Human Papillomavirus-16 and 18 in Penile Carcinomas: DNA Methylation, Chromosomal Recombination, and Genomic Variation

Abstract

Penile carcinomas are frequently associated with high-risk human papillomavirus (HPV) types. Since little is known about the molecular biology of this association, we investigated three properties of HPV genomes in penile carcinomas from Brazilian patients: (i) HPV DNA methylation, (ii) junctions between HPV and cellular DNA, and (iii) genomic variation. In cervical carcinogenesis, recombination between HPV and chromosomal DNA is frequent and likely necessary for progression, and DNA hypermethylation - specifically of the L1 gene - is a biomarker for cancerous progression. The same mechanisms apparently occur during penile carcinogenesis, because 95 HPV-16 molecules derived from 19 penile lesions had 58% of the CpGs in L1 and 22% in the 5′ part of the long control region methylated, more than the percentages found in cervical carcinomas. In addition, two out of three HPV-18 infections, all present in double-infections with HPV-16, showed L1 specific methylation typical of malignant cervical lesions. In 11 out of 15 HPV-16 lesions we confirmed chromosomal integration by reverse ligation inverted PCR, while four samples had concatemeric integrations or episomes. Nine of 17 penile carcinomas contained HPV-16 AA variants, and eight E variants. As AA variants are relatively rare in Brazilian cohorts of asymptomatic women, the high prevalence in penile carcinomas may indicate a higher risk of progression of AA lesions, as suspected for cervical infections. Our observations of frequent viral DNA methylation, chromosomal integration, and the prevalence of high-risk variants suggest that HPV dependent carcinogenesis of the penis and cervix follow similar etiological and epidemiological parameters.

Introduction

Penile carcinomas are malignancies derived from the squamous epithelium of the foreskin or the glans penis. They are rare neoplasms, with 0.2% lifetime risk of a man in the United States or Europe developing invasive penile cancer.^1–3 Much higher rates are found in some developing countries, for example 2.1% in Brazil.⁴ 30 to 60% of penile carcinomas are reported to harbor human papillomavirus (HPV) DNA.^5–11 The presence or absence of HPV DNA in these lesions suggests two different etiological pathways of penile carcinogenesis, one influenced by the HPV oncoproteins and one independent of their functions. This explanation correlates with the detection of a pathologically distinct subset of penile cancers, as basaloid and warty penile cancers contain HPVs very frequently (about 80 to 100%), while the viral DNA is found only in a very small fraction of keratinizing and verrucous penile carcinomas.¹²

In the case of female genital cancers, the DNA of “high-risk” HPV types,¹³ a subset of about 18 of more than 100 known HPVs, is detected in nearly all cervical carcinomas and in a smaller fraction of vaginal and vulval carcinomas. The association between HPVs and genital carcinomas in females is considered causal because (i) HPV DNA is present in all precursor lesions and malignancies, (ii) HPVs express the oncogenes E6 and E7 in all cells of these lesions, (iii) these E6 and E7 proteins have pleiotropic molecular properties, which explain the properties of the transformed cell,¹⁴ and (iv) repression of the E6 and E7 genes annihilates transformation and viability of the HPV containing cell lines.¹⁵ HPVs may initiate neoplasia of the penis in a similar manner as genital carcinomas of females, but this is supported by relatively few data, since much more research has addressed HPV lesions of women than lesions of men.

HPV genomes replicate as episomes, i.e. circular double-stranded DNA, during the normal life cycle. However, in cancer and cancer derived cell lines they often recombine with the chromosomal DNA ^16–18 with an increase of the integrated viral DNA correlating with disease progression.^19–21 Recombination apparently favors progression as it leads to increased oncogene expression, relieves E6–E7 transcription from repression by E2,²² stabilizes E6–E7 transcripts,²³ and activates a nuclear matrix dependent enhancer of E6–E7 transcription.²⁴ Recombination is therefore frequent and possibly even necessary to trigger the progression of cervical HPV infections to malignancies, although one of these studies suggested that recombination might be a consequence rather than cause of progression.¹⁸ Recombination between HPV and chromosomal DNA also occurs in the progression of anal and oral carcinomas.^25,26 In the research presented here we attempted to study whether a similar correlation exists in penile carcinogenesis. Specifically, we examined whether the methylation of HPV-16 and 18 DNA, a biomarker of HPV integration into the chromosomes in cervical carcinomas,^27–29 occurs in penile carcinomas in similar diagnostically relevant patterns as found in cervical carcinomas, and whether these patterns correlate with recombination measured by detection of junction fragments between HPV and chromosomal DNA.

