04 (2007) 629–635www.elsevier.com/locate/ygyno

Gynecologic Oncology 1

Hypermethylation of two consecutive tumor suppressor genes, BLU andRASSF1A, located at 3p21.3 in cervical neoplasias

Hung-Cheng Lai a,1, Ya-Wen Lin b,1, Cheng-Chang Chang a, Hui-Chen Wang a,Ta-Wei Chu c, Mu-Hsien Yu a, Tang-Yuan Chu b,d,⁎

a Department of Obstetrics and Gynecology, Tri-Service General Hospital, Taipei, Taiwan, ROCb Graduate Institutes of Medical Sciences, National Defense Medical Center, Taipei, Taiwan, ROC

c Department of Obstetrics and Gynecology, Army Forces Taoyuan General Hospital, Taoyuan, Taiwan, ROCd Department of Obstetrics and Gynecology, Buddhist Tzu Chi General Hospital and Tzu Chi University, Hualien, Taiwan, ROC

Received 22 June 2006Available online 13 November 2006

Abstract

Objectives. Although initiated by human papillomavirus (HPV), cervical carcinogenesis demands other cofactors to shape its natural course.Epigenetic effects such as DNA methylation, are considered to contribute to carcinogenesis process.

Methods. The methylation status of BLU and RASSF1A, as well as the HPV infection status, were assessed in a full spectrum of cervicalneoplasia, including 45 low-grade squamous intraepithelial lesions (LSIL), 63 high-grade squamous intraepithelial lesions (HSIL), 107 squamouscell carcinomas (SCC), 23 adenocarcinomas (AC), and 44 normal control tissues.

Results. The BLU was methylated in 76.9% of SCC, 57.4% of HSIL, 20.0% of LSIL and 12.5% of normal tissues (P<0.001). The RASSF1Awas methylated in 15% of SCC, 17.5% of HSIL, but not in LSIL or normal tissues (P<0.001). In AC, 43.5% of patients showed BLU methylationand 26.1% RASSF1A methylation, significantly higher than the corresponding control frequencies of 12.5% (P=0.005) and 0% (P=0.001),respectively. There was an insignificant trend toward loss of BLU methylation with advancing clinical stages of SCC (84.8%, 67.7%, and 63.6%in stages I, II, and III/IV, respectively; P=0.08). Patients with LSIL infected with high-risk HPV showed a higher rate of BLU methylation thanthose without HPV (38.8% vs 9.1%, respectively; P=0.057). The methylation of RASSF1Awas inversely related to HPV infection in patients withHSIL/SCC (P=0.003).

Conclusions. These results suggest that the methylation of BLU and RASSF1A genes is associated with cervical carcinogenesis, which couldbe clinically important in the future molecular screening of cervical neoplasia.© 2006 Elsevier Inc. All rights reserved.

Keywords: Cervical cancer; HPV; BLU; RASSF1A; Methylation

Introduction

Cervical cancer is one of the major causes of death of womenworldwide [1]. Infection with oncogenic human papillomavirus(HPV) is the most significant risk factor in its etiology. Withadvances in detection methods and when tumor tissue integrity ismaintained, HPV DNA can be detected in virtually all cases of

⁎ Corresponding author. Department of Obstetrics and Gynecology, BuddhistTzu Chi Medical Center, 707, Sec.3, Chung-Yang Rd., Haulien City, Hualien,970, Taiwan, ROC. Fax: +886 2 87927199.

E-mail address: tychu@mail2000.com.tw (T.-Y. Chu).1 These authors equally contributed to this work.

0090-8258/$ - see front matter © 2006 Elsevier Inc. All rights reserved.doi:10.1016/j.ygyno.2006.10.003

cervical cancer [2]. While persistence of HPV infection is detri-mental in the development of high-grade squamous intraepitheliallesions (HSIL) and cancer [3], the immediate cervical lesion ofHPV infection is low-grade squamous intraepithelial lesions(LSIL). In follow up, about 60% of LSIL regress, 30% persist, 5–10% progress to HSIL, and less than 1% become cervical cancer[4]. The molecular mechanism underlying the persistant viralinfection and the long term, inefficient carcinogenesis processremainsmostly elusive [5]. Environmental or genetic factors otherthan HPV may also play a decisive role in the malignant conver-sion of cervical keratinocytes and the progression to cancer [6,7].

