Desmoglein 3 (Dsg3), the pemphigus vulgaris antigen, has recently been shown to be upregulated in squamous cell carcinoma (SCC) and has been identified as a good tumor-specific marker for clinical staging of cervical sentinel lymph nodes in head and neck SCC. However, little is known about its biological function in cancer. The actin-binding protein Ezrin and the activator protein 1 (AP-1) transcription factor are implicated in cancer progression and metastasis. Here, we report that Dsg3 regulates the activity of c-Jun/AP-1 as well as protein kinase C (PKC)-mediated phosphorylation of Ezrin-Thr567, which contributes to the accelerated motility of cancer cells. Ectopic expression of Dsg3 in cancer cell lines caused enhanced phosphorylation at Ezrin-Thr567 with concomitant augmented membrane protrusions, cell spreading and invasive phenotype. We showed that Dsg3 formed a complex with Ezrin at the plasma membrane that was required for its proper function of interacting with F-actin and CD44 as Dsg3 knockdown impaired these associations. The increased Ezrin phosphorylation in Dsg3-overexpressing cells could be abrogated substantially by various pharmacological inhibitors for Ser/Thr kinases, including PKC and Rho kinase that are known to activate Ezrin. Furthermore, a marked increase in c-Jun S63 phosphorylation, among others, was found in Dsg3-overexpressing cells and the activation of c-Jun/AP-1 was further supported by a luciferase reporter assay. Taken together, our study identifies a novel Dsg3-mediated c-Jun/AP-1 regulatory mechanism and PKC-dependent Ezrin phosphorylation that could be responsible for Dsg3-associated cancer metastasis.
Desmoglein 3 (Dsg3), the pemphigus vulgaris antigen, is one of seven transmembrane desmosomal cadherins.1, 2 Although a large body of evidence suggests that Dsg3 has a paramount role in cell–cell adhesion,1 an in vitro study showed that it mediates weak homophilic cell–cell adhesion.3 There are two pools of Dsg3 protein existing in epithelial cells, non-junctional (Triton soluble) and junctional (Triton insoluble); the latter is associated with the desmosomes.4, 5, 6, 7, 8 Our previous studies show that non-junctional Dsg3 is involved in E-cadherin signaling via Src and Rac1/Cdc42, suggesting cross talk between the desmosomal and classical cadherins in epithelial cells.8, 9, 10
Alterations in the expression of desmosomal components could contribute to cancer progression similar to E-cadherin;11, 12, 13 however, the correlation between the expression levels of desmosomal components and the state of tumor development has not been fully established. Recent studies have shown upregulation of Dsg3 in squamous cell carcinoma (SCC) in various tissues, particularly in the head and neck and lungs, which is correlated with T stage, N stage and overall stage.14, 15, 16, 17, 18 Moreover, DSG3 messenger RNA and protein have been identified as the best discriminatory (with 100% accuracy) biomarker among 40 potential candidate genes for accurate intraoperative staging of cervical sentinel lymph nodes in the head and neck SCC.18, 19, 20 These findings suggest strongly a pro-metastatic role for Dsg3 in tumor cell biology; however, the underlying molecular mechanism remains poorly understood.
