Human Papillomavirus Type 16 E2 Protein Transcriptionally Activates the Promoter of a Key Cellular Splicing Factor, SF2/ASF

Abstract

Human papillomavirus (HPV) gene expression is regulated in concert with the epithelial differentiation program. In particular, expression of the virus capsid proteins L1 and L2 is tightly restricted to differentiated epithelial cells. For HPV16, the capsid proteins are encoded by 13 structurally different mRNAs that are produced by extensive alternative splicing. Previously, we demonstrated that upon epithelial differentiation, HPV16 infection upregulates hnRNP A1 and SF2/ASF, both key factors in alternative splicing regulation. Here we cloned a 1-kb region upstream of and including the transcriptional start site of the SF2ASF gene and used it in in vivo transcription assays to demonstrate that the HPV16 E2 transcription factor transactivates the SF2/ASF promoter. The transactivation domain but not the DNA binding domain of the protein is necessary for this. Active E2 association with the promoter was demonstrated using chromatin immunoprecipitation assays. Electrophoretic mobility shift assays indicated that E2 interacted with a region 482 to 684 bp upstream of the transcription initiation site in vitro. This is the first time that HPV16 E2 has been shown to regulate cellular gene expression and the first report of viral regulation of expression of an RNA processing factor. Such E2-mediated control during differentiation of infected epithelial cells may facilitate late capsid protein expression and completion of the virus life cycle.

Human papillomavirus type 16 (HPV16) infects cervical epithelial cells, causing mainly benign lesions (cervical dysplasia). However, in some rare cases upon persistent infection, lesions can progress to cervical cancer (66). The life cycle of this 7.9-kb double-stranded DNA virus is highly dependent upon the differentiation status of the epithelial tissue it infects. Of particular importance is restriction of expression of the highly immunogenic capsid proteins L1 and L2 to the most differentiated cells of the structure, where immune surveillance is low. However, while L1 and L2 RNAs have been detected in less-differentiated epithelial cells (5, 60), fully processed messenger RNAs (mRNAs) are found only in differentiated epithelial cells (42), and capsid protein expression is restricted to the granular layer cells (48). Thus, expression of the virus capsid proteins is regulated at least partly at posttranscriptional levels (20). cis-Acting RNA regulatory motifs have been identified in the HPV16 genome that may regulate capsid protein expression at one or more posttranscriptional levels (10, 11, 12, 29, 45, 50, 55, 65). The elements are proposed to act via interactions with cellular RNA splicing factors, including U1 snRNP, hnRNP A1, SF2/ASF, PTB hnRNP C1/C2, and CUG-BP1 (9, 10, 12, 19, 30, 41, 56, 64). Such RNA-protein interactions are proposed to be essential in regulating virus late gene expression, for example by regulating alternative splicing and polyadenylation of virus late mRNAs (20).

There are two key protein families involved in regulation of alternative splicing in human cells. These are the SR (serine/arginine-rich) and hnRNP protein families. SF2/ASF is the key SR splicing protein (7), and hnRNP A1 is the main antagonistic counterpart of SF2/ASF (15). SF2/ASF has roles in constitutive splicing by bringing U1 snRNP to the 5′ splice site and by bridging spliceosome complexes across exons and introns. This leads to efficient recognition of intron/exon junctions and determination of RNA splicing patterns. In relation to this, SF2/ASF has extremely well-documented roles in alternative splicing by binding intronic and exonic sequence enhancers to modulate the efficiency of recognition of upstream and downstream splice sites by the splicing machinery. The binding pattern of SR and hnRNP proteins has been termed the “splicing code,” as it determines which mRNA isoforms and therefore proteins are ultimately expressed from a gene.

Surprisingly, our recent studies indicated that 13 different late mRNAs encode the virus capsid proteins (42) in differentiated W12 cells, cervical epithelial cells that contain nuclear episomal copies of the HPV16 genome (57). These mRNAs that are the products of extensive alternative splicing of late pre-mRNAs also encode E6/E7 isoforms, E1^E4 and E5. Overlapping mRNAs may be required to ensure that each open reading frame is first in one of the mRNAs to ensure efficient translation of each protein (31). Previously, comparing undifferentiated and differentiated cells, we showed that abundance of SF2/ASF increased approximately four- to eightfold in differentiated W12 cells (41) and in cervical tissue isolated from patients with HPV16-positive, low-grade cervical lesions (unpublished data). In contrast, SF2/ASF levels diminished in uninfected differentiated epithelial cells such as HaCaT, a spontaneously immortalized epithelial cell line, and normal human keratinocytes (41). This indicated that HPV16 infection upregulated SF2/ASF expression in a differentiation stage-specific manner. Furthermore, SF2/ASF was found to be upregulated in U2OS cells stably expressing HPV16 E2, the virus transcription factor, suggesting it could transcriptionally regulate SF2/ASF (41).

E2 is a typical transcription factor with a DNA binding and a transactivation domain separated by a flexible hinge region (22). Normally, it regulates the virus early promoter in the long control region (LCR) via four binding sites (ACCN₆GGT), which differ in E2 affinity (6). In addition to viral regulatory activity, transactivation of cellular promoters has been shown by a number of bovine papillomavirus (BPV) and HPV E2 molecules. For example, HPV18 E2 interaction with Sp1 is involved in transcriptional downregulation of the cellular hTERT promoter (34), whereas both HPV8 and HPV18 E2s transactivate p21 expression via Sp1 interactions (59). Interaction of HPV18 E2 with C/EBP transcription factors transactivates the involucrin promoter (21). HPV8 E2 has also been reported to downregulate β4-integrin expression by causing displacement of one or more important cellular factors from its promoter (46). Coactivators CBP/p300 and p/CAF, which regulate gene expression via indirect DNA association, also interact with a number of papillomavirus E2 proteins and, in the case of HPV18 and HPV8, synergize transcription (33, 35). Some papillomavirus E2 proteins have been shown to interact with components of the basal transcription machinery to regulate preinitiation complex formation, which may affect both viral gene expression, via the LCR, and cellular promoters. However, as yet no cellular genes controlled by HPV16 E2 have been reported.

