Human Papillomavirus Type 16 E6* Induces Oxidative Stress and DNA Damage

ABSTRACT

High-risk types of human papillomavirus (HPV) are the causative agents of virtually all cases of cervical cancer and a significant proportion of other anogenital cancers, as well as both oral and pharyngeal cancers. The high-risk types encode two viral oncogenes, E6 and E7, which work together to initiate cell transformation. Multiple steps involving the activities and interactions of both viral and cellular proteins are involved in the progression from HPV infection to cell transformation to cancer. The E6 oncoprotein is expressed as several isoforms: a full-length variant referred to as E6 and a few shorter isoforms collectively referred to as E6*. In this study, we found that expression of E6* increased the level of reactive oxygen species (ROS) in both HPV-positive and HPV-negative cells. This increased oxidative stress led to higher levels of DNA damage, as assessed by the comet assay, quantification of 8-oxoguanine, and poly(ADP-ribose) polymerase 1. The observed increase in ROS may be due to a decrease in cellular antioxidant activity, as we found that E6* expression also led to decreased expression of superoxide dismutase isoform 2 and glutathione peroxidase. These studies indicate that E6* may play an important role in virus-induced mutagenesis by increasing oxidative stress and DNA damage.

IMPORTANCE Our findings demonstrate for the first time that an HPV gene product, E6*, can increase ROS levels in host cells. This ability may play a significant role both in the viral life cycle and in cancer development, because an increase in oxidative DNA damage may both facilitate HPV genome amplification and increase the probability of HPV16 DNA integration. Integration, in turn, is thought to be an important step in HPV-mediated carcinogenesis.

INTRODUCTION

High-risk (HR) types of human papillomavirus (HPV) are the causative agents of virtually all cases of cervical cancer as well as a significant percentage of other anogenital and oropharyngeal cancers. In fact, current estimates indicate that HPV infection may be associated with as many as 93% of anal cancers, 63% of oropharyngeal cancers, 40% of penile cancers, 64% of vaginal cancers, and 51% of vulvar cancers (1). HPV infection accounted for approximately 26,700 cases of HPV-related cancers in the United States (2, 3), and it is estimated that 5.2% of all cancers worldwide can be attributed to HPV infection (4). While the incidence of cervical cancer has declined in the last 30 years due to Pap smear screening, the incidence rates of anal, oropharyngeal, and vulvar cancers steadily increased within the same period (1). These numbers underscore the need for ongoing research into the mechanisms behind HPV-related carcinogenesis.

The high-risk types of HPV encode two viral oncogenes, E6 and E7, that together serve as the major initiators of cell transformation (5). Multiple steps are involved in the progression from HPV infection to cellular transformation to cancer. Virus-related factors influencing this progression include virus persistence, viral load, and the reprogramming of target cell function by HPV early genes to favor virus production. In rare cases, infection plus subsequent events can lead to HPV genome integration. The significance of viral genome integration in HPV-mediated carcinogenesis is illustrated by the fact that most cases of HPV-mediated cervical cancer present with the genome in an integrated form (6). Frequently, this integration allows the unregulated expression of the viral oncogenes E6 and E7 (5).

In addition to these virus-related factors, genetic susceptibility to viral infection, increasing age of the host, and other epigenetic and lifestyle factors, such as smoking, chronic inflammation, and coinfection with other sexually transmitted organisms, particularly Chlamydia trachomatis, have been shown to increase the risk of progression to cervical cancer in HPV-infected women (7). Several of these factors can be logically linked to increased oxidative stress and DNA damage. Extensive DNA damage usually leads to apoptosis (8), but in cells infected with HPV, the viral oncogenes E6 and E7 rescue cells from this pathway, resulting in mutagenesis, increased cell proliferation, and in rare instances, cancer (9). One of the factors shown to promote cellular transformation is oxidative stress. Oxidative stress during viral infection can be a result of the immune response to viral proteins and/or a consequence of viral gene expression. Oxidative stress causes oxidative DNA damage, which may facilitate HPV DNA integration (10).

A link between virus-induced oxidative stress and viral pathogenesis has been demonstrated in several viral infections, including Epstein-Barr virus (EBV), hepatitis C virus (HCV), and hepatitis B virus (HBV) infections. For example, in the case of HBV-associated hepatocellular cancer, it has been shown that the accumulation of HBV mutant surface proteins in the endoplasmic reticulum of infected cells induces oxidative DNA damage in the late stages of infection. In HCV-induced hepatocarcinogenesis, chronic infection with HCV is characterized by increased oxidative stress. In the case of EBV and nasopharyngeal cancer, the lytic life cycle and the viral oncogene EBNA-1 have both been shown to induce oxidative stress (11–15). In these cases, an increase in oxidative stress both causes direct oxidative DNA damage and also participates in various signaling pathways that can lead to chromosomal aberrations and cell transformation.

In the case of HPV infection, the host immune response is generally limited and viral infection itself does not induce a state of chronic inflammation. The primary reason for this is that the virus infects basal epithelial cells, which are shielded from circulating immune cells during the initial stages of infection. Nevertheless, reactive oxygen species (ROS) and reactive nitrogen species (RNS) have potential significance to the development of viral carcinogenesis. For example, one study showed that exposing cells infected with HPV16 to the reactive nitrogen species nitric oxide increased the levels of E6 and E7 and increased the level of DNA damage (16). Also, previous research has shown that the expression of HPV16 E6 in L929 cells increases ROS accumulation (17).

