A Synthetic E7 Gene of Human Papillomavirus Type 16 That Yields Enhanced Expression of the Protein in Mammalian Cells and Is Useful for DNA Immunization Studies

Abstract

A synthetic E7 gene of human papillomavirus (HPV) type 16 was generated that consists entirely of preferred human codons. Expression analysis of the synthetic E7 gene in human and animal cells showed levels of E7 protein 20- to 100-fold higher than those obtained with wild-type E7. Enhanced expression of E7 protein resulted from highly efficient translation, as well as increased stability of the E7 mRNA due to its codon optimization. Higher levels of E7 protein in cells transfected with synthetic E7 correlated with significant loss of cell viability in various human cell lines. In contrast, lower E7 protein expression driven by the wild-type gene resulted in a slight induction of cell proliferation. Furthermore, mice inoculated with plasmids expressing the synthetic E7 gene produced significantly higher levels of E7 antibodies than littermates injected with wild-type E7, suggesting that synthetic E7 may be useful for DNA immunization studies and the development of genetic vaccines against HPV-16. In view of these results, we hypothesize that HPVs may have retained a pattern of G + C content and codon usage distinct from that of their host cells in response to selective pressure. Thus, the nonhuman codon bias may have been conserved by HPVs to prevent compromising viability of the host cells by excessive viral early protein expression, as well as to evade the immune system.

Human papillomaviruses (HPVs) are small double-stranded-DNA viruses that infect stratified epithelia and cause benign and malignant proliferative lesions. A subset of HPVs with mucosal tropism, the so-called “high-risk” HPVs, has been linked to cancers of the uterine cervix (55), which account for about 11% of the global cancer incidence in women. Of these, more than 90% contain HPV DNA, most notably HPV type 16 (HPV-16) and HPV-18, integrated into the host cell genome, where they express two viral early genes, E6 and E7, whose products block p53 and retinoblastoma protein-mediated cell cycle control pathways (38, 56). Indeed, the tumorigenic phenotype of HPV relies on continuous expression of the E6 and E7 genes (9). E7 is an oncoprotein which can transform rodent fibroblasts (24), cooperate with activated ras to transform primary cells (34), and, in association with E6, immortalize keratinocytes (10, 19, 30).

Progression from detectable HPV infection to invasive cancer occurs in less than 1% of cases and usually takes more than 2 decades, indicating that additional factors are involved in the process of carcinogenesis. Several lines of evidence suggest that such factors may act by intensifying HPV early gene expression, as illustrated in the following examples. First, disruption or mutation of the viral E2 gene, which is known to repress the HPV early promoter and hence expression of E6 and E7, is observed in cervical carcinogenesis (7, 36, 42, 47, 52). Moreover, the site of chromosomal integration of the viral DNA influences expression of viral early genes (48). Second, viral load appears to be a determinant for the development of cervical carcinoma (23), and viral amplification is a frequent phenomenon in cervical tumors and tumor cell lines (3, 5, 42). Third, tumor formation in nude mice by cervical cancer cells is inhibited by antisense E6-E7 RNA (49). Fourth, repression of HPV gene expression by tumor necrosis factor alpha (TNF-α) is observed in nontumorigenic cervical cancer cell lines, while tumorigenic lines are not sensitive to TNF-α (41), suggesting a correlation between tumorigenicity and deregulation of HPV expression. Fifth, constitutive expression of E6 and E7 driven by keratin promoters causes epidermal hyperplasia and skin cancer in transgenic mice (2, 28). Altogether, these observations suggest that progressive intensification of viral gene expression is required for malignant transformation.

At the transcriptional level, besides the viral E2 gene product a number of cellular proteins, including AP-1 family members, SP-1, YY1 glucocorticoid, and progesterone receptors, regulate the HPV early promoter, located in the viral long control region, in cervical cancer cells as well as in transgenic mice. Since efforts to identify tissue-specific transcription factors have failed, it is conceivable that an interplay between those factors may account for the cell type specificity of the HPV-16 early promoter-enhancer, which is strictly epithelial in vivo (8), and for the continuous expression of E6 and E7 in cancer tissues.

Transcriptional analysis of HPV-infected cervical cancer tissues and cell lines has revealed that viral transcription is restricted to the early region and that the most abundant mRNA species encodes the E7 protein (11, 40, 43). Intriguingly, in contrast with the high levels of E7 mRNA that can be demonstrated by Northern blotting and in situ hybridization in carcinoma cell lines and tissues, the E7 protein is hardly detectable by immunostaining, immunoprecipitation, or Western blot analysis (15). Possible explanations for this discrepancy may be that the protein is expressed at concentrations too low for immunodetection and the fact that it is rapidly degraded by the ubiquitin-proteasome pathway (35). Additional evidence in support of a restricted presence of E7 in infected cells comes from patients with HPV-associated genital dysplasia, who, in most cases, generate a poor antibody response to E7, suggesting that this protein is expressed in small amounts and hence remains inaccessible to the immune system. Moreover, it has been proposed that epitopes of E7 could remain masked due to interaction with host cell proteins in the nucleus, where E7 is mainly localized (15, 25).

