High-Risk Alphapapillomavirus Oncogenes Impair the Homologous Recombination Pathway

ABSTRACT

Persistent high-risk genus human Alphapapillomavirus (HPV) infections cause nearly every cervical carcinoma and a subset of tumors in the oropharyngeal tract. During the decades required for HPV-associated tumorigenesis, the cellular genome becomes significantly destabilized. Our analysis of cervical tumors from four separate data sets found a significant upregulation of the homologous-recombination (HR) pathway genes. The increased abundance of HR proteins can be replicated in primary cells by expression of the two HPV oncogenes (E6 and E7) required for HPV-associated transformation. HPV E6 and E7 also enhanced the ability of HR proteins to form repair foci, and yet both E6 and E7 reduce the ability of the HR pathway to complete double-strand break (DSB) repair by about 50%. The HPV oncogenes hinder HR by allowing the process to begin at points in the cell cycle when the lack of a sister chromatid to serve as a homologous template prevents completion of the repair. Further, HPV E6 attenuates repair by causing RAD51 to be mislocalized away from both transient and persistent DSBs, whereas HPV E7 is only capable of impairing RAD51 localization to transient lesions. Finally, we show that the inability to robustly repair DSBs causes some of these lesions to be more persistent, a phenotype that correlates with increased integration of episomal DNA. Together, these data support our hypothesis that HPV oncogenes contribute to the genomic instability observed in HPV-associated malignancies by attenuating the repair of damaged DNA.

IMPORTANCE This study expands the understanding of HPV biology, establishing a direct role for both HPV E6 and E7 in the destabilization of the host genome by blocking the homologous repair of DSBs. To our knowledge, this is the first time that both viral oncogenes were shown to disrupt this DSB repair pathway. We show that HPV E6 and E7 allow HR to initiate at an inappropriate part of the cell cycle. The mislocalization of RAD51 away from DSBs in cells expressing HPV E6 and E7 hinders HR through a distinct mechanism. These observations have broad implications. The impairment of HR by HPV oncogenes may be targeted for treatment of HPV⁺ malignancies. Further, this attenuation of repair suggests HPV oncogenes may contribute to tumorigenesis by promoting the integration of the HPV genome, a common feature of HPV-transformed cells. Our data support this idea since HPV E6 stimulates the integration of episomes.

INTRODUCTION

High-risk genus human Alphapapillomavirus (α-HPV) infections can cause tumors throughout the anogenital tract, as well as in the oropharynx (1–3). The term “high risk” refers to the relative ability of a subset of HPV viruses to cause cancer (4). Although HPV is a very diverse family of viruses, in the context of this work “HPV” specifically refers to high-risk members of the α-HPV genus (particularly HPV16). Although rigorous screening prevents the majority of deaths from HPV-associated cervical cancers in the developed world, they remain a significant risk in the developing world (5, 6). Changing sexual behaviors in developed countries are driving the expansion of the number of HPV⁺ oropharyngeal malignancies (7, 8). The frequency of these tumors has nearly tripled in the last 30 years (7). HPV infections cause a death every 2 to 3 min (6). Although lethal, these cancers often take decades to develop following initial infection. During this time, the cellular genome becomes markedly destabilized.

Some of this genomic instability can be contributed to well understood properties of HPV oncogenes (HPV E6 and E7) that are expressed in each cell of HPV-associated tumors. HPV E6 forms a complex with E6AP, a cellular E3 ubiquitin ligase (9). This association allows HPV E6 to promote the ubiquitination and degradation of the tumor suppressor p53 (9–11). HPV E7 binds and inactivates the cell cycle regulator, pRB (12). This releases E2F driving unregulated S-phase entry (13). Because both p53 and pRB are necessary to pause cell cycle and coordinate the repair of damaged DNA, HPV E6 and E7 have previously been indirectly connected to the perturbation of DNA repair (14, 15). HPV E6 and E7 also interact with components of multiple repair pathways, suggesting they more directly contribute to the genome destabilization associated with HPV-mediated transformation by blocking the resolution of DNA damage (16).

Typically, HPV E2 acts a regulator of HPV E6 and E7 expression through repression of the viral promoter responsible for their transcription (17, 18). However, integration of the viral episomal genome into the host genome can interrupt the HPV E2 reading frame (19). Although this integration event most likely occurs at random, the ability of HPV E6 and E7 to promote proliferation offers a significant growth advantage to cells where integration disrupts HPV E2 (20). As a result, disruption of HPV E2 by integration of HPV is a frequently observed step in HPV-associated oncogenesis (19).

Integration of an episome, like the HPV genome, requires a double-strand break (DSB) in the cellular genome, as well as in the episome. These lesions are the most deleterious type of damage a cell faces, with the potential to induce large deletions as well as chromosome fusions (21). The cellular response to DSBs can be grouped based on whether or not the homologous sequence is used as a template for repair. Nonhomologous end joining occurs without a sister chromatid (22), whereas a homologous sequence is an absolute requirement for the completion of homologous recombination (HR) (23). Previous reports show that HPV E6 and E7 interact with BRCA1, an essential protein in the HR pathway (24, 25). Further, disruption of DSB repair by HPV E7 has been reported (26). Together, this led us to the hypothesis that HPV E6 and E7 block the resolution of DSBs by the HR pathway. Our analysis of DSB repair by immunofluorescence (IF) microscopy supports this hypothesis, demonstrating that resolution of DSBs is significantly impaired by HPV E6 and E7, when expressed either individually or together, although notably HPV E6 has a greater ability to prevent repair of these lesions.

Further dissection of this phenotype indicates that both HPV E6 and E7 promote the initiation of HR during portions of the cell cycle when the pathway cannot be successfully completed. In addition to this hindrance of DSB repair, HPV oncogenes also impair the localization of RAD51 to DSBs. A green fluorescent protein (GFP)-based HR assay confirms that HPV E6 and E7 impair the HR pathway. Our data show that the loss of robust DSBs repair correlates with an increased rate of episome integration. The sum effect of this repair attenuation is that DSBs are more frequently present in cells expressing HPV oncogenes and DSBs are less efficiently repaired when induced. Together, this work describes novel genome destabilizing properties of both HPV E6 and E7. The ability to disrupt DSB repair may explain the increased integration of HR-HPV episomes found in tumors and be exploited for targeted therapies against HPV⁺ tumors. Finally, our bioinformatics analysis of cervical tumors suggests that HPV oncogenes continue to alter the homologous repair pathway during tumorigenesis.

RESULTS

HPV oncogenes disrupt DSB repair.

