[CANCER RESEARCH 55, 4531—4535, October 15, 1995J

Advances in Brief

Methylation of the 5' CpG Island of the p16/CDKN2 Tumor Suppressor Gene in

Normal and Transformed Human Tissues Correlates with Gene Silencing'

Mirella Gonzalez-Zulueta, Christina M. Bender, Allen S. Yang, TuDung Nguyen, Robert W. Beart,Jan M. Van Tornout, and Peter A. Jones2

Urological Research Laboratory, USC/Norris Comprehensive Cancer Center (M. G-Z, C. M. B., A. S. 1'., T. N., J. M. V., P. A. fl, and Department of Colorectal Surgery[R. W. B.J, University ofSouthern California, School ofMedicine, Las Angeles, Cal(fornia 90033

Abstract

Loss of heterozygosity on 9p21, where thepl6/CDKN2 tumor suppressor and the pJ5INK4Bcell cycle regulator genes are located, is a commongenetic alteration in bladder cancer. However, it has been difficult todemonstrate homozygous deletions and intragenic mutations in either ofthese two genes in primary transitional cell carcinomas (TCC) of thebladder. Similarly, colon cancer-derived cell lines have shown no homozygous deletions of the p16/CDKN2 locus in contrast to a wide variety oftumor-derived cell lines. We have investigated abnormal methylation ofthe5' CpGislandsofthepl6/CDKN2andp15@4'@genesasanalternativemechanism of inactivation of these genes in bladder and colon cancers. Denovo methylation of the 5' CpG island ofpl6/CDKN2 was observed in 12of 18 (67%) uncultured bladder TCCS and in 2 of 3 (67%) bladder celllines. In contrast, only 1 of 10 (10%) colon cardnomas showed methyladon of the 5' CpG lsland ofpl6/CDKN2. It was striking to fInd that thisregion was extensively methylated and the gene not expressed in thenormal colonic mucosa of 6 of 10 (60%) patients with colon cancer,whereas 5 of the corresponding colon tumors showed no methylation andhigh levels ofpl6/CDKN2 expression. Our data show a significant correlatlon (P = 0.00001, two-sided) between the absence of p16/CDKN2 expresslon and methylation ofits 5' CpG island in bladder tumors, cell lines,and normal colon mucosa. In contrast, no association was observed between expression and methylation status ofthe 5' CpG lsland 0f@J5@4B•Our resultssuggestthat thep16/CDKN2tumor suppressorgenemay beinactivated by methylation of its 5' CpG lsland in TCCs of the bladder.We also present evidence of methylation of the 5' CpG island in thisautosomal gene in normal colonic tissue.

Introduction

The p16 protein inhibits CDK43 and CDK6, which are key regulators of the progression of eukaryotic cells through the G@phase ofthe cell cycle (1). The p16/CDKN2 gene resides on chromosome band9p2l, a region frequently altered in diverse tumor types (2, 3). Thehigh frequency of p16/CDKN2 alterations initially reported in tumorderived cell lines (4, 5) and in some tumor types (6, 7), along with therole ofpi6/CDKN2 in cell cycle control, made this gene an excellentcandidate for a tumor suppressor. Emerging evidence suggests thatpi6/CDKN2 alterations are involved in the progression of certaintumor types (8, 9). Another related cell cycle regulator, the pJ5INK4Bgene, is also located on 9p21, and it has been shown to be upregulated in human keratinocytes following treatment with TGF-@3,suggesting that it may be an effector of TGF-@3-mediated cell cycle

Received 6/20/95; accepted 8/30/95.The costs of publication of this article were defrayed in part by the payment of page

charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

I Supported by USPHS Grant R35CA49758 from the National Cancer Institute.2 To whom requests requests for reprints should be addressed, at USC/Norris Cancer

Center, 1441 Eastlake Avenue, Los Angeles, CA 90033. Phone: (213) 764—0816; Fax:(213) 764—0102.

3 The abbreviations used are: CDK, cyclin-dependent kinase; TGF, transforming

growthfactor;TCC, transitionalcell carcinoma;RT, reversetranscription;GAPDH,glyceraldehyde 3-phosphate dehydrogenase.

arrest (10). Although most of the p16/CDKN2 homozygous deletions

also include the @J5INK4Blocus, no intragenic mutations in @j5@1(4Bhave been reported to date.

Hemizygous deletions of 9p2l are a frequent genetic alteration inTCC of the bladder (1 1, 12), which suggests that a tumor suppressorgene for bladder cancer may reside in this area. However, pitS!

CDKN2 homozygous deletions and intragenic mutations occur at lowfrequencies in uncultured TCCs (13, 14), although they are commonin uncultured squamous cell carcinomas of the bladder (15). Nohomozygous deletions of the p16!CDKN2 locus have been describedin colon cancer-derived lines in contrast to a wide variety of tumorderived cell lines (4). This relative low rate and absence of p16!CDKN2 alterations in bladder and colon tumors, respectively, mayreflect that p16!CDKN2 inactivation occurs in these tumor types by

alternative mechanisms. Transcriptional repression by DNA methylation of promoter and 5 â€r̃egulatory sequences may be a pathway toinactivate the p16!CDKN2 and @j5INK4Bgenes. Global changes inDNA methylation patterns are known to occur during tumorigenesis,and gene silencing has been associated with methylation of CpG

islands located in, or near, promoters and 5' regulatory regions (16,17). CpG islands are G+C rich regions that show a higher frequency

of CpG dinucleotides than is normally seen in the vertebrate genomeand that are not methylated in the germline (18). With the exception

of islands in genes on the inactive X chromosome (19), Alu and Lisequences (20), and some imprinted genes (21), CpG islands are

usually unmethylated in normal somatic cells. In contrast, widespreadmethylation of CpG islands occurs on autosomal genes during oncogenic transformation (22, 23). Promoters silenced by methylation canbe reactivated in many cases by treatment with the drug 5-aza-2'-deoxycytidine, which is a well-established inhibitor of DNAmethylation (24).

