Abstract

E-cadherin is critical for the maintenance of tissue architecture and is a major component of adherens junctions. Its role in tumour development is well established, with many human carcinomas exhibiting E-cadherin loss at the invasive front. In many invasive carcinomas, the mechanisms leading to the loss of E-cadherin remains elusive. Here, we hypothesize that mechanisms of protein quality control play a key role in E-cadherin regulation. As a cell model system, we used CHO cells stably expressing E-cadherin germline missense mutations R749W and E757K, which are associated with hereditary diffuse gastric cancer. An abnormal pattern of E-cadherin expression was observed, with protein accumulating mainly in the endoplasmic reticulum (ER). We demonstrated that E-cadherin missense mutants are subjected to Endoplasmic Reticulum Quality Control (ERQC) and that their loss is due to ER-associated degradation. Treatment of these mutant cells with specific chemical chaperones restored E-cadherin to the cell membrane and rescued its function. We show that ERQC plays a major role in E-cadherin regulation and propose that overcoming this regulation may represent an approach to rescue E-cadherin expression and functionality in cancer.

INTRODUCTION

E-cadherin is the major component of adherens junction (AJ) and is crucial for the establishment and maintenance of polarized and differentiated epithelia, in development and in adult tissues (1). A role for E-cadherin in tumour development is well established, with many human carcinomas, such as skin, head and neck, lung, breast, thyroid, gastric, colon and ovarian cancer, exhibiting reduced E-cadherin expression relative to their normal cellular counterparts (2). In sporadic diffuse gastric cancers and lobular breast cancers, E-cadherin inactivation is associated with somatic point mutations of CDH1, the gene encoding E-cadherin, as well as loss of heterozygosity, promoter hypermethylation or overexpression of transcriptional repressors (3–5). In most carcinomas, loss of E-cadherin is usually a late event associated with invasion and metastasis; nevertheless, the study of early hereditary diffuse gastric cancer (HDGC) lesions in the carriers of the germline E-cadherin (CDH1) mutation suggest that E-cadherin loss is an early or initiating event in tumourigenesis (6).

Despite the well-documented association between E-cadherin mutations and diffuse gastric cancer, E-cadherin mutations are rather an exception in epithelial cancers. This low frequency of E-cadherin mutations stands in contrast to the ubiquitous disturbance or loss of E-cadherin in invasive and pre-invasive cancers, reinforcing the idea that mechanisms of transcriptional or post-transcriptional regulation could be responsible for the observed protein loss (7).

In normal epithelial cells, newly synthesized E-cadherin binds to beta-catenin and is transported from the endoplasmic reticulum (ER) to the lateral membrane, where it is assembled into the AJ (8). Although static at first, AJ undergoes rearrangement: cadherin molecules enter the endocytosis pathway, and they are either recycled to the plasma membrane (PM) or ubiquitinated for lysosomal degradation (9). Arf6, a known regulator of endosome traffic, and Hakai, an E3 ubiquitin ligase, have emerged as key modulators of this process, together with different receptor and non-receptor tyrosine kinases capable of promoting E-cadherin internalization (9). It was shown that in cells in which catenin p120 is knocked-down, E-cadherin rapidly undergoes lysosomal degradation following arrival at the cell surface, indicating that p120 catenin might keep cadherins away from the degradative pathway. The juxtamembrane domain on the cytoplasmic tail of E-cadherin plays a major role in this dynamic trafficking, mainly by interacting with the catenins and with the vesicle-trafficking machinery (9).

We recently identified two HDGC-associated CDH1 missense mutations, R749W (10) and E757K (reported here), both affecting the E-cadherin juxtamembrane domain. Therefore, we decided to characterize their effect on protein cellular trafficking and function. We found that CHO cells expressing these mutant E-cadherin proteins exhibit low levels of surface-exposed E-cadherin, whereas the majority of the mutant proteins are retained in the ER. Based on this observation, we explored the hypothesis that E-cadherin levels and activity are critically modulated by ER-associated mechanisms of protein quality control (PQC), and therefore their subversion could be an alternative mechanism for E-cadherin loss in cancer progression.

RESULTS

Expression pattern of E-cadherin mutants

Cell lines stably expressing human E-cadherin were established by lentiviral infection of the E-cadherin-negative mammalian cell line CHO (Chinese hamster ovary). Cells expressing the R749W and E757K mutants and controls [MOCK cells and wild-type (WT) human E-cadherin-expressing cells] were selected by antibiotic resistance to blasticidin. Total expression of E-cadherin was analyzed by western blot (WB) with anti-human E-cadherin antibody (Fig. 1A). Mutants R749W and E757K expressed lower levels of total E-cadherin, corresponding to 60% and 25% of the WT expression, respectively (Fig. 1A). To test whether the decrease in total E-cadherin expression was reflected in the decrease in surface expression, we used flow cytometry (FCM) and surface biotinylation to analyze E-cadherin expression at the cell membrane. Both FCM (Fig. 1B and C) and surface biotinylation (data not shown) showed that the mutants exhibited not only an overall decrease in the E-cadherin levels, but also a statistically significant reduction in surface expression. To rule out the possibility that differences in protein levels were a consequence of differences in transduction efficiency or in selection, human E-cadherin mRNA levels were characterized in the different stable cell lines by both semi-quantitative reverse transcriptase-polymerase chain reaction (RT-PCR) (Fig. 1D) and quantitative real-time PCR (Fig. 1E). Both WT and mutant cell lines show comparable levels of human E-cadherin mRNA. These results indicate that a post-transcriptional mechanism of regulation could be responsible for the differences in protein expression.

Figure 1.

E-cadherin expression pattern is abnormal in CHO cells stably transduced with hereditary diffuse gastric cancer-associated mutations due to a post-transcriptional silencing mechanism. (A–D) CHO cells were stably transduced with the empty vector (Mock) or with wild-type, R749W or E757K hE-cadherin. (A) E-cadherin in whole cell lysate was detected by western blot using anti-E-cadherin antibody (Ab); anti-Actin Ab was used as a loading control. Results shown are representative of three independent experiments. (B) E-cadherin cell surface expression was assessed by flow cytometry after staining with an extracellular anti-E-cadherin Ab. Each histogram represents 50,000 cell counts. The black area in the histogram represents the cell surface expression of E-cadherin in mock CHO cells. Surface expression of WT E-cadherin (yellow), R749W (red), and E757K (blue) is shown. Results are representative of four independent experiments. For each sample, the mean expression was calculated, and in the graph (C), the bars represent the average ± SD of four independent experiments. (D) E-cadherin mRNA levels were analyzed by semi-quantitative RT-PCR, with GAPDH as a control. (E) In three independent experiments, RNA was extracted and quantified using real-time PCR. E-cadherin mRNA levels were normalized to 18S mRNA. Experiments were performed in triplicates and the graph shows the means ± SD.