To begin determining whether HPV oncogenes can destabilize the cellular genome by blocking DSB repair, we expressed HPV16 E6 and E7 (HPV E6 and E7) together and separately in primary human keratinocytes (HFKs). Expression was confirmed by immunoblot against established targets of HPV16 E6 and E7, p53 and pRB, respectively, as well as by qRT-PCR of 16 HPV E6 and E7 (Fig. 1A and B). The long terminal repeat (LTR) promoter driving HPV16 E6 and E7 expression in transduced cells is a conservative model of HPV-associated cancers since it results in lower levels of HPV oncogene expression than those found in HPV-transformed cell lines such as HeLa and SiHa. To examine the presence of DSBs in cells expressing HPV oncogenes, immunofluorescence microscopy was performed with antibodies against a standard DSB marker, phospho-H2AX (27). Analysis of phospho-H2AX foci in nonirradiated cells suggests that HPV16 E6 and E7 impair DSB repair. DSBs appear approximately twice as often in nonirradiated cells expressing HPV16 E6 and/or E7 (LXSN control, 8.6% foci positive; HPV16 E6, 18.0% foci positive; HPV16 E7, 19.4% foci positive; HPV16 E6 and E7, 12.1% foci positive) (Fig. 1C). Double-strand DNA breaks were then introduced in these cells by exposure to ionizing radiation (IR). Immediately after IR, phospho-H2AX foci were observed in nearly all cells in every cell line tested (Fig. 1D). Control (LXSN) cells resolved the majority of these foci in the first 8 h after IR, returning to background staining after 24 h (Fig. 1D and F). In contrast, HPV16 E6 and E7, both separately and together, significantly delayed this process from 4 to 24 h after IR (Fig. 1D and F). The attenuation of DSB repair was most notable in cells expressing HPV16 E6, with ca. 75% of these cells still containing H2AX foci 24 h after IR (Fig. 1D and F). The relative changes in the number of H2AX foci between nonirradiated cells and cells after 24 h of irradiation are also consistent with impaired DSB repair (no significant difference for LXSN or HPV16 E7, but significantly more DSBs after 24 h irradiation in HPV16 E6 and HPV16 E6,E7-expressing keratinocytes).

FIG 1.

HPV oncogenes attenuate DSB repair. (A) The relative expression of HPV16 E6, HPV16 E7 and GAPDH as determined by qRT-PCR is presented. (B) Representative immunoblot confirming HPV oncogene expression by demonstrating reduction cellular proteins established to be degraded by HPV16 E6 (p53) and HPV16 E7 (pRB). HPV E6 denotes HPV16 E6. HPV E7 denotes HPV16 E7. HPV E6,E7 denotes combined expression of HPV16 E6 and E7. (C) Percentages of p-H2AX foci in nonirradiated cells. *, Statistically significant difference from the control (LXSN). P ≤ 0.05 (Student t test). (D) After HFK cells were exposed to 4 Gy of IR, IF microscopy was used to monitor repair over a 24-h period. This graph depicts the percentage of cells with p-H2AX foci after irradiation. *, Statistically significant difference of all HPV oncogene-expressing cells from control cells (LXSN). P ≤ 0.05 (Student t test). (E) As an indication of focus size, this graph depicts the average intensity of p-H2AX foci relative to the intensity of nonirradiated vector (LXSN) that is set to 1. *, Statistically significant difference from similarly treated control cells (LXSN). P ≤ 0.05 (Student t test). (F) Representative pictures of p-H2AX foci (green) in cells 6 h after exposure to 4 Gy of IR. DAPI-stained nuclei are shown in blue. All data points in this figure represent experiments repeated at least five times. Error bars represent the standard errors of the mean.

Until a DSB is repaired, H2AX phosphorylation spreads in either direction along the chromosome (28). This can be detected by immunofluorescence as a brighter and/or larger focus. As confirmation that HPV oncogenes disrupted DSB repair, the intensity of p-H2AX staining was measured in HFK cells after IR. The expression of HPV oncogenes resulted in more intense p-H2AX foci at all times after IR (Fig. 1D and E). Moreover, the intensity of these foci was similarly increased without induction of DSBs, demonstrating that HPV E6 and E7 prevent the repair of uninduced DSBs (Fig. 1E).

HPV oncogenes attenuate homologous recombination.

Betapapillomavirus E6 proteins disrupt DSB repair by attenuating the homologous recombination (HR) repair pathway, suggesting a potential explanation for the disruption described above (29). To evaluate this possibility using the α-HPV genus, HPV16 E6 and E7 were expressed (separately or together) in DR-GFP U2OS cells. HPV oncogene expression was confirmed indirectly by confirming that the established degradation targets of E6 and E7 (p53 and pRB, respectively) were reduced in cells (data not shown). This system has limitations as the U2OS cells are not the cell types naturally infected by HPV, but they are frequently used to determine the efficiency of HR. These U2OS cells also have a single, clonal integration of the DR-GFP reporter cassette developed in Maria Jasin's lab that has become the gold standard for measuring HR efficiency (29) (Fig. 2A). Briefly, this cassette consists of a promoter facing two nonfunctional GFP genes each separated by a linker sequence. The recognition site for a rare cutting endonuclease, I-SceI, and multiple in-frame stop codons prevent expression of the first GFP gene. The second GFP gene is nonfunctional due to 5′ and 3′ truncations. Exogenous expression of I-SceI causes a break in the first GFP gene that if repaired using the second gene as a template will result in a gene conversion event that leads to a functional GFP gene. The frequency of these GFP positive cells can be readily detected by fluorescence-activated cell sorting (FACS) analysis and serves as a well-characterized measure of HR efficiency. Supporting the hypothesis that the reduction in ability to repair DSBs is due to attenuated HR, HPV16 E6 alone decreased the prevalence of GFP positive cells by 50%, while no inhibition was seen when HPV16 E7 was expressed (Fig. 2B and C). HPV16 E6 retained its ability to repress HR when expressed in combination with HPV16 E7 (Fig. 2B and C). It has previously been reported that HPV oncogenes do not alter transfection efficiency, so these differences are unlikely to be the result of decreased I-SceI (30).

FIG 2.

HPV E6 attenuates homologous recombination of a reporter cassette. (A) A representation of the DR-GFP cassette used to measure homologous recombination. (B) Main images are representative profiles of GFP⁺ cells analyzed by flow cytometry. Insets in the upper right of each these profiles are forward (x axis)- versus side (y axis)-scatter plots for the corresponding flow cytometry analysis used to select cell populations from debris. (C) The frequency of GFP+ cells representing HR events is shown relative to the vector control (LXSN) set to 100. All data points represent experiments repeated at least five times. *, Statistically significant difference from the similarly treated control (LXSN). P ≤ 0.05 (Student t test).

HR gene expression is not impaired by HPV oncogenes.

