EGF induces efficient Cx43 gap junction endocytosis in mouse embryonic stem cell colonies via phosphorylation of Ser262, Ser279/282, and Ser368

Abstract

Gap junctions (GJ) traverse apposing membranes of neighboring cells to mediate intercellular communication by passive diffusion of signaling molecules. We have shown previously that cells endocytose GJs utilizing the clathrin machinery. Endocytosis generates cytoplasmic double-membrane vesicles termed annular gap junctions or connexosomes. However, the signaling pathways and protein modifications that trigger GJ endocytosis are largely unknown. Treating mouse embryonic stem cell colonies - endogenously expressing the GJ protein connexin43 (Cx43) - with epidermal growth factor (EGF) inhibited intercellular communication by 64% and activated both, MAPK and PKC signaling cascades to phosphorylate Cx43 on serines 262, 279/282, and 368. Upon EGF treatment Cx43 phosphorylation transiently increased up to 4 fold and induced efficient (66.4%) GJ endocytosis as evidenced by a 5.9 fold increase in Cx43/clathrin co-precipitation.

Introduction

Direct intercellular communication by gap junction (GJ) channels is a hallmark of normal cell and tissue physiology. GJs are the only cell-cell junction type that allows direct cell-cell communication via the transfer of molecules between cells. Examples include small metabolites such as glucose, amino acids, and ATP; ions such as Na⁺, Ca²⁺, and Cl⁻; cell signaling molecules such as IP3 and cAMP; and potentially functional RNAs, such as miRNAs in glioma cells [1] and siRNAs in NRK cells ([2], reviewed in [3]). Complete double-membrane spanning GJ channels are formed when two hexameric hemi-channels (connexons) dock in the extracellular space. In addition, based on their double-membrane configuration GJs likely contribute significantly to cell-cell adhesion. Clearly, these cellular GJ functions require precise modulation. Remarkably, docked GJ channels cannot be separated into individual hemi-channels under physiological conditions [4,5]. Yet, analyses in many different cell and tissue types revealed that cells endocytose their GJs constitutively, and efficiently after treatment with inflammatory mediators such as thrombin and endothelin, in response to treatment with the non-genomic carcinogen lindane, and under many physiological and pathological conditions that require cell-cell uncoupling and/or physical cell-cell separation such as cell migration in development and wound healing, tissue differentiation, mitosis, apoptosis, leukocyte extravasation, ischemia, hemorrhage, edema, and cancer cell metastasis [6–14]. Constitutive and acute GJ endocytosis correlates with the short half-life of 1–5 hours reported for GJ proteins (connexins, Cxs) and GJs [7,15–17]. We have previously shown that GJs are endocytosed as a whole in a clathrin-mediated endocytic process [6,12,18,19]. However, the specific post-translational modifications such as phosphorylation, ubiquitination, etc. that may render Cx proteins in GJs endocytosis competent are still poorly understood.

Twenty Cx isoforms are found in mouse with Cx43 being the most prominent isoform. Cx43 is a well-known phospho-protein. Numerous serine residues in the Cx43 C-terminus are phosphorylated to up-regulate (Ser325, Ser328, Ser330, Ser364/365, and Ser373) or down-regulate (Ser255, Ser262, Ser279/282, and Ser368) GJ mediated intercellular communication (GJIC) (reviewed in [20]). Protein kinase C (PKC) is thought to phosphorylate Cx43 at Ser368 to down-regulate GJIC [21,22]. Cx43 has also been shown to be a substrate of mitogen activated protein kinase (MAPK) that upon mitogen stimulation phosphorylates Cx43 at Ser255, Ser262, and Ser279/Ser282 to down-regulate GJIC [23,24]. Epidermal growth factor (EGF), a well-studied mitogen, binds to the EGF receptor (a receptor tyrosine kinase [RTK] family member) to activate both MAPK and PKC signaling pathways, to for example promote cell proliferation. Mouse embryonic stem (mES) cells are known to express the GJ proteins Cx31, Cx43, and Cx45 and to form functional GJs [25,26]. ES cells are actively proliferating cells with a relatively short cell cycle (see Supplemental Movie 1). They can infinitely self-renew while maintaining their pluripotency that is mediated via GJIC among the cells in the colony [27–29]. Treatment of mES cells with EGF is known to down-regulate GJIC [30]. However, the mechanism/s that lead to GJIC inhibition are not known. Since EGF can stimulate cell proliferation, and mitotic cells are known to remove their GJs at onset of mitosis [13], we hypothesized that EGF-treatment may lead to GJ endocytosis to down-regulate GJIC. To test this hypothesis, and to characterize signals that may lead to GJ endocytosis, we treated mES cell colonies with EGF. We found that EGF-treatment induced inhibition of GJIC that correlated with clathrin recruitment and Cx43-GJ endocytosis, and that GJ endocytosis was initiated by phosphorylation of Cx43 at serines 262, 279/282, and 368.

