2337

Journal of General Virology (1992), 73, 2337 2345. Printed in Great Britain

Expression of human papillomavirus type 16 E6 protein by recombinant baculovirus and use for detection of anti-E6 antibodies in human sera Simon N. Stacey, 1. Jennifer S. Bartholomew, 2 Anna Ghosh, 2 Peter L. Stern, 2 Michael M a c k e t t 1 and John R. Arrand I Cancer Research Campaign Departments o f 1Molecular Biology and 2immunology, Paterson Institute for Cancer Research, Christie Hospital, Wilmslow Road, Manchester M20 9BX, U.K.

Existing assays to detect antibodies to human papillomavirus type 16 (HPV-16) proteins in sera from cervical carcinoma patients rely primarily on bacterially produced recombinant proteins or synthetic peptides for use as target antigens. These methods have had limited success in the detection of antibodies against the E6 protein. To produce more authentic E6 protein for use in serological assays, we have employed a recombinant baculovirus vector to synthesize the protein in insect cells. Cells infected with the vector containing E6 gene sequences expressed a stable protein doublet comprising 18-5K and 19-1K bands. This protein reacted in Western blots with an antiserum raised against a purified E6 fusion protein produced in Escherichia coli. This antiserum, and several others raised against E. coli-derived E6 fusion

proteins, were unable to recognize the baculovirus E6 protein in radioimmunoprecipitation assays (RIPAs). However, serum from a cervical carcinoma patient readily immunoprecipitated the baculovirus E6 protein, suggesting that the baculovirus-derived protein represented a realistic antigenic target. A RIPA was developed for the detection of anti-E6 protein antibodies in human sera. The assay was tested on a selected group of sera from carcinoma patients and controls, in comparison with a Western blotting method using bacterial fusion proteins. The baculovirus E6 protein-based RIPA showed a marked increase in detection rate over the Western blotting method. These findings suggest that serum antibodies to HPV-16 E6 protein may be more prevalent than has previously been shown.

Introduction

Since the E6 and E7 proteins are virus-encoded, their expression might be expected to expose the tumours to immunological attack. In model systems it has been shown that the E6 and E7 proteins can be targets for several different arms of the immune response (Chen et al., 1991 ; Meneguzzi et al., 1991 ; Comerford et al., 1991 ; Strang et al., 1990; Banks et al., 1991). In man, one indication that HPV proteins may present immunological targets is the appearance of serum antibodies against them. However, evaluation of the antibody responses in patients exposed to high risk HPV types has been hampered by the lack of a readily available source of viral proteins for use as target antigens in serological assays. Investigations have been made possible only through the use of prokaryotic gene expression systems or synthetic peptides to provide target antigens. Using Western blots of immobilized bacterial fusion proteins containing HPV-16 proteins, Jochmus-Kudielka et al. (1989) found that the presence of serum antibodies against E7 protein correlates with cervical carcinoma. Using similar methods, Kochel et at. (1991) found a significant association between cervical carcinoma and

Carcinoma of the cervix has been strongly linked with infection by so-called high risk types of human papillomavirus (HPV). These high risk HPV types, usually HPV-16 or HPV-18, can be found in over 8 0 ~ of cervical carcinomas (for reviews see Pfister, 1987; Galloway & McDougall, 1989; Vousden, 1989; Niedobitek & Herbst, 1991). These tumours tend to retain and actively transcribe HPV early region open reading frames (ORFs) (Baker et al., 1987; Schneider-Gadicke & Schwarz, 1986; Schwarz et al., 1985; Smotkin & Wettstein, 1986). The E6 and E7 proteins have been implicated in tumorigenesis owing to their prevalence in cervical tumours and cell lines (Androphy et al., 1987; Banks et al., 1987; Schneider-Gadicke et al., 1988; Seedorf et al., 1987; Smotkin & Wettstein, 1986), transforming activity in vitro (reviewed in Matlashewski, 1989; DiMaio, 1991) and their ability to bind to the products of the cellular tumour suppressor genes p53 and Rb, respectively (Dyson et al., 1989; Munger et al., 1989; Werness et al., 1990). 0001-0992 © 1992SGM

2338

S. N. Stacey and others

serum antibodies to either the E4 or E7 protein. Peptide ELISA has also been used to identify associations between cervical carcinoma and serum antibodies to HPV- 16 E7 and E2 proteins (Krchfi~tk et al., 1990; Mann et al., 1990; Dillner, 1990). However, the frequency of seropositivity rarely exceeds 40 ~ of carcinoma patients, which is far less than would be expected based on the expected frequency of HPV-16 DNA in this tumour type. Comparatively little attention has been given to the E6 ORF product. Neither peptide-based ELISA nor bacterial fusion protein Western blotting methods have been very successful in the detection of anti-E6 protein antibodies in human sera (Dillner, 1990; Kochel et al., 1990; Jenison et al., 1988). It is not known whether the lack of apparent anti-E6 serological responses is due to the sensitivity of the techniques used. Using optimized Western blotting techniques, we have found that anti-E6 antibodies can be detected in sera from cervical carcinoma patients (this report and A. Ghosh et al., unpublished results). However, as with E7 the frequency of this response is less than would be expected based on HPV DNA typing data. Bacterially expressed fusion proteins and synthetic peptides, however, are not ideal target antigens for serological surveys, because they are often highly insoluble and can present only linear epitopes. It has been established from studies on bovine papillomaviruses that antibodies generated against denatured capsid proteins are broadly cross-reactive with a wide range of papillomavirus types, whereas antibodies raised against intact virus particles tend to be type-specific (see Cowsert et al., 1987; Lim et al., 1990). Similarly, where bacterial fusion proteins or synthetic peptides have been used to generate antibodies against HPV coat proteins, broadly cross-reactive antisera are obtained (Strike et al., 1989; Christensen et al,, 1990). This factor has been shown to have practical importance in serological surveys. Strike et al. (1990) and Bonnez et al. (1990) have found that Western blots using HPV-6b capsid antigen fusion proteins can not discriminate serologically between condylomata acuminata patients and normal controls. However, an ELISA-based method using whole particles of the closely related HPV-11 is able to do so (Bonnez et al., 1991). The use of immobilized bacterial fusion protein antigens might restrict assay sensitivity by not identifying antibodies targeted to conformational epitopes. At the same time, specificity might be reduced by the exposure of potentially cross-reactive epitopes which normally would be hidden. It is not yet known whether these considerations apply to the early proteins of HPV. However, these reports would seem to indicate that the true serological picture might better be revealed using assays which present target antigens in more

realistic conformational states. This requires the synthesis of antigens at high levels in a soluble and correctly folded form. To generate improved HPV serological target antigens that fulfil these criteria, we have employed a eukaryotic expression system based on an insect baculovirus (Summers & Smith, 1987). We describe here the construction and characterization of a recombinant baculovirus vector expressing the HPV-16 E 6 0 R F . The baculovirus-derived E6 protein displayed more authentic antigenic characteristics than Escherichia coli fusion proteins, and was used to detect the presence of anti-E6 antibodies in human sera.

