Enhanced Activation of Epidermal Growth Factor Receptor Caused by Tumor-Derived E-Cadherin Mutations

Classic cadherins are a family of transmembrane glycoproteins that mediate calcium-dependent cell-cell adhesion (1). Cadherins play a critical role in normal development, and alterations in cadherin function have been implicated in tumorigenesis (2). In addition to their adhesive functions, cell adhesion molecules can modulate several signal transduction pathways.

E-cadherin, the prototype of the classic cadherin family, functionally interacts with receptor tyrosine kinases (RTK), components of the Wnt signaling pathway, and Rho-family GTPases (3). Activation of tyrosine kinases, such as epidermal growth factor receptor EGFR, c-Met, or Src, provokes tyrosine phosphorylation of components of the cell adhesion complex such as E-cadherin and β-catenin, resulting in disruption of cell adhesion and endocytosis of E-cadherin (4, 5). On the other side, functional adhesion junctions can affect the activity and localization of RTKs. EGFR has been shown to colocalize with E-cadherin and to form a multicomponent complex that includes E-cadherin (6). Furthermore, recent results show that E-cadherin–mediated adhesion inhibits ligand-dependent activation of EGFR, also pointing to a crosstalk between EGFR and E-cadherin with potential relevance for tumor and metastasis formation (7, 8). Complex formation of E-cadherin with EGFR and other RTKs was shown to be mediated through the extracellular domain of E-cadherin, thereby decreasing receptor motility and ligand binding affinity (7). In contrast, E cadherin has been found to activate transiently EGFR when cell-cell contacts were formed by switching from low-calcium conditions to high calcium (9). Taken together, these findings imply that the regulation of RTKs and E-cadherin is bidirectional.

In 50% of diffuse-type gastric carcinomas, somatic E-cadherin mutations are found (10). We reported previously that gastric carcinoma–derived somatic E-cadherin mutations are preferentially located in a mutational hotspot region within the second and third extracellular domains of E-cadherin (10). These mutations are mainly splice site and point mutations, affecting putative calcium binding sites (11). Typical gastric cancer-associated in frame deletions of exons 8 (del 8) or 9 (del 9) resulted in decreased adhesion and increased motility of the cells (11–14). The motility enhancement by mutant E-cadherin was sensitive to treatment with an EGFR tyrosine kinase inhibitor (12), identifying the EGFR signaling network as a mediator of the effect. However, the molecular mechanism linking E-cadherin mutations and the EGFR signaling pathway remained unresolved.

The aim of the present study was to understand the functional relationship between E-cadherin harboring a deletion of exon 8 and EGFR. In particular, we tested the hypothesis that activation of EGFR signaling might be observed in the presence of the somatic E-cadherin mutation. We also investigated whether the E-cadherin mutation still permits association with EGFR because complex formation between E-cadherin and the EGFR was described to depend on the extracellular domain of E-cadherin (7). Moreover, the effect of the mutation on endocytosis of E-cadherin and EGFR was examined. Finally, activation of EGFR was analyzed in gastric cancer samples with E-cadherin exon 8 or 9 deletion mutations.

Cell cultivation and transfection. The human E-cadherin–negative breast cancer cell line MDA-MB-435S [American Type Culture Collection (ATCC)] or mouse L929 fibroblasts (ATCC) were previously transfected with wild-type or del 8 E-cadherin cDNA, and stable transfectants were established (11). Cells were cultivated in DMEM (Life Technologies) supplemented with 10% FCS (PAN Biotech) and penicillin-streptomycin (50 IU/mL and 50 mg/mL; Life Technologies) at 37°C and 5% CO₂. The MDA-MB-435S strain evolved from the parent line MDA-MB-435, isolated from the pleural effusion of a female with metastatic ductal adenocarcinoma of the breast (15). Two different clones were investigated for each E-cadherin expression construct in this study.

