Recombination-Dependent Oligomerization of Human Papillomavirus Genomes upon Transient DNA Replication

Abstract

We describe the extensive and progressive oligomerization of human papillomavirus (HPV) genomes after transfection into the U2OS cell line. The HPV genomic oligomers are extrachromosomal concatemeric molecules containing the viral genome in a head-to-tail orientation. The process of oligomerization does not depend on the topology of the input DNA, and it does not require any other viral factors besides replication proteins E1 and E2. We provide evidence that oligomerization of the HPV18 and HPV11 genomes involves homologous recombination. We also demonstrate oligomerization of the HPV18 and HPV11 genomes in SiHa, HeLa, and C-33 A cell lines and provide examples of oligomeric HPV genomes in clinical samples obtained from HPV-infected patients.

INTRODUCTION

Human papillomaviruses (HPV) are important pathogens that cause different epithelial hyperplastic lesions, most commonly manifesting as benign warts or papillomas. Over 100 HPV types have been identified to date (1). These epitheliotropic viruses can be categorized based on their ability to infect mucosal or cutaneous keratinocytes. The mucosal viruses can be further subdivided into low- and high-risk HPVs. A potential for malignant progression is characteristic of high-risk HPV types, such as HPV18, HPV16, HPV31, and HPV45, whereas types such as HPV6 and HPV11 do not show similar associations and are considered low risk (2). Essentially all cervical carcinomas (3) and a quarter of reported head and neck cancers (4) are associated with HPV infections.

HPV is a small DNA virus with an approximately 8-kbp genome. During infection of stratified cutaneous or mucosal epithelia, the viral genomes replicate as multicopy extrachromosomal genetic elements in the nuclei of host cells. HPV genomes undergo a three-phase replication cycle linked to the host cell differentiation program (5). The first stage of HPV DNA replication occurs in undifferentiated basal keratinocytes after infection and is referred to as transient amplificational replication. During the first phase, the viral replication factors are produced and the HPV genome is amplified up to 100 of copies per cell during the S phase of the cell cycle. After initial amplification, the expression of viral replication proteins is downregulated to a level sufficient for the stable maintenance phase of episomal genomes in HPV-infected basal cells. Upon differentiation of the infected cells, the regulated expression of viral proteins initiates a second amplification of the viral genome, the production of capsid proteins, and the assembly of viral particles in the uppermost layers of terminally differentiated epithelium. The mechanisms regulating the switch from the initial HPV genome replication to HPV genome maintenance and, subsequently, to vegetative amplification are not entirely understood.

Replication of the HPV genome is carried out by the cellular replication machinery, which is directed to the viral origin by the viral replication proteins E1 and E2. The mechanism underlying the initiation of DNA replication is well described for papillomaviruses at the molecular level, particularly regarding the recognition of the origin by the E1 and E2 proteins and the assembly and movement of the viral helicase during DNA synthesis (6–15). However, the viral genome replication process, including the engagement of host cellular factors into the replication of the viral DNA and the regulation of this process through cellular functions, such as the cell cycle phases, DNA repair, and recombination or cellular differentiation, is still poorly understood. Replication studies have demonstrated that HPV genome replication is initiated bidirectionally (16), although the regulation of elongation of synthesized DNA strands, termination of replication, and maturation of the final products have not been sufficiently studied. It has been suggested that the HPV genome also undergoes bidirectional replication during the stable maintenance period; however, there is evidence that DNA replication switches to a rolling-circle mode after the vegetative amplification phase of the viral life cycle has begun (16). Rolling-circle replication has also been implicated as a mechanism of replication for bovine papillomavirus (BPV) type I (17).

Papillomavirus replication machinery has been shown to associate with the cellular DNA repair and recombination machinery. Several studies have demonstrated that expression of the E1 protein induces DNA breaks (18–20) and activates ATM- and ATR-dependent signal transduction pathways (18, 19, 21, 22). HPV recruits many cellular DNA repair and homologous recombination (HR) factors to viral replication centers (22, 23). The compartmentalization of the DNA damage response (DDR) machinery at HPV replication centers depends on specific functions of the E1 protein, such as the ability to hydrolyze ATP and unwind DNA, that are essential for viral DNA replication (18, 19, 21).

We have developed a cellular assay system for the study of HPV genome replication based on the human osteosarcoma U2OS cell line having a unique capability to support the transient, stable, and late amplificational replication of both cutaneous and mucosal HPV genomes (24). Although this system is not suitable for the reconstruction of the entire viral intracellular life cycle, especially expression of the capsid proteins and assembly of the viral particles, our data suggest that these cells provide a relevant environment for studies, supported by the facts that transcription maps of the HPV18 and HPV5 genome early regions in U2OS cells are similar to those in keratinocytes (A. Männik, G. Kivi, M. Toots, M. Ustav, Jr., E. Ustav, and M. Ustav, unpublished data; E. Sankovski, A. Männik, J. Geimanen, E. Ustav, and M. Ustav, unpublished data). The possibility of using the continuously proliferating homogeneous cell population for replication studies enables us to reveal the detailed mechanisms of HPV DNA replication, which have been difficult to determine thus far, especially for low-risk and cutaneous papillomaviruses, mainly due to their poor immortalization of primary keratinocytes (25, 26). After transfection of the religated HPV genomes into the U2OS cells, we observed effective initiation of viral DNA replication; however, we also detected extensive progressive formation of HPV genomic oligomers (24). In this study, we demonstrate that during HPV18 and HPV11 genome replication, head-to-tail oligomeric molecules are formed in a time-dependent manner irrespective of the topology of the transfected DNA. The formation of concatemeric oligomers is dependent on the viral replication proteins E1 and E2 and the replication process initiated from the HPV origin. We also provide evidence that oligomerization is induced through homologous recombination.

We describe the replication and oligomerization of the HPV11 and HPV18 genomes in the HPV-positive HeLa and SiHa cells as well as in the HPV-negative C-33 A cell line, and we give examples of patient-derived tissue samples containing head-to-tail HPV genomic oligomers which suggest that oligomerization of the HPV18 genome occurs in vivo. The previous detection of episomal HPV16 and HPV18 DNA oligomers in clinical samples (27–32) and in primary keratinocytes following transfection of religated viral genomes (33–36) suggests that the oligomerization of the viral genome is a natural process and a characteristic feature of HPV genome replication.

Our data suggest that HPV genomes use a unique recombinational replication mechanism for multiplication of the viral DNA during the establishment period of infection, and that the formation of HPV genomic oligomers is a reflection of that recombinational replication mechanism.

MATERIALS AND METHODS

Cell lines and transfection.

U2OS, SiHa, HeLa, and C-33 A cells were grown in Iscove's modified Dulbecco's medium (IMDM) supplemented with 10% fetal calf serum (FCS). The cells were transfected through electroporation using a Bio-Rad Gene Pulser II apparatus supplied with a capacitance extender (Bio-Rad Laboratories). The capacitance was set to 975 μF, and the voltage was set to 220 V (U2OS, SiHa, and HeLa cells) or 180 V (C-33 A cells) in all experiments.

Plasmids.

The HPV18 minicircle parental plasmid was generated following insertion of the pMC.BESPX vector (37) into the HPV18 genome between positions 7473 and 7474, the same position previously used in the production of HPV18 genomes via the Cre-Lox technology (38). The minicircle technology produces covalently closed circular HPV genomic plasmids, with the minor 36-bp remnant of the recombination site inserted into HPV genomic sequence but otherwise free from the bacterial sequences of the pMC.BESPX vector (37, 39). The following methods were used to generate HPV18 genomic mutants from the HPV18 minicircle parental plasmid, pMC.BESPXHPV18: for the HPV18 E1 mutant, the nucleotides CTCGAGA were added after genomic position 943; for the HPV18 E2 mutant, the ATG start codons at positions 2818, 3058, 3214, and 3247 were mutated to ACG; for the HPV18 E4 mutant, the ATG start codon at position 3419 was mutated to ACG; for the HPV18 E1E4 mutant, a stop codon was introduced after the splice site at position 3434 by mutating the nucleotides at positions 3452 (A to G) and 3453 (T to A); for the HPV18 E5 mutant, the nucleotides TAACTCGAG were added after genomic position 4011; for the HPV18 E6 mutant, the nucleotides CTCGAGG were added after genomic position 134; and for the HPV18 E7 mutant, the nucleotides AACTCGAG were added after genomic position 610. The HPV18 E8 mutant has been previously described (40), and a similar mutation was introduced into the pMC.BESPXHPV18 plasmid. The HPV18 upstream regulatory region (URR) minicircle parental plasmid contains the BamHI-BamHI fragment of the wild-type HPV18 genome (termed HPV18wt) cloned into the BglII site of the pMC.BESPX vector. The BamHI fragment of HPV18wt contains 206 nucleotides from the end of the L1 open reading frame (ORF), 14 nucleotides from the beginning of the E6 open reading frame, and the full URR region. The HPV18 E1 and E2 protein expression vectors have been described previously (41). To generate the HPV18E minicircle parental plasmid, the pUC vector was removed from the pUCeHPV18 construct (24), followed by insertion of the pMC.BESPX vector into HindIII linkers. The HPV11 minicircle parental plasmid (pMC.BESPXHPV11) was produced following insertion of the pMC.BESPX vector into the BamHI site of the wild-type HPV11 genome. To generate the HPV11 E1 mutant, the HPV11 genome was digested with the double-cutter AccI, and the resulting 7.7-kbp HPV genomic fragment was gel purified and religated after Klenow treatment. The E1 mutant has a 182-bp deletion in the E1 reading frame, resulting in a frameshift mutation. To generate the HPV11 E2 mutant, the wild-type HPV11 genome was linearized using SdaI, treated with T4 DNA polymerase to remove 3′ overhangs, and religated. Treatment with T4 DNA polymerase generated a 4-bp deletion and a frameshift mutation in the E2 open reading frame. Both HPV11 mutants were cloned into the minicircle production vector pMC.BESPX in a manner similar to that for the HPV11 wild-type genome.

