PLEKHA7 Recruits PDZD11 to Adherens Junctions to Stabilize Nectins*

PLEKHA7 is a junctional protein implicated in stabilization of the cadherin protein complex, hypertension, cardiac contractility, glaucoma, microRNA processing, and susceptibility to bacterial toxins. To gain insight into the molecular basis for the functions of PLEKHA7, we looked for new PLEKHA7 interactors. Here, we report the identification of PDZ domain-containing protein 11 (PDZD11) as a new interactor of PLEKHA7 by yeast two-hybrid screening and by mass spectrometry analysis of PLEKHA7 immunoprecipitates. We show that PDZD11 (17 kDa) is expressed in epithelial and endothelial cells, where it forms a complex with PLEKHA7, as determined by co-immunoprecipitation analysis. The N-terminal Trp-Trp (WW) domain of PLEKHA7 interacts directly with the N-terminal 44 amino acids of PDZD11, as shown by GST-pulldown assays. Immunofluorescence analysis shows that PDZD11 is localized at adherens junctions in a PLEKHA7-dependent manner, because its junctional localization is abolished by knock-out of PLEKHA7, and is rescued by re-expression of exogenous PLEKHA7. The junctional recruitment of nectin-1 and nectin-3 and their protein levels are decreased via proteasome-mediated degradation in epithelial cells where either PDZD11 or PLEKHA7 have been knocked-out. PDZD11 forms a complex with nectin-1 and nectin-3, and its PDZ domain interacts directly with the PDZ-binding motif of nectin-1. PDZD11 is required for the efficient assembly of apical junctions of epithelial cells at early time points in the calcium-switch model. These results show that the PLEKHA7-PDZD11 complex stabilizes nectins to promote efficient early junction assembly and uncover a new molecular mechanism through which PLEKHA7 recruits PDZ-binding membrane proteins to epithelial adherens junctions.

The molecular organization of cell-cell contacts is of central importance in understanding the morphogenesis, architecture, and physiology of vertebrate organs, as well as the mechanisms underlying the pathogenesis of human disease. In epithelia, cell-cell junctions are essential to establish and maintain tissue integrity and barriers between tissue compartments, and they also play key roles in the transmission of signals to and from adjacent cells, and within tissues, to regulate proliferation and differentiation (1)(2)(3)(4). The apical junctional complex includes two structurally and functionally distinct units, tight junctions (TJ) 3 and adherens junctions (AJ), whose canonical functions are paracellular barrier and intramembrane fence (TJ) (1) and cell-cell adhesion/mechanical tissue integrity (AJ), respectively (2,3). AJ comprise cadherins and nectins as transmembrane adhesion molecules, both linked to the submembrane cortical cytoskeleton by cytoplasmic complexes of adaptor proteins: catenins (p120-catenin and ␤-catenin/␣-catenin) for E-cadherin and afadin-ponsin for nectins (2,5).
PLEKHA7 is a recently discovered cytoplasmic component of AJ, which was identified through its interaction with the AJ protein p120-catenin (6) and with the TJ/AJ protein paracingulin (CGNL1) (7)(8)(9). PLEKHA7 is specifically associated with the zonula adhaerens (ZA) (7), and it interacts with the microtubule-interacting protein nezha (CAMSAP3) to provide a molecular linkage between the cadherin-associated protein complex and the microtubule cytoskeleton (6). Genetic and genome-wide association studies have implicated PLEKHA7 in the regulation of cardiac contractility in a zebrafish model (10), in hypertension in human cohort studies (11)(12)(13) and in a rat model (14), and in primary angle closure glaucoma (15,16). In addition, PLEKHA7 has recently been involved in the control of microRNA processing (17) and susceptibility to staphylococcal ␣-toxin (18). However, the molecular and cellular mechanisms through which PLEKHA7 is implicated in the above-mentioned diseases and in its cellular functions are still largely unknown.
To gain further insight into the biochemical basis for the functions of PLEKHA7, it is essential to identify and characterize new PLEKHA7 interacting partners. Toward this goal, we describe here the identification of PDZ domain-containing protein 11 (PDZD11) as a novel PLEKHA7 interactor. PDZD11 is recruited to AJ by PLEKHA7, contributing to the stabilization of nectins at AJ and ensuring efficient early steps of junction assembly.
Yeast Two-hybrid Screen and QUantitative BAC-Intera Ctomics (QUBIC)-For the yeast two-hybrid screen, the fulllength sequence of human PLEKHA7 (residues 1-1121) was fused C-terminally to LexA in the pB27 vector. This construct was used to screen a human placenta library, in the presence of 0.5 mM 3-amino-1,2,4-triazole (Hybrigenics, Paris, France). A very high confidence interaction (PBS Score "A", based on the PIM predicted global biological score) was detected with 24 distinct clones of PDZD11, all of which contained the fulllength coding sequence of PDZD11. In addition, a PBS score A was also obtained with six clones of the known interactor paracingulin (CGNL1) (8), whereas no PLEKHA7-interacting clones were detected with scores "B" (high confidence) or "C" (good confidence). For affinity purification of PLEKHA7 prior to mass spectrometry, a BAC clone harboring the mouse Plekha7 gene (RP23-350A9, BACPAC Resources Center) was recombined with an N-terminal GFP BAC tagging cassette (NFLAP: GFP_ PreScission_S-peptide_TEV_FLAG) (22). Precise incorporation of the tagging cassette was confirmed by PCR and sequencing. Next, the GFP-tagged BAC was isolated from bacteria using the NucleoBond PC100 kit (Macherey-Nagel, Germany), and HeLa Kyoto cells were transfected using Effectene (Qiagen) and cultured in DMEM selection media containing 400 g/ml geneticin (G418, Invitrogen). The pool of HeLa cells stably expressing the tagged Plekha7 transgene was analyzed by Western blot and immunofluorescence using an anti-GFP antibody (Roche Applied Science) to verify correct protein size and localization of the tagged transgene. Affinity purification and mass spectrometry were carried out as described (23). PDZD11 (Target ID Q5EBL8-2) and PLEKHA7 (Target ID Q6IQ23-2) were identified in the immunoprecipitate with the highest identical score of 10.296 (three peptides).