Western blot analysis. Extraction of proteins from cultured cells was performed with L-CAM lysis buffer [140 mmol/L NaCl, 4.7 mmol/L KCl, 0.7 mmol/L MgSO₄, 1.2 mmol/L CaCl₂, and 10 mmol/L HEPES (pH 7.4), containing 1% (v/v) Triton X-100; ref. (16)]. The lysis buffer contained 2 mmol/L phenylmethyl-sulfonylfluoride, 2 mmol/L orthovanadate, 19 μg/mL aprotinin, 20 μg/mL leupeptin, 10 mmol/L sodium phosphate, and 100 mmol/L sodium fluoride. For the calcium-depletion procedure, EGTA was added to the medium (2 mmol/L) and cells were incubated for 6 min. If cells were stimulated before lysis, they were treated with 100 ng/mL epidermal growth factor (EGF) for the indicated period of time. Proteins were separated by SDS-polyacrylamid gel electrophoresis followed by a transfer to a nitrocellulose membranes (Schleicher & Schuell). Cell lysates were investigated with polyclonal anti-EGF receptor phosphospecific antibodies directed against tyrosine 1068, 1148, and 1173, respectively (Biosource; purchased from Invitrogen), polyclonal anti-EGFR (Ab-6) antibody (Calbiochem; Merck KGaA), and monoclonal anti–E-cadherin antibody (clone 36; BD Biosciences). Monoclonal anti–α-tubulin antibody (Sigma) was applied to stain α-tubulin as a loading control. For signal quantification, blots were scanned and densitometric analysis was performed with Scion Image Software from Scion Corporation (Version Beta 4.0.2; Frederick).

Immunoprecipitation. Cells were seeded at a density of 2.5 × 10⁶ on 6-, 10-, and 15-cm tissue culture plates and cultured for 24 h in DMEM with 10% FCS or 8 h in DMEM with 10% FCS, followed by a 16-h incubation in DMEM without FCS. If cells were stimulated before lysis, they were treated with 100 ng/mL EGF for 1 min. As a control, EGFR activation was blocked by application of 6.3 μmol/L Tyrphostin AG 1487 (Calbiochem), which was added during the 16-h incubation in DMEM without FCS. Five hundred micrograms of protein lysate and polyclonal anti-EGFR (1005; sc-03 antibody; Santa Cruz) was used to precipitate EGF receptor with Catch and Release v2.0 Reversible Immunoprecipitation System (Upstate, now part of Millipore) according to the manufacturer's instructions. Immunocomplexes were eluted and then subjected to immunoblotting analysis using monoclonal anti–E-cadherin, anti-growth factor receptor binding protein 2 (Grb2), and anti-Shc antibodies (BD Biosciences).

Glutathione S-transferase-Raf1-Ras binding domain of Raf1 pull-down assay. Cells were seeded at a densitiy of 3 × 10⁶ on 10- or 15-cm tissue culture plates. After 8 h, cells were washed twice with PBS and cultured for another 16 h in DMEM without FCS before they were lysed. If cells were stimulated before lysis, they were treated with 100 ng/mL EGF for 1 min. Using equal amounts of protein lysate, the glutathione S-transferase (GST)-Raf1-Ras binding domain of Raf1 (RBD) pull-down assay was performed according to the manufacturer's instructions using the EZ-Detect Ras Activation kit (Pierce). Thirty microliters of eluate were separated by SDS-polyacrylamid gel electrophoresis followed by a transfer to a nitrocellulose membrane. Activated Ras was detected with monoclonal anti-Ras antibody (Pierce).

Quantitative indirect immunofluorescence assay. To quantify cell-surface E-cadherin by flow cytometry, an indirect immunofluorescence assay (QIFIKIT; Dako Diagnostika GmbH) was performed according to the manufacturer's instructions. MDA-MB-435S and L929 transfectants were harvested with versene, and 5 × 10⁵ cells were incubated with 40 μg/mL monoclonal antibody directed to E-cadherin (SHE78-7; Alexis Deutschland) for 1 h on ice. Then, cells and calibration beads that were coated with different quantities of mouse monoclonal antibody molecules were labeled in parallel with FITC-conjugated goat anti-mouse IgG (diluted 1:50; delivered with the QIFIKIT) for 1 h on ice. Cells and calibration beads were analyzed on a Beckman Coulter Epics XL (Beckman Coulter). The antigen quantity was expressed in antibody binding capacity (ABC) units. The calibration beads were used for construction of the calibration curve against ABC. ABC of the analyzed cells was calculated based on the equation of the calibration curve.