Production of HPV genomic and subgenomic minicircles.

HPV minicircle genomes were produced from the minicircle parental plasmids as previously described (18, 37). The HPV minicircles extracted from bacterial cells were additionally gel purified in order to use covalently closed monomeric circular HPV genomes in transfections.

LMW DNA extraction.

Low-molecular-weight (LMW) DNA was extracted at the indicated time points using the modified Hirt method (42). Prior to analysis the samples were treated with DpnI (Thermo Scientific) to remove bacterially produced input HPV DNA. We observed nicking of the LMW DNA samples during incubation in buffers containing Mg²⁺ ions, which improved the detection of concatemeric HPV genomic molecules on the Southern blots. Therefore, the gels were treated with hydrochloric acid prior to Southern analysis in order to nick the DNA through depurination and increase the efficiency of the detection of the large HPV oligomers.

DNA extraction from patient samples.

The samples were collected from women previously diagnosed with persistent HPV infections who were undergoing follow-up studies at the Women's Clinic of Tartu University Hospital. Gynecological brush or colposcopy samples were collected and stored in 1× phosphate-buffered saline (PBS; 137 mM NaCl, 2.7 mM KCl, 4.3 mM Na₂HPO₄, and 1.47 mM KH₂PO₄) at 4°C. The samples were centrifuged at 16,200 × g, after which the 1× PBS was removed, and treatment with proteinase K (2 μg/ml) in SolIV was performed for 24 h at 65°C. Following proteinase K treatment, DNA was extracted using the phenol-chloroform method and precipitated with ethanol. The obtained DNA pellet was dissolved in 1× Tris-EDTA (TE) and treated with RNase A at 37°C for 1 h. DNA was precipitated with ethanol in the presence of 200 mM NaCl and then dissolved in 1× TE, and the concentration was measured using a NanoDrop ND-1000 spectrophotometer (Thermo Scientific). The Research Ethics Committee of the University of Tartu granted permission (209T-14) to use patient samples in this study. All patients gave their informed consent prior to participating in the study.

Total DNA extraction from cell lines.

Total cellular DNA was extracted by following the same protocol as that for the DNA extraction from patient samples, with the omission of the first centrifugation step.

1D gel electrophoresis.

For the one-dimensional (1D) replication product analysis, LMW DNA was resolved in a 0.5 or 0.8% agarose gel in 1× Tris-acetate-EDTA (TAE) buffer containing ethidium bromide (0.3 μg/ml). Electrophoresis was performed at room temperature at 0.8 V/cm for 20 h.

2D gel electrophoresis.

For the two-dimensional (2D) replication product analysis, the first dimension was run using a 0.4% agarose gel in 0.5× Tris-borate-EDTA (TBE) buffer without ethidium bromide at 0.3 V/cm for 48 h at room temperature. The second dimension was run using a 1% agarose gel in 0.5× TBE at 6 V/cm for 5 h at 4°C. Ethidium bromide (0.3 μg/ml) was added to the gel and buffer when running the second dimension.

Southern transfer and hybridization.

The DNA fragments in the agarose gels were denatured through treatment with Sol A (0.5 M NaOH and 1.5 M NaCl) for 30 min. The gels were subsequently rinsed with double-distilled water (ddH₂O) and neutralized with Sol B (1 M Tris, pH 7.4, 1.5 M NaCl), and the DNA was transferred to a nylon transfer membrane (Membrane Solutions LLC) through capillary transfer using 10× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) buffer for 4 to 18 h. The DNA was then cross-linked with the filter using a UV Stratalinker 1800 apparatus (Stratagene). To avoid nonspecific binding to the filter, it was treated with a prehybridization solution (30% 20× SSC, 10% 50× Denhardt's, 5% SDS, and 200 μg/ml denatured carrier DNA) at 65°C for at least 45 min. HPV genomic DNA fragments were labeled using the DecaLabel DNA labeling kit (Thermo Scientific) and radioactive [α-³²P]dCTP (PerkinElmer/Hartmann Analytic), and they were employed as specific probes. The filter was hybridized for at least 18 h at 65°C. Posthybridization the filter was washed twice with washing buffer I (2× SSC and 0.1% SDS) for 5 min each time, once with washing buffer II (1× SSC and 0.1% SDS) for 15 min, and twice with washing buffer III (0. 1× SSC and 0.1% SDS) for 10 min each time. The resultant signals were detected via exposure to X-ray film.

RESULTS

Multimerization of HPV genomic DNA in U2OS cells is independent of the transfected DNA topology.

The first phase of the HPV life cycle is the viral establishment period, during which HPV DNA replication generates a large number of copies (20 to 100) of identical HPV molecules from a unit-length circular genome within the infected host cell. Using a transient replication assay mimicking the phase of the establishment of the viral infection, we have previously noticed that in addition to the monomeric HPV genomes, dimeric and oligomeric forms of HPV DNA appeared in U2OS cells. These oligomers were effectively maintained in proliferating HPV-positive U2OS cell lines over many passages (24). Thus, we analyzed the kinetics and the nature of HPV genomic oligomer formation in greater detail.

First, we investigated whether the topology of the input DNA influences the oligomerization process. In a previous U2OS-based replication assay, HPV DNA was cleaved out from the vector backbone, and the linear HPV genomes (∼8 kbp) were then religated at low DNA concentrations and transfected into the cells (24). In principal, it is possible that during ligation small amounts of oligomeric molecules form which replicate more efficiently than monomeric genomes. To exclude this possibility and examine the effect of the conformation of the input DNA on oligomerization, linear, religated (relaxed circular), and covalently closed circular (ccc) forms of the HPV18E8⁻ mutant and HPV11 wild-type genomes were prepared and introduced into U2OS cells under identical conditions (Fig. 1A and B).

Fig 1.

(A) Southern blot analysis of the transient replication of HPV18E8⁻ mutant genomes in U2OS cells. U2OS cells were transfected with 2 μg of linear (lanes 1 to 7), recircularized (lanes 8 to 15), or minicircle (lanes 16 to 23) HPV18E8⁻ genomic DNA. In the case of the linear molecules, the break in the HPV18 genome is located in the E1 open reading frame. LMW DNA was extracted at the indicated time points (days) and analyzed through 1D gel electrophoresis, followed by Southern blotting. Prior to analysis, the samples were treated with DpnI. Marker plasmids for the linear (wtH18 lin) and DpnI-treated (wtH18xDpnI) HPV18E8⁻ genomes are shown in lanes 24 and 25. A supercoiled circular DNA ladder (M ccc; 2 to 16 kbp) (Sigma-Aldrich) and linear DNA ladder (M lin; 1 kbp) (Naxo) are shown in lanes 26 and 27. (B) Southern blot analysis of the transient replication of HPV11 genomes in U2OS cells. U2OS cells were transfected with 5 μg of linearized (lanes 1 to 4), 5 μg of recircularized (lanes 5 to 8), or 1.5 μg of minicircle (lanes 9 to 12) HPV11 wild-type genomic DNA. In the case of the linear molecules, the break in the HPV11 genome is located in the L1 open reading frame. A total of 50 μg of salmon sperm carrier DNA was added to each transfection. LMW DNA was extracted at the indicated time points (days) and analyzed through 1D gel electrophoresis, followed by Southern blotting. Prior to analysis, the samples were treated with DpnI. Marker plasmids for the uncut (H11 mc), linear (wtH11 lin), and DpnI-treated (wtH11xDpnI) HPV11 genomes are shown in lanes 13, 14, and 15. A supercoiled circular DNA ladder (Sigma-Aldrich) and linear DNA ladder (Naxo) are shown in lanes 16 and 17. (C) Southern blot analysis of the oligomerization of the HPV18wt genome in U2OS cells. U2OS cells were transfected with 2 μg of the HPV18wt minicircle genome and 5 μg of the pauxoMSC carrier DNA plasmid (Icosagen) and seeded onto 100-mm cell culture dishes (8 × 10⁵ cells per dish). After transfection, half of the plates (lanes 1 to 8) were passaged every 2 days from day three onward to keep the cells under nonconfluent culture conditions, and the other half of the plates (lanes 9 to 12) were allowed to grow dense while adding 2 ml of fresh media every 2 days. Total DNA was extracted at the indicated time points (days) and analyzed through 1D gel electrophoresis, followed by Southern blotting. Prior to analysis the samples were treated with DpnI. Marker plasmids for the uncut (wtH18 mc), linear (wtH18 lin), and DpnI-treated (wtH18xDpnI) 8-kbp HPV18 monomeric genomes are shown in lanes 14 to 16. A supercoiled circular DNA ladder (M ccc; 2 to 16 kbp) (Sigma-Aldrich) and linear DNA ladder (M lin; 1 kbp) (Naxo) are shown in lanes 17 and 18.