The exon 1 coding sequences of the p16!CDKN2 and @15INK4Bgenes reside within 5 â€C̃pG islands, and their exon 2 regions are 83%homologous and also constitute CpG islands. We hypothesized thatabnormal DNA methylation might be an alternative mechanism ofinactivation of the p16!CDKN2 and @J5INK4Bgenes in bladder andcolon tumors. In this study, we examined the methylation status andexpression levels of the p16!CDKN2 tumor suppressor gene and thepJ5INK4B cell cycle regulator in normal tissues, cell lines, and bladder

and colon cancer. Samples from normal tissues included sperm

(n 1), colon epithelium from individuals without colon cancer(n = 4), bladder urothelium (n = 1), kidney (n 1), and WBC(n 1). We also obtained 18 bladder transitional cell carcinomaspecimens, as well as normal colonic mucosa (n = 10) and corresponding tumor specimens (n = 10) from patients with colon cancer,and one ulcerative colitis specimen. Our data suggest that expressionof the pi6!CDKN2 tumor suppressor gene, but not expression of the

@15INK4B cell cycle regulator, is controlled by methylation of its 5'

CpG island, and that de novo methylation of this island is a mechanism for p16!CDKN2 inactivation in bladder TCCs. In contrast, the

4531

on July 28, 2018. © 1995 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

pItS/CDKN2 METHYLATION IN BLADDER AND COLON CANCERS

p16!CDKN2 5' CpG island is sometimes methylated and the genesilenced in normal colonic mucosa.

Materials and Methods

Tissue Samples. Sperm DNA (n = 1) was isolated from a donated sample.

Peripheral blood lymphocytes (n = 1) and normal kidney (n = 1) were

obtained from a patient with bladder cancer. Normal colon mucosa (n = 4) wasobtained from individuals with normal sigmoidoscopy, who had undergone

coleostomy for a noncancerous process. One ulcerative colitis specimen was

obtained from a patient affected with ulcerative colitis. Colon carcinomaspecimens (n = 10) and matched normal colonic mucosa (n = 10) were

obtained from 10 patients with colon cancer. Colon tumors were gradedaccording to Dukes' classification (25) and ranged from stage B1 to C@. The

collection process of the normal colon mucosas prevented major contaminationof epithelial cells with cells from the rest of the colon wall since the epithelialcell layer was dissected from submucosa, muscle, and surrounding adipose

tissue. Bladder TCCs (n = 18) were obtained from 18 patients, and their gradesranged from Ta to T4, according to the Bergkvist classification (26). DNA wasisolated by proteinase K digestion and phenol/chloroform extraction.

Cell Lines. El, HT1376 and J82 (bladderTCC-derivedcell lines) wereobtained from the American Type Culture Collection (Rockville, MD). DNA

and total RNA were isolated from nonconfluent, exponentially growing cells.RT-PCR. Total RNA was isolated as described (27). RNA was reverse

transcribed using random hexamers. cDNA was amplified using primers specific for the pJOICDKN2gene, the pJ5INK4Bgene, or the GAPDH gene, whichwas used as control. The primers used for p16/CDKN2 amplifications werelocated in exon 1 and exon 2 of the gene, flanking intron 1. Their sequences

were 5'-AGC CVF COG CTG ACT GGC TGG-3' (sense) and 5'-CFG CCC

ATC ATC ATG ACC TGG A-3' (antisense). Conditions for p16/CDKN2amplifications were 94°C for 3 mm, 22 cycles of 94°C for 1 mm, 56°C for 30

s, and 72°C for 40 s, followed by incubation at 72°C for 1 mm. Primers for

@J5tNK4Bwere 5'-TGA TGA TGG GCA GCG CCC GC-3' (sense) and

5'-CGC Cr0 GGG AAC CTG GCG TCA-3' (antisense), which were locatedwithin exon 2; the antisense primer was complementary to @J51NK4Bexon 2 in

the region where pI5@'@411 and p16/CDKN2 diverge. Conditions for pJ5INK4B

amplifications were 95°Cfor 3 mm, 22 cycles of 95°Cfor 1 mm 40 s, 60°Cfor 45 s, and 72°Cfor 40 s, followed by incubation at 72°Cfor 2 mm. Primersfor the GAPDH gene were 5'-CAG CCG AGC CAC ATC 0-3' (sense) and5'-TGA GGC TOT TOT CAT ACF TCT C-3' (antisense). Conditions for

GAPDH amplifications were 94°Cfor 1 mm, 22 cycles of 94°Cfor 1 mm,58°C for 30 s, 72°C for 45 s, followed by incubation at 72°C for 1 mm. Each

