[CANCER RESEARCH 58, 5489-5494, December 1, 1998]

Aging and DNA Methylation in Colorectal Mucosa and Cancer1

Nita Ahuja, Qing Li, Avinash L. Mohan, Stephen B. Baylin, and Jean-Pierre J. Issa2

The Johns Hopkins Oncology Center ¡N.A.. Q. L, A. L. M.. S. B. B., J-P. J. /./ and Departments of Surgery ¡N.A./ and Medicine ¡S.B. B.], The Johns Hopkins University School

of Medicine, Baltimore, Maryland 21231

ABSTRACT

DNA methylation of promoter-associated CpG islands may function as

an alternate mechanism of silencing tumor suppressor genes in multipleneoplasias including colorectal cancer. De novo methylation of genesappears to be an early and frequent event in most neoplasias. For the ERand IGF2 genes, we have previously shown that methylation actuallybegins in the normal colon mucosa as an age-related event and progresses

to hypermethylation in cancer. In this study, we have determined thefrequency of age-related methylation in normal colonie mucosa among the

genes hypermethylated in colorectal cancer. We studied six genes, including N33, MYOD,p¡6, HIC-l, THBSI, and CALCA. The N33 gene showedpartial methylation in normal colon mucosa, which was age-related(r = 0.7; P = 0.003 using regression analysis). Adenomas and cancers

showed further hypermethylation at this locus. Similarly, the MYOD geneshowed age-related methylation in normal colon mucosa (r = 0.7;/' < 0.00001 using regression analysis) and hypermethylation in cancers.

Age-related methylation seems to be gene specific, because pI6, THBSI,HIC-l, and CALCA were not affected. Furthermore, this process may alsobe modulated by tissue-specific factors. Our study suggests that aging is a

major contributing factor to hypermethylation in cancer.

INTRODUCTION

CGIs3 are areas rich in CpG dinucleotides that are found within the

promoter region of most housekeeping genes ( 1). These islands arenormally unmethylated regardless of the transcriptional status of thegene, except for genes on the inactivated X chromosome in females(2) and certain imprinted genes (3). De novo methylation of promoter-associated CGIs has been associated with the transcriptional inacti-

vation of the gene (4) and seems to be functionally equivalent to aninactivating mutation for the silencing of multiple tumor suppressorgenes (5, 6) such as pI6, VHL, and RB. Abnormal CpG islandmethylation seems to be a frequent event in most neoplasias, includingCRCs (5, 6).

Aging is one of the most important risk factors for the developmentof neoplasia (7). The incidence of cancer rises sharply after the age of60 years (8). For example, CRC, which is one of the common solidhuman malignancies, appears at a median age of 62 years in both menand women. This risk factor has been attributed partly to the cumulative exposure to carcinogens over time and the multiple hits requiredfor the onset of neoplasia (7). Epigenetic modifications of DNA, suchas altered DNA methylation patterns, have long been postulated toplay a role in aging and the associated increased incidence of neoplasia (9). However, most of the focus has been on the overall genomichypomethylation, which has been proposed to function as a "countingmechanism" for cellular senescence (10, 11), or on regional areas of

hypermethylation outside the promoter region of genes (12), such as

Received 6/15/98; accepted 10/5/98.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.

1Supported in part by Colon Cancer Spore Grant CA62924 and National Cancer

Institute Grant CA433I8. N. A. is supported by NIH Training Grant 1-T32-DK07713.J-P. J. I. is a Kimme] Foundation Scholar.

2 To whom requests for reprints should be addressed, at The Johns Hopkins Oncology

Center. 424 North Bond Street. Baltimore. MD 21231. Phone: (410) 955-8506; Fax: (410)614-9884; E-mail: jpissa@>welchlink.welch.jhu.edu.

' The abbreviations used are: CGI, CpG islands; CRC. colorectal cancer; MI, micro-

satellite instability.

those in c-MYC (13) orc-FOS(\4, 15). The functional significance of

these events is somewhat unclear at present (16).However, hypermethylation of promoter-associated CGIs has been

more clearly associated with gene silencing (5, 6). We have recentlyshown that the estrogen receptor (£/?)gene on 6q gets hypermethylated at its promoter CGI in all colonie neoplasms (17). This processis associated with decreased ER expression that appears to start in thenormal colonie epithelium as a function of age. No methylation isapparent in young individuals, but partial methylation appears in olderindividuals, and the ER gene is hypermethylated in all colonie adenomas and neoplasms. Thus, aging itself or age-dependent events

appear to be a significant risk factor for hypermethylation of this gene.We have subsequently observed that the insulin-like growth factor II(IGF2) gene also shows age-related methylation in the colon, which

eventually progresses to hypermethylation in cancers (18). Thesestudies illustrate that aging itself may be a significant risk factor forthe hypermethylation of these genes in cancers.

In the current study, we have tried to determine the frequency ofage-related methylation for multiple additional genes that are hypermethylated in CRC. Our data indicate that aging-related methylation

is a common event in CRCs, being present in four of the eight genesknown to be hypermethylated in this tumor type, and suggest thataging is a major contributor to hypermethylation in CRC.