The HPV18E8⁻ mutant carries an ATG-to-ACG mutation in the initiator codon of the E8 ORF; therefore, it does not produce the E8E2 repressor protein, resulting in much more efficient replication of the viral genome (18, 40). The overall oligomerization patterns of HPV18wt and the HPV18E8⁻ mutant are identical (Fig. 2 and 3), so the mutant form can be used in experiments benefitting from its increased replication capability. The linear HPV18E8⁻ genome was prepared from the pBRHPV18E8⁻ parental construct using EcoRI (the restriction site is located in the E1 ORF), while the linear HPV11 genome was cleaved from the pUCHPV11 parental plasmid using BamHI (the restriction site is in the L1 ORF). Linear papillomavirus DNA is recircularized in cells after transfection and becomes fully functional regarding gene expression and replication (43). Relaxed circular HPV genomes were generated through the religation of linear HPV18E8⁻ and HPV11 genomes. The HPV18E8⁻ mutant genome and the HPV11wt genome were cloned into the pMC.BESPX minicircle producer vector (37), as described in Materials and Methods, facilitating the production of viral genomes in the form of ccc molecules. The minicircle technology provides an efficient and cost-effective system for the production of HPV genomic plasmids that are similar to the native viral genomes (18). Papillomavirus genomic episomes can be purified from bacterial cells free of the contaminating bacterial vector (37).

Fig 2.

HPV18 genomic oligomers are extrachromosomal concatemeric DNA molecules. (A) Restriction analysis of HPV18wt genomic oligomers. U2OS cells were transfected with 0.5 μg of minicircle HPV18 genomic DNA and 50 μg of salmon sperm carrier DNA. LMW DNA was extracted after 7 days. After purification, the samples were linearized with SdaI (lane 1), XmaJI (lane 2), Bpu1102I (lane 3), and BglI (lane 4) and analyzed by 1D gel electrophoresis, followed by Southern blotting. Uncut (wtH18 mc) (lane 5) and linearized (wtH18 lin) (lane 6) HPV18wt genomes were used as marker plasmids. The positions of linear DNA ladder (Naxo) fragments are indicated on the right. (B, C, E, and F) U2OS cells were transfected with 2 μg of either wild-type (B, lanes 1 to 8; C, lanes 1 to 7; E and F) or E8⁻ mutant (B, lanes 9 to 16; C, lanes 8 to 14) HPV18 genomic DNA. LMW DNA was extracted via Hirt lysis at 5 days posttransfection and analyzed through 1D (B and C) or 2D (E and F) gel electrophoresis, followed by Southern blotting. (B) Partial digestion analysis of HPV18wt and HPV18E8⁻ mutant genomic oligomers. Equal amounts of HPV18wt (lanes 1 to 8) and HPV18E8⁻ (lanes 9 to 16) LMW DNA were incubated under identical conditions for 30 min at 37°C, with increasing concentrations of BglI in each reaction mixture. Uncut (wtH18 mc) (lane 17) and linear (wtH18 lin) (lane18) HPV18 monomeric genomes were used as marker plasmids. The sizes of the fragments of a supercoiled circular ladder (Sigma-Aldrich) and linear DNA ladder (Thermo Scientific) are indicated on the right. (C) Effect of topoisomerase I treatment on HPV18wt and HPV18E8⁻ genomic oligomers. Equal amounts of HPV18wt (lanes 1 to 7) and HPV18E8⁻ (lanes 8 to 14) LMW DNA were treated with 1 U (lanes 2 to 4 and 9 to 11) or 2 U (lanes 5 to 7 and 12 to 14) of topoisomerase I at 37°C for the indicated time periods (hours). The control samples (lanes 1 and 8) were incubated for 3 h at 37°C under similar buffer conditions but were not treated with topoisomerase I. Supercoiled ladder (Sigma-Aldrich) was used as a size marker. (D) A schematic depiction of the expected migration patterns following two-dimensional (2D) neutral-neutral (N/N) gel electrophoresis. (E and F) 2D N/N gel electrophoresis of HPV18wt genome. The uncut (E) and nicked (F) HPV18wt genomes were analyzed through two-dimensional neutral-neutral gel electrophoresis. Nicking of the HPV18wt genome was achieved through digestion of the LMW DNA sample with Nb.Mva1269I (Thermo Scientific) according to the manufacturer's protocol. HPV18 genomes of different sizes and conformations were identified using marker plasmids run in both dimensions.

Fig 3.

Role of viral proteins in the formation of concatemers. (A) A schematic overview of the single mutations introduced into the HPV18wt early open reading frames. (B) A total of 2 μg of HPV18wt (lanes 1 to 3), HPV18E6⁻ (lanes 4 to 6), HPV18E7⁻ (lanes 7 to 9), HPV18E1⁻ (lanes 10 to 12), HPV18E8⁻ (lanes 13 to 15), HPV18E2⁻ (lanes 16 to 18), HPV18E4⁻ (lanes 19 to 21), HPV18E1E4⁻ (lanes 22 to 24), or HPV18E5⁻ (lanes 25 to 27) minicircle genomic DNA was transfected into U2OS cells, together with 50 μg of salmon sperm carrier DNA. LMW DNA was extracted via Hirt lysis at the indicated time points (days), and the samples were digested with DpnI for 2 h at 37°C and analyzed through 1D gel electrophoresis, followed by Southern blotting. Uncut (wtH18 mc) (lane 28) and DpnI-treated (wtH18xDpn) (lane 29) HPV18 minicircles were used as marker plasmids.

Linear, religated, and ccc forms of the HPV18E8⁻ and HPV11 genomes were transfected into U2OS cells through electroporation (Fig. 1A and B). At the indicated time points, low-molecular-weight (LMW) DNA was extracted from the cells via Hirt lysis, subsequently purified, and analyzed through Southern blotting. Experiments designed to study the HPV DNA topology during oligomerization require analysis of uncut DNA forms; however, cleavage with DpnI was also performed to remove bacterially produced input DNA. The HPV DNA replication signal consists of a series of discrete bands that increase over time. Monomeric HPV structures include supercoiled covalently closed circles (1ccc), linear molecules (1lin), and open circular molecules with single-strand nicks (1oc), all of which are indicated with arrows. Larger slower-migrating molecules also appeared, indicating the formation of oligomeric molecules. The input linear HPV18E8⁻ DNA recircularized, replicated, and oligomerized in the U2OS cells after transfection (Fig. 1A, lanes 1 to 7), although at a lower efficiency than the religated (lanes 8 to 15) or minicircle ccc (lanes 16 to 23) forms of HPV18E8⁻ genomic DNA. This may be explained by the fact that the linearized genome contains a disrupted E1 open reading frame; therefore, the expression of the functional replication protein was possible only after the perfect recircularization of the HPV18 genome. In the case of HPV11, input linear (Fig. 1B, lanes 1 to 4), religated (lanes 5 to 8), and minicircle (lanes 9 to 12) genomes replicated and oligomerized at similar efficiencies. At early time points (Fig. 1A, lanes 1 to 3, 8 to 10, 16 to 18, and B, lanes 1 and 2, 5 and 6, and 9 and 10), monomeric genomes were predominant; however, at later time points (Fig. 1A, lanes 4 to 7, 11 to 15, and 19 to 23, and B, lanes 3 and 4, 7 and 8, and 11 and 12), much larger slower-migrating molecules appear, indicating the formation of oligomeric molecules.

The results showed that the overall oligomerization patterns of the linear, religated, and ccc HPV18E8⁻ and HPV11 genomes were identical (Fig. 1A and B). Therefore, multimerization is not dependent on the topology of the input HPV DNA. Oligomerization of the HPV11 genome after transfection into U2OS cells has not been previously described, and the observation of this phenomenon indicates that oligomerization is a universal process common to most, if not all, HPV genomes.