Methods Cells. Sf21 cells were generously provided by Dr R. Possee, Institute of Virology, Oxford, U.K. They were maintained at 27 °C in TC-100 medium (Gibco-BRL) supplemented with heat-inactivated foetal calf serum (FCS) (to 10~) and gentamicin (to 50 lag/ml). Cells were used between passages 170 and 190. Periodic screening for mycoplasma contamination gave consistently negative results. Construction and selection o f recombinant baculovirus. Established cloning methodology (Sambrook et al., 1990) was used to construct an E6 recombinant baculovirus insertion vector. A 630 bp DdeI fragment from coordinates 24 to 654 (Seedorf et al., 1985) was isolated, bluntended using the Klenow fragment of DNA polymerase I, and phosphorylated BamHI linkers were attached. After cleavage with BamHI, the fragment was ligated into the BamHI site of the baculovirus insertion vector p36c (Page, 1989). This resulted in the E6 ORF, including 59 bp of 5' untranslated leader sequence, being inserted into a position corresponding to + 35 bp relative to the polyhedrin start codon (see Fig. 1). The resulting plasmid insertion vector was designated pE636. General methods for the production and handling of baculovirus recombinants were taken from Summers & Smith (1987) and from Bailey & Possee (1991). Insertion vector pE636 DNA was transfected using lipofectin (Gibco-BRL) into Sf21 cells along with high M r DNA from a recombinant baculovirus, NPBV, which expresses fl-galactosidase from within the polyhedrin locus. Screening for recombinants was done by the methods of Fung et al. (1988) and P. Ertle (personal communication) as follows. 96-well plates were seeded with Sf21 cells at 1.5 × 10~ cells/well and then infected by adding serial dilutions (10 -1 to 10-8) of transfection supernatant taken at 6 days post-transfection. After 6 days of incubation at 27 °C, supernatants from the 96-well plate were transferred to replica 96-well plates and stored at 4 °C until required. Cells remaining on the original plate were lysed by adding 100 pl 0-2 M-NaOH followed 15 min later by 100 ~tl 2 M-NaC1. Lysates were dot-blotted onto nylon membranes using a vacuum manifold (BioRad). The membranes were then hybridized to a 32p-labelled E6specific probe comprising the 630 bp Ddel fragment described. For each dilution series, the lowest dilution giving a positive signal was identified and the corresponding stored supernatant was recovered. These supernatants were serially diluted (10-2 to 10-7) and used to infect Sf21 cells plated onto six-well dishes at 2 x 106 cells/well. Plaques were allowed to form under agarose, and then stained with neutral red and 100 ~tg/ml X-Gal for 3 to 4 h. Plaques were selected for loss of the blue plaque phenotype and underwent at least three further rounds of plaque purification using the same selection. Three clones were chosen for further analysis. These clones were examined for the presence and location of the E 6 0 R F within the baculovirus genome

Baculovirus expression o f H P V - 1 6 E 6 protein

using Southern blotting (data not shown). A single clone, designated bE6, was selected and used for all subsequent experiments. Large-scale stocks of bE6 were prepared and titrated as described (Summers & Smith, 1987; Bailey & Possee, 1991).

Purification of HPV-16 MS2-E6 fusion protein from E. coli. E. coli strain C600/537 containing an MS2-E6 fusion protein expression vector was induced and lysed according to the method of Seedorf et al. (1987). Differential urea extraction (Seedorf et al., 1987) resulted in the majority of the MS2-E6 protein being present in the insoluble fraction, which was then dissolved in 4 M-guanidine hydrochloride. DTT and PMSF were added to 50 mM and 1 raM, respectively, and the lysate was incubated at 37 °C for 3 h. The lysate was then applied to an iminoSepharose metal affinity chromatography column (using Sigma 1-4510 iminodiacetic acid-epoxy activated Sepharose 6B fast flow resin) which had been charged with 3 mg/ml ZnSO4 and equilibrated with Buffer A (4 M-guanidine hydrochloride, 50 mM-Tris-acetate, 0.2 M-NaC1, 10 mMDTT, 0.1 mM-PMSF, pH 7-8). The column was washed with several column volumes of Buffer A, then 20 ml of a pH and salt elution gradient using Buffer A and Buffer B (4 M-guanidine hydrochloride, 50 mM-Tris-acetate, 1 M-NaC1, 10 mM-DTT, 0.1 mM-PMSF, pH 5.8) was run through the column. Fractions of 500 p.l were taken and 50 ~tl aliquots were analysed by 12 % PAG E followed by silver staining. The MS2-E6 fusion protein eluted in its most pure form at pH 6-6. Six of the fractions judged most pure by PAGE and silver staining were pooled and repurified on the metal chelating column as above. The resulting highly purified preparation of MS2-E6 fusion protein was quantified by the Bradford assay then stored at - 70 °C until required. Preparation of polyclonal anti-MS2-E6 fusion protein antisera. Approximately 10 txg of affinity-purified MS2-E6 fusion protein was precipitated with 2 volumes of saturated ammonium sulphate, centrifuged and the washed pellet redissolved in PBS containing 0-1 SDS. This was mixed with an equal volume of Freund's incomplete adjuvant and injected subcutaneously into a New Zealand White rabbit at four sites. Three further immunizations were given similarly at monthly intervals. Serum was tested by Western blotting against lysates of E. coli containing the MS2-E6 plasmid.

