Isolation of Isogenic Subclones from RL-CL9 Cells Expressing E-Cadherin and N-Cadherin

One hundred RL-CL9 cells from passages 4 and 30 were seeded in 10-cm dishes in complete culture medium and allowed to grow into colonies. Between 63–100 clones from early and late passage cells were isolated using glass cylinders and expanded and frozen. Each clone was analyzed separately by Western blot and immunocytochemical analysis for the expression of E-Cad, N-Cad, Cx43 as well as for subcellular localization. Subclones that expressed only E-Cad and no detectable level of N-Cad were designated as RL-EΔN cells whereas subclones that expressed N-Cad and no detectable level of E-Cad were designated as RL-NΔE cells. Further characterization of these clones is described in Results.

Plasmids, Retroviral Vectors, and Other Recombinant DNA Constructs

We used the following retroviral vectors for our studies. 1. LXSN, LZRS-neo and LZRS-pac (control retroviruses). 2. LXSNCx32 harboring rat Cx32. 3. LXSNE-Cad (human), which was constructed by subcloning E-Cad into the EcoR1 and BamH1 sites of LXSN. 4. LZRS-E-Cad-Myc and LZRS-N-Cad-Myc expressing Myc-tagged human E-Cad and N-Cad. 5. ShN-Cad, ShP-Cad, and ShEGFP in pSuper.Retro.puro. The construction of these retroviral vectors has been described (Mehta et al., 1999; Nieman et al., 1999; Kim et al., 2000a; Govindarajan et al., 2002; Maeda et al., 2005; Maeda et al., 2006; Mitra et al., 2006; Shintani et al., 2008). GFP-tagged Rab5 and Rab7 and their constitutively active mutants, Rab5Q-L and Rab7Q-L, were kind gifts from Dr. Caplan. These retroviral vectors were used to produce recombinant retroviruses in HEK293T, PA317 and PTi67 amphotropic packaging cell lines as described (Mitra et al., 2006; Chakraborty et al., 2010). RL-CL9 cells were multiply (2–4 times) infected with various recombinant retroviruses and selected in either G418 (400 μg/ml) or puromycin (2 μg/ml) for 2–3 wk in complete medium. Pooled cultures from 1000–2000 colonies obtained from three to four dishes were expanded, frozen, and maintained in G418 (200 μg/ml) and in G418 and puromycin (2 μg/ml). Pooled polyclonal cultures were used within two to three passages for immunocytochemical and biochemical analyses.

Cell Surface Biotinylation Assay

Cells (2 × 10⁵) were seeded in 6-cm dishes in 4 ml of complete medium and grown to confluence. Cell surface biotinylation was performed at 4°C by incubation with freshly prepared EZ-LinkSulfo-NHS-SS Biotin reagent (Pierce; Rockford, IL) at 0.5 mg/ml in phosphate buffered saline (PBS) supplemented with 1 mM CaCl₂ and 1 mM MgCl₂ (PBS-PLUS) for 1 h as described (VanSlyke and Musil, 2000; Chakraborty et al., 2010). The reaction was quenched with PBS-PLUS containing 20 mM glycine and cell lysis and affinity precipitation of biotinylated proteins were performed with 100 μg of total protein using 50 μl of streptavidin agarose beads (Pierce, Rockford, IL) on a rotator overnight at 4°C. The streptavidin-bound biotinylated proteins were eluted by incubating the beads for 30 min in 1× SDS loading buffer and resolved by SDS-PAGE followed by Western blotting. As an input, equal amount of total protein (10 μg) was also subjected to SDS-PAGE and Western blot analysis. To determine the kinetics of degradation of cell-surface associated Cx43, cells were biotinylated at 4°C as described above and, after washing and quenching biotin, were chased at 37°C for various times before affinity precipitation of the biotinylated proteins with streptavidin. The protein concentration was determined using BCA reagent (Pierce).

Detergent Extraction and Western Blot Analysis of Cx43 and Cx32

Cell lysis, detergent solubility assay with 1% Triton X-100 (TX-100), and Western blot analysis were performed as described (VanSlyke and Musil, 2000; Mitra et al., 2006; Chakraborty et al., 2010). Briefly, 5 × 10⁵ RL-EΔN and RL-NΔE cells were seeded per 10-cm dish in 10 ml of complete medium and grown to confluence. Cells were then lysed in buffer SSK (10 mM Tris, 1 mM EGTA, 1 mM PMSF, 10 mM NaF, 10 mM NEM, 10 mM Na₂VO₄, 10 mM iodoacetamide, 0.5% TX-100, pH 7.4) supplemented with the protease inhibitor cocktail (Sigma, St. Louis, MO). To determine the detergent solubility of Cx43, the concentration of TX-100 was raised to 1% before ultracentrifugation at 100,000 × g for 60 min (35,000 rpm in analytical Beckman ultracentrifuge; Model 17–65 using a SW50.1 rotor). The detergent-insoluble pellets were dissolved in buffer C (70 mM Tris/HCl, pH 6.8, 8 M urea, 10 mM NEM, 10 mM iodoacetamide, 2.5% SDS, and 0.1 M DTT). After normalization based on cell number, the total, TX-100-soluble, and -insoluble fractions were mixed with 4× SDS-loading buffer to a final concentration of 1× and boiled at 100°C for 5 min (for Cx43) or incubated at room temperature for 1 h (for Cx32) before SDS-PAGE analysis.

Detergent (TX-100) Extraction of Cells In Situ

RL-EΔN and RL-NΔE cells were extracted in situ with 1% TX-100 essentially as described previously (Govindarajan et al., 2002; Chakraborty et al., 2010). In brief, 2.5 × 10⁵ cells were seeded per well in six-well clusters containing glass cover slips and were grown for 2–3 d when they attained confluence. The cells were rinsed once in PBS and then incubated in isotonic medium (30 mM HEPES, pH 7.2; 140 mM NaCl; 1 mM CaCl₂, 1 mM MgCl₂) supplemented with the protease inhibitor cocktail (Sigma, MO) for 45 min at 4°C in the presence and absence of 1% TX-100. The dishes were swirled gently and intermittently. After incubation, cells were fixed and immunostained with appropriate primary and secondary antibodies as described below. The detergent-solubility of Cxs in single RL-EΔN and RL-NΔE cells was examined 24–36 h after seeding between 1 and 2 × 10³ cells per well in six-well clusters. Under these conditions between 50 and 60% of seeded cells remained single and well-spread.