HPV16 E6 is expressed in cells as two main isoforms: a full-length variant (E6) and a few similarly truncated variants frequently referred to collectively as E6* due to their similarity. The function and activities of full-length oncoprotein E6 have been intensively studied over the last 2 decades (9, 18, 19). In contrast, relatively little is known regarding the activities of the truncated E6* isoform, and its significance both in the viral life cycle and in carcinogenesis has been disputed. The early transcripts produced from the early promoter located upstream of the E6 gene can undergo alternative splicing from a donor site located within the E6 gene (nucleotide [nt] 226 in the case of HPV16) to an acceptor site that can be located either within or outside the E6 gene. This results in the production of several E6* splice variants. This splicing pattern is a unique characteristic of all high-risk HPV types and is not restricted to HPV16 and HPV18. Rather, it is also present in the high-risk types HPV31, HPV33, and HPV45, which, together with HPV16 and HPV18, are responsible for almost all cases of cervical cancer. Interestingly, E6* is the most abundant splice variant produced during the early stages of infection (20). In contrast, low-risk types do not express E6* due to the absence of the consensus splice site. Work done in both our laboratory and that of L. Banks suggests significant and independent roles for the E6* splice variant. For instance, we observed that overexpression of E6* in SiHa cells sensitizes these cells to both tumor necrosis factor- and Fas-induced apoptosis (21) and that full-length E6 and E6* bind to different regions on procaspase 8 and have opposite effects on the stability of that protein (22). Work done by the Banks group has shown that HPV18 E6* regulates the ability of the full-length isoform to degrade p53, with an inverse relationship existing between the level of E6* and the ability of full-length E6 to degrade p53 (23). This group has also shown that E6* can suppress the growth of transformed cells (24). Taken together, these observations suggest that E6* possesses important functions distinct from those of the full-length isoform of E6.

In the current study, we discovered that E6*, but not E6, increases cellular ROS and leads to higher levels of DNA damage. Modulation of the ratio of E6 isoforms can change these ROS levels, with increased proportions of E6* consistently promoting oxidative stress. We also investigated the mechanisms responsible for these changes and found that the truncated versions of E6 may modulate the expression of enzymes involved in ROS metabolism, thereby leading to higher levels of ROS and DNA damage.

MATERIALS AND METHODS

Reagents.

Monoclonal anti-superoxide dismutase isoform 1 (anti-SOD1) and anti-glutathione peroxidase (anti-Gpx) antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Monoclonal anti-SOD2 antibody was obtained from BD Biosciences (Franklin Lakes, NJ), and monoclonal anti-poly(ADP-ribose) polymerase 1 (anti-PARP1) antibody was purchased from Calbiochem (Billerica, MA). Monoclonal anti-β-actin was obtained from Sigma-Aldrich (St. Louis, MO). Monoclonal antibodies directed against the HPV16 E6 N terminus were obtained from Euromedex (France). MG132 was purchased from Sigma-Aldrich (St. Louis, MO).

Cell culture.

CaSki and SiHa cells (derived from human cervical carcinomas), L929 (mouse fibrosarcoma) cells, and U2OS (human osteosarcoma) cells were obtained from the ATCC (Manassas, VA). CaSki, SiHa, and L929 cells were cultured in modified Eagle medium (MEM; Cellgro, Manassas, VA), U2OS cells were cultured in McCoy's 5a medium modified, and normal oral keratinocytes (NOKs) were grown in keratinocyte serum-free medium (Invitrogen, Carlsbad, CA). The medium for all cells was supplemented with penicillin (100 U/ml) and streptomycin (100 μg/ml) (Sigma-Aldrich, St. Louis, MO). MEM and McCoy's 5a medium modified were supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA).

Plasmids, small interfering RNA (siRNA) inhibition, and transfections.

Plasmids pFlag-E6L and pFlag-E6* were obtained by cloning the E6 wild type, E6, and E6* in frame with the N-terminal Flag tag and the C-terminal c-Myc tag into the pFlag-Myc CMV-22 vector (Sigma-Aldrich, St. Louis, MO). Cloning of Flag-E6* and Flag-E6L into the retroviral vector pLNCX (BD Clontech, Mountain View, CA) and production of retroviral stocks have been described previously (21). To inhibit the expression of E6, target sequences for shE6large, as well as the scrambled sequence, were cloned into pSilencer (version 3.1; Invitrogen, Carlsbad, CA) (21).

Transfections were carried out using the TransIT-2020 reagent (Mirus Bio, Madison, WI), as directed by the manufacturer. For transient transfections, cells were analyzed at 48 h posttransfection. SiHa cell- and CaSki cell-derived stable cell lines were obtained by transfection of the parental cells with the corresponding plasmids, followed by G418 or puromycin selection for 2 to 3 weeks. NOK-derived stable cell lines were produced by transduction of retrovirus pLNCX, pLNCX-E6*, or pLNCX-E6, followed by isolation of clones derived from single cells. Individual clones were selected, grown, and analyzed for protein expression by immunoblotting and/or real-time PCR (RT-PCR).

Expression of SOD2 in NOKs was decreased by siRNA inhibition, employing the oligonucleotides for the siRNA control (siControl) and siRNA against SOD2 (siSOD2) obtained from Invitrogen (Grand Island, NY). The X-tremeGENE siRNA transfection reagent (Roche Diagnostics, Mannheimm Germany) was used to transfect these cells with the siControl and siSOD2 oligonucleotides according to the manufacturer's protocol.

Measurement of ROS by flow cytometry.