Western blotting. Preparation of E. coli MS2 fusion proteins for Western blotting was done according to Jochmus-Kudielka et al. (1989). For Western blotting of baculovirus-infected cell lysates, approximately 1-5 × 107 Sf21 cells were infected with bE6 or the parental virus NPBV at a multiplicity of 10 p.f.u./cell. Cultures were incubated at 27 °C for 48 h then lysed using ice-cold NP40 lysis buffer (0.5% NP40, 25 mM-Tris-HC1, 137 mM-NaC1, 100 ~tg/ml aprotinin, 0"5 mM-PMSF, pH 7.9). An equal volume of 2 x PAGE loading buffer was added and lysates were loaded onto polyacrylamide gels at 4 x 105 cells/lane. In all Western blotting experiments, extracts were fractionated by SDS-PAGE and proteins were transferred to nitrocellulose using a semi-dry blotting apparatus (Novablot; Pharmacia/LKB). Filters were blocked in 5 % non-fat dried milk (Marvel) in Tris-buffered saline (TBS) at 4 °C overnight, followed by incubation with specific antibody at a 1:50 dilution in the same buffer for 2 h at room temperature. Filters were then washed for 30 min in 0.05 % Tween 20 in TBS. In experiments using human sera, a 1 : 1500 dilution of rabbit antihuman IgG (Dako) was added and incubated for 1 h at room temperature, followed by another 30 min wash. All filters were incubated for 40 min with the biotinylated F(ab')2 fragment of swine anti-rabbit Ig (Dako) diluted 1:1000 in Marvel/TBS. Filters were then incubated for 30 min with streptavidin-biotinylated alkaline phosphatase, washed and developed using naphthol/AS-MX phosphate/Fast blue as a substrate. Metabolic labelling of bE6 infections with [35S]cysteine. T75 flasks containing approximately 1.5 x 10v cells were infected with virus at 10 p.f.u./cell in complete TC100 medium including 10% FCS. At 23 h

2339

post-infection the medium was removed and the cells were washed gently twice with cysteine-free TC100 medium without FCS. Then, 4 ml of cysteine-free TC100 containing 10~ dialysed FCS and 0-5 mCi [35S]cysteine (Amersham) was added, followed by incubation for 2 h at 27 °C. Cells were washed twice in cysteine-free TC100 and lysed in 7.5 ml/flask ice-cold NP40 lysis buffer, radioimmunoprecipitation assay (RIPA) buffer (1 ~ NP40, 0.5~ deoxycholate, 0.1 ~ SDS, 50 mRTris-HC1, 150 mM-NaC1, 10 ~tg/ml aprotinin, 0.5 mM-PMSF, pH 7-8) or PAGE loading buffer as indicated. RIPA and PAGE loading buffer lysates were passed three times through a fine-gauge needle. NP40 lysates were incubated on ice for 20 min with frequent agitation. All lysates were then centrifuged at 10000 g for 10 rain. The supernatant fractions were taken and stored in aliquots at - 7 0 °C. Aliquots were thawed only once. For direct visualization of 35S-labelled bE6 protein, 5 ~tl of labelled cell lysate was fractionated by 15 ~ PAGE and exposed to autoradiographic film without fluorographic enhancement. In experiments to determine protein stability, the period of labelling was reduced to 1 h and times were calculated from the end of the labelling period.

Immunoprecipitation. 35S-labelled cell lysate (100 ~tl)was used for each immunoprecipitation. Each sample was precleared with 5 ktl of normal rabbit serum (Dako) at 4 °C for I h. Protein A-Sepharose (Pharmacia) was added to a concentration of 8 ~ and the sample incubated for a further 1 h at 4 °C with frequent agitation. Protein A-Sepharose was then pelleted by centrifugation for 15 s and the supernatant was transferred to a fresh tube. Ten p.1 of 10~ non-fat dried milk in lysis buffer was added, followed by 5 I11 of specific antiserum, and the sample was incubated overnight at 4 °C. Protein A-Sepharose beads were added to 8% and the mixture was incubated at 4 °C for 1 h with frequent agitation. Beads were pelleted and washed three times with lysis buffer. PAGE loading buffer (50 ~d) was added and the sample heated to 95 °C for 5 min, after which 25 ~tl was fractionated by 15~ PAGE. The gels were treated with fluorographic reagent (Amplify; Amersham), dried and exposed to autoradiographic film. Antisera used for specific immunoprecipitation were (i) RPaE6, rabbit polyclonal antiserum raised against affinity-purified MS2-E6 protein (see above), (ii) ~tMS2-E6, a rabbit polyclonal antibody raised against the MS2-E6 protein (a gift of L. Gissmann), (iii) ~flgal-E6, a rabbit polyclonal antibody raised against a fl-galactosidase-E6 fusion protein (a gift of M. Stanley), (iv) CIP5, a monoclonal antibody raised against an HPV-18 E6-fl-galactosidase fusion protein (Banks et al., 1987) and (v) cancer patients' serum samples taken from patients attending Christie Hospital for carcinoma of the cervix, prior to their treatment. Control sera were taken from ostensibly normal female laboratory personnel or normal women attending a family planning clinic. These sera were taken as part of a larger scale seroepidemiological study (A. Ghosh et al., unpublished results). Selection of sera for testing using bE6 RIPA was based on the performance of these sera in Western blotting experiments, as described in Results. HPVDNA detection. D N A extraction from tumours and detection of HPV DNA were done as described previously (Connor & Stern, 1990).

