Vaccine 32 (2014) 2610–2617

Contents lists available at ScienceDirect

Vaccine journal homepage: www.elsevier.com/locate/vaccine

A three component mix of thioredoxin-L2 antigens elicits broadly neutralizing responses against oncogenic human papillomaviruses Hanna Seitz a,1 , Elena Canali b , Lis Ribeiro-Müller a , Anikó Pàlfi a , Angelo Bolchi b , Massimo Tommasino c , Simone Ottonello b , Martin Müller a,∗ a

German Cancer Research Center, Heidelberg, Germany Department of Life Sciences, Biochemistry and Molecular Biology Unit, University of Parma, Parma, Italy c International Agency for Research on Cancer, Lyon, France b

a r t i c l e

i n f o

Article history: Received 30 October 2013 Received in revised form 30 January 2014 Accepted 7 March 2014 Available online 21 March 2014 Keywords: Papillomavirus minor capsid protein L2 Cross-protection Thioredoxin-L2 In vitro neutralization In vivo protection

a b s t r a c t Current human papillomavirus (HPV) vaccines based on major capsid protein L1 virus-like particles (VLP) provide potent type-specific protection against vaccine-type viruses (mainly HPV16 and 18), but crossprotect against only a small subset of the approximately 15 oncogenic HPV types. It is estimated that L1-VLP vaccines, which require a fairly complex production system and are still quite costly, fail to cover 20–30% of HPV cervical cancers worldwide, especially in low-resource countries. Alternative antigens relying on the N-terminal region of minor capsid protein L2 are intrinsically less immunogenic but capable of eliciting broadly neutralizing responses. We previously demonstrated the enhanced immunogenicity and cross-neutralization potential of an easily produced recombinant L2 antigen bearing the HPV16 L2(20–38) peptide epitope internally fused to bacterial thioredoxin (Trx). However, antibodies induced by Trx-HPV16 L2(20–38) failed to cross-neutralize notable high-risk HPV types such as HPV31. In the present work, the Trx-L2 design was modified to include L2 sequence information from the highly divergent HPV31 and HPV51 types in addition to HPV16, with the aim of extending cross-neutralization. Multivalent antigens comprising L2(20–38) peptides from all three HPV types on a single Trx scaffold molecule were compared to a mixture of the three type-specific monovalent Trx-L2 antigens. While multivalent antigens as well as the mixed antigens elicited similar anti-HPV16 neutralization titers, cross-reactive responses against HPV31 and HPV51 were of higher magnitude and more robust for the latter formulation. A mixture of monovalent Trx-L2 antigens thus represents a candidate lead for the development of a broadly crossprotective, low-cost second-generation anti-HPV vaccine. © 2014 Elsevier Ltd. All rights reserved.

1. Introduction Approximately 16% of cancer cases worldwide are attributable to infections and human papillomaviruses (HPV), along with hepatitis B and C virus are the main viral culprits [1]. Infection by a subset of high-risk HPVs is linked to approximately 50% of vulvar, vaginal and penile cancers, ∼85% of anal carcinomas, ∼10% of laryngeal and ∼20% of oropharyngeal cancers and, most notably, ∼99% of cervical cancers [2,3]. After breast and colorectal cancer, cervical cancer is the third most frequent type of cancer in women worldwide [4].

∗ Corresponding author. Tel.: +49 6221 424628. E-mail address: [email protected] (M. Müller). 1 Present address: Laboratory of Cellular Oncology, NCI, 37 Convent Drive, Bethesda, MD 20814, USA. http://dx.doi.org/10.1016/j.vaccine.2014.03.033 0264-410X/© 2014 Elsevier Ltd. All rights reserved.

