Introduction

Gastric cancer is the most common cancer in east Asian countries such as Korea (Bae et al. 2001). CDH1 germline mutations have been reported in patients with the diffuse type of familial gastric cancers (Guilford et al. 1998; Yoon et al. 1999). However, CDH1 mutations are not recognized as a major cause of familial gastric cancer in Asian countries (Yoon et al. 1999; Kim et al. 2003). Recent reports indicate truncation mutations predominate in patients of Western origin, whereas only a few missense mutations are found in patients from Asian countries (Guilford et al. 1998; Yoon et al. 1999). Recently, we reported a MET germline missense mutation in a diffuse type of familial gastric cancer patient (Kim et al. 2003). However, the MET mutation frequency is low (5%, 1/21), suggesting that it is not a major cause of familial gastric cancer. Although somatic mutations have been reported in some genes, such as p53 and β-catenin (Fenoglio-Preiser et al. 2003; Woo et al. 2001), the genetic mechanisms underlying the development of gastric cancers have not been sufficiently identified. BRAF mutations have recently been found in a variety of human cancers (Davies et al. 2002). BRAF is one of three serine/threonine RAF kinases, is RAS-regulated, and participates in cell growth and malignant transformation kinase pathways (Brose et al. 2002; Smith et al. 2003). BRAF somatic mutations have been identified in 66% of malignant melanomas (Davies et al. 2002), 35.8%–69% of papillary thyroid carcinomas (Cohen et al. 2003; Kimura et al. 2003), 5.1%–10% of colorectal tumors (Rajagopalan et al. 2002; Yuen et al. 2002), 33% of low-grade ovarian serous carcinoma (Singer et al. 2003), and at a relatively low frequency in a wide range of other cancers (Davies et al. 2002). A high frequency of BRAF mutations has also been reported in 82% of nevi (Pollock et al. 2003). Reported BRAF mutations are confined to exons 11 and 15 (kinase domain), and up to 80% of BRAF mutations involve a V599E substitution (Davies et al. 2002). Interestingly, this most common of BRAF mutations has not been identified in tumors with K-ras mutations (Rajagopalan et al. 2002; Yuen et al. 2002; Kimura et al. 2003; Singer et al. 2003). This mutually exclusive relationship supports the hypothesis that BRAF (V599E) and K-ras mutations exert equivalent effects in tumorigenesis (Rajagopalan et al. 2002; Singer et al. 2003).

Whereas the majority of human cancer types have been screened for BRAF mutations, no primary gastric cancers have been examined, and only six gastric cancer cell lines have been investigated (Davies et al. 2002). Since between 0% and 28% of gastric cancers have K-ras mutations (Lee et al. 1995; Arber et al. 1997), we decided to investigate the incidence of BRAF mutations in gastric cancers, and the relationship between BRAF and K-ras mutations in such cancers. Our approach was to screen for BRAF and K-ras mutations in gastric tumors and cell lines to determine whether BRAF was involved in the development of gastric cancers. In addition, 20 familial gastric cancer patients without CDH1 and MET germline mutations were investigated for the presence of BRAF germline mutations.

