Identification of Human Papillomavirus Type 16 L1 Surface Loops Required for Neutralization by Human Sera

Abstract

The variable surface loops on human papillomavirus (HPV) virions required for type-specific neutralization by human sera remain poorly defined. To determine which loops are required for neutralization, a series of hybrid virus-like particles (VLPs) were used to adsorb neutralizing activity from HPV type 16 (HPV16)-reactive human sera before being tested in an HPV16 pseudovirion neutralization assay. The hybrid VLPs used were composed of L1 sequences of either HPV16 or HPV31, on which one or two regions were replaced with homologous sequences from the other type. The regions chosen for substitution were the five known loops that form surface epitopes recognized by monoclonal antibodies and two additional variable regions between residues 400 and 450. Pretreatment of human sera, previously found to react to HPV16 VLPs in enzyme-linked immunosorbent assays, with wild-type HPV16 VLPs and hybrid VLPs that retained the neutralizing epitopes reduced or eliminated the ability of sera to inhibit pseudovirus infection in vitro. Surprisingly, substitution of a single loop often ablated the ability of VLPs to adsorb neutralizing antibodies from human sera. However, for all sera tested, multiple surface loops were found to be important for neutralizing activity. Three regions, defined by loops DE, FG, and HI, were most frequently identified as being essential for binding by neutralizing antibodies. These observations are consistent with the existence of multiple neutralizing epitopes on the HPV virion surface.

Human papillomaviruses (HPVs) are a family of double-stranded DNA viruses that infect epithelial cells in a tissue-specific fashion. Infection with certain “high-risk” types that infect the genital mucosa, such as types 16 (HPV16), 18, and 31, has been shown to be a necessary step in the progression to cervical cancer (1). In nations without effective cervical cancer screening programs, HPVs are the cause of considerable morbidity and mortality (25). In the United States, more than $6 billion is spent annually on evaluation and management of low-grade lesions caused by HPV infection (9). For these reasons, the newly developed vaccines that can prevent HPV infection and that hold promise for eradication of cervical cancer have been greeted with enthusiasm. Currently, vaccines for types 16 and 18 and for the “low-risk” types 6 and 11 (which cause genital warts) are in phase 3 clinical trials (11, 24).

The vaccines that are now being evaluated are composed of the HPV major late protein (L1) for each type. This protein self-assembles into empty capsids, also referred to as virus-like particles (VLPs) (10, 13). Early indications are that VLP vaccines are safe and provide protection from persistent HPV infection in a type-specific fashion (11, 24). Animal studies suggest that protection from papillomavirus infection is mediated by antibodies (2, 23).

Type-specific antibodies recognize conformation-dependent epitopes involving the surface-exposed loops of L1 proteins that exhibit considerable amino acid sequence variation between types (7, 8, 17, 18). It has been suggested that an epitope composed of the FG and HI loops is immunodominant (8) for HPV16. However, binding of HPV16-specific immune human sera was not transferred to HPV11 VLPs that had HPV16 substitutions for these regions (26). Studies of HPV6 and -11 monoclonal antibody (MAb) binding specificity indicated that the BC, DE, and HI loops were often important for these types (14-17). The DE loop of HPV6 was also found to be important for recognition by some human sera (18).

Neutralizing epitopes have been mapped to one or a combination of the BC and EF loops of HPV6 (17), the DE and HI loops of HPV11 (14, 15, 16), and the FG and HI loops of HPV16 (8, 20). Neutralizing MAbs have been found that recognize conformation-dependent epitopes consisting of only one loop (15), but more commonly two noncontiguous loops constituted the epitope (8, 17). An interesting study by Sadeyen et al. (22), in which an HBV epitope was inserted into each of the five loops of HPV16 L1, indicated that amino acid changes on any of the loops diminished the HPV16-specific immunogenicity of VLPs. However, insertions into the FG loop reduced the HPV16-specific immunogenicity to a greater extent than insertions into other loops.

