Combined prophylactic and therapeutic immune responses against human papillomaviruses induced by a thioredoxin-based L2-E7 nanoparticle vaccine

Global burden of cervical cancer, the most common cause of mortality caused by human papillomavirus (HPV), is expected to increase during the next decade, mainly because current alternatives for HPV vaccination and cervical cancer screening programs are costly to be established in low-and-middle income countries. Recently, we described the development of the broadly protective, thermostable vaccine antigen Trx-8mer-OVX313 based on the insertion of eight different minor capsid protein L2 neutralization epitopes into a thioredoxin scaffold from the hyperthermophilic archaeon Pyrococcus furiosus and conversion of the resulting antigen into a nanoparticle format (median radius ~9 nm) upon fusion with the heptamerizing OVX313 module. Here we evaluated whether the engineered thioredoxin scaffold, in addition to humoral immune responses, can induce CD8+ T-cell responses upon incorporation of MHC-I-restricted epitopes. By systematically examining the contribution of individual antigen modules, we demonstrated that B-cell and T-cell epitopes can be combined into a single antigen construct without compromising either immunogenicity. While CD8+ T-cell epitopes had no influence on B-cell responses, the L2 polytope (8mer) and OVX313-mediated heptamerization of the final antigen significantly increased CD8+ T-cell responses. In a proof-of-concept experiment, we found that vaccinated mice remained tumor-free even after two consecutive tumor challenges, while unvaccinated mice developed tumors. A cost-effective, broadly protective vaccine with both prophylactic and therapeutic properties represents a promising option to overcome the challenges associated with prevention and treatment of HPV-caused diseases.


Introduction
Cervical cancer is the fourth most common cancer in women worldwide. It is estimated that more than one million women are currently suffering from cervical cancer, and there were 570,000 new cases in 2018 [1]. According to current projections, the global burden of cervical cancer will continue to rise and will reach up to 700,000 cases and 400,000 deaths by 2030 [2,3]. Nearly 90% of the current death cases occur in low-and middle income countries (LMIC) [2]. The main cause of cancerous cervical lesions is persistent infection by an oncogenic HPV type [4]. At least 14 oncogenic HPV types are known to induce cervical carcinogenesis [5]. While the carcinogenic process usually progresses from initial infection to the invasive carcinoma stage over one to three decades, precancerous lesions occur much earlier [6].
An effective therapeutic strategy or post-exposure prophylaxis capable of eliminating HPVinfected cells should involve immune system activation with the aim to target viral antigens. An effective therapy should thus induce virus-specific T-cell responses. The HPV oncoprotein E7, due to its prominent role during the onset and progression of malignancy and expression limited to HPV-positive neoplastic cells, stands out among the most promising viral targets for vaccine-induced therapy [6,12,13]. Together with E6, the other major HPV oncoprotein, E7 has been the target of various immunotherapeutic approaches [13,14]. Some of these have proven to be at least partially successful, but no immunotherapeutic protocol for the treatment of pre-cancerous lesions or established HPV tumors has thus far reached a clinically consolidated stage of development. A variety of E6/E7 antigen formulations has been set-up and tested, but only a few of them were actually aiming at both HPV prophylaxis and the immunotherapeutic treatment of pre-existing HPV-related malignancies [15][16][17][18][19][20]. Interestingly, some of these dual-purpose vaccine formulations, include HPV16 L2 as the monotypic prophylactic component of the vaccine, linearly fused to E7 alone [16][17][18]21], or in combination with E6 [19,20], in the form of either chimeric virus-like particles (VLPs) [16][17][18] or linear polypeptides [19,20].
As cross-presentation represents a crucial step for the activation of T-cell responses, a therapeutic vaccine should rely on a highly immunogenic and possibly multivalent antigen for efficient delivery to dendritic cells. Following-up to our earlier work on the development of a broadly protective, multimeric L2-based vaccine prototype [22,23] and to the encouraging results obtained with some of the above dual-purpose vaccines [15], we describe, here, the construction and testing of a combined vaccine relying on L2-and E7-specific epitopes grafted on the surface of a hyper-stable thioredoxin (Trx) scaffold. In particular, HPV-L2 immune reactivity was provided by a polytopic repeat derived from eight different HPV types [22,23], coupled to the extended version of an E7-specific CTL epitope [24]. The resulting basic antigen was then converted into a nanoparticle format with the use of a heptamerization-promoting module (OVX313) that has previously proven to be a very effective super-scaffold for the construction of multivalent vaccines targeting various infectious agents in addition to HPV [25][26][27][28][29]. Our data document the pivotal role of the L2 8mer polytope and the OVX313 heptamerization domain in the induction of an E7-specific T-cell response, without any appreciable interference with the humoral production of HPV16 and HPV18 L2 neutralizing antibodies. We also provide a proof-of-concept demonstration of the strong therapeutic capacity of the Trx-8mer-flank E7-OVX313 vaccine in a double-challenge mouse model of HPV16 E7-induced carcinogenesis. Trx-8mer-flank E7-OVX313 thus lends itself as a robust and costeffective vaccine with protective and therapeutic capacity, potentially capable of resolving productive infection as well as HPV-related malignancies, and thus benefitting both uninfected and already infected individuals.
https://doi.org/10.1371/journal.ppat.1008827.g001 superior adjuvant for peptide immunization (S1 Fig). AddaVax was thus used as adjuvant in all subsequent experiments performed with PfTrx-based antigens. As shown in Fig 1D, even with AddaVax as adjuvant, immunogenicity of the E7 epitope bearing Trx construct (PADRE-Trx-E7; see Fig 1A), was not significantly higher than that of the corresponding free peptides. In this construct, the three-fold repeated E7 epitopes are separated by a Gly-Gly-Pro linker and we reasoned that there might be insufficient proteasomal epitope processing. Therefore, a second antigen (PADRE-Trx-flank E7) was designed, in which the E7 epitopes were individually flanked, both C-and N-terminally, by the five amino acids that in the HPV16 E7 protein are located upstream and downstream to the sequence of the standard E7 49-57 epitope (see Fig 1A legend for details). As shown in Fig 1D, a robust response, outperforming the peptide-induced responses, was observed against the HPV16 E7-specific CTL epitope upon immunization with the PADRE-Trx-flank E7 antigen. In conclusion, the PADRE-supplemented PfTrx scaffold appears to be an effective inducer of cytotoxic T-cell responses and was thus considered as a promising building block for the construction of a therapeutic vaccine.