Genetic changes with resultant genomic instability havelong been recognized as an important mechanism in cervical

Table 1Primers and conditions used in the present study

Primers Sequence 5′→3′ Ampliconsize (bp)

Annealingtemperature(°C)

MSPRASSF1A MF CGAGAGCGCGTTTAGTTTCGTT 192 58RASSF1A MR CGATTAAACCCGTACTTCGCTAARASSF1A UF GGGGGTTTTGTGAGAGTGTGTTT 204 58RASSF1A UR CCCAATTAAACCCATACTTCACTAABLU MF GCGGGTTAGAGATTCGTTC 231 55BLU MR TCGAAACCGAAATCCGACGBLU UF GGTGGGTTAGAGATTTGTTT 235 53BLU UR ATATCAAAACCAAAATCCAACA

Bisulfite sequencingRASSF1A BGS1 AGTTTTTGTATTTAGGTTTTTATTG 465 55RASSF1ABGS2 CTACCCCTTAACTACCCCTTCCBLU BGS1 GGGGTTATTTTTATTTTTGTGTAGG 406 53BLU BGS2 ACAACAATTCCAAATCTCCCATAT

630 H.-C. Lai et al. / Gynecologic Oncology 104 (2007) 629–635

carcinogenesis. Several molecular genetic studies have identi-fied a few frequent alleleic loss or loss-of-heterozygosity chro-mosomal sites, indicating the loss of tumor suppressor genes(TSGs) in the development of cervical cancer [8–11]. In additionto genetic changes, epigenetic alterations such as DNAmethylation and histone modifications are recognized as impor-tant driven forces of cancer [12–14]. Methylation of cytosine iswidely found in mammal genomes in the context of thepalindromic sequence 5′-CpG-3′. Most CpG dinucleotidepairs are methylated except at some areas called “CpG island”where the methylation is developmentally controlled. CpGislands are CpG-rich areas of approximately 1 kb, usuallylocated in the vicinity of genes and often found near the promoterof widely expressed genes [15,16]. Controlled by DNAmethyltransferases, global DNA hypomethylation and site-specific hypermethylation have been reported as the hallmarkof cancer [12]. It has become increasingly apparent that DNAhypermethylation with subsequent epigenetic silencing of TSGsthrough chromatin remodeling is associated with loss offunction, which may constitute the second hit of the “two hit”hypothesis, providing a selective advantage during carcinogen-esis [12,13,17]. The epigenetic silencing of tumor suppressorgenes by DNA hypermethylation is commonly seen in humancancers, including cervical cancer [18–21].

Our previous work and that of others have shown that thedeletion of chromosome 3p21, which has been reported inseveral cancers, is a frequent event in cervical carcinogenesis[22–24]. The deletions are usually monoalleleic, and no muta-tions of the remaining allele in several candidate genes could beidentified, indicating epigenetic changes such as hypermethyla-tion may be responsible for the total shut down of the candidatetumor suppressor genes. Several TSGs are known to occur onchromosome 3p, among which are two recently characterizedgenes, BLU and RASSF1A [25–28].

BLU and RASSF1A are two tandemly head-to-tail liked geneslocated within a 25 kb region at 3p21.3. The present study wasundertaken to examine the association between the methylationof BLU and RASSF1A promoters and cervical neoplasia of dif-ferent severities. The association between the methylationphenotypes and HPV genotypes was also investigated.

Materials and methods

Patients

A hospital-based case-control study was conducted. The study includedpatients with LSIL (n=45), HSIL (n=63), invasive squamous cell carcinoma(SCC; n=107), or adenocarcinoma (AC; n=23) of the uterine cervix, diagnosedand treated at the Tri-Service General Hospital, Taipei, Taiwan, between 1993and 2000. Cytological, histological, and clinical data for all patients were panel-reviewed by a group of staff members including colposcopists, cytologists, andpathologists, to reach a final diagnosis. All patients were investigated and treatedwith a standard protocol for cervical neoplasia at the same hospital. Controls wererecruited from healthy women undergoing routine Pap screening during the sameperiod. Exclusion criteria included previous pregnancy, chronic or acute viralinfection, a history of cervical neoplasia, skin or genital warts, an immune-compromised state, other cancers, and operations on the uterine cervix. Speci-mens from subjects were collected and genomic DNA was extracted with anestablished protocol for tissue banking as previously described [29]. Theconcentration of DNA was determined by the PicoGreen fluorescence

absorption method, and its quality was checked by agarose gel electrophoresis.Cervical scrapings of controls, LSIL and HSIL, and tumor tissues from SCCand AC were used in the present study. The study was approved by theInstitutional Review Board of the Tri-Service General Hospital. Informedconsent was obtained from each patient and control subject.