Homologous recombination (HR) is disrupted by β-HPV E6 by decreasing the expression and activation of BRCA1 and BRCA2, two proteins required for HR (31). To determine whether HPV16 E6 and E7 blocked HR through a similar mechanism, the abundance of four HR proteins (BRCA1, BRCA2, RAD51, and RPA70) were measured by immunoblotting. Similar to these results and in notable contrast to the effect of β-HPV E6 expression, HPV16 E6 and E7 either do not alter or increase the abundance of these HR proteins (Fig. 3A). Specifically, BRCA1 and RPA70 are more plentiful when HPV16 E6 is expressed separately or together with HPV16 E7, but unaltered or minimally increased by HPV16 E7 expression alone (Fig. 3A). While RAD51 and BRCA2 are overexpressed in HPV16 E6-, E7-, and E6E7-expressing cells compared to LXSN control cells (Fig. 3A).

FIG 3.

HPV oncogenes activate the HR pathway. (A) Representative immunoblot of four homology recombination proteins (BRCA1, BRCA2, RAD51, and RPA70) and actin as a loading control. (B to E) Graphs depict IF microscopy analysis after exposure to 4 Gy of ionizing radiation at the indicated times. The bars represent the percentages of focus-positive cells in panels B (BRCA1), C (BRCA2), D (RPA70), and E (RAD51). All data points represent experiments repeated at least five times. Error bars represent the standard errors of the mean. *, Statistically significant difference from the similarly treated control (LXSN). P ≤ 0.05 (Student t test).

Impairment of repair complex formation or resolution could provide an alternative mechanism to explain the observed attenuation of HR. This possibility was evaluated using IF microscopy to examine the formation of BRCA1, BRCA2, RPA70, and RAD51 foci after IR. This analysis clearly demonstrated that HR repair focus formation is not attenuated by HPV oncogene expression (Fig. 3B to D). Often these foci occur in a higher proportion of cells when HPV E6 and/or E7 is expressed (Fig. 3B to D). Cells expressing both HPV E6 and E7 were consistently more likely to have BRCA1 foci at each time point after IR, as well as in unexposed cells (Fig. 3B). There was a generally modest increase in BRCA2 foci in cells expressing HPV E7 alone or with HPV E6, but there was no significant change when HPV E6 was expressed by itself (Fig. 3C). The maximum HPV E7-induced increase in BRCA2 foci occurred 1 h after exposure to IR (43.1% versus 67.6% BRCA2-positive cells for LXSN and HPV E7, respectively). A significant and early (30 min after IR) increase in RPA70 foci was observed in HPV oncogene-expressing cells compared to control cells (Fig. 3D). The prevalence of RPA70 focus-positive cells gradually decreased after reaching this early peak but remained significantly elevated compared to controls at each time point observed. RAD51 foci occurred more often in cells expressing HPV oncogenes as well (Fig. 3D and E). The magnitude of RAD51 foci in these cells increased with time, before reaching a peak 8 h after IR exposure.

To better understand whether these foci represented activated repair complexes, as has been previously reported, the kinetic relationship between repair focus formation and resolution were analyzed more precisely. Specifically, because BRCA1 and BRCA2 activity is required for exchanging RPA70 for RAD51, the peak of RAD51 foci would be expected to occur after the other foci reach their maxima. Supporting the interpretation of focus formation as indicative of pathway activation, RPA70 foci are at their highest point in all four cell lines 30 min after irradiation (Fig. 4). The apex of BRCA1 and BRCA2 foci occurs after RPA but before RAD51 (Fig. 4). In contrast to these early events and in line with RAD51's known dependence on the other repair proteins' activity, the highest number of RAD51 foci are not seen until 4 to 8 h after the induction of DSBs (Fig. 4). Although these data are consistent with reported literature and indicative of active RPA70, BRCA1, and BRCA2, we cannot rule out the possibility that the focus formation occurs through a mechanism independent of active repair complexes. Although there is a generalized increase in HR protein abundance and HR focus formation, these two are not necessarily expected to correlate. HR proteins, like many DNA repair genes, are only visible by IF microscopy when they congregate at sites of DNA damage, making it difficult to gauge their abundance using this approach.

FIG 4.

Kinetic analysis of homologous recombination focus formation. These graphs depict IF microscopy analysis after exposure to 4 Gy of ionizing radiation at the indicated times. The lines represent the percentage foci positive cells for BRCA1 (purple), BRCA2 (black), RPA70 (blue), and RAD51 (red). Arrows of the corresponding colors are used to denote the maximum observed percentages of foci for these repair proteins. Four cell lines (LXSN, HPV16 E6, HPV16 E7, and HPV16 E6,E7) are shown as indicated by the titles at for each of the four graphs. All data points represent experiments repeated at least five times. Error bars represent the standard errors of the mean.

HPV-associated cervical carcinomas upregulate the HR pathway.

To determine whether the HR pathway was also upregulated in HPV-associated malignancies, a total of 86 HPV⁺ cervical tumor tissue and 58 healthy control samples from the NCBI GEO database were analyzed. The data stem from four data sets comparing mRNA expression of HPV⁺ cervical cancer patients to healthy control subjects. Across all data sets, the gene ontology enrichment analysis tools GOrilla and GSEA identified a strong upregulation of the cellular response to DNA damage in general (P < 10⁻⁹) and an upregulation of the HR pathway specifically (P < 10⁻⁴) as shown in Fig. 5A. A group of genes required at different stages of HR was then selected based on a literature review (32) (KEGG PATHWAY; homologous recombination, Homo sapiens [human]), including the genes for RPA70, BRCA1, BRCA2, and RAD51 discussed extensively here. The comparison of normalized relative mRNA expression levels between control tissue and HPV⁺ cervical tumor tissue and across the four data sets displays a clear mRNA upregulation pattern in cervical tumor tissue (Fig. 5B). The HR genes selected for analysis are involved in the formation of the MRN complex (MRE11A, RAD50, and Nbs1), single-strand protection (RPA70), filament formation and strand invasion (BRCA1/2, RAD51/54, and PALB2), and resolution of the Holiday junction (GEN1). When the four data sets are combined, analysis of the relative mRNA expression for RPA70, BRCA1, BRCA2, and RAD51 shows a significant increase (P < 10⁻⁷ or less, Fig. 5C) in cervical cancer tissue. This is in agreement with immunoblot results described in Fig. 3.

FIG 5.