Two prophylactic, L1 virus-like particle (VLP) HPV vaccines, Gardasil® (Merck) and Cervarix® (GlaxoSmithKline), are available and prevent infection by HPV16 and HPV18, which together are responsible for approximately 70% of the cervical cancer burden worldwide. In addition to HPV16 and HPV18 VLPs, Gardasil® contains L1 VLPs from HPV6 and HPV11, two types responsible for benign genital warts. Both VLP vaccines are safe and immunogenic with neutralizing immune responses detectable 8 years after vaccination [5]. Also, both vaccines elicit nearly 100% effective protection against the parent HPV16 and HPV18 types and less consistent protection against some non-vaccine types including HPV31, HPV33 and HPV45 [6,7]. Cost, in addition to the limited set of targeted highrisk HPV types and the lack of comprehensive cross-protection, is a major drawback of the current VLP vaccines. In 2012, the cost for a three-dose regimen of either VLP vaccine was approximately 390 U.S. dollars [8]. Therefore, second-generation HPV vaccines should strive to be more cross-protective and cost-effective.

H. Seitz et al. / Vaccine 32 (2014) 2610–2617

In contrast to L1, immunization with polypeptides derived from the N-terminal region of the minor capsid protein L2 elicits broadly cross-neutralizing antibody responses [9–13]. However, L2 is poorly immunogenic because conserved regions eliciting crossprotective responses are made of linear peptides and unlike L1 it cannot self-assemble into VLPs. Various strategies have been employed to increase the immunogenicity of L2 neutralization epitopes, most notably amino acids (aa) 17–36 [14]. Most of these rely on presenting neutralization epitopes as repetitive, high molecular weight entities. For example, cross-protective responses have been elicited by virus-like display of L2 epitopes on bacteriophage PP7 and MS2 VLPs [15,16] or on HPV16 L1 VLPs [17,18]. Another potent, non VLP-display approach is based on bacterially produced, concatenated multi-type L2 fusion proteins [19,20]. We have successfully adopted the small, non-toxic and highly soluble Escherichia coli thioredoxin (Trx) protein (109 aa) as a vaccine scaffold to display various L2 multipeptides, including the cross-neutralization aa 20–38 L2 epitope [12]. Insertion of L2 ablates the native function of thioredoxin while presumably allowing proper surface exposure of L2. Monotypic Trx-HPV16 L2(20–38)3 antigens elicited neutralizing HPV16 responses with additional cross-neutralization of HPV18, HPV45 and HPV58 pseudovirions [12]. However, poor neutralization of other high-risk HPV types, such as HPV31, was observed with polyclonal antibodies and a cross-neutralizing monoclonal antibody (K18L220–38 ) [12,21]. This sub-optimal immune performance is due to type-specific primary sequence differences within L2. Knowing that sequence differences can mediate escape of cross-neutralizing antibodies, we examined the entire set of mucosal high-risk HPV L2 sequences encompassing the cross-neutralizing (aa 20–38) epitope. Ultimately, HPV16, HPV31 and HPV51 L2(20–38) peptides were chosen and tested as epitopes for the construction of a prototype multivalent anti-HPV vaccine. Not only does inclusion of HPV51, in addition to HPV16 and HPV31, cover most of the primary sequence differences within the cross-neutralizing epitope, but this type belongs to a divergent species, alpha-5, while HPV16 and HPV31 belong to alpha-9. Here we show that immunization of mice with a mixture of monovalent Trx-HPV16, Trx-HPV31 and Trx-HPV51 L2 antigens provides more comprehensive cross-neutralization than (a) immunization with type-specific monovalent antigens or (b) multivalent Trx-L2 antigens bearing the same HPV16, HPV31 and HPV51 L2(20–38) epitopes on a single Trx scaffold protein. 2. Materials and methods 2.1. Construction of monovalent and multivalent Trx-L2 antigens Monovalent Trx-HPV 16 L2(20–38)3 was generated as previously described [12]. The monovalent Trx-HPV 31 L2(20–38)3 and Trx-HPV 51 L2(20–38)3 constructs were produced with the same procedure. DNA encoding multivalent 1 and multivalent 2 L2 sequences was synthesized by Mr Gene (Regensburg, Germany; now Mr Gene/GeneArt/Life Technologies). 2.2. Expression and purification of Trx-L2 proteins The monovalent and multivalent Trx-L2 proteins were expressed and purified as recently described [22]. The composition of purified Trx-L2 was confirmed by immunoblotting with the use of anti-Trx (K12Trx) and anti-L2(K4L220–38 and K18L220–38 ) antibodies [21]. 2.3. Mouse immunization Six- to eight-weeks-old female Balb/c mice were obtained from Charles River (Sulzfeld, Germany) and kept in the animal facility