To identify residues on the surfaces of HPV16 virions important for neutralizing activity, we employed a series of HPV16/31 hybrid VLPs. These particles, which had one or more loops of HPV16 replaced with an HPV31 loop(s) or, in one case, an HPV52 loop, were used to adsorb antibodies from human sera. Treated sera were then tested for activity in a pseudovirus neutralization assay. This approach permitted us to determine which loops were recognized by neutralizing antibodies and to address the question of whether there is a dominant neutralizing epitope on HPV16.

(These data were presented in part at the 22nd International Papillomavirus Meeting, Vancouver, Canada, May 2005.)

MATERIALS AND METHODS

Cell lines.

Sf9 cells (ATCC) were grown in SF-900 II medium (Invitrogen, Carlsbad, CA) augmented with 10% fetal bovine serum (Invitrogen). Virus-like particles were produced in Sf9 cells and purified by centrifugation as described previously (6). 293TT cells (generously provided by John Schiller, National Cancer Institute) were grown in Dulbecco's modified Eagle's medium (Invitrogen) with supplements as previously described (19).

Antibodies and sera.

Human sera were obtained from two studies: a natural history study of HPV infection among university women (27) and a case control study of anogenital cancers (5). All sera had been previously screened for HPV16 antibodies by enzyme-linked immunosorbent assay (ELISA) (3). Sera that had reacted most strongly by ELISA to HPV16 VLPs and for which there was sufficient volume were chosen for evaluation. Two sera were selected from one woman for whom there was over 9 years of follow-up. Ascites or hybridomas were generously provided by Neil Christensen (Hershey Medical Center, Hershey, PA) (H16.V5, H31.A4, and H16.U4) and Richard Roden (Johns Hopkins University, Baltimore, MD) (H16.E70). The institutional review board of the Fred Hutchinson Cancer Research Center or the University of Washington approved all research protocols.

Virus-like particles.

The creation and purification of the VLPs used here have been previously described (6). Briefly, mutations were created in the HPV16 L1 sequence in the pDest8 vector (Invitrogen) using primers that were designed to replace the HPV16 amino acid sequences with homologous sequences from HPV31 (16:DE, 16:EF, 16FG, 16:HI, 16:410, and 16:430) or HPV52 (16:BC). The rational for generating 16:BC using sequences from HPV52 was that we had been interested in residue 50 (6) (residue 50 is Phe for types 16 and 52, but the type 31 residue 50 is Tyr). However, residue 50 was not of particular interest in this study. A similar procedure was followed to alter the HPV31 L1 sequence to contain HPV16 on the FG and HI loops. After the plasmids were sequenced, recombinant bacmids were created using the Bac-to-Bac expression system (Invitrogen). Sf9 cells were transfected with bacmid DNA, and after several rounds of amplification, a baculovirus stock was generated. Sf9 cells were infected, and particles were purified by centrifugation over sucrose and CsCl gradient centrifugation (6).

Neutralization assay.

The preparation of pseudovirus and neutralization assays were performed essentially as described by Pastrana et al. (19). Briefly, pseudovirus stocks were prepared by transfection of 293TT cells with vectors expressing HPV16 L1, HPV16 L2, and secreted alkaline phosphatase (p16L1h p16L2h and pYSEAP; provided by John Schiller). Three days later, the cells were harvested and pseudovirus was isolated over Optiprep (Accurate Chemical, Westbury, NY) gradients. Fractions from the Optiprep gradients were tested for the presence of pseudovirus by infection of 293TT cells in 96-well plates and assayed for secreted alkaline phosphatase activity in the supernatant after 3 days of incubation.