L2 neutralization epitopes can be combined with CTL epitopes
After assessing the potential of PfTrx as a scaffold capable of inducing CD8 + cytotoxic T-cell responses, we asked whether L2 neutralization epitopes could be incorporated in the construct in order to induce a B-cell response without compromising the ability to drive T-cell immunity. The L2 proteins of papillomaviruses harbor a major cross-neutralization epitope in their N-terminal region, corresponding to amino acids 20-38 of HPV16 L2 [33]. Starting with the PADRE-Trx-OVA construct, we added three copies of the HPV16 L2 20-38 epitope (Fig 2A) and used this combined antigen to immunize C57BL/6 and BALB/c mice.

PLOS PATHOGENS
Combined prophylactic and therapeutic immune responses against human papillomaviruses To confirm the ability to induce CTL responses, we performed IFN-γ ELISpot assays after a single immunization of C57BL/6 mice with PADRE-Trx-OVA and PADRE-Trx-L2-OVA constructs, using PADRE-Trx-L2 as a negative control. As shown in Fig 2B, where the cellular immunity induction performance of PADRE-Trx-L2-OVA is compared with that of the L2-lacking construct PADRE-Trx-OVA, insertion of the L2 epitopes did not significantly affect the ability to induce OVA-specific CTL responses. The ability of the combined vaccine to induce a L2-targeted, anti-HPV humoral response was determined, in parallel, in PADRE-Trx-L2 and PADRE-Trx-L2-OVA immunized BALB/c mice. As shown by the results of the anti-HPV16 and HPV18 pseudovirion-based neutralization assays in Fig 2C, the presence of the OVA-derived CTL epitopes slightly enhanced the induction of neutralizing antibodies, although this effect was only statistically significant in the case of HPV16 pseudovirions.