While the molecular mechanisms underlying the oncogenic functions of E7 through its interaction with various cellular regulatory proteins have been studied extensively, very little is known about factors that limit E7 expression. As the transcriptional rate from the HPV LCR is maintained at relatively high levels (40), it is likely that posttranscriptional mechanisms account for the low expression of E7. One such mechanism may result from the different codon usage of HPV and human genes. The extensive third-position degeneracy of the genetic code permits ample variation in the set of codons preferred by viruses with respect to that prevalent in human cells (45). This may result, however, in less efficient synthesis of viral protein due to insufficient amount of the tRNA species necessary to translate the corresponding mRNAs, as has been observed in other organisms (14, 21, 50). Indeed, limitations in protein expression due to codon usage divergence from human genes have been postulated for the latent genes of Epstein-Barr virus (26) and have been demonstrated for the env gene of human immunodeficiency virus type 1 (17), and the L1 and L2 capsid genes of bovine papillomavirus type 1 (54). In both cases, synthetic genes designed to have codon usage resembling that of human genes yielded enhanced expression of the corresponding proteins. It seems likely that a nonhuman codon bias leading to inefficient expression of viral proteins resulted from selective pressure to escape the defense mechanisms of the host. Should this apply to HPV, it would help explain the need for a high transcriptional activity to overcome the limitation in viral translation observed in cancer tissues.

In view of these observations, we decided to examine the role of the divergent codon bias of HPV-16 and human genes in the restricted expression of E7 in mammalian cells. We report a synthetic E7 gene engineered to contain exclusively codons preferred by highly expressed human genes, thereby yielding expression of remarkably high levels of E7 protein compared with the wild-type E7 gene. Enhanced expression of E7 from the synthetic gene appeared to result from both increased translation and higher stability of its mRNA. High expression levels of E7 protein had a negative effect on cell viability, which was not observed at the lower levels of protein expressed by the wild-type gene. Furthermore, DNA immunization experiments showed higher immunogenicity to E7 elicited by the synthetic gene, suggesting that it may be useful for the development of new DNA vaccines against HPV-16. In addition, our data suggest that selective pressure may have forced HPVs to adopt a suboptimal codon usage in order to persist as commensal agents and carry out their slow transforming effects in the cells.

MATERIALS AND METHODS

Synthetic genes and plasmid constructions.

A synthetic HPV-16 E7 gene was engineered in vitro by introducing 64 silent mutations (22% of the E7 sequence), which created an open reading frame consisting entirely of preferred human codons. In addition, a sequence encoding a Flag tag was fused to its 3′ end. The synthetic E7 gene was generated by two consecutive PCRs. Platinum Pfx polymerase (Invitrogen) was used in all PCR amplifications. The following overlapping oligonucleotides were used (numbers in parentheses correspond to nucleotide positions in the sequence given in Fig. 1C): A, 5′-GACAGCTAGC-(1 to 89)-3′ (forward primer); B, 5′-(156 to 65)-3′ (reverse primer); C, 5′-(138 to 232)-3′ (forward primer); D, 5′-ACACGAATTC-(318 to 211)-3′ (reverse primer); E, 5′-AACAGCTAGC-ATG-(106 to 131)-3′ (forward primer); and F, 5′-ACACGAATTC-AAT-(291 to 268)-3′ (reverse primer). In a first PCR, primers A and B and primers C and D were used to generate two fragments, AB and CD, by chain extension using no template. In a second reaction, the overlapping products AB and CD were mixed together and used as a template to produce a synthetic E7-Flag fusion gene by using primers A and D. Furthermore, the resulting E7-Flag gene was used as a template to amplify a synthetic E7 gene with no Flag tag (with primers A and F) and a truncated E7-Flag fusion gene (with primers E and D) lacking the first 35 codons of E7. The synthetic E7 genes were isolated from the PCR mixtures by agarose gel electrophoresis, purified with QiaexII (Qiagen), double digested with NheI and EcoRI, and cloned into the polylinker of pIRES-neo2 (Clontech), which was digested with the same restriction enzymes. Single clones were identified by transformation of DH5α bacteria and sequenced. A green fluorescent protein (GFP)-expressing plasmid (pQBI-25-fPA; Q-BIOgene) was used to monitor efficiency of transfections. Plasmids were purified with commercial kits (DNA Maxiprep, endotoxin free; Qiagen).

FIG. 1.

(A) Comparison of codon usage frequencies in highly expressed human genes (Hu∗), according to Haas et al. (17), and HPV-16 E6 and E7 (transforming genes) and L1 and L2 (capsid genes). (B) (Top) Comparison of the G+C content of average (Hu) and highly expressed (Hu∗) human genes (17, 31) with the coding regions of the most prevalent genital HPVs: HPV-16 and -18 (high risk, oncogenic) and HPV-6 and -11 (low risk, nononcogenic). The frequencies of A and T in the third position in human and HPV codons (middle) and among HPV-16 genes are also compared (bottom). (C) Nucleotide sequence of the synthetic HPV-16 E7 gene (eE7) generated in this work. Silent mutations introduced to create an open reading frame of preferred human codons are in capitals. The sequence encoding a Flag tag is underlined, and an asterisk indicates the stop codon. The encoded amino acid sequence is given below the nucleotide sequence.

Mice, cell lines, and transfections.