Materials and Methods

Cell culture

E14TG2a mouse embryonic stem (mES) cells (ATCC, Cat. No. CRL-1821) were seeded on mouse embryonic fibroblasts (MEFs) (Millipore, Cat. No. PMEF-NL). Under established culture conditions described below mES cells remain undifferentiated and grow into 3-dimensional colonies that are only loosely attached to the culture dishes. Colonies were passaged and maintained in 0.1% gelatin-coated dishes (MEF-free) in humidified atmosphere containing 5% CO₂ at 37°C in KO DMEM (Gibco, Cat. No. 10829). Media were supplemented for a final concentration of 15% with KO serum replacement (Gibco, Cat. No. 10828), 3mM L-glutamine (Gibco, Cat. No. 25030), 50 I.U/ml penicillin and 50μg/ml streptomycin (Gibco, Cat. No. 15070), 1mM sodium pyruvate (Gibco, Cat. No. 11360 stock 100mM), 1x non-essential amino acids (Millipore, Cat. No. TMS-001-C), 1x β-mercaptoethanol (Millipore, Cat. No. ES-007-E), and 1000 U/ml ESGRO mLIF (Millpore, Cat. No. ESG1106) to prevent cells from differentiating. Prior to EGF treatment (100ng/ml EGF, Sigma, Cat. No. E4127), media were replaced with serum free media (to starve cells of growth factors) supplemented either with 50μM PD98059 (MEK, MAPK pathway inhibitor; Sigma, Cat. No. P215), 100–500nM chelerythrine-Cl (PKC inhibitor; Cayman Chemical, Cat. No. 11314), or both for 1 hour and cultured under standard conditions.

Dye transfer assays

mES cell colonies were cultured in 3.5 cm dishes for 2 days, then pre-incubated with serum free media as described above. Appropriate culture dishes were then treated with PD98059, chelerythine- Cl, or both and cultured for 1 hour before adding 100ng/ml EGF and incubating for additional 30 minutes. Media were replaced with 0.1% Lucifer Yellow (LY; Invitrogen, Cat. No. L682) in OPTIMEM (Gibco, Cat. No. 31985). To wound cells and allow for LY dye uptake, mES cell colonies were cut with a sharp scalpel in the presence of LY, incubated at room temperature (RT) for 5 minutes to allow dye to transfer to neighboring cells, then washed 3 times with OPTIMEM followed by 3.7% paraformaldehyde fixation for 10 minutes. Paraformaldehyde was removed by washing 3 times with 1xPBS containing Ca²⁺ and Mg²⁺. LY fluorescence and Phase Contrast images were acquired using a 20x objective. Dye spreading from the injured cells to the farthermost receiving cells was measured using MetaVue software version 6.1r5 (Molecular Devices, Sunnyvale, CA), averaged and plotted. In addition, fluorescence intensities along lines placed perpendicular to the cut at representative areas were measured and plotted as well.

