ABSTRACT
The functional significance of sequence variation within the upstream regulatory region (URR) of six human papillomavirus type 16 (HPV16) cervical cancer isolates from Australia was investigated. Specific changes in transcription factor binding sites leading to increased promoter activity may explain the transforming ability of some episomal HPV16 isolates.
Human papillomavirus type 16 (HPV16) isolates show considerable sequence diversity within the upstream regulatory region (URR) which controls the expression of the viral oncogenes (E6 and E7) (2). Evidence is accumulating that variation affecting the level of E6 and E7 expression may influence the oncogenic potential of individual HPV16 isolates (5, 6, 19). We recently sequenced the URR of 34 HPV16 cervical cancer isolates from Australian and New Caledonian women and found that 28 were typical of the European lineage while 4 were Asian (As) variants and two were Asian-American (AA) variants (reference 2 and unpublished data). Six of these isolates were selected for functional analysis of URR variations on the basis of an 81-bp duplication of the enhancer (isolate O2) and/or sequence variation in transcription factor binding sites (TFBS) potentially affecting E6 and E7 expression (isolates O2, K2, K4, H1, R1, and S1) (Table1).
URR sequence variation of six HPV16 isolates compared with the HPV16 prototype (HPV16R) and location within known TFBS
Initially, the promoter activities of variant URRs were compared with that of the HPV16 prototype cloned from a German cervical cancer sample (10, 15) using luciferase activity as reporter of E6 and E7 expression. PCR products from the entire URRs (PCR A [Table2]) were cloned into pCR-Blunt (Invitrogen) and then subcloned into pALuc (5). Base changes were confirmed by forward and reverse sequencing of different PCR products. Transient transfections were performed in duplicate using 60 to 80% confluent cultures of cervical cancer-derived cells (HeLa [HPV18 positive] and HT3 [HPV negative]) by calcium phosphate coprecipitation (5) with 6 μg of luciferase-URR plasmid and 1 μg of pCMV β-galactosidase expression vector (Promega) per 60-mm-diameter dish. Luciferase and β-galactosidase assays were performed on equivalent amounts of protein harvested from 48-h cultures (4). Individual URR promoter activities, determined by calculating the luciferase/protein ratios adjusted for the variation in transfection efficiencies, were based on the results of at least three independent experiments. Identification of the specific changes mediating elevated promoter activity was done by using PCR-based site-directed mutagenesis (1) to convert base changes potentially affecting promoter activity to the corresponding base in the HPV16 prototype (Table 2) and retesting promoter activity.
Primers used in the PCRs, genome positions, product sizes, nucleotides, and TFBS altered by site-directed mutagenesis, and constructs generated
Three of the six variants, i.e., K2, K4, and H1, were found to have significantly upregulated promoter activity in HeLa cells (11.3-, 7.3- and 3.5-fold, respectively), while the activities of the isolate with the large-scale duplication (O2) and of isolates R1 and S1 were similar to that of the HPV16 prototype (Fig. 1). The results obtained using HT3 cells were comparable. Site-directed mutagenesis showed that the 11-fold increase associated with K2 was primarily due to a substitution in a Yin Yang 1 (YY1) motif (12) at nucleotide (nt) 7792 but a change in overlapping octamer-1/papillomavirus enhancer factor 1 (Oct-1/PEF-1) sites (16) at nt7676 also contributed. Upregulation of K4 activity was partly due to the change in the papillomavirus silencing motif (PSM) (13) at nt 7894. In contrast, variations in a YY1 site at nt 7786 (K4 and H1) and in the TEF-1 site (7) at nt 7743 (H1) did not alter promoter activity.
Expression of luciferase under the control of HPV 16 URR in HeLa cells. The promoter activities of the H1, K2, K4, R1, O2, and S1 URR (black bars) were assessed in the context of the prototype URR, which was assigned the value 1 (white bar). The promoter activities of H1, K2, and K4 altered by site-directed mutagenesis are designated by hatched bars. The change in the YY1 motif at nt 7792 explained 9-fold of the 11-fold increase associated with K2 (shown by the difference between pALuc16K2 and pALuc16K2YY1P); the change in overlapping Oct-1/PEF-1 sites at nt 7676 was also significant, as shown by the almost twofold reduction in activity between pALuc16K2 and pALuc16K2Oct-1P. The variation in the PSM at nt 7894 upregulated the promoter threefold (as shown by the difference between pALuc16K4 and pALuc16K4PSMP). Variations in the YY1 site and TEF-1 sites at nt 7743 and 7786 in K4 and H1, respectively, did not alter promoter activity (pALuc16H1 activity was similar to pAluc16H1TEF-1P activity, and pALuc16K4 activity was similar to pALuc16K4YY1P activity).