HR gene mRNA expression is upregulated in cervical cancer patients. (A) Visualized output of the gene expression analysis tools GOrilla and GSEA. The cellular response to DNA damage is notably significantly upregulated (P ≤ 10⁻⁹) indicated by a red box. DNA damage repair and DSB repair are both highly significantly upregulated (P ≤ 10⁻⁸ and P ≤ 10⁻⁷, respectively), as indicated by orange boxes. The HR pathway is also significantly upregulated (P ≤ 10⁻⁴) and is indicated by a light orange box. (B) Heatmap of the relative expression of relevant HR genes. The heatmap is organized by control (left) and cervical cancer (right) samples, and each data set is represented by an individual column. These data sets are matched such that the first column of control tissues comes from the same data set as the first column of the cervical cancer tissues. Warmer, more yellow/red colors indicate a higher relative expression, whereas cooler colors represent lower relative expression. (C) The relative expressions of RPA70, BRCA1, BRCA2, and RAD51 from the combined data sets are displayed as boxplots. Blue boxplots represent combined control groups, while red boxplots represent combined cervical cancer groups. The upregulation is highly significant in all four cases, with all P < 10⁻⁷.

HPV oncogenes abrogate homologous recombination by allowing the pathway to initiate during G₁.

A plausible way for these viral oncogenes to disrupt repair of DSBs without preventing repair focus formation is to cause repair foci to form during the wrong part of the cell cycle. The timing of repair focus formation with respect to the cell cycle is critical because homologous recombination is dependent on the availability of homologous sequence. As a result, it cannot be completed outside the S and G₂ phases of the cell cycle (33). The IF microscopy data described above demonstrate that 80 to 90% of cells expressing HPV E6 and/or E7 contain RPA foci 30 min after exposure to IR (Fig. 3D). Because neither HPV oncogene-induced proliferation nor damage-induced cell cycle arrest are likely to result in just 10 to 20% of cells in G₁ only 30 min after IR, HPV E6 and E7 may allow RPA foci to form when sister chromosomes are not available. To evaluate this possibility, IF microscopy and flow cytometry were used to determine the point of the cell cycle when RPA foci were forming.

For IF microscopy, cyclin A and cyclin E were used as markers of cells in the S/G₂ and G₁ phases, respectively. Cyclin A served as a marker for times during the cell cycle when sister chromosomes would be present to act as the templates for repair (34, 35). To ensure that the increase of RPA⁺ cells was not simply the result of HPV oncogene-induced increases in DSBs, only the relative frequency of cells was analyzed. Validating the use of cyclin A as a marker for appropriate times in the cell cycle for HR, nearly all RPA⁺ control cells also expressed cyclin A (Fig. 6A and B). In contrast, RPA⁺ cyclin A⁻ cells were readily observed in cells when HPV E6 and E7 were expressed either individually or jointly (Fig. 6A and B). Not only did aberrant RPA foci form in these cells following IR, but they were also present in nonirradiated cells (Fig. 6A and B). The observed frequency of RPA⁺ cyclin A⁻ cells in HPV E6 and E7 cells was also in line with the attenuation of DSB repair described in Fig. 1.

FIG 6.

IF microscopy analysis show HPV oncogenes allow RPA focus formation during the G₁ phase. (A) Representative images of cells with nuclei stained with DAPI (blue), cyclin A (green), and RPA70 (red), as well as an image merging these colors. White arrows point out cells that are cyclin A negative but RPA70 positive. (B) Quantification of RPA70⁺ cyclin A⁻ cells. (C) Representative images of cells with nuclei stained with DAPI (blue), cyclin E (green), and RPA70 (red), as well as an image merging these colors. (D) Quantification of RPA70⁺ cyclin E⁺ cells at the indicated time points after exposure to 4 Gy of IR. All data points represent experiments repeated at least five times. Error bars represent the standards error of the means. *, Statistically significant difference from the similarly treated control (LXSN). P ≤ 0.05 (Student t test). ***, Statistically significant difference from the similarly treated control (LXSN). P ≤ 0.001 (Student t test).

To confirm that HPV oncogene expression caused HR to initiate earlier in the cell cycle than appropriate, the propensity of cells to initiate HR during G₁ was determined by immunofluorescence microscopy. Cyclin E was used as a marker of cells in G₁, whereas RPA70 staining was used as an indicator of cells committed to repair DSBs through the HR pathway. For this analysis, cyclin E status was only considered in cells that had RPA foci. Cyclin E staining was not observed in cells with RPA foci in nonirradiated control cells (Fig. 6C and D). However, cyclin E staining occurred in 5 to 10% of RPA⁺ cells expressing HPV E6 and/or E7 without irradiation (Fig. 6C and D). Exposure to radiation increased the frequency of cyclin E⁺ RPA⁺ cells in all cell lines. Although these events occurred in only a small minority (<10%) of control cells, HPV E6 and E7 (individually and together) significantly increased the frequency of cyclin E⁺ RPA⁺ cells 1 h after IR (Fig. 6C and D). Similar observations were seen 10 min after IR, but the differences among control cells and cells expressing HPV oncogenes only approached statistical significance after 1 h after IR.

HPV oncogene ability to allow RPA70 to occur in G₁ phase was further confirmed by flow cytometric analysis. After 1 h of IR irradiation, pulse-labeling of cells with bromodeoxyuridine (BrdU; an analog of thymidine) was performed for 30 min. Analysis of BrdU incorporation was combined with the simultaneous analysis of RPA70 levels. Flow cytometric analysis show that the percentages of RPA70-positive cells are greater in E6- and E6E7-expressing HFKs compared to LXSN control cells, even without irradiation, which corresponded with the immunoblot results discussed above (Fig. 7A). Approximately 40 to 60% of RPA70-positive cells are in G₁ phase in HPV16 E6-expressing cells compared to the LXSN control (5 to 10%), with or without irradiation. E7-expressing cells showed an ∼2-fold increase in percent RPA70 in G₁ phase (Fig. 7B and C) than did the LXSN control. Cells expressing both E6 and E7 also showed a significant increase in %RPA70 in G₁ phase following IR. The results of the flow cytometry and IF microscopy demonstrate that HPV16 E6 and E7 increase the likelihood that HR initiates during G₁, but there are differences between the two assays. Particularly, HPV16 E6 seems to be the dominant driver of the phenotype when measured by flow cytometry, whereas the opposite is true when a microscopy-based approach is used. This could suggest unaccounted nuances in the approaches, but both assays demonstrate the ability of HPV oncogenes to allow HR to initiate during G₁.

FIG 7.

Flow cytometric analysis showing HPV oncogenes allow RPA70 to occur during G₁ phase. (A) Percentages of RPA70-positive HFK cells analyzed by flow cytometry. (B) Representative profiles of HFK cells stained with anti-RPA70, anti-BrdU, and DAPI and analyzed by flow cytometry. The inset in this image shows representative gatings for side scatter (y axis) and forward scatter (x axis) to select cell populations from debris. (C) Quantification of percentage of RPA70 in G₁ phase in nonirradiated cells and 1 h after exposure to 4 Gy of IR. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001 (statistically significant differences from the similarly treated LXSN control cells as determined by Student t test). ns, nonsignificant difference. Error bars represent the standard errors of the mean. Quantification was based on data observed from three independent experiments.

