Silencing of HTR1B and Reduced Expression of EDN1 in Human Lung Cancers, Revealed by Methylation-Sensitive Representational Di€Erence Analysis

Silencing of HTR1B and reduced expression of EDN1 in human lung cancers, revealed by methylation-sensitive representational dierence analysis

Daiya Takai1,2, Yukiko Yagi1, Kuniko Wakazono1, Nobuya Ohishi2, Yutaka Morita2, Takashi Sugimura1 and Toshikazu Ushijima*,1

1Carcinogenesis Division, National Cancer Center Research Institute 5-1-1 Tsukiji, Chuo-ku, Tokyo 104-0045, Japan; 2Department of Respiratory Medicine, Graduate School of Medicine, University of Tokyo, 7-3-1, Hongo, Bunkyo-ku, Tokyo 113-8655, Japan

Aberrantly hypermethylated genes in human lung cancers Introduction were searched for by a genome scanning technique, methylation-sensitive-representational dierence analysis Lung cancer is one of the major causes of cancer death (MS-RDA). A total of 59 DNA fragments were isolated in many countries. Both genetic and epigenetic as those methylated more heavily in either/both of two alterations have been identi®ed as components of lung squamous cell carcinoma cell lines, EBC-1 and LK- alterations involved in multistep lung carcinogenesis. 2, than in a primary culture of normal human bronchial Mutations of the ras, p53 and RB genes are found in epithelium, NHBE. Thirty-four DNA fragments, whose approximately 15 ± 20, 40 ± 60 and 15 ± 30%, respec- hypermethylation was con®rmed in primary squamous tively, of human non-small cell lung cancers (NSCLCs) cell carcinomas, were sequenced. By database searches, (Sekido et al., 1998). Hypermethylation and transcrip- 17 of them were shown to be located within 2 kb of tional repression of the p16 and HIC-1 (hypermethy- putative CpG islands, and ®ve of the 17 DNA fragments lated in cancer) genes have been reported in 28 and had transcribed regions of known genes in their vicinities. 33% of NSCLCs (Eguchi et al., 1997; Merlo et al., By RT ± PCR of the ®ve genes in the carcinoma cell lines 1995). However, when the necessity of multiple and NHBE, decreased expression of HTR1B (5- alterations for the development of a cancer is taken hydroxytryptamine receptor 1B) and EDN1 (endothe- into account, a signi®cant portion of genetic alterations lin-1) was observed. Sequencing after bisul®te modi®ca- involved in NSCLCs still remains unknown. Consider- tion showed that the CpG island in the promoter region ing that hypermethylation of various genes plays wide of HTR1B was hypermethylated, while that of EDN1 roles in various human cancers (Baylin et al., 1998; was not. Demethylation and re-expression of HTR1B Jones and Laird, 1999; Sugimura and Ushijima, 2000), were observed after treatment of LK-2 cells with 5-aza- it can be hypothesized that there will be more genes 2'-deoxycytidine. In primary lung cancers, decreased that are involved in NSCLCs by their hypermethyla- mRNA expression of HTR1B was observed in 11 of 20 tion and silencing. cases, and that of EDN1 was in 16 of 20 cases. To scan the genome for such genes, we previously Immunohistochemical analysis of endothelin-1 con®rmed established the methylation-sensitive-representational that its immunoreactivity was reduced in squamous cell dierence analysis (MS-RDA) method (Ushijima et carcinoma cells compared with that in normal bronchial al., 1997). In this method, the whole genome is epithelial cells. Considering that endothelin-1 induces represented by DNA fragments (HpaII-amplicon) apoptosis in melanoma cells and that silencing of derived from unmethylated CpG-rich regions. This is endothelin receptor B is observed in prostate cancers, accomplished by the digestion of genomic DNA with its reduced expression was speculated to confer a growth HpaII, which recognizes the 5'-CCGG-3' sequence and advantage to lung cancer cells. MS-RDA was shown to is blocked by methylation of the internal C residue, isolate DNA fragments that are hypermethylated and followed by ampli®cation of the total digestion product silenced, such as HTR1B, and those whose expressions by PCR after ligation of a universal adaptor. When a are altered and the methylation statuses outside the speci®c genomic region is aberrantly methylated in a promoter region are altered, such as EDN1. Oncogene cancer, DNA fragments derived from the region are (2001) 20, 7505 ± 7513. absent in the HpaII-amplicon from the cancer, though they are present in the HpaII-amplicon from the Keywords: DNA methylation; lung neoplasms; normal counterpart. A series of subtractive hybridiza- endothelin-l; serotonin IB receptor; MS-RDA tions between the two HpaII-amplicons are known to reveal DNA fragments derived from the aberrantly methylated region. The MS-RDA method was recently shown to be capable of identifying all the known *Correspondence: T Ushijima; E-mail: [email protected] Received 28 June 2001; revised 14 August 2001; accepted 30 imprinted regions on mouse distal chromosome 2 August 2001 (Kelsey et al., 1999). Silencing/repression of HTR1B/EDN1 in lung cancers D Takai et al 7506 In this study, to identify genes aberrantly hyper- EBC-1 and LK-2 cell lines (Figure 2b). The putative methylated and transcriptionally silenced in NSCLCs, CGI 