Ezrin is a member of the Ezrin/Radixin/Moesin (ERM) proteins that act as plasma membrane-actin linkers and are enriched in microvilli, ruffles, filopodia, uropods and cell adhesion sites. The ERM proteins exist in two forms, active and inactive. The inactive proteins predominantly localize in the cytosol21, 22 and are present in a closed conformation. Recruitment of these proteins to the plasma membrane leads to the release of intramolecular interactions, exposing the binding site for actin at their C-terminus and for plasma membrane proteins such as CD44,23, 24 or adaptor proteins such as EBP5025 at the N-terminus. Phosphatidylinositol 4,5-bisphosphate binding and phosphorylation of Ezrin on Thr567 are involved in this activation process.26, 27, 28 Protein kinase C (PKC) and Rho kinase (ROCK) are responsible for Ezrin-Thr567 phosphorylation29, 30, 31, 32, 33, 34, 35, 36 and also activation of the transcriptional factor activator protein 1 (AP-1) downstream that regulates the expression of many genes, including those encoding for Ezrin and CD44, which are necessary for implementation of the invasion program.37 Ezrin participates in diverse cellular functions including cell adhesion, polarization and morphogenesis,21, 38, 39 and more importantly has been identified as a potent regulator of tumor cell invasion and metastasis.32, 33, 35, 40, 41, 42, 43 A positive correlation between Ezrin expression and cervical lymph node metastasis and clinical staging has been found in various tumors.43, 44, 45, 46, 47 In vitro RNA interference studies in cancer cell lines have demonstrated that knocking down Ezrin expression suppresses cell growth, invasion, tumor progression and metastasis,48, 49, 50 a phenotype that overlaps strikingly with that of Dsg3 silencing.14
In this study, we investigated the potential link between Dsg3 signaling and Ezrin activation, and describe a novel phosphorylation mechanism of the ERM proteins, primarily Ezrin-Thr567 that is regulated by Dsg3 in a PKC-dependent manner in A431 cells. We showed that non-junctional Dsg3 forms a complex with and is capable of activating Ezrin, which regulates actin-based cell shape change and migratory behavior. We confirmed that this pathway can be abrogated by inhibition of various signaling molecules including PKCs. Finally, we demonstrated a strong correlation between AP-1 activity and Dsg3 expression levels in cells. Collectively, our study uncovers a novel signaling role of Dsg3 that acts as a cell surface activator of AP-1 and the PKC/Ezrin pathway that promotes cancer cell migration and invasion.
Dsg3 induces morphological change and promotes membrane protrusion and cell spreading
To establish the role of Dsg3 in adhesion-induced membrane morphogenesis,51 we analyzed A431-hDsg3.myc cells10 and its derived clones with high or low Dsg3 levels and also an oral SCC line, SqCC/Y113 with transduced hDsg3.myc (Sq-D3). Recently, we have shown that Dsg3 functions as an upstream regulator of Rac1/Cdc42 in the regulation of actin organization and dynamics.9 In this study, we performed immunostaining for Ezrin and analyzed membrane coverage with Ezrin-positive protrusions. We showed that cells with high Dsg3 levels exhibited exaggerated membrane protrusions and cell spreading (Figure 1a and Supplementary Figure 1). Consistently, the transient transfection of hDsg3.myc construct in 293T cells resulted in striking membrane branching and micro-spikes compared with green fluorescent protein control cells (Supplementary Figure 1b). To demonstrate the specificity of the phenotype, Dsg3 in both A431-V and -D3 were depleted by RNA interference, and subsequent immunostaining showed that cells with Dsg3 depletion displayed remarkable collapse of the membrane protrusions and a concomitant inhibition of E-cadherin junction assembly (Figures 1b and c). These findings are consistent with our hypothesis that Dsg3 is involved, at least in part, in the regulation of actin organization.9 To address whether Dsg3 overexpression enhances cell–cell adhesion, we performed the hanging drop assay (Supplementary Figure 2) and consistently showed that overexpression failed to enhance cell–cell adhesion, as we reported previously.10
Dsg3 promotes cell migration and invasion
Previously, we have shown that overexpression of Dsg3 enhanced cell migration by a wound scratch assay.10 To complement this observation, we analyzed both A431 and SqCC/Y1 stably overexpressing Dsg3 under various conditions. Transwell migration assays indicated that all Dsg3-overexpressing cells exhibited greater cell motility, especially the A431-C2- and -C7-cloned cells (Figure 2a) while knockdown of Dsg3 had no effect in this migration assay. Consistent with this result, a transwell invasion assay showed that Dsg3 overexpression significantly promoted cell invasion compared with control in SqCC/Y1 cells (Figure 2b) and furthermore, we demonstrated that the enhanced migratory activity in Sq-D3 cells could be recapitulated in an organotypic cell invasion assay52 (Figure 2c) that showed a significantly greater number of cell clusters invading into the Matrigel:collagen matrix compared with control cells (Figures 2c–e). Collectively, these data suggest that Dsg3 promotes cancer cell migration and invasion.