PCR was performed with an amount of eDNA equivalent to 100 ng of RNA,and PCRs for the p16/CDKN2 and pJ5INK4B genes contained 10% DMSO.Under the above conditions, all amplifications were in the linear range of theassay (data not shown). PCR products were resolved on 2% agarose gels. DNA

was transferred to a nylon membrane (Zetaprobe; Bio-Rad, Richmond, CA) via

alkali transfer. p16/C'DKN2 blots were hybridized to radiolabeled p16/CDKN2

eDNA. @J5INK4Hblots were hybridized to a radiolabeled or digoxigeninlabeled internal oligonucleotide. GAPDH blots were hybridized to radiolabeled

GAPDH cDNA. Quantitation of PCR products was performed by scanningautoradiographs with an LKB UltraScan XL Laser densitometer. The ratiobetween p16/CDKN2 and GAPDH and @J51NK4Band GAPDH signals was

obtained for each sample. All reactions were done at least twice, and all werecontrolled with the omission of reverse transcriptase.

PCR-based Methylation Assay. A PCR assay relying on the inability ofsome restriction enzymes to cut methylated sequences (28) was used to analyze

the methylation status of the first and second exons of the p16/CDKN2 and@J5tNK4Bgenes. The sites examined were: one FnuDII, one SaclI, one HpaII,

and two CfoI sites in exon 1 ofpi6/CDKN2; and two NaeI sites in exon 1 ofp151NK4H, Due to the high sequence similarity, identical restriction sites could

be examined in exon 2 of p16/CDKN2 and @J5INK4B which included one

SnwI, four HpaII, and six CfoI sites. DNA digests were performed accordingto the manufacturer's directions (Boehringer Mannheim, Indianapolis, IN).

DNA (1 1.Lg)was digested for 2 h, with 10 units of enzyme/@xg of DNA. Fiftyng of the digested DNA were amplified with primers flanking the restriction

sites. The primer set used for methylation analysis ofpltS/C'DKN2 exon 1 was

5'-AGC Crc CGG CTG ACT GGC TGG-3' (sense) and 5'-CTG GAT COG

CCT CCG ACC OTA-3' (antisense), under the following conditions: 94°Cfor

3 mm,21cyclesof94°Cfor 1mm,55°Cfor30s,and72°Cfor40s,followedby incubation at 72°C for 1 mm. Primers used for p16/CDKN2 exon 2 were

5'-CFG CU GGC GOT GAG 000 0-3' (sense) and 5'-CCF CAC CTO

AGO GAC err C-3' (antisense); conditions were the same as for exon 1,except for an annealing temperature of 57°C.The primer set used for methylation analysis of pJ5INK4Bexon 1 was 5'-AGC TOA GCC CAG GTC TCCTA-3' (sense) and 5'-CGC CTC CCG AAA COG iTO AC-3' (antisense).Conditions were: 95°Cfor 3 mm, 21 cycles of 94°Cfor 1 mm 40 s, 58°Cfor45 5, and 72°Cfor 45 s, followed by incubation at 72°Cfor 2 mm. Primers for

@J5tNK4Bexon 2 were 5'-TCT TTA AAT GGC TCC ACC T-3' (sense) and

5'-CTC CCC OTT GGC AGC Cli.' C-3' (antisense); conditions were the sameas for @J51NK4Bexon 1 amplifications. Under the indicated PCR conditions,

amplifications with each primer set were in the linear range of the assay as

determined by cycle curves and DNA concentration curves performed to

establish the optimal conditions of the assay (data not shown). PCR productswere resolved on 2% agarose gels, transferred to a nylon membrane, andhybridized as described for the RT-PCR products.

PCR-based methylation an@(ysis using restriction enzymes may be subjectto variability if the DNA digestions are not complete and if amplifications are

performed for an excessive number of cycles. To rule out the possibility of

incomplete restriction, all samples were digested twice with each of the

enzymes in independent experiments. PCR amplifications from each of theduplicate digests were repeated at least twice to ensure reproducibility of theresults. We also performed PCR with various sets of primers flanking the samerestriction sites in other regions to assess complete digestion. To prevent

overcycling, the cycle number was determined for each of the primer sets byperforming a cycle curve using undigested template and template digested with

MspI restriction enzyme, which is methylation insensitive. The optimal cycle

number was determined to be that which produced a detectable band ofminimal intensity after hybridization of PCR products from the undigested

template and that did not produce any signal from the MspI-digested DNA. An

undigested DNA control and a MspI-digested DNA control were included forevery site examined.

Results

Methylation and Expression of the p16/CDKN2 and @J5INK4BGenes in Primary Bladder TCCs and Cell Lines. The p16!CDKN2

and @J5INK4Bgenes both contain CpG islands within their first andsecond exons based on the criteria of Gardiner-Garden and Frommer

(29),sincetheyhavea G+Ccontentover50%,andanobserved!expected CpG dinucleotide ratio greater than 0.6. The methylation

status of several CpG sites within the first and second exons of

p16!CDKN2 and @J5INK4Bwas examined in normal bladder mucosa,18 uncultured TCCs of the bladder, and 3 bladder carcinoma-derived

cell lines. The sites were analyzed by a PCR-based assay (28) using

methylation-sensitive endonucleases for DNA cleavage, followed by

PCR amplification of the digested DNAS (Fig. 1). The normal urothehum showed no methylation in either exon of either gene, because noPCR product was generated following digestion with the corresponding enzyme (Fig. 1). In contrast, 2 of 3 (67%) cell lines and 12 of 18

(67%) uncultured TCCs showed de novo methylation of exon 1 of

p16!CDKN2, since PCR products were generated after cutting withthe methylation-sensitive enzymes FnuDII and Sad! (Fig. 1). Fig. 1

also shows that while no methylation was observed in exon 2 ofp16!CDKN2 in the normal bladder mucosa, extensive methylation ofthis CpG island occurred in the bladder tumors. Exon 2 of pi6!