Antigen heptamerization and addition of an L2 polytope within the display site of thioredoxin boost T-and B-cell immunogenicity
According to our previous results [30,33], incorporation into PfTrx of only the major L2 cross-neutralization epitope of HPV16 fails to provide protection against certain oncogenic HPV types, some of which, for example HPV31 and 51, partially escape neutralization by the antibodies elicited by HPV16 L2 [30,33]. Therefore, in order to induce a more comprehensive humoral response against all oncogenic mucosal HPV types plus the mucosal low-risk types HPV6 and 11 which are responsible for > 80% of genital warts, we developed a polytope comprising the L2 epitopes of seven oncogenic HPV types plus HPV6 (named 8mer [22,23]). When coupled with the heptamerization domain OVX313, the resulting antigen (Trx-8mer-OVX313) exhibited a broader and stronger cross-protection in mice and guinea pigs compared to its monomeric version (Trx-8mer) [23]. Additionally, we documented the presence of a T-helper epitope derived from the OVX313 domain in BALB/c mice [23]. Based on this body of evidence, we set out to replace the promiscuous T-helper epitope PADRE and the HPV16 L2 trimer with the OVX313 module and the 8mer polytope, respectively (see Fig 3), in order to determine whether the resulting construct would lead to an increased T-cell response. As shown in Fig 4A, compared to the monomeric PADRE-Trx-L2-OVA antigen, the heptameric form of the antigen containing the 8mer polytope (Trx-8mer-OVA-OXV313) induced a stronger OVA-specific T-cell response. In the IFN-γ ELISpot assay, the stimulatory capacity of the double-modified Trx-8mer-OVA-OXV313 antigen was increased by more than two-fold. Incorporation of the OVX313 domain together with the 8mer polytope into the antigen harboring the E7 multiepitope (Trx-8mer-flank E7-OVX313, Fig 3) also positively affected T-cell induction, albeit to a lesser extent compared to the OVA antigen ( Fig 4A). Importantly, as shown in Fig 4B, both heptameric constructs induced T-cell responses specifically recognizing syngeneic target cells endogenously expressing OVA or E7, including TC-1 cells, thereby demonstrating the capacity of these vaccines to prime tumor-reactive T cells in vivo.
Next, we asked whether the presence of the L2 polytope, the OVX313 domain, or both, also leads to an enhanced induction of CD8 + T-cell responses. To directly document a possible contribution of the L2 polytope, we compared the immunogenicity of the Trx-8mer-OVA-OVX313 and the Trx-OVA-OVX313 constructs (see Fig 3) in C57BL/6 mice. According to the results of the IFN-γ ELISpot assay in Fig 4C and the OVA-specific tetramer staining data shown in Fig 4D, the presence of the L2 polytope in the heptamerized antigen substantially increased the immunogenicity of the OVA-derived epitopes. In both assays, a more than two-fold increase in spot numbers and in the frequency of OVA-specific CD8 + T cells was observed with splenocytes from mice immunized with the antigen harboring the 8mer polytope. To better delineate the importance of multimerization on the increased immunogenicity induced by the OVX313 module, in a complementary approach, we further compared the immune performance of the Trx-8mer-OVA-OVX313 and the Trx-8mer-OVA antigens (see Fig 3 for details on construct design). As shown in Fig 4E, the presence of the OVX313 module significantly improved OVA-specific CTL responses. In contrast, no such effect was observed for constructs bearing the E7 epitopes, for which the induction of an E7-specific CTL response did not differ significantly between the Trx-8mer-flank E7-OVX313 and the Trx-8mer-flank E7 antigens.
Nevertheless, in keeping with the results of our previous studies, we observed a clear contribution of OVX313-induced heptamerization to the strength of the L2-specific antibody responses. Indeed, as shown in Fig 5A, neutralizing antibody titers against HPV types 16 and 18 were approximately 10-15-fold higher in mice immunized with the heptameric antigen (Trx-8mer-flank E7-OVX313) compared to its monomeric counterpart (Trx-8mer-flank E7). The two antigens also differed in the isotype profile of the elicited antibodies. In fact, as shown in Fig 5B, a distribution with a prevalent IgG1 isotype, that correlates with a Th2 response, was observed for the OVX313-lacking antigen, whereas a pattern skewed toward the Th1 response-correlated IgG2a and IgG2b isotypes was apparent for the OVX313-containing antigens, regardless of the presence of the E7 epitope. A similar isotype distribution of L2-specific antibodies was observed for both HPV16 (Fig 5B, left panel) and HPV18 (Fig 5B, right panel). Based on this data, we conclude that OVX313-mediated heptamerization enhances both humoral and cellular immunogenicity, even though the latter effect is influenced by the specific CTL epitope.