Dsg3 associates with Ezrin at the plasma membrane
Our preliminary observation of the Dsg3/Ezrin colocalization10 led us to hypothesize that Dsg3 may have a role in regulating cell morphology and migratory behavior through a mechanism involving Ezrin. To address this question, we, first of all, analyzed protein–protein interaction by the proximity ligation assay (PLA) and demonstrated a marked increase of PLA signals that were generated by close proximity of Dsg3/Ezrin in Dsg3-overexpressing cells compared with control (Figure 3a). In line with this, the fluorescence resonance energy transfer (FRET) analysis of A431-D3 cells showed positive FRET (∼14.6% of regions of interest) predominantly localized to the membrane projections (Figure 3b). Furthermore, the association between Dsg3/Ezrin was verified by co-immunoprecipitation assay, demonstrating reproducibly that both endogenous and ectopic Dsg3 formed a complex with Ezrin in A431 cells in a Dsg3 dose-dependent manner (Figures 3c and d). To determine whether the association exists in the detergent soluble pool, that is, non-junctional Dsg3, we analyzed the association in Triton-soluble and -insoluble (NP-40 soluble) fractions by immunofluorescence and co-immunoprecipitation. For immunofluorescence, cells were treated with or without Triton buffer before fixation and immunostaining. Cells treated with Triton showed significant reduction of phospho-ERM and its colocalization with Dsg3, particularly at the cell borders (Supplementary Figure 3a and b). In support, co-immunoprecipitation demonstrated that the association predominantly existed in Triton-soluble fractions in all cell lines (Supplementary Figure 3c). Together, these results suggest that non-junctional Dsg3 forms a complex with Ezrin outside of desmosomes. In order to search for other potential-binding partners, mass spectrometry was applied. A construct of the cytoplasmic tail of Dsg3 tagged with Halo at the N-terminus was cloned into pFN21A (Promega, Southampton, UK) and the plasmid or Halo vector control (Vect) was transfected into A431 parental cells. The lysates were subjected to Halo tag purification and the resulting complex was analyzed using one-dimensional SDS–polyacrylamide gel electrophoresis proteomics approach. Proteins were separated using SDS–polyacrylamide gel electrophoresis and bands were digested by trypsin before further analyzed by liquid chromatography-tandem mass spectrometry (Supplementary Figure 4). Actin was identified as the major component present in the test sample but not in the empty vector (Supplementary Figure 4d).
Image analyses showed that changes in Dsg3 levels directly affected its colocalization with Ezrin (also phosphorylated form; Supplementary Figures 5 and 6) indicating an association between them. Detailed analysis of confocal image stacks revealed that the interaction was predominantly located at the basolateral domain of the plasma membrane where membrane protrusions were pronounced (Supplementary Figure 7). Quantitation of immunofluorescence intensity of Ezrin at the plasma membrane showed approximately a twofold increase in A431-D3 compared with control cells (immunofluorescence intensity/pixel in D3 vs Vect is 66.0±15.6 vs 36.7±16.1), implying that Dsg3 overexpression enhances Ezrin activity. Consistent with this notion, triple staining for Dsg3/pERM/F-actin showed a positive correlation and significantly enhanced colocalization of the three proteins at the cell borders in D3 compared with Vect cells, particularly in those stratified cells located in the center of the colonies (Figure 4a arrows). As Ezrin activation is associated with its interaction with CD44 and F-actin at the plasma membrane,23, 24 we evaluated the colocalization of Ezrin/F-actin or CD44/F-actinin A431-D3 with or without Dsg3 knockdown. As shown in Figures 4a and b, a significant reduction of their colocalization was seen in RNA interference-treated cells compared with controls.