BALB/c mice (WiGa-Charles River, Hamburg, Germany) 6 to 8 weeks old were used for DNA immunization experiments with endotoxin-free plasmid complexed to polyethylenimine (in vivo jet-PEI; Q-Biogene) as suggested by the manufacturer. Equal amounts of plasmid (4 μg) were injected subcutaneously three times at 2-week intervals. Vero, HeLa, SiHa, and C-33 cells were grown in Dulbecco's Modified Eagle Medium with Glutamax (Invitrogen-Life Technologies) supplemented with 10% fetal calf serum (heat inactivated; Invitrogen-Life Technologies) and penicillin-streptomycin. Cells were transfected at 50 to 70% confluence by using Effectene (Qiagen) or FuGENE (Roche) by following the manufacturer's instructions.

Western blotting.

Cells growing on 6-cm plates were transfected with 2 μg of plasmid and allowed to express the protein for 24 h. Then cells were washed twice with phosphate-buffered saline (PBS) and lysed in sodium dodecyl sulfate (SDS) loading buffer containing 1 mM dithiothreitol. Cellular proteins were separated on 15% polyacrylamide gels by SDS-polyacrylamide gel electrophoresis (PAGE), blotted onto polyvinylidene difluoride membranes, and hybridized with the corresponding specific antibodies, followed by detection with horseradish peroxidase (HRP)-conjugated anti-mouse or -rabbit immunoglobulin G antibodies (Molecular Probes). Antibodies used were anti-Flag M2 conjugated to HRP (Sigma), anti-HPV-16 E7 antibody (Oncogene), anti-neomycin phosphotransferase II (anti-NPT-II) (Upstate), antiactin (Santa Cruz), and anti-GFP (Clontech). Proteins were visualized after reaction with enhanced-chemiluminiscence reagents (Renaissance; NEN Life Science Products).

Immunofluorescence.

Cells growing on glass coverslips were transfected in separate wells of 24-well plates with 0.4 ml of growth medium. At various times after transfection, the medium was aspirated and the cells were washed once with PBS, fixed with 4% paraformaldehyde for 15 min at room temperature, and washed once with PBS. Background fluorescence was reduced by quenching free aldehydes in 50 mM NH₄Cl-PBS for 10 min. After a brief washing with PBS, the cells were permeabilized by immersing coverslips in PBS-0.2% Triton X-100 for 5 min at room temperature, washed again with PBS, and blocked in blocking solution (2% fetal calf serum, 2% bovine serum albumin, and 0.2% gelatin in PBS) for 30 min. Primary and secondary antibodies were diluted in a 1:10 dilution of blocking solution in PBS. Incubation times were 1 h for each antibody. Coverslips were washed six times in PBS after incubation with each antibody. Finally, nuclei were stained with DAPI (4′,6-diamidino-2-phenylindole) or propidium iodide (Molecular Probes). Primary antibodies were anti-Flag M2 (Sigma) and anti-HPV-16 E7 (Oncogene). Secondary antibodies were Cy2- or Cy3-conjugated anti-mouse and anti-rabbit antibodies (Jackson ImmunoResearch). Cells were analyzed with a confocal laser scanning microscope, Leica TCS SP, with the detector pinhole set at 1 Airy disk unit.

Expression of E7 protein.

The complete HPV-16 E7 protein with an N-terminal six-His tag was expressed by using a T7 expression plasmid [pET28a(+); Novagen] and purified over nickel-agarose columns (Qiagen). The purified protein was quantified by using the Bradford assay (Bio-Rad).

ELISA.

Enzyme-linked immunosorbent assay (ELISA) plates were coated overnight with 1 μg of protein per ml in carbonate buffer, pH 9.3, at 4°C and blocked in 5% fetal calf serum in PBS for 2 h at room temperature. Plates were subsequently washed, and 100 μl of serum (diluted in 0.5% milk-0.1% Tween 20 in PBS) was added to each well and incubated for 1 h at 37°C. Bound antibody was detected with goat anti- mouse immunoglobulin G conjugated to HRP (Molecular Probes) for 1 h at 37°C. Plates were developed by adding 100 μl of O-phenylenediamine (Sigma). After 30 min at room temperature in a darkened area, the reaction was stopped by the addition of 50 μl of 1 M sulfuric acid per well. Plates were read at 450 nm.

Northern blotting.

Northern blotting was performed essentially as described previously (8). Briefly, total RNA was extracted from HeLa cells 24 h after transfection with plasmids expressing either synthetic or wild-type HPV-16 E7, by using an RNA extraction kit (RNeasy; Qiagen) as directed by the manufacturer. Purified RNA samples were separated on 1% agarose-formaldehyde gels, blotted onto nylon membranes (Genescreen; NEN), and hybridized with mixed (1:1 molar ratio) E7 and enhanced E7 (eE7; see below) or with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) probes labeled by primer extension.

Studies on protein stability.

Cells growing on six-well plates were transfected with pIRES-Neo2-wtE7 or pIRES-Neo2-eE7 by using Effectene. After 4 h, the transfection medium was aspirated and replaced with 2 ml of fresh growth medium. The medium was aspirated 24 h later, the cells were washed once with PBS, and fresh medium containing either 40 μg of cycloheximide (Sigma) per ml, 5 mM clasto-lactacystin β-lactone (Calbiochem), or both was added. At various times, the cells were washed with cold PBS, scraped from the plates, washed again with cold PBS, and lysed with SDS loading buffer. A control dish of untreated, nontransfected cells was processed similarly.

Cell viability assay.