Immunofluorescence microscopy and image analyses

mES cell colonies were grown on glass cover slips pre-treated with 0.1% gelatin (Fisher Scientific, Cat. No. G-7) to improve colony adhesion; fixed and permeabilized in pure ice-cold ethanol for 10 minutes; blocked with 10% FBS in PBS at RT for 1 hour, and incubated with primary rabbit polyclonal anti-Cx43 antibodies (Cell Signaling Technology, Cat. No. 3512) at 1:500 dilution at 4°C overnight. Secondary antibodies (Alexa488-conjugated goat anti-rabbit, Molecular Probes/Invitrogen, Cat. No. A11008) were used at 1:500 dilution at RT for 1 hour. Plasma membranes were visualized using 1 μg/ml Alexa594-conjugated wheat germ agglutinin (WGA; Molecular Probes/Invitrogen, Cat. No. W11262), or a monoclonal antibody directed against the membrane-associated protein, ZO-1 (at 1:300 dilution, Zymed Laboratories, Cat. No. 33-9100) combined with an Alexa568-conjugated secondary antibody (goat anti-mouse, at 1:500 dilution, Molecular Probes/Invitrogen, Cat. No. A-11031) after fixation in 3.7% formaldehyde and permeabilization in 0.2% Triton X-100. Cell nuclei were stained with 1μg/ml DAPI. Cells were mounted using Fluoromount-G™ (SouthernBiotech, Cat. No. 0100-01) and examined. Wide-field fluorescence microscopy was performed on a Nikon Eclipse TE 2000E inverted fluorescence microscope equipped with a 40x, NA 1.4, Plan Apochromat oil immersion objective. Images were acquired using MetaVue software. Confocal microscopy was performed on a Zeiss Axiovert 200 M inverted fluorecence microscope equipped with an LSM510 META scan head and a 63x, NA 1.4, Plan Apochromat oil immersion objective. Argon-ion and Helium-Neon lasers were used to generate 488 and 543nm excitation lines, and pinholes were typically set to 1 airy unit. Images were acquired and analyzed using ZEN software. Quantitative analyses were performed using ImageJ (National Institutes of Health, USA).

Electron microscopic analyses

mES cell colonies were cultured under feeder-free conditions in 3.5 cm diameter dishes, then fixed with 2.5% glutaraldehyde (Sigma, Cat. No. G7651) in 0.1M sodium cacodylate buffer at RT for 2 hours. Cells were washed, treated with tannic acid, dehydrated, uranyl acetate-stained and flat-embedded in the dishes as described in Falk [31]. Embedded cells were mounted, trimmed, thin-sectioned and examined with a Phillips CM100 electron microscope.

Western blot analyses

Denatured protein samples derived from mES cell lysates were analyzed on 10% SDS-PAGE mini-gels (BioRad). Biotinylated protein ladder (Cell Signaling Technology, Cat. No. 7727S) was used as a molecular weight marker. Proteins were transferred onto nitrocellulose membranes (Whatman, Cat. No. 10439396) on ice at 120 V for 1 hour before blocking with 5% non-fat dry milk, or 5% BSA in TBST at RT for 1 hour. Membranes were then incubated with primary antibodies at 4°C overnight. Antibodies used were: rabbit anti-Cx43 (Cell Signaling Technology, Cat. No. 3512) at 1:2500 dilution, phospho-specific rabbit anti-Cx43 Ser255 (Santa Cruz, Cat. No. sc-12899-R), phospho-specific rabbit anti-Cx43 Ser262 (Santa Cruz, Cat. No. sc-17219-R) both at 1:500, phospho-specific rabbit anti-Cx43 Ser279/282 [32] at 1:1000, phospho-specific rabbit anti-Cx43 Ser368 (Cell Signaling Technology, Cat. No. 3511) at 1:2000, phospho-ERK1/2 (Cell Signaling Technology, Cat. No. 9102) at 1:3000, mouse anti-clathrin heavy chain (BD Transduction Laboratories, Cat. No. 610499), at 1:2000, and mouse anti-α-tubulin (Sigma, Cat. No. T9026) at 1:5000 dilution. Membranes were washed with TBST followed by incubation with HRP-conjugated secondary antibodies (Zymed Laboratories, Cat. No. 81-6520 or 81-6120) at 1:10,000 dilution at RT for 1 hour. Proteins were detected with enhanced chemiluminescent (ECL) reagent. NIH ImageJ was used to quantify protein band intensities.

Co-Immunoprecipitation assays

mES cells were lysed with 1x SDS-free RIPA buffer (Cell Signaling Technology, Cat. No. 9806) supplemented with protease and phosphatase inhibitor cocktail (Cell Signaling Technology, Cat. No. 5872). Supernatants were incubated with rabbit anti-Cx43 antibodies (Cell Signaling Technology) bound to protein-G magnetic beads (Dynabeads®, Invitrogen, Cat. No. 10003) at 4°C overnight or RT for 2 hours, then washed 3 times with 1x TBS. Cx43 and bound proteins were eluted using hot SDS-PAGE sample buffer containing 50 mM DTT. Samples were analyzed by Western blot as described above.

