The Upstream Regulatory Region of Human Papillomavirus Type 31 Is Insensitive to Glucocorticoid Induction

Abstract

The upstream regulatory region (URR) of various types of human papillomaviruses (HPVs) has been shown to contain functional glucocorticoid response elements (GREs), including HPV type 11 (HPV11), HPV16, and HPV18. Glucocorticoids have been demonstrated to induce the transcriptional activity of the early promoters of these HPV types. Although it has been assumed that the URR of HPV31 contains at least one GRE, no functionality has been demonstrated. We attempt to show here inducibility of the URR of HPV31 by the synthetic glucocorticoid dexamethasone (dex). By sequence analysis we identified three potential GREs in the URR of HPV31. Gel shift analysis indicated that each of these three sites has the potential to be a functional GRE. However, constructs containing the full-length URR, 5′ deletions of the URR, and an internal fragment of the URR containing all three putative GREs were only weakly inducible by dex. Linker scanning mutants, whereby each potential GRE was replaced individually, in double combination, or in triple combination by a unique polylinker, had no effect on dex inducibility. Replacement of each of the three HPV31 GREs with the GRE of HPV18 failed to induce a response to dex. Placement of the HPV18 GRE into the URR of HPV31 in a region similar to its location in the HPV18 URR was also unable to result in a strong dex induction of the HPV31 URR. These data suggest that the lack of dex inducibility is due to the overall context of the HPV31 URR and may be dependent on the requirements of the major early promoter for transcriptional activation. Finally, replacement of the HPV18 GRE with each of the HPV31 GREs in HPV18 only showed weak inducibility, indicating that the three GREs of HPV31 are in fact only weak inducers of dex. Overall, these data suggest that dex responsiveness, along with oncogenic potential, may provide a possible explanation for the classification of HPV31 as an intermediate-risk virus and demonstrate the complexity of transcriptional regulation of the URR of HPV.

Human papillomaviruses (HPVs) are small, double-stranded DNA viruses that specifically infect mucosal and cutaneous epithelium. Over 100 types of HPV have been identified, with roughly one-third infecting the genital epithelium. With the identification of HPV as the etiologic agent of cervical cancer (74, 76), HPV types that infect the genital epithelium are separated based on their associated risk. High-risk types, most commonly HPV type 16 (HPV16) and HPV18, are frequently found in high-grade cervical lesions and invasive cancer. Intermediate-risk types, including HPV31, HPV33, and HPV35, are associated with all grades of cervical lesions and are often associated with invasive cancer. Low-risk HPV types, most frequently HPV6 and HPV11, are commonly associated with low-grade cervical lesions and are rarely found in invasive cancer (35).

The genomes of all of the genital HPV types are approximately 8 kb in length and are similarly organized into early, late, and noncoding regions. The noncoding region, or upstream regulatory region (URR), contains cis-enhancer elements to many cellular and viral factors. Binding of these factors to the URR modulates both viral replication and viral gene transcription (33). Binding sites have been identified for the viral proteins E1 and E2 (39, 56, 66, 68), as well as for the cellular factors AP1 (12, 25, 30, 34), Oct-1 (25), YY1 (30), Sp1 (25), NF1 (21), TEF-1 (30), and CDP/Cut (48, 55), among others.

Steroid hormones have been suggested to be risk factors for HPV-positive cervical cancer. This is primarily due to epidemiological evidence that indicates oral contraceptive users are at an increased risk for developing cervical cancer, with long-term users at an even greater risk (4, 5, 69, 71). In vitro, hydrocortisone has been shown to enhance the transformation of human primary keratinocytes transfected with HPV16 and activated ras (16), whereas baby rat kidney cells transformed with HPV16 have a threefold increased transformation rate with dexamethasone (dex) treatment (44). Studies involving reporter constructs containing the URRs of HPV11, HPV16, or HPV18 demonstrated that both dex and progesterone treatment can upregulate the activity of the major early promoters of these HPV types (7, 9, 11, 40) and that estrogen can upregulate HPV16 gene expression through the URR (11, 43). The major early promoter of HPV controls transcription of the early, nonstructural genes of HPV, including the viral oncogenes E6 and E7 (26, 52, 63, 67). These studies indicate the oncogenic potential of steroid hormones and implicate them as possible therapeutic targets.

In addition to the cellular and viral factors mentioned that bind the URR of HPV, glucocorticoid response elements (GREs) have also been identified in the URRs of several HPV types, including HPV11, HPV16, and HPV18 (7, 9, 20, 31, 44, 56, 72). Furthermore, the presence of GREs has been proposed in the URRs of HPV6, HPV31, and HPV33 (9). Since the publication of the HPV31 sequence in 1989, it has been assumed that the URR of HPV31 contains at least one GRE (22). Since no functionality has been assigned to this or any other GRE in the URR of HPV31, we chose to analyze potential GREs and the induction of transcription by glucocorticoids in the URR of HPV31. Previous studies by our laboratory have uncovered a great deal of information about promoter usage and transcript expression of HPV31 (49-52, 61). These studies have utilized the CIN612-9E cell line (9E), a clonal isolate of a low-grade cervical lesion that maintains approximately 50 episomal copies of HPV31b per cell (3). Organotypic (raft) cultures of this cell line produce infectious virus, indicating that all factors necessary for the complete viral life cycle are present and expressed at the proper levels and times in this cell line (42). Therefore, we chose to study dex inducibility of the URR of HPV31 in the 9E cell line.

