBE cells with siRNA disrupted apical junctions as expected (Figures 4B and 4D) and this could be fully rescued by WT or S143A Par6B. However, S413E or ∆P136 Par6B could not rescue (Figures 4B and 4D).
Phosphorylated Par6B is unable to support apical junction assembly in 16HBE cells
Phosphorylation and subcellular localization of Par6B
Our analysis showing that Ser143 phosphorylation by PAK4 prevents Par6B from interacting with Cdc42 raises several possible scenarios. Confocal analysis of mature 16HBE monolayers appears to show co-localization of PAK4 and Par6B at tight junctions (Figure 5A). Since Par6B is believed to be a structural component of junctions, whereas PAK4 plays a regulatory role, this could suggest that PAK4 is involved in the turnover of Par6B at the membrane, although we cannot exclude the possibility that the proteins are segregated at the nano-scale. It is also clear that Par6B, but not PAK4, localizes to the apical membrane (Figure 5A z-stack images, right panels). This suggests that recruitment of Par6B to apical junctions may subsequently lead to Par6B transitioning into the apical membrane, through its interaction with other proteins.
Effect of Par6B phosphorylation on subcellular localization and protein–protein interactions
To examine the effect of Ser143 phosphorylation on its interaction with other proteins, we first examined Par3, since Bazooka, the Drosophila orthologue of Par3, has been reported to play a role in targeting Par6 . WT, S143A and S143E versions of Par6B were co-expressed with Par3 in HEK293T cells and co-immunoprecipitations were performed. Par3 co-precipitated with Par6B and this was not affected by either the S143A or the S143E substitutions (Figure 3C). Since the interaction of Par3 with Par6 is known to be independent of Cdc42, this was not unexpected . Par6B also interacts with aPKC, but co-immmunoprecipitation of Par6B and aPKC co-expressed in HEK293T cells was also unaffected by the S143E substitution (Figure 3C).
Two additional proteins have been proposed to facilitate apical membrane localization of Par6 in epithelial cells, Morg1 (mitogen-activated protein kinase organizer 1), a WD40 scaffold protein and CRB3 . Interestingly, Morg1 and Cdc42 have been shown to bind to Par6 in a mutually exclusive manner. To examine whether Ser143 phosphorylation affects the Morg1 interaction, WT and Par6B mutants were co-expressed in HEK293T cells together with Morg1 and constitutively activated Cdc42. Morg1 binding to WT and S143A Par6B was relatively weak in the presence of Cdc42, whereas it interacted strongly with phospho-mimetic S143E Par6B (Figure 5B). This raises the possibility that Ser143 phosphorylation favours an interaction of Par6B with Morg1 rather than Cdc42. CRB3, on the other hand, has been reported to bind to Par6–aPKC at the apical membrane and this is promoted by Cdc42. To examine the effect of Ser143 phosphorylation on CRB3 binding, WT and Par6B mutant were co-transfected in HEK293T cells with constitutively activated Cdc42 and CRB3. In the absence of Cdc42 (Figure 5C, lanes 1–4), very little CRB3 was co-precipitated with Par6B. However, Cdc42 significantly promoted the binding of CRB3 to both WT and S143A Par6B, but not to S143E Par6B (Figure 5C, lanes 5–8). The enhanced interactions seen between CRB3 and WT or S143A Par6B were not due to a direct interaction between Cdc42 and CRB3, since when co-expressing these two proteins in HEK293 cells, neither of them was able to co-precipitate the other (Supplementary Figure S3).
In previous work using human airway epithelial cells, we identified Par6B and PAK4 as two target proteins acting downstream of Cdc42 required for the maturation of primordial cell–cell adhesions into apical junctions and the establishment of apical–basal polarity . We now show that Par6B is a substrate for PAK4. PAK4 can phosphorylate Par6B at residue Ser143, which is close to the Cdc42-binding site (CRIB motif) and is highly conserved throughout metazoans. Phosphorylation of Par6B at Ser143 prevents Cdc42 binding, suggesting that this could play an important regulatory role.