Combined prophylactic and therapeutic immune responses against human papillomaviruses
We then wished to find out whether the enhanced immunogenicity observed in the presence of the heptamerization module only depends on OVX313-induced antigen multimerization or also involves an improved T-helper cell activation. To this end, we used a set of long overlapping peptides spanning the entire OVX313 sequence and, in keeping with our previous The diagram on the right shows the cumulative data obtained from the experiments performed with the tetramers; each symbol represents one mouse; the mean percentage (OVA 257-264specific CD8 + T cells/total CD8 + T cells) is indicated by horizontal bars. (E) Influence of the OVX313 domain on OVA-and E7-specific CTL responses. In the left graph, the numbers of IFN-γ spots per 10 6 splenocytes, measured upon stimulation with the OVA 257-264 peptide, are compared in two groups of mice immunized with Trx-8mer-OVA-OVX313 or Trx-8mer-OVA, as indicated. The results of the same assay, performed with splenocytes from Trx-8mer-flank E7-OVX313 or Trx-8mer-flank E7 immunized mice, stimulated with the E7 49-57 peptide, are shown on the right. Shown are the mean and SD of triplicate values on each mouse. P-value � 0.05 was considered as significant and are labeled as follows: � , P-value < 0.05; �� , P-value < 0.01; ��� , P-value < 0.001; ���� , P-value < 0.0001.
The above data, together with the lack of interference of the CTL epitopes with the humoral responses induced by the 8mer component in a single construct shown by PBNA ( Fig 6A) and also by protection in the cervico-vaginal mouse model (Fig 6B), clearly point to the potential (A) Sera collected from vaccinated mice (10 animals per group) one month after the last immunization with Trx-8mer-flank E7 or Trx-8mer-flank E7-OVX313, were analyzed against HPV16 (left) and HPV18 (right) pseudovirions using the L1-PBNA. Each symbol represents the neutralization titer (EC50) of an individual mouse; geometric means of the titers for each group are indicated by horizontal lines. EC50 values of three Trx-8merflank E7 immunized mice are lower than 50 (the EC50 value set as threshold for neutralization positivity); the corresponding data are thus not shown in the graph. P-value � 0.05 was considered as significant. � , P-value < 0.05; �� , P-value < 0.01. (B) Analysis of isotype distribution of L2-specific antibodies measured by antibody isotype ELISA; the values are the mean ±SD of data obtained from duplicate assays performed on the pooled sera from each immunization group. (C) A T-helper response is induced by the 'OVX313-I5' peptide. Mice were immunized twice at 5 days intervals with the Trx-8mer-flank E7-OVX313 antigen. The splenocytes were stimulated with the OVX313 peptide panel (OVX313-I1 to OVX313-I9; see 'Materials and methods' for details). Data, representing the numbers of IFN-γ spots per 10 6 splenocytes, are the mean ±SD of three mice for each peptide stimulation. https://doi.org/10.1371/journal.ppat.1008827.g005

PLOS PATHOGENS
Combined prophylactic and therapeutic immune responses against human papillomaviruses

PLOS PATHOGENS
Combined prophylactic and therapeutic immune responses against human papillomaviruses of the Trx-8mer-CTL epitope-OVX313 design for constituting a prophylactic/therapeuticcombined anti-HPV vaccine.