To demonstrate a function for E2 in HPV16-mediated regulation of the key SR protein, SF2/ASF, we cloned the SF2/ASF putative promoter region into a reporter gene construct and conducted transactivation assays. The SF2/ASF promoter was transactivated in E2-expressing cells. This was dependent on the amount of E2 expressed in the cells. The transactivation domain of E2 was required for transcription stimulation but not the DNA binding domain. Furthermore, a transactivation-dead E2 expressed from HPV31 genomes transfected into human foreskin keratinocytes reduced SF2/ASF levels. Association of E2 with the promoter was proven in vivo, using chromatin immunoprecipitation (ChIP) assays, and in vitro, using electrophoretic mobility shift assays (EMSAs). However, specific competitor DNA could not fully compete off the E2 complex observed in EMSA, suggesting E2 may be tethered to the promoter via another transcription factor, such as Sp1, which binds the same region of the SF2/ASF promoter. This novel regulation of a key RNA processing factor by HPV E2 is significant because cellular and viral gene expression is controlled to a large extent by RNA processing events.

MATERIALS AND METHODS

Cell lines.

HeLa, HaCaT, and U2OS cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 2 mM l-glutamine, 50 U/ml penicillin-50 μg/ml streptomycin. G418 (500 μg/ml) was added to medium when culturing U2OS A4 and B1 cell clones stably expressing HPV16 E2 (41). A differentiation-competent subclone of W12 cells (clone 20863) containing only episomal copies of the HPV16 genome (27) and CIN612 9E cells containing mainly episomal copies of the HPV31 genome (2) were grown as described previously (42).

Plasmids and DNA transfections.

Wild-type HPV16 E2 DNA from nucleotides (nt) 2735 to 3852 in the HPV16 genome was cloned from pCMVE2 into pBluescript by blunt end cloning and then into the XbaI/SalI sites of pCI-neo (Promega) to produce pCI-E2. pCMV-E2, pCMV-R37AE2, and pWEB-E2(DBDmut) are described elsewhere (63). pET15b;E2(16) (a gift of Julie Burns, University of York) was used to in vitro purify His-tagged, full-length HPV16 E2 from Escherichia coli BL21(DE3)plysS. The SF2/ASF promoter was PCR amplified from HeLa cell genomic DNA from nt −1024 to 25 of the transcriptional start site, using primers SF2pr −1024 F and SF2pr +5 R (Table 1), and cloned into KpnI/SacI sites of pGCAT3-B (modified from pGL3-B [Promega] to replace the luciferase gene with a chloramphenicol acetyltransferase [CAT] reporter gene) to produce pGCAT3-B-SF2pr. Cells were transfected in 24-well plates using Lipofectamine 2000 (Invitrogen), following the manufacturer's instructions. Cotransfection with pCMV-βgal was used to control for transfection efficiency. After a 24-h incubation, cells were scraped into 30 μl 0.25 M Tris, pH 7.5. Cell lysates were obtained by being freeze-thawed three times before cell debris was pelleted by centrifugation. The supernatant was harvested and stored at −20°C.

TABLE 1.

Primer sequences used for cloning, RT-PCR, ChIP assays, and EMSAs^a

Primer	Sequence	Application
SF2pr −1024F	GTTACGGTTCTCACATCCATTTTCG	Cloning of SF2pr EMSA probe 1
SF2pr −829 R	TGACACTAGTGTGGGACTAGCGCAGAATGC	EMSA probe 1
SF2pr −852 F	TGACACTAGTCTTCATTAAGTACCGTTCC	EMSA probe 2
SF2pr −650 R	CAGGCATTCTGCGCTAGTC	EMSA probe 2
SF2pr −684 F	TGACACTAGTGTTCTGACGAGAAGGCGGAAC	EMSA probe 3
SF2pr −482 R	GAAACCCGGGTATCTTCGTAG	EMSA probe 3
SF2pr −524 F	TGACACTAGTCAGCTCTGGATTAGACGCAC	ChIP assays, EMSA probe 4
SF2pr −325 R	GCAATGAGGATCTTTGAAAGC	ChIP assays, EMSA probe 4
SF2pr −402 F	ATGGTGGGACAACGCTTTAG	Endogenous ChIP assays
SF2pr −222 R	TTTGCGAACAGAGTGACCAG	Endogenous ChIP assays
SF2pr −364 F	TGACACTAGTCTTTGAAGGCGCCGAGTTGC	EMSA probe 5
SF2pr −162 R	GACGTCACCCTCCCCACGAAG	EMSA probe 5
SF2pr −199 F	TGACACTAGTGGGACTTTTTTACCCCCTTC	EMSA probe 6
SF2pr +5R	CTCCCGCGGCCCCTCCAAAATG	Cloning of SF2pr EMSA probe 6
CAT5′ F	AATCACTGGATATACCACCGTTGA	ChIP assays
CAT5′ R	TGAACGGTCTGGTTATAGGTAC	ChIP assays
5S RNA F	GGCCATACCACCCTGAACGC	ChIP assays
5S RNA R	CAGCACCCGGTATTCCCAGG	ChIP assays
SF2 F	CTCATCATTCCCCAGAAACC	ChIP assays, RT-PCR
SF2 R	GGCAGGAATCCACTCCTATG	ChIP assays, RT-PCR
GAPDH F	TCCACCACCCTGTTGCTGTA	RT-PCR
GAPDH R	ACCACAGTCCATGCCATCAC	RT-PCR
CAT3′ F	CTGGCCTATTTCCCTAAAGG	RT-PCR
CAT3′ R	CAAACGGCATGATGAACCTG	RT-PCR

^aRestriction enzyme sites are underlined. F; forward primer, R; reverse primer.