Figure 1.

E-cadherin expression pattern is abnormal in CHO cells stably transduced with hereditary diffuse gastric cancer-associated mutations due to a post-transcriptional silencing mechanism. (A–D) CHO cells were stably transduced with the empty vector (Mock) or with wild-type, R749W or E757K hE-cadherin. (A) E-cadherin in whole cell lysate was detected by western blot using anti-E-cadherin antibody (Ab); anti-Actin Ab was used as a loading control. Results shown are representative of three independent experiments. (B) E-cadherin cell surface expression was assessed by flow cytometry after staining with an extracellular anti-E-cadherin Ab. Each histogram represents 50,000 cell counts. The black area in the histogram represents the cell surface expression of E-cadherin in mock CHO cells. Surface expression of WT E-cadherin (yellow), R749W (red), and E757K (blue) is shown. Results are representative of four independent experiments. For each sample, the mean expression was calculated, and in the graph (C), the bars represent the average ± SD of four independent experiments. (D) E-cadherin mRNA levels were analyzed by semi-quantitative RT-PCR, with GAPDH as a control. (E) In three independent experiments, RNA was extracted and quantified using real-time PCR. E-cadherin mRNA levels were normalized to 18S mRNA. Experiments were performed in triplicates and the graph shows the means ± SD.

E-cadherin mutants are retained in the endoplasmic reticulum

We characterized the subcellular distribution pattern of E-cadherin protein by immunofluorescence, both in WT and mutant cells, using an antibody against E-cadherin, together with markers for the Golgi (Fig. 2A) or ER (Fig. 2B). Staining with the E-cadherin antibody shows that WT E-cadherin localizes at the PM, whereas mutants R749W and E757K are often found in perinuclear regions. Co-immunofluorescence with the Golgi marker GM130 (Fig. 2A) indicates that E-cadherin mutants do not accumulate in the Golgi. On the contrary, double-staining with the ER marker calnexin indicates that perinuclear accumulation of the E-cadherin mutants results mainly from ER retention.

Figure 2.

E-cadherin mutants R749W and E757K accumulate immature E-cadherin in the endoplasmic reticulum. (A–C) CHO cells were stably transduced with the empty vector (Mock) or with wild-type (WT), R749W or E757K hE-cadherin. (A, B) Cells were fixed, permeabilized and co-immunostained with anti-human E-cadherin and anti-GM130 (A), or with anti-calnexin antibodies (Ab). (B) E-cadherin was visualized with FITC- (A) or Texas Red (B)-conjugated Ab. GM130 was visualized with Texas Red-conjugated Ab and calnexin with FITC-conjugated Ab. Images were acquired by confocal (A) or conventional fluorescence (B) microscopy. The arrows indicate areas with no overlap. Scale bar represents 10 µm. (C) After biotinylation of intact cells, whole cell lysates were incubated with streptavidin beads, and the supernatants (cytoplasmic fractions) were collected. Total cell lysates (T) and cytoplasmic fractions (CT) were western-blotted and probed with anti E-cadherin Ab. Whole cell lysate of CHO cells stably transduced with mutant E757K and treated with 0.5 µg/ml Brefeldin A for 8 h was used as a positive (Pos) control to distinguish the immature form. Each band in the western blots (WB) was quantified by densitometry. The top band is the immature form (Im approximately 135 kDa) and the other is the mature form (Mat approximately 120 kDa). Percentage of immature form was calculated based on the ratio of immature:total, and the WT was normalized to 1. The numbers in the WB depict the relative increase of immature form in that experiment. The blots are representative of three independent experiments.

Figure 2.

E-cadherin mutants R749W and E757K accumulate immature E-cadherin in the endoplasmic reticulum. (A–C) CHO cells were stably transduced with the empty vector (Mock) or with wild-type (WT), R749W or E757K hE-cadherin. (A, B) Cells were fixed, permeabilized and co-immunostained with anti-human E-cadherin and anti-GM130 (A), or with anti-calnexin antibodies (Ab). (B) E-cadherin was visualized with FITC- (A) or Texas Red (B)-conjugated Ab. GM130 was visualized with Texas Red-conjugated Ab and calnexin with FITC-conjugated Ab. Images were acquired by confocal (A) or conventional fluorescence (B) microscopy. The arrows indicate areas with no overlap. Scale bar represents 10 µm. (C) After biotinylation of intact cells, whole cell lysates were incubated with streptavidin beads, and the supernatants (cytoplasmic fractions) were collected. Total cell lysates (T) and cytoplasmic fractions (CT) were western-blotted and probed with anti E-cadherin Ab. Whole cell lysate of CHO cells stably transduced with mutant E757K and treated with 0.5 µg/ml Brefeldin A for 8 h was used as a positive (Pos) control to distinguish the immature form. Each band in the western blots (WB) was quantified by densitometry. The top band is the immature form (Im approximately 135 kDa) and the other is the mature form (Mat approximately 120 kDa). Percentage of immature form was calculated based on the ratio of immature:total, and the WT was normalized to 1. The numbers in the WB depict the relative increase of immature form in that experiment. The blots are representative of three independent experiments.

We then quantified the fraction of immature mutant and WT E-cadherin molecules. To this end, surface proteins were separated by biotin–streptavidin precipitation, followed by densitometric quantification. Cells expressing mutant R749W or E757K contained a larger fraction of immature forms than the WT-expressing cells, both in total and in the cytoplasmic fraction (Fig. 2C). This finding further strengthens the hypothesis of ER retention.

Endoplasmic reticulum-associated degradation regulates E-cadherin mutants

Our results show that the HDGC-associated mutant E-cadherins, R749W and E757K, are retained in the ER and exhibit lower level of total and surface expression. We hypothesized that this could be the result of ER-associated degradation (ERAD), a mechanism by which misfolded proteins are translocated to the cytosol and degraded by the ubiquitin–proteasome machinery (11,12). We therefore analyzed the ubiquitination status of E-cadherin variants by performing double-fluorescence staining of E-cadherin and ubiquitin. As shown in Figure 3A, mutant E-cadherin staining and ubiquitin staining overlapped mainly in perinuclear regions, indicating that the protein is highly ubiquitinated. Interestingly, WT E-cadherin at the cell membrane level hardly shows any overlap with ubiquitin.