MATERIALS AND METHODS

Tissue Samples. Samples of all primary cancers including CRCs, meta-

static liver lesions, and normal colon and liver tissues were obtained tronisurgical resections of patients operated on at the Johns Hopkins Hospital. Thesamples were immediately fro/.en in liquid nitrogen after resection and storedat —70°Cuntil processing. For the separation of epithelium from stroma.

freshly resected samples were lightly scraped with a razor blade to collectsurface epithelium and incubated for 2 h with gentle agitation in HBSS withoutcalcium and phosphate supplemented with 30 IDM EDTA (pH 7.0). Afterincubation, epithelial cells were floating in the solution and collected bycentrifugation. The remaining tissue contained stromal cells exclusively. Thepurity of each fraction was greater than 90% as determined by immunohisto-

chemistry. Specimen collection procedures were approved by the Joint Committee on Clinical Investigation of the Johns Hopkins Hospital in accordancewith the policies of the Department of Health and Human Services. Cell lineswere obtained from the American Type Culture Collection (Rockville. MD).

Southern Blot Analysis. DNA was extracted using standard techniques.Five pig of genomic DNA were digested with 100 units of the appropriaterestriction enzyme as specified by the manufacturer (New England Biolabs).DNA was precipitated, and the DNA fragments were separated on a 1%agarose gel. transferred to a nylon membrane (Zetaprobe; Bio-Rad), and

hybridized with the various probes described below. The blots were thenexposed on a phosphor screen for 2-5 days and developed using a Phospho-

rlmager (Molecular Dynamics). Using ImageQuant software, image analysiswas carried out, and the density of methylation was determined as the densityof the higher molecular weight band as a percentage of the total density for thatsample.

Hybridization Probes. Probes used for the analysis of promoter methylation have been described previously and included a 0.3-kb 5' ER probe (17).a 0.65-kb 5' c-ABL probe ( 18), a 0.35-kb 5' pió probe ( 19), a 0.6-kb 5' THBSIprobe (19), a 1.1-kb 5' N33 probe (20) containing exon 1, a 0.6-kb CALCA

probe (21), a 0.65-kb exon 1 MYOD probe (21), and the highly polymorphicpYNZ22.1 probe near the HIC-l locus (22). Information on the IR, CNP,HASH!, hMSH2, and c-MYC probes is available upon request. All probes were

5489

Research. on December 18, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from

AGING AND DNA METHYLATION IN CRC

labeled by random priming (Boehringer Mannheim) using [a-'2P]dCTP (Am-

ersham).Statistics. All statistical analyses were performed using the Microsoft

Excel software program.

RESULTS

Genes Studied. We have previously shown age-related methyla-

tion of the normal colorectal mucosa for the ER and the IGF2 genes(17, 18). In the present study, we analyzed six additional genes thathave previously been shown to be hypermethylated in CRC: (a) N33,a ubiquitously expressed gene on 8p22 that was recently cloned froman area of frequent allelic loss in prostate carcinomas and has beenshown to display promoter hypermethylation in several CRC cell linesassociated with a lack of expression of the N33 transcript (20); (b)MYOI), a muscle-specific transcription factor located on 1Ip that was

one of the first genes shown to be hypermethylated in immortalizedcell lines and has subsequently been shown to be hypermethylated inseveral cancers, including bladder, breast, and CRCs (23); (e) piò, acyclin-dependent kinase inhibitor on 9p21 with tumor-suppressor

properties that has been shown to be hypermethylated in most common human cancers, including CRCs (24); (d) THBS1, a p53- and/?/>-regulaied angiogenesis inhibitor on 15q that we have previouslyshown to be methylated in sporadic CRC with MI (19); (e) HIC-1, azinc-finger transcription factor on I7pl3.3 that was cloned on the

basis of the presence of a large CGI in this area that becomesfrequently methylated in various cancers, including CRC (25); and (/)CALCA (calcitonin), on 1Ipl5 that shows CGI hypermethylation in

lung and CRCs as well as in leukemias (26). Control genes that werenot known to be hypermethylated in CRC included c-MYC, IR (insulinreceptor), CNP (C-type natriuretic peptide), c-ABL, HASH! (human

achaete schule homologue), and !iMSH2.N33 Methylation in Normal Colon Mucosa and CRCs. The N33

gene has a typical CGI spanning its promoter and first exon withmultiple methylation-sensitive restriction sites, including an Eagl site,

a Bi\yHlI site, and a Sacll site (Fig. \A). We investigated the status ofN33 methylation in both normal colon mucosa and in CRCs.

We first examined the methylation of the N33 CGI-associated

promoter in 44 normal colorectal mucosa samples from subjectsranging in age from 8 to 90 years. We used Southern blot analysis ofgenomic DNA that was first restricted with methylation-sensitiveenzymes and then hybridized with a 5' region probe. The N33 gene

showed a partial methylation in the normal colorectal mucosa that wasage-related, being barely detectable in young individuals and progres

sively increasing in older individuals (see the examples in Fig. Iß).Tobetter study this relationship, we used densitometric analysis to measure the relative ratio of the methylated ( 1.1-kb) band in each lane. The

intensity of methylation was plotted as a function of age (Fig. 1C) andshowed that N33 methylation increased as a function of age (r = 0.7;P = 0.003 using regression analysis). The intensity of methylationaveraged 15.1% in patients less than 20 years of age (n = 4), 37.1%in patients 20-39 years of age (n = 10), 48.0% in patients 40-59years of age (n = 9), and 58.4% in patients more than 60 years of age(n = 21 ; Fig. 1C). Next we did a subset analysis comparing the extent

of N33 methylation with various clinical and pathological parameters.

A.

