• Open Access

Germline alterations in the CDH1 gene in familial gastric cancer in the Japanese population


To whom correspondence should be addressed.
E-mail: hsugimur@hama-med.ac.jp


Germline point or small frameshift mutations of the CDH1 (E-cadherin) gene are known to cause familial gastric cancer (FGC), but the frequency of CDH1 mutations is low in Japanese patients with FGC. Because recent studies have reported germline large genomic deletions of CDH1 in European and Canadian patients with FGC, in the present study we examined DNA samples from 13 Japanese patients with FGC to determine whether similar germline changes were present in CDH1 in this population. Using a sequencing analysis, a 1-bp deletion (c.1212delC), leading to the production of a truncated protein (p.Asn405IlefsX12), was found in an FGC family; immunohistochemical analysis revealed the loss of CDH1 protein expression in the tumors in this family. Using a combination of multiplex ligation-dependent probe amplification (MLPA) and RT-PCR analyses, we also found a large genomic deletion (c.164-?_387+?del), leading to the loss of exon 3 and the production of a truncated protein (p.Val55GlyfsX38), in another FGC family. The functional effects of the detected mutations were examined using a slow aggregation assay. Significant impairment of cell–cell adhesion was detected in CHO-K1 cells expressing Ile405fsX12- and Gly55fsX38-type CDH1 compared with cells expressing wild-type CDH1. Our results suggest that the p.Asn405IlefsX12 and p.Val55GlyfsX38 mutations of the CDH1 gene contribute to carcinogenesis in patients with FGC. This is the first report of CDH1 germline truncating mutations in Japanese patients with FGC. Screening for large germline rearrangements should be included in CDH1 genetic testing for FGC. (Cancer Sci 2011; 102: 1782–1788)

Gastric cancer is one of the most common cancers worldwide and is divided histopathologically into two types: intestinal (differentiated) and diffuse (undifferentiated).(1) Hereditary diffuse gastric cancer (HDGC) is a cancer predisposition syndrome dominated by diffuse gastric cancer, and the CDH1 (E-cadherin) gene is known to be responsible.(2,3) The CDH1 gene is a calcium-dependent cell adhesion molecule and is one of the most important tumor suppressor genes in gastric cancer.(4) Sequencing analyses have revealed germline CDH1 mutations in approximately 30% of patients with HDGC, as determined using the criteria for HDGC defined by the International Gastric Cancer Linkage Consortium (IGCLC).(5,6) However, the detection rate of CDH1 germline mutations in Japanese patients with familial gastric cancer (FGC) is low compared with that in European patients.(7–10) The difference in the detection rate may be due to differences in the contribution of environmental factors, because the incidence of gastric cancer is relatively high in Japan.(1,5) As another possibility, gross CDH1 genomic rearrangements that cannot be detected using conventional sequencing analysis may be responsible for FGC in Japanese patients. Large genomic rearrangements have recently been reported to cause susceptibility to several hereditary cancers, such as those of the MLH1 or MSH2 genes in patients with Lynch syndrome,(11) the APC gene in patients with familial adenomatous polyposis,(12) and the BRCA1 gene in patients with familial breast cancer.(13) Based on these findings, we hypothesized that large genomic rearrangements of the CDH1 gene may be responsible for a subset of FGC in the Japanese population. In agreement with this hypothesis, Oliveira et al.(14) recently collected European and Canadian FGC patients in which no germline CDH1 mutations were detected using a sequencing analysis and identified some FGC families with germline large genomic deletions of the CDH1 gene. Therefore, in the present study, we examined 13 Japanese FGC families for the possible presence of large genomic rearrangements, as well as germline point or small frameshift mutations, in the CDH1 gene. We identified a germline 1-bp deletion (c.1212delC) and a germline large genomic deletion (c.164-?_387+?del), both of which led to the production of a truncated CDH1 protein, and functionally characterized these CDH1 mutant proteins.