Figure 3.

E-cadherin hereditary diffuse gastric cancer-associated mutants are more ubiquitinated and degraded by the proteasome in endoplasmic reticulum-associated degradation. (A–C) CHO cells were stably transduced with the empty vector (Mock) or with wild-type (WT), R749W or E757K hE-cadherin. (A, C) Cells were fixed, permeabilized and co-immunostained with anti-human E-cadherin antibody (Ab) and anti-ubiquitin Ab (A), or with E-cadherin alone Ab (C). E-cadherin was visualized with FITC-conjugated Ab, and ubiquitin was visualized with Texas Red-conjugated Ab (A). The arrows indicate areas of overlap. (B, C) Cells were treated with 10 µm MG132 (or the corresponding amount of solvent dimethyl sulphoxide, 0.25%) for 16 h. Whole cell lysates were analyzed by western blot (C), using the anti-E-cadherin Ab and anti-actin Ab as the loading control. The blots are representative of three independent experiments. Scale bars represent 5 µm in (A) and 20 µm in (D).

Figure 3.

E-cadherin hereditary diffuse gastric cancer-associated mutants are more ubiquitinated and degraded by the proteasome in endoplasmic reticulum-associated degradation. (A–C) CHO cells were stably transduced with the empty vector (Mock) or with wild-type (WT), R749W or E757K hE-cadherin. (A, C) Cells were fixed, permeabilized and co-immunostained with anti-human E-cadherin antibody (Ab) and anti-ubiquitin Ab (A), or with E-cadherin alone Ab (C). E-cadherin was visualized with FITC-conjugated Ab, and ubiquitin was visualized with Texas Red-conjugated Ab (A). The arrows indicate areas of overlap. (B, C) Cells were treated with 10 µm MG132 (or the corresponding amount of solvent dimethyl sulphoxide, 0.25%) for 16 h. Whole cell lysates were analyzed by western blot (C), using the anti-E-cadherin Ab and anti-actin Ab as the loading control. The blots are representative of three independent experiments. Scale bars represent 5 µm in (A) and 20 µm in (D).

To further investigate whether ERAD was responsible for the decreased expression of mutants, R749W and E757K, CHO stable cell lines were incubated with 10 µm of the proteasome inhibitor MG132, or with its solvent [dimethyl sulphoxide (DMSO)]. Different incubation times were tested (data not shown) and 16 h was chosen because of a stronger effect without evidence for cell toxicity. After incubation, cells were either lysed or prepared for immunofluorescence. WB analysis of total lysates (Fig. 3B) shows convincingly that, upon treatment with the proteasome inhibitor MG132, expression of E-cadherin protein mutants R749W and E757K is restored and even exceeds WT levels. Accumulation of E-cadherin was also observed in MG132-treated WT cells, indicating that the proteasome machinery is also involved in regulating WT E-cadherin. These results were also confirmed by immunofluorescence analysis (Fig. 3C). Interestingly, treatment with 0.25% of DMSO (solvent used for MG132) seems to mediate some mutant recovery for the PM (compare –MG132 panels of Fig. 3C with Fig. 2A). The ubiquitin–proteasome-dependent degradation, together with the observation that mutant E-cadherins, R749W and E757K, are retained in the ER, strongly suggests that these two mutant proteins are degraded by ERAD.

Chemical chaperones can rescue the expression and functions of mutant E-cadherin

Chemical chaperones (CC) are small molecules that have been described to bind nascent proteins, promoting the folding of mutant proteins to a folded state resembling the WT protein, thus allowing them to escape ERAD. CCs like glycerol, DMSO, trimethylamine-N-oxide (TMAO) and 4-phenylbutyrate (4-PBA) have been to shown to promote the folding of several mutant proteins associated with diseases (13). To test the effect of CCs on the folding of E-cadherin mutants, cells were incubated for 24 h with different concentrations of the abovementioned chemicals. Total expression was assessed by WB analysis (Fig. 4A) and surface expression by FCM (Fig. 4B). We used Brefeldin A (Bref A) treatment for 8 h as a positive control for accumulation of immature protein (higher molecular weight band), and MG132 treatment for 24 h as a control for total protein rescue. For total expression analysis, the same amount of protein (25 µg) was loaded for all samples and treatments. Figure 4A shows a representative WB for E-cadherin mutant cells and WT cells following various treatments. As mentioned above, WT expression is partially affected by treatment with MG132 and 2% DMSO, likely due to proteasomal regulation and increased folding efficacy, respectively. Both mutants exhibit a clear increase in total E-cadherin expression upon treatment with 2% DMSO and 4-PBA, whereas no obvious effect is observed for the other treatments tested. To determine weather the CC-induced recovery of E-cadherin expression was accompanied by protein translocation to the membrane, we performed surface analysis by FCM (Fig. 4B). The mean expression of each cell line was calculated following each treatment. The fractional change was calculated from the ratio between the mean expression of the treated and non-treated cells. Three independent experiments were performed, and the mean ratio of increase (±SD) is presented in Figure 4B. We observe that MG132 induces a three-fold increase in E757K surface expression compared with non-treated cells. R749W is also rescued at the surface (two-fold increase) upon proteasome inhibition. Interestingly, inhibiting lysosomes with NH4Cl did not alter surface E-cadherin expression. DMSO was the most effective CC, promoting a 2.5-fold increase in E757K and a 1.7-fold increase in R749W. FCM histograms show the surface recovery of both mutants upon treatments with MG132 and 2% DMSO, in comparison with WT (Fig. 4B, bottom).

Figure 4.