Antibodies and Immunostaining

Rabbit polyclonal and mouse monoclonal antibodies against Cx43 and Cx32 have been described earlier (Mehta et al., 1999). Rabbit anti-EEA-1 (PA1–063) was obtained from Affinity BioReagents (Golden, CO). Rabbit anti–caveolin-1 and anti–caveolin-2 were purchased from BD Transduction laboratories (San Jose, CA). Mouse anti-occludin (clone OC-3F10) was from Zymed Laboratories (South San Francisco, CA). Rabbit anti–α-catenin, rabbit anti–β-catenin, rabbit anti-Cx32 (loop and tail), and mouse anti–β-actin (clone C-15) antibodies were from Sigma (St. Louis, MO). Mouse anti-ZO1 (610967) and mouse anti-GM130 (610822) antibodies were obtained from BD Transduction laboratories (San Jose, CA). Mouse anti-clathrin heavy chain (MAI-065) was purchased from Affinity Bioreagents (Rockford, IL). Mouse anti-N-Cad, mouse anti-E-Cad, mouse anti–α-catenin, and mouse anti–β-catenin antibodies have been previously described (Maeda et al., 2005; Mitra et al., 2006; Shintani et al., 2008; Chakraborty et al., 2010). We also used rabbit anti–E-Cad antibody for immunostaining rat E-Cad. The production of this antibody has been described (Wheelock et al., 1987). Alexa-488 and Alexa-594 phalloidins as well as Alexa-488 and Alexa-594 conjugated cholera toxin, transferrin, and EGF were purchased from Invitrogen (Molecular Probes/Invitrogen, Eugene, OR).

RL-EΔN and RL-NΔE cells (see Results), seeded in six-well clusters at a density of 2.5 × 10⁴/well and allowed to grow to confluence, were immunostained at room temperature with various antibodies at appropriately calibrated dilutions after fixing with 2% para-formaldehyde for 15 min as described previously (Mitra et al., 2006; Chakraborty et al., 2010). Secondary antibodies (rabbit or mouse) conjugated with Alexa-488 and Alexa-594 were used as appropriate. Images of immunostained cells were acquired with a Leica DMRIE microscope (Leica Microsystems, Wetzler, Germany) equipped with Hamamatsu ORCA-ER CCD camera (Hamamatsu City, Japan). For colocalization studies, serial z sections (0.5 μm) were collected and analyzed after iterative deconvolution using image-processing software (Volocity; Improvision, Lexington, MA). SlowFade antifade (Molecular Probes/Invitrogen) was used to mount cells on glass slides.

Cell Growth on Transwell Filters

RL-CL9, RL-EΔN, and RL-NΔE cells (2 × 10⁴) were plated onto 12-mm transwell filters (pore size, 0.4 μm; Corning Life Sciences, MA) and grown for 7–21 d as described (Chakraborty et al., 2010). Medium of the upper and lower filter chambers was changed on alternate days. The immnunocytochemical analyses were performed directly on filters, after which the filters were cut with a sharp scalpel and mounted on glass slides with their cell side facing up. A drop of SlowFade antifade was placed on top followed by a glass coverslip, which was pressed gently using a 50 gm weight overnight. The edges were sealed with nail polish.

Cell Aggregation Assay

The hanging drop suspension culture method was used to test aggregation of RL-EΔN and RL-NΔE cells as described (Chakraborty et al., 2010). After harvesting with trypsin/EDTA, cells were resuspended at 2.5 × 10⁵ cells per ml in complete culture medium and a 20-μl drop of medium containing 5 × 10³ cells was suspended as a hanging drop from the lid of a 10 cm² Petri dish (40 hanging drops per lid) filled with PBS to prevent drying of the drops. The cells were then allowed to aggregate in a humidified 5% CO₂ incubator at 37°C for 16–20 h after which cells and aggregates were triturated five times with a standard 200 μl pipette tip to disperse loosely associated cells and a coverslip was placed gently on top of the cells. The cells were visualized by phase-contrast light microscopy at ×5 magnification and images captured using a CCD camera (Retiga 2000R, FAST 1394) mounted on a Leica DMRIE2 microscope with the aid of Volocity software (Improvision, Lexington, MA). For quantifying the size, the outline of each aggregate was drawn using an ROI tool and the area of each aggregate was measured. All images were analyzed at the same magnification and the area was recorded as “relative units” for each aggregate. The average area of the aggregates was considered as a measure of aggregate formation.

Cell Migration Assay

RL-EΔN and RL-NΔE cells were seeded on LabTek two-well chamber slides and allowed to grow to confluence. Confluent monolayers of cells were wounded by manually scratching with a pipette tip. We chose scratch areas of equal width for these assays. After marking the positions of the wounds in the X-Y plane, cells were imaged every 30 min at 37°C in an atmosphere of 5% CO₂/95% air for 24 h in a live cell imaging chamber mounted on an Olympus IX81 motorized inverted microscope (Olympus America Inc; Center Valley, PA). The microscope was controlled by IX2-UCB U-HSTR2 motorized system with a focus drift compensatory device IX1-ZDC. Images were captured using a Hamamatsu ORCA ER2 CCD camera and processed by imaging software Slidebook version 5.0 (Intelligent Imaging Innovations, Denver, CO).

