The endothelin receptor B (EDNRB) promoter displaysheterogeneous, site specific methylation patterns innormal and tumor cellsMartha M. Pao1, Masakazu Tsutsumi2, Gangning Liang1, Eva Uzvolgyi3,Felicidad A. Gonzales1 and Peter A. Jones1,+
1Department of Biochemistry and Molecular Biology, University of Southern California/Norris Comprehensive CancerCenter, Los Angeles, CA 90089-9176, USA, 2Department of Urology, Hitachi General Hospital, 2-1-1 Jonancho,Hitachi-shi, Ibaraki, 317-0077, Japan and 3Department of Pathology and Microbiology, University of NebraskaMedical Center, Omaha, NE 68198, USA
Received 12 December 2000; Revised and Accepted 19 February 2001
The 5′ region for the endothelin receptor B (EDNRB)gene is a complex CpG island giving rise to fourindividual transcripts initiating within the island.Here, for the first time, we analyze the relationshipbetween methylation and gene expression in a CpGisland located in the 5′ region of a gene with multipletranscription start sites. The CpG island was unmethyl-ated in normal prostate and bladder tissue, whereasit became methylated in apparently normal colonicepithelium. Tumors derived from these tissues werefrequently hypermethylated relative to the respectivenormal tissues. Analysis of 11 individual CpG siteslocated throughout the CpG island showed thatspecific sites with high methylation levels in severaltumors were also methylated in normal tissues,suggesting that they might serve as foci for furtherde novo methylation. This region also had high levelsof methylation in several cancer cell lines, and wefound that a low methylation level in a small regionwithin the 5′ region correlated with expression of the5′-most transcript. Interestingly, almost completemethylation 200–1000 bp downstream of the tran-scriptional start site did not block expression of thistranscript. Finally, we show that treatment with 5-aza-2′-deoxycytidine can induce transcriptional activationof the four EDNRB transcripts. Our results show theexistence of differential, tissue-dependent methyla-tion at the EDNRB 5′ region, suggest the existence ofa spreading mechanism for de novo methylation,starting from methylation hotspots, and show thathypermethylation immediately 3′ to a transcriptionalstart site does not prevent initiation.
INTRODUCTION
DNA methylation has been implicated in transcriptionalrepression (1) and the formation of condensed inactive chromatin
(2,3). Methylation is required for normal mammalian develop-ment (4) and generally occurs at CpG dinucleotides (5), whichare found in lower than expected frequency in the genomeexcept in CpG islands. These are regions of ∼1 kb in lengthwhich are often associated with promoters or transcribedregions of genes and are generally not methylated (6). It hasbecome increasingly clear that autosomal genes can besilenced in cancer by abnormal de novo methylation of CpGislands, leading to transcriptional downregulation of geneexpression (7,8). Examples of genes that are frequent targetsfor de novo methylation include CDKN2A (p16/INK4A) (9,10),RB1 (11,12) and CDKN2B (p15/INK4B) (13), the mismatchrepair gene MLH1 (14–17) and the estrogen receptor gene(ESR1) (18). The possible number of genes that may betargeted for de novo methylation in cancer is in all probabilitymuch higher than currently known, and scanning techniques,such as arbitrarily primed PCR (AP-PCR) (19), provide a rapidway to determine general patterns of methylation changes incancer cells and allow us to find new genes that have under-gone abnormal methylation changes during oncogenesis.Using the AP-PCR technique we found previously that the 5′region of the endothelin receptor B (EDNRB) gene is hyper-methylated in cancer compared with white blood cell (WBC)DNA (20). Other investigators had previously shown that theEDNRB gene is abnormally methylated in prostate cancer (21),but the relationship of this methylation with expression was notexamined.
The role of EDNRB in mediating vaso-constriction has beenwell established. Mutations in the EDNRB gene also give riseto Hirschsprung’s disease (22), a condition characterized bymegacolon and abnormal skin pigmentation (23). Furthermore,recent findings show that EDNRB signaling is necessaryduring development for proper migration of cells derived fromthe neural crest, including melanoblasts and enteric neuro-blasts (24). A role for EDNRB in carcinogenesis has not yetbeen established; however, EDNRB joins a growing number ofgenes that are important for normal development and maybecome disregulated in cancer (25).
The EDNRB gene has a complex 5′ region which we haveshown recently to give rise to four different transcripts (20).
+To whom correspondence should be addressed. Tel: +1 323 865 0816; Fax: +1 323 865 0102; E-mail: [email protected]
904 Human Molecular Genetics, 2001, Vol. 10, No. 9
Three of these transcripts encode for the EDNRB protein, buthave unique splicing of the 5′-untranslated region (5′-UTR),whereas a more upstream transcript has the potential of generatinga new EDNRB protein with a unique N-terminus. The rolesand regulation of these novel transcripts are not known;however, the known 5′ region of the EDNRB gene is a CpGisland spanning ∼1 kb, and the transcriptional start sites for allfour EDNRB transcripts are located within this CpG island.DNA methylation in promoter regions is known to decreasetranscriptional activity, but its effect on multiple transcriptswithin the same 5′ region has not been examined previously.