CAT assays.

One to 20 μl of lysate was added to a reaction mixture of 625 μM acetyl coenzyme A (Sigma), 0.05 μCi [¹⁴C]chloramphenicol, and 0.25 M Tris-HCl, pH 7.5, in a final volume of 40 μl and incubated at 37°C for 1 h. Two hundred microliters of ethyl acetate was added, and the solution was vortexed for 10 s before centrifugation at 13,000 rpm for 5 min at room temperature. The upper organic layer was transferred to a fresh tube and dried in an Eppendorf Concentrator 5301. Following resuspension in 25 μl ethyl acetate, the solution was applied to a thin-layer chromatography plate (Sigma) and samples were separated in 95% chloroform, 5% methanol. Plates were visualized by phosphorimaging (Bio-Rad) and were analyzed using QuantityOne software.

EMSAs.

Probe DNA was PCR amplified from pGCAT3-B-SF2pr using SF2pr −684 F and SF2pr −482 R primers (Table 1) and digested with SpeI. DNA was purified from acrylamide gels by elution in 0.5 M NaCl, 1 mM EDTA at 4°C overnight, followed by ethanol precipitation. DNA was resuspended in 30 μl distilled water and radiolabeled by incubation in 10 mM Tris-HCl, pH 7.9; 50 mM NaCl; 10 mM MgCl₂; 1 mM dithiothreitol; 3.3 μM dATP, dGTP, and dTTP; 2.5 U DNA polymerase I; large (Klenow) fragment (NEB); and 10 μCi [α-³²P]dCTP. Reactions were carried out at room temperature for 20 min, and then 3.3 μM dCTP was added, followed by incubation for a further 5 min at room temperature. The probe was purified using a mini Quick Spin column (Roche), following the manufacturer's instructions. EMSAs were performed as described previously (3).

ChIP.

ChIP assays were performed as described previously (18) using 1 ml anti-TATA-binding protein (anti-TBP) (4C8/26; a gift of R. J. White, Beatson Institute, Glasgow, United Kingdom), 5 μl anti-HPV16 E2SCT (a gift of J. Burns, University of York, United Kingdom), 5 μl anti-HPV16 E2 polyclonal antibody (E2P; a gift of L. Banks, Trieste, Italy) or 5 μl anti-involucrin (clone SY5; Sigma). The 5′ region of the CAT gene was amplified using CAT5′ F and CAT5′ R. The SF2/ASF promoter region was amplified using either SF2pr −524 F and SF2pr −325 R, for the plasmid-contained promoter, or SF2pr −402 F and SF2pr −222 R, for the endogenous promoter. The 5S RNA promoter was amplified using 5S F and 5S R. The SF2/ASF 3′ coding sequence was amplified using SF2/ASF F and SF2/ASF R. Primer sequences are shown in Table 1. Amplification conditions were 94°C for the initial denaturation, followed by 25 cycles of 94°C for 30 s, 55°C (SF2/ASF promoter and coding sequence), 58°C (5S RNA), or 60°C (CAT5′) for 30 s, 72°C for 30 s, and finally 10 min at 72°C.

Western blotting.

Cells were lysed in 12.5 mM Tris-HCl, pH 6.8, 0.4% (wt/vol) sodium dodecyl sulfate (SDS), 143 mM (vol/vol) β-mercaptoethanol, 0.06% (wt/vol) bromophenol blue, 4% (vol/vol) glycerol and stored at −20°C. Following resolution by SDS-polyacrylamide gel electrophoresis (SDS-PAGE), proteins were transferred to polyvinylidene difluoride membrane using a Bio-Rad Trans-Blot system. Electroblotting was performed with 250 mA for 2 h or 100 mA overnight at 4°C. Blocking of the membrane was carried out in phosphate-buffered saline-Tween 20 with 5% dried skimmed milk (PBS-T, 5% milk) at room temperature for 1 h or at 4°C overnight. The membrane was then incubated in the appropriate dilution of primary antibody in PBS-T, 5% milk for 1 h at room temperature or 4°C overnight. The antibodies used were anti-SF2/ASF (clone 96; Zymed Laboratories) at a dilution of 1:500, anti-GAPDH (glyceraldehyde-3-phosphate dehydrogenase, 6CS; Biodesign International) at a dilution of 1:2,000, and TVG261 against HPV16 E2 at a dilution of 1:250. Following washing with PBS-T, the membrane was incubated in secondary antibody, either horseradish peroxidase-conjugated anti-mouse or anti-rabbit immunoglobulin G at a dilution of 1:1,000 in PBS-T, 5% milk for 1 h at room temperature. The membrane was then washed in PBS-T and subjected to enhanced chemiluminescence using an enhanced chemiluminescence system (Pierce).

Reverse transcriptase-PCR (RT-PCR).

Total RNA was prepared from W12 cells using Trizol reagent (Invitrogen) according to the manufacturer's instructions. RNA was digested with DNase I to remove any residual DNA and reverse transcribed using Superscript II (Invitrogen) as previously described (41). cDNA was subjected to PCR to simultaneously amplify SF2/ASF and GAPDH mRNAs using SF2 F and SF2 R with GAPDH F and GAPDH R. Amplification conditions were 94°C for the initial denaturation, followed by 25 cycles of 94°C for 30 s, 50°C for 30 s, 72°C for 30 s, and finally 10 min at 72°C.