HPV oncogenes cause RAD51 to mislocalize from double-strand DNA breaks.

In addition to beginning during the correct part of the cell cycle, repair foci must be properly localized to sites of damage. The sustained prevalence of RAD51 foci after IR suggests that RAD51 may form unproductive repair complexes. To determine whether HPV oncogenes have overlapping abilities to disrupt HR, RAD51 foci localized to DNA lesions were observed. Phospho-H2AX foci were used as markers for DSBs. The localization of RAD51 to DSBs was monitored by IF microscopy following IR. After this radiological insult, H2AX foci were readily visible in all cell lines tested. Although RAD51 could be seen localized with p-H2AX in cells regardless of the presence or absence of HPV oncogenes, colocalization studies also suggested that HPV oncogenes may cause repair complexes to be localized away from DSBs. The ability of HPV oncogenes to prevent RAD51 localization to DSBs was seen in irradiated HFKs expressing HPV16 E6, E7, and E6,E7. A total of 50 to 60% of RAD51 appear to be mislocalized away from p-H2AX in cells expressing E6 and/or E7 compared to ca. 10 to 20% in LXSN controls (Fig. 8). To complement these colocalization studies, we took advantage of the DR-GFP U2OS cells that contain a single I-SceI recognition site. Exogenously expressed I-SceI cleaves its recognition site, producing a single DSB that it will continue recut unless the recognition sequence is mutated during repair. This persistent DSB allows integration of repair protein recruitment independent of the kinetic differences resulting from HPV16 oncogene expression (Fig. 9A).

FIG 8.

HPV16 oncogenes cause mislocalization of RAD51 away from DSBs in primary keratinocytes. (A) Representative images of HFK cells with nuclei stained with DAPI (blue), RAD51 (green), and pH2AX (red). White arrows denotes mislocalization of RAD51 away from pH2AX. (B) Quantification of the percentages of RAD51 focus mislocalization away from pH2AX at the indicated time points after exposure to 4 Gy of IR. *, P ≤ 0.05; **, P ≤ 0.01 (statistically significant differences from the similarly treated LXSN control cells as determined by Student t test). Error bars represent standard errors of the mean. Quantification was based on data observed from ≥15 nuclei from three independent experiments.

FIG 9.

HPV E6 hinders localization of RAD51 to DSBs in transformed U2OS cells. (A) Schematic of localization assay. Exogenous expression of I-SceI in U2OS cells with one integrated copy of the I-SceI recognition site produces a single but a persistent DSB. The lesion is readily visible by IF microscopy as an enlarged p-H2AX focus (green). Recruitment of repair proteins to this lesion can similarly be visualized (red) and analyzed for localization to the DSB by merging these images (yellow). (B and C) Representative images of this analysis for RPA70 and RAD51 in control (LXSN) and HPV16 E6-expressing U2OS cells. (D) Chart depicting the frequency of colocalization of the indicated repair proteins with p-H2AX foci. All data points represent experiments repeated at least five times. Error bars represent the standard errors of the mean. *, Statistically significant difference from the similarly treated control (LXSN). P ≤ 0.05 (Student t test).

Indicative of a persistent DSB, I-SceI expression induced large phospho-H2AX foci that were readily visible by IF microscopy in all of the cell lines observed (Fig. 9B to D). In control cells, the four HR proteins of interest (BRCA1, BRCA2, RPA70, and RAD51) were localized to large phospho-H2AX foci (Fig. 9B to D). Similarly, the majority of RPA70, BRCA1, and BRCA2 remained colocalized with phospho-H2AX foci in the presence of HVP E6 and E7 (Fig. 9B to D). In contrast, RAD51 foci were found at these DSB about half as often in HPV16 E6-expressing cells compared to controls (Fig. 9B to D). HPV E7 also caused mislocalization of RAD51, but in a significantly less robust manner. This phenotype persists when HPV16 E6 and E7 expression is combined, but the difference does not reach statistical significance (Fig. 9B to D).

HPV E6 increases the integration of episomal DNA.

The data presented above show that while both HPV16 E6 and E7 prevent DSB repair, HPV16 E6 has a more pronounced ability to block the HR pathway. As a result, the impact of disrupted DSB repair by HPV16 E6 was investigated further. To understand the relative ability of different HPV E6s to disrupt DSB repair, phospho-H2AX foci were observed in HFK cells expressing three E6 proteins from betapapillomaviruses (HPV5, -8, and -38 E6), as well as the high-risk HPV E6 examined in above (HPV16 E6). Confirming previous reports, each HPV E6 examined impaired DSB repair after exposure to IR (Fig. 10A). Only HPV16 E6 was able to increase the frequency of phospho-H2AX foci in nonirradiated cells (0-h time point), as well as 24 h after IR.

FIG 10.

Integration of episomal DNA. (A) Chart depicting the frequency of p-H2AX positive cells before and after exposure to ionizing radiation. (B) Graph depicting the relative integration of episomal DNA in HT1080 cells stably expressing either an empty vector control (LXSN), one of three β-HPV E6s (HPV5, -8, and -38 E6), or HPV16 E6. Integration rates are set relative to LXSN at 100. (C) Chart depicting the relative integration rates after exposure to a gradient of hydrogen peroxide concentrations. Integration is set relative to untreated control within each cell line. Untreated cells are set at 100. For each sample, the transition from white to black indicates increasing exposures to hydrogen peroxide. White bars represent untreated samples. All others were exposed to 25 μM H₂O₂ for 15 min (light gray), 30 min (dark gray), or 60 min (black). All data points represent experiments repeated at least three times. Error bars represent the standard errors of the mean. *, Statistically significant difference from the similarly treated control (LXSN). P ≤ 0.05 (Student t test). A “!” symbol denotes a statistically significant difference from all other similarly treated cells. P ≤ 0.001 (Student t test). A “+” symbol denotes a statistically significant increase in integration compared to untreated samples from the same cell line.

Integration of the HPV episomal genome into the host genome is often a key event in HPV-associated tumorigenesis (19). Since the integration of episome DNA requires two simultaneously occurring DSBs (one in the episome and one in the genome), it is likely to be increased by the abundance of DSBs in untreated HPV16 E6-expressing cells. The relative rate of episomal integration was tested in cells expressing HPV16 E6 by transfection of a vector containing a selectable marker but lacking an origin of replication. In order for cells to survive selection, the resistance vector must be integrated into the host genome, or else it will be diluted and lost with passaging. As a result, after selection, integration events were quantified by crystal violet staining. In untreated cells, expression of HPV16 E6 significantly increased the integration of episomes over control integration (Fig. 10B). In contrast, HPV5, -8, and -38 E6 did not increase the integration of episomes. Although the expression of HPV5, -8, and -38 E6 cannot be confirmed by immunoblotting, their expression in these cells has been previously confirmed by qPCR (30).