In the present study we studied the relationship betweenexpression and DNA methylation in a CpG island withmultiple transcriptional start sites, as this would extend ourunderstanding of the connection between methylation and tran-scriptional silencing. Specifically, analysis of the 5′ region ofthe EDNRB gene in a series of tumor and adjacent normaltissues of prostate, bladder and colon origin showed thatnormal prostate and bladder tissues are unmethylated, asexpected, whereas normal colonic tissue revealed a moderatelevel of methylation. Furthermore, the tumor tissue was gener-ally hypermethylated relative to the normal tissue. Our analysisincluded methylation data on 11 individual CpG sites spanningthe whole island and, interestingly, several non-adjacent CpGsites showed high levels of methylation in tumor tissues andsome of the normal samples. This pattern is suggestive of thesesites serving as a seeding point for methylation. Methylationcould potentially spread from these sites through the remainderof the CpG island, providing a potential mechanism to explainde novo methylation of CpG islands found in cancer cells. Toelucidate the relationship between the 5′ region methylationand expression of the four EDNRB transcripts, we analyzedexpression and methylation in a panel of cancer cell lines. Lowmethylation levels in a small region within the CpG islandcorrelated with expression of the EDNRB ∆3 transcript, the5′-most transcript in this cluster. Surprisingly, high levels ofmethylation immediately downstream (200–1000 bp) from
EDNRB ∆3 did not block its transcriptional activity, which isin disagreement with other studies (26,27) performed inplasmid and episomal vectors, where methylation along thevector backbone does inhibit transcriptional activity. Lastly,using 5-aza-2′-deoxycytidine (5-Aza-CdR) we showed thatexpression of the four EDNRB transcripts could be induced byDNA demethylation, further supporting the role of methylationin the control of this complex promoter. Our findingssuggest the existence of a spreading mechanism for de novomethylation starting from distinct methylation hotspots andshow that hypermethylation in the 5′ region of a gene does notnecessarily equate with transcriptional silencing.
RESULTS
The EDNRB 5′ region is a CpG island.
Our earlier AP-PCR analysis showed a fragment that washypermethylated in tumor tissue compared with WBC DNAand which corresponded to the 5′ region of the EDNRB gene.Sequencing upstream of the known sequence (GenBankaccession no. D13162) showed that the 2000 bp encompassingthe EDNRB 5′ region and exon 1 is a CpG island, as defined byAntequera et al. (28) (Fig. 1). The CpG island has a GC contentof 55% and a CpG observed/expected ratio of 0.74. Figure 1also shows the transcriptional start sites for the four transcriptsof the EDNRB gene located within this CpG island (20).
EDNRB 5′ region methylation in normal and tumor tissue
A panel of tumor and adjacent normal tissues was screened todetermine whether there were any changes in methylationlevels at the EDNRB locus (Fig. 2). The methylation levels of11 individual CpG sites in the EDNRB 5′ region (Fig. 1) werefirst measured in tissues derived from the prostate, bladder andcolon using the quantitative methylation sensitive-singlenucleotide primer extension (Ms-SNuPE) technique. In normalprostate tissue the levels of methylation were generally low, as is
Figure 1. The EDNRB promoter and the CpG island. Schematic representation of the CpG island associated with EDNRB and the additional 5′ sequence. Theenlarged area shows the CpG island only. Arrows indicate transcriptional start sites of each EDNRB mRNA. Black tick marks represent CpG dinucleotides. Trianglesshow the exact location of CpGs analyzed with Ms-SNuPE technique. Numbering is derived from GenBank accession no. D13162.
Human Molecular Genetics, 2001, Vol. 10, No. 9 905
expected for normal tissues (Fig. 2A) with increased methylationat CpG -130, the 5′-most CpG dinucleotide analyzed, which islocated at the fringe of the CpG island. Prostate tumor tissuehad mostly low methylation levels, nevertheless the levelswere higher than in the normal adjacent tissue. Higher methyl-
ation levels measured in the 3′ end of the CpG island wereconsistent with conclusions from a previous study (21).Normal bladder tissue was generally unmethylated, with theexception of CpG -130, where methylation was high (Fig. 2B)and two distinct patterns of methylation could be discerned in
Figure 2. DNA methylation profile of EDNRB promoter in primary tissue samples. (A) Methylation in normal and tumor prostate tissue. (B) Methylation in normaland tumor bladder tissue. (C) Methylation in normal and tumor colon tissue. The left column lists the sample identification number. Sample IDs followed by -1, -2 or-3 indicate that multiple samples were taken from different areas of the same tumor. Sample ID followed by M indicates that sample was from a metastatic node.The top row lists CpG sites analyzed by Ms-SNuPE. Methylation levels were divided into four groups and color coded for easy visual recognition: blue, nomethylation/background methylation (0–10%); green, low methylation level (11–25%); yellow, intermediate methylation level (26–50%); red, high methylationlevel (>51%). The number in each colored box is the exact methylation level as determined by Ms-SNuPE. White boxes, not done. The top bar is a schematicrepresentation of the EDNRB promoter region, arrows indicate approximate location of transcriptional start sites relative to the CpGs analyzed.
