Patients with advanced stage invasive cervical cancer (CC) exhibit highly complex genomic alterations and respond poorly to conventional treatment protocols. In our efforts to understand the molecular genetic basis of CC, we examined the role of Fanconi Anemia (FA)-BRCA pathway. Here, we show that FANCF gene is disrupted by either promoter hypermethylation and/or deregulated gene expression in a majority of CC. Inhibition of DNA methylation and histone deacetylases induces FANCF gene re-expression in CC cell lines. FANCF-deregulated CC cell lines also exhibit a chromosomal hypersensitivity phenotype after exposure to an alkylating agent, a characteristic of FA patients. We also show the involvement of BRCA1 gene by promoter hypermethylation or down-regulated expression in a small subset of CC patients. Thus, we have found inactivation of genes in the FA-BRCA pathway by epigenetic alterations in a high proportion of CC patients, suggesting a major role for this pathway in the development of cervical cancer. Thus, these results have important implications in understanding the molecular basis of CC tumorigenesis and clinical management in designing targeted experimental therapeutic protocols.
Approximately 500,000 new cases of cervical cancer (CC) are diagnosed worldwide each year, and the majority of affected women with advanced cancer die (1) . Human papillomavirus is considered as a common risk factor for CC. Most women with early-stage tumors can be cured by surgery and radiotherapy. However, treatment of late-stage and recurrent CC remains largely ineffective. The poor prognosis of advanced CC is due to lack of understanding of its biology at the molecular level and of effective treatment regimens. CC cells harbor complex genomic alterations and are highly unstable (2) . A better understanding of molecular alterations could therefore facilitate the design of new and targeted therapies for individualized treatment.
Fanconi anemia (FA) is an autosomal recessive chromosomal instability syndrome characterized by hypersensitivity to DNA cross-linking agents and predisposition to cancer, especially leukemia (3) . FA patients are also prone to various solid malignancies, including squamous cell carcinoma. The lifetime risk for developing CC in FA patients is also significantly higher than in control populations (3 , 4) . FA is a genetically heterogeneous disease with genes for seven FA complementation (FANC) groups identified (5) . FANC genes are essential in DNA repair pathways in a normal cellular response to cisplatin and other DNA cross-linking agents. FANC proteins interact with BRCA genes in a pathway that involves a number of other genes (6) . Recently, it has been shown that promoter hypermethylation of FANCF gene disrupts the FA-BRCA pathway, resulting in cisplatin resistance (7) . FANCF promoter hypermethylation has also shown to occur in squamous cell carcinomas of lung and oral cavity (8) .
Because FA patients exhibit an increased risk for the development of CC, we investigated whether the FA-BRCA pathway is altered in this tumor.
Materials and Methods
Patients, Tumor Tissues, and Cell Lines.
A total of 100 CC DNA samples derived from 91 at-diagnosis tumor biopsies from invasive CCs and nine cell lines was used in these studies. The tumor biopsies were ascertained from patients evaluated at the Instituto Nacional de Cancerologia (Santa Fe de Bogota, Colombia) and from the Department of Obstetrics and Gynecology of Friedrich Schiller University (Jena, Germany) after appropriate informed consent and approval of protocols by Institutional Review Boards. The primary tumors were clinically classified as International Federation of Gynecologists and Obstetricians stage IB (18 tumors), IIB (23 tumors), IIIB (47 tumors), and IV (3 tumors). Histologically, 86 tumors were classified as squamous cell carcinoma and 5 as adenocarcinoma. Clinical information was collected as described previously (9) . Follow-up ranged from 1 to 72 months. Cytologically diagnosed cervical swabs from 18 normal, 11 atypical squamous cells of undetermined significance, 19 low- and 7 high-grade squamous intraepithelial lesions were collected in PBS from patients attending the Gynecologic Oncology Clinic at Columbia University Medical Center (New York, NY), after appropriate informed consent. The CC cell lines HeLa, SiHa, SW756, C-4I, Ca Ski, C-33A, HT-3, MS751, and ME-180 were obtained from the American Type Culture Collection (Manassas, VA) and grown in tissue culture according to the supplier’s recommendations. DNA and/or RNA were isolated from frozen tumor tissues or cell pellets by standard methods. RNA was obtained from 20-μm sections with H&E staining of adjacent sections to evaluate for tumor content. Only sections that contained >70% tumor cells were used for RNA preparation.