Neutralization assays were performed by incubation of diluted human sera or ascites with diluted pseudovirus stocks for 1 h on ice in a total volume of 120 ml of tissue culture medium without phenol red. These incubations were performed in silanized polypropylene 96-well plates (Fisher Scientific, Pittsburgh, PA). One hundred microliters from each well was transferred to a 96-well tissue culture plate that had previously been seeded with 293TT cells. The method for detection of secreted alkaline phosphatase in supernatants was a modification of that of Pastrana et al. (19). Their procedure recommended the use of a luminescent assay; however, we found that use of a standard alkaline phosphatase substrate gave equivalent results at considerably reduced cost (data not shown). To detect alkaline phosphatase, 30 μl of supernatant was transferred from the infected 293TT cells into a standard ELISA plate. One hundred microliters of developer (0.1 M CO₃, 10 mM MgCl₂, 4.3 mg/ml Sigma 104-Phosphate [Sigma Chemicals Inc., St Louis MO], pH 9.5) was added to each well, and the plates were incubated at room temperature. Optical density readings (405 nm) were recorded at 30 min and 1 h using a microplate reader (EL_x 808; Bio-Tek Instruments Inc., Winooski, VT).

Experiments designed to identify amino acids recognized by neutralizing sera were performed in one of two ways. The first approach was to serially dilute VLPs on polypropylene plates in tissue culture medium (30 μl per well). Sera diluted in tissue culture medium were then added in 30 μl at a final concentration of 1:100 (human sera) or 1:75,000 (MAb 16.V5). The plates were sealed and incubated overnight at 4°C. The following day, 60 μl of diluted pseudovirus stock was added to each well, and the plates were incubated on ice for 1 h. The contents of each well (100 μl) was transferred to a tissue culture dish (96-well plate) previously seeded with 293TT cells. Neutralization assays were conducted as described above. The second approach was to serially dilute sera (or MAb) on silanized polypropylene plates in tissue culture medium before adding VLPs at a concentration determined to be optimal (final volume, 60 μl). The plates were sealed and incubated overnight at 4°C. Diluted pseudovirus stock solution was added, and neutralization assays were conducted as described above. Because of the propensity of VLPs to aggregate, the mass of protein used may not reflect the quantity of available epitopes. Therefore, the quantity of VLPs in each preparation was compared by ELISA rather than by protein assays.

The reproducibility of the assay was assessed by repeatedly performing the neutralization assay using several VLP preparations and the H16.V5 antibody (because of the large amount of serum used in these experiments, we were not able to routinely assay human sera repeatedly). The assay was performed once in triplicate using four antigens on separate plates to assess intra-assay variation and on 4 days to assess interassay variation. The coefficient of variation of triplicates was 7.38% (range, 5.82% to 8.97% for four antigens). The interassay coefficient of variation was 18.73% (range, 15.90% to 22.75% for four antigens), with the same pattern of inhibition consistently observed.

RESULTS

Inhibition of neutralization by H16.V5 MAb.

A recently developed pseudovirus-neutralizing assay (19) was modified to identify regions on the surfaces of VLPs important for adsorbing neutralizing activity (Fig. 1). In this assay, antibodies were pretreated with VLPs, either wild-type or hybrid VLPs that had one or more HPV16 loops replaced with homologous sequences from HPV31 or HPV52. Wild-type and hybrid VLPs that retain epitopes necessary for HPV16 neutralization activity adsorb antibodies, thus allowing pseudovirus infection and high alkaline phosphatase activity. Wild-type HPV31 and hybrid HPV16 VLPs that had neutralizing epitopes disrupted by replacement with HPV31 or HPV52 sequences failed to adsorb neutralizing antibodies, resulting in reduced pseudoviral infection and low levels of alkaline phosphatase activity (Fig. 1).

FIG. 1.

Diagram of the method used for identification of neutralizing epitopes. Antibodies and VLPs were incubated together overnight as described in Materials and Methods. In the presence of wild-type HPV16 VLPs or hybrid VLPs that retained neutralizing epitopes, the effective concentrations of neutralizing antibodies are reduced. Pseudovirus was then added, the cells were infected, and secreted alkaline phosphatase activity was measured 3 days postinfection. If pretreatment of sera failed to adsorb neutralizing activity, low alkaline phosphatase activity was detected. Conversely, high alkaline phosphatase activity indicated a low concentration of neutralizing antibodies and that the VLPs used for pretreatment retained neutralizing epitopes.