By sequence analysis we have identified three potential GREs in the 5′ domain of the URR of HPV31. Gel shift analysis indicated that all three GREs induced a bound complex in 9E nuclear extracts and with recombinant human glucocorticoid receptor (rhGR). We found that dex treatment did not result in a strong induction of transcriptional activity of the URR of HPV31. Linker scanning mutational analysis failed to alter the response to dex in both the full-length URR or an internal fragment, suggesting that the GREs are not acting as antagonists of each other to prevent dex inducibility. Placement of the GRE of HPV18 in the URR of HPV31 did not result in a transcriptional induction with dex treatment, even though the GRE from HPV18 is functional in our system. Replacement of HPV18 GRE within the HPV18 URR with each of the HPV31 GREs only showed mild dex inducibility, further suggesting that the HPV31 GREs are only weakly active. Taken together, these studies suggest a potential explanation for why HPV31 is considered an intermediate-risk virus and demonstrates the complexity of transcriptional regulation of the URR of HPV.

MATERIALS AND METHODS

Cell culture.

The CIN612-9E (9E) cell line is a clonal isolate from a biopsy of a CIN-1 lesion that maintains 50 episomal copies of HPV31b per cell (3). Cells were cultured in epithelial medium (E medium) (41) with 5% fetal bovine serum along with mitomycin C-treated J2 fibroblasts. To remove endogenous steroid hormones from the media and serum for transfection experiments, cells were grown in E medium lacking phenol red and hydrocortisone and supplemented with 5% dextran-coated, charcoal-stripped fetal bovine serum (Cocalico, Reamstown, Pa.). We termed this medium Deficient medium. Cells were maintained in a 37°C, 5% CO₂ incubator.

Plasmids and constructs.

For the wild-type pGL2-31URR construct, the HPV31 URR, nucleotides (nt) 6921 to 121, was cloned into the KpnI/HindIII sites of the promoterless, enhancerless pGL2-Basic (pGL2-B) firefly luciferase vector (Promega Corp., Madison, Wis.) as described previously (61). pGL2-31URR linker scanning mutants (LSMs) were created as described previously (61). Single (p31LSMGRE1, p31LSMGRE2, and p31LSMGRE3), double (p31LSMGRE1,2, p31LSMGRE2,3, and p31LSMGRE1,3), and triple (p31LSMGRE1,2,3) LSM constructs were created within the pGL2-31URR construct, replacing 31GRE1 (nt 7188 to 7205), 31GRE2 (nt 7253 to 7270), and 31GRE3 (nt 7404 to 7421) with the unique polylinker NdeI-ApaI-BclI. 5′ deletion constructs of pGL2-31URR were created as follows: p31URRΔ6921-7238 was constructed by cloning the AccI-PstI fragment of HPV31 (nt 7238 to 121) into pGEM-3zf- (Promega), with subsequent cloning into the KpnI-HindIII sites of pGL2-B. p31URRΔ6921-7557 and p31URRΔ6921-7799 were created by cloning the SpeI-HindIII (HPV31 nt 7557 to 121) and StyI-HindIII (HPV31 nt 7799 to 121) fragments, respectively, of pGL2-31URR into pGL2-B. pAd constructs were created by replacing the simian virus 40 early promoter of pGL2-Pr (Promega) with the adenovirus 2 major late promoter (Ad2MLP; nt −38 to +3) at the BglII/HindIII sites by using oligonucleotides engineered with sticky BglII (5′) and HindIII (3′) ends. pAd31URR7170-7440 wild-type and LSM constructs (pAdLSMGRE) were made by PCR with primers just upstream of 31GRE1 (5′ primer) and just downstream to 31GRE3 (3′ primer) and using pGL2-31URR wild-type and LSM constructs (p31LSMGRE) as templates. PCR products were then cloned into the polylinker of the pAd vector. pGL2-18URR was created by subcloning the URR of HPV18 from pXP2LCR18 (a kind gift from Saleem Khan) into pBSsk+ at the BamHI site, with subsequent cloning into pGL2B at the KpnI/SacI sites. This construct contains the full-length URR of HPV18 from nt 6929 to 119. The p18LSMGRE construct was created in a manner similar to that of the single LSM constructs of pGL2-31URR, replacing the GRE of HPV18 (nt 7839 to 7853) with the unique polylinker NdeI-ApaI-BclI. 18URR-C7843A was created by site-directed mutagenesis of pGL2-18URR to create a 1-bp change in the GRE of HPV18 from the sequence 5′-AGCACATACTATACT-3′ to 5′-AGCAAATACTATACT-3′, which has been shown by others to abolish dex induction of the HPV18 URR (40). Replacement of each of the three HPV31 GREs with the GRE of HPV18 was performed in both pGL2-31URR single LSMs (p31LSMGRE/18GRE) and pAd31URR7170-7440 single LSMs (pAdLSMGRE/18GRE) via NdeI-BclI digestion of each mutant, and insertion of an 18GRE oligonucleotide with sticky NdeI (5′) and BclI (3′) ends. Replacement of the HPV18 GRE with each of the HPV31 GREs in the p18LSMGRE construct [p18URRLSM(31GRE)] was created in a similar fashion by using 31GRE oligonucleotides with sticky NdeI and BclI ends. Positioning of the HPV18 GRE in the URR of HPV31 at a location that corresponds to the native location in HPV18 was accomplished by creating an LSM spanning nt 7883 to 7900 in the URR of HPV31 (p31URR-3′LSM) and inserting the 18GRE oligonucleotide in the NdeI-BclI sites [p31-3′LSM(18GRE)]. Similar constructs were created for all 5′ deletion constructs of pGL2-31URR as well in an identical fashion. All oligonucleotides were purchased from Operon Technologies (Alameda, Calif.). All PCR-generated constructs were verified by sequencing.

Transfection and luciferase assays.