Materials and Methods

Tissue samples, cell line, and nucleic acid extraction.  Blood samples and paraffin-embedded tissues were collected from 13 FGC families fulfilling the revised clinical criteria for HDGC.(5) The Chinese hamster ovary (CHO)-K1 cell line was purchased from Human Science Research Resource Bank (Osaka, Japan). Cells were cultured in α-minimum essential medium (α-MEM; Gibco BRL, Grand Island, NY, USA) supplemented with 10% FBS under a 5% CO2 atmosphere at 37°C. Genomic DNA was extracted using a QIAamp DNA Blood Maxi Kit (Qiagen, Valencia, CA, USA) or a DNeasy Tissue Kit (Qiagen). Total RNA was extracted with a PAXgene Blood RNA Kit (Qiagen) or with an RNeasy Plus Mini Kit (Qiagen). All samples were analyzed in a blinded manner. The research protocol was approved by the institutional review boards of Hamamatsu University School of Medicine and the relevant institutes.

Polymerase chain reaction and sequencing analysis.  Fragments covering all coding exons and boundary regions of the CDH1 gene were amplified using PCR. Platinum Taq PCRx DNA polymerase (Invitrogen, Carlsbad, CA, USA) was used for the amplification of exon 1, whereas HotStarTaq DNA polymerase (Qiagen) was used for the amplification of exons 2–16. Information on the primer sequences and PCR conditions have been described previously.(7,10,15) The PCR products were purified with Exo-SAP-IT (GE Healthcare Bio-Science, Piscataway, NJ, USA) and sequenced directly using a BigDye Terminator Cycle Sequencing Reaction Kit (Applied Biosystems, Tokyo, Japan) and the ABI 3100 Genetic Analyzer (Applied Biosystems).

Immunohistochemical analysis.  Immunohistochemical analysis was performed as described previously.(16) Briefly, paraffin block sections derived from patients who underwent a gastrectomy and/or autopsy were immunostained with a monoclonal antibody (mAb) against CDH1 (clone 36B5; epitope, N-terminal amino acid sequence; Novocastra, Newcastle, UK). Sections were also stained with H&E.

Multiplex ligation-dependent probe amplification analysis.  Twelve FGC patients negative for germline point or small frameshift mutations in the CDH1 gene were tested for large genomic deletions in the CDH1 gene using the SALSA P083-B1 CDH1 multiplex ligation-dependent probe amplification (MLPA) kit (MRC-Holland, Amsterdam, The Netherlands). The reactions were performed according to the manufacturer’s instruction. Probe ratios below 0.7 and above 1.3 were regarded as indicative of a decrease and increase, respectively, in the gene dosage.

Reverse transcription–polymerase chain reaction.  Total RNA was converted to cDNA using the SuperScript First-Strand Synthesis System for RT-PCR (Invitrogen) according to the manufacturer’s instructions. The following sets of primers were used in the PCR amplification: 5′-AGG TCT CCT CTT GGC TCT GC-3′ and 5′-CAG CTG ATG GGA GGA ATA ACC-3′ for the CDH1 transcripts; and 5′-TGG GCC AGA AGG ACT CCT AC-3′ and 5′-GCA TGA GGG AGA GCG TAG C-3′ for the β-actin transcripts. The PCR products were fractionated using electrophoresis on a 2.0% agarose gel and stained with ethidium bromide; the gel was then examined under UV light. A 100-bp DNA ladder (New England Biolabs, Beverly, MA, USA) was used. Any PCR products exhibiting multiple bands were sequenced after subcloning with a pGEM-T Easy vector system (Promega, Madison, WI, USA).

Plasmid construction.  Wild-type and exon 3 deletion-type CDH1 cDNA were inserted into a pIRESpuro2 mammalian expression vector (Clontech, Palo Alto, CA, USA). The expression vectors for the Ile405fsX12-type and Leu415-type CDH1 were generated using site-directed mutagenesis with the QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA). The sequences of all vectors were confirmed by sequencing.

Establishment of CHO-K1 cell lines stably expressing human CDH1.  The wild-type or mutant-type CDH1 plasmid vector was transfected into CDH1-negative CHO-K1 cells(17) using Lipofectamine 2000 reagent (Invitrogen). Puromycin-resistant clones were isolated by culturing in medium containing 5 μg/mL puromycin (Clontech). Positive clones were confirmed using a combination of RT-PCR, western blot, and immunofluorescence analyses.

