HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 280, Issue 20, 19925-19936, May 20, 2005

This Article Abstract Full Text (PDF) All Versions of this Article: 280/20/19925    most recent M412921200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Wei, C.-J. Articles by Lo, C. W. Search for Related Content PubMed PubMed Citation Articles by Wei, C.-J. Articles by Lo, C. W. Social Bookmarking                      What's this?

Connexin43 Associated with an N-cadherin-containing Multiprotein Complex Is Required for Gap Junction Formation in NIH3T3 Cells^*

Chih-Jen Wei, Richard Francis, Xin Xu, and Cecilia W. Lo

From the Laboratory of Developmental Biology, NHLBI, National Institutes of Health, Department of Health and Human Services, Bethesda, Maryland 20892

Received for publication, November 15, 2004 , and in revised form, February 23, 2005.

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

 
Previous studies have indicated an intimate linkage between gap junction and adherens junction formation. It was suggested this could reflect the close membrane-membrane apposition required for junction formation. In NIH3T3 cells, we observed the colocalization of connexin43 (Cx431) gap junction protein with N-cadherin, p120, and other N-cadherin-associated proteins at regions of cell-cell contact. We also found that Cx431, N-cadherin, and N-cadherin-associated proteins were coimmunoprecipitated by antibodies to either Cx431, N-cadherin, or various N-cadherin-associated proteins. These findings suggest that Cx431 and N-cadherin are coassembled in a multiprotein complex containing various N-cadherin-associated proteins. Studies using siRNA knockdown indicated that cell surface expression of Cx431 required N-cadherin, and conversely, N-cadherin cell surface expression required Cx431. Pulse-chase labeling and cell surface biotinylation experiments indicated that in the absence of N-cadherin, Cx431 cell surface trafficking is blocked. Surprisingly, siRNA knockdown of p120, an N-cadherin-associated protein known to modulate cell surface turnover of N-cadherin, reduced N-cadherin cell surface expression without altering Cx431 expression. These observations suggest that in contrast to the coregulated cell surface trafficking of Cx431 and N-cadherin, N-cadherin turnover at the cell surface may be regulated independently of Cx431. Functional studies showed gap junctional communication is reduced and cell motility inhibited with N-cadherin or Cx431 knockdown, consistent with the observed loss of both gap junction and cadherin contacts with either knockdown. Overall, these studies indicate that the intracellular coassembly of connexin and cadherin is required for gap junction and adherens junction formation, a process that likely underlies the intimate association between gap junction and adherens junction formation.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

 
Gap junctions are specialized cell junctions containing hydrophilic channels that mediate intercellular communication. They play an essential role in electrical and metabolic coupling by regulating the movement of ions and small molecules between cells (1). Gap junctions are comprised of two hemichannels, each consisting of a hexameric assembly of transmembrane proteins known as the connexins (2). In the mouse genome, there are 20 connexin genes and each generate gap junction channels with different gating and regulatory properties that are likely dictated by the intracellular domains of the proteins (3).

The gap junction gene, Gja1, is one of the most widely expressed connexin genes. It encodes a 43-kDa protein known as connexin43 or ₁ connexin, referred to here as Cx431.¹ Cx431 is synthesized in the endoplasmic reticulum and transported to the trans-Golgi network, where it is oligomerized (4). This is followed by trafficking to the cell surface and incorporation of the hexameric arrays into gap junction plaques at regions of cell-cell contact (5). The Cx431 protein has a half-life of only 1–5 h (for review, see Ref. 6), and is turned over through the endosomal/lysosomal pathway (7–9). The cytoplasmic terminus (CT) of Cx431 is subject to modification by a variety of protein kinases (10, 11). In Western blots, there is typically a faster migrating species of Cx431 that co-migrates with nonphosphorylated Cx431 (NP), and one or more slower migrating bands containing phosphorylated isoforms. The NP form is intracellular, whereas the phosphorylated P2 form is localized in gap junction plaques (12–14).

Various studies have indicated that gap junction formation is dependent on the assembly of adherens junctions. This was first indicated by studies in which antibody-mediated disruption of cadherin-containing cell adhesion contacts were shown to block gap junction formation (15–17). More recent studies showed that N-cadherin knockout cardiomyocytes are Cx431 gap junction-deficient (18), whereas expression of a dominant-negative N-cadherin construct in cardiomyocytes was shown to disrupt Cx431 gap junctions (19). A reciprocal requirement for gap junctions in adherens junction formation also has been noted. For example, when gap junction antibodies were injected into 2-cell mouse embryos, the embryos failed to undergo compaction, a process regulated by cadherins (20). In Xenopus embryos, expression of a dominant negative Cx321/Cx431 chimeric construct inhibited gap junction communication, and led to the extrusion of cells from the embryo because of a loss of cell adhesion contacts (21). In another study, treatment of cells with Fab fragments targeting Cx431 or Cx321 blocked both gap junction as well as adherens junction formation (17). Together these findings indicate that formation of gap junction and adherens junction are intimately linked. It was previously suggested that this might reflect the close membrane-membrane apposition required for gap junction and adherens junction formation (17).