The chemical chaperone dimethyl sulphoxide (DMSO) rescues surface expression of mutant E-cadherin in both mutants, as well as functionality in R749W. (A–E) CHO cells were stably transduced with the empty vector (Mock) or with vectors expressing wild-type (WT), R749W or E757K E-cadherin proteins. (A) Cells were incubated for 24 h with 5% glycerol, 100 mm trimethylamine-N-oxide (TMAO), 5 mm 4-phenylbutyrate (4-PBA) or 2% DMSO, or for 16 h with 0.25% DMSO ± 10 µm MG132, or for 8 h with 0.5 µg/ml Brefeldin A (Bref A). Total lysates were prepared and equal amounts (25 µg) of each protein were loaded. Western blot analysis was performed using the anti-E-cadherin antibody (Ab). (B) Cells were incubated for 24 h with 5% glycerol, 100 mm TMAO, 5 mm 4-PBA or 2% DMSO or for 16 h with 25 mm NH4Cl or 10 µm MG132. E-cadherin cell surface expression was assessed for each treatment by flow cytometry (FCM). The ratio between the mean expression of treated and non-treated cells was calculated. The mean of three independent experiments is presented in the graph. FCM histograms show the surface recovery of both mutant E-cadherins upon treatment with 10 µm MG132 and 2% DMSO. The black area in the histogram represents the surface expression of cells transduced with WT E-cadherin. (C) Cells were incubated with 2% DMSO or with normal medium for 24 h and then the Matrigel invasion assay was performed. After 24 h, the filters were fixed and the invasive cells were counted. The percentage of invasion was calculated, setting invasion by Mock cells as 100%. (D) Cells were incubated with 2% DMSO or with normal medium for 24 h and then slow aggregation in soft agar was analyzed. After 48 h of incubation with the corresponding medium, the cells were photographed and one representative image is shown. Triplicates were performed for each experiment and the experiment was repeated three times. In the graph, bars represent the average ± SD of three independent experiments.

Figure 4.

The chemical chaperone dimethyl sulphoxide (DMSO) rescues surface expression of mutant E-cadherin in both mutants, as well as functionality in R749W. (A–E) CHO cells were stably transduced with the empty vector (Mock) or with vectors expressing wild-type (WT), R749W or E757K E-cadherin proteins. (A) Cells were incubated for 24 h with 5% glycerol, 100 mm trimethylamine-N-oxide (TMAO), 5 mm 4-phenylbutyrate (4-PBA) or 2% DMSO, or for 16 h with 0.25% DMSO ± 10 µm MG132, or for 8 h with 0.5 µg/ml Brefeldin A (Bref A). Total lysates were prepared and equal amounts (25 µg) of each protein were loaded. Western blot analysis was performed using the anti-E-cadherin antibody (Ab). (B) Cells were incubated for 24 h with 5% glycerol, 100 mm TMAO, 5 mm 4-PBA or 2% DMSO or for 16 h with 25 mm NH4Cl or 10 µm MG132. E-cadherin cell surface expression was assessed for each treatment by flow cytometry (FCM). The ratio between the mean expression of treated and non-treated cells was calculated. The mean of three independent experiments is presented in the graph. FCM histograms show the surface recovery of both mutant E-cadherins upon treatment with 10 µm MG132 and 2% DMSO. The black area in the histogram represents the surface expression of cells transduced with WT E-cadherin. (C) Cells were incubated with 2% DMSO or with normal medium for 24 h and then the Matrigel invasion assay was performed. After 24 h, the filters were fixed and the invasive cells were counted. The percentage of invasion was calculated, setting invasion by Mock cells as 100%. (D) Cells were incubated with 2% DMSO or with normal medium for 24 h and then slow aggregation in soft agar was analyzed. After 48 h of incubation with the corresponding medium, the cells were photographed and one representative image is shown. Triplicates were performed for each experiment and the experiment was repeated three times. In the graph, bars represent the average ± SD of three independent experiments.

We further addressed the adhesive and invasion suppressor function of the mutant protein recovered at the PM upon treatment with CCs. To this end, Matrigel invasion assays (Fig. 3C) and slow aggregation assays in soft agar (Fig. 4D) were performed for both treated and untreated mutant and control cell lines. The results show that the untreated mutants exhibit an increased capacity for invading a Matrigel layer than WT cells (two-fold increase in R749W and four-fold increase in E757K), which supports the concept of their pathogenic role in gastric cancer. When cells are treated with 2% DMSO, the invasiveness of both mutants in Matrigel is decreased to the level of WT cells, thus indicating rescue of the invasion suppressor function (Fig. 4C). The aggregation assay shows that mutants R749W and E757K lack adhesive capacity, as they do not form compact aggregates (Fig. 4D). When E-cadherin surface expression is induced by DMSO treatment, mutant R749W completely recovered adhesive function, whereas the effect was weaker on mutant E757K.

The adhesive function of mutant E757K is not completely recovered due to insufficient plasma membrane localization and inadequate complex formation

We characterized the adhesion complex both in the absence and the presence of 2% DMSO. Immunoprecipitation (IP) analysis showed that in the absence of DMSO, the interaction of both mutant E-cadherins with each of the catenins is weaker than that of WT E-cadherin (Fig. 5B). Immunofluorescence (IF) analysis confirmed that E-cadherin and catenin signals overlap only at the membrane (Fig. 5C). Interestingly, a pool of β-catenin co-localizes with perinuclear E-cadherin. IP analysis following 2% DMSO treatment shows that the adhesion complex in R749W is correctly assembled (Fig. 5C). On the contrary, the interaction between E-cadherin and each of the catenins is weaker for cells harbouring the E757K mutation (Fig. 5B) than in cells with WT or R749W E-cadherin. IF analysis indicates that the DMSO treatment is not sufficient to relocate all E757K to the membrane. As expected, the pool of perinuclear-retained mutant E-cadherin does not co-localize at all with p120 and α-catenins (Fig. 5C).

Figure 5.

Upon dimethyl sulphoxide (DMSO)-induced E-cadherin recovery to the plasma membrane, a pool of E757K does not assemble a proper cadherin–catenin complex due to residual perinuclear localization. (A, B) CHO cells were stably transduced with the empty vector (Mock) or with wild-type (WT), R749W or E757K hE-cadherin. In (B), cells were incubated for 24 h with complete medium (no DMSO) or with 2% DMSO to induce E-cadherin recovery. Whole cell lysate (A) or the E-cadherin immunoprecipitated fraction (B) were detected by western blot using the anti-E-cadherin antibody (Ab) and anti-p120, anti-β- and anti-α-catenins Abs. Anti-actin-Ab was used as a loading control for whole cell lysate analysis. (C) Cells were seeded and, at 80% confluence, they were incubated with 2% DMSO or with complete medium. After 24 h, cells were fixed, permeabilized and immunostained with anti-human E-cadherin and anti-p120, anti-β- or anti-α-catenin antibodies, and visualized with Texas Red-conjugated and FITC-conjugated secondary antibodies, respectively. Red fluorescence represents E-cadherin and green fluorescence represents each catenin, according to the top label. Overlap of E-cadherin with each catenin is indicated by yellow fluorescence. Scale bar represents 10 µm.

