6) 175–183www.elsevier.com/locate/yviro

Virology 349 (200

Methylation of the human papillomavirus-18 L1 gene:A biomarker of neoplastic progression?

Tolga Turan a, Mina Kalantari a, Itzel E. Calleja-Macias a, Heather A. Cubie b, Kate Cuschieri b,Luisa L. Villa c, Hanne Skomedal d, Hugo A. Barrera-Saldaña e, Hans-Ulrich Bernard a,⁎

a Department of Molecular Biology and Biochemistry, University of California Irvine, Irvine, CA 92697, USAb Royal Infirmary of Edinburgh, Edinburgh, Scotland

c Ludwig Institute for Cancer Research, Sao Paulo, Brazild Norchip Nuclisense Inc., Oslo, Norway

e Department of Biochemistry at the School of Medicine, Universidad Autonoma de Nuevo Leon, Mexico

Received 19 October 2005; returned to author for revision 8 December 2005; accepted 23 December 2005Available online 10 February 2006

Abstract

Epigenetic transcriptional regulation plays an important role in the life cycle of human papillomaviruses (HPVs) and the carcinogenicprogression of anogenital HPV associated lesions. We performed a study designed to assess the methylation status of the HPV-18 genome,specifically of the late L1 gene, the adjacent long control region (LCR), and part of the E6 oncogene in cervical specimens with a range ofpathological diagnoses. In asymptomatic infections and infections with precancerous (precursor) lesions, HPV-18 DNAwas mostly unmethylated,with the exception of four samples where hypermethylation of L1 was detected. In contrast, L1 sequences were strongly methylated in all cervicalcarcinomas, while the LCR and E6 remained unmethylated. HeLa cells, derived from a cervical adenocarcinoma, contain chromosomally integratedHPV-18 genomes. We found that L1 is hypermethylated in these cells, while the LCR and E6 are unmethylated. Treatment of HeLa cells with themethylation inhibitor 5-Aza-2′-deoxycytidine (5-Aza-CdR) led to the expected reduction of L1 methylation. After removal of 5-Aza-CdR, L1methylation resumed and exceeded pretreatment levels. Unexpectedly, the LCR and E6 also became methylated under these conditions, albeit atlower levels than L1. We hypothesize that L1 is preferentially methylated after integration of the HPV genome into the cellular DNA, possibly sincelinearization prohibits its normal transcription, while the enhancer and promoter may be protected from methylation by transcription factors. Sinceour data suggest that HPV-18 L1 methylation can only be detected in carcinomas, except in some few precancerous lesions and asymptomaticinfections, L1 methylation may constitute a powerful molecular marker for detecting this important step of neoplastic progression.© 2006 Elsevier Inc. All rights reserved.

Keywords: Papillomavirus; Cervical carcinoma; Epigenetic; Transcription regulation

Introduction

Human papillomaviruses (HPV) are implicated as causativeagents in the etiology of cervical cancer (International Agencyfor Research on Cancer, 1995). Eighteen HPV types areassociated with nearly 100% of all cervical carcinomas, andamong these types, HPV-16 and HPV-18 account for approx-imately two thirds of these malignant lesions worldwide(Munoz et al., 2003). HPV-18 has been associated with more

⁎ Corresponding author. Fax: +1 949 824 8551.E-mail address: hbernard@uci.edu (H.-U. Bernard).

0042-6822/$ - see front matter © 2006 Elsevier Inc. All rights reserved.doi:10.1016/j.virol.2005.12.033

aggressive forms of cervical intraepithelial neoplasia (Schwartzet al., 2001; Burk et al., 2003) and has a higher prevalence thanHPV-16 in adenocarcinomas (Andersson et al., 2001). In vitro,the oncoproteins of HPV-16 and -18 have similar molecularproperties by binding the p53 and pRB tumor suppressorproteins and interfering with their function (Dyson et al., 1989,Werness et al., 1990). However, HPV-18 has been found to havean increased potential for the immortalization of epithelial cells(Schlegel et al., 1988), possibly due to the higher transcriptionalactivity of its enhancer-promoter system (Villa and Schlegel,1991; Romanczuk et al., 1991). In view of these findings andthe taxonomically remote relationship of HPV-16 and -18 (deVilliers et al., 2004), our research pertaining to gene expression

176 T. Turan et al. / Virology 349 (2006) 175–183

has aimed to investigate consistencies between these twoviruses in addition to type-specific molecular and pathologicaldistinctions.

For many years, investigations of HPV gene expression haveconcentrated on the identification of sequence-specific tran-scription factors (reviewed in Bernard, 2002). More recently, ithas become clear that epigenetic mechanisms, which influencechromatin conformations that favor or repress gene expression,play a major role in modulating HPV transcription (Stünkel andBernard, 1999; Zhao et al., 1999). One of several mechanismsthat affect chromatin conformation is the methylation of DNA,specifically at cytidine-guanidine (CpG) dinucleotides. Meth-ylated CpG dinucleotides (meCpGs) bind repressors such asMeCP2, which alter the conformation of nucleosomes throughtheir interaction with histone deactylases in a mannerunfavorable to transcription (reviewed in Bird, 1992; Goll andBestor, 2005). Methylation of the genomes of HPV-1 and thecottontail rabbit PV was observed more than 20 years ago(Burnett and Sleeman, 1984; Sugawara et al., 1983; Wettsteinand Stevens, 1983). It has also been reported that, as would beexpected, in vitro methylated HPV-16 genomes are transcrip-tionally repressed after transfection into cell culture (Rosl et al.,1993). However, more recently, it was shown that methylationof the DNA of HPV-16 and -18 takes place regularly in vivo incervical cells, in patients in situ in addition to cell cultures.Present evidence suggests that the genomes of both viruses canbecome methylated but apparently in different patterns and insomewhat different pathological contexts (Badal et al., 2003,2004; Kim et al., 2003; Van Tine et al., 2004; Kalantari et al.,2004; Wiley et al., 2005). It raises questions of where and howHPV genomes are recognized by the cellular DNA methylationmachinery and about the role of this mechanism during the HPVlife cycle and HPV-dependent carcinogenesis.