The term DNA methylation refers to the transfer of a methyl group to cytosines that are part of CpG dinucleotides (meCpGs), which results in the binding of meCpG specific transcriptional repressors, for example MeCP2.^30–33 In undifferentiated cervical cells, HPV-16 acquires a low and sporadically distributed CpG methylation, which disappears completely upon differentiation.^28,34 During carcinogenesis, upon integration of HPV-16 and 18 into cellular DNA, the L1 gene, and to a lower extent adjacent long control (LCR) sequences, become hypermethylated,^25,27,28,35 a fate of HPV DNA shared with most unrelated external DNA sequences that enter mammalian cells.³⁶ As most HPV infected cells contain a large number of viral genomes, most of the viral DNA can become hypermethylated and thereby transcriptionally repressed, as long as one or a few viral oncogenes remain unmethylated and transcriptionally active to maintain the transformed state of the infected cell.^27–29 The L1 gene is a particularly effective target of methylation, likely since recombination places this gene upstream of the viral promoter, and lack of transcription is known to facilitate methylation. As a consequence, HPV-16 and 18 L1 DNA methylation can serve as a biomarker of recombination between HPV and cellular DNA, which is useful to confirm the cancerous state of the lesion or to identify cell populations that have acquired a higher propensity to progress to the cancerous state. In this study we included experiments that validated this biomarker by adding to the methylation analysis the direct determination of integration by isolating the junctions between viral and human DNA by reverse ligation inverted PCR (rliPCR).³⁷ While this technique is powerful to confirm the integration of single or few copies of viral DNA, it cannot safely differentiate between high copy number concatemeric integrations and episomes.

We added to this etiological study an investigation of the nucleotide sequence variation of HPV-16 genomes. It is known that the HPV-16 genomes diversied during the linked evolution of the virus with the human host. While HPV-16 variants of the E branch evolved in European and those of the AA branch in American Indians, both are now spread in mixed populations of the American continent regardless to the ethnicity of the host.^38,39 Epidemiological studies in Brazil, Costa Rica and the United States suggested that infections with HPV-16 AA variants have a higher propensity toward malignant progression than E variants due to higher persistence and increased transcription of the viral oncogenes.^40–43 The high prevalence of AA variants in penile carcinomas that we observed adds further supports that genomic variation of HPVs is a risk factor not only in cervical but also in penile anatomic sites.

Materials and Methods

Clinical specimens

All DNA samples were obtained from penile carcinomas of a cohort of Brazilian patients that has been published.⁴⁴ 24 samples entered this study, and 19 of these contained HPV-16. Three samples contained HPV-18 (P6, P20, P23) in addition to HPV-16. It is not known whether HPV-16 and 18 occurred in these samples in the same cells or in different cell populations. Five samples contained other HPV types but HPV-16 and 18 and became deleted from this study.

Bisulfite modification

DNA sequencing subsequent to bisulfite modification and PCR amplification (with or without additional cloning into E. coli) is the most sensitive technique to measure methylcytosines.⁴⁵ For bisulfite treatment, 50–1000 ng of sample DNA supplemented with 1 μg of salmon sperm DNA in a total volume of 18 μl were denatured with 2 μl of 3 M NaOH and incubated at 37 °C. After denaturation, 278 μl of 4.8 M sodium bisulfite and 2 μl of 100 mM hydroquinone were added and the mixture was incubated in a thermal cycler for 20 cycles of 55 °C for 15 minutes and 95 °C for 30 seconds. The modified DNA was desalted with the QIAquick PCR purification protocol and desulfonated by adding 5.5 μl of 3 M NaOH and 5 μg glycogen prior to 15 min incubation at 37 °C. The DNA was precipitated with 5.6 μl of 3 M sodium acetate and 150 μl of 100% ethanol and centrifuged. The pellet was washed with 70% ethanol and dissolved in 30–50 μl TE buffer (10mM Tris-HCl pH 8, 1 mM EDTA).