Bisulfite modification, methylation-specific PCR (MS-PCR) andbisulfite sequencing

Of the genomic DNA, 1 μg was bisulfite modified using the DNA modi-fication Kit (Chemicon, Ternecula, CA, USA) according to the manufacturer'srecommendations. The final precipitate was eluted in 70μl of pre-warmed (55°C)TE buffer. MSP was performed according to Herman et al. [47]. In short, 1 μl ofmodified DNA was amplified using MSP primers (Table 1) that specificallyrecognized either the unmethylated or methylated RASSF1A and BLU genesequence after bisulphite conversion [25,30]. Methylation-specific PCR wasdone in a total volume of 25μl, containing 1 μl modified template DNA, 1.5 pmolof each primer, 0.2 mmol/L deoxynucleotide triphosphates and 1 unit Gold TaqDNA polymerase (PE Applied Biosystems, Foster City, CA). MSP reactionswere subjected to initial incubation at 95°C for 5 min, followed by 35 cycles of95°C for 30 s, and annealing at the appropriate temperature for 30 s and 72°C for30 s. Final extension was done by incubation at 72°C for 5 min. Normal DNAfrom human peripheral blood was bisulphite modified to serve as a control forthe unmethylated promoter sequence. Normal human DNAwas treated in vitrowith SssI methyltransferase (New England Biolabs, Beverly, MA, USA) in orderto generate a positive control for methylated alleles [31]. Amplification productswere visualized on 2.5% agarose gel containing ethidium bromide andilluminated under UV light. All MSP data were done on at least two inde-pendent modifications of DNAwith PCR cycles of 35. Signals that were weakerthan 1:10 dilution of in vitro methylated DNA with unmethylated peripheralblood DNA and signals that were not detectable in repeated experiments werescored as negative of methylation. Bisulfite-treated genomic DNAwas amplifiedusing primers for human RASSF1A and BLU (Table 1). Amplified PCR productwas purified and cloned into pCR4-TOPO vector (Invitrogen, Carlsbad, CA).DNA sequencing was performed on at least 5 individual clones using the 377automatic sequencer (Applied Biosystems, Foster City, CA, USA).

Re-expression of BLU and RASSF1A mRNA by5′-aza-2′-deoxycytidine treatment in cervical cancer cells

The methylation status of BLU and RASSF1Awas tested in HeLa and CaSkicervical cancer cell lines by MSP. The re-expression of BLU and RASFF1A incervical cancer cell lines after treatment with 10 μM of 5′-aza-2′-deoxycytidine(Sigma Chemical Co.) for 4 days was assessed by RT-PCR. Total RNA was

631H.-C. Lai et al. / Gynecologic Oncology 104 (2007) 629–635

extracted by using Qiagen RNeasy kit (Qiagen, Valencia, CA). An additionalDNase I digestion procedure (Qiagen) was included in the isolation of RNA toremove DNA contamination following the manufacturer's protocol. Onemicrogram of total RNA from each sample was subject to cDNA synthesisusing Superscript II reverse transcriptase and random hexamer (Invitrogen). ThecDNA generated was used for PCR amplification with the reagents in the PCRmaster mix reagents kit (Applied Biosystems, Foster city, CA, USA) asrecommended by the manufacturer. The reactions were carried out in a thermalcycler (GeneAmp 2400, PE Applied Biosystems). The primers and conditionsfor the PCR were previously described [25,30].

HPV detection and genotyping

The presence of HPV DNA in SCC was detected by L1 consensus PCRfollowed by a reverse line blot [32]. In brief, extracted DNA (100 ng) was PCRamplified with biotin-labeled primers that hybridized with HPV (PGMY pri-

Fig. 1. Methylation analysis of BLU and RASSF1A and effect of 5-Aza-CdR treatmenthe indicated concentration of 5-Aza-CdR. (A) MS-PCR assay of BLU (upper panel)cancer cells. The peripheral blood lymphocyte (PBL) DNAwas used as a negative conMA) was used as a positive control (P). M, methylation-specific PCR product; U, unm(B) Expression of BLU and RASSF1A transcript was analyzed by RT-PCR. Expressibisulfite sequencing. The amplified 465-bp and 406-bp products correspond to −162region, respectively. A total of 41 CpG dinucleotides (CpGs) for RASSF1A and 31 CpEach line represents one of the five clones that were sequenced. Black and white circlCpG sites covered by MSP primers.

mers) or β2-microglobulin. An aliquot of the amplified product was visualizedby ethidium–bromide staining after agarose (1.5%) gel electrophoresis. Theintegrity of the extracted DNA and the HPV genotype were confirmed byhybridization with a strip containing probes for 27 HPV types and for the β2-microglobulin control, and visualized with streptavidin and alkaline phosphatasestaining. DNA sequencing was used to verify novel HPV types beyond thedetection spectrum of the hybridization.