Analysis of Dynamics of Junction Assembly by Calcium Switch-A modification of the classical protocol (29) was used here, because mCCD cells undergo complete junction disassembly upon removal of extracellular calcium much more rapidly than MDCK cells. mCCD cells (300,000 cells inoculated on each round coverslip in 24-well plates 24 h earlier) were rinsed twice with PBS and incubated for 30 min in S-MEM low calcium medium (Gibco catalog no. 11380-052), containing 10 mM Hepes, pH 7.4, 1ϫ non-essential amino acids, 5% FBS, 2 mM L-glutamine, 2 M EGTA, and 1ϫ penicillin/ streptomycin, to disassemble junctions. To initiate junction assembly, the S-MEM medium was replaced with normal medium, and cells were fixed for immunofluorescent labeling at 15 and 30 min and 1 and 4 h after the switch. Junctional assembly of E-cadherin and cingulin was evaluated by immunofluorescence as described below.
Generation of Knock-out (KO) mCCD Cells by CRISPR/Cas9 Genome Editing-A construct for the CRISPR/Cas9-mediated knock-out (KO) of PDZD11 KO in mouse mCCD cells was generated by cloning a guide-RNA targeting exon-1 of the mouse Pdzd11 gene (sequence: GCCGGCCTATGAAAACC-CTC) into the BbsI site of px458 CRISPR plasmid (Addgene catalog no. 48138, containing GFP). Cells were transfected by Lipofectamine2000, and isolated clones were obtained 2 days later by fluorescence-activated cell sorting (using a Beckman Coulter MoFlo Astrios sorter, Flow Cytometry Service, University of Geneva Medical School). Single PDZD11-KO clones were screened by immunoblotting, and the genomic sequence of the Pdzd11 locus was determined in distinct KO clones by PCR amplification and subcloning of a fragment spanning a 1-kb region, from the 5Ј-untranslated region of the Pdzd11 gene to exon-2. Clone 1B2 showed the insertion of a G in both alleles (resulting sequence: GCCGGGCCTATGAAAACC-CTC). Clone 2C7 showed a 2-bp deletion in allele 1 (G-GGCCT-ATGAAAACCCTC) and a 164-bp insertion in allele 2 (GCC-164 bp-GGCCTATGAAAACCCTC). Clone 1C8 showed a 2-bp deletion in allele 1 (G-GGCCTATGAAAACCCTC) and a 1-bp insertion in allele 2 (GCCGGGCCTATGAAAACCCTC). Rescue clones were obtained by expression of either GFP or GFP-tagged PDZD11 (cloned in pTre-2Hyg), and selection of clones was made in hygromycin-containing medium. The three clones behaved similarly, but only data from one clone (2C7) are shown in Figs. 4 and 6 for the sake of space. The generation of PLEKHA7-KO mCCD cells by CRISPR-mediated genome editing, the generation of rescue clones, and their full characterization will be described elsewhere. 4 Transfection was carried out using either Lipofectamine 2000 (Invitrogen) or the calcium phosphate method (for HEK293T cells).
Immunofluorescence Microscopy, Immunohistochemistry, and Proximity Ligation Assay-Conventional fixation of cells for labeling of junctional proteins was with methanol (10 min at Ϫ20°C), and incubations with antibodies were for 30 min at 37°C. To detect cytoplasmic GFP in PLEKHA7-KO cells, cells were fixed with Triton X-100 and paraformaldehyde (26). Immunohistochemistry was carried out as described previously (7). Proximity ligation assay (Sigma, catalog no. DUO92101) was carried out according to the manufacturer's instructions, using rabbit anti-nectin1/3 and mouse anti-PLEKHA7. Coverslips were mounted with Vectashield (Reactolab) and imaged with a Zeiss LSM700 confocal microscope, equipped with ϫ63, 1.3NA objective (Zeiss LSM software). LSM 8-bit files were converted to TIFF files separately for each channel using ImageJ software and linearly adjusted and cropped using Adobe Photoshop. Quantification of immunofluorescent junctional labeling for nectin-3 was carried out with ImageJ software, as described previously (8), using either E-cadherin or cingulin as an internal reference standard for junctional labeling (E-cadherin is not affected by PDZD11-KO, and cingulin is not affected by PLEKHA7-KO). Quantification of junction assembly during the calcium switch was carried out by drawing the outline of the cell periphery of 20 cells/field (three fields for each sample) (each cell was identified by DAPI-labeled nuclei), and by measuring the length of the regions of cell-cell contact that were labeled by either anti-E-cadherin or anti-cingulin antibodies. The percentage of junction assembly was obtained by ratioing the labeled lengths over the total calculated lengths of the cell peripheries.
