Virus-like particles and capsomeres are potent vaccines against cutaneous alpha HPVs

Abstract

The potential as prophylactic vaccines of L1-based particles from cutaneous genus alpha human papillomavirus (HPV) types has not been assessed so far. However, there is a high medical need for such vaccines since HPV-induced skin warts represent a major burden for children and for immunocompromised adults, such as organ transplant recipients. In this study, we have examined the immunogenicity of capsomeres and virus-like particles (VLPs) from HPV types 2, 27, and 57, the most frequent causative agents of skin warts. Immunization of mice induced immune responses resembling those observed upon vaccination with HPV 16 L1-based antigens. The antibody responses were cross-reactive but type-restricted in their neutralizing capacities. Application of adjuvant led to an enhanced potential to neutralize the respective immunogen type but did not improve cross-neutralization. Vaccination with capsomeres and VLPs from all four analyzed HPV types induced robust IFNγ-associated T-cell activation. Immunization with mixed VLPs from HPV types 2, 27, and 57 triggered an antibody response similar to that after single-type immunization and capable of efficiently neutralizing all three types.

Our results imply that vaccination with combinations of VLPs from cutaneous HPV types constitutes a promising strategy to prevent HPV-induced skin lesions.

1. Introduction

Cutaneous human papillomaviruses (HPVs) represent a large fraction of the more than 100 types identified so far [1]. Most cutaneous HPV types including those that are associated with the rare hereditary disease Epidermodysplasia verruciformis (EV) belong to genus beta [2,3]. Evidence is accumulating that some EV types may play a co-factor role in the development of non-melanoma skin cancer (NMSC) [4,5]. In contrast, there is only scarce information on the association of cutaneous genus alpha HPVs and NMSC [6]. Yet it is undisputed that they cause benign skin lesions, such as common, plantar, or flat warts [7]. Skin warts are common during childhood and constitute a particular burden for immunocompromised adults, such as organ transplant recipients who frequently suffer from their confluent occurrence at multiple body sites [8–10]. In particular, types 2, 27, and 57 were shown to rank among the most prevalent HPVs detected in cutaneous warts from renal transplant recipients [11]. A prophylactic vaccine protecting against infections by cutaneous HPV types would alleviate the burden represented by HPV-induced skin lesions. Such a measure could be applied before patients are treated with immunosuppressive drugs, as is already common-practice with a number of other vaccines [12].

In comparison to the cutaneous HPV types, some mucosal types are well characterized. In fact, two commercially available vaccines protecting from infections by mucosal types have recently been developed [13]. Gardasil® (Merck Sharpe & Dohme) and Cervarix® (GlaxoSmithKline) are both composed of virus-like particles (VLPs), which assemble spontaneously from the viral major capsid protein L1 [14]. Upon immunization with the vaccines, high titers of largely type-specific neutralizing antibodies are induced protecting from infection by either of the respective HPV types [15–17]. Both vaccines are formulated with aluminium-based adjuvants: for Gardasil®, the proprietary adjuvant amorphous aluminum hydroxysulfate (AAHS) is used, whereas Cervarix® contains aluminum hydroxide and monophosphoryl lipid A (MPL), a detoxified form of lipopolysaccharide (LPS). Both formulations were shown to achieve excellent efficacies [18,19]. However, VLPs for both vaccines are produced using eukaryotic expression systems, which are relatively costly [20,21]. Moreover, transportation of the vaccines requires a cold-chain further complicating their worldwide distribution. On the contrary, capsomeres, the pentameric subunits of VLPs, can be produced in bacteria, which are easier and more economical to maintain. As capsomeres are considered more stable than VLPs and can induce similar immune responses, they represent promising candidates as second-generation vaccines [22–25].

We reported previously that large amounts of VLPs from HPV types 2, 27, and 57 can be produced upon expression of full length L1 in baculovirus-infected insect cells [26]. We aimed to analyze the immunogenicity of these particles in comparison to capsids from HPV type 16. We compared VLP capture ELISA, GST-L1-based multiplex serology, and neutralization assays to gauge the biologically relevant humoral immune response. Furthermore, we addressed the question of whether bacteria constitute an alternative for the expression of L1-based antigens from cutaneous species alpha HPVs.

2. Materials and methods

2.1. Recombinant baculovirus stocks

Recombinant baculoviruses were generated using the Multi-Bac system as previously reported [26]. A detailed description of the expression system is provided in [27]. Briefly, the L1 open reading frames (ORFs) were introduced into the polyhedrin and p10 promoter-controlled multiple cloning sites of a pFBDM plasmid [28] by PCR amplification introducing the restriction sites EcoRI/HindIII and XhoI/SphI respectively. All constructs were sequenced and 10 ng of each plasmid were transformed into DH10MultiBac cells. Positive clones were identified by blue/white selection. Bacmid DNA was isolated and 1 µg was transfected into 5 × 10⁶ Sf9 cells by calcium phosphate precipitation. Baculoviruses were amplified at least three times and their titers were determined using a plaque assay [29] before they were used for productive infections.

