Detection of Human Papillomavirus Type 18 E7 Oncoprotein in Cervical Smears: a Feasibility Study

Abstract

Persistent infections by high-risk human papillomaviruses (HPVs) are the main etiological factor for cervical cancer, and expression of HPV E7 oncoproteins was suggested to be a potential marker for tumor progression. The objective of this study was to generate new reagents for the detection of the HPV18 E7 oncoprotein in cervical smears. Rabbit monoclonal antibodies against recombinant E7 protein of HPV type 18 (HPV18) were generated and characterized using Western blotting, epitope mapping, indirect immunofluorescence, and immunohistochemistry. One clone specifically recognizing HPV18 E7 was used for the development of a sandwich enzyme-linked immunosorbent assay (ELISA). The assay was validated using recombinant E7 proteins of various HPV types and lysates from E7-positive cervical carcinoma cells. A total of 14 HPV18 DNA-positive cervical swab specimens and 24 HPV DNA-negative-control specimens were used for the determination of E7 protein levels by the newly established sandwich ELISA. On the basis of the average absorbance values obtained from all 24 negative controls, a cutoff above which a clinical sample can be judged E7 positive was established. Significant E7 signals 6- to 30-fold over background were found in 7 out of 14 abnormal HPV18 DNA-positive cervical smear specimens. This feasibility study demonstrates for the first time that HPV18 E7 oncoprotein can be detected in cervical smears.

INTRODUCTION

Persistent infections by human papillomaviruses (HPVs) are the main etiologic factor for cervical cancer (41). Approximately 85% of cervical cancers are squamous cell carcinomas (SCCs) which arise from dividing keratinocytes in the squamous epithelium of the ectocervix and 15% are adenocarcinomas (ACs) which arise from glandular cells located in the endocervix (3, 25). About 40 HPV genotypes that can infect epithelial squamous and glandular cells in the cervical mucosa have been described. On the basis of epidemiological and biochemical data, only a subgroup of HPV types, referred to as high-risk HPVs, is associated with intraepithelial lesions that have a high potential for progression to invasive carcinoma. Infections by high-risk HPV genotypes have been detected in virtually all cervical cancers (37). At least 15 high-risk HPV types have been associated with these cancers. HPV type 16 (HPV16) and HPV18 are the most prevalent genotypes worldwide in SCCs as well as in ACs (23).

A persistent infection with oncogenic HPVs is necessary for the development of cervical precancer and cancer (3, 37, 40, 41). Initial events of cervical carcinogenesis after viral infection by high-risk HPV types are specific changes that overcome the transcriptional control of viral gene expression in the infected keratinocytes (40). Inactivation of these cellular control functions permits deregulated transcription of the early viral genes E6 and E7. This event triggers reprogramming of cell proliferation, apoptosis, differentiation, metabolism, epigenetic reorganization, and genomic instability (21). These changes can support the integration of episomal HPV genomes into chromosomes of the host cell (19, 33) and contribute to further overexpression of the viral genes E6 and E7 (20, 28). This is consistent with an increase of the high-risk E7 protein levels during early steps of carcinogenesis in cells of the cervical squamous epithelium (11). High-risk E7 in cooperation with high-risk E6 can efficiently immortalize human primary keratinocytes (16, 22), and the consistent overexpression of these two oncogenes is required to induce and to maintain the transformed phenotype of cervical cancer cells (36). High-risk E7 is the major HPV oncoprotein (21). It acts efficiently in transformation of immortalized rodent cells (4), and E7 alone can immortalize primary human cells (38). Moreover, the expression of E7 alone is sufficient to induce invasive cervical cancers in transgenic mice treated with low doses of estrogens (27). Early studies have shown that immortalization by the E7 oncoprotein involves its ability to bind and thereby functionally inactivate cell cycle regulatory proteins such as the retinoblastoma tumor suppressor protein (9, 11). Further work has demonstrated that E7 is located in both the cytoplasm and the nucleus (1, 6, 8, 11, 14, 17, 24, 26, 29, 34, 39, 42). In keeping with these findings, it has been shown that HPV16 E7 is an integral part of many cellular protein complexes in the cytoplasm as well as in the nucleus and has multiple biochemical functions in the deregulation of pathways necessary for the oncogenic potential of the virus (reviewed in references 21 and 42). Thus, the levels of E7 oncoproteins of carcinogenic HPV types could be specific markers for the detection of cervical precancer and cancer.

Even though screening with cervical cytological testing (Papanicolaou [Pap] test) has been available for over 50 years, cervical cancer remains the second most common cancer in women worldwide (5). Cytological analysis of cervical smears is characterized by high rates of false-positive and false-negative results, which might be improved by the introduction of molecular tumor markers for cervical carcinoma. Thus, there is a need for new technologies for cervical cancer screening. We have demonstrated in previous studies that high-risk HPV E7 proteins are regularly expressed in cervical SCCs and ACs and in their high-grade precursor lesions, suggesting that high-risk E7 oncoproteins are necessary for this cancer and may serve as new tumor markers (8, 11, 12, 26). In the current communication, we addressed the question if E7 oncoproteins of the high-risk virus HPV18 can be detected in cervical smears by an innovative enzyme-linked immunosorbent assay (ELISA).

MATERIALS AND METHODS

PCR-based HPV typing. (i) Processing of tissue samples.