CDKN2 was found to be de novo methylated in 15 of 18 (83%)uncultured TCCs and in 3 of 3 cell lines (Table 1).

RNA for expression analysis was available from normal urothehum, three of the uncultured tumors, and three bladder cell lines. Agood correlation between methylation of exon 1 of p16!CDKN2 andtranscriptional inactivity of the gene was found as one of the threetumors (Fig. 1), and two of three cell lines showed methylation of

exon 1 with no expression ofpi6!CDKN2, whereas the normal urothehum (Fig. 1), two of three tumors, and one of three cell lines showed

4532

on July 28, 2018. © 1995 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Table I De novo methylation ofpl6/CDKN2 in bladder and coloncancerTumor

type Methylation of p16 exon 1― Methylation of p16 exon2Bladder

TCCS 12/18(67%)15/18(83%)Bladder

cell lines 2/3 (67%) 3/3(100%)Colon

carcinomas 1/10 (10%)―7/10(70%)aNumberofspecimensshowingdenovomethylationofatleasttheFnuDIIandSacIIsites/number of specimens analyzed.

b The corresponding normal colonic mucosa of this tumor showed an identical

methylationpattem.

FH H H IIs' cc ccc SH1CCC I

p16 @Jill Ii ii 14—

p15@ I 11 1 11 t 4-@ PCR-based Methylation Analysis ExpressionNH

.3 4 1 Exonl Exon2

UF sHCM USHCM

NORMAL @16 @@jlll?iIlll Fà 1 i. I I@@I

MUCOSA pls @,@lll 91111I@ j , I I@ IBLADDER U N M ___________

p16_@L@lll!!lll 1::::: I F@:@s I@ I

BLADDER

TUMOR @151•!@i' ii ¶!@@-i@ I@ I [@ I

NORMAL@ @Ill¶lttl L F— --@--@ I I@ I T@l@ ___

COLON _____ ___________MUCOSA @15 @@@llt!lltl I@@ IS I

COLON@@ @,, F. I .@ @-..@@ . i __

TUMOR p15@ ¶!!@@ 1 1- - — — I [

pIWCDKN2 METHYLATION IN BLADDER AND COLON CANCERS

Fig. 1. PCR methylation analysis and RT-PCRexpression analysis of the p16ICDKN2 andpJSINK4B genes in normal and tumor tissues. Re

suits for some of the specimens examined areshown. Schematic maps of exon 1 (first solid box)and exon 2 (second solid box) of the p16/CDKN2and @j5I@4Bgenes are presented in the boxedportion; the sites analyzed and the primer sets used(arrows) are indicated. DNA was digested with

either FnuDII (F), SacII (s), HpalI (H), CfoI (C),Smal (5), NaeI (N), or MspI (M). One hundred ngof digested or undigested (U) DNA was amplifiedwith the primer sets indicated by the arrows in themaps. •,methylated sites; partially solid circles,partial methylation; 0, unmethylated sites. The absence of a circle on a site indicates that, although itwas examined, its precise methylation status couldnotbedetermineddueto thepresenceof morethanone restriction site for the same enzyme in theregion flanked by the primer sets used in the assay.The GAPDHexpressioncontrolsin the RT-PCRexpression analysis were not included in the figurefor simplicity.

Methylation of exon 1 of p16!CDKN2 was observed in the normalmucosa of 6 of the 10 colon cancer patients examined, as well as in1 of the 10 colon tumors (Fig. 2). Fig. 2 shows that in 7 of 10 cases,

p16!CDKN2 expression was undetectable in the normal colonic epithelium of patients with colon cancer, and only 3 patients (F, I, and J)showed expression of the gene in their normal mucosa. All sampleswere found to contain a wild-type p16!CDKN2 gene by single-strandconformation polymorphism analysis (data not shown), and the tumors showed no up-regulation of CDK4 expression (data not shown).Furthermore, the levels of p16!CDKN2 expression were markedlyincreased in the tumors of 50% (5 of 10) of the patients (A to E) whencompared to the normal mucosa (Fig. 2). In the remaining five tumors,p16!CDKN2 expression varied from undetectable (tumors G and H) tolow levels (tumors F, I, and J). In cases F, I, and J, in which there wasno detectable methylation in the normal mucosa and both normal andtumor tissue expressed p16!CDKN2, it is possible that the low expression levels detected in the tumor were due to contamination withnormal cells. Importantly, in no case was exon 1 methylated and thep16!CDKN2 gene expressed. De novo methylation of the CpG islandin exon 2 ofpi6!CDKN2 was observed in 7 of 10 (70%) colon tumors(Table 1). Interestingly, an inverse pattern of expression of pitS!CDKN2 and @J5INK4Bwas also observed in cases A, C, D, and E, i.e.,p16!CDKN2 expression was undetectable in the normal mucosa butsignificantly up-regulated in the corresponding tumor, whereas

@15INK4B expression was higher in the normal tissue when compared

to the tumor. In contrast to p16!CDKN2, no clear correlation wasobserved between methylation of exon 1 of @J5INK4Band expressionof the gene. In addition, @J5INK4BmRNA was detected in most (9 of10) of the normal colonic mucosas and, in general, decreases in thelevel of expression were observed in the tumors.