Genomic DNA was treated with sodium bisulfite as described previously (9) . Placental DNA treated in vitro with SssI methyltransferase (New England Biolabs, Beverly, MA) and normal lymphocyte DNA treated similarly with sodium bisulfite were used as controls for methylated and unmethylated templates, respectively. The primers for amplification of methylated and unmethylated promoters of BRCA1, BRCA2, and FANCF (two sets covering regions −86 to +97 and +279 to +431 bp) were designed as described previously (9) . PCR products were run on 2% agarose gels and visualized after ethidium bromide staining.
Mutation Analysis and Human Papillomavirus Detection.
Six sets of primers covering the coding region of FANCF gene were designed to amplify genomic sequences up to 250 bp. Single-strand conformation polymorphism analysis was performed on PCR products generated in the presence of α [32P]dCTP and running 6% nondenaturing polyacrylamide gels containing 10% glycerol. Purified PCR products from representative tumors were sequenced to identify mutations. Human papillomavirus types were identified as described earlier (9) .
Drug Treatment and Chromosome Preparations.
Cells in culture were treated with mitomycin C at a concentration of 20, 40, or 80 ng for 48 h and exposed to Colcemid for the last 2 h. Metaphases were prepared by standard protocols. A total of 100–150 metaphases was analyzed from replicate experiments to identify chromatid- and chromosome-type aberrations. Cell lines were treated with 5-aza-2′deoxycytidine (Aza-C) and trichostatin (TSA) as described previously (9) .
Reverse Transcription-PCR Analysis.
Total RNA isolated from treated and untreated cell lines, tumor tissues, and three normal cervix uteri obtained from different commercial sources was reverse transcribed as described previously (9) . A multiplex semiquantitative analysis of gene expression of BRCA1, BRCA2, FANCFA, FANCC, FANCE, FANCF, and FANCG was performed in replicate experiments as described previously (9) . A given gene was considered down-regulated in a tumor when the level of mRNA was less than mean-2SD calculated from the expression in normal cervix. All primers used in the present study are available from the authors upon request.
Oligonucleotide Microarray Gene Expression Analysis.
Biotinylated cRNA preparation and hybridization to Affymetrix U133A oligonucleotide microarray (Affymetrix, Santa Clara, CA), which contains 33,000 genes, was performed by the standard protocols supplied by the manufacturer. Arrays were subsequently developed and scanned to obtain quantitative gene expression levels. Expression values for the genes were determined using the Affymetrix GENECHIP software and the Global Scaling option, which allows a number of experiments to be normalized to one target intensity to account for the differences in global chip intensity. To perform the supervised gene expression analysis, we used the [email protected] software platform, which is a gene expression analysis tool based on the pattern discovery algorithm Structural Pattern Localization Analysis by Sequential Histograms (10 , 11) . The support value for the analysis was chosen as n = no – 2 (no, number of samples in each of the two groups).
Association between promoter hypermethylation and various clinical and genetic characteristics was performed using a χ2 test.
Results and Discussion
FANCF and BRCA1 Genes Are Hypermethylated in CC.
We previously reported that the 11p15 region to which the FANCF gene maps exhibit frequent loss of heterozygosity (LOH) in CC (12 , 13) . Because the FA-BRCA pathway has been shown to be disrupted by promoter hypermethylation of FANCF gene in ovarian, lung, and oral cancers (7 , 8) , we reasoned that FANCF gene may be the target tumor suppressor at 11p15 LOH in CC. To test this hypothesis, we examined the status of FANCF promoter hypermethylation in DNA samples from 18 normal cervical epithelia and 100 CC. FANCF promoter hypermethylation was not detected in normal cervical epithelium. However, hypermethylation was found in 30% of primary tumors (27 of 91 specimens) as well as in three of nine cell lines (Fig. 1 ⇓ ; Table 1 ⇓ ). We previously reported the status of 11p LOH in 42 of the primary tumors included in the present study (13) . Fourteen of the 42 had an 11p LOH. Promoter hypermethylation of FANCF was observed in 5 of the 14 (35.7%) tumors that exhibited 11p LOH compared with 6 of 28 (21.4%) tumors without evidence of 11p LOH. These differences were statistically not significant, suggesting that the changes affecting 11p by LOH and promoter hypermethylation in CC may be independent events.
Correlation of FANCF hypermethylation with clinicopathological features showed no relationship with stage, size of the tumor, clinical outcome, and human papillomavirus type. However, patients <45 years showed a significantly higher frequency of promoter methylation compared with the patients >45 years of age (44.4%; 20 of 45 tumors versus 15.2%; 7 of 46 tumors; P = 0.001). These data, therefore, suggest that FANCF promoter hypermethylation plays a role in initiation or progression of CC in younger patients. It has also been previously reported that inherited or somatic mutations in FANCC and FANCG genes predispose to pancreatic cancer at younger age (14) . Taken together, these results suggest that alterations of genes in the FA-BRCA pathway may result in increased risk for the development of cancer in younger age group patients.