Formalin-fixed, paraffin-embedded tissues were processed with a QIAamp tissue kit according to the manufacturer's instructions (Qiagen, Hilden, Germany).

(ii) Processing of cervical smears.

Two consecutive cytobrush samples were taken under colposcopic inspection. The first was used for diagnostic Pap and HPV testing. The second was immediately placed in sample buffer (0.1% Tween 20, 1× phosphate-buffered saline [PBS]) and frozen at −80°C for analysis in sandwich ELISA. For DNA extraction, the first sample was soaked for 10 min in 2 ml PBS and vortexed vigorously. Extracts were centrifuged at 6,000 rpm for 5 min, the supernatant was discharged, and the sediments were resuspended in 400 μl PBS. For DNA extraction, 200 μl was processed according to the QIAamp minikit protocol (Qiagen, Hilden, Germany). Purified DNA was eluted 2 times with 80 μl, and 10 μl was used per 50-μl PCR mixture. Total cellular DNA was used in the GP5+/GP6+ general primer PCR, and the amplicons were used for HPV typing using an enzyme immunoassay (EIA) with different HPV type-specific oligonucleotides as described previously (13, 18). Alternatively, HPV genotyping was performed by the BSGP5+/BSGP6+ PCR/multiplex HPV genotyping (MPG) assay comprising the BSGP5+/BSGP6+ PCR, which homogeneously amplifies all known genital HPV types, generating biotinylated amplimers of ∼150 bp from the L1 region and β-globin (31, 32), and an MPG assay with bead-based xMAP Luminex suspension array technology, which is able to simultaneously identify and quantify 51 HPV types and the β-globin gene (30, 31) (M. Schmitt et al., submitted for publication). By normalizing the HPV18 median fluorescence intensity (MFI) signal with the β-globin MFI signal, viral load values were determined. Copy numbers per cell were calculated on the basis of external standards previously quantified with an HPV18-specific quantitative PCR and BSGP5+/BSGP6+ PCR/MPG (Schmitt et al., submitted).

Tissue specimens.

Paraffin-embedded specimens with cervical SCCs and ACs were obtained from the departments of Obstetrics and Gynecology, Medical University of Innsbruck, Innsbruck, Austria. Cervical smears were obtained using cytobrush sampling from the intra- and extracervical area under colposcopic inspection after treatment with 5% acetic acid. Two samples were taken, and the second was used for biochemical analysis in this study.

Purification of HPV E7 oncoproteins and generation of rabbit monoclonal antibodies (RabMAbs) against HPV18 E7 protein.

E7 proteins of HPV types 6, 11, 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, and 59 were purified as described previously (10). Purified HPV18 E7 protein was used to generate anti-HPV18 E7 RabMAbs in collaboration with Epitomics (Burlingame, CA).

Indirect immunofluorescence experiments and quantification of E7 fluorescence.

Confocal immunofluorescence microscopy was performed essentially as described previously (12, 30). Briefly, all cells (transiently transfected U-2OS, CaSki, and HeLa cells) were fixed with 4% (wt/vol) phosphonoformic acid (PFA)–1× PBS and permeabilized for 3 min with 0.1% (wt/vol) sodium citrate–0.3% (vol/vol) Triton X-100. After blocking with 1× PBS–1% bovine serum albumin (BSA) for 30 min at room temperature, the cells were incubated for 1 h at 37°C with the rabbit monoclonal anti-HPV18 E7 antibody or a mixture (1:1) of polyclonal goat anti-HPV16 E7 and anti-HPV18 E7 antibodies in 1× PBS–1% BSA. Cells were washed in 1× PBS, incubated for 45 min with fluores-cein isothiocyanate (FITC)-conjugated swine polyclonal anti-rabbit immunoglobulin (IgG; DakoCytomation, Heidelberg, Germany), FITC-conjugated donkey anti-goat IgG (Jackson ImmunoResearch Laboratories, West Grove, PA), or Alexa Fluor 488-conjugated goat anti-rabbit IgG (Invitrogen, Carlsbad, CA). TO-PRO3 (Invitrogen, Carlsbad, CA) was used as nucleic acid counterstain. After fixation, slides were processed for fluorescence microscopy using a confocal laser scanning system. Quantification of E7 fluorescence intensity was carried out using representative confocal images and ImageJ software. An area of 549 by 549 pixels per picture was selected and displayed as a surface plot representing a three-dimensional visualization of intensities for each colored pixel.

siRNA-mediated knockdown of HPV18 E7 in HeLa cells.

HeLa cells were grown in 6-well plates and transfected with small interfering RNA (siRNA) targeting HPV18 E7 or scrambled siRNA (Dharmacon Inc., Lafayette, CO), upon reaching 50% confluence. The transfection procedure was carried out using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) and 100 pmol siRNA per well. After 24 h and 48 h, cells were either scraped off and lysed in NP-40 buffer containing protease inhibitors (complete, EDTA free; Roche Diagnostics, Germany) for the generation of Western blotting lysates or fixed with 4% (wt/vol) PFA–1× PBS and processed for immunofluorescence staining as described above.

Immunohistochemical detection of HPV18 E7 in cervical cancer biopsy specimens.