RESULTS

HPV16 E2 transactivates the SF2/ASF promoter.

We showed previously that SF2/ASF was upregulated four- to eightfold in differentiated W12 cells compared to undifferentiated W12 cells (41), where only episomal virus genomes were present (clone 20863) (27). Although levels of the protein were somewhat higher in undifferentiated HaCaT cells than in undifferentiated W12 cells, neither HaCaT nor normal human keratinocytes showed upregulation of SF2/ASF upon differentiation. In addition, a clone of W12 cells with only integrated genomes (20861) (27) showed downregulation of this splicing factor upon differentiation. These data suggested that the presence of the HPV16 genome in episomal form was responsible for the increased levels of SF2/ASF (41). We hypothesized that the HPV16 E2 transcription factor could mediate this regulation because it is expressed from episomal but not integrated virus genomes and because E2 levels have been observed to increase in the upper layers of low-grade cervical lesion tissues (40). To determine if the upregulation of SF2/ASF in W12 20863 cells was achieved transcriptionally, RT-PCR was used to detect levels of mRNAs encoding SF2/ASF in undifferentiated and differentiated cells. Figure 1A shows that levels of SF2/ASF RNA were high in differentiated W12 cells compared to those in undifferentiated W12 cells. GAPDH mRNA levels did not change significantly in W12 cells upon differentiation. No PCR products were observed in the absence of RT, as expected. Quantification of three separate experiments indicated that levels of SF2/ASF mRNA increased around threefold upon epithelial differentiation (Fig. 1B). Figure 1C confirms the upregulation of SF2/ASF protein in the same batch of cells from which RNA was prepared for the RT-PCR experiment.

FIG. 1.

SF2/ASF is transcriptionally upregulated in differentiated, infected epithelial cells. (A) RT-PCR amplification of SF2/ASF and GAPDH mRNAs in undifferentiated (U) and differentiated (D) W12 cells (episomal HPV16 genomes) and HaCaT cells (no virus genomes). RT−, reaction in the absence of RT; RT+, reaction in the presence of RT. (B) Quantification of data from three separate RT-PCR experiments. (C) Western blot titration analysis of levels of SF2/ASF protein detected with monoclonal antibody 96 in undifferentiated (U) and differentiated (D) W12 and HaCaT cells. ×1, ×2, and ×4 indicate relative amounts of extract applied to the gel (×1 = 10 μg protein).

When stably expressed at a low level in U2OS cells, HPV16 E2 upregulates SF2/ASF expression (41). A three- to fourfold level of upregulation was observed consistently in three individual E2-expressing cell clones (A4, B1, and B3). In order to test if HPV16 E2 could transactivate, the SF2/ASF promoter, a region from nt −1024 to 25 with respect to the transcriptional start site, was PCR amplified from HeLa cell genomic DNA and inserted upstream of the CAT gene in pGCAT3B to give pGCAT3-B-SF2pr. This region should cover the transcription initiation site and the proximal and distal promoter regions. Transient transfection transcription assays were carried out in U2OS E2-expressing clones A4, B3, and B1. CAT activity from pGCAT3-B-SF2pr increased nearly threefold in the presence of E2 in the A4 clone (and the B3 clone; data not shown), but no activation was observed with the B1 clone (Fig. 2A). It is known that the level of E2 protein in cells determines its transactivation/repression activities (6, 44, 58). Clones A4 and B3 expressed lower levels of E2 than clone B1 (Fig. 2C) but transactivated the SF2/ASF promoter to a greater extent, indicating that only low levels of E2 were able to transactivate the SF2/ASF promoter in this assay. No CAT activity was observed in cells transfected with pGCAT3-B without the SF2/ASF promoter either in the presence or absence of E2 (data not shown). E2 failed to transactivate a CAT expression vector driven by the cytomegalovirus (CMV) promoter (Fig. 2A), although some variation was observed in the level of reporter activity between cell clones, suggesting that there are potentially minor differences between the cells clones with respect to general transcriptional activity. Transfection efficiency as determined by cotransfection of a β-galactosidase expression plasmid was similar between clones at around 40%.

FIG. 2.

The SF2/ASF promoter is transactivated in the presence of HPV16 E2. (A) U2OS cells stably expressing E2 (A4 and B1 clones) or stably transfected with the vector construct (−ve) were transiently transfected with the SF2/ASF promoter construct pGCAT3-B-SF2pr or a control construct with the CMV promoter. CAT assays were performed on protein lysates, and the graph shows the means and standard deviation of the means from three experiments. (B) HeLa cells were transfected with pGCAT3-B-SF2pr or a control herpes simplex virus immediate early 5 promoter, and 50 ng, 150 ng, or 250 ng pCI-E2. Transfections were performed in triplicate, and the graph represents the means and standard deviation of the means from three separate experiments. (C) Representative Western blots using equal quantities of protein lysates from the transfected U2OS or HeLa cells probed with anti-E2 and anti-GAPDH antibodies.