2.2. Virus-like particle production and purification

For the production and purification of VLPs, a protocol described in [30] was employed. Briefly, 2 × 10⁸ Trichoplusia ni (TN) High Five cells (Invitrogen) were infected with recombinant baculovirus at an multiplicity of infection (MOI) of 2 [31,32]. Cells were incubated for three days at 27 ◦C and lysed by sonication. The cleared lysate was further purified by a two-step gradient consisting of 7 mL 30% sucrose on top of 7 mL 57.5% CsCl, which was centrifuged at 96,500 × g and 10 ◦C in a Beckman SW32 rotor for 3 h. The interphase between the sucrose and the CsCl layer was collected, mixed, and transferred into Quick-Seal tubes (Beckman), and centrifuged again for 16 h at 184,000 × g and 20 ◦C in a Sorval TFT 65.13 rotor. The gradient was fractionated (1 mL fractions) from the bottom of the tubes. The L1 protein content of each fraction was determined by Bradford assay using Rotiquant® (Roth) and by comparison to a BSA standard on a Coomassie-stained SDS PAGE gel. The peak fractions were purified further by centrifugation for 10 min at 20,000 × g and 4 ◦C following dialysis against 50 mM Hepes (pH 7.4, 0.3 M NaCl). VLPs were loaded on 1 mL HiTrapTM columns (GE Healthcare) over night and eluted with 50 mM Hepes (pH 7.4, 1 M NaCl). VLPs were analyzed by SDS-PAGE followed by staining with colloidal Coomassie (Thermo Scientific) and Western blot analysis. The concentration of L1 protein in the eluates was determined by Bradford assay and by comparing to a BSA standard on a Coomassie-stained SDS-PAGE gel. The structure of the VLPs was confirmed by electron microscopy.

2.3. Capsomere production and purification

For production of capsomeres in bacteria, truncated L1 genes (ΔN10L1ΔC27 for HPV types 2 and 27; ΔN10L1ΔC26 for HPV type 57; ΔN10L1ΔC32 for HPV type 16) were cloned into a pGEX 4T-2 vector (Stratagene) and expressed in Escherichia coli Rosetta cells as described for HPV 16 ΔN10L1 in [33]. LB medium containing 1 mM ampicillin was inoculated and incubated overnight shaking at 37 ◦C. At an optical density (OD_{600 nm}) of 0.3–0.5, the bacteria cultures were cooled down for 5 min on ice. Subsequently, IPTG was added to a final concentration of 0.2 mM to induce protein expression. Bacteria were incubated shaking for 16 to 18 h at room temperature (RT) and harvested by centrifugation. Bacteria pellets obtained from one litre culture were resuspended in 40 mL buffer L (50 mM Tris, pH 8.2, 0.2 M NaCl, 1 mM EDTA, 5 mM DTT) supplemented with a complete protease inhibitor cocktail tablet (Roche) and lyzed using a high-pressure homogenizer (Avestin). ATP and MgCl₂ were added to final concentrations of 2 and 5 mM, respectively. After 1 h incubation at RT, urea was added to a final concentration of 3.5 M. Following 2 h incubation at RT, lysates were dialysed for 16 h against buffer L at 4 ◦C with three buffer exchanges. Lysates were centrifuged at 51,200 × g for 30 min and loaded onto 1 mL GSTrap columns (GE Healthcare) over night at 4 ◦C. Columns were washed with 10 to 20 mL of buffer L and 40 U of thrombin protease (GE Healthcare) in 1 ml buffer L was added. The cleavage reaction was performed at RT for 12 h. Cleaved L1 proteins were dialyzed against modified buffer L (50 mM Tris, pH 8.2, 0.5 M NaCl, 1 mM EDTA, 5 mM DTT, 0.01% Tween80) for 2h at RT. A size exclusion chromatography was carried out using a Superdex200 column (GE Healthcare). Residual lipopolysaccharide (LPS) was removed by treatment with Triton X-114 as described in [34]. Successful LPS removal was verified using the limulus amoebocyte lysate (LAL)-based colorimetric assay QCL-1000 (Lonza) following the manufacturer’s protocol. The concentration of L1 protein was determined by Bradford assay and by comparing to a BSA standard on a Coomassie-stained SDS-PAGE gel. The structural integrity of the capsomeres was verified by electron microscopy.

2.4. SDS-PAGE and Western blot analysis

Protein samples were supplemented with loading buffer (35 mM Tris–HCl, pH 8.6, 60 mM DTT, 3.6% glycerol, 1% SDS) and boiled for 10 min at 95 ◦C. The denatured proteins were separated by SDS-PAGE with 12% polyacrylamide gels and either stained using colloidal Coomassie (Thermo Scientific) or transferred to a PVDF membrane (Millipore). For Western blot analysis, L1 proteins were detected using the MAb MD2H11 and enhanced chemiluminescence for HRP detection (AppliChem).

2.5. Electron microscopy

VLPs (100 ng) and capsomeres (150 ng) were applied onto carbon-coated grids and stained with 2% uranyl acetate. Grids were analysed using a transmission electron microscope CM200 FEG (FEI) operating at 200 kV. Pictures were taken at a 27,000-fold magnification using a 2k × 2k CCD camera.

2.6. Mouse immunizations

C57BL/6 mice (Charles River Wiga Breeding Laboratories) were maintained under pathogen-free conditions. All mice were immunized at an age of 8 to 12 weeks.

Mice were immunized three times in 2-week intervals subcutaneously with 10 µg of VLPs or capsomeres. Antigens were administered either without any additives or they were adsorbed to 500 µg of aluminium hydroxide (A8222; Sigma) and supplemented with 50 µg of monophosphoryl lipid A and 50 µg of synthetic trehalose dicorynomycolate in squalene and Tween 80 (Sigma). Prior to each immunization, blood samples were taken by puncture of the superficial temporal vein. One week after the final immunization, mice were sacrificed, spleens were isolated, and blood was collected by cardiac puncture.