Immunohistochemistry was performed on paraffin-embedded tissue sections derived from cervical conization or hysterectomy specimens. Two-micrometer sections were mounted on slides, deparaffinized in xylene (2 × 12 min), incubated for 5 min each in 100%, 90%, 80%, 70%, and 50% isopropanol for rehydration, and processed for antigen retrieval by treatment for 1 h in a steamer in target retrieval solution (pH 6.1; S1700; DakoCytomation, Heidelberg, Germany) for the detection of HPV E7. Slides were cooled for 20 min at 4°C. Endogenous peroxidase activity was blocked by incubation in 20% H₂O₂ for 15 min. Sections were washed twice in water and subsequently incubated for 15 min in blocking buffer (10% goat or rabbit serum [Dako-Cytomation, Heidelberg, Germany] in 1× Tris buffer [640 mM Tris-HCl, 150 mM NaCl, pH 7.5]). The sections were incubated with the monoclonal rabbit anti-HPV18 E7 antibody for 1 h at room temperature in 5% BSA–1× TBS in a wet chamber. Samples were rinsed twice in 1× TBS–0.1% Tween 20 and incubated with secondary IgGs (rabbit anti-goat biotinylated; 1:600; DakoCytomation, Heidelberg, Germany) for 45 min at room temperature. After washing in 1× TBS–0.1% Tween 20 buffer, samples were incubated with an ultrasensitive streptavidin-peroxidase polymer (S2438; diluted 1:500; Sigma, Vienna, Austria) for 30 min at room temperature. Bound antibodies were visualized with diaminobenzidine solution (Sigma, Vienna, Austria) as substrate. Counterstaining was performed with Hemalaun solution (Merck, Vienna, Austria). The specimens were incubated in increasing concentrations of isopropanol and mounted using Entellan mounting medium (Merck, Vienna, Austria).

Preparation of lysates from cells and clinical samples. (i) Preparation of lysates from cultured tumor cells.

Mixtures of trypsinized cells were resuspended in 1 ml ice-cold lysis buffer (0.1% Tween 20, complete EDTA-free protease inhibitors [Roche Diagnostics, Germany], 1× PBS) and frozen at −80°C. Samples were thawed immediately before use and centrifuged (20 min, 13,000 rpm, 4°C). Resulting supernatants were applied to sandwich ELISA or used for measurement of green fluorescence after addition of EvaGreen DNA stain (Jena Bioscience, Jena, Germany).

(ii) Preparation of lysates from clinical samples.

The clinical samples were thawed immediately before use, and 20 μl of 50× protease inhibitor stock solution (complete EDTA-free protease inhibitors; Roche Diagnostics, Germany) was added per collecting tube (1 ml sample). Remaining liquid was wiped off the brush at the edge of the tube, and the brush was discarded. The sample was centrifuged (20 min, 13,000 rpm, 4°C), and the supernatant was used for measurement of the E7 expression level in the sandwich ELISA and for measurement of its fluorescence after addition of EvaGreen.

Detection of HPV18 E7 in cell lysates and cervical swabs by sandwich ELISA.

One hundred microliters coating buffer (0.1 M NaHCO₃, pH 9.6) containing 0.25 μg polyclonal goat anti-HPV16 E7 antibody and 0.25 μg polyclonal goat anti-HPV18 E7 antibody was added to each well, and the 96-well plate (Maxisorp F; Nunc, Vienna, Austria) was incubated overnight at 4°C. Wells were washed 3 times with washing buffer (0.05% Tween 20, 1× PBS). Three hundred microliters universal casein diluent/blocker (SDT, Baesweiler, Germany) was added to each well, followed by 2 h of incubation at room temperature. Wells were washed 3 times with washing buffer. Recombinant E7 protein, cell lysate, or lysate from clinical samples was added to each well. In the case of recombinant E7 protein, 200 μl recombinant E7 protein diluted with lysis buffer containing no protease inhibitors (0.1% Tween 20, 1× PBS) was added to each well, followed by 30 min incubation at room temperature. Subsequently, the wells were aspirated and the process was repeated. In the case of cell lysate and lysate from clinical samples, 100 μl lysate was added per well, followed by 1 h of incubation at room temperature. After incubation of recombinant E7 protein, cell lysates, or lysate from clinical samples, wells were washed 3 times with washing buffer. One hundred microliters biotinylated anti-HPV18 E7 RabMAb diluted with universal casein diluent/blocker (0.2 μg/ml) was added to each well, followed by 1 h of incubation at room temperature. Wells were washed 3 times with washing buffer. One hundred microliters streptavidin–poly-horseradish peroxidase (poly-HRP; PolyHRP40) conjugate (SDT, Baesweiler, Germany) diluted with universal casein diluent/blocker (0.1 μg/ml) was added to each well, followed by 1 h of incubation at room temperature. Wells were washed 6 times with washing buffer. One hundred microliters tetramethylbenzidine (TMB) detection reagent es(HS)TMB (SDT, Baesweiler, Germany) was added to each well, followed by 30 min incubation in the dark at room temperature. Fifty microliters stop solution (2 N H₂SO₄) was added to each well, and absorbance was measured at 450 nm using a multilabel plate reader (Victor X5; Perkin Elmer, Vienna, Austria).

EvaGreen fluorescence of cell lysates and clinical samples.