We previously showed that Cx431 gap junctions also can modulate cell motility. Thus altered neural crest cell motility was shown to contribute to the cardiac phenotype of the Cx431 knockout mouse (22–24). We found in migrating neural crest cells, the colocalization of Cx431 and N-cadherin at many regions of cell-cell contact (24). To further examine the interactions between Cx431 and N-cadherin, and investigate their role in regulating gap junction communication and cell motility, in this study we used NIH3T3 cells as a model system. NIH3T3 cells are reminiscent of neural crest cells, as they are a highly motile mesenchymal cell type, and express both Cx431 and N-cadherin. Analyses by immunohistochemistry, and coimmunoprecipitation and Western immunoblotting indicated that Cx431 and N-cadherin are coassembled in a large multiprotein complex. Consistent with this, siRNA-mediated knockdown of either Cx431 or N-cadherin resulted in the apparent loss of both proteins from the cell surface. Surface biotinylation and pulse chase experiments showed that cell surface trafficking of Cx431 is blocked with siRNA-mediated N-cadherin knockdown. However, siRNA knockdown of p120, a N-cadherin-associated protein known to modulate cell surface turnover of N-cadherin (25), eliminated cell surface expression of N-cadherin without affecting Cx431 expression. This suggests that the cell surface turnover of N-cadherin may be regulated independently of Cx431. We also found with either N-cadherin or Cx431 knockdown, not only was gap junctional communication reduced, but cell motility was inhibited. This is consistent with the observed loss of both gap junction and cell adhesion contacts with either knockdown. Overall, these studies indicate that the intracellular coassembly of connexin and cadherin is required for gap junction and adherens junction formation. This process of coassembly likely accounts for the intimate linkage between gap junction and adherens junction formation, a process that is also functionally important in the modulation of cell motility.

View larger version (70K): [in this window] [in a new window]   FIG. 1. Cx431 and N-cadherin protein interactions. Panels a–d, double immunostaining of NIH3T3 cells showed at regions of cell-cell contact, colocalization of Cx431 (red) with other proteins (green) including N-cadherin (panel a), p120 (panel b), -catenin (panel c), and ZO-1 (panel d) (see regions denoted by white arrows). These are merged phase contrast and darkfield immunofluorescence images. Boxed area is shown in magnified view in upper right inset, and corresponds to a region showing Cx431 colocalization with N-cadherin or N-cadherin-associated protein. Scale bar, 25 µm. Panel e, cell lysates were immunoprecipitated with a Cx431 antibody, then Western immunoblotted using antibodies to -catenin, -catenin, N-cadherin, and p120. With each antibody, a band of the expected size for respective proteins was obtained, showing that N-cadherin and N-cadherin-associated proteins were coimmunoprecipated with Cx431. Panel f, Cx431 antibody precipitates analyzed by Western immunoblotting using antibodies to paxillin or FAK showed neither of these two proteins was in the Cx431 immunoprecipitates. Panel g, immunoprecipitates obtained from cell lysates incubated with antibodies to either N-cadherin, or various N-cadherin-associated proteins wereanalyzed by Western immunoblotting using a Cx431 antibody. Cx431 was observed in all of these immunoprecipitates. Panel h, affinity pull-down assays were carried out using purified GST-N-cad-CT bound to glutathione beads as bait (left panel). These were incubated with purified His-Cx431-CT, which served as prey (right panel, lane 1). Bait-prey interaction was examined by immunoblotting proteins pulled down by the bait-loaded beads using a His tag antibody. Controls included glutathione beads alone (lane 2), glutathione beads with bovine serum albumin (lane 3), prey alone (lane 4), and prey with GST-loaded beads (lane 5). Some His-Cx431-CT protein was pulled-down with the bait-loaded beads incubated with prey (lane 6), and also in control samples containing either prey alone (lane 4), or prey incubated with GST loaded beads (lane 5).

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Immunohistochemistry—NIH3T3 cells were fixed with 4% paraformaldehyde for 15 min or 4% paraformaldehyde containing 0.15% Triton X-100 for 10 min, then washed and incubated with primary antibody at 4 °C overnight, followed by incubation in FITC- or Cy3-conjugated secondary antibodies. Epifluorescence was carried out on a Leica DMRE microscope with a x63 PL oil objective and imaged using an ORCA-ER camera (Hamamatsu). Z stacks comprised of 0.2-µm optical slices were obtained and deconvolved using Volocity iterative deconvolution algorithm (Improvision, Ltd.). All images shown in this study are derived from merging 3 deconvolved optical sections.

Immunoprecipitation and Western Immunoblotting—NIH3T3 cells were harvested and lysed in radioimmune precipitation assay buffer (RIPA, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 1 mM EDTA, 0.1% SDS, 50 mM Tris, pH 7.5) containing complete protease inhibitors (no. 8140, Sigma) and 1 mM phenylmethylsulfonyl fluoride for 15 min at 4 °C. Protein concentration was determined by BCA assay (Pierce). Lysates were precleared using protein G-agarose beads (Invitrogen) and centrifuged at 15,000 x g for 10 min. The supernatants were incubated with desired antibodies (10 µg) bound to protein G-agarose beads at 4 °C for overnight, washed extensively with cold RIPA buffer, and resolved by SDS-PAGE. The gel was electroblotted and processed for chemiluminescence detection using horseradish peroxidase-conjugated secondary antibodies. For some experiments, cells were treated with the lysosomal inhibitor ammonium chloride (Sigma) for 24 h prior to lysis in RIPA buffer, followed by immunoblotting analysis.