Immunoprecipitation, GST Pulldown, and Immunoblotting-Immunoprecipitation of endogenous proteins from Caco2, mCCD, and meEC cells was carried out by rinsing cells with cold PBS and incubating them in 0.5 ml (coIP) of buffer (150 mM NaCl, 20 mM Tris-HCl, pH 7.5, 1% Nonidet P-40, 1 mM EDTA, 5 mg/ml antipain, 5 mg/ml leupeptin, 5 mg/ml pepstatin, 1 mM PMSF) for 10 min at 4°C. After sonication (8 s at 40% power, Branson sonifier), the lysate was centrifuged for 15 min at 13,000 rpm. The supernatant was kept (cytoskeleton-soluble fraction). The pellet (cytoskeleton-insoluble fraction) was resuspended in 50 l of SDS-buffer (1% SDS, 10 mM Tris-HCl, pH 7.5, 2 mM EDTA, 0.5 mM DTT, 0.5 mM PMSF), sonicated 2 s at 10% power, incubated at 100°C for 5 min, clarified by centrifugation, brought to a volume of 0.5 ml with co-IP buffer, and mixed with the cytoskeleton-soluble fraction to make a "total" IP lysate. Dynabeads protein-G or protein-A (Invitrogen) (20 l) were coupled to antibodies (either 2 l of serum or 2 g of purified antibody) at 4°C for 90 min, washed with PBS containing 5% bovine serum albumin and 1% Nonidet P-40, and incubated with 0.1-0.3 ml of total IP lysate for 16 h at 4°C. After washing three times with coIP buffer, beads were incubated in 20 l of SDS sample buffer at 100°C for 5 min, and between 5 and 15 l of eluate were loaded on SDS gels (12% polyacrylamide for PDZD11 and 8% for PLEKHA7). SDS-PAGE and immunoblotting were carried out as described previously (21,27). Protein loadings were normalized by immunoblotting with anti-tubulin antibody. Detection of proteins after Western blotting was performed by Odyssey Imager (LI-COR). Quantification of nectin-1 and nectin-3 protein levels by immunoblotting in normalized samples was carried out using ImageJ software analysis of scanned blots. Sf21 cells expressing His-tagged PLEKHA7 were lysed in LBT buffer (150 mM NaCl, 20 mM Tris-HCl, 5 mM EDTA, 1% Triton X-100, 5 mg/ml antipain, 5 mg/ml leupeptin, 5 mg/ml pepstatin, 1 mM PMSF), sonicated four times for 30 s at 30% power and centrifuged for 20 min at 13,000 rpm (8). Purification of GST fusion proteins and GST pulldowns with full-length His-tagged PLEKHA7 expressed in baculovirus-infected insect cells was performed as described previously (24,25).
Data Analysis-All experiments were carried out at least in triplicate, and similar results were obtained when using distinct clones of either PLEKHA7-KO or PDZD11-KO cells. For immunoblots and immunofluorescence data, one representative example out of three independent experiments is shown. Statistical significance of quantitative data was determined by Student's t test.

PDZ Domain-containing Protein 11 Is a Novel
Interactor of PLEKHA7-We sought to identify novel interactors of PLEKHA7 by yeast two-hybrid screen and by mass spectrometry analysis of affinity-purified PLEKHA7 immunoprecipitates. Both approaches identified PDZD11, also known as PISP (plasma membrane calcium ATPase-interacting single-PDZ protein) or AIPP1 (ATPase-interacting PDZ protein) (30 -32) as the highest confidence interactor of PLEKHA7. We generated a rabbit polyclonal antiserum against PDZD11, which specifically labeled a polypeptide of 17 kDa in lysates of MDCK, mCCD, human keratinocytes (HaCaT), human lung carcinoma cells (A427 and A549), mouse endothelial cells (bEnd.3, meEC), human colon carcinoma cells (Caco-2), and mouse kidney tissue, consistent with the predicted size of PDZD11 (Fig. 1, A and B). The antiserum specifically immunoprecipitated exogenously expressed PDZD11, containing N-terminal GFP and C-terminal Myc tags, from MDCK cell lysates (Fig. 1C). To validate the findings of the two-hybrid screen and mass spectrometry analysis, we explored the PLEKHA7-PDZD11 interaction by co-immunoprecipitation and GST-pulldown experiments. First, PDZD11 was detected in PLEKHA7 immunoprecipitates from lysates of endothelial (aorta, meEC) and epithelial (kidney, mCCD) cells (Fig. 1D). Second, PLEKHA7 was detected in PDZD11 immunoprecipitates from lysates of Caco2 colon epithelial cells (Fig. 1E). Third, bacterially expressed full-length PDZD11 interacted with fulllength His-tagged PLEKHA7, isolated from insect cells (Fig.  1F). Taken together, these results demonstrate that PDZD11 is expressed in epithelial and endothelial cells and tissues, where it forms a complex and directly interacts with PLEKHA7.