Flow cytometry. To assess the surface localization of EGFR, cells were harvested with versene and 5 × 10⁵ cells were incubated with 20 μg/mL monoclonal antibody directed to EGFR (BD Biosciences) for 1 h on ice in PBS, washed with 0.1% sodium azide and 0.1% bovine serum albumin (Sigma), and stained with dichlorotriazinylaminofluorescein (DTAF)-conjugated anti-mouse IgG for 1 h on ice. To examine the internalization of EGFR in response to EGF, cells were seeded at a density of 1 × 10⁶ cells per 10-cm dish, serum starved overnight, and treated with 100 ng/mL EGF for 4 h. Cells were harvested with versene, and 5 × 10⁵ cells were incubated with 5 ng/mL monoclonal antibody directed to EGFR (Upstate Biotechnology) for 1 h on ice in PBS, washed with 0.1% sodium azide and 0.1% bovine serum albumin (Sigma), and stained with DTAF-conjugated anti-mouse IgG DTAF-coupled secondary antibody (goat anti-mouse IgG; Jackson ImmunoResearch Laboratories purchased from Dianova) for 1 h on ice. Purified mouse IgG1 (PharMingen) were used as κ immunoglobulin isotype controls. Cells were analyzed on a Beckman Coulter Epics XL (Beckman Coulter).

Immunohistochemical analysis. The following antibodies were used for immunohistochemistry: antiphosphorylated EGFR [rabbit polyclonal antibody reacting with EGFR only when phosphorylated at tyrosine residue 1086, #36-9700 obtained from Zymed Laboratories, and mouse monoclonal antibody reacting with EGFR only when phosphorylated at tyrosine residue 1068 (1H12) and #2236 from Cell Signaling Technology obtained from New England Biolabs], monoclonal rat E-cad delta 8-1 detecting mutant E-cadherin protein lacking exon 8 (del 8 mutation; ref. 17), monoclonal rat E-cad delta 9-1 detecting mutant E-cadherin protein lacking exon 9 (del 9 mutation), clone 7E6 (18), and anti–E-cadherin antibody HECD-1 (Alexis Deutschland).

Manual staining protocols were used for phosphorylated EGFR (pEGFR) and E-cadherin staining with the following antibody dilutions: pEGFR (Y1086), 1:50; pEGFR (Y1068), 1:400; del 8 E-cadherin, 1:5; and del 9 E-cadherin undiluted and normal E-cadherin, 1:500. Antigen retrieval was performed using citrate buffer (pH 6) using a microwave (for pEGFR, 2 × 10 min) or a pressure cooker (for HECD-1 for 7 min). A peroxidase block (3% H₂0₂ for 15 min at room temperature), an avidin biotin block (Vectastain; twice for 15 min at room temperature), and blocking with 5% antigoat serum in Dako dilution solution (1 h at room temperature) were performed. Staining was carried out with LSAB-3,3′-diaminobenzidine or Fast red from Dako Diagnostika GmbH.

Immunohistochemical analysis of 10 preselected formalin-fixed, paraffin-embedded tumor specimen with the somatic del 8 or del 9 E-cadherin mutations was performed using antibodies E-cad delta 8-1 and E-cad delta 9-1. The mutation-specific E-cad delta 8-1 and delta 9-1 antibodies stained exclusively tumor cells. As positive controls, we used formalin-fixed and paraffin-embedded cell pellets from MDA-MB-435S cells transfected with del 8 or del 9 E-cadherin (11). pEGFR was detected using anti-EGFR phosphospecific antibodies directed against tyrosine 1068 or 1086, respectively. The pEGFR staining was cytoplasmic and membranous. As positive control, we used formalin-fixed and paraffin-embedded cell pellets from EGF-treated A431 cells. Immunohistochemical staining of 10 diffuse and 10 intestinal gastric cancers without del 8 or del 9 E-cadherin reactivity was performed using a gastric cancer tissue microarray. pEGFR was detected as described above, and E-cadherin was stained with the HECD-1 antibody. Tumors were considered as E-cadherin positive when they showed strong membranous staining. The protocol was reviewed and approved by the local ethic committee, and informed consent was obtained according to institutional regulations.