PLOS PATHOGENS
Combined prophylactic and therapeutic immune responses against human papillomaviruses population (with an average value of 3.2%; right panel) was found to be specific for the E7 epitope, thus indicating a potent E7-specific immunogenicity of the Trx-8mer-flank E7-OVX313 antigen.
Having demonstrated the CTL immunogenicity of our combined antigen ex vivo, we next tested its therapeutic capacity in the TC-1 murine model of HPV16 E7-induced carcinogenesis. To this end, C57BL/6 mice were first challenged subcutaneously with E7-expressing syngeneic TC-1 cells [34], followed by two immunizations with the Trx-8mer-flank E7-OVX313 antigen 7 and 12 days after the challenge (see Fig 7B for a schematic representation of this experiment). As indicated by the tumor growth and Kaplan-Meier survival curves reported in Fig 7C and 7D, 10 out of 12 vaccinated mice remained tumor-free, whereas 10 out of 12 control mice developed tumors within about two months. Moreover, complete protection was observed in five mice that had successfully responded to Trx-8mer-flank E7-OVX313 immunotherapy and were re-challenged with TC-1 cells. In contrast, all mice of the same age belonging to the non-vaccinated control group developed tumors (Fig 7E). These results clearly indicate that the T-cell responses we have found to be induced by the Trx-8mer-flank E7-OVX313 antigen (see Fig 7A) are sufficient not only to induce regression of TC-1 tumors, but also to prevent tumor outgrowth upon subsequent re-challenge. We then tested the CTL responses of cured mice that had remained tumor free upon re-challenge by ex-vivo IFN-γ ELISpot assays and included an E6-derived CTL epitope in addition to the E7-specific epitope in order to test for possible epitope spreading [35,36]. A potent E7-specific T-cell response, but no response against E6, was detected in these animals (Fig 7F). To further evaluate therapeutic efficacy, tumor-bearing mice at a further advanced stage were immunized with either the Trx-8mer-flank E7-OVX313 antigen, or with the E7-lacking, Trx-8mer-OVX313 construct as a control, (Fig 8A). As reported in Fig 8B and 8C, a curbed tumor growth and an improved survival rate were observed in mice immunized with the Trx-8mer-flank E7-OVX313 antigen compared to the Trx-8mer-OVX313 control. In conclusion, vaccination with Trx-8mer-flank E7-OVX313 exerted a strong therapeutic effect in the TC-1 tumor model of HPV16-induced carcinogenesis.

Discussion
Not many diseases highlight how global inequity impacts humanity as cervical cancer. Currently available measures for prevention, including prophylactic vaccination, have being poorly implemented on a national scale in low and middle income countries, which resulted in twice and three times higher cervical cancer incidence and mortality, respectively, compared to high-income countries, where such measures are widely available. In addition, a vast number of already infected women are expected to develop cervical precancerous and cancerous lesions in the years to come. While there is no approved immunotherapy for HPV-driven malignancies, current treatments vary from surgical intervention, chemotherapy, radiation, to a combination of these three approaches, which although often effective, do not clear HPV infection. In this context, a prophylactic vaccine also endowed with a therapeutic activity, and thus administrable to exposed individuals (e.g., HIV-positive subjects) regardless of their HPV-infection status would be highly desirable. Such a dual-purpose vaccine would be particularly valuable for post-exposure prophylaxis. Early stages of HPV infection are characterized by the predominant expression of virus early proteins (including E6 and E7), which do not sustain virion production, and very little or no expression of late capsid proteins (including L2). Under these conditions, anti-early HPV protein, in particular anti-E6 or anti-E7 T-cell responses are critical for clearing virus-infected cells, whereas humoral antibody responses to capsid proteins have no proven effect against a prevalent infection. This scenario changes within days as virus-infected cells follow-on in the epithelial differentiation program, with the overexpression of virus capsid proteins and subsequent virion formation in the upper layers of the epithelium (upon skin desquamation). Infection of other epithelial cells by these newly released virions can only be prevented by humoral antibody responses directed against capsid proteins, while cellular immunity does not play a significant role at this stage of the viral replication cycle. We think, therefore, that the concomitant stimulation of B-and T-cell responses is crucial for mounting an effective immunity against the dynamic and complex chain of events that accompany the HPV life cycle. Moreover, after virus clearance, such combined vaccines would shield epithelial cells from recurrent HPV infections. Appreciation of the scope and potential benefits of this kind of comprehensive anti-HPV weapon has driven the search for effective dual-purpose vaccines capable of a prophylactic and therapeutic action against HPV.
One approach in this direction is to use VLPs as carriers of early HPV antigens. VLPs containing an L2-E7 fusion protein, and thus capable of inducing anti-L1 prophylactic and anti-E7 therapeutic responses, have been produced and tested in mouse models of HPV-induced carcinogenesis [18,38]. Another approach was to produce chimeric VLPs composed of the HPV16 L1 protein, C-terminally fused to the N-terminal region of HPV16 E7 [16]. After promising preclinical prophylactic and therapeutic effects [39,40], this combined vaccine was administered to patients with HPV-induced CIN 2/3 lesions, but no significant clinical effects were observed [41]. Another combined antigen, TA-CIN (tissue antigen-cervical intraepithelial neoplasia) containing HPV16 L2, E6 and E7 was applied to vulvar intraepithelial