Age

100-

90-

80-

70-

«H

I30

20-

10'

0

R-0.7p=0.003

Age

S 15 20 29 37 46 56 61 72 82 F F TI T2 T3 T4 T5 T6 T7 T8 T9 MAC MC

B. D.

OJ

Fig. l. Methylation status of the N33 gene CGI in normal colorectal mucosa and CRCs. A, partial restriction map of the N33 gene showing the first exon (•)and flanking region.The CpG density of the area is plotted ahove the map, and the high density of CpG dinucleotides indicates a CGI. The predicted restriction fragments using ///ndlll/Si/rll are shownal the bottom: H. Hináltt; S, Sai II; £.Eagl; Sm, Smal; B. BssHll. B, partial methylation of the N33 gene in normal colorectal tissues. Representative Southern blots of DNA from normalcoloreclal mucosa are shown. /.<•/;.approximate sizes of the hands (in kh). Tup. age of the patient at the time the mucosa sample was obtained. All lanes show patients' DNA restrictedwith //('.'itili, a flanking enzyme, and .S'nrII. a melhylation-sensitive restriction enzyme, except for Lane F. which shows DNA restricted with //(milIf alone. The l.l-kb band reflects

Ihe methylalion of the Sudi site. C. methylation of N33 CGI increases with age. The percentage of N33 methylation of DNA from normal colorectal mucosa was plotted against theage of the patient including the 10 patients shown in Fig. \H and 34 other control patients. A regression analysis of all samples gave a linear fit, as shown by the straight line, witha correlation coefficient of 0.7. which is statistically significant (P - 0.003). D, methylation of the N33 CGI in colonie tumors. Left, the size of the various bands (in kb) is shown.The Ume F shows DNA restricted with the ////idlll flanking cn/yme alone, whereas all other lanes show samples restricted with ///ndlll/Si/rll. iMiiex T1-T8 are cancers, whereas LaneTV is an adenoma. Lanes T1-T3 and TV are hypermethylated. Limes T4-T6 appear to be equivalently methylated to normal colonie mucosa, and Lanes T7-T8 are cancers that appearto be hypomethylated at N33. On the right are two colorectal tumor/adjacent mucosa pairs. M, uninvolved mucosa; A. adenoma; C, carcinoma. Note that the small 0.3-kb unmethylatedband is visible in only some samples, depending on the length of time the samples were electrophoresed.

5490

Research. on December 18, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from

AGINO AND DNA METHYLATION IN CRC

There were no differences in the intensity of N33 methylation basedon the gender of the patient (means: females = 44% (n = 21) andmales = 50% (n = 23); P = 0.3 by / test) or the site of origin of thecolorectal mucosa [means: left side = 46% and right side = 48%(n = 22 each); P = 0.7 by / test]. Of the 44 cases examined, 22 werefrom patients without neoplasia. Age-related methylation was also

observed in this group (from 10% methylation in patients less than 20years of age to 43% methylation in patients over 40 years of age).However, control patients were significantly younger than those withneoplasia, precluding an accurate comparison of the two groups.

Next we studied the pattern of N33 methylation in 44 primarysporadic CRCs. The median age of these patients was 68 years. Basedon the above data for aging and N33, the normal surrounding colonmucosa of these samples would be predicted to have a high degree ofN33 methylation, averaging greater than 50%. The majority of CRCsappeared to either get more hypermethylated (20 of 43 samples) orremain equivalently methylated (19 of 43 samples) when compared tosimilar age-matched normal colon samples (see the examples in Fig.

ID). However, there are some tumors that are clearly hypomethylated,ranging from 2-20% methylation (4 of 43 samples) at the N33

promoter, but the data on the normal colon mucosa from theseparticular samples were unavailable (Fig. ID). To further understandthe extent of N33 methylation in CRCs, we studied the pattern of N33methylation in 11 paired samples containing both cancer and normalmucosa from the same patient. The average age of these 11 pairedsamples was 65.9 years. As expected, the normal colon tissue fromthese samples showed a high degree of N33 methylation, averaging67%. However, the paired tumor tissue from these samples showed asignificantly higher amount of N33 methylation, averaging 83%(P = 0.03 using / test; see the examples in Fig. ID). We also measured

N33 methylation in 10 adenomatous colonie polyps. These polyps alsoshowed a high degree of N33 methylation, averaging 80% (n = 10;

see the examples in Fig. ID), which was equivalent to that seen incarcinomas. Finally, we had previously observed an association between MI and CGI methylation in CRC for some (P16 and THBS1)but not all (ER) genes (19). MI data were available for 24 of the CRCsanalyzed. N33 methylation was not affected by the MI status of thetumors; N33 methylation averaged 68% in 14 MI+ cases and 75% in10 MI- cases (P = 0.33). Thus, N33 hypermethylation seems to be

a very early event in CRC pathogenesis, starting as an age-related

event in the normal colorectal mucosa and evolving further in adenomas and carcinomas.

Age-related MYOD Methylation. The MYOD gene has a largeCGI of about 3.0 kb that spans its promoter as well as the first threeexons of the gene. The MYOD CGI contains five Sacll sites as well asmultiple Smal sites (Fig. 2A). We looked at the methylation status ofthe MYOD gene CGI in normal colorectal mucosa and CRCs usingSouthern blot hybridization. There were 37 samples of normal colorectal mucosa of various ages, ranging from 8-90 years. Throughoutits entire CGI, the MYOD gene had age-related partial methylation in

the normal colorectal mucosa samples as measured by two differentmethylation-sensitive restriction enzyme sites (see the examples of

Southern blots in Fig. 2B). The intensity of MYOD methylation wasmost dense within the heart of the CGI in exon 1 but also involved theCpG dinucleotides near the transcription start site in an age-relatedfashion, as indicated by the 1.7-kb band in Fig. 2B. To quantitate this

relationship, the intensity of methylation at the Sacll site closest to thetranscription start site, designated as SI on Fig. 2A, was measured bydensitometric analysis and plotted as a function of age. As shown inFig. 2C, the MYOD gene promoter showed increased methylation ofits CGI as a function of age (r = 0.7; P < 0.00001 using regression

analysis) at the SI site. The intensity of methylation averaged 3.5% inpatients less than 20 years of age (n = 4), 5.5% in patients 20-39

A. III H«IIIHIIS2i"ÎS»JSJ B

HIM

7» O tt 15 » F

R=0.8p<0.00001

D.