906 Human Molecular Genetics, 2001, Vol. 10, No. 9
bladder tumors. Some bladder tumors showed low levels ofmethylation in the EDNRB 5′ region, but substantial hyper-methylation was observed in a subset of bladder tumors. Somesites were consistently heavily methylated (CpG 377, CpG618, CpG 654, CpG 1115) in these tumors and might thereforerepresent sites targeted for preferential de novo methylation.Conversely, CpG 336 remained resistant to hypermethylation,even when adjacent CpGs were highly methylated.
In contrast to prostate and bladder normal tissues, apparentlynormal colonic epithelium obtained from patients with colorectalcancer was extensively methylated (Fig. 2C). Hypermethyla-tion of other CpG islands in normal colonic mucosa has beendocumented previously and has been associated with aging(29). The EDNRB 5′ region was also a target for increasedmethylation in colon tumors and CpG 377, CpG 618, CpG 654and CpG 1115 had the highest levels of methylation, whereasCpG 336 again appeared to be protected from increased methyla-tion, in a pattern analogous to the one observed in bladdertumors. These findings showed that in the EDNRB 5′ regula-tory region, prostate, bladder and colon normal tissue havemethylation patterns that are particular to each tissue type. Ourfindings also showed that hypermethylation at the EDNRB5′ region is common in tumors, regardless of the tissue type.Furthermore, some sites within the CpG island appeared to bepreferential targets for de novo methylation, whereas othersseemed to be protected from hypermethylation changes.
EDNRB expression and methylation profile in cancer celllines
Next, we asked if the hypermethylation in the EDNRB 5′region correlated with expression of any of the EDNRB tran-scripts. The methylation profiles of the 5′ region in a panel ofcancer cell lines were measured and compared with the expres-sion of each of the four EDNRB transcripts in the same celllines (Fig. 3). Figure 3A shows that of nine cell lines analyzedonly the SK-Mel-28 melanoma cell line strongly expressed allfour transcripts as was reported previously (20). EDNRBexpression was detected in two additional cell lines. J82expressed EDNRB ∆3 strongly and expressed EDNRB ∆1 andthe original EDNRB transcript weakly, but did not expressEDNRB ∆2. Weak expression of EDNRB ∆3 was detected inSW48 and the remaining cell lines did not express any EDNRBtranscripts.
Methylation was not detected throughout the EDNRB 5′region in the SK-Mel-28 cell line (Fig. 3B), which correlatedwell with the high levels of expression of all EDNRB tran-scripts in this cell line. The methylation profile for the J82 cellline showed a bipartite pattern with the upstream regioncomprising five CpG sites (CpG -130 to CpG 336) having alower level of methylation (26%) compared with the moredownstream region comprising the latter six CpG sites, wherethe average methylation was considerably higher (75%).Therefore, the low level of methylation in the upstream region
Figure 3. Expression and methylation profiles in cancer cell lines. A panel of nine cancer cell lines was analyzed. SK-Mel-28 is a melanoma cell line. J82, 5637and T24 are bladder cancer cell lines. SW48, SW837, HCT-15, LoVo and HCT116 are colon cancer cell lines. (A) Expression of individual EDNRB transcripts wasdetermined by RT–PCR: +, strong expression; +/–, weak expression; –, no detectable expression. (B) Methylation profile of the EDNRB promoter region in celllines. The top row lists CpG sites analyzed by Ms-SNuPE. The number in each box is the exact methylation level as determined by Ms-SNuPE. The top bar is aschematic representation of the EDNRB promoter region, arrows indicate approximate location of transcriptional start sites relative to the CpGs analyzed.
Human Molecular Genetics, 2001, Vol. 10, No. 9 907
correlated with strong expression of EDNRB ∆3, but not of theother transcripts and high methylation levels downstream ofthe transcriptional start site did not block transcription ofEDNRB ∆3.
EDNRB expression can be induced by 5-Aza-CdRtreatment
We reasoned that if EDNRB expression was inhibited by DNAmethylation, then treatment of cells with the demethylatingagent 5-Aza-CdR would induce expression of EDNRB tran-scripts. The bladder cancer cell line T24 which had high methyla-tion levels in the 5′ region of EDNRB (Fig. 3B) and did notexpress any EDNRB transcript (Fig. 3A), was therefore treatedwith the drug and expression of EDNRB transcripts monitoredas a function of time after treatment (Fig. 4A). Figure 4Ashows a representative analysis of the kinetics of reactivationof EDNRB transcripts following 5-Aza-CdR treatment. Theoriginal, ∆1 and ∆3 transcripts were coordinately induced withpeak expression seen at day 21 after treatment, suggesting acommon regulatory mechanism. The EDNRB ∆2 transcriptwas induced more weakly and expression was detected on days21 and 35 after treatment. The kinetics of expression wereconfirmed in a separate experiment (data not shown) whichshowed minor variations in maximal expression levels andtiming. An additional five cell lines with low or no expression
of EDNRB transcripts (SW48, 5637, SW837, HCT-15,HCT116) were treated with 5-Aza-CdR. In all cases reactiva-tion of EDNRB transcripts was detected after treatment withthe drug (data not shown) showing that 5-Aza-CdR-mediatedtranscript reactivation of EDNRB transcripts was not limited tothe T24 cell line.