Polymerase chain reactions, primers, T/A cloning and DNA sequencing

We have learned in previous studies that each HPV associated lesion contains HPV genomes with diverse methylation patterns. Therefore, we analyzed from each sample several, in this study five, HPV-16 genomes between the genomic positions 7079 and 85, which contains the 3′ part of the L1 gene and the LCR. Since bisulfite treated DNA is partially degraded, we had to dissect this segment into three amplicons. The sequences of the three primer pairs were designed according to the genomic sequence of HPV-16 assuming conversion of all cytosine residues into uracils as published.²⁷ We amplified part of the L1 gene and the 5′-LCR with the primers 16msp3F (position 7049–7078, AAGTAGGATTGAAGGTTAAATTAAAATTTA) and 16msp3r (position 7590-7560, AACAAACAATACAAATCAAAAAAACAAAAA); the HPV-16 enhancer with the primers 16msp4F (position 7465–7493, TATGTTTTTTGGTATAAAATGTGTTTTT) and 16msp7R (position 7732-7703, TAAATTAATTAAAACAAACCAAAAATATAT); and the HPV-16 promoter with the primers 16msp5F (position 7748–7777, TAAGGTTTAAATTTTTAAGGTTAATTAAAT) and 16msp8R (position 115-86, ATCCTAAAACATTACAATTCTCTTTTAATA).

Three samples contained HPV-18 DNA, which was examined as published³⁵ by dividing part of the HPV-18 L1 gene and the HPV-18 LCR into three amplicons between the genomic positions 6845–7186, 7282–7747, and 7753-186. The segment 6845–7186 encompassed the 3′ part of the L1 gene and was amplified with the primers Msp6Fa, AATTATTAGTTTGGTGGATATATAT (position 6845–6869) and Msp6R, AAAACATACAAACACAACAATAAATA (position 7186-7161). The segment 7282–7747 with the 5′ part of the LCR and the viral enhancer, was amplified with the primers Msp10F, TAAAATATGTTTTGTGGTTTTGTG (position 7282–7293) and Msp10R, ATAATTATACAAACCAAATATACAATT (position 7747-7721). The genomic segment 7743-186 with the viral replication origin, the E6 promoter and the 5′ part of the E6 gene was amplified with the primers Msp8F TGTTTAATATTTTGTTTATTTTTAATATG (position 7753–7781) and Msp8R, TATCTTACAATAAAATATTCAATTCC (position 186-161).

The PCR reaction was done in 25 μl with 0.2 mM of each of the four dNTPs, 10 pmol of each primer, 2 mM MgCl₂ and 1 unit of AmpliTaqGold (Perkin-Elmer). The PCR conditions were 94°C for 1 min, followed by 40 cycles of 94 °C for 10 sec, 54 °C for 30 sec and 68 °C for 1 min with a final extension at 68 °C for 7 min. The presence of PCR products was verified by agarose gel electrophoresis, and confirmed amplicons were cloned with the TOPO TA cloning kit for sequencing (Invitrogen). Cloned DNAs were sequenced by Big Dye terminator v3.1 Cycle Sequencing (Applied Biosystems).

Reverse ligation inverted PCR (rliPCR)

The principles of directly determining the recombination between HPV genomes and cellular DNA by reverse ligation inverted PCR (rliPCR) have been published by one of us.³⁷ 0.5–2 μg DNA was digested with 20 units TaqI in a total volume of 50 μl at 65 °C overnight. TaqI has a single target site in HPV-16 genomes at the genomic position 505. TaqI was heat-inactivated, and the DNA was circularized by addition of 5 units T4 DNA ligase (Invitrogen) in a total volume of 100 μl, followed by incubation at 16°C overnight. Circularized DNA was purified using phenol/chloroform extraction and ethanol precipitation and dissolved in 50 μl distilled H₂O. PCR was run with 5 μl of circularized DNA (200ng) as well as undigested DNA using platinum Taq DNA polymerase High Fidelity (Invitrogen) in a 50 μl volume containing 200 μM of each dNTP, 2mM MgSO₄, and 10 pmoles of each primer. The DNA was amplified with the primer pairs H16Lon-F and R, composed of E6 gene sequences, and E1-invP1 and P2, composed of sequences from the 5′ end of the E1 gene. The primers had the sequences: H16Lon-F: 5′-ACCCACAGGAGCGACCCAGAAAGTTACCAC-3′; H16Lon-R: 5′-AAAAACAATGGAATGGTTGGA\CAAGCAGTGC; E1-invP1: 5′-CGGGGTGAGTTTTTCAGAATTAGTA-3′; E1-invP2: 5′-TTCCACTACAGCCTCTACATAAAAC-3′. In cases with no PCR product, a second PCR was performed with 5 μl of the first PCR product and the following primers: H16Lon-3: 5′-GGGGTCGGTGGACCGGTCGATGTATGTCTTGTT-3′; H16Lon-4: 5′-GGCAAGCAGTGCAGGTCAGGAAAACAGGGATTTG-3′; E1-invP3: 5′-GGACTTACACCCAGTATAGCTGACAGTATAA-3′; E1-invP4: 5′-AGCCTCTACATAAAACCATCCATTACAT-3′. The thermocycling conditions were: 94°C for 2 min, followed by 35 cycles of denaturing at 94°C for 20 sec and annealing and extension at 68°C for 8 min. The PCR products were analyzed on 0.8% agarose gels. This strategy leads for episomal DNA to a 7463 bp fragment in the case of the H16Lon primers, and to a 7252 fragments in the case of the E1-inv primers. Amplicons of different sizes suggest viral-chromosomal junction fragments and were subjected to gel purification (Qiaquick gel extraction kit, Qiagen) and analyzed by sequencing with BigDye terminator.