N1T1 N2T2N3T3 N4T4N5T5

«l

0 20

T6-12

40 60Age

80 100

Fig. 2. Methylation status of the MYOD gene CGI in normal colorectal mucosa andCRC. A. partial restriction map of the MYOD gene showing the first three exons (•)andthe flanking region. Vertical barf, restriction sites; B. BatnHl: H. //mdlll; SI-S5. Sacllsites. Sml-Sm6, Smal sites. The CpG density of the area is plotted above the map. and thehigh density of CpG dinucleotides indicates a CGI. B. partial methylation of the MYODgene in normal colorectal tissues. Representative Southern blots of normal colorectalmucosa using two different methylation-sensitive restriction enzymes are shown. On theleft of each panel are the approximate sizes of the bands (in kbl. Urne F shows DNA cutwith the flanking enzyme alone. U-fl panel. DNA of 10 patients of various ages (indicatedon top of each lane) restricted with BamHl. a flanking enzyme, and Suri I. a methylation-sensitive restriction enzyme. The 0.36- and 0.71-kb bands are present if there is nomethylation of any Sudi site, whereas the other bands reflect the methylation of either S2(1.1 kb), S3 (0.53 kb). S2 and S3 (1.3 kb), orali of S1-S3 (1.7 kb). RiKhi panel. DNA ofthe same 10 patients restricted with BamHl and Srnai. another methylation-sensitiverestriction enzyme. The 0.34- and 0.56-kb bands arc present if there is no methylation ofthe Smal sites, whereas the other bands reflect the methylation of Sm2 (0.9 kb). Sm2 andSm.3 (I.I kb). Sml and Sm2 (1.5 kb). and Sml-Sm3 (1.7 kb). C methylation of theMYOD CGI increases with age. The percentage of MYUD melhylation of DNA fromnormal colorectal mucosa was plotted against the age of the patient. /,<'// panel, themethylation of the 5' SacII site. SI, plotted as a function of age for the 10 patients shown

above and 27 other patients. A regression analysis of all samples gives a linear fit, asshown by the straight line, with a correlation coefficient of 0.7. which is statisticallysignificant (P < 0.00001). Righi panel, the melhylation of the 5' Smal site. Sml. plotted

asa function of age for a total of 23 patients ranging in age from 10-90 years. Again, thedata result in a linear fit by regression analysis, with a correlation coefficient of 0.8(P < 0.00001 ). D, methylation of the MYOD CGI in CRCs. On the left are shown the siz.esof the various bands (¡nkb). In this analysis of DNA restricted with ///»dill and Saill. the0.36- and 1.1-kb bands are seen if there is no methylation at the MYOD locus, whereas thehigher molecular weight bands represent increasing methylation. The left panel shows fivepairs of normal (Wtumor (D specimens. Tumors show hypermethylalion compared to thesurrounding normal tissue. In the middle, Lanes T6-TI2 represent additional tumors

showing almost complete methylation. and the panel on the extreme righi shows hypermethylation in various CRC cell lines (indicated at the top of each lane). Note that CaCO2,SW480, OLD 1. and HT29 show almost 100% melhylation. and none of the cell lines havealíelesdevoid of methylation (i.e.. no bands at 0.36 and 1.1 kb).

years of age (n = 6), 11.8% in patients 40-59 years of age (n = 10),and 20.0% for patients more than 60 years of age (n = 17; Fig. 2C).

The methylation of the Sacll sites within exon 1, namely S2 and S3,

5491

Research. on December 18, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from

AGING AND DNA METHYLATION IN CRC

90 85 82 82 Svii S5yn ÃŒMlyo

F N N N

.--— * —

c-abl THBSI IR ( '¡licitomi!

Fig. 3. Melhylation stalus of other genes in normal colorectal tissues. Representative Southern blots of multiple genes that show no evidence of age-related methylation are shown.Lane F represents DNA cut with the Hanking enzyme alone, whereas the other lanes are DNA cut with both the flanking enzyme and a methylation-sensitive restriction enzyme (£.ktixl: lì./f.v.vHII;S. Siu'll: M. Mlul; Nr. Nrul; N. Not]). The ages of patients are shown on the lop of each panel for the various samples of normal colorectal mucosa. In each case,

Ihe arrowheads point to the size of the methylated band (which is absent in all cases). In the THBSI blot, a nonspecific band can be seen in all of the lanes, including the flank lane.

also showed an age-related increase, as demonstrated by multiple

higher molecular weight bands on Southern blot analysis (Fig. 2B). Infact, when we considered the methylation of either SI, S2, or S3, theMYOD CGI showed a striking age-related methylation (r = 0.9;

P < O.CKXXXMusing regression analysis): the intensity of methylationaveraged 28.2% in patients less than 20 years of age (n = 4), 39.8%in patients 20-39 years of age (n = 6), 61.7% in patients 40-59 yearsof age (n = 10), and 70.3% in patients more than 60 years of age(;i = 17). On subset analysis, there was no difference in the intensity

of MYOD methylation at the SI site based on gender (means: 13.5%for females (n = 18) and 13.7% for males (n = 19); P = 0.9 by / test).