Statistical analysis

The SPSS program (version13) for Windows (SPSS, Chicago, IL, USA) wasused for statistical analyses. Associations between the methylation of genes andclinical parameters, including HPV status, were analyzed using a χ2 test andFisher's exact test, where necessary. The kappa coefficient was computed to testthe interdependence of the methylation of the two genes. The patients werestratified by methylation status and analyzed for their survival probabilities, as a

t in cervical cancer cell lines. HeLa and CaSki cells were treated for 4 days withand RASSF1A (lower panel) in DNA isolated from untreated or treated cervicaltrol, and PBL DNA treated with SssI Methylase (New England Biolabs, Beverly,ethylation-specific PCR; P/10, 1/10 dilution of in vitromethylated DNA in PBL.on of GAPDH was determined as a control for RNA integrity. (C) Summary ofto 303 in the RASSF1A promoter region and −384 to 22 in the BLU promoter

Gs for BLU site within the CpG islands were analyzed and represented as circles.es represent methylation or unmethylation, respectively. The arrows indicate the

632 H.-C. Lai et al. / Gynecologic Oncology 104 (2007) 629–635

function of months and median survival, using Kaplan–Meier survival analysisand the log-rank test. The alpha level of statistical significancewas set at P=0.05.

Results

Methylation and reactivation of the BLU and RASSF1A genesin cervical cancer cell lines

Methylation of the BLU promoter was detected in HeLa andCaSki cells, whereas theRASSF1A promoter remained unmethy-lated (Fig. 1A). Treatment with 10 μM 5′-aza-2′-deoxycytidineresulted in the demethylation of the BLU promoter, which. wasaccompanied by the re-expression of BLU mRNA in HeLa andCaSki cells (Fig. 1B). Bisulfite sequencing confirmed themethylation pattern of BLU and RASSF1A in HeLa and CaSkicells (Fig. 1C).

Clinical correlations with BLU and RASSF1A genemethylation

The methylation of BLU and RASSF1A promoters was de-tected in DNA from clinical samples by a semiquantitativeMSP. Examples of MSP were shown in Fig. 2A. As described inMaterials and methods, we used stringent criteria for scoring asample positive for methylation and disregarded low levels ofmethylation (MSP signal<1/10 dilution of in vitro methylated

Fig. 2. Examples of methylation analysis of the BLU and RASSF1A promoter clintemplate DNA; P, in vitro methylated DNA of PBL; 1/5 and 1/10, dilutions of in vitroof bisulfite sequencing. Tumor case numbers and normal control are indicated at th

DNA). The MSP results were validated by bisulfite sequencing(Fig. 2B). The frequencies of methylation are listed in Table 2.Failures of MSP were excluded from the analysis. Aberranthypermethylation of the BLU gene was detected in 12.5% ofcontrols, 20% of LSIL, 57.4% of HSIL, and 76.9% of SCC(P<0.001). The methylation of the RASSF1A gene was detectedin 17.5% of HSIL and 15.0% of SCC but not in LSIL or thecontrols (P=0.001). In AC, methylation of BLU and RASSF1Awas detected in 43.5% and 26.1% of cancer DNA, respectively;these incidences are significantly higher than those in controls(12.5% and 0%; P=0.005 and P=0.001, respectively). Asshown in Table 3, there was a trend toward the loss of BLUmethylation with advancing clinical stages of SCC (84.8%,67.7%, and 63.6% in stages I, II, and III/IV, respectively).However, this trend was not statistically significant (P=0.080).There was no significant difference in RASSF1A genemethylation in patients at different clinical stages. No associa-tion was identified between the methylation of these two genesand lymph-node metastasis. Survival analysis found noprognostic value for BLU or RASSF1A gene methylation inpatients with SCC of the uterine cervix (data not shown).

BLU and RASSF1A gene methylation and HPV infection

The association between BLU/RASSF1A methylation andHPV infection was assessed (Table 4). In patients with LSIL,

ical specimens. (A) Examples for typical MSP results. H2O, reaction withoutmethylated DNA; PBL, DNA from peripheral blood lymphocyte. (B) Summarye top.

Table 2Methylation of BLU and RASSF1A in cervical neoplasms a

Diagnosis BLU (%) P RASSF1A (%) P

Normal tissue 5/40 (12.5) 0/44 (0.0)LSIL 9/45 (20.0) 0/45 (0.0)HSIL 35/61 (57.4) 11/63 (17.5)SCC 80/104 (76.9) <0.001b 16/107 (15.0) <0.001AC 10/23 (43.5) 0.005 c 6/23 (26.1) 0.001a Detected by methylation-specific PCR.b Compare normal, LSIL, HSIL and SCC.c Compare normal and AC.

