Gap junctions, composed of Cxs (connexins), allow direct intercellular communication. Gap junctions are often lost during the development of malignancy, although the processes behind this are not fully understood. Cx43 is a widely expressed Cx with a long cytoplasmic C-terminal tail that contains several potential protein-interaction domains. Previously, in a model of cervical carcinogenesis, we showed that the loss of gap junctional communication correlated with relocalization of Cx43 to the cytoplasm late in tumorigenesis. In the present study, we demonstrate a similar pattern of altered expression for the hDlg (human discs large) MAGUK (membrane-associated guanylate kinase) family tumour suppressor protein in cervical tumour cells, with partial co-localization of Cx43 and hDlg in an endosomal/lysosomal compartment. Relocalization of these proteins is not due to a general disruption of cell membrane integrity or Cx targeting. Cx43 (via its C-terminus) and hDlg interact directly in vitro and can form a complex in cells. This novel interaction requires the N- and C-termini of hDlg. hDlg is not required for Cx43 internalization in W12GPXY cells. Instead, hDlg appears to have a role in maintaining a cytoplasmic pool of Cx43. These results demonstrate that hDlg is a physiologically relevant regulator of Cx43 in transformed epithelial cells.

INTRODUCTION

Intercellular junctions and various adhesion molecules in mammals regulate cell–cell interactions. The membrane junction complexes comprise gap junctions, tight junctions and adherens junctions [1]. Gap junctions are specialized cell membrane channels that allow direct intercellular diffusion of critical regulatory ions and small molecules <1 kDa between the cytoplasmic interiors of contiguous cells, providing a direct co-ordination of electrical and metabolic functions, proliferation and differentiation [2,3]. Gap junction channels are formed of paired connexons, each composed of six Cx (connexin) subunits, docking with their counterparts in the membrane of the neighbouring cell to form a continuous aqueous channel [4]. The core of each Cx is buried in the plasma membrane and consists of four hydrophobic α-helices connected by short loops. However, Cx43 (and other members of the α connexon subgroup [5]) have a long cytoplasmic C-terminal tail that appears to be involved in several protein–protein interactions [4].

GJIC (gap junctional intercellular communication) is tightly regulated by many mechanisms including Cx transcriptional and post-transcriptional control [6], and also post-translational regulation related to Cx protein turnover, trafficking and hemichannel and gap junction activity [3,7,8]. Cxs have a short half-life estimated to be 1–5 h and are involved in a constant cycle of synthesis, intracellular trafficking and degradation [4]. A body of evidence has accumulated to show that GJIC may be lost during malignant progression, for instance, in HPV (human papillomavirus)-positive cervical cancer [912]. Cx43, the most widespread Cx and a major component of gap junctions in stratified epithelia, has been seen to be down-regulated in epithelial carcinoma cells [13,14]. Nevertheless, the steps leading to the loss of GJIC in cancer remain largely unknown.

At cell-contact sites, the MAGUK [membrane-associated GUK (guanylate kinase)] family, members of which have several protein-interaction domains, can form protein scaffolds and comprise macromolecular complexes with their protein partners, which are thought to be involved in cell-signalling cascades and cell-morphology organization [15,16]. The terminal five amino acids of Cx43 have been shown to bind to the middle of three adjacent PDZ domains in ZO-1 (zona occludens-1) protein [17,18]. ZO-1 was first found to regulate gap junction size [19] and more recently to control the rate of formation of gap junctions from connexons in the plasma membrane [20]. Cx45 also interacts with ZO-1 through its C-terminal domain [21,22], but the role of this interaction has not been fully elucidated.

In contrast, Cx32 does not interact with ZO-1 [23], but instead interacts with the C-terminal SH3 (Src homology 3) domain of another MAGUK protein, the human homologue of the Drosophila discs large protein (hDlg) [24]. In Drosophila, Dlg is defined as a tumour suppressor and is an essential septate junction (similar to tight junctions) structural protein, which also acts as a cell growth regulator and has an important role in cell-membrane integrity, cell differentiation and cell polarity [25]. hDlg is located at intercellular contact sites in epithelial cells where it can bind to other cellular proteins including APC (adenomatous polyposis coli), another tumour suppressor, involved in the regulation of the Wnt signalling pathway [26,27]. The structure of hDlg is similar to that of ZO-1. It contains several protein-interaction domains including an SH3 domain, a HOOK domain, a GUK domain and three PDZ domains [26]. However, it also contains an N-terminal protein-interaction domain [28].

Cervical cancer progression is caused by persistent infection with anogenital-infective ‘high risk’ HPVs [29]. The W12 model of cervical cancer progression [12] was established from an immortal but untransformed HPV16-positive cervical epithelial cell line, W12 20861 (named W12G here) [30], and comprises two derivative cell lines, W12GPX and W12GPXY [12]. We showed previously that in the parental cell line, W12G, Cx43 localizes to cell junctions, providing extensive GJIC, but in W12GPXY cells that are fully transformed and invasive, Cx43 relocates from the membrane to the cytoplasm, and these cells are deficient in GJIC [12]. In the present study, we demonstrate hDlg relocation to the cytoplasm in W12GPXY cells where it co-localizes with Cx43. hDlg can co-immunoprecipitate Cx43 in cervical epithelial cells and interact directly with Cx43 in vitro. Cx43 binds by its C-terminal domain to both the N- and C-termini, but not the PDZ domains of hDlg. siRNA (small interfering RNA) depletion of hDlg showed that hDlg seems to maintain a cytoplasmic pool of Cx43, possibly antagonizing lysosomal degradation of this protein.

EXPERIMENTAL

Cell and culture conditions and drug treatment

W12G (clone 20861) cells are immortal but non-transformed cervical epithelial cells [30]. They were grown using two different culture methods, either grown on a feeder layer of mitomycin C-treated 3T3 fibroblasts in F-medium [30] or without a fibroblast feeder layer in KGM (keratinocyte growth medium) (Lonza). Culturing of 3T3 cells was in DMEM (Dulbecco's modified Eagle's medium) containing 10% (v/v) DCS (donor calf serum) (Invitrogen). W12G cells were seeded on to 3T3 monolayers at 2×105 per 10 cm dish and grown for up to 10 days with media changes at 2 day intervals. 3T3 fibroblasts were removed by trypsinization, and W12 cell layers were washed twice with PBS before any further procedures were carried out. For immunofluorescence, W12G cells were grown in serum-free KGM until colonies formed. W12GPXY cells, derived from W12G cells, are fully transformed keratinocytes with invasive properties [12]. They do not require feeder cells or mitogens for growth and were routinely cultured either in DMEM containing 10% (v/v) FBS (fetal bovine serum) or in KGM for immunofluorescence. However, in case the culture conditions altered the cellular phenotype, experiments comparing W12G and W12GPXY cells were carried out with cells grown in both F-medium and in KGM. CaSki, cervical tumour cells, were also cultured in DMEM containing 10% FBS. Cell lines were maintained at 37°C in a 5% (v/v) CO2 humidified incubator. Lysosomal inhibition was achieved using either 10 mM NH4Cl or 200 μM chloroquine and incubation for 8 h at 37°C. Solution vehicle alone was added to mock-treated cells. All chemicals were purchased from Sigma–Aldrich.

Co-IP (immunoprecipitation)

Cells at a confluency of 70% were washed twice with ice-cold PBS and scraped into 5 ml of chilled IP buffer [50 mM Tris/HCl, pH 8.0, 0.5% NP-40 (Nonidet P40), 150 mM NaCl, one tablet of PhosSTOP phosphatase inhibitor cocktail (Roche) and one tablet of Complete™ mini protease inhibitor cocktail (Roche) per 10 ml]. Cell lysates were incubated on ice for 30 min and passed through an 18-gauge syringe needle then cleared of cellular debris by centrifugation at 12000 g for 10 min at 4°C. Protein concentration was determined by Bradford's assay (Bio-Rad Laboratories). Cell extracts were pre-cleared with Protein G–Sepharose beads (Sigma) for 1 h at 4°C. Primary antibodies {negative control antibody, 1 μg of anti-SRPK1 polyclonal antibody clone 11305 (Santa Cruz Biotechnology); anti-hDlg polyclonal antibody H60 (Santa Cruz Biotechnology); and anti-ZO-1 polyclonal antibody clone ZMD436 (Invitrogen)} were then added to 100 μg of protein from each cell lysate and incubated for 2 h at 4°C with agitation. Subsequently, fresh Protein G–Sepharose was added and the volume was adjusted to 400 μl. The samples were mixed by rotation at 4°C overnight. Immunocomplexes were harvested by centrifugation (800 g, 4°C, 5 min) and washed five times with 500 μl of ice-cold IP buffer. Proteins were solubilized by adding protein-loading buffer [125 mM Tris/HCl, pH 6.8, 4% (w/v) SDS, 20% (v/v) glycerol, 10% (v/v) 2-mercaptoethanol and 0.2% Bromophenol Blue], boiled for 5 min and resolved by SDS/PAGE followed by Western blotting.

Western blotting

Cells were scraped into protein-loading buffer containing PhosSTOP phosphatase inhibitor cocktail and Complete™ mini protease inhibitor cocktail and passed through an 18-gauge syringe needle. Proteins (10 μg) were resolved by SDS/PAGE and subsequently transferred on to nitrocellulose membrane. For standard Western blotting, membranes were blocked for 1 h at room temperature (20°C) in 5% (w/v) dried skimmed milk powder dissolved in PBST (PBS containing 0.1% Tween 20), before overnight incubation at 4°C with primary antibodies diluted in the same buffer. HRP (horseradish peroxidase)-conjugated secondary antibodies [anti-(mouse IgG) or anti-(rabbit IgG); GE Healthcare] were diluted 1:2000 in PBST containing 5% (w/v) skimmed milk powder and incubated for 1 h. Following washing in PBS, the blot was developed using an ECL (enhanced chemiluminescence) kit (Thermo Scientific) and exposed to Kodak X-OMAT film. For co-IP, a CleanBlot IP detection kit (Thermo Scientific) was used. Western blotting was carried out exactly as detailed in the manufacturer's protocol. Blots were blocked in StartingBlock™ blocking buffer (Thermo Scientific) for 1 h before overnight incubation with antibody diluted in StartingBlock™ buffer. The proprietary HRP CleanBlot IP detection reagent that detects only the native antibody was used. Anti-Cx43 rabbit polyclonal antibody was purchased from Sigma–Aldrich (C-6219). A mouse monoclonal antibody against ZO-1 was purchased from BD Transduction Laboratories (610966). Mouse monoclonal clone 2D11 and rabbit polyclonal clone H-60 anti-hDlg antibodies were purchased from Santa Cruz Biotechnology. Anti-SRPK1 mouse monoclonal antibody clone 12 was purchased from BD Biosciences and goat polyclonal antibody clone sc11305 was from Santa Cruz Biotechnology. Anti-E6 antibody was obtained from Arbor Vita Corporation, Switzerland and anti-GAPDH (glyceraldehyde-3-phosphate dehydrogenase) antibody was from AMS Biodesign International. Anti-GST (glutathione transferase) antibody was from Sigma. All of these antibodies were used at 1:1000 dilution. Anti-Cx26 and anti-Cx30 antibodies (Zymed) and anti-E-cadherin antibody (BD Transduction Laboratories) were used at 1:500. hDlg clone 2D11 was used at 1:250. HPV16 E2 TVG 261 polyclonal rabbit antibody was used at 1:5000. Quantification of levels of proteins was carried out using Image J and normalized to levels of GAPDH detected on the same membrane. A Student's t test was used to calculate statistical significance.