Gap Junctional Communication Assays

Gap junctional communication was measured by microinjecting fluorescent tracers, Alexa Fluor 488 (MW 570 Da; A-10436) and Alexa Fluor 594 (MW 760 Da; A-10438), and by scrape loading as described (Govindarajan et al., 2002; Chakraborty et al., 2010). Fluorescent tracers were microinjected with an Eppendorf InjectMan and FemtoJet microinjection systems (Models 5271 and 5242, Brinkmann Instrument, Westbury, NY) mounted on a Leica DMIRE2 microscope. The images of microinjected cells were captured using a CCD camera (Retiga 2000R, FAST 1394) with the aid of Volocity software (Improvision, Lexington, MA) one and five minutes after microinjection for RL-EΔN and RL-NΔE cells, respectively, and the number of fluorescent cells, excluding the microinjected one, was counted from the captured images that were stored as TIFF files as described previously (Govindarajan et al., 2002; Chakraborty et al., 2010).

Cells were scrape-loaded as described (Govindarajan et al., 2002; Chakraborty et al., 2010). Briefly, RL-EΔN and RL-NΔE cells were seeded in six-well clusters at a density of 2.5 × 10⁴/well and allowed to grow to confluence. Cell culture medium from freshly confluent cells was removed and replaced with 1 ml of medium containing rhodamine-conjugated fluorescent dextrans (10 kDa, 1 mg/ml; fixable) and Lucifer yellow (0.5%). Cells were scrape-loaded using a sterile scalpel by two longitudinal scratches and incubated for one minute at room temperature. Cells were washed quickly two to three times with warm PBS containing calcium and magnesium and returned to the incubator for five minutes, after which the medium was removed, cells were washed two times with PBS and fixed with 3.7% buffered Formalin at room temperature. The autofluorescence of cells was quenched with 0.1 M glycine and after washing once with PBS, the cover slips were mounted on glass slides in a droplet of SlowFade. Images of the scrape-loaded cells were captured as described above.

Altered Cadherin Expression and Gap Junction Assembly

We fortuitously observed that upon serial propagation of RL-CL9 cells from passage 4 to greater than passage 20, a population of cells emerged that began to lose typical cobblestone-like epithelial morphology (Figure S-1, top row). Moreover, immunocytochemical analysis showed that this population of cells began to express N-Cad (Figure S-1, middle row, left panel) in addition to E-Cad (Figure S-1, bottom row, middle panel). Furthermore, in this population of cells Cx43 was not assembled into GJs and was retained in the cytoplasm (Figure S-1, middle row). In addition, E-Cad expression at the areas of cell–cell contact of these cells appeared to be less robust and at times discontinuous (Figure S-1, bottom row, arrows). On systematic examination, we found that serial propagation of early passage RL-CL9 cells for 4–6 mo was gradually accompanied by an increase in the expression level of N-Cad (Figure S-2A, lower panels; Figure S-2C, blots) and vimentin (Figure S-2B, upper panels; Figure S-2C, blot), an increase in the proportion of cells in which actin was organized into stress fibers (Figure S-2B, bottom panels), and an increase in the proportion of cells in which Cx43 remained intracellular and failed to form GJs (not shown). Moreover, the expression level of E-Cad was not significantly altered (Figure S-2C), although some E-Cad began to accumulate in the cytoplasm (Figure S-2A, upper panels).

Gap Junction Assembly in E-Cadherin and N-Cadherin Expressing RL-CL9 Isogenic Subclones

To test whether gain of N-Cad expression was associated with the loss of GJ forming ability, we isolated between 63 and 92 clones from early and late passage RL-CL9 cells and examined the expression level and localization of Cx43, E-Cad, and N-Cad by immunocytochemical and Western blot analyses in each clone. We found that 70% of clones isolated from early passage (passage 4) RL-CL9 cells expressed only E-Cad and no detectable level of N-Cad, whereas 79% of clones isolated from late passage (passage 20) cells expressed only N-Cad and no detectable levels of E-Cad (Table S-1). Moreover, as assessed immunocytochemically, 61% of clones isolated from early passage RL-CL9 cells formed GJs whereas only 25% of clones isolated from late passage cells formed GJs (Table S-1). Furthermore, in clones in which both N-Cad and E-Cad were expressed to varying level, a fraction of Cx43 remained intracellular whereas a fraction was assembled into GJs (data not shown). For further study, we selected two E-Cad expressing clones lacking N-Cad expression, designated as RL-EΔN, and two N-Cad expressing clones lacking E-Cad expression, designated as RL-NΔE, from early passage RL-CL9 cells to minimize the phenotypic and epigenetic changes that might have ensued upon serial propagation of early passage cells in vitro and which might have occurred in late passage RL-CL9 cells. Qualitatively similar data were obtained with both clones selected from each category and data from only one representative clone are shown in subsequent studies.

Gap Junction Assembly and Function in RL-EΔN and RL-NΔE Cells

As assessed by Western blot analysis, RL-EΔN cells expressed only E-Cad and no N-Cad, whereas RL-NΔE cells expressed only N-Cad and no E-Cad; moreover, the expression level of Cx43, α- and β-catenins, occludin and ZO-1, and cytokeratin 8 and vimentin was not significantly different (Figure 2, A and B). In RL-EΔN cells, Cx43 was assembled into GJs, whereas in RL-NΔE cells it was predominantly intracellular as discrete puncta although small puncta, which were barely detectable immunocytochemically, were also observed at cell–cell contact regions (Figure 2 B). Although no discernible difference in the expression level of actin was observed between RL-EΔN and RL-NΔE cells, actin was organized into cortical bundles in RL-EΔN cells whereas it was organized into more stress fibers in RL-NΔE cells as was observed in early passage and late passage RL-CL9 cells (data not shown, but see Figure 1 and Figure S-2B). Moreover, as was observed in early passage RL-CL9 cells, when grown on transwell filters, RL-EΔN cells grew as a monolayer, became columnar, and expressed ZO-1 robustly at the areas of cell–cell contact (Figure 2C, top row), whereas RL-NΔE cells formed a monolayer in which cell borders seemed to partially overlap. Furthermore, the pattern of localization of ZO-1 at areas of cell–cell contact was discontinuous and diffuse (Figure 2C, bottom row). In addition, although Cx43 appeared to have formed GJ like puncta due to partial overlap of the cell peripheries of the adjacent cells, such puncta were rarely observed at the areas of cell–cell contact upon careful examination (Figure 2C, bottom row). Compared with RL-EΔN cells which became columnar and could be maintained as a monolayer on filters for 2–3 mo, RL-NΔE cells became moribund and died within 10 d after attaining confluence. To corroborate the immunocytochemical data, we measured the junctional transfer of fluorescent tracer, Lucifer Yellow (MW 443), by scrape-loading (Figure 3A), and of Alexa 488 (MW 570) and Alexa 594 (MW 760), by microinjection (Figure 3B), in cells grown on glass cover slips and transwell filters. We found that RL-EΔN cells communicated extensively whereas RL-NΔE cells did not communicate either on plastic or on transwell filters (Figure 3 and Table 1).