Construction and Purification of His- and GST-tagged Fusion Proteins—To generate His-tagged Cx431 CT-expressing vector, cDNA encoding the CT domain of Cx431 (amino acids 227–382) was generated by PCR using primers 5'-CCGGAATTCGCGCTCTTCTATGTCTTCTTC-3' and 5'-CCGCTCGAGTTAAATCTCCAGGTCATCAGGC-3', and ligated into the EcoRI/XhoI sites of pET28a vector (Novagen). The recombinant vector was transformed into Escherichia coli BL21(DE3) pLysS, and the fusion protein was purified as previously described (26).

To generate GST-tagged fusion containing the CT domain of N-cadherin (amino acids 753–904), the cDNA for mouse N-cadherin (kindly provided by Dr. Glenn L. Radice, University of Pennsylvania) was PCR-amplified using primers 5'-CCGGAATTCATGCGCCAAGCCAAGCAGCTTTTAATTGAC-3'and 5'-CCGCTCGAGTCAGTCGTCACCACCGCCGTACATGTCCGC-3', and the resulting PCR fragment was inserted into the EcoR I/XhoI sites of pGEX-T4–1 expression vector (Amersham Biosciences). After expression in E. coli DH5, GST-tagged fusion proteins was purified by affinity chromatography on a glutathione-Sepharose 4B column (Amersham Biosciences).

Affinity Binding Assay—Purified GST-tagged N-cadherin CT (GST-N-cad-CT, bait) was immobilized on glutathione-agarose beads (Pierce), and after extensive washing, was incubated with previously purified His-tagged Cx431-CT (His-Cx431-CT, prey) for 2 h at 4 °C. His-Cx431-CT was also incubated with the glutathione beads as prey control, and GST-glutathione beads were used as bait control. Bound GST or GST-N-cad-CT fusion proteins was eluted in elution buffer (50 mM reduced glutathione, 50 mM Tris-HCl, pH 8.0), and subjected to immunoblot analysis with a GST antibody.

siRNA-mediated Knockdown—For siRNA-mediated knockdown of Cx431 and N-cadherin, siRNAs (Dharmacon, Inc.) were synthesized containing nucleotides 143–163 (AACAGTCTGCCTTTCGCTGTA) and 859–879 (AAGCTGGTCACTGGTGACAGA) of the mouse Cx431 coding sequence, and nucleotides 139–159 (AAGGATGTGCACGAAGGACAG) and 1049–1069 (AAGCCACAGACATGGAAGGCA) of the mouse N-cadherin coding sequence. Scrambled siRNAs contained the same sequences except for 3 mismatched nucleotides (for Cx431: AACAGTATGCCGTTCGCCGTA, and AAGCTGTTCGCTGGGGACAGA; for N-cadherin: AAGGATGGGAACGCAGGACAG, and AAGCCACGGCCATAGAAGGCA).

Oligofectamine (Invitrogen) was used to transfect NIH3T3 cells with the siRNAs for immunofluorescence microscopy and Western immunoblotting. For time lapse videomicroscopy, cells were electroporated with siRNAs and a GFP expression vector, pEGFP-C1 (Clontech, Inc.), using Cell Line Nucleofector Kit R (Amaxa Biosystems). Alternatively, siRNA-mediated knockdown was performed using a mammalian expression vector directing intracellular synthesis of siRNA-like transcripts (27). They were made by inserting into pSuper.neo+gfp vector (OligoEngine), an oligonucleotide sequence corresponding to mouse Cx431 (CAGTCTGCCTTTCGCTGTA), p120 (CGAAGTTATCGCTGAGAAC), or N-cadherin (GCCACAGACATGGAAGGCA). We also generated control plasmids with 3 nucleotide mismatches in the 19-nucleotide targeting sequences.

Dye Coupling Assay—NIH3T3 cells were co-transfected with Cx431 or N-cadherin siRNAs and a pDsRed2-N1 expression vector (Clontech, Inc.), and used 48 h after transfection. Prior to microelectrode impalements, cells were transferred to L-15 medium (with L-glutamine, Invitrogen) and maintained at 37 °C on the heated stage of Leica DM-LFSA microscope equipped with a x40 objective. Dark field Cy3 imaging was used to locate DsRed-containing cells, and these were dye-injected while being monitored using a FITC filter. Micropipettes were backfilled with 2% 6-carboxyfluorescein, and ionotophoretic injection was carried out using 1 nA current pulses of 0.5-s duration at 1 Hz for 1 min. Following injection, a post-impalement DIC image was collected, and passive dye spread was recorded for another 2 min. Dye spread was quantitated by counting total number of dye filled cells and calculating percentage of dye-filled primary neighbors (directly adjacent to injected cell), secondary neighbors (adjacent to primary neighbors), and ternary neighbors (adjacent to secondary neighbors).