Table 4Association between HPV status and BLU/RASSF1A gene methylation

BLU (%) P RASSF1A (%) P

LSIL (n=41)HPV negative 2/22 (9.1) 0/22 (0.0)HPV positive 7/19 (36.8) ns ⁎ 0/19 (0.0) na

HSIL (n=60)HPV negative 12/23 (52.2) 7/23 (30.4)HPV positive 22/37 (59.5) ns 4/39 (10.3) ns ⁎⁎

SCC (n=104)HPV negative 10/11 (90.9) 4/11 (36.4)HPV positive 70/93 (75.3) ns 12/96 (12.5) ns ⁎⁎⁎

AC (n=23)HPV negative 4/10 (40.0) 3/10 (30.0)HPV positive 6/13 (46.2) ns 3/13 (23.1) ns

HSIL/SCC (n=164)HPV negative 22/34 (64.7) 11/34 (32.4)HPV positive 92/130 (70.8) ns 16/135 (11.9) 0.003

ns: not significant; na: not applicable.⁎ P=0.057.

⁎⁎ P=0.082⁎⁎⁎ P=0.058.

633H.-C. Lai et al. / Gynecologic Oncology 104 (2007) 629–635

high-risk HPV infection tended to be associated with higher rateof BLU methylation than those without HPV (36.8% vs 9.1%,P=0.057). No association was identified between BLU methy-lation and HPV infection in HSIL, SCC, or AC. In patients withHSIL, 30.4% with HPV infection showed methylation of theRASSF1A gene, whereas it was methylated in only 10.3% ofpatients without detectable high-risk HPV (P=0.082). Thistrend was similar in patients with SCC (36.4% methylation withHPV vs 12.5% methylation without HPV; P=0.058). Theinverse relation between RASSF1A gene methylation and HPVinfection was statistically significant when these two sets ofpatients (HSIL and SCC) were combined (32.4% methylationwithout HPV vs 11.9% methylation with HPV; P=0.003).When stratified by HPV genotype, no association was iden-tified between HPV genotype and the methylation of theBLU/RASSF1A genes in either SCC or AC.

Association between BLU and RASSF1A gene methylation

Because of the tandem location of the two genes in closeproximity, the interdependence of methylation at the BLU andRASSF1A genes was assessed.We found no association betweenthe methylation of these two genes in HSIL (kappa coefficient=−0.078, P=0.377), SCC (kappa coefficient=0.047, P=0.275),or AC (kappa coefficient=0.0726, P=0.707).

Discussion

In addition to HPVinfection, genetic susceptibility [7,33] andvarious environmental cofactors are also important in cervicalcarcinogenesis [34–36]. However, individual susceptibilityseems to be a minor component and environmental exposure

Table 3Clinical correlations with BLU/RASSF1A gene methylation a

Clinical status BLU (%) P RASSF1A (%) P

SCCStage I 50/59 (84.8) 9/62 (14.5)Stage II 23/34 (67.7) 5/34 (14.7)Stage III and IV 7/11 (63.6) ns ⁎ 2/11 (18.2) ns

Lymph-node metastasisNegative 34/45 (75.6) 8/45 (17.8)Positive 24/29 (82.8) ns 3/31 (9.7) ns

a Detected by methylation-specific PCR; ns: not significant.⁎ P=0.080.

is very difficult to estimate in such a lengthy disease course.Epigenetic modifications affecting gene expression withoutchanges in the DNA sequences have become a focus of intenseresearch because DNA modifications may serve as an interfaceof host–environmental interactions [13,14]. The present studyreported the methylation of a recently characterized putativetumor suppressor gene, BLU, and a less well-characterized one,RASSF1A, in a full spectrum of cervical neoplasia.

BLU is normally expressed in the uterine cervix [25].Limited evidence suggests that BLU may be a stress-responsivegene [25]. The mechanism of BLU methylation in cervicalcarcinogenesis is far from understood. A recent study demon-strated that BLU is an E2F-regulated gene and is therefore cell-cycle regulated [25]. The binding of E2F may facilitate therecruitment of DNA methyltransferase 1 (DNMT1) through itsinteraction with retinoblastoma protein (RB), resulting in themethylation of the promoter region of the target gene [37]. Themost recent reports have demonstrated that transcription-factor-directed methylation of target promoters does occur [38]. Wespeculate that the binding of E2F to the promoter of the BLUgene under various stresses may result in BLU silencingthrough DNA methylation, which could partly explain its site-specific methylation in carcinogenesis. The independence of themethylation of the BLU and RASSF1A genes, which has alsobeen reported in other cancers, indicates that DNA methylationin cancer development can be localized within a specificsequence, rather than spreading over a long delimited distance[25,39]. Elucidating the role of BLU and the network of otherE2F-responsive genes in response to HPV infection and otherenvironmental factors should greatly facilitate our understand-ing of cervical carcinogenesis.