Results Construction, isolation and expression o f the baculovirus recombinant

The E 6 0 R F from HPV-16 was excised as a 630 bp DdeI fragment containing the ORF and 59 bp of untranslated leader sequences. The fragment was inserted into the baculovirus transfer vector p36c, which contains the first 33 bp of the polyhedrin coding sequences rendered

2340

S. N. Stacey and others

59 b p H P V

~

+~

ATC CGG

+3a

E60RF

|

ATG CAG CAA

.177 TAA

PH-R

I

Fig. 1. Structure of the expression cassette used to produce the E6 recombinant baculoviruspE636. HPV-16 sequences from positions 24 to 654 (Seedorfet al., 1985)were inserted using BamHl linkers into a position correspondingto + 33 bp of the polyhedrincodingsequences. The polyhedrinsequenceshad been rendered untranslatableby a G to C mutation at position +3 (underlined) (Page, 1989). Flanking polyhedrin(marked PH) sequences allowfor insertion and expression in the polyhedrinlocus of the recombinant baculovirus. untranslatable by in vitro mutation of the start codon (Page, 1989). The resulting construct, pE636 (Fig. 1), was cotransfected into Sf21 cells in combination with high Mr parental baculovirus DNA. To simplify selection for recombinant baculovirus, the parental virus was itself a recombinant (designated NPBV) which expressed flgalactosidase from within the polyhedrin locus, and produced a blue plaque phenotype when cultured in the presence of X-Gal. Replacement of the fl-galactosidase sequences of NPBV with those of the E 6 0 R F was expected to result in reversion of the blue plaque phenotype. Initial screening subsequent to transfection was done by limiting dilution dot blot assay (Fung et al., 1987). Pools containing E6-positive virus were then plaquepurified by at least three rounds of selection for revertants of the blue plaque phenotype. Three clones were tested for the presence of E6 sequences using Southern blotting and all proved positive, although no novel bands were detected on Coomassie blue-stained polyacrylamide gels of recombinant-infected cell lysates from all three clones (data not shown). One clone, designated bE6, was selected and used for all subsequent analyses. To investigate the expression of the bE6 encoded protein further, Sf21 cells were infected with either bE6 or NPBV, or were mock-infected. At 23 h post-infection, the cells were pulse-labelled with [35S]cys_ teine for 2 h. Cells were then washed and lysed in RIPA buffer. P A G E fractionation of labelled cell extracts revealed the presence of a novel doublet comprising 18.5K and 19.1K bands in bE6-infected cells compared to cells infected with the parental virus NPBV (Fig. 2a). This doublet was usually resolved only on high resolution gels which had not been treated with fluorographic enhancers prior to autoradiography. A strong hostderived band of similar size was seen in mock-infected cells (Fig. 2a, lane 3), but not in baculovirus-infected cells, owing to the shut-down of host cell protein synthesis resulting from infection. The stability of the bE6 doublet was investigated by pulse-chase experiments (Fig. 2b). Sf21 cells were infected with bE6 then labelled for 1 h with [35S]cysteine

starting at 23 h post-infection. Label was then removed and chased with complete medium. At various times after the termination of labelling, aliquots of cells were lysed in P A G E loading buffer and fractionated by 15 PAGE. This analysis revealed no appreciable decline in the intensity of either member of the bE6 doublet, up until the termination of the experiments at 48 h postlabelling (Fig. 2b, lanes 1 to 8). Therefore, the bE6 doublet appeared to be relatively stable in vivo (cell culture). We also investigated the stability of the bE6 protein in vitro by incubating NP40 lysates of pulselabelled bE6-infected cells for various times at different temperatures, followed by P A G E fractionation and autoradiography. We found that under the NP40 lysis conditions described here, the bE6 protein could withstand at least 48 h incubation at 37 °C without substantial degradation (data not shown). The solubility of the bE6 protein in different lysis buffers was investigated using 35S pulse-labelled material. A gentle lysis protocol using buffer containing the non-ionic detergent NP40 liberated a substantial proportion of the total bE6 protein into the supernatant (Fig. 2b, lane 11), some insoluble material remaining in the pellet (lane 12). We considered the NP40-soluble fraction to contain primarily native bE6 protein and this fraction was used in subsequent immunoprecipitation experiments. Antigenic characterization o f the bE6 protein

To identify the bE6 18.5K and 19.1K protein doublet as products of the E 6 0 R F , it was necessary to generate antisera against E6 protein derived from an independent source. To generate such an independent specific antiserum, we used the bacterial expression vector pEXMS2-E6, which expresses an MS2 replicase-HPV-16 E6 fusion protein (Seedorf et al., 1987). The innate zincbinding property of the E6 protein (Barbosa et al., 1989; Grossman & Laimins, 1989) was utilized to purify the protein by zinc-charged metal chelating affinity chromatography. Fig. 3 shows a silver-stained gel of an elution series to recover the purified MS2-E6 protein. The MS2-E6 protein had an Mr of approximately 30K, and an approximately 15K protein, presumably a degradation product, was consistently co-purified. The yield of highly purified MS2-E6 protein was approximately 8 rag/1 of culture. The purified protein was used to obtain a rabbit polyclonal anti-MS2-E6 antiserum (designated RP~E6). In Western blots, the RPeE6 antiserum identified an approximately 18.5K protein present in lysates from cells infected with bE6 but not with parental virus (Fig. 4a, lanes 1 and 2). This same antiserum recognized the MS2-E6 fusion protein in extracts from producer bacteria (Fig. 4a, lane 3).

Baculovirus expression of HPV-16 E6 protein

2341

(b)

(a) 2

1

3

4

1

2

3

4

5

6

7

8

9

10

1!

12

13

Fig. 2. (a) Autoradiograph showing expression of the E6 protein by recombinant baculovirus bE6. Sf21 cells were pulse-labelled with [35S]cysteineat 23 h post-infection, lysed in PAGE loading buffer and separated by 15% PAGE. Viruses used for infection were: lane 1, bE6; lane 2, NPBV; lane 3, mock-infected; lane 4, Mr markers. The sizes of the Mr markers are given to the right. (b) Autoradiograph showing stability (lanes 1 to 8) and solubility (lanes 10 to 13) of bE6 protein. Cells were infected with bE6 then pulse-labelled for 1 h starting at 23 h post-infection, followed by a chase with complete unlabelled TC 100 medium. Samples were taken at the followingtimes during the chase : lane 1,0 min; lane 2, 15 min; lane 3, 30 min; lane 4, 90 min; lane 5, 3 h; lane 6, 8 h; lane 7, 24 h; lane 8, 48 h; lane 9, Mr markers. Lanes 10 to 13 show the solubility of the E6 doublet under the following lysisconditions: lane 10, RIPA buffer; lane I 1, NP40 buffer supernatant; lane 12, N P40 buffer pellet; lane 13, PAGE buffer. The position of the E6 doublet is indicated (arrow). Sizes of Mr markers are as in (a). 1

2

3

4

5

6

7

8

9

10

M

Fig. 3. Purification of MS2 E6 fusion proteins from E. coli by metal chelate affinity chromatography. A silver-stained polyacrylamide gel of second-round elution fractions is shown. Lane 1, run-through; lanes 2 to 10, fractions of a pH 7.8 to pH 5.8 elution gradient. The pH of fraction 8 is approximately 6-6. The position of the MS2 E6 fusion protein is indicated (arrow). Sizes of Mr markers (lane M) are as in Fig. 2.