PLOS PATHOGENS
Combined prophylactic and therapeutic immune responses against human papillomaviruses neoplasia patients together with the innate immunity modulator Imiquimod [19]. The trial was not placebo-controlled, but lesions regressed more frequently compared to previous intervention studies and a local infiltration of CD4 + and CD8 + T cells was observed.
In the present work, we aimed to generate a therapeutic antigen with broadly protective ability by combining L2 from various HPV types and E7 from HPV16. Our concept of vaccine design followed an increasing complexity development pipeline. Initially, we validated the ability of our monomeric PADRE-supplemented Trx scaffold to induce cytotoxic T-cell responses against the OVA CTL epitope, utilized as a prototype target epitope, and subsequently confirmed also for the HPV16 E7 CTL epitope. Building on our previous work, we showed that multiple prophylactic epitopes as well as a three-fold repeated HPV16 E7 CTL epitope could be incorporated into the basic Trx scaffold, which upon fusion with the OVX313 hepatmerization domain was converted into a highly multivalent nanoparticle-like format. The resulting antigen, Trx-8mer-flank E7-OVX313, was then shown to be capable of inducing HPV neutralizing responses as well as potent antitumor effects in C57BL/6N mice.
In addition to the combined prophylactic and therapeutic activity, our dual-purpose vaccine presents some unique advantages compared to previous similar HPV vaccines. One of the main aims that inspired and motivated Trx-8mer-flank E7-OVX313 development was, in fact, the reduction of the economic burden on global health. Our vaccine was produced in E. coli, which is the most cost-effective and easy-to-use recombinant protein production host. Further, the Trx scaffold utilized for our antigen is derived from the hyperthermophile Pyrococcus furiosus, and it has previously been shown to be an extremely robust protein, highly resistant to heat, freeze-drying as well as proteolytic damage, and with a large capacity to accept long polypeptide inserts within its display site [30,31,42]. Due to this sturdy scaffold, our vaccine, a prophylactic variant of which is currently undergoing GMP production in view of a phase I clinical study, does not require cold-chain transportation, which cuts costs significantly and makes it applicable to many underdeveloped countries worldwide. In addition, it was reported that T-helper responses against E7 are considerably correlated with self-regression of HPVinduced lesions in healthy individuals [43,44]. We thus envision a possible application of our vaccine in individuals with early-stage HPV infection who could possibly benefit from a boosting of E7-specific T-helper responses for the clearing of HPV-transformed cells.
We found that a 10-amino acids extension of the published CTL epitope [24], from RAH-VYNIVTF (E7 [49][50][51][52][53][54][55][56][57] to QAEPDRAHVYNIVTFCCKCD (E7 44-62 , namely flank E7), significantly improved E7-specifc T-cell responses. We believe that this enhanced T-cell response can be attributed to the sequences flanking the CTL epitopes, which may facilitate proteasomal processing, thus promoting epitope presentation. Indeed, there have been several claims on the importance of the design of the linkers or flanking sequences interposed between individual epitopes for optimal vaccine immunogenicity and the induction of enhanced T-cell responses [45][46][47][48][49]. This is especially important if the vaccine harbors multi-epitopes. In addition to the inter-epitope flanking sequences, we have shown that the oligomerization domain OVX313 plays a pivotal role in overall immunogenicity. OVX313 is a positively charged derivative of the C4-binding protein (C4 bp) involved in complement system inhibition that promotes the autonomous heptamerization of any protein fusion partner. Also, it is a chimeric variant of avian C4 bp sharing less than 20% similarity with human C4 bp, a feature that has been engineered in order to minimize auto-antibody induction. It contains an amphipathic αhelix region, which is necessary and sufficient for heptamerization, as well as two cysteine residues which upon inter-subunit disulfide bond formation covalently stabilize the resulting heptameric structure [50]. OVX313 has been successfully employed for the construction of multiple prophylactic vaccines in addition to HPV [25-29, 51], some of which are progressing into clinical trial investigation studies (e.g., tuberculosis vaccine, NCT01879163; malaria vaccine, NCT02532049; influenza vaccine, NCT03594890). Due to the density of positive charges it provides, the OVX313 module may also enhance overall immunogenicity by promoting cellular internalization of the antigen. However, the immunological mechanisms underlying OVX313-mediated immunogenicity enhancement are not fully understood. We observed a Tcell response induced by stimulation with the 'OVX313-I5' peptide in IFN-γ ELISpot assay. However, this response could not be identified as either CD8 + or CD4 + T cell-specific by intracellular cytokine staining. Nevertheless, because of the fairly large size (20 amino acids) of the stimulating peptide, we assume it represents a T-helper response. Additionally, a T-cell response induced by the same OVX313 peptide was observed in BALB/c [23] but also in A2DR1 (S2 Fig) mice. This strongly suggests that the 'OVX313-I5' peptide is likely a universal T-helper epitope.
We observed enhanced OVA specific T-cell responses upon incorporation of the OVA-specific CTL epitope within the L2 8mer polytope displayed on Trx. Nonetheless, by employing a set of L2 8mer-specific, 20mer-long, overlapping peptides we could not identify any T-helper epitope clearly associated to the L2 polytope. Other assays will be required in order to precisely localize T-helper epitopes in the L2 8mer.
Epitope spreading, that is the development of immune responses against epitopes noncross-reactive with the original inducing epitope, has been observed in patients upon therapeutic HPV vaccination [52,53]. The potent therapeutic effect exerted by the Trx-8mer-flank E7-OVX313 antigen in TC-1 tumor-bearing mice was accompanied by an epitope-specific anti-E7 CTL response without any detectable response against E6, thus suggesting that epitope spreading was not involved in TC-1 tumor regression.
There are also a few shortcomings and potential limitations associated with our combined vaccine. For example, at variance with the extremely broad prophylactic capacity conferred by the L2 polytope [23], the presence of a monotypic E7 epitope may limit its therapeutic activity to one (HPV16) or only a few HPV types. Also, our dual-purpose vaccine was developed and tested in the genetic background of C57BL/6N mice. Since our ultimate goal is to translate the Trx-8mer-flank E7-OVX313 prototype antigen from animals to humans, more efforts will have to be made toward the development of an HPV vaccine that can induce both humoral and cellular immune responses in a human genetic background.