We extended our observation by performing transient transfection assays using pGCAT3-B-SF2pr in HeLa cells along with either an empty expression vector (pCI) or an HPV16 E2 expression vector (pCI-E2) (Fig. 2B). As E2 causes cell growth arrest and apoptosis of HeLa cells, titration experiments were performed to ensure the level of E2 expression was optimal for transactivation at the SF2/ASF promoter. It was difficult to detect E2 expression in HeLa cells transiently transfected with pCIE2, suggesting that levels of expression were very low (Fig. 2C). However, a twofold increase in CAT activity was observed from pGCAT3-B-SF2pr when cotransfected with 50 ng pCI-E2 compared to cotransfection with the empty vector (Fig. 2B). In contrast to the SF2/ASF promoter, transcription of CAT constructs driven by the herpes simplex virus immediate early 5 promoter was unaffected by E2 expression as expected (Fig. 2B). The low level of E2 expression achieved may explain in part the relatively low level of transactivation of the SF2/ASF promoter compared to what was obtained in E2-expressing U2OS cells. The reduction in CAT activity with increasing E2 levels in HeLa cells is not statistically significant.

The transactivation domain of E2 is necessary for stimulation of the SF2/ASF promoter.

To determine which domain of the E2 protein was necessary for transactivation of the SF2/ASF promoter, further CAT assays were performed using extracts of HeLa cells transfected with pGCAT3-B-SF2pr together with either pWEB (vector only control) or pWEB-E2 (expressing wild-type E2) or pWEBDBDmut (DNA binding domain mutant) (63). Similar experiments were carried out with either pCMV (vector only control) or pCMVE2 or pCMV-R37E2 (transactivation mutant pCMVADmut) (51). Figure 3 shows that levels of transactivation of the SF2/ASF promoter with E2 expressed from pWEB-E2 (Fig. 3A) were similar to those seen with pCMV-E2 (Fig. 3B) and pCIE2 (Fig. 2B). pWEB-E2DBDmut also gave very similar results. Again transactivation levels decreased with increasing E2 expression levels. This suggests that the DNA binding domain of E2 is not involved in SF2/ASF transactivation. In contrast, expression of pCMVADmut, the transactivation domain mutant, was unable to transactivate significantly the SF2/ASF promoter (Fig. 3B). There was little increase over background transcription levels, even for the smallest amount of transfected DNA. Increasing the amounts of transfected pCMVADmut to 250 ng caused repression of CAT activity compared to that with pCMVE2. The data suggest that the transactivation domain is necessary for E2 to stimulate transcription of the SF2/ASF promoter, perhaps by binding a coactivator, although large amounts of E2 may transrepress the promoter.

FIG. 3.

The transactivation but not the DNA binding domain of E2 is necessary for regulation of the SF2/ASF promoter. (A) Data from CAT assays performed using protein lysates from HeLa cells transfected with pGCAT3-B-SF2pr and 50 ng, 150 ng or 250 ng pWEB, pWEB-E2 or pWEB-E2DBDmut. (B) Data from CAT assays performed using protein lysates from HeLa cells transfected with pGCAT3-B-SF2pr and 50 ng, 150 ng, or 250 ng pCMV, pCMVE2, and pCMVADmut. Transfections were performed in triplicate, and the graphs show means and standard deviation of the means from three separate experiments.

E2 actively associates with the SF2/ASF promoter in vivo.

The presence of E2 at the SF2/ASF promoter was detected using ChIP assays. HeLa cells were cotransfected with pGCAT3-B-SF2pr and either empty vector (pCI) or an E2-expressing vector (pCI-E2). Two anti-E2 antibodies were used, both polyclonal, one detecting only C-terminal epitopes (E2SCT) and one raised against the entire protein (E2poly). Following antibody purification of E2/DNA complexes, presence of the SF2/ASF promoter region in immunoprecipitated DNA was assayed using CAT5′ F and CAT5′ R primers, which amplify a region within the 5′ region of CAT gene, and either SF2pr −524 F and SF2pr −325 R or SF2pr −402 F and SF2pr −222 R, which amplify two overlapping regions within the middle of the promoter, and as a negative control, 5S RNA F and 5S RNA R, which amplify a region of the 5S RNA gene (Fig. 4A). The 5S RNA promoter is controlled by RNA polymerase III, which is not reported to be regulated by E2. E2 interaction (Fig. 4A lanes 3 and 4) with the SF2/ASF promoter was detected only in cells transfected with pCI-E2 but not pCI, using CAT5′ and SF2pr primers. Each set of primers gave very similar results. Only data from primer set SF2pr-524F and −325R are shown in Fig. 4A. TBP interaction (Fig. 4A, lane 2) with promoter DNA was detected in each case, as this protein binds all eukaryotic promoters and acts as a control for promoter detection in the assay. So, while TBP interacted with the 5S promoter region in cells transfected with pCI-E2, E2 did not interact with this promoter as expected. This indicates that E2 interacts specifically with the SF2/ASF promoter. Finally, ChIP experiments were performed using cells transfected with pCI-E2 and pGCAT3-B to ensure that E2 was not associated with the CAT construct at a region other than that of the SF2/ASF promoter. While CAT DNA could be detected from input samples, CAT DNA fragments bound by TBP and E2 could not, suggesting that E2, like TBP, associates with pGCAT3-B-SF2pr via the SF2/ASF promoter region.

FIG. 4.

E2 associates with the SF2/ASF promoter in vivo. (A) HeLa cells were cotransfected with pCI-E2 and pGCAT3-B-SF2pr or pGCAT3-B. DNA-protein complexes from cell extracts were sonicated into fragments of 200 bp to 1 kb and subjected to ChIP assays using an antibody (αb) against TBP as a positive control, two test polyclonal E2 antibodies (E2SCT and E2P), an antibody against involucrin as a negative control (N/S), and beads only (beads). Primers were used to amplify a region of the CAT gene (CAT5′ F and CAT5 R) and SF2/ASF promoter (SF2pr −524F and SF2pr −325R) and 5S RNA (5S RNA F and 5S RNA R) as a negative control. (B) HeLa cells transfected with pCI-E2 or pCI control were subjected to ChIP assays to amplify the endogenous SF2/ASF promoter and a 3′ region of the SF2/ASF coding sequence (SF2 F and SF2 R) as a negative control. Panels A and B, lane 1, show input DNA with no antibody.