900 bp upstream of the translation initiation site we performed modi®ed MS-RDA using two squamous of the HTR1B gene (HTR1B-Pro in Figure 2c; the lung cancer cell lines and normal human bronchial transcriptional start site is unknown) and a CpG-rich epithelial cells, considering that squamous cell carcino- region within the fragment H6 (HTR1B-H6) were mas arise from the bronchial epithelia (Harris et al., sequenced after bisul®te modi®cation. The HTR1B-Pro 1991; Sekido et al., 1998). region was almost completely methylated in LK-2 and EBC-1, while it was demethylated in NHBE (Figure 2c). The HTR1B-H6 region was almost completely methylated in the two carcinoma cell lines, while it was Results partially demethylated in NHBE (Figure 2c). To examine the role of the hypermethylation in the loss Isolation of DNA fragments hypermethylated in of HTR1B expression, LK-2 cells were treated with a human lung cancers demethylating agent, 5-aza-2'-deoxycytidine (5-aza-dC). Two series of MS-RDA were performed, using NHBE After the treatment, it was observed that the H6 as the tester and either of human squamous cell lung fragment was demethylated and expression of HTR1B cancer cell lines, EBC-1 or LK-2, as the driver. A total were restored (Figure 2d,e). From these ®ndings, the of 96 clones, 48 clones from each subtraction, were hypermethylation of the CGI in the upstream region of analysed for their independence by cross-hybridization, HTR1B was considered to be playing a causal role in and 66 of them were found to be independent. The 66 the loss of its expression. clones were further analysed by Southern blot whether or not they really detected dierential methylation in Decreased expression of EDN1 in carcinoma cell lines the carcinoma cell lines. Fifty-nine clones were found to detect hypermethylation, four clones detected The DNA fragment B3 was located in the junction of polymorphisms, and three clones were not dierentially intron 1 and exon 2 of EDN1. The DNA fragment was methylated in the tester and the drivers. To exclude completely methylated in EBC-1 and LK-2 cell lines, arti®cial hypermethylation during the establishment of while it was demethylated in NHBE (Figure 3a). EDN1 the cell lines, the 59 DNA fragments were examined for expression was partially and completely lost in EBC-1 their methylation status in ®ve primary squamous cell and LK-2, respectively (Figure 3b). A putative CGI carcinomas. Thirty-four clones were shown to be 200 bp upstream of the transcriptional initiation site of hypermethylated in two or more clinical samples by the EDN1 gene (EDN-Pro in Figure 3c) and a CpG- Southern blot analysis using HpaII (Table 1). rich region in intron 1 (EDN-B3) were sequenced after bisul®te modi®cation. The CGI was not methylated in any of NHBE, EBC-1 and LK-2 (Figure 3c). On the Identification of genes with decreased expression other hand, the EDN-B3 region was completely These 34 dierentially methylated DNA fragments methylated in EBC-1 and LK-2 while it was were sequenced, and, by GenBank searches, the demethylated in NHBE. ¯anking genomic sequences were identi®ed for 30 fragments. Seventeen of them had CGIs (Gardiner- Reduced expression of HTR1B and EDN1 in Garden and Frommer, 1987; Larsen et al., 1992) in primary lung cancers their 2 kb vicinities (Table 1), and ®ve had known genes in their vicinities (Figure 1). Fragment B3 was To exclude artifacts only observed in cell lines, the derived from intron 1 of Endothelin-1 (EDN1); D3 methylation status of the DNA fragments and from exon 1 of Protocadherin g-a12 (PCDHGA12); expression levels of HTR1B and EDN1 were examined F12 from exon 5 of Myeloid ecotropic viral integration in primary human lung cancers and the corresponding site 1 (MEIS1); G4 from exon 1 of Collagen12-a1 normal tissues. Since the primary normal lung tissue (COL12A1); and H6 from the upstream region of 5- consisted of cells of various origins and primary cancer Hydroxytryptamine receptor 1B (HTR1B). The expres- samples had contaminating stromal cells, an increase in sion levels of these ®ve genes were analysed by RT ± the proportion of methylated DNA molecules and PCR using NHBE and the two carcinoma cell lines, decrease of expression levels were taken as positive and those of EDN1 and HTR1B were found to be (Figure 4). The DNA fragment H6 was hypermethy- decreased in carcinoma cell lines. lated in 14 of 20 primary cancers, and decreased HTR1B expression was observed in 11 of 20 (Table 2). The DNA fragment B3 was hypermethylated in 14 of Transcriptional silencing of HTR1B in 20 primary cancers, and decreased EDN1 expression carcinoma cell lines was observed in 16 of 20 (Table 2). DNA fragment H6 was located 3 kb upstream of an Immunohistochemistry of endothelin-1, an antibody intronless gene, HTR1B (Demchyshyn et al., 1992). against which was available, was performed to con®rm The DNA fragment was completely methylated in two that it was expressed in normal alveolar and bronchial cancer cell lines while it was demethylated in NHBE epithelial cells and reduced in cancer cells. Immuno- (Figure 2a). HTR1B expression was completely lost in reactivity of endothelin-1 was strongly positive in the