Open in a separate window

Figure 2.

Localization and expression of Cx43, adherens and tight junction proteins, and cytoskeletal proteins in RL-EΔN and RL-NΔE cells. (A) Western blot analysis of the expression level of E-Cad (E-cad) and N-Cad (N-cad), Cx43, α- and β-catenins, occludin and ZO-1, cytokeratin 8 and vimentin and actin. Note that RL-EΔN cells express only E-Cad, but no detectable level of N-Cad, whereas RL-NΔE cells express only N-Cad but no E-Cad. Note also that the expression level of Cx43 and other proteins is not discernibly different in the two cell lines. (B) RL-EΔN and RL-NΔE cells were seeded on glass coverslips and immunostained for Cx43, E-Cad (E-cad), N-Cad (N-cad), and occludin (Occl). Note that Cx43 is assembled into GJs in RL-EΔN cells, whereas it remains intracellular as vesicular puncta in RL-NΔE cells. Note also that E-Cad, N-Cad, and occludin are localized at the areas of cell–cell contact in both cell types. (C) RL-EΔN and RL-NΔE cells were seeded on transwell filters for 5 d and immunostained for Cx43 and ZO-1. Note a more robust expression of ZO-1 at cell–cell contact areas in RL-EΔN cells compared with RL-NΔE cells. Note also partial overlap of cell boundaries in RL-NΔE cells as well as GJ like punctate structures. Nuclei (Blue) were stained with DAPI. Bar = 10 μm in B and 15 μm in C.

Open in a separate window

Figure 3.

Gap junctional communication in RL-EΔN and RL-NΔE cells. (A) Cells were grown on glass coverslips and scrape-loaded with Lucifer Yellow and Alexa-594 conjugated dextrans. (B) Cells were microinjected with the indicated tracers as described in Materials and Methods. Note that in RL-EΔN cells, Lucifer Yellow has transferred from scrape-loaded (red) cells to neighboring cells, whereas no transfer is observed in RL-NΔE cells. Note also that the transfer of Alexa-488 and Alexa-570 from microinjected cells (indicated by arrows) to neighboring cells is observed only in RL-EΔN cells.

Because in contrast to RL-NΔE cells, Cx43 was assembled into GJs in RL-EΔN cells, we also investigated whether cell–cell adhesion mediated by E-Cad and N-Cad remained functional in both cell types. A hanging drop assay, which measures the rate and strength of cell–cell adhesion based on the size of cell aggregates formed over time as well as the resistance of the formed aggregates to a shearing force was used (Kim et al., 2000b). As shown in Table S-2 both RL-EΔN and RL-NΔE cells aggregated efficiently as assessed by the mean area of the aggregates. We also found that RL-EΔN and RL-NΔE cells failed to invade in a standard Matrigel invasion assay (data not shown). Standard scratch (wound) assays were performed to determine whether RL-NΔE cells were more motile compared with RL-EΔN cells. Live cell imaging of the scratches of equal width in three independent experiments showed that wounds were filled within 14 ± 2 h in RL-NΔE cells whereas 22 ± 2 h were required to fill the wounds in RL-EΔN cells (Figure S-3). The mean (± SE) rate of migration was significantly faster in RL-NΔE cells as compared with RL-EΔN cells (p = 0.04).

Taken together, the data presented in Figures 2, ​,3,3, and S-3 and in Tables 1 and S-1 suggest that a switch in expression from E-Cad to N-Cad is associated with the attenuation of assembly and/or maintenance of GJs and their function, while the expression of E-Cad is associated with the facilitation of assembly and function. Moreover, these data also suggest that in RL-NΔE cells GJ assembly was not disrupted because of lack of cell–cell adhesion but by some other defect. Furthermore, these findings suggest that in RL-NΔE cells, N-Cad expression is more closely related with increased motility and with the loss of ability to grow and sustain a polarized monolayer on transwell filters than with invasion in Matrigel.

Detergent-Solubility of Cx43 in RL-EΔN and RL-NΔE Cells

To substantiate the immunocytochemical and functional data, and to elucidate the possible mechanism of attenuation of GJ assembly in RL-NΔE cells, we determined the assembly of Cx43 into GJs biochemically by Western blot analysis of total and Triton X (TX)-100-soluble and TX-insoluble cell fractions (VanSlyke and Musil, 2000). Intriguingly, a significant proportion of Cx43 was detected in detergent-insoluble fractions in both RL-EΔN and RL-NΔE cells (Figure 4A). Detergent solubility of live RL-NΔE and RL-EΔN cells by in situ extraction with 1% TX-100 showed that intracellular Cx43 puncta in RL-NΔE cells and intercellular gap junctional puncta in RL-EΔN cells were detergent-insoluble (Figure 4B). We also found that in RL-NΔE cells small intercellular puncta at areas of cell–cell contact, which were barely detectable immunocytochemically, remained detergent-resistant. The solubility of β-catenin, an adherens junction–associated protein, was not significantly affected in RL-EΔN and RL-NΔE cells (Figure 4B), whereas EEA1, a marker for the early endosomes, was lost upon in situ extraction (data not shown). These data suggest that a significant proportion of Cx43 in RL-NΔE cells is detergent-insoluble and that the detergent-insoluble fraction corresponds to discrete puncta that lie scattered throughout the cytoplasm, as well as to very small puncta that are seen at areas of cell–cell contact.