The H16.V5 MAb is known to be neutralizing (21), and the variable loops required for recognition have been determined (26). To see if the surface loops important for binding were identical to regions required for neutralization, wild-type and hybrid VLPs were used to pretreat H16.V5 ascites prior to use in a pseudovirus neutralization assay. Hybrid VLPs were composed of either the HPV16 or HPV31 backbone, onto which one or two regional substitutions were made using homologous sequences from the other type (or type 52 for the BC region substitution). The regions replaced were the variable surface loops identified by Chen et al. (7) and two variable regions between residues 400 and 450, one of which had been found to be important in H16.U4 binding (6).

It was important to confirm that all VLPs were in a native conformation. We have shown previously that all of the hybrid VLPs used in these experiments folded into a native conformation by binding to MAbs that recognize conformation-dependent epitopes and by resistance to proteolysis by trypsin (6). To confirm that the hybrid VLPs used here retained native epitopes, they were reacted with MAbs that are known to bind HPV16 or HPV31 only in a native conformation (Fig. 2A). Although all hybrid and wild-type VLPs reacted with MAbs as predicted, it should be noted that the relative concentrations of 16:HI and 16:FG/HI were much lower than for 16:FG (Fig. 2A, H16U4 graph).

FIG. 2.

(A) Identifying surfaces recognized by antibodies and required for neutralizing activity by H16.H5. To determine if the VLPs were correctly folded, they were tested in direct binding assays (ELISAs) using MAbs known to recognize the various hybrids. H16.V5 and H16.U4 are HPV16 MAbs that recognize epitopes known to depend on the native conformation of the HPV16 VLPs. H31.A4 is a specific HPV31 MAb that recognizes a conformation-dependent epitope on HPV31 VLPs. (B) H16.V5 was used to inhibit pseudovirus infection following incubation with VLPs. H16.V5 was titrated across a plate, and hybrid or wild-type VLPs were added. The following day, HPV16 pseudovirions were added to each well and incubated on ice for 1 h. Those samples were transferred to a 96-well tissue culture dish seeded with 293TT cells. After 3 days, 30 μl was removed from each well and tested for alkaline phosphatase activity.

At the highest concentration of H16.V5 tested, all of the samples neutralized the pseudovirus infections, regardless of the VLP type used to pretreat the antibody (Fig. 2B). At lower H16.V5 concentrations, some of the samples neutralized (lower optical densities) while others did not (higher optical densities). Higher optical-density values indicated that the VLPs used for pretreatment adsorbed neutralizing activity. In the example shown (Fig. 2B), the following VLPs adsorbed neutralizing antibodies: wild-type 16, 16:BC, 16:DE, 16:EF, 16:410, and 16:430. In contrast, wild-type 31, 16:FG, 16:HI, 16FG/HI, and 31:FG/HI could not adsorb neutralizing antibodies. These findings indicated that sequences on the FG and HI loops were necessary for adsorbing neutralizing activity. This is the same region recognized as the H16.V5 binding site (26). It was previously shown that H16.V5 binding could be transferred to HPV31 by the HPV16 FG and HI loops alone (6). In this experiment, however, the neutralizing epitope was not transferred to HPV31 by substitution of these two regions, in spite of the fact that H16:V5 bound 31:FG/HI in an ELISA (Fig. 2A).

The 16:EF and 16:DE VLPs were relatively inefficient at adsorbing 16.V5 antibodies compared with wild-type 16 VLPs and 16:BC (Fig. 2B). This observation suggested either that the wild-type and 16:BC VLP preparations were more concentrated (∼5 times) than the 16:EF and 16:DE preparations or that the neutralizing antibodies did not bind as well to the 16:DE or 16:EF VLPs. It is interesting that although 16:DE VLPs had a lower apparent concentration than 16:BC, 16:410, and 16:430 VLPs in the neutralization assay, 16:DE had a concentration similar to that of 16:BC and a higher concentration than 16:410 and 16:430 by ELISA (compare the curves in Fig. 2A with Fig. 2B). Thus, the neutralization assay might detect subtle differences in antibody-antigen interactions not measured by binding assays. These data suggest that amino acid changes on DE (and perhaps EF) could directly or indirectly reduce binding to the neutralizing epitope.