We previously reported that 10–20% of the CpGs of the longcontrol region (LCR) of HPV-16 DNA are found methylated inasymptomatic infection. Methylation decreases strongly inprecursor lesions, while remethylation occurs in high-gradelesions and carcinomas. It exceeds 50% in the case of someCpG dinucleotides within the L1 gene (Kalantari et al., 2004;Wiley et al., 2005). The transition between hypermethylation inasymptomatic infection and hypomethylation in low-gradelesions correlates with evidence that W12 cell cultures withepisomal HPV-16 genomes maintain the viral DNA in ahypermethylated state in undifferentiated cells with hypomethy-lation occurring upon differentiation (Kim et al., 2003). Thisfinding supports the hypothesis that HPV-16 can be transcrip-tionally repressed upon infection. In order to initiate thetransforming process, it has to become demethylated as hasbeen found in clinical samples (Kalantari et al., 2004) and cellcultures (Kim et al., 2003). In carcinomas, HPV-16 DNA isnormally not found in the episomally replicating form but ratheris integrated into the cellular DNA (Schwarz et al., 1985; Danielet al., 1995). It is probably not surprising that this correlateswith increased methylation (Kalantari et al., 2004; Wiley et al.,2005), since DNA integrating into cellular chromosomes is apreferential target for methylation and transcriptional inactiva-tion (Doerfler et al., 2001) per se. Therefore, there may not be a

specific recognition of HPV-16 DNA in this scenario. Thepossibility of a specific recognition mechanism has beenproposed, however, leading to the selective activation of afew HPV-16 genomes in each cell for maintenance of theneoplastic state (Van Tine et al., 2004).

The project reported here aimed to refine our knowledge ofthe role of methylation in HPV-18. Our findings are based on astrategy in which we purified DNA from clinical samples andcultured cells, bisulfite treated and PCR amplified this DNA,and cloned the product into an E. coli vector for analysis ofindividual HPV-18 segments. Our investigations of HPV-16 hadpointed to the need for this refined approach, as we had learntthat individual samples and even cell lines contain mixtures ofdifferentially methylated HPV-16 genomes (Kalantari et al.,2004; Wiley et al., 2005). Therefore, analyses based entirely onrestriction digestions or direct bisulfite sequencing (Badal et al.,2004) do not lead to robust analyses of this phenomenon. Thefindings of this study taken together with our previouspublications suggest that significant differences exist in theepigenetic regulation of HPV-16 and -18 and that L1methylation could prove to be a powerful biomarker by whichto study tumor progression.

Results

Hypomethylation of the HPV-18 L1 gene and the LCR inasymptomatic patients

The objective of our study was to establish the diversity ofmethylation patterns in the 3′ part of the L1 gene andthroughout the LCR into the 5′ part of the E6 gene of HPV-18covering a segment of about 1.2 kb with 31 CpG dinucleotidesserving as potential methylation targets. This was done afterbisulfite treatment of sample DNA, which converts cytosineresidues into uracils, while not affecting methyl-cytosineresidues. PCR converts the uracils into thymine and themethyl-cytosines into cytosine residues, and DNA sequencingof the PCR product allows for the determination of themethylation pattern before treatment. We analyzed five clonesfrom each clinical sample in order to observe any potentialheterogeneity between different HPV-18 genomes withinindividual samples as detected in our analyses of HPV-16(Kalantari et al., 2004). Since bisulfite treatment precludes theamplification of large DNA segments, we split the 1.2-kbtarget into three amplicons of approximately equal size, thefirst correlating with the 3′ part of the L1 gene, the second withthe enhancer, the third with the promoter and 5′ end of the E6gene.

Fig. 1A shows the results obtained with eleven DNA extractsfrom cervical smears diagnosed as pathologically normal. Threeand eight of these samples came from Mexican and Brazilianpatients, respectively. It was evident that these samples wereonly very lightly methylated, with meCpGs observed in only 64of 1671 positions (3.8%). Altogether, 36 out of 55 HPV-18genomes were completely unmethylated implying they weretranscriptionally unaffected. Most of the methylation wasevident in only three samples, which accounted for 49 of the

177T. Turan et al. / Virology 349 (2006) 175–183

64 meCpGs. One of these samples (top of Fig. 1A) had 28meCpGs in the L1 gene, and only a single meCpG in the LCR.The methylation rate was 7.5% (44 meCpGs among 595 CpGs)in L1, 2.8% (14 meCpGs among 492 CpGs) in the enhancersegment, and 1% (6 meCpGs among 594 CpGs) in the promotersegments. This sample and the second one in Fig. 1A withsubstantial L1 methylation came from the Brazilian cohort,which otherwise contained hypo- and unmethylated samples.We obtained very similar data for 13 other DNA samples fromasymptomatic patients of the same two cohorts, although wecould not establish complete data sets because of a shortage ofDNA (data not shown). We conclude that methylationis generally very light in asymptomatic HPV-18 infections,with occasional sample-specific increases, preferentially occur-ring in L1.