A colorimetric assay for quantification of cellular proliferation, viability, and cytotoxicity was used. Cells were plated at a density of 10⁵/well in a 96-well plate with 200 μl of medium per well. At 24 h after transfection, 10 μl of the tetrazolium salt MTT [3(4,5-dimethylthiazol-2-yl)2,5-diphenyl-tetrazolium bromide; Roche] (5 mg/ml in PBS) per well was added, and the plate was incubated for another 2 h. Then the plate was centrifuged at 250 × g for 5 min, and the supernatant was discarded. The precipitated MTT was dissolved with 100 μl of acidic isopropanol (0.1 N HCl) per well until crystals were completely dissolved (as monitored under the microscope). Measurement of absorbance was at 570 nm. Negative control wells containing medium but not cells and positive controls consisting of nontransfected cells were run in parallel.

Nucleotide sequence accession number.

The GenBank accession number for the synthetic E7 gene is AF373107.

RESULTS

Construction of a synthetic HPV-16 E7 gene consisting entirely of preferred human codons.

HPVs have a codon usage that differs substantially from that of most human genes. A detailed analysis of the codon composition of the E7 gene showed that 13 amino acids are encoded by codons rarely used by highly expressed human genes, as occurs with other early (transforming) and late (capsid) HPV-16 genes (Fig. 1A). This can be explained by the particular base composition of HPV genomes. As shown in Fig. 1B, A and T are preferred in the third position of a majority of their codons (65 to 74%). In contrast, only 42% of codons in human genes contain A or T at the third position (31), and the number is even lower among highly expressed genes (23%). This is in good correlation with the low G+C content of these viruses (35 to 43%), which is markedly lower than the content in human genes (53 to 57%) (Fig. 1B). There are only minor differences between the main oncogenic (HPV-16 and -18) and nononcogenic HPVs (HPV-6 and -11). However, HPV-16 has the highest rate of A or T as the third base and lowest G+C content among genital HPVs (35%). Such biased use of the genetic code by HPV-16 could account, at least in part, for the small amount of E7 protein detected in transfected cells even when the E7 gene is expressed under the control of strong promoters.

To investigate whether differences in codon usage between the viral and the human genomes could account for the poor expression of E7, we generated a mutant synthetic E7 gene containing exclusively codons that are found preferentially in highly expressed human genes (17) (Fig. 1A). The resulting gene contained 63% G+C (20% more than wild-type E7) and no A or T in the third position of its codons (Fig. 1C). The synthetic E7 gene was named eE7 (for “enhanced E7”; see below and Fig. 2 to 4) and was constructed by PCR based on the sequence of the prototype virus (37) (GenBank accession number K02718) by using oligonucleotides with overlapping sequences as described in Materials and Methods. As a control, a wild-type gene which was amplified from a plasmid containing the prototype E7 gene of HPV-16 was used. In addition, a 5′ deletion mutant lacking the first 35 amino acids of E7 was constructed by PCR using the full-length synthetic eE7 and wild-type E7 genes as templates, which were named eE7Δ1 and E7Δ1, respectively. Upstream sequences flanking the coding regions of the genes were converted to a Kozak consensus translation initiation site (27). In addition, to facilitate detection of the protein, a sequence encoding the Flag epitope (DYKDDDDK) was fused to the 3′ ends of the genes. The wild-type and mutant chimeric genes were inserted separately in a mammalian expression vector (pIRESneo2) under the control of the immediate early promoter of the human cytomegalovirus (CMV) and expressed as a bicistronic mRNA, with the second part of the transcript being an NPT-II gene.

FIG. 2.

Expression of synthetic and wild-type HPV-16 E7 genes in transient-transfection assays demonstrated by Western blot analysis. HeLa cells growing on six-well plates (10⁵ cells/well) were transfected with 2 μg of plasmid pIRES-neo2 (Clontech) (pIn2) or plasmids expressing synthetic eE7 or wild-type E7 genes or their truncated derivatives eE7Δ1 and E7Δ1, respectively. The efficiencies of transfection, monitored by cotransfecting 0.5 μg of a plasmid expressing a GFP gene (pQBI-25-fPA) with each of the above plasmids and counting fluorescent cells, were identical in all cases. The cells were harvested 24 h after transfection and lysed, and half of the lysate was subjected to SDS-PAGE. After blotting, complete and truncated E7 proteins were detected with anti-Flag antibodies, and the filter was rehybridized with antiactin antibodies to show that equal amounts of protein had been loaded. A representative blot is shown.

FIG. 4.

Immunofluorescence analysis of HPV-16 E7 expression in Vero cells transfected with plasmids expressing synthetic (eE7 and eE7Δ1) or wild-type (E7 and E7Δ1) E7 genes. Cells were fixed 48 h after transfection and incubated with specific antibody as indicated, followed by Cy2-conjugated goat anti-mouse antibody. (A) The synthetic eE7 gene expressed higher levels of protein than the wild type. E7 protein was detected with anti-Flag antibody. Immunofluorescence was analyzed with a Leica TCS SP confocal microscope using a 63× objective and exactly the same laser and photomultiplier intensities, with the pinhole set at 1 Airy disk unit. Single sections corresponding to the plane of the highest intensity of fluorescence are shown. (B) Confocal sections of cells transfected with plasmids expressing full- length (eE7) and truncated (eE7Δ1) synthetic genes. Anti-Flag antibody shows the intracellular distribution of E7 (left); nuclei were counterstained with propidium iodide (middle), which labels particularly strongly the nucleoli, as seen in the merged picture (right). Both forms of E7 protein are scattered throughout the nuclear matrix and excluded from the nucleoli. (C) Confocal section of a cell transfected with a plasmid expressing a truncated E7 mutant lacking the first 35 amino acids (eE7Δ1), without the Flag tag. The protein localized to the nucleus, as detected with a monoclonal anti-E7 antibody followed by a Cy2-conjugated goat anti-mouse antibody.