Identification of variable surface loops important for neutralization by human sera (titration of VLPs).

To identify surface loops on VLPs important for neutralization activity by human sera, the sera were pretreated with VLPs as described above and tested for neutralization activity in a pseudovirus assay. In these experiments, the concentration of human sera was held constant (1:100) and the concentration of each of the VLPs was varied. Figure 3 shows the results from four sera. Serum “A” did not bind to HPV16 VLPs in an ELISA (data not shown) and, as expected, was not able to neutralize pseudovirus infection regardless of the VLPs used for pretreatment. At higher concentrations of VLPs, there was a gradual diminution of optical density values, suggesting that the VLPs might compete with pseudovirus for binding to the cells, but the effect was modest (Fig. 3A). Neutralizing antibodies in serum “B” were adsorbed by VLPs on which the FG and HI loops were type 16. Wild-type HPV16, 16:DE, 16:EF, 16:410, and 16:430 VLPs effectively adsorbed neutralizing antibodies from serum “B.” The 16:FG and 16:HI hybrid VLPs partially adsorbed neutralizing antibodies, and the double hybrid VLPs 16:FG/HI failed to adsorb neutralizing antibodies. Compared with HPV31 wild-type VLPs, the 31:FG/HI hybrid VLPs acquired the ability to partially adsorb neutralizing antibodies. Thus, the neutralizing activity of serum “B” recognized an epitope consisting of residues on the FG and HI loops. For serum “C,” the following VLPs did not have the ability to adsorb neutralizing antibodies: 31 wild type, 31:FG/HI, 16:DE, 16:FG, 16:HI, and 16:FG/HI. Neutralizing activity in this serum required HPV16 sequences on the DE, FG, and HI loops. Surprisingly, substitution of any one of those regions alone totally ablated the neutralization adsorption activity of the VLPs. Serum “D” required the same loops as “C”; however, the concentration of serum “D” was lower, and VLP titering failed to reach the end point (this was also true for serum “B”). The concentration of anti-HPV16 antibodies in serum “C” appeared to be higher because at a VLP dilution of 1:1,000, wild-type HPV16 could not adsorb neutralizing antibodies from serum “C” but the same concentration of wild-type VLPs effectively adsorbed neutralizing antibodies from sera “B” and “D” (Fig. 3).

FIG. 3.

Identification of VLP surface loops important for neutralizing activity in human sera by titration of VLPs. Wild-type and hybrid VLPs were titrated and incubated with one of four human sera overnight. These samples were then tested for residual activity in a pseudovirus neutralization assay. Lower optical density readings indicated that neutralization activity was not adsorbed by pretreatment with VLPs. Higher optical density readings indicated that neutralization activity was adsorbed or that there was no neutralizing activity in the serum. Serum A was known to be nonreactive with HPV16 VLPs (not shown). A control for each experiment was serum that was not pretreated with VLPs and that was used to neutralize pseudovirus infection (on the right side of each graph). The symbols represent the same VLP preparations as in Fig. 2.

Table 1 summarizes the results from these experiments for the four sera shown in Fig. 3 (sera A to D) and an additional four sera (see Fig. S1 in the supplemental material for graphs). Since serum A had high alkaline phosphatase activity when treated with HPV31 VLPs, this indicated that the serum had no HPV16-specific neutralizing antibodies. The VLPs with the DE replaced from HPV31 (16:DE) could not adsorb neutralizing antibodies from four of the seven sera (which had neutralizing antibodies) and partially adsorbed neutralizing activity from two of the other three sera. This indicated that neutralizing antibodies in six of seven sera tested recognized the DE loop. The HI loop was recognized by all of the immune sera tested, but in four sera, 16:HI VLPs adsorbed less neutralizing activity than did 16:FG/HI VLPs that had two regions replaced. One possible explanation for this result is that amino acid substitutions made on the HI loop alone might disrupt the conformation of a third region (perhaps the DE loop) that might be important for neutralization by some sera; however, when both FG and HI were HPV31, the induce conformational change might be relieved.