Figure 5.

Upon dimethyl sulphoxide (DMSO)-induced E-cadherin recovery to the plasma membrane, a pool of E757K does not assemble a proper cadherin–catenin complex due to residual perinuclear localization. (A, B) CHO cells were stably transduced with the empty vector (Mock) or with wild-type (WT), R749W or E757K hE-cadherin. In (B), cells were incubated for 24 h with complete medium (no DMSO) or with 2% DMSO to induce E-cadherin recovery. Whole cell lysate (A) or the E-cadherin immunoprecipitated fraction (B) were detected by western blot using the anti-E-cadherin antibody (Ab) and anti-p120, anti-β- and anti-α-catenins Abs. Anti-actin-Ab was used as a loading control for whole cell lysate analysis. (C) Cells were seeded and, at 80% confluence, they were incubated with 2% DMSO or with complete medium. After 24 h, cells were fixed, permeabilized and immunostained with anti-human E-cadherin and anti-p120, anti-β- or anti-α-catenin antibodies, and visualized with Texas Red-conjugated and FITC-conjugated secondary antibodies, respectively. Red fluorescence represents E-cadherin and green fluorescence represents each catenin, according to the top label. Overlap of E-cadherin with each catenin is indicated by yellow fluorescence. Scale bar represents 10 µm.

DISCUSSION

Despite the advance in our understanding of the mechanisms that regulate E-cadherin expression, the cause of the loss of E-cadherin observed in invasive carcinomas has not been identified in many cases. In sporadic diffuse gastric cancers and lobular breast cancers, E-cadherin inactivation has been associated with CDH1 somatic point mutations, as well as with loss of heterozygosity, promoter hypermethylation and overexpression of transcriptional repressors (3–5). Nevertheless, mutations are an exception rather than a rule.

Before entering the secretory pathway, transmembrane glycoproteins are co-translationally translocated into the ER, where molecular chaperones fine-tune the folding process until a properly folded native state is achieved. The protein is sent to its final destination only when a correct native conformation is achieved. When molecular chaperones of the ER quality control (ERQC) identify a protein as misfolded, the protein is retained in the ER (14). A complex network of chaperones then binds to the misfolded protein in an attempt to correct the folding problem. If the protein cannot be folded properly, it is dislocated to the cytoplasm to be eliminated by the proteasome in a process termed ERAD (15). This quality control system of the ER is a cellular process that is meant to protect the cell from the accumulation of toxic unfolded proteins. But sometimes this control is overzealous and prevents mutants, which could still be biologically active, from leaving the ER (or to escape ERAD), thus leading to the disease (13,14,16).

To the best of our knowledge, very little is known about E-cadherin regulation by mechanisms of PQC and the role of these mechanisms in cancer. In this work, we address this phenomenon using HDGC-associated CDH1 germline missense mutations. Such mutations are the recognized genetic cause of the HDGC syndrome. Most of the mutations identified to date are of the nonsense type, leading to premature termination codons and transcripts that are commonly downregulated by the nonsense-mediated decay, a mechanism of mRNA surveillance (17). Nevertheless, germline missense mutations have also been identified in approximately 20% of affected HDGC families (18,19). These mutations are a clinical burden in genetic counselling because their pathogenicity is not straightforward; on the other hand, they also represent a unique scientific tool to dissect specific protein functions and domains. How CDH1 missense mutations are regulated in the cell and lead to loss of function is yet to be understood in most cases (19).

The idea of PQC-driven E-cadherin downregulation in cancer was prompted by our initial results obtained by characterizing the pattern of E-cadherin protein expression in cells transduced with the HDGC-associated missense mutants, R749W and E757K. We found that these cells, despite levels of E-cadherin RNA comparable with WT-expressing cells, display low levels of total and surface E-cadherin protein. Double immunostaining with the ER marker calnexin showed that, for both mutants, E-cadherin frequently accumulates in the ER. Furthermore, cells expressing mutant E-cadherin accumulate more of the immature form of E-cadherin than WT-expressing cells. As E-cadherin processing in the ER involves transformation of the immature (135 kDa) to the mature (120 kDa) form (20), an increase of the immature form is another indication of ER retention. We therefore hypothesized that the retention of mutant E-cadherin in the ER is the result of ERQC (14) and that reduced mutant E-cadherin expression is a consequence of ERAD (11,12). To explore this hypothesis, we investigated E-cadherin ubiquitination in mutant cells in comparison with WT-expressing cells. Double immunostaining of E-cadherin and ubiquitin showed that while ubiquitin does not co-localize with E-cadherin in WT cells, E-cadherin mutants are strongly ubiquitinated in the perinuclear region. This supports the idea that E-cadherin mutants are prone to ERAD, whereas WT E-cadherin undergoes endocytosis-dependent proteasome degradation. In accordance, it has been shown that, upon their clustering, AJ are dynamically rearranged: cadherin molecules enter the endocytosis pathway, and then they are either recycled to the PM or ubiquitinated and sent for degradation (9).

To further confirm a role for ERAD in the regulation of E-cadherin mutants, we treated cells with the proteasome inhibitor MG132. Both total and membrane E-cadherin expression levels were increased, demonstrating that ERAD is responsible for the partial silencing of mutants R749W and E757K. These results also indicate that E-cadherin mutants suffer from a folding problem that leads to their retention in the ER and their degradation by proteasomes, thus inhibiting their proper localization and function. Interestingly, solvent treatment (0.25% DMSO) by itself induced partial restoration of both mutants to the PM, indicating that this organic solvent could be used to manipulate the regulation of E-cadherin. CCs like DMSO (21) are small molecules that bind to nascent proteins, and are thought to promote folding of mutant proteins into a form that resembles the WT protein, which allows them to escape ER retention and degradation (21). This observation prompted the idea that CCs could be used to rescue E-cadherin expression and function in mutant cells. We thus tested the effect of different CCs on mutant E-cadherin expression and localization. We found that both 4-PBA and DMSO are effective in rescuing high expression of E-cadherin, but it was only upon DMSO treatment that expression at the PM was achieved. It is not that surprising that among several CCs tested only DMSO has an effect on the expression and localization of E-cadherin. Protein specificity of different CCs, which can also be mutation-specific, has already been described in other models (21,22).