Expression of synthetic and wild-type E7 in mammalian cells.

To ascertain the efficacy of codon replacement in enhancing synthesis of HPV-16 E7 protein, we compared the expression of synthetic and wild-type E7 genes after transient transfection into human, mouse, and monkey cell lines. In all cell types tested, the level of E7 protein was significantly higher from the synthetic gene. This was determined, in transient-transfection experiments, by Western blot hybridization (Fig. 2) and by protein labeling and immunoprecipitation assays (data not shown). Transfection efficiencies were monitored by cotransfecting the E7 plasmids with a plasmid expressing a GFP gene. As illustrated in Fig. 2 for HeLa cells, the synthetic eE7 gene expressed markedly higher levels of E7 protein than the wild-type E7 gene. The expression of E7 protein by the eE7 gene in different transfection experiments was estimated to be 20- to 100-fold higher than that by the wild-type E7 gene, depending on the experimental conditions. Similar differences in amounts of protein were observed in transfections with the deletion mutants eE7Δ1 and E7Δ1 (Fig. 2), indicating that subfragments of eE7 yield enhanced protein expression with the same efficiency as the full-length gene.

The full-length E7 protein was detected in Western blots as a doublet, the upper band of which represents a protein of about 19 kDa (Fig. 2), as in previous studies using anti-E7 antibodies (15). This doublet was also observed by immunoprecipitation (data not shown). It has been speculated that the less-abundant, higher-mobility peptide could represent a protein that coprecipitates with E7 or shares an epitope with it (15). However, this seems unlikely in our experiments, since we used anti-Flag antibodies for detection of E7. Rather, it may be a product resulting from intracellular processing of E7.

In a different series of experiments, we investigated whether higher protein expression by the eE7 gene was due to enhanced translation of its message. To this end, RNA and protein levels of E7 and NPT were compared after transient transfections of HeLa cells (Fig. 3). As described above, wild-type E7 as well as synthetic eE7 coding sequences were expressed with the pIneo2 plasmid from a CMV promoter as bicistronic mRNAs joined to the NPT coding sequence through an internal ribosome, entry site. Results of a representative experiment are shown in Fig. 3. As expected, the amount of E7 protein expressed by plasmid pIneo2-eE7-IRES-NPT was markedly higher (over 50-fold in this particular experiment) than that by pIneo2-E7-IRES-NPT. In contrast, the amount of NPT protein expressed from the eE7-IRES-NPT message was only slightly higher (less than 1.5-fold) (Fig. 3A). The amounts of bicistronic mRNA transcribed from each plasmid were compared in Northern blots of total and cytoplasmic RNA isolated from transfected cells of the same experiment by hybridization with a radioactive NPT probe (Fig. 3B). The eE7-IRES-NPT mRNA was more abundant than the E7-IRES-NPT mRNA (about two- to three-fold, as determined with the NIH Image program). Nearly the same differences were observed for total and cytoplasmic RNA isolates.

FIG. 3.

Expression of HPV-16 E7 and NPT mRNA and protein. HeLa cells growing on 6-cm plates (3 × 10⁵ cells/plate) were transfected with the GFP-expressing plasmid pQBI-25-fPA (−) alone or cotransfected with this plasmid and a plasmid carrying either synthetic (pIn2-eE7) or wild-type (pIn2-E7) E7. The latter plasmids are transcribed into bicistronic mRNAs harboring sequences for expression of the E7 and NPT proteins. After 24 h, cells were harvested and lysates were processed for total or cytoplasmic RNA purification and for protein extraction. (A) Western blot analysis. Note the remarkably small amount of E7 protein expressed in cells transfected with pIn2-E7 compared with that in cells transfected with pIn2-eE7. In contrast, the difference in the amount of NPT expressed by the two plasmids is much lower (about 1.5 times higher by pIn2-eE7, as estimated densitometrically). Hybridization with GFP antibodies showed equal efficiency of transfection. (B) Northern blot analysis showing levels of eE7-IRES-NTP and E7-IRES-NTP bicistronic RNA in transiently transfected cells. Total RNA was loaded at 10 μg per lane. Detection was with [α-³²P]dCTP-labeled probes. A labeled NPT probe was used to detect the eE7-IRES-NPT and E7-IRES-NPT messages (arrowhead). The blot was stripped and rehybridized with a radioactive GFP probe to show nearly equal transfection efficiencies (middle). Staining of the gel with ethidium bromide after electrophoresis is shown in the lower panel. The positions of the 28S (4.7-kb) and 18S (1.9-kb) rRNAs are indicated.

These data suggested that mRNA containing synthetic eE7 was translated more efficiently into E7 protein (but not NPT protein) than the message bearing wild-type E7. The differences seen in the amounts of the respective mRNAs were too low to explain the higher levels of E7 protein expressed from eE7 and may reflect a higher stability of the eE7 mRNA resulting from its secondary structure or its longer binding to polysomes, thus remaining inaccessible to RNase.

