CANCER RESEARCH 57, 594-599. February 15. 1997]

Advances in Brief

Identification and Characterization of Differentially Methylated Regions of GenomicDNA by Methylation-sensitive Arbitrarily Primed PCR1

Mark L. Gonzalgo,2 Gangmng Liang,2 Charles H. Spruck III,@Jean-Marc Zingg, William M. Rideout ifi,―andPeter A. Jones5

Department of Biochemistry and Molecular Biology, Urological Cancer Research Laboratory, University of Southern California/Norris Comprehensive Cancer Center, Universityof Southern California. School of Medicine, Los Angeles, California 90033 [M. L G.. G. L, J-M. Z, P. A. J.]

Abstract

We have developeda simple and reproducible fingerprinting methodfor screeningthegenomefor regionsof DNA thathavealteredpatternsofDNA methylationassociatedwith oncogenictransformation.Restrictionenzymes with different sensitivities to cytosine methylation in their rec

ognition sites were used to digest genomic DNAs from primary tumors,cell lines, and normal tissuesprior to arbitrarily primed PCR amplification.Fragmentsthatshoweddifferentialmethylationwereclonedandsequencedafter resolvingthe PCR productson high-resolutionpolyacrylamidegels.The clonedfragmentswerethenusedasprobesfor Southernanalysisto confirm differential methylationof theseregionsin colontissues and cell lines. Forty-four DNA fragments associated with a total of

five differentregionsof genomicDNA containingmethylationsitesweredetectedin 10matchedsetsof normalandtumorcolonDNAsand7 coloncancercell lines.A novelCpG islandwasalsoisolatedthat wasfoundtobe frequently hypermethylated in bladder and colon tumors. We havedemonstratedthat thistechniqueisa rapidandefficientmethodthatcanbeusedtoscreenfor alteredmethylationpatternsin genomicDNA andtoisolatespecificsequencesassociatedwith thesechanges.

Introduction

Approximately 1% of cytosines in vertebrate DNA are methylatedat CpG dinucleotides ( 1). The presence of 5-methylcytosine at CpGdinucleotides may contribute to tumorigenesis either by generatingpoint mutations or by altering gene expression (2, 3). DNA methylation has been shown to be essential for normal embryonic development (4), and alterations in both the levels and patterns of methylationduring transformation may affect the normal regulation of gene expression (3, 5). The number of CpG dinucleotides in the humangenome is underrepresented by a factor of 5 (6), presumably due to the

conversion of methylated cytosine to thymine via deamination (7);

however, certain areasof the genome do not show such suppression,and these areasare known as CpG islands (1, 8, 9). CpG islands arecommonly associatedwith housekeepinggenesand may regulate theirtranscriptional activities. Hypomethylation of these regions is usuallyassociated with gene activity, whereas methylation of CpG islandssuch as those on the inactive X chromosome (10) may suppresstranscription. Methylation of CpG islands typically occurs only inparentally imprinted genes (1 1, 12) and in genes that have been

Received I 1/14/96; accepted 1/2/97.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 United States Public Health Service Grant R35 CA49758 from the

National Cancer Institute. W. M. R. is funded by a Burroughs-Wellcome Fund HitchingsElion Postdoctoral Fellowship.

2 These authors contributed equally to this work.

3 Current address: The Scripps Research Institute, 10666 North Torrey Pines Road,

MB-7, La Jolla, CA 92037.4 Current address: Whitehead Institute, Massachusetts Institute of Technology, Nine

Cambridge Center, Cambridge, MA 02142.S To whom requests for reprints should be addressed. Phone: (213) 764—0816; Fax:

(213) 764—0102.

silenced on the inactive X chromosome. Autosomal CpG islands in

nonimprinted genesremain unmethylated in the normal state but maybecome methylated upon oncogenic transformation(13, 14). It hasbeendemonstratedpreviously that alterations in the methylation statusof CpG islands during transformation are correlated with transcrip

tional changesin a number of genesassociatedwith growth regulation(15—17),and it is possible that other genes involved in cell cyclecontrol may also be influenced by these epigenetic mechanisms.