Cellular levels of hydrogen peroxide and hydroxyl and peroxyl radicals (H₂O₂, OH⁻, and ROO⁻) and cellular levels of superoxide (O₂⁻) were estimated using 5-(and-6)-carboxy-2′,7′-dichlorodihydrofluorescein diacetate (DCF) or dihydroethidium (DHE), respectively (25). All fluorescent probes were purchased from Invitrogen (Carlsbad, CA). Stock solutions (20 mM) were diluted into culture medium and added to cells to a final concentration of 10 μM DCF and 5 μM DHE. Cells were incubated at 37°C in the dark for 25 to 30 min. After treatment, cells were trypsinized, washed, and collected in phosphate-buffered saline (PBS). Cells were analyzed using a Becton, Dickinson FACSCalibur flow cytometer (BD, Franklin Lakes, NJ). A total of 10,000 events were measured per sample. DCF was detected in the FL-1 channel (530/30 nm), while DHE was detected in the FL-2 channel (650 nm). Data were collected in log scale and analyzed using CellQuest Pro and FlowJo software.

Immunoblotting and immunoprecipitation.

For immunoblot analysis, 10⁶ of the indicated cells were lysed in 200 μl Laemmli lysis buffer, and lysates were separated by SDS-PAGE. After transfer of protein onto Immobilon P membranes (Millipore, Billerica, MA) and blocking of the membrane with 1% bovine serum albumin, primary antibodies diluted in Tris-buffered saline–Tween 20 (TBST) were applied. After incubation at 4°C overnight, membranes were washed with TBST and secondary ImmunoPure antibody (antimouse or antirabbit) conjugated with horseradish peroxidase (HRP; Thermo Scientific) was applied. The signal was detected using chemiluminescent SuperSignal West Femto or Pico maximum-sensitivity substrate (Thermo Scientific, Rockford, IL). For detection of the E6 isoforms in SiHa pSilencer, SiHa siE6, SiHa pFlag, and SiHa pE6* cells, 10⁶ cells were pretreated with 10 μM MG132 for 16 h prior to preparation of the lysates. The cells were collected and lysed in 100 μl of radioimmunoprecipitation assay buffer (Sigma-Aldrich).

For immunoprecipitation, 10⁷ cells from each of the NOK clones NOK pLNCX, NOK pLNCX-E6*, and NOK pLNCX-E6 were treated with 10 μM MG132 for 16 h prior to preparation of lysates. Flag-tagged proteins were then precipitated using Flag-agarose, and bound proteins were subjected to SDS-PAGE and then transferred to a polyvinylidene difluoride (PVDF) membrane and detected by immunoblotting. Detection of Flag-E6* and Flag-E6 was performed using anti-Flag-HRP antibodies.

Comet assay.

The comet assay was performed using a Trevigen kit (Trevigen, Gaithersburg, MD) under alkaline conditions. Nuclei were stained with SYBR gold (Invitrogen, Carlsbad, CA). For each sample tested, 100 DNA tails were photographed and analyzed. The length of each tail was measured from the center of the comet to the end of the tail using ImageJ software, and each tail was categorized into one of four tail types reflecting the severity of DNA damage. DNA damage was classified into four classes of tail lengths (0 to 50, undamaged; 50 to 100, minimum damage; 100 to 150, medium damage; >150, maximum damage) such that the severity of DNA damage increases proportionately with tail length.

8-Oxyguanine DNA damage analysis.

DNA damage was determined by binding the fluorescein isothiocyanate (FITC) conjugate to 8-oxodeoxyguanosine (8-oxodG). Cells were fixed in 1% paraformaldehyde and permeabilized in methanol. After fixation and permeabilization, cells were washed, blocked, and incubated with OxyDNA-FITC conjugate (Calbiochem, San Diego, CA) for 1 h in the dark. Cells were resuspended in PBS and analyzed by flow cytometry for fluorescence (excitation, 495 nm; emission, 515 nm) on a BD FACSCalibur flow cytometer (BD, Franklin Lakes, NJ). Data were analyzed using FlowJo software. A total of 10,000 events were measured per sample.

RNA isolation and RT-PCR.

Cells were plated in a 10-cm tissue culture plate and allowed to grow to 80% confluence. RNA was isolated using the Tri Reagent according to the manufacturer's protocol (Sigma-Aldrich, St. Louis, MO). cDNA was synthesized using ImPromII reverse transcriptase (Promega, Madison, WI) and an oligo(dT) primer. Primers for the 5′ and 3′ ends of Flag (5′-ATGGGCGGTAGGCGTGTAC-3′ and 5′-GGTCACAGGGATGCCAC-3′) were used to amplify PCR products for the E6 full-length and splice variants expressed from plasmids. Primers for the 5′ and 3′ ends of E6 (5′-AATGTTTCAGGACCCACAGG-3′ and 5′-CACACAACGGTTTGTTGTATTGCTG-3′) were used to amplify PCR products for full-length E6. The E6* splice variant was amplified using the same primer for the 5′ end with a different primer for the 3′ end (5′-CTTTTGACAGTTAATACACCTCACG-3′) (26). The PCR product from PGK1, obtained using primers 5′-CTGTGGGGGTATTTGAATGG-3′ and 5′-CTCCAGGAGCTCCAAACTG-3′, was used for normalization.

qRT-PCR.

Quantitative real-time PCR (qRT-PCR) was conducted to measure the levels of the E6 isoforms using primers designed as described previously by Hafner et al. (26), along with an Absolute QPCR Sybr green kit according to the manufacturer's protocol (Bio-Rad Laboratories, Inc.). The observed E6 isoform concentrations were normalized using the level of β-actin or 36B4 expression.

Statistics.

All assays were repeated at least three times. Results are reported as means ± standard deviations. Differences were analyzed by Student's t test. A P value of <0.05 was regarded as significant.

RESULTS

ROS levels are higher in CaSki cells than in SiHa cells.