Analysis of HPV-16 variants

A 363 bp segment of the HPV-16 LCR (genomic position 7478–7841) was amplified with primers and PCR conditions as described.³⁸ Successful amplification was confirmed by visualization of an aliquot in agarose gel electrophoresis in the presence of ethidium bromide. For sequencing, another aliquot was purified with ExoI and Shrimp Alkaline Phosphatase enzymes (USB corporation, Ohio), followed by cycle sequencing reaction with the ABI Prism BigDye Terminator version 3.1 (Applied Biosystems, Foster City, California) and sequenced on an Automate ABI Prism 3100 DNA Sequencer (Applied Biosystems). All data were confirmed twice by repeat PCR amplification and sequence analysis. Sequence similarities were searched with LALIGN.⁴⁶ AA variants of HPV-16 were distinguished from E variants by C to T mutations at the genomic positions 7667, 7762 and 7784 as published.^38,39

Results

Study design

This study was designed to compare the DNA methylation patterns of HPV-16 and 18 DNA in penile carcinomas with those in cervical neoplasia and cervical cell lines, and to investigate whether hypermethylation and chromosomal integration of HPVs would correlate by cloning of junction fragments between HPV and chromosomal DNA. Our previous studies have shown that hypermethylation of L1 and the 5′-part of the LCR correlates with chromosomal integration of the virus and malignant progression.^27,28,35 Following the same approach, we modified penile carcinoma DNA with bisulfite, cloned PCR amplicons into E. coli vectors and recorded the DNA sequences of 19 samples with HPV-16. As individual lesions are known to contain mixtures of HPV-16 genomes with different methylation patterns, we analyzed five clones from each sample. Specifically, we established sequences of a 913 bp genomic segment between the positions 7079 and 85 with the 3′ part of the L1 gene and the LCR. The segment has 19 CpGs, three overlapping with the L1 gene (7091, 7136, and 7145), five with the 5′-segment of the LCR (7270 to 7461), five with the enhancer (7535 to 7695), one with the replication origin (7862) and five with the E6 promoter (31–58) (Fig. 1). As it is not possible to amplify efficiently genomic segments with sizes exceeding a few hundred base pairs from bisulfite modified DNA, we dissected the 913 bp segment into three amplicons (see materials and methods). As an unavoidable consequence, data for these three segments represent the methylation status of different, non-contiguous molecules, as indicated by white vertical bars (Fig. 2). The numbers P1 to P24 identify those 19 of the 24 carcinomas that were HPV-16 positive, and each of the five lines of rectangles the five cloned amplicons of each sample. White rectangles indicate CpGs, black rectangles meCpGs. Altogether, Fig. 2 represents a database of the methylation state of 1805 individual CpG dinucleotides.

Fig. 1. Topographic relationship between 19 CpG dinucleotides and genomic organization of the L1 gene and the long control region (LCR) of HPV-16.

The numbers below the line indicate the genomic position of the first C of each CpG dienucleotide.

Fig. 2. Distribution of methylated CpG residues at the 3′ flank of L1 and through the LCR of HPV-16 in penile carcinomas.

Each vertical set of rectangles represents one of 19 specific CpG dinucleotides, the number on the top of the bar the position of this CpG in the genome of HPV-16 (see Fig. 1). Each horizontal set of rectangles represents a 913 bp segment of the HPV-16 genome, covering the 3′ end of the L1 gene and the complete long control region. Unmethylated CpGs are indicated by white rectangles, methylated by black ones. The two vertical white separators indicate the borders between amplicons, and discontinuities between unlinked HPV-16 genome segments.