The First WW Domain of PLEKHA7 Interacts with the Nterminal 44 Residues of PDZD11 (P7-ID)-To identify the sequences involved in PLEKHA7-PDZD11 interaction, we generated mutated constructs of the proteins to use in GST-pull-down assays. PLEKHA7 comprises WW and PH domains in its N-terminal half and coiled-coil and proline-rich domains in its C-terminal half ( Fig. 2A). PLEKHA7 prey (N-and C-terminal and central regions) containing N-terminal YFP and C-terminal Myc tags were obtained from lysates of stably transfected MDCK cell lines (Fig. 2, A and C) (21). Another PLEKHA7 prey consisted of His-tagged full-length PLEKHA7 obtained from lysates of insect cells infected with baculovirus ( Fig. 2H) (8). PDZD11 consists of a 44-residue N-terminal peptide, followed by a 81-residue PDZ domain and a 14-residue C-terminal peptide (Fig. 2B). PDZD11 baits consisted of GST fused N-terminally to 1) full-length sequence, 2) N-terminal peptide, 3) PDZ domain, 4) the 21 C-terminal residues of the PDZ domain, followed by the 14-residue C-terminal peptide (Fig. 2, B and H). Immunoblot analysis of GST pulldowns showed that fulllength PDZD11 interacts with the YFP-tagged N-terminal region of PLEKHA7, but not with C-terminal or central domains of PLEKHA7 (Fig. 2C), indicating that the PDZD11interacting region in PLEKHA7 lies within residues 1-282 of PLEKHA7. This region includes Trp-Trp (WW) and PH domains ( Fig. 2A). GST fusions of these domains, together or separately, were generated to prepare baits for pulldown assays ( Fig. 2A). Pulldown experiments using HA-tagged PDZD11 as a prey showed that the N-terminal fragment of PLEKHA7 (residues 1-162) comprising the WW domains, either alone or in combination with the PH domain, but not the PH domain alone, specifically binds to PDZD11 (Fig. 2, D and E, for control pulldowns with CFP-HA). PLEKHA7 actually contains two WW domains (33). The first contains the canonical Trp-Trp signature, and the second contains a Trp and a Phe residue; the two domains are separated by a spacer region of 11 residues. We generated GST fusions comprising either the first WW (W1, Fig. 2F) or the second (W2, Fig. 2F), or both (W1 ϩ 2), and we examined their interaction with full-length PDZD11 by pulldown assays (Fig. 2G). Only the first, but not the second, WW domain could interact, when isolated, with PDZD11. However, a stronger interaction was detected when both W1 and W2 domains were present in the fusion protein (Fig. 2G). Finally, GST fusion proteins were generated to map the region of PDZD11 involved in interaction with PLEKHA7 (Fig. 2, B and H). GST fusions of full-length PDZD11 and N-terminal peptide interacted with the full-length His-tagged PLEKHA7 in insect cell lysates, but no interaction of PLEKHA7 was detected with GST fusions of either the PDZ domain of PDZD11 or the C-terminal construct (Fig. 2H). Because its sequence is unique, and it binds specifically to PLEKHA7, we propose to name the 44-residue N-terminal sequence of PDZD11 P7-ID (PLEKHA7 interacting domain).
PLEKHA7 Recruits PDZD11 to Epithelial Adherens Junctions-Because PLEKHA7 is accumulated at the zonula adhaerens (7), the interaction between PLEKHA7 and PDZD11 predicts that PDZD11 should also be localized at AJ in epithelial cells and tissues. To test this, we first labeled epithelial cell lines and kidney tissue by immunofluorescence with the anti-PDZD11 antiserum. Junctional labeling for PDZD11 that co-localized with TJ and AJ markers was detected in MDCK, A427, meEC, and additional epithelial cell types and kidney tissue (Fig. 3A, data not shown). To establish more clearly the localization of PDZD11, we imaged serial apico-basal confocal z-sections of MDCK cells expressing GFP-tagged PDZD11 and double-labeled with markers of either TJ (cingulin and ZO-1) or AJ (␤-catenin) (Fig. 3, B and C). Immunofluorescence analysis showed that GFP-PDZD11 labeling only partially overlapped with endogenous cingulin and ZO-1 labeling along apicolateral junctions (arrow and arrowhead in Fig. 3B), whereas it overlapped more precisely with endogenous labeling for junctional ␤-catenin (arrows in Fig. 3C). This is very similar to the localization of endogenous and exogenous PLEKHA7 in epithelial cells (8,21). Next, we asked whether PDZD11 is recruited to adherens junctions by PLEKHA7, by examining the localization of endogenous PDZD11 in either wild-type (WT) mCCD or in cells where both PLEKHA7 alleles were targeted by CRISPRmediated gene disruption, and which do not express PLEKHA7 (PLEKHA7-KO) (Fig. 3, D-F). Junctional labeling for PDZD11 was detected in WT cells but not in PLEKHA7-KO cells (arrow and arrowhead in Fig. 3D, respectively). Next, we transiently expressed either GFP or GFP-tagged PDZD11 in either WT or PLEKHA7-KO mCCD cells (Fig. 3E). Labeling for the GFP control protein was detected diffusely in the cytoplasm both in WT and in PLEKHA7-KO cells (asterisks in Fig. 3E). Importantly, exogenous GFP-tagged PDZD11 was detected exclusively at junctions in WT cells (arrows in Fig. 3E, WT), where it co-localized precisely with PLEKHA7. In contrast, exogenous GFPtagged PDZD11 was localized exclusively in the cytoplasm in PLEKHA7-KO cells (asterisks and arrowheads in Fig. 3E, PLEKHA7-KO). PLEKHA7-KO mCCD cells form junctions similar to WT cells, as determined by labeling with most markers of TJ and AJ, and establish a normal TJ barrier. 4 This indicates that the lack of association of exogenous PDZD11 with junctions in PLEKHA7-KO cells is specifically due to the absence of PLEKHA7, rather than to secondary effects on other components of zonular junctions. To confirm the role of PLEKHA7 in recruiting PDZD11, we carried out a rescue experiment by re-expressing either GFP-or YFP-tagged PLEKHA7 and either WT or mutant in PLEKHA7-KO mCCD cells (Fig. 3F). Immunofluorescence analysis showed that the junctional localization of PDZD11 was rescued in PLEKHA7-KO cells, upon exogenous expression of either full-length PLEKHA7 (arrows in Fig. 3F, YFP-PLEKHA7) or PLEKHA7 lacking the PH domain (arrows in Fig. 3F, YFP-P7-⌬PH), but not upon expression of PLEKHA7 lacking the N-terminal WW domains (arrowhead in Fig. 3F, YFP-P7-⌬WW). In addition, junctional labeling for endogenous PDZD11 was increased in WT cells that overexpressed exogenous PLEKHA7 (data not shown), supporting the idea that the accumulation of PDZD11 at junctions is promoted by PLEKHA7. Taken together, these results demonstrate that PLEKHA7 recruits PDZD11 to epithelial junctions through its WW domain-containing N-terminal region.