Taken together, our findings demonstrate that non-junctional Dsg3 forms a complex with Ezrin at the plasma membrane and regulates its activity that is required for membrane projections and cell migration/invasion.31, 53
Dsg3 activates the ERM proteins, particularly Ezrin via the phosphorylation at Ezrin-Thr567
The phosphorylation of Ezrin-Thr567 leads to its activation, which has been implicated in cell invasion and metastasis.32, 35, 54, 55, 56, 57 To show that Ezrin was activated in response to Dsg3 overexpression, we analyzed the phosphorylation of ERM proteins including Ezrin-Thr567. Using a rabbit antibody (pERM) specific for the conserved C-terminal phospho-threonine of ERM proteins, we observed a significant increase in pERM levels in Dsg3-overexpressing cells compared with Vect or C11 that expressed low Dsg3 levels (Figure 5a). No change was seen for total ERM or Ezrin. To delineate the increased pERM was mainly due to activation of Ezrin-Thr567, we knocked down Ezrin or moesin in A431–C7 using an Ezrin- or moesin-specific small interfering RNA. Cell lysates were analyzed by western blotting for pERM. In the control the upper band—Ezrin and radixin (∼81 kDa) appeared to be almost double that of the lower (77 kDa) phospho-moesin (Figure 5b). The RNA interference-mediated Ezrin knockdown caused a significant reduction in the upper band, suggesting that Ezrin is the predominant phosphorylated isoform in A431 cells (Figure 5b), as reported previously.58 Thus, it is reasonable to conclude that the elevated pERM in A431-D3 and C7 (Figure 5a) was due to increased phospho-Ezrin rather than phospho-radixin or -moesin. To further validate this finding, we performed PLA using an antibody pair for Ezrin/pERM and showed a significant increase of PLA signals in D3 compared with the control, indicating the enhanced pERM indeed was phospho-Ezrin-Thr567 (Figure 5c). Ezrin activation was also reported to induce shape changes including cell rounding in addition to the membrane projections in A431 cells,21 and such a rounded cell phenotype coupled with pronounced membrane protrusions was also observable in A431-D3 compared with control cells or cells with low levels of Dsg3 in the same sample (Figure 5d). These results further support the notion that Dsg3 regulates Ezrin activity through phosphorylation at Thr567.
Dsg3-induced phosphorylation of Ezrin-Thr567 can be abrogated by inhibitors of PKC as well as various signaling molecules
To investigate whether Ezrin activation by Dsg3 was PKC-dependent, a dose–response experiment with the broad spectrum PKC inhibitor, bisindolylmakeimide I59 was performed. Although a significant inhibition of pERM was seen in all cell lines in a dose-dependent manner, the increased levels of pERM in Dsg3-overexpressing cells was abrogated effectively only at higher concentrations (>10 μM) comparable to that in control cells (only 5 μM; red dotted box in Figure 6a). This result implied that the increased pERM in Dsg3-overexpressing cells was PKC-dependent. Consistent with this observation, a time course study with bisindolylmakeimide I showed inhibition of pERM to a lesser degree in D3 and C7 compared with vector and C11 controls (Figure 6b).
Ezrin-Thr567 is also known as a substrate of ROCK,35, 36 and PKC-mediated ROCK activation increases actomyosin contractility and cell rounding,60 the phenotype also observed in our study (Figure 5d). Thus, a dose-dependent experiment with the ROCK inhibitor Y-27632 was performed and similar inhibition of pERM was observed in Dsg3-overexpressing cells. To further explore what other potential signaling molecules are likely involved in the Ezrin-Thr567 phosphorylation in our system, we tested various inhibitors including Rö-31-7549 for conventional PKC,61 PP2 for Src,10 C3 for RhoA, Y-27632 for ROCK,62 NSC23766 for Rac163 and SB202190 for p38,64 and found inhibition at different degrees for most inhibitors except for PP2 (Figure 6c). As conventional PKCs are Ca2+-dependent, whereas novel and atypical isoforms do not require Ca2+ for their activation, we wanted to determine which group of PKCs was responsible for the ERM phosphorylation in our system. We performed a dose-dependent experiment with three PKC inhibitors bisindolylmakeimide I (broad spectrum), Rö-31-7549 and Gö6976 (conventional), and showed significant inhibition of pERM by all inhibitors tested in a dose-dependent manner (Supplementary Figure 8a), suggesting both conventional and nonconventional PKCs are involved in ERM phosphorylation. In addition, we conducted an experiment with these inhibitors in conjunction with a calcium switch and observed that in Ca2+-free conditions, the serum-induced pERM could be abrogated effectively by bisindolylmakeimide I or Rö-31-7549 but to a lesser extent by Gö6976 (Supplementary Figure 8b). Addition of calcium in the absence of inhibitors caused an increase of pERM compared with dimethyl sulfoxide control in calcium-free conditions indicative of calcium-induced activation of PKCs. Calcium addition also enhanced the expression of Dsg3, as we reported previously.8 We observed that the calcium-induced pERM could only be partially abrogated by PKC inhibition (Supplementary Figure 8b, plus Ca2+) and this incomplete inhibition suggests that multiple mechanisms (as shown in Figure 6c), either mediated by calcium or Dsg3, are involved in ERM phosphorylation.