2.7. GST-L1 multiplex serology assay

The Luminex-based multiplex serology assay to determine antibody titers in sera using GST-L1-tag fusion proteins as antigens of various PV types was previously described [35,36]. Briefly, GST-L1-tag fusion proteins were expressed via pGEX4T3 plasmids (GE Healthcare) in E. coli Rosetta (Merck) as in [37]. Bacterial lysates were cleared, loaded on glutathione-casein-coupled, distinctly fluorescence-labelled polystyrene beads (Luminex), and co-incubated with serially diluted pre-blocked mouse sera for 1h at RT. Bound antibodies were detected by biotinylated goat anti-mouse antibody and streptavidin-R-phycoerythrin (Molecular Probes). The individual bead sets were identified and their reporter fluorescence was quantified using an xMAP Luminex 100 analyser (Luminex). The background was determined using a GST-tag fusion protein as negative control antigen. Median fluorescence intensities were considered positive when exceeding the cut-off defined as autofluorescence of the individual bead sets (fluorescence upon incubation with culture supernatant of 293T cells (Invitrogen)) plus twice the background determined for each mouse serum in the respective dilution (fluorescence for GST-tag as antigen).

2.8. VLP capture ELISA

Polyclonal rabbit anti-HPV 57 VLP serum was blocked with wild type (WT) Autographa californica nuclear polyhedrosis virus (AcNPV)-infected insect cell lysate, protein G-purified (GE Healthcare), and used to coat immunosorbent 96-well plates (Nunc) over night at 4 ◦C. Plates were blocked with PBS plus 0.05% Tween 20 (PBS-T) and 300 ng VLPs in 5% skim milk were applied per well. After 1 h incubation at 37 ◦C, plates were washed four times with PBS-T and mouse sere were applied serially diluted with 5% skim milk in duplicates. Following four washes with PBS-T, horseradish peroxidase-conjugated goat anti-mouse antibody (Dianova) was applied for 1h at 37 ◦C. Assays were developed with and 2,2’-azino-bis[3-ethylbenzthiazoline-6-sulphonic acid] (ABTS, Sigma) substrate and absorbance at 405 nm was measured using an ELISA reader. Signals were defined positive when absorbance values exceeded values obtained with pre-immunization sera of the same mice at the same dilution plus three times standard deviation.

2.9. Preparation of pseudovirions and neutralization assays

For the production of pseudovirions, 293TT cells [38] were co-transfected with expression constructs carrying codon-modified L1 and L2 genes along with a reporter plasmid encoding Gaussia luciferase [39] using PolyFect™ transfection reagent (Qiagen). HPV type 16 pseudovirions were produced using the bicistronic L1/L2 expression plasmid, p16sheLL as described in [40]. The nucleotide sequences for L1 and L2 from HPV types 27 and 57 were codon-modified employing the as-different-as-possible (ADAP) method [38] and synthesized (HPV2 ORFs: Blue Heron, GenBank accession numbers FJ976666 and FJ976667; HPV27 and HPV57 ORFs: GenScript, GenBank accession numbers FJ976662–FJ976665). The codon-modified genes were cloned into the vector described for HPV type 2 as pRaLw at http://home.ccr.cancer.gov/Lco/plasmids.asp. Seventy-two hours post transfection, cells were harvested and lysed with Brij58 (0.35% final concentration, Sigma), benzonase was added (250 U, Merck), and capsids were matured overnight at 37 ◦C. Subsequently, NaCl was added to a final concentration of 0.85 mol/L.

For the analysis of their neutralizing potentials, mouse sera were pre-incubated in serial dilutions with pseudovirions (HPV type 2, 27, and 57 pseudovirions 1:2000 diluted; HPV 16 pseudovirions 1:50,000 diluted) for 15 min at RT, and applied to 6 × 10⁴ 293TT cells per 96-well in Dulbecco’s modified Eagle’s medium (Sigma) supplemented with 10% fetal calf serum (Gibco) in duplicates. Five days post inoculation, the enzymatic reporter reaction was induced using Gaussia luciferase substrate (New England Biolabs) and reporter activity was quantified with a luminescence reader (VICTOR³ 1420 multilabel plate reader, PerkinElmer). The neutralization at a given mouse serum dilution in percent was calculated by the following equation: (neutralization at dilution X [%] = 100–100* mean luminescence upon infection with pre-incubated pseudovirions at dilution X/mean luminescence upon infection with untreated pseudovirions).

2.10. IFN-γ ELISPOT assay

For ex vivo enzyme-linked immunospot (ELISPOT) analysis, 96well-MultiScreen-HTS™ ELISPOT plates were activated with 70% ethanol, washed, and coated over night at 4 ◦C with 600 ng of anti-mouse IFN-γ antibody (BD Pharmingen) in PBS per well. Plates were washed and blocked with RPMI medium (Sigma) supplemented with 10% FCS (Gibco) and 1% penicillin and streptomycin. Splenocytes were seeded in triplicates as twofold dilution series from 1 × 10⁶ cells to 1.25 × 10⁵ cells per well and T cell stimulation was performed with 500 ng of VLPs/well. As positive control, cells were stimulated unspecifically with 1 µg Pokeweed mitogen (Sigma) per well, as negative control, only RPMI medium was added. After an 18 h-incubation at 37 ◦C, cells were removed and 0.1 µg per well of biotinylated rat anti-mouse IFNγ antibody (BD Pharmingen) was applied for 2h at RT. Subsequently, 100 µl per well of streptavidin-alkaline phosphatase (BD Pharmingen) diluted 1:1000 in PBS were added and incubated for 30 min at RT. Finally, 5-bromo-4-chloro-3-indolylphosphate (BCIP)/Nitro Blue Tetrozolium substrate (Sigma) was added and reaction was stopped with water. Spots were counted using an ELISPOT reader controlled by an AID ELISPOT software (Zeiss).