Plasmid construction

The human Cx43-containing plasmid pcDNA3-Cx43 (kindly provided by Dr Dale Laird, Department of Anatomy and Cell Biology, University of Western Ontario, London, ON, Canada) was used as the template for PCR-mediated site-directed mutagenesis. Primers were used to generate a C-terminal portion of Cx43 corresponding to amino acids 263–382, with or without the last five amino acids (PDZ-binding motif) [17,18]. Both sense (5′-GGGAAAGGGGGATCCATGCAAAAATATGCTTAT-3′) and antisense wild-type (5′-GGGAAAGGGGGATCCCTAGATCTCCAGGTCATCAGGCCG-3′) and antisense Δ5 (5′-GGGAAAGGGGGATCCCTAAGGCCGAGGTCTGCTGCTGGCACG-3′) primers contained a BamHI site (underlined) and a 9 nt flanking sequence. PCR reactions were carried out using the Expand High Fidelity PCR System (Roche) according to the manufacturer's protocol. BamHI-cut PCR fragments were ligated into calf-intestinal phosphatase-treated, BamHI-cut pGEXT2T (GE Healthcare), a GST expression vector. Ligated plasmids were transformed into DH5α Escherichia coli and clones were analysed by restriction digestion. Positive clones were confirmed by DNA sequencing. FLAG–Cx43 was constructed by subcloning the C-terminal portion of Cx43 into the bacterial expression vector pT7-FLAG-1 (Sigma P1118). Specifically, a HindIII/BamHI C-terminal Cx43 fragment was cut from a derivative of the vector p3XFLAGCMV10 (Sigma) which contained the C-terminal portion of Cx43 cloned between the HindIII and BamHI sites. This fragment was then subcloned into the HindIII/BglII site of pT7-FLAG-1. Ligated plasmids were transformed into DH5α E. coli and clones were analysed by restriction digestion. Positive clones were confirmed by DNA sequencing.

GST-protein expression and purification

Plasmids expressing GST-fusion proteins were transformed by electroporation into E. coli BL21. Protein expression was under the control of an IPTG (isopropyl β-D-thiogalactopyranoside) inducible tac promoter. Cultures were grown to a D600 of 0.4–0.6 before being induced for 3 h at 30°C with 1 mM IPTG. Cells were pelleted by centrifugation at 5500 g at 4°C for 5 min. Cell pellets were resuspended in 2 ml of PBS/1% Triton X-100 and suspensions were sonicated for 10×30 s on ice. Cellular debris was removed by centrifugation at 12000 g at 4°C for 15 min and protease and phosphatase inhibitors were added to the supernatant. Proteins were batch-purified using Glutathione Sepharose 4B (GE Healthcare). Briefly, 500 μl of a 75% slurry was added to each sample and then incubated by rotation for 1 h at 4°C to allow GST-fusion-binding to the beads. Samples were washed with 5×10 ml of wash buffer, then beads were pelleted by centrifugation (500 g, 4°C, 5 min). Beads were resuspended in 1 vol. glycerol and stored at −80°C. Purity of the proteins was assessed by SDS/PAGE.

FLAG protein expression and purification

FLAG–Cx43 constructs were transformed into BL21(DE3)-T1R competent E. coli (Sigma). This strain expresses an IPTG-inducible T7 polymerase and allows high-level induction and expression of genes driven by the T7 promoter. Expression was induced with 1 mM IPTG for 3 h at 30°C, as described above. FLAG-tagged protein was captured on anti-FLAG M2 affinity resin (Sigma) and purified by elution with FLAG peptide (Sigma).

His-tagged protein expression and purification

His–HPV16 E2 was expressed in BL21 E. coli and purified as described using His·Bind resin (Novagen) [31].

GST-pull-down and in vitro binding experiments

GST–hDlg preparation and purification was carried out as described previously [32]. GST–Cx43 C-terminal domain was purified from bacterial lysates by a very similar protocol. Protein purity and concentration was assessed using a Protein 200 Lab-Chip kit (Agilent Technologies). Similar amounts of purified protein (determined by Coomassie Blue visualization) were added to 50–100 μg of protein extract in IP buffer and incubated for 2 h with rotation at 4°C. Unbound protein was removed by repeated washing in IP buffer and pellets were resuspended in protein-loading buffer. Samples were boiled at 100°C for 5 min before being resolved by SDS/PAGE. For in vitro binding studies, equal amounts (5 μg) of purified GST–hDlg and His–Cx43 were mixed and rocked in 1 ml of IP buffer for 3 h at 4°C. Beads were washed and samples eluted as above. Interacting proteins were detected by Western blotting using specific antibodies.

Confocal immunofluorescence microscopy

Cell were grown on sterile 16 mm diameter coverslips until 70% confluent, washed three times with PBS and fixed and permeabilized with 100% ice-cold methanol for 5 min at −20°C or with 2% sucrose and 4% formaldehyde for 10 min at room temperature. Methanol fixation generally yielded better images with the antibodies we used. The coverslips were blocked with 5% (v/v) horse serum in PBS for 30 min at room temperature. Cells were incubated with primary antibodies diluted 1:1000 (anti-Cx43; a gift from Dr Edgar Rivedal, Department of Cancer Prevention, Institute for Cancer Research, Oslo University Hospital, Oslo, Norway), 1:500 [anti-MPR (mannose 6-phosphate receptor), Ab 2733, Abcam; anti-Cx26 and anti-Cx30, Zymed; and anti-E-cadherin, Santa Cruz Biotechnology], 1:250 (ZO-1, clone ZMD437; Invitrogen), 1:100 (β-catenin; BD Transduction Laboratories) or 1:25 (hDlg, sc-9961; Santa Cruz Biotechnology) in 5% (v/v) horse serum in PBS for 1 h at room temperature. Coverslips were washed in PBS six times before incubation for 1 h with secondary antibodies (diluted 1:200) conjugated to Alexa Fluor® 488 or Alexa Fluor® 555 (Invitrogen). After washing in PBS six times then once in water, coverslips were dried then mounted with Vectashield hard-set mounting medium [with a DAPI (4′,6-diamidino-2-phenylindole) nuclear stain]. Negative controls (no primary antibody) were included in all experiments. Images were captured using either a Leica SP2-AOBS or Zeiss LSM 510 confocal microscope. For the lysosomal inhibition experiments, W12GPXY cells were either mock-treated or treated with 10 mM NH4Cl for 8 h and then incubated with Lysotracker Red DND-99 (Invitrogen) in pre-warmed medium (100 nM) for 2 h. Cells were fixed (2% sucrose and 4% formaldehyde), permeabilized (10% sucrose and 0.5% NP-40 solution), blocked (5% horse serum) and incubated for 1 h at room temperature with the primary antibodies rabbit anti-(mouse Cx43) and anti-hDlg. Coverslips were incubated for 1 h with the secondary antibodies (diluted 1:500) donkey anti-(mouse Alexa Fluor® 647) and donkey anti-(rabbit Alexa Fluor® 488) (Invitrogen). Coverslips were air-dried then mounted with Vectashield hard-set mounting medium with DAPI (H-1500, Vector Laboratories). Images were captured using a Zeiss LSM 510 confocal microscope with appropriate settings, and colocalization was analysed with Zeiss software.

Cellular Triton X-100 fractionation assay

Cells were washed with ice-cold PBS and extracted in 0.5 ml of ice-cold PBS/1% Triton X-100 containing PhosSTOP phosphatase inhibitor cocktail, Complete™ mini protease inhibitor cocktail and 1 mM PMSF, then left for 1 h on ice with intermittent vortex-mixing before centrifugation at 10000 g for 5 min at 4°C. The supernatant containing Triton-soluble fractions was removed and the Triton X-100-insoluble pellets resuspended in 500 μl of protein-lysis buffer. Equal volumes of fractions were then resolved by gel electrophoresis followed by immunoblotting.

Scrape-loading of cells

Cells were grown to confluence in 60 mm dishes. Cell layers were washed twice in PBS, then 3 ml of 0.05% Lucifer Yellow was added and gentle ‘rolling’ cuts were made with a round-ended scalpel blade. After 4 min at 37°C, cells were washed five times in PBS, fixed in formaldehyde and examined under a fluorescence microscope.

RESULTS

The PDZ domain protein hDlg relocates from the membrane to the cytoplasm of cervical tumour cells in concert with Cx43

We showed previously that Cx43 is detected in both the cytoplasm and in the membrane, in typical gap junction plaques, in non-transformed W12G cervical epithelial cells. In contrast, in the fully transformed W12GPXY epithelial cell line derived from W12G, gap junction plaques were not observed and Cx43 relocated to the cytoplasm, concomitant with a loss of GJIC [12].

Scaffolding proteins have been postulated to take part in Cx targeting to the membrane. Cx43 has been shown to bind the middle of three adjacent PDZ domains in ZO-1 protein [17,18]; Cx45 also interacts with the PDZ domains of ZO-1 [21,22] and Cx32 interacts with the C-terminal SH3 domain of hDlg, which has also been identified as a potential binding partner for Cx43 [33].

We used confocal microscopy to investigate whether hDlg or ZO-1 scaffolding proteins had any relationship with Cx43 relocalization in cervical epithelial cells. hDlg was mainly identified on the cell membrane and the cell margin in untransformed W12G cells, partially co-localizing with Cx43 (Figure 1A). Cx43 was located in discrete regions of the plasma membrane with a morphology indicative of gap junctions. In addition, some cytoplasmic Cx43 and hDlg was observed. In contrast, Cx43 was consistently located in the cytoplasm of transformed W12GPXY cells and little cell membrane staining was found (Figure 1B), as we observed previously [12]. Most hDlg was also relocated to the cytoplasm of W12GPXY cells and displayed a speckled pattern of staining. Moreover, in many W12GPXY cells both hDlg and Cx43 were found in a perinuclear location and some Cx43 and hDlg co-localized in the cytoplasm (Figure 1B, white arrowheads). A similar pattern of Cx43 and hDlg cytoplasmic location was observed in CaSki cells, another cervical tumour cell line (results not shown). Next we examined the location of ZO-1 in the two W12 cell lines as a positive control for a PDZ domain protein known to bind Cx43. ZO-1 was more difficult to detect than hDlg in W12G cells, but it displayed some cell margin/membrane (white arrowhead) and some cytoplasmic staining (Figure 1C). In the W12GPXY cells, ZO-1 gave a similar pattern of staining. In contrast with the staining pattern observed with hDlg, there was little evidence of perinuclear staining with the anti-ZO-1 antibody (Figure 1D).