We previously showed that cells expressing E-cadherin missense mutations, despite a pattern of membrane expression that resembles the WT protein pattern, exhibit loss of function resulting in impaired cell–cell adhesion and acquisition of an invasive phenotype (23). In view of this, we investigated whether induction of PM expression of R749W and E757K upon DMSO treatment also rescues protein function. We performed Matrigel invasion and slow aggregation assays upon 2% DMSO treatment: the adhesive and anti-invasive functions were completely restored for the mutant, R749W. The results were less striking for the mutant, E757K: despite a substantial reduction in cell invasion, cell–cell adhesion was still impaired. As shown by IP analysis and immunostaining, the interaction between the E757K mutant and all catenins tested are destabilized and, despite the DMSO treatment, part of the protein is still retained in the perinuclear region, which accounts for the incomplete recovery of functionality.

In conclusion, we have shown in this work that E-cadherin is regulated by ERAD, which targets mutated, unfolded proteins for proteasomal degradation. Treatment of cells harbouring cytoplasmic E-cadherin mutations with specific CCs could force correct folding of the protein and rescue protein expression, indicating that these mutations affect E-cadherin folding. Upon treatment with the most effective CC (2% DMSO), mutant R749W completely recovered PM expression and the adhesion and invasion suppressor functions, while mutant E757K was still partially retained in the perinuclear region and not fully restored at the functional level. Taken together, our results indicate that the recognition of misfolded mutant E-cadherin proteins by the ERAD machinery may be somehow non-discriminatory, in some cases leading to the degradation of otherwise neutral mutants. Furthermore, our results also suggest that CCs could have an effective therapeutic value for tumours harbouring specific E-cadherin mutations, by stabilizing their folding and inhibiting ERAD-mediated degradation. In this regard, an increasing number of inherited diseases are found to result from mutations that lead to misfolded proteins. The therapeutic value of pharmacological chaperones (small molecules designed to specifically target misfolded protein), in endocrine and metabolic disorders such as hyperinsulinemic hypoglycemia, hypogonadotropic hypogonadism and nephrogenic diabetes insipidus is already well established. This further supports the idea that such type of therapeutic interventions could also be successfully applied to cancer treatment, at least in cases where protein misfolding and ERAD degradation play a major role.

MATERIALS AND METHODS

Identification of mutant E757K

A heterozygous germline missense mutation leading to the amino acid substitution E757K (2269 G>A) was found, by direct sequencing, in a Portuguese 38-year-old male with signet ring carcinoma of the stomach. His 33-year-old sister had died earlier with the same type of neoplasia. The finding of early-onset diffuse gastric cancer in two <50-year-old first-degree relatives classifies this kindred as a HDGC family. The same mutation was then identified in three other first-degree relatives, two of whom had histologically confirmed diffuse gastric cancer. This mutation was not found in over 100 normal individuals of Portuguese origin. For initial sequencing, all 16 coding regions, intron–exon boundaries and the promoter region of the E-cadherin gene were amplified by PCR using the DNA extracted from the peripheral blood of the HDGC patient. DNA extraction and PRC primers and conditions were as previously reported (10).

Plasmids construction

WT hE-cadherin was amplified from pECAD1 (23) and subcloned in Plenti vector (Invitrogen, Barcelona, Spain) according to the manufacturer's instructions. A fragment of E-cadherin mutants (from 1285 bp to stop codon) was amplified with mutation-specific primers: R749W (F: 5′-GAT GAC ACC TGG GAC AAC G; R: 5′-CGT TGT CCC AGG TGT CAT C) and E757K (F: 5′-ATT ACT ATG ATA AAG AAG GAG G; R: 5′-GCC TCC TTC TTT ATC ATA GTA), and cloned in TOPO (Invitrogen). Restriction with BspEI and XhoI was used to subclone the mutant fragment in WT-plenti. All the clonings were verified by direct sequencing.

Cell culture and transduction

CHO cells were transduced with ViraPower Lentiviral Expression kit (Invitrogen) using the following vectors: empty vector (Mock), WT hE-cad, R749W and E757K. Lentiviral transduction was performed following the manufacturer's instructions, and the transduced cells were selected by antibiotic resistance to blasticidin (5 µg/ml). Cells were grown in alpha MEM medium (Gibco, Invitrogen, Barcelona, Spain) supplemented with 10% foetal bovine serum (FBS; Gibco, Invitrogen), 1% penicillin-streptomycin (Gibco, Invitrogen) and 5 µg/ml blasticidin (Gibco, Invitrogen), in a humidified incubator with 5% CO2 at 37°C.

Reagents and cell treatments

Cells were plated in six-well plates and incubated with 10 µm of the proteasome inhibitor MG132 (CalBioChem, Darmstedt, Germany) or 25 mm of the lysosome inhibitor NH4Cl (Sigma, Sintra, Portugal) for 16 h, and with several CCs for 24 h. CCs used were 4-PBA (Sigma; 5 mm), TMAO (Sigma; 100 mm), 2% DMSO (Sigma) and 5% glycerol (Sigma). Incubation with 0.5 µg/ml Bref A (Sigma) for 8 h was used as a positive control of ER retention of the immature form. Other reagents were bought from Sigma unless mentioned otherwise.

Flow cytometry

Cells were grown to a confluent monolayer, detached with Versene (Gibco, Invitrogen) and resuspended in ice-cold phosphate-buffered saline (PBS) with 0.05 mg/ml CaCl2. An aliquot of 5 × 105 cells was centrifuged for 5 min at 1500 rpm and 4°C and washed in PBS with 0.05 mg/ml CaCl2 and 3% bovine serum albumin (BSA). Cells were incubated for 30 min with the extracellular primary antibody against E-cadherin, HECD1 (Zymed Laboratories, Barcelona, Spain) at 1:50 dilution. Cells were washed twice and incubated with rabbit polyclonal anti-mouse FITC (1:100; DAKO) in the dark for 20 min. Finally, the cells were washed and resuspended in 1 ml of washing solution. At least 5 × 104 cells were analyzed in a Coulter Epics XL-MCL flow cytometer (Beckman-Coulter, Fullerton, CA, USA). The data were analyzed with WinMDI (Joe Trotter, TSRI, San Diego, CA, USA).