PDZD11 Is Required to Stabilize Nectin-1 and Nectin-3 at Epithelial Adherens Junctions-
To determine the role of PDZD11 in the molecular architecture of epithelial junctions, we analyzed the localization and expression of zonular (ZO-1, afadin, paracingulin, cingulin, and PLEKHA7) and zonular/lateral (occludin, E-cadherin, ␣-catenin, ␤-catenin, and p120catenin) junctional markers in cells where both alleles of Pdzd11 were inactivated by CRISPR-mediated genome editing. When mixed cultures of WT and PDZD11-KO cells were analyzed by immunofluorescence (to identify better the differences between WT and KO cells in the same field), the localization of PLEKHA7 and ZO-1 was not detectably altered in PDZD11-KO cells, when compared with neighboring WT cells (Fig. 4A, top row). No labeling for PDZD11 was detected in KO cells by either immunofluorescence (Fig. 4A) or immunoblotting (Fig.  4B), confirming the specificity of the anti-PDZD11 antiserum. The immunofluorescent localization ( Fig. 4A and data not shown) and the levels of expression (Fig. 4B), not only of PLEKHA7 but also of the other zonular and zonular/lateral tight and adherens junction markers (E-cadherin, cingulin, occludin, afadin, p120-catenin, ␣-catenin, ␤-catenin, and paracingulin), were not detectably altered in confluent PDZD11-KO cells (Fig. 4, A and B). Next, we examined the localization and expression of nectins, Ig-like cell adhesion molecules localized at adherens junctions, that display PDZbinding motifs at their C-terminal cytoplasmic tail (34,35). In three distinct PDZD11-KO clones, we observed a decrease in junctional labeling for nectin-3 (Fig. 4, C, top row, showing only clone 2C7, and quantification in D). The decreased junctional labeling for nectin-3 in PDZD11-KO cells was rescued by re-expression of GFP-tagged PDZD11 but not of GFP (double arrows in Fig. 4C, 2nd row). By immunoblotting, we observed decreased levels of expression of both nectin-1 and nectin-3 in PDZD11-KO cells (Fig. 4E and quantification in F), with no significant changes in nectin mRNA levels, as determined by qRT-PCR (Fig. 4G). Nectin-1 and nectin-3 protein levels in PDZD11-KO cells were rescued by re-expression of GFPtagged PDZD11, but not of GFP alone (Fig. 4E and quantifications in F). To ask whether the decreased levels of nectin-1 and nectin-3 in PDZD11 KO cells were due to decreased stability, e.g. increased degradation, we treated either WT or KO cells with the proteasome inhibitor MG132 (Fig. 4H and I for quantification). Immunoblot analysis showed that normal levels of nectin expression were rescued in PDZD11 KO cells upon treatment with MG132 (Fig. 4, H and I), revealing a proteasome-dependent degradation of nectin. Next, we tested the hypothesis that the junctional localization and expression of nectin-1 and nectin-3 should also be altered in PLEKHA7-KO cells, because PLEKHA7 recruits PDZD11 to junctions. Immunofluorescence (Fig. 4C, 3rd and 4th rows) and immunoblotting (Fig. 4J and quantifications in K) analyses showed that PLEKHA7-KO cells phenocopied PDZD11KO cells, because junctional nectin-3 labeling and nectin-1 and nectin-3 protein levels were reduced in PLEKHA7-KO cells and rescued upon re-expression of PLEKHA7. These data demonstrate that although PDZD11 is not required to maintain the molecular organization of either TJ or the E-cadherin-associated complex at steady state, the PLEKHA7-PDZD11 complex is required for the efficient recruitment and stabilization of nectins at AJ.
PDZD11 Binds Directly to Nectin-1-To examine the molecular basis for the regulation of nectin junctional accumulation and stability by the PDZD11-PLEKHA7 complex, we studied the physical interaction of PDZD11 and PLEKHA7 with nectins. Immunoprecipitation experiments show that endogenous nectin-1 and nectin-3 immunoprecipitate PDZD11 (Fig. 5A). In contrast, we were unable to detect PLEKHA7 in nectin immunoprecipitates. To determine whether PDZD11 or PLEKHA7 interacts directly with nectins, GST fusion constructs of PDZD11 and PLEKHA7 were used as baits to bind Myc-tagged nectin, expressed in HEK293T cells. Immunoblot analysis showed that full-length nectin-1 is detected specifically in GST pulldowns of PDZD11 but not in GST pulldown using fragments of PLEKHA7 from the N-terminal, central, or C-terminal domains (Fig. 5B). Next, we asked whether the interaction between nectin-1 and PDZD11 depends on the interaction between the PDZ domain of PDZD11 and the PDZ-binding motif of nectin-1. No binding to the PDZD11 GST fusion bait was observed when the four-residue PDZ-binding motif of nectin-1 was deleted (Myc-nectin1-⌬4 in Fig. 5C). Furthermore, only the PDZ fragment of PDZD11 could bind nectin-1 but not the N-terminal P7-ID nor the C-terminal fragments (Fig. 5D). Finally, we used the PLA to detect the physical association between PLEKHA7 and either nectin-1 or nectin-3. In both cases, a strong junctional accumulation of labeling was detected in WT cells (double arrows and magnified insert in Fig. 5E, WT panels). However, only background labeling was observed in PLEKHA7-KO cells (arrowhead and magnified inset in Fig. 5E, PLEKHA7-KO panels), demonstrating that the proximity between nectins and PLEKHA7 is specifically due to PLEKHA7. Importantly, in PDZD11 KO cells there was a strong reduction in junctional signal with respect to WT cells (arrow and magnified inset in Fig. 5E, PDZD11-KO panels). This indicates that PDZD11 is critical to ensure maximal physical proximity between PLEKHA7 and nectins and that residual association between nectins and PLEKHA7 may be indirect, through afadin. Taken together, the GST pulldown and PLA results demonstrate that a complex of PLEKHA7 and PDZD11 is required for the efficient recruitment and stabilization of nectins at AJ, through the binding of the PDZ-binding motif of nectins to the PDZ domain of PDZD11.