View larger version (47K): [in this window] [in a new window]   FIG. 2. Intracellular accumulation of Cx431 with N-cadherin knockdown. A, cells treated with either N-cadherin siRNAs (bottom panel) or control scrambled siRNAs (upper panel) were immunostained with an N-cadherin antibody. Note the absence of N-cadherin immunostaining with N-cadherin siRNA treatment (bottom panel), whereas in scrambled siRNAs-treated cells, N-cadherin expression remained abundant along cell processes delineating regions of cell-cell contact (see white arrows). B, Western immunoblotting revealed a reduction in N-cadherin protein abundance in cells transfected with N-cadherin siRNAs, but not with scrambled siRNA or mock-transfected control cells. Using mouse monoclonal P2C4 antibodies, which recognizes predominantly nonphosphorylated Cx431 (top panel) or rabbit polyclonal antibody recognizing all forms of Cx431(bottom panel), no difference was seen with N-cadherin knockdown in either total Cx431, or the faster migrating versus slower-migrating forms of Cx431. Tubulin antibody detection served as a loading control. C, cells transfected with N-cadherin siRNAs were double-immunostained with antibodies to N-cadherin (green) and Cx431 (red). Little or no N-cadherin expression was detected, whereas Cx431 was localized almost exclusively in the cytoplasm. Scale bar, 25 µm.

Time Lapse Videomicroscopy and Motion Analysis—For time lapse videomicroscopy, cells were maintained at 37 °C in L-15 medium supplemented with 10% fetal bovine serum (Hyclone), and visualized using Leica DMIRE2 inverted microscope with a x63 objective. Images were captured every 2 min over a 1-h interval using an ORCA-ER camera. Motion analysis was carried out using Dynamic Image Analysis Software (Solltech, Inc.) to compute: area, perimeter, roundness (100x 4 (area/perimeter²)) (29), speed, directionality (net path length divided by total path length), positive flow (percentage cell area in later images not overlapping earlier image), and negative flow (percentage cell area in earlier image not overlapping later image). Data were analyzed using Student's t test with Prism 3 (GraphPad).

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

 
Using immunofluorescence microscopy, we examined the distribution of Cx431 and N-cadherin in NIH3T3 cells. As expected, both proteins are localized in punctate spots at cell-cell contact sites along cell processes. Double immunolabeling with antibodies to Cx431 and N-cadherin showed regions of cell surface colocalization (Fig. 1 panel a). A similar pattern of colocalization was observed for Cx431 and p120, -catenin, and ZO-1 (Fig. 1, panels b–d). These results indicate a close association between Cx431, N-cadherin, and N-cadherin-interacting proteins in NIH3T3 cells. This is consistent with the results of previous studies in other cell types (24, 26, 30–36).

Immunoprecipitation and Western immunoblotting experiments were carried out to further characterize the interactions between Cx431 and N-cadherin (Fig. 1, panels e and g). Immunoprecipitations with a Cx431 antibody followed by Western immunoblotting with a Cx431 antibody yielded a predominant 43-kDa band, the molecular mass expected for Cx431 (Fig. 1, panel g). Immunodetection using antibodies to N-cadherin, -catenin, -catenin, and p120 showed that each of these proteins also coimmunoprecipitated with Cx431 (Fig. 1, panel e). In contrast, immunoblot analysis of Cx431 immunoprecipitates with antibodies to paxillin and focal adhesion kinase (FAK), two proteins found at focal complexes, showed neither proteins were associated with Cx431 (Fig. 1, panel f). To further examine the specificity of the interactions detected in the Cx431 immunoprecipitates, reciprocal experiments were carried out involving immunoprecipitations using antibodies to either N-cadherin, -catenin, -catenin, p120, or ZO-1. This was followed by Western immunoblotting using a Cx431 antibody (Fig. 1, panel g). These studies showed that Cx431 was indeed coimmunoprecipitated by antibodies to N-cadherin, -catenin, -catenin, p120, or ZO-1. Together these results indicate that Cx431 is localized in a large multiprotein complex containing N-cadherin and many N-cadherin-interacting proteins.

View larger version (61K): [in this window] [in a new window]   FIG. 3. Loss of cell surface Cx431 and N-cadherin with Cx431 knockdown. A, cells were immunostained with a Cx431 antibody 48 h after transfection with Cx431 siRNAs (bottom panel) or scrambled siRNAs (upper panel). Cx431 expression was greatly reduced in Cx431 siRNA-treated cells, whereas the control scrambled siRNA-treated cells show abundant Cx431 at regions of cell-cell contact (white arrow in top panel). B, Cx431 and N-cadherin protein expression after transfection with Cx431 siRNAs was analyzed by Western immunoblotting using antibodies to Cx431 and N-cadherin, with tubulin antibody serving as loading control. Cx431 protein abundance in Cx431 siRNA-transfected cells was substantially reduced, but N-cadherin protein abundance was not changed. C, cells transfected with Cx431 siRNAs were double-immunostained with antibodies to Cx431 (green) and N-cadherin (red). Little or no Cx431 expression was detected, whereas a low level of N-cadherin expression was observed in the cytoplasm. Scale bar, 25 µm.

To investigate this further, we carried out pull-down assays performed in reverse, i.e. with GST-N-cad-CT serving as prey, and His-Cx431-CT as bait. These studies also showed no direct interaction between the CT domains of Cx431 and N-cadherin (data not shown). We also performed GST-Cx431-CT affinity binding assays using total cell lysates as prey, and again found no evidence for N-cadherin binding (data not shown). Finally, Far-Western blotting was carried out using total protein extracts prepared from adult mouse heart, a tissue containing Cx431 and N-cadherin in abundance. The heart extract was separated by SDS-PAGE, blotted onto a polyvinylidene difluoride membrane, then incubated with GST-Cx431-CT. Immunodetection with a GST antibody showed no band in the position corresponding to N-cadherin (data not shown). Overall, these experiments suggest no direct interaction between the CT domains of Cx431 and N-cadherin.