Our initial studies examining the influence of E6 and E6* on ROS levels were done using CaSki and SiHa cells, which are well-known cellular models of cervical cancer derived from HPV16-positive cervical carcinomas. ROS levels in SiHa and CaSki cells were estimated using flow cytometry following staining with the fluorescent dyes 5-(and-6)-carboxy-2′,7′-dichlorofluorescein diacetate (DCF; which detects hydrogen peroxide and hydroxyl and peroxyl redicals) and dihydroethidium (DHE; which detects superoxide radicals) (25). The flow cytometry results clearly demonstrated that the levels of both species were higher in CaSki cells than in SiHa cells (Fig. 1A). These assays were repeated three times to generate the bar graphs shown in Fig. 1B and C. Because previous studies suggested that E6 may be responsible for the increase in ROS (17), we postulated that the difference in ROS levels between these cell lines might be due to differences in E6 expression. Interestingly, the level of expression of full-length E6 was similar in both cell lines, while the level of expression of E6*I (27), the most abundant splice product (referred to as E6* from this point forward), was much higher in CaSki cells than in SiHa cells (Fig. 1D and E). These findings are consistent with the results that we obtained earlier concerning the ratio between E6 and E6* protein levels (21). Since both the ratio of E6*/E6 expression and the absolute level of E6* expression differ between these cell lines, it is possible that either or both factors could contribute to the elevated ROS levels.

FIG 1.

CaSki cells express higher levels of E6* and display higher levels of ROS than SiHa cells. (A to C) The levels of ROS are higher in CaSki cells than in SiHa cells. Cells were treated with either 10 μM DCF or 5 μM DHE in medium and incubated in the dark at 37°C for 30 min. Cells were washed, resuspended in PBS, and analyzed by flow cytometry. (A) Representative flow cytometry histograms (cell number/DCF [FL1] or cell number/DHE [FL2]). (B and C) Triplicate measurements of the mean fluorescence intensity (MFI) of DCF (B) and DHE (C) were performed to generate the bar graphs. The mean fluorescence intensity of CaSki cells was set at 100%. (D and E) The expression levels of E6 and E6* transcripts in SiHa and CaSki cells were analyzed by RT-PCR (D), and the ratio between them was estimated by qRT-PCR (E). RNA was isolated from 10⁶ CaSki or SiHa cells, cDNA was synthesized using a high-capacity cDNA reverse transcription kit (Applied Biosystems, Foster City, CA), and PCR was performed using primers for E6 5′ and E6 3′. Primers for PGK1 were used for normalization (D). (E) The levels of the E6 and E6* transcripts were estimated using primers specific for full-length E6 and primers designed to be specific for the E6* exon-exon region and were normalized by the β-actin level (fold versus β-actin × 10,000). Error bars represent the standard deviations.

Increased E6*/E6 ratios cause higher levels of ROS in SiHa cells.

If a difference in the ratio between E6 isoforms was the major factor determining the difference in ROS levels, one resulting prediction is that changing the ratio between E6 and E6* should alter the level of ROS. To test this idea, we decreased E6 expression in SiHa cells such that the ratio of E6*/E6 would increase. We designed short hairpin RNA to target E6 (shE6), cloned it into the pSilencer vector, and then stably transfected either the empty vector or pshE6 into SiHa cells to generate the SiHa pSilencer control cell line and the stable cell line SiHa shE6, respectively. Both qRT-PCR analysis (Fig. 2A) and immunoblotting data (Fig. 2B) showed that E6 expression was indeed decreased in these cells. Figure 2C shows a bar graph generated following staining of these SiHa-derived cells with DCF, demonstrating that increasing the relative ratio of E6*/E6 expression in SiHa cells leads to a parallel increase in the level of ROS compared to that for the control cell line SiHa pSilencer.

FIG 2.

Manipulation of the ratio between E6 isoform levels in SiHa cells alters ROS levels in these cells. (A) Knockdown of full-length E6 increased the ratio of E6* to E6. RNA was isolated from SiHa pSilencer and SiHa shE6, and cDNA was synthesized using oligo(dT). Real-time PCR was performed using primers specific for E6 and E6*. The levels of the E6 and E6* transcripts in SiHa pSilencer cells were set at 100%. (B) Cell lysates were prepared from SiHa pSilencer and SiHa shE6 cells and subjected to SDS-PAGE. To detect the expression of the full-length and short isoforms of E6, cells were treated with 5 μM MG132 for 16 h prior to lysate preparation. Expression of E6 and β-actin in lysates was detected by immunoblotting using HPV16 E6 N terminus antibodies and monoclonal β-actin antibodies. β-Actin was used to normalize the input. (C) Reduction of full-length E6 expression resulted in increased levels of ROS. The ROS levels in SiHa pSilencer and SiHa shE6 cells were analyzed by flow cytometry as described in the legend to Fig. 1A. Triplicate measurements of the mean fluorescence intensity of DCF were performed to generate the bar graph, and the mean fluorescence intensity of SiHa pSilencer cells was set at 100%. (D) Overexpression of E6* increased the ratio of E6* to E6. Real-time PCR was performed using primers specific for the full-length E6 and E6*. The levels of the E6 and E6* transcripts in SiHa pFlag cells were set at 100%. Error bars represent the standard deviations. (E) Cell lysates were prepared from SiHa pFlag and SiHa pE6* cells and subjected to SDS-PAGE. Expression of E6* and β-actin in lysates was detected by immunoblotting using HPV16 E6 N terminus antibodies and monoclonal β-actin antibodies. To detect the expression of the full-length and short isoforms of E6, cells were treated with 5 μM MG132 for 16 h prior to lysate preparation, and loading was normalized by blotting for β-actin. (F) Overexpression of E6* increased ROS levels in SiHa cells. SiHa cells were stably transfected with plasmid pFlag or pE6*. The levels of ROS in these cells were analyzed by flow cytometry. Triplicate measurements of the mean fluorescence intensity of DCF were performed to generate the bar graph. The mean fluorescence intensity of SiHa pFlag cells was set at 100%. Error bars represent the standard deviations. ***, a 0.99 level of confidence.