Western blot analysis.  Cells were harvested in lysis buffer containing 10 mM HEPES (pH 7.5), 1.0% Nonidet P-40, 1 mM EDTA, 1 mM DTT, and 0.1 mg/mL PMSF. The whole-cell extracts were mixed with an equal volume of 2× SDS sample buffer and boiled. A 25-μg aliquot of the extract was subjected to SDS-PAGE and the proteins obtained were transferred electrophoretically to a PVDF membrane (GE Healthcare Bio-Science). Membranes were blocked with non-fat milk and incubated with an anti-CDH1 mAb (clone SHE78–7; epitope, the first extracellular domain;(18) Takara Bio, Shiga, Japan) or anti-β-actin mAb (Abcam, Cambridge, UK). After washing with Tris-buffered saline containing 0.1% Tween-20 (TBS-T), membranes were incubated with anti-mouse HRP-conjugated secondary antibody (GE Healthcare Bio-Science). After washing with TBS-T, immunoreactivity was visualized with an ECL chemiluminescence system (GE Healthcare Bio-Science).

Indirect immunofluorescence analysis.  The CHO-K1 clones were fixed, permeabilized, and blocked with goat serum. Cells were incubated with anti-CDH1 mAb SHE78–7 (epitope, the first extracellular domain;(18) Takara Bio) at room temperature for 1 h, and indirect immunofluorescence labeling was performed at room temperature for 1 h with an Alexa Fluor 488-conjugated goat anti-mouse secondary antibody (Molecular Probes, Eugene, OR, USA). Nuclei were stained with 4′,6′-diamidino-2-phenylindole (DAPI; Sigma, St Louis, MO, USA). Immunostained cells were examined, digitized, and stored as described previously.(19)

Slow aggregation assay.  Slow aggregation assays were performed as described previously.(20) Cells were trypsinized and transferred to an agar gel in a 96-well plate. After 48 h, aggregate formation was evaluated using an inverted microscope. The entire area was divided into “units” using a square mesh, with each unit containing a maximum of 30–40 cells. Quantification of the aggregate was estimated using the following formula:

Aggregate (%) = (no. units with cells occupying ≥50% of the unit)/(total no. units) × 100

Statistical analysis.  Statistical analyses were performed using Dunnett’s multiple comparison test with JMP version 7.01 software (SAS Institute, Cary, NC, USA).


Identification of two germline mutations of the CDH1 gene in Japanese FGC patients.  Of the 13 FGC families evaluated in the present study, we had shown previously that six were negative for germline point or small frameshift mutations in the CDH1 gene;(10,15) therefore, in the present study, we screened the probands of seven FGC families for germline CDH1 mutations using PCR and subsequent sequencing analysis. Representative pedigrees are shown in Figure 1. One heterozygous c.1212delC mutation at the CDH1 gene locus was found in one male proband (III-3) who was affected with signet-ring cell carcinoma of the stomach at 32 years of age (Figs 1a,2a). A deletion of one nucleotide (c.1212C) in exon 9 resulted in a frameshift at codon 404, the introduction of 11 novel amino acids, and the premature termination of a 415-amino acid protein (p.Asn405IlefsX12). A gastric cancer had also been recorded in four other family members (I-2, II-3, II-4, and II-5), and the c.1212delC mutation was detected in III-3’s aunt (II-4), who had gastric cancer, using sequencing analysis (Fig. 2a). The c.1212delC mutation had been found previously in one northern European FGC family,(5) meaning that this is the first case of the germline mutation in Asia. Immunohistochemical analysis showed the loss of CDH1 protein expression in cancerous gastric tissue from the proband and his aunt, indicating that a second hit had occurred in the remaining wild-type CDH1 allele (Fig. 2b).

Figure 1.