Intracellular localization of E7.

HPV-16 E7 is a phosphoprotein associated with the nuclear matrix (15) that has been also found in the cytoplasm (39). The high expression of E7 protein from the eE7 gene allowed us to assess its intracellular localization in transiently transfected cells. As shown in Fig. 4, E7 localized mainly to the nucleus, although it could also be detected with a pattern of intensely fluorescent dots in the cytoplasm surrounding the nucleus. Within the nucleus, the protein appeared to be uniformly distributed through the nuclear matrix. However, a localization of E7 in the nucleoli was not observed. Rather, E7 appeared to be completely excluded from nucleolar structures, as shown in confocal sections of specimens in which DNA and RNA were stained with propidium iodide to label the nucleoli (red fluorescence in Fig. 4B). The same localization pattern was observed in HeLa, SiHa, and other human cell lines. Like the full-length E7, the E7Δ1 deletion mutant localized mainly in the nuclei of transfected cells and was also excluded from the nucleoli (Fig. 4B, lower panel). The possibility that E7 was excluded from the nucleoli due to the Flag tag was discarded by expressing E7 without tag and detecting the protein with anti-E7 antibodies. As shown in Fig. 4C, E7 was again excluded from the nucleoli. These data are in contrast with a previous report describing nucleolar localization of E7 (53) and in agreement with previous work showing association of E7 with the nonchromatin nuclear structure (15).

Half-life of the E7 protein in eE7-transfected cells.

Next, we examined whether the enhanced steady-state levels of E7 protein in cells transfected with eE7 might be due, at least in part, to delayed degradation of the protein. To this end, the stability of the protein was investigated in cells in the absence and presence of the protein synthesis inhibitor cycloheximide and the proteasome inhibitor lactacystin. As the levels of E7 protein expressed by the eE7 gene were sufficiently high for detection in Western blots, a nonradioactive protocol was chosen for analysis of protein expression after SDS-PAGE. Cotransfection with plasmid pQBI-25-fPA expressing GFP was used to confirm equality of transfection efficiency by counting green fluorescent cells. Actin was used as an internal control for the amount of protein loaded per lane. Inhibition of both protein synthesis and degradation caused a reduction in levels of the E7 protein to about 50% in 1 h (Fig. 5, compare lane 3 with lanes 2 and 4). During the same time, no change in the actin content of the cells was observed, as could be expected from its longer half-life (nearly 6 h) (6). These results are in agreement with previous work showing that E7 is degraded via ubiquitin-mediated proteolysis, with an estimated half-life of about 30 to 40 min (35, 39).

FIG. 5.

The half-life of E7 was not affected by its enhanced expression. HeLa cells were transiently cotransfected with plasmids pIneo2-eE7 and pQBI-25-fPA (expressing GFP). At 24 h after transfection (Tf), protein synthesis was inhibited with cycloheximide (Chx), and protein degradation was blocked with clasto-lactacystin β-lactone (β-Lact), a cell-permeative proteasome inhibitor, as indicated. Then the cells were lysed, subjected to SDS-PAGE, and transferred to polyvinylidene difluoride membranes. Anti-Flag antibody was used to detect E7 protein, and antiactin was used to show equality of protein loading. Equal transfection efficiency was monitored by counting green fluorescent cells. A representative blot is shown.

Effect of high expression of E7 protein on cell viability.

In the course of transient-transfection experiments with the synthetic eE7 gene, we observed that increasing numbers of dead cells detached from the plates after the second day of transfection, a phenomenon that was not seen in transfections with the wild-type E7 gene. This effect was observed in HeLa cells as well as C-33 cells, another cervical carcinoma cell line that does not carry any known HPV DNA. Furthermore, our attempts to obtain a stable cell line with the plneo2-eE7 plasmid failed. In these experiments, over 30 clones resistant to neomycin were found not to contain E7 (data not shown), suggesting selection against cells expressing this protein. Silencing of E7 expression in those clones might be due to point mutations, since no deletions were noted in fragments amplified by reverse transcriptase PCR. These results prompted us to use the MTT assay to investigate the possibility that sustained elevated levels of E7 protein may cause cells to die (18, 29). MTT is a tetrazolium salt that is cleaved to formazan by the succinate reductase system, which is part of the respiratory chain of mitochondria and is active only in viable cells. Accordingly, the amount of formazan formed, which can be measured in a colorimetric assay, is in direct correlation with the number of living cells in a culture.

HeLa cells were transiently cotransfected with plasmids expressing synthetic HPV-16 eE7 (pIneo2-eE7), wild-type E7 (pIneo2-E7), or plasmid pIneo2 as a control and with a plasmid expressing GFP (pQBI25-fPA) as an indicator of the efficiency of transfection. This ranged from 20 to 25%, as determined by counting the number of green fluorescent cells at 24 h and dividing it by the number of seeded cells. The cells were processed for the MTT viability assay 24, 48, and 72 h after transfection. As shown in Fig. 6, comparable numbers of viable cells were observed with the three plasmids 24 h after transfection. However, pIneo2-eE7-transfected cells underwent decreases in viability of 15 and 26% after 48 and 72 h, respectively, compared to mock (pIneo2)-transfected cells. In contrast, pIneo2-E7-transfected cells showed survival rates of 12 and 15% over that of the control after 24 and 48 h, respectively. Considering the transfection efficiency rates, it can be concluded that most cells carrying pIneo2-eE7 were not viable after 72 h. Furthermore, cells transfected with pIneo2-E7 seemed viable and proliferating at about a 37% higher rate than nontransfected cells, as the 12% increase in the total number of viable cells should be attributed to 25% transfected cells.