ZO-1, hDlg and Cx43 are present on the plasma membrane of W12G immortal cervical cells, but are found in the cytoplasm of W12GPXY transformed cervical epithelial cells

Figure 1
ZO-1, hDlg and Cx43 are present on the plasma membrane of W12G immortal cervical cells, but are found in the cytoplasm of W12GPXY transformed cervical epithelial cells

Confocal immunofluorescence microscopy showing the location of Cx43 and hDlg in (A) W12G cells (immortal but not transformed cervical epithelial cells) compared with (B) W12GPXY cells (transformed cervical epithelial cells). W12GPXY cells grew in small colonies. The image shows some cells in close contact and some isolated cells. White arrowheads indicate examples of co-localization of Cx43 and hDlg. Green, Cx43; red, hDlg. (C) Confocal immunofluorescence microscopy showing the location of ZO-1 and Cx43 in W12G cells. (D) Confocal immunofluorescence microscopy showing the location of Cx43 and ZO-1 in W12GPXY cells. White, Cx43; green, ZO-1. In all cases, cells were counterstained with DAPI (blue). Scale bar, 10 μm.

Figure 1
ZO-1, hDlg and Cx43 are present on the plasma membrane of W12G immortal cervical cells, but are found in the cytoplasm of W12GPXY transformed cervical epithelial cells

Confocal immunofluorescence microscopy showing the location of Cx43 and hDlg in (A) W12G cells (immortal but not transformed cervical epithelial cells) compared with (B) W12GPXY cells (transformed cervical epithelial cells). W12GPXY cells grew in small colonies. The image shows some cells in close contact and some isolated cells. White arrowheads indicate examples of co-localization of Cx43 and hDlg. Green, Cx43; red, hDlg. (C) Confocal immunofluorescence microscopy showing the location of ZO-1 and Cx43 in W12G cells. (D) Confocal immunofluorescence microscopy showing the location of Cx43 and ZO-1 in W12GPXY cells. White, Cx43; green, ZO-1. In all cases, cells were counterstained with DAPI (blue). Scale bar, 10 μm.

To confirm these data we carried out biochemical cell fractionation using Triton X-100 extraction [34]. In this approach, whole cells are transferred directly into Triton X-100 extraction buffer, and soluble and insoluble fractions are obtained. Therefore, because there is no pre-extraction input control, we used as ‘input’ whole-cell lysates in SDS/PAGE loading buffer from populations of cells grown at the same time and under the same conditions as those from which Triton X-100 extracts were made. Figure 2(A) demonstrates input levels of the various proteins detected in the Triton X-100 fractionation experiment in Figure 2(B). We found signifcantly lower levels of Cx43 in W12GPXY cells than in W12G cells. Figure 2(C) shows quantification of levels of Cx43 and the PDZ proteins hDlg and ZO-1 in both cell lines. In contrast with Cx43, levels of hDlg and ZO-1 were similar and higher respectively in the transformed cell line. Although ZO-1 was difficult to detect by Western blotting (Figures 2A and 2B), it was clearly present in W12G cells as shown by immunofluorescence in Figure 1(C). Figure 2(B) shows that in untransformed W12G cells, Cx43 was extracted in both the soluble and insoluble fractions, whereas in fully transformed W12GPXY cells, Cx43 was detected only at a very low level in the insoluble fraction and Cx43 levels were reduced overall (soluble plus insoluble fractions) compared with W12G cells, reflecting the 40% reduction in Cx43 level in W12GPXY cells compared with W12G cells (Figure 2C). The Cx43 blot was subject to longer exposure than the other blots in Figure 2(B) in order to visualize any bands in the insoluble fraction from W12GPXY cells. hDlg was found mainly in the Triton X-100-soluble fraction in both W12G and W12GPXY cells. In contrast, ZO-1 was found in the Triton X-100-insoluble fraction of both cell clones although levels were much lower in W12G cells (Figure 2C). ZO-1 displayed a cytoplasmic staining in W12GPXY cells that might indicate removal from the plasma membrane so the retention of this protein in insoluble fraction of W12GPXY cells indicates that it must reside in cytoskeletal or membranous compartments of the cytoplasm [35]. E-cadherin was extracted in both the soluble and insoluble compartments [36], whereas GAPDH was also found mainly in the soluble fractions as expected.

Triton X-100 cell extraction confirms Cx43 relocation in W12GPXY cells

Figure 2
Triton X-100 cell extraction confirms Cx43 relocation in W12GPXY cells

(A) Western blot analysis of levels of various membrane-associated proteins in W12G and W12GPXY cells. The extracts used in the present study were isolated from the same batch of cells grown at the same times as the extracts used in (B). Molecular mass in kDa is shown on the left-hand side. (B) Fractionation of Cx43, hDlg, ZO-1, GAPDH and E-cadherin into Triton X-100-soluble (S) and -insoluble (I) fractions of W12G and W12GPXY cells. Cx43 is lost from the Triton X-100-insoluble fraction in W12GPXY cells compared with W12G cells. Several isoforms of hDlg can be observed as expected [64]. (C) Quantification of relative total levels of hDlg, ZO-1, Cx43 and GAPDH protein in W12G cells (set at 100%) compared with W12GPXY cells. Results are means±S.D. for three independent experiments. **P<0.005 (Student's t test).

Figure 2
Triton X-100 cell extraction confirms Cx43 relocation in W12GPXY cells

(A) Western blot analysis of levels of various membrane-associated proteins in W12G and W12GPXY cells. The extracts used in the present study were isolated from the same batch of cells grown at the same times as the extracts used in (B). Molecular mass in kDa is shown on the left-hand side. (B) Fractionation of Cx43, hDlg, ZO-1, GAPDH and E-cadherin into Triton X-100-soluble (S) and -insoluble (I) fractions of W12G and W12GPXY cells. Cx43 is lost from the Triton X-100-insoluble fraction in W12GPXY cells compared with W12G cells. Several isoforms of hDlg can be observed as expected [64]. (C) Quantification of relative total levels of hDlg, ZO-1, Cx43 and GAPDH protein in W12G cells (set at 100%) compared with W12GPXY cells. Results are means±S.D. for three independent experiments. **P<0.005 (Student's t test).

For both the biochemical fractionation and immunofluorescence experiments, each W12 cell line was grown both in F-medium and in KGM. Very similar data were obtained regardless of the growth conditions used for the W12 cell clones. In case the relocation of Cx43 in W12GPXY cells was due to a global down-regulation of gap junction formation, we also examined localization of Cx26 and Cx30 in W12G and W12GPXY cells. Figure 3(A) shows the distribution of Cx26 in W12G cells. Not every cell expressed Cx26, but where Cx26 was present it was found at the junctions between adjacent cells (white arrowhead). Similarly, Cx30 was found on the membrane of some W12G cells (Figure 3C, white arrowhead). For both of these Cxs in the fully transformed W12GPXY cells, we observed increased cytoplasmic and membrane staining with relatively large membrane puncta compared with W12G cells (Figures 3B and 3D). Although the pattern of membrane staining with Cx26 and Cx30 indicated possible formation of gap junction plaques, these were not functional, at least as observed with Lucifer Yellow (443 Da) dye injection [12] and scrape loading. Figure 3(E) shows that the dye was extensively spread into surrounding cells upon scrape loading of W12G cells, but there was only limited dye transfer when W12GPXY cells were used (Figure 3F).

Cx relocation to the cytoplasm of W12GPXY cervical epithelial cells is specific for Cx43

Figure 3
Cx relocation to the cytoplasm of W12GPXY cervical epithelial cells is specific for Cx43

W12G and W12GPXY cells were stained with antibodies against two other Cx proteins, Cx26 (A and B) and Cx30 (C and D). Confocal microscopy reveals membrane staining with each of these antibodies in W12G and W12GPXY cells. Cells stained with anti-Cx antibodies (green) were co-stained with an anti-hDlg antibody (red). Cells were counterstained with DAPI (blue). White arrowheads show membrane Cx26 and Cx30. Scale bar, 10 μm. Large merged images of a single cell in each multiple cell image (white arrows indicate the chosen cells) are shown on the right-hand side. W12G cells can undergo Lucifer Yellow dye transfer following scrape loading (E), whereas W12GPXY cells do not transfer the dye efficiently (F), showing that W12GPXY cells do not possess good gap junctional communication for molecules larger than 443 Da.

Figure 3
Cx relocation to the cytoplasm of W12GPXY cervical epithelial cells is specific for Cx43

W12G and W12GPXY cells were stained with antibodies against two other Cx proteins, Cx26 (A and B) and Cx30 (C and D). Confocal microscopy reveals membrane staining with each of these antibodies in W12G and W12GPXY cells. Cells stained with anti-Cx antibodies (green) were co-stained with an anti-hDlg antibody (red). Cells were counterstained with DAPI (blue). White arrowheads show membrane Cx26 and Cx30. Scale bar, 10 μm. Large merged images of a single cell in each multiple cell image (white arrows indicate the chosen cells) are shown on the right-hand side. W12G cells can undergo Lucifer Yellow dye transfer following scrape loading (E), whereas W12GPXY cells do not transfer the dye efficiently (F), showing that W12GPXY cells do not possess good gap junctional communication for molecules larger than 443 Da.

In case there was some disruption of the plasma membrane, perhaps due to the tumorigenic phenotype of W12GPXY cells that could cause specific relocation of Cx43 we examined the location of E-cadherin and β-catenin (Figure 4). Both proteins were observed at the cell periphery in both W12G and W12GPXY cells although cell–cell contacts were not as smooth in W12GPXY cells compared with W12G cells (see β-catenin staining pattern).

Cx43 relocation is not due to disruption of the W12GPXY cell membrane

Figure 4
Cx43 relocation is not due to disruption of the W12GPXY cell membrane

Staining of W12G (A) and W12GPXY (B) with antibodies against two membrane proteins, E-cadherin and β-catenin. Cells stained with E-cadherin (red) were co-stained with an antibody against Cx43 (green). β-Catenin is shown with green staining. Cells were counterstained with DAPI (blue). Scale bar, 10 μm.

Figure 4
Cx43 relocation is not due to disruption of the W12GPXY cell membrane

Staining of W12G (A) and W12GPXY (B) with antibodies against two membrane proteins, E-cadherin and β-catenin. Cells stained with E-cadherin (red) were co-stained with an antibody against Cx43 (green). β-Catenin is shown with green staining. Cells were counterstained with DAPI (blue). Scale bar, 10 μm.