EGF-induced phosphorylation of EGFR by a somatic E-cadherin mutation. It has been reported recently that E-cadherin can negatively regulate ligand-dependent activation of several RTKs in an adhesion-dependent manner (7). To determine whether the somatic del 8 E-cadherin mutation influences EGFR activation, we first analyzed the phosphorylation level of EGFR by immunoblotting using phosphoepitope-specific antibodies. Detailed kinetic analysis of EGF-dependent activation of EGFR was performed in wild-type and del 8 E-cadherin–expressing MDA-MB-435S cells at 40% confluency. Cells were treated with EGF for different times within the range of 1 to 60 min. Transient phosphorylation of EGFR at tyrosine residue 1173 was observed with a maximum at 1 min stimulation time, followed by a rapid decrease (Fig. 1A). The course of EGFR activation was strongly influenced by the presence of the E-cadherin mutation because the level of phoshorylated EGFR at Y1173 was 47.3-fold elevated in del 8 cells compared with 17.6-fold in wild-type cells (Fig. 1B). This result was confirmed with a panel of antibodies recognizing EGFR phosphorylated at tyrosine residues 845, 1068, and 1148 (data not shown). Notably, the expression levels of EGFR and E-cadherin remained unchanged during the time of stimulation (Fig. 1C). Taken together, these results suggest that the activity level of EGFR is increased in the presence of the somatic E-cadherin mutation.

EGF-induced EGFR activation is cell density dependent. Next, the influence of cell density and calcium depletion on EGF-induced phosphorylation of EGFR was determined to clarify the role of adhesive cell-cell contacts. EGFR phosphorylation was dependent on cell density and was, in general, higher in low-dense (40% confluency) than in high-dense (95% confluency) wild-type and del 8 cells (Fig. 2A and B). Again, EGF-dependent tyrosine phosphorylation of EGFR was stronger in del 8 than in wild-type E-cadherin–expressing cells. This effect was prominent in low-dense cells, but it was still detectable in confluent cells.

To examine whether the lower response to EGF at high density was dependent on E-cadherin (wild-type or del 8) cell-cell adhesion, high density cells were incubated with EGTA to deplete calcium. Reduction of extracellular calcium disrupts cell-cell contacts and induces the internalization of E-cadherin. Under these conditions, EGFR activation was increased compared with untreated high density cells and was almost as strong as in low density cells with little surface E-cadherin (Fig. 2A and B). Together, in our cell system, we confirmed previous reports that wild-type E-cadherin can negatively regulate EGFR activation (7). Moreover, we show that this cell density–dependent EGFR-regulating function is preserved in mutant E-cadherin. The largest differences in EGFR activation between wild-type and mutant E-cadherin were observed under subconfluent conditions where wild-type cells form more E-cadherin–mediated cell-cell contacts than del 8 cells (11). This result together with the calcium depletion experiment suggests that the effect of E-cadherin on EGFR is adhesion-dependent.

Activation of EGFR downstream signaling. Binding of EGF to the extracellular domain of the EGFR causes dimerization of the receptor and its phosphorylation on several tyrosine residues within the cytoplasmic domain. The phosphorylated tyrosine residues serve as specific docking sites for the Src homology-2 or phosphotyrosine binding domains of intracellular signal transducers and adaptors, leading to their colocalization and the assembly of multicomponent signaling complexes. For further evaluation of the enhanced EGFR activation in the presence of mutant E-cadherin, the interaction of EGFR and the adaptor proteins Shc and Grb2, which are recruited to the receptor after its activation, was analyzed.