The ability to detect methylation changes associated with oncogenic transformationis of critical importancein understandinghowDNA methylation may contribute to tumorigenesis. A variety oftechniques have been previously developed that can be used to investigate patterns of DNA methylation, such as genomic sequencing (18,

19), PCR-basedmethylation analysis (20), and Southern blotting (21).However, a fundamental limitation of both genomic sequencing andPCR-based methylation analysis is that these methods require someknowledge of the DNA sequencebeing studied. Techniques includingSouthern blotting, high-performance liquid chromatography analysis(22), and the methyl-acceptance assay (23) have been used to studyglobal levels of DNA methylation. None of these methods, however,can be used to isolate specific and unknown DNA sequences fromgenomic DNAs that are differentially methylated between normal andtumor tissues. RLGS6 (24) can be used to identify specific methylation differences between genomes; however, this technique is ratherlabor-intensive, and deletions, amplifications, and rearrangementsmay also be resolved. AP-PCR has been used previously to identifygenetic changes including loss of heterozygosity and chromosomalgains in colorectal tumors (25, 26). The application of this method foridentifying and characterizing methylation differences between genomes, however, has not been demonstrated. We describe a methylation-sensitive AP-PCR technique that uses methylation-sensitive restriction digestion coupled with AP-PCR to identify random butspecific methylation changes at multiple sites in genomic DNA in a

rapid and efficient manner.

Materials and Methods

DNA IsolationfromTissuesandCellLines. Matchedpairsofnormalandtumorcolonandbladderspecimenswereobtainedfrom patientstreatedat theLos Angeles County-University of Southern California Medical Center and the

University of Southern California/Norris Comprehensive Cancer Center (Los

Angeles,CA). Mucosaltissuewasfirst removedfrom surroundingmuscleandfat, and DNA was then isolated using standard procedures by treatment with

proteinase K and phenol extraction (27). Seven colon cancer cell lines(SW837,SW480,HT-29, LoVo, SW48, HCT 116, and HCT-15) were obtamedfrom the AmericanType CultureCollectionand grown in the appropriate medium as suggested by the American Type Culture Collection. DNAwas isolated from these cell lines in the same manner as described previously(27).

6 The abbreviations used are: RLGS, restriction landmark genomic scanning; AP-PCR,

arbitrarily primed PCR.

594

on May 4, 2020. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

METHYLATION-5ENSITIVE AP-PCR

MLG2 primer was designed with a 3' end complementary to therecognition sequenceof HpaII and MspI. The possible formation ofprimer dimers produced as a result of the interaction between thepalindromic 3' ends of the primers (-CCGG) did not inhibit theamplification of a large number of DNA fragments (see Fig. I). NoPCR products would be produced if the region of DNA between twoprimer annealing sites contained at least one unmethylated HpaII siteor if at least one of the primer annealing sites contained an unmethylated CCGG sequence. DNA digested with RsaI only and DNAdigested with RsaI + MspI served ascontrols for determining whetherbands observed in the AP-PCR of RsaI + HpaII-digested DNA weredue to differential methylation of CCGG sequenceswithin the regionof amplification. AP-PCR products were considered to contain one ormore CCGG sites if DNA digested with only RsaI produced a PCRproduct and if RsaI + MspI digestion@yielded no PCR productbecause MspI should cut its recognition site regardless of the meth

ylation statusof the internal cytosine and prevent DNA amplification.Fig. 1A shows DNA bands generated by methylation-sensitive

AP-PCR on matched pairs of normal and tumor colon DNAs. At leastfive fragments (A—E)of different sizeswere present in RsaI digestionsof normal and tumor DNAs, absent in RsaI + MspI digestions of both

normal and tumor DNAs, present in most of the RsaI + HpaIIdigestions of normal DNAs, and absent in RsaI + HpaII digestions oftumor samples. These findings suggested that these bands were hypomethylated in colon tumor DNA compared to normal colonicepithelium. Fragments were excised from dried polyacrylamide gels,reamplified, cloned, and sequenced.Interestingly, none of these fragments contained HpaII sites located between the primer annealingsites. The absenceof internal HpaII sites in fragments A—Esuggestedthat one or both of the primer annealing sites contained CCGGsequences that were differentially methylated between normal andtumor DNA.

Methylation-sensitive fingerprints for seven colon cancer cell linesare shown in Fig. lB. All of the major PCR products previously

identified in the AP-PCR analysis of normal and tumor colon DNAswere also identified in this separate experiment. All five AP-PCRfragments that were previously identified as being hypomethylated incolon tumor DNA compared to normal colon tissue were also foundto be unmethylated in some of the colon cancer cell lines examined.Each of these AP-PCR bands was isolated, and all fragments wereconfirmed to be the same PCR products that were previously identifled in the AP-PCR gel of normal and tumor colon DNAs by sequenceanalysis. A total of 44 DNA fragments associatedwith approximatelyfive different regions of genomic DNA containing methylation siteswere identified by methylation-sensitive AP-PCR in 10 matched sets

of normal and tumor colon tissues and 7 colon cancer cell lines. Ofthese fragments, 27 (61%) were differentially methylated in normal

and tumor DNAs and 17 (39%) were found to be unmethylated indifferent colon cancer cell lines.