PDZD11 Is Required for the Efficient Early Assembly of Apical Junctions in the Calcium-switch Experimental Model-To
address the cellular functions of PDZD11, we hypothesized, MAY 20, 2016 • VOLUME 291 • NUMBER 21 based on the effects of the PLEKHA7-PDZD11 complex on nectin junctional localization and stability, that we might detect a phenotype related to the function of nectins, which are implicated in early junction assembly (see under "Discussion") (5). Analysis of junctions at steady state showed that WT and PDZD11-KO cells are indistinguishable when examining a large number of markers (Fig. 4). Therefore, we examined the role of PDZD11 in the dynamic assembly of apical junctions, by using the calcium-switch protocol, which is the most commonly used experimental model to investigate the biogenesis of junctions in epithelial cells (29). Cells were immunofluorescently labeled with antibodies against cingulin (TJ marker) and E-cadherin (AJ marker) at different times after the start of junction assembly by the calcium switch (15 and 30 min and 1 and 4 h, Fig. 6, A and D). Junction assembly was scored by quantifying linear junctional labeling for each marker versus putative total junctional length (Fig. 6, B, C, E, and F). At 15 and 30 min after the calcium switch, WT cells showed 20 and 35% of junction assembly, respectively, corresponding to clearly identifiable segments of E-cadherin and cingulin labeling at the cell periphery (arrows in Fig. 6, A, WT, and quantifications in B and C). In contrast, in PDZD11 KO cells only 8 and 20% of junctions were assembled after 15 or 30 min, respectively, corresponding to fewer segments of the cell peripheries labeled by the cingulin and E-cadherin antibodies (arrows in Fig. 6, A, and quantifications in B and C). In PDZD11-KO cells, E-cadherin and cingulin labeling were detected mostly in a granular perinuclear pattern (arrowheads in Fig. 6A), suggesting an accumulation in endoplasmic reticulum/Golgi compartments. The differences between WT and PDZD11-KO cells were no longer observed at 1 and 4 h after the beginning of the calcium switch (WT and KO in Fig. 6D and quantifications in E and F). After 4 h, junctions were fully assembled (Fig. 6, D and F). The slight early delay in junction assembly observed in PDZD11-KO cells was specifically due to the lack of PDZD11 and not to clone-dependent variations, because the phenotype was rescued by the stable re-expression of GFP-PDZD11 but not by the re-expression GFP alone (KOϩGFP-PDZD11 and KOϩGFP in Figs. 6, A-C, and see 4E for immunoblots of transgenes). Therefore, the absence of PDZD11 causes a slight delay in the very initial phases of junction assembly, but this delay is eventually overcome, allowing the formation of junctions with localization and expression of junctional proteins indistinguishable from that of WT cells.

Discussion
Here, we identify PDZD11 as a new interactor of PLEKHA7, and we show that PDZD11 is recruited by PLEKHA7 to AJ, to promote the efficient junctional recruitment and stabilization of nectins and the efficient early phases of assembly of apical junctions in epithelial cells. These results uncover a new function for PLEKHA7, in organizing the junctional clustering of PDZ-interacting proteins, through PDZD11. They also show that PLEKHA7 stabilizes both major adhesion transmembrane proteins of adherens junctions (cadherins and nectins), through binding to components of the respective cytoplasmically associated complexes: p120-catenin/paracingulin for E-cadherin and PDZD11-afadin for nectins (Fig. 7).
Considering the potentially critical role of PLEKHA7 in major human pathologies, such as hypertension and glaucoma, it is essential to learn more about its molecular interactors and its cellular functions. Here, we report that two different types of screening technologies identify PDZD11 as the top hit among PLEKHA7 interactors, and we confirm the relevance of the PLEKHA7-PDZD11 interaction, as well as of the interaction FIGURE 4. PDZD11 and PLEKHA7 are required for the efficient recruitment and stabilization of nectins at adherens junctions. A, PDZD11 is not required for the integrity of adherens and tight junctions. Top panel, immunofluorescence analysis of mixed cultures of wild-type (WT) and PDZD11 knock-out (KO) cells with antibodies against PDZD11 (rabbit, Alexa488), PLEKHA7 (guinea pig, Cy3), and ZO-1 (rat, Cy5). Merge images show nuclei labeled in blue (DAPI). Arrows indicate junctional labeling; arrowheads indicate lack of junctional labeling; asterisks indicate KO cells. Bottom panels, individual cultures (not mixed) of either WT or PDZD11-KO cells were immunofluorescently labeled with antibodies against E-cadherin, afadin, ␣-catenin, ␤-catenin, p120-catenin, cingulin, paracingulin, and occludin. B, immunoblot analysis of lysates of wild-type (WT) and PDZD11-KO (KO) mCCD cells, using antibodies against PDZD11, PLEKHA7, afadin, paracingulin, p120-catenin (p120), E-cadherin, ␣-catenin, ␤-catenin, ZO-1, cingulin, occludin, and tubulin (for loading control). C, PDZD11 and PLEKHA7 are required for the junctional accumulation of nectin-3. Immunofluorescence analysis of (from top to bottom): 1st row: mixed cultures of WT cells ϩ 2C7 PDZD11-KO clone to identify differences in neighboring WT versus KO cells; 2nd row: mixed cultures of the PDZD11-KO clone (2C7) rescued with GFP together with cells of the same clone rescued with GFP-PDZD11: again, to compare directly nectin-3 labeling in mock-rescued cells (GFP) or in cells rescued with the GFP-PDZD11 construct. 