The methylation profile of the CpG island in 5-Aza-CdR-treated T24 cells was measured over a period of 35 days (Fig. 4B)to determine the possible relationship between methylation andexpression. The initial methylation at this locus in the T24 cellswas high, as was determined previously (Fig. 3B) and maximaldemethylation was reached between days 3 and 6 after drugtreatment. At day 21 post treatment remethylation was detect-able and increased over time. This increasing methylation wasfollowed by decreasing expression levels of the EDNRB tran-scripts. It is of note that the CpG sites within this island did notremethylate with equal kinetics, but that site specificity wasdetected. CpGs 165, 377 and 654 showed the most rapid rate ofremethylation, whereas CpG 336 seemed protected. Thispattern was similar to the one observed in tumor samples.Therefore, measurement of the kinetics of remethylation at theEDNRB 5′ region showed that methylation was correlated withtranscriptional activity and that there was preferential targetingof some CpG dinucleotides for remethylation.
Figure 4. Kinetics of EDNRB mRNA expression following chemical demethylation. (A) T24 cells were treated with the demethylating agent 5-Aza-CdR for 48 h.Expression of each EDNRB transcript was measured in untreated cells (day 0) and at various timepoints after drug treatment by semi-quantitative RT–PCR. Controlfor EDNRB amplification was SK-Mel-28 RNA. The labels on the right indicate the transcripts which were amplified. mRNA original + ∆1 denotes simultaneousamplification of both EDNRB original and EDNRB ∆1 transcripts. Due to the structure of the original transcript it was not possible to generate primers that wouldspecifically amplify the EDNRB original transcript only. On the other hand, EDNRB ∆1 could be amplified unequivocally by taking advantage of a unique 5′ end.GAPDH expression controls for presence of RNA in sample. (B) Methylation levels in the EDNRB 5′ region were measured using Ms-SNuPE in T24 cells thatwere treated with the demethylating agent 5-Aza-CdR. CpG dinucleotides and numbering are as in Figure 3.
908 Human Molecular Genetics, 2001, Vol. 10, No. 9
DISCUSSION
The purpose of this study was to investigate the role of DNAmethylation in the regulation of the complex EDNRB 5′ region,a CpG island containing three transcription start sites givingrise to four distinct mRNA transcripts. We have foundincreased levels of methylation in tumor compared withnormal samples derived from colon, bladder and prostatetissues. A few individual CpG dinucleotides exhibited highermethylation levels in normal tissue and might representseeding points from which methylation spreads in tumorsamples. Transcription of EDNRB mRNA was detected even inthe presence of high levels of methylation downstream of thetranscriptional start site.
Hypermethylation in tumors relative to normal tissue hasbeen described before in many instances (7,8) and it has beensuggested that these changes can be useful as markers for theearly detection of cancer cells. Hypermethylation in theEDNRB 5′ region could potentially be used as a marker formalignancy and our extensive study of the EDNRB 5′ regionshows that selected CpGs within the island may be morereliable markers for malignancy. However, background noiseparticular to a tissue type must be taken into account. Forexample, a previous study in the EDNRB promoter foundfrequent hypermethylation in prostate tumors at the more3′ end of the EDNRB promoter (21), yet, we show that thishypermethylation seldom spreads to the more upstream sites inprostate tumors. Conversely, bladder and colon tumors areoften highly methylated in the 5′ region, with the highest methyla-tion levels found around a few ‘hotspot’ sites. These hotspotscould possibly represent the most appropriate CpGs to use asmarkers for the detection of early methylation changes. Theintrinsic basal levels of methylation varied in each tissue sothat prostate and bladder normal tissue were mostly unmethylatedas expected. However, colon normal tissue had levels of methyla-tion which sometimes exceeded the levels in prostate orbladder tumors. This situation is analogous to the one in theestrogen receptor (ESR1) promoter, where hypermethylation innormal colon tissue has been correlated to ageing (18). There-fore, differences in basal methylation levels vary betweentissue types and must be taken into account when measuringhypermethylation events. Although the EDNRB 5′ regiondisplays abnormal methylation patterns in tumors of the colon,bladder and prostate, additional experiments will be needed toconfirm its value as a prognostic marker for the detection ofneoplastic changes.
Little is known about the mechanisms that result in CpGisland hypermethylation in cancer. Proposed mechanismsinclude the binding of proteins, such as SpI, that preventmethylation of the island (30,31) or the presence of boundarysequences that demarcate methylated and unmethylatedregions (32,33). We detected sites within the EDNRB 5′ regionthat were methylated in normal tissue and showed furtherhypermethylation in tumor tissue. This result is suggestive ofthe presence of ‘hotspots’ for methylation, rather than adefined boundary between methylated and unmethylatedsequences, and supports a model for a gradual increase ofmethylation originating from the core of the EDNRB island.