To further evaluate the possibility that differences in the E6*/E6 ratio may affect ROS levels, we also manipulated the ratio of E6 isoform expression by increasing the relative amount of E6* in SiHa cells. E6* was cloned into the pFlag vector in frame with the Flag tag at the N terminus and c-Myc at the C terminus, and stable cell lines expressing the vector-derived E6* were produced. To eliminate the effect of clonal integration, six clones with the highest level of E6* expression were combined to produce SiHa pE6* pooled cells. These cells, together with SiHa pFlag control cells, were stained with DCF and analyzed by flow cytometry. DCF staining was carried out in triplicate to produce the bar graph shown in Fig. 2F. The results clearly demonstrate that the SiHa pE6* pooled cell line displayed higher levels of ROS than control cells transfected with the empty vector. Thus, overexpression of the E6* isoform in SiHa cells was able to further increase ROS levels in these cells. qRT-PCR analysis and immunoblotting data (Fig. 2D and E, respectively) confirmed the higher level of E6* expression in SiHa pE6* cells than vector control cells.

Overexpression of E6*, but not that of full-length E6, increases ROS levels in cells.

The experiments described above demonstrated that manipulating E6 isoform levels so that the E6*/E6 ratio increased also increased ROS levels. However, in these experiments, both isoforms were present, potentially complicating interpretation. To address this limitation, we individually expressed E6 and E6* in the HPV-negative noncervical cancer cell lines U2OS (human osteosarcoma) and L929 (mouse fibrosarcoma). Cells were transiently transfected with the pFlag plasmid coding for E6, E6*, or the empty vector (control). At 48 h posttransfection, these cells were stained with DCF for ROS analysis using a fluorimetric plate reader. In both U2OS- and L929-derived cell lines, ROS levels were increased in cells expressing pE6* compared to the levels in those expressing the vector control or pE6 (Fig. 3A and C, respectively). Figure 3B and D show RT-PCR data, confirming the expression of E6 and E6* in U2OS and L929 cells, respectively. These findings indicate that the actual level of E6* expression is responsible for the increase of ROS levels in cells and also that the effect of E6* on ROS levels is independent of the presence or absence of other HPV proteins.

FIG 3.

Transient expression of E6* in noncervical cancer cells U2OS and L929 increased ROS. pFlag, pE6*, and pE6 were transiently expressed in U2OS (A and B) and in L929 (C and D) cells. (A and C) The level of ROS at 48 h posttransfection was estimated by measuring the generation of fluorescent oxidized DCF and measured in cells using a fluorescence plate reader. Autofluorescence in unstained control cells was subtracted. The mean fluorescence intensity of three repeats is presented in the graphs, and error bars represent the standard deviations. (B and D) RT-PCR data show expression of these isoforms in U2OS (B) and L929 (D) cells. (Top) PCR products obtained using primers for the E6 gene; (bottom) PCR products obtained with primers for the PGK1 gene. ***, a 0.99 level of confidence; **, a 0.95 level of confidence.

ROS levels are higher in NOKs expressing E6*.

The findings discussed above were demonstrated in transformed cancer cells. To ask whether the activities of full-length E6 and/or E6* might contribute to differences in cellular ROS in nontransformed target cells, the influence of each isoform on ROS levels was investigated individually in immortalized nontransformed normal oral keratinocytes (NOKs). These cells were chosen because they are natural targets of HPV infection and can be transformed by HPV16 (28).

The E6* and full-length E6 isoforms tagged with the Flag epitope were cloned into the pLNCX retroviral vector. NOKs immortalized by human telomerase reverse transcriptase expression (29) were then infected with the retroviral stocks (pLNCX [control], pLNCX-E6*, and pLNCX-E6), and several stable cell lines were obtained for each isoform. The level of ROS was measured in 2 cell lines expressing pLNCX-E6* (cell lines 1 and 5) and in 2 cell lines expressing pLNCX-E6 (cell lines 1 and 3) and compared to that seen in control pLNCX cells. Figure 4A shows the flow cytometry results following DCF staining in these cells. Figure 4B shows PCR data confirming the expression of E6 and E6* in the NOK stable cell lines. Representative cell lines (NOK pLNCX, NOK pE6* cell line 1, and NOK pE6 cell line 5) were chosen, and the experiment was then repeated three times following DCF and DHE staining to generate the bar graphs shown in Fig. 4C. qRT-PCR data (Fig. 4D) and immunoblotting assay results (Fig. 4E) confirmed the higher level of E6 isoform expression in the selected NOK clones. Overall, NOKs expressing E6* displayed higher levels of ROS than NOKs expressing the empty vector, pLNCX. In contrast, NOKs expressing the full-length isoform demonstrated no significant change in ROS levels relative to those in control cells.

FIG 4.