Electrophoretic mobility shift assays (EMSAs) using nuclear extracts of 106 HeLa cells and 250 pg of end-labeled (20,000 cpm) prototype or variant oligonucleotides (Table3) (12, 17) were then undertaken to determine whether the increased promoter activities associated with the K2 and K4 URRs were due to altered protein-DNA binding at nt 7792 (YY1), 7676 (Oct-1/PEF-1), or 7894 (PSM). Competition assays were performed using a labeled prototype backbone and a 12.5- to 200-fold excess of unlabeled prototype or variant oligonucleotides (18). Supershifts were carried out using 2 μg of corresponding antibodies (Santa Cruz). As shown in Fig.2 and 3, the changes at nt 7792 and 7676 in K2 had no effect on the binding of YY1 or Oct-1, respectively. Binding of PEF-1 to nt 7676 was also unaffected (data not shown). However, the change in the PSM of K4 at nt 7894 substantially reduced the binding of the dimer, but not the monomer, form of the PSM-binding protein (PSM-BP) (Fig.4).
Oligonucleotides used in EMSA, genome positions, and TFBSs affected
Autoradiograph showing the binding of HeLa nuclear extract to 32P-labeled AAV P5-60 YY1 (lane 1) and HPV 16 prototype oligonucleotide nt 7790 to 7804 not competed (lane 2) or competed with a 200-fold (lanes 3 and 6), 100-fold (lanes 4 and 7), or 50-fold (lanes 5 and 8) excess of unlabeled prototype oligonucleotides (16P7792) with a C at nt 7792 (lanes 3 to 5) or unlabeled mutant oligonucleotide (16M7792) with a T at nt 7792 (lanes 6 to 8). The unlabelled YY1 prototype (16P7792) and mutant (16M7792) oligonucleotides competed almost equally effectively for binding with the labelled prototype. Evidence that the band represents YY1 binding is provided in lane 9, where polyclonal YY1 immunoglobulin G (Santa Cruz H-414 sc1703x) has almost completely abolished binding. FP, free probe.
Autoradiograph showing the binding of HeLa nuclear extract to 32P-labeled prototype oligonucleotide nt 7667 to 7693. Evidence that the uppermost band in lane 2 represents Oct-1 binding is provided in lane 1 by the addition of polyclonal Oct-1 antibody (Santa Cruz C-21 sc232x), which has abolished binding of the labelled prototype. The effect of the C-to-A mutation at nt 7676 on binding to Oct-1 and an uncharacterized ladder of faster-migrating complexes is shown by competition with a 200-fold (lanes 3 and 8), 100-fold (lanes 4 and 9), 50-fold (lanes 5 and 10), 25-fold (lanes 6 and 11) and 12.5-fold (lanes 7 and 12) excess of unlabeled prototype (16P7676) (lanes 3 to 7) or unlabelled mutant oligonucleotide (16M7676) (lanes 8 to 12). FP, free probe. The unlabelled Oct-1/PEF-1 prototype (16P7676) and mutant (16M7676) oligonucleotides competed equally effectively with labeled prototype for binding of Oct-1.
EMSA autoradiograph showing the binding of HeLa nuclear extract to 32P-labeled prototype oligonucleotide nt 7893 to 7915. Lane 1 shows two slowly migrating complexes, C1 and C2, analogous to those reported by O'Connor et al. (13) and believed to represent the monomer and dimer forms of the PSM-BP, respectively. The effect of the A-to-C mutation at nt 7894 is shown by competition with a 200-fold (lanes 2 and 6), 100-fold (lanes 3 and 7), 50-fold (lanes 4 and 8), and 25-fold (lanes 5 and 9) excess of unlabeled prototype (16P7894) (lanes 2 to 5) or unlabeled mutant oligonucleotide (16M7894) (lanes 6 to 9). The PSM-BP bound to both prototype and mutant oligonucleotides, but the unlabeled mutant competed for binding to the dimeric form of PSM-BP (to give C2) less effectively than the prototype did. Binding of the monomer form of PSM-BP (to give C1) was unaffected. Evidence of the specificity of C1 and C2 is provided in lane 10 by competition with a 200-fold excess of an unrelated unlabelled oligonucleotide (16P7676), where binding of both forms of the PSM-BP has been largely unaffected. UC, uncharacterized bands; FP, free probe.