View larger version (74K): [in this window] [in a new window]   FIG. 4. Cx431 siRNA treatment alters distribution of N-cadherin-associated proteins. Cells transfected with Cx431 siRNAs were double-immunostained for Cx431 (panels a, c, e) and either p120 (panel b), -catenin (panel d), or ZO-1 (panel f). Cx431 immunostaining is shown in green and all other antibodies are shown in red. In Cx431 siRNA-treated cells, little Cx431 expression was detected. Cell surface expression of p120 (panel b) and -catenin (panel d) were also greatly reduced or eliminated. In contrast, ZO-1 continued to be expressed abundantly, and was found in small puncta along regions of cell-cell contact (panel f). Note the prominent nuclear localization of p120 as well as some cytoplasmic p120 immunostaining. Scale bar, 25 µm.

View larger version (81K): [in this window] [in a new window]   FIG. 5. N-cadherin siRNA treatment alters expression of N-cadherin-associated proteins. Control cells were transfected with scrambled siRNAs and double-immunostained for N-cadherin and N-cadherin-associated proteins, including -catenin (panel a), -catenin (panel c), and ZO-1 (panel e). Each of these N-cadherin-associated proteins (red) shows extensive colocalization with N-cadherin (green) at regions of cell-cell contact (yellow puncta denoted by white arrows). In cells transfected with N-cadherin siRNAs, N-cadherin-deficient cells also showed little or no -catenin (panel b) or -catenin (panel d) at the cell surface. ZO-1 expression, though persisting, was significantly reduced (panel f). Images were generated by merging the phase contrast and deconvolved darkfield immunofluorescence images. Scale bar, 25 µm.

To determine if the loss of Cx431 may reciprocally perturb N-cadherin cell surface expression, NIH3T3 cells were transiently transfected with two siRNAs targeting different regions of Cx431. Immunostaining showed a marked reduction in Cx431 expression (Fig. 3A), with >80% reduction in Cx431 protein abundance indicated by Western immunoblotting analysis (Fig. 3B). In contrast, two scrambled siRNAs had no effect (Fig. 3, A and B). When Cx431 siRNA-treated cells were double-immunostained with Cx431 and N-cadherin antibodies, N-cadherin was observed predominantly in the cytoplasm in the Cx431-deficient cells (Fig. 3C). Analysis by Western immunoblotting showed no significant change in N-cadherin protein abundance in the Cx431 siRNA-treated cells (Fig. 3B). The combined results of these N-cadherin and Cx431 siRNA studies suggest that the cell surface expression of Cx431 requires N-cadherin, and conversely, the cell surface expression of N-cadherin also appears to require Cx431.

Cx431 siRNA Alters Distribution of N-cadherin-associated Proteins—To determine if Cx431 knockdown may also affect the distribution of N-cadherin-associated proteins, cells transfected with Cx431 siRNAs were double-immunostained with antibodies to Cx431 and p120. Cells deficient in Cx431 showed a marked reduction in cell surface expression of p120 (Fig. 4, panels a and b), whereas p120 expression in the cytoplasm and nuclei were elevated (compare Figs. 4, panel b with 1, panel b). These results are reminiscent of our previous finding that p120 is redistributed to the nuclei in Cx431 knockout neural crest cells (24). We also observed with Cx431 knockdown, the loss of cell surface -catenin (Fig. 4, panels c and d). A parallel analysis of N-cadherin siRNA-treated cells revealed a similar loss or reduction of cell surface expression of -catenin, -catenin, as well as p120 with N-cadherin knockdown (Fig. 5, panels b and d, and data not shown). Despite the overall similarities between the effects of Cx431 versus N-cadherin siRNA treatment on the reduction in expression of N-cadherin-associated proteins, some differences were noted. Thus Cx431 siRNA-transfected cells showed the retention of more cytoplasmic -catenin (Figs. 4, panel d versus 5, panel d), whereas N-cadherin, but not Cx431 siRNA, caused an apparent reduction in ZO-1 expression (Figs. 4, panel f versus 5, panel f). In control studies carried out with scrambled siRNA treatments, we observed no change in the usual pattern of cell surface colocalization of N-cadherin with -catenin, -catenin, and ZO-1 (Fig. 5, panels a, c, and e).

View larger version (35K): [in this window] [in a new window]   FIG. 6. Lysosomal turnover and cell surface trafficking of Cx431 in N-cadherin siRNA-transfected cells. A, cells transfected with N-cadherin siRNAs or control scrambled siRNAs were cell surface-biotinylated, lysed, and the cell surface Cx431 was recovered by immunoprecipitation using streptavidin-coated agarose beads. Some cells were first treated with 5 or 10 mM ammonium chloride, a lysosomal inhibitor, prior to surface biotinylation. In control scrambled siRNA-transfected cells, ammonium chloride treatment elevated total and cell surface localized Cx431. Untreated N-cadherin siRNA-transfected cells when compared with scrambled siRNA-transfected cells showed a reduction in cell surface expression of Cx431 even as total Cx431 expression levels remained unchanged. With ammonium chloride treatment, there was either no change or a small decrease in cell surface expression of Cx431. B, Cx431 cell surface trafficking was examined by pulse chase labeling of newly synthesized Cx431. N-cadherin or scrambled siRNA-transfected cells were labeled with [35S]methionine for 20 min followed by a chase interval of up to 4 h. Surface biotinylated Cx431 was isolated, and gel separated and blotted, and imaged using a phosphorimager to visualize the cell surface 35S-labeled Cx431. In scrambled siRNA-treated cells, the Cx431 bands show strong labeling up to 3 h, with a visible decrease in intensity detected at 4 h. Quantification of the blot provided a half-life estimate of 2.8 h versus 2.3 h for the slower migrating versus faster-migrating species of Cx431. In the N-cadherin siRNA-treated cells, these Cx431 bands show only background levels of labeling, with a decrease in intensity also seen at 4 h. The latter may correspond to Cx431 turnover associated with cells that were not transfected by the N-cadherin siRNA.