Dsg3 enhances the phosphorylation of c-Jun and activates the transcriptional activity of AP-1
To explore the involvement of other kinases, we performed a phospho-kinase profiler array study and observed that a number of kinases were increased in Dsg3-overexpressing cells compared with Vect. In addition to the Src family as we have previously reported,8, 10 increased phosphorylation was also seen for c-Jun S63 (2.6-fold), PLCγ-1-Y783 (1.6-fold) and p70 S6 Kinase-T229 (1.8-fold; Figure 7a). C-Jun is part of the AP-1 transcription factor, a proto-oncogene that has been implicated in the regulation of a variety of cellular responses including cell migration and oncogenesis. The phosphorylation of S63 at the N-terminal transactivating domain of c-Jun is known to be mediated by Jun kinase,65 suggesting the Jun kinase pathway is activated by Dsg3. To confirm that the AP-1 activity is indeed induced by Dsg3, we performed a luciferase reporter assay in A431 parental cells with transient co-transfection of an AP-1 reporter construct together with various concentrations of the pBABE-hDsg3.myc construct and observed a strong dose-dependent change in AP-1 activity in relation to Dsg3 expression levels with ∼14-fold increase in cells with the highest amount of Dsg3 complementary DNA (Figure 7b). Furthermore, this finding was validated by loss-of-function analysis showing a more than twofold reduction in luciferase activity in Dsg3 knockdown cells compared with those treated with scrambled control small interfering RNA (Figure 7c). Finally, we tested whether the enhanced luciferase activity can be abrogated by various inhibitors and observed partial inhibition with inhibitors of PKC, ROCK and p38.
This study identifies the desmosomal cadherin Dsg3 as an upstream cell surface activator for AP-1 and the PKC/Ezrin pathway in the control of cell motility and invasion in cancer (Figure 8). Dsg3 and Ezrin have been independently implicated in cancer progression and metastasis.32, 33, 35, 40, 41, 42, 43 This study provides direct evidence of cross talk between Dsg3 and Ezrin at the plasma membrane that is responsible, at least in part, for promoting cell migration and invasion in Dsg3-associated SCC. We demonstrated that Dsg3 is capable of forming a complex with Ezrin at the plasma membrane and regulates its phosphorylation at the Thr567 residue, the activity required for proper interaction with F-actin and CD44 that is a prerequisite for cell migration and invasion. We also showed that the Dsg3-induced Ezrin-Thr567 phosphorylation can be abrogated by various inhibitors including those for PKC and ROCK, indicating the likelihood that both kinases are partially regulated by Dsg3. Furthermore, we discovered that Dsg3 also regulates the activity of AP-1, via phosphorylation of c-Jun S63, a transcription factor that has a pivotal role in tumor metastasis.