 Pedigrees of familial gastric cancer (FGC) families with germline CDH1 alterations, specifically (a) a c.1212delC (p.Asn405IlefsX12) germline mutation of the CDH1 gene and (b) a c.164-?_387+?del (p.Val55GlyfsX38) germline mutation of the CDH1 gene. Squares indicate males; circles indicate females. Solid symbols indicate gastric cancer patients. Symbols with a slash indicate deceased individuals. Arrows point to the probands. The numbers below the symbols indicate the age at diagnosis for affected family members or the age at the time of analysis for unaffected family members.

Figure 2.

 Identification of the c.1212delC (p.Asn405IlefsX12) germline mutation of the CDH1 gene. (a) Detection of the germline CDH1 mutation in DNA from the male proband (III-3) and his aunt (II-4) using direct sequencing analysis. Both electropherograms of CDH1 exon 9 show a heterozygous 1-bp deletion (c.1212delC). An arrow points to the position of the 1-bp deletion at codon 404. (b) Immunohistochemical analysis of the CDH1 protein in gastric cancers from this family. Non-cancerous tissue and cancerous gastric tissue samples from patient II-4 and cancerous gastric tissue from patient III-3 are shown. Scale bars, 50 μm.

Next, we screened the remaining 12 FGC patients who were negative for germline point or small frameshift mutations of CDH1 for large genomic rearrangements in the CDH1 gene using MLPA analysis. A decreased signal at CDH1 exon 3, suggesting a heterozygous deletion of a portion of CDH1, was detected in one female proband (IV-2) affected with signet-ring cell carcinoma of the stomach at 25 years of age (Figs 1b,3a). The decreased signal was also detected in a blood sample from the proband’s mother (III-4) and a paraffin-embedded duodenal sample from the proband (Fig. 3a). Among the proband’s family members, her brother (IV-1) and a male cousin of her mother (III-1) were diagnosed with gastric cancer at 22 and 28 years of age, respectively. To determine the effect of a reduced signal for exon 3, as detected using MLPA analysis, on the CDH1 mRNA transcript, RT-PCR was performed with a set of primers for the sequences on CDH1 exons 2 and 4. An aberrant band was detected in cDNA from the proband, and subsequent sequencing analysis confirmed that exon 3 of the CDH1 transcript was heterozygously deleted in this case (Fig. 3b–d); thus, a heterozygous c.164-?_387+?del mutation existed in the proband. The c.164-?_387+?del mutation was predicted to cause the production of a truncated protein (p.Val55GlyfsX38). As far as we know, the c.164-?_387+?del mutation observed in this case has not been reported previously as a germline mutation, indicating that it is a novel CDH1 germline mutation.

Figure 3.

 Identification of large genomic deletion (c.164-?_387+?del) of the CDH1 gene. (a) Detection of a decreased signal (asterisk) at CDH1 exon 3 in DNA from the female proband (IV-2) and her mother (III-4) using multiplex ligation-dependent probe amplification (MLPA) analysis. The names of the MLPA probes are shown below the panels. Data are shown as the mean ± SD. (b–d) Identification of CDH1 exon 3 deletion using RT-PCR and sequencing analyses. (b) A blood sample from patient IV-2 was subjected to RT-PCR with a set of primers for a sequence spanning exons 2 and 4 of CDH1; the products were subsequently electrophoresed on an agarose gel. The arrow indicates a band that is smaller than the band corresponding to the calculated size of the wild-type sequence. An individual not showing an abnormal signal in the MLPA analysis for CDH1 was used as a control. M, DNA size marker. (c) Direct sequencing analysis of the RT-PCR product from Patient IV-2. (d) Sequencing analysis of the subcloned RT-PCR product from patient IV-2.