Methylation of the L1 gene and the 5′ LCR of HPV-16

The amplicon representing the 5′ part of the targeted segment includes eight CpGs, three within the L1 gene (position 7091, 7136, and 7145), and five in the 5′ part of the LCR, which are known to be hypo-methylated in the episomal state in asymptomatic and precancerous infections of the cervix, but hyper-methylated in the integrated state in cervical carcinomas. The leftmost panel of Fig. 2 represents the methylation patterns that we observed in this genomic segment of HPV-16 in penile cancers, and the leftmost eight columns in Fig. 3 represent the corresponding methylation frequencies. While two samples contained unmethylated molecules (P21 and P23), there was substantial methylation of most molecules in all other 17 samples ranging from 27% (P1) to 100% (P14). On the average of all 95 molecules, 58% the CpGs in the three positions within L1 were methylated and 22% of the five CpGs located in the 5′ part of the LCR.

Fig. 3. Frequency of methylated CpG residues at the 3′ flank of L1 and through the LCR of HPV-16 in penile carcinomas.

The graph represents the average methylation of the CpGs represented in Fig. 2. The numbers below the x-axis represent the genomic positions of each CpG.

Methylation of the enhancer and promoter of HPV-16

The central amplicon between the genomic positions 7535 and 7695 contains five CpGs that overlap with the transcriptional enhancer. The 3′ amplicon between the genomic positions 7862 and 58 includes six CpGs, one (position 7862) being part of the E2 binding site of the viral replication origin, the other five CpGs between the genomic positions 31 and 58 being part of the Sp1 and E2 binding sites of the E6 promoter. The two right panels of Fig. 2 and the eleven rightmost columns of Fig. 3 represent the methylation of this segment. We measured for ten of these 11 CpGs methylation rates ranging from 5 to 13%. The average methylation rate was 8.7%, lower than that found in cervical carcinomas (about 16%), but higher than that found in cervical precursors and asymptomatic infections (about 7.5%). The CpG at position 7862 was always unmethylated as in cervical, anal and oral lesions, an unexplained phenomenon as discussed in past studies.^25–28

Integration of HPV-16 DNA measured by rliPCR

The rliPCR protocol described in the materials and methods section and depicted in Fig. 4 generates with two alternative primer pairs amplicons with sizes of 7463 and 7252 bp, respectively, when targeted at circular episomal HPV-16 genomes. From integrated HPV-16 DNA both primer pairs generate amplicons with different sizes. Integrated concatemeric HPV-16 genomes lead to the 7463 and 7252 bands plus an additional fragment from each of the junctions. If the number of tandem repeated HPV-16 genomes is large compared to the junction fragments, the intensities of the full-length HPV-16 bands may exceed those of the junction fragments to such an extent that they become undetectable, and it may be impossible to distinguish concatemeric and episomal viral genomes.

Fig. 4. Integration of HPV-16 DNA into chromosomal sequences can be revealed by restriction ligation inverted PCR (rliPCR).

The scheme represents how alternative products are generated by targeting episomal or integrated HPV-16 DNA by rliPCR with either of two alternative primer sets A or B. Dashed lines: HPV-16 genome, black lines: flanking human chromosomal DNA.

We analyzed 15 of the 19 HPV-16 positive samples by rliPCR, but could not study four samples (P6, P10, P15, and P20) due to insufficient sample size. Six of these 15 samples (P3, P9, P11, P14, P22, and P24) generated exclusively amplicons with sizes different from the 7463 and 7252 bp amplicons, suggesting recombination between single copies of HPV-16 and human chromosomal DNA. Five additional samples (P1, P2, P5, P7, and P21) led to 7463 and 7252 amplicons and additional amplicons of different molecular weights, suggesting either concatemeric HPV-16 genomes integrated into the human DNA, or simultaneous existence of episomal and integrated HPV-16 DNA. Four samples generated only 7463 and 7252 bands (P4, P8, P17, P23) indicating either episomal HPV-16 DNA or large concatemers outnumbering the junction fragments. Fig. 5 documents a subset of our data, including a SiHa, CaSki, and W12 control with single copy and multiple copy integrated copies and episomal HPV-16, respectively.

Fig. 5. Analysis of the integration of HPV-16 DNA in five representative penile carcinomas and three control cell lines by rliPCR and agarose gel electrophoresis.