Up to one-third of HPV16-positive cervical cancers carry the virus in episomal form (14). Since certain sequence variations, notably those in YY1 sites, seem to cluster in isolates carried episomally, the physical state of five of the isolates was determined by Southern hybridization (H1 could not be analyzed due to lack of tumor tissue). A 10-μg portion of nucleic acids digested withBglII (no cut) and BamHI (single cut) were hybridized with a full-length HPV16 probe under high-stringency conditions. Since integration of the HPV genome into cellular DNA frequently disrupts the viral E2, PCR for the integrity of the E2 gene (3) was also performed (PCR L [Table 2]). Overall, the results indicated that O2, K2, and K4 were entirely episomal, R1 was entirely integrated, and S1 existed as both episomal and integrated forms.
It is noteworthy that the promoter activity of the episomal isolate O2 (containing the large duplication and base substitutions characteristic of As variants) was comparable to that of the prototype. Duplications are uncommon in the HPV16 URR, but Hall et al. (6) have reported that the transforming ability of a European episomal HPV16 isolate was dependent on a similar duplication. Our findings indicate that malignant conversion without integration is not necessarily dependent on elevated promoter activity.
In the episomal isolate K2, the 11-fold increase in promoter activity was primarily due to a change in a YY1 site at nt 7792, consistent with reports that other natural YY1 mutations released the promoter from YY1 repression, allowing malignant conversion in the absence of integration (5, 9). However, in contrast to these previous reports, this particular change did not affect YY1 binding affinity.
The other isolates showing increased promoter activity, K4 and H1, displayed changes typical of AA variants. The sevenfold increase in the promoter activity of K4 was found to be partly due to a change at nt 7894 located in the PSM. The PSM-BP, identified as CCAAT displacement protein, appears to be a master regulator of HPV transcription and replication during epithelial differentiation (11). Experimental deletion of either of the two PSM motifs derepresses promoter activity (13). Our study has shown for the first time that a single, naturally occurring single-base change can have the same effect. Since K4 was episomal, there are clear parallels between the functional effects of PSM and YY1 sequence variations. However, our EMSAs showed that the change impacting a YY1 site at nt 7786, present in both K4 and H1, had no influence on promoter activity, thus highlighting the need to evaluate YY1 changes individually. In a recent study (8), the C-to-A variation in an unidentified TFBS at nt 7729, present in both K4 and H1, enhanced the promoter activity of AA variant URR containing changes very similar to those in K4 and H1. This change may well have been responsible for the increased the promoter activity of K4 and H1, with the change in the PSM accounting for the difference between them.
Our study contributes to the understanding of the biological significance of naturally occurring sequence changes in the HPV16 URR by showing that previously unreported sequence variations produce derepression of the E6-E7 promoter without integration or involvement of YY1. Large multicenter studies comparing patterns of URR variation in benign cervical lesions and dysplasias, as well as cancers, are needed to confirm the biological relevance of viral variation.
ACKNOWLEDGMENTS
This study was supported in part by grants from the National Health and Medical Research Council of Australia (grant 34407) and the Cancer Research Fund of Royal Prince Alfred Hospital, Sydney, Australia.
We thank Martin Tattersall from the Department of Cancer Medicine at the University of Sydney and Christopher Dalrymple, Jonathon Carter, Peter Russell, and other members of the Departments of Gynaecological Oncology and Anatomical Pathology at King George V/Royal Prince Alfred Hospital, Sydney, Australia, for their valued collaboration.
FOOTNOTES
- Received 1 August 2000.
- Returned for modification 13 December 2000.
- Accepted 5 March 2001.
- Copyright © 2001 American Society for Microbiology