9E cells were transfected with Lipofectamine Plus reagent (Invitrogen/Life Technologies, Carlsbad, Calif.) by using 1 μg of DNA construct, 6 μl of PLUS reagent, and 4 μl of Lipofectamine. Briefly, cells were seeded at a density of 2 × 10⁵ cells per 35-mm dish in 5 ml of Deficient medium. At 12 h postseeding, cells were washed with 1 ml of phosphate-buffered saline and transfected with serum-free media (KGM-2; BioWhittaker, Walkersville, Md.) containing 1 μg of empty vector or each experimental construct according to the manufacturer's instructions. After a 3-h incubation, the medium was replaced, and cells were treated with vehicle (0.1% ethanol [EtOH]) or with 10⁻⁶ M dex. Mitomycin C-treated J2 fibroblasts were then seeded at a density of 5 × 10⁴ cells per dish. After 48 h, cell lysates were prepared by using Passive lysis buffer (Promega), and luciferase assays were performed with a Turner Designs TD 20/20 luminometer by using the Luciferase Assay System (Promega) as recommended by the manufacturer. Luciferase activity values were normalized so that the relative luciferase activity is shown as a fold change over the wild-type construct under vehicle treatment, which was set to 1. Any effect of treatment on the empty vector was taken into account for all normalized values. All transfection experiments were done in duplicate three times.

Electrophoretic mobility shift assays (EMSAs).

Nuclear protein extraction was performed as described previously (34). Protein concentration was determined by the Bradford assay with Bio-Rad protein assay dye reagent (Bio-Rad Laboratories, Hercules, Calif.). Gel shifts were performed as described previously (1) with some modifications. Briefly, 10 μg of nuclear extract were incubated with 1 ng of [γ-³²P]ATP-labeled double-stranded DNA probe, 1.5 μg of poly(dI-dC), and reaction buffer containing 12% glycerol, 20 mM HEPES (pH 7.9), 60 mM KCl, 1 mM EDTA, and 1 mM dithiothreitol. Reactions were performed at room temperature for 30 min. For the competition gel shift, a 10× or 50× molar excess of cold competitor was added prior to addition of protein. GRE consensus (GREc) and GRE consensus mutant (GREmt) oligonucleotides were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, Calif.), whereas 18GRE and 31GRE oligonucleotides were constructed and purchased from Operon. Gel shift with purified protein was performed under similar reaction conditions where 25 ng of rhGR (Panvera, Madison, Wis.) was incubated with 1 ng of labeled probe, 1.5 μg of poly(dI-dC), 400 ng of BSA/ml, and reaction buffer containing 10% glycerol, 20 mM Tris (pH 7.5), 1 mM EDTA, 0.05% NP-40, 50 mM KCl, and 1 mM dithiothreitol (23). All reactions were run on a 6% 1× TAE polyacrylamide gel (37.5:1 acrylamide-bisacrylamide). Gels were then dried at 80°C for 2 h and subjected to autoradiography.

RESULTS

Identification of GREs in the URR of HPV31.

It has always been assumed that, like HPV16 and HPV18 (7, 9, 12, 20, 31, 40), HPV31 contains GREs in its URR (9). In fact, in the published report of the HPV31 sequence, a potential GRE is highlighted as being present in the URR (22). Characteristics of GREs include a 15-bp sequence containing an incomplete palindrome separated by a 3-bp spacer (8, 65). Based on sequence analysis, we have identified three potential GREs in the URR of HPV31 which have a large degree of homology to the consensus GRE sequence of 5′-GGTACAnnnTGTTCT-3′. All three GREs identified are located in the 5′ URR domain, within a 270-bp region (Fig. 1A). For simplicity, we have named the GREs as follows: 31GRE1 (at nt 7188), 31GRE2 (at nt 7254), and 31GRE3 (at nt 7406). The location of these GREs is in direct contrast to the location of the HPV18 GRE, which lies very close to the p105 promoter in the 3′ end of the URR (Fig. 1B) (7). The 12-bp incomplete palindrome of the GRE is considered to be responsible for GR binding (2, 65). While GR is able to bind GRE as both a monomer and a dimer, transcriptional activation via GR binding to GRE is thought to be due to dimer binding only (8). 31GRE1 has 9 of 12 nt in common with the consensus GRE, whereas 31GRE2 and 31GRE3, as well as the GRE of HPV18, have 8 of 12 nt in common with the consensus sequence (Fig. 1C). The lack of complete identity to the consensus has also been seen with the other natural GREs, such as those from the tyrosine aminotransferase gene (20, 28), human metallothionein II gene (20), and the mouse mammary tumor virus long terminal repeat (10, 37).

FIG. 1.

Schematic representations comparing the URR of HPV31 (A) and HPV18 (B). Organization of functional domains of the HPV31 URR (25) and HPV18 URR (33) are indicated. Transcription factor binding sites are shown for YY1 (30), TEF-1 (30), AP1 (7, 33, 34), Oct-1 (24), NF1 (7, 21), KRF-1 (7), E2BS (30, 57), E1BS (30, 53), Sp1 (7), and GRE (7, 40). The location of the three GREs identified in the 5′ URR domain of HPV31 and the known GRE in the URR of HPV18 are highlighted. (C) Sequence comparison of GREc, 18GRE, and the three HPV31 GREs (31GRE1, 31GRE2, and 31GRE3). Boldfaced and boxed nucleotides indicate homology to the GREc sequence.

HPV31 GREs appear to be functional as determined by EMSA.

To determine whether the three GREs we identified in the URR of HPV31 are in fact true GREs, we first attempted to show functionality by EMSA or gel shift assay. We used 15-bp oligonucleotides of the GREc, the HPV18 GRE (18GRE) or each of the three HPV31 GREs (31GRE1, 31GRE2, and 31GRE3) as probes in a binding reaction with 9E nuclear extracts. The 9E cell line is a clonal isolate of a CIN-1 lesion maintaining 50 episomal copies of HPV31b per cell (3). This cell line has been shown to support the complete life cycle of HPV31 in raft cultures (42), making the 9E cell line ideal in the study of HPV31. As shown in Fig. 2A, all three HPV31 GREs were able to form a bound complex identical to that of GREc and 18GRE. Interestingly, 31GRE1 also induces another bound complex, which probably consists of a monomer of the GR, as opposed to the other bound complex, which presumably contains GR bound as a dimer (Fig. 2A, lane 6). In addition, all three GREs from HPV31 were able to compete off this bound complex from the consensus GRE (Fig. 2B), with 31GRE3 being the strongest competitor, even compared to 18GRE. These data suggest that the same protein is binding to the HPV31 GREs we identified as to the consensus GRE and the HPV18 GRE.