We t h e n investigated w h e t h e r the R P ~ E 6 a n t i s e r u m would recognize the n a t i v e c o n f o r m a t i o n b a c u l o v i r u s E6 protein using R I P A . Despite its ability to recognize the bE6 protein in W e s t e r n blots, the R P ~ E 6 a n t i s e r u m was u n a b l e to i m m u n o p r e c i p i t a t e the n a t i v e b E 6 protein. A n u m b e r of other antisera p r e p a r e d against other E. col# derived E6 fusion proteins were at best able to i m m u n o p r e c i p i t a t e the n a t i v e b E 6 p r o t e i n only very weakly (see Fig. 4b, lanes 1 a n d 2). W e speculated that if the baculovirus E6 p r o t e i n closely resembled the n a t u r a l E6 protein, t h e n h u m a n sera from cervical c a r c i n o m a p a t i e n t s with a n t i b o d y responses a g a i n s t the E6 p r o t e i n m i g h t be better able to recognize the b E 6 protein. To investigate this we identified a cervical c a r c i n o m a p a t i e n t - d e r i v e d s e r u m which c o n t a i n e d a n t i b o d i e s a g a i n s t H P V - 1 6 E6, using E. coli-derived E 6 - M S 2 fusion p r o t e i n in a high-sensitivity W e s t e r n b l o t t i n g protocol (Fig. 4c). As s h o w n in Fig. 4(b) (lanes 4 to 6), this s e r u m i m m u n o p r e c i p i t a t e d a n a p p r o x i m a t e l y 18-5K p r o t e i n specifically from b E 6 - i n f e c t e d cells, whereas the n o r m a l

2342

S. N. Stacey and others

(a)

(b) 1

2

3

(e) 1

2

3

4

5

6

1

2

3

4

5

m

Fig. 4. (a) Western blot of: lane 1, NPBV; lane 2, bE6; lane 3, MS2-E6 lysates using RP~E6 antiserum directed against the E. coilderived fusion protein. (b) RIPA using the lysates of bE6-infected (lanes 1, 3 and 5) or NPBV-infected (lanes 2, 4 and 6) cells. Antisera were : lanes 1 and 2, rabbit polyclonal anti-E6 fusion protein antibody; lanes 3 and 4, cervical carcinoma patient's serum; lanes 5 and 6, normal human control serum. (c) Western blot using sera from two cervical carcinoma patients against E. coli-derived MS2 fusion proteins. Lanes 1 to 3, serum MX05 ; lanes 4 to 6, serum B44. Fusion proteins were E6 (lanes 1 and 4), E4 (lanes 2 and 5) and E7 (lanes 3 and 6). Sizes of M~ markers (indicated by marker bars) are as in Fig. 2.

human serum was unable to do so. Taken together, these data indicated that (i) the target epitope(s) of the RP~E6 antiserum are present on the bE6 protein, but they are not available for antibody binding when the protein is in its native conformation, and (ii) cervical carcinoma patients' sera can contain antibodies that recognize epitopes that are exposed on the native bE6 protein.

Screening patients" sera by immunoprecipitation using native bE6 protein Based on the preceding observations, it might be anticipated that use of the bE6 protein in immunoprecipitation assays would provide a more natural target for human anti-E6 antibodies than immobilized bacterial fusion proteins. Accordingly, we conducted a comparison of MS2 fusion protein Western blotting and baculovirus-based RIPA for the detection of anti-E6 serum antibodies. In a concurrent seroepidemiological study (A. Ghosh et al., unpublished results) we found, using Western blotting to E6-MS2 fusion proteins, antiE6 antibodies in cervical carcinoma patients at a frequency of approximately 35 ~, similar to that of E7. The response in normal individuals was less than 5 ~. We wanted to use the baculovirus E6 RIPA to investigate whether similar seropositive and seronegative class assignments would be obtained using the more authentic bE6 target antigen and, in particular, whether the bE6

RIPA would detect responses in the classes (both normal and carcinoma patients) which had scored seronegative by Western blotting. Such additional responses might be detected in the bE6 RIPA owing to its potential to present non-linear target epitopes. Thirty samples were selected from the seroepidemiological study, comprising 10 E6 seronegative controls, 10 E6 seronegative carcinoma patients and 10 E6 seropositive carcinoma~ patients. Examples of the reactivities revealed by MS2 fusion protein Western blotting are shown in Fig. 4(c). The samples were selected to show a range of intensities of E6 seropositivity and a variety of different anti-E4, -E7 and -E6 antibody profiles (see Table 1). For the carcinoma patients, the corresponding tumours were examined for the presence of HPV-16 and HPV-18 DNA by Southern blotting. DNA status was not taken into consideration in the selection of samples. The samples were encoded and blind-tested at 1:50 dilutions in the bE6 RIPA. A representative bE6 RIPA experiment is shown in Fig. 5. Each experiment included standard positive and negative control sera. To give some indication of the strength of the antibody responses observed, several serial dilutions of the positive control serum were used in each experiment. Samples were scored positive if the intensity of their signal exceeded that of the lowest dilution of positive control serum which gave a signal distinguishable from that of the negative control. This

Baculovirus expression of HPV-16 E6 protein Table 1. Anti-HPV-16 antibody status of the test population

M

1

2

3

4

5

6

2343

7

8

determined by Western blotting to MS2 fusion proteins and by baculovirus E6 RIPA Antiserum Carcinoma patient MX35