The lanes 2/3, 4/5, 6/7, 8/9 and 10/11 contain the DNA from the penile carcinomas P1, P2, P3, P5, and P7, respectively. In each case, the left lanes show PCR amplifications without TaqI digestion, and each right lane the amplifications after Taq digestion and self-ligation. As controls, the lanes 14/15, 16/17, and 18/19 contain PCR amplified undigested (left lanes) and TaqI/ligation treated SiHa (single HPV-16 genome integration), CaSki (concatemeric HPV-16 genomes integration), and W12-20863 (episomal HPV-16) DNA. Detection of a 7252 bp band (without or with Taq1 digestion/religation) is diagnostic either of episomal copies or concatemeric repeats, while bands of other sizes identify integration and recombination events. Absence of a band in lane 14 combined with an aberrant sized band in lane 15 documents the single integrated HPV-16 DNA in SiHa. Presence of a 7252 bp band in lane 16 and the same band plus an aberrant sized band in lane 17 confirms the concatemeric integrations in CaSki, and a 7252 band in lane 18 and 19 the episomes in W12-20863²⁸. Based on detection of 7252 bands, P1 (lane 3), P2 (lane 5), and P5 (lane 8/9) have full-length episomal or concatemeric HPV-16 DNAs, which are accompanied by junction fragments or recombined fragments (lane 2/3, 5, 9). Aberrant sized fragments are the only products found in P3 and P7. Lane 1, 12, 13: size markers.

In order to confirm the recombination between HPV-16 and human DNA, we sequenced one of the two junction fragments of P1, P5, P7, P11, P21 and P24, and could detect integration into the human chromosomes 20, 3, 9, 8, 9 and 20, respectively. In the case of P1 and P5, we sequenced the complete early region of HPV-16, which confirmed that the HPV-16 genomes were discontinued at the positions 2627 and 3399, respectively, followed by human sequences. In the case of the other four clones, we established insertion into human DNA by sequencing into the 5′ direction of the TaqI side, a strategy that did not reveal the HPV-16 early gene junction site. The sequencing of clones from P9 suggested that most of the HPV-16 DNA existed in form of a deleted episome, and the junction was not detected. The nucleotide sequence of all junction segments between HPV-16 and human DNA have been deposited in GenBank (Accession numbers EU528866-EU528871).

Genomic variants of HPV-16 in penile carcinomas

Previous studies of the prevalence of HPV-16 variants in female Brazilian patients have found E variants as most prevalent, followed by AA variants, and much rarer, by Af variants. These variants originated in European, native American, and African ancestors of this multi-ethnic society, and are now spread throughout this country’s population irrespective of the ethnicity of the infected individual. Epidemiological studies from Brazil, Costa Rica and the United States suggested that AA and Af variants are more carcinogenic than E variants, possibly due to higher persistence of the infection.^40,42,43 Such a conclusion is warranted if one observes that the prevalence of AA and Af variants in high-grade lesions relative to that in asymptomatic infections. In asymptomatic female Brazilian patients, E variants have been found to outnumber AA and Af HPV-16 variants by a factor of three, while E and AA/Af variants have similar prevalence in high-grade lesions.⁴²

Such a direct comparison is not possible for our male cohort, as we did not have access to asymptomatic infections of the penis. In spite of this limitation, we determined the variant status in our cohort, and found that nine of 17 samples (P1, P3, P4, P7, P8, P11, P14, P22, and P23) (53%) contained AA and eight E variants (P2, P5, P6, P15, P16, P20, P21, P24). The variant status of two samples could not be confirmed due to small sample size.

Methylation of HPV-18 genomes

We also measured the methylation of HPV-18 DNA in penile carcinomas by examining eleven CpGs of the 3′ part of the HPV-18 L1 gene and 19 CpGs located in the LCR. Our previous studies of HPV-18 in cervical samples had found sporadic or complete lack of methylation throughout this genomic segment in asymptomatic infections and precursor lesions, while carcinomas and some high-grade lesions showed hypermethylation of L1 and some methylation of the LCR and E6 gene. In homology with HPV-16, hypermethylation likely correlates to a chromosomally integrated state as confirmed for the HPV-18 genomes in HeLa cells.³⁵

Only three of our penile carcinoma samples contained HPV-18 DNA, and they happened to exist in the form of double-infections with HPV-16 (sample P6, P20, P23). It is not known whether both viruses were associated with the neoplastic process in these carcinomas, or whether one virus was the “driver” and the other one a fortuitous coinfection. Fig. 6 shows that one of the samples, P6 contained mostly unmethylated HPV-18 DNA. A second sample, P20, had hyper-methylated L1 and promoter-E6 segments, with methylation in some clones extending throughout the LCR. A third sample, P23 contained hypermethylated and unmethylated L1 segments, and no methylation throughout the enhancer, promoter and E6 sequences.