TABLE 1.

Alkaline phosphatase activities of cells infected with pseudovirus following neutralization with serum pretreated with various concentrations of wild-type or hybrid VLPs

Serum	Activity for VLP used to pretreat seraa
	16wt	16:BC	16:DE	16:EF	16:FG	16:HI	16:FG/HI	16:410	16:430	31wt
A	+	ND	+	+	+	+	+	+	+	+
B	+	ND	+	+	±	±	−	+	+	−
C	+	ND	−	+	−	−	±	+	+	−
D	+	ND	−	+	−	−	−	+	+	−
E	+	ND	±	−	ND	−	±	+	+	−
F	+	ND	−	+	ND	−	−	+	+	−
G	+	ND	−	+	ND	−	±	+	+	−
H	+	ND	±	±	ND	−	±	+	+	−

^aND, not done; +, high optical-density readings; −, low optical-density readings; ±, intermediate optical-density readings.

Identification of variable surface loops important for neutralization by human sera (titration of sera).

The method used in the experiments described above did not work well for all sera that had neutralizing antibodies (see Fig. S1 in the supplemental material). We suspected that the reason for the failure to identify loops important for neutralizing activity in some sera was because the serum concentration was not optimum. The data presented in Fig. 3 indicated that VLPs were effective at adsorbing neutralizing antibodies over a broad range of concentrations. Suitable dilutions of VLPs were selected based on those observations and used to pretreat sera that had been titrated and tested in a pseudovirus neutralization assay (Fig. 4; see Fig. S2 in the supplemental material). Again, it was found that some VLPs were very efficient at adsorbing neutralizing antibodies (higher optical density readings), whereas other VLPs were not (lower optical density readings). Sera with lower antibody titers were neutralizing only at the highest concentrations of antibody tested (Fig. 4, serum J), while sera with high titers neutralized pseudovirus infection regardless of the VLPs used for pretreatment at high concentrations of serum (Fig. 4, serum M). To quantify the efficiency of neutralization inhibition by different hybrid VLPs, data from the three dilutions of sera that had the greatest optical density differences between HPV16 VLPs and HPV31 VLPs were used. The proportion of alkaline phosphatase activity for each VLP type was normalized to that of HPV16 wild-type VLPs after subtracting the HPV31 wild-type VLP optical density values. The bar graphs (Fig. 4) represent the normalized values computed from the raw data presented in the line graphs on the left. Sequences important for recognition by serum “K” were found on the DE and FG loops (and potentially the HI loop, but the 16:HI VLPs could not adsorb antibodies from any sera tested and it was suspected that the preparation had denatured) (data not shown). Sequences important for recognition by serum “J” were on the EF loop, and for serum “M” on the DE and FG loops. Sequences important for recognition by serum “L” were not clear. The BC, DE, and EF regions all showed some reduction in the ability to adsorb neutralization activity, suggesting that the epitope may be a combination of sequences on all of these regions or that this serum contained antibodies with a variety of specificities.

FIG. 4.

Identification of surface loops required for neutralizing activity in human sera by titration of sera. Human sera were titrated, mixed with wild-type or hybrid VLPs, and incubated overnight. These samples were then tested for residual activity in a pseudovirus neutralization assay. Lower optical density readings indicated that neutralization activity was not adsorbed by pretreatment with VLPs. Relatively high optical density readings indicated that neutralization activity was adsorbed or that there was no neutralizing activity in the serum. The bar graphs represent the optical density values from the graphs on the left that have been normalized for HPV16 wild-type activity. Only the three dilutions of VLPs that showed the greatest difference between HPV16 wild type and HPV31 wild type were used, except for serum J, where only the highest concentration of VLPs was used. The error bars are standard deviations for the three values.