Hypomethylation of the HPV-18 L1 gene and the LCR inprecancerous (precursor) lesions

Fig. 1B shows the results obtained with twelve DNA extractsof cervical smears diagnosed with cytological abnormalities.Specifically, five samples were graded as mild dyskaryosis,which equates to a low-grade abnormality/lesion (the uppermostfive data groups in Fig. 1B), and three samples showed“borderline nuclear changes” (equating most closely to ASCUS,i.e., atypical squamous cells of undetermined significance) (thesixth to eighth data group in Fig. 1B). Four samples were gradedas severe dyskaryosis, which equates to a high-grade abnor-mality/lesion. All of these samples came from patients fromEdinburgh. Similar to the findings with pathologically normalsamples, HPV-18 was only very lightly methylated in themajority of these precursors. meCpGs were observed in only 77of a total of 1849 positions (4.0%). Altogether, 33 out of 60HPV-18 genomes were completely unmethylated and thereforewere likely to be unaffected transcriptionally. Most of themethylation was evident in only two samples, which accountedfor 57 of the 77 meCpGs. The HPV-18 genomes in these twosamples (the ninth and the eleventh sample in Fig. 1B) had mostof the methylation in the L1 gene. It was also interesting to notethat the two, heavily methylated genomes, had been extractedfrom samples that had high-grade abnormalities. Ordered bygenomic region, the methylation rate was 8.9% (59 meCpGsamong 660 CpGs) in L1, 1.5% (8 meCpGs among 529 CpGs) inthe enhancer segment, and 1.5% (10 meCpGs among 660CpGs) in the promoter segments. We conclude that methylationis generally very light in HPV-18 containing low-gradeprecursor lesions, with a potential for intense methylation,however, in high-grade lesions.

Hypermethylation of the HPV-18 L1 gene in carcinomas

Fig. 1C shows the results obtained with 14 DNA prepara-tions from cervical squamous carcinomas. Thirteen of thesesamples came from patients from Norway and one from apatient in Mexico. Fig. 1C clearly depicts the high methylationrate of the HPV-18 L1 gene in all of these samples in contrast tothe sporadic methylation in the LCR. Altogether, meCpGs were

observed in 647 among a total of 2170 positions (29.8%). Onlyfour of the 70 HPV-18 genomes were completely unmethylated.Ordered by genomic region, the methylation rate was 76.8%(592 meCpGs among 770 CpGs) in L1, 2.7% (17 meCpGsamong 630 CpGs) in the enhancer segment, and 4.8% (37meCpGs among 770 CpGs) in the promoter segments. Weconclude that there is a strong bias for methylation of L1 incarcinomas. Figs. 2A to C represent the summary of themethylation patterns observed in clinical samples in form of bardiagrams.

Preferential methylation of the HPV-18 L1 gene in HeLa cells

In order to evaluate the stability of the methylated/unmethylated state across the L1-LCR junction, we decidedto study methylation changes in an HPV-18 containing cancercell line in the presence and absence of a methylation inhibitor.We have previously reported that in HeLa cells, which containabout 50 chromosomally integrated HPV-18 copies, the L1gene is methylated, while the LCR is devoid of meCpGs (Badalet al., 2004). These data were based on two fairly crudetechniques, namely digestion with the restriction enzymeMcrBC and direct sequencing of PCR amplicons generatedafter bisulfite modification. Here, we aimed to establish aprecise pattern of methylation across the same 31 CpGdinucleotides between the 3′ part of the L1 gene and the 5′part of E6 that were analyzed in clinical samples. Fig. 3A showsan analysis of three amplicons covering this genomic region, asdescribed above. In the L1 gene, we found 100 meCpGs among264 CpG dinucleotides, which cumulatively corresponds to anoverall methylation rate of 38%. There were only five of 216CpGs (1.9%) methylated in the enhancer segment and none inthe promoter segment.

We considered it unlikely that the methylation discontinuityobserved between the 3′ end of L1 and the 5′ part of the LCRobserved in HeLa cells and clinical samples could be an artifactoccurring during the generation of the two separate amplicons.However, in order to validate this assumption, we createdamplicons across this boundary that included CpG residuesfrom the L1 gene as well as some from the 5′ side of the LCR(amplicons Msp7F and R, see Materials and methods section).All amplicons derived from samples with L1 methylationgenerated by this strategy had the same discontinuity of themethylated state across the positions 7317 or 7122, respectively(data not shown), as evident from Fig. 2A. We conclude that thespecific methylation patterns that we observed in this genomicregion are not artifacts of primer design or primer position.

The effect of 5-Aza-CdR treatment and withdrawal on HPV-18L1 methylation in HeLa cells

5-Aza-CdR is an inhibitor of the maintenance DNAmethyltransferase DNMT1, and can be used in cell culture tointerfere with CpG methylation and to assess its reestablishmentafter removal of the drug. Here, we attempted to monitor theeffects that this drug had on the methylation of HPV-18 L1sequences, specifically, to observe which genomic regions are

178 T. Turan et al. / Virology 349 (2006) 175–183

targeted after subsequent culture of the HeLa cells in theabsence of the drug.