Identificationof Novel CpG Island Hypermethylationin Tumors by Methylation-sensitive AP-PCR. We also used a combination of 2 primers (MGCO + MGF2) in the methylation-sensitiveAP-PCR technique to screen for methylation differences in matched

pairs of normal and tumor colon and bladder samples. Fig. 2 showsbands generated by AP-PCR, which amplified a region of DNA thatwas hypermethylated in tumors compared to normal tissues. Bandingpatterns suggestedthat this region of DNA, which was approximately580 bp in length, was hypermethylated in three of six (50%) colontumors and three of seven (43%) bladder tumors. Further analysis ofthis band revealed the presence of two internal HpaII sites, a G+Ccontent of 57%, and an observed/expected CpG ratio of 0.62, whichfulfills the sequencecriteria for being a CpG island (9). GenBank database searchesproduced no significant matches with any known se

595

RestrictionEnzymeDigestionofGenomicDNA. Two @geachofnormaland tumor DNA from tissuesor DNA from colon cancercell lines wereseparatelydigestedwith either20 unitsof RsaI,20 unitseachof RsaIandthemethylation-sensitiverestrictionenzymeHpaH,or 20 unitseachof RsaIandMspI (Boebringer Mannheim, Indianapolis, IN) at 37°Cfor 16 h. Doublerestrictionenzymedigestionswereusedto reducethe numberof PCRfragmentsandpotentialartifactsthat might be amplified in the AP-PCR.HpaIIdoesnot cut DNA if the internal cytosineof its restrictionsite (CCGG) ismethylated,whereasMspI is insensitiveto the methylation statusof theinternalcytosine.Restrictionenzymeswereheatinactivatedby incubatingthereactionsat 65°Cfor 20 mm andthenstoredat —20°C.

Methylation-sensitiveAP-PCR. Restriction-digestedDNA (200 ng) wasamplified usingAP-PCR(25, 26) with a singleprimer (MLG2: AAC CD'CACCD' AACCCCGG)oracombinationof2primers(MGCO:AACCD'CACCD' AACCGCOC)+ (MGF2:AACCCTCACCCTAACCCGCG).PCRreactionswereperformedin a total volumeof 25 pAunderthefollowingconditions: 10 mM Tris-HC1(pH 8.3), 1.5 mM MgCI2, 50 mistKC1,0.1%gelatinperml, 200 @LMeachof thefour deoxynucleotidetriphosphates,2 @Ciof 33P-labeleddATP, 25 pmol of primer, and 0.8 units of Taq polymerase(BoehringerMannheim).Fivecyclesat low-stringencyconditionsof 94°Cfor30 5,40°Cfor 60 5,72°Cfor 90 s weredonebeforeincreasingthestringencyofthe PCRto 94°Cfor 15s,55°Cfor 15s,72°Cfor 60sfor another30cycles.PCR products were resolved on high-resolution 5% polyacrylamide gels underdenaturingconditions(7 Murea),dried,andexposedto autoradiographicfilm(Amersham Corp., Cleveland, OH) overnight. Sequence data for all DNA

fragmentsdescribedin this paperhavebeensubmittedto GenBank(accessionnumbers:Yl0088-Y10093).

Isolation and Sequencing of DNA Fragments. Candidate bands thatappearedto be differentially methylatedby AP-PCRanalysiswere excisedfrom driedpolyacrylamidegelsandplacedin a microcentrifugetubecontaining 50 @dof sterileH20. Themicrocentrifugetubewasheatedto 80°Cfor 10mm andvortexedto facilitatedissolutionof DNA. Theeluate(2 p1)wasthenused in a PCR reaction with the same primer(s) used in the original AP-PCRto generatesufficientamountsof templatefor plasmidcloningandsequencing.PCR ingredientswere the sameas previouslydescribedwith the followingamplificationparametersfor a totalof 30cycles:94°Cfor 60 s,56°Cfor 30s,72°Cfor 60 s. PCRproductswereclonedinto plasmidvectorsusingtheTAcloning kit (Invitrogen,SanDiego,CA). Multiple coloniesderivedfrom thecloning procedurewere sequencedaccordingto manufacturer'sinstructionsfor double-strandedplasmid DNA using the SequenaseVersion 2.0 kit(Amersham Corp.).