Cell extracts of untreated and EGF-stimulated wild-type and del 8 E-cadherin–expressing MDA-MB-435S cells were immunoprecipitated with anti-EGFR antibody, and binding of Grb2 and Shc was analyzed by Western blot analysis. Consistent with the enhanced tyrosine phosphorylation of EGFR in del 8 cells, both Grb2 and Shc bound stronger to EGFR in del 8 cells than in wild-type E-cadherin transfectants (Fig. 3A). Interestingly, in unstimulated cells, binding of EGFR to both adaptors was detectable, suggesting that there is a basal level of activated EGFR in the cells. To show specific involvement of EGFR, adaptor recruitment of Shc as an example was shown to be sensitive to treatment with the EGFR inhibitor Tyrphostin AG 1478 (Fig. 3B). These data suggest that EGFR activation is reflected by enhanced binding of Shc and Grb2 to the receptor.

Mutant E-cadherin increases the level of active Ras. After the activation of the EGFR, Grb2, which is constitutively bound to the Ras exchange factor Sos, is recruited from the cytosol to the receptor at the membrane. This relocation of the Grb2/Sos complex facilitates the interaction of membrane-associated Ras with Sos, resulting in the exchange of Ras-bound GDP for GTP. To further analyze the influence of mutant E-cadherin on downstream intracellular signaling after EGFR activation, the level of active Ras was determined in the presence of both wild-type and del 8 E-cadherin.

We performed Raf1-RBD pull-down assays using untreated and EGF-stimulated MDA-MB-435S transfectants seeded at different cell densities (Fig. 3C). In the unstimulated situation, higher levels of active Ras were found in del 8 than in wild-type E-cadherin–expressing cells. After EGF-stimulation, the level of Ras-GTP in wild-type E-cadherin–expressing cells increased to the level observed in del 8 E-cadherin transfectants. In both cell lines, the Ras-GTP level was elevated in low density cells compared with high density cells. The differences at low versus high density were not related to changes in the expression levels of Ras or E-cadherin (Fig. 3C and data not shown). These results indicate that mutant E-cadherin–dependent EGFR activation is mirrored by a higher level of active Ras.

EGFR forms a complex with wild-type and del 8 E-cadherin. E-cadherin was described to form a multicomponent complex with EGFR (6). Several groups have shown that the extracellular domain of E-cadherin is critical for this interaction (7, 19). In the light of the involvement of E-cadherin in regulation of EGFR activation, the physical interaction between both molecules was characterized. Complex formation of wild-type and del 8 E-cadherin was observed by immunoprecipitation under serum starvation conditions as well as during EGF treatment (Fig. 4), indicating that deletion of exon 8 in the extracellular domain of E-cadherin still permits interaction of mutant E-cadherin with EGFR. Of note, under EGF stimulation, association of del 8 E-cadherin with EGFR was weaker compared with the wild-type protein. Expression levels of EGFR and E-cadherin remained constant under these conditions (data not shown). Reduced binding of mutant E-cadherin to EGFR in a multicomponent complex or reduced stability of the complex may enhance EGFR surface motility, thereby facilitating EGFR dimerization and activation.

Mutant E-cadherin shows decreased cell surface localization. Cell-cell contact-dependent inhibition of endocytosis contributes to the stabilization of E-cadherin at the cell surface (20). Because the del 8 mutation results in reduced cell-cell contact formation (11), we expected to observe decreased del 8 E-cadherin surface levels. To test this hypothesis, the number of cell surface del 8 E-cadherin molecules compared with wild-type E-cadherin was quantified in MDA-MB-435S cells by flow cytometry using indirect immunofluorescence analysis. The amount of mutant E-cadherin at the plasma membrane was found to be only half of the amount of the wild-type protein (Fig. 5A). In addition, there was no difference between the expression levels of wild-type and mutant E-cadherin (Supplementary Fig. S1). Together, these results suggest that mutation of E-cadherin, namely loss of exon 8, influences the internalization of E-cadherin and destabilizes the protein at the plasma membrane. This result was confirmed with an independent cell system. L929 fibroblasts previously transfected with wild-type or del 8 E-cadherin (11) were included into the experiment and led to the same result (Fig. 5A).

