Twenty six squamous cell carcinomas (SCC) of uterine cervix and 14 specimens of adjacent normal tissues from patients with I, II and III FIGO stages of disease were collected under the approval of the Institutional Review Board of the Blochin Cancer Research Center. Informed consent was obtained from the patients prior to surgery. All tumors were HPV16/18-positive and E7 viral genes were expressed in all of them according to PCR analysis.

The human cervical carcinoma cells HeLa, SiHa, CaSki, C-4-I and C33A (American Type Culture Collection, Rockville, MD) were maintained in DMEM supplemented with 10% FCS and 2mM glutamine.

Serial sections (4–5 μm thick) of neutral formalin-fixed and paraffin embedded tissues were used. The first section was stained with hematoxylin-eosin to confirm the diagnosis. Three independent pathologists (G.F., L.Z. and Y.A.) verified the diagnosis. Next sections were stained with antibodies. Normal cervical epithelium was included into every set of slides for staining as a negative control 18. The procedures of epitope retrieval, blocking of nonspecific binding and endogenous peroxidase activity, incubation with the p16^ink4a-specific monoclonal antibodies (CINtec™ Histology Kit, DakoCytomation) were conducted according to Klaes et al 17. To visualize the results, the slides, which had been exposed to the primary antibodies, were treated with the EnVision™ +Dual Link DakoCytomation reagent. The reaction was developed with either AEC or DAB chromogen. The slides were counterstained with Mayer's hematoxylin. To detect p16^ink4a in cells, they were grown on the HistoBond adhesive coated slides (SMT Geraetehandel GmbH, Germany), fixed with 4% paraformaldehyde, washed with PBS and processed similarly to the deparaffinized histological slides.

DNAs and RNAs were isolated from frozen tissues or cells by homogenization in the 4, 5 M solution of guanidine isothiocyanate and centrifugation through cesium chloride cushion as described previously 28.

DNAs (1.5–3.0 μg) were denaturated by incubation in solution of 0.35 M NaOH for 10 minutes at 37°C. The reaction of bisulfite conversion was carried out in a solution of sodium bisulfite (3.1 M) and hydroquinone (0.6 mM) at pH 5.5 and 50°C during 4 hours. After desalting DNAs were desulfonated by incubation in solution of 0.3 M NaOH at 37°C for 10 min, then the solution was neutralized with 5 M ammonium acetate (pH 7.0) and DNAs were ethanol-precipitated. The pellets were dissolved in distilled H₂O and the DNA aliquots (100–300 ng) were used for the PCR amplifications by JumpStart RedTaq DNA polymerase according to manufactur's procedure (Sigma, USA) in 30-μkl reaction mixtures under the following conditions: 94°C, 4 min; 35 amplification cycles (94°C, 30 s; 64°C, 60 s; 72°C, 30 s) and finally at 72°C, 5 min. The p16INK4a primers that corresponded to the upper strand of bisulfite-modified DNA and did not contain CpG dinucleotides in the original sequences were used: sense 5' GAG GAG GGG TTG GTT GGT TAT TAG AG (1115) and antisense 5' TAC CTA ATT CCA ATT CCC CTA CAA AC (1418). The positions of the primers are given in base pairs related to the gene sequence [GenBank: X94154 + U12818]. In some cases the combined primer (5' M13+ sense primer) was used for direct sequencing of PCR product 29. The resulting PCR products were fractionated in low melting agarose (Sigma), purified using QIAEX II kit (Qiagen) and either used for automated sequencing on DNA Sequencer (ABI Prism 3100-Avant Genetic Analyzer) or were cloned into pGEM-T Easy vector (Promega, USA) and then DNAs of individual clone were sequenced. The same total PCR products were used for determination of methylation status of p16 INK4a promoter region by of MSP 30. The primer sequences specific for methylated (sense 5 ' TTA TTA GAG GGT GGG GCG GAT CGC, antisense 5' GAC CCC GAA CCG CGA CCG TAA) and unmethylated (sense 5' TTA TTA GAG GGT GGG GTG GAT TGT, antisense 5' CAA CCC CAA ACC ACA ACC ATA A) variant of p16INK4a and the conditions of amplification were as described by Herman et al 30. The positions of these primers are indicated at the map of p16INK4 promoter region (fig. 1a.)