Interaction of Cx43 with hDlg is detected in vivo in W12 cervical epithelial cells

To investigate whether relocation of hDlg with Cx43 into the cytoplasm might represent a direct protein–protein interaction we first performed a series of co-IP experiments. Because the molecular mass of Cx43 at 43 kDa is close to that of the antibody heavy chain fragment we carried out the experiment using the CleanBlot system (Thermo Scientific) that detects only the antibody used in Western blotting and does not detect heavy and light chain antibody fragments. Figure 5(A) shows precipitation of Cx43 specifically with hDlg in W12G (lane 3) and W12GPXY cells (lane 6). Anti-SRPK1 antibody (lanes 2 and 5) was used as an additional non-specific control to beads alone (lanes 1 and 4). hDlg and SRPK1 were capable of precipitating themselves in the reaction (Figures 5B and 5C). Figure 5(D) shows GAPDH probing of the supernatants from the co-IP experiment as an estimate of input. The results of a standard co-IP and Western blotting protocol are shown in Supplementary Figure S1 at http://www.BiochemJ.org/bj/446/bj4460009add.htm.

Direct interation of Cx43 and hDlg in cervical epithelial cells

Figure 5
Direct interation of Cx43 and hDlg in cervical epithelial cells

(A) Co-IP of endogenous Cx43 from W12G and W12GPXY cervical epithelial cells using antibodies to proteins named above the blots. beads, beads without added antibody used in co-IP. SRPK1, control for non-specific antibody binding: goat antibody against SRPK1 used in coIP. hDlg, co-IP with rabbit anti-hDlg antibody. A CleanBlot detection system (Thermo Scientific) was used to avoid visualization of contaminating antibody fragments from these blots. (B) The upper half of the blot in (A) was probed with an anti-hDlg antibody to show IP of hDlg with its own antibody. (C) A very similar blot with the same protein samples was probed with an anti-SRPK1 mouse monoclonal antibody to reveal IP of SRPK1. (D) A one-tenth volume of supernatants from the co-IP experiment was electrophoresed and Western blotted with an anti-GAPDH antibody to give an estimate of protein loading. (E) GST-pull-down of purified bacterially expressed N-terminal-FLAG-tagged C-terminus (CT) of Cx43. GST–hDlg, purified, bacterially expressed, GST–hDlg fusion protein (lanes 3 and 4) or GST–TopBP1 (lanes 1 and 2) used in pull-down of FLAG-tagged C-terminal Cx43 and probed with an anti-FLAG antibody. (F) A second Western blot of the same reaction products probed with anti-Cx43 antibody. (G) The blot in (E) stripped and reprobed with anti-GST antibody. (H) GST pull-down of GST–hDlg or GST–TopBP1 with His-tagged HPV16 E2 protein and probed with HPV16 E2 antibody. Molecular mass in kDa is shown on the left-hand side. Ab, antibody; I, FLAG-tagged protein purification from induced bacterial cell lysate; U, FLAG-tagged protein purification from uninduced bacterial cell lysate.

Figure 5
Direct interation of Cx43 and hDlg in cervical epithelial cells

(A) Co-IP of endogenous Cx43 from W12G and W12GPXY cervical epithelial cells using antibodies to proteins named above the blots. beads, beads without added antibody used in co-IP. SRPK1, control for non-specific antibody binding: goat antibody against SRPK1 used in coIP. hDlg, co-IP with rabbit anti-hDlg antibody. A CleanBlot detection system (Thermo Scientific) was used to avoid visualization of contaminating antibody fragments from these blots. (B) The upper half of the blot in (A) was probed with an anti-hDlg antibody to show IP of hDlg with its own antibody. (C) A very similar blot with the same protein samples was probed with an anti-SRPK1 mouse monoclonal antibody to reveal IP of SRPK1. (D) A one-tenth volume of supernatants from the co-IP experiment was electrophoresed and Western blotted with an anti-GAPDH antibody to give an estimate of protein loading. (E) GST-pull-down of purified bacterially expressed N-terminal-FLAG-tagged C-terminus (CT) of Cx43. GST–hDlg, purified, bacterially expressed, GST–hDlg fusion protein (lanes 3 and 4) or GST–TopBP1 (lanes 1 and 2) used in pull-down of FLAG-tagged C-terminal Cx43 and probed with an anti-FLAG antibody. (F) A second Western blot of the same reaction products probed with anti-Cx43 antibody. (G) The blot in (E) stripped and reprobed with anti-GST antibody. (H) GST pull-down of GST–hDlg or GST–TopBP1 with His-tagged HPV16 E2 protein and probed with HPV16 E2 antibody. Molecular mass in kDa is shown on the left-hand side. Ab, antibody; I, FLAG-tagged protein purification from induced bacterial cell lysate; U, FLAG-tagged protein purification from uninduced bacterial cell lysate.

Cx43 cytoplasmic tail binds hDlg directly in vitro

Having shown that hDlg and Cx43 can exist as part of a complex in epithelial cells, we next wanted to determine whether there was any potential direct interaction between the two proteins. We focused upon the C-terminal tail of Cx43, because this domain is cytoplasmic and appears to mediate regulatory protein–protein interactions [4]. To do this, we conducted in vitro binding experiments between bacterially expressed and purified proteins (Supplementary Figure S2 at http://www.BiochemJ.org/bj/446/bj4460009add.htm). GST–hDlg was incubated with a Cx43 C-terminal domain carrying an N-terminal FLAG-tag (FLAG–Cx43CT) purified from uninduced or induced bacterial lysates. As a control, we incubated FLAG–Cx43CT with an irrelevant GST-tagged protein, GST–TopBP1 (topoisomerase binding protein 1). Bound proteins were eluted and fractionated by SDS/PAGE followed by Western blotting. GST–hDlg (Figures 5E and 5F, lane 4), but not GST–TopBP1 (Figure 5E, lane 2), pulled down the C-terminal domain of Cx43. With either GST–hDlg or GST–TopBP1, no protein was precipitated using uninduced bacterial FLAG lysate (Figure 5E, lanes 1 and 3). Similar amounts of GST proteins were used in each reaction (Figure 5G). GST–TopBP1 was capable of protein–protein interaction, because it was able to precipitate the HPV16 E2 protein (Figure 5H), a known TopBP1 interactor [37].

Cx43 cytoplasmic tail binds the N- and C-terminal domains, but not the PDZ domains of hDlg

ZO-1 has many structural and functional similarities with hDlg (both proteins are in the same MAGUK protein subgroup) [17]. ZO-1 binds Cx43 through its second PDZ domain, so hDlg might be expected to bind Cx43 in a similar manner. However, another Cx, Cx32, has been shown to interact with the C-terminal SH3/HOOK domain of hDlg [24]. To determine which domain of hDlg bound Cx43, GST-fusion proteins of the N-terminus (NT: amino acids 1–122), the N-terminus plus one PDZ domain (NT+1PDZ), the three PDZ domains alone (3PDZ) and C-terminal (CT) domains (SH, HOOK and GUK domains) were prepared (Supplementary Figure S2). GST-pull-down experiments in W12GPXY cells indicated that, although full-length hDlg precipitated a large proportion of endogenous Cx43 from W12GPXY cell extracts (Figure 6B, lane 3), the isolated PDZ domains could only interact very inefficiently (Figure 6B, lane 6). We detected weak binding to the C-terminal domain (Figure 6B, lane 7). However, binding was clearly observed with the hDlg N-terminal domain (Figure 6B, lanes 4 and 5). The upper bands in some of the tracks are non-specific proteins that interact with GST and react with the Cx43 polyclonal antibody. To control for binding to the hDlg-3PDZ region, we tested interaction with a known PDZ binder, HPV16 E6, which is expressed in W12GPXY cells. E6 was precipitated as efficiently with full-length hDlg (Figure 6C, lane 3) as with the 3PDZ domains of hDlg (Figure 6C, lane 6) but not with the N- or C-termini as expected. To ensure that the C-terminal domain of hDlg that we prepared was capable of interacting with another protein, we tested binding of Cx32, previously reported to interact with this domain [24]. Figure 6(D) shows that Cx32 was able to bind full-length hDlg (Figure 6D, lane 3) and the hDlg C-terminal domain (Figure 6D, lanes 4 and 6), but not the 3PDZ domain region (Figure 6D, lane 5). The upper band in lane 3 that interacts with anti-Cx32 antibody is a contaminating cross-reacting protein from the bacterial lysate.

GST-pull-down analysis of regions of interaction between hDlg and Cx43

Figure 6
GST-pull-down analysis of regions of interaction between hDlg and Cx43

(A) Diagram of the modular structure of hDlg showing the N-terminal domain (hatched box), three PDZ domains (light grey and numbered 1, 2 or 3) stretching from amino acids (aa) 220 to 545 and the C-terminal (C-term) SH, HOOK and GUK domains (striped, dark grey and stippled boxes respectively) stretching from amino acid 560 to the end of the molecule at amino acid 911. (B) GST-pull-down of Cx43 from W12GPXY cell extracts. Lane 1, input, one-fifth volume of lysate used in GST-pull-downs. Lane 2, GST alone, negative control. Lane 3, GST–hDlg fusion protein. Lane 4, GST–hDlg N-terminal domain (NT) (amino acids 1–122). Lane 5, GST–hDlg N-terminal domain plus the first PDZ domain (NT+1PDZ). Lane 6, GST–hDlg-3PDZ, amino acids 220–545 containing the three PDZ domains. Lane 7, GST–hDlg-CT, amino acids 560–911 covering the C-terminal domain of hDlg. The blot was probed with anti-Cx43 antibody. Cross-reactive antibody bands were observed above the Cx43 band in lanes 1, 2, 6 and 7. (C) A very similar Western blot of the same samples was probed with an anti-HPV16 E6 antibody. (D) GST-pull-down of Cx32 from W12GPXY cell extracts. Lane 1, input, one-fifth volume of lysate used in GST-pull-downs. Lane 2, GST alone, negative control. Lane 3, GST–hDlg fusion protein purified from bacterial cell lysates. The lower band indicated with a white asterisk is Cx32. Lanes 1–6 were from the same gel. The blot has been cut (white line) to remove a second GST–hDlg fusion protein purified from bacterial cell lysates, but not used in the experiments in lanes 4–6. Lane 4, GST–hDlg-CT, amino acids 560–911 covering the C-terminal domain of hDlg. Lane 5, GST–hDlg-3PDZ, amino acids 220–545 containing the three PDZ domains. Lane 6, GST–hDlg-CT+3PDZ domains. Cross-reacting non-specific bands of higher molecular mass than 32 kDa can be observed in lanes 1, 3, 4 and 6. (E) GST-pull-down of hDlg from W12GPXY cells using GST–C-terminal (CT) Cx43 or GST–C-terminal Cx43 with a deletion of the final five amino acids (Cx43Δ5). (F) GST-pull-down of ZO-1 from W12GPXY cells using GST-C-terminal (CT) Cx43 or GST-C-terminal (CT) Cx43 with a deletion of the final five amino acids (Cx43Δ5). Lane 1, input, one-tenth volume of lysate used in GST-pull-downs. Lane 2, GST-pull-down using GST alone as a negative control. Lane 3, GST–CT-Cx43, N-terminal GST-tagged C-terminal domain of Cx43 used to pull down ZO-1- and hDlg-containing complexes. Lane 4, GST–CT-Cx43Δ5, N-terminal GST-tagged C-terminal domain of Cx43 with the last five amino acids deleted used to pull down ZO-1- and hDlg-containing complexes. Molecular mass in kDa is shown on the left-hand side in (BE).