FIG. 2.

EMSAs of HPV31 GREs. (A) Comparative binding of the three HPV31 GREs (31GRE1, 31GRE2, and 31GRE3) to the GREc and 18GRE in 9E nuclear extracts. Odd-numbered lanes are reactions without protein, and even-numbered lanes contain 10 μg of 9E nuclear extract. (B) Competition gel shift of 9E nuclear extracts on a GREc-labeled probe with increasing amounts of cold competitor. A 10× or 50× molar excess was added as indicated cold competitor. (C) Comparative binding of the three HPV31 GREs (31GRE1, 31GRE2, and 31GRE3) to the GREc, a consensus mutant GRE (GREmt), and 18GRE using purified rhGR in place of 9E nuclear extracts. Odd-numbered lanes show reactions without protein, and even-numbered lanes contain 25 ng of rhGR.

To ensure that the bound complex seen with the three HPV31 GREs was due to binding by GR to the GRE, purified rhGR was used in place of the 9E nuclear extracts. As shown in Fig. 2C, 31GRE1, 31GRE2, and (to a lesser extent) 31GRE3 were all able to induce a bound complex similar to GREc and 18GRE, whereas a mutated form of GREc (GREmt) failed to bind. 31GRE3 appeared to be the strongest competitor as determined by competition EMSA (Fig. 2B) but only weakly bound purified GR (Fig. 2C). This may be explained by the fact that nuclear extract was used for the competition EMSA as opposed to purified protein. Steroid hormone receptors act as part of a large multiprotein complex to affect gene transcription, and we suggest that these proteins are present in the nuclear extracts. These data indicate that all three GREs from HPV31 are good candidates to be functional GREs.

The GRE of HPV18 is inducible by dex in our system.

In order to be certain that the 9E cell line would support glucocorticoid-inducible transcriptional activation, we first tested inducibility of the HPV18 URR, which has previously been shown by others to contain a functional GRE (7, 9, 40). The BamHI fragment of HPV18 (nt 6929-121) was cloned in front of a luciferase reporter gene. The resulting construct, pGL2-18URR, along with an LSM construct that replaced the GRE of HPV18 (nt 7839 to 7853) with the unique polylinker NdeI-ApaI-BclI (p18LSMGRE), were transfected into 9E cells which were then treated with either vehicle (0.1% EtOH final concentration) or the synthetic glucocorticoid dex at a final concentration of 10⁻⁶ M. In addition, a more subtle GRE mutant, 18URR-C7843A, which changes one nucleotide within the GRE, was similarly transfected. This mutation has been shown to result in a loss of dex inducibility of the URR of HPV18 (40). As shown in Fig. 3, the wild-type construct was induced almost sevenfold upon dex treatment, indicating that the 9E cell line can support dex-inducible transcriptional activation. Based upon published reports (7, 40), we expected both the LSM and the subtle mutant of the GRE of HPV18 to result in a complete loss in the dex induction seen with the wild-type construct. Instead, we observed only a 60% reduction in activity, suggesting that the URR of HPV18 can be induced by dex at sites other than the known GRE in the 9E cell line. This is not surprising, since another GRE at position 7756 was identified by computer search, although it was unresponsive to dex treatment in the C33 cell line (40).

FIG. 3.

Activation of HPV18 URR constructs in 9E monolayer cells upon dex treatment. Schematic diagrams of constructs are shown at the left, where GRE indicates the location of the GRE and LSM indicates substitution with the unique polylinker NdeI-ApaI-BclI. The wild-type HPV18 URR construct pGL2-18URR, the LSM p18LSMGRE (which replaces the GRE of HPV18 with the unique polylinker), and the subtle mutant 18URR-C7843A (which changes one nucleotide in the GRE) were transfected into 9E cells and treated with 0.1% EtOH (veh; open bars) or 10⁻⁶ M dex (solid bars) and assayed for luciferase activity. The relative luciferase activity is shown as the fold change over that of pGL2-18URR under vehicle conditions, which was set to 1. The fold change in activity upon dex treatment is shown on the right.

HPV31 URR is weakly inducible by dex.