The ability to increase the number of DSBs present in untreated cells correlates with the ability to induce episome integration. This is likely because the breaks provide a greater opportunity for episomal DNA to integrate into chromosomes. A corollary of this hypothesis is that HPV E6s that cause DSBs to be available following damage should also elevate the frequency of integration events that are induced by DNA-damaging agents (36–38). To test this possibility, integration was measured after the induction of DSBs by hydrogen peroxide. First, concentrations of hydrogen peroxide low enough to avoid toxic responses from cells were defined (data not shown). After normalizing for differences in the uninduced integration, these concentrations were also low enough not to increase the frequency of integration in control cells. Nevertheless, hydrogen peroxide at these concentrations were capable of inducing a >12-fold increase in integration in HPV16 E6-expressing cells (Fig. 10C). Supporting the hypothesis that the changes in integration frequency are the result of greater access to DSBs, the β-HPV E6s (HPV5, -8, and -38 E6) that restricted the repair of DSBs also enhanced integration after the induction of DSBs (Fig. 10A and C) (31).

DISCUSSION

Over the multidecade course of HPV-induced tumorigenesis, cellular genome integrity has been substantially impaired. Although HPV E7 has been shown to delay DSB repair (26), the data presented here show that both HPV oncogenes contribute to genome instability (Fig. 1). HPV E6 and E7 attenuate DSB repair through multiple mechanisms that target the HR pathway (Fig. 2 to 9). Unlike E6 proteins from the betapapillomaviruses (31), HPV E6 and E7 from high-risk alphapapillomaviruses maintain or increase the abundance of HR proteins and enhances their ability to form repair foci (Fig. 3). Instead, HPV16 E6 and E7 allow the initiation of HR during portions of the cell cycle when there is no homologous template available to complete repair (Fig. 6 and 7). Despite the fact that both HPV E6 and E7 allow HR to initiation during G₁, only HPV E6 prevented homologous recombination as measured by the DR-GFP assay. In addition to permitting HR during G₁, HPV E6 and E7 also mislocalize RAD51 away from sites of damage (Fig. 8 and 9). The DR-GFP reporter cassette provides a homologous template for repair that is present during all phases of the cell cycle. As a result, HR repair of the DR-GFP cassette can occur in the G₁ portion of the cell cycle. Thus, HR, as measured by the DR-GFP cassette, would not be affected by the ability of HPV E6 and E7 to allow HR to begin outside the S and G₂ phases (Fig. 3). However, the mislocalization of RAD51 by HPV E6 and E7 attenuates repair independent of the availability of homologous sequence (Fig. 8 and 9). The inability to fully utilize HR causes DSBs to be more prevalent in nonirradiated cells and more persistent following IR (Fig. 1). HPV E6 is the dominant inhibitor of DSB repair and increases the frequency of extrachromosomal DNA integration (Fig. 10).

DNA repair pathway choice.

Competition between 53BP1 and BRCA1 regulates resection in response to DSBs (39, 40). During the S and G₂ phases, BRCA1 preferentially gains access to the break inducing extensive resection of one strand of DNA near the break. Although this provides a template for HR, it also impairs repair by nonhomologous end joining (NHEJ). In contrast, during G₁, 53BP1 generally promotes the smaller scale resection needed to facilitate NHEJ. Inhibition of BRCA1 or 53BP1 makes it more likely the other protein gains access to the lesion. As a result, defects in HR can lead to increases in NHEJ and vice versa. HPV oncogenes increase BRCA1 activation, as measured by damage-induced focus formation, suggesting that BRCA1 is more frequently promoting the large resection events that inhibit NHEJ (Fig. 3B). Our RPA70 focus formation also support this scenario (Fig. 3D). Therefore, we speculate that HPV E6 and E7 attenuate both the NHEJ and the HR DSB repair pathways.

Implications for the HPV life cycle.

Productive HPV replication requires HR pathway proteins and an ATM-dependent damage response (16, 41–43). This includes the increased abundance of two HR proteins, RAD51 and BRCA1, as well as the association of RAD51, BRCA1, and components of the RPA complex (RPA32, RPA70, and RPA14) and NBS1 with the viral genomes. We demonstrate that RAD51 and BRCA2 levels are increased by both HPV16 oncogenes, whereas only HPV16 E6 increased the levels of BRCA1 and RPA70. Further, our bioinformatics analysis demonstrates that these genes are overexpressed in cervical cancers. Together, these data suggest that the changes occur at the transcriptional level and increased HR protein expression in HPV-associated tumors is driven by HPV oncogenes. Because our data were obtained in the absence of a viral origin of replication, there is no replication site for HPV oncogenes to take these repair factors. It is important to note that the mislocalization of RAD51 that we observe is not complete and may represent a fruitless attempt to bring the repair protein to a nonexistent of viral replication origin. This would attenuate repair regardless of the presence or absence of the viral genome, however, since repair proteins would be removed from sites of genomic damage. The relocalization of RAD51 away from pH2AX does not occur when the whole genome is present, suggesting either that the whole virus induces aberrant phosphorylation of the histone or that there is a dominant activity of another HPV protein that prevents improper RAD51 localization by HPV E6 and E7.

Although HPV16 E6 hindered RAD51 localization to DSBs regardless of their origin, HPV16 E7 was only able to impede RAD51 recruitment to DSBs that could readily be resolved. When persistent DSBs were induced by continual I-SceI cleavage of its recognition site, HPV16 E7 was no longer able to attenuate RAD51's localization. Further, the relocalization of RAD51 is much more significant in transient DSBs compared to persistent DSBs. One explanation for these results is that the transient lesions occurred in primary keratinocytes, whereas the other experiments were conducted in transformed cells. However, we believe the difference in the kinetics of the DSB offer insight into HPV16 E7's role in mislocalizing RAD51. Specifically, our data are consistent with a scenario wherein HPV16 E7 delays recruitment of RAD51 to breaks long enough that a normal lesion is repaired by other mechanism, but the viral protein cannot delay RAD51 long enough to prevent it from localizing to the continual DSB induced by I-SceI cleavage.

HPV oncogenes promote the initiation of HR but hinder completion of the pathway, suspending cells in a state where HR proteins are more abundant (Fig. 3 to 7). Because HR proteins are required for HPV replication (41–43), we postulate that this creates an environment where HR proteins are readily available to support replication. Alternatively, preventing efficient RAD51 filament formation could be independently important for the viral life cycle. Although these explanations are not mutually exclusive, the fact that the HR pathway is abrogated by both HPV E6 and E7 suggests an evolutionary pressure to prevent its completion.