Fig. 3B shows that HPV-18 L1 sequences analyzed aftergrowth of HeLa cells in the presence of 5-Aza-CdR exhibited a

moderate reduction in the amount of methylated L1 sequencesto 28% (75 meCpGs among 264 CpG residues). There were nomethylated CpGs in the enhancer and promoter amplicons.After subsequent growth of the 5-Aza-CdR treated cells for 20days in the absence of this methylation inhibitor, L1methylation resumed and reached a level of 61% (162 meCpGsamong 264 CpG residues), significantly exceeding thepretreatment level of 38% (Fig. 3C). While the enhancer andpromoter had been nearly completely unmethylated in theabsence of 5-Aza-CdR, treatment with this drug and itssubsequent removal led to an unexpected and surprising denovo methylation of 38 out of 216 CpG residues (17.6%) of theenhancer segment and 122 of 264 CpGs (46%) of the promotersegment, indicating a lack of complete protection frommethylation of the LCR, but a maintenance of limited protectionof the segment that binds important enhancer factors like AP-1.Fig. 4 summarizes these observations.

Discussion

Research outside the papillomavirus field has led to thetheory of an “epigenetic conversation network” involvinghistone acetylation, histone deacetylation, cytosine methylation,and histone methylation. These reactions covalently modifynucleosomes and DNA and lead either to a transcriptionallycompetent or silenced chromatin (Fuks, 2005). Some of theseregulatory mechanisms are well established in papilloma-viruses: the LCRs of HPV-6, -11, -16, and -18 genomes eachcontain a silencer recognized by the transcriptional repressorsCDP and YY1, which interact with histone deacetylases(HDACs) and thereby repress the enhancer and E6 promoter(Bauknecht et al., 1992; O'Connor et al., 1996, 2000; Pattison etal., 1997; Zhao et al., 1999; Ai et al., 2000). It is likely that thisprocess links HPV transcription with epithelial differentiation(Stünkel and Bernard, 1999; Ai et al., 2000). CDP and YY1 alsobind histone methyltransferases, leading to methylated histones,association with DNA methyltransferases, and methylatedcytosines (Fuks, 2005). The resulting meCpG residues bindfactors such as MeCP2, which repress transcription throughinteractions with HDACs. While histone methylation has not

Fig. 1. CpG methylation of the L1 gene, the enhancer, promoter, and the 5′ partof the E6 gene of HPV-18 in asymptomatic infections, precancerous lesions andcarcinomas of the cervix. Methylated CpGs are shown as black rectangles,unmethylated CpGs as white rectangles. Each vertical bar of rectanglesidentifies a genomic position, which is specified by number at the top of thefigure. Each horizontal bar of rectangles identifies an E. coli clone of HPV-18obtained after bisulfite modification and PCR amplification of the sample DNA.Each sample is represented by five E. coli clones, as grouped in the figure.Occasional omissions, for example in the first vertical bar, indicate poor DNAsequence quality. The gaps between the L1, enhancer, and promoter groups ofCpGs separate three different amplicons, since it is not possible to obtain 1.2-kbamplicons with bisulfite modified DNA due to degradation. Therefore, eachhorizontal bar is derived from three differentHPV-18 genomes as separated by thegaps. (A) DNA from asymptomatic cervical infections. (B) DNA fromprecancerous lesions. For pathological explanations, see the Results section. (C)DNAfromcervical carcinomas.Thediscontinuityofmethylation in someselectedcarcinomas across the positions 7122 and 7317 was confirmed by contiguousamplicons as explained the Materials and methods and Results sections.

Fig. 2. CpG methylation of the L1 gene, the enhancer, promoter, and the 5′ partof the E6 gene of HPV-18 in asymptomatic infections, precancerous lesions andcarcinomas of the cervix. Representation of the data of Fig. 1 in form of a bardiagram.

179T. Turan et al. / Virology 349 (2006) 175–183

yet been studied in papillomaviruses, the correlation betweenthe decreasing concentrations of CDP, YY1 (Pattison et al.,1997; Ai et al., 2000), and HPV-16 DNA methylation (Kim etal., 2003; Kalantari et al., 2004) during epithelial differentiationsuggests an involvement of this mechanistic network in therepression of HPV-16 in undifferentiated epithelial cells in vitroand in asymptomatic infections in situ. This model accom-modates the data available for HPV-16 (Badal et al., 2003; Kimet al., 2003; Kalantari et al., 2004), which indicate thatmethylation targets HPV-16 genomes in basal layers ofsquamous epithelia, and a balance between methylation anddemethylation appears to influence the viral life cycle anddecisions between latent and progressing infections.

In contrast, the findings that we report here do not provideevidence for a similar scenario in HPV-18 as we detected onlyvery low methylation, and often none, in the HPV-18 LCR inasymptomatic infections. The small sample size that wasavailable to us precludes a final conclusion that HPV-16 and

HPV-18 really do respond differentially to the cellularmethylation machinery (potentially in line with their fairlyremote relationship), or whether divergence may originate fromdifferent cellular contexts, as HPV-18 may preferentially targetcolumnar rather than squamous epithelia. Nevertheless, thesediscrepancies are surprising in view of the extensive similaritiesin the transcription biology of HPV-16 and -18, which bind thesame transcription activators and repressors and show a similarchromatin structure with specifically positioned nucleosomes(Stünkel and Bernard, 1999).