Promoter hypermethylation is also known to inactivate other genes in the FA-BRCA pathway, including BRCA1 and BRCA2 (15) . In the same cohort of cases studied here, we previously reported BRCA1 promoter hypermethylation in 6.1% of CC patients (9) . Promoter hypermethylation of FANCF and BRCA1 genes was mutually exclusive in these tumors. In the present study, we also studied BRCA2 promoter methylation and found it in none of the tumors. The frequency of FANCF promoter hypermethylation seen in the present study is the highest in any tumor reported thus far (7 , 8) . Thus, in ≥35% of CC patients, either FANCF or BRCA1 was inactivated by promoter hypermethylation, suggesting a major role for the FA-BRCA pathway in this tumor. To identify the role of FANCF in CC progression, we studied DNA obtained from 37 pap smears diagnosed at various stages of precancerous lesions by methylation-specific PCR and found no evidence of hypermethylation of the promoter, suggesting that the FANCF inactivation is a late event in the tumor development.
FANCF Expression Is Down-Regulated in Most CC Cell Lines.
The FA proteins (Fanca, Fancc, Fance, Fancf, and Fancg) that are ubiquitously expressed in dividing normal cells play a major role by forming nuclear complex mediating monoubiquitination of Fancd2 in response to DNA damage (6) . To test the role of FA-BRCA pathway genes, BRCA1, BRCA2, and five FANC genes (FANCA, FANCC, FANCE, FANCF, and FANCG) were studied for the expression by semiquantitative reverse transcription-PCR analysis in CC cell lines and/or primary tumors. Examination of steady-state levels of BRCA1 and BRCA2 genes in nine cell lines and 7 primary tumors, which did not exhibit promoter methylation of these genes, showed no evidence of down-regulated expression of BRCA2. However, a relative decrease in the expression of BRCA1 was found in 3 of 7 (43%) primary tumors but not in the cell lines (Fig. 2B) ⇓ .
We then examined the levels of mRNA of five FA genes (FANCA, FANCC, FANCE, FANCF, and FANCG) in nine CC cell lines. The FANCA, FANCC, FANCE, and FANCG genes showed no evidence of decreased expression (data not shown). However, the FANCF gene showed a down-regulated expression in all three cell lines (SiHa, SW756, and ME-180) that exhibited promoter hypermethylation (Fig. 2A) ⇓ . In addition, three other cell lines (C-4I, MS751, and HeLa), which did not show promoter methylation, also showed down-regulated FANCF mRNA compared with normal cervix (Fig. 2A ⇓ ; Table 1 ⇓ ). Overall, six of nine (66.6%) cell lines exhibited FANCF gene down-regulation. Thus, the present results suggest that FANCF is inactivated by mechanisms other than complete methylation of its promoter such as partial methylation or mutation. To evaluate the latter possibility, we examined mutations in the coding region in a panel of 16 methylated and 16 unmethylated CC. No nonsense or frameship mutations could be found. However, this analysis identified five different sequence variants present both in tumor as well as in the corresponding peripheral blood DNA, four in the coding region and one at the 5′-noncoding region, that were not reported in the single nucleotide polymorphism (SNP) database. 8 These were TCG→TTG at the −10 position, CGC→CGT at codon Arg32 without change in amino acid, GCC→GTC that changes Ala186→Val, AAG→AGG that changes Lys203→Arg, and CCT→CTT that changes Pro320→Leu. The Pro320Leu has previously been reported as a normal variant (16) . Whether these germ-line variants truly represent polymorphisms or inherited mutations that predispose the patients to cancer remain to be tested.
Because FANCF is a component of FA complex and plays a crucial role in DNA damage repair by mediating downstream FANCD2 monoubiquitination, we wanted to examine if gene expression differences exist between tumors carrying hypermethylated and unmethylated FANCF promoter (6) . Supervised analysis of expression profile of 7 FANCF methylated tumors and 8 unmethylated tumors identified a small number of 12 genes differentially expressed between the two groups (Fig. 3) ⇓ . Microarray analysis confirmed FANCF expression is associated with promoter hypermethylation (Fig. 3) ⇓ . Nucleoplasmin (NPM1) is the only gene that showed higher levels of expression in promoter hypermethylated tumors compared with unmethylated tumors. Eleven genes, most of which are members of immune response family, showed down-regulated expression in FANCF-hypermethylated tumors (Fig. 3) ⇓ . Role of these genes in relation to DNA repair pathway is unclear. Therefore, these results suggest that FANCF inactivation by promoter hypermethylation is only upstream regulator in DNA damage response, and no consistent quantitative changes in expression of downstream genes occur in normal cellular conditions.