Previous studies using plasmid and episomal vectors (26,27)have used patch methylation techniques to show that methyla-tion downstream of a promoter can decrease transcription
levels. In the J82 cell line EDNRB ∆3 was expressed even inthe presence of almost complete methylation in the region200–1000 bp downstream of the transcriptional start site.Therefore, a promoter in its native chromosomal context maybe subject to levels of transcriptional control not present inplasmids or episomes. We do not know whether this extensivemethylation attenuated the transcription level, but the dataclearly show that extensive methylation closely downstream ofthe initiation site does not completely silence a gene.
Demethylation using 5-Aza-CdR in T24 cells was of interestsince it was revealed that activated transcripts could be coord-inately expressed and there was a considerable lag timebetween the demethylation event and maximal expressionlevels. All four EDNRB transcripts were activated by treatmentwith 5-Aza-CdR and in three of the four transcripts theexpression patterns observed over 35 days were surprising;expression was clearly coordinated between the individualtranscripts. We expected to see preferential expression of onetranscript, following the model of transcriptional interference(34–36). However, it should be noted that all EDNRB tran-scription start sites are located within one CpG island, and thisis the first time a promoter of its type has been analyzed. It ispossible that the close proximity of the transcription start sitesopens up the chromatin of the whole region when one tran-script is activated, thus resulting in coordinate expression of allthe mRNAs. Another interesting observation was the fact thatmaximal expression levels were not obtained until several daysafter the treatment. Previous reports showed that p16/INK4Areactivation closely followed treatment with 5-Aza-CdR, andmaximal expression was obtained 3–6 days after drug treat-ment (37). The different kinetics of expression and methylationat this 5′ region together with the unique promoter architecturesuggests that in addition to DNA methylation, other layers ofcontrol are at work in the regulation of expression of thiscomplex 5′ region.
MATERIALS AND METHODS
Tissue samples
Human bladder, colon and prostate tumor tissue and adjacentnormal tissue were obtained from patients from the LosAngeles County/University of Southern California MedicalCenter and the University of Southern California/NorrisComprehensive Cancer Center (Los Angeles, CA), followinginstitutional guidelines. To ensure purity of sample, mucosaltissue was dissected from surrounding muscle and adiposetissue, as previously described (10,38,39). DNA and RNA wasisolated as described previously (40).
Cell culture
Colorectal cancer cell lines (HCT116, LoVo, HCT-15,SW837, SW48) and bladder cancer cell lines (T24, J82, 5637)were obtained from the American Type Culture Collection(Rockville, MD). The melanoma cell line SK-Mel-28 wasobtained from Memorial Sloan Kettering (New York, NY). Allcolorectal cancer cell lines and T24 and 5637 were cultured inMcCoy’s 5A medium (Life Technologies). J82 was cultured inMEM (Life Technologies) supplemented with 0.1 mM non-essential amino acids and 1mM sodium pyruvate. SK-Mel-28
Human Molecular Genetics, 2001, Vol. 10, No. 9 909
was cultured in DMEM (Life Technologies). All media weresupplemented with 10% heat-inactivated FCS, 100 U/mlpenicillin and 10 µg/ml streptomycin. Cell lines were main-tained at 37°C, 5% CO2-95% air in standard tissue cultureincubators under sterile conditions.
5-Aza-CdR treatment
Cells (3 × 105) were seeded in a 100 mm cell culture dish 24 hprior to treatment with 5-Aza-CdR (Sigma). Cells wereincubated with 10–6 M 5-Aza-CdR for 24 h, medium waschanged and cells were incubated for another 24 h with freshlyprepared 10–6 M 5-Aza-CdR. At the conclusion of the treat-ment, medium was changed, and cells kept non-confluent andgrowing with medium changes every 3 days. The 5-Aza-CdRtreatments were performed multiple times to ensure reproduci-bility of results. DNA and RNA were harvested at varioustimepoints as described (40).
RT–PCR
Reverse transcription was performed using total RNA (2.5 µg),MMLV-reverse transcriptase (Life Technologies) and randomhexamers (Amersham Pharmacia Biotech) in a volume of25 µl, as described previously (40). The sequences of the fourEDNRB transcript variants have been described (20,41,42). Wegenerated primers that would specifically amplify EDNRB ∆1,EDNRB ∆2 and EDNRB ∆3. Due to sequence constraints it isnot possible to design primers that amplify only the originalEDNRB transcript, so we designed primers that amplify boththe original EDNRB transcript and EDNRB ∆1 and called theproduct EDNRB original + ∆1. PCR amplifications wereperformed using 100 ng cDNA, 1 U Expand High FidelityPCR system enzyme mix (Roche Molecular Biochemicals),0.2 mM each dNTP, 0.5 µM each sense and antisense primerand 1× Expand HF buffer in a total volume of 25 µl. Addition-ally, amplification of EDNRB original + ∆1 and EDNRB ∆2required 10% DMSO.