One hundred microliters cell lysate or lysate of clinical samples was transferred to wells of a black 96-well plate (F96 MicroWell plate black; Nunc, Vienna, Austria). The autofluorescence of the lysate was measured, 2 μl 50× EvaGreen DNA stain solution (Jena Bioscience, Jena, Germany) was added, and the plate was incubated for 10 min in the dark at room temperature. Subsequently, the EvaGreen fluorescence was measured. The autofluorescence and EvaGreen fluorescence of the samples were measured using a multilabel plate reader (Victor X5; Perkin Elmer, Vienna, Austria) with the following settings: excitation, 480 ± 31 nm; emission, 535 nm; emission side, above; measurement time, 0.1 s; and measurement height, 4 mm.

RESULTS

Production and characterization of HPV18 E7-specific rabbit monoclonal antibodies.

Recombinant E7 protein of HPV18 was expressed in bacteria, purified to homogeneity as described previously (10), and used for the immunization of rabbits. Rabbit hybridoma cell lines were generated, and their supernatants were screened by standard procedures for the presence of RabMAbs that would recognize HPV18 E7. Out of 80 clones, RabMAb 143-7 was selected. The epitope recognized by 143-7 was determined by PepScan technology. This antibody recognized one major epitope in the N-terminal part of HPV18 E7 (Fig. 1A) but did not recognize any epitopes of E7 proteins of the HPV16 subfamily (data not shown). The specificity of RabMAb 143-7 was also analyzed by Western blotting. Purified recombinant E7 protein of HPV types 16 and 18 was separated by SDS-PAGE and analyzed by Western blotting using RabMAb 143-7. No signal was obtained for recombinant E7 protein of HPV16, whereas positive signals were obtained for recombinant HPV18 E7 (Fig. 1B). Similarly, specific signals were obtained with RabMAb 143-7, when extracts of transiently transfected U-2OS cells expressing FLAG-tagged versions of HPV18 E7 were analyzed. Finally, endogenous E7 proteins of HPV18-positive HeLa cells but not of HPV16-positive CaSki cells were detectable by Western blotting (Fig. 1B). To analyze the suitability of RabMAb 143-7 for immunofluorescence analysis, U-2OS cells were transfected with expression vectors coding for E7 proteins of HPV18 and HPV16 and analyzed by indirect immunofluorescence. HPV18 E7 protein gave rise to strong and highly specific signals in transiently transfected cells (Fig. 2A). RabMAb 143-7 was also able to detect endogenous E7 protein in HeLa cells by immunofluorescence (Fig. 2B). The heterogeneous pattern of E7 detection in HeLa cell populations probably reflects changes in the cell cycle state of such cells (A. Kaiser et al., unpublished findings). We cannot, however, exclude the possibility that the pattern is additionally shaped by variable accessibility of antibodies to E7, dependent on its interaction with cellular proteins. Consistent with the results obtained by Western blotting, no cross-reactivity was obtained by immunofluorescence using U-2OS cells expressing HPV16 E7 (Fig. 2A), and accordingly, no signal was obtained using CaSki cells (HPV16 positive) and U-2OS cells (HPV negative) (Fig. 2B). When HPV18 E7 expression was ablated using E7-specific siRNA, Western blotting analysis demonstrated a significant (roughly 2-fold; Fig. 2C) reduction in the level of HPV18 E7 protein in E7 siRNA-treated cells relative to controls. This was confirmed by immunofluorescence staining and subsequent confocal analysis, which revealed a significant reduction of E7 fluorescence signal in a large number of cells relative to control cells (Fig. 2C). Together these data establish the high sensitivity and specificity of RabMAb 143-7 in indirect immunofluorescence assays. In the next step, we analyzed whether RabMAb 143-7 would be suitable to detect HPV18 E7 in cervical cancer biopsy specimens. Sections of HPV18 DNA-positive squamous cell carcinoma and adenocarcinoma biopsy specimens were stained with RabMAb 143-7. This experiment revealed a strong and specific staining of the tumor islets in each case, with no staining of the adjacent stroma (Fig. 2D), and HPV18 E7-positive cells were also detected in cervical intraepithelial neoplasia grade 3 (CIN3) and adenocarcinoma in situ (ACIS) biopsy specimens (Fig. 2D). Normal epithelium stained negative in each case, suggesting that RabMAb 143-7 is suitable for immunohistochemical detection of E7 in clinical samples.

Fig 1.

Biochemical characterization of RabMAb 143-7. (A) (Upper) RabMAb 143-7 as well as two unrelated monoclonal antibodies were used for epitope mapping on the HPV18 E7 sequence provided as multiple 11-mer overlapping peptides; (lower) the data establish the N-terminal peptide PKATLQDIVL of HPV18 E7 to be the epitope recognized by RabMAb 143-7. (B) Purified, recombinant E7 protein of HPV16 and HPV18 was prepared, and 10 ng each was loaded on a 12.5% SDS gel and probed with RabMAb 143-7 in Western blots (left). Cellular extracts were prepared from HeLa, CaSki, and U-2OS cells, with U-2OS cells being transfected with an expression vector encoding FLAG-tagged HPV18 E7. Cell lysates were separated by SDS-PAGE and probed in Western blots with RabMAb 143-7.

Fig 2.