Southern Blot Analysis of DNA from Tissues and Cell Lines. Ten .tgeachof normal and tumor DNA from tissuesor DNA from cell lines wereseparatelydigestedwith either40 units of RsaI,40 units eachof RsaI andHpall, or 40 unitseachofRsal andMspI (BoehringerMannheim,Indianapolis,IN) at 37°Cfor 16 h followed by an additional 1 unit of each enzyme per @gof DNA for 8 h. DigestedgenomicDNA waselectrophoresedon 1%agarosegels and Southerntransferredto Zeta-Probe(Bio-Rad,Hercules,CA) membranes overnight. Cloned DNA fragments that were previously isolated fromAP-PCR polyacrylamide gels were subsequently used as probes for hybridization to thesefilters. Approximately 100ng of plasmidinsert DNA were32P-labeledby randomprimingandusedto probethefilters.All hybridizationswere performed in 1% BSA, 1 mrsi EDTA, 0.5 M NaHPO4, 5% SDS, 50%

formamideat 42°Cfor 24 h. Membraneswerewashedtwice at roomtemperature with 0.2X SSC,0.5% SDS and exposedto autoradiographicfilm at—80°C.

Results

Methylation-sensitive AP-PCR Identifies Differences betweenNormal and Tumor Colon Samples and Cell Lines. Restrictionenzyme digestionswith RsaI, RsaI + Hpall, and RsaI + MspI wereused as a rapid and efficient means ofpreparing DNA for methylation

sensitive AP-PCR analysis. The RsaI restriction enzyme was used inaddition to HpaII and MspI to generate smaller fragments of DNAprior to AP-PCR and to reduce the number of potential artifacts thatmight be generated. AP-PCR was performed on digested DNA usinga single 20-mer primer (MLG2) under low-stringency conditions. The

on May 4, 2020. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

p A

METHYLATION-SENSfl'IVE AP-PCR

A

700 bp

500bp

400bp

300bp

200 bp

BPatient

Colonc&Iline

A

II,

“p

. @I

11

Fig. 1. A, methylation-sensitive fingerprints of matched pairs of normal and tumor colon samples using primer MLG2 in the AP-PCR reactions after a 16-h restriction enzymedigestion with 20 units each of RsaI. RsaI + HpaII, or RsaI + MspI. Bands that appeared to be differentially methylated between matched normal and tumor samples are indicatedby arrows. N, normal DNA; T, tumor DNA. B, methylation-sensitive fingerprints of colon cancer cell line DNAs using primer MLG2 in the AP-PCR reactions after a 16-h restrictionenzyme digestion with 20 units each of RsaI, RsaI + HpaII, or RsaI + MspI. Bands that appeared to be unmethylated in cell lines are indicated by arrows. Major band groups thatwere isolated and sequencedare indicated (A—E).R, RsaI digestion; H, RsaI + HpaII digestion; M, RsaI + MspI digestion.

quences. Our findings demonstrate that methylation-sensitive AP- plasmid clones were sequencedfrom each of the bands isolated fromPCR canthereforebe usedfor the identificationof novelCpG islands the AP-PCR gelsof coloncell line DNAs. Theseresultsindicatedthethat may become differentially methylated during tumorigenesis. presenceof only one speciesof PCR product for each of the isolated

Purity of FragmentsIsolatedfrom AP-PCR Gels. Fragments fragmentsbasedon sequenceanalysisof multipleplasmidclones.that were of potential interest were excised from the AP-PCR poly- Southern Analysis Confirms Differential Methylation. Confiracrylamide gels and reamplified before cloning for DNA sequence mation of the potential methylation differences identified by methyanalysis to assess their purity. The five different PCR fragments lation-sensitive AP-PCR by another technique was necessary to suppreviously identified as being differentially methylated in both pri- port the validity of the assay. Southern analysis was performed tomary tissues and colon cancer cell lines were excised and reamplifled determine whether PCR fragments representing putative hypomethy

with the sameprimer used in the original AP-PCR reactions. Multiple lated regions of DNA in colon tumors and cell lines resulted from596

,-.

S.S

I, @‘.@ @:z

..s

S

S

0

I, V@.@ —

-@ -@

B-@@ --

00

C.@

..,@ -@S@

aa

a.

a.0

a.C

C

000S

E

.w

00C

@•::@@B

ma

D

I

0

C

0

E

200bp

1 2 3 4 5 6 7 8 9 10RHM RHM RHM RHM RHM RHM RHM RHM RHM RHMTNTNTN TNTNTN TNTNTN TNTNTN TNTNTN TNTNTN TNTNTN TNTNTN TNTNTN TNTNTN

It@@@

$

@.