Expression of E6* increases ROS levels in NOKs. (A) NOKs stably expressing pLNCX, pLNCX-E6* (cell lines 1 and 5), or pLNCX-E6 (cell lines 1 and 3) were analyzed for their levels of ROS by flow cytometry, as described in the legend to Fig. 1A. (B) RT-PCR shows expression of E6 isoforms in NOKs. (Top) PCR products obtained using primers for the E6 gene; (bottom) PCR products obtained with primers for the 36B4 gene. (C) Levels of hydrogen peroxide and hydroxyl and peroxyl radicals detected with DCF (left) and superoxide detected with DHE (right) are higher in the NOK pLNCX-E6* clone than in the NOK pLNCX and NOK pLNCX-E6 clones. The mean fluorescence intensity of three repeats expressed as a percentage compared to that for the NOK pLNCX clone is presented for DCF (left) and DHE (right). Error bars represent the standard deviations. ***, a 0.99 level of confidence; **, a 0.95 level of confidence. (D) Amplification plots of E6 variants in NOK pLNCX, NOK pLNCX-E6*, and NOK pLNCX-E6 clones using a quantitative RT-PCR SYBR green assay are shown. Amplification plots with threshold cycle (C_T) values greater than 30 were shown to be nonspecific and were not detected on the agarose gels. RFU, relative fluorescence units. (E) Overexpression of E6 and E6* in NOKs. The indicated NOK clones were treated with 5 μM MG132 for 16 h prior to sample preparation. Expression of E6* and E6 in these cells was analyzed by immunoprecipitation using anti-Flag agarose, followed by detection using antibodies directed against Flag-HRP. PVDF membranes carrying the SDS-separated proteins were probed with anti-Flag-HRP antibodies, and β-actin antibodies were used to normalize for the protein load.

SOD and glutathione expression levels decrease with E6* expression.

A number of factors may contribute to the observed increase in cellular ROS following E6* expression. One possibility is that E6*-expressing cells possess a reduced antioxidant capacity, thereby allowing ROS levels to increase due to the cell's reduced ability to adequately dispose of them. To determine whether decreases in the cellular antioxidant capacity might contribute to E6*-mediated increases in ROS, we examined the expression of two important antioxidant enzymes, superoxide dismutase (SOD) and glutathione peroxidase (Gpx). SOD can be expressed as three isoforms, SOD1, SOD2, and SOD3. Of the three, only SOD1 and SOD2 are intracellular and were therefore chosen for analysis.

We observed that SiHa and CaSki cells differed not only in the levels of ROS but also in the levels of expression of antioxidant proteins (Fig. 5A). In particular, ROS levels were higher, while expression of the antioxidant enzymes SOD2 and Gpx1/2 was lower in CaSki cells than SiHa cells. To further investigate the relationship between antioxidant expression levels and E6 isoform expression, we repeated these experiments in the NOK system. Consistent with our findings in the cervical cancer cell lines, we observed a distinct reduction in SOD2 expression in cells of the NOK pLNCX-E6* clone compared to that in the control (NOK pLNCX), while no change in SOD1 levels was detected. Furthermore, NOKs expressing E6* displayed lower levels of Gpx expression (Fig. 5B). In contrast, NOKs expressing the full-length isoform pLNCX-E6 did not display significantly changed levels of these proteins. The data described above suggest that E6* may have the ability to affect the levels of SOD2 and/or Gpx1/2 and that this in turn affects the levels of ROS in these cells. To further explore this possibility, we asked whether decreases in the expression of antioxidant enzymes such as SOD2 would also lead to increased ROS levels in NOKs. NOKs were transiently transformed with either siControl or siSOD2 for 72 h. Cells were then collected and either stained with DCF to detect ROS or used to prepare protein lysates for the determination of SOD2 levels. Figure 5C demonstrates a decrease in SOD2 expression in siSOD2 NOKs compared to siControl NOKs, and Fig. 5D shows that this decrease in SOD2 resulted in an increase in ROS levels in siSOD2 NOKs. Together, these findings support the idea that lower expression of antioxidant proteins may contribute to the higher ROS levels observed in E6*-expressing cells.

FIG 5.

Expression of E6* decreases expression of the antioxidant enzymes SOD2 and Gpx1/2, and knockdown of SOD2 levels in NOKs leads to an increase in ROS. (A) SOD2 and Gpx1/2 expression levels are lower in CaSki cells than SiHa cells. (B) E6* expression in NOKs decreases expression of SOD2 and Gpx proteins, while no significant change in SOD1 was observed. Cell lysates were prepared from the indicated cells and subjected to SDS-PAGE. Expression of SOD2, SOD1, Gpx1/2, and β-actin in lysates was detected by immunoblotting using the appropriate antibodies. β-Actin was used to normalize the input. (C) SOD2 expression in NOKs was decreased by siSOD2 and then detected by immunoblotting using mouse anti-SOD2 antibodies. β-Actin was used to normalize the input. siCont refers to siControl. (D) The level of ROS in siSOD2-expressing NOKs at 72 h posttransfection was estimated by flow cytometry. The mean fluorescence intensity of three repeats is presented in the graphs, and the mean fluorescence intensity of siControl-expressing NOKs was set at 100%. Error bars represent the standard deviations. **, a 0.95 level of confidence.

E6* expression results in higher levels of DNA damage in both cervical cancer cell and NOK lines.

The results described above demonstrate that E6* expression leads to an increase in ROS and a decrease in the level of at least two antioxidant enzymes, SOD2 and Gpx. Both findings point toward the induction of oxidative stress as a result of E6* expression. Oxidative stress, in turn, is known to induce DNA damage. To determine whether E6* expression also increases DNA damage, we employed two assays: the comet assay (Sigma-Aldrich), which detects single-strand DNA breaks, and the flow cytometric OxyDNA assay (Calbiochem), which measures the levels of 8-oxoguanine, one of the most commonly modified bases during oxidative stress in cells (30).