Open in a separate window

Figure 4.

Detergent-solubility of Cx43 in RL-EΔN and RL-NΔE cells. (A) Total (T), TX-100-soluble (S), and TX–insoluble (I) fractions from cells were analyzed by Western blot analysis as described in Materials and Methods. The blots were stripped and reprobed with anti–β-actin antibody to verify equal loading. Note that Cx43 is detected in both detergent-soluble and detergent-insoluble fractions in both cell types. (B) Detergent solubility of Cx43 was examined in cells upon in situ extraction with 1% Triton X-100 at 4°C as described in Materials and Methods. Cells were immunostained for Cx43 (green) and β-catenin (red). Note that both junctional Cx43 in RL-EΔN cells and intracellular Cx43 in RL-NΔE cells were detergent-insoluble. Note also that the intensity of β-catenin at the areas of cell–cell contact is not significantly affected by the detergent treatment. Bar = 15 μm.

Opposing Effects of E-Cadherin and N-Cadherin on Gap Junction Formation

To explore further whether the expression of E-Cad and N-Cad affects GJ assembly in a diametrically opposed way, we undertook two complementary experimental approaches. First, we knocked down N-Cad in RL-NΔE cells using ShRNA. Second, we introduced human N-Cad in RL-EΔN cells and E-Cad in RL-NΔE cells. We used recombinant retroviruses to express and knock down cadherins. Also, for these experiments both N-Cad and E-Cad were myc-tagged to distinguish them from their endogenously expressed rat counterparts. An independent approach to knock down E-Cad using SiRNA was not successful as it failed to reduce the expression of E-Cad in RL-EΔN cells (unpublished). To minimize clonal heterogeneity, GJ assembly and cellular localization of Cx43 were examined in pooled polyclonal cultures within 1–3 passages after G418 or puromycin selection (see Materials and Methods). Figures 5 and ​and66 show the representative data from three independent sets of experiments and Table 2 provides the summary of the quantitative data.

Open in a separate window

Figure 5.

N-cadherin knockdown restores gap junction assembly in RL-NΔE cells. Cells were infected with ShEGFP and ShN-Cad and selected as described in Materials and Methods. Expression level of total (T), detergent-soluble (S), and detergent-insoluble (I) N-Cad (A) and Cx43 (B) were determined in cells infected with ShEGFP and ShN-Cad, respectively. Ten micrograms of total protein and normalized soluble and insoluble fractions were resolved on 12% SDS-PAGE and immunoblotted for Cx43 and N-Cad. The blots were reprobed with β-actin to verify equal loading. Note that the N-Cad expression level is significantly reduced in ShN-Cad infected pooled RL-NΔE polyclonal cultures (A), whereas the expression level of Cx43 (B) is not discernibly affected. Note also that knock down of N-Cad decreased the detergent soluble fraction, and increased the detergent-insoluble fraction, of Cx43. (C) Pooled polyclonal cultures of RL-NΔE cells infected with ShEGFP and ShN-Cad were immunostained for Cx43 and N-Cad. Note the reduced expression of N-Cad at areas of cell–cell contact and in the intracellular pool, the disappearance of intracellular Cx43, and GJ formation in cells infected with ShN-Cad but not with control retrovirus (ShEGFP). Bar = 15 μm.

Open in a separate window

Figure 6.

E-cadherin and N-cadherin have opposite effects on gap junction assembly. Myc-tagged N-Cad and E-Cad were introduced, respectively, into RL-EΔN and RL-NΔE cells as described in Materials and Methods. Pooled polyclonal cultures were immunostained for Cx43 (green) and either endogenous cadherin or Myc (red). Note that E-Cad expression (+E-Cad) in RL-NΔE cells restores GJ formation (A) and N-Cad expression (+N-Cad) in RL-EΔN cells partially disrupts GJ assembly (B) and induces intracellular accumulation of Cx43. Bar = 20 μm.

As assessed immunocytochemically and by the densitometric scanning of Western blots, ShN-cad decreased the expression level of N-Cad by 80 ± 7% (n = 3) in RL-NΔE cells (Figure 5 A) with a concomitant decrease in the localization of N-Cad at the areas of cell–cell contact (Figure 5C, compare top left panel with the bottom left panel), decrease in the intracellular accumulation of Cx43 and an increase in GJ assembly (Figure 5C, compare top middle and right panels with the corresponding bottom panels). Knock down of N-Cad had no discernible effect on the expression level of Cx43 but decreased the detergent-soluble fraction without affecting the detergent-insoluble fraction significantly (Figure 5B). These effects were not observed when RL-NΔE cells were infected with ShEGFP (Figure 5, A and B, left panels) or ShP-Cad, an ShRNA directed against placental cadherin (not shown). Moreover, knock down of N-Cad neither induced E-Cad expression nor affected the expression level of ZO-1 and α- and β-catenins (data not shown).

In the next series of experiments, we introduced myc-tagged N-Cad in RL-EΔN cells and myc-tagged E-Cad in RL-NΔE cells. As assessed by immunocytochemical and Western blot analyses, E-Cad (Figure S-4A, bottom right panel; Figure S-4C) and N-Cad (Figure S-4B, bottom right panel; Figure S-4D) were expressed robustly and appropriately at the areas of cell–cell contact when introduced into RL-NΔE and RL-EΔN cells, respectively. As assessed by quantitative immunocytochemical analysis, we found that the expression of E-Cad in RL-NΔE cells induced GJ assembly and decreased intracellular Cx43 puncta (Figure 6A, Table 2), whereas the expression of N-Cad in RL-EΔN cells partially attenuated GJ assembly and caused intracellular accumulation of Cx43 (Figure 6B, Table 2). The expression of E-Cad in RL-NΔE cells had no discernable effect on the expression level of N-Cad and its localization (Figure S-4, A and C). Similarly, the expression of N-Cad had no detectable effect on the expression level of E-Cad and its localization in RL-NΔE cells (Figure S-4, B and D). Taken together, these data document that the presence of N-Cad at the cell surface is associated with the attenuation of GJ assembly, whereas the presence of E-Cad with the facilitation of the assembly.