All PCR amplifications started with an initial denaturationstep at 94°C for 4 min and ended with a final extension step at72°C for 10 min. PCR cycling conditions for EDNRB original+ ∆1 were 36 cycles of 94°C for 45 s, 65°C for 60 s and 72°Cfor 60 s. The primers were EDNRB orig. + ∆1, 5′-GGC GGTATT AGC GTT TGC AGC GAC TT-3′ (sense), and EDNRBorig. + ∆1, 5′-GCC AGT CCT CTG CCA GCA GC-3′ (anti-sense). PCR cycling conditions for EDNRB ∆1 were 38 cyclesof 94°C for 45 s, 67°C for 60 s and 72°C for 60 s. Primers wereEDNRB ∆1, 5′- AGC TTT GCC TGG GAC CCC CAT C-3′(sense), and EDNRB ∆1, 5′-GCC AGT CCT CTG CCA GCAGC-3′ (antisense). PCR cycling conditions for EDNRB ∆2were 35 cycles of 94°C for 45 s, 63°C for 60 s and 72°C for60 s. Primers were EDNRB ∆2, 5′-GGA GCT GTA GCT CAGCCA GC-3′ (sense), and EDNRB ∆2, 5′-GAG ATG GTG CGTGGC GGA GA-3′ (antisense). PCR cycling conditions forEDNRB ∆3 were 34 cycles of 94°C for 45 s, 62°C for 30 s and72°C for 45 s. Primers used were EDNRB ∆3, 5′-CGA GCAAAC GGT GGA GGC TAC A-3′ (sense) and EDNRB ∆3,5′-CGG CTG CAT GCT GCT ACC TG-3′ (antisense). PCRproducts were resolved on a 2% agarose gel and transferredonto a Zeta-Probe membrane (Bio-Rad Laboratories) via alka-line transfer. Membranes were probed using a 5′ end labeledoligonucleotide with sequence 5′-CTC TGA AAC TGC GGA
GCG GCC AC-3′ and exposed to autoradiographic BioMaxMR film (Eastman Kodak).
Methylation analysis
Methylation status in the 5′ region of the EDNRB gene wasdetermined by Ms-SNuPE (43). This technique depends onchemical modification of DNA using sodium bisulfite, whereunmethylated cytosine residues are converted to uracil residuesbut methylated cytosines remain unchanged. DNA sampleswere bisulfite-converted as described previously (44).
Ms-SNuPE was performed as described previously (43). Inorder to analyze a representative number of CpG dinucleotidesspanning the 5′ region of the EDNRB gene, four separatebisulfite-PCR products were generated. EDNRB segment 0allowed for analysis of CpG dinucleotides at positions –130 and–8. EDNRB segment 1 allowed for analysis of CpG dinucleotidesat positions 165, 291 and 336. EDNRB segment 2 allowed foranalysis of CpG sites at positions 336, 377, 431, 618 and 654.EDNRB segment 3 allowed for analysis of CpG 994 and 1115.Numbering is relative to GenBank accession no. D13162. PCRand Ms-SNuPE conditions have been described previously(40).
Ms-SNuPE primers were as follows:CpG-130: 5′-GAA GTA GAG TAA AGA GTA G-3′;CpG-8: 5′-GTT TTG TTT TAG TTT GGA GTT GT-3′;CpG 165: 5′-AGG GGA AAG GTT GTA G-3′;CpG 291: 5′-AGG TTA TAT TGT TTG GTA TTT T-3′;CpG 336: 5′-TTT GTA GTT TAA GGG AGG-3′;CpG 377: 5′-TGG AAT TTA GTT GGG TTT-3′;CpG 431: 5′-TTG TAT TTG GTT TGT TAG ATT-3′;CpG 618: 5′-GTT TGG AGG GAA TAG-3′;CpG 654: 5′-GTT GAT TTG AGA AGT TTT TG-3′;CpG 994: 5′-TGT ATA TTA TTT ATT TTT TTT GGT TA-3′;CpG 1115: 5′-TAA ATT TGA GTT ATT TTT GAG-3′.
ACKNOWLEDGEMENTS
We thank Daniel Weisenberger for helpful discussions. Thiswork was supported by grant USPHS R35 CA 49758 from theNational Cancer Institute.
REFERENCES
1. Bird, A.P. (1986) CpG-rich islands and the function of DNA methylation.Nature, 321, 209–213.
2. Bird, A.P. and Wolffe, A.P. (1999) Methylation-induced repression—belts, braces, and chromatin. Cell, 99, 451–454.
3. Keshet, I., Lieman-Hurwitz, J. and Cedar, H. (1986) DNA methylationaffects the formation of active chromatin. Cell, 44, 535–543.
910 Human Molecular Genetics, 2001, Vol. 10, No. 9
4. Li, E., Bestor, T.H. and Jaenisch, R. (1992) Targeted mutation of the DNAmethyltransferase gene results in embryonic lethality. Cell, 69, 915–926.