Cell invasion is one of the hallmarks of cancer. Our previous studies showed that overexpression of human Dsg3 in cancer cell lines causes dissolution of both classical and desmosomal cadherins via Src activation.8, 10 However, to enable tumor cells to invade into the neighboring tissues, additional properties that involve multigenic processes and coordinated rearrangements of the actin cytoskeleton are required. AP-1 is known as a proto-oncogene and a critical regulator of a complex gene expression program that defines the invasive phenotype.65 Ezrin is one of its functional effectors and has a key role in remodeling actin dynamics and cell adhesion that associates with invasion. AP-1 comprises heterodimers of Fos and Jun or homodimers of Jun that regulate gene expression by binding to a consensus DNA motif in the promoter region of target genes. Sustained activation of AP-1 is required for transformation by many oncogenes. It is well known that the growth factor signaling pathway is an upstream regulator for sustained activation of AP-1.66 In this study, we report that the cell adhesion protein Dsg3 also functions as an upstream activator of AP-1 and is capable of activating one of its target effectors, Ezrin,37 at the plasma membrane to promote cell shape change and migratory capacity.
Dsg3 (pemphigus vulgaris antigen) has recently been implicated in tumor metastasis and progression of SCC in multiple organs, especially the head and neck and lungs,14, 15, 16, 18, 19, 20 suggesting it has a pivotal role in cancer cell biology. In support, our gain-of-function studies demonstrated that, rather than promoting cell–cell adhesion, overexpression of this gene results in compromised intercellular adhesion and a reduction in the expression of classical and desmosomal cadherins.8, 10 In this study, we describe an additional mechanism by which Dsg3 promotes cell invasive capability by activating AP-1- and the PKC-dependent Ezrin-Thr567 pathway that are likely to be responsible for tumor metastasis in SCC. Evidence that supports this conclusion are as follows: (1) overexpression of Dsg3 enhances membrane protrusions, cell spreading and rounding that are the necessary prerequisites for cell migration/invasion; (2) overexpression of Dsg3 accelerates cell migration/invasion in both A431 and SqCC/Y1 cancer cell lines; (3) Dsg3 forms a complex with and activates Ezrin at the Thr567 residue that is dependent upon several kinases including PKC; (4) modulation of Dsg3 levels directly influenced the activity of Ezrin that directly associates with its interaction with F-actin and CD44 at the plasma membrane; and finally, (5) Dsg3 functions as an upstream activator of AP-1that could be abrogated by PKC inhibition.
The positive role of Ezrin in cancer progression and metastasis has been shown by many studies in the context of cortical cytoskeleton organization, filopodia extension, cell spreading, proliferation, survival and motility.32, 67, 68 Ezrin activation is dynamically regulated in a spatiotemporal manner during metastatic progression.33 Given that upregulation of Dsg3 and Ezrin has been independently found in lymph node metastasis of the head and neck cancer,18, 33, 40, 41, 42 our study suggests it is likely that the Dsg3/PKC/Ezrin-Thr567 pathway is activated at the leading edge or invasive front, in lymph node metastasis of the head and neck cancer.18, 19, 20
A number of transmembrane proteins including the hyaluronan receptor, CD44, have been characterized as ERM-binding proteins.24, 69 This study has identified Dsg3 as a novel membrane protein that associates with Ezrin and is able to regulate its activation. We showed that both endogenous and ectopic Dsg3 form a complex with Ezrin at the plasma membrane where membrane projections are pronounced. Dsg3 knockdown not only affected its association with Ezrin but also impaired E-cadherin adhesion and the interaction between Ezrin and CD44/F-actin, indicating the functional significance of such a complex. Although the actual nature of the binding between Dsg3 and Ezrin remains to be elucidated, our mass spectrometric analysis suggests it could be indirect that warrants further investigation. Collectively, the findings from this study and our previous reports8, 9, 10 support the view that non-junctional Dsg3 has a signaling role in epithelial cells, acting as a membrane receptor in regulating a diverse array of cellular events that control actin-based cell–cell