Antibodies, immunofluorescence and microscopy

Cells were seeded on glass coverslips and grown to at least 80% confluence. They were fixed in ice-cold methanol for 20 min, washed and incubated with primary antibody diluted in PBS containing 5% BSA for 1 h at room temperature. The primary antibodies were mouse monoclonal anti E-cadherin (1:300; BD Biosciences, Missouga, ON, Canada), rat anti E-cadherin ECCD2 (1:200; Takara), mouse monoclonal anti p120 (1:250; BD Biosciences), rabbit anti-α-catenin (1:500; Sigma), rabbit anti-β-catenin (1:1000; Sigma), rabbit anti Calnexin (1:250; Stressgen, Michigan, USA), mouse anti GM130 (1:250; BD Biosciences) and mouse anti-ubiquitin (1:100; Zymed Laboratories). Secondary antibodies were anti-mouse FITC and Texas Red (DAKO, Glostrup, Denmark), anti-rabbit FITC (DAKO) and anti-rat Texas Red (Jackson ImmunoResearch, Suffolk, UK). The coverslips were mounted on slides using Vectashield with DAPI (Vector Laboratories, Burlingame, CA, USA). Images were acquired using a Leica DM2000 microscope with objectives of 40X or 100X, and processed with Leica Application Suite Version 2.7.1 R1 software. Golgi images were obtained with a Leica SP5 AOBS Confocal Microscope, with a 63X 1.4 oil objective and the pinhole set on 1 airy unit.

Reverse transcriptase-PCR and real-time quantitative PCR

Cells were grown to a confluent monolayer and RNA was extracted with Tripure (Roche, Amadora, Portugal). cDNA was produced from 1 µg of RNA with Superscript Reverse Transcriptase (Invitrogen) using random primers (Invitrogen). Semi-quantitative RT-PCR was performed with E-cadherin primers—F: 5′-TTC CCT GCG TAT ACC CTG GT and R: 5′-GCG AAG ATA CCG GGG GAC ACT CAT GAG, and GAPDH primers—F: TCA AGG CTG AGA ACG GAA G and R: AGA GGG GGC AGA GAT GAT GA. Real-time PCR was performed with primers for E-cadherin (F: GTC ACT GAC ACC AAC GAT AAT CCT; R: TTT CAG TGT GGT GAT TAC GAC GTT A) and 18S (F: CGC CGC TAG AGG TGA AAT TC; R: CAT TCT TGG CAA ATG CTT TCG) in an AbiPrism 7000 Sequence Detection System.

Immunoprecipitation, SDS–PAGE and western blotting

Cells were lysed in cold catenin lysis buffer (1% Triton X-100, 1% Nonidet P-40 in PBS) enriched with a protease inhibitor cocktail (Roche) and a phosphatase inhibitor cocktail (Sigma). The proteins were quantified using a modified Bradford assay (Bio-Rad, Amadora, Portugal). To immunoprecipitate E-cadherin, 500–750 µg of protein was incubated overnight with a mouse monoclonal anti-E-cadherin antibody (BD Biosciences, Transduction Laboratories). Immunocomplexes were incubated for 50 min with protein G-Sepharose beads (Amersham Biosciences, Piscotoway, NJ, USA), washed, eluted in sample buffer and loaded in one gel. Immunoprecipitated proteins were separated by 7.5% sodium dodecylsulphate–polyacrylamide gel electrophoresis (SDS–PAGE) and electroblotted onto a Hybond ECL membrane (Amersham Biosciences). For analysis of total protein samples, 25 µg of proteins were loaded in 6.5% or 7.5% SDS–PAGE, depending on the mass of the molecules. Membranes were blocked with 5% non-fat milk and 0.5% Tween-20 in PBS and immunoblotted with antibodies against E-cadherin (1:1000; BD Biosciences), p120 (1:1000; BD Biosciences), α-catenin (1:250; BD Biosciences), β-catenin (1:1000; Sigma) or actin (1:1000; Santa Cruz Biotechnology, Heidelberg, Germany). Donkey anti-rabbit (Amersham Biosciences), sheep anti-mouse (Amersham Biosciences) or donkey anti-goat (Santa Cruz Biotechnology), HRP-conjugated secondary antibodies were used, followed by ECL detection (Amersham Biosciences). Immunoblots were quantified with the Quantity One Software (Bio-Rad).

Biotinylation of cell surface proteins

A confluent cell monolayer was incubated with 0.5 mg/ml of Sulpho-NHS-biotin (Pierce Biotechnology, Rockford, IL, USA) in PBS for 30 min at 4°C. Whole cell lysates were prepared according to the protocol described above. An aliquot of 400 µg protein was incubated overnight with 50 µl of Streptavidin sepharose beads (Amersham Biosciences). To separate the surface fraction from the cytoplasmic fraction, the samples were spun down, and the supernatant (cytoplasmic fraction) was recovered and quantified. The pellet (surface fraction) was washed and eluted in sample buffer. The samples were loaded on 6.5% SDS–PAGE.

Slow aggregation assay

The functional significance of the E-cadherin missense changes was assessed with respect to aggregation and invasion suppression capacity as described in Suriano et al. (23). For the aggregation assay, the wells of a 96-well-plate were coated with 50 µl of an agar solution (100 mg Bacto-agar in 15 ml of sterile PBS, dissolved at 40–50°C). Cells were detached with trypsin and resuspended in culture medium. A suspension of 1 × 105 cells/ml was prepared and 2 × 104 cells were seeded in each well. The plate was incubated at 37°C in a humidified atmosphere with 5% CO2 for 48 h. Aggregation was evaluated under an inverted microscope (objective of 4X) and photographed with a Nikon digital camera.

Matrigel invasion assay

For the invasion assay (23), Matrigel invasion chambers with 24 wells (BD Biocoat) were hydrated by filling the inner and outer compartments with α-MEM medium and incubating them for 1 h at 37°C. Cells were detached with trypsin and 5 × 104 cells were then incubated for 24 h at 37°C in 5% CO2. The cells and Matrigel from the upper side of the filters were removed with a pre-wet ‘cotton swab’. The filters were washed in PBS, fixed in ice-cold methanol for 15 min and mounted in glass coverslips with Vectashield/DAPI. The total number of invasive nuclei was counted using a Leica DM2000 microscope.

FUNDING

Fundação para a Ciência e Tecnologia, Portugal (PTDC/SAU-OBD/64319/2006, SFRH/BD/15239/2004); National Fund for Scientific Research, Flanders (FWO).

ACKNOWLEDGEMENTS

We appreciate the technical help of Cecília Durães (real-time PCR), Wies Deckers and Eef Parthoens (confocal microscopy). Joana Simões-Correia is part of the Graduate Programme in Areas of Basic and Applied Biology (GABBA), Porto, Portugal.