FANCF Is Reactivated in Response to Cellular Exposure to Demethylating Agents.
To test whether other epigenetic mechanisms such as partial methylation and histone deacetylation play a role, we examined FANCF expression after treatment with Aza-C, TSA, or both. Aza-C or TSA induced reactivation of FANCF in all three cell lines (SiHa, SW756, and ME-180) that exhibited complete promoter hypermethylation and down-regulated expression. Two additional cell lines (C-33A and MS751) without evidence of promoter hypermethylation also induced FANCF expression in response to exposure to Aza-C or TSA (Fig. 2C ⇓ ; Table 1 ⇓ ). Two other cell lines (C4-I and HeLa) with unmethylated promoters also induced expression of FANCF after exposure to TSA but not with Aza-C. Thus, seven of nine (78%) CC cell lines exhibited deregulated expression by epigenetic mechanisms of inactivation.
DNA hypermethylation-mediated gene silencing is closely associated with histone modifications such as methyl-H3-K9. In this regard, DNA-demethylating agents Aza-C and histone deacetylase inhibitor TSA reactivates expression of epigenetically silenced genes (17) . Although DNA hypermethylation is essential to maintain repressive state of histone code, histone modifications precede DNA hypermethylation in silencing specific genes (18 , 19) . In the present study, reactivation of FANCF after exposure to TSA in the absence of promoter methylation suggests that key histone modifications, either by direct or indirect involvement of promoter methylation, also play a role in down-regulating FANCF gene expression in CC.
CC Cell Lines Exhibit Chromosomal Hypersensitivity to Mitomycin C.
FA cells exhibit increased chromosome breakage and multiradial formation after exposure to DNA-damaging agents (20) . To examine the chromosomal sensitivity in relation to FANCF inactivation, we exposed four cell lines (one without evidence of defects in FANCF and three with FANCF deregulated expression) to various concentrations of mitomycin C. The cell lines ME-180 and SW756 with methylated promoters and the C-4I cell line without a methylated promoter but with down-regulated and reactivated FANCF expression upon treatment to TSA showed a several-fold increase in chromosomal breakage, including multiradials compared with CaSki, a cell line that did not exhibit any FANCF defect (Fig. 4) ⇓ . Frequency of aberrant metaphases was increased 2–3.5-fold in FANCF-defective cell lines (C-4I, ME-180, and SW756) compared with the CaSki cell line (Fig. 4C) ⇓ . However, the metaphases exhibiting multiradial chromosomes were 14–25-fold higher in cell lines with FANCF epigenetic alteration (Fig. 4D) ⇓ . We also found that the cell lines C-4I and SW756 exhibit a high frequency of polyploid metaphases with extensive chromosomal breakage (Fig. 4B) ⇓ . Thus, this pronounced mitomycin C-induced chromosomal hypersensitivity exhibited by CC cell lines with FANCF epigenetic alterations is similar to the phenotype seen in FA patients.
Our results suggest that epigenetic inactivation of FA-BRCA pathway is common in CC. Tumor cells carrying FANCF or BRCA1 promoter hypermethylation are hypersensitive to DNA-damaging drugs and may result in pronounced tumor death because of their underlying defect in FA-BRCA pathway. It has been shown earlier that progression of ovarian cancer is related to FANCF promoter hypermethylation and demethylation of the promoter results in cisplatin resistance (7) . Additional studies are required to determine the role of FANCF promoter hypermethylation in causing cisplatin resistance in CC. However, identification of this important molecular pathway may have implications in designing appropriate chemotherapy regimen to be used in combination with radiotherapy and a genotype-based therapy may produce a long-term improvement in the treatment of advanced stage CC.
We thank Dr. Raju S. K. Chaganti, Memorial Sloan-Kettering Cancer Center (New York, NY) for critical reading of the manuscript.
Grant support: Entertainment Industry Foundation Award from the Herbert Irving Comprehensive Cancer Center (Columbia University) and NIH Grant CA095647.
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 with 18 U.S.C. Section 1734 solely to indicate this fact.
Requests for reprints: Murty Vundavalli, Department of Pathology, College of Physicians & Surgeons of Columbia University, 630 West 168th Street, New York, New York 10032. Phone: (212) 305-7914; Fax: (212) 305-5498; E-mail:
- Received January 25, 2004.
- Revision received February 23, 2004.
- Accepted March 3, 2004.
- ©2004 American Association for Cancer Research.