In carcinomas, however, HPV-16 and -18 show similarly apreferential methylation of the L1 gene, and the particularlyintense methylation of the HPV-18 L1 gene in all carcinomas,exceeding 80% at some CpG positions, is the most intriguingfinding of this study. In HPV-16 associated carcinomas themethylation frequency of three CpGs overlapping with L1 wasalso pronounced but only slightly exceeded 50%. We hadreported in a previous paper that the HPV-18 L1 gene ispreferentially methylated (Badal et al., 2004). This latter studyincluded a few cell lines and clinical samples, and was based onrestriction digestion and direct sequencing, two techniques withlimited resolution. In that past paper, we had reported thatmethylation was often absent from all parts of the viral genomeexcept L1 (Badal et al., 2004). In the present study, we havedescribed refined HPV-18 methylation maps, which has led tothe observation that preferential methylation of L1 extendsthrough the 3′ part of the L1 gene precisely to position 7122,which is the last CpG within L1, ten nucleotides upstream of thetermination codon of this gene. Occasionally methylationextends to position 7317, 185 bp within the LCR. This regionis normally referred to as 5′ LCR and does not includesequences relevant for enhancer function, while the enhanceroverlaps with the CpG at position 7460 (part of an E2 bindingsite) and all CpGs 3′ of this position. An explanation for theprecise methylation of L1 may be the fact that this gene isunlikely to be transcribed after recombination in the cancerousstate, where L1 is located upstream of the viral transcriptionregulatory elements. However, the LCR is also not transcribed,and one may have to postulate that its hypomethylated state ismaintained through interaction with transcription factors.

The striking difference between intense L1 methylation in allcarcinomas and lack of L1 methylation in most asymptomaticsmears and low-grade precursor lesions suggests that oneshould explore whether L1 methylation may be a usefulbiomarker to study carcinogenic progression. It is noteworthythat the only two precursor lesions with intense L1 methylationhad been pathologically diagnosed as high-grade lesions, and itis conceivable that they contained an HPV-18 genome in amolecular state indicative of cancer. Our observation of twosmears with L1 methylation, which had been pathologicallydiagnosed as normal, may point to the power of this approach.Intense methylation of HPV-18 L1 may identify cell popula-tions with integrated HPV-18 genomes and therefore with amolecular predisposition toward high-grade lesions or carcino-mas and may highlight pathological misdiagnoses. However,we are aware of the alternative that the early-late switch ofpapillomaviruses may be accompanied by changes of the

Fig. 4. CpG methylation of the L1 gene, the enhancer, promoter, and the 5′ partof the E6 gene of HPV-18 in HeLa cells and changes during and after treatmentwith 5-Aza-CdR. Representation of the data of Fig. 2 in form of a bar diagram tovisualize the methylation discontinuity across the genomic position 7122 (at the3′ end of L1) and 7317 (in the 5′ part of the LCR), and de novo hypermethylationof the enhancer and promoter after withdrawal of the methylation inhibitor.

Fig. 3. CpG methylation of the L1 gene, the enhancer, promoter, and the 5′ partof the E6 gene of HPV-18 in HeLa cells (A) and changes during (B) and after (C)treatment with 5-Aza-CdR. Methylated CpGs are shown as black rectangles,unmethylated CpGs as white rectangles. Each vertical bar of rectanglesidentifies a genomic position, which is specified by number at the top of thefigure. Each horizontal bar of rectangles identifies an E. coli clone of HPV-18obtained after bisulfite modification and PCR amplification of the sample DNA.The gaps between the L1, enhancer, and promoter groups of CpGs separate threedifferent amplicons, since it is not possible to obtain 1.2-kb amplicons withbisulfite modified DNA due to degradation. Therefore, each horizontal bar isderived from three different HPV-18 genomes as separated by the gaps. Thediscontinuity of methylation across the positions 7122 and 7317 was confirmedby contiguous amplicons as explained the Materials and methods and Resultssections. The number of clones studied in part of Fig. 2B was reduced foreconomic and not for scientific reasons, as there was no good reason to study theinfluence of a methylation inhibitor on the demethylated DNA segment.

180 T. Turan et al. / Virology 349 (2006) 175–183

methylation status of late genes, and it is possible that ouranalysis may differentiate between HPV-18 infections withexpressed or repressed late genes. We are in the process ofimproving the technical power of the study of HPV-18methylation by procedures that selectively detected methylatedCpGs, such as meCpG-specific amplification primers andmethylation-specific molecular beacons included in real-timePCR experiments.

We observed partial demethylation of HPV-18 L1 in HeLacells under the influence of 5-Aza-CdR, as expected fromsimilar studies of methylated CpGs in cellular chromosomes(Velicescu et al., 2002). It is striking that a re-methylationpattern at the border of L1 and the LCR flanking the position7317 becomes reestablished after discontinuation of 5-Aza-CdR treatment. Methylation exceeded 60% at position 7122and at all CpGs to the 5′ side and was present at 3–20% levelsat the eight CpGs to the 3′ side, overlapping with the enhancer.

This observation strengthens the case for L1 as a uniquemethylation target, as it is apparently selectively recognizedsubsequent to 5-Aza-CdR treatment. We do not have anexplanation for the methylation of LCR segments, particularlyat the promoter of HPV-18, under these conditions, which arespared from methylation during the regular culture of HeLacells. It has been proposed, however, that remethylation afterremoval of CpGs may show cell type-specific idiosyncrasiesdue to differential expression of both the de novo methylaseDNMT3a and the maintenance methylase DNMT1 (Velicescuet al., 2002). It is evident from this and previous studies thatthe concept of “epigenetic conversation” applies to papilloma-virus gene regulation, and that continuing studies of HPVgenome methylation will likely have considerable impact both

181T. Turan et al. / Virology 349 (2006) 175–183

to the understanding of the HPV life cycle as well as tomolecular insight into and diagnosis of cancer progressionevents.