The methylation status and expression levels of the p16!CDKN2and @J5INK4Bgenes were also examined in several normal tissuesincluding sperm, bladder urothelium, kidney, WBC, and colonic epithelium from patients without colorectal cancer (results not shown).Partial methylation of the two NaeI sites within the @J51N@@@4Bexon 1

4533

no methylation of exon 1 and high levels of p16!CDKN2 mRNA. Incontrast, no correlation was found between the methylation status of

exon 2 and expression levels of p16!CDKN2 or between expressionand methylation of either exon of the pJ5INK4Bgene in unculturedTCCS and bladder cell lines.

Methylation and Expression of the p16/CDKN2 and @j5@4BGenes in Colon Carcinomas and Adjacent Normal Mucosa. Fig. 1shows a representative example of the 10 colon cancer cases exammcd, from which normal colonic epithelium and the correspondingcolon tumor were available. In contrast to the lack of methylation seenin the normal bladder mucosa, all five sites examined in exon 1 ofpi6!CDKN2 were fully methylated in this normal colonic mucosapresented in Fig. 1. In contrast, no PCR product was obtained from thecolon tumor DNA of the same individual following enzyme digestionwith any of the four methylation-sensitive enzymes used or the controlendonuclease MspI (Fig. 1). Thus, extensive methylation in exon 1 ofpJ6JCDKN2 was present in the normal colonic epithelium and wasassociated with lack of expression of p16!CDKN2, while the corresponding tumor showed no methylation of this region and expressedthe gene (Fig. 1). In contrast, the CpG island in exon 2 of p16!CDKN2underwent extensive de novo methylation in this tumor. While exon 1of pJ5INK4B was methylated at both NaeI sites in normal and tumortissue, exon 2 of @J5INK4Balso underwent extensive de novo methylation in the tumor.

on July 28, 2018. © 1995 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Table 2 Methylation of exons I and 2 and expression of the p16./CDKN2geneExon

1Exon2Unmethylated

MethylatedUnmethylatedMethylatedpitS

expressed17 0712p16

not expressed2 11―3 8!)

‘p ‘1

COLON CANCER CASES. Normalmucosa

PATIENT@ Colon tumor

pIó/CDKN2 METHYLATION IN BLADDER AND COLON CANCERS

was seen in sperm. Normal bladder epithelium, kidney, and WBCwere not methylated at the sites examined in either exon of eithergene. However, the normal colonic epithelium from the individualswithout colon cancer showed methylation of exon 1 ofpl6!CDKN2 (4of 4; data not shown), which correlated with the lack of p16!CDKN2expression in this tissue. On the other hand, @15INK4Bwas not methylated in either exon in this normal colon mucosa. The results showingmethylation of exon 1 ofpl6!CDKN2 in the normal colon mucosas ofindividuals with normal sigmoidoscopy were similar to those obtainedin one case of ulcerative colitis (data not shown) as well as in thenormal colonic epithelium of 6 of the 10 patients with colon cancerexamined (Fig. 2), where exon 1 of p16!CDKN2 was extensivelymethylated (example shown in Fig. 1; summary of results in Fig. 2).

Methylation of Exon 1, but not of Exon 2, of p16/CDKN2 IsAssociated with Transcriptional Silencing. Table 2 is a compilationof the analyses of mRNA expression and methylation status of exons1 and 2 of the p16!CDKN2 gene conducted in a total of 30 specimensfrom which both RNA and DNA was available, including 4 normaltissues, 10 colon cancer cases (normal and tumor from each case), 3bladder tumors, and 3 bladder cell lines. Overall, exon 1 of pl6!CDKN2 was methylated in 11 of 30 (37%) specimens, and remarkably, no case revealed expression of the gene with methylation of thisexon. On the other hand, exon 1 was not methylated in 17 of 30(57%)

specimens that expressed pi6!CDKN2, while the remaining 2 specimens did not show methylation of exon 1 nor expression of the gene.The correlation between methylation of both the FnuDII and SacIIsites in exon 1 and transcriptional silencing of pi6!CDKN2 wasstatistically significant (P = 0.00001). In contrast, there was not asignificant association between methylation of exon 2 and transcriptional repression of p16!CDKN2 (P = 0.340), since 12 of 30 (40%)specimens showed extensive methylation of exon 2 with p16!CDKN2expression (Table 2).

METHYLATION OF p16 EXON 1

FSH CCI I I 11

@ 8@ 8

!@!@!!

888 8888 8::: 8888 8888 8

apo.0000i,two-sided.b@ 0.340, two-sided.