In the U2OS-based model system, the initial transient amplification phase develops into the stable maintenance of HPV genomes, which can progress to the secondary amplification phase when dense cell culture conditions are achieved for host cells (24). To analyze the persistence of the HPV oligomers over time, U2OS cells were transfected with the HPV18wt minicircle genome and DNA replication was monitored for up to 2 weeks under two different experimental conditions (Fig. 1C). An equal number of transfected cells were seeded onto 100-mm cell culture dishes and cultivated with regular passaging under subconfluent conditions (Fig. 1C, lanes 1 to 8) or under dense cell culture conditions (Fig. 1C, lanes 9 to 12) without splitting the cells during this time period. At the indicated time points DNA was extracted, and equal amounts of the total DNA were analyzed by Southern blotting. Prior to analysis the samples were treated with DpnI. Under subconfluent cell culture conditions, there is a clear shift in the predominant form of HPV18 genomes. At early time points (1 to 2.5 days; Fig. 1C, lanes 1 to 4), HPV18 genomes are mainly in monomeric form; however, at later time points (4 to 14 days; Fig. 1C, lanes 5 to 8) oligomeric forms prevail. When HPV18-transfected cells are cultivated as dense cell cultures (Fig. 1C, lanes 9 to 12), both mono- and multimeric HPV18 episomes undergo amplification. The prevalence of oligomeric forms increases at the later time points, but monomeric HPV18 genomes also remain (Fig. 1C, lanes 10 and 11).

These results indicate that under subconfluent cell culture conditions, where active cell division takes place, HPV18 oligomeric genomes have an advantage in maintenance over the monomeric genomes. In the case of the confluent cells, where cell division is slowed down, the replication as well as maintenance of the oligomeric forms of the genome are more effective; therefore, the oligomeric forms of the genome are prevalent at later time points.

Head-to-tail concatemeric molecules are the major form of HPV multimers.

Extrachromosomal circular molecules can exist as different topological invariants. Characterization of these forms through simple agarose gel analysis does not provide conclusive results. Thus, to identify the actual forms of the HPV18 genomic oligomers, additional analysis is required. First, we performed a simple restriction analysis (Fig. 2A). LMW DNA was extracted from U2OS cells 7 days after transfection with the wild-type HPV18 minicircle genome and digested with 4 different linearizing enzymes (shown in the scheme in Fig. 2A). Oligomeric forms present in the undigested DNA (as shown in Fig. 1C, lane 6) were converted to single-unit-length linear fragments with all four linearizing endonucleases (Fig. 2A, lanes 1 to 4), which is indicative of either head-to-tail concatemers or catenated episomes.

To distinguish between head-to-tail concatemers and catenated episomes, a partial digestion analysis was performed. LMW DNA was extracted from U2OS cells transfected with wild-type HPV18 and HPV18E8⁻ DNA 5 days after electroporation via Hirt lysis. Equal amounts of LMW DNA were treated with the indicated concentrations of the BglI enzyme, which serves as a single cutter for both HPV18 genomes, for 30 min at 37°C under identical conditions (Fig. 2B, lanes 1 to 8 and 9 to 16). The samples were analyzed through one-dimensional (1D) gel electrophoresis, and the results were visualized via Southern blotting using an HPV18-specific radiolabeled probe. When catenated interlocking rings are subjected to partial digestion, the resulting linear molecules are the length of the monomer, and there can be an increase in the amount of unit-length circular molecules. We observed the generation of large linear molecules that were not present in the untreated LMW DNA samples (Fig. 2B, lanes 2 to 6 and 10 to 13) which corresponded to dimeric and trimeric linear genome lengths (marked with arrows 2lin and 3lin). As the concentration of BglI increases, these molecules will eventually disappear due to complete digestion. The experiment was repeated with the enzymes Bpu1102I and XmaJI (data not shown), and identical results were obtained. The HPV18wt (Fig. 2B, lanes 1 to 8) and HPV18E8⁻ mutant genomes (Fig. 2B, lanes 9 to 16) showed no differences regarding the partial restriction patterns. The stepwise size reduction of the linear molecules generated through partial digestion indicated that the multimeric genomes were organized as episomal head-to-tail concatemers.

Type I topoisomerases catalyze DNA knotting and unknotting (44) and the relaxation of negatively supercoiled DNA (45). Topoisomerases cleave DNA after binding to the double-stranded phosphate backbone, and this intramolecular break facilitates the untangling or unwinding of DNA. To further confirm the concatemeric nature of the HPV18 genomic oligomers, LMW DNA similar to that used in the previous experiment was subjected to treatment with topoisomerase I (Fig. 2C). The samples were treated with 1 U (Fig. 2C, lanes 2 to 4 and 9 to 11) or 2 U (Fig. 2C, lanes 5 to 7 and 12 to 14) topoisomerase I for the indicated time periods (Fig. 2C) according to the manufacturer's protocol. The control samples were incubated for 3 h under similar buffer conditions but were not treated with topoisomerase I (Fig. 2C, lanes 1 and 8). The HPV18wt and HPV18E8⁻ samples showed similar responses to topoisomerase I treatment. The covalently closed circular molecules present in the control samples (Fig. 2C, lanes 1 and 8) disappeared after the minimal topoisomerase I treatment (Fig. 2C, lanes 2 and 9), although the linked molecules remained in these samples. Even after the almost complete relaxation of all of the linked molecules following prolonged topoisomerase I treatment (Fig. 2C, lanes 7 and 14), the oligomeric forms remained, with an increasing prevalence of the relaxed forms being observed. The insensitivity of the multimers to topoisomerase I treatment provides additional proof of the concatemeric nature of these molecules.

We also analyzed the HPV18 replication intermediates via neutral-neutral (N/N) two-dimensional (2D) gel electrophoresis. This type of two-dimensional analysis facilitates the separation of linear, open circular, and covalently closed molecules. A scheme representing the expected migration of molecules of different sizes and topologies is provided in Fig. 2D (46). The LMW DNA used in the analysis was identical to that used in the previous experiments. Untreated DNA (Fig. 2E) or DNA treated with nicking enzyme Nb.Mva1269I (Fig. 2F) was analyzed through 2D N/N gel electrophoresis, and the results were visualized via Southern blotting using an HPV18-specific radioactive probe. Analysis of the untreated LMW DNA (Fig. 2E) revealed that the HPV18 genome primarily exists in infected cells as circular, oligomeric, extrachromosomal molecules. To distinguish between circular molecules in different topological conformations, particularly the ccc and open (oc) circular forms, LMW DNA samples were treated with the nicking enzyme Nb.Mva1269I (Fig. 2F), which nicks HPV18 genomic DNA in a site-specific manner, thereby converting covalently closed circular molecules into open circular molecules. Analysis of the LMW DNA revealed that the HPV18 genome is present in cells in several oligomeric forms (Fig. 2E and F) in addition to its initial monomeric form. Signals indicating large linear molecules, including the 8-kbp monomeric genome, were also present, but they were not prevalent (Fig. 2E and F). As the signal indicating linear molecules also contained discrete bands with sizes that appeared to be similar to those of the circular monomeric and oligomeric genomes, we speculate that these bands represent circular HPV18 genomes that were linearized during the DNA extraction and treatments rather than integrated forms of HPV18 genomic DNA. These data convincingly demonstrate that HPV genomic oligomers are in the form of head-to-tail concatemeric episomes.

Role of viral proteins in the formation of concatemers.

The mechanism underlying the oligomerization of HPV genomes has not been described previously. Given the limited number of proteins expressed from the HPV genome, it is possible that oligomerization is influenced by a specific viral factor(s). Therefore, we examined the role of different HPV proteins in the induction of the oligomerization of the HPV18 genome. We generated a series of single mutations in all of the HPV early open reading frames in the context of the complete HPV18 genome (Fig. 3A; also described in Materials and Methods). The two late HPV open reading frames encoding the viral capsid proteins L1 and L2 (47) were omitted from the analysis based on previous data showing efficient DNA replication and oligomerization for a HPV18 subgenomic fragment (HPV18E) containing only the early reading frames with the upstream regulatory region (URR) (see Fig. 5B) (24).

Fig 5.

(A and B) U2OS cells were transfected with 2 μg of minicircle HPV18E genomic DNA (A and B, lanes 1 to 4), 2 μg of HPV18E together with 1 μg of HPV18 genomic DNA (A and B, lanes 5 to 8), and 1 μg of HPV18 minicircle genomic DNA (A, lanes 9 to 12; B, lanes 9 to 16) in the presence of 10 μg of pauxoMCS (A) or salmon sperm (B) carrier DNA. LMW DNA was extracted via Hirt lysis at the indicated time points (days) and analyzed through 1D gel electrophoresis, followed by Southern blotting. (A) Southern blot analysis of the transient replication of the HPV18E and HPV18 genomes in U2OS cells. Prior to analysis, the samples were treated with DpnI and BspTI, which linearized both the HPV18E and HPV18wt genomes. Linear and DpnI-treated marker plasmids for HPV18E are shown in lanes 13 and 14, and marker plasmids for HPV18 are shown in lanes 15 and 16. (B) Southern blot analysis of the transient replication of the HPV18E and HPV18 genomes in U2OS cells. Prior to analysis, the samples in lanes 1 to 12 were treated with DpnI and XmaJI, which only acts as a linearizing enzyme for the full-length HPV18 genome. The HPV18wt samples in lanes 13 to 16 were treated with DpnI. Uncut, linear, and DpnI-treated marker plasmids for HPV18E are shown in lanes 17 to 19, and marker plasmids for HPV18wt are shown in lanes 20 to 22. The positions of linear DNA ladder fragments (Naxo and Thermo Scientific) are indicated on the right. The 13.4-, 18.8-, and 24.2-kbp linear fragments are indicated with arrows. (C) A schematic representation of monomeric and dimeric HPV18E and HPV18wt genomes.