adhesion, cell shape determination and migratory behavior. It is possible that cells use Dsg3 to recruit Ezrin to the basolateral domain to form membrane projections, such as filopodia and lamellipodia, in order to facilitate contacts with neighboring cells that is a prerequisite for junction formation and cell polarization in normal keratinocytes.51 However, this process is somehow hijacked in transformed tumor cells where the stability and organization of intercellular junctions are drastically altered. Evidence also suggesting the signaling roles of Dsg3 comes from many in vitro studies based on pemphigus vulgaris (PV-IgG).70, 71, 72 Nevertheless, no report has yet shown its association with Ezrin, although the link between actin and pemphigus acantholysis is just beginning to be appreciated.73
Serine/threonine kinases such as ROCK, PKC, p38 and LOK have been identified as the ERM kinases for phosphorylating the C-terminal threonine residue in a different cellular context.32, 74, 75, 76, 77 Among them, PKC family isoforms seem to be ubiquitously involved in all cell types. PKC isoforms such as α, γ and ι are capable of phosphorylating Ezrin-Thr567 at the plasma membrane, particularly the membranous protrusions that control cell motility33 and agents that block this phosphorylation also abrogate PKC α-dependent migration.32 Dsg3 is a calcium-dependent protein and becomes stabilized (increase in expression levels) upon calcium switching8 that also activates PKC (conventional). Our results with the PKC inhibitors that showed significant inhibition of pERM in Dsg3-overexpressing cells indicate it is likely that the Dsg3-mediated Ezrin-Thr567 phosphorylation involves PKCs, particularly calcium-dependent conventional isoforms. There are seven PKC isoforms, that is, α, βII, δ, η, ɛ, λ, ζ, which were shown to be expressed in A431 cells.78 To delineate which isoform(s) is responsible for Dsg3-mediated Ezrin activation merits further investigation. In addition, our studies (this and the study by Tsang et al.9) also support the general notion that other kinases such as ROCK, p38 MAPK, Rac1 and RhoA may also have a role in Dsg3-mediated Ezrin activation, as our phospho-kinase array and the inhibitor analyses showed a wide range of signaling pathways that had been activated in Dsg3-overexpressing cells (Figures 6c and 7a). Positive feedback from Ezrin to Rho GTPases and ROCK was not ruled out here.60 Evidence indicating that Rac1 is coupled with Ezrin activation in the control of E-cadherin-mediated adherens junction assembly is documented in the literature.79 In summary, our findings suggest that Dsg3 might serve as a master cell surface regulator for a variety of signal transduction pathways that influence the activity of Ezrin, one of the effector proteins of these signaling cascades.
In conclusion, this study provides the first evidence of the desmosomal cadherin Dsg3 that functions as a key membrane activator for AP-1 and Ezrin activation that are associated with various cellular responses such as membrane dynamics, cell spreading, migration and invasion. This pathway may involve several signaling molecules including PKCs that are regulated by Dsg3.8, 9, 10 These findings will advance our understanding of the role of Dsg3 in tumor progression and invasion, and likely facilitate development of new therapeutic strategies for the head and neck and lung cancers.
Materials and methods
Cell culture and reagents
A431 cells were maintained in Dulbecco’s modified Eagle medium (DMEM) plus 10% fetal calf serum. Human oral SCC cell line SqCC/Y1 was cultured in EpiLife (Invitrogen, Life Technologies Ltd, Paisley, UK) or keratinocyte growth medium.13 Primary human oral fibroblasts were cultured in DMEM plus 10% donor calf serum and cells with passage number <10 were used. A431 clones with transduction of hDsg3.myc10 were obtained using the limited dilution assay. The protocol, as described previously,10 was used for the generation of stable SqCC/Y1 lines with transduction of human Dsg3 as well as the matched Vects. Antibodies and small interfering RNA sequences are listed in Supplementary Tables 1 and 2.
Immunofluorescence and confocal microscopy
Immunofluorescence was performed as described previously.10 The coverslips were examined with a Zeiss Meta 510LSM/510LSM laser scanning confocal (Carl Zeiss, Cambridge, UK) or Leica DM5000 epi-fluorescence microscope (Leica Microsystems (UK) Ltd. Milton Keynes, UK).