Pp.p.@

on May 4, 2020. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

11 12 13 14 15 16RHM RHM RHMRHMRHMRHM

NTNTNT NTNTNT NTNTNTNTNTNTNTNTNTNTNTNT

@0bP_F@.@_i-—

Bladder

5@ 4 1R H M R H@WR H M

r@@ fFi @Fin@@@@@METHYLATION-5ENSmvE AP-PCR

Colon

Fig. 2. Methylation-sensitive fingerprints ofmatched pairs of normal and tumor colon andbladdersamplesusingprimersMGCO+ MGF2inthe AP-PCR reactionsafter a 16-h restrictionenzymedigestionwith 20 unitseachof RsaI,RsaI + HpaII, or RsaI + MspI. Bands that appeared to be differentially methylated between

matched normal and tumor samples are indicatedbyarrows.N, normalDNA;T, tumorDNA;R,RsaI digestion; If, RsaI + HpaII digestion; M,RsaI + MspI digestion. 17 181920212223RHMRHMRHMRHMRHMRHMNTNTNTNTNTNTNTNTNTNTNTNTNTNTNTNTNTNT,w;;-w_

@ Iauthentic methylation changes. Several of the AP-PCR fragmentsidentified by this technique did not contain internal HpaII sites asdetermined by sequence analysis, and it was believed that themethylation changes observed on the AP-PCR gels were due to

changes iti iiicthylation of putative CCGG site(s) to which theprimers may have annealed. Fig. 3A shows the results obtainedfrom Southernanalysisof threematchedsetsof normalandtumorcolon DNA@ previously analyzed by AP-PCR (Fig. 1). Thesesamples were digested with either RsaI, RsaI + HpaII, orRsaI + MS/)l and plasmid cloned fragment A was used as a probefor hyhridiiation to theseDNAs. The Southernblot clearly showedbands of strutiiz intensity at approximately 1500 bp in normal andtumor DNA digested with RsaI only, absenceof a strong hybridization signal at 1500 bp in tumor samples digested withRsaI + HpaII (patients 1 and 5), and presence of a band at 1500 bpin all of the matched normal colon DNAs. A 1500-bp band waspresentin the RsaI + HpaII-digested lane of colon tumor DNA forpatient 4. All normal and tumor DNAs digestedwith RsaI + MspIproduced hybridization signals at approximately 900 bp. Bandingpatterns were consistentwith the hypomethylation of the genomicDNA regionassociatedwith fragmentA in at least66% (two ofthree) of these colon tumors compared to normal tissues.

A

Southern analysis also showed similar hybridization patterns forcolon cancer cell lines probed with band A (Fig. 3B). Absence of al500-bp band and presence of a 900-bp band in the RsaI + HpaIIdigested lanes of HCT-l 5, HT-29, and SW837 DNAs indicated thatthis region was not methylated in these cell lines. These resultsconfirmed that fragment A was unmethylated in 75% (three of four)of colon cancer cell lines previously analyzed by methylation-sensitive AP-PCR. The design of the primer used for AP-PCR made itpossible to amplify genomic DNA fragments that could produce PCRproducts with flanking CCGG sequencesdefined by the primer annealing sites. Southern analysis of both primary tissues and cell linesrevealed the presenceof a band in the RsaI + MspI-digested lanes thatwas approximately 900 bp in size. The size of this fragment on theSouthern blot matched the size of the PCR product that was producedby methylation-sensitive AP-PCR analysis of theseDNAs. The resultsof the Southern analysis of both primary tissues and cell lines mdicated that there must be two CCGG sites at the termini of thesefragments that were hypomethylated in the genomic DNAs examinedby methylation-sensitive AP-PCR. The clear differences observed bySouthern analysis using fragment A as a probe confirmed that methylation-sensitive AP-PCR is capable of identifying methylationchanges at random but specific sites in genomic DNA.

B

PatientColoncellline

HCT-15 HT-29 SW8V SW480

RHM RHM RHM RHM

597

RHMNTNTNT

580b@jis TW

l500bp

900bp

@,.

.@

Fig. 3. SOUtl@Crl1blot analysis of primary colon tissues and colon cancer cell line DNAs using isolated fragment A as a probe. A, hybridization patterns indicating hypomethylationof fragmentA in tumor. comparedto normalcontrols.B, hybridizationpattemsindicatinglack of methylationof fragmentA in threeof four coloncancercell linespreviouslyanalyzedh@niethv]atio,i.@ei@tivc AP-PCR. Ten p@geach of matched normal and tumor colon DNA were digested separately with either RsaI (R), Rsal + HpaII (H), or RsaI + MspI (M),cli-ctr@phore@cd. blotted, and probed with plasmid cloned bands to confirm methylation changes.

on May 4, 2020. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

METHYLATION-SEN5ITIvE AP-PCR

digestion and AP-PCR to identify random but specific methylationchanges at multiple sites in genomic DNA.