Statistical analyses

Unpaired student t-tests were performed to analyze the distance of dye transfer (Figure 1), and to calculate efficiency of EGF-mediated GJ endocytosis (Figures 3A, S1) using GraphPad software (GraphPad Inc., La Jolla, CA). ANOVA with Dunnett's post-hoc tests were used to compare the levels of serine phosphorylation in Cx43, and the amount of co-precipitated clathrin heavy chains (Figures 2, 4). Data are presented as mean ± SEM. In all analyses, a p-value ≤ 0.05 (depicted with *, **, or ***) was considered statistically significant.

Figure 1.

EGF inhibits GJIC in mES cell colonies. mES cell colonies were treated with EGF in the presence or absence of the MEK-inhibitor PD98059 (MAPK signaling cascade), the PKC-inhibitor chelerythrine-Cl, or both followed by Lucifer yellow (LY) scrape loading dye transfer assays. Representative overlays of merged Phase Contrast and LY fluorescence images of three independent experiments are shown in (A), quantitative analyses of relative dye transfers (representative of GJIC efficiencies) are shown in (B). mES cell colonies are outlined by dotted lines. Scalpel-cuts through the colonies leading to cell wounding and dye-uptake are depicted by red lines. Fluorescence intensity profiles along lines (white) placed in representative areas perpendicular to the cuts are shown beside the images. In average, EGF-treatment reduced GJIC to 36%, while the MAP kinase cascade inhibitor, PD98059, counteracted the inhibitory EGF effect and GJIC almost reverted to the level detected in untreated colonies (90%). Chelerythrine-Cl counteracted the inhibitory EGF effect only to some extent and GJIC efficiency remained low (49%). Addition of both inhibitors reverted GJIC to about the level observed for PD98059 alone, suggesting that the inhibitory EGF effect is mainly MAPK driven.

Figure 3.

EGF induces GJ endocytosis in mES cells. mES cell colonies grown on cover glasses were fixed and stained using antibodies specific for Cx43 (Alexa488-conjugated secondary antibodies, green). Alexa594–conjugated WGA (red) and DAPI (blue) were used to decorate PMs and label cell nuclei. (A) Representative wide-field immunofluorescence micrographs of 4 independent experiments are shown. Magnified views of boxed areas are shown on the right. An increased number of cytoplasmic Cx43 fluorescent puncta, indicative of internalized AGJ vesicles (labeled with arrowheads) were observed in EGF treated cells (middle panels). GJs, localized in the PM and appearing as puncta and short lines, are depicted by arrows. Noticeably less AGs are detected in untreated colonies (top panels), and in colonies that were treated with PD98059 plus EGF (bottom panels). Statistical analyses in EGF-treated and untreated colonies revealed an average GJ endocytosis efficiency of 66.4%. (B) Transmission electron micrographs revealed typically appearing, double-membrane (penta-laminar striped) GJs (marked with arrows) and AGJ vesicles (marked with arrowheads) in treated and untreated mES cell colonies, supporting the conclusion that cytoplasmic puncta detected in (A) probably are endocytosed GJs. A low-magnification overview image showing the densely arranged morphology of mES cells within a colony is shown in panel 1. N = cell nuclei.

Figure 2.

EGF activates MAPK and PKC signaling cascades to phosphorylate Cx43 at serine residues 262, 279/282, and 368. mES cell colonies were treated with EGF before cell lysates were prepared at indicated times. Phosphorylated Cx43 was detected using phopho-specific antibodies directed against Ser255, Ser262, Ser279/282, and Ser368. Membranes were stripped and re-probed with anti α-tubulin antibodies as a loading control. Representative blots are shown in (A), normalized quantitative analyses of phosphorylated Cx43 of 4 independent experiments are shown in (B). EGF induced a significant, 4-fold increase of Cx43 phosphorylation at serines 262, 279/282 (MAPK sites), and 368 (PKC site), but not on Ser255 (another MAPK site). Phosphorylation peaked between 10–30 minutes after EGF treatment.

Figure 4.