Next, association of E2 with the endogenous SF2/ASF promoter was assayed. This time, ChIP experiments were performed in HeLa cells transfected with either pCI-E2 or pCI, but without pGCAT3-B-SF2pr, using primers SF2pr −402 F and SF2pr −222 R (Table 1) to determine association. Furthermore, primers amplifying within the 3′ region of the SF2/ASF coding sequence approximately 2.5 kb downstream from the promoter, SF2 F and SF2 R (Table 1), were used as negative controls. Figure 4B indicates that E2 associated with the endogenous SF2/ASF promoter in the presence but not in the absence of E2. However, this time, only the E2 polyclonal antibody (Fig. 4B, lane 4) could precipitate E2 bound to the endogenous SF2/ASF promoter, as no PCR product was detected when immunoprecipitation was carried out with the E2SCT antibody (Fig. 4B, lane 3) against the E2 carboxyl-terminal domain. The failure to detect E2 bound to the endogenous SF2/ASF promoter using the antibody against the E2 C-terminal domain could reflect differences in chromatin conformation in the endogenous and exogenous SF2/ASF promoters. In the case of the endogenous promoter, the E2 C-terminal domain may be buried in a transcription complex, making it inaccessible to the antibody. SF2 F and SF2 R primers failed to detect precipitated DNA using anti-TBP or anti-E2 antibodies. This suggests that E2 interacts specifically with the SF2/ASF promoter region and not with the downstream coding region.

E2 interacts with the SF2/ASF promoter in vitro.

In the SF2/ASF promoter, at −782 nt from the transcriptional start site, a consensus E2BS was identified which could have the potential to bind E2 (Fig. 5A). To investigate this possibility, EMSAs were performed. A 202-bp probe containing the potential E2BS was produced (13, 24). His-tagged HPV16 E2 was purified from BL21 bacterial cells transfected with pET15bE2(16); however, a number of contaminants were copurified (Fig. 5B, lanes 1 to 3). To take this into account, wash 3 from the column purification (Fig. 5B, lane 3) was used as a negative non-E2-containing fraction to control for any nonspecific binding, while elution 1 (Fig. 5B, lane 4) was used as the E2-containing fraction for all EMSAs. Figure 5C shows that while purified 16E2 bound the positive control probe (100 bp of the LCR from HPV18 containing E2BS 1 and E2BS 2) (62) and occupation of one and two binding sites was observed (Fig. 5C, lane 6), E2 failed to interact with either the negative control probe (similar to the positive control but with E2 binding sites mutated by C-to-T substitutions at position 10 in E2BS1 and positions 2 and 3 in E2BS2) (62) (Fig. 5C, lane 3) or the SF2/ASF promoter region containing the potential E2BS (Fig. 5C, lane 9). Wash 3 did not interact with the E2 binding sites as expected (Fig. 5C, lane 5). This indicates that E2 does not interact with the potential E2BS within the SF2/ASF promoter in this assay.

FIG. 5.

E2 associates with the SF2/ASF promoter in vitro. (A) Positions of probes used in EMSA experiments are shown below a diagram of the SF2/ASF promoter with positions relating to the transcriptional start site indicated as +1 with an arrow. putE2BS, putative E2 binding site. (B) Coomassie blue-stained SDS-PAGE gel of wash and elution samples from column purification of bacterial expressed His-tagged E2. Washes 1 to 3 and elutions 1 to 5 are shown. (C) EMSAs using a region of the HPV18 LCR containing two E2 binding sites either mutated (mE2BS) or intact (2xE2BS) and the SF2/ASF promoter probe, which contained a potential E2BS (probe 2). EMSAs were performed using probe only, wash 3 (lacking E2), and elution 1 (containing E2). (D) EMSAs using probe 3 from the SF2/ASF promoter, with wash 3 (lacking E2) and elution 1 (containing E2), with and without HeLa nuclear extract (NE). The arrow indicates an E2-associated complex. (E) EMSA using probe 3 with wash 3 (lanes 2 to 4) and elution 1 (containing E2) (lanes 5 to 7), with increasing concentrations of nonradiolabeled probe 3 (one-, two-, and fourfold molar excess). Nonradiolabeled probe was added prior to radiolabeled probe. The E2-associated complex is indicated.

Next, we determined which portion of the SF2/ASF promoter could bind E2. Six overlapping DNA probes of ∼200 bp each, covering the entire cloned promoter, were synthesized (Fig. 5A). EMSAs were performed as above with purified E2 protein (elution 1) and control wash 3. To detect whether E2 bound DNA in complex with any cellular proteins, assays were also performed in the presence of HeLa nuclear extract, with and without E2. EMSAs using DNA probes 1 and 2 spanning the region of nt −1024 to −650 from the SF2/ASF transcriptional start site showed no complex formation (data not shown). E2 did not interact with the three most proximal probes (probes 4 to 6) spanning nt −524 to 5 from the transcriptional start site; however, a number of complexes were observed upon addition of nuclear extract, as expected for a proximal promoter region (data not shown). In contrast, in EMSAs using probe 3 spanning nt −684 to −482 from the start site, a band shift was observed upon addition of E2 (Fig. 5D, lane 3). Furthermore, addition of nuclear extract increased the potential E2 binding (Fig. 5D, lane 6). Although a faint band shift was observed when the probe was incubated with control wash 3, this was of a different mobility than that observed with the E2 fraction (Fig. 5D, compare lanes 2 and 3). To determine whether E2 bound the SF2/ASF probe 3 promoter region specifically, cold competition EMSA was carried out (Fig. 5E). In a specific competition assay, when increasing concentrations of nonradiolabeled probe were added to reactions 15 min prior to radiolabeled probe, only a slight reduction in binding was observed to probe 3 (Fig. 5E, compare lanes 5 and 6), suggesting that E2 bound this SF2/ASF promoter region indirectly or in a dynamic fashion. This fits with the observation that the DNA binding domain of E2 is not required for E2 transactivation of the SF2/ASF promoter.