Oncogene Table 1 List of clones hypermethylated in human lung cancers Clone Flanking genomic region Name Length (bp) % (G+C) Obs CpG/Exp CpG Flanking CGI (distance*) Large-insert clone Chromosomal position Gene/EST

1 A6 338 55.9 0.69 No AL117192 Ch. 14 cDNA clone/NCI CGAP synovical sarcoma 2 A8 403 46.7 0.44 Yes (1.5 kb) AL357134 1p35.1-36.12 3 A9 470 44.7 0.30 No AL390030 1p31.3-32.3 4 A11 363 54.0 0.57 Yes (0 kb) AL136295 Ch.14 cDNA clone/NCI CGAP gastric tumor 5 B1 391 41.7 1.12 Yes (0 kb) AC005631 Ch.7? 6 B2 447 51.7 0.54 Yes (0.2 kb) AL133467 Ch.14 cDNA clone/NCI CGAP breast tumor 7 B3 414 54.1 0.40 Yes (1.8 kb) Z98050 6p24 EDN1 /Exon 2 8 B5 484 47.5 0.18 No AL390063 10q23 9 B7 475 58.3 0.25 No AL022326 22q12.1-12.3 10 B8 395 58.5 0.18 ± ± ± 11 B9 432 37.7 0.52 No AL161640 Ch.1 12 B10 390 50.0 0.34 ± ± ± 13 B11 440 49.8 0.50 Yes (0 kb) AC008640 Ch.5 14 C5 346 46.2 0.43 ± ± ± 15 C9 427 50.8 0.57 Yes (0 kb) AL355385 Ch.6 16 C12 485 49.9 0.30 No AL356459 Ch.1 17 D3 439 46.7 0.63 Yes (0 kb) AC008579 Ch.5 PCDHGA12/Exon 1

18 D12 452 54.2 0.24 No Z98743 22q13.2-13.33 Takai D of Silencing/repression 19 E2 477 47.6 0.07 No AC008164 Ch.2? 20 E10 412 51.5 0.92 Yes (0 kb) AC002407 Ch.X 21 E11 341 47.2 0.64 ± ± ± al et 22 F2 476 43.5 0.49 No AL034451 6q23.2-24.2 23 F3 450 43.1 0.58 Yes (0.3 kb) AL138691 Ch.13 24 F6 439 42.6 0.86 Yes (0.2 kb) AC012400 Ch.8 25 F7 440 47.5 0.32 No AL445529 Ch.6

26 F8 452 46.2 0.56 Yes (0 kb) AC016057 Ch.8 HTR1B/EDN1 27 F12 457 38.3 0.54 Yes (2 kb) AC068404 Ch.2 MEISI /Exon 6 28 G4 484 53.5 0.47 Yes (0.2 kb) AL136448 Ch.6 COL12A1 /Exon 2 29 G8 421 48.5 0.49 Yes (0.1 kb) AC067804 Ch.4 30 G11 340 42.9 0.39 No AC073490 Ch.X

31 H6 388 51.6 0.55 Yes (0 kb) AL049595 6q13-14.3 HTR1B cancers lung in 32 H8 429 47.3 0.19 No AC074383 Ch.8 33 H10 430 47.4 0.91 No AC024527 Ch.3 34 H11 474 35.2 0.34 Yes (2 kb) AC003035 Ch.X