FIG. 6.

Effect of HPV-16 E7 expression on viability of HeLa cells. HeLa cells were seeded in 96-well microplates (5,000 cells per well in 200 μl of medium). One day later, cells were cotransfected with plasmid pIneo2, pIneo2-eE7, or pIneo2-E7 and with plasmid pQBI25-fPA expressing GFP, which served to monitor efficiency of transfection. At the indicated time points after transfection, cell viability was measured by the MTT method (see the text). The efficiency of transfection (number of GFP-expressing cells divided by number of seeded cells), estimated at 24 h, ranged from 20 to 25%, with minimal variation from well to well. Results are averages of three independent transfection experiments. Note the lower survival rates of cells transfected with pIneo2-eE7.

DNA immunization of mice with plasmid expressing E7.

The ability of synthetic and wild-type E7 genes to induce effective anti-E7 humoral immunity in mice was tested after injection of the corresponding plasmids. Three groups of naïve mice were injected three times subcutaneously at 2-week intervals with either PBS or plasmid DNA expressing eE7 or wild-type E7. The titers of antibody elicited against E7 were markedly higher in mice injected with plasmid expressing eE7 than in those inoculated with the wild-type E7 gene, in which the antibody was almost undetectable (Fig. 7). Titers ranged from 100 to 400. These relatively low titers are likely due to the small amounts of plasmid injected with the method used and the fact that the plasmid solutions were injected subcutaneously. However, this approach was preferred to methods using larger amounts of plasmids and more invasive injections (e.g., intraperitoneal), because it resembles better the standard immunization method used in clinical practice.

FIG. 7.

Plasmid expressing eE7 induces anti-E7 specific antibodies in mice. Animals were inoculated three times subcutaneously either with no plasmid (PBS) or with 4 μg of plasmid expressing wild-type E7 or synthetic eE7 genes. Six weeks after the first inoculation, mice were bled, and the sera were analyzed for E7-specific antibodies by ELISA. Antibody titers are expressed as the reciprocal of the lowest serum dilution that gave an absorbance of 0.1 unit above the background. Each symbol represents an antibody titer for an individual mouse.

DISCUSSION

Compared to human coding sequences, HPV genomes have at least three distinct features, as can be deduced from Fig. 1A and B. First, the overall G+C content of HPVs is significantly lower than that of human genes. Second, HPV genes contain A or T in the third position of two-thirds of their codons, while only 40% (average) and 23% (highly expressed) of human codons have A or T in the third position (17, 31). Third, HPV genes have a codon usage that differs substantially from that of human genes. For instance, all codons used preferentially by HPV-16 fall among those less frequently used by highly expressed human genes (Fig. 1A), with slight differences between early and late genes (e.g., codons for amino acids A, E, and P). Interestingly, HPV-16 is the genital HPV type with the lowest G+C content and highest proportion of A+T at the third position of its codons. Nevertheless, although they are markedly different from human genes, the HPV types compared in Fig. 1B have only minor differences between them, in spite of their very different biological and clinical significance (types 16 and 18 are oncogenic and types 6 and 11 are nononcogenic).

A likely consequence of having codon biases divergent from that of human genes is that the tRNA species required for translation of HPV mRNAs may be underrepresented in the host cell and hence viral protein expression severely limited. This would explain why high levels of viral mRNA are accompanied by low levels of viral protein in HPV-bearing tumors and cell lines. Furthermore, HPV-16 variants carrying mutations that optimize codon usage are not rare in cervical cancers (4, 13).

It was the aim of this study to determine to what extent such striking differences between HPV and human codon usage could influence expression of HPV-16 genes. The HPV-16 E7 gene was chosen for this purpose because it is a major HPV oncogene, has a short sequence, and is known to express very small amounts of the protein (15) (Fig. 2). A synthetic E7 gene (eE7) whose codon usage was conveniently optimized to contain exclusively codons used preferentially by highly expressed human genes was generated and characterized (Fig. 1A). The resulting gene (Fig. 1C) has a G+C content above that of human genes and does not contain A or T in the third position of its codons, in contrast to human genes, which have A or T in that position in 23% of their codons. As expected, the eE7 gene expressed larger amounts of E7 protein than the wild-type E7 gene. The difference in expression was estimated, in transiently transfected cells, as 1 to 2 orders of magnitude higher (Fig. 2 to 4). Interestingly, enhancement of protein expression was observed in human, monkey, and mouse cells.