B44 B92

HPV aMS2-E4

ctMS2-E7

aMS2-E6

DNA

status

bE6

RIPA

-

--

+

--

+/-

-

+

16

+

+

-

-

+

ND*

+ +

B99

-

-

+

16

B104

+/--

-

+

16

+

MX22

+/-

-

+

16

+

MX30

+

-

+

16

+

MXll

-

+

+

16

+

B74

+

+

+

-

+

MX26

--

--

--

18

--

MX27

-

-

-

16

+ /-

MX28

-

-

-

16

+ /-

MX24

-

-

-

16

-

MX19

-

-

-

16

-

MX15

+

-

-

16

+

MX20

+

-

-

18

-

MX01

-

+

-

-

MX05

+

+

-

16

MX13

+

+

-

-

-

-

Normal control 27B + 25B 21B +/Other (7) * ND, Not

ND ND ND ND

+/+

+ +/-

F i g . 5. R I P A

using

E6 antibodies

in human

35S-labelled

bE6-infected

sera. Lanes

cell lysates to detect

1 to 4 show

anti-

immunoprecipitations

- -

--

determined.

was typically at a 1 : 1250 or 1 : 625 dilution of the positive control serum. Samples were scored as equivocal if they gave a signal which was intermediate between the negative control and the lowest positive control dilution. Each sample was tested in at least two independent experiments. Samples which were eventually scored as equivocal were tested in at least three independent experiments. The results are shown in Table 1. All of the cervical carcinoma patients' sera that were scored positive by Western blotting were also positive by bE6 R I P A . However, of the 10 negative carcinoma patients, only five scored negative by bE6 R I P A . The remaining samples scored either positive or equivocal by bE6 R I P A . O f the 10 normal sera, eight scored negative by both methods, whereas one was positive and one was equivocal by bE6 R I P A . Notably, the normal sample that scored positive by bE6 R I P A had previously shown strong antibody responses to the E. coli-derived E4 fusion protein in Western blots, and had been selected for that reason (see Table 1). One seropositive carcinoma patient's serum did not give consistent results on repeated testing and was excluded from the study. Overall, we found that the use of native bE6 protein in the R I P A afforded a considerably increased rate of detection of anti-E6 antibodies in h u m a n serum.

using various patients' sera at 1:50 dilutions; lane 5, negative control serum; lanes 6 to 8, positive control serum at 1:2500, 1:250 and 1:50 dilutions, respectively.Lane M, Mr markers; sizes indicated to the left.

Discussion Previous studies have had little success in the detection of antibodies against HPV-16 E6 in h u m a n sera (Jenison et al., 1988; DiUner, 1990; Kochel et al., 1990). In the present study we found that both optimized Western blotting and bE6 R I P A could be used in the identification of anti-E6 serum antibodies. Soluble native HPV-16 E6 protein produced using a baculovirus vector was readily recognized by antibodies present in some cervical carcinoma patients' sera, suggesting that the baculovirus-derived E6 protein closely resembles the original immunogen, i.e. E6 protein encountered during the course of a natural infection. On the other hand, R P a E 6 and several other antisera raised against bacterially derived E6 fusion proteins were unable to recognize the native bE6 protein in immunoprecipitation assays. The functional integrity of the R P a E 6 antiserum was demonstrated by its ability to detect both MS2-E6 fusion protein and bE6 protein in Western blots. Moreover, this showed that the RP0tE6 antiserum recognized linear epitopes present on both forms of the E6 protein. However, these linear epitopes are not available as targets on the native E6 molecule, as evidenced by the failure of this antiserum to react in R I P A . Since the

2344

S. N. Stacey and others

human sera did react with the native protein, at least some of the targets for the human sera must be exposed on the native bE6 protein, and must necessarily be distinct from the RPctE6 antiserum targets. The targets specific to the human sera may be conformational epitopes, or distinct linear epitopes that are exposed on the surface of the molecule. To generate antisera against the HPV-16 E6 protein, we employed an MS2 replicase-E6 fusion protein expressed in E. coll. We were able to purify this protein to near homogeneity by zinc-charged metal chelating affinity chromatography, confirming the zinc-binding nature of this protein (Barbosa et al., 1989; Grossman & Laimins, 1989). The MS2-E6 fusion protein was found to be of very low solubility and we had to use strong denaturing agents (4 M-guanidine hydrochloride) before, during and after purification. As discussed above, the requirement for strongly denaturing agents for solubilization of MS2-E6 fusion protein probably limits its value as an immunogen, producing antisera that recognize only a limited set of linear epitopes. The value of the MS2-E6 fusion protein as a target antigen in serological assays is also constrained because the protein must be immobilized by a technique such as Western blotting. The denatured state of the protein would then restrict its serological target presentation to linear epitopes. As mentioned in the Introduction, this may cause loss of both sensitivity and specificity of the assay. The bE6 RIPA identified as positive or equivocal five of 10 carcinoma patients' sera and two of 10 normal sera designated E6 seronegative by MS2-E6 Western blotting. This was probably due to an increased sensitivity of the bE6 RIPA relative to Western blotting, produced by the presentation of more authentic target epitopes. Alternatively, this could be due to a reduction in specificity of the bE6 RIPA, resulting in the detection of cross-reactive antibody specificities. It is notable that all five of the discrepant samples which scored E6-negative by Western blotting but positive by bE6 RIPA had antibodies to at least one other HPV-16 ORF product or were from patients with HPV-16 DNA-positive tumours (see Table 1). These observations support the interpretation that increased sensitivity rather than reduced specificity of the bE6 RIPA accounts for the increased detection rate. At present it is not possible to confirm this unequivocally because the HPV type-specificity of the MS2-E6 Western blotting assay is not known, and a positive reaction is therefore not proof of genuine exposure to HPV-16. The presence of HPV DNA in tumours can be taken as positive proof of exposure to the virus. However, examples were seen where no serum antibodies were detected in HPV-16 DNA-positive cancer patients, and where strong antibody responses were seen in the absence of detectable HPV DNA.