Cx43 Traffics Normally to Cell Surface in RL-NΔE and RL-EΔN Cells

We next explored the molecular basis of attenuation of GJ assembly and intracellular accumulation of Cx43 in RL-NΔE cells. Attenuation of GJ assembly and/or intracellular accumulation of Cx43 could be caused either by its impaired trafficking or by its endocytosis before GJ assembly upon arrival at the cell surface or by the endocytosis of minuscule GJs that remain very small. We used cell surface biotinylation to examine whether Cx43 trafficked normally to the cell surface in RL-NΔE and RL-EΔN cells. Figure 7A shows that Cx43 was biotinylated significantly in both RL-NΔE and RL-EΔN cells. To determine whether cell surface–associated Cx43 was degraded more rapidly in RL-NΔE cells compared with RL-EΔN cells, we determined the kinetics of its degradation after biotinylation. We found that cell surface biotinylated Cx43 degraded with a similar kinetics in both cell types (Figure 7, B–D). These findings suggest that intracellular accumulation of Cx43 in N-Cad expressing RL-NΔE cells is neither caused by impaired trafficking nor by altered kinetics of its degradation at the cell surface, but by some undefined mechanism that interferes with its assembly into GJs upon arrival at the cell surface resulting in its accumulation in intracellular vesicles or minuscule annular GJs.

Open in a separate window

Figure 7.

Cx43 traffics to cell surface in RL-EΔN and RL-NΔE cells. Confluent cultures of RL-EΔN and RL-NΔE cells (A) were biotinylated with sulfo-NHS-SS-biotin at 4°C, and biotinylated proteins from 100 μg of total cell lysate were detected by Streptavidin pull down followed by Western blotting for Cx43, E-Cad, and N-Cad (see Materials and Methods). In the lanes labeled Input, 10 μg of total protein was used and probed for Cx43. Note that in both RL-EΔN and RL-NΔE cells (A; labeled as Biotinylated), nearly the same amount of Cx43 was detected in the Streptavidin “Pull down” lanes. Note also that E-Cad and N-Cad were pulled down, respectively, only in RL-EΔN cells and RL-NΔE cells. A nonbiotinylated dish (labeled as Control) was used as a negative control. (B–D). Cell surface associated Cx43 degrades with similar kinetics in RL-EΔN and RL-NΔE cells. Cells were biotinylated at 4°C and were incubated for various times at 37°C before Streptavidin pull down and Western blotting as described in Materials and Methods. A nonbiotinylated dish was kept as a control (−). The blots were quantified after normalization of the input for each time point, and the values from three independent experiments were plotted graphically using Sigma plot as described in Materials and Methods. Note that cell surface associated biotinylated Cx43 degrades with similar kinetics in both RL-EΔN and RL-NΔE cells.

Cx43 Is Endocytosed by Nonclathrin-Mediated Pathway in RL-NΔE Cells

In RL-NΔE cells, Cx43 trafficked to the cell surface based on cell surface biotinylation, but failed to form functional GJs; moreover, a significant proportion of Cx43 remained detergent-insoluble as discrete intracellular puncta as assessed biochemically and immunocytochemically upon in situ extraction with 1% TX-100 (Figures 2​2–4). This raised the possibility that N-Cad expression attenuated GJ assembly either by triggering the endocytosis of connexons into detergent-insoluble puncta before their assembly into GJs or by triggering endocytosis of nascent, minuscule, and nonfunctional GJ plaques before their maturation into immunoctyochemically detectable, larger, and functional plaques (Jordan et al., 2001; Piehl et al., 2007; Falk et al., 2009). Hence, we investigated whether Cx43 was endocytosed by a clathrin-mediated pathway or a nonclathrin-mediated pathway. We used chlorpromazine and hypertonic medium, which prevent the formation of clathrin coated pits, to block the clathrin-mediated pathway (Heuser and Anderson, 1989; Wang et al., 1993), and filipin and methyl-β-cyclodextrin (MβCD), which impair the structure and function of sphingolipid-cholestrol-rich membrane microdomains (lipid rafts), to inhibit nonclathrin-mediated pathway (Subtil et al., 1994; Orlandi and Fishman, 1998; Skretting et al., 1999). Moreover, the specificity of these drugs in inhibiting each pathway was assessed by measuring the uptake of Alexa-594– or Alexa-488–labeled transferrin and cholera toxin, which are predominantly endocytosed by clathrin-mediated and nonclathrin-mediated pathways, respectively (Wang et al., 1993; Subtil et al., 1994; Orlandi and Fishman, 1998). For these experiments, RL-EΔN and RL-NΔE cells were treated with chlorpromazine (5 μg/ml), filipin (2.5 μg/ml), and MβCD (5 mM) for 2 h. These concentrations were determined based on several pilot experiments in which the dose and the duration of treatment were varied while the internalization of Alexa-488/594-labeled transferrin and cholera toxin was monitored microscopically. We found that both in RL-EΔN and RL-NΔE cells, chloropromazine and hypertonic media inhibited the uptake of only Alexa-488/594–labeled transferrin but did not affect the uptake of Alexa-488/594–labeled cholera toxin. Similarly, filipin and MβCD inhibited the uptake of only Alexa-488/594–labeled cholera toxin but did not affect the uptake of Alexa-488/594–labeled transferin (data not shown). Consistent with the effect of filipin and MβCD on the inhibition of uptake of cholera toxin, we found that they induced GJ assembly profoundly in RL-NΔE cells (Figure 8, bottom row, compare panel 1 with panels 3 and 4) but had no discernible effect on the assembly in RL-EΔN cells (Figure 8, upper row, compare panel 1 with panels 3 and 4). On the other hand, chlorpromazine had no effect on GJ assembly in RL-NΔE cells (Figure 8, bottom row, compare panel 1 and 2) but appeared to decrease the preponderance of intracellular puncta in RL-EΔN cells as assessed visually (Figure 8, top row, compare panels 1 and 2). A rigorous quantitative analysis has to be performed to assess if chloropromazine enhanced GJ assembly in RL-EΔN cells. Similar data were obtained when RL-NΔE and RL-EΔN cells were treated with hypertonic medium (data not shown).