Once it was determined that the URR of HPV18 was inducible by dex in our system, we next sought to determine whether the URR of HPV31 was similarly inducible. The PstI fragment of HPV31 (nt 6921-121) was cloned in front of a luciferase reporter gene and transfected into 9E cells. Compared to pGL2-18URR, pGL2-31URR was only weakly inducible upon dex treatment (Fig. 4). Lack of a strong dex induction could be due to the presence of three GREs squelching or competing for factors necessary for a glucocorticoid response. In addition, there are reports of negative GREs (nGRE) that can bind GR but result in a glucocorticoid-dependent repression of transcriptional activity (13, 45, 58). Therefore, we created LSMs of the three GREs in the URR of HPV31 in which each GRE was replaced individually (p31LSMGRE1, p31LSMGRE2, and p31LSMGRE3), in double combination (p31LSMGRE1,2, p31LSMGRE2,3, and p31LSMGRE1,3), or in triple combination (p31LSMGRE1,2,3) with the unique polylinker NdeI-ApaI-BclI. As seen in Fig. 4, none of the LSMs resulted in a significant change in transcriptional activity upon dex treatment compared to the wild-type construct. One possible explanation for these results could be the presence of a repressor within the URR of HPV31, which could prevent a dex-inducible response from any or all of the GREs. It has been well established that the cellular transcription factor AP1 and GR are antagonists to each other's actions (29, 60, 73). Furthermore, the URR of HPV31 contains four AP1 sites, while that of HPV18 contains only two AP1 sites (Fig. 1A and B) (7, 19, 33), which could explain why the URR of HPV18 is inducible by dex, whereas the URR of HPV31 is not. Therefore, we chose to analyze the dex inducibility of an internal fragment of the HPV31 URR. This 270-bp fragment (nt 7170 to 7440) contains all three HPV31 GREs. We placed this fragment in front of the luciferase reporter gene under the control of the minimal Ad2MLP. This promoter has been used by others to study dex-inducible elements (27, 36). The resulting construct, pAd31URR7170-7440, removes the GREs from the context of the rest of the URR, including the four AP1 sites. Therefore, if sites outside this fragment, such as AP1, are acting as repressors, this construct should be inducible by dex. To ensure that Ad2MLP can support dex inducibility, a fragment of the URR of HPV18 that contains the GRE (nt 7620-60) was placed under the control of this promoter. The dex treatment of 9E cells transfected with this construct resulted in a threefold induction in activity, indicating that constructs utilizing Ad2MLP can be used to study dex inducibility (data not shown). As seen in Fig. 5, pAd31URR7170-7440 is still only weakly inducible by dex treatment. We created LSMs of the pAd31URR7170-7440 construct in single (pAdLSMGRE1, pAdLSMGRE2, and pAdLSMGRE3), double (pAdLSMGRE1,2, pAdLSMGRE2,3, and pAdLSMGRE1,3), and triple (pAdLSMGRE1,2,3) combinations to determine whether any of the GREs were themselves acting as repressors within this 270-bp fragment. Loss of any or all of the GREs in pAd31URR7170-7440 did not result in an induction of activity by dex (Fig. 5), further demonstrating the weak functionality of the HPV31 GREs.

FIG. 4.

Transcriptional activation of the HPV31 URR and GRE LSM constructs. Schematic diagrams of constructs and fold change values are shown for each construct. Activation of pGL2-18URR is included for comparison. Wild-type HPV31 URR or GRE LSMs created in single, double, or triple combinations were transfected into 9E cells, treated with 0.1% EtOH (veh; open bars) or 10⁻⁶ M dex (solid bars), and assayed for luciferase activity. The relative luciferase activity is shown as a fold change over that of pGL2-31URR under vehicle conditions, which was set to 1.

FIG. 5.

dex-induced activation of the 270-bp internal fragment of the HPV31 URR and GRE LSM constructs. Schematic diagrams of constructs and fold change values are shown for each construct. This 270-bp fragment (nt 7170 to 7440), contains all three GREs under the control of Ad2MLP. Wild-type or single, double, or triple GRE LSMs were transfected into 9E cells, treated with 0.1% EtOH (veh; open bars) or 10⁻⁶ M dex (solid bars), and assayed for luciferase activity. The relative luciferase activity is shown as the fold change over that of pAd31URR7170-7440 under vehicle conditions, which was set to 1.

HPV18 GRE function is abrogated when placed in the URR of HPV31.

Thus far, we have demonstrated that, while the three GREs we have identified appear to be functional based on gel shift analyses (Fig. 2), the URR of HPV31 is only weakly inducible by dex (Fig. 4). Furthermore, none of the GREs appears to be acting as a repressor to prevent the other GREs or unknown glucocorticoid-inducible sites from functioning, either in the context of the whole URR (Fig. 4) or within the 270-bp fragment that contains all three GREs (Fig. 5). To more closely examine why the three GREs in the URR of HPV31 do not confer dex inducibility when the gel shift data indicate positive functionality, we sought to determine whether the context of the URR of HPV31 plays a role in the weak inducibility of the URR of HPV31 by dex. We chose to replace each HPV31 GRE with 18GRE within the context of the entire URR (within the pGL2-31URR construct) or within the 270-bp internal fragment (within the pAd31URR7170-7440 construct). We have previously shown that the GRE of HPV18 is functional in our system (Fig. 3). Therefore, if the overall context of the URR of HPV31 is preventing transcriptional activation by dex, placement of the HPV18 GRE in place of the HPV31 GREs should not alter dex inducibility of the URR of HPV31. On the other hand, if the context of the URR plays no role in the lack of dex inducibility but in fact the three GREs identified in the URR of HPV31 are simply nonfunctional or negative elements, replacement with the HPV18 GRE should result in a similar dex induction as is seen with HPV18 URR. Using single LSMs of pGL2-31URR (p31LSMGRE1, p31LSMGRE2, and p31LSMGRE3) and pAd31URR7170-7440 (pAdLSMGRE1, pAdLSMGRE2, and pAdLSMGRE3), we replaced the unique polylinker with an 18GRE oligonucleotide engineered with NdeI and BclI ends (see Materials and Methods). The resulting 18GRE replacements of pGL2-31URR (p31GRE1/18GRE, p31GRE2/18GRE, and p31GRE3/18GRE) and pAd31URR7170-7440 (pAdGRE1/18GRE, pAdGRE2/18GRE, and pAdGRE3/18GRE) were transfected into 9E cells. Replacement of each of the HPV31 GREs within the context of the entire URR with the HPV18 GRE had no effect on luciferase activity upon dex treatment (Fig. 6A), nor did replacement within the 270-bp fragment construct (Fig. 6B). These data indicate that the overall context of the URR of HPV31 plays a role in glucocorticoid inducibility.