Fig. 6. CpG methylation of the 3′ part of the L1 gene, the enhancer, promoter, and the 5′ part of the E6 gene of HPV-18 in penile carcinomas and HeLa cells.

Details of the notation are explained in the legend to Fig. 1. Detailed interpretations of these methylation patterns can be found in the references 35 and 52.

The lack of methylation in P6/HPV-18 coincides with a P6/HPV-16 pattern of HPV-16 L1 hypermethylation, and may indicate that HPV-16 was integrated and therefore responsible for this tumor, while HPV-18 may be a superinfection, possibly of other cells in this sample. The methylation patterns of HPV-18 in P20 and P23 would be compatible with a chromosomally integrated state, and HPV-18 may have induced these two tumors. This correlated with a completely unmethylated state of HPV-16 in P23, suggesting that P23 was caused by HPV-18 integration, while HPV-16 may have been a superinfection.

Discussion

The role of DNA methylation in HPV biology

This project had two principal objectives, namely to investigate (i) whether DNA methylation affects HPV-16 and 18 genomes in penile and cervical carcinomas in a similar manner, and (ii) whether HPV DNA methylation and integration data would strengthen the presumed causal role of HPVs in the etiology of penile carcinomas.

DNA methylation occurs in two different and unrelated contexts of HPV-16 and HPV-18 biology. In the first instance, scattered, low methylation (average methylation rate up to 7.4%,²⁸ throughout the LCR likely represses HPV transcription⁴⁷ in undifferentiated epithelial cells, and its loss during differentiation relieves repression.^27,34 This phenomenon, a normal part of the HPV life cycle was discussed in detail in a recent cell culture based study²⁸ is mentioned here for clarification, but was not the target of this research on penile carcinomas. The second scenario, methylation in linkage to recombination between HPV and chromosomal DNA, is clearly not part of the HPV life cycle, as it is specific for carcinogenesis.^18,19 Most external DNA of any source becomes methylated and thereby transcriptionally silenced upon recombination with the chromosomal DNA of the recipient cell.³⁶ The same mechanism targets HPV DNA, with two possible outcomes. In those cases where the carcinogenic process depends on numerous HPV genomes, as it is the case in most cancers and exemplified by the CaSki cell line, most HPV genomes can become methylated and repressed, as long as a single one remains transcriptionally active and maintains the transforming process.^27,29 Most of the genomes become contiguously methylated, and, for yet unknown reasons, methylation is most prevalent in the L1 gene. On the other side, if the carcinogenic process depends on a single HPV genomes, as exemplified by the SiHa cell line, the LCR of this single genome must remain transcriptionally active and therefore unmethylated, while methylation can readily target the L1 gene, whose function is irrelevant for carcinogenesis.

The HPV-16 and 18 L1 genes show similar patterns of hypermethylation in penile and cervical carcinomas

We have structured the first part of this study in a way similar to previous analyses of cervical, anal, and oral carcinomas, with the goal to establish a catalogue of the methylation patterns encountered in HPV-16 DNA associated with penile carcinomas (Fig. 2) and to condense these data in form of a bar diagram (Fig. 3). The hypermethylation of the three CpGs of the L1 gene (58%) is conspicuous and resembles a similar graphic representation of L1 hypermethylation in cervical carcinomas (47%),²⁷ and both diverge from the lower L1 methylation of cervical precursor lesions (20%) and asymptomatic infections (25%). Similarly, the methylation rates of the five CpGs in the 5′ part of the LCR were higher in penile (22%) and cervical carcinomas (18%) than in cervical precursor lesions (11%) and asymptomatic infections (6%). Hypermethylation was also observed in the HPV-18 L1 gene in two penile carcinomas, similar to the very high hypermethylation observed in cervical carcinomas.³⁵ We conclude that high HPV-16 methylation rates in penile carcinomas resemble those reported in cervical malignant lesions, and propose as the most likely explanation that they depend on chromosomal recombination. Therefore, they are molecular surrogate markers for the integration of viral DNA into cellular chromosomes and useful to diagnose a progression event.