Materials and methods

We screened BRAF and K-ras mutations in 66 gastric cancer tissues and 16 gastric cancer cell lines (SNU-1,-5, -16, -216, -484, -520, -601, -620, -638, -668, -719, AGS, KATO III, MKN45, MKN74, and NCI-N87). Criteria for selection of familial gastric cancers were at least two first- or second-degree relatives affected with gastric cancer, at least one of which was diagnosed with cancer prior to age 50 (Kim et al. 2003). Blood samples from probands of each of these families were collected from Seoul National University Hospital. Informed consent was obtained from all participants before testing. Pathological data were available on 62 of the 66 gastric cancer tissues and showed that 32 were diffuse types of gastric cancers, 24 were intestinal types, and six were mixed types. Of 20 familial cancer probands, eight represented families suffering from diffuse types of gastric cancer, four represented families suffering from intestinal types, and there was no histological data available for the remaining eight. DNA from tumor samples and from peripheral blood lymphocytes was extracted by using TRI reagent (Molecular Research Center, Cincinnati, Ohio, USA) according to manufacturers' instructions. Codons 12 and 13 of the K-ras gene were screened by bi-directional sequencing with the Taq dideoxy terminator cycle sequencing kit and an ABI 3100 DNA sequencer (Applied Biosystems, Foster City, Calif., USA). The following polymerase chain reaction (PCR) primer sequences were used for amplification of K-ras exon 1; F: 5'-GGTGGAGTATTTGATAGTGTA-3', R: 5'-GGTCCTGCACCAGTAATATGC-A-3'. Exons 11 and 15 of the BRAF gene were screened with previously described primer sets (Kimura et al. 2003) by both PCR-SSCP (single-strand conformational polymorphism) and DHPLC (denaturing high performance liquid chromatography; WAVE, Transgenomic, Omaha, Nb., USA), as previously described (Kim et al. 2000, 2003). The melting temperatures of each exon were optimized by analyzing melting curves with WAVEMAKER software (Transgenomic). All samples with abnormal PCR-SSCP bands or DHPLC patterns were subsequently sequenced. Lung adenocarcinoma cell line NCI-H1395 and colorectal cancer cell line HT-29 were used as positive controls for exon 11 and exon 15 BRAF gene mutations, respectively (Smith et al. 2003). Reverse transcription (RT)-PCR was performed with two primer sets to examine alternative splicing of the BRAF gene; F1: 5'-AAATGTTGAATGTGACAGCA-3', R1: 5'-CAAAATGGATCCAGACAACT-3', F2: 5'-TCCACAGAGACCTCAAGAGT-3', R2: 5'-GCACTCTGCCATTAATCTCT-3'. Complementary DNA was synthesized by using the SuperScript RT II system (Invitrogen, Carlsbad, Calif., USA).

Results and discussion

We screened 20 familial gastric cancer patients in order to identify BRAF germline mutations. No such mutations were found. We also screened 66 gastric cancer tissues and 16 gastric cancer cell lines and found a total of three BRAF polymorphisms. No clear pathogenic BRAF mutations were identified in exons 11 and 15. The SNU-638 gastric cancer cell line harbored a silent P452P mutation (CCT→CCC) in exon 11. Gastric tissues S40 and S45 showed the same sequence changes (IVS14-10T→C) 10 bp upstream of the acceptor site invariant AG of intron 14. These variations were also found in the matched normal tissues of S40 and S45 (Fig. 1). Ninety-six unrelated healthy individuals with two positive controls (S40 and S45) were screened by DHPLC in order to identify possible IVS14-10T→C polymorphisms. However, none of these healthy controls showed an aberrant diagram in DHPLC, whereas two positive controls showed abnormal patterns. Analysis involving RT-PCR followed by direct sequencing showed this intronic change did not result in alternative splicing. All 16 gastric cancer cell lines were re-screened by direct sequencing of exons 11 and 15 of BRAF to confirm the results obtained by PCR-SSCP and DHPLC analysis. In the K-ras mutation analysis, we identified seven K-ras (9%, 7/82) somatic mutations in the 66 gastric cancers and 16 gastric cancer cell lines. All seven K-ras mutations were found in codon 12. The results of the BRAF and K-ras mutation analyses are summarized in Table 1. Most K-ras mutations in gastric cancers were found in codon 12, in accordance with previous reports (Lee et al. 1995; Arber et al. 1997).

Fig. 1a, b
figure 1

BRAF polymorphism in intron 14 of tissue sample S45. a DHPLC chromatograms. Arrow Abnormal pattern in intron 14 of S45. The other two peaks originate from normal controls. b Automatic sequencing showing the nucleotide substitution (IVS14-10T→C)

Full size image
Table 1 BRAF and K-ras mutations in gastric cancers
Full size table

Although BRAF mutations have been found in many human cancer types, little is known of their possible presence in gastric cancers. Moreover, the exclusive relationship between BRAF and K-ras mutations suggests gastric cancers without K-ras mutations may contain BRAF mutations. We found that, although only 9% of our gastric cancer panels had K-ras mutations, there was not a high incidence of BRAF mutations. Thus, BRAF mutations, which are found in a variety of other human cancers, do not seem to be involved in gastric cancer development.