Reproducibility and artifacts are important considerationswhendealing with any PCR-based assay. Because many different DNAfragments may be generatedby methylation-sensitive AP-PCR, it wascritical to determine which species resulted from true methylationchanges. The use of a methylation-sensitive restriction enzyme withan isoschizomer was important for controlling for the presence ofPCR fragments unrelated to methylation changes.We have also useda double restriction enzyme strategy with RsaI combined with eitherHpaII or MspI to generate smaller fragments of DNA that potentiallycontain cccx; ait@. Digestion of DNAs with RsaI would also reducethe number of possible regions of DNA that could be amplified byAP-PCR. The use of other restriction enzymes in addition to Hpafland MspI in a doubledigestionstrategycould also be usedto detectadditional regions of genomic DNA, which may contain differentiallymethylated sites. Some of the fragments that were amplified byAP-PCR showed no differences between normal and tumor digestedDNAs and bands were present in both the normal and tumor Hpalland MspI digestions. These types of fragments clearly did not represent methylation changes,but they did serve as controls for assessingthe consistency of PCR reactions between samples. Furthermore, byusing fragments identified by methylation-sensitive AP-PCR asprobes on Southern blots containing RsaI, RsaI + HpaII, and

RsaI + MspI-digested DNA, we confirmed that bands isolated by thismethodrepresentedtrue methylationdifferencesbetweennormal andtumor samples.

Methylation-sensitive AP-PCR can be used to identify and isolateregions of genomic DNA that have undergone methylation changesassociated with tumorigenesis, much as differential display RT-PCR(29, 30) can be used to identify and isolate differentially expressedsequences.In the present study, we have isolated regions of DNA thatshowed methylation differences between normal and tumor colontissuesand cell lines. Using various primer combinations and a doublerestriction enzyme digestion strategy, we have detected regions ofDNA hypo- and hypermethylationin the samplesanalyzed.Previousstudies have reported that changes such as global hypomethylationand regional hypermethylationof CpG islands are common eventsthat occur during tumorigenesis (reviewed in Ref. 2). Hypomethylation of genessuchashumangrowthhormone,y-globin, and a-globinin colon cancers has also been described (28).

In additionto isolatingunknown sequencesassociatedwith methylation changes in genomic DNA, methylation-sensitive AP-PCR canalso be applied to rapidly estimate the variability of methylation atmultiple sites between different cell lines or between normal andtumor tissues.We haveusedmethylation-sensitiveAP-PCR to screenfor methylation differences between normal and tumor DNA. Basedon the numberof putativemethylationsitesdetectedby methylationsensitive AP-PCR using a single primer in matched setsof normal andtumor colon DNAs, the major differences observed were bands representing putative regions of hypomethylation in tumors compared tonormal samples. A total of 44 DNA fragments associated with ap

proximatelyfive different regionsof genomicDNA containingmethylation siteswere identified in a simpleandefficient mannerthat hasnot been previouslydescribed.Furthermore,we have identified andisolated a novel CpG island that was frequently hypermethylated inboth bladder and colon tumors by using a dual primer strategy forAP-PCR. Different primer sets were used to demonstrate that otherdifferentially methylated fragments could be identified by this technique. It is interesting to note that regardless of the primer combina

tion used for AP-PCR, identification of hypomethylatedregionsofDNA was consistentlyassociatedwith CpG-poor sequences,whereasidentificationof hypermethylatedregionsof DNA yielded CpG-rich

598

The remaining 4 fragments (B—E)identified by AP-PCR were alsoused as probes for Southern analysis. Fragments B and C producedhigh background on Southern blots, which could not be clearly evaluated perhaps due to the presence of sequences associated withrepetitive elements or pseudogenes. Blots of colon cancer cell lineDNAs probed with fragment D produced methylation patterns consistent with those observed by AP-PCR, and fragment E was not usedasa probe becauseit was known to contain homology to Alu repetitivesequences. A total of three other DNA fragments that were isolated

from AP-PCR polyacrylamide gels but showed no methylation differences between normaland tumorcolon tissues and cell lines werealso used as probes on Southern blots to verify the methylationpatterns identified by AP-PCR. Hypermethylation of Hpall siteswithin the novel CpG island in tumors previously detected by APPCR was also confirmed in colon tissues by using the cloned DNAfragment as a probe (data not shown).