Conflict of Interest statement. The authors declare no conflict of interest.

REFERENCES

1
Gumbiner
B.M.
Cell adhesion: the molecular basis of tissue architecture and morphogenesis
Cell
 , 
1996
, vol. 
84
 (pg. 
345
-
357
)
2
Berx
G.
Becker
K.F.
Hofler
H.
van Roy
F.
Mutations of the human E-cadherin (CDH1) gene
Hum. Mutat.
 , 
1998
, vol. 
12
 (pg. 
226
-
237
)
3
Machado
J.C.
Oliveira
C.
Carvalho
R.
Soares
P.
Berx
G.
Caldas
C.
Seruca
R.
Carneiro
F.
Sobrinho-Simoes
M.
E-cadherin gene (CDH1) promoter methylation as the second hit in sporadic diffuse gastric carcinoma
Oncogene
 , 
2001
, vol. 
20
 (pg. 
1525
-
1528
)
4
Peinado
H.
Portillo
F.
Cano
A.
Transcriptional regulation of cadherins during development and carcinogenesis
Int. J. Dev. Biol.
 , 
2004
, vol. 
48
 (pg. 
365
-
375
)
5
Becker
K.F.
Atkinson
M.J.
Reich
U.
Becker
I.
Nekarda
H.
Siewert
J.R.
Hofler
H.
E-cadherin gene mutations provide clues to diffuse type gastric carcinomas
Cancer Res.
 , 
1994
, vol. 
54
 (pg. 
3845
-
3852
)
6
Carneiro
F.
Oliveira
C.
Suriano
G.
Seruca
R.
Molecular pathology of familial gastric cancer, with an emphasis on hereditary diffuse gastric cancer
J. Clin. Pathol.
 , 
2008
, vol. 
61
 (pg. 
25
-
30
)
7
Mareel
M.
Leroy
A.
Clinical, cellular, and molecular aspects of cancer invasion
Physiol. Rev.
 , 
2003
, vol. 
83
 (pg. 
337
-
376
)
8
Chen
Y.T.
Stewart
D.B.
Nelson
W.J.
Coupling assembly of the E-cadherin/beta-catenin complex to efficient endoplasmic reticulum exit and basal-lateral membrane targeting of E-cadherin in polarized MDCK cells
J. Cell Biol.
 , 
1999
, vol. 
144
 (pg. 
687
-
699
)
9
Bryant
D.M.
Stow
J.L.
The ins and outs of E-cadherin trafficking
Trends Cell Biol.
 , 
2004
, vol. 
14
 (pg. 
427
-
434
)
10
Kaurah
P.
MacMillan
A.
Boyd
N.
Senz
J.
De Luca
A.
Chun
N.
Suriano
G.
Zaor
S.
Van Manen
L.
Gilpin
C.
, et al.  . 
Founder and recurrent CDH1 mutations in families with hereditary diffuse gastric cancer
JAMA
 , 
2007
, vol. 
297
 (pg. 
2360
-
2372
)
11
Ahner
A.
Brodsky
J.L.
Checkpoints in ER-associated degradation: excuse me, which way to the proteasome?
Trends Cell Biol.
 , 
2004
, vol. 
14
 (pg. 
474
-
478
)
12
Yoshida
H.
ER stress and diseases
FEBS J.
 , 
2007
, vol. 
274
 (pg. 
630
-
658
)
13
Chaudhuri
T.K.
Paul
S.
Protein-misfolding diseases and chaperone-based therapeutic approaches
FEBS J.
 , 
2006
, vol. 
273
 (pg. 
1331
-
1349
)
14
Sitia
R.
Braakman
I.
Quality control in the endoplasmic reticulum protein factory
Nature
 , 
2003
, vol. 
426
 (pg. 
891
-
894
)
15
Meusser
B.
Hirsch
C.
Jarosch
E.
Sommer
T.
ERAD: the long road to destruction
Nat. Cell Biol.
 , 
2005
, vol. 
7
 (pg. 
766
-
772
)
16
Anelli
T.
Sitia
R.
Protein quality control in the early secretory pathway
EMBO J.
 , 
2008
, vol. 
27
 (pg. 
315
-
327
)
17
Karam
R.
Carvalho
J.
Bruno
I.
Graziadio
C.
Senz
J.
Huntsman
D.
Carneiro
F.
Seruca
R.
Wilkinson
M.F.
Oliveira
C.
The NMD mRNA surveillance pathway downregulates aberrant E-cadherin transcripts in gastric cancer cells and in CDH1 mutation carriers
Oncogene.
 , 
2008
, vol. 
27
 (pg. 
4255
-
4260
)
18
Oliveira
C.
Seruca
R.
Carneiro
F.
Genetics, pathology, and clinics of familial gastric cancer
Int. J. Surg. Pathol.
 , 
2006
, vol. 
14
 (pg. 
21
-
33
)
19
Suriano
G.
Ferreira
P.
Mateus
A.R.
Correia
J.
Henriques
L.
Seruca
R.
Genetics of hereditary diffuse gastric cancer: progress and future challenges
Future Oncol.
 , 
2006
, vol. 
2
 (pg. 
363
-
370
)
20
Shore
E.M.
Nelson
W.J.
Biosynthesis of the cell adhesion molecule uvomorulin (E-cadherin) in Madin-Darby canine kidney epithelial cells
J. Biol. Chem.
 , 
1991
, vol. 
266
 (pg. 
19672
-
19680
)
21
Perlmutter
D.H.
Chemical chaperones: a pharmacological strategy for disorders of protein folding and trafficking
Pediatr. Res.
 , 
2002
, vol. 
52
 (pg. 
832
-
836
)
22
Tveten
K.
Holla
O.L.
Ranheim
T.
Berge
K.E.
Leren
T.P.
Kulseth
M.A.
4-Phenylbutyrate restores the functionality of a misfolded mutant low-density lipoprotein receptor
FEBS J.
 , 
2007
, vol. 
274
 (pg. 
1881
-
1893
)
23
Suriano
G.
Oliveira
C.
Ferreira
P.
Machado
J.C.
Bordin
M.C.
De Wever
O.
Bruyneel
E.A.
Moguilevsky
N.
Grehan
N.
Porter
T.R.
, et al.  . 
Identification of CDH1 germline missense mutations associated with functional inactivation of the E-cadherin protein in young gastric cancer probands
Hum. Mol. Genet.
 , 
2003
, vol. 
12
 (pg. 
575
-
582
)