Levels of SF2/ASF are reduced in human foreskin keratinocytes transfected with HPV31 genomes containing the activation domain mutant E2:IL-73.

A transactivation mutant of HPV16 E2 is available, but as far as we know, the mutation has not yet been made in the whole-virus genome. However, such a genome mutation has been made for the E2 protein (E2:IL-73) of the closely related virus, HPV31 (61). When expressed from the HPV31 virus genome transfected into human foreskin keratinocytes, the transactivation-deficient mutant HPV31 E2 protein, E2:IL-73, is able to support viral DNA replication and induce late gene expression. However, cells expressing the mutant protein show moderately reduced levels of L1 and L2 RNAs. One possibility is that this is due to disruption of E2-mediated transactivation of cellular gene promoters. Our data indicate that SF2/ASF may be one such promoter. First, we determined whether SF2/ASF is upregulated upon differentiation of HPV31 genome-containing cervical keratinocytes. Figure 6A shows Western blot analysis of similar amounts of protein extracts isolated from undifferentiated (Fig. 6A, lane 1) and differentiated (Fig. 6, lane 2) CIN612 9E cells (2). There was a similar upregulation of SF2/ASF in these cells compared to that observed with W12E cells (Fig. 1), suggesting that HPV31 infection also regulates expression of SF2/ASF. Having confirmed upregulation of SF2/ASF upon differentiation of HPV31-containing epithelial cells, we obtained matched cell lysates from two separate experiments where human foreskin keratinocytes were transfected with either wild-type HPV31 or E2:IL-73 mutant virus genomes (a gift of L. A. Laimins, Northwestern University, Chicago [61]). Western blot analysis of each matched set showed reduced levels of SF2/ASF in cells expressing E2:IL-73 compared to those in cells expressing wild-type E2 (Fig. 6B). Similar levels of decreased SF2/ASF protein expression were detected in each matched set, indicating that clonal differences were not responsible for the change in SF2/ASF levels. These data indicate that the E2 activation domain is at least partly responsible for regulating expression of cellular SF2/ASF upon HPV infection.

FIG. 6.

SF2/ASF levels are regulated by the activation domain of HPV31 E2 in virus genome-containing human foreskin keratinocytes. (A) Cell extracts were prepared from CIN612 9E cells either undifferentiated (lane 1) or differentiated (lane 2). Twenty micrograms of protein extracts per track were fractionated on SDS-PAGE and analyzed using Western blotting. Blots were probed with monoclonal antibody 96 against SF2/ASF and 6CS against GAPDH. (B) Twenty micrograms of protein extracts per track from normal human keratinocytes transfected with HPV31 genomes expressing wild-type E2 (wt) (lanes 1) or the activation domain mutant E2:IL-73 (mut) (lanes 2) was fractionated on SDS-PAGE and analyzed using Western blotting. Antibodies were used as described above. Extracts from two separate transfection experiments (sample sets 1 and 2) were used, and a representative Western blot analysis result from each experiment is shown. The graph shows quantification of levels of SF2/ASF protein with respect to levels of GAPDH from three Western blots for each experiment.

DISCUSSION

Papillomavirus E2 proteins regulate viral gene transcription from the viral LCR, but they are also known to regulate cellular promoters. In this paper, we demonstrate transcriptional regulation of the key splicing-related SR protein SF2/ASF by HPV16 and HPV31 E2. This is the first example of transactivation of a cellular promoter by HPV16 E2. Furthermore, SF2/ASF is the first RNA regulatory protein identified whose expression is controlled by HPV proteins. Although HPV5 E2 has been demonstrated to interact via its hinge domain with four splicing factors, including SF2/ASF, it has not been shown to regulate their expression (32). Upregulation of SF2/ASF and other members of the SR protein family (unpublished data) in W12 cells with episomal HPV16 genomes is specific, as other nonrelated splicing proteins are not similarly induced (9). One exception is hnRNP A1, the antagonistic partner of SF2/ASF in alternative splicing, which is upregulated to a similar extent like SF2/ASF in differentiated cervical epithelial cells (9). Furthermore, SF2/ASF levels fall slightly upon differentiation of normal human keratinocytes and HaCaT cells that do not contain virus genomes and in W12 cells where only integrated HPV16 DNA is present (41). This indicates that virus infection, as opposed to differentiation or transformation, causes the change in levels of the protein. This finding underscores the importance of virus-mediated cellular control, as splicing regulation is now known to be a major controller of cellular and viral protein expression. Indeed, it is becoming clear that viral life cycles can require interaction with the host cell RNA processing machinery; for example, influenza virus requires SF2/ASF to regulate M2 ion channel protein levels (54). Moreover, host-splicing SR proteins are known to be regulated by adenovirus and human immunodeficiency virus infection (17, 43).