FIG. 6.

dex inducibility of HPV31 URR constructs which replace the GREs of HPV31 with the HPV18 GRE, either in the full-length construct (A) or in the 270-bp internal fragment construct (B). Wild-type, single LSMs, or single 18GRE replacements were transfected into 9E cells, treated with 0.1% EtOH (veh; open bars) or 10⁻⁶ M dex (solid bars), and assayed for luciferase activity. The relative luciferase activity is shown as the fold change over that of pGL2-31URR (A) or pAd31URR7170-7440 (B) under vehicle conditions, which was set to 1. The fold change of activity upon dex treatment is shown for each construct.

Another possibility is that within the context of the whole URR of HPV31, the location of the GREs in the 5′ region may be responsible for the inability of the HPV18 GRE to function at those positions, since the GRE of HPV18 is located close to the major early promoter, p105, in the URR (Fig. 1B). Therefore, we chose to place the GRE of HPV18 in the 3′ end of the HPV31 URR at a location similar to the position in the HPV18 URR. We first created an LSM in the 3′ end of the HPV31 URR, replacing nt 7883 to 7990 with the unique polylinker NdeI-ApaI-BclI. This location falls between the second E2 binding site (E2BS#2; nt 7868 to 7879) and the E1 binding site (E1BS; nt 7903 to 7910), which corresponds to the exact location of the HPV18 GRE in the HPV18 URR (Fig. 1A and B). This construct, 31URR-3′LSM, showed no change in either basal activity or activity upon dex treatment compared to the wild-type pGL2-31URR construct (Fig. 7). We then replaced the polylinker with the HPV18 GRE, creating 31URR-3′LSM(18GRE). If the location of the HPV31 GREs in the 5′ region of the URR is the reason for the lack of dex inducibility and for why the HPV18 GRE is not functional at those sites, then 31URR-3′LSM(18GRE) should be dex inducible. Upon transfection into 9E cells, we found that placement of the HPV18 GRE in a location in the HPV31 URR comparable to its native location in the HPV18 URR was unable to confer transcriptional activity upon dex induction (Fig. 7). To rule out any inhibition or repression of dex induction by the HPV18 GRE in the 3′ end of the HPV31 URR by the HPV31 GREs in the 5′ region, we replaced the GREs of HPV31 with the unique polylinker in 31URR-3′LSM(18GRE). None of the resulting constructs tested were found to be inducible by dex [see p31LSMGRE1,2,3-3′LSM(18GRE) in Fig. 7; other data not shown], even with the HPV18 GRE in a location known to be functional in the HPV18 URR. These data suggest that GREs, either from HPV31 or HPV18, are not dex inducible in the URR of HPV31 due to the overall context of the URR.

FIG. 7.

Activation of various GRE replacement constructs upon dex treatment. Schematic diagrams of constructs and fold change values are shown for each construct. See Materials and Methods for a description of each construct. Constructs were transfected into 9E cells, treated with 0.1% EtOH (veh; open bars) or 10⁻⁶ M dex (solid bars), and assayed for luciferase activity. The relative luciferase activity is shown as the fold change over that of pGL2-31URR under vehicle conditions, which was set to 1.

In order to determine whether we could identify repressive elements or regions of the URR of HPV31 responsible for the lack of dex inducibility, we created 5′ deletion constructs of the HPV31 URR, either wild type or with the HPV18 GRE located in the 3′ end of the URR. As shown in Fig. 8, none of the 5′ deletions resulted in a strong dex induction, including a 5′ deletion that contains only the minimal origin and promoter of the HPV31 URR (p31URRΔ6921-7799). Furthermore, the 5′ deletions that contained the HPV18 GRE in a 3′ location comparable to its own location in the HPV18 URR were not inducible by dex as well (Fig. 8). These data suggest that the promoter region of the URR of HPV31, including the major early promoter p99, is responsible for the lack of a strong dex induction.

FIG. 8.

Effect of dex treatment on the activation of 5′ deletion constructs of the HPV31 URR. Schematic diagrams of constructs and fold change values are shown for each construct. 5′ deletion constructs or constructs with placement of 18GRE in the 3′ region of the HPV31 URR were transfected into 9E cells, treated with 0.1% EtOH (veh; open bars) or 10⁻⁶ M dex (solid bars), and assayed for luciferase activity. The relative luciferase activity is shown as the fold change over that of pGL2-31URR vehicle conditions, which was set to 1.

GREs of HPV31 are weakly inducible in the URR of HPV18.

To definitively address the question of whether the three GREs we have identified in the URR of HPV31 that appear to be functional by EMSA are truly functional or not, we replaced the GRE of HPV18 with each of the GREs from HPV31 in the URR of HPV18. We have shown that the context of the HPV18 URR and location of the GRE of HPV18 allow for an induction of the basal activity upon dex treatment (Fig. 3). Therefore, if any or all of the HPV31 GREs are truly functional but are being repressed either by other cis factors in the URR of HPV31 or by their location in the 5′ end of the URR, their placement in the HPV18 URR would result in an induction of activity upon dex treatment. We replaced the unique polylinker in the p18LSMGRE construct with each of the HPV31 GREs [p18URRLSM(31GRE1), p18URRLSM(31GRE2), and p18URRLSM(31GRE3)]. We tested whether any or all of these constructs would result in an induction of basal activity upon dex treatment over the p18LSMGRE construct, which itself showed a two- to threefold induction (Fig. 3 and 7). We first replaced the polylinker of p18LSMGRE with the HPV18 GRE to ensure that the extra nucleotides created by the polylinker had no effect on dex inducibility. This construct resulted in a dex induction comparable to pGL2-18URR, indicating that we could indeed determine functionality of the HPV31 GREs by using these constructs (data not shown). As seen in Fig. 7, none of the constructs were able to significantly induce the activity of the HPV18 URR over the p18LSMGRE construct (2.3-fold versus 2.6- to 3.2-fold), indicating that the three GREs we identified in the URR of HPV31 are unable to support a glucocorticoid induction in transcriptional activity.