However, there was a marginally significant higher intensity ofMYOD methylation at the SI site in samples originating from left-sided colorectal mucosa (mean, 14.2%; n = 24) compared to right-sided colorectal mucosa [mean, 8.6%; n = 11 (P = 0.05 by t test)]. Of

the 37 samples studied, 22 were from controls, and 15 were frompatients with CRC. Age-related methylation was equivalent in the two

groups.We also studied the methylation of the Sma\ sites in the MYOD CGI

and found a similar age-related increase in methylation at all of the

Símilsites around exon 1 (examples of Southern blot, Fig. 2B). Theintensity of methylation at the Snuil site closest to the transcriptionstart site, which is designated as Sml ¡nFig. 2A, was measured bydensitometric analysis and plotted as a function of age. As shown inFig. 2C, the MYOD gene promoter also showed increased methylationat the Sml site as a function of age (r = 0.8; P < 0.00001; range,10-90 years; sample size, 23). The intensity of methylation averaged0.7% in patients less than 20 years of age (n = 2), 1.1% in patients20-39 years of age (n = 4), 6.7% in patients 40-59 years of age(n = 7), and 19.9% for patients more than 60 years of age (n = 10;

Fig. 2C).To determine the impact of age-related methylation at the MYOD

locus on hypermethylation in cancer, we studied 11 cases of pairednormal/tumor DNA from the same patients. In all cases, methylationwas significantly more extensive in the tumors (including 4 adenomasand 10 carcinomas), with most alíelesshowing hypermethylation ofall of the methylation-sensitive restriction enzyme sites in the area

(Fig. 2D). Methylation seemed to be similar in adenomas and carcinomas, suggesting that MYOD hypermethylation is a very early eventin CRC pathogenesis. We also studied an additional 20 carcinomas,and all showed high degrees of methylation (>80%), suggesting thatMYOD hypermethylation is nearly universal in CRCs (see Fig. 2D).There was no difference in MYOD methylation by MI status because100% of the tumors were hypermethylated. Finally, six of eight CRCcell lines (RKO, CaCO2, SW480, SW48, DLD1, and HT29) analyzedhad more than 95% of the alíelesmethylated at all Sodi sites in theregion, whereas the remaining two (Lovo and HCT116) had more than50% of the alíelessimilarly methylated. None of the cell lines hadevidence of alíelescompletely devoid of methylation (i.e., bands at

0.36 and 1.1 kb; Fig. ID). Overall, these data suggest that MYODmethylation starts in the normal-appearing colorectal mucosa as a

function of age and becomes more prominent in cancers, suggestingeither that the cells with preexisting methylation are selected for in theneoplastic process, or that tumors greatly extend methylation at thislocus.

Age-related Methylation Is Gene Specific. In contrast to N33 andMYOD, none of the other genes examined showed evidence of age-

related methylation using similar Southern blot analysis. In particular,THBSI was analyzed in 47 samples of normal colon mucosa rangingin age from 20-98 years, and no methylation was observed (see theexamples in Fig. 3). Similarly, HIC-1 (n = 36; range, 20-98 years),pió (n = 26; range, 20-98 years), and CALCA (n = 23; range, 10-90

years) had no evidence of methylation in any normal colonie mucosaexamined (Fig. 3; data not shown). Finally, we analyzed the normalcolonie mucosa at several gene loci that have shown no hypermethylation in CRC, including IR, c-ABL, c-MYC, CNP, and liMSH2 (Fig.

3; data not shown). These genes were similarly unmethylated in thenormal colon in 23 patients ranging in age from 10-90 years.

Age-related Methylation Is Modulated by Tissue-specific Factors. Whereas age-related methylation seems to be a common event

within the normal colon mucosa, it is unknown whether this processis present to a similar degree in all aging tissues, or whether it isrestricted to the colon alone. We therefore examined the methylation

D stroma

•epithelium

Fig. 4. Age-related melhylation in various fractions of the colon mucosa. The mucosa

was separated into stroma and epithelial components, and the density of methylalion wascalculated for each fraction. The two colonie fractions for each of the genes are plotted onthe x-axis. and the average density of methylation for each fraction is plotted on the y-axis.Slroma contains primarily stroma! components, whereas the epithelium contains epithelialcells only. Both N33 and MYOD show methylation in both the epithelial and straniaicomponents, whereas the ER CGI primarily shows methylation in its epithelial component.

5492

Research. on December 18, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from

AGINO AND DNA METHYLATION IN CRC

60

504030•

N33•MyoD* ER«

20

10

100

Fig. 5. Concordancy of age-related methylation. Data on N33, MYOD, and ER CGImethylation was available in 16 patients. The density of methylaüonwas plotted as afunction of age for each of these genes in all 16 patients. The data for each gene showsa linear fit by regression analysis that seems broadly similar. The overall density ofmethylation varies for each gene, being highest for the N33 gene and lower for ER andMYOD. The MYOD data reflect the methylation of only the 5' SucII site, SI. The linear

fit curves for MYOD and ER were nearly superimposable, and only one (ER) is shown forclarity.

status of the N33 and ER promoter CGIs in different normal tissuesfrom the same patients. In three patients for whom samples of normalliver and colonie mucosa were available, N33 methylation was higherin the colon than in the liver (74 versus 17%), which was statisticallysignificant (P = 0.01 by t test). By contrast, ER methylation was

much higher in the liver than it was in the normal colonie mucosa (81versus 40%; P = 0.02 by i test). CRC that metastasized to the liver

showed high levels of methylation at all loci, as expected. Thus, theage-related methylation of genes such as N33 and ER is modulated bytissue-specific factors.