2.11. Statistical analyses

Data sets were tested for significance using the Wilcoxon rank sum test. For the analysis of antibody titer development over the course of an immunization, the area under the curve was quantified and tested for significance using the Wilcoxon rank sum test. p-Values below 0.05 were considered statistically significant.

3. Results

3.1. Production of the L1 antigens

We employed a modified baculovirus expression system to produce VLPs in insect cells. We have previously shown that it allows the generation of large amounts of homogeneous VLPs [26]. For the production of capsomeres, we chose to express the L1 genes in bacteria as GST-fusion proteins and purified them using glutathione sepharose chromatography followed by cleavage of the GST-tag as described previously [33]. Expression of full length L1 in E. coli proved unsuitable as the resulting proteins formed mostly insoluble aggregates when the GST-tag was removed. In contrast, N- and C-terminal truncations (ΔN10L1ΔC27 for HPV types 2 and 27; ΔN10L1ΔC26 for HPV type 57) led to increased protein solubility and, hence improved antigen recovery (at least 10-fold: 0.3–2.5 mg protein/liter bacteria culture). As reference antigen, we expressed the accordingly truncated L1 gene from HPV type 16 (ΔN10L1ΔC32). Structural characterization of the L1 proteins produced in E. coli by sedimentation analysis and transmission electron microscopy demonstrates that homogeneous capsomere preparations can be generated with this expression system (Fig. 1).

Fig. 1.

Characterization of capsomeres produced in bacteria. HPV type 2, 27, 57, and 16 capsomeres were produced in bacteria. They were characterized by velocity gradient sedimentation and Western blotting (A) and transmission electron microscopy (B). For the sedimentation analysis, HPV 16 VLPs and catalase were used as calibration markers.

All L1 proteins were of the expected size (about 55 kDa and 52 kDa for full length and truncated L1 proteins, respectively) and of high purity (at least 90% of total protein amount; Fig. 2). HPV 2, 27, and 57 VLPs produced in insect cells contained not only full length L1 proteins but also two smaller L1 populations. Further analysis by Q-TOF mass spectrometry revealed that the additional L1 species are C-terminally truncated by about 30 and 100 amino acids, respectively (data not shown), possibly reflecting premature stops during translation of the corresponding transcripts or degradation products.

Fig. 2.

Purity of VLP and capsomere preparations. Equal amounts of the purified L1 proteins were loaded on SDS-PAGE gels and analyzed by Coomassie staining (top) or immunoblotting using the L1-specific monoclonal antibody MD2H11 (bottom).

3.2. Immunization with VLPs and capsomeres induces high antibody titers

To investigate the immunogenicity of capsomeres and VLPs from HPV types 2, 27, and 57, C57BL/6 mice (5 per group) were immunized three times at biweekly intervals subcutaneously with 10 µg of the different antigens. As reference, HPV 16 capsomeres and VLPs were injected. To exclude that LPS originating from bacteria compromises the immune response to capsomeres, all capsomere preparations were freed from LPS contaminations as previously described [34] and their L1 content per immunization was adjusted. Immunogens were administered with or without a combination of aluminum hydroxide and an adjuvant system, resembling the adjuvants used in the Cervarix® vaccine (see Section 2). Negative control mice were injected with PBS. The development of antibody titers specific to L1 of the immunogen HPV type was monitored by VLP capture ELISA. Upon immunization with capsomeres or VLPs from HPV types 2, 27, 57, and 16, L1-specific antibodies developed in a similar pattern (Fig. 3a–d). In the presence of the adjuvant, antibodies developed faster and to higher titers in case of HPV type 2 (p ≤ 0.016 for VLPs and capsomeres), 57 (p ≤ 0.008 for VLPs and capsomeres), HPV 27 VLPs (p ≤ 0.016) and HPV 16 capsomeres (p ≤ 0.008). Typically, antibodies developed faster and to higher titers after immunization with VLPs than with capsomeres of the same HPV type (for all 4 types without adjuvant: p ≤ 0.008 for either type; for HPV types 27 and 57 with adjuvant: p ≤ 0.03 and p ≤ 0.008, respectively).

Fig. 3.

Development of L1-specific antibody titers after immunization with antigens from HPV types 2, 27, 57, and 16. Using VLP capture ELISA, L1-specific antibody titers for HPV types 2, 27, 57, and 16 (A–D) were determined over the course of the immunizations with VLPs (filled circles) and capsomeres (empty circles) with or without adjuvant (dashed and continuous lines, respectively) from the according immunogen type.

These results indicate that capsomeres and VLPs from HPV types 2, 27, and 57 trigger the development of L1-specific antibodies to a similar extent as capsomeres or VLPs from HPV type 16.

3.3. Antibodies produced upon immunization with VLPs and capsomeres are cross-reactive but type-restricted in their neutralizing capacity

HPV types 2, 27, and 57 are very closely related members of the same species [41]. Therefore, we aimed to examine whether immunization with an L1-based antigen triggers strong cross-reactivity to L1 from the other types of the same species. The bead-based multiplex assay described in [35] allows for high-throughput analysis of serum reactivity with GST-L1 proteins from multiple PV types. We employed this assay to determine the reactivity of the mouse sera with GST-L1 from 18 PV types representative of the genera alpha (HPV types 2, 3, 6, 7, 10, 11, 16, 18, 27, 32, 57, 77), beta (HPV types 5, 38), gamma (HPV type 4), delta (BPV type 1), mu (HPV type 1), and nu (HPV type 41). Median antibody titers determined for sera from mice immunized with capsomeres were higher than for sera from VLP-immunized mice (Table 1). Addition of the adjuvant system to capsomeres and VLPs increased the maximum median antibody titers from 1.8 × 10⁷ to 4.8 × 10⁸. Extent of cross-reactivity of the mouse sera mostly corresponded with phylogenic relatedness of the antigens tested (Supplementary Tables S1–S4). For example, cross-reactivity with GST-L1 from any of the Alphapapillomavirus species 4 types was much stronger when mice had been immunized with L1 from either of the closely related HPV types 2, 27, or 57 than upon immunization with L1 from HPV type 16 (member of Alphapapillomavirus species 9; see upper panels of Fig. 4 for immunization with adjuvant).