Transwell cell migration and invasion assays
The assays were conducted with 24-well transwell inserts, which contain a permeable membrane with 8 μm pore size (VWR International Ltd, West Sussex, UK). Cells (1 × 105) in basal medium (alpha-DMEM or DMEM:Ham’s F12 at 3:1) plus 1% bovine serum albumin were placed on the upper layer of an insert, which was placed into a 24-well plate containing 500 μl alpha-minimum essential medium plus 10% fetal calf serum for A431 cells and keratinocyte growth medium for SqCC/Y1, which acted as a chemoattractant. The plate was kept in an incubator for 2–3 days to allow cells to migrate into the bottom chamber. The total cell number migrating to the bottom chambers was determined using the CASY cell counter (Roche Innovatis AG, Reutlingen, Germany). For the cell invasion, Matrigel diluted 1:3 in DMEM was loaded on the upper layer of the inserts and allowed to gel uniformly before addition of the cell suspension.
The Matrigel:collagen organotypic raft culture is described elsewhere.80
Histology, immunohistochemistry and quantitative analysis
Raft cultures were collected after 14 days and fixed in Trumps (100 mM NaH2PO4, 67.5 mM NaOH, 4% formaldehyde, 1% glutaraldehyde). Specimens were subjected to routine procedures for histology. Five-micrometer sections were stained with hematoxylin and eosin or for keratin with LP34, followed by Dako Cytomation EnVision+dual Link System Peroxidase. The peroxidase was visualized using 3,3′-Diaminobenzidine (DAB) substrate (DAKO, Dako UK Ltd, Cambridgeshire, UK). Quantitation of cell invasion is described elsewhere.80 Images were acquired in a MBF stereology system (MBF Bioscience, Williston, VT, USA) and analyzed by ImageJ (Wayne Rasband, NIH, Bethesda, MD, USA). Parameters such as particle number, total area and average size of particles, and invasion depth were analyzed. The ‘invasion index’ was calculated.80
A Duolink in situ PLA for protein–protein interactions
The PLA was conducted following the protocol described in the Duolink PLA kit (Cambridge Bioscience Limited, Cambridge, UK).
FRET efficiency was measured using acceptor photobleaching as described previously.10 Eighty-nine regions of both junctional and free edges were analyzed and FRET efficiency was calculated, and the FRET efficiency >5% was considered to be positive.10
Co-immunoprecipitation, protein fractionation and western blotting
Human phosphokinase profiler array
A431-Vect and -D3 cell lines seeded at equal density in 10-cm culture dishes were extracted and the phosphokinase array was carried out using 500 μg of total protein in each sample, according to the protocol in the kit (Catalog number: ARY003, R&D Proteome Profiler Array, R&D Systems Europe Ltd., Abingdon, UK).
An AP-1 reporter construct was made using the pGL4.26 vector (Promega) containing a minimal eukaryotic promoter. An oligo containing six repeats of the consensus AP-1-binding motif (TGAC[G]TCA) was cloned into KpnI and BglII sites of pGL4.26.AP-1 promoter. A431 parental cells seeded in 6-well plate were co-transfected with equal amount (1 μg) of AP-1 reporter plasmid or empty pGL Vect along with pBABE-hDsg3.myc10 at various concentrations using Fugene HD reagent (Promega). The total amount of pBABE plasmid was topped up by pBABE-GFP to 1 μg in each transfection. Cells were extracted after 2 days or were subjected to serum starvation for 1 day before treatment with various inhibitors. Luciferase assays with cell lysates were performed within 3 days of transfection using the Luciferase Assay System (Promega). Luciferase activities were normalized against protein concentration determined by Bio-Rad DC protein assay (Bio-Rad Laboratories Ltd., Hertfordshire, UK).
We thank Dr MT Teh, Professor K Parkinson, Dr SM Tsang, Dr MS Ikram and members of CDOS for helpful discussion and technical assistance. This work was supported by a British Skin Foundation-funded studentship awarded to HW and in part by the Institute of Dentistry as well as The Facial Surgery Research Foundation-Saving Faces.
About this article
Supplementary Information accompanies this paper on the Oncogene website (http://www.nature.com/onc)
British Journal of Cancer (2015)