Figure 6
GST-pull-down analysis of regions of interaction between hDlg and Cx43

(A) Diagram of the modular structure of hDlg showing the N-terminal domain (hatched box), three PDZ domains (light grey and numbered 1, 2 or 3) stretching from amino acids (aa) 220 to 545 and the C-terminal (C-term) SH, HOOK and GUK domains (striped, dark grey and stippled boxes respectively) stretching from amino acid 560 to the end of the molecule at amino acid 911. (B) GST-pull-down of Cx43 from W12GPXY cell extracts. Lane 1, input, one-fifth volume of lysate used in GST-pull-downs. Lane 2, GST alone, negative control. Lane 3, GST–hDlg fusion protein. Lane 4, GST–hDlg N-terminal domain (NT) (amino acids 1–122). Lane 5, GST–hDlg N-terminal domain plus the first PDZ domain (NT+1PDZ). Lane 6, GST–hDlg-3PDZ, amino acids 220–545 containing the three PDZ domains. Lane 7, GST–hDlg-CT, amino acids 560–911 covering the C-terminal domain of hDlg. The blot was probed with anti-Cx43 antibody. Cross-reactive antibody bands were observed above the Cx43 band in lanes 1, 2, 6 and 7. (C) A very similar Western blot of the same samples was probed with an anti-HPV16 E6 antibody. (D) GST-pull-down of Cx32 from W12GPXY cell extracts. Lane 1, input, one-fifth volume of lysate used in GST-pull-downs. Lane 2, GST alone, negative control. Lane 3, GST–hDlg fusion protein purified from bacterial cell lysates. The lower band indicated with a white asterisk is Cx32. Lanes 1–6 were from the same gel. The blot has been cut (white line) to remove a second GST–hDlg fusion protein purified from bacterial cell lysates, but not used in the experiments in lanes 4–6. Lane 4, GST–hDlg-CT, amino acids 560–911 covering the C-terminal domain of hDlg. Lane 5, GST–hDlg-3PDZ, amino acids 220–545 containing the three PDZ domains. Lane 6, GST–hDlg-CT+3PDZ domains. Cross-reacting non-specific bands of higher molecular mass than 32 kDa can be observed in lanes 1, 3, 4 and 6. (E) GST-pull-down of hDlg from W12GPXY cells using GST–C-terminal (CT) Cx43 or GST–C-terminal Cx43 with a deletion of the final five amino acids (Cx43Δ5). (F) GST-pull-down of ZO-1 from W12GPXY cells using GST-C-terminal (CT) Cx43 or GST-C-terminal (CT) Cx43 with a deletion of the final five amino acids (Cx43Δ5). Lane 1, input, one-tenth volume of lysate used in GST-pull-downs. Lane 2, GST-pull-down using GST alone as a negative control. Lane 3, GST–CT-Cx43, N-terminal GST-tagged C-terminal domain of Cx43 used to pull down ZO-1- and hDlg-containing complexes. Lane 4, GST–CT-Cx43Δ5, N-terminal GST-tagged C-terminal domain of Cx43 with the last five amino acids deleted used to pull down ZO-1- and hDlg-containing complexes. Molecular mass in kDa is shown on the left-hand side in (BE).

In the above experiments, the C-terminal domain of Cx43 was used to demonstrate interaction with hDlg. The extreme C-terminal sequence DLEI can bind the second PDZ domain of ZO-1 [17]. Therefore we tested whether this motif was required for binding of Cx43 to hDlg in GST-pull-down assays. Figure 6(E) shows that, in W12GPXY cells, wild-type C-terminal Cx43 bound efficiently to full-length hDlg (Figure 6E, lane 3). Binding between hDlg and Cx43 was still observed when the last five amino acids of Cx43, containing the putative PDZ-binding domain were deleted (Figure 6E, lane 4). In contrast, the same deletion abrogated interaction of Cx43 with ZO-1 in the same experiment (Figure 6F, lane 4).

Cx43 and hDlg are located in the endosomal/lysosomal compartment in W12GPXY cells

The distribution pattern of Cx43 and hDlg in W12GPXY cells suggested a possible location in the ER (endoplasmic reticulum)/Golgi where newly synthesized or aberrantly folded or mutated Cxs can be located [38]. However, co-staining with a Golgi marker (GM 130) revealed that most Cx43 and hDlg was not in the Golgi (results not shown). Cxs can be removed from the plasma membrane by endocytosis then finally degraded by the endo-lysosomes [39]. The perinuclear location of Cx43 and hDlg in the cytoplasm of W12GPXY cells [12] could also correspond to endosomal/lysosomal compartments. To test this hypothesis, W12GPXY cells were stained with the anti-EEA (early endosome antigen) and anti-MPR antibodies that detect early and late endosomes respectively. Cx43 and hDlg showed partial co-localization with both endosomal markers, in a perinuclear location (results not shown). To test whether Cx43 or hDlg co-localized in the endosomal/lysosomal compartment for degradation, W12GPXY cells were mock-treated or treated with the lysosomal inhibitor NH4Cl at 10 mM for 8 h. Cells were then stained with Cx43 (green) and hDlg (magenta) and location was examined in relation to Lysotracker Red (red). In mock-treated cells, Cx43 and hDlg located to the cytoplasm with a proportion in a concentrated perinuclear location where some of the proteins co-localized with the lysosomes (Figure 7A, rightmost two panels, and Supplementary Figure S3 at http://www.BiochemJ.org/bj/446/bj4460009add.htm). Upon NH4Cl treatment, cells were no longer stained with Lysotracker Red as expected and hDlg moved from a perinuclear location to a more diffuse cytoplasmic location with some located to the cell margin (Figure 7B and Supplementary Figure S3). Although some Cx43 remained perinuclear in NH4Cl-treated cells, more Cx43 was located throughout the cytoplasm compared with untreated cells. However there was no apparent trafficking to the membrane. Some hDlg and Cx43 co-localized in a diffuse pattern in the cytoplasm (Figure 7B, rightmost panel). To test whether Cx43 or hDlg were targeted for lysosomal degradation, protein extracts isolated from W12G and W12GPXY cells mock-treated or NH4Cl-treated were fractionated by SDS/PAGE and probed with antibodies against hDlg and Cx43 (Figure 7C). Quantification of three separate experiments revealed that NH4Cl treatment had little effect on the levels of either protein in W12G cells (Figures 7D and 7E). In contrast, levels of both hDlg and Cx43 increased significantly (P<0.05) in the presence of NH4Cl in W12GPXY cells. Experiments were repeated with chloroquine (200 μM) as the lysosomal inhibitor with very similar results (not shown).

hDlg and Cx43 relocate to the endosomal/lysosomal compartments in W12GPXY cells

Figure 7
hDlg and Cx43 relocate to the endosomal/lysosomal compartments in W12GPXY cells

Confocal immunofluorescence microscopy of the location of Cx43 (green) and hDlg (magenta) in W12GPXY cells co-staining using Lysotracker Red to detect the lysosomal compartment (red). Cells were either mock-treated (A) or treated (B) with 10 mM NH4Cl for 8 h to inhibit lysosomal degradation. Cells were counterstained with DAPI. The rightmost panels in (A) and (B) show colocalization of Cx43 and hDlg calculated using Zeiss confocal software. Scale bar, 10 μm. (C) Western blot analysis of hDlg, Cx43 and GAPDH controls in W12G and W12GPXY cells either mock-treated (−) or treated (+) with 10 mM NH4Cl as above. Molecular mass in kDa is shown on the left-hand side. (D) Quantification of Cx43 levels upon NH4Cl treatment. (E) Quantification of hDlg levels upon NH4Cl treatment. Results in (D and E) are means±S.D. for three independent experiments. *P<0.05 (Student's t test). Cntrl, control.

Figure 7
hDlg and Cx43 relocate to the endosomal/lysosomal compartments in W12GPXY cells

Confocal immunofluorescence microscopy of the location of Cx43 (green) and hDlg (magenta) in W12GPXY cells co-staining using Lysotracker Red to detect the lysosomal compartment (red). Cells were either mock-treated (A) or treated (B) with 10 mM NH4Cl for 8 h to inhibit lysosomal degradation. Cells were counterstained with DAPI. The rightmost panels in (A) and (B) show colocalization of Cx43 and hDlg calculated using Zeiss confocal software. Scale bar, 10 μm. (C) Western blot analysis of hDlg, Cx43 and GAPDH controls in W12G and W12GPXY cells either mock-treated (−) or treated (+) with 10 mM NH4Cl as above. Molecular mass in kDa is shown on the left-hand side. (D) Quantification of Cx43 levels upon NH4Cl treatment. (E) Quantification of hDlg levels upon NH4Cl treatment. Results in (D and E) are means±S.D. for three independent experiments. *P<0.05 (Student's t test). Cntrl, control.

To investigate a functional impact of hDlg on cytoplasmic Cx43 levels, W12GPXY cells were treated with siRNA against hDlg. If hDlg was necessary for delivery of Cx43 to the endosomal/lysosomal degradation pathway, loss of hDlg might result in increased levels of Cx43. Effective knockdown of hDlg was obtained in every experiment (Figure 8A and Supplementary Figure S4 at http://www.BiochemJ.org/bj/446/bj4460009add.htm). Levels of the control proteins E-cadherin and GAPDH were unaffected by a reduction in hDlg levels (Figure 8B). However, no increase in Cx43 was detected. In fact, Cx43 was consistently present at reduced levels of approximately 50% of the control (control siRNA) in W12GPXY cells treated with siRNA against hDlg (Figures 8A and 8B). These results suggest that hDlg is required to maintain a part of the cytoplasmic pool of Cx43 in the cervical tumour cells. Next, to investigate a possible contribution of hDlg in lysosomal targeting of Cx43, levels of Cx43 were determined in cells where hDlg levels were reduced by siRNA treatment and where lysosomal degradation was inhibited. If hDlg were involved in antagonizing Cx43 lysosomal trafficking then lysosomal inhibition might overcome the loss of Cx43 caused by hDlg knockdown. As already shown in Figure 7(D), upon lysosomal inhibition with NH4Cl (Figure 8C, lanes 4—6) levels of Cx43 in W12GPXY cells were always higher compared with mock-treated cells (Figure 8C, lanes 1—3) confirming lysosomal targeting of Cx43 in W12GPXY. However, although hDlg knockdown in the presence of NH4Cl was still able to reduce levels of Cx43 (compare lanes 5 and 6) quantification of three separate experiments (Figure 8D) revealed that less depletion of Cx43 (31% reduction compared with NH4Cl control, P=0.056) occurred than in mock-treated cells (51% reduction compared with mock control, P=0.01). As controls, E-cadherin and GAPDH levels were unaffected by hDlg knockdown and drug treatment (Figure 8C). Finally, we investigated whether hDlg had any role in membrane trafficking of Cx43. Confocal microscopy revealed that siRNA-mediated reduction in hDlg levels did not result in appearance of membrane Cx43 (Figure 8E). These results indicate that hDlg can maintain part of the cytoplasmic pool of Cx43, possibly in opposition to lysosomal degradation.