3rd row: mixed cultures of WT cells ϩ PLEKHA7-KO clone 1 (KO-1); 4th row: mixed cultures of the PLEKHA7-KO-1 clone rescued with GFP together with cells of the same clone rescued with GFP-PLEKHA7. Cells were labeled with antibodies against either PDZD11 (1st and 2nd rows) or PLEKHA7 (3rd and 4th rows) to identify WT and KO cells (or cells expressing exogenous GFP-PDZD11 or GFP-PLEKHA7 rescue), with antibodies against nectin-3 to examine nectin-3 accumulation, and with either E-cadherin (E-cad) (1st and 2nd rows) or cingulin (CGN) (3rd and 4th rows) to identify junctional regions. The same results were obtained with additional clonal lines of PDZD11-KO cells (1B2 and 1C8) and PLEKHA7 (KO-5 and KO-7) (data not shown). Arrows and arrowheads indicate normal or decreased junctional labeling, respectively; double arrows indicate increased junctional labeling in cells expressing either the GFP-PDZD11 or GFP-PLEKHA7 constructs; asterisks identify the nuclei of either KO cells (1st and 3rd row) or KO cells rescued with GFP alone (2nd and 4th rows), which is not visible in the cytoplasm, due to methanol fixation. D, histograms showing the quantification of junctional labeling for nectin-3 in the corresponding panels, expressed as a ratio between nectin-3 and either E-cadherin (rows 1 and 2) or cingulin (rows 3 and 4) labeling in junctions between WT cells (black) or between KO cells (gray). E, immunoblot analysis of nectin-1 and nectin-3 levels in lysates from WT cells and from the PDZD11 2C7 clonal line, either without or with re-expression of exogenous GFP-PDZD11 (GFP-P11) (or GFP, as a control) for rescue. Immunoblotting with anti-GFP is also shown, to detect expression of rescuing proteins. between PDZD11 and nectin, by biochemical direct in vitro binding assays. We show that the first WW domain of PLEKHA7 is sufficient for direct interaction with PDZD11 and that the 44-res-idue N-terminal sequence of PDZD11 interacts with PLEKHA7. Moreover, the PDZ-binding four-residue motif of nectin-1 mediates the interaction of nectin with the PDZ domain of PDZD11. The sequence of the P7-ID shows no homology to the N-terminal sequences of other single PDZ domain proteins and may function as a PDZ "supramodule," to allow higher specificity of interaction of PDZD11 with its target sequences (36). PLEKHA7 binding to this region may also modulate the affinity of binding of the PDZ domain of PDZD11 with its target sequences. Conversely, the P7-ID of PDZD11 comprises eight proline residues, including two PP (Pro-Pro) residues, but none of them shows a consensus ligand sequence belonging to any of the four previously identified WWbinding motif groups (37). Additional experiments are required to determine whether the proline motifs of PDZD11 define a new group of WW-interacting sequences and which residues of the WW1 domain of PLEKHA7 are implicated in its interaction with PDZD11. In summary, we assigned a specific protein-interaction function to the most N-terminal WW region of PLEKHA7, and we dissected the structure-function relationships in PDZD11.
PDZD11 was previously described as PISP, based on its binding to the cytoplasmic domain of all plasma membrane Ca 2ϩ -ATPase b-splice variants (30). PDZD11 is also known as AIPP1 through its binding to the cytoplasmic tail of the Menkes copper ATPase ATP7A (31) and as an interactor of the sodium-dependent multivitamin transporter (32). The proposed function of PDZD11, a ubiquitously expressed protein, was to provide a scaffold for these transmembrane proteins, by binding through its PDZ domain, to the PDZ-binding motif at their C terminus. Supporting this idea, depletion of PDZD11 reduced the cell surface expression of the sodium-dependent multivitamin transporter, as assessed by cell surface biotinylation assay (32). Other single PDZ proteins, such as MALS (mammalian homolog of , are implicated in intracellular transport and targeting of their partner proteins (38), and future studies should determine whether PDZD11 also plays this role, for example by promoting the efficient transport of PDZ-binding proteins such as nectins to the cell surface. Here, we discover a new role for PDZD11. It promotes the efficient recruitment and stabilization of nectins at AJ of epithelial cells. The C-terminal cytoplasmic domain of nectins interacts with the PDZ domains of several junctional partners as follows: afadin (35), which is important for the accumulation of nectin-1 at AJ (39), and depending on nectin isoform, also Par-3 (40), PICK-1 (41), MUPP1 (42), PATJ (42), and MPP3 (43). Therefore, redundant interactions can stabilize nectins at junctions (44). This is consistent with our observation that KO of PDZD11 in epithelial cells does not result in the complete loss of junctional nectins. PDZD11 may function not only through direct scaffolding of nectins but also indirectly by affecting the conformation of the N-terminal region of PLEKHA7, which interacts with afadin, hence affecting the PLEKHA7-afadin interaction and the ability of afadin to form an efficient scaffold for nectins.