Moreover, evidence from more extensive surveys has shown that the relationship between DNA status and seropositivity is far from clear (Mann et al., 1990; Cason et al., 1992). In the absence of a consistent, independent marker of exposure to HPV-16, we cannot prove that the additional antibody specificities detected using the bE6 RIPA are HPV-16-specific. The present study was undertaken to determine whether antibodies to the E6 protein could be detected using baculovirus-produced E6 protein in comparison with an optimized Western blotting protocol using E6MS2 fusion protein targets. We find that both techniques can be used for the detection of anti-E6 serological responses, but the baculovirus-based assay may be substantially more sensitive. The data suggest that antiE6 responses may be more common than previously detected, particularly amongst cervical carcinoma patients, and that seroepidemiological surveys that rely solely on linear target antigens run the risk of substantial misclassification errors. This information encourages new surveys to be carried out on defined populations to determine the prevalence of E6 seropositivity detected using baculovirus-expressed E6 target antigens. We would like to thank D. Jordan and N. Smith for technical assistance, Professor H. zur Hausen for HPV clones, Drs L. Gissmann and M. Stanley for antisera, and Drs R. Possee, P. Ertle, M. Page and E. Littler for methods, advice and materials for producing baculovirus recombinants. This work was supported by the Cancer Research Campaign.

References ANDROPHY, E. J., HUBBERT, N. L., SCHILLER, J. T. & LOWY D. R. (1987). Identification of the HPV-16 E6 protein from transformed mouse cells and human cervical carcinoma cell lines. EMBO Journal 6, 989-992. BAILEY, M. J. & POSSEE, R. D. (1991). Manipulation of baculovirus vectors. In Methods in Molecular Biology, vol. 7, pp. 147-168. Edited by E. J. Murray. Clifton, New Jersey: The Humana Press. BAKER, C. C., PHELPS, W. C., LINDGREN, V., BRAUN, M. J., GONDA, M. A. & HOWLEY, P. M. (1987). Structural and transcriptional analysis of human papillomavirus type 16 sequences in cervical carcinoma cell lines. Journal of Virology 61, 962 971. BANKS, L., SPENCE, P., ANDROPHY, E., HUBBERT, N., MATLASHEWSKI, G., MURRAY, A. & CRAWFORD, L. (1987). Identification of human papiUomavirus type 18 E6 polypeptide in cells derived from human cervical carcinomas. Journal of General Virology 68, 1351-1359. BANKS, L., MOREAU, F., VOUSDEN, K., PIM, D. & MATLASHEWSKI,G. (1991). Expression of the human papillomavirus E7 oncogene during cell transformation is sufficient to induce susceptibility to lysis by activated macrophages. Journal of Immunology 146, 2037-2042. BARBOSA,M. S., LOWY, D. R. 8z SCHILLER, J. T. (1989). Papillomavirus polypeptides E6 and E7 are zinc-binding proteins. Journalof Virology 63, 1404-140 BONNEZ, W., ROSE, R. C., REICHMAN, R. C. & STRIKE, D. G. (1990). The PstI XholI restriction fragment of the HPV-6b L 1 0 R F lacks immunological specificity as determined by sera from HPV-6 condyloma acuminatum patients and controls. In UCLA Symposia on Molecular and Cellular Biology, New Series, vol. 124, pp. 77-80. Edited by P. Howley & T. Broker. New York: Allan R. Liss.

Baculovirus expression o f H P V - 1 6 E 6 protein

BONNEZ, W., DA RIN, C., ROSE, R. C. & REICHMAN, R. C. (1991). Use of human papillomavirus type 11 virions in ELISA to detect specific antibodies in humans with condylomata acuminata. Journal of General Virology 72, 1343-1347. CASON, J., KAMBO, P. K., BEST, J. M. & MCCANCE, D. J. (1992). Detection of antibodies to a linear epitope on the major coat protein (LI) of human papillomavirus type-16 (HPV-16) in sera from patients with cervical intraepithelial neoplasia and children. International Journal of Cancer 50, 349-355. CHEN, L., KINNEY-THOMAS, E., HU, S.-L., HELLSTROM, I. & HELLSTROM, K. E. (1991). Human papillomavirus type 16 nucleoprotein E7 is a tumor rejection antigen. Proceedings of the National Academy of Sciences, U.S.A. 88, 110-114. CHRISTENSEN, N. D., KREIDER, J. W., CLADEL, N. M. & GALLOWAY, D. M. (1990). Immunological cross-reactivity to laboratory-produced HPV-I 1 virions of polysera raised against bacteriaUy derived fusion proteins and synthetic peptides of HPV-6b and HPV-16 capsid proteins. Virology 175, 1-9. COMERFORD, S. A., MCCANCE, D. J., DOUGAN, G. & TITE, J. P. (1991). Identification of T- and B-cell epitopes of the E7 protein of human papillomavirus type 16. Journal of Virology 65, 4681-4690. CONNOR, M. E. & STERN, P. L. (1990). Loss of MHC class-I expression in cervical carcinomas. International Journal of Cancer 46, 1029-1034. COWSERT, L. M., LAKE, P. ~. JENSON, A. B. (1987). Topographical and conformational epitopes of bovine papillomavirus type 1 defined by monoclonal antibodies. Journal of the National Cancer Institute 79, 1053-1057. DILLNER, J. (1990). Mapping of linear epitopes of human papillomavirus type 16: the El, E2, E4, E5, E6 and E7 open reading frames. International Journal of Cancer 46, 703-711. DIMAIO, D. (1991). Transforming activity of bovine and human papillomaviruses in cultured cells. Advances in Cancer Research 56, 133-159. DYSON, N., HOWLEY, P. M., MUNGER, K. & HARLOW, E. (1989). The human papillomavirus-16 E7 oncoprotein is able to bind to the retinoblastoma gene product. Science 243, 934-937. FUNG, M. C., CHIU, K. Y., WEBER, T., CHANG, T. W. & CHANG, N. T. (1988). Detection and purification of a recombinant human B lymphotropic virus (HHV-6) in the baculovirus expression system by limiting dilution and DNA dot-blot hybridization. Journal of Virological Methods 19, 33-42. GALLOWAY, D. A. & McDOUGALL, J. K. (1989). Human papillomaviruses and carcinomas. Advances in Virus Research 37, 125 171. GROSSMAN, S. R. & LAIMINS, L. A. (1989). E6 protein of human papillomavirus type 18 binds zinc. Oncogene 4, 1089-1093. JENISON, S. A., FIRZLAFF, J. M., LANGENBERG,A. & GALLOWAY,D. A. (1988). Identification of immunoreactive antigens of human papillomavirus type 6b by using Escherichia coli-expressed fusion proteins. Journal of Virology 62, 2115 2123. JOCHMUS-KUDIELKA, 1., SCHNEIDER, A., BRAUN, R., KIMMING, R., KOLDOVSKY, U., SCHNEWEIS, K. E., SEEDORF, K. & GISSMANN, L. (1989). Antibodies against the human papillomavirus type 16 early proteins in human sera: correlation of anti-E7 reactivity with cervical cancer. Journal of the National Cancer Institute 81, 16981703. KOCHEL, H. G., SIEVERT, K., MONAZAHIAN, M., MITTELSTADTDETERDING, A., TEICHMANN, A. & THOMSSEN, R. (1991). Antibodies to human papillomavirus type 16 in human sera as revealed by the use of prokaryotically expressed viral gene products. Virology 182, 644-654. KRCHlgLSd(, V., WAGNER, J., SUCFIANKOVJ,,A., KRCM/~, M., RITTEROV.~, L. & VONKA, V. (1990). Synthetic peptides derived from E7 region of human papillomavirus type 16 used as antigens in ELISA. Journal of General Virology 71, 2719-2724. LIM, P. S., JENSON, A. B., COWSERT, L., NAKAI, Y., LIM, L. Y., JIN, X. W. & SUNOBERG, J. P. (1990). Distribution and specific identification of papillomavirus major capsid protein epitopes by immunocytochemistry and epitope scanning of synthetic peptides. Journal of Infectious Diseases 162, 1263-1269.