Calculation of 6Obs CpG/Exp CpG' and judgment of an CpG island were performed according to Kelsey et al. (1999). *Distance between CGI and DNA fragment isolated by MS-RDA Oncogene 7507 Silencing/repression of HTR1B/EDN1 in lung cancers D Takai et al 7508

Figure 1 Physical map of the ®ve fragments that had known genes in their vicinities. Closed boxes represent known exons, and open boxes represent DNA fragments isolated by MS-RDA. GpC, CpG and HpaII sites are represented by tick marks

brush border of bronchial epithelial cells and moder- including its distribution in air-way systems. It is of ately positive in their cytoplasm and in alveolar great interest to examine the role of shut-o of 5- epithelium (Figure 5a ± c). It was also strongly positive hydroxytryptamine (serotonin) and its signaling from in alveolar macrophages (Figure 5c; shown by arrow- its type B receptor. heads), and in vascular endothelium (Figure 5d), as In the case of EDN1, the CGI in its promoter region previously reported (Ehrenreich et al., 1990; Markewitz was not methylated in two carcinoma cell lines. On the et al., 1995; Mattoli et al., 1990). In all the three cases other hand, methylation in intron 1, where no CGI was (cases 1, 3 and 8), immunoreactivity was absent in present, was observed in the two carcinoma cell lines carcinoma cells (Figure 5e). while it was absent in NHBE. EDN1 mRNA expression was lost in LK-2 and moderately decreased in EBC-1. Considering that methylation of gene body Discussion and introns can be involved in the regulation of gene expression (Costello et al., 1994; Sengupta and Smith, Two genes were identi®ed as those whose expressions 1998), the role of the methylation of EDN1 intron 1 in were decreased in human lung carcinoma cell lines and its decreased expression cannot be concluded at this primary cancers. In the case of HTR1B, CGI in its stage. Decreased expression was also observed in the promoter region was methylated in LK-2 and EBC-1, primary cancers (16/20; 80%), but correlation between and its expression was restored by treatment with aza- the decreased expression and methylation of intron 1 dC. The methylation of the upstream region of the was not clear. Endothelin-1 was originally identi®ed as transcription start site is known to block transcription a potent vasoconstrictive peptide (Yanagisawa et al., signi®cantly (Baylin et al., 1998), and these facts 1988), and subsequent studies revealed its diverse strongly indicated that HTR1B gene was silenced by biological roles (Masaki et al., 1992). A recent study the promoter hypermethylation at least in the two showed that endothelin-1 induces apoptosis of human carcinoma cell lines. As for the role of HTR1B melanoma cells and inhibits their growth (Okazawa et silencing in lung carcinogenesis, little is known, al., 1998). This signal is mediated by endothelin

Oncogene Silencing/repression of HTR1B/EDN1 in lung cancers D Takai et al 7509

Figure 3 Hypermethylation of EDN1 intron 1 and its decreased expression in two carcinoma cell lines. (a) Methylation status of DNA fragment B3 isolated by MS-RDA. Each DNA of NHBE, EBC-1 and LK-2 was digested with HpaII (H) or MspI (M), and was hybridized with DNA fragment B3 (414 bp). The fragment B3 was completely methylated in the two carcinoma cell lines (lanes 3 and 5) while it was partially demethylated in NHBE. (b) Decreased expression of EDN1. EDN1 expression was partially and completely decreased in EBC-1 and LK-2, respectively. (c) Methylation status of the EDN1 promoter region and the EDN- B3 region. The fraction of methylated DNA molecules at each CpG site is shown by the proportion of ®ll in each circle. Although the EDN-B3 region was hypermethylated in the carcinoma cell lines, the EDN1 promoter region was not methylated in either NHBE, LK-2 or EBC-1