Figure 6A (top) demonstrates the effect of incremental increases in DNA damage on comet appearance in control cells following alkaline electrophoresis. Figure 6A (bottom) shows the results of a representative comet assay performed on NOKs transfected with the empty vector, E6*, or E6. When CaSki and SiHa cells were analyzed using the comet assay, the results (Fig. 6B) showed that CaSki cells, with their higher level of oxidative stress, sustained a higher level of DNA damage than SiHa cells. This is consistent with a model in which the observed increase in oxidative stress corresponds to downstream changes in the level of DNA damage. An alternative method of determining DNA damage is to assess the levels of 8-oxoguanine, since this is the most commonly modified base. The OxyDNA assay results for the cervical cancer cells CaSki and SiHa (Fig. 6C) reflected the results of the comet assay, namely, that CaSki cells exhibited a higher level of DNA damage than SiHa cells.

FIG 6.

E6* expression coincides with higher levels of DNA damage both in cervical cancer cells and in NOKs. (A) Representative comet images of cells stained with intercalating dye and visualized by epifluorescence microscopy following alkaline electrophoresis. The length of the comet tail is an indicator of the level of DNA damage. (Top) Control cells (provided by Trevigen) for comparison of DNA damage using tail lengths are presented. CC0, a healthy cell population; CC1 and CC3, increasing levels of DNA damage, as evidenced by their increasing tail lengths. (Bottom) NOK populations expressing E6* have a larger number of cells with the longest tail lengths and thus greater levels of DNA damage than control populations and those expressing E6. (B and D) DNA damage detected by the comet assay is higher in CaSki cells than SiHa cells (B) and in the NOK pLNCX-E6* clone than the NOK pLNCX or NOK pLNCX-E6 clone (D). The comet assay was performed and the results were analyzed as described in Materials and Methods. One hundred cells were counted per cell line, and the percentage of cells with each tail length was calculated. (C and E) Levels of 8-oxodG were higher in CaSki cells than SiHa cells (C) and in the NOK pLNCX-E6* clone than the NOK pLNCX or NOK pLNCX-E6 clone (E). The 8-oxodG level was estimated using a fluorometric OxyDNA assay kit (Calbiochem) as directed by the manufacturer. Cells were incubated with the FITC probe conjugated to 8-oxodG and analyzed by flow cytometry. Experiments were performed in triplicate, the mean fluorescence intensity for CaSki cells was expressed as a percentage compared to the mean fluorescence intensity of SiHa cells (which was set at 100%), and error bars represent the standard deviations. ***, a 0.99 level of confidence; **, a 0.95 level of confidence.

To determine whether increased E6* expression can also lead to DNA damage in nontransformed cells, comet and OxyDNA assays were performed on the NOK pLNCX, NOK pLNCX-E6*, and NOK pLNCX-E6 clones. We observed (Fig. 6D and E) that for both assays, pLNCX-E6* cells displayed a higher level of DNA damage than either the control (NOK pLNCX) or the NOK pLNCX-E6 clone. Taken together, these findings indicate that E6* expression in cells can lead both to increased levels of ROS and to a corresponding increase in oxidative DNA damage.

Poly(ADP-ribose) polymerase 1 (PARP1) is an enzyme that ADP ribosylates nuclear proteins and is involved in the repair of DNA breaks. Therefore, we used it as an indirect marker of DNA damage to validate our findings (31). Our data revealed higher levels of PARP1 expression in CaSki cells (Fig. 7A) than in SiHa cells and in the NOK pLNCX-E6* clone than in the NOK pLNCX or NOK pLNCX-E6 clone.

FIG 7.

The DNA damage marker PARP1 is increased in cells with higher E6* expression, and overexpression of E6* in SiHa cells increases DNA damage and decreases SOD2 expression. (A and B) PARP1 levels are higher in CaSki cells than SiHa cells (A) and in the NOK pLNCX-E6* clone than the NOK pLNCX or NOK pLNCX-E6 clone (B). Cell lysates were prepared from the indicated cells and subjected to SDS-PAGE. Expression of PARP1 and β-actin in lysates was detected by immunoblotting using the appropriate antibodies. β-Actin was used to normalize the input. (C) Levels of 8-oxoguanine in SiHa pFlag and SiHa pE6* cells were estimated as described in the legend to Fig. 6. ***, a 0.99 level of confidence. The mean fluorescence intensity of SiHa pFlag cells was set at 100%. (D) Expression of SOD2 in SiHa-derived cells was estimated by immunoblotting using mouse anti-SOD2 antibodies. β-Actin was used to normalize the input.

If E6* causes oxidative damage leading to DNA damage, we would predict that overexpression of E6* in SiHa cells would result in an increase in DNA damage. Analyzing the levels of 8-oxoguanine in SiHa pE6* cells compared to those in SiHa pFlag control cells confirmed this expectation (Fig. 7C). Consistent with these findings, SOD2 levels were lower in the SiHa pE6* cells than in the SiHa pFlag control cells (Fig. 7D).

DISCUSSION

In this study, we demonstrated that expression of E6*, the truncated splice variant of the HPV16 E6 protein, can increase ROS levels in host cells. This may be due to decreased expression of antioxidant enzymes and leads to downstream DNA damage.