Mutations eliminating the expression of E1, E2, E4, E1E4, E5, E6, E7, or E8 were introduced into the wild-type genome of HPV18, and the corresponding covalently closed monomeric minicircle genomes were produced, purified, and transfected into U2OS cells through electroporation (Fig. 3B). Three independent experiments were performed, generating similar results. LMW DNA was extracted at the indicated time points and analyzed by Southern blotting. Prior to the analysis, the samples were treated with DpnI to remove input DNA.

The HPV18E1⁻ and HPV18E2⁻ mutants (Fig. 3B, lanes 10 to 12 and 16 to 18) were predictably replication defective, as these mutants do not express the essential viral replication proteins E1 and E2, respectively. Thus, these mutants served as controls to show that transient HPV replication in the U2OS cell line follows the thoroughly described E1- and E2-dependent mechanism of papillomavirus DNA replication initiation (6–8, 11, 14, 15, 43, 48).

As expected, the HPV18E8⁻ genome replicates much more effectively than the wild-type genome (Fig. 3B, compare lanes 13 to 15 to lanes 1 to 3) but with a similar oligomerization pattern.

HPV E6 and E7 are viral oncoproteins that are capable of inducing immortalization of the host cell (49). E6 induces the degradation of the cellular tumor suppressor protein p53 (50), while E7 binds (51) and induces the degradation of the cellular tumor suppressor protein pRB (52). Together, these proteins create a supportive cellular milieu for viral DNA replication while also modulating the activity of some of the key cellular proteins involved in cell cycle control and the DNA damage response (53–55). We observed more intensive replication, leading to more efficient oligomerization for the E7 mutant compared to that of the E6 mutant and the wild-type HPV18 genome (Fig. 3B, compare lanes 7 to 9 to lanes 4 to 6 and 1 to 3), but the overall pattern of oligomerization for the E6 and E7 mutants remained unchanged (Fig. 3B, lanes 4 to 6 and 7 to 9).

HPV E4 and E1E4 have been implicated in inducing the collapse of the intermediate filament network of the host cell (56, 57). The E1E4 protein also exhibits cell cycle-regulatory properties (58). The HPV E5 protein associates with several host cell transmembrane receptors (59–61), supports cell cycle progression, and activates late viral functions upon host cell differentiation (62). Mutations in E4 (Fig. 3B, lanes 19 to 21), E1E4 (Fig. 3B, lanes 22 to 24), and E5 (Fig. 3B, lanes 25 to 27) did not change the oligomerization pattern, although mutations in E4 and E5 resulted in a slight reduction of the overall level of HPV replication.

These data suggest that no viral proteins, with the exception of E1 and E2, are crucial to the oligomerization of the HPV18 genome. Therefore, it is likely that an association exists between the replication and oligomerization of the viral genome.

URR-dependent replication and oligomerization of HPV18.

To better understand the potential overlap between replication and oligomerization, we examined the effects of the crucial replication proteins E1 and E2 on oligomerization. For this purpose, we reverted from using the full-length viral genome, which allows simultaneous analysis of the efficiency of replication factor expression and the origin functions, and instead provided these two replication factors from expression vectors (8, 10). The HPV origin of replication is located in the URR of the viral genome. The HPV18 URR minicircle (mc) construct was generated from the BamHI-BamHI fragment of the HPV18wt genome as a source of the papillomavirus origin. The HPV18 URR minicircle plasmids were produced by following the standard minicircle production protocol (described in Materials and Methods), and the molecules are in the form of covalently closed ∼1-kbp episomes. Efficient expression plasmids for E1 and E2 have been previously described (41). The HPV18 URR minicircle plasmid was cotransfected into U2OS cells in the presence of various concentrations of E1 and E2 expression vectors (Fig. 4A). The results indicated that transfection of 10 ng of the E1 and E2 expression plasmids was sufficient to trigger efficient replication (Fig. 4A, lanes 1 and 2) and oligomerization (Fig. 4A, lanes 13 and 14) of the HPV18 URR minicircle.

Fig 4.

(A) Effect of E1 and E2 protein levels on transient replication of the HPV18 URR plasmid. U2OS cells were transfected with HPV18 E1 and E2 expression plasmids in the presence of 0.5 μg of HPV18 URR minicircle plasmid DNA and 50 μg of salmon sperm carrier DNA. E1 and E2 were titrated to obtain final plasmid DNA concentrations of 10 or 100 ng (lanes 1 to 8 and 13 to 20). Low-molecular-weight DNA was extracted at 24 and 48 h posttransfection, separated via 1D gel electrophoresis, and analyzed by Southern blotting. Prior to analysis, the samples were treated with the linearizing enzymes BglI and DpnI (lanes 1 to 8) or DpnI (lanes 13 to 20). Uncut (H18URR mc) (lanes 9 and 21), DpnI-treated (HPV18URRxDpn) (lanes 10 and 22), linear (H18URR lin) (lanes 11 and 23), and linear DpnI-treated (H18URR linxDpn) (lanes 12 and 24) HPV18 URR minicircles were used as marker plasmids. (B) Southern blot analysis of the transient replication of HPV18 and HPV18 URR minicircle DNA in U2OS cells. U2OS cells were transfected either with 1 μg of HPV18 and 0.5 μg of HPV18 URR minicircle DNA (lanes 1 to 4) or with 0.5 μg of HPV18 URR minicircle DNA and 50 ng of the E1 and E2 expression vectors (lanes 5 to 10) in the presence of 10 μg of the pauxoMCS carrier DNA plasmid (Icosagen). LMW DNA was extracted at the indicated time points (days) and analyzed by 1D gel electrophoresis, followed by Southern blotting. Prior to analysis, the samples were treated with DpnI and PaeI, which linearized the full-length HPV18 genome. The linearized (wtH18 lin) (lane 11) and DpnI-treated (wtH18xDpn) (lane 12) minicircle HPV18 genomes and uncut (H18URR mc) (lane 13), linearized (H18URR lin) (lane 14), and DpnI-treated (H18URRxDpn) (lane 15) HPV18URR minicircle plasmids were used as markers. Lanes 7 to 10 represent 5 times shorter exposure than lanes 1 to 6. (C) 2D N/N gel electrophoresis analysis of the transient replication of HPV18 URR minicircle DNA in U2OS cells in the presence of E1 and E2 expression plasmids. U2OS cells were transfected with 0.5 μg of HPV18 URR minicircle DNA together with 10 ng of the E1 and E2 expression constructs in the presence of 10 μg of pauxoMCS carrier DNA plasmid. Low-molecular-weight DNA was extracted at 72 h posttransfection. Prior to analysis, the samples were treated with DpnI and analyzed through 2D N/N gel electrophoresis, followed by Southern blotting. HPV18 URR oligomeric molecules were identified using marker plasmids for both dimensions. (D) A schematic depiction of the expected migration pattern of open circular, covalently closed circular, and linear molecules. (E) 2D N/N gel electrophoresis analysis of the transient replication of HPV18 URR minicircle DNA in U2OS cells in the presence of the HPV18E8⁻ mutant genome. U2OS cells were transfected with 0.5 μg of HPV18 URR minicircle DNA together with 0.5 μg of HPV18E8⁻ in the presence of 50 μg of salmon sperm carrier DNA plasmid. Low-molecular-weight DNA was extracted at 48 h posttransfection. Prior to analysis, the samples were treated with XmaJI, which linearized the full-length HPV18E8⁻ genome, and analyzed with 2D N/N gel electrophoresis, followed by Southern blotting. The 8-, 9-, 10-, and 11-kbp linear fragments are indicated. HPV18 URR oligomeric molecules were identified using marker plasmids for both dimensions.

The HPV18 URR minicircle construct also effectively coreplicates and cooligomerizes when the wild-type HPV18 genome is the source of E1 and E2 (Fig. 4B). The HPV18 URR minicircle plasmid was transfected into U2OS cells in the presence of either the HPV18wt genome (lanes 1 to 4) or the E1 and E2 expression vectors (lanes 7 to 10, with longer exposures for the two time points repeated in lanes 5 and 6). In the experiment involving cotransfection with HPV18wt, the full-length genome was linearized with PaeI to distinguish the replication products of the URR minicircle from HPV18wt minicircles (Fig. 4B, lanes 1 to 4). In both cases, the HPV18 URR oligomers formed a clearly visible series of discrete bands. Upon cotransfection of the URR minicircle with the E1 and E2 expression vectors, the oligomeric molecules showed strong signals from 2 days after transfection (Fig. 4B, lanes 5 and 6 and/or 7 to 10). Following cotransfection with the HPV18wt genome, oligomerization of the URR mc was apparent at later time points (4 and 5 days) (Fig. 4B, lanes 3 and 4). This potentially reflects the delayed onset of E1 and E2 expression from the viral genome and the reduced expression of E1 and E2 proteins in cells transfected with the HPV18 genome as opposed to the cells subjected to transfection with the expression vectors. Taken together, these results show that the efficiency of oligomerization depends on the efficiency of replication, but the pattern of oligomerization itself remains unchanged independent of the source or levels of E1 and E2 expression.