Characterization of RabMAb 143-7 in immunofluorescence and immunohistochemistry. (A) U-2OS cells were transfected with expression vectors coding for HPV18 E7, HPV16 E7, and empty vector, as indicated. The transfected cells were seeded on coverslips and analyzed by indirect immunofluorescence using RabMAb 143-7, followed by Alexa Fluor-conjugated secondary antibodies. DNA was counterstained in red using TO-PRO3. (B) HeLa cells, CaSki cells, and U-2OS cells were seeded on coverslips and subjected to immunofluorescence staining, using RabMAb 143-7. TO-PRO3 was used as a DNA counterstain. (C) HeLa cells were transfected with siRNA against HPV18 E7 or scrambled siRNA (Dharmacon, Inc.^, Lafayette, CO), as indicated. At 24 h and 48 h after transfection, cells were lysed and subjected to Western blotting analysis, implicating RabMAb 143-7 and anti-glyceraldehyde-3-phosphate dehydrogenase (anti-GAPDH) antibody (upper set). Simultaneously, transfected cells grown on coverslips were analyzed by immunofluorescence using RabMAb 143-7 and Alexa Fluor-conjugated secondary antibody (middle set). TO-PRO3 served as a nuclear counterstain. Representative areas within both confocal images were used for quantification of HPV18 E7-originating fluorescence (green). Each peak represents one cell, and fluorescence intensity is reflected by peak height (lower set). scr., scrambled; rec., recombinant. (D) (Upper set) Sections of HPV18 DNA-positive SCC and HPV18-positive AC of the cervix uteri were stained with RabMAb 143-7, as indicated. Normal squamous epithelia (NSE) and normal glandular epithelia (NGE) were stained under the same conditions, as indicated. (Lower set) Sections of HPV18 DNA-positive CIN3 and ACIS samples were stained with RabMAb 143-7, as indicated. Normal epithelia (NSE, NGE) were stained under the same conditions, as indicated.

ELISA-based quantification of HPV18 E7 protein in cervical smears.

So far, no procedures to quantify the level of E7 proteins in cervical carcinoma cells or clinical samples have been described. To address this point, we established a specific ELISA suitable for detection of HPV18 E7 using RabMAb 143-7. As capture antibodies, we used goat polyclonal sera raised against E7 proteins of HPV16 and HPV18. A 1:1 mixture of these antibodies was tested under immunofluorescence for the detection of high-risk and low-risk E7 proteins using transiently transfected U-2OS cells. We found that these polyclonal antibodies did not produce any signal in cells transfected with E7 proteins of low-risk HPV types 1, 6, and 11. However, strong and specific signals were obtained for cells transfected with E7 proteins of HPV16 and -18 (Fig. 3A). These data suggest that the goat polyclonal antibodies may be suitable as capture antibodies for an E7 ELISA. Ninety-six-well plates were coated with goat anti-HPV16 E7 and HPV18 E7 polyclonal antibodies, incubated with recombinant HPV18 E7 protein, and subsequently developed with RabMAb 143-7, using conventional secondary antibodies. In this setting, we were able to detect HPV18 E7 at a protein concentration of roughly 10 pg/well (Fig. 3B). To increase the specificity and sensitivity of the assay, RabMAb 143-7 was coupled to biotin. The biotinylated RabMAb 143-7 was used in combination with streptavidin–poly-HRP conjugates for additional ELISA experiments. To establish the detection limit of this improved ELISA setup, we performed titration experiments with recombinant HPV18 E7 protein at concentrations ranging from 100 fg to 1 ng per well. Using this setup, we established a detection limit of 0.25 to 0.5 pg/well for HPV18 E7 (Fig. 3C). Multiple runs of this titration curve were performed and used to determine the intra-assay and interassay variances of the ELISA (Table 1). The improved ELISA setup was used to establish the cross-reactivity pattern with respect to other high-risk and low-risk E7 proteins. Recombinant E7 proteins of the high-risk HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, and 59 as well as of the low-risk HPV types 6 and 11 were produced in bacteria and purified to homogeneity as described previously (36). ELISA plates were coated with goat polyclonal antibodies as before, and 100 pg of the respective high-risk E7 protein and 250 pg of the respective low-risk E7 protein were added per well. Subsequently, biotinylated RabMAb 143-7 was added and the signal was determined in an ELISA reader (Fig. 3D). This experiment revealed that the detection system established here is highly specific for HPV18 E7 and HPV45 E7, whereas no E7 protein of low-risk viruses is detected. Furthermore, E7 proteins of 10 other high-risk HPV types were not detected by the ELISA described herein.

Fig 3.

Sandwich ELISA for detection of HPV18 E7 oncoprotein. (A) U-2OS cells were transiently transfected with empty vector or expression vectors encoding E7 proteins of HPV1, -6, -11, -16, and -18, as indicated. Transfected cells were seeded on coverslips and analyzed by indirect immunofluorescence using a mixture (1:1) of goat polyclonal antibodies directed against HPV16 E7 and HPV18 E7. (B) ELISA plates were coated with a mixture (1:1) of polyclonal goat anti-HPV16 E7 and anti-HPV18 E7 antibodies. Subsequently, recombinant HPV18 E7 protein was added, as indicated. This was followed by the addition of RabMAb 143-7 and a secondary anti-rabbit antibody coupled to HRP. After incubation with TMB, absorbance at 450 nm was measured using a Victor ELISA reader. (C) ELISA plates were coated with a mixture (1:1) of polyclonal goat anti-HPV16 E7 and anti-HPV18 E7 antibodies. Subsequently, increasing amounts of recombinant HPV18 E7 protein were applied. This was followed by addition of biotinylated RabMAb 143-7. Streptavidin-linked poly-HRP conjugates were used for visualization. After incubation with TMB substrate, absorbance was measured at 450 nm using an ELISA reader. Results are represented in nonlogarithmic (left) and semilogarithmic (right) graphs. Detection of 1 pg E7 was highly significant (P < 0.0001) compared to the control (no E7). (D) ELISA plates were coated as described for panel B. E7 proteins from various high-risk HPV types at 100 pg/well (left) or from HPV18, HPV6, and HPV11 at 250 pg/well (right) were added. This was followed by addition of biotinylated RabMAb 143-7. Streptavidin-linked poly-HRP conjugates were used for visualization. After incubation with TMB substrate, absorbance was measured at 450 nm using an ELISA reader.