Protein expression and purification
Synthetic DNA encoding target proteins were inserted into the pET 24 plasmid for expression in E. coli BL21. "PADRE-Trx" derived proteins contain dual 6xHis-tag and nickel affinity chromatography was applied for purification of these antigens. "OVX313-Trx" recombinant proteins were purified by cation exchange chromatography (HiTrap SP FF column, GE Healthcare) based on an arginine-rich motif at the C-terminus of the OVX313 heptamerization domain (OligoDOM technology, OSIVAX). The concentration and purity of the proteins were analyzed by SDS-PAGE-Coomassie blue staining. For endotoxin removal, all proteins were detoxified twice by Triton X-114 separation before immunization.

PLOS PATHOGENS
Combined prophylactic and therapeutic immune responses against human papillomaviruses

Mouse immunization
The 6 to 8 weeks-old female C57BL/6N mice were purchased from Envigo (Gannet, France; animal permit G20/17, Regierungspräsidium Karlsruhe, Germany) and the 6 to 8 weeks-old female BALB/c mice were obtained from Charles River Laboratories (Sulzfeld, Germany; animal permit G248/16, Regierungspräsidium Karlsruhe, Germany). Mice were kept at the German Cancer Research Center under specific-pathogen-free conditions. C57BL/6N mice were used for assessment of cellular immune responses. Twenty μg of the protein adjuvanted with 50% (vol/vol) AddaVax (InvivoGen; 50 μl) or Incomplete Freund's adjuvant (Sigma-Aldrich; 50 μl), or 50 μg aluminum hydroxide (Brenntag) and 10 μg synthetic monophosphoryl lipid A (Avanti Lipids) (5 μl aluminum hydroxide, 10 μl monophosphoryl lipid A, in a total volume of 50 μl) was injected subcutaneously (s.c.) into the base of a mouse tail. For peptide administration, the adjuvant formulation and the immunization route were the same as protein vaccination, but the injected dose was different. A peptide mixture containing 100 μg MHC-I restricted peptide and 140 μg MHC-II restricted peptide was injected s.c. in a total volume of 50 μl. Mice received a single injection and were sacrificed 7 days later for ex vivo analysis of Tcell responses by IFN-γ ELISpot assay. Alternatively, mice were immunized twice at weekly intervals and splenocytes of sacrificed mice were analyzed by multimer staining 7 days later after the last immunization. Female BALB/c mice were employed for evaluation of humoral immune responses. Twenty μg of protein adjuvanted with 50% (vol/vol) AddaVax (25 μl) were injected intramuscularly (i.m.) into the caudal thigh muscle. Mice were immunized 4 times at biweekly intervals. One month after the last immunization, blood was collected by cardiac puncture and serum was analyzed against HPV16 and HPV18 pseudovirions using L1-PBNA.