Discussion

Our results show a highly significant correlation (P = 0.00001,two-sided) between methylation of exon 1 of p16!CDKN2 andtranscriptional silencing of the gene, since none of the normal andtumor specimens that showed methylation of this exon (1 1 of 30)expressed pi6!CDKN2, and 17 of 30 specimens expressed p16!CDKN2 without evident methylation of exon 1 (Table 2). These dataagree with results reported recently (30, 31) and suggest that methylation of the 5' CpG island of pi6!CDKN2 might be a mechanism tocontrol expression of this tumor suppressor gene. It remains to be seenwhether the methylation is causally related to the silencing of expression; however, there are numerous examples in the literature wheremethylation of 5' CpG islands strongly suppresses gene activity (16,17, 22, 23). Additional evidence that methylation of the 5' CpG island

is involved in p16!CDKN2 silencing comes from experiments showing the induction of p16!CDKN2 expression in cell lines after treatment with the inhibitor of DNA methylation 5-aza-2'-deoxycytidine.4We are presently examining exon 1 and the upstream region in moredetail to investigate the potential for this de novo methylation tosilence pi6!CDKN2 expression.

Our analysis of uncultured bladder TCCs and bladder cell linesindicates that the 5' CpG island of the p16!CDKN2 tumor suppressorgene is frequently methylated in this tumor type. The high frequency(67%) of uncultured tumors showing de novo methylation of the

p16!CDKN2 5' CpG island could explain the low rate at whichp16!CDKN2 homozygous deletions and intragenic mutations arefound in bladder TCC (13, 14). Thus, our data indicate that, in contrastto squamous cell carcinomas of the bladder in which p16/CDKN2 isfrequently deleted (15), de novo methylation of the 5' CpG island ofpi6!CDKN2 may be a more common mechanism of inactivation ofthis tumor suppressor gene in TCC.

Our study is the first to show that methylation of the 5' CpG island

of p16!CDKN2 occurred not only in bladder tumors and bladder celllines but also in normal colon mucosa. In contrast to other normaltissues examined, methylation of this region wa@ observed in thenormal colonic epithelium from four patients who had segments of thecolon removed for reasons other than cancer development and also inthe normal colon mucosa of several (6 of 10) patients with coloncancer (Fig. 2). This is striking as methylation of CpG islands inautosomal genes is generally restricted to genes located on the mactive X chromosome (19), imprinted genes (21), or genes that undergode novo methylation during cell line inmortalization and tumorigenesis (22, 23). A study by Herman et a!. (31) in this issue of CancerResearch reports the absence of p16!CDKN2 methylation in normalcolon mucosa based on Southern analysis. Several possibilities mayexplain this apparent discrepancy: (a) the CpG sites examined werenot the same in both studies, although they are all located within the

5' CpG island of p16/CDKN2; (b) Southern analysis may be lesssensitive in detecting methylation changes than the PCR-based assayused in our study; and (c) the normal colon samples examined in our

4 M. Gonzalez-Zulueta, C. M. Bender, M. Pao, Y. Tsai, J-M. Zingg, and P. A. Jones,

unpublished data.

p16 EXPRESSION

A

B

C

D

E

F

G

H

J

Fig. 2. Summary of mRNA expression and methylation status of the p16/CDKN2genein colon carcinomas and adjacent normal mucosa of patients A to J. mRNA levels weredetected by RT-PCR; GAPDH expression was used as control. Bars and numbers,p16:GAPDH ratio as determined with an LKB UltraScan XL Laser densitometer. Methylation status of exon 1 was determined by a PCR-based assay with primers flanking theFnuDII (F), SacII (5), HpaII (H), and CfoI (C) sites in exon 1 ofpl6/CDKN2. Diagramsof exon 1 and restriction sites are not drawn to scale. 0, unmethylated site. •,fullymethylated site. Partially filled circles, partially methylated sites, as determined byscanning with an LKB UltraScan XL Laser densitometer.

4534

on July 28, 2018. © 1995 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

phVCDKN2 METHYLATION IN BLADDER AND COLON CANCERS

study were enriched for colon mucosa cells (see “MaterialsandMethods―).To date, methylation of CpG islands of autosomal genesin normal tissues has only been reported for the estrogen receptor genein normal colonic mucosa, where methylation of this gene was observed to arise as a direct function of age (17). However, we did notobserve any association between methylation of p16!CDKN2 in normal colon mucosas and the age of the patients in our study. It will beimportant to determine if the methylation and transcriptional silencingof pi6!CDKN2 observed in some colon specimens are involved innormal cell function or are precursors to tumorigenesis.

The results obtained with the normal colon and matched tumorpairs were surprisingly different to those obtained with the bladderTCCS. In 50% of the colon normal!tumor matched cases, p16!CDKN2

was methylated and not expressed in the normal tissue but unmethy

lated and highly expressed in the tumor tissue. Variable results regarding the expression of the gene in the tumor and the normalmucosa were seen in the other patients; however, in no case wasexpression seen in samples that demonstrated methylation of the sitesexamined in exon 1. Furthermore, in four of the five colon carcinomasin which p16!CDKN2 expression was strongly up-regulated, we observed a clear inverse expression pattern of the p16!CDKN2 and

@15INK4B genes, with p16!CDKN2 up-regulation and pJ5INK4B down

regulation in the tumors when compared to the corresponding normalmucosa. p16!CDKN2 appears to accumulate in cells in which the

function of the retinoblastoma susceptibility gene product (pRb) hasbeen compromised (1). It would be of interest to examine the status ofpRb in our panel of colon carcinomas, although pRb alterations havenot been reported in colon cancer. Hannon and Beach (10) haveshown a strong up-regulation of @J5INK4Bexpression by TGF-j3 in

human keratinocytes and have proposed a role for @J5INK4Bas aneffector of TGF-@3-mediated cell cycle arrest. In normal colonic epi

thelium, the upper 2!3 of the crypts are composed of quiescent cellsthat are rapidly being replaced by new cells migrating from the lower1!3 of the crypt (25). The appreciable levels of @J5INK4Bexpressiondetected in the normal colon mucosas in our study may indicate thatthis cell cycle regulator has a role in inducing quiescence in themajority of cells of the normal colonic epithelium. Markowitz et a!.(32) have recently reported a high frequency of mutations in theTGF-f3 type II receptor in colon carcinomas; therefore, it would be

important to examine the status of this receptor in our panel of coloncarcinomas to determine if the observed @J5hT@@4Hdown-regulation in

a subset of these tumors correlates with TGF-(3 type II receptormutations.