To confirm that the HPV URR oligomers are comparable to the full-length genomic oligomers, two-dimensional neutral/neutral gel electrophoresis was performed. LMW DNA was extracted from U2OS cells transfected with the HPV18 URR minicircle and E1 and E2 expression vectors, treated with DpnI, and analyzed by 2D agarose gel electrophoresis (Fig. 4C). The results indicated the presence of the same basic arcs of linear, open circular, and covalently closed circular molecules (schematic depiction in Fig. 4D), as detected in the experiments performed with the full-length HPV18 genome (Fig. 2D to F). More than 10 separate molecules with consecutively increasing sizes can be counted on each of the arcs. Two-dimensional neutral/neutral gel electrophoresis was also performed to analyze the replication products that arise after the cotransfection of HPV18 URR minicircle with the HPV18E8⁻ genome (Fig. 4E). Prior to analysis the samples were digested with XmaJI, a linearizing enzyme for the HPV18E8⁻ genome, resulting in the presence of a prominent 8-kbp dot representing the full-length HPV18E8⁻ genome on the linear arc. The analysis of the uncut HPV18 URR replication products revealed the existence of the same arcs of linear, open circular, and covalently closed circular HPV18 URR molecules as those detected in the experiments performed with the expression vectors (compare Fig. 4E [long exposure] to C). Taken together, these findings show that replication of the HPV18 URR plasmid together with the E1 and E2 expression vectors creates oligomeric molecules with the same basic characteristics as those obtained following coreplication of the full-length genome, further indicating that oligomerization is dependent on viral DNA replication.

On closer inspection, however, the analysis revealed the appearance of additional 9-, 10-, and 11-kbp linear fragments that could represent molecules containing the HPV18E8⁻ genome linked with increasing numbers of copies of the HPV18 URR minicircles (Fig. 4E, short exposure). The appearance of such linked oligomers suggests that homologous recombination takes place between two replicating HPV molecules.

HPV replication-dependent oligomers arise through homologous recombination.

In principle, there are several mechanisms associated with replication that could be responsible for generating oligomeric HPV genomes. For example, rolling-circle replication has been implicated as a mechanism for carrying out secondary amplificational replication during the later stages of the PV life cycle (16, 17). The Mre11-Rad50-Nbs1 (MRN) complex, associated with homologous recombination, is connected with the generation of adenoviral genomic oligomers (63). The recruitment of components of the MRN complex to HPV replication foci also has been observed (23). One of the differences between oligomers generated through rolling-circle replication and those generated through homologous recombination is that recombination takes place between two separate (partially or fully homologous) molecules, but rolling-circle replication uses a single molecule as a replication template to create multimers. To elucidate whether homologous recombination is involved in oligomerization or whether rolling-circle replication is implicated in this process, additional cotransfections were performed using two distinguishable HPV genome molecules.

U2OS cells were transfected with the subgenomic 5.4-kbp HPV18E minicircle construct (Fig. 5A and B, lanes 1 to 4), the 8-kbp wild-type HPV18 minicircle genome (Fig. 5A, lanes 9 to 12, and B, lanes 9 to 16), or the HPV18E construct cotransfected with the HPV18 genome (Fig. 5A and B, lanes 5 to 8). The HPV18E construct lacks the L1 and L2 open reading frames but contains URR and early regions of the HPV18 genome identical to those in the wild-type genome (24). The transfections were performed under identical conditions, and LMW DNA was extracted from the cells at the indicated time points and analyzed by Southern blotting. The DNA samples were digested with BspTI, which acts as a linearizing enzyme for both the HPV18E and HPV18wt genomes (Fig. 5A). This analysis confirmed that both molecules retain their ability to independently replicate in host cells upon cotransfection (Fig. 5A, lanes 5 to 8), similar to what is seen following separate transfection of these cells (Fig. 5A, lanes 1 to 4 and 9 to 12).

To examine the possible recombination events that occur between HPV18E and HPV18wt, we took advantage of the unique restriction sites present in HPV18wt that do not exist in HPV18E (Fig. 5B). LMW DNA samples were digested with XmaJI, which cleaves the full-length HPV18 genome once within the L2 reading frame, linearizing the HPV18wt genome (Fig. 5B, lanes 9 to 12) but leaving the HPV18E construct uncleaved (Fig. 5B, lanes 1 to 4). Dimers and higher oligomeric forms (Fig. 5C) were observed in the uncut HPV18E (Fig. 5B, lanes 1 to 4) and HPV18wt (Fig. 5B, lanes 13 to 16) DNA samples. XmaJI treatment of the LMW DNA samples extracted from cells cotransfected with both HPV18E and HPV18wt predictably generated an 8-kbp linear fragment, increasingly present with time, representing the full-length HPV18 genome (Fig. 5B, lanes 5 to 8). However, treatment with XmaJI also generated 13.4-, 18.8-, and 24.2-kbp linear fragments (Fig. 5B, lanes 5 to 8) not observed in the analysis of samples from the single-molecule transfections (Fig. 5B, lanes 1 to 4 and 9 to 16). The 13.4-kbp fragment corresponds to dimeric forms containing one HPV18wt and one HPV18E genome (Fig. 5C). The 18.8-kbp linear fragment indicates the presence of oligomeric molecules containing the HPV18wt genome with two copies of the HPV18E genome; thus, the 24.2-kbp linear fragment should contain a single copy of the HPV18wt genome and three copies of the HPV18E genome.

Because digestion with BspTI (Fig. 5A, lanes 5 to 8) produced unit-length HPV18E (5.4-kbp) and HPV18wt (8-kbp) molecules from oligomeric molecules generated after cotransfection, the different hetero-oligomers must contain the HPV18E and HPV18wt genomes in a head-to-tail orientation. These results suggest that hetero-oligomeric forms of HPV18 are generated through intermolecular homologous recombination, which is the only mechanism capable of producing these types of joint molecules. Homo-oligomers containing only the HPV18wt or HPV18E genome likely arise through the same mechanism, but involvement of rolling-circle replication cannot be entirely excluded for these molecules based on current results.

A similar experiment was performed with HPV11 genomic mutants (Fig. 6). Two HPV11 full-length genome mutants were generated through deletions in the E1 and E2 open reading frames. Alone, these mutants are inactive regarding replication (Fig. 6, lanes 17 and 18). However, cotransfection of the E1 and E2 mutant genomes resulted in complementation of their genetic defects, and replication and oligomerization were observed for both mutants (Fig. 6, lanes 5 to 8), similar to the HPV11 wild-type genome (Fig. 6, lanes 1 to 4). In addition to not synthesizing crucial viral replication proteins, the E1 mutant lacks the PaeI restriction site common to wild-type HPV11 genomes, and the E2 mutant lacks the SdaI restriction site. When LMW DNA extracted from U2OS cells cotransfected with the HPV11E1⁻ and HPV11E2⁻ mutant genomes is simultaneously treated with the PaeI and SdaI restriction enzymes and no joint molecules containing both mutants are evident, linear unit-length molecules similar to linearized wild-type genomic molecules should be observed (Fig. 6, lanes 9 to 12). In addition to linear unit-length molecules, digestion with the PaeI and SdaI single cutters also generates 10- and 6-kbp linear fragments, which are produced only when hetero-oligomers containing both the E1 and E2 mutant genomes are present in cells. These results suggest that HPV genomic oligomers are produced through the homologous recombination of replicating molecules. These data provide no evidence of the involvement of rolling-circle replication in this process, but theoretically it cannot be entirely excluded as an alternative partial mechanism.

Fig 6.

Southern blot analysis of the transient replication of HPV11 mutants in U2OS cells. U2OS cells were transfected with 2 μg of HPV11wt (lanes 1 to 4 and 9 to 12), HPV11E1⁻ mutant (lane 17), and HPV11E2⁻ mutant DNA (lane 18) or were cotransfected with 2 μg of HPV11E1⁻ mutant DNA together with 2 μg of HPV11E2⁻ mutant DNA (lanes 5 to 8 and 13 to 16). LMW DNA was extracted via Hirt lysis at the indicated time points (days) and analyzed through 1D gel electrophoresis, followed by Southern blotting. HPV11wt (lanes 9 to 12) and HPV11E2⁻ (lane 18) samples were digested with the linearizing enzyme PaeI and with DpnI, HPV11E1⁻ (lane 17) sample was digested with the linearizing enzyme SdaI and with DpnI, while HPV11E1⁻ plus HPV11E2⁻ samples (lanes 13 to 16) were digested with linearizing enzymes PaeI (only cuts the E2 mutant) and SdaI (only cuts the E1 mutant) and DpnI. HPV11wt and HPV11E1⁻ plus HPV11E2⁻ uncut samples were digested with DpnI (lanes1 to 8). Uncut (lane 19) and linear (lane 20) HPV11 minicircle plasmids were used as molecular size markers in addition to supercoiled (lane 21) (Sigma-Aldrich) and linear (lane 22) (Thermo Scientific) ladders. A schematic representation of monomeric and dimeric HPV11 genomes is provided.