It should be pointed out that some samples show strong methylation of the promoter sequences of HPV-16, as previously observed in our study of anal carcinomas.²⁶ It has been discussed by others that this pattern may also be diagnostic of carcinogenic progression,⁴⁸ as methylation of these sequences would stimulate oncogene expression, as it interferes with binding of the transcriptional repressor E2, while favoring binding of the transcriptional activator Sp1.⁴⁹

Correlation between direct integration measurements and DNA methylation

It is generally appreciated that there is no straightforward test to unequivocally differentiate between the episomal maintenance of an HPV genome and its physical linkage to cellular chromosomal DNA, although some techniques have acceptable analytic power for selected samples.^37,50,51 Here we attempted to compare the methylation analysis with clones generated by rliPCR.³⁷ As discussed above, this technique can unequivocally recognize the integration of single copies of HPV DNA, but may have sensitivity problems when used to distinguish between samples with numerous concatemers and samples with episomes. In consideration of this technical limitation, it is satisfying that 11 of the 15 samples, that could be analyzed by rliPCR clones, yielded HPV-16-human junction fragments, unequivocally confirming the integrated state of most (or potentially all) of the viral genomes. Specifically, the five clones that contained exclusively individual HPV-16 copies showed very high L1 methylation as expected (P9, 67%; P11, 93%; P14, 100%; P22, 73%; P24, 87%). Only three clones deviated from the rule that samples with integrated HPV DNA should be methylated in L1, and samples without methylation should be episomal. P4 and P17 were heavily methylated but seemed to contain episomes judged by rliPCR, and P21 was unmethylated in L1, but contained integrated copies. We do not interpret these three clones as contradictions of our hypothesis, but as examples where these laborious techniques reached their limits. We also published recently that one of several concatemeric HPV-16 integrations in a cell line remained unmethylated, possibly as a consequence of association with euchromatin.²⁸ In summary, the combination of DNA methylation and rliPCR analysis establishes a very strong case that HPV is the cause of penile cancer, and that chromosomal recombination is a necessary step in the etiological process as in cervical cancer. The use of methylation analysis as clinical diagnosis will require more labor-efficient modifications of our approach, such as the use of real-time PCR and methylation specific primers.⁵²

HPV-16 AA and E variants in penile carcinomas

Based on research of three cohorts from Brazil, Costa Rica and the State of Washington, USA, it is believed that non-European HPV-16 variants are more carcinogenic than European variants,^40,42,43 a conclusion that is based on the relative prevalence of variants in asymptomatic infections or lesions of the cervix. The term “non-European” variants is normally used to lump the two African and the AA branch for merely statistical reasons. Detailed mechanisms of this higher carcinogenicity are not well understood, but may include higher transcriptional activity and longer persistence of the infection.^41,42,43,53 African HPV-16 variants are also highly prevalent in penile carcinomas from Uganda, but possibly due to the fact that these variants are the most prevalent even in healthy patients of African cohorts.^39,54,55

The HPV-16 variant study most relevant for a comparison with data was done in the laboratory of one of us (L.L. Villa) and addressed female Brazilian patients. In this cohort, the prevalence of E variants vastly exceeded that of AA patients in asymptomatic patients, while the inverse was found in high-grade lesions and cancers,⁴² a finding that strongly supports the hypothesis that AA variants are more carcinogenic than E variants. While we have no information about the prevalence of these variants in asymptomatic infection of male Brazilian patients, it seems reasonable to assume that this will be similar to the prevalence in female infections. As a consequence we believe that the observed slight excess of AA variants over E variants in penile carcinomas may have originated from infections, where E variants were more common in asymptomatic infections. This would support a similarly higher carcinogenicity of AA variants in penile infections. The higher prevalence of HPV-16 AA variants in cervical cancers from Brazil as compared to asymptomatic women (ref. ⁴²; and L.L.Villa, manuscript in prep.) are further support for the etiological similarities between tumors from these two anatomical sites.

In summary, our comparison of penile and cervical carcinomas with HPV-16 and 18 DNA observed numerous striking similarities, recombination between viral and cellular DNA, frequent viral DNA methylation leading to similar DNA methylation patterns in response to the integration, and the prevalence of the same high-risk variants. These data strongly support the causality of the HPV infection in the etiology of penile cancers and suggest similar etiological and epidemiological parameters for HPV dependent cervical and penile carcinogenesis.