Modeling HPV episome integration.

The presence of persistent unrepaired DSBs can be very deleterious to genomic fidelity since it provides the opportunity for chromosomal rearrangements, fusions, and deletions (21). As a result, extrachromosomal DNA often will be inserted into these lesions as a stopgap method of preventing worse damage (36). Because the HPV genome typically exists in cells as an episome, it is a source of extrachromosomal DNA that can be inserted into unrepaired breaks. The integration of the HPV genome into the host genome is a common step in HPV-associated oncogenesis because it can disrupt the HPV E2 gene, leading to deregulation of HPV oncogene expression (19). Although there is little evidence of a hot spot for the integration of the HPV genome at the HPV E2 gene, such an event would be selected for since it provides an obvious growth advantage (17, 18, 20, 44). HPV E6 increases the frequency of episome integration more than five times that of controls (Fig. 10A). HPV E6-induced integration is further enhanced by exogenous sources of DNA damage, similar to the damage produced by cigarette smoke and medical procedures such as X-rays (Fig. 10B and C). The integration of HPV genomes in clinical samples mirrors our findings (19, 45). Our model system eliminates the possibility that differences in the ability to promote proliferation among the various HPV E6 and E7 proteins results in a selection bias. This suggests the viral oncogene may also play a more direct role in the observed integration of HPV genomes and that environmental toxins could directly contribute to the risk of HPV-associated oncogenesis. If, as we suspect, HPV oncogenes attenuate NHEJ, as well as HR, our results would indicate that the primary limitation for episome integration is the availability of a DSB. This is not particularly surprising given the number of alternative DSB repair mechanisms that come into play more frequently when the more canonical systems are impaired (46). An important caveat to the interpretation of these experiments is that integration in our system takes place over a relatively short period of time, whereas the integration of the HPV genome occurs over a much longer time scale. These temporal differences must be considered when interpreting these data.

Potential for therapeutic intervention.

The translational aspects of this work are particularly exciting. HPV-associated tumors are dependent on continued viral oncogene expression (47–49). Our data demonstrate that these oncogenes disrupt the repair of DSBs by impairing the homologous recombination pathway. Because the increased expression of HR genes is similarly seen in HPV-associated cervical cancers, these malignancies and other HPV-associated tumors likely also lack the ability to efficiently repair DSBs through this pathway. Such a deficiency could be targeted by existing PARP inhibitors, to provide an alternative to existing therapeutic approaches (50). Existing DNA damage repair inhibitors could further enhance the synthetic lethality of PARP inhibitors (51, 52). This would be particularly beneficial in tumors that have developed resistance to cisplatin, a drug commonly used in the treatment of cervical cancer. In support of this approach, the use of this class of chemotherapeutics is an emerging direction in the treatment of cervical cancer (53), including Olaparib in combination with Carboplatin in an ongoing clinical trial (Olaparib in combination with carboplatin for refractory or recurrent women's cancers [ClinicalTrials.gov]). Finally, the work described here represents a significant advance in the understanding of the biology of HPV oncogenes, as well as their etiological contributions to tumorigenesis.

MATERIALS AND METHODS

Tissue culture.

HT1080 cells are derived from a fibrosarcoma, and U2OS cells are derived from an osteosarcoma. U2OS DR-GFP cells contain a single integrated copy of the DR-GFP cassette (29). The primary HFKs were generated from neonatal human foreskins and grown in EpiLife medium supplemented with calcium chloride (60 μM), human keratinocyte growth supplement, and penicillin-streptomycin. The results from HFK cells presented in the manuscript were repeated in HFKs derived from multiple donors. These cells underwent viral transduction and selection to achieve the desired expression of HPV oncogenes. The expression of HPV16 E6 and E7 was confirmed by immunoblot to well-characterized cellular targets of these proteins, p53 and pRB, respectively, and by quantitative reverse transcription-PCR (qRT-PCR) of 16 E6 and E7 (9–12). Transient transfections of U2OS and HT1080 cells were performed with Xtreme Gene (Roche) according to the manufacturer's instructions.

Immunofluorescence microscopy.

HFKs and U2OS DR-GFP cells (transduced with LXSN, HPV16 E6, HPV16 E7, and HPV16 E6,E7) were grown on coverslips overnight. After exposure to 4 Gy of ionizing radiation (IR) or transfection with I-SceI expression vector (Addgene, catalog no. 26477) (54), cells were washed and fixed with 4% paraformaldehyde at the indicated times. They were then permeabilized with 0.5% Triton-X in phosphate-buffered saline (PBS). A solution of 3% bovine serum albumin and 0.1% Tween 20 in PBS was used as a blocking buffer. After blocking, the coverslips were stained with the primary antibodies indicated in the text and appropriate Alexa Fluor secondary antibodies (Molecular Probes). Primary antibodies were diluted in blocking buffer and incubated overnight at 4°C. The following day, the coverslips were washed with PBS and then incubated for 1 h with secondary antibodies and Hoechst 33342 diluted in blocking buffer. The cells were then visualized using a DeltaVision Elite microscope.

The intensity of IF microscopy is a measure of the number of photons detected from one omission channel (55). Intensity was calculated based on the number of photons at a specific location. The intensity of immunofluorescence staining is commonly used to determine the local concentration of fluorophores (secondary antibodies) found at any position in the image. In this way, it is the equivalent to measuring densitometry to approximate the protein concentration from an immunoblot.

To study the localization of RAD51 to pH2AX, HFK cells (transduced with LXSN, E6, and/or E7) were fixed and permeabilized at the indicated times after IR. After blocking, the cells were costained with antibodies against pH2AX and RAD51 at 4°C for overnight, followed by three washes with 1× PBST (1× PBS plus 0.1% Tween 20) for 5 min each. After incubation with appropriate Alexa Fluor-conjugated secondary antibodies for 1 h at room temperature, the cells were washed and mounted using ProLong Diamond antifade mountant with DAPI (4′,6′-diamidino-2-phenylindole). Images were acquired using a DeltaVision Elite microscope and analyzed by using ImageJ. Manual counting was done, and the percentage of RAD51 mislocalization was quantified by dividing the number of RAD51 away from pH2AX by the total number of RAD51 foci present in a cell. Quantification was based on data observed from ≥15 nuclei from at least three independent experiments.

Antibodies.