DISCUSSION

Steroid hormones have been suggested to act as cofactors for HPV in the development of cervical cancer. Glucocorticoids in particular, such as the synthetic glucocorticoid dex, have been shown to enhance the transformation rate of high-risk HPV-transfected cells (16, 44, 54), suggesting the oncogenic potential of glucocorticoids in conjunction with HPV. GREs have been identified in the URRs of HPV11, HPV16, and HPV18 and have been suggested in others, including that of HPV31 (7, 9, 11, 20, 31, 40, 44, 72). In the present study we attempted to show dex inducibility of the HPV31 URR. By sequence analysis we identified three GREs in the URR of HPV31 located within the 5′ URR domain. Gel shift analyses indicated that these three potential GREs were functional elements, as all could bind purified GR and could compete off a complex bound to the consensus GRE (Fig. 2B and C). However, we were surprised to find that the URR of HPV31 was only weakly inducible by dex in monolayer 9E cell cultures (Fig. 4). To rule out the possibility that the lack of strong inducibility was due to the cell line used, the method of transfection, or the treatment, we tested the URR of HPV18, which has been shown by others to contain a functional GRE (7, 9, 40). The HPV18 URR was inducible by dex in our system as well, and we were able to show that this inducibility was due in part to the GRE (Fig. 3). Therefore, these data suggest two possiblities: (i) that there is something intrinsic to the HPV31 URR that prevents dex inducibility or (ii) that the three GREs we identified are in fact not functional elements and the URR of HPV31 is not responsive to dex.

We thought perhaps a possible reason for the lack of strong inducibility by dex on the URR of HPV31 was the presence of a repressor within the URR, either one of the GREs, or another transcription factor binding site. The presence of nGREs has been reported for the proliferin gene, the corticotropic-releasing hormone gene, and the prolactin gene (13, 38, 58). Studies have shown that GR can bind to nGREs, but binding results in transcriptional repression rather than activation upon dex treatment (13). DNA sequence is not the sole determinant of negative regulation, and nGREs can also act as positive regulators, depending on the physiological context (58). We found that the three GREs we identified in the URR of HPV31 were in fact not acting as repressors, either on their own or in conjunction with each other, since loss of any combination of the GREs did not result in a strong dex induction. This was true for both the full-length URR and the 270-bp internal fragment of the URR that contains all three GREs. This did not rule out the possibility of other transcription factor binding sites acting as repressors to prevent any or all of the three GREs from mediating a dex response. For example, GR is inhibited by, and in turn inhibits, the transcriptional activation of the well-known transcription factor AP1 (29, 60, 73). Interestingly, this mutual antagonism does not seem to be due to an inhibition of DNA binding but results from a direct protein-protein interaction between GR and AP1 (29, 60, 73). Even so, we noticed that the URR of HPV31 contains four AP1 sites, whereas the URR of HPV18 contains only two (Fig. 1A, B) (7, 19, 33). These AP1 sites are critical for the basal activity of the URR, and we and others have shown that mutation of these sites results in a repression of the basal activity of the URR of HPV31 (34, 61). To rule out a repressive effect of any other transcription factor binding sites on the URR, we analyzed the dex inducibility of the 270-bp internal fragment of the URR that contains all three GREs but contains none of the AP1-binding sites. We discovered that this fragment was still only weakly inducible by dex, suggesting that other sites are not acting as repressive elements to inhibit a dex induction by the GREs in the HPV31 URR unless they are contained within this 270-bp region of the URR. Although these results do not discount direct protein-protein interactions between GR and other factors such as AP1, it seems unlikely that another transcription factor would be directly inhibiting GR from acting on the HPV31 GREs while allowing GR to upregulate HPV18 URR transcriptional activity through its GRE.

Location of the GREs in the URR could also play a role in the strength of steroid hormone inducibility. The GREs we identified in the URR of HPV31 are all located within the 5′ URR domain. This region has been shown to be nonessential for viral replication and differentiation-dependent DNA amplification (25). In fact, studies suggest a negative regulatory effect for this region of the URR (25, 30, 55). The GRE of HPV18, on the other hand, is located very close to the major early promoter, p105 (7, 40). Steroid hormone response elements are generally considered to act in a distance- and orientation-independent manner, sometimes functioning as far as 1 kb upstream or even downstream of an intended promoter (17, 28). Since the HPV18 GRE was unable to induce a dex response in the HPV31 URR in any of the three HPV31 GRE locations, it seemed plausible to assume that the GREs of HPV31 were located in a region of the URR that prevented a dex response. However, placement of the HPV18 GRE in a 3′ location in the HPV31 URR that corresponds to its native location in the URR of HPV18 also failed to result in a dex induction. This indicates that location is not the reason for the lack of a strong dex induction but that the overall context of the URR of HPV31 may somehow prevent dex inducibility.

While our data thus far did not rule out the possibility that the lack of dex inducibility of the HPV31 URR was due to the overall context of the URR, it was important to determine whether the three GREs identified in the URR of HPV31 were functional outside the context of the URR of HPV31. We placed each of the three HPV31 GREs in a context we knew to allow for dex inducibility: the HPV18 URR. Since the HPV18 GRE is able to mediate a dex response in the context of its URR within its specific location in the URR, placement of the HPV31 GREs in that location should indicate whether the HPV31 GREs are functional or not. We found that the three HPV31 GREs mediated only a weak response to dex when located in the HPV18 URR in place of the HPV18 GRE. These data indicate that the three GREs of HPV31 are in fact unable to mediate a strong dex response in monolayer culture, even though GR can bind these sites.