Significantly increased amounts of clathrin, a direct indicator of GJ endocytosis, co-precipitates with Cx43 in EGF-treated mES cell colonies. (A) Western blot analyses of EGF-treated cell lysates show increased activity of ERK (as indicated by increased levels of phosphorylated ERK, pERK1/2), and increased phosphorylation of Cx43 at Ser262, Ser279/282, and Ser368 in cells treated for 30 min with EGF (lane 2) consistent with results shown in Figure 2. Treatment with PD98059 (MEK/MAPK pathway inhibitor) plus EGF inhibited EGF-mediated phosphorylation of Ser262 and Ser279/282 (lanes 3, 5), while chelerythrine-Cl (PKC inhibitor) treatment inhibited EGF-mediated phosphorylation of Ser368 (lanes 4, 5). Untreated mES cells and non-Cx43 expressing MDCK cell lysates were analyzed in control (lanes 1, 6). (B) Cx43 polypeptides from lysates in (A) of three independent experiments were immunoprecipitated, and co-precipitated clathrin (CHC) was detected by Western blot analyses and quantified. Cx43 from EGF treated cells in average co-precipitated 5.9 fold more clathrin than in untreated control cells (lanes 1, 2). PD98059-treatment counteracted EGF-induced GJ internalization by inhibiting ERK1/2 (lanes 3, 5). Moderately increased levels of clathrin (4.9 fold) co-precipitated in cells treated with chelerythrine-Cl alone (lane 4). No Cx43 or CHC was precipitated in MDCK control cells.

Results

EGF treatment inhibits GJIC in mES cell colonies by activating the MAPK signaling cascade

GJIC has been shown to play a crucial role in embryonic development, tissue function, and cellular homeostasis (reviewed in [26,33]). GJIC in pre-implantation mouse embryos as early as at the 8-cell stage was described by Lo and Gilula [14]. Cx protein expression and GJIC in mES cell colonies has also been described [25]. In 2008, inhibition of GJIC in mES cells treated with EGF has been described [30]. However, the mechanism/s that lead to GJIC-inhibition were not explored. Thus, to specifically examine the effects of EGF-treatment on Cx43, and to test whether specific Cx43 phosphorylation events may be responsible for the inhibition of GJIC and/or GJ endocytosis, we pre-incubated mES cells either with or without PD98059 (a specific MEK inhibitor, an upstream kinase of the MAPK signaling cascade, see Figure 5), chelerythrine-Cl (a PKC inhibitor), or both in serum free media before treating cells with EGF. To assess GJIC inhibition, scrape loading/dye transfer assays using the GJ-permeable dye, Lucifer Yellow (LY) were performed, and GJIC-efficiency in EGF-treated cell colonies was compared to untreated cell colonies in 3 independent experiments. GJIC in EGF-treated cell colonies was significantly reduced to 36±1.1% (p=0.0001, n=41) compared to untreated controls (100±5.9%, n=35) (Figure 1A, B, columns 1, 2). Cell colonies pre-incubated in PD98059, or PD98059 and chelerythrine-Cl exhibited a similar level of GJIC comparable to untreated control cells (90±2.5%, p=0.078, n=59 and 87±6.1%, p=0.164, n=19, respectively), suggesting that the MEK inhibitor significantly abolished the GJIC inhibitory effect of EGF (Figure 1A, B, columns 3, 5). Pre-treating cells with chelerythrine-Cl did only slightly reduce the inhibitory effect of EGF, resulting in a 49±2.5% (p=0.0001, n=11) efficient GJIC (Figure 1A, B, column 4). Together, these results indicate that EGF inhibits GJIC in mES cell colonies by activating the MAPK signaling cascade.

Figure 5.

Schematic representation illustrating EGF-induced, MAPK/PKC-mediated GJ internalization in mES cells. EGF binds to, phosphorylates and activates its receptor, a receptor tyrosine kinase (RTK) located in the PM of cells. Activated EGF-receptor activates MAPK and PKC signaling cascades to phosphorylate Cx43 at Ser262, Ser279/282, and Ser368. Phosphorylated-Cx43 recruits clathrin and clathrin-adaptors to endocytose GJs.

EGF activates MAPK and PKC in mES cell colonies to phosphorylate Cx43 on serines 262, 279/282, and 368

In order to examine which serine residues in the Cx43 sequence might be phosphorylated in response to EGF treatment, we implemented phospho-specific antibodies to detect the level of Cx43 phosphorylation at specific serine residues. mES cell colonies were cultured and serum-starved as described, then lysed at 10, 20, 30, 45, and 60 minutes post EGF treatment. Levels of phospho-Cx43 were analyzed by Western blots and intensities of protein bands were quantified. α-tubulin levels were analyzed as a loading control. Results indicate that EGF induced Cx43 phosphorylation at specific serine residues located in its C-terminus. Levels of pSer262, pSer279/282, and pSer368 at 10, 20 and 30 minutes post EGF treatment were significantly higher than in the untreated control (Figure 2A, B, red, green and purple columns). At later time points (45 and 60 min), serine phosphorylation levels again dropped to levels that were not significantly different from untreated controls. No significant increase in phospho-Ser255, another Cx43 MAPK site that is known to be phosphorylated in mitotic cells [34,35], was detected (see Discussion) (Figure 2A,B, blue column).