Establishment of various stable CDH1 transfectants.  To better understand the relationship between the CDH1 mutations detected and the familial clustering of gastric cancer, we attempted to characterize the mutant CDH1 proteins (Gly55fsX38-type and Ile405fsX12-type) functionally. The Leu415-type CDH1 was also included in this analysis, because: (i) a germline mutation causing the production of the Leu415-type mutant CDH1 protein was reported previously in Japanese patients with FGC;(5) (ii) this amino acid substitution is predicted to affect protein function according to the Sorting Intolerant From Tolerant (SIFT) program;(21) and (iii) it has never been characterized in vitro. We transfected a human CDH1 expression vector into CHO-K1 cells lacking CDH1 expression(17) and CDH1 stable transfectants were isolated using puromycin selection (Fig. 4). The expression of the mRNA transcript and the protein of ectopic CDH1 was confirmed using a combination of RT-PCR, western blot, and immunofluorescence analyses (Fig. 4b–d). Regarding intracellular localization, membranous expression was detected in cells expressing the wild-type or Leu415-type CDH1, but not in cells expressing the Ile405fsX12-type or Gly55fsX38-type CDH1 (Fig. 4d). Through all the above analyses, five different CHO-K1 clones, including an empty vector-transfected clone, were established successfully.

Figure 4.

 Establishment of CHO-K1 cell lines stably expressing human CDH1 protein. (a) Scheme of the wild-type (Wt) and mutant (G55fsX38, I405fsX12, and L415) human CDH1 proteins. Amino acid sequences newly created by CDH1 genetic alterations are denoted in red in each mutant. Sig, signal peptide; EC, extracellular domain; TM, transmembrane domain; Cyto, cytoplasmic domain. (b) Detection of the expression of wild-type and mutant human CDH1 mRNA transcripts in stable CHO-K1 clones using RT-PCR analysis, which was performed for each clone isolated using a set of primers for a sequence spanning exons 2 and 4 of CDH1 to confirm ectopically expressed CDH1. Empty, empty vector-transfected clone. In addition, mRNA transcripts of the Chinese hamster β-actin were amplified as an internal control. M, DNA size marker. (c,d) Detection of the expression of wild-type and mutant human CDH1 protein in stable CHO-K1 clones using western blot analysis (c) and immunofluorescence analysis (d). Chinese hamster β-actin protein was used as an internal control. In the immunofluorescence analysis, CDH1 protein (green) was immunostained with an anti-CDH1 primary antibody and an Alexa Fluor 488-conjugated goat anti-mouse IgG secondary antibody. Nuclei were counterstained with 4′,6′-diamidino-2-phenylindole (blue).

Functional characterization of mutant CDH1 proteins.  Because cell–cell adhesion is a major function of the CDH1 protein, we compared levels of homotypic cell adhesion between the five CHO-K1 clones using a slow aggregation assay. Clones expressing the Gly55fsX38-type or Ile405fsX12-type CDH1 and an empty vector-transfected clone produced significantly fewer aggregations than the clone expressing wild-type CDH1 (Fig. 5). The quantity of cell–cell aggregation did not differ significantly between clones expressing wild-type and Leu415-type CDH1 (Fig. 5). These results suggest that Gly55fsX38-type and Ile405fsX12-type CDH1, but not Leu415-type CDH1, have an impaired cell–cell adhesion function.

Figure 5.

 Evaluation of the cell adhesion function of mutant-type CDH1 using a slow aggregation assay. Representative photographs of the results of a slow aggregation assay of CHO-K1 clones stably expressing wild-type (Wt) or mutant (G55fsX38, I405fsX12, and L415) human CDH1 are shown above the graph. Empty, empty vector-transfected clone. Data are the mean ± SD of four independent experiments. *< 0.05. Scale bars, 100 μm.


In the present study, DNA samples from 13 Japanese FGC families were examined for germline alterations in the CDH1 gene. Using sequencing analysis, a 1-bp deletion (c.1212delC: p.Asn405IlefsX12) was detected in an FGC family; in another FGC family, a large genomic deletion (c.164-?_387+?del: p.Val55GlyfsX38) was detected using MLPA and RT-PCR analyses. Both germline changes led to the production of a truncated CDH1 protein. When the level of cell–cell adhesion was compared between wild-type CDH1 and these two mutant CDH1 proteins using a slow aggregation assay, significantly impaired cell–cell adhesion function was detected for the two mutant proteins. These results suggest that the c.1212delC mutation in the CDH1 gene, which is associated with the production of p.Asn405IlefsX12, and the c.164-?_387+?del mutation in the CDH1 gene, which is associated with the production of p.Val55GlyfsX38, are involved in carcinogenesis in Japanese patients with FGC.