The co-localization of PAK4 and Par6B is highly spatially restricted in airway epithelial cells and is seen only at tight junctions. Furthermore, Par6B is thought to be an important structural component of apical junctions, whereas PAK4 probably plays a regulatory role. This may account for our inability to detect Ser143-phosphorylated Par6B in 16HBE monolayers using a phospho-specific antibody, although interestingly, phosphorylation of this residue (in Par6A) has been identified in a proteomic study using renal cells . We find that a phosphomimetic Par6B mutant (S143E) is unable to localize at apical junctions when expressed in 16HBE cells and cannot rescue junction formation after RNAi depletion of endogenous Par6B. This is in agreement with an essential requirement for Cdc42 in localizing Par6B during the establishment of apical–basal polarity . The fact that the Par6B and PAK4 appear to co-localize raises the possibility therefore that PAK4 regulates dynamic exchange of Par6B at apical junctions. Interestingly, PAK4 overexpression has been reported in many human cancers, in particular the gene is amplified in 6% of human squamous cell lung carcinomas and, although it is likely to have multiple substrates under these conditions, Par6 phosphorylation could potentially contribute to loss of epithelial architecture and tumour progression [47–52].
Par6 isoforms interact with numerous other proteins in addition to Cdc42. However, we found no effect of Ser143 phosphorylation on its interaction with aPKC or another polarity protein, Par3. Unlike aPKC or Par3, the interaction of Par6 with two other proteins, CRB3 and Morg1, has been reported to be differentially influenced by Cdc42 . CRB3 is a key polarity protein in epithelial cells that localizes to the apical membrane. CRB3 binding to Par6 has been reported to be Cdc42-dependent and, in agreement with this, we find that Ser143 phosphorylation of Par6B prevents binding to CRB3 [20,45]. Morg1 is a poorly understood scaffold protein reported to interact with diverse signalling proteins, including the hypoxia-inducible factor prolyl hydroxylase 3 (PHD3) and several components of the extracellular-signal-regulated kinase (ERK) mitogen-activated protein kinase (MAPK) pathway [53,54]. In addition, it was recently shown to participate in apical–basal polarity establishment in epithelial cells . Morg1 binding to Par6 has been reported to be inhibited by Cdc42 binding and, in agreement with this, we find that, in the presence of active Cdc42, Par6B phosphorylated at Ser143 interacts strongly with Morg1. These results raise an additional possibility, namely that Par6B phosphorylation locally controls its association and disassociation with specific binding partners. Thus, although Cdc42 is essential to recruit Par6B to apical junctions, where it could promote an interaction with CRB3, subsequent phosphorylation at Ser143 could lead to dissociation of CRB3 and Cdc42 and subsequent association with Morg1. We are currently exploring these different possibilities.
auto-inhibitory switch domain
atypical protein kinase C
cell division cycle 42
human bronchial epithelial
human embryonic kidney
mitogen-activated protein kinase organizer 1
p21 protein (Cdc42/Rac)-activated kinase 4
protein kinase C-related kinase 2
zonula occludens protein 1
Dan Jin and Joanne Durgan performed the experiments. Alan Hall and Dan Jin wrote the paper. All authors read and approved the final paper.
We thank Dr Hediye Erdjument-Bromage (Sloan–Kettering Microchemistry and Mass Spectrometry Core Facility) for MS sample analysis and Dr Frances Weis-Garcia (Sloan–Kettering Monoclonal Antibody Core Facility) for helpful suggestions on phospho-specific antibody generation. We also thank Dr Sean W. Wallace (Rockefeller University, New York), Dr Franck Pichaud (University College London, London) and all Hall laboratory members for valuable discussions.
This work was supported by the National Institutes of Health [grant numbers GM081435 and CA008748 (to A.H.)].