Materials and methods

Clinical specimens

DNA samples extracted from cervical carcinomas wereobtained from an archival collection of tumors maintained at theDepartment of Pathology, Norwegian Radium Hospital, Oslo,Norway, and from a specimen obtained during a publishedstudy on the genomic diversity of HPVs in Monterrey, Mexico(Calleja-Macias et al., 2004). All of these samples had beendiagnosed as squamous carcinomas and none as adenocarcino-mas, which are frequently associated with HPV-18. DNAextracted from the residual material of normal and abnormalcervical smear preparations was obtained from archivalcollections established as a result of previous epidemiologicalstudies at the Ludwig Institute for Cancer Research in SaoPaulo, Brazil (L.L.V.), at the Department of Pathology, NationalHospital, Oslo, Norway (B.H.), the Royal Infirmary ofEdinburgh, Edinburgh, Scotland (H.A.C. and K.C.), and froma collection of samples (Fajardo and Barrera-Saldaña, unpub-lished) that reconfirmed a recent investigation from Monterrey,Mexico (Calleja-Macias et al., 2004). No patient was sampledfor the purpose of the specific research described here.Unfortunately, the fact that only very few laboratories in theworld have large collections of HPV-18 positive samples (due tothe relatively low prevalence of this virus compared to HPV-16)led to a bias of geographic overrepresentation in somepathological categories. Thirteen out of fourteen cancer samplescame from Oslo, and all DNA extracted from abnormal smears(consistent with evidence of precursor lesions) came fromEdinburgh. We do not believe, however, that these geographicbiases influenced the study outcome, as we found no correlationbetween methylation state and geographic origin in previouslypublished research on HPV-16 (Kalantari et al., 2004).

Cell culture and DNA preparations

HeLa cells have been derived from a cervical adenocarci-noma and have been well studied to contain approximately 50HPV-18 copies in several chromosomal loci (Ambros andKarlic, 1987; Mincheva et al., 1987; Popescu et al., 1987; Lazoet al., 1989). We used a strain of HeLa cells that has beenmaintained in our laboratory for many years. Although strainsof HeLa cells have been cultured separately for many decades,our recent comparison of several HeLa strains did not point tomajor differences in the composition and epigenetic status oftheir endogenous HPV-18 genomes (Badal et al., 2004). HeLacells were grown in standard DME medium with 10% fetalbovine serum. The methylation inhibitor 5-Aza-2′-deoxycyti-dine (5-Aza-CdR) was used at a concentration of 1 μM for 3days. All DNA preparations from HeLa cells and clinicalsamples were purified with Qiagen genomic tips followingprotocols suggested by the supplier.

Bisulfite modification

For bisulfite treatment, 50–1000 ng of sample DNAsupplemented with 1 μg of salmon sperm DNA in a totalvolume of 18 μl in water was denatured with 2 μl of 3 MNaOH and incubated at 37 °C. After denaturation, 278 μl of4.8M sodium bisulfite and 2 μl of 100 mM hydroquinonewere added with the mixture being incubated in a thermalcycler for 20 cycles of 55 °C for 15 min and 95 °C for 30 s.The modified DNA was desalted with the QIAquick PCRpurification protocol and desulfonated thereafter by adding 5.5μl of 3M NaOH and 5 μg glycogen prior to 15-minincubation at 37 °C. The DNA was precipitated with 5.6 μlof 3 M sodium acetate and 150 μl of 100% ethanol, followedby centrifugation. The pellet was washed with 70% ethanoland dissolved in 30–50 μl TE buffer (10 mM Tris–HCl pH 8,1 mM EDTA).

Polymerase chain reactions, primers, T/A cloning and DNAsequencing

Since bisulfite-treated DNA is partially degraded, largeamplicons cannot be generated, so we divided the HPV-18segment that we wanted to study into three amplicons betweenthe genomic position 6845–7186, 7282–7747, and 7753–186.The sequences of the primers were designed according to thegenomic sequence of HPV-18 (Myers et al., 1994) assumingconversion of all cytosine residues into uracils. The genomicsegment 6845–7186 encompassed the 3′ part of the L1 geneand was amplified with the primers Msp6Fa, AATTATTA-GTTTGGTGGATATATAT (position 6845–6869) and Msp6R,AAAACATACAAACACAACAATAAATA (position 7186–7161). The genomic segment 7282–7747 with the 5′ part ofthe LCR and its center, including the viral enhancer, wasamplified with the primers Msp10F, TAAAATATGT-TTTGTGGTTTTGTG (position 7282–7293) and Msp10R,ATAATTATACAAACCAAATATACAATT (position 7747–7721). The genomic segment 7743–186 with the viralreplication origin, the E6 promoter and the 5′ part of the E6gene was amplified with the primers Msp8F TGTTTAA-TATTTTGTTTATTTTTAATATG (position 7753–7781) andMsp8R, TATCTTACAATAAAATATTCAATTCC (position186–161). In selected samples, we amplified a fourth segmentthat spanned the border between the 6845–7186 and 7282–7747amplicons, using the primers Msp7F, AGATTTAGATTAA-TATTTTTTTGGA (position 6986–7010) and Msp7R, AAAT-TAAAATTTACAATAATACCAAC (position 7537–7512).

PCR was carried out in a 25 μl volume containing 0.2 mM ofeach of the four dNTPs, 10 pmol of each primer, 2 mM MgCl2and 1 U of Ampli Taq Gold (Perkin-Elmer). The PCRconditions were 94 °C for 1 min, followed by 40 cycles of94 °C for 10 s, 54 °C for 30 s, and 68 °C for 1 min with a finalextension at 68 °C for 7 min. The presence of PCR products wasverified by agarose gel electrophoresis, and confirmed ampli-cons were cloned with the TOPO TA cloning kit for sequencing(Invitrogen). Cloned DNAs were sequenced by Big Dyeterminator v3.1 Cycle Sequencing (Applied Biosystems).