Table 1.

Humoral immune responses after administration of capsomeres and VLPs of HPV types 2, 27, 57, and 16 with and without adjuvant.

Injected antigen	Adj	VLP capture ELISA		GST-L1 multiplex assay		Neutralization assay
		Median L1-specific Ab titer	IQR	Median L1-specific Ab titer	IQR	Median neutralization IC50 value	IQR
HPV2 Caps	+	109,350	36,450–328,050	159,432,300	159,432,300–159,432,300	3,000	3,000–5,000
	−	12,150	12,150–12,150	17,714,700	17,714,700–53,144,100	2,000	1,500–2,500
HPV2 VLPs	+	984,150	328,050–984,150	159,432,300	159,432,300–159,432,300	6,000	5,000–14,000
	−	102,400	51,200–102,400	656,100	656,100–1,968,300	7,000	6,000–8,000
HPV27 Caps	+	12,150	12,150–328,050	5,904,900	5,904,900–5,904,900	4,000	3,000–7,000
	−	36,450	12,150–102,600	17,714,700	1,968,300–17,714,700	2,000	1,500–3,000
HPV27 VLPs	+	328,050	328,050–984,150	17,714,700	5,904,900–17,714,700	15,000	10,000–14,000
	−	102,400	102,400–204,800	1,968,300	1,968,300–17,714,700	15,000	10,000–20,000
HPV57 Caps	+	2,952,450	984,150–2,952,450	53,144,100	53,144,100–1,434,890,700	38,000	30,000–90,000
	−	328,050	109,350–984,150	1,968,300	1,968,300–17,714,700	4,500	4,500–12,500
HPV57 VLPs	+	8,857,350	8,857,350–8,857,350	17,714,700	17,714,700–53,144,100	460,000	450,000–850,000
	−	204,800	102,400–204,800	1,968,300	1,968,300–5,904,900	25,000	15,000–30,000
HPV16 Caps	+	109,350	109,350–984,150	478,296,900	478,296,900–1,434,890,700	50,000	50,000–820,000
	−	12,150	12,150–12,150	17,714,700	17,714,700–17,714,700	20,000	12,000–22,000
HPV16 VLPs	+	109,350	109,350–328,050	17,714,700	17,714,700–159,432,300	200,000	150,000–900,000
	−	204,800	204,800–204,800	5,904,900	5,904,900–17,714,700	150,000	100,000–200,000

Ab: antibody Adj: adjuvant Caps: capsomeres IQR: Interquartile range VLPs: virus-like particles

Fig. 4.

Immunization with HPV 2, 27, and 57 VLPs or capsomeres induces a cross-reactive humoral immune response that is type-restricted in its neutralizing potential. Sera from mice immunized with VLPs or capsomeres together with adjuvant (A and B, respectively) were analyzed for reactivity with GST-L1 from HPV types 2, 27, 57, and 16 by multiplex serology (upper panels), for reactivity with VLPs from HPV types 2, 27, 57, and 16 by VLP capture ELISA (middle panels), and for neutralization of pseudovirions from HPV types 2, 27, 57, and 16 (lower panels). Data obtained upon immunization without adjuvant are not shown.

In comparison to GST-L1 fusion proteins, VLPs are generally thought to present a higher density of conformational epitopes. Therefore, reactivity with VLPs from HPV types 2, 27, 57, and 16 was analyzed by VLP capture ELISA. In contrast to the results obtained by GST-L1-based multiplex serology, slightly stronger reactivity was observed upon immunization with VLPs than with capsomeres (Table 1). Employment of the adjuvant system led to cross-reactivity with capsids from all types and sera reacted to heterologous HPV types at titers of up to 10⁵ (Fig. 4, middle panels).

To analyze whether antibodies upon immunization with VLPs or capsomeres of HPV types 2, 27, and 57 are neutralizing, we developed luciferase reporter pseudovirions for these three HPV types according [38]. The pseudovirions were treated with the serially diluted mouse sera before applying to 293TT cells. As a measure for their neutralizing capacities, the median IC₅₀ value for each mouse group was determined with respect to neutralization of HPV 2, 27, 57, and 16 (Fig. 4 lower panels).

All sera from mice immunized with VLPs or capsomeres efficiently neutralized pseudovirions of the cognate HPV type with median IC₅₀ values ranging from 2000 to 460,000 (Table 1). Application of adjuvant entailed increased neutralizing capacities of the respective mouse sera. The differences in IC₅₀ values reached significance for VLPs and capsomeres of HPV 57 (p ≤ 0.008 for both) and capsomeres of HPV type 16 (p ≤ 0.008). Consistent with the determined L1-specific antibody titers, immunization with VLPs led to significantly elevated neutralizing abilities compared to immunization with capsomeres of the same HPV type (without adjuvant: p ≤ 0.039 for HPV type 2, p ≤ 0.008 for HPV type 27, p ≤ 0.031 for HPV type 57, p ≤ 0.008 for HPV type 16; with adjuvant: p 0.039 for HPV type 27, p 0.019 for HPV type 57). However, despite the relatively strong cross-reactivity detected by VLP capture ELISA and by GST-L1-based multiplex serology assay, the mouse sera showed only marginal cross-neutralization of the heterologous HPV types. A comparison of the results obtained using GST-L1-based Luminex multiplex serology, VLP capture ELISA, and neutralization assays for sera from mice immunized with VLPs or capsomeres together with adjuvant is shown in Fig. 4.