The progressive increase in RASSF1A gene methylationwith increasing disease severity also indicates a role for theRASSF1A gene in the development of cervical cancer. Althoughaberrant methylation of the RASSF1A promoter has frequentlybeen detected in several tumor types, reports of this in cervical

634 H.-C. Lai et al. / Gynecologic Oncology 104 (2007) 629–635

cancer and precancerous lesions have been quite limited andcontroversial [40]. An earlier report showed no methylation ofthe RASSF1A promoter in any of 31 cases of SCC from theKorean population [41]. However, reports from other groupsshowed 30% methylation in 33 SCCs in a Chinese population[42] and 10% in 42 SCCs in the United States [43]. This studydemonstrated RASSF1A gene methylation in 15% of 104 SCCpatients and in 18.3% of 60 HSIL patients, but none in LSILpatients. In AC, the promoter of RASSF1Awas hypermethylatedin 26.1% of patients, which indicates that RASSF1A hyper-methylation may play an especially important role in the deve-lopment of cervical adenocarcinoma [41].

The association between HPV infection and RASSF1Amethylation is also interesting. Studies have shown a trend foran inverse correlation between RASSF1A methylation and high-risk HPV (type 16/18) infection in cervical SCC [43] and in headand neck cancers [44]. The underlying mechanism is not wellunderstood. There is evidence that RASSF1A can induce cell-cycle arrest by engaging the Rb family cell-cycle checkpoint[45]. RASSF1A inhibits the accumulation of native cyclin D1,and RASSF1A-induced G1/S arrest can be relieved by HPV E7.The similar functions in cell-cycle checkpoint regulation ofRASSF1Amethylation and the HPVE7 oncoprotein may explainthis inverse correlation. This study confirms the inversecorrelation between HSIL/SCC and RASSF1A methylation in alarger series and across a broader spectrum of high-risk HPVgenotypes.

The current concept of methylation in cancer biology is thatthe role of site-specific methylation is causal or promotional incarcinogenesis. However, as shown in this study, the signifi-cance of changes in methylation late in cancer developmentremains largely unclear. We have identified a trend in the loss ofBLU methylation in advanced stages of cervical cancer, whichcontradicts the concept of monoclonal evolution in cancerdevelopment and the failure to detect DNA demethylase in vivoto date. Recently, it was shown that the epigenetic regulation byDNA methylation of the ER-β promoter in prostate cancer istumor-stage specific, with the promoter highly methylated in insitu cancers but much less methylated in lymph-node and bonymetastases [46]. The phenomenon and significance of stage-specific methylation in cancer remains to be confirmed andwarrants further investigation.

From a clinical perspective, the present study demonstratesan association between BLU/RASSF1A gene methylation andcervical cancer, which supports the concept of using a panel ofDNA methylations in the screening of cervical neoplasias [11].Further identification of novel genes methylated during cervicalcarcinogenesis and improvement in quantitative rather thanqualitative analysis of methylation will be of great value in themolecular screening of cervical cancer.

Acknowledgments

The authors are grateful to Rui-Len Huang for excellenttechnical assistance. This work was supported in part by thefollowing grants: TSGH-C93-5-S02∼3 and TSGH-C95-7-S01∼4 from the Tri-Service General Hospital, NSC94-2314-B-

016-046 and NSC94-2622-B-016-001from the National ScienceCouncil, ROC, AFTYGH-9401-OB9402∼4 from Army ForcesTaoyuan General Hospital, the C. Y. Chai Foundation forAdvancement of Education, Sciences and Medicine and Teh-Tzer Study Group for Human Medical Research Foundation.

References

[1] Estimated cervical cancer mortality in selected countries, 2000. J NatlCancer Inst 2002;94:246.

[2] Walboomers JM, Jacobs MV, Manos MM, Bosch FX, Kummer JA, ShahKV, et al. Human papillomavirus is a necessary cause of invasive cervicalcancer worldwide. J Pathol 1999;189:12–9.

[3] Ylitalo N, Josefsson A, Melbye M, Sorensen P, Frisch M, Andersen PK,et al. A prospective study showing long-term infection with humanpapillomavirus 16 before the development of cervical carcinoma in situ.Cancer Res 2000;60:6027–32.

[4] Syrjanen KJ. Spontaneous evolution of intraepithelial lesions according tothe grade and type of the implicated human papillomavirus (hpv). Eur JObstet Gynecol Reprod Biol 1996;65:45–53.

[5] zur Hausen H. Papillomaviruses causing cancer: evasion from host–cellcontrol in early events in carcinogenesis. J Natl Cancer Inst 2000;92:690–8.

[6] Ylitalo N, Sorensen P, Josefsson A, Frisch M, Sparen P, Ponten J, et al.Smoking and oral contraceptives as risk factors for cervical carcinoma insitu. Int J Cancer 1999;81:357–65.

[7] Magnusson PK, Gyllensten UB. Cervical cancer risk: is there a geneticcomponent? Mol Med Today 2000;6:145–8.