2611

of the German Cancer Research Center under specific-pathogenfree conditions. Animals were immunized subcutaneously (27G1/2 needle, Becton Dickinson, USA) four times at biweekly intervals with 25–50 ␮g of total Trx-L2 antigen adjuvanted, initially, with 50% v/v Montanide ISA720 (Seppic, France) and subsequently with 25% v/v Montanide ISA720 + 25 ␮g PolyI:C (Enzo Life Sciences, Lörrach, Germany) for increased immunogenicity. For example, a group receiving a single monovalent or multivalent Trx-L2 antigen received 25 ␮g or 50 ␮g (depending on the experiment) while an animal receiving the Trx-L2 mix received either 8.3 ␮g (25 ␮g/3) or 16.7 ␮g (50 ␮g/3) of each of the monovalent Trx-L2 antigens. Eight weeks after the fourth immunization, blood samples were collected by cardiac puncture and the resulting sera were used for in vitro and in vivo neutralization assays. 2.4. In vitro standard (L1) pseudovirion-based neutralization assay (L1-PBNA) HPV pseudovirions (PSV) were prepared as described previously [23] with some modifications [22]. Standard, L1-neutralization assays were performed as recently described [22] with a pseudovirion input of ∼0.5 ng L1 (per well) so as to reach similar and comparable absolute infectious values for each PSV type. 2.5. In vitro L2 pseudovirion-based neutralization assay (L2-PBNA) The L2 neutralization assay was exactly performed as recently described by Day et al. [24]. 2.6. In vivo neutralization in the cervicovaginal mouse model The cervicovaginal mouse model was performed as originally described by Roberts et al. [25] with minor modifications. Briefly, both passive transfer and challenge experiments were run over the course of eight days. On day 1, Balb/c male cage bedding was transferred to the cages of female mice to initiate hormonal synchronization (Whitten effect). On day 3, 100 ␮l of 30 mg/ml Medroxyprogesteronacetat (Pharmacia, Ireland) was administered subcutaneously to the mice. If a passive transfer experiment was performed, 17 ␮l of serum diluted to 100 ␮l with 1 × PBS were delivered intraperitoneally to each mouse on day 5. No treatment was performed on day 5 for challenge experiments. On day 6 mice were treated with 50 ␮l of 4% Nonoxynol-9 (N9) (Spectrum, USA) in 4% carboxymethylcellulose (CMC). Approximately 4 h after N9 treatment, HPV PSVs were instilled intra-vaginally. On day 8 imaging was performed with a Xenogen IVIS imager (Xenogen Corporation/Perkin Elmer, USA). Images of the mice were acquired before substrate addition and following intra-vaginal instillation of 20 ␮l of the luciferin substrate (15 mg/ml; Promega, USA). After 3 min of incubation, luminescence images were acquired (30 s–1 min exposure time, medium binning). A region-of-interest (ROI) analysis was performed with Living Image 2.50.1 software (Xenogen/Perkin Elmer, USA). 2.7. Cladogram construction Sequence alignment and the resulting cladogram tree were generated with ClustalW2 (http://www.ebi.ac.uk/Tools/msa/ clustalw2/); the logo representation of clade consensus sequences was created with the MEME program (http://meme.nbcr.net/ meme/cgi-bin/meme.cgi).