5. Bird, A. (1992) The essentials of DNA methylation. Cell, 70, 5–8.6. Cooper, D.N. and Krawczak, M. (1989) Cytosine methylation and the fate
of CpG dinucleotides in vertebrate genomes. Hum. Genet., 83, 181–188.7. Baylin, S.B. and Herman, J.G. (2000) DNA hypermethylation in tumori-
genesis: epigenetics joins genetics. Trends Genet., 16, 168–174.8. Jones, P.A. and Laird, P.W. (1999) Cancer epigenetics comes of age.
Nature Genet., 21, 163–167.9. Merlo, A., Herman, J.G., Mao, L., Lee, D.J., Gabrielson, E., Burger, P.C.,
Baylin, S.B. and Sidransky, D. (1995) 5′ CpG island methylation is asso-ciated with transcriptional silencing of the tumour suppressor p16/CDKN2/MTS1 in human cancers. Nat. Med., 1, 686–692.
10. Gonzalez-Zulueta, M., Bender, C.M., Yang, A.S., Nguyen, T., Beart,R.W., Van Tornout, J.M. and Jones, P.A. (1995) Methylation of the 5′CpG island of the p16/CDKN2 tumor suppressor gene in normal andtransformed human tissues correlates with gene silencing. Cancer Res.,55, 4531–4535.
11. Ohtani-Fujita, N., Fujita, T., Aoike, A., Osifchin, N.E., Robbins, P.D. andSakai, T. (1993) CpG methylation inactivates the promoter activity of thehuman retinoblastoma tumor-suppressor gene. Oncogene, 8, 1063–1067.
12. Stirzaker, C., Millar, D.S., Paul, C.L., Warnecke, P.M., Harrison, J.,Vincent, P.C., Frommer, M. and Clark, S.J. (1997) Extensive DNA meth-ylation spanning the Rb promoter in retinoblastoma tumors. Cancer Res.,57, 2229–2237.
13. Herman, J.G., Jen, J., Merlo, A. and Baylin, S.B. (1996) Hypermethyla-tion-associated inactivation indicates a tumor suppressor role forp15INK4B. Cancer Res., 56, 722–727.
14. Veigl, M.L., Kasturi, L., Olechnowicz, J., Ma, A.H., Lutterbaugh, J.D.,Periyasamy, S., Li, G.M., Drummond, J., Modrich, P.L., Sedwick, W.D.et al. (1998) Biallelic inactivation of hMLH1 by epigenetic gene silenc-ing, a novel mechanism causing human MSI cancers. Proc. Natl Acad.Sci. USA, 95, 8698–8702.
15. Deng, G., Chen, A., Hong, J., Chae, H.S. and Kim, Y.S. (1999) Methyla-tion of CpG in a small region of the hMLH1 promoter invariably correlateswith the absence of gene expression. Cancer Res., 59, 2029–2033.
16. Herman, J.G., Umar, A., Polyak, K., Graff, J.R., Ahuja, N., Issa, J.P.,Markowitz, S., Willson, J.K., Hamilton, S.R., Kinzler, K.W. et al. (1998)Incidence and functional consequences of hMLH1 promoter hypermethyl-ation in colorectal carcinoma. Proc. Natl Acad. Sci. USA, 95, 6870–6875.
17. Kane, M.F., Loda, M., Gaida, G.M., Lipman, J., Mishra, R., Goldman, H.,Jessup, J.M. and Kolodner, R. (1997) Methylation of the hMLH1promoter correlates with lack of expression of hMLH1 in sporadic colontumors and mismatch repair-defective human tumor cell lines. CancerRes., 57, 808–811.
18. Issa, J.P., Ottaviano, Y.L., Celano, P., Hamilton, S.R., Davidson, N.E. andBaylin, S.B. (1994) Methylation of the oestrogen receptor CpG islandlinks ageing and neoplasia in human colon. Nature Genet., 7, 536–540.
19. Liang, G., Salem, C.E., Yu, M.C., Nguyen, H.D., Gonzales, F.A., Nguyen,T.T., Nichols, P.W. and Jones, P.A. (1998) DNA methylation differencesassociated with tumor tissues identified by genome scanning analysis.Genomics, 53, 260–268.
20. Tsutsumi, M., Liang, G. and Jones, P.A. (1999) Novel endothelin Breceptor transcripts with the potential of generating a new receptor. Gene,228, 43–49.
21. Nelson, J.B., Lee, W.H., Nguyen, S.H., Jarrard, D.F., Brooks, J.D.,Magnuson, S.R., Opgenorth, T.J., Nelson, W.G. and Bova, G.S. (1997)Methylation of the 5′ CpG island of the endothelin B receptor gene is com-mon in human prostate cancer. Cancer Res., 57, 35–37.
22. Puffenberger, E.G., Hosoda, K., Washington, S.S., Nakao, K., deWit, D.,Yanagisawa, M. and Chakravart, A. (1994) A missense mutation of theendothelin-B receptor gene in multigenic Hirschsprung’s disease. Cell,79, 1257–1266.
23. Jackson, I.J. (1997) Homologous pigmentation mutations in human,mouse and other model organisms. Hum. Mol. Genet., 6, 1613–1624.