Figure 2 Silencing of HTR1B by its promoter hypermethylation in the carcinoma cell lines. (a) Methylation status of DNA fragment H6 isolated by MS-RDA. Each genomic DNA from prostate cancers (Nelson et al., 1996, 1997). These NHBE and two carcinoma cell lines was digested with HpaII or reports and our ®ndings suggest that endothelin-1 and MspI, and hybridized with DNA fragment H6 (388 bp). The ETRB may mediate apoptotic signals also in bronchial fragment H6 was completely methylated in the two carcinoma cell epithelium, and that inactivation of this signaling lines (lanes 2 and 3) while it was completely demethylated in NHBE (lane 1). (b) Expression levels of HTR1B analysed by pathway may help the bronchial epithelial cells escape RT ± PCR. HTR1B expression was completely lost in EBC-1 and from apoptosis and play a role in lung carcinogenesis. LK-2. (c) Methylation status of the HTR1B-H6 region and the In the clinical sample analysis, hypermethylation of CGI in the HTR1B promoter region. The fraction of methylated HTR1B and its decreased expression did not correlate DNA molecules at each CpG site is shown by the proportion of ®ll in each circle. The CGI in the HTR1B promoter region was clearly, in spite that correlation was clear in the cell almost completely methylated in LK-2 and EBC-1 while it was lines and re-expression by aza-dC was observed in LK- almost completely demethylated in NHBE. HTR1B-H6 region 2 cells. In clinical samples, the presence of contaminat- was relatively hypermethylated in the two carcinoma cell lines, ing non-cancerous cells in cancer samples and the compared with NHBE. (d) Demethylation of DNA fragment H6 diverse origins of cells in the normal samples should be by 5-aza-dC treatment. LK-2 cells were treated with 5-aza-dC twice for 24 h. Methylation status of genomic DNA was analysed considered. Especially, although NHBE is supposed to by Southern blot hybridization after digestion with HpaII, and be the normal counterpart of lung cancers (Al appearance of an unmethylated band (388 bp) was observed after Moustafa et al., 1999; Jakowlew et al., 1997; Moghal 5-aza-dC treatment. (e) Induction of HTR1B expression by 5-aza- and Neel, 1998), there still remains a possibility that dC. RT ± PCR showed that HTR1B was re-expressed after 5-aza- dC treatment lung cancers are originated from a minor population of cells in the lung, dierent from NHBE, and that the population has a methylated 5' CGI for the HTR1B gene. If this were the case, analysis of methylation receptor B (ETRB) (Okazawa et al., 1998). Hyper- patterns in various cell types would greatly facilitate methylation of the 5' CGI of the ETRB gene and its clari®cation of the origin of cancer cells. For this decreased expression have been reported in human purpose, development of a technique capable of in situ

Oncogene Silencing/repression of HTR1B/EDN1 in lung cancers D Takai et al 7510 Table 2 Methylation and expression levels of HTR1B and EDN1 in primary lung carcinomas HTR1B EDN1 Methylation Methylation Case of DNA Decreased of DNA Decreased No. Age Sex fragment H6 mRNA fragment B3 mRNA

Squamous cell carcinomas 171M++++ 260M 7 + 77 369F 7 +++ 478M+ 7 ++ 549M++++ 679M++++ 774M++++ 851F++++ 971M+ 7 ++ 10 77 M 77++ 11 52 M + 7 + 7 12 60 M 77++

Adenocarcinomas 13 75 F + 77+ 14 74 F + + + 7 15 56 F 777+ 16 55 M + + 7 + 17 52 M + 7 ++ 18 50 M + 77+ 19 65 M 7 +++ 20 59 M + + 77

whose expressions are silenced by the hypermethylation is expected to lead to identi®cation of new tumor- suppressors. It is noteworthy that recent public access to the human genome draft sequence greatly facilitated the power of MS-RDA by enabling us to identify CGIs and genes ¯anking the DNA fragments isolated by MS-RDA. Figure 4 Methylation and expression of HTR1B and EDN1 in primary lung cancers. (a,c) Southern blot analysis of DNA from pairs of normal adjacent tissues (N) and primary lung cancers (Ca) after digestion with HpaII (H) or MspI (M). Hybridization was performed with DNA fragment H6 (388 bp) for HTR1B (a), Materials and methods and with DNA fragment B3 (414 bp) for EDN1 (c). Reduction of the hybridization signal (closed arrow heads) in HpaII digest in a Cell lines and tissue samples cancer sample, compared with that in a normal sample, was taken Normal human bronchial epithelial cells (NHBE) were as hypermethylation of the respective DNA fragments. The appearance of hypermethylated bands (open arrow heads) was purchased from Clonetics (Walkersville, MD, USA). Two also taken into account. Hypermethylation of H6 was observed in human lung squamous cell carcinoma cell lines (EBC-1, LK- cases 1 and 4, and that of B6 was in cases 1, 3 and 4. Sq: 2) were obtained from the Japanese Collection of Research squamous cell carcinoma. (b,d) Expression levels of HTR1B and Bio-resources. Twenty pairs of a primary lung cancer and EDN1 were examined by RT ± PCR. Decreased expression of adjacent non-cancerous lung tissue were obtained from 15 HTR1B was observed in cases 1, 2 and 3. That of EDN1 was in male and ®ve female patients (average: 63.9 years old) with cases 1, 3 and 4 informed consent. DNA was extracted by standard phenol/ chloroform procedures (Sambrook, 2000). RNA was isolated using ISOGEN (Nippon gene, Tokyo, Japan).