The generation of oxidative stress during a viral infection is a common occurrence during the inflammatory response to infection, due to the release of ROS from neutrophils and macrophages during the oxidative burst. In addition, activated phagocytes release prooxidant cytokines. RNA viruses, such as influenza A virus and members of the Paramyxoviridae family, have been shown to activate monocytes and polymorphonuclear leukocytes, which respond to infection with a respiratory burst and generate ROS. However, some viruses, such as HPV, EBV, and HBV, are weakly immunogenic and do not necessarily induce a pronounced inflammatory response (32–34). Interestingly, oxidative stress plays a significant role in viral pathogenesis in these infections as well. Several groups have demonstrated that HBV can induce oxidative stress both in vivo in mice and in vitro in cells (35). The human hepatitis virus X protein targets the mitochondria to alter the membrane potential and increase endogenous ROS levels, while HBV-infected cells carrying the HBV pre-S mutant exhibit enhanced levels of ROS and oxidative DNA damage through endoplasmic reticulum stress pathways (36). Although the significance of oxidative stress in the context of the viral life cycle has not been elucidated in each of these cases, biologically significant effects of these increases in oxidative stress have been well documented.

E6/E7 mRNA from high-risk HPV types undergoes alternate splicing to produce an E6* transcript, which is often the most prevalent species found both in cervical tumors and in the early stages of HPV infection. Here, we present data demonstrating that the E6* isoform present in high-risk HPV types increases oxidative stress in both HPV-positive (Fig. 1 and 2) and HPV-negative (Fig. 3 and 4) cell lines. In contrast, the full-length isoform, E6, displayed no significant effect on ROS.

ROS and antioxidants exist in a delicate balance in cells. Oxidative stress occurs when this balance is disrupted due to either an increase in ROS production, a decrease in cellular antioxidant levels, or both (37). The reactive oxygen species released during oxidative stress have the ability to directly damage DNA (38). One reactive oxygen species in particular, the highly reactive hydroxyl radical, interacts with DNA directly, damaging both purine and pyrimidine bases (39). The reaction of the hydroxyl ion with guanine, the most easily oxidized base, leads to formation of the most common base modification, 8-oxo-7,8-dihydro-2′-deoxyguanosine (8-oxodG) (38, 40). In addition to creating modified bases, ROS-induced DNA damage leads to single- and double-strand breaks, abasic sites, and DNA cross-linking (41, 42). This damage to DNA may lead to mutations, aberrations, and genomic rearrangements. Consistent with these observations, our data show that DNA damage is the highest in HPV-positive cells with higher relative E6* expression (Fig. 6B and C), as well as in NOKs expressing E6* (Fig. 6D and E). PARP, a cellular marker of DNA damage, was also increased in CaSki cells compared to SiHa cells and in cells of the NOK E6* clone compared to cells of the NOK pLNCX and NOK pLNCX-E6 clones, further validating our data showing that DNA damage is higher in these cell lines (Fig. 7).

One of the mechanisms by which cells can counteract the effects of increased oxidative stress is through the expression of antioxidant enzymes. Endogenous ROS originates from the mitochondria, cytochrome P450 metabolism, peroxisomes, and inflammatory cell activation (43). Among these, the mitochondria are the primary source of the free radicals hydrogen peroxide and superoxide in cells. Since mitochondrial oxidative phosphorylation is the major source of free radical generation, mitochondria are enriched with antioxidant enzymes, such as SOD and Gpx (44, 45). In this report, we demonstrate that changes in cellular ROS are associated with changes in the expression of antioxidant enzymes, such as SOD2 and Gpx (Fig. 5 and 7D). Events that cause a decrease in SOD are predicted to result in an increase in the levels of ROS. We found that expressing E6* in NOKs resulted in both a decrease in SOD2 expression and a corresponding increase in ROS (Fig. 4 and 5). In addition, CaSki cells displayed lower levels of SOD2 than SiHa cells, and these lower levels of SOD2 were associated with higher levels of ROS (Fig. 1 and 5). Gpx also functions in antioxidant defense, and its inhibition results in an increase in the level of cellular ROS as well (46). Our findings with Gpx mirror those for SOD2, with an increase in E6* leading to a decrease in Gpx and an increase in ROS (Fig. 1, 4, and 5). Expression of antioxidant enzymes is regulated by several transcription factors, including AP-1, SP-1, Nf-κB, p53, and NRF2, among others (47, 48). It is possible that E6* may modulate the expression of antioxidant enzymes either directly or indirectly by influencing transcriptional factor expression or activity. However, this question requires further investigation.

In conclusion, we have now demonstrated that expression of the HPV16 E6* isoform increases oxidative stress and induces oxidative DNA damage in host cells. The significance of this increase in ROS to the HPV life cycle requires further exploration. At present, we can only speculate that this increase in ROS and oxidative DNA damage may contribute to viral genome production. Recent studies suggest that the introduction of double-strand DNA breaks into HPV DNA during productive replication is an important step in HPV genome amplification and genome maturation (49). Studies have also shown that exposure of cells that normally maintain episomal copies of the HPV genome to physiologically high doses of nitric oxide can lead to upregulation of early E6 and E7 oncogene expression, as well as to a significant increase in DNA double-strand breaks (16). This increased oxidative stress likely plays a role in HPV-mediated carcinogenesis. For example, studies in W12 cells demonstrated that increased DNA double-strand breaks are associated with HPV16 integration in cervical keratinocytes (50). In addition, ROS-induced DNA damage results in single- and double-strand breaks, abasic sites, modified bases, and DNA cross-linking (41, 42). Interestingly, expression of the E6* variant of E6 coincides with E7 expression during the early stages of HPV infection, and E6* has been demonstrated to be the most prevalent species in cervical tumors (20), suggesting that a clear understanding of its activities and roles is likely to contribute to our understanding of both the virus life cycle and carcinogenesis. With this study, we have now demonstrated a link between E6* expression and oxidative stress in cells, by showing that E6* expression can increase ROS levels, resulting in increased levels of DNA damage. Further work will focus on the impact of this E6*-mediated oxidative stress on the virus life cycle and carcinogenesis.