The neutralizing capacities of the mouse sera and the L1-specific antibody titers were compared by correlation coefficients (R²). Neutralizing IC₅₀ values for all 4 analyzed HPV types are in better accordance with the antibody titers measured by the VLP capture ELISA ($R_{\text{HPV2}}^2$= 0.723; $R_{\text{HPV27}}^2$= 0.814; $R_{\text{HPV57}}^2$= 0.809; $R_{\text{HPV16}}^2$= 0.816) than by GST-L1-based multiplex serology ($R_{\text{HPV2}}^2$= 0.521; $R_{\text{HPV27}}^2$= 0.445; $R_{\text{HPV57}}^2$= 0.039; $R_{\text{HPV16}}^2$= 0.301).

3.4. Immunization with VLPs and capsomeres activates the cellular immune response

To assess the cellular immune response to capsomeres and VLPs of HPV types 2, 27, 57, and 16, we isolated spleen cells 7 days after the final immunization and performed IFNγ-ELISPOT analyses following their ex vivo stimulation with VLPs from HPV types 2, 27, 57, and 16 (Fig. 5a–d).

Fig. 5.

Cellular IFNγ response to antigens from HPV types 2, 27, 57, and 16. Mice were immunized with VLPs (dark grey bars) or capsomeres (light grey bars) with or without adjuvant (striped or plain) and, one week after the final immunization, splenocytes were stimulated ex vivo with VLPs from HPV types 2, 27, 57, and 16 (A–D) and analyzed for specific T lymphocytes by IFN-γ ELISPOT.

The cellular IFNγ responses in mice immunized with VLPs or capsomeres from any of the species 4 alpha HPV types were at least as strong as by HPV type 16-immunized mice. Furthermore, pronounced cross-reactivity among HPV types 2, 27, and 57 was observed. The amount of IFNγ-producing cells was enhanced significantly by the addition of adjuvant (capsomere immunizations: p ≤ 0.008 for HPV 2, p ≤ 0.008 for HPV 27, p ≤ 0.008 for HPV 57, p ≤ 0.008 for HPV 16; VLP immunizations: p ≤ 0.008 for HPV 2, p ≤ 0.008 for HPV 27).

Different immune cell populations might contribute to the assayed IFNγ response. We examined the extent to which CD4⁺ and CD8⁺ T-cells respond to VLP stimulation by depletion of either cell population from the spleen cell pools obtained from immunized mice before carrying out an IFNγ-ELISPOT assay upon ex vivo stimulation. We found that the majority of the observed IFNγ response was attributable to CD4⁺ T-cells (data not shown).

In summary, these results indicate that a profound cellular immune response is triggered by vaccination with VLPs and capsomeres from HPV types 2, 27, and 57. Mostly CD4⁺ cells respond to ex vivo stimulation with VLPs and the extent of this response is enhanced significantly by the employment of adjuvant.

3.5. A trivalent HPV type 2, 27, and 57 vaccine triggers an immune response protecting against all three types

Even in the presence of adjuvant, immunization with VLPs or capsomeres of either HPV type 2, 27, or 57 did not induce high titers of antibodies that cross-neutralize the heterologous types. Therefore we aimed to investigate whether a trivalent vaccine induces a response directed against all three antigens. Since in the initial analyses we had observed slightly stronger antibody responses to VLPs compared to capsomeres (Figs. 3 and 4, Table 1), we chose to immunize with different combinations of VLPs.

In groups of five, C57BL/6 mice were immunized three times at biweekly intervals subcutaneously either with 10 µg of VLPs from HPV type 2, 27, or 57 alone, or with all 4 possible combinations of VLPs from two or all three types (10 µg of each). Negative control mice were injected with PBS. Blood samples were drawn prior to each immunization and one week after the final immunization and analyzed by VLP capture ELISA. The neutralizing activity of the mouse sera was examined only after the final bleed. One week after the final immunization, splenocytes were isolated for IFNγ-ELISPOT analysis.

The L1-specific antibody titers developed as expected: reactivity to the immunogen HPV types developed in a similar pattern when VLPs had been administered alone or in mixtures (data not shown).

Analysis of sera from the final bleed revealed that median antibody titers were slightly higher when VLPs of single types had been applied as compared to this particular type given in mixtures (Fig. 6, Table 2). In case of HPV 27 and 57, sera from mice that had received the trivalent vaccine showed up to three-fold lower median antibody titers compared to the bivalent mixtures. However, only in a few cases did the administration of pooled VLPs entail a significantly reduced antibody titer when compared with the single-type immunization (HPV2+27 versus HPV2: p < 0.024; HPV2+27+57 versus HPV27: p < 0.032; HPV2+27+57 versus HPV57: p < 0.008).

Fig. 6.

Humoral immune response to mixed VLPs from HPV types 2, 27, and 57. Mice were immunized with VLPs from HPV types 2, 27, and 57 either alone or in all possible combinations. One week after the final immunization, L1-specific antibody titers were determined with VLPs from types 2, 27, and 57 (A–C) by capture ELISA (black columns, left y-axis) and the capacities to neutralize pseudovirions from HPV types 2, 27, and 57 (A–C) were quantified (white columns, right y-axis).