2612

H. Seitz et al. / Vaccine 32 (2014) 2610–2617

Fig. 1. Antigen design and production. (A) Cladogram tree of the 15 high-risk HPV types based on the sequence of the L2(20–38) peptide. The cladogram was generated by neighbor-joining analysis of aligned L2(20–38) peptide sequences from the indicated HPV types; a logo peptide consensus sequence was then calculated for each of the three major clusters (see Section 2 for details) (B) Monovalent, multivalent and mix antigen design; L2(20–38) peptides from HPV16, 31 and 51 are indicated with black, grey and white squares, respectively. (C) Representative SDS–PAGE image of the various recombinant antigens expressed in E. coli and purified by metal-affinity chromatography; the size (kDa) and migration distance of molecular mass markers are shown in the left-side lane (M).

2.8. Statistical analysis Statistical significance of neutralization assay results was determined with the non-parametric Mann-Whitney test performed with GraphPad Prism 5.00 (GraphPad Software, USA); differences were considered significant at p ≤ 0.05.

(A)

L1 PBNA Titer 10,000

1,000

3. Results 100

With the aim of extending cross-neutralizing responses elicited by Trx-HPV16 L2(20–38)3 , we selected two further HPV types, 31 and 51, as additional sources of L2(20–38) peptide epitopes. Based on sequence alignment and clustering (Fig. 1A) the 20–38 L2 epitopes of these two types appear to be most distantly related to HPV16, thereby potentially covering the entire set of highrisk HPVs. Two multivalent antigens bearing L2(20–38) peptides from HPV31 and HPV51 in addition to HPV16, in an alternate (multivalent 1) or type-homogeneous (multivalent 2) tripeptide arrangement, were constructed and compared with individual type-specific monovalent antigens and with a mixture (“mix”) of the three monovalent antigens (Fig. 1B). The monovalent and multivalent Trx-L2 antigens were all expressed at high levels in E. coli and purified by metal-affinity chromatography (Fig. 1C). 3.1. Type-specific monovalent Trx-L2 antigens do not mediate cross-protection against some HPV types First, the ability of monovalent antigens to elicit crossneutralizing responses was determined. Mouse immune sera were analyzed on HPV16, HPV31 and HPV51 pseudovirions with the standard PSV-based neutralization assay (L1-PBNA) and a recently developed, more sensitive assay optimized for neutralizing L2 antibody detection (L2-PBNA) [24]. L1- and L2-PBNA serum titers from animals immunized with the HPV16 monovalent antigen clearly showed that cross-neutralization was not efficiently extended to

10

HPV type:

1

16

31

51

16 monovalent

16

31

51

31 monovalent

16

31

51

51 monovalent

(B) L2 PBNA Titer 1,000,000 100,000 10,000 1,000 100 10 HPV type:

1

16

31

51

16 monovalent

16

31

51

31 monovalent

16

31

51

51 monovalent

Fig. 2. L1 and L2 PBNA titers of sera from mice immunized with the monovalent Trx-L2 antigens. Mice (n = 5/group) were immunized four times at biweekly intervals with 50 ␮g each of the monovalent HPV16, 31 or 51 Trx-L2 antigens formulated in 50% v/v Montanide ISA720. Final sera were collected eight weeks after the last immunization and titrated on HPV16, HPV31 and HPV51 pseudovirions in the L1 PBNA (A) and L2 PBNA (B).

H. Seitz et al. / Vaccine 32 (2014) 2610–2617

2613

L1 PBNA Titer

(A) p = 0.0015

100,000

p = 0.0007 ns

p = 0.0355 ns

10,000

p = 0.0014

ns

ns

p = 0.0049

1,000 100 10 1

HPV type:

16

31

51

16 monovalent

16

31

51

Multivalent #1

16

31

51

16

Multivalent #2

31

51

Mix

L2 PBNA Titer

(B) p = 0.0002

1,000,000

ns

p = 0.0029 ns

ns

100,000

p = 0.0168 ns

ns

ns

10,000 1,000 100 10 HPV type:

1

16

31

51

16 monovalent

16

31

51

Multivalent #1

16

31

51

Multivalent #2

16

31

51

Mix

Fig. 3. L1 and L2 PBNA titers of sera from mice immunized with monovalent or multivalent formulations of the Trx-L2 antigen. Mice (n = 10/group) were immunized four times at biweekly intervals with 25 ␮g of the indicated antigen formulations adjuvanted with 25% v/v Montanide ISA720 and 25 ␮g PolyIC; the monovalent HPV16 Trx-L2 antigen (same dose and adjuvants as the multivalent formulations) was used as a reference for this experiment. Final serum was collected eight weeks after the last immunization and titrated on HPV16, HPV31 and HPV51 pseudovirions in the L1 PBNA (A) and L2 PBNA (B). ns = not significant.