Table 1.

Intra- and interassay variances of the E7 ELISA with recombinant HPV18 E7 protein and HeLa cells

ELISA range	Variancea
	HPV18 E7	HeLa cells
Low	2/10	5/21
Medium	3/12	3/15
High	2/7	4/13

^aData represent % intra-assay variance/% interassay variance. For HPV18 E7, ELISA experiments whose results are shown in Fig. 3C were repeated five times independently, and the intra-assay and interassay variances were determined in the low (2.5 pg/well), medium (10 pg/well), and high (25 pg/well) ranges of the ELISA. For HeLa cells, ELISA experiments whose results are shown in Fig. 4A were repeated five times independently, and the intra-assay and interassay variances were determined in the low (2,500 cells/well), medium (10,000 cells/well), and high (25,000 cells/well) ranges of the ELISA.

To explore the possibility that the newly established ELISA system could detect endogenous E7 proteins in cervical cancer cells, lysates were produced from HPV18-positive HeLa cells in the background of HPV-negative U-2OS cells. In this experiment, we also attempted to establish the minimal number of cells required in a well to obtain a positive signal in the ELISA. For this purpose, decreasing numbers of HPV18-positive HeLa cells were complemented with HPV-negative U-2OS cells to achieve a total and constant cell number of 25,000 cells per well. Lysates of these cell mixtures were prepared and analyzed by ELISA using biotinylated RabMAb 143-7. The background signal, obtained with no addition of HeLa cells, was taken as a control, and the mean value of the control plus 3 times its standard deviation (SD) was defined as the cutoff, above which a sample was considered positive. Using this algorithm, we found that the current ELISA is able to detect 500 to 1,000 HeLa cells/well (Fig. 4A). As before, the ELISAs were independently conducted multiple times, and the results were used to calculate the interassay and intra-assay variances (Table 1).

Fig 4.

E7 detection in cervical carcinoma cells and cervical smears. (A) An increasing number of HPV18-positive HeLa cells (0 to 25,000 cells/well) was complemented with HPV-negative U-2OS cells to achieve a total cell concentration of 25,000 cells/well. Cell lysates were prepared and analyzed in the HPV18 E7 ELISA using biotinylated RabMAb 143-7. Results of one representative experiment are represented in nonlogarithmic (left) and semilogarithmic (right) graphs. The detection limit for HeLa cells was determined to be roughly 500 to 1,000 cells/well. (B) HPV18-positive HeLa cells and HPV-negative U-2OS cells were analyzed in various compositions, as indicated. Cellular lysates were prepared, and an aliquot was analyzed by the HPV18 E7 ELISA (left). DNA-based fluorescence of the same lysates was determined after addition of EvaGreen (right). Note that the signal obtained with the E7 ELISA is linearly decreasing with the number of HeLa cells, whereas the overall cell number is appropriately estimated by the EvaGreen signal. (C) Different amounts of HPV18-positive HeLa, HPV16-positive CaSki, and HPV-negative C33a cells were used for generation of cellular lysates, which were subsequently analyzed by 18E7 ELISA, implicating biotinylated RabMAb 143-7. (D) Lysates were prepared from 24 HPV DNA-negative cervical smears, all classified PapII. Patients were numbered from 15 to 38, as indicated. Lysates were analyzed by the HPV18 E7 ELISA and EvaGreen assay in parallel. Shown in the figure are normalized values. The red line indicates the cutoff value, defined as the average ELISA signal of all 24 patients displayed, plus 3 SDs. (E) Lysates were prepared from 14 HPV18 DNA-positive cervical smear specimens. Patients were numbered from 1 to 14, as indicated. For each sample, the cytological assessment at the time of diagnosis is indicated. Lysates were analyzed by the HPV18 E7 ELISA and EvaGreen assay in parallel. Shown in the figure are normalized values. The cutoff value determined for panel D is shown here for reference.

As a loading control, the total number of cells in a sample was estimated by independent measurement of free DNA by addition of the DNA-specific dye EvaGreen. Intercalation of EvaGreen into DNA yields green fluorescence, the intensity of which is proportional to the number of lysed cells. This can serve as a control for efficient cell lysis, as well as a validation for the cellularity of clinical samples. To validate the procedure, we performed both measurements in parallel for lysates containing a decreasing number of HeLa cells in the background of HPV-negative U-2OS cells. The total cell concentration of the lysates was kept constant at 25,000 cells per well. Decreasing numbers of HeLa cells caused a decrease in E7 ELISA signals, whereas the EvaGreen signal did not significantly change since the overall cell number was kept constant (Fig. 4B). Together, these results validate the normalization of ELISA data by EvaGreen fluorescence. The ability of RabMAb 143-7 to detect HPV18 E7 was independent of the cell line used for reference. Thus, signals obtained with U-2OS cells were indistinguishable from signals obtained with both HPV-negative C33a cells and HPV16-positive CaSki cells (Fig. 4C). Comparing ELISAs performed with recombinant E7 proteins (Fig. 3C) and ELISAs performed with cervical carcinoma cells, such as HeLa cells (Fig. 4A), allows determination of the concentration of E7 protein present per HeLa cell, which was estimated to be roughly 1 fg/cell.