Table 2.

Humoral immune responses after immunization with mixed VLPs of HPV types 2, 27, and 57.

Injected antigen	Assayed reactivity with	Median L1-specific Ab titer	IQR	Median neutralization IC50 value	IQR
HPV2 VLPs	HPV2	109,350	109,350–109,350	12,000	10,000–25,000
HPV27 VLPs		36,450	12,150–36,450	200	200–200
HPV57 VLPs		4,050	4,050–12,150	200	150–200
HPV2+27 VLPs		36,450	36,450–109,350	3,200	3,000–8,000
HPV2+57 VLPs		12,150	12,150–36,450	8,000	6,000–10,000
HPV27+57 VLPs		12,150	12,150–12,150	200	200–400
HPV2+27+57 VLPs		36,450	36,450–36,450	5,000	3,000–10,000
PBS		1,350	450–1,350	150	150–200
HPV2 VLPs	HPV27	12,150	12,150–12,150	200	200–300
HPV27 VLPs		109,350	109,350–328,050	10,000	9,000–13,500
HPV57 VLPs		12,150	1,350–12,150	150	100–200
HPV2+27 VLPs		328,050	109,350–328,050	10,000	3,500–10,000
HPV2+57 VLPs		12,150	12,150–36,450	200	170–300
HPV27+57 VLPs		109,350	36,450–109,350	12,000	12,000–14,000
HPV2+27+57 VLPs		36,450	36,450–36,450	6,000	3,500–9,000
PBS		450	450–1,350	200	200–200
HPV2 VLPs	HPV57	4,050	4,050–4,050	200	200–300
HPV27 VLPs		36,450	36,450–36,450	200	200–200
HPV57 VLPs		328,050	328,050–328,050	130,000	130,000–200,000
HPV2+27 VLPs		36,450	36,450–36,450	200	100–200
HPV2+57 VLPs		109,350	36,450–109,350	220,000	60,000–250,000
HPV27+57 VLPs		109,350	109,350–109,350	190,000	180,000–200,000
HPV2+27+57 VLPs		36,450	36,450–36,450	50,000	50,000–100,000
PBS		150	150–150	100	100–200

Abtiter: antibody titer IQR: Interquartile range VLPs: virus-like particles.

With respect to neutralizing capacities of the sera, immunization with pooled VLPs from HPV types 2, 27, and 57 led to the development strong immune responses against all types. Only statistically non-significant differences in the neutralizing potential were observed between mouse sera after single HPV type VLP immunizations and sera upon immunization with pooled VLPs from different HPV types (Fig. 6, Table 2). Hence, VLPs from HPV types 2, 27, and 57 do not mutually impair their ability to induce neutralizing antibodies.

The cellular immune response was most pronounced after stimulation of the spleen cells with VLPs of the homologous HPV type. When mice had received mixtures of VLPs similar responses were obtained against all types (data not shown).

In summary, these data show that immunization with mixed VLPs from HPV types 2, 27, and 57 triggers the generation of high titers of antibodies that efficiently neutralize all immunogen types, highlighting the promising prophylactic potential of the trivalent HPV vaccine against cutaneous papillomaviruses of species 4.

4. Discussion

The potential of L1-based particles from cutaneous species alpha HPV types as prophylactic vaccines has not been assessed so far. We report here that HPV types 2, 27, and 57 capsomeres and VLPs produced in bacteria and insect cells, respectively, elicit high titers of neutralizing antibodies following s.c. immunization even without adjuvant. Therefore, they are promising vaccine candidates for prevention of skin lesions induced by these HPV types.

For the production of capsomeres in E. coli we expressed N- and C-terminally truncated versions of L1 from HPV types 2, 27, and 57 as these constructs were found to be most soluble and, hence, most efficient to purify. In parallel, VLPs were produced upon expression of full-length L1 in insect cells. The capsomere and VLP preparations may differ in nature and quantity of residual contaminating components originating from the different producer cells. Moreover, capsomeres consisting of L1 deletion mutants may lack epitopes that contribute to an immune response when challenged by infectious virus particles composed of full-length L1. Therefore, we acknowledge that a stringent comparison between the immune responses induced by immunization with capsomeres and VLPs is not possible. However, we only aimed to assess the general applicability of bacteria as alternative expression system for capsomeres to be used for immunizations. We found that such capsomeres elicit an immune response comparable in quality to the one triggered by VLPs. These results are consistent with previous reports that capsomeres from HPV types 11 and 16 induce neutralizing antibodies [22,24,42]. In our study, sera from mice immunized with capsomeres showed a 2- to 13-fold decrease in neutralizing antibodies to the respective immunogen HPV type when compared to sera from VLP-immunized mice. These results differ from data by Thönes and colleagues, who reported that at least 20–40 times more capsomeres are required to achieve a comparable antibody response [43]. However, unlike Thönes and colleagues, we used L1 deletion mutants to produce capsomeres in E. coli rather than to express in insect cells L1 constructs in which two conserved cysteine residues involved in intercapsomeric disulfide bond formation had been replaced with alanine residues. We showed recently that an increased intrinsic ability of capsomeres to assemble into stable larger particles correlates with higher immunogenicity so that other capsomere constructs such as N-terminal L1 deletion mutants induce an immune response resembling the one after VLP immunization [44]. This concept is supported by the results of the present study.