HPV31 and HPV51 (Fig. 2A and B, Supplementary Table 1). Likewise, the HPV31 monovalent antigen did not elicit neutralizing HPV16 and HPV51 responses. Interestingly, the HPV51 monovalent antigen elicited significant neutralization titers against non-cognate HPV16 pseudovirions, but very low titers against homologous HPV51 PSVs and negligible HPV31 titers. Initially, we hypothesized that HPV51 pseudovirions were behaving differently from HPV16 pseudovirions in the in vitro neutralization assays and that L2 epitope exposure of HPV51 was not as complete or prolonged as for HPV16. However, recent immunization experiments in guinea pigs revealed HPV51 monovalent and, surprisingly, HPV16 monovalent antigens to be potent elicitors of HPV51 neutralizing responses (Seitz et al., in preparation). Still, while not completely comprehensive, single molecule Trx-L2 monovalent antigens do elicit cross-neutralization in mice. In fact, the HPV16 and HPV51 monovalent antigens elicited very effective anti-HPV18, HPV45 and HPV58 cross-neutralizing responses, while sera from mice immunized with the monovalent HPV31 antigen effectively neutralized HPV59, whose L2 epitope bears a similar primary sequence (data not shown). 3.2. The mix of Trx-L2 monovalent antigens elicits more robust and broadly neutralizing responses than single-molecule multivalent Trx-L2 antigens Having confirmed the shortcomings of monovalent Trx-L2 antigens with regard to the induction of comprehensive neutralizing responses, we compared the HPV16 monovalent antigen with two differently formulated multivalent antigens born by a single Trx scaffold molecule (multivalent 1 and multivalent 2, see Fig. 1A)

and the mix of monovalent antigens. While multivalent 1 and 2 induced comparable HPV16 L1-PBNA titers and these were lower than those elicited by HPV16 monovalent antigen, they did not extend neutralization to HPV31 and HPV51 in most of the mice (Fig. 3A, Supplementary Table 2). However, when employing the more sensitive L2-PBNA, nine mice immunized with multivalent 1 yielded potent HPV31 titers and three of them had measurable titers against HPV51 (Fig. 3B). Based on L2-PBNA titers, multivalent 2 appeared to be marginally superior to multivalent 1 since it elicited HPV51 neutralizing responses in a larger fraction of animals. Importantly, however, it was the mix formulation that displayed the most potent immune performance, with HPV31 and HPV51 L1-PBNA titers mostly significantly higher than multivalent 1/2 titers. Also, comparable HPV16 titers in mix immunized mice were maintained. In the L2-PBNA all ten animals immunized with the mix formulation yielded substantial HPV31 neutralizing titers and eight of them also displayed measurable HPV51 titers. Of note, in the L2-PBNA even the HPV16 monovalent sera yielded HPV51 titers comparable to the multivalent and mix immune sera. For a more comprehensive cross-neutralization analysis, the HPV16 monovalent, multivalent 1 and 2, and mix immune sera were tested on the 10 most prevalent high-risk HPV pseudovirions in both the L1 and L2-PBNAs (Fig. 4A–J). The results of this analysis clearly pointed to the mix as the formulation of choice. In addition to eliciting the most potent HPV31 and HPV51 neutralizing responses (Fig. 4D and J) the mix also extended neutralization to HPV59 (Fig. 4I). The aa 20–38 region of HPV59 L2 closely resembles the corresponding region of HPV31 with a conserved serine residue at position 30. This possibly explains why the HPV16 monovalent