The mRNA level of p16INK4a gene was determined by reverse transcription combined with polymerase chain reaction (RT-PCR). RT was carried out in 20 μl reaction mixture containing 1 μg of total cellular RNA, treated with DNAase I (Invitrogene), 1× buffer (50 mM Tris, pH 8.3, 61.5 mM KCl, 3 mM MgCl₂); 1 mM of each of dATP, dTTP, dGTP and dCTP; 20 pmol of random hexamer; 200 U of reverse transcriptase (Superscript II Rnase H^-, Invitrogen, USA) at 25°C for 10 min and then at 42°C for 50 min. PCR was performed with the primers: sense, 5' AGC CTT CGG CTG ACT GGC TGG; antisense, 5' GCG CTG CCC ATC ATC ATG AC under the following conditions: 94°C for 2 min; 26 cycles (94°C, 30 s; 58°C, 30 s; 72°C, 40 s), and finally at 72°C for 3 min. To control the content of the initial total mRNA, PCRs were carried out with the primers for GAPDH (sense 5' ACC ACA GTC CAT GCC ATC AC, antisense 5' TCC ACC ACC CTG TTG CTG TA) and HPRT1 (sense 5' CTG GAT TAC ATC AAA GCA CTG, antisense 5' GGA TTA TAC TGC CTG ACC AAG). The conditions of PCR were as follow: 94°C, 3 min; (94°C, 30 s; 58°C, 30 s; 72°C, 1 min,) and 72°C, 10 min, 23 cycles for GAPDH and 30 cycles for HPRT1. PCR products were resolved on 2% agarose gels in the presence of ethidium bromide and visualized by fluorescence in transmitted ultraviolet light or transferred to membrane (Hybondand-N+, Amersham Biosciences) and probed with p16INK4a-specific P³² labeled internal oligonucleotide 5'CTG CCC AAC GCA CCG AAT AGT TAC G 4.

Analysis of the oncogene E7 transcripts was performed with the following primers: HPV16 sense, 5' CGG ACA GAG CCC ATT ACA AT; antisense, 5' GAA CAG ATG GGG CAC ACA AT; HPV18 sense, 5' GCT GAA CCA CAA CGT CAC AC; antisense, 5'-GGT CGT CTG CTG AGC TTT CT. The conditions of PCR were as follow: 94°C, 4 min; (94°C, 30 s; 58°C, 30 s; 72°C, 90 s) 35 cycles and 72°C, 6 min.

First the methylation status of the region around the p16 INK4a translation start site was analyzed by methylation-specific PCR (MSP) with commonly used primers (see scheme, fig. 1) 30. Exactly this region of the p16 INK4a CpG island was reported to be methylated in many types of tumors including cervical squamous cell carcinomas 322232425262730. The presence of methylated alleles was revealed in 6 out of 26 (23%) squamous cell carcinomas and in 3 out 14 of samples of normal tissues adjacent to tumors. There were methylated alleles in none of the 5 cervical cancer cell lines. Then methylation status of half of tumor and normal samples and 5 cell lines were verified by the bisulfite-based DNA sequencing. The methylation status of 26 CpG dinucleotides localized within the 200 bp-long fragment of DNA around the p16 INK4a translation start site was evaluated. Samples containing methylated alleles of gene as determined by MSP are marked by + (fig. 1a). Bisulfite-modified DNA sequencing analysis has not revealed extensive methylation of this region in either 13 carcinomas or 5 carcinoma cell lines. Sequencing analysis of total PCR product generated from this region revealed a varied number of solitary partially methylated CpG sites in DNA of some tumors and carcinoma cell lines, as well as in DNA of some normal tissues. Examples of partially methylated CpG sites are presented on fig. 1b. There was no essential difference in mosaic patterns of partially methylated CpG sites of tumors and normal tissues. In some cases the number of partially methylated CpG sites was greater in normal tissues than in tumors (fig. 1a, samples 2 and 12). Moreover, the number of partially methylated CpG sites in some of MSP-positive carcinomas was less than in MSP-negative carcinomas had (fig 1a, 7T, 10 T and 13T respectively). The distribution of partially methylated CpG dinucleotides within examined region varied in different samples. There was no preferential association of partially methylated CpG dinucleotides with any transcriptional factor binding site. The analysis of the cloned PCR products showed that 5 out of 5 clones contained only unmethylated CpG sites (fig. 1a and 1b, cl. 10T; Caski, cl.; one out of 5 identical clones is presented). These results are compatible with those for the population of DNA molecules (total PCR product) and demonstrate that methylation of certain CpG sites within the p16 INK4a 5'CpG islands is a rare event. Thus, no extensive methylation of the examined region was revealed in HR-HPV-positive cervical carcinomas, including carcinomas positive by MSP in our hands.