Open in a separate window

Figure 8.

Inhibition of nonclathrin-dependent endocytosis restores gap junction assembly in RL-NΔE cells. RL-EΔN and RL-NΔE cells, seeded on glass cover slips in six-well clusters and allowed to grow to confluence, were treated with chlorpromazine (Chlorprom; 5 μg/ml), filipin (2.5 μg/ml), or methyl-β-cyclodextrin (MβCD, 5 mM) for 60 min. After washing and fixation, cells were immunostained for Cx43. Note that treatment with filipin and MβCD restored GJ assembly in RL-NΔE cells but had no significant effect on RL-EΔN cells. Bar = 20 μm.

To further define the subcellular fate of intracellular puncta, we immunostained RL-NΔE cells with the markers specific for the endocytic pathway and examined whether they colocalized with Cx43. These analyses showed that Cx43 did not detectably colocalize with clathrin, a major structural component of clathrin-coated vesicles, with the early endosome marker, EEA-1, and with caveolin-2 (Figure 9) although some colocalization was observed with caveolin-1 both in the perinuclear region and at the cell surface consistent with what has been reported in earlier studies (Langlois et al., 2008). Because none of the commercially available antibodies recognized rat Rab5 and Rab7, we transiently expressed GFP-tagged, constitutively active, Rab5 and Rab7 in RL-NΔE cells and immunostained cells for Cx43 to examine whether it colocalized with these markers. We found that Cx43 did not colocalize with Rab5-GFP–positive vesicles, however some colocalization was observed with Rab7-GFP positive late endosomes (Figure S-5). Moreover, significant colocalization of Cx43 was observed with Lamp-1 (Figure S-5, bottom row), which is localized on lysosomes (Rohrer et al., 1996). Taken together, the above data suggest that in RL-NΔE cells connexons that arrive at the cell surface are either internalized before their docking and assembly into GJs or are assembled into minuscule GJs, which are endocytosed by nonclathrin dependent pathway into detergent-resistant puncta that eventually are targeted to lysosomes for degradation.

Open in a separate window

Figure 9.

Cx43 does not colocalize with known endocytic markers in RL-NΔE cells. RL-NΔE cells, seeded as described in Figure 8 legend, were immunostained for Cx43 (green) and endocytic markers, EEA-1, clathrin, caveolin-1, and caveolin-2 (red). Note that Cx43 does not colocalize with any of these markers as shown in the enlarged images of the marked boxes on the right. W-Box = white box; R-Box = Red box. Bar = 20 μm.

On the Origin of Detergent-Insoluble Intracellular Puncta in RL-EΔN and RL-NΔE Cells

In the next series of experiments, we explored the possible mechanisms by which detergent-resistant intracellular puncta might be generated. To test whether these puncta arose from the endocytosis of connexons that failed to dock with the adjacent connexons or from the formation of minuscule GJs, which acquired resistance to detergent extraction, as has been observed during the formation of adherens junctions (Adams et al., 1996), we examined the subcellular fate of intracellular puncta in single RL-NΔE and RL-EΔN cells 24–36 h after seeding and determined their detergent solubility upon in situ extraction with 1% TX-100 in 3 independent experiments. We rationalized that given the short half-life of Cx43 (2–5 h), intracellular puncta in single cells plausibly represent endocytosed undocked connexons that trafficked to the cell surface or vesicles along the secretory pathway. We found that in RL-NΔE and RL-EΔN single cells, Cx43 was localized as discrete puncta both at the cell periphery and in the cytoplasm; moreover, as shown in representative images in Figure 10, these puncta colocalized with clathrin, EEA-1 and Cav-1 albeit to a variable extent. The extent of colocalization appeared to be more robust at the perinuclear region as compared with cell periphery. However, we found that in contrast to what was observed in confluent cells, all puncta—both intracellular and at the cell periphery—were lost upon in situ extraction with 1% TX-100 as assessed immunocytochemically (data not shown).

Open in a separate window

Figure 10.

Cx43 is endocytosed into detergent-soluble puncta in RL-NΔE and RL-EΔN single cells. Between 1- 2 × 10³ RL-EΔN and RL-NΔE cells were seeded on six-well clusters containing glass cover slips. The subcellular fate of Cx43 was examined in single cells upon in situ extraction with 1% Triton X-100 at 4°C as described in Materials and Methods. Cells were immunostained for Cx43 (green) and clathrin, EEA-1 and caveolin-1 (red). Enlarged images marked by the white boxes (W-box) and red boxes (R-box) are shown on the right. Note that significant colocalization with all the markers is observed in the perinuclear regions in both cell types, while partial localization to a variable extent is observed in the cell peripheries. All, punctuate and nonpunctate, immunostaining was lost upon detergent extraction (not shown). Bar = 7 μm.