HPV31 is considered an intermediate-risk virus that is occasionally associated with cervical cancer compared to high-risk types, such as HPV16 and HPV18, which are frequently associated with cervical cancer (75, 76). A major determinant thought to be involved in the risk status of genital HPVs is the functionality of the viral oncogenes E6 and E7. E6 has been shown to bind p53 and target it for ubiquitin-mediated degradation (32, 70), while E7 is known to bind Rb and dissociate it from E2F (32). High risk E6 and E7 bind their respective targets with higher affinity than do E6 and E7 of low-risk HPV types (18, 46, 70). Furthermore, there is a different mechanism in regulating the expression of E6 and E7 between high-risk and low-risk types. High-risk HPV types transcribe polycistronic E6 and E7 mRNAs from one major promoter, whereas the oncogenes are transcribed from two separate promoters in low-risk HPV (15, 47, 62). HPV31 is considered to have high-risk-like oncogenes based on sequence similarities, the presence of motifs important for function, and the same mechanism of gene expression (22, 64). Since cervical cancer is thought to arise from other cofactors in addition to a high-risk HPV infection, the lack of glucocorticoid inducibility could be one reason to explain why HPV31 is considered an intermediate-risk virus. Furthermore, HPV11, a low-risk type, although it contains a functional GRE, contains E6 and E7 that bind p53 and Rb, respectively, with lower affinity than do high-risk E6 and E7 and are transcribed from independent promoters (15, 18, 46, 47, 62). Therefore, it is possible that functional GREs and oncogenes that bind targets with high affinity are two factors necessary for an HPV type to be considered a high-risk type.

The cell line we chose for the present study, CIN612-9E (9E), is ideal in that it maintains episomal HPV31b and produces infectious virus in organotypic (raft) cultures (42). Therefore, the 9E cell line could be used to study the effect of steroid hormones on the complete life cycle of HPV31 in addition to demonstrating the inducibility of the URR in monolayer cell culture. HPV is an epitheliotropic virus and depends on the differentiation of that epithelium for its complete viral life cycle. The fact that the URR of HPV31 was not strongly inducible by dex in monolayer culture could indicate that differentiation is necessary for transcriptional activation of the URR of HPV31 by steroid hormones. We are currently pursuing this idea and have preliminary data to suggest that GRE functionality in the HPV31 URR is altered upon differentiation, indicating the importance of studying HPV in the context of its complete life cycle (unpublished data).

Why would the URR of HPV18 be inducible by dex in 9E monolayer cell culture, whereas the URR of HPV31 is not, even though the GREs identified in the URR of HPV31 appear to be comparable to the GRE of HPV18, both by sequence and gel shift analysis? The presence of HPV31b in the 9E cell line is a potential reason. The endogenous HPV31b and the exogenous HPV31 URR construct could be competing for similar factors necessary for glucocorticoid-dependent transcriptional activation that the URR of HPV18 either does not need or can more strongly compete away from the endogenous HPV31b. A glucocorticoid-mediated response is probably due to other factors in addition to GR, as evidenced by the fact that GREs can act as negative or positive regulators, depending on the cell type. For HPV, dex inducibility has also been shown to be a cell-specific response. The HPV18 GRE is not inducible in C33 cells without exogenous GR (40), whereas the HPV11 GRE is only weakly inducible in laryngeal keratinocytes but strongly inducible in the HeLa cell line (14). It is clear that other factors contribute to transcriptional regulation by steroid hormones, and it is likely that HPV31 and HPV18 utilize different factors based on the context of their respective URRs.

Although the basal activities of the HPV18 URR and the HPV31 URR constructs are comparable in our system and in others (33, 57), the early promoters of these HPV types may have different requirements for both basal activity and dex-inducible activity. Our 5′ deletion data indicated that removal of all but the minimal origin and promoter of the HPV31 URR (p31URRΔ6921-7799) results in a strong enhancement in basal activity compared to the full-length URR, suggesting the presence of a repressor in the region between nt 7557 and 7799. Surprisingly, the basal activity of p31URRΔ6921-7799-3′LSM(18GRE), where the HPV18 GRE was placed in between E2BS#2 and E1BS (the 3′ location corresponding to the position of the HPV18 GRE in the HPV18 URR), was greatly repressed and was not inducible by dex. In the full-length HPV31 URR construct, placement of the HPV18 GRE in this 3′ position had no effect on either the basal activity or on dex inducibility. Therefore, it is intriguing that, in the p31URRΔ6921-7799 construct, the HPV18 GRE could be a transcriptional repressor of the HPV31 URR and its major early promoter p99. The transcription factor binding sites that are lost within the region from nt 7557 to 7799 include two AP1 sites and an Oct-1 site. There are reports that Oct-1 acts as a negative regulator in the URR of HPV18, as well as the URR of HPV31 (24, 34, 53). This could explain why the basal activity of the 5′ deletion mutant p31URRΔ6921-7799 is upregulated compared to the full-length HPV31 URR and the other 5′ deletion mutants. Furthermore, Oct-1 has been shown to act synergistically with GREs to promote dex-mediated transcriptional activation, although mutation of the Oct-1-binding site had no effect on basal activity (6, 59). Interestingly, another study reported that an HPV18 URR 5′ deletion mutant almost identical to p31URRΔ6921-7799 had no enhancer activity when placed in front of the thymidine kinase promoter compared to the full-length HPV18 URR (19). Therefore, repression of the basal activity of p31URRΔ6921-7799-3′LSM(18GRE) indicates that placement of the HPV18 GRE in the 3′ location of p31URRΔ6921-7799 resulted in the activity of p99 in HPV31 to resemble that of p105 in HPV18. This suggests that in fact the major early promoters of different HPV types have different requirements for transcriptional activation, which could explain why the URR of HPV18 is dex inducible, whereas the URR of HPV31 is not, even with the presence of GREs in both URRs.

In summary, the URR of HPV31 is only weakly inducible by the synthetic glucocorticoid dex in monolayer cell culture. Furthermore, not only are the three GREs identified themselves only weakly functional, but the URR of HPV31 is able to repress functional GREs as well, suggesting that the context of the URR plays an important role in transcriptional regulation by steroid hormones.