Internalization of EGFR is affected by E-cadherin. Studies using EGFR fused to green fluorescent protein revealed that although EGF receptors are primarily localized at the cell surface, they constantly shuttle and recycle through the cell (21). After ligand binding, the rate of EGFR endocytosis is increased 5- to 10-fold. Colocalization of E-cadherin and EGFR and the observed differences in cell surface distribution of wild-type and mutant E-cadherin (Fig. 5A) raised the question if mutant E-cadherin alters the localization and internalization of EGFR. Fluorescence-activated cell sorting (FACS) analysis revealed enhanced surface localization of EGFR in del 8 compared with wild-type cells. The mean percentage of EGFR positive cells was 52.4% in wild-type and 98.7% in del 8 cells (Fig. 5B). After EGF stimulation, the internalization of EGFR was decreased in del 8 cells (Fig. 5B). These data suggest that mutant E-cadherin influences the endocytosis of EGFR, which could have an effect on its activation.

Expression and localization of E-cadherin and pEGFR in gastric cancer. To determine the phosphorylation status of EGFR in gastric carcinomas in the presence of mutations of E-cadherin within the hotspot region comprising exons 8 and 9, 10 gastric adenocarcinomas of the diffuse type with exon 8 or 9 deletions were examined immunohistochemically (Supplementary Table S1). We detected pEGFR in 8 of 10 cases with del 8 or del 9 E-cadherin mutations. Conversely, pEGFR was detected only in 3 of 10 diffuse type and in 3 of 10 intestinal type gastric adenocarcinomas without del 8 or del 9 E-cadherin mutations (Suplementary Tables S1 and S2). One tumor displaying strong membranous staining of del 9 E-cadherin (Fig. 6A) and pEGFR (at tyrosine residue 1086; Fig. 6B) is shown as an example. In contrast, a gastric adenocarcinoma of the intestinal type showed strong wild-type E-cadherin staining (Fig. 6C) and complete absence of pEGFR staining (Fig. 6D). To determine whether enhanced EGFR activation was dependent on a dominant effect of the exon 8 or 9 mutations or whether it was also present in tumors that have lost E-cadherin due to other mechanisms, tumors were stratified according to the E-cadherin expression status (Supplementary Table S3). We obtained evidence that pEGFR was present in 30% of tumor cells independent of the presence or absence of E-cadherin. These results clearly indicate that enhanced EGFR activaton observed in the presence of del 8 or del 9 E-cadherin is due to a dominant effect of the mutations.

Mutations in the tumor suppressor gene E-cadherin and overexpression of the RTK EGFR are among the most frequent genetic alterations associated with gastric carcinoma (10, 22). We report the first study of a functional relationship between EGFR and E-cadherin harboring a gastric carcinoma–derived somatic mutation. The investigated E-cadherin mutation was an in frame deletion of exon 8, leading to partial loss of the adhesive function and up-regulation of cellular motility (11, 12). EGFR activation was accompanied with enhanced binding of the adaptor proteins Shc and Grb2 and an increased level of active Ras. Complex formation of wild-type and del 8 E-cadherin was observed, indicating that deletion of exon 8 in the extracellular domain of E-cadherin still permits interaction of mutant E-cadherin with EGFR. Of note, association of del 8 E-cadherin with EGFR was weaker compared with the wild-type protein. We conclude that reduced binding of mutant E-cadherin to EGFR in a multicomponent complex or reduced stability of the complex may enhance EGFR surface motility, thereby facilitating EGFR dimerization and activation. Moreover, reduced surface localization due to enhanced endocytosis of mutant E-cadherin compared with the wild-type protein was observed. Moreover, the internalization of EGFR was decreased in response to EGF stimulation in cells expressing mutant E-cadherin, suggesting that mutation of E-cadherin also influences the endocytosis of EGFR. Finally, we show increased activation of EGFR in gastric carcinoma samples with mutant E-cadherin lacking exons 8 or 9.