EGF induces GJ endocytosis in mES cell colonies

We have previously demonstrated that cells utilize the clathrin-mediated endocytic (CME) machinery to endocytose GJs either constitutively or acutely after treatment with natural inflammatory mediators such as thrombin and endothelin [6,12,18,19]. GJ channel endocytosis at onset of mitosis in rat fibroblasts and NRK cells has also been described [13]. Thus, to examine whether EGF-induced Cx43 phosphorylation may also induce GJ internalization, mES cell colonies were treated with EGF and cultured together with appropriate untreated controls. Colonies were then fixed and stained with Cx43 specific antibodies, followed by Alexa488-conjugated secondary antibodies (Figures 3A, S1, columns 1). Plasma membranes (PMs) were visualized either by using Alexa594-conjugated wheat germ agglutinin (WGA) (Figures 3A, S1, columns 2), or antibodies specific for the protein ZO-1 followed by Alexa568-conjugated secondary antibodies (Figure S1, column 2). When cell colonies were examined by wide-field fluorescence microscopy (Figure 3A) or confocal microscopy (Figure S1), an obvious increase in cytoplasmically located punctate Cx43 structures (presumably AGJ vesicles) was detected in EGF-treated cell colonies (Figures 3A, S1, middle panels, labeled with arrowheads) when compared with untreated cells (Figures 3A, S1, top panels). GJs appeared as puncta and short lines clearly located in the membranes between cells (Figures 3A, S1, labeled with arrows). Note that ES cells only have small cytosolic volumes and hence AGJs often appear in close proximity to the PM. When GJs and AGJs were counted in 6 representative images, and ratios calculated and compared between treated and untreated cells, average GJ/AGJ ratios in untreated cells dropped from 68.8±2.4% to 23.1±8.0% in EGF-treated cells, representing a GJ endocytosis rate of 66.4% (p=0.0007, n=68 GJs and AGJs counted in total) in mES cell colonies when treated with EGF. When colonies were pre-incubated with PD98059 prior to EGF-treatment, the average GJ/AGJ ratio was only slightly lower (63.3±1.7%, n=60 GJs and AGJs counted in 4 images) than the ratio observed in untreated cells correlating with the previously observed efficient inhibitory effect of PD98059 (Figure 3A, bottom panel).

To further support our conclusion that the cytoplasmic puncta represent internalized AGJ vesicles, we treated cells with EGF or PD98059 followed by EGF, and then performed EM analyses in comparison with untreated cells. Typical GJs were detected in the PMs between mES cell pairs in untreated and PD98059 plus EGF treated cells (Figure 3B, left and right panels, marked with arrows). Consistent with our immunofluorescence data, we detected numerous AGJ vesicles in the cytoplasm of the EGF-treated cells (Figure 3B, middle panels, marked with arrowheads). Note the typical, penta-laminar striped appearance of GJs and of the AGJ vesicle `wall' that is representative of the double-membrane configuration of GJs. Also consistent, clearly less AGJs and more GJ plaques were detected in PD98059 plus EGF-treated cells (Figure 3B, right panel). No AGJ vesicles were found in untreated cells. Taken together, our results suggest that EGF-treatment induces efficient GJ internalization in mES cells. Representative images from 6 independent immunofluorescence experiments and 2 independent EM studies are shown.