Because PDZD11 promotes the efficient recruitment of nectins to ZA, we hypothesized that one cellular function of PDZD11 may be related to the function of nectins. Few studies have been carried out to address the function of nectins in cultured epithelial cells. Nectins independently initiate, subsequently followed by E-cadherin, the formation of AJ, by trans- FIGURE 5. Nectin forms a complex and interacts directly with PDZD11. A, PDZD11 and nectins form a complex. Immunoblotting analysis of either nectin-1 or nectin-3 immunoprecipitates (from Caco2 whole cell lysates) with antibodies against PDZD11 (top), nectin-1 (middle), and nectin-3 (bottom). B, PDZD11 but not PLEKHA7 fragments interact directly with nectin-1. Immunoblotting analysis of GST pulldowns using either GST or GST fused to the following fragments of PLEKHA7: WW (residues 1-162, see Fig. 2); PH (residues 120 -300); proline-rich and coiled-coil domains (PCC, residues 500 -844); C-terminal (residues 821-1121 (7)), and Myc-tagged-nectin-1 as a prey. Dotted boxes indicate recombinant proteins stained by Ponceau-S. C, PDZD11 binds to the PDZ-binding motif of nectin-1. Immunoblotting analysis of GST pulldowns using either GST or GST-PDZD11 as baits and either Myc-tagged nectin-1 WT (Myc-nectin1) or the nectin-1 mutant lacking the last PDZ-binding motif 4 residues (Myc-nectin1-⌬4) (both expressed in HEK293T cells) as preys. The Ponceau-S-stained blot shows the amounts of recombinant proteins used as baits. Numbers on the left indicate size (kDa) and migration of prestained markers. D, PDZ domain of PDZD11 binds to nectin-1. Immunoblotting analysis of GST pulldowns using either GST or different fragments of PDZD11 (see Fig. 2) as baits, and Myc-tagged nectin-1 as a prey. E, detection of PDZD11-dependent PLEKHA7-nectin association by PLA. Cells were labeled with mouse anti-PLEKHA7 antibodies, and either rabbit anti  interacting at protrusions of neighboring cells, giving rise to spot-like junctions that subsequently mature into belt-like AJ (45). Nectins and E-cadherin cooperatively organize AJ, by interacting through their cytoplasmic partners (46), although nectins can efficiently drive the formation of AJ and TJ even in the absence of E-cadherin (47). Nectin mutations that prevent their trans-interaction affect junction formation in MDCK cells, causing a delay in AJ formation (48). Consistent with the reduction of junctional nectin levels, and the role of nectins in initiating junction assembly, the KO of PDZD11 resulted in a reduction of the efficiency of assembly of zonular junctions, which was detectable at the 15-and 30-min time points in the calcium switch experimental model. Junction assembly resumed to normal kinetics at later time points in PDZD11-KO cells, probably due to the cooperative and redundant adhesive functions of E-cadherins and the residual nectin molecules.
We propose a new model for the architecture of the ZA (Fig.  7), based on the results presented here. By binding to p120catenin and paracingulin on one side and to PDZD11 and afadin on the other side, PLEKHA7 establishes a molecular bridge between nectin-and cadherin-based protein complexes at the ZA (Fig. 7), distinct from the connection provided through the afadin-ponsin complex (49). Through PDZD11, PLEKHA7 contributes to recruiting nectins to the ZA. In addition, by binding both to afadin and to PDZD11, PLEKHA7 further strengthens the cytoplasmic scaffold for nectins, by bridging together these two cytoplasmic scaffolding proteins. Finally, by binding to actin filaments through afadin (and potentially other actin-binding proteins (17)), and to microtubules through nezha (6), PLEKHA7 contributes to reinforcing the cytoskeletal anchoring of both nectin and cadherin adhesion receptors.
The PLEKHA7-PDZD11 module that we describe provides a new type of molecular machinery that brings PDZ-binding membrane proteins to the AJ in epithelial and endothelial cells. This is important to explore new mechanistic hypotheses to understand at the molecular level the role of PLEKHA7 in tissue physiology and pathology. In fact, it has not escaped our attention that the interaction of PDZD11 with the plasma membrane Ca 2ϩ -ATPase (30) may be relevant to the role of PLEKHA7 in the control of calcium homeostasis. Studies on zebrafish show that loss of the PLEKHA7 homologue hadp1 leads to a reduced rate of extrusion of cytoplasmic Ca 2ϩ during diastole (10). Furthermore, in a model of PLEKHA7-KO rats, increased endothelial NOS signaling and increased intracellular calcium detected in PLEKHA7-KO aortic endothelial cells suggest that reduced blood pressure in KO animals may depend on PLEKHA7-and calcium-mediated increase of endothelial NOS signaling to vascular smooth muscle, leading to its relaxation (14). Significantly, both PDZD11 and PLEKHA7 are localized at endothelial junctions (Fig. 3A). Thus, we are currently investigating whether PLEKHA7 regulates plasma membrane Ca 2ϩ -ATPase, and hence calcium homeostasis, through PDZD11, to clarify mechanistically its role in the regulation of blood pressure, cardiac contractility, and glaucoma. It is also possible that altered calcium homeostasis in PLEKHA7-KO cells underlies the phenotype of decreased susceptibility and recovery after intoxication by staphylococcal ␣-toxin, shown by cells and mice lacking PLEKHA7 (18). Interestingly, PDZD11 was one of the proteins identified in the screen on Hap1 cells whose KO reduced the cytotoxic effects of ␣-toxin (18).
In summary, here we describe a new protein complex module comprising PDZD11 and PLEKHA7, whereby the single PDZ domain protein PDZD11 connects nectins to the cadherin-and cytoskeleton-associated protein complex, through its binding to PLEKHA7 (Fig. 7), to stabilize nectins at AJ and promote efficient early assembly of junctions. The results presented here also raise the hypothesis that PLEKHA7 orchestrates the membrane organization and function of proteins regulating calcium homeostasis, through PDZD11, providing a new potential molecular mechanism to explain the implication of PLEKHA7 in vascular and cardiac pathophysiology.
Author Contributions-D. G. characterized anti-PDZD11 antibodies, carried out immunofluorescence, immunoprecipitation, and pulldown analyses, and generated and characterized PDZD11-KO mCCD cells. J. S. generated PLEKHA7-KO mCCD cells and rescue cells and constructs, characterized anti-PLEKHA7 antibodies, and carried out immunofluorescence and pulldown analyses. E. V. and S. S. carried out immunoblotting, immunofluorescence, and immunoprecipitation analysis of additional cultured cell lines. I. M. and L. J. generated cDNA constructs and helped in recombinant protein expression, GST pulldown, and immunoblotting experiments. I. P., M. M., and A. A. H. conducted QUBIC experiments. S. C. initiated, organized, and supervised the project. D. G., J. S., and S. C. analyzed the data. S. C. wrote the paper.