Oligomerization of the HPV genomes can be detected in the SiHa, HeLa, and C-33 A cell lines.

Analysis of previous publications suggests that HPV genomic concatemers can be detected in HPV-immortalized human keratinocytes (64, 65). The W12 cells after passage 10 revealed the presence of monomeric HPV16 genomes together with high-molecular-weight viral genomic DNA, which was proved to represent head-to-tail episomal HPV16 multimers (65). Actively replicating and amplifying high-molecular-weight oligomeric episomal HPV genomes have also been described in the HPV31b-positive cell line CIN612 (64). Both cell lines were derived from low-grade cervical lesions histologically diagnosed as CIN I and are considered to represent a natural model of HPV infection (66, 67).

SiHa and HeLa cells contain transcriptionally active HPV16 and HPV18 integrated sequences (68, 69), and both have been initiated from cervical carcinomas (70, 71). We have previously described efficient transient replication of the HPV18 and HPV16 genomes after transfection into the SiHa and HeLa cell lines (20); however, we have not monitored the physical state of HPV DNA in these cells. The C-33 A cell line has been propagated from a cervical carcinoma but does not contain HPV genomic material (72).

Transient replication assays were performed in SiHa, HeLa, and C-33A cells to determine the ability of these cell lines to support the oligomerization of HPV genomes. Cells were transfected with the HPV18E8⁻ mutant (Fig. 7A, B, and D) or HPV11wt (Fig. 7C and E) genome. LMW DNA was extracted at the indicated time points via Hirt lysis and analyzed by Southern blotting. Concentration-dependent replication and oligomerization were observed for both the HPV18E8⁻ (Fig. 7A, lanes 1 to 3, B, lanes 5 to 7, and D, lanes 7 to 9) and HPV11wt (Fig. 7C, lanes 4 to 9 and 13 to 18, and E, lanes 1 to 3 and 7 to 9) genomes in all three cell lines. These results indicate that oligomerization of the HPV genomes is not common to the U2OS cell line alone.

Fig 7.

(A and B) Southern blot analysis of the transient replication of the HPV18E8⁻ genome in SiHa and HeLa cells. SiHa (A) and HeLa (B) cells were transfected with 2 (B) or 5 (A) μg of the HPV18E8⁻ mutant genome in the presence of 50 μg of salmon sperm carrier DNA. LMW DNA was extracted at the indicated time points (days) via Hirt lysis and analyzed by 1D gel electrophoresis, followed by Southern blotting. Prior to analysis, the samples were digested with DpnI (A, lanes 1 to 3; B, lanes 5 to 7) or with linearizing enzyme BglI and DpnI (A, lanes 4 to 6; B, lanes 8 to 10). In the case of HeLa cells, 100- to 1,000-pg samples (B, lanes 1 to 4) of total DNA extracted from untransfected cells were used as controls. Marker plasmids for the uncut (H18E8⁻ mc), linear (H18E8⁻ lin), and DpnI-treated (H18E8⁻xDpnI and H18E8⁻xlin/DpnI) 8-kbp HPV18E8⁻ monomeric genomes are shown in lanes 7 to 10 (A) and 11 to 14 (B). A supercoiled circular DNA ladder (M ccc; 2 to 16 kbp) (Sigma-Aldrich) and linear DNA ladder (M lin; 1 kbp) (Naxo) are shown in lanes 11 and 12 (A) and 15 and 16 (B). (C) Southern blot analysis of the transient replication of HPV11 genomes in SiHa and HeLa cells. SiHa (lanes 1 to 9) and HeLa (lanes 10 to 18) cells were transfected with 2 (lanes 1 to 3 and 10 to 12), 5 (lanes 4 to 6 and 13 to 15), and 10 (lanes 7 to 9 and 16 to 18) μg of the HPV11wt genome in the presence of 50 μg of salmon sperm carrier DNA. LMW DNA was extracted at the indicated time points (days) via Hirt lysis and analyzed by 1D gel electrophoresis, followed by Southern blotting. Prior to analysis the samples were digested with DpnI. Monomeric HPV11 genomes in covalently closed circular, linear, and open circular forms were used as markers (wtH11 mc) (lane 19). (D) Southern blot analysis of the transient replication of the HPV18E8⁻ mutant genomes in C-33 A cells. C-33 A cells were transfected with 2 (lanes 1 to 6) or 5 (lanes 7 to 12) μg of HPV18E8⁻ genomic DNA. LMW DNA was extracted at the indicated time points (days) and analyzed through 1D gel electrophoresis, followed by Southern blotting. Prior to analysis the samples were digested with DpnI (lanes 1 to 3 and 7 to 9) or linearizing enzyme PsyI and DpnI (lanes 4 to 6 and 10 to 12). Monomeric HPV18E8⁻ genomes in covalently closed circular, linear, and open circular forms were used as markers (lane 13, H18E8⁻ mc). (E) Southern blot analysis of the transient replication of the HPV11wt genomes in C-33 A cells. C-33 A cells were transfected with 2 (lanes 1 to 6) or 5 (lanes 7 to 12) μg of HPV11wt genomic DNA. LMW DNA was extracted at the indicated time points (days) and analyzed by 1D gel electrophoresis, followed by Southern blotting. Prior to analysis the samples were digested with DpnI (lanes 1 to 3 and 7 to 9) or with linearizing enzyme PaeI and DpnI (lanes 4 to 6 and 10 to 12). Monomeric HPV11 genomes in covalently closed circular, linear, and open circular forms were used as markers (lane 13, wtH11 mc).

HPV genomic oligomers can be observed in HPV-infected patient tissue samples.

The final proof that oligomerization of HPV genomes is a natural process, however, can only be provided by demonstrating the existence of multimeric HPV genomes in vivo. To address the relevance of HPV genomic concatemers in the viral life cycle, we have initiated a collaborative study with the Womens's Clinic of the University of Tartu. Cervical brush or colposcopy samples were collected from women previously diagnosed with persistent HPV18 or HPV16 infections who were undergoing follow-up studies. We selected some of these samples to provide examples of the existence of different forms of HPV DNA in vivo. The physical state of HPV DNA was characterized by following the example set in Cullen et al. (29).

Reference analysis was performed with (i) LMW DNA from HPV18-transfected U2OS cells (Fig. 8A, lanes 1 to 6), which serves as an example of the coexistence of mono- and multimeric episomal HPV genomes; (ii) HPV18-positive U2OS subclone 1.13 (Fig. 8A, lanes 7 to 12) and HPV16-positive U2OS subclone 3.16 (Fig. 8B, lanes 1 to 6), which carry episomal head-to-tail multimers (24); (iii) total DNA from the CaSki cell line (Fig. 8B, lanes 20 to 24), which contains tandem head-to-tail repeats of integrated HPV16 DNA (73); and (iv) monomeric HPV18 and HPV16 genomes in covalently closed circular, linear, and open circular forms (Fig. 8A, lanes 13 and 14 as well as 22 and 23, and B, lanes 18 and 19).

Fig 8.

Southern blot analysis of HPV-positive subclones, patient samples, HPV18-transfected U2OS cells, and the CaSki cell line. A total of 50 (B, lane 18) or 500 (A, lanes 14 and 22) pg of uncut and 10 (B, lane 19) or 300 (A, lanes 13 and 23) pg of linear HPV18wt (A) or HPV16wt (B) minicircle genomic DNA was used as a plasmid marker. The positions of 8-kbp linear (lin), open circular (oc), and covalently closed circular (ccc) molecules and oligomeric forms are marked at the left side of the panels. (A) Southern blot analysis of HPV18-transfected U2OS cells, HPV18-positive subclone 1.13, and patient sample 18.1. LMW DNA extracted 5 days posttransfection from HPV18-transfected U2OS cells (lanes 1 to 6), LMW DNA extracted from stable HPV18-positive cell line 1.13 (lanes 7 to 12), and total DNA extracted from patient sample 18.1 (lanes 17 to 21) was digested with appropriate noncutter (lanes 2 and 3, 8 and 9, and 18 and 19) or single-cutter (lanes 4 to 6, 10 to 12, and 20 and 21) enzymes and analyzed by 1D gel electrophoresis, followed by Southern blotting. (B) Southern blot analysis of the HPV16-positive subclone 3.16, patient samples 16.1 and 16.2, and the HPV16-positive CaSki cell-line. Two μg of total DNA extracted from HPV16-positive cell line 3.16, 5 μg of total DNA extracted from patient sample 16.1, 4 μg of total DNA extracted from patient sample 16.2, and 2 μg of total DNA extracted from the HPV16-positive CaSki cell line were digested with appropriate noncutter (lanes 2 and 3, 8 and 9, 14 and 15, and 20 and 21) or single-cutter (lanes 4 to 6, 10 to 12, 16 to 17, and 22 to 24) enzymes and analyzed by 1D gel electrophoresis, followed by Southern blotting.