24. Shin, M.K., Levorse, J.M., Ingram, R.S. and Tilghman, S.M. (1999) Thetemporal requirement for endothelin receptor-B signalling during neuralcrest development. Nature, 402, 496–501.
25. Ford, H.L. (1998) Homeobox genes: a link between development, cellcycle, and cancer? Cell Biol. Int., 22, 397–400.
26. Kass, S.U., Goddard, J.P. and Adams, R.L. (1993) Inactive chromatinspreads from a focus of methylation. Mol. Cell. Biol., 13, 7372–7379.
27. Hsieh, C.L. (1997) Stability of patch methylation and its impact in regionsof transcriptional initiation and elongation. Mol. Cell Biol., 17, 5897–5904.
28. Antequera, F., Boyes, J. and Bird, A. (1990) High levels of de novomethylation and altered chromatin structure at CpG islands in cell lines.Cell, 62, 503–514.
29. Ahuja, N., Li, Q., Mohan, A.L., Baylin, S.B. and Issa, J.P. (1998) Agingand DNA methylation in colorectal mucosa and cancer. Cancer Res., 58,5489–5494.
30. Macleod, D., Charlton, J., Mullins, J. and Bird, A.P. (1994) Sp1 sites inthe mouse aprt gene promoter are required to prevent methylation of theCpG island. Genes Dev., 8, 2282–2292.
31. Brandeis, M., Frank, D., Keshet, I., Siegfried, Z., Mendelsohn, M.,Nemes, A., Temper, V., Razin, A. and Cedar, H. (1994) Sp1 elementsprotect a CpG island from de novo methylation. Nature, 371, 435–438.
32. Melki, J.R., Vincent, P.C. and Clark, S.J. (1999) Cancer-specific region ofhypermethylation identified within the HIC1 putative tumour suppressorgene in acute myeloid leukaemia. Leukemia, 13, 877–883.
33. Millar, D.S., Paul, C.L., Molloy, P.L. and Clark, S.J. (2000) A distinctsequence (ATAAA)n separates methylated and unmethylated domains atthe 5′-end of the GSTP1 CpG island. J. Biol. Chem., 275, 24893–24899.
34. Bateman, E. and Paule, M.R. (1988) Promoter occlusion during ribosomalRNA transcription. Cell, 54, 985–992.
35. Wu, J., Grindlay, G.J., Bushel, P., Mendelsohn, L. and Allan, M. (1990)Negative regulation of the human epsilon-globin gene by transcriptionalinterference: role of an Alu repetitive element. Mol. Cell. Biol., 10, 1209–1216.
36. Proudfoot, N.J. (1986) Transcriptional interference and terminationbetween duplicated alpha-globin gene constructs suggests a novel mecha-nism for gene regulation. Nature, 322, 562–565.
37. Bender, C.M., Gonzalgo, M.L., Gonzales, F.A., Nguyen, C.T., Robertson,K.D. and Jones, P.A. (1999) Roles of cell division and gene transcriptionin the methylation of CpG islands. Mol. Cell. Biol., 19, 6690–6698.
38. Salem, C.E., Markl, I.D., Bender, C.M., Gonzales, F.A., Jones, P.A. andLiang, G. (2000) PAX6 methylation and ectopic expression in humantumor cells. Int. J. Cancer., 87, 179–185.
39. Nguyen, T.T., Nguyen, C.T., Gonzales, F.A., Nichols, P.W., Yu, M.C. andJones, P.A. (2000) Analysis of cyclin-dependent kinase inhibitor expres-sion and methylation patterns in human prostate cancers. Prostate, 43,233–242.
40. Pao, M.M., Liang, G., Tsai, Y.C., Xiong, Z., Laird, P.W. and Jones, P.A.(2000) DNA methylator and mismatch repair phenotypes are not mutuallyexclusive in colorectal cancer cell lines. Oncogene, 19, 943–952.
41. Ogawa, Y., Nakao, K., Arai, H., Nakagawa, O., Hosoda, K., Suga, S.,Nakanishi, S. and Imura, H. (1991) Molecular cloning of a non-isopeptide-selective human endothelin receptor. Biochem. Biophys. Res. Commun.,178, 248–255.
42. Arai, H., Nakao, K., Takaya, K., Hosoda, K., Ogawa, Y., Nakanishi, S.and Imura, H. (1993) The human endothelin-B receptor gene. Structuralorganization and chromosomal assignment. J. Biol. Chem., 268, 3463–3470.
43. Gonzalgo, M.L. and Jones, P.A. (1997) Rapid quantitation of methylationdifferences at specific sites using methylation-sensitive single nucleotideprimer extension (Ms-SNuPE). Nucleic Acids Res., 25, 2529–2531.
44. Frommer, M., McDonald, L.E., Millar, D.S., Collis, C.M., Watt, F., Grigg,G.W., Molloy, P.L. and Paul, C.L. (1992) A genomic sequencing protocolthat yields a positive display of 5-methylcytosine residues in individualDNA strands. Proc. Natl Acad. Sci. USA, 89, 1827–1831.