2345

MANN, V. M., LOO DE LAD, S., BRENES, M., BRINTON, L. A., PAWLS, J. A., GREEN, M., REEVES, W. C. & RAWLS, W. E. (1990). Occurrence of IgA and IgG antibodies to select peptides representing human papillomavirus type 16 among cervical cancer cases and controls. Cancer Research 50, 7815-7819. MATLASHEWSKI, G. (1989). The cell biology of human papillomavirus transformed cells. Anticancer Research 9, 1447-1556. MENEGUZZI, G., CERNI, C., KIENY, M. P. & LATHE, R. (1991). Immunization against human papillomavirus type 16 tumor cells with recombinant vaccinia viruses expressing E6 and E7. Virology 181, 62-69. MUNGER, K., WERNESS, B. A., DYSON, N., PHELPS, W. C., HARLOW, E. & HOWLEY, P. M. (1989). Complex formation of human papillomavirus E7 proteins with the retinoblastoma tumor suppressor gene product. EMBO Journal 8, 4099-4105. NIEDOmTEK, G. & HERaST, H. (1991). Human papillomaviruses and cancer. Cancer Journal 4, 163-167. PAGE, M. J. (1989). p36C: an improved baculovirus expression vector for producing high levels of mature recombinant proteins. Nucleic Acids Research 17, 454. PFISTER, H. (1987). Human papillomaviruses and genital cancer. Advances in Cancer Research 48, 113-147. SAMBROOK, J., FRITSCH, E. F. & MAmATIS, T. (1989). Molecular Cloning: A Laboratory Manual 2nd edn. New York: Cold Spring Harbor Laboratory. SCHNEIDER-GADICKE, A. & SCHWARZ, E. (1986). Different human cervical carcinoma cell lines show similar transcription patterns of human papillomavirus type 18 early genes. EMBO Journal 5, 22852292. SCHNEIDER-GADICKE, A., KAUL, S., SCHWARZ, E., GAUSEPOHL, H., FRANK, R. & BASTERT, G. (1988). Identification of the human papillomavirus type 18 E6* and E6 proteins in nuclear protein fractionations from human cervical carcinoma cells grown in the nude mouse or in vitro. Cancer Research 48, 2969-2974. SCHWARZ, E., FREESE, O. K., GISSMANN, L., MAYER, W., ROGGENBUCK, B., STREMLAU, A. & ZUR HAUSEN, H. (1985). Structure and transcription of human papillomavirus sequences in cervical carcinoma cells. Nature, London 314, 111-114. SEEDORF, K., KRAMMER,G., DURST, i . , SUHAI, S. & ROWENKAMP, W. (1985). Human papiUomavirus type 16 DNA sequence. Virology 145, 181-185. SEEDORF, K., OLTERSDORF,T., KRAMMER, G. & ROWEKAMP,W. (1987). Identification of early proteins of the human papillomaviruses type 16 (HPV-16) and type 18 (HPV 18) in cervical carcinoma cells. EMBO Journal 6, 139-144. SMOTKIN, O. & WETTSTEIN, F. O. (1986). Transcription of human papillomavirus type 16 early genes in a cervical cancer and a cancer derived cell line and identification of the E7 protein. Proceedings of the National Academy of Sciences, U.S.A. 83, 4680-4684. STRANG, G., HICKLING, J. K., MCINDOE, G. A. J., HOWLAND, K., WILKINSON, D., IKEDA, H. & ROTHBArtD, J. B. (1990). Human T celt responses to human papillomavirus type 16 LI and E6 synthetic peptides: identification of T cell determinants, HLA-DR restriction and virus type specificity. Journal of General Virology 71, 423-431. STRIKE, D. G., BONNEZ, W., ROSE, R. C. & REICHMAN, R. C. (1989). Expression in Escherichia coil of seven DNA fragments comprising the complete L1 and L2 open reading frames of human papillomavirus type 6b and localization of the 'common antigen' region. Journal of General Virology 70, 543 555. SUMMERS, M. D. & SMITH, G. E. (1987). A Manual of Methods for Baculovirus Vectors and Insect Cell Culture Procedures. Texas Agricultural Experimental Station, Texas A&M University, Texas. VOUSDEN, K. H. (1989). Human papillomaviruses and cervical carcinoma. Cancer Cells 1, 43-50. WERNESS, B. A., LEVINE, A. J. & HOWLEY, P. M. (1990). Association of human papillomaviruses types 16 and 18 E6 proteins with p53. Science 248, 76-79.

(Received 20 March 1992; Accepted 3 June 1992)