182 T. Turan et al. / Virology 349 (2006) 175–183

Acknowledgments

Our research was supported by NIH grant ROI CA-91964and by funds from the Chao Family Comprehensive CancerCenter of the University of California Irvine to H.U.B. We aregrateful for samples identification and experimental advice toIrene Krau, Tor Molden, and Frank Karlsen from NorChipNuclisense Inc. Oslo and Ruth Holm, Agnes Kathrine Lie, andGunnar B. Kristensen from the Norwegian Radium Hospital.

References

Ai, W., Narahari, J., Roman, A., 2000. Yin yang 1 negatively regulates thedifferentiation-specific E1 promoter of human papillomavirus type 6.J. Virol. 74, 5198–5205.

Ambros, P.F., Karlic, H.I., 1987. Chromosomal insertion of human papilloma-virus 18 sequences in HeLa cells detected by nonisotopic in situhybridization and reflection contrast microscopy. Hum. Genet. 77, 251–254.

Andersson, S., Rylander, E., Larsson, B., Strand, A., Silfversvard, C., Wilander,E., 2001. The role of human papillomavirus in cervical adenocarcinomacarcinogenesis. Eur. J. Cancer 37, 246–250.

Badal, V., Chuang, L.S.H., Badal, S., Tang, E., Villa, L.L., Wheeler, C.M., Li,B.F.L., Bernard, H.U., 2003. CpG methylation of human papillomavirus-16 DNA in cervical cancer cell lines and in clinical specimens: genomichypomethylation correlates with carcinogenic progression. J. Virol. 77,6227–6234.

Badal, S., Badal, V., Calleja-Macias, I.E., Kalantari, M., Chuang, L.S.H., Li,B.F.L., Bernard, H.U., 2004. The human papillomavirus-18 genome isefficiently targeted by cellular DNA methylation. Virology 324, 483–492.

Bauknecht, T., Angel, P., Royer, H.D., zur Hausen, H., 1992. Identification ofa negative regulatory domain in the human papillomavirus type 18promoter: interaction with the transcriptional repressor YY1. EMBO J. 11,4607–4617.

Bernard, H.U., 2002. Gene expression of genital human papillomaviruses andpotential antiviral approaches. Antiviral Ther. 7, 219–237.

Bird, A.P., 1992. The essentials of DNA methylation. Cell 70, 5–8.Burk, R.D., Terai, M., Gravitt, P., Brinton, L.A., Kurman, R.J., Barnes, W.A.m.,

Greenberg, M.D., Hadjimichael, O.C., Fu, L., McGowan, L., Mortel, R.,Schwartz, P.E., Hildesheim, A., 2003. Distribution of human papillomavirustypes 16 and 18 variants in squamous cell carcinomas and adenocarcinomasof the cervix. Cancer Res. 63, 7215–7220.

Burnett, T.S., Sleeman, J.P., 1984. Uneven distribution of methylation siteswithin the human papillomavirus la genome: possible relevance to viral geneexpression. Nucleic Acids Res. 12, 8847–8860.

Calleja-Macias, I.E., Kalantari, M., Huh, J., Ortiz-Lopez, R., Rojas-Martines,A., Gonzales-Guerrero, J.F., Williamson, A.L., Hagmar, B., Wiley, D.J.,Villarreal, L., Bernard, H.U., Barrera-Saldana, H.A., 2004. High prevalenceof specific variants of human papillomavirus-16, 18, 31, and 35 in a Mexicanpopulation. Virology 319, 315–323.

Daniel, B., Mukherjee, G., Seshadri, L., Vallikad, E., Krishna, S., 1995. Changesin the physical state and expression of human papillomavirus type 16 in theprogression of cervical intraepithelial neoplasia lesions analysed by PCR.J. Gen. Virol. 76, 2589–2593.

de Villiers, E.M., Fauquet, C., Broker, T.R., Bernard, H.U., zur Hausen, H.,2004. Classification of papillomaviruses. Virology 324, 17–27.

Doerfler, W., Remus, R., Muller, K., Heller, H., Hohlweg, U., Schubbert, R.,2001. The fate of foreign DNA in mammalian cells and organisms. Dev.Biol. (Basel) 106, 89–97.

Dyson, N., Howley, P., Munger, K., Harlow, E., 1989. The human papillomavirus-16 E7 oncoprotein is able to bind to the retinoblastoma gene product.Science 243, 934–937.

Fuks, F., 2005. DNA methylation and histone modifications: teaming up tosilence genes. Curr. Opin. Genet. Dev. 15, 490–495.

Goll, M.G., Bestor, T.H., 2005. Eukaryotic cytosine methyltransferases. Annu.Rev. Biochem. 74, 481–514.

International Agency for Research on Cancer Monograph on the evaluation ofcarcinogenic risks to humans, 1995. Human Papillomaviruses, vol. 64.IARC, Lyon.

Kalantari, M., Calleja-Macias, I.E., Tewari, D., Hagmar, B., Barrera-Saldana,H.A, Wiley, D.J., Bernard, H.U., 2004. Conserved methylation patterns ofhuman papillomavirus-16 DNA in asymptomatic infection and cervicalneoplasia. J. Virol. 78, 12762–12772.

Kim, K., Garner-Hamrick, P.A., Fisher, C., Lee, D., Lambert, P.F., 2003.Methylation patterns of papillomavirus DNA, its influence on E2 function,and implications in viral infection. J. Virol. 77, 12450–12459.

Lazo, P.A., DiPaolo, J.A., Popescu, N.C., 1989. Amplification of the integratedviral transforming genes of human papillomavirus 18 and its 5′-flankingcellular sequence located near the myc protooncogene in HeLa cells. CancerRes. 49, 4305–4310.