Using a slow aggregation assay, the present study detected a defect in the cell–cell adhesion function in cell lines harboring Gly55fsX38-type and Ile405fsX12-type CDH1. Both mutants are truncated CDH1 proteins; therefore, they lack the entire or part of the extracellular domain, transmembrane domain, and cytoplasmic domain of the mature CDH1 protein. The absence of these domains in both types of mutant proteins is considered responsible for their functional impairments. Because these results are in agreement with previous findings that germline CDH1 changes leading to a truncated protein are present in FGC patients of various ethnicities,(3) we considered that the p.Val55GlyfsX38 and p.Asn405IlefsX12 mutations are pathogenetic towards the familial occurrence of gastric cancers. All previously reported CDH1 germline mutations in Japan are missense types (p.Gly62Val, p.Ile415Leu, and p.Val832Met);(7,9,10) therefore, the present report is the first description of a CDH1 germline mutation associated with the production of a truncated CDH1 protein in Japanese patients with FGC.

In the recent review by Cisco et al.,(22) it was reported that the penetrance of diffuse gastric cancer in patients who carry a CDH1 mutation is estimated to be 63–83% for women and 40–67% for men. In the family with the c.164-?_387+?del CDH1 mutation (Fig. 1b), III-1, IV-1, and IV-2 were affected with gastric cancer in their 20s, however, II-3, II-4, and III-4 were unaffected, implying the penetrance of gastric cancer in this family seems slightly lower than the penetrance described by Cisco et al.(22) However, because some information, such as endoscopic surveillance data for II-3, II-4, and III-4, are unavailable and it is possible that new cases of gastric cancer may develop in the family members in future, it is important to investigate this kind of family more precisely and for a longer period of time to determine the correct penetrance.

In the present study, CDH1 germline alterations were detected in two of 13 Japanese FGC families (15.4%) who fulfilled the revised clinical criteria for HDGC according to the IGCLC. What is responsible for the gastric cancer in the remaining FGC families? One possible explanation is the contribution of environmental factors, such as infection with Helicobacter pylori (H. pylori) and the intake of salted/smoked and pickled/preserved foods (rich in salt, nitrites, and preformed nitroso compounds),(23) to the development of FGC. In fact, the incidence of H. pylori infection is known to be high in Japan.(24) Another possible explanation is the existence of other genes responsible for the development of FGC. So far, germline mutations of the p53, MET, and STK11 genes and genes causing Lynch syndrome have been detected in the familial clustering of gastric cancer.(25–29) However, the incidence of these germline mutations is likely to be lower than that of the CDH1 germline mutation, suggesting that previously unidentified genes responsible for FGC may exist. Future investigation using the “omics” approach or whole genome sequencing may identify novel genes responsible for FGC.

In the present study, MLPA analysis was performed to detect large genomic rearrangements, which are difficult to detect using conventional PCR sequencing, at the CDH1 gene locus. Indeed, an abnormal signal was detected in MLPA analysis in one case in which no germline CDH1 mutations had been detected by sequencing analysis. Therefore, similar to the detection of genomic rearrangements in the MLH1 and MSH2 genes, which has been proposed to be included in genetic testing for Lynch syndrome,(30) we propose that screening for germline large rearrangements of CDH1 should be included in CDH1 genetic testing for FGC in the Japanese population.


The authors acknowledge Mr T. Kamo (Hamamatsu University School of Medicine) for technical assistance. This work was supported by Grants-in-Aid from the Ministry of Health, Labour and Welfare (21–1), the Japan Society for the Promotion of Science (21790383, 22590356, and 22790378), the Scientific Support Programs for Cancer Research, a Grant-in-Aid for Scientific Research on Innovative Area, the Ministry of Education, Culture, Sports, Science and Technology (221S0001), the Princess Takamatsu Cancer Research Fund, and the Smoking Research Foundation.

Disclosure statement

The authors have no conflicts of interest.