E7-specific ELISAs may be developed into clinical tools for the early diagnosis of cervical precancer and cancer. As a first step in this direction, we performed a small feasibility study to investigate if the ELISA developed in the study described in this communication would be able to detect HPV18 E7 protein in cervical smears from HPV18 DNA-positive patients with abnormal cytology. To this end, 272 smear specimens were collected and typed for HPV DNA. Out of 272 smear specimens, 16 were found to be positive for HPV18 DNA (and several other HPV types in some cases; data not shown), of which 2 could not be used due to inadequate storage conditions. The remaining 14 HPV 18-positive samples were used for the analysis. In addition, we included a total of 24 normal (Pap classification II [PapII]) cervical smear specimens which were also negative for any high-risk HPV DNA. Samples were lysed, analyzed by ELISA, and normalized by EvaGreen fluorescence. Signals obtained by HPV18 E7 ELISA with samples from HPV-negative, cytologically normal (PapII) smears were generally weak (Fig. 4D). The values obtained with HPV-negative, cytologically normal (PapII) smears were defined as background, and a cutoff line was arbitrarily defined as the value of the average background signal plus 3 SDs. For 7 out of 14 HPV18-positive samples, E7 values were highly elevated, whereas E7 was not detectable in the remaining 7 HPV18 DNA-positive smear samples (Fig. 4E).

DISCUSSION

There is increasing evidence that E7 oncoproteins of high-risk human papillomaviruses play a crucial role in cervical carcinogenesis (21, 40, 41). The continuous presence of such proteins during cervical carcinoma development suggests that these proteins may be good markers for the progression of precancerous and cancerous lesions. However, although HPV18 is the second most prevalent HPV type worldwide, no immunoassays based on HPV18 E7 proteins are available for early detection of cervical disease. In this study, using highly specific RabMAbs to HPV18 E7, we show for the first time that E7 oncoproteins can be readily detected in cervical smears of high-risk HPV DNA-positive women with abnormal cytology. These data suggest that strategies to use E7 oncoproteins of high-risk viruses such as HPV18 as tools for the detection of ongoing cervical precancer and cancer are feasible.

E7 proteins of high-risk HPV, including HPV18 E7, display transforming activity in cell culture as well as transgenic mouse models (recently reviewed in reference 21). In agreement with these findings, the E7 proteins of the most prevalent high-risk HPV types, types 16, 18, and 45, have been detected in biopsy specimens of cervical precancers and cancers (8, 11, 12, 26). In the present study, we generated monoclonal antibodies against HPV18 E7 in rabbits (RabMAbs), to establish a reliable source of high-quality immunological reagents suitable for the quantitative determination of HPV18 E7 oncoprotein in cervical smears by an ELISA system. Therefore, hybridomas were selected by ELISA. The epitope recognized by one clone (RabMAb 143-7) was mapped to the amino acids 5PKATLQDIVL13 in conserved domain 1 (cd1) of the HPV18 E7 oncoprotein. Epitope mapping experiments and experiments with the HPV18 E7 ELISA demonstrate that RabMAb 143-7 cross-reacts only with the closely related HPV45 E7 protein and does not react either with the E7 proteins of 10 additionally analyzed high-risk HPV types or with the low-risk E7 proteins of HPV6 and -11. Whereas the epitope recognized by RabMAb143-7 in the N-terminal domain of E7 is important for the ability of E7 to transform cells, the epitope does not encompass the Rb binding domain, representing one of the major functional interaction sites of E7. However, we cannot formally exclude the possibility that antibody recognition is affected by cellular proteins binding to the E7 N-terminal domain. RabMAb 143-7 was further extensively characterized by Western blotting and immunofluorescence analysis. Results of these experiments established the high specificity of RabMAb 143-7, and immunohistochemistry data presented here indicate that RabMAb 143-7 is suitable to detect HPV18 E7 protein in clinical samples. To allow the quantitative assessment of E7 oncoproteins, a sandwich ELISA was established. Using biotinylation and streptavidin-coupled HRP multimers, we were able to improve the sensitivity by roughly 20-fold. Using this ELISA, E7 proteins of low-risk HPV types were not recognized, and out of 12 high-risk HPV types, only HPV18 and HPV45 E7 proteins were detected. Both HPV types belong to HPV species 7 in the genus Alphapapillomavirus and are phylogenetically closely related (7). With the sandwich ELISA established here, 0.5 pg of recombinant HPV18 E7 protein can be detected, and the respective ELISA signal was found to be equivalent to approximately 500 HeLa cells. The ELISA detection system described here is characterized by high sensitivity and high analytical precision, shown by the relatively low intra-assay and interassay variances. Whereas a relatively mild detergent is used for cell lysis in our standard protocol, we cannot exclude the possibility that the extraction conditions used here fail to disrupt all E7 interactions with cellular proteins. However, the linear dependence of E7 ELISA signals on the number of HeLa cells, shown in Fig. 4A, provides clear evidence that quantitative extraction of E7 from cells is possible under our conditions and that any inhibiting interactions, if they occur, would not affect the linearity of the assay.