Table 2 summarizes data from these experiments for 15 sera (also see Fig. S2 in the supplemental material). The FG/HI region was essential for neutralizing activity for 4 of 15 sera. There was a partial reduction in the ability of FG/HI VLPs to adsorb in 10 of the remaining 11 sera. The HPV16 DE loop was essential for adsorption of neutralizing antibodies among 10 of 15 sera tested, with an additional 3 sera showing partial adsorption. The FG loop was essential for adsorption of neutralizing activity for 2 of 10 sera, with 5 of the remaining 8 showing partial reduction of adsorption. The EF loop was essential for 2 and partially responsible for neutralization for 6 of 15 sera. Replacement of the BC, 410, or 430 region rarely disrupted the ability of HPV16 VLPs to adsorb neutralizing antibodies.

TABLE 2.

Alkaline phosphatase activities of cells infected with pseudovirus following neutralization with various concentrations of serum pretreated with wild-type or hybrid VLPs

Serum	Activity for VLP used to pretreat seraa
	16wt	16:BC	16:DE	16:EF	16:FG	16:FG/HI	16:410	16:430	31wt	No capsids
I	+	+	±	±	ND	−	+	+	−	−
J 2001	+	±	+	−	+	±	±	+	−	ND
K	+	+	−	+	±	±	+	+	−	ND
L	+	+	+	±	+	+	+	+	−	ND
M	+	+	±	+	−	−	+	+	−	ND
N	+	+	−	±	ND	±	+	+	−	−
O	+	+	−	±	ND	−	+	+	−	−
P	+	+	−	+	ND	±	+	+	−	−
Q	+	±	−	+	ND	±	+	+	−	−
R	+	+	−	+	±	±	+	+	−	ND
S	+	+	±	+	±	±	+	+	−	ND
T	+	±	−	±	+	±	+	+	−	ND
U	+	+	−	+	±	±	+	+	−	ND
V	+	±	−	±	±	±	+	+	−	ND
J 1992	+	+	−	−	−	−	+	+	−	ND

^aND, not done; +, normalized optical-density values ≥2/3 of wild-type 16 value; −, normalized optical-density values <1/3 of wild-type 16 value; ±, normalized optical-density values >1/3 but <2/3 of wild-type 16 value.

DISCUSSION

To identify regions on HPV16 virions important for neutralization, a series of HPV16/31 hybrid VLPs were used to preadsorb neutralizing antibodies from human sera before testing them in pseudovirion neutralization assays. The sera used in these experiments were from women naturally infected with HPVs. Although sera were selected for having relatively high titers, natural HPV infections do not induce strong antibody responses (4). Loops important for neutralization were identified when wild-type VLPs adsorbed neutralization activity but hybrid VLPs (in which one or more loops of HPV16 were replaced with homologous regions from HPV31) failed to do so.

One of the interesting findings from this study was that, for the great majority of sera tested, the ability of VLPs to adsorb HPV16-neutralizing antibodies was destroyed by substitution with a single loop from HPV31. It might have been anticipated that disruption of a single loop on VLPs would have decreased, but not ablated, the ability of particles to adsorb neutralizing activity. That pattern was observed in only 2 of 23 sera examined here (sera “L” and “S”) and prevented identification of the specific loop(s) required for neutralizing activity. For the other 21 sera, adsorption of neutralizing activity was destroyed by substitution of a single region (loop) from HPV31 onto the HPV16 backbone. Thus, although sera recognized a variety of different combinations of loops, it appeared that for most sera, a single domain of the L1 surface, composed of multiple discontinuous regions, was recognized by neutralizing antibodies.

A second interesting finding was that, of the sequences required for neutralization defined here, none consisted of a single loop. In this analysis, for a serum to have all neutralizing residues on a single loop, one VLP hybrid (with a single loop substitution) would have had to reduce the normalized optical densities by greater than two-thirds, but no other VLP hybrid could reduce the values by more than one-third. Most sera appeared to require sequences on two loops or three loops, but two sera were identified that required sequences on four loops (sera “T” and “V”). This finding is in contrast to binding studies that have identified epitopes consisting of one or two loops (8, 14, 15, 17). It is possible (as with all experiments of this type) that amino acid sequence changes on one loop altered the conformations of adjacent loops. Although the five loops do not contact a single site, each loop is in close proximity to at least one other loop (7). Induced conformation changes on adjacent loops could reduce neutralization adsorption activity. Changes in antibody binding might have been detected in these assays but not in ELISAs, because the ELISA might detect only antibodies with higher avidity (19). Thus, our estimation of the number of loops involved might overestimate the number of loops that were in direct contact with neutralizing antibodies. This problem is being addressed by using L1 protein with single-amino-acid substitutions in the regions of interest.