As expected, reactivity of sera from mice immunized with capsomeres or VLPs from HPV types 2, 27, 57, or 16 with GST-L1 fusion proteins from 18 different PV types reflects their individual phylogenetic relatedness. Nevertheless, our data suggest that results obtained with neutralization assays are more authentic to reflect the biologically relevant humoral immune response. This finding is in agreement with a report by Smith et al., who showed that neutralization assays allow to distinguish type-specific as well as neutralizing antibodies in the sera of nonhuman primates vaccinated with VLPs from HPV types 6, 11, 16, and 18, whereas an ELISA did not [45].

In this study neutralization IC₅₀ values correlated more closely with antibody titers determined by VLP capture ELISA than by GST-L1-based multiplex serology. Sera from capsomere-immunized mice were found to react more strongly with GST-L1 fusion proteins than those from VLP-immunized mice whereas with VLP capture ELISA highest antibody titers were observed for mouse sera after VLP immunization. Rizk et al. have demonstrated that the GST-L1 fusion proteins utilized in the multiplex assay display conformational as well as linear epitopes [46]. The elevated reactivity of sera from capsomere-immunized mice with GST-L1 fusion proteins may therefore be due to the presence of relatively high amounts of denatured L1 protein in the capsomere preparations. However, as capsomeres are thought to be more stable than VLPs, this explanation seems rather unlikely. Instead, we favour the hypothesis that both, capsomeres and GST-L1 fusion proteins, display a subset of epitopes that are inaccessible on VLPs as they reside in the interior of the capsid structure. The superior correlation of the results obtained by VLP capture ELISA with neutralization titers is consistent with this conclusion because the neutralizing antibodies bind to surface-exposed epitopes, which are also presented by VLPs [46–53].

In case of the commercially available vaccines Cervarix® and Gardasil® VLPs from the mucosal HPV types 16 and 18 proved to elicit a certain degree of cross-protection against non-vaccine HPV types in clinical studies [16,54,55] but the immunologic correlate of this observation is not yet published. For the cutaneous species beta HPV types 5 and 8, vaccination with VLPs emulsified in Freund’s adjuvant caused cross-neutralizing antibodies in rabbits [56]. In our study, no substantial cross-neutralization of the closely related non-vaccine HPV types was found upon immunization with VLPs from HPV types 2, 27, or 57, not even when they were administered together with adjuvant. An analysis of the B-cell epitopes on VLPs using a panel of 94 monoclonal antibodies generated upon immunization with capsids from HPV types 2, 27, and 57 has revealed that the type-specificity of the neutralizing antibodies attributes to the nature of their recognized epitopes that reside mostly in the hypervariable surface loop regions, which consist of type-specific amino acid sequences [57]. However, the cause of the different results is not clear. We used luciferase reporter pseudovirions to determine neutralization rather than haemagglutination inhibition assays or analyses of the clinical trial outcomes. Inconsistencies may result from different specificity of the assays, modifications of the antigen as consequence of the production processes, adjuvants used in the studies and/or the fact that different species (man, mice or rabbits) had been immunized. Ultimately, we cannot exclude that the relatively low cross-neutralization observed in the present study reflects a biological peculiarity of the analyzed HPV types.

For the HPV VLP vaccine formulations, adjuvants have been shown to boost the immune response to up to 100-fold higher antibody titers [58,59]. However, the amount of VLPs in these vaccines has been well titrated and adjusted to optimized scales. In contrast, immunizations of excessive amounts of VLPs have been reported to elicit a maximum plateau immune response [60]. For the present study, we chose to immunize with rather large amounts of antigens to gather preliminary insights into the immunogenicity of the analyzed HPV particles and their potential to induce a cross-reactive immune response. Hence, the observed antibody responses were already exhausted when antigens had been immunized without adjuvant, so that complementation with adjuvant did not always lead to significantly increased immune responses. However, in most cases, highest antibody titers and strongest potential to neutralize the respective immunogen HPV type were achieved upon immunization of VLPs together with adjuvant.

Previous studies have shown that immunization with HPV 16 capsomeres or VLPs triggers a substantial T-cell response [42,43,61]. The cellular immune response upon immunization with capsomeres or VLPs from cutaneous species alpha HPV types observed here is in accordance with these reports. We determined that the cellular immune response is attributed mostly to CD4⁺ cells. Even though the IFN-γ ELISPOT only allows for the detection of helper T cells type 1, the CD4⁺ cells responsive to ex vivo stimulation with VLPs are likely to consist of both, helper T cells of type 1 and 2, as cytokine and chemokine profiling of human peripheral blood mononuclear cells following immunization with HPV type 16 VLPs suggests that it activates both arms of the adaptive immune system [62,63].

Immune interference has been described for multiple vaccine formulations [64,65]. It has been defined as the reduction in immunogenicity of one or several vaccine antigens when administered jointly due to an immunodominant response to a single antigen. We have shown that administration of pooled VLPs from HPV types 2, 27, and 57 entailed only slightly decreased immune responses when compared to immunization with one HPV VLP type only. Thus, at least in mice, no major immune interference between these antigens occurs. Our results are consistent with data by Garland et al., who reported that HPV type 16 VLPs do not exert immune interference against the other vaccine HPV VLP types when combined into the quadrivalent vaccine Gardasil® [66].

In summary, our data provide the first analysis of the potential of capsomeres and VLPs from cutaneous species alpha HPVs as prophylactic vaccines. We show that a trivalent vaccine containing capsids from HPV types 2, 27, and 57 elicits an immune response with broad neutralizing potential, suggesting that it represents a promising strategy to prevent infections by cutaneous HPV types.