methylation analysis, such as the one developed Modified MS-RDA (Nuovo et al., 1999), is important. In addition, there is a possibility that the H9 fragment was not exactly To enrich the DNA fragments derived from CpG islands located in the region that was critical for transcrip- (CGIs), we modi®ed the original method of MS-RDA (Ushijima et al., 1997). Genomic DNA of the tester and tional control and would show better correlation. driver were digested by HpaII (New England Biolabs, MS-RDA isolated 17 DNA fragments ¯anked by Beverly, MA, USA), and then fractionated by reverse CGIs. Since the methylation status of the ¯anking CGI electrophoresis (Kadokami and Lewis, 1995) to enrich was also altered for one of the two genes analysed in DNA fragments shorter than 500 bp (tester) and those this study, a signi®cant proportion of the 17 CGIs will shorter than 600 bp (driver). The RHpa adaptor was ligated have altered methylation status. Identi®cation of genes to the fractionated tester and driver DNA, and HpaII-

Oncogene Silencing/repression of HTR1B/EDN1 in lung cancers D Takai et al 7511

Figure 5 Reduced expression of EDN1 in cancer cells, compared with normal bronchial and alveolar epithelial cells. Cancer tissue and adjacent normal tissue were stained using anti-endothelin-1 monoclonal antibody, and photographs were taken from the same section (a ± e). (a,b) Normal bronchial epithelial cells, especially in the brush border, showed strong immunoreactivity of endothelin- 1(a, 2006; b, 4006). (c) Normal alveolar cells and macrophages (shown by arrowheads) (4006). (d) Vascular endothelial cells (4006). (e) Squamous cell carcinoma (2006). (f) H&E staining of the same area

amplicon was prepared by PCR in the presence of 1 M a 1% agarose gel, and transferred to a Nylon membrane betaine (Sigma, St. Louis, MO, USA), which facilitated (Hybond N+; Amersham-Pharmacia Biotech, Uppsala, ampli®cation of G+C rich sequences (Henke et al., 1997; Sweden) and UV-cross linked. The membranes were McDowell et al., 1998). The RHpa adaptor of the tester hybridized and denatured probes prepared by Gene Images HpaII-amplicon was switched to the JHpa adaptor, and the Random Prime Labeling Kit (Amersham-Pharmacia Bio- RHpa adaptor of the driver HpaII-amplicon was removed by tech). The hybridization was detected by a Gene Images MspI digestion and gel ®ltration. The tester HpaII-amplicon CDP-Star detection kit (Amersham-Pharmacia Biotech). was mixed with an excess amount of the driver HpaII- amplicon, and competitive hybridization was performed. DNA sequencing and database search Then, PCR with the JHpa primer was performed in the presence of 1 M betaine. The products after two cycles of Cycle-sequencing was performed with a BigDye Terminator competitive hybridization were cloned into the pGEM-T kit (Applied Biosystems, Foster City, CA, USA) using an Easy vector (Promega, Madison, WI, USA). After transfor- ABI 310 automated DNA sequencer (Applied Biosystems). mation, insert-positive clones were selected by PCR using Sp6 Repetitive sequences were screened by the Repeat Masker and T7 primers. Clones were dot-blotted, and cross- Web site (University of Washington Genome Center, http:// hybridization was performed with the above PCR products ftp.genome.washington.edu/cgi-bin/RepeatMasker), and se- after removing the adaptor and plasmid portions. quence homology was searched for by the BLASN algorithm.

Southern blot analysis RT ± PCR analysis Two mg DNA was digested with 20 units of HpaII or MspI cDNA was synthesized by a Superscript II kit (Life (New England Biolabs). Puri®ed DNA was eletrophoresed in Technologies, Rockville, MD, USA) using 1 mg of total