To rule out the possibility that the detergent-resistant puncta arose during the transit of Cx43 along the secretory pathway, we blocked ER to Golgi transport with brefeldin (Klausner et al., 1992; Chardin and McCormick, 1999) and cis-Golgi to medial-Golgi transport with monensin (Tartakoff, 1983). For these experiments, confluent RL-NΔE and RL-EΔN cells were treated with brefeldin (10 μM) and monesin (10 μM) for 30 min to 4 h and immunostained for Cx43 and GM130, a cis-Golgi resident protein (Marra et al., 2007). We found that in RL-NΔE cells, treatment with brefeldin for 1–4 h disrupted Golgi structure but did not diminish the number of Cx43 puncta discernibly (Figure 11A, compare panels in the top row with the middle row). Moreover, these puncta neither colocalized with GM130 (Figure 11A, middle row) nor with calreticulin, an ER-resident protein (Caramelo and Parodi, 2008) (data not shown). We also found that treatment of RL-NΔE cells with monensin for 2–4 h blocked the transport of Cx43 in the cis-Golgi, as assessed by its extensive colocalization with GM130 in the perinuclear regions, but did not appear to diminish the number of intracellular puncta. Moreover, these puncta still persisted when brefeldin and monensin treated RL-NΔE cells were extracted in situ with 1% TX-100 although the perinuclear staining specific for Cx43 and GM130 was lost (data not shown).

Open in a separate window

Figure 11.

Detergent-resistant intracellular puncta are not generated along the secretory pathway in RL-NΔE cells. RL-NΔE (A) and RL-EΔN (B) cells were seeded in 6-cm Petri dishes and allowed to attain confluence. Cells were treated with brefeldin or monensin (10 μM) for 2 h and immunostained for Cx43 and GM130, a cis-Golgi resident protein (see Materials and Methods). Enlarged images marked by the white boxes are shown on the right. Note that both in RL-NΔE cells (A) and RL-EΔN cells (B), brefeldin disrupted the Golgi. Note also that brefeldin did not decrease the number of Cx43 puncta in RL-NΔE (A) cells but seemed to have stabilized or increase GJ plaques in RL-EΔN cells (B). Note a significant colocalization of Cx43 with GM130 due to the blockade of intraGolgi transport by monensin. Note also that monensin did not decrease the number of Cx43 puncta in RL-NΔE cells. Bar = 8 μm.

Intriguingly, treatment of RL-EΔN cells with brefeldin for 1–4 h also fragmented Golgi, but, in contrast to N-Cad expressing RL-NΔE cells, seemed to enhance GJ assembly or stabilize GJ plaques as assessed visually by the appearance of larger puncta at areas of cell–cell contact. As has been observed by others, in untreated RL-EΔN cells, Cx43 appeared to colocalize with GM130 near the perinuclear regions (Figure 11B, top row). Also, treatment of RL-EΔN cells with monensin not only caused accumulation of Cx43 in perinuclear regions, as assessed by its extensive colocalization with GM130 but also appeared to increase the number of intracellular puncta in some cells, which possibly represent endocytosed or degraded GJs (Figure 11B, compare panels in to top row with the bottom row). We also found that when RL-EΔN cells were treated with brefeldin and monensin for more than 4 h GJ assembly was drastically disrupted, with concomitant increase in the number of intracellular puncta (data not shown). Taken together, the data shown in Figures 10 and ​and11,11, combined with those shown in Figures 4, ​,8,8, and ​and9,9, suggest that detergent resistant intracellular puncta in N-Cad expressing RL-NΔE cells are likely generated at the site of cell–cell contact—and not en route from ER to cell surface via Golgi and trans-Golgi network—and possibly represent minuscule annular GJs, which become detergent-insoluble and eventually degrade in the lysosomes. The data obtained from single cells suggest that undocked connexons in both RL-NΔE and RL-EΔN cells most likely are endocytosed by both a clathrin-dependent and nonclathrin-dependent pathway.

Cadherin Expression and Assembly of Cx32 into Gap Junctions in RL-EΔN and RL-NΔE Cells

To investigate whether the assembly of other Cxs into GJs was also regulated differentially by E- and N-Cad, we introduced rat Cx32 into RL-EΔN and RL-NΔE cells, which do not express this Cx, by recombinant retroviruses and examined its subcellular fate in pooled polyclonal cultures within 2–3 passages after selection in G418. We chose Cx32 because it is expressed only by well-differentiated and polarized cells (Bavamian et al., 2009). We found that Cx32 was assembled efficiently into GJs in RL-EΔN cells but not in RL-NΔE cells where it remained intracellular as discrete puncta that also remained detergent-insoluble as assessed immunocytochemically upon in situ extraction with 1% TX-100 (Figure 12A) and biochemically by the detergent insolubility assay (Figure 12B) similar to what was observed with Cx43. Moreover, we found that both junctional and nonjunctional Cx43 and Cx32 did not appear to colocalize significantly in RL-EΔN and RL-NΔE cells. Taken together, these data suggest that N-Cad expression also prevents the assembly of Cx32 into GJs, possibly by triggering its endocytosis into detergent-insoluble puncta.

Open in a separate window

Figure 12.

Differential assembly of Cx32 in RL-EΔN and RL-NΔE cells. (A) RL-EΔN and RL-NΔE cells were infected with recombinant retrovirus harboring rat Cx32 and pooled polyclonal cultures were immunostained for Cx43 and Cx32 after in situ extraction with 1% TX-100. Note that both Cxs assemble into GJs in RL-EΔN cells but remain intracellular in RL-NΔE cells. (B) Cell lysates prepared from pooled polyclonal cultures were analyzed by the detergent-solubility assay. Total (T), TX-100-soluble (S), and TX–insoluble (I) fractions were analyzed by Western blot analysis as described in Materials and Methods. The blots were stripped and reprobed with anti–β-actin antibody to verify equal loading. Note that both Cx43 and Cx32 are detected in detergent-soluble and detergent-insoluble fractions in both cell types. Bar = 20 μm.

We thank Linda Kelsey for her expert technical help. We thank Tom Dao for help with live cell imaging. We thank Drs. S. Caplan and Naava Naslavsky for helpful suggestions and discussion throughout the course of this study. This research was supported by National Institutes of Health Grant CA113903, Department of Defense Prostate Cancer Research Program Grant 081198, and Nebraska State Grant LB506 (to P.P.M.) and RO-1DE12308 (to K.R.J.). We gratefully acknowledge support from the Nebraska Center for Cellular Signaling and Cancer Graduate Research Training Program of Eppley Institute in the form of fellowships (to S.C. and K.E.J., respectively).