Pseudovirion preparation
Pseudovirions (PsV) were prepared by cotransfection of the human fibroblast cell line 293TT with plasmids encoding humanized HPV L1 and L2 proteins of different HPV types plus a reporter plasmid. Purification was performed by iodixanol gradient ultracentrifugation following a previously described protocol [54,55].

In vitro standard (L1) neutralization assay (L1-PBNA)
Unless otherwise stated, the L1-PBNA was used to detect neutralizing antibodies in sera against human papillomaviruses. L1-PBNA was performed as described previously [54]. Briefly, 50 μl of serially diluted sera were combined with 50 μl of diluted PsV and incubated at room temperature for 20 min. Next, 50 μl of HeLa T cells [56] (2.5×10 5 cells/ml) were added to the PsV-antibody mixture and incubated for 48 h at 37˚C humidified incubator. The amount of secreted Gaussia luciferase in 10 μl of cell culture supernatant was determined using the Gaussia glow juice kit according to the manufacturer's instructions (PJK GmbH; Germany). The light emissions of samples were measured 15 minutes after substrate addition. Serum concentrations (titers) inhibiting 50% of the PsV infection (EC50) were calculated from median of duplicates. EC50 values > 50 were defined as neutralizing positive.

In vivo neutralization in the cervico-vaginal mouse model
The mouse cervico-vaginal model experiment was performed as described previously [57] with minor modifications (animal permit number G-93/17). The passive transfer assay was performed over the course of 8 days. On day 1, BALB/c male mouse cage bedding was transferred to the cages of female mice to induce hormonal synchronization (Whitten effect). On day 3, 100 μl of 30 mg/ml medroxyprogesterone acetate (Depo-Provera; Pharmacia) was injected subcutaneously. One hundred microliters of serum from immunized mice (diluted 1:1 with PBS) was delivered i.p. to each mouse on day 5. On day 6, mice were treated intravaginally with 50 μl of 4% Nonoxynol-9 (N9; Spectrum) in 4% carboxymethyl cellulose (Sigma). Four hours after N9 treatment, HPV pseudovirions encapsidating a firefly luciferase plasmid in 4% carboxymethyl cellulose (Sigma) were instilled intravaginally. Luminescence-based imaging using a Xenogen in vivo imaging system (IVIS) imager (Xenogen Corporation; Perki-nElmer) was performed on day 8, and the efficiency of L2-directed immune responses was assessed after intravaginal instillation of 20 μl luciferin substrate (15 mg/ml; Promega). A region-of-interest (ROI) analysis was performed using Living Image 2.50.1 software (Xenogen; PerkinElmer). Background signal was obtained and subtracted by imaging each group of mice prior to instillation of luciferin.

PLOS PATHOGENS
Combined prophylactic and therapeutic immune responses against human papillomaviruses Germany) in FACS buffer at 4˚C for 1 hour protected from the light. The number of OVA 257-264 tetramer + CD8-α + or E7 49-57 streptamer + CD8-α + T cells was determined by flow cytometry.

Tumor regression assay
Six weeks-old C57BL/6N female mice were inoculated with 0.2×10 6 TC-1 cells in 100 μl of PBS on their right flank. When tumor size was between 3-5 mm diameter, half of the tumor bearing animals were immunized at the base of the tail s.c. with 20 μg of antigen Trx-8mer-flank E7-OVX313 adjuvanted with 50% (vol/vol) AddaVax (50 μl), and two doses of the vaccine were received with 5 days apart. The rest of tumor bearing mice were used as negative control without any vaccination. When the tumor of vaccinated mice had completely regressed, half of the regressors were re-challenged with 0.3×10 6 TC-1 cells to detect the memory T-cell response induced by the vaccine. Tumor volume was measured with a digital caliper. Mice were excluded from the experiment when the tumor diameter exceeded over 1.5 cm.