Oncogene Silencing/repression of HTR1B/EDN1 in lung cancers D Takai et al 7512 RNA treated with DNase I (Life Technologies). b-actin was TTT ATT TGT GTT GGG AGG GGA GT-3';A:5'-CAC ampli®ed as a control, and the amounts of cDNA used for ATT CCA AAA TTT TCA TTC CC-3'. The PCR product PCR of target genes were adjusted. All reactions were was cloned into pGEM-T Easy, and used as a template for performed with RT-free controls. The PCR conditions cycle sequencing. The proportion of methylated cytosine was (product length, annealing temperature, number of cycles) determined after sequencing 10 clones. and primers (S: sense, A: antisense) are as follows. EDN1 (224 bp, 658C, 28 cycles), S: 5'-GCC CTC CAG AGA GCG 5-aza-dC treatment TTA TG-3',A: 5'-GAT GGT TGG GGG AAC TCC TT-3'. PCDHGA12 (210 bp, 628C, 25 cycles), S: 5'-GAT TCT ACC LK-2 cells were seeded at a density of 16105 cells/10 cm AGA TGG AAG TGC-3',A:5'-ACC TGT CCG TTT TCC plate on day 0, and exposed to 5 mM 5-aza-dC (Sigma) for TCA G-3'. MEIS1 (224 bp, 658C, 28 cycles), S: 5'-CAA ATC 24 h on days 3 and 5. The cells were cultured in a fresh media TAA CTG ACC AGC CCT CTT-3',A:5'-TGT GCC AAC after each treatment, and harvested on day 8. TGC TTT TTC TGT T-3'. COL12A1 (161 bp, 628C, 25 cycles), S: 5'- TCG TAA TGT TGG AGT TGA AGT T-3', Immunohistochemistry A: 5'-AGC ACA CCT GGG AGA TG-3'. HTR1B (169 bp, 658C, 32 cycles, in the presence of 1 N betaine), S: 5'-TTC Deparanized sections were treated by microwave, and CTC AAG CCA ACT TAT CCT-3',A:5'-CGG TAC ACT endogenous peroxidase was inactivated by 0.3% hydrogen GTG GCA ATC AC-3'. b-actin (460 bp, 628C, 25 cycles), S: peroxide. Mouse monoclonal antibody against the carboxyl 5'-TTT GAG ACC TTC AAC ACC CCA G-3',A:5'-TTT terminal of human endothelin-1 (Yamasa Soy Sauce, Choshi, CGT GGA TGC CAC AGG A-3'. Japan) was used for the primary antibody in a dilution of 1:200 in 1% normal horse serum in PBS. Mouse IgG1 (DAKO, Copenhagen, Denmark) was used as a negative Sodium bisulfite modification and sequencing control in a dilution of 1:100. The primary antibody was According to the protocol by Rein et al. (1997), 1 mg of DNA detected by anti-mouse IgG antibody conjugated with biotin, was incubated with 0.6 N NaOH in a ®nal volume of 20 ml using a Vectastain ABC peroxidase staining kit (Vector for 15 min at 378C. One hundred and twenty microliters of a Laboratories, Burlingame, CA, USA) and diaminobenzidine solution containing 3.6 M sodium bisul®te (freshly prepared, tetrahydrochloride as the chromogen. Slides were counter- Sigma) and 0.6 mM hydroquinone (freshly prepared, Sigma) stained with Mayer's hematoxylin. was added, and the sample underwent 20 cycles of denaturation of 958C for 30 s and incubation at 508C for 15 min. The sample was desalted using the Wizard DNA Clean-Up System, and desulfonated by incubation with 0.6 N NaOH for 5 min at room temperature. The DNA was Acknowledgments ethanol-precipitated and re-suspended in 10 mlofH2O. The authors are grateful to Drs T Niki and T Yamada for Using 1 ml of the solution, each region was ampli®ed with providing tumor samples and preparing paran block the following primers and conditions (product length, sections; to Dr M Nagao for supportive discussions; and to annealing temperature, number of cycles). HTR1B-H6 Drs T Kuramoto and K Asada for critical reading of the (195 bp, 608C, 40 cycles), S: 5'-ATT TGT TTG GAG GTT manuscript. D Takai is a recipient of a Research Resident TTT TTT AG-3',A:5'-TTC CTT TCT CAC TTT AAA Fellowship from the Foundation for Promotion of Cancer ACA CA-3'. HTR1B-Pro (244 bp, 558C, 40 cycles), S: 5'- Research. This study was supported by a Grant-in-Aid for GGA GTT TTT TTG GTT AGG AAA GG-3';A:5'-CGA Cancer Research from the Ministry of Health, Labor and CAC ACT AAA AAA AAA CAA AT-3'. EDN1-Pro Welfare; by a Grant-in-Aid for Human Genome and Gene (234 bp, 558C, 40 cycles), S: 5'-CGG GGT TGG TAG Therapy; by a Grant-in-Aid from the Ministry of Educa- TTT GTA AAG GG-3';A:5'-CGA CTT TAT TCC AAC tion, Science, Culture and Sports; and by a grant from the CCC AAA CA-3'. EDN1-B6 (173 bp, 608C, 40 cycles), S: 5'- Hishinomi Foundation for Cancer Research.