Discussion

Digestion of genomic DNA with methylation-sensitive restrictionenzymes followed by low-stringency AP-PCR allowed us to identifybands that were differentially methylated in normal and tumor colonDNAs and cell lines. The random association of primers with genomicDNA at low annealing temperatures would be expected to generatemultiple PCR fragments. If two primer annealing sites flanked at leastone unmethylated HpaII restriction site, then no PCR product wouldbe expected to be generated due to HpaII cutting. The presence ofmethylation at HpaII sites located between primers during AP-PCRwould allow for the amplification of certain regions of DNA. TheMLG2 primer was designed with a CCGG sequenceat its 3' end toincrease the probability of primer annealing to HpaII and MspIrestriction sites under low-stringency PCR conditions. This primerdesign increases the potential number of methylation sites that can be

analyzed by the preferential annealing of primers to methylatedCCGG sequences,asdemonstratedby Southernanalysisof primarytissues and cell lines (Fig. 3). Absence of methylation at CCGG sitesto which the primers may anneal would produce no amplificationproduct due to HpaII cutting, whereas primer annealing to thesesequencescould occur in the presenceof methylation becauseof theinability of HpaH to cut methylated CCGG. Methylation changesidentified by this technique are therefore not limited to internal CCGGsites within isolated PCR products, but may also include CCGGsequenceslocated at primer annealing sites. The possible formation ofprimer dimers when using a primer with a 3' ending in -CCGG did notinhibit the AP-PCR amplification of a large number of DNA fragments (Fig. 1).

Methylation-sensitive AP-PCR is a simple and rapid method thatcan be used to screen for methylation changes and to isolate specificfragments of DNA associatedwith thesechangesin normal and tumorDNA. Southern analysis can be used to study methylation differencesbetween known genes, but its application for isolating specific andunknown regions of DNA associated with methylation changes islimited. In contrast to Southern blotting, which requires large amountsof DNA (5—10p.g), methylation-sensitive AP-PCR can detect methylation changes in as little as 200 ng of genomic DNA. Other techniques such as conventional PCR-based methylation analysis andgenomic sequencing require at least partial knowledge of the DNAsequencebeing studied. Global levels of methylation can be studiedusing more labor intensive techniques such as high-performance liquid chromatography analysis and RLGS, but methylation-sensitiveAP-PCR can be used to screen for genome-wide methylation changesin a simple and rapid manner. This technique usesrestriction enzyme

on May 4, 2020. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

METHYLATION-SENSITIVEAP-PCR

during the establishmentof immortal cell lines. Proc. Natl. Acad. Sci. USA, 87:6117—6121,1990.

15. Greger, V., Passarge,E., Hopping, W., Messmer, E., and Horsthemke, B. Epigeneticchanges may contribute to the formation and spontaneous regression of retinoblastoma. Hum. Genet., 83: 155—158,1989.

16. Merlo, A., Herman, J. G., Mao, L., Lee, D. J., Gabrielson,E., Burger, P. C., Baylin,S. B., and Sidransky, D. 5' CpG island methylation is associated with transcriptionalsilencing of the tumour suppressorpl6/CDKN2/MTSI in human cancers. Nat. Med.,1: 686—692,1995.

17. Gonzalez-Zulueta, M., Bender, C. M., Yang, A. S., Nguyen, T., Beast, R. W., VanTornout,J.M., andJones,P.A. Methylationof the5' CpGislandof thep16/CDKN2tumor suppressorgene in normal and transformed human tissues correlates with genesilencing. Cancer Res., 55: 4531-4535, 1995.

18. Church, G. M., and Gilbert, W. Genomic sequencing. Proc. Natl. Acad. Sci. USA, 81:1991—1995,1984.

19. Pfeifer, G. P., Steigerwald,S. D., Mueller, P. R., Wold, B., andRiggs,A. D. Genomicsequencing and methylation analysis by ligation mediated PCR. Science (WashingtonDC), 246: 810—813,1989.

20. Singer-Sam, J., LeBon, J. M., Tanguay, R. L., and Riggs, A. D. A quantitativeHpaII-PCR assayto measuremethylation of DNA from a small number of cells.Nucleic Acids Res., 18: 687, 1990.

21. Southern, E. M. Detection of specific sequencesamong DNA fragments separatedbygel electrophoresis. J. Mol. Biol., 98: 503—517,1975.

22. Gama-Sosa,M. A., Slagel,V. A., Trewyn, R. W., Oxenhandler,R., Kuo, K. C.,Gehrke,C. W., andEhrlich, M. The 5-methylcytosinecontentof DNA from humantumors. Nucleic Acids Res., 11: 6883—6894,1983.

23. Bestor,T. H., andIngram,V. M. Two DNA methyltransferasesfrom murineerythroleukemia cells: purification, sequence specificity, and mode of interaction withDNA. Proc. Nail. Acad. Sci. USA, 80: 5559—5563,1983.

24. Hatada, I., Hayashizaki, Y., Hirotsune, S., Komatsubara, H., and Mukai, T. Agenomicscanningmethodfor higher organismsusingrestrictionsitesas landmarks.Proc.Nail. Acad.Sci. USA, 88: 9523—9527,1992.