More clathrin is recruited to Cx43 GJs in EGF treated mES cell colonies

As described above, cells utilize the clathrin endocytic machinery to internalize their GJs [6,12,18,19]. Thus, the amount of clathrin that co-immunoprecipitates with Cx43 can be used as a direct measurement to quantify GJ endocytosis. We thus pre-treated mES cell colonies in serum free media for one hour with PD98059 (MEK inhibitor) and/or chelerythrine-Cl (PKC inhibitor), followed by treatment with EGF, and immunoprecipitation of Cx43 protein using magnetic beads. Mardin Darby canine kidney (MDCK) cells known not to express endogenous Cx43 protein [36], and untreated mES cell colonies were included as controls. Fractions of the cell lysates were collected for input analyses of Cx43 phosphorylation, ERK1/2 activity (= MAP kinase cascade activation), and protein loading (Figure 4A). Consistent with the results shown in Figure 2, Cx43 was substantially phosphorylated at Ser262, Ser279/282, and Ser368 after 30 minutes post EGF treatment (Figure 4A, lanes 1, 2). Note, that the activity of ERK1/2, a late kinase of the MAPK signaling cascade (see Figure 5), was significantly increased in EGF treated cells as indicated by the elevated levels of phosphorylated ERK1/2 that was detected with phospho-ERK-specific antibodies (Figure 4A, row 5, lanes 2, 4). Also note that ERK1/2 activation/phosphorylation was effectively inhibited by PD98059, a specific inhibitor of the upstream MAP kinase MEK (Figure 4A, row 5, lanes 3, 5). As a result and consistent with ERK1/2 inhibition, in PD98059 treated colonies, Cx43 phosphorylation at Ser262 and Ser279/282 (known Cx43 MAPK target sites [24]) was inhibited (Figure 4A, rows 2, 3, lanes 3, 5). Likewise, in chelerythrine-Cl-treated colonies a significant reduction of EGF-induced Ser368 phosphorylation (a known Cx43 PKC target site [21]) was observed (Figure 4A, row 4, lanes 4, 5).

Proteins that co-precipitated with Cx43 under these experimental conditions were analyzed by Western blot using antibodies specific for clathrin heavy chain (CHC) (Figure 4B). The amount of CHC that co-immmunoprecipitated with Cx43 in EGF-treated mES cell colonies was increased in average by 5.9±1.5 fold (p<0.01, n=3) when compared to untreated colonies (Figure 4B, lanes 1, 2), strongly supporting our earlier finding that EGF induced GJ endocytosis. Only a somewhat increased (insignificant) clathrin co-precipitation was observed in cells that were pre-incubated with PD98059 (2.7±0.9 fold; Figure 4B, lane 3), or PD98059 and chelerythrine-Cl combined (2.8±0.8 fold, Figure 4B, lane 5), consistent with effective but incomplete blockage of respective kinase activity in these drug-treated cells. Interestingly, when cells were pre-incubated with chelerythrine-Cl alone followed by EGF treatment, still a significant 4.91±1.3 (p<0.05) fold increase in GJ internalization was observed (Figure 4B, lane 4). Thus, chelerythrine-Cl did not significantly inhibit EGF-induced GJ endocytosis, suggesting that PKC-mediated phosphorylation of Ser368 is not sufficient to render Cx43-GJs internalization competent (see Discussion). No Cx43 or clathrin was immunoprecipitated in Cx43 negative MDCK control cells (Figure 4A, B, lanes 6).

Taken together, our data (summarized in Figure 5) suggest that EGF induces activation of the MAP kinase signaling cascade (including MEK and ERK1/2), and of PKC to phosphorylate Cx43 at Ser262, Ser279/282, and Ser368. These specific phosphorylation events render Cx43 protein subunits in GJs clathrin-binding capable to initiate the endocytosis of GJs in mES cells. Inhibiting MAPK activation by treating cells with the upstream MEK kinase specific inhibitor, PD98059, impedes Cx43 phosphorylation and in consequence, GJ internalization.

Discussion

EGF-mediated inhibition of GJIC in mES cell colonies has previously been observed by Park and colleagues [30]. However, no molecular mechanism/s explaining this observation have been reported. In WB-F344 and T51B rat liver epithelial cells, it was described that EGF-treatment induced inhibition of GJIC that was mediated by MAPK [23,37]. In principle, down-regulation of GJIC can be achieved by either GJ channel closure, and/or GJ endocytosis [38]. In 2004, by treating IAR20 rat liver epithelial cells in hypertonic medium, Leithe and Rivedal gained initial evidence suggesting that the EGF-induced inhibition of GJIC was mediated by the internalization of Cx43-GJs, and that internalization may had been mediated by a clathrin-driven endocytic process [39]. However, again detailed molecular signals that may be involved in this process have not been elucidated. In order to better understand the mechanism of EGF-induced GJIC-inhibition, we treated mES cell colonies with EGF, confirmed GJIC inhibition using dye-transfer assays (Figure 1), examined whether EGF-treatment would lead to GJ endocytosis using imaging and clathrin-co-immunoprecipitation based assays (Figures 3, 4, S1), and analyzed whether potential concomitant phosphorylation events on specific serine residues located in the known regulatory Cx43 C-terminal domain would correspond to EGF treatment (Figures 2, 4). Implementing phospho-specific antibodies and kinase inhibitors, we uncovered that EGF-treatment induced specific Cx43 phosphorylation, and that activation of ERK1/2 (MAPK signaling pathway) is responsible for EGF-mediated Cx43 phosphorylation at Ser262 and Ser279/282, whereas PKC activation is responsible for Ser368 phosphorylation (see Figure 5). Recently, Johnson et al. reported that in pancreatic cancer cells mutation of serine residues 279 and 282 that prevented phosphorylation of these residues led to GJ internalization-deficiency [40]. Consistent with their findings, our data demonstrate that ERK1/2-mediated phosphorylation of Cx43 on Ser279/Ser282 also occurs upon EGF-treatment in mES cell colonies, and that this phosphorylation is required for GJ internalization in these cells.