Vaccination with L1 VLPs induces high-titer neutralizing serum (2, 12, 23); however, because there are approximately 30 genital HPV types, making a vaccine specific for each type is impractical. This has led to an interest in creating polyvalent VLPs that could induce immunity to multiple types. Chimeric L1 VLPs have been produced that induce neutralizing sera specific for two types (8). The findings presented here, indicating that most neutralizing sera required sequences on two or more loops (and given the fact that the number of antigenic sites is limited to five loops [7]), implies that creating chimeric L1 VLPs that can induce immunity for more than two types will be difficult, if not impossible.

Finally, there was not a single neutralizing epitope that all of the sera recognized. The regions most frequently identified as influencing a neutralizing activity were the DE loop (82.6% of sera) (combined data from Tables 1 and 2), the FG loop (71.3% of sera), and the EF loop (43.5% of sera). The VLPs with two regions replaced with HPV31 sequence (16:FG/HI) were identified as being important for neutralization for 91.3% of sera tested. Comparison of results using FG/HI with results using FG VLPs suggests that the HI region was important for recognition by a minimum of 20.0% of sera. Unfortunately, this question could not be addressed directly because of technical difficulties in producing native 16:HI VLPs. Pretreatment of sera with 16:FG/HI VLPs more often showed a partial loss of the ability to adsorb neutralizing antibodies (14 of 23 sera; 60.9%) rather than a reduction to less than one-third of wild-type VLPs (7 of 23 sera; 30.4%). In comparison, pretreatment of sera with 16:DE VLPs more often showed a greater disruption in the ability to adsorb neutralization antibodies (to <1/3 of wild type; 14 of 23 sera; 60.9%) than a partial disruption (5 of 23 sera; 21.7%). This again indicated that the region most important for neutralizing human sera included the DE loop. The other regions (the BC loop and the variable regions between residues 410 and 430) were infrequently involved in neutralizing epitopes.

A previous report described the FG and HI loops as the immunodominant epitopes on HPV16 (8) because they were the regions most frequently detected by mouse monoclonal antibodies. However, another study found that binding by most HPV16-reactive human sera was not transferred to HPV11 VLPs with HPV16 FG and HI loop substitutions (26). In the latter study, a comparison of HPV16 immune human serum binding was also made between wild-type HPV16 VLPs and HPV11/16 hybrid VLPs in which the first 171 residues (which encode loops BC and DE) were from HPV11, with the remainder of the molecule being HPV16 sequences. Again, many HPV16-reactive sera bound to wild-type but not to hybrid VLPs, suggesting that perhaps the BC or DE loops were important for HPV16 binding by human sera. Thus, our results and those of Wang et al. (26) indicate that the FG and HI loops alone do not define the HPV16 immunodominant epitope and implicate DE as often being involved in human serum binding.

The assay used here to detect sequences important for neutralizing activity found subtle differences between antibody binding and neutralization activity. For instance, 31:FG/HI VLPs did not adsorb neutralizing H16.V5 but were recognized by this MAb in ELISA (Fig. 2). These differences may be due to the fact that one assay takes place in solution while the other is done on a solid surface. It is also possible that the neutralization assay was more sensitive to antibody-VLP avidity than the ELISA.

In summary, the use of hybrid VLPs to adsorb HPV16-specific antibodies from human sera permitted the identification of surface loops required for neutralizing activity. Two or more variable regions on the VLP surface were required for neutralization activity for all sera tested. There was not a single dominant epitope, but sequences on DE and FG loops were required for neutralizing activity for all but two of the sera tested.