References

AlMoustafaAE,YansouniC,Alaoui-JamaliMAand Harris CC, Reddel R, Pfeifer A, Iman D, McMenamin M, O'Connor-McCourt M. (1999). Clin. Cancer Res., 5, Trump BF and Weston A. (1991). IARC Sci. Publ., 105, 681 ± 686. 294 ± 304. Baylin SB, Herman JG, Gra JR, Vertino PM and Issa JP. Henke W, Herdel K, Jung K, Schnorr D and Loening SA. (1998). Adv. Cancer Res., 72, 141 ± 196. (1997). Nucleic Acids Res., 25, 3957 ± 3958. Costello JF, Futscher BW, Tano K, Graunke DM and Pieper Jakowlew SB, Mariano JM, You L and Mathias A. (1997). RO. (1994). J. Biol. Chem., 269, 17228 ± 17237. Biochim. Biophys. Acta., 1353, 157 ± 170. Demchyshyn L, Sunahara RK, Miller K, Teitler M, Homan Jones PA and Laird PW. (1999). Nat. Genet., 21, 163 ± 167. BJ,KennedyJL,SeemanP,VanTolHHandNiznikHB. Kadokami Y and Lewis RV. (1995). Anal. Biochem., 226, (1992). Proc. Natl. Acad. Sci. USA, 89, 5522 ± 5526. 193 ± 195. Eguchi K, Kanai Y, Kobayashi K and Hirohashi S. (1997). Kelsey G, Bodle D, Miller HJ, Beechey CV, Coombes C, Cancer Res., 57, 4913 ± 4915. Peters J and Williamson CM. (1999). Genomics, 62, 129 ± Ehrenreich H, Anderson RW, Fox CH, Rieckmann P, 138. Homan GS, Travis WD, Coligan JE, Kehrl JH and Larsen F, Gundersen G, Lopez R and Prydz H. (1992). Fauci AS. (1990). J. Exp. Med., 172, 1741 ± 1748. Genomics, 13, 1095 ± 1107. Gardiner-Garden M and Frommer M. (1987). J. Mol. Biol., Markewitz BA, Kohan DE and Michael JR. (1995). Am. J. 196, 261 ± 282. Physiol., 268, L192 ± L200.

Oncogene Silencing/repression of HTR1B/EDN1 in lung cancers D Takai et al 7513 Masaki T, Yanagisawa M and Goto K. (1992). Med. Res. Okazawa M, Shiraki T, Ninomiya H, Kobayashi S and Rev., 12, 391 ± 421. Masaki T. (1998). J. Biol. Chem., 273, 12584 ± 12592. Mattoli S, Mezzetti M, Riva G, Allegra L and Fasoli A. Rein T, Zorbas H and DePamphilis ML. (1997). Mol. Cell. (1990). Am. J. Respir. Cell. Mol. Biol., 3, 145 ± 151. Biol., 17, 416 ± 426. McDowell DG, Burns NA and Parkes HC. (1998). Nucleic Sambrook J and Russel DW. (2000). Molecular Cloning, 3rd Acids Res., 26, 3340 ± 3347. edn. Cold Spring Harbor Laboratory Press, New York. MerloA,HermanJG,MaoL,LeeDJ,GabrielsonE,Burger Sekido Y, Fong KM and Minna JD. (1998). Biochim. PC, Baylin SB and Sidransky D. (1995). Nat. Met., 1, Biophys. Acta., 1378, F21 ± F59. 686 ± 692. Sengupta PK and Smith BD. (1998). Biochim. Biophys. Acta., Moghal N and Neel BG. (1998). Mol. Cell. Biol., 18, 6666 ± 1443, 75 ± 89. 6678. Sugimura T and Ushijima T. (2000). Mutat. Res., 462, 235 ± Nelson JB, Chan-Tack K, Hedican SP, Magnuson SR, 246. Opgenorth TJ, Bova GS and Simons JW. (1996). Cancer Ushijima T, Morimura K, Hosoya Y, Okonogi H, Tatematsu Res., 56, 663 ± 668. M, Sugimura T and Nagao M. (1997). Proc. Natl. Acad. Nelson JB, Lee WH, Nguyen SH, Jarrard DF, Brooks JD, Sci. USA, 94, 2284 ± 2289. Magnuson SR, Opgenorth TJ, Nelson WG and Bova GS. Yanagisawa M, Kurihara H, Kimura S, Tomobe Y, (1997). Cancer Res., 57, 35 ± 37. Kobayashi M, Mitsui Y, Yazaki Y, Goto K and Masaki Nuovo GJ, Plaia TW, Belinsky SA, Baylin SB and Herman T. (1988). Nature, 332, 411 ± 415. JG. (1999). Proc. Natl. Acad. Sci. USA, 96, 12754 ± 12759.

Oncogene