Total DNA was extracted from the patient samples as described in Materials and Methods, and restriction analysis followed by 1D gel electrophoresis was performed (Fig. 8A and B). Two noncutting and two to three linearizing enzymes were chosen for the restriction analysis. Uncut DNA was also analyzed when a sufficient amount of the DNA was available. Episomal HPV molecules should display a similar signal when treated with different noncutting or linearizing enzymes (Fig. 8A, lanes 1 to 6). When integrated or rearranged HPV sequences are present in samples, restriction analysis should generate differences between the treatments performed with the different enzymes (Fig. 8B, lanes 22 to 24).

The results of the restriction analysis of patient sample 18.1 (Fig. 8A, lanes 17 to 21) mirrored those of subclones 1.13 (Fig. 8A, lanes 7 to 12) and 3.16 (Fig. 8B, lanes 1 to 6), indicating the presence of at least dimeric episomal HPV18 genomic molecules.

The results of the restriction analysis of patient sample 16.1 (Fig. 8B, lanes 7 to 12) were similar to those of the LMW DNA extracted from HPV-transfected U2OS cells (Fig. 8A, lanes 1 to 6), indicating the presence of mono- and dimeric HPV genomes. As minor non-unit-length HPV16 DNA signals appeared when the samples were treated with linearizing enzymes (Fig. 8B, lanes 10 and 12), there is also a possibility of the coexistence of integrated HPV16 genomic material. Comparison of the restriction pattern of the CaSki cell line to that of the patient sample, however, indicates that the majority of the HPV16 genomic material in the patient tissue is in episomal form (Fig. 8B, compare lanes 7 to 12 to lanes 20 to 24).

The results of the restriction analysis of patient sample 16.2 (Fig. 8B, lanes 13 to 17) mirror those of the LMW DNA extracted from HPV-transfected U2OS cells (Fig. 8A, lanes 1 to 6), indicating the presence of mono-, di-, and multimeric forms of HPV16 DNA, with oligomeric forms being prevalent.

Additional studies of larger numbers of patient samples have to be conducted in order to make final conclusions. However, together with the findings reported in previous publications, particularly those from a number of years ago (27–29, 31, 74), these results strongly suggest that the oligomerization of the HPV genome is a naturally occurring phenomenon.

DISCUSSION

Oligomeric episomal forms of the HPV genomes were first demonstrated almost 30 years ago in clinical samples obtained from HPV-associated cervical lesions (27–32, 74). However, these oligomers have not been adequately studied and described, and the relevance of oligomerization in the viral life cycle and the mechanisms underlying the formation of oligomers remained elusive. Even more, integration of the high-risk virus genomes and correlation of this process with formation of cancer were primarily studied. The appearance of the high-molecular-weight forms of HPV genomes on the Southern blots were interpreted as integration of the viral genomes without further analysis.

We used the U2OS cell line-based assay system to show that HPV oligomeric genomes are episomal concatemers containing the viral genome in a head-to-tail configuration (Fig. 2A). The observed oligomerization does not depend on the topology of the viral genomes transfected into the cells (Fig. 1), and no viral proteins are required other than replication proteins E1 and E2 (Fig. 3). However, E1- and E2-driven replication initiated from the replication origin of the viral genome is crucial for the triggering of oligomerization, and the expression level of these proteins directly affects the efficiency of this process (Fig. 4A and B). This last point could also explain the positive influence made by the E7 and E8 open reading frame mutations on the replication and oligomerization of the viral genome (Fig. 3), as these mutations increase the functional levels of E1 and E2 expression, respectively.

The existence of oligomeric episomal HPV genomes has previously been observed in the case of HPV16 genomes in the W12 cell line (65) and in the case of HPV31-b genomes in the CIN612 cell line (64). Both cell lines are presumed to represent a natural model of HPV infection (66, 67). We demonstrated the oligomerization of HPV18E8⁻ and HPV11 genomes in SiHa, HeLa, and C-33 A cell lines (Fig. 7). All three of the cell lines have been derived from cervical carcinomas. SiHa cells contain a single integrated copy of the HPV16 genome, which is transcriptionally active (68, 75). HeLa cells contain transcriptionally active HPV18-integrated DNA (69, 74). C-33 A cells represent an HPV-negative human cervical cell line (72, 73). Together with the detection of oligomeric HPV genomes in samples obtained from HPV-infected patients (Fig. 8), these results suggest that oligomerization is not a phenomenon specific to U2OS cells but rather is a general feature coupled with viral DNA replication.

The progressive formation of HPV18 URR episomal oligomeric plasmids in the presence of the HPV18 E1 and E2 replication proteins (Fig. 4) indicates that the oligomerization of HPV genomes takes place during the E1- and E2-dependent replication initiated from the viral origin. In general, it is believed that replication of the papillomavirus genomes occurs via theta structures, which is indicative of the bidirectional initiation of DNA replication. However, head-to-tail oligomers can be formed by two mechanisms: first by rolling-circle replication and second by recombinational replication. A combination of these mechanisms in this process is also possible. Rolling-circle replication can generate long linear concatemeric molecules, which later must be processed and assembled into head-to-tail oligomeric plasmids by the DNA processing and modification systems. However, rolling-circle replication generates the precursors of the oligomers only from single species of the molecules, which means that intermolecular oligomers between different replicating homologous molecules cannot be generated in this way. The cotransfection experiments of the two different replicating HPV genomic and subgenomic plasmids allowed us to demonstrate that hetero-oligomers can be generated. The HPV18 genome forms hetero-oligomers with the truncated 5.4-kbp subgenomic plasmid as well as with the 1-kbp URR minicircle plasmid. In a complementation assay with the HPV11 E1 and E2 mutants, we also demonstrated formation of the hetero-oligomers of the E1 and E2 mutant genomes. These data suggest that homologous recombination of the replicating molecules not only takes place within the same molecules but also occurs by the intermolecular recombination of homologous sequences (Fig. 4E, 5, and 6). Recruitment of components of the MRN complex to HPV replication foci has been previously demonstrated (23). Our data suggest that the appearance of HPV oligomeric genomic structures is due to the involvement of homologous replication in viral genome replication.

Homologous recombination is a cellular repair mechanism for double-stranded DNA breaks (reviewed in reference 76). The components of the homologous replication machinery bind to DNA strands near the break, recruit nucleases to create short 3′ single-stranded DNA (ssDNA) ends, and mediate the invasion of those ssDNA ends into a homologous DNA duplex, resulting in assembly of the replication complex and the resynthesis of damaged DNA sequences using the invading 3′ ssDNA ends as primers. The end products of homologous recombination contain crossover structures between the two molecules, known as Holliday junctions, which might be resolved in a manner that leads to the creation of oligomeric molecules. We have studied the replication intermediates and detected the molecules connected by the Holliday junctions (data not shown), and the data additionally support our conclusion.

Our data demonstrate clearly that E1- and E2-dependent replication of the HPV genomes recruits cellular homologous recombination machinery at the replicating molecules, which results in formation of the oligomeric HPV genomes. It is highly likely that E1 protein is one of the key players in recruiting the homologous recombination apparatus at the HPV genome, and formation of the oligomers is rather indicative of this process. The induction of double-stranded breaks is absolutely necessary for the recruitment of the HR machinery to a molecule. These breaks can be random, but in the case of the HPV genome, they might also be generated by the E1 protein, which has been shown to produce such breaks (18, 19, 21). Furthermore, the HR machinery is recruited to breaks induced through collapsed replication forks (76), which could occur during replication of the viral genome. It is unclear so far how the HR machinery is recruited at the replicating genome, how this process is regulated at the molecular level, and how the switch from the bidirectional replication to the homologous recombination can occur. The data presented here clearly suggest that further studies about the linkage of the initiation of viral DNA replication and homologous replication machinery into the process of HPV genome replication are needed.

The potential role of oligomeric genomes in the viral life cycle remains unknown. It has been demonstrated that larger DNA molecules are maintained with better efficiency in proliferating mammalian cells than smaller ones (77). Therefore, we speculate that oligomerization provides an advantage through retaining the viral genomes in the host cell during the stable maintenance phase, when the efficiency of viral genome replication is low due to the reduced expression level of the replication proteins. Oligomeric molecules do become the prevalent form of viral genomes during the 2-week cultivation of HPV-transfected cells under subconfluent cell culture conditions (Fig. 1C). The previous observation that positive selection for multimeric HPV genomes occurs during the long-term cultivation of HPV-positive cell lines supports this speculation (24). When secondary amplification of these cell lines was induced, all of the forms of oligomeric genomes were found to be capable of intense replication. These results were mirrored by the 2-week cultivation of HPV18-transfected cells under dense cell culture conditions (Fig. 1C) in the current publication.

It has been suggested that BPV1 (78) and HPV (24) genomic oligomers carry only one active replication origin. Under these conditions, it would be possible to replicate a large number of viral genomes in the presence of low levels of viral replication proteins, thereby maintaining a high genomic copy number while the infected host cells remain undetected by the adaptive immune system. Therefore, we hypothesize that the oligomerization of HPV genomes is a vital step for entering the stable maintenance phase of the viral replication cycle. The exact mechanisms underlying oligomerization are currently unknown, but they have become an increasingly interesting subject of ongoing study in our research group.