hDlg is required to maintain a cytoplasmic pool of Cx43

Figure 8
hDlg is required to maintain a cytoplasmic pool of Cx43

(A) Western blot analysis of levels of Cx43 in W12GPXY cells mock transfected (lane 1), transfected with a control scrambled hDlg siRNA (lane 2) or transfected with hDlg siRNA (lane 3). Levels of E-cadherin and GAPDH are shown as control proteins not affected by hDlg siRNA knockdown. (B) Quantification of levels of Cx43 relative to levels of GAPDH in the hDlg knockdown experiments. Results are mean±S.D.for three independent experiments. *P<0.05 (Student's t test). (C) Western blot analysis of levels of Cx43 with E-cadherin and GAPDH as controls, in W12GPXY cells mock-transfected (lanes 1 and 4), transfected with a control scrambled hDlg siRNA (control siRNA; lanes 2 and 5) or transfected with hDlg siRNA (lanes 3 and 6). At 48 h after transfection, cells were mock-treated (lanes 1–3) or treated with 10 mM NH4Cl for 8 h (lanes 4–6) before harvesting for protein isolation. (D) Quantification of the results in (C) showing means±S.D. for three independent experiments. *P<0.05 (Student's t test). (E) Confocal immunofluorescence microscopy of W12GPXY cells showing the location of Cx43 (green) upon treatment for 48 h with control siRNA or treatment with siRNA against hDlg. Nuclei are stained with DAPI (blue). Scale bar, 10 μm.

Figure 8
hDlg is required to maintain a cytoplasmic pool of Cx43

(A) Western blot analysis of levels of Cx43 in W12GPXY cells mock transfected (lane 1), transfected with a control scrambled hDlg siRNA (lane 2) or transfected with hDlg siRNA (lane 3). Levels of E-cadherin and GAPDH are shown as control proteins not affected by hDlg siRNA knockdown. (B) Quantification of levels of Cx43 relative to levels of GAPDH in the hDlg knockdown experiments. Results are mean±S.D.for three independent experiments. *P<0.05 (Student's t test). (C) Western blot analysis of levels of Cx43 with E-cadherin and GAPDH as controls, in W12GPXY cells mock-transfected (lanes 1 and 4), transfected with a control scrambled hDlg siRNA (control siRNA; lanes 2 and 5) or transfected with hDlg siRNA (lanes 3 and 6). At 48 h after transfection, cells were mock-treated (lanes 1–3) or treated with 10 mM NH4Cl for 8 h (lanes 4–6) before harvesting for protein isolation. (D) Quantification of the results in (C) showing means±S.D. for three independent experiments. *P<0.05 (Student's t test). (E) Confocal immunofluorescence microscopy of W12GPXY cells showing the location of Cx43 (green) upon treatment for 48 h with control siRNA or treatment with siRNA against hDlg. Nuclei are stained with DAPI (blue). Scale bar, 10 μm.

DISCUSSION

The disorganization of epithelial junctions can lead to defective cell–cell adhesion, loss of cell polarity and unregulated cell proliferation, and therefore may represent a crucial step in tumorigenesis. Cx43 is a common unit of stratified epithelial cell gap junctions and, in keratinocytes in culture, its loss from the plasma membrane leads to loss of GJIC [12,40]. Our studies demonstrate a relocation of Cx43 from the plasma membrane to the cytoplasm upon tumorigenic transformation of W12GPXY cervical epithelial cells. The relocation of Cx43 did not appear to be due to a significant loss of cell–cell contact, because other proteins were maintained at the cell membrane. One important factor to be considered in our experiments was the phenotype of the two cervical epithelial cell lines. W12G cells are immortal epithelial cells representative of low grade dysplastic cervical epithelial cells. Because they are not transformed they are still capable of some differentiation. The growth conditions we used in our W12G experiments generated a mixed population of cells both undifferentiated and differentiating [41]. We used this cell population because we and others have shown previously that Cx43 is most highly expressed in normal and dysplastic differentiating epithelial cells [42,43]. On the other hand, W12GPXY cells are transformed and unable to differentiate. Lack of cell–cell communication and reduction in Cx43 expression could reflect the undifferentiated status of these cells, a component of their tumorigenic phenotype. However, the W12-derived transformed cell line, W12GPX, the precursor to W12GPXY, that has also lost differentiation capacity was found to display good GJIC and membranous Cx43 [12]. Therefore we propose that loss of GJIC in W12GPXY reflects a late event in cervical tumour progression [12] and not a loss of Cx expression due to an inability of the cells to differentiate.

Our results demonstrate that Cx43 relocation to the cytoplasm of W12GPXY cells is specific, as other Cx proteins, Cx26 and Cx30, remained on the membrane. Indeed, there appeared to be increased Cx26 and Cx30 on the membrane of W12GPXY cells (Figure 3). It is possible that the progressed malignant transformation of W12GPXY cells results in increased expression or stability of Cx26 and Cx30. Studies have revealed such an increase in breast cancer [44], ovarian tumours [45], melanoma [46] and pancreatic tumours [47]. W12GPXY cells displayed punctate membrane staining of these two Cxs, yet do not display competent GJIC, at least as assayed by transfer of Lucifer Yellow [12], consistent with the lower size exclusion of Cx26/Cx30 in comparision with Cx43 gap junction channels. Only some W12G cells in the cell populations studied in Figure 3 expressed membrane Cx26 and Cx30, whereas others did not stain with antibodies against Cx26 and Cx30. This heterogeneity is probably due to the presence of both differentiating epithelial cells (high levels of Cx26/Cx30 expression) and undifferentiated cells (low levels of Cx26/Cx30 expression) in the culture [42,43].

Cx43 binds to the second PDZ domain of the MAGUK protein ZO-1 and it has been proposed that this scaffolding interaction may organize gap junctions at the plasma membrane by facilitating insertion of Cxs to the edge of gap junction plaques aiding the connexon/gap junction transition and maintaining gap junction stability [18,21,48,49]. ZO-1 may also be involved in facilitating gap junction disassembly [50,51]. Our results confirm that ZO-1 can interact with Cx43 in cervical epithelial cell extracts. However, we detected little co-localization of ZO-1 with Cx43 in confocal microscopy of W12GPXY cells, and ZO-1 was extracted mainly in the Triton X-100-insoluble fraction, whereas Cx43 was found mainly in the soluble fraction of these cells. Therefore ZO-1 is unlikely to regulate Cx43 trafficking to the cytoplasm in the cervical tumour cells.

hDlg, another MAGUK protein, was previously shown to bind Cx32 [24] and was also identified as a potential binding partner for the Cx43 C-terminal tail in a MS/MS (tandem MS) analysis screen [52]. In contrast with ZO-1, hDlg is a good candidate for regulating Cx43 in cervical epithelial cells. We observed co-localization of the two proteins in confocal images and neither protein was found in the Triton X-100-insoluble fraction in W12GPXY cells. Moreover, we have direct and indirect evidence that hDlg interacts with the Cx43 C-terminal domain. The MAGUK family proteins (hDlg, ZO-1, ZO-2, PSD-95/SAP 90, p55 and others) all display a similar modular structure composed of two to three PDZ domains, an SH3 domain and GUK domain. Interestingly, hDlg also contains an N-terminal domain, which is not found in other MAGUKs [53,54]. This N-terminal domain controls subcellular location of hDlg [55]. Our studies indicate that the C-terminal domain of Cx43 forms a complex with hDlg within cervical epithelial cells. Moreover, it binds hDlg directly and protein–protein association is through the hDlg N- and C-terminal domains. The relatively weak interaction of hDlg and Cx43 with either terminus in GST pull-downs implies that full interaction requires simultaneous binding to both domains. Therefore Cx43 binds hDlg in a different manner to that through which it binds ZO-1: whereas the second PDZ domain of ZO-1 is required for Cx43 binding, hDlg-Cx43 binding does not require the PDZ domains. In support of a different mode of interaction between Cx43 and hDlg is our observation that the last five amino acids of Cx43 that comprise the PDZ-binding domain, including those required for ZO-1 binding [18], were dispensable for hDlg interaction.

In a scaffolding model, the different domains of hDlg serve as docking modules for interactions with different proteins/polypeptides [56]. hDlg could function in the transport cycle of Cx43 by providing a docking platform for molecules involved in assembling or disassembling gap junctions. This is similar to what has been proposed for Cx43 interaction with ZO-1 and raises the intriguing possibility that both PDZ proteins may co-operate to regulate gap junction trafficking.

The transport cycle of Cxs involves trafficking from the ER/Golgi to the plasma membrane and formation of gap junction plaques. Internalization of membrane Cxs may be mediated through the ‘connexosome’ and targeted to either the lysosomal or proteasomal degradation pathways. It has been proposed that different populations of Cxs may be differentially targeted [4]. Our results demonstrate a significant increase in lysosomal degradation of Cx43 in transformed W12GPXY cells compared with the untransformed W12G cells. Although co-localization of hDlg and Cx43 in W12GPXY cells suggested that hDlg might target Cx43 to the lysosomes, and indeed the two proteins do co-localize with a lysosomal marker, siRNA knockdown of hDlg revealed a decrease, rather than an increase in Cx43 levels. Our results suggest that although hDlg may accompany a pool of Cx43 to the lysosomes, some hDlg is required to maintain another pool of Cx43 in the cytoplasm in transformed cervical epithelial cells. The two pools of hDlg may represent different splice isoforms that can be detected with the anti-hDlg antibodies we used [55,57] or differentially post-translationally modified forms of the protein [58].

There is conflicting evidence for the roles of Cxs in cancer. In some cases, particularly during tumorigenesis, it is clear that Cxs may act as tumour suppressors. However, at later disease stages, Cxs may play a role in the various processes between cancer-cell migration and metastasis [59]. W12GPXY cells have invasive properties [12], and it is possible that an hDlg-maintained Cx43 pool in cervical tumour cells could serve to reform gap junctions in cells disseminated to distant metastatic sites.

In conclusion, our findings highlight hDlg as a physiological regulator of Cx43 and suggest that hDlg may contribute to control of the Cx43 life cycle. It has been recognized for some time that, in line with disruption of GJIC, hDlg is a common target in epithelial cancers, displaying mutations, changes in protein levels and changes in subcellular localization [6062]. Although W12 cells express HPV16 E6 protein, which can bind and target PDZ proteins for proteasomal degradation, it has only a minimal role in hDlg degradation [63], meaning that E6 regulation of hDlg levels is unlikely to control Cx43 relocation. We showed previously a significant alteration in Cx43 isoform profile in W12GPXY cells compared with W12G cells [12]. Dephosphorylation and phosphoprotein purification studies have revealed that at least some of these isoforms are due to altered phosphorylation (results not shown). Since many sites of phosphorylation are located in the C-terminal domain of Cx43 [8], they could modulate Cx43 interaction with hDlg or Cx43 trafficking leading to loss of GJIC during malignant progression of keratinocytes. Future studies will shed light on the relationship of the phosphorylation status of Cx43 and alteration of its functions in these tumour cells.