The use of a bicistronic expression system proved useful for determining the different translation rates of synthetic and wild-type E7 having NPT as a common internal control. Thus, translation of synthetic eE7 from bicistronic eE7-IRES-NPT mRNA produced about 50-fold more protein than translation of the wild-type gene contained in the E7-IRES-NPT message (Fig. 3A). In contrast, the amount of NPT protein was only 1.5 times higher, indicating that translation of eE7 but not NPT was significantly enhanced from the eE7-IRES-NPT mRNA. Additional evidence in support of enhanced translation of eE7 came from quantitative analysis of Northern blots using NPT as probe, which should hybridize equally with both bicistronic mRNA species, showing a two- to threefold larger RNA amount for eE7-IRES-NPT than E7-IRES-NPT (Fig. 3B). This difference was too small to account for the 50-fold increase in E7 protein expression seen in Fig. 3A. However, this result was somewhat surprising, as the transcription rates of these plasmids were expected to be the same, since they carry the same CMV promoter. Therefore, we concluded that the eE7-IRES-NPT mRNA should have a longer half-life than its E7-IRES-NPT counterpart, likely because it remains bound to polyribosomes longer, and hence inaccessible to degradation, and/or because the eE7 sequence, with a G+C content exceeding by 20% that of wild-type E7, confers stability on the bicistronic mRNA.

Another parameter that could contribute to the increase in the steady-state levels of E7 protein could be an extended half-life resulting from its higher expression levels. However, the half-life of E7 in cells transfected with pIneo2-eE7 was estimated to be close to that described previously for cells expressing the wild-type E7 gene (35) (Fig. 5). Therefore, we concluded that enhanced translation of the eE7 message should account for the high levels of E7 protein that observed.

The results of three series of experiments carried out in this study illustrated the utility of our synthetic eE7 gene. In the first series, the eE7 gene was used to assess the intracellular localization of the protein (Fig. 4). The main functional property of E7 is to deregulate control of cell cycle progression, thereby inducing cells to enter S phase. This is achieved through interaction of E7 with nuclear proteins such as retinoblastoma protein, which suggests its nuclear localization. However, E7 has also been shown to interact with M2-pyruvate kinase (57), suggesting that it also has cytoplasmic functions. Indeed, E7 has been localized to the cytoplasm (39) but has also been reported to be in the nucleus (15) and to have a nuclear localization signal (12). We found that E7 localized mainly in the nucleus with a pattern suggestive of its association with the nuclear matrix (Fig. 4), in agreement with previous reports (15). However, in a previous study the E7 protein was found primarily bound to nucleoli (53). This discrepancy may be due to artifactual binding of the antibody used in that study to nucleoli. This view is supported by previous studies by Greenfield et al. (15) and Kanda et al. (25), who observed that epitopes in two immunodominant regions of E7 could not be detected with monoclonal antibodies unless chromatin was removed or a polyclonal antibody was used instead, suggesting that the epitopes were masked by cellular proteins. Furthermore, in a recent study, Guccione et al. (16) described a pattern of expression for a hemagglutinin-tagged E7 that matches the one described here.

The nuclear localization of the E7Δ1 deletion mutant was somehow unexpected, since it was reported previously that a peptide consisting of amino acids 16 to 41 of E7 localized to the nucleus (12) and residues 1 to 35 are absent in the E7Δ1 mutant. However, E7 does not contain any consensus nuclear localization signal, and the mechanisms of its import into the nucleus are still unknown.

Second, the eE7 gene was used to determine the effect of elevated levels of E7 protein on cell viability. Transient-transfection experiments showed that the high levels of E7 protein expressed by the eE7 gene eventually caused the death of all transfected cells after 72 h of transfection. In contrast, lower levels of protein expressed by wild-type E7 seemed not to cause damage but rather stimulated cell proliferation (Fig. 6). These observations suggest that, above a threshold, the ability of the E7 protein to interfere with essential cellular functions exceeds its capacity to stimulate cell proliferation. It has been shown that expression of HPV-16 E7 induces apoptosis in human fibroblasts and keratinocytes (1, 20, 44, 51) and in the lens of transgenic mice (22, 33). However, E7 has also been found to inhibit apoptosis induced by TNF-α in human fibroblasts (46), and E7-binding peptide aptamers induced apoptosis in cervical cancer cells (32). According to our results, differences in the level of E7 protein expressed in the various systems used for those studies may provide an explanation for such discrepancies. Nevertheless, in view of the transforming properties of E7 and the relevance of the cellular proteins with which it interacts, the elucidation of mechanisms by which E7 induces cell death is of considerable interest. To this end, the synthetic eE7 gene described here could be very useful if the appropriate promoters are used for its expression.

In a third use of the eE7 gene in this study, it proved effective for DNA immunization studies in mice. The results of the analysis of the humoral response in animals inoculated with the plasmids described here showed that the eE7 gene had a higher immunogenic capability than wild-type E7 (Fig. 7). The higher antibody response in mice injected with plasmids carrying the synthetic eE7 revealed the importance of ensuring high levels of expression of the target protein. Further work should determine the ability of eE7 to trigger cytotoxic immune responses to E7. Since much effort is being applied to the development of vaccines against oncogenic HPVs, the eE7 synthetic gene described here demonstrates that codon optimization may be an essential strategy.

Finally, taken together, our data provide support for the hypothesis that selective pressure has forced HPVs to evolve a codon usage substantially different from that of human genes in order to minimize deleterious effects to the host due to excessive expression of early viral genes, thereby evading the immune system. While it may seem contradictory that human viruses have conserved codon biases distinct from that of their host cells, this may be the consequence of selective pressure to minimize the pleiotropic effects of the viral proteins and presentation of viral antigen that may result in more effective immune response. In order to prove this hypothesis, these studies need to be extended to other HPV synthetic genes generated with the same criteria as the eE7 gene.