[8] Harris CP, Lu XY, Narayan G, Singh B, Murty VV, Rao PH.Comprehensive molecular cytogenetic characterization of cervical cancercell lines. Genes Chromosomes Cancer 2003;36:233–41.

[9] Mullokandov MR, Kholodilov NG, Atkin NB, Burk RD, Johnson AB,Klinger HP. Genomic alterations in cervical carcinoma: losses of chromo-some heterozygosity and human papilloma virus tumor status. Cancer Res1996;56:197–205.

[10] Mitra AB. Genetic deletion and human papillomavirus infection in cervicalcancer: loss of heterozygosity sites at 3p and 5p are important geneticevents. Int J Cancer 1999;82:322–4.

[11] Feng Q, Balasubramanian A, Hawes SE, Toure P, Sow PS, Dem A, et al.Detection of hypermethylated genes in women with and without cervicalneoplasia. J Natl Cancer Inst 2005;97:273–82.

[12] Feinberg AP, Tycko B. The history of cancer epigenetics. Nat Rev, Cancer2004;4:143–53.

[13] Jones PA, Baylin SB. The fundamental role of epigenetic events in cancer.Nat Rev, Genet 2002;3:415–28.

[14] Laird PW. The power and the promise of DNA methylation markers. NatRev, Cancer 2003;3:253–66.

[15] Bird AP. Cpg-rich islands and the function of DNA methylation. Nature1986;321:209–13.

[16] Larsen F, Gundersen G, Lopez R, Prydz H. Cpg islands as gene markers inthe human genome. Genomics 1992;13:1095–107.

[17] Knudson AG. Two genetic hits (more or less) to cancer. Nat Rev, Cancer2001;1:157–62.

[18] Widschwendter A, Muller HM, Fiegl H, Ivarsson L, Wiedemair A, Muller-Holzner E, et al. DNA methylation in serum and tumors of cervical cancerpatients. Clin Cancer Res 2004;10:565–71.

[19] Widschwendter A, Gattringer C, Ivarsson L, Fiegl H, Schneitter A,Ramoni A, et al. Analysis of aberrant DNA methylation and humanpapillomavirus DNA in cervicovaginal specimens to detect invasivecervical cancer and its precursors. Clin Cancer Res 2004;10:3396–400.

[20] Baylin SB, Herman JG. DNA hypermethylation in tumorigenesis:epigenetics joins genetics. Trends Genet 2000;16:168–74.

[21] Jones PA, Laird PW. Cancer epigenetics comes of age. Nat Genet 1999;21:163–7.

[22] Hamoudi RA, El-Hamidi A, Du MQ. Identification of novel prognosticmarkers in cervical intraepithelial neoplasia using ldmas (loh datamanagement and analysis software). BMC Bioinformatics 2005;6:18.

635H.-C. Lai et al. / Gynecologic Oncology 104 (2007) 629–635

[23] Wistuba II, Montellano FD, Milchgrub S, Virmani AK, Behrens C, ChenH, et al. Deletions of chromosome 3p are frequent and early events in thepathogenesis of uterine cervical carcinoma. Cancer Res 1997;57:3154–8.

[24] Chu TY, Shen CY, Lee HS, Liu HS. Monoclonality and surface lesion-specific microsatellite alterations in premalignant and malignant neoplasiaof uterine cervix: a local field effect of genomic instability and clonalevolution. Genes Chromosomes Cancer 1999;24:127–34.

[25] Qiu GH, Tan LK, Loh KS, Lim CY, Srivastava G, Tsai ST, et al. Thecandidate tumor suppressor gene blu, located at the commonly deletedregion 3p21.3, is an e2f-regulated, stress-responsive gene and inactivatedby both epigenetic and genetic mechanisms in nasopharyngeal carcinoma.Oncogene 2004;23:4793–806.

[26] Agathanggelou A, Dallol A, Zochbauer-Muller S, Morrissey C, HonorioS, Hesson L, et al. Epigenetic inactivation of the candidate 3p21.3suppressor gene blu in human cancers. Oncogene 2003;22:1580–8.

[27] Dreijerink K, Braga E, Kuzmin I, Geil L, Duh FM, Angeloni D, et al. Thecandidate tumor suppressor gene, rassf1a, from human chromosome3p21.3 is involved in kidney tumorigenesis. Proc Natl Acad Sci U S A2001;98:7504–9.

[28] Burbee DG, Forgacs E, Zochbauer-Muller S, Shivakumar L, Fong K, GaoB, et al. Epigenetic inactivation of rassf1a in lung and breast cancers andmalignant phenotype suppression. J Natl Cancer Inst 2001;93:691–9.

[29] Yang YF, Tsao YP, Yin CS, Chen SL, Chu TY. Overexpression of theproto-oncogene c-jun in association with low-risk type specific humanpapillomavirus infection in condyloma acuminata. J Med Virol 1996;48:302–7.