Abbreviations

     
  • Cx

    connexin

  •  
  • DAPI

    4′, 6-diamidino-2-phenylindole

  •  
  • Dlg

    discs large

  •  
  • DMEM

    Dulbecco's modified Eagle's medium

  •  
  • ER

    endoplasmic reticulum

  •  
  • FBS

    fetal bovine serum

  •  
  • GAPDH

    glyceraldehyde-3-phosphate dehydrogenase

  •  
  • GJIC

    gap junctional intercellular communication

  •  
  • GST

    glutathione transferase

  •  
  • GUK

    guanylate kinase

  •  
  • hDlg

    human Dlg

  •  
  • HPV

    human papillomavirus

  •  
  • HRP

    horseradish peroxidise

  •  
  • IP

    immunoprecipitation

  •  
  • IPTG

    isopropyl β-D-thiogalactopyranoside

  •  
  • KGM

    keratinocyte growth medium

  •  
  • MAGUK

    membrane-associated guanylate kinase

  •  
  • MPR

    mannose 6-phosphate receptor

  •  
  • NP-40

    Nonidet P40

  •  
  • PBST

    PBS containing 0.1% Tween 20

  •  
  • SH3

    Src homology 3

  •  
  • siRNA

    small interfering RNA

  •  
  • TopBP1

    topoisomerase binding protein 1

  •  
  • ZO-1

    zona occludens-1

AUTHOR CONTRIBUTION

Alasdair MacDonald carried out all of the experiments in Figures 2, 6 and 8, some of the experiments in Figures 3, 4 and 7, and helped to write the paper. Peng Sun carried out the experiments in Figures 1, 3 and 4. Hegel Hernandez-Lopez carried out the experiment in Figure 7(A). Trond Aasen developed the W12GPXY cell line and carried out the experiment in Figures 3(E) and 3(F). Malcolm Hodgins was a co-supervisor of Peng Sun and Trond Aasen, a co-applicant on the grant that supported Alasdair MacDonald, gave advice on experiments, and helped to write and revise the paper. Michael Edward was a co-supervisor of Peng Sun and Trond Aasen. Sally Roberts had the original idea of Cx43–hDlg interaction and gave advice on developing the project. Paola Massimi generated important background data. Miranda Thomas generated most of the Dlg/E6 constructs. Lawrence Banks was a co-applicant on the grant that supported Alasdair MacDonald, supervised Paola Massimi and Miranda Thomas, gave advice on experiments, and helped to write and revise the paper. Sheila Graham carried out the experiments in Figure 5, co-ordinated the project, was a co-supervisor of Peng Sun and Trond Aasen, was the principal investigator on the grant that supported Alasdair MacDonald, supervised his research, and wrote and revised the paper.

We are grateful to Margaret Stanley (Department of Pathology, University of Cambridge, Cambridge, U.K.) for the original W12 cell line and Paul Lambert (School of Medicine and Public Health, University of Wisconsin, Madison, WI, U.S.A.) for the W12E and W12G clones of this line. We thank Patricia Martin (School of Health and Life Sciences, Glasgow Caledonian University, Glasgow, Scotland, U.K.) for generously sharing reagents. Mary Donaldsdon (Centre for Virus Research, University of Glasgow, Glasgow, Scotland, U.K.) kindly provided purified GST–TopBP1. We are very grateful to Colin Loney for technical assistance with confocal microscopy. We thank Arbor Vita Corporation for provision of the E6 antibody.

FUNDING

This work was funded by the Association for International Cancer Research [grant number 08–0159 (to S.V.G., M.B.H. and L.B.)]. P.S. and T.A. were supported by Glasgow University postgraduate scholarships and Overseas Research Studentship awards. H.H.-L. is supported by a CONACyT (Consejo Nacional de Ciencia y Tecnología) scholarship from the Mexican Government.