Interestingly, we did not find a significant increase in Ser255 phosphorylation (another MAPK Cx43 target site) upon EGF-treatment of mES cell colonies (Figures 2, 4A), as this was described to occur in 12-O-tetradecanoylphorbol 13-acetate (TPA)-treated epithelial IAR20 cells [35]. TPA is an analog of the second messenger, diacylglycerol (DAG) that is known to activate both, MAPK and PKC signaling pathways and to induce elevated phosphorylation and ubiquitination of Cx43 [35,41]. Observed differences might be explained by unalike experimental conditions (e.g. EGF versus TPA specificity) or they might be cell type specific (rat liver epithelial versus mES cells). In addition, besides its known cell-proliferative activity, EGF might have significant yet to be characterized additional effects on mES cells that require cellular uncoupling as well as GJ internalization (e.g. ES cell differentiation, migration, or channel-independent Cx43 signaling). Indeed, in recent years, the role of GJIC in mouse and human ES cell maintenance and differentiation, as well as its potential role in the reprogramming and differentiation of induced pluripotent stem (iPS) cells has attracted increasing attention. For example, it has been shown that GJIC is required for ES cells to maintain their pluripotent state [28]. In addition, Cx43 expression has recently been found to be important for the generation of iPS cells [27]. In 2008, Buehr et al. defined specific rat ES cell culture conditions that capture their authentic ES cell phenotype by using pharmacological inhibitors including MEK/ERK inhibitors [42]. Their findings together with our own results presented here indicate that ES cells utilize the EGF pathway, for example to regulate cell proliferation, differentiation as well as other essential events by modulating GJIC within the colonies. A clear understanding of the molecular mechanisms that lead to growth factor-induced inhibition of GJIC and GJ endocytosis are therefore of eminent importance.

We used PD98059, a MEK specific inhibitor, and monitored corresponding activation (phosphorylation) of ERK1/2 (upstream and downstream kinases of the MAPK signaling cascade, see Figure 5) to inhibit EGF-induced GJIC. In addition, our results show that EGF-receptor activation in mES cells also stimulates PKC to phosphorylate Cx43 on Ser368. However, inhibiting PKC-activation alone (by chelerythrine-Cl treatment) did not appear sufficient to inhibit EGF-induced GJ internalization, suggesting that activation of ERK1/2 (and the MAPK signaling cascade) may be more critical for EGF-induced GJ endocytosis than activation of PKC. Clearly, further studies will be required to fully understand a likely crosstalk between PKC and MAPK pathways in regulating GJIC and GJ internalization. Although the exact order of Cx43 phosphorylation events still needs to be determined, for the first time we provide compelling evidence that suggests that specific phosphorylation events renders Cx43 protein subunits in GJs clathrin-binding competent to induce GJ endocytosis.

• --The first signals that mediate gap junction endocytosis have been characterized
• --Specific phosphorylation events in Cx43 trigger gap junction endocytosis
• --MAPK and PKC phosphorylate connexin43 to mediate gap junction endocytosis
• --EGF induces gap junction endocytosis in mouse embryonic stem cell colonies
• --EGF induces MAPK and PKC mediated phosphorylation of connexin43 to endocytose gap junctions

Abbreviations used

AGJ

annular gap junction

CME

clathrin mediated endocytosis

EGF

epidermal growth factor

GJ

gap junction

GJIC

gap junction intercellular communication

LY

Lucifer yellow

MAPK

mitogen activated protein kinase

MEF

mouse embryonic fibroblast

mES cells

mouse embryonic stem cells

PKC

protein kinase C

PM

plasma membrane

RTK

receptor tyrosine kinase