Mincheva, A., Gissmann, L., zur Hausen, H., 1987. Chromosomal integrationsites of human papillomavirus DNA in three cervical cancer cell linesmapped by in situ hybridization. Med. Microbiol. Immunol. (Berl.) 176,245–256.

Munoz, N., Bosch, F.X., de Sanjosé, S., Herrero, R., Castellsagué, X., Shah,K.V., Snijders, P.J.F., Meijer, C.J.L.M., 2003. Epidemiological classifica-tion of human papillomavirus types associated with cervical cancer. N. Engl.J. Med. 348, 518–527.

Myers, G., Bernard, H.U., Delius, H., Favre, M., Iconogle, J, van Ranst,M., Wheeler, C. (Eds.), 1994. Human papillomaviruses. Compendium.pp. 1-C-4–1-A-7. Los Alamos National Laboratory, Los Alamos, NewMexico, USA.

O'Connor, M.J., Tan, S.H., Tan, C.H., Bernard, H.U., 1996. YY1 represseshuman papillomavirus type 16 transcription by quenching AP-1 activity.J. Virol. 70, 6529–6539.

O'Connor, M.J., Stünkel, W., Koh, C.H., Zimmermann, H., Bernard, H.U.,2000. The differentiation-specific factor CDP/Cut represses transcriptionand replication of human papillomaviruses. J. Virol. 74, 401–410.

Pattison, S., Skalnik, D.G., Roman, A., 1997. CCAAT displacement protein, aregulator of differentiation-specific gene expression, binds a negativeregulatory element within the 5′ end of the human papillomavirus type 6long control region. J. Virol. 71, 2013–2022.

Popescu, N.C., DiPaolo, J.A., Amsbaugh, S.C., 1987. Integration sites of humanpapillomavirus 18 DNA sequences on HeLa cell chromosomes. Cytogenet.Cell. Genet. 44, 58–62.

Romanczuk, H., Villa, L., Schlegel, R., Howley, P., 1991. The viraltranscriptional regulatory region upstream of the E6 and E7 genes is amajor determinant of the differential immortalization activities of humanpapillomavirus types 16 and 18. J. Virol. 65, 2739–2744.

Rosl, F., Arab, A., Klevenz, B., zur Hausen, H., 1993. The effect of DNAmethylation on gene regulation of human papillomaviruses. J. Gen.Virol. 74,791–801.

Schlegel, R., Phelps, W., Zhang, Y., Barbosa, M., 1988. Quantitativekeratinocyte assay detects two biological activities of human papillomavirusDNA and identifies viral types associated with cervical carcinoma. EMBO J.7, 3181–3187.

Schwartz, S.M., Daling, J., Shera, K.A., Madeleine, M.M., McKnight, B.,Galloway, D.A., Porter, P.L., McDougall, J.K., 2001. Human papillomavirusand prognosis of invasive cervical cancer: a population-based study. J. Clin.Oncol. 19, 1906–1915.

Schwarz, E., Freese, U.K., Gissmann, L., Mayer, W., Roggenbuck, B.,Stremlau, A., zur Hausen, H., 1985. Structure and transcription of humanpapillomavirus sequences in cervical carcinoma cells. Nature 314,111–114.

Stünkel, W., Bernard, H.U., 1999. The chromatin structure of the long controlregion of human papillomavirus type 16 represses viral oncoproteinexpression. J. Virol. 73, 1918–1930.

Sugawara, K., Fujinaga, K., Yamashita, T., Ito, Y., 1983. Integration andmethylation of shope papilloma virus DNA in the transplantable Vx2 andVx7 rabbit carcinomas. Virology 131, 88–99.

Van Tine, B.A., Kappes, J.C., Banerjee, N.S., Knops, J., Lai, L., Steenbergen,R.D., Meijer, C.L., Snijders, P.J., Chatis, P., Broker, T.R., Moen Jr., P.T.,Chow, L.T., 2004. Clonal selection for transcriptionally active viraloncogenes during progression to cancer. J. Virol. 78, 11172–11186.

183T. Turan et al. / Virology 349 (2006) 175–183

Velicescu, M., Weisenberger, D.J., Gonzales, F.A., Tsai, Y.C., Nguyen, C.T.,Jones, P.A., 2002. Cell division is required for de novo methylation of CpGislands in bladder cancer cells. Cancer Res. 62, 2378–2384.

Villa, L., Schlegel, R., 1991. Differences in transformation activity betweenHPV-18 and HPV-16 map to the viral LCR-E6-E7 region. Virology 181,374–377.

Werness, B., Levine, A., Howley, P., 1990. Association of humanpapillomavirus types 16 and 18 E6 proteins with p53. Science 248,76–79.

Wettstein, F.O., Stevens, J.G., 1983. Shope papilloma virus DNA is

extensively methylated in non-virus-producing neoplasms. Virology 126,493–504.

Wiley, D.J., Huh, J., Chang, C., Kalantari, M., Rao, J.Y., Goetz, M., Msongsong,E., Poulter, M., Bernard, H.U., 2005. Methylation of human papillomavirusDNA in samples of HIV-1 infected men screened for anal cancer. J. AcquiredImmune Defic. Syndr. 39, 143–151.

Zhao, W., Noya, F., Chen, W.Y., Townes, T.M., Chow, L.T., Broker, T., 1999.Trichostatin A up-regulates human papillomavirus type 11 upstreamregulatory region-E6 promoter activity in undifferentiated primary humankeratinocytes. J. Virol. 73, 5026–5033.