Mechanism of EGFR activation by somatic E-cadherin mutation. The mechanism for ligand-induced dimerization of the extracellular domain of EGFR has been unraveled (23). Formation of an asymmetrical dimer with one kinase domain in the EGF-mediated dimer activating the other through an allosteric mechanism has shed some light on the mechanism by which the receptor is activated (24). There are several possible explanations for the increase in EGFR activation observed in our study in the presence of the E-cadherin mutation. First, reduced complex formation between del 8 E-cadherin and EGFR might be involved in the effect. E-cadherin has been shown to sequester EGFR by complex formation, which leads to reduced receptor mobility and subsequent inhibition of receptor activation; this complex formation required the extracellular region of E-cadherin (7). The deletion of exon 8 of E-cadherin is located within the hotspot region comprising exons 8 and 9, affecting the region connecting extracellular domains 2 and 3. This region seems to be the most critical domain in the process of tumor formation and progression, resulting both in loss of function (reduced cell-cell adhesion) and gain of function (increased motile behavior; refs. 11, 25). The del 8 mutation affects a putative calcium-binding motif and may result in reduced binding of calcium. Because calcium binding is required for E-cadherin function and protease resistance (26), the E-cadherin mutation might impair the overall structure and stability of the protein. A conformational change of E-cadherin might affect its capacity to bind to EGFR. In accordance with our results, a reduction in complex formation between EGFR and E-cadherin with germline mutations in the extracellular region (T340A and A634V) that abrogate cell adhesion was detected (27). These germline E-cadherin mutations were also associated with increased activation of EGFR. In contrast, intracellular E-cadherin germline mutations (P799R and V832M) had no detectable effect on the strength of binding in that study. Taken together, our results point to a crucial role of a mutation within the extracellular region of E-cadherin for complex formation with EGFR and activation of EGFR signaling. Reduced binding of mutant E-cadherin to EGFR might increase receptor mobility and facilitate receptor dimerization.

Second, the deletion of exon 8 changes the surface localization and endocytosis of E-cadherin, possibly contributing to the effect of the mutation on EGFR signaling. del 8 E-cadherin is located in the cytoplasm, in the perinuclear region, and in lamellipodia, besides being present at the remaining cell-cell contacts that are formed despite the mutation (28). The amount of cell surface del 8 E-cadherin determined by quantitative indirect immunofluoresence analysis using FACS was found to be only half of the amount of the wild-type protein. Reduced surface localization associated with enhanced internalization of mutant E-cadherin might influence the endocytosis of EGFR because both proteins colocalize (6). Consistent with this hypothesis, we observed that after EGF stimulation, the internalization of EGFR was decreased in del 8 cells, suggesting that mutant E-cadherin influences the endocytosis of EGFR, which could have an effect on its activation. Taken together, our data suggest that both mechanisms, reduced complex formation between mutant E-cadherin and EGFR, as well as reduced surface localization of E-cadherin combined with decreased internalization of EGFR contribute to enhanced EGFR activation observed in the presence of the E-cadherin mutation.

Significance of EGFR activation by E-cadherin mutations for tumor progression. The EGFR signaling pathway is one of the most important networks that regulate cell proliferation, differentiation, survival, and motility. Enhanced signaling of EGFR due to its overexpression is well-known in several carcinomas, including gastric carcinoma (22). Here, we describe a novel mechanism of regulating EGFR activation. We show that extracellular mutation of E-cadherin contributes to the frequently observed activation of EGFR in tumors and explains the motile and invasive behavior of tumor cells with E-cadherin mutations.

Grant support: A grant to Drs. B. Luber and H. Höfler from the Deutsche Forschungsgemeinschaft (SFB 456).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank U. Buchholz, A. Voss, and A. Brütting for excellent technical assistance and J. Schlegel for helpful discussions.