Acknowledgments

We thank Drs. StephenB. Baylinand James 0. Hermanfor communicationofresults prior to publication, Dr. Jane W. Fountain for many helpful discussions andcritical reviewof the manuscript,Dr. RobertStellwagenfor criticalreviewof thismanuscript,and Dr. David Beach for kindly providingp16 eDNA.

References

1. Serrano, M., Hannon, G. J., and Beach, D. A new regulatory motif in cell-cycle controlcausing specific inhibitionof cyclin D/CDK4. Nature (Land.), 366: 704—707,1993.

2. Fountain, J. W., Karayiogou, M., Ernstoff, M. S., K1rkWOOd,J. M., Viock, D. R.,Titus-Emstoff,L, Bouchard,B., Vijayasaradhi,S., Houghton,A. N., Lahti,J., Kidd,V. J.,Housman, D. E, and Dracopoli, N. C. Homozygous deletions within human chromosomeband 9p21 in melanoma.Proc. NatI. Acad. Sci. USA, 89: 10557—10561,1992.

3. van der Riet, P., Nawroz, H., Hruban, R. H., Corio, R., Tokino, K., Koch, w., andSidransky, D. Frequent loss of chromosome 9p2l—22early in head and neck cancerprogression. Cancer Res., 54: 1156—1158, 1994.

4. Kamb, A., Gruis, N. A., wcaver-Feldhaus, J., Liu, 0., Harshman, K., Tavtigian, S. V.,Stocker,E.,Day,R.S.,III,Johnson,B.E.,andSkolnick,M.H.Acellcycleregulatorpotentially involved in genesis of many tumor types. Science (Washington DC), 264:436—440, 1994.

5. Nobori, 1., Miura, K., Wu, D. J., Lois, A., Takabayashi, K., and Carson, D. A.Deletions of the cycle-dependent kinase-4 inhibitor gene in multiple human cancers.Nature (Land.), 368: 753—756,1994.

6. Caldas, C., Hahn, S. A., da Costa, L. T., Redston, M. S., Schutte, M., Seymour. A. B.,Weinstein, C. L., Hruban R. H., Yeo C. J., and Kern, S. E. Frequent somatic mutationsand homozygous deletions of the pitS (MTSJ ) gene in pancreatic adenocarcinoma.Nat. Genet., 8: 27—32,1994.

7. Mori, T., Miura, K., Aoki, T., Nishihira, T., Mori, S., and Nakamura, Y. Frequentsomatic mutation of the MTSI/CDK41 (multiple tumor suppressor/cyclin-dependentkinase 4 Inhibitor) gene in esophageal squamous cell carcinoma. Cancer Res.. 54:3396—3397, 1994.

8. Okamoto, A., Hussain, S. P., Hagiwara, K., Spillare, E. A., Rusin, M. R.. Demetrick,D. J., Serrano,M., Hannon,G. J., Shiseki, M., Zariwala.M., and Harris,C. C.Mutations in the p!6INK4IMTSI/CDKN2, pl5INK4B@KaH, and p18 genes in primaryand metastatic lung cancer. Cancer Res., 55: 1448—1451,1995.

9. Reed, J. A., Loganzo, F., Shea C. R., Walker, G. J., Flores, J. F., Glendening, M.,Bogdany, J. K., Shiel, M. J., Haluska, F. G., Fountain, J. W., and Albino, A. P. Lts.Sof expression of the p16/cyclin-dependent kinase inhibitor 2 tumor suppressor gene inmeianocytic lesions corre. :tes with invasive stage of tumor progression. Cancer Res.,55:2713—2718,1995.

10. Hannon, G. J., and Beach, D. p15INK4B is a potential effector ofTGF-@3-inducedcellcycle arrest. Nature (Land.), 371: 257—260,1994.

I 1. Cairns, P., Tokino, K., Eby, Y., and Sidransky, D. Homozygous deletions of 9p21 inprimary human bladder tumors detected by comparative multiplex polymerase chainreaction. Cancer Res., 54: 1422—1424,1994.

12. Stadler, W. M., Sherman, J., Bohlander, S. K., Roulston, D., Dreyling. M., Rukstalis,D., and Olopade, 0. 1.Homozygous deletions within chromosomal bands 9p2l—22inbladder cancer. Cancer Res., 54: 2O60—2()63,1994.

13. Cairns, P., Mao, L., Merlo, A., Lee, D. J., Schwab, D., Eby, Y., Tokino, K.. van dciRiet, P., Blaugrund, J. E., and Sidransky, D. Rates of p16 (MTS/) mutations inprimary tumors with 9p loss. Science (Washington DC), 265: 415—416, 1994.