We next determined the specific patterns of methylation within thecolonie mucosa itself. The normal colon is composed of both epithelial and stromal components, and it was unclear which fractions areinvolved in age-related methylation. Therefore, we separated the

epithelial and stromal components of colorectal mucosa to determinethe differences in methylation patterns between the components ingenes showing age-related methylation. The N33 gene showed similar

levels of methylation in both epithelia (mean, 60%) and stroma (mean,75%) in four samples. MYOD also showed equivalent levels of methylation in both the epithelial (mean, 20%) and stromal components(mean, 19%; P = 0.8) in two samples. This was in marked contrast to

the ER gene, which showed four times higher methylation in thecolonie epithelium (mean, 31%) than in the stroma (mean, 8%) in fivecases, and this difference was statistically significant (P = 0.02; data

summarized in Fig. 4).Concordancy of Age-related Methylation. ER, N33, and MYOD

all show age-related methylation in the normal colon. Therefore, we

looked at the concordancy of methylation of these three genes. Dataon the methylation status of all three genes was available for 16samples of normal colonie mucosa. As shown in Fig. 5, whereas theamount of methylation varied for each gene, the overall pattern ofage-related methylation was very similar for all three genes. Of the 16

cases, 4 appeared to have concordant deviations from the average; 1case (a 30-year-old man) had relatively high methylation levels for all

three genes, whereas 3 cases (all of whom were over 60 years of age)had relatively low methylation levels for all three genes.

DISCUSSION

In the current study, we have shown a strikingly similar pattern ofage-related methylation in normal colorectal mucosa and frequent

hypermethylation in CRC affecting the N33 gene on chromosome 8pand the MYOD gene on chromosome 1Ip. In fact, of eight genes thatare known to be hypermethylated in CRC, four (N33, MYOD, ER, andIGF2) display this remarkable pattern of age- and neoplasia-related

methylation. Interestingly, all four of these genes are very frequently(>90%) hypermethylated in CRC, whereas the methylation frequencyof the other genes ranges from 5-38%. These data suggest that many

hypermethylation events observed in cancer are related to the fact thatneoplasia begins in aging cell populations.

Although we have limited our observations to CRC in this study,our data should have important implications for all hypermethylationevents in cancer. In fact, in a recent study of hypermethylation ofmultiple genes in glioblastoma multiforme, we found a striking concordance in the methylation status of ER and N33, which were bothmuch more frequently methylated in the tumors of older patients (27).By contrast, just as seen in the colon, pi6 and THBSÃŒmethylation wasnot related to the age of the patients studied.4 Furthermore, HlC-1,

which is frequently methylated in prostate cancer, is also partiallymethylated in the normal prostate as well, and preliminary observations suggest that this methylation is also age related.5 An importantpoint to note in these analyses is that the "normal" tissue adjacent to

cancer may not always be representative of the cell populations thatare predisposed to neoplasia. For example, normal breast tissue islargely composed of supportive stromal cells and is unmethylated atthe HIC-1 locus, whereas purified breast epithelium is highly methylated at this CGI.5 This factor may lead to an underestimation of the

contribution of age-related methylation in normal tissues to the ulti

mate frequency of methylation in neoplasia.These observations raise the question of whether all hypermethy

lation events in cancer are related to initial methylation in normaltissues, whether age related or stem cell related. Using the verysensitive methylation-specific PCR (MSP) technique, which can detect a methylated cell diluted with 104-105 normal cells (28), we wereunable to find methylation of the piò gene in colorectal mucosa;6

similarly, we found no evidence of pi5 gene methylation in purifiedhematopoietic progenitor cells (29). Thus, there seem to be twodistinct types of CGI methylation events in cancer: (a) methylationthat can be detected in both normal aging cells and neoplastic cells;and (b) methylation that can only be detected in the neoplastic cells.

At present, it is unclear why only certain genes show age-relatedmethylation changes. This process seems to be tissue-specific and

involves both genes that are expressed in the colon (ER and N33) andgenes that are not expressed or are expressed at low levels (IGF2 andMYOD). Age-related methylation of the ER gene may provide a

selective growth advantage for the normal colonie cells, but this doesnot seem to hold true for MYOD or N33. MYOD is not expressed inthe normal colon, and N33, which seems to be an oligo-saccharyl-

transferase enzyme, has not been shown to affect the growth of cancercells in vitro. Thus, it is not readily apparent why the loss of N33 geneexpression would provide a selective advantage for tumor growth.Furthermore, age-related methylation does not seem to be a simple

random event with selection of affected cells (based on a growthadvantage), because hypermethylation is not seen for the pl6 locus innormal colon. It seems possible, therefore, that different CGIs havedifferent intrinsic rates of de novo methylation and protection against

4 Q. Li, N. Ahuja, P. C. Burger, and J-P. J. Issa, Methylation and silencing of the

thrombospondin-1 promoter in human cancer, submitted for publication.5 J-P. J. Issa, unpublished observation.6 J. Herman and J-P. J. Issa, unpublished observation.