[30] Tischoff I, Markwarth A, Witzigmann H, Uhlmann D, Hauss J,Mirmohammadsadegh A, et al. Allele loss and epigenetic inactivation of3p21.3 in malignant liver tumors. Int J Cancer 2005;115:684–9.

[31] Esteller M, Hamilton SR, Burger PC, Baylin SB, Herman JG. Inactivationof the DNA repair gene o6-methylguanine-DNA methyltransferase bypromoter hypermethylation is a common event in primary humanneoplasia. Cancer Res 1999;59:793–7.

[32] Gravitt PE, Peyton CL, Apple RJ, Wheeler CM. Genotyping of 27 humanpapillomavirus types by using l1 consensus pcr products by a single-hybridization, reverse line blot detection method. J Clin Microbiol1998;36:3020–7.

[33] Horng JT, Hu KC, Wu LC, Huang HD, Lin FM, Huang SL, et al.Identifying the combination of genetic factors that determine susceptibilityto cervical cancer. IEEE Trans Inf Technol Biomed 2004;8:59–66.

[34] Smith JS, Herrero R, Bosetti C, Munoz N, Bosch FX, Eluf-Neto J, et al.Herpes simplex virus-2 as a human papillomavirus cofactor in the etiologyof invasive cervical cancer. J Natl Cancer Inst 2002;94:1604–13.

[35] Moreno V, Bosch FX, Munoz N, Meijer CJ, Shah KV, Walboomers JM,et al. Effect of oral contraceptives on risk of cervical cancer in womenwith human papillomavirus infection: the iarc multicentric case-controlstudy. Lancet 2002;359:1085–92.

[36] Haverkos HW, Soon G, Steckley SL, Pickworth W. Cigarette smoking andcervical cancer: Part i: a meta-analysis. Biomed Pharmacother2003;57:67–77.

[37] Robertson KD, Ait-Si-Ali S, Yokochi T, Wade PA, Jones PL, Wolffe AP.Dnmt1 forms a complex with rb, e2f1 and hdac1 and repressestranscription from e2f-responsive promoters. Nat Genet 2000;25:338–42.

[38] Zhang Q, Wang HY, Marzec M, Raghunath PN, Nagasawa T, WasikMA. Stat3- and DNAmethyltransferase 1-mediated epigenetic silencing ofshp-1 tyrosine phosphatase tumor suppressor gene in malignant tlymphocytes. Proc Natl Acad Sci U S A 2005;102:6948–53.

[39] Hesson L, Bieche I, Krex D, Criniere E, Hoang-Xuan K, Maher ER, et al.Frequent epigenetic inactivation of rassf1a and blu genes located within thecritical 3p21.3 region in gliomas. Oncogene 2004;23:2408–19.

[40] Pfeifer GP, Dammann R. Methylation of the tumor suppressor gene rassf1ain human tumors. Biochemistry 2005;70:576–83 [Mosc].

[41] Cohen Y, Singer G, Lavie O, Dong SM, Beller U, Sidransky D. The rassf1atumor suppressor gene is commonly inactivated in adenocarcinoma of theuterine cervix. Clin Cancer Res 2003;9:2981–4.

[42] Yu MY, Tong JH, Chan PK, Lee TL, Chan MW, Chan AW, et al.Hypermethylation of the tumor suppressor gene rassfia and frequentconcomitant loss of heterozygosity at 3p21 in cervical cancers. Int J Cancer2003;105:204–9.

[43] Kuzmin I, Liu L, Dammann R, Geil L, Stanbridge EJ, Wilczynski SP, et al.Inactivation of ras association domain family 1a gene in cervicalcarcinomas and the role of human papillomavirus infection. Cancer Res2003;63:1888–93.

[44] Dong SM, Sun DI, Benoit NE, Kuzmin I, Lerman MI, Sidransky D.Epigenetic inactivation of rassf1a in head and neck cancer. Clin CancerRes 2003;9:3635–40.

[45] Shivakumar L, Minna J, Sakamaki T, Pestell R, White MA. The rassf1atumor suppressor blocks cell cycle progression and inhibits cyclin d1accumulation. Mol Cell Biol 2002;22:4309–18.

[46] Zhu X, Leav I, Leung YK, Wu M, Liu Q, Gao Y, et al. Dynamicregulation of estrogen receptor-beta expression by DNA methylationduring prostate cancer development and metastasis. Am J Pathol2004;164:2003–12.

[47] Herman JG, Graff JR, Myohanen S, Nelkin BD, Baylin SB. Methylation-specific PCR: a novel PCR assay for methylation status of CpG islands.Proc Natl Acad Sci U S A 1996(93):9821–6.