14. Spruck, C. H., III, Gonzalez-Zulueta, M., Shibata, A., Simoneau, A. R., Lin, M-F.,Gonzales, F., Tsai, Y., and Jones, P. A. plo gene in uncultured tumors. Nature(Lond.), 370: 183—184,1994.

15. Gonzalez-Zulueta, M., Shibata. A.. Ohneseit, P.. Spruck. C. H.. III, Busch, C.,Shamaa, M., Elbaz, M., Nichols, P. W., Malmström,P. U.. and Jones, P. A. Highfrequency of chromosome 9p allelic loss and alterations of the CDKN2 locus distinguish squamous cell carcinoma from transitional cell carcinoma of the bladder. J.NatI. Cancer Inst., in press, 1995.

16. Hermann, J. G., Latif, F., Weng, W., Lerman, M. I. ‘thar.B., Liu, S., Samid, D..Duan, D-S., Gnarra, J. R., Linehan, W. M., and Baylin, S. B. Silencing of the VHLtumor suppres.sorgene by DNA methylation in renal carcinoma. Proc. NatI. Acad.Sd. USA, 91: 9700—9704,1994.

17. Issa, i-P. 1., Ottaviano, Y. L., Celano, P., Hamilton, S. R., Davidson, N. E., andBaylin, S. B. Methylation of the estrogen receptor CpG island links aging andneoplasia in human colon. Nat. Genet., 7: 536—540, 1994.

18. Bird, A. P. CpG-rich islands and the function of DNA methylation. Nature (Lond.).321: 209—213,1986.

19. Singer-Sam, J., and Riggs, A. D. X chromosome inactivation and DNA methylation.Exper. Suppl. (Ba.sel),64: 358—384,1993.

20. Liu, W-M., and Schmid, C. W. Proposed roles for DNA methylation in Alu Iranscriptional expression and mutational inactivation. Nucleic Acids Res.. 2/: 1351—1359, 1993.

21. Ferguson-Smith, A. C., Sasaki, H., Cattanach. B. M., and Surani, M. A. Parentalorigin-specific epigenetic modification of the mouse 1-119gene. Nature (Lond.). 362:751—755,1993.

22. Jones, P. A., and Buckley, J. D. The role of DNA methylation in cancer. Adv. CancerRes., 54: 1—23,1990.

23. Laird, P. W., and iaenisch, R. DNA methylation and cancer. Hum. Mol. Genet.. 3:1487—1495,1994.

24. Jones, P. A., and Taylor, S. M. Cellular differentiation, cytidinc analogs and DNAmethylation. Cell, 20: 85—93,1980.

25. Crissman, I. D., and Barwick, K. W. Pathologic staging of colon and rectal cancer. In:H. Wanebo(ed), ColorectalCancer,pp. 57—74.St. Louis: Mosby, 1993.

26. Bergkvist, A., Ljungquist, A., and Moberger, G. Classification of bladder tumorsbased on the cellular pattem. Acta Chir. Scand., 130: 371—378,1965.

27. Chomczynski, P., and Sacchi, N. Single-step method of RNA isolation by acidguanidinium thiocyanate-phenol-chloroform extraction. Anal. Biochem., 162: 156—159, 1987.

28. Singer-Sam, J., Yang, T. P., Mori, N., Tanguay, R. L., La Bon, I. M., Flores, I. C.,and Riggs, A. D. DNA methylation in the 5' region of the mouse PGK-I gene and aquantitative PCR assay for methylation. In: G. A. Clawson. D. B. Willis. A.Weissbach, and P. A. Jones (eds.), Nucleic Acid Methylation, Vol. 128, pp. 285—289.New York: Alan R. Liss, Inc., 1990.

29. Gardiner-Garden, M., and Frommer, M. CpG islands in vertebrate genomes. J. Mol.Biol., 19t5: 261—282,1987.

30. Merlo, A., Herman, I. 0., Mao, L., Lee, D. I., Gabrielson. E., Burger, P. C., Baylin,S. B., andSidransky,D. 5' CpGislandmethylationis associatedwith transcriptionalsilencing of the tumor suppressor pl6/CDKN2/MTS1 in human cancers. Nat. Mcd.,I: 686—692, 1995.

31. Herman, I. G., Merlo A., Mao L., Lapidus R. G., Issa, i-P. 1., Davidson N. E.,Sidransky, D., and Baylin, S. B. Inactivation of the CDKN2/pI6IMTSJ gene isfrequently associated with aberrant DNA methylation in all common human cancers.Cancer Res., 55: 4525—4530,1995.

32. Markowitz, S., Wang, J., Myeroff, L., Parsons, R., Sun, L., Lutterbaugh. J.. Fan, R. S.,Zborowska, E., Kinzler, K., Vogelstein, B., Brattain, M., and Willson, I. K. V.Inactivation of the type II TGF-@3receptor in colon cancer cells with microsatelliteinstability. Science (Washington DC), 268: 1336—1338,1995.

4535

on July 28, 2018. © 1995 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

1995;55:4531-4535. Cancer Res   Mirella Gonzalez-Zulueta, Christina M. Bender, Allen S. Yang, et al.   Correlates with Gene SilencingSuppressor Gene in Normal and Transformed Human Tissues

Tumorp16/CDKN2 CpG Island of the ′Methylation of the 5

  Updated version

  http://cancerres.aacrjournals.org/content/55/20/4531

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/55/20/4531To request permission to re-use all or part of this article, use this link

on July 28, 2018. © 1995 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from