To determine if the reduction in cell surface-localized Cx431 in N-cadherin-deficient cells is due to altered Cx431 cell surface trafficking, we used pulse chase labeling together with cell surface biotinylation to track the cell surface translocation of newly synthesized Cx431. For these studies, cells were [³⁵S]methionine labeled for 20 min, and at timed intervals spanning 4 h, surface biotinylation was carried out followed by harvesting of biotinylated Cx431. After Western blotting, the radiolabeled biotinylated Cx431 was then visualized using a phosphorimager. In cells transfected with scrambled siRNA, [³⁵S]methionine-labeled Cx431 was detected at the cell surface soon after labeling, but by 4 h, the intensity of labeling was decreased (Fig. 6B, upper panel). This indicates that the pulse-labeled Cx431 moved off the cell membrane between 3 and 4 h after translocation to the cell surface. This is consistent with previous reports of a half-life of 1–5 h for Cx431 in cultured cells (for review, see Ref. 6). In contrast, in cells transfected with N-cadherin siRNAs, over the same 4-h chase interval only baseline levels of [³⁵S]methionine labeling were seen in the cell surface pool of Cx431. This would suggest that with N-cadherin knockdown, most newly synthesized Cx431 failed to traffic to the cell surface (Fig. 6B, bottom panel). The observed residual cell surface Cx431 may represent Cx431 expression in cells not successfully transfected with the N-cadherin siRNAs. Consistent with this, we note this residual cell surface Cx431 band decreased in intensity at 4 h, similar to the profile in control scrambled siRNA-transfected cells and likely reflects normal Cx431 turnover. Overall, these results indicate that Cx431 cell surface trafficking is inhibited by N-cadherin knockdown.

View larger version (66K): [in this window] [in a new window]   FIG. 7. Differential effects of p120 siRNA knockdown on Cx431 versus N-cadherin expression. A, Western immunoblot analysis of cells transfected with a p120-siRNA expression vector significant showed a significant decrease in p120 protein abundance. B, cells transfected with p120 siRNA were immunostained with antibodies to p120, Cx431, or N-cadherin. In control cells, p120 is abundantly expressed at regions of cell-cell contact (white arrows), whereas little cell surface expression was seen in p120 siRNA-treated cells. In siRNA-transfected cells, N-cadherin continued to be found at regions of cell-cell contact (white arrows), but its abundance was greatly reduced. Surprisingly, Cx431 expression was unchanged (see white arrows). Scale bar, 10 µm. C, reverse transcriptase-PCR analysis for expression of ARVCF or -catenin transcripts in NIH3T3 cells. These assays showed ARVCF and -catenin transcripts are both expressed in NIH3T3 cells. Thus a product of the expected size was observed after reverse transcription and PCR amplification with primer pairs for both genes, whereas no products were obtained in the absence of reverse transcriptase (Control).

View larger version (44K): [in this window] [in a new window]   FIG. 8. Knockdown of Cx431 or N-cadherin inhibit cell-cell communication. A, darkfield fluorescence images showing dye spread in siRNA-transfected NIH3T3 cells after intracellular microelectrode impalement and iontophoretic injection of 6-carboxyfluorescein. The extent of dye spread was significantly reduced in Cx431 or N-cadherin siRNA-transfected cells as compared with scrambled siRNA-transfected cells. White outlines delineate the position of cells seen using phase contrast optics. Scale bar, 50 µM. B, bar graphs show the number of primary, secondary, and ternary neighboring cells to which the injected dye has spread from the site of impalement. Note the marked decrease in dye spread in the Cx431 and N-cadherin siRNA-transfected cells. Error bar in B is S.E. (*, p < 0.05).

View larger version (144K): [in this window] [in a new window]   FIG. 9. Time lapse images show reduced cell motility with Cx431 and N-cadherin knockdown. Cells transfected with EGFP expression vector alone or in combination with Cx431 or N-cadherin siRNAs were analyzed by time lapse imaging. Shown are phase-contrast images taken at 30-min intervals starting 48 h after siRNA transfection. Cells treated with Cx431 or N-cadherin siRNAs showed less displacement than control cells (see position of arrows). Scale bar, 25 µm.