5493

Research. on December 18, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from

AGING AND DNA METHYLATION IN CRC

methylation, and that the weakly protected islands are those thatdisplay age-related methylation. Thus, age-related methylation of

genes may initially arise from local triggering factors (chromatinstructure or the proximity of Alu sequences) and be modulated byrra/is-acting factors such as described in the APRT gene promoter,

where the CGI is protected from methylation by a cluster of bindingsites for the Spl transcription factor (30). It is also possible that forsome genes, this protection against de novo methylation may be lostduring aging. Additionally, several factors have been shown to modulate CGI methylation in cancers such as defects in DNA repairsystems (19) or different types of carcinogen exposure (31). Theremay also be genetic differences in the rate of age-related methylation

between different individuals, as suggested perhaps by the fact that afew patients in our study appeared to have concordant differences inmethylation for all three genes when compared with the age-adjustedaverage. Age-related methylation of promoter-associated CGIs may

then be related to both endogenous triggering factors and exogenousmodulating factors, such as carcinogen exposure.

Overall, our data suggest that many genes become methylated as afunction of age in the normal colon. Whereas some of the genesaffected are not expressed in the normal colon and may be of littlephysiological relevance, others, such as the ER gene, seem to modulate growth and differentiation in the normal colon (17). Thus, methylation and the loss of expression of these latter genes may impart agrowth-selective advantage to the affected cells, which then become

more susceptible to acquiring further genetic defects that ultimatelylead to neoplasia. Therefore, we propose that age-related methylationconstitutes a type of field-defect in the colon, and that it partiallyexplains the dramatic age-related increase in CRC incidence. This

hypothesis predicts that patients with high levels of methylation intheir colorectal mucosa may be at higher risk for developing colorec-

tal adenomas and cancer, and that normal mucosa from CRC patientswould have higher levels of methylation than mucosa from patientswithout cancer. In the current study, the limited number of patientsstudied precludes such an analysis, which will require a carefullydesigned epidemiological investigation. This hypothesis also providesa potential explanation for the important finding that the reduction ofDNA-methyltransferase in a mouse model of CRC results in a markedreduction in polyp incidence (32). Thus, reducing DNA-methyltransferase levels may inhibit or slow the development of age-related

methylation, thus limiting the extent of the field defect and ultimatelydecreasing the formation of tumors.

In summary, our study shows that age-related methylation is in fact

a frequent event among genes that are hypermethylated in cancer.Age-related methylation may be one of the key events, accounting for

the fact that aging is the most important risk factor for most of thecommon types of cancers in humans.

ACKNOWLEDGMENTS

We thank Dr. R. Bookstein (Canji Inc., San Diego, CA) for kindly providingthe N33 probe. Dr. S. Pearson-White (University of Virginia, Charlottesville,

VA) for kindly providing the MYOD probe, and Dr. S. Hamilton (JohnsHopkins University, Baltimore, MD) for providing samples of normal andneoplastic colon.

REFERENCES

1. Cross. S. H.. and Bird, A. P. CpG islands and genes. Curr. Opin. Genet. Dev., 5:309-314, 1995.

2. Singer-Sam, J.. and Riggs. A. D. X chromosome inaciivation and DNA methylation.EXS, 64: 358-384, 1993.

3. Barlow, D. P. Camelie imprinting in mammals. Science (Washington DC), 270;1610-1613. 1995.

4. Eden. S., and Cedar, H. Role of DNA methylation in the regulation of transcription.Curr. Opin. Genet. Dev.. 4: 255-259. 1994.

5. Baylin. S. B.. Herman, J. G., Graff, J. R., Venino, P. M., and Issa, J. P. J. Alterationsin DNA methylation: a fundamental aspect of neoplasia. Adv. Cancer Res., 72:141-196, 1997.

6. Jones. P. A. DNA methylation errors and cancer. Cancer Res., 56: 2463-2467, 1996.

7. Ershler, W. B.. and Longo, D. L. Aging and cancer: issues of basic and clinicalscience. J. Nati. Cancer Inst., 89: 1489-1497. 1997.

8. tandis. S. H.. Murray, T., Bolden, S.. and Wingo. P. A. Cancer statistics, 1998. CA.Cancer J. Clin., 48: 6-29, 1998.

9. Holliday, R. The inheritance of epigenelic defects. Science (Washington DC), 238:163-170, 1987.

10. Mazin, A. L. Life span prediction from the rate of age-related DNA demethylation innormal and cancer cell lines. Exp. Gerontol.. 30: 475-484, 1995.

11. Cooney. C. A. Are somatic cells inherently deficient in methylation metabolism? Aproposed mechanism for DNA methylation loss, senescence and aging. Growth Dev.Aging. 57: 261-273, 1993.

12. Ghazi. H., Gonzales, F. A., and Jones, P. A. Methylation of CpG-island-containinggenes in human sperm, fetal and adult tissues. Gene (Amst.), 114: 203-210, 1992.

13. Halle, J. P., Schmidt, C.. and Adam, G. Changes of the methylation pattern of thec-myc gene during in vitro aging of imr90 human embryonic fibroblasts. Mutât.Res..316: 157-171, 1995.

14. Uehara. Y.. Ono. T., Kurishita, A., Kokuryu, H., and Okada, S. Age-dependent andtissue-specific changes of DNA methylation within and around the c-fos gene in mice.Oncogene. 4: 1023-1028, 1989.