Primary antibodies against BRCA1 (D-9; Santa Cruz Biotechnology), BRCA2 (Cell Signaling Technology), RAD51 (Santa Cruz Biotechnology/Cosmo Bio Co., Ltd.), p-H2AX (Millipore), RPA70 (EMD Bioscience), cyclin A (Abcam), fluorescein isothiocyanate-conjugated cyclin E (HE12; Santa Cruz Biotechnology), Alexa Fluor 488-conjugated anti-RPA70 (Abcam), and PerCP-Cy5.5-conjugated anti-BrdU (BD Biosciences) was used alone or in combination with appropriate Alexa Fluor- or horseradish peroxidase (HRP)-conjugated secondary antibodies (Cell Signaling Technology or Life Technologies).

Automated immunofluorescence microscopy.

HFK cells (transduced with LXSN, HPV16 E6, HPV16 E7, and HPV16 E6, E7) were seeded on Greinger glass-bottom 96-well dishes (Sigma) and grown overnight. After exposure to 4 Gy of IR, the cells were fixed and stained as described above. After completion of this staining process, images of the cells were obtained using a Cellomics ArrayScan VTI with a 20× × 0.40 NA objective. Image analysis (% positive and staining intensity) was performed with a Cellomics ArrayScan HCS reader. Hoechst-stained nuclei were used to delineate cells. After background correction, nuclear foci were quantified so that approximately 400 cells were analyzed per well. Each data point was determined in three wells for each of at least three independent experiments. For BRCA1 and BRCA2, cells were considered positive if they had more than 25 foci. For RAD51 and RPA70, the cutoff to be considered positive was 10 foci in a cell.

Immunoblotting.

As previously described (30, 31, 56, 57), protein extracts from whole cells were prepared by mechanically detaching cells and resuspending them in WE16th lysis buffer. Lysates were then sonicated and clarified by centrifugation. Equal amounts of protein lysates (15 to 30 μg) were electrophoresed in SDS-polyacrylamide gels and transferred to Immobilon-P membranes. These membranes were then exposed to primary antibodies against target proteins listed in the text. Next, membranes were incubated in a dilute solution of corresponding HRP-conjugated secondary antibody, and proteins were visualized using an HRP substrate.

DR-GFP assay.

The development of the DR-GFP assay was performed as originally described (29). We determined the relative GFP⁺ populations 24 h after transfection with an I-SceI expression vector. The resulting data were quantified using a BD Canto 1 instrument (BD Biosciences) and analyzed using FlowJo software (FlowJo Enterprise).

Episome integration assay.

HT1080 cells (transduced with LXSN, HPV5 E6, HPV8 E6, HPV38 E6, and HPV16 E6) were grown on a six-well plate and transfected with a selection vector (pBabe Puro). After 24 h, the cells were grown in media containing puromycin until an untransfected control was killed. Growth in selection medium was continued until colonies were visible by crystal violet staining. The relative number of resistant colonies (representing episome integration events) was then determined by using our previously described spectrophotometer-based measure of crystal violet eluted with acetic acid (30).

I-SceI colocalization assay.

U2OS DR-GFP cells (transduced with LXSN, HPV16 E6, HPV16 E7, and HPV16 E6,E7) were grown on glass coverslips and transfected with an I-SceI expression vector. At 24 h after transfection, the cells were fixed, permeabilized, and stained with antibodies to p-H2AX and the indicated protein. Cells with single large p-H2AX foci were selected by visual inspection and analyzed for colocalization of the indicated repair protein with the p-H2AX focus.

To ensure that the repair complexes analyzed formed similarly across experiments, the following selection criteria were used. Only cells with large foci were analyzed because the sizes of these foci reflect the spreading of phosphorylation along the chromosome, indicating the persistent break induced by I-SceI expression (28). Further, only cells with a single focus were analyzed since the DR-GFP cassette containing the recognitions site for I-SceI is clonally integrated once in the U2OS genome.

RPA70 in cell cycle phases by flow cytometric analysis.

HFK cells (transduced with LXSN, HPV16 E6, HPV16 E7, and HPV16 E6,E7) were grown overnight. Cells were then irradiated with 4 Gy of IR, followed by BrdU pulse-labeling for 30 min after 1 h of IR irradiation. The cells were then fixed using BD Cytofix/Cytoperm fixation and permeabilization solution (DNA damage and cell proliferation kit; BD Biosciences) and stained with PerCP-Cy5.5-conjugated anti-BrdU antibody, Alexa Fluor 488-conjugated anti-RPA70 antibody, and DAPI. The cells were then assessed by using an SP6800 spectral analyzer (Sony), and the data were analyzed using FlowJo software.

RNA extraction and qRT-PCR analysis.

Total RNA was isolated from HFK cells (transduced with LXSN, E6, and/or E7) by using a PureLink RNA minikit (Thermo Fisher Scientific), with DNase I treatment, according to the manufacturer's instructions. Single-stranded cDNA was synthesized from 1 μg of total RNA using SuperScript VILO cDNA synthesis kit (Thermo Fisher Scientific). To evaluate the expression of HPV16 E6, HPV16 E7, and GAPDH genes, qRT-PCR was performed in triplicates using 200 ng of cDNA as the template, the gene-specific forward and reverse primers (0.3 μM each), and Power SYBR green Supermix (Thermo Fisher Scientific) in separate 20-μl reaction volumes in an Applied Biosystems StepOnePlus real-time PCR system (Thermo Fisher Scientific).

Bioinformatics analysis of gene expression in cervical cancers.

The data sets for gene expression bioinformatics were selected based on the availability of a comparison between control patients (having normal cervical epithelia) and HPV⁺ cervical cancer patients within one data set. Four data sets were selected (58–61), and the raw data, as well as available fold changes and significance values, were loaded into the statistical analysis software R using the GEO2R webtool. Altogether, 86 cervical cancer cases were analyzed and compared to 58 controls. Using an established approach (62), a ranked file containing the gene identifier and the ranking metric were generated for each data set, ranking the genes from the most significantly upregulated (top) to the most significantly downregulated (bottom) with insignificant regulation changes in the middle. The gene ontology enrichment analysis tools GOrilla and GSEA were used to identify upregulated DNA damage response pathways using a threshold of P < 10⁻³ within each data set.

For the analysis of the HR pathway on the gene level, HR-relevant genes were identified based on the literature (6, 7, 32) (see also the KEGG PATHWAY), and raw data were extracted from all four data sets. The fold change data were feature scaled to allow intradata set comparison. A heatmap displaying the relative fold change patterns across all data sets was generated using the R “image.plot” function from the Fields package (63). Finally, feature scaled mRNA expression data for all four genes (RPA70, BRCA1, BRCA2, and RAD51) discussed here in detail were combined from all data sets. Boxplots comparing control and cervical cancer groups were generated, and the significance of the fold change was calculated by using the Student t test.