N-cadherin and Cx431 Knockdown Inhibits Cell Motility— We also examined the motile behavior of NIH3T3 cells transfected with Cx431 or N-cadherin siRNA, as previous studies indicated a role for Cx431 and N-cadherin in modulating cell motility (24, 40, 41). For these studies, cells were transfected with an EGFP expression vector together with siRNA for either Cx431 or N-cadherin, and the motile behavior of individual cells was examined by time lapse videomicroscopy, with phase contrast and darkfield images captured at 4-min intervals over a period of 1 h. EGFP expression visualized in darkfield images was used to identify cells that are likely cotransfected with the Cx431 or N-cadherin siRNAs. Control cells showed dynamic changes in cell shape and position, whereas Cx431 or N-cadherin siRNA-transfected cells exhibited less cell shape changes and little net displacement (Figs. 9 and 10B). Quantitative assessment of cytoplasmic protrusions and retractions, referred to as positive and negative flow, respectively, revealed a reduction in protrusive activity with either Cx431 or N-cadherin knockdown, but we noted a greater reduction in both positive and negative flow with N-cadherin knockdown (Fig. 10, A and C). Consistent with the reduced protrusive activity, we also found an increase in the roundness of the cells, which is calculated from the cell area and perimeter (see "Experimental Procedures," Fig. 10C). With either Cx431 or N-cadherin siRNA treatment, we also observed a reduction in the overall speed of cell locomotion (Fig. 10C), whereas the directionality of cell movement was unchanged (data not shown). The combined reduction in the speed of cell locomotion and protrusive activity is expected to reduce net cell movement, and this is indeed evident, either by examining the migratory paths of individual cells over the 60-min interval (Fig. 10B), or by comparing the relative cell position in individual time lapse frames (Fig. 9). Overall, these observations show that Cx431 and N-cadherin siRNA treatment inhibited cell motility in a similar manner, by inhibiting protrusive cell activity and reducing the speed of locomotion.

View larger version (29K): [in this window] [in a new window]   FIG. 10. Motion analysis of cells treated with Cx431 and N-cadherin siRNA. A, shown are time lapse image stacks obtained from individual control (blue), Cx431 siRNA (green), and N-cadherin-siRNA (red)-treated NIH3T3 cells captured at 4-min intervals over 1 h. The image on top represents the cell at time 0, with the successive time points indicated by the underlying gray images. Note the much greater net displacement for the control cell, suggesting that N-cadherin or Cx431 siRNA treatment inhibited cell motility. B, actual migratory paths of the cells shown in A are plotted from a common origin. Much shorter migratory paths were observed for the N-cadherin and Cx431 siRNA-treated cells as compared with the control cell, indicating a reduction in cell motility with N-cadherin or Cx431 siRNA treatment. C, effects of siRNA treatment were quantitatively assessed by measuring the roundness of cells, the speed of cell locomotion, and positive and negative cytoplasmic flow. A reduction was observed in the speed of cell locomotion, and in positive/negative flow. In contrast, the roundness of the cells was increased. These findings suggest a reduction in protrusive activity. Error bar in C corresponds to S.E. When compared with control: *, p < 0.05; **, p < 0.01; ***, p < 0.005. When compared with Cx431 siRNA: +, p < 0.05.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Recent studies have indicated that N-cadherin coassembles with catenins in the ER/Golgi compartments prior to its cell surface translocation (43). Interestingly, the forced expression of -catenin in an -catenin-deficient prostate cancer cell line has been reported to rescue Cx431 and Cx321 trafficking to the cell surface (44). Moreover, targeting of Cx431 to the plasma membrane in cardiomyocytes have been shown to require the association of Cx431 with -catenin and ZO-1 (35). These findings further point to interactions between Cx431 and N-cadherin/N-cadherin-associated proteins playing a critical role in gap junction formation. However, the precise role of cadherins in gap junction formation may be context or cell-type dependent. Thus in a previous study, the ectopic expression of N-cadherin in L cells was reported to inhibit Cx431 localization at cell junctions, even as N-cadherin expression in hepatoma cells enhanced gap junction formation (45). Analysis of Cx431 knockout cardiomyocytes also showed no change in the expression of cadherins and catenins (46), nor were any changes seen in Cx431 expression in N-cadherin knockout neural crest cells (24). In contrast, N-cadherin knockout cardiomyocytes exhibited significant reductions in the expression of p120, -catenin, and Cx431 (18).

We noted some nonreciprocal effects elicited by N-cadherin versus Cx431 siRNA treatment. For example, ZO-1 expression was reduced with N-cadherin knockdown, but not with Cx431 knockdown. With both N-cadherin and Cx431 knockdown, cell surface expression of -catenin was eliminated, but this was accompanied by an accumulation of cytoplasmic -catenin only with Cx431 knockdown. These observations suggest a possible hierarchy in interactions between Cx431, N-cadherin, and N-cadherin-associated proteins. One discrepancy we noted in our studies was that N-cadherin immunostaining was not detected in Cx431 siRNA-transfected cells even though N-cadherin protein expression was evident in Western immunoblots. We suggest this may reflect a failure of N-cadherin to cluster in Cx431-deficient cells, making it difficult to detect by immunohistochemistry. This same discrepancy was noted with p120 siRNA knockdown, which disrupted N-cadherin cell surface clustering without affecting total N-cadherin protein expression level (25). A similar situation was previously reported for emerging chick neural crest cells, which showed abundant N-cadherin protein expression by Western immunoblotting but not by immunohistochemistry (47). These observations suggest that perhaps the cell surface clustering of N-cadherin may require both p120 and Cx431, a possibility that will be further investigated in future studies. It should be noted that p120 has been shown to play an important role in modulating the cell surface turnover of N-cadherin (25, 39, 48). Our p120 knockdown experiments showed that the loss of p120 expression was accompanied by a reduction in the cell surface expression of N-cadherin, but with no effects on Cx431 expression. This suggests that N-cadherin turnover may be regulated independently of Cx431.