The amount of E7 protein expressed by tumor cells in situ is currently unknown; on the basis of results with immunohistochemical analysis of cervical carcinoma biopsy specimens (8, 11, 12, 26), it can be expected to vary between different tumors. According to our data, the current ELISA may be suitable to identify precancerous lesions in smears containing at least 500 tumor cells, provided that the expression level of HPV18 E7 in HeLa cells is representative for primary tumors. Hence, further improvements of the sensitivity are desirable.

Out of 272 cervical smear samples, we identified 16 HPV18-positive samples. Two HPV18 DNA-positive smears were excluded due to inadequate storage conditions, while 14 samples were analyzed by the HPV18 E7 sandwich ELISA, along with 24 HPV DNA-negative, cytologically normal smear samples. Very weak signals were detected with cytologically normal, HPV DNA-negative smears, which were used to define an arbitrary background level. Of the 14 HPV18 DNA-positive samples analyzed by sandwich ELISA, 7 samples yielded signals well above the background in the E7 ELISA, including 1 sample cytologically classified PapV (patient 14), 2 samples classified PapIVa (patients 12 and 13), 1 sample classified PapIIID (patient 7), 1 sample classified PapII/III D (patient 1), and 2 samples classified PapII (patients 1 and 2). The results of this feasibility study do not allow conclusions on the clinical utility of the ELISA developed in this study to be drawn but raise several interesting points for discussion which may be useful for the design of future clinical studies. First, an HPV18 E7 protein level well over background was found in smears with abnormal cytology, such as samples 12, 13, and 14, providing proof of the principle. Patient 11, negative for HPV18 E7 protein, despite being diagnosed with PapIIID/IVa, was positive for DNA of both HPV16 and HPV18, suggesting a multiple infection. We cannot exclude the possibility that in this case the dysplasia CIN2 was associated with HPV16 and not HPV18 and HPV18 infection was just transient. In this respect it may be worth noting that HPV E7 expression requires viral E6E7 RNA splicing (35, 37a). Accordingly, the samples with HPV18 E7-negative ELISA results might just not express E6E7 mRNA, or even so, the viral RNA might not be spliced. Second, HPV18 E7 protein was detected in two HPV18-positive samples with apparently normal cytology derived from patients 1 and 2. However, immunohistological analysis of a biopsy specimen from patient 1 revealed that it stained positive for the surrogate marker p16^INK4a in lower sections of the biopsy specimen; moreover, high proliferative activity was also detected in the suprabasal epithelium (data not shown), whereas histological analysis of patient 2 revealed small cell carcinoma of the cervix uteri (data not shown). Together, these findings suggest the presence of high-grade lesions in patients 1 and 2, although both smears were classified PapII by cytological analysis. Third, out of eight HPV18 DNA-positive samples with cytological diagnosis of PapII/IIID or PapIIID, five samples (from patients 5, 6, 8, 9, and 10) were clearly negative for HPV18 E7 protein in our ELISA. Sample 6 was quadruple positive for HPV types 18, 56, 33, and 70, and we cannot exclude the possibility that HPV type 56 and/or 70 was driving the lesion in this case. Supporting the absence of high-grade lesions in some of these cases, histological examination of patient 5 revealed that this sample, suspicious of CIN1, stained negative for the histological surrogate marker p16^INK4a (data not shown), whereas histology of patients 9 and 10 revealed the presence of relatively low-grade lesions (recurrent vaginal intraepithelial neoplasia and CIN2, respectively). If HPV18 E7-expressing tumor cells were present in any of these samples, their number and/or E7 expression level was apparently too low to be recognized as HPV18 E7 positive in the ELISA. It was reported that, in the case of HPV16, viral load was weakly associated with cervical disease stage (2), whereas viral load assessment of HPV types 16, 18, 31, and 33 had no additive value to stratify high-risk HPV-positive women for risk of CIN2 or CIN3 in a cervical screening setting (15). To address the question if HPV18 load may correlate with E7 ELISA signals in our study, HPV18 copy number was determined in selected HPV18-positive smears, revealing viral loads ranging from 0.3 copy to 1,600 copies per cell. Values obtained for viral load did not correlate with the signals obtained for the same patient in the HPV18 E7 ELISA (data not shown). Further studies with larger patient cohorts are warranted to address this question. Some of the results mentioned above could be due to sampling problems. A limitation could be that the smears used for the ELISA were generally second samples, while the first smear sample was used for cytology and HPV detection purposes. On the other hand, it is also possible that the number of tumor cells and/or their E7 level is too low in a given sample, despite ongoing carcinogenesis. To address this point and to further validate the ELISA developed here, larger studies will be needed. In particular, it will be important to address the question whether E7 positivity at the time of diagnosis correlates with a negative clinical outcome in a large number of cases. In this respect, attempts to also further increase the sensitivity of the ELISA are warranted. Nevertheless, the data reported here provide proof of the principle that immunological detection of E7 oncoproteins in cytologically abnormal smears is feasible.