References

References
1
Fleming
T. P.
Ghassemifar
M. R.
Sheth
B.
Junctional complexes in the early mammalian embryo
Semin. Reprod. Med.
2000
, vol. 
18
 (pg. 
185
-
193
)
2
Evans
W. H.
Martin
P. E.
Gap junctions: structure and function
Mol. Membr. Biol.
2002
, vol. 
19
 (pg. 
121
-
136
)
3
Saez
J. C.
Berthoud
V. M.
Brañes
M. C.
Martínez
A. D.
Beyer
E. C.
Plasma membrane channels formed by connexins: their regulation and functions
Physiol. Rev.
2003
, vol. 
83
 (pg. 
1359
-
1400
)
4
Laird
D. W.
Life cycle of connexins in health and disease
Biochem. J.
2006
, vol. 
394
 (pg. 
527
-
543
)
5
Kumar
N. M.
Gilula
N. B.
The gap junction communication channel
Cell
1996
, vol. 
84
 (pg. 
381
-
388
)
6
Oyamada
M.
Oyamada
Y.
Takamatsu
T.
Regulation of connexin expression
Biochim. Biophys. Acta
2005
, vol. 
1719
 (pg. 
6
-
23
)
7
Moreno
A. P.
Lau
A. F.
Gap junction channel gating modulated through protein phosphorylation
Prog. Biophys. Mol. Biol.
2012
, vol. 
94
 (pg. 
107
-
119
)
8
Solan
J. L.
Lampe
P. D.
Connexin43 phosphorylation: structural changes and biological effects
Biochem. J.
2009
, vol. 
419
 (pg. 
261
-
272
)
9
McNutt
N. S.
Hershberg
R. A.
Weinstein
R.
Further observations on the occurrence of nexuses in benign and malignant human cervical epithelium
J. Cell Biol.
1971
, vol. 
51
 (pg. 
805
-
825
)
10
Krutovskikh
V.
Yamasaki
H.
The role of gap junctional intercellular communication (GJIC) disorders in experimental and human carcinogenesis
Histol. Histopathol.
1997
, vol. 
12
 (pg. 
761
-
768
)
11
King
T. J.
Fukushima
L.
Hieber
A. D.
Shimabukuro
A. D.
Sakr
W. A.
Bertram
J. S.
Reduced levels of connexin 43 in cervical dysplasia: inducible expression in a cervical carcinoma line decreases neoplastic potential with implications for tumour progression
Carcinogenesis
2000
, vol. 
21
 (pg. 
1097
-
1109
)
12
Aasen
T.
Hodgins
M. B.
Edward
M.
Graham
S. V.
The relationship between connexins, gap junctions, tissue architecture and tumour invasion, as studied in a novel in vitro model of HPV-16-associated cervical cancer progression
Oncogene
2003
, vol. 
22
 (pg. 
7969
-
7980
)
13
Cronier
L.
Crespin
S.
Strale
P.-O.
Defamie
N.
Mesnil
M.
Gap junctions and cancer: new functions for an old story. Antioxid
Redox Signaling
2009
, vol. 
11
 (pg. 
323
-
338
)
14
Naus
C. C.
Laird
D. W.
Implications and challenges of connexin connections to cancer
Nat. Rev. Cancer
2010
, vol. 
10
 (pg. 
435
-
441
)
15
Dimitratos
S. D.
Wioods
D. F.
Stathakis
D.
Bryant
P. J.
Signalling pathways are focused at specialised regions of the plasma membrane by scaffolding proteins of the MAGUK family
Bioessays
1999
, vol. 
21
 (pg. 
912
-
921
)
16
Kim
E.
Sheng
M.
PDZ domain proteins of synapses
Nat. Rev. Neurosci.
2004
, vol. 
5
 (pg. 
771
-
781
)
17
Toyofuku
T.
Yabuki
M.
Otsu
K.
Kuzuya
T.
Hori
M.
Tada
M. H.
Direct association of the gap junction protein connexin-43 with ZO-1 in cardiac myocytes
J. Biol. Chem.
1998
, vol. 
273
 (pg. 
12725
-
12731
)
18
Geipmans
B. N. G.
Moolenar
W. H.
The gap junction protein connexin-43 interacts with the second PDZ domain of the zona occludens-1 protein
Curr. Biol.
1998
, vol. 
8
 (pg. 
931
-
934
)
19
Hunter
A. W.
Barker
R.
Zhu
C.
Gourdie
R.
Zonal occludens-1 alters connexin43 gap junction size and organisation by influencing channel accretion
Mol. Biol. Cell
2005
, vol. 
16
 (pg. 
5686
-
5698
)
20
Rhett
J. M.
Jourdan
J.
Gourdie
R. G.
Connexin 43 connexon to gap junction transition is regulated by zonula occludens-1
Mol. Biol. Cell
2011
, vol. 
22
 (pg. 
1516
-
1528
)
21
Laing
J. G.
Manley-Marowski
R. N.
Koval
M.
Civitelli
R.
Steinberg
T. H.
Connexin45 interacts with zonula occludens-1 and connexin43 in osteoblastic cells
J. Biol. Chem.
2001
, vol. 
276
 (pg. 
23051
-
23055
)
22
Kausalya
P. J.
Reichert
M.
Hunziker
W.
Connexin45 binds to ZO-1 and localises to the tight junction region in epithelial MDCK cells
FEBS Lett.
2001
, vol. 
505
 (pg. 
92
-
96
)
23
Laing
J. G.
Chou
B. C.
Steinberg
T. H.
ZO-1 alters the plasma membrane localization and function of Cx43 in osteoblastic cells
J. Cell Sci.
2005
, vol. 
118
 (pg. 
2167
-
2176
)
24
Duffy
H. S.
Iacobas
I.
Hotchkiss
K.
Hirst-Jensen
B. J.
Bosco
A.
Dandachi
N.
Dermietzel
R.
Sorgen
P. L.
Spray
D. C.
The gap junction protein connexin32 interacts with the Src homology 3/hook domain of discs large homolog
J. Biol. Chem.
2007
, vol. 
282
 (pg. 
9789
-
9796
)
25
Woods
D. F.
Bryant
P. J.
The disc-large tumor supressor gene of Drosophila encodes a guanylate kinase homolog localized at septate junctions
Cell
1991
, vol. 
66
 (pg. 
451
-
464
)
26
Matsumine
A.
Ogai
A.
Senda
T.
Okumura
N.
Satoh
K.
Baeg
G. H.
Kawahara
T.
Kobayashi
S.
Okada
M.
Toyoshima
K.
, et al. 
Binding of APC to the human homolog of the Drosophila disc large tumor suppressor protein
Science
1996
, vol. 
272
 (pg. 
1020
-
1023
)
27
Goode
S.
Perrimon
N.
Inhibition of patterned cell shape change and cell invasion by discs large during Drosophila oogenesis
Genes Dev.
1997
, vol. 
11
 (pg. 
2532
-
2544
)
28
Funke
L.
Dakoji
S.
Bredt
D. S.
Membrane-associated guanylate kinases regulate adhesion and plasticity at cell junctions
Annu. Rev. Biochem.
2005
, vol. 
74
 (pg. 
219
-
245
)
29
zur Hausen
H.
Papillomaviruses in the causation of human cancers: a brief historical account
Virology
2009
, vol. 
384
 (pg. 
260
-
265
)
30
Jeon
S.
Allen-Hoffman
B. L.
Lambert
P. F.
Integration of human papillomavirus type 16 into the human genome correlates with a selective growth advantage of cells
J. Virol.
1995
, vol. 
69
 (pg. 
2989
-
2997
)
31
Mole
S.
Milligan
S. G.
Graham
S. V.
Human papillomavirus type 16 E2 protein transcriptionally activates the promoter of a key cellular splicing factor, SF2/ASF
J. Virol.
2009
, vol. 
83
 (pg. 
357
-
367
)
32
Gardiol
D.
Kühne
C.
Glausinger
B.
Lee
S.
Javier
R.
Banks
L.
Oncogenic human papillomavirus E6 proteins target the discs large tumour suppressor for proteasome-mediated degradation
Oncogene
1999
, vol. 
18
 (pg. 
5487
-
5496
)
33
Singh
D.
Lampe
P. D.
Identification of connexin-43 interacting proteins
Cell Commun. Adhes.
2003
, vol. 
10
 (pg. 
215
-
220
)
34
Musil
L. S.
Goodenough
D. A.
Biochemical analysis of connexin43 intracellular transport, phosphorylation, and assembly into gap junction plaques
J. Cell Biol.
1991
, vol. 
115
 (pg. 
1357
-
1374
)
35
Hartsock
H.
Nelson
W. J.
Adherens junctions: structure, function and connections to the actin cytoskeleton
Biochim. Biophys. Acta
2008
, vol. 
1778
 (pg. 
660
-
669
)
36
Näthke
I. S.
Hinck
L.
Swedlow
J. R.
Papkoff
J.
Nelson
W. J.
Defining interactions and distribution of cadherin and catenin complexes in polarised epithelial cells
J. Cell Biol.
1994
, vol. 
125
 (pg. 
1341
-
1352
)
37
Boner
W.
Taylor
E. R.
Tsirimonaki
E.
Yamana
K.
Campo
M. S.
Morgan
I. M.
A functional interaction between the human papillomavirus 16 transcription/replication factor E2 and the DNA damage response protein TopBP1
J. Biol. Chem.
2002
, vol. 
277
 (pg. 
22297
-
22303
)
38
Musil
L. S.
Le
A.-C. N.
VanSlyke
J. K.
Roberts
L. M.
Regulation of connexin degradation as a mechanism to increase gap junction assembly and function
J. Biol. Chem.
2000
, vol. 
275
 (pg. 
25207
-
25215
)
39
Leithe
E.
Brech
A.
Rivedal
E.
Endocytic processing of connexin43 gap junctions: a morphological study
Biochem. J.
2006
, vol. 
393
 (pg. 
59
-
67
)
40
Shore
L.
McLean
P.
Gilmour
S. K.
Hodgins
M. B.
Finbow
M. E.
Polyamines regulate gap junction communication in connexin 43-expressing cells
Biochem J.
2001
, vol. 
357
 (pg. 
489
-
495
)
41
Milligan
S. G.
Veerapraditsin
T.
Ahamat
B.
Mole
S.
Graham
S. V.
Analysis of novel human papillomavirus type 16 late mRNAs in differentiated W12 cervical epithelial cells
Virology
2007
, vol. 
360
 (pg. 
172
-
181
)
42
Aasen
T.
Graham
S. V.
Edward
M.
Hodgins
M. B.
Reduced expression of multiple gap junction proteins is a feature of cervical dysplasia
Mol. Cancer
2005
, vol. 
4
 (pg. 
1
-
5
)
43
Steinhoff
I.
Leykauf
K.
Bleyl
U.
Dürst
M.
Alonso
A.
Phosphorylation of the gap junction protein connexin43 in CIN III lesions and cervical carcinomas
Cancer Lett.
2006
, vol. 
235
 (pg. 
291
-
297
)
44
Jamieson
S.
Going
J. J.
D'Arcy
R.
George
W. D.
Expression of gap junction proteins connexin 26 and connexin 43 in normal human breast and in breast tumours
J. Pathol.
1998
, vol. 
184
 (pg. 
37
-
43
)
45
Zhai
Y.
Wu
R.
Schwartz
D. R.
Darrah
D.
Reed
H.
Kolligs
F. T.
Nieman
M. T.
Fearon
E. R.
Cho
K. R.
Role of the β-catenin/T-cell factor-regulated genes in ovarian endometrioid adenocarcinomas
Am. J. Pathol.
2002
, vol. 
160
 (pg. 
1229
-
1238
)
46
Ito
A.
Katoh
F.
Kataoka
T. R.
Okada
M.
Tsubota
N.
Asada
H.
Toshikawa
K.
Maeda
S.
Kitamura
Y.
Yamasaki
H.
, et al. 
A role for heterologous gap junctions between melanoma and endothelial cells in metastasis
J. Clin. Invest.
2000
, vol. 
105
 (pg. 
1189
-
1197
)
47
Kyo
N.
Yamamoto
H.
Takeda
Y.
Ezumi
K.
Ngan
C. Y.
Terayama
M.
Miyake
M.
Takemasa
I.
Ikeda
M.
Doki
Y.
, et al. 
Overexpression of connexin 26 in carcinoma of the pancreas
Oncol. Rep.
2008
, vol. 
19
 (pg. 
1189
-
1197
)
48
Rhett
J. M.
Jourdan
J.
Gourdie
R. G.
Connexin 43 connexon to gap junction transition is regulated by zonula occludens-1
Mol. Biol. Cell
2010
, vol. 
22
 (pg. 
1516
-
1528
)
49
Chakraborty
S.
Mitra
S.
Falk
M. M.
Caplan
S. H.
Wheelock
M. J.
Johnson
K. R.
Mehta
P. P.
E-cadherin differentially regulates the assembly of connexin43 and connexin32 into gap junctions in human squamous carcinoma cells
J. Biol. Chem.
2010
, vol. 
285
 (pg. 
10761
-
10776
)
50
Akoyev
V.
Takemoto
D. J.
ZO-1 is required for protein kinase Cγ-driven disassembly of connexin 43
Cell. Signalling
2007
, vol. 
19
 (pg. 
958
-
967
)
51
Gilleron
J.
Fiorini
C.
Carette
D.
Avondet
C.
Falk
M. M.
Segretain
D.
Pointis
G.
Molecular reorganisation of Cx43, ZO-1 and Src complexes during the endocytosis of gap junction plaques in response to a non-genomic carcinogen
J. Cell Sci.
2008
, vol. 
121
 (pg. 
4069
-
4078
)
52
Singh
D.
Lampe
P. D.
Identification of connexin-43 interacting proteins
Cell Commun. Adhes.
2003
, vol. 
10
 (pg. 
215
-
220
)
53
Lue
R. A.
Marfatia
S. M.
Branton
D.
Chishti
A. H.
Cloning and characterization of hDlg: the human homologue of the Drosophila discs large tumour suppressor binds to protein 4.1
Proc. Natl. Acad. Sci. U.S.A.
1994
, vol. 
91
 (pg. 
9818
-
9822
)
54
Lue
R. A.
Brandin
E.
Chan
E. P.
Branton
D.
Two independent domains of hDlg are sufficient for subcellular targeting: the PDZ1–2 conformational unit and an alternatively spliced domain
J. Cell Biol.
1996
, vol. 
114
 (pg. 
4285
-
4292
)
55
Wu
H.
Reuver
S. M.
Kuhlendahl
S.
Chung
W. J.
Garner
C. C.
Subcellular targeting and cytoskeletal attachment of SAP97 to the epithelial lateral membrane
J. Cell Sci.
1998
, vol. 
111
 (pg. 
2365
-
2376
)
56
Hung
A. Y.
Sheng
M.
PDZ domains: structural modules for protein complex assembly
J. Biol. Chem.
2002
, vol. 
277
 (pg. 
5699
-
5702
)
57
McLaughlin
M.
Hale
R.
Ellston
D.
Gaudet
S.
Lue
R. A.
Viel
A.
The distribution and function of alternatively spliced insertions in hDlg
J. Biol. Chem.
2001
, vol. 
277
 (pg. 
6406
-
6412
)
58
Mantovani
F.
Banks
L.
Regulation of the discs large tumor suppressor by a phosphorylation-dependent interaction with the β-TrCP ubiquitin ligase receptor
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
42477
-
42486
)
59
Naus
C. C.
Gap junctions and tumour progression
Can. J. Physiol. Pharmacol.
2002
, vol. 
80
 (pg. 
136
-
141
)
60
Watson
R. A.
Rollason
T. P.
Reynolds
G. M.
Murray
P. G.
Banks
L.
Roberts
S.
Changes in expression of the human homologue of the Drosophila discs large tumour suppressor protein in high grade premalignant cervical neoplasias
Carcinogenesis
2002
, vol. 
23
 (pg. 
1791
-
1796
)
61
Lin
H.-T.
Steller
M. A.
Aish
L.
Hanada
T.
Chishti
A. H.
Differential expression of human hDlg in cervical intraepithelial neoplasias
Gynecol. Oncol.
2004
, vol. 
93
 (pg. 
422
-
428
)
62
Cavatorta
A. L.
Fumero
G.
Chouhy
D.
Aguirre
R.
Nocito
A. L.
Giri
A. A.
Banks
L.
Gardiol
D.
Differential expression of the human homologue of Drosophila discs large oncosuppressor in histological samples from human papillomavirusassociated lesions as a marker for progression to malignancy
Int. J. Cancer
2004
, vol. 
111
 (pg. 
373
-
380
)
63
Thomas
M.
Dasgupta
J. C. X.
Banks
L.
Analysis of specificity determinants in the interactions of different HPV E6 proteins with their PDZ domain-containing substrates
Virology
2008
, vol. 
376
 (pg. 
371
-
378
)
64
Roberts
S.
Calautti
E.
Vanderweil
S.
Nguyen
H. O.
Foley
A.
Baden
H. P.
Viel
A.
Changes in localization of human discs large (hDlg) during keratinocyte differentiation is associated with expression of alternatively spliced hDlg variants
Exp. Cell Res.
2007
, vol. 
313
 (pg. 
2521
-
2530
)

Author notes

1

Present address: MD Anderson Cancer Centre, Houston, TX 77030, U.S.A.

2

Present address: Pathology Department, Fundació Institut de Recerca Hospital Vall d'Hebrón, 08035 Barcelona, Spain

Supplementary data