15. Choi. E. K.. Uyeno. S.. Nishida, N.. Okumoto, T.. Fujimura. S., Aoki, Y.. Nata, M..Sagisaka, K., Fukuda, Y.. Nakao. K., Yoshimoto. T., Kim, Y. S., and Ono, T.Alterations of c-fos gene methylation in the processes of aging and tumorigenesis inhuman liver. Mutai. Res., 354: 123-128, 1996.

16. Mays-Hoopes, L. L. DNA methylation in aging and cancer. J. Gerontol., 44: 35-36,

1989.17. Issa. J. P.. Ottaviano, Y. L., Celano, P., Hamilton, S. R., Davidson, N. E., and Baylin,

S. B. Methylation of the oestrogen receptor CpG island links ageing and neoplasia inhuman colon. Nat. Genet.. 7: 536-540, 1994.

18. Issa. J. P. J., Vertino. P. M.. Boehm, C. D., Newsham, 1. F., and Baylin, S. B. Switchfrom mono-allelic to bi-allelic human IGF2 promoter methylation during aging andcarcinogenesis. Proc. Nati. Acad. Sci. USA, 93: 11757-11762. 1996.

19. Ahuja, N., Mohan. A. L.. Li. Q.. Stoiker, J. M., Herman. J. G.. Hamilton. S. R..Baylin, S. B.. and Issa, J. P. Association between CpG island methylation andmicrosatellite instability in colorectal cancer. Cancer Res., 57: 3370-3374, 1997.

20. MacGrogan, D.. Levy, A., Bova. G. S., Isaacs. W. A., and Bookstein, R. Structure andmethylation-associated silencing of a gene within a homozygously deleted region ofhuman chromosome band 8p22. Genomics, 35: 55-65, 1996.

21. Venino, P. M., Issa, J. P., Pereira-Smith, O. M.. and Baylin, S. B. Stabilization ofDNA methyltransferase levels and CpG island hypermethylation precede SV40-induced immortalization of human fibroblasts. Cell Growth Differ., 5: 1395-1402,

1994.22. Issa, J. P. J., Zehnbauer, B. A.. Kaufmann, S. H.. Biel. M. A., and Baylin. S. B. HIC1

hypermethylation is a late event in hematopoietic neoplasms. Cancer Res., 57:1678-1681. 1997.

23. Rideout, W. M., Ill, Eversole-Ciré.P., Spruck. C. H.. Ill, Hustad. C. M.. Coetzee.

G. A., Gonzales. F. A., and Jones, P. A. Progressive increases in the methylationstatus and heterochromatinization of the myoD CpG island during oncogenic transformation. Mol. Cell. Biol.. 14: 6143-6152, 1994.

24. Herman. J. G., Merlo. A.. Mao. L., Lapidus, R. G.. Issa. J. P.. Davidson. N. E..Sidransky. D.. and Baylin. S. B. Inactivalion of the CDKN2/pl6/MTSI gene isfrequently associated with aberrant DNA methylation in all common human cancers.Cancer Res.. 55: 4525-4530. 1995.

25. Wales, M. M., Biel. M. A., el Deiry. W.. Nelkin, B. D.. Issa, J. P., Cavenee, W. K.,Kuerbitz. S. J.. and Baylin, S. B. p53 activates expression of HIC-l. a new candidatetumour suppressor gene on I7pl3.3. Nat. Med., /: 570-577. 1995.

26. Baylin, S. B.. Fearon. E. R.. Vogelstein. B., de Bustros. A.. Sharkis, S. J., Burke, P. J..Staal. S. P.. and Nelkin. B. D. Hypermethylation of the 5' region of the calcitonin

gene is a property of human lymphoid and acute myeloid malignancies. Blood, 70:412-417, 1987.

27. Li. Q., Jedlicka. A.. Ahuja. N.. Gibbons, M. C., Baylin, S. B., Burger, P. C.. and Issa.J. P. J. Concordant methylation of the ER and N33 genes in glioblastoma multiforme.Oncogene. 16: 3197-3202, 1998.

28. Herman, J. G.. Graff, J. R., Myohanen, S., Nelkin, B. D.. and Baylin, S. B. Methy-lation-specific PCR: a novel PCR assay for methylation status of CpG islands. Proc.Nati. Acad. Sci. USA, 93: 9821-9826. 1996.

29. Herman, J. G., Civin, C. I., Issa. J. P. J., Collector. M. !.. Sharkis, S. J., and Baylin,S. B. Distinct patterns of /)/5""""1 and pl6INK4* characterize the major types of

hématologiemalignancies. Cancer Res., 57: 837-841, 1997.

30. Turker, M. S.. and Bestor, T. H. Formation of methylation patterns in the mammaliangenome. Mutât.Res.. 386: 119-130, 1997.

31. Issa. J. P. J.. Baylin. S. B.. and Belinsky. S. A. Methylation of the estrogen receptorCpG island in lung tumors is related to the specific type of carcinogen exposure.Cancer Res., 56: 3655-3658, 1996.

32. Laird, P. W., Jackson-Grusby, L., Fazeli, A., Dickinson. S. L., Jung, W. E., Li, E.,

Weinberg. R. A., and Jaenisch. R. Suppression of intestinal neoplasia by DNAhypomethylation. Cell, 81: 197-205. 1995.

5494

Research. on December 18, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from

1998;58:5489-5494. Cancer Res   Nita Ahuja, Qing Li, Avinash L. Mohan, et al.   Aging and DNA Methylation in Colorectal Mucosa and Cancer

  Updated version

  http://cancerres.aacrjournals.org/content/58/23/5489

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/58/23/5489To request permission to re-use all or part of this article, use this link

Research. on December 18, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from