2614

H. Seitz et al. / Vaccine 32 (2014) 2610–2617

Fig. 4. In vitro (cross-) neutralization efficiency of sera from mice immunized with the monovalent HPV16 Trx-L2 antigen and with different multivalent formulations of the HPV16, 31 and 51 Trx-L2 antigens. Immune sera were tested against 10 different HPV pseudovirions at a fixed 1:200 dilution using the L1 and the L2 PBNAs as indicated. The tested HPV types were: HPV16 (A), HPV18 (B), HPV45 (C), HPV31 (D), HPV33 (E), HPV52 (F), HPV58 (G), HPV35 (H), HPV59 (I) and HPV51 (J). Four groups of mice (n = 10/group) were immunized four times at biweekly intervals with a total of 25 ␮g antigen adjuvanted with 25% v/v Montanide ISA720 and 25 ␮g PolyIC. Each dot represents one mouse serum. Data are presented as mean neutralization percentages ± SD. ns = not significant.

H. Seitz et al. / Vaccine 32 (2014) 2610–2617

ul M

passively transferred sera Fig. 5. Passively transferred sera from mice immunized with multivalent 1, multivalent 2 and the mix formulation prevent infection by HPV16 pseudovirions. The indicated sera (multivalent 1, 2 and mix) were passively transferred to naïve mice that were subsequently infected with HPV16 pseudovirions. Negative control animals (n = 5, black bar) received 1 x PBS; test mice (n = 3/group, white bars) received 17 ␮l each of the indicated sera (effective ∼1:100 dilution); numbers in brackets indicate the specific animal whose immune-serum was utilized for the passive transfer assay (see also Supplementary Table 2). Data are presented as average radiance values ± s.e.m.

antigen failed and the mix succeeded at extending neutralization to this particular HPV type. Notably, multivalent 2 performed better than the HPV16 monovalent and the multivalent 1 formulation at eliciting HPV31, HPV59 and HPV51 neutralizing responses (Fig. 4D, I and J). For the remaining seven oncogenic HPV types (HPV16, HPV18, HPV45, HPV33, HPV52, HPV58 and HPV35), the HPV16 monovalent, the multivalent 1 and 2 and the mix antigens elicited similarly cross-neutralizing responses (Fig. 4A–C and E–H). 3.3. In vivo HPV protection by different Trx-L2 antigens To gain insight on how in vitro neutralization titers parallel in vivo protection capacity, three HPV16 L1- and L2-PBNA-positive sera were randomly selected and tested in a small-scale passive transfer experiment in mice (see Supplementary Table 2). Due to the small sample size of this experiment and two of five animals not infected properly in the negative control group, significance could not be determined. Despite this, in vivo neutralization activity was observed in all three sample groups receiving Trx-L2 immune sera, indicating that L1/L2 PBNA results correlate well with in vivo neutralization (Fig. 5). Even though inclusion of HPV31 L2 sequence information in Trx-L2 antigens appears to be essential for eliciting in vitro detectable HPV31 neutralizing responses, we wished to determine whether mice immunized with the HPV16 monovalent antigen are protected from challenge with HPV31 pseudovirions. To this end we analyzed sera from mice immunized with HPV16 monovalent antigen (n = 9) shortly before HPV31 pseudovirion challenge. No anti-HPV31 neutralizing titers could be detected with the L1-PBNA and only three animals yielded low titers in the L2-PBNA. Nonetheless, almost all animals (with the exception of animal 5) showed measurable levels of protection upon challenge with HPV31 pseudovirions despite poor infection of two out of five animals in the negative control group (Fig. 6). This supports our observation that even when in vitro titers fall below detection,

A three component mix of thioredoxin-L2 antigens elicits broadly neutralizing responses against oncogenic human papillomaviruses.

Current human papillomavirus (HPV) vaccines based on major capsid protein L1 virus-like particles (VLP) provide potent type-specific protection agains...
881KB Sizes 1 Downloads 3 Views