We also showed that Cx431 and N-cadherin knockdown inhibited cell motility. Studies by others have shown that N-cadherin expression can enhance cell motility and also promote tumor metastasis (40, 41). The mechanism by which Cx431 and N-cadherin modulates cell motility is not known, although it is interesting to note that in Cx431 knockout neural crest cells and NIH3T3 cells transfected with Cx431 siRNA, p120 appeared to redistribute from the cell surface to the nucleus (24). Hence, alterations in p120/Rho-GTPase signaling may have a role in mediating cross-talk between the actin cytoskeleton and these two specialized cell junctions (24). Although our studies showed that Cx431 or N-cadherin knockdown had similar effects on NIH3T3 cell motility, cell protrusive activity was inhibited to a greater extent with N-cadherin knockdown. We previously found that cell motility was inhibited in both N-cadherin and Cx431 KO neural crest cells, but the precise change in motile behavior differed (24). The basis for the differential effects of Cx431 versus N-cadherin knockdown/knockout on cell motility is not known, but among various possibilities, it could reflect a hierarchical relationship in connexin-cadherin interactions. Together these observations suggest that interactions between connexin and cadherin may play an important role in modulating motile cell behavior.

Overall, these studies indicate that Cx431 cell surface trafficking and gap junction formation are dependent on the coassembly of Cx431 in an N-cadherin-containing multiprotein complex. This process of coassembly may account for the close linkage between gap junction and adherens junction formation. Using GST and His-tagged fusion protein binding assays, we found no evidence for direct binding between Cx431 and N-cadherin, but weak interactions would have been difficult to detect in such in vitro assays. The CT domain of Cx431 protein has been shown to bind proteins such as -catenin and ZO-1 (26, 31, 33, 36). As the latter proteins also are known to bind N-cadherin, such interactions could serve as a bridge to anchor Cx431 to N-cadherin-containing protein complexes. As our fusion protein construct included only the CT domain of Cx431, it leaves open the possibility of interactions mediated by the short N terminus or intracellular loop domains of Cx431. It is interesting to note that recent studies in Drosophila showed that the insect gap junction protein innexin 2 can interact with core proteins in adherens and septate junctions, and likely play an essential role in epithelial morphogenesis (49). Together with our present findings, it would suggest that interactions between connexin and cadherins may be conserved in evolution and may have an important role in regulating cell adhesion, cell-cell communication, and cell motility.

	FOOTNOTES

 
^* This work was supported by National Institutes of Health Grant Z01-HL005701. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

To whom correspondence should be addressed: 50 South Dr. MSC 8019, Rm. 4537, Bethesda, MD 20892-8019. Tel.: 301-451-8041; Fax: 301-480-4581; E-mail: loc{at}nhlbi.nih.gov.

¹ The abbreviations used are: Cx431, connexin43; N-cadherin, neuronal cadherin; p120, p120-catenin; siRNA, small interfering RNA; GST, glutathione S-transferase; FITC, fluorescein isothiocyanate; CT, cytoplasmic terminus; NP, nonphosphorylated.

	ACKNOWLEDGMENTS

 
We thank P. Lampe for providing antibody reagents and D. Walker for helpful suggestions.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

 

1. Goodenough, D. A., Goliger, J. A., and Paul, D. L. (1996) Annu. Rev. Biochem. 65, 475–502[CrossRef][Medline] [Order article via Infotrieve]
2. Kumar, N. M., and Gilula, N. B. (1996) Cell 84, 381–388[CrossRef][Medline] [Order article via Infotrieve]
3. Sohl, G., and Willecke, K. (2003) Cell Commun. Adhes. 10, 173–180[Medline] [Order article via Infotrieve]
4. Musil, L. S., and Goodenough, D. A. (1993) Cell 74, 1065–1077[CrossRef][Medline] [Order article via Infotrieve]
5. Yeager, M., Unger, V. M., and Falk, M. M. (1998) Curr. Opin. Struct. Biol. 8, 517–524[CrossRef][Medline] [Order article via Infotrieve]
6. Berthoud, V. M., Minogue, P. J., Laing, J. G., and Beyer, E. C. (2004) Cardiovasc. Res. 62, 256–267[Abstract/Free Full Text]
7. Musil, L. S., Le, A. C., VanSlyke, J. K., and Roberts, L. M. (2000) J. Biol. Chem. 275, 25207–25215[Abstract/Free Full Text]
8. Laing, J. G., and Beyer, E. C. (1995) J. Biol. Chem. 270, 26399–26403[Abstract/Free Full Text]
9. Laing, J. G., Tadros, P. N., Westphale, E. M., and Beyer, E. C. (1997) Exp. Cell Res. 236, 482–492[CrossRef][Medline] [Order article via Infotrieve]
10. Herve, J. C., Plaisance, I., Loncarek, J., Duthe, F., and Sarrouilhe, D. (2004) Eur. Biophys. J., 33, 201–210[Medline] [Order article via Infotrieve]
11. Lampe, P. D., and Lau, A. F. (2004) Int. J. Biochem. Cell Biol. 36, 1171–1186[CrossRef][Medline] [Order article via Infotrieve]
12. Musil, L. S., and Goodenough, D. A. (1991) J. Cell Biol. 115, 13