25. Welsh, J., and McClelland, M. Fingerprinting genomes using PCR with arbitraryprimers.NucleicAcids Res.,18: 7213—7218,1990.

26. Peinado, M. A., Malkhosyan, S., Velazquez, A., and Perucho, M. Isolation andcharacterizationof allelic lossesandgainsin colorectaltumorsby arbitrarily primedpolymerase chain reaction. Proc. Natl. Acad. Sd. USA, 89: 10065—10069,1992.

27. Michalowsky,L. A., andJones,P.A. Genestructureandtranscriptionin mousecellswith extensively demethylated DNA. Mol. Cell Bio., 9: 885—892,1989.

28. Feinberg, A. P., and Vogelstein, B. Hypomethylation distinguishes genes of somehumancancersfrom their normal counterparts.Nature (Lond.), 301: 89—92,1983.

29. Liang, P., and Pardee, A. B. Differential display of eukaryotic messenger RNA bymeans of the polymerase chain reaction. Science (Washington DC), 257: 967—971,1992.

30. Welsh, J., Chada, K., Dalal, S. S., Cheng, R., Ralph, D., and McClelland, M.Arbitrarily primed PCR fingerprinting of RNA. Nucleic Acids Res., 20: 4965—4970,1992.

599

sequences.These findings are consistent with the types of methylationchanges associated with tumorigenesis as previously described (2).The degree of genome-wide DNA methylation at multiple sites innormal and tumor tissues can be assayedarbitrarily by this technique,and we have shown that it may be possible to examine a number ofadditionalregionsof DNA by alteringtherestrictionenzymesusedfordigestion of DNAs, primer sequences,or PCR conditions. Methylation-sensitive AP-PCR can therefore be used to rapidly screen formethylation differences between genomes and to isolate specific genetic elementsthat have undergonemethylationchangesassociatedwith tumorigenesis.

References

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

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

3. Jones, P. A. DNA methylation errors and cancer. Cancer Res., 56: 2463—2467,1996.4. Li, E., Bestor, 1. H., and Jaenisch, R. Targeted mutation of the DNA methyltrans

ferase gene results in embiyonic lethality. Cell, 69: 915—926,1992.5. Counts,J. L., andGoodman,J. I. Alterationsin DNA methylationmayplaya variety

of roles in carcinogenesis.Cell, 83: 13—15,1995.6. Schorderet, D. F., and Gartler, S. M. Analysis of CpG suppression in methylated and

nonmethylatedspecies.Proc.Nail. Aced.Sci. USA, 89: 957—961,1992.7. Rideout, W. M., ifi, Coetzee, G. A., Olumi, A. F., and Jones,P. A. 5-Methylcytosine

as an endogenous mutagen in the human LDL receptor and p53 genes. Science(WashingtonDC), 249: 1288—1290,1990.

8. McClelland, M., and Ivarie, R. Asymmetrical distribution of CpG in an “average―mammaliangene.NucleicAcids Res.,10: 7865—7877,1982.

9. Gardiner-Garden, M., and Frommer, M. CpG islands in vertebrate genomes. J. Mol.Biol., 196: 261—282,1987.

10. Riggs, A. D., and Pfeifer, 0. P. X-chromosome inactivation and cell memory. TrendsGenet.,8: 169—174,1992.

11. Barlow, D. P., Stoger,R., Hemnann, B. 0., Saito, K., and Schweifer,N. The mouseinsulin-like growth factor type-2 receptor is imprinted and closely linked to the Tmelocus.Nature(Lond.),349: 84—87,1991.

12. Ferguson-Smith, A. C., Sasaki, H., Cattanach, B. M., and Surani, M. A. Parentalorigin-specific epigenetic modification of the mouse H19 gene. Nature (Lond.), 362:751—755,1993.

13. Antequera, F., Boyes, J., and Bird, A. High levels of de novo methylation and alteredchromatin structure at CpG islands in cell lines. Cell, 62: 503—514,1990.

14. Jones, P. A., Wolkowicz, M. J., Rideout, w. M., Gonzales, F. A., Marziasz, C. M.,Coetzee,0. A., andTapscott,S. J. De novomethylationof theMyoDl CpG island

on May 4, 2020. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

1997;57:594-599. Cancer Res   Mark L. Gonzalgo, Gangning Liang, Charles H. Spruck III, et al.   Primed PCRRegions of Genomic DNA by Methylation-sensitive Arbitrarily Identification and Characterization of Differentially Methylated

  Updated version

  http://cancerres.aacrjournals.org/content/57/4/594

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/57/4/594To request permission to re-use all or part of this article, use this link

on May 4, 2020. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from