Why would HPV16 infection result in upregulation of alternative splicing factors? Epithelial cells become less metabolically active as they enter terminal differentiation (16), and many nuclear proteins, including RNA processing proteins, are downregulated. During differentiation of HPV16-infected epithelial cells, a suite of 13 different mRNAs that encode the virus late proteins, including the capsid proteins, are produced by extensive alternative splicing (42). We propose that HPV16 regulates key alternative splicing factors during differentiation to ensure appropriate virus gene expression and completion of the virus life cycle. However, in addition to splicing, SF2/ASF also regulates mRNA export, stability, and translation (26, 36, 53). So, E2 upregulation of SF2/ASF may impinge on other facets of the RNA expression pathway. Alternatively, it is clear that SF2/ASF has other roles in cellular metabolism. For example, it is involved in the maintenance of genome stability (38), while loss of SF2/ASF can induce cell cycle arrest and apoptosis (39). So, an E2-induced increase in SF2/ASF levels could protect HPV16-infected, differentiated epithelial cells from apoptosis and DNA damage to the host chromosomes and possibly also to amplified virus genomes.

E2 can transrepress or transactivate viral and cellular promoters, and this can be dependent on levels of the protein in cells. Generally, we observed elevated levels of reporter gene activity driven by the SF2/ASF promoter region in cells expressing E2. However, the extent of upregulation differed depending on the levels of E2 expressed in the transfected cells. High E2 expression observed in U2OS cell clone B1 transactivated the promoter to a much lesser extent compared to the lower levels observed in U2OS clones A4 and B3. High levels of HPV16 E2 are known to result in inhibition of cell growth and promotion of apoptosis (47, 52, 63). Therefore, with higher E2 levels, cellular metabolism may be compromised, leading to a reduced ability of cellular factors to control expression of SF2/ASF. Alternatively, E2 may begin to transrepress the promoter at higher concentrations, similar to the mechanism by which E2 has been shown to regulate the BPV4 LCR (44).

E2 associates with the SF2/ASF promoter as determined using ChIP assays, suggesting it transactivates SF2/ASF directly. However, direct interaction between E2 and the promoter could not be determined definitively in vitro. In EMSA experiments, purified His-tagged E2 appeared to bind a region of the promoter located −684 to −482 nt upstream of the transcription initiation site. The binding of His-tagged E2 was augmented in the presence of HeLa nuclear extract. However, a specific competitor could reduce the binding to the probe only partially, indicating that the binding was indirect or dynamic in fashion. E2 may require another protein(s) to bind and activate the SF2/ASF promoter. A known cellular binding partner of E2, Sp1 (37), was predicted to bind the same region of the promoter as E2, and a supershift EMSA indicated that Sp1 interacted with the promoter in vitro (data not shown). It is possible that E2 associates with Sp1 to become tethered to the SF2/ASF promoter. As expected, the basal transcription factor TBP also associates with the promoter. BPV1, HPV8, and HPV18 E2 proteins can interact with TBP and/or a number of components of the basal transcription machinery (8, 14, 49). Furthermore, HPV16 E2 interacts with yeast TFIIB (4). So if HPV16 E2 binds human TBP or TFIIB, it may enhance formation of a transcription initiation complex on the SF2/ASF promoter.

Transcription assays demonstrated that the DNA binding domain of E2 was not essential for its transactivation of the SF2/ASF promoter. In contrast, an HPV16 E2 transactivation mutant was unable to transactivate the SF2/ASF promoter, and it caused transrepression at higher concentrations. Experiments with virus genomes containing an HPV31 E2 transactivation mutant, E2:IL-73, confirmed that a wild-type E2 activation domain is required to achieve high levels of SF2/ASF in virus-infected cells. Stubenrauch et al. (61) reported that human keratinocytes expressing E2:IL-73 in the context of the HPV31 genome had reduced levels of spliced L1 and L2 mRNA expression. Our results suggest that the explanation for this observation may lie in the reduced levels of the key splicing factor SF2/ASF caused by an activation domain-mutated E2 protein that can no longer transactivate the SF2/ASF promoter efficiently.

A near-consensus E2 binding site was found in the SF2/ASF distal promoter. However, E2 did not bind this sequence. It differs from the consensus at position 4, where a C is present rather than a G, and at position 9, where a T is found instead of a C [the canonical binding site is ACCG(N₄)CGGT]. Furthermore, HPV16 E2 binding is found to be susceptible to the spacer (N₄) sequence, with preferential binding to sites with AT-rich spacers (24) that may provide a prebent DNA molecule to which E2 binds more easily (23). However, the SF2/ASF site contains CTTC within the spacer. Finally, HPV16 E2 has more recently been shown to bind larger sites composed of AACCG(N₄)CGGTT more efficiently (13). Although a T is found directly downstream of the near-consensus SF2/ASF potential E2BS, the immediate upstream nucleotide is a T, not an A. Therefore, it is possible that while one of these discrepancies may be tolerated, all three within the same site prevent E2 from binding.

Recently, SF2/ASF has been classified as an oncoprotein, as it is found to be upregulated in a range of cancer-causing viruses (28). However, we believe that tumorigenic upregulation of SF2/ASF is not what we are observing here. In the W12 20863 clone we have used, there is evidence that the virus can complete its replication cycle, as measured by increased virus genome copy numbers (25) and capsid protein expression (42). W12 20863 cells retain the ability to differentiate and require specific growth conditions, including mouse fibroblast feeder layers and mitogens, indicating that they are not transformed. In addition, W12 cells where the HPV16 genome has integrated (W12 20861 cells) require similar growth conditions (27) and are also not transformed, as they do not form tumors in nude mice (1). Genome integration results in disruption of the E2 open reading frame, no E2 protein can be detected in these cells (data not shown), and differentiation results in a decrease in SF2/ASF levels (41). Finally, there is a significant response of SF2/ASF expression to W12 epithelial differentiation. Much-higher levels of the protein are detected in differentiated cells. SF2/ASF may prove to be a suitable marker for staging cervical disease in the future, especially since a similar upregulation of the protein can be detected in HPV31 genome-containing cells.