Next we examined methylation status of 3 HpaII recognition sites located around putative multiple sites of p16 transcription initiation (300 bp upstream of translation start site indicated on scheme, fig. 1). It has been demonstrated previously by CAT assay and bisulfite genomic sequencing that methylation of these critical HpaII sites significantly down-regulates p16 promoter activity 431. Analysis of this region by HpaII-methylation-sensitive PCR revealed the presence of methylated alleles only in 1 out of 26 carcinomas. Methylation of these critical HpaII sites was not detected in any cancer cell lines or normal tissues (data are not presented). These findings demonstrate that methylation of critical HpaII sites in the region of putative transcription start sites is a rare event in HPV-positive cervical carcinomas.

To understand whether mosaic methylation of 5'CpG island can modulate p16 INK4a transcription, the levels of mRNA were analyzed in all samples listed in fig 1. A representative example of the RT-PCR analysis is shown in fig 2. All carcinomas expressed HPV-E7 oncogene. The up-regulation of p16 INK4a transcription was detected in all carcinomas and carcinoma cell lines except for case No 1 (table 1 and fig 2, 1T). Independently of the number of partially methylated CpG sites within 5'CpG island, the mRNA levels were three- to 30-fold higher in carcinomas then in normal adjacent tissues. A rare case of the complete absence of the p16 INK4a mRNA in tumor No 1 expressing E7 viral oncogene was not related to hypermethylation of the examined regions, which was completely unmethylated in this carcinoma (fig. 1, 1T). Thus, random methylation of some CpG sites in 5'CpG islands does not prevent the p16 INK4a transcriptional activation in HR-HPV-positive cervical carcinomas.

Unlike RT-PCR analysis, the p16^ink4a immunostaining allows to estimate the amount of tumor cells expressing the protein. Rpresentative examples of the p16^ink4a immunoreactivity are shown in fig. 3 and the results of the analysis are summarized in table 1. Cervical normal cells are not stained for p16^ink4a, as expected (fig. 3g). On the other hand, the technical procedure used in this study allows to reveal the p16^ink4a expression in all HR-HPV-positive cell lines SiHa, HeLa, Caski, C4-I and in HPV-negative line C33A, in which pRb is inactivated by mutation 8. As expected, a strong predominantly cytoplasmic p16^ink4a immunoreactivity was detected in a majority of cells 17. The relative number of stained nuclei varied in different cell lines from 5 to more than 25% (fig 3a,b,c). A cytoplasmic immunoreactivity in the majority of cells was also revealed for all tumors with p16 INK4a transcriptional activation (fig. 3d,e,f) Tumors differed from each other by the number of stained nuclei like cell lines (from less than 1% to more than 25%). Carcinoma No 1 with the complete absence of mRNA did not show immunohistochemically detectable p16^ink4a expression either in the cytoplasm or in the nuclei, confirming specificity of the p16^ink4a immunoreactivity (table 1 and fig 2, 1T). The tumors that contained methylated alleles of gene (detected by MSP) exhibited the p16^ink4a immunoreactivity. In some cases they displayed strong diffuse p16^ink4a staining (fig. 3f). The percentage of stained nuclei did not correlate with the mRNA levels. The lowest number of stained nuclei was detected in carcinoma with the highest mRNA level (table 1 and fig. 2, sample 3). Thus, the transcriptional activation of p16 INK4a correlates with p16^ink4a cytoplasmic staining but does not influence the level of nuclear staining. The random methylation of some CpG sites within 5'CpG islands does not interfere with the p16^ink4a immunostaining in the majority of cells of HR-HPV-positive cervical tumors and cancer cell lines.