Identification of Vaccine Candidates by Immunization with a Complex Pool of Recombinant VLPs

Abstract

Identifying the regions of a protein that can be readily translated into an effective vaccine remains a major challenge in epitope-focused vaccine design, especially when a protein is poorly characterized. Here, we describe an immune-driven strategy for epitope identification which uses the immune system to reveal regions within a candidate protein capable of eliciting targeted antibody responses. Using IL-17A, a pro-inflammatory cytokine, as a model antigen, we engineered a comprehensive library of MS2 bacteriophage virus-like particles (VLPs) displaying peptides representing every possible linear 10-amino acid peptide spanning the 133-amino acid protein. We then used the library to immunize mice, with the hypothesis that a subset of the IL-17A VLP mixture would elicit anti-IL-17A antibody responses. The IL-17A-reactive antibodies elicited by this complex vaccine were then used to select individual recombinant VLPs from the starting library, which were then tested for their ability to elicit anti-IL-17A antibodies in mice. Selected recombinant VLPs were able to elicit anti-IL-17A antibody responses, and antibody levels could be improved by conjugating synthetic peptides representing the selected epitopes to another bacteriophage platform (Qβ VLPs). Overall, this work describes a new method for antigen-agnostic epitope discovery and vaccine development that does not require pre-existing antibodies or rely on computational approaches to identify potential epitopes.

1. INTRODUCTION

The rational design of epitope-targeted vaccines is often a time-consuming, empirical process, especially when the immunogenic regions, or even the native three-dimensional structure, of a target protein are poorly characterized. Even when potentially immunogenic epitopes have been identified, it can be particularly challenging to produce vaccines which display those epitopes in a structural context that mimics the epitope’s native conformation - a critical consideration for vaccines that aim to induce targeted antibodies. Thus, there is a need for innovative strategies to not only systematically identify potential epitopes, but also to readily translate these epitopes into highly immunogenic vaccines. Here, we present a new approach for epitope identification and vaccine development which, unlike traditional phage display techniques, utilizes the immune system to select peptide epitopes that are able to elicit antibody responses that recognize native antigen. This scheme offers a novel framework for identifying epitope-based vaccines which circumvents the need for prior knowledge of complex protein structure or the reliance on computational approaches to identify potential epitopes.

Overexpression of viral structural proteins often results in the spontaneous formation of virus-like particles (VLPs), non-infectious nanoparticles that resemble their parental virus. VLPs have been employed in a variety of applications, but most commonly they have been used directly as vaccines or as vaccine platforms [1]. VLPs make excellent vaccines because of their multivalent, particulate structure, which is highly immunostimulatory, especially for B cells [2, 3]. VLP-based vaccines elicit high-titer and durable antibody responses, often without requiring exogenous adjuvants[4]. We have been particularly interested in utilizing bacteriophage VLPs as a vaccine platform. VLPs derived from single-stranded RNA bacteriophage such as MS2, Qβ, PP7, and AP205 are formed from a single viral structural protein called coat protein [5]. These VLPs are exceptionally thermostable can be rapidly produced at high concentrations in bacterial expression systems. There are numerous techniques for displaying antigens on the surface of bacteriophage VLPs; one approach is to produce recombinant VLPs in which sequences encoding target epitopes are genetically inserted into the bacteriophage coat protein. For example, the coat protein of bacteriophage MS2 has been engineered so that diverse heterologous sequences can be inserted into a site that encodes a beta-hairpin structure which is prominently displayed on the surface of the protein [6]. When overexpressed in bacteria, these recombinant coat proteins self-assemble into VLPs which display the guest peptide in a multivalent format (90 peptides per VLP). Recombinant MS2 VLPs displaying heterologous peptides within the beta-hairpin structure are highly immunogenic and can elicit strong and durable antibody responses [7, 8].

MS2 VLPs can also be employed in a technology analogous to filamentous phage display. Large libraries of VLPs displaying random or defined peptides can be constructed and then screened by affinity selection using antibodies to identify VLPs with specific binding activity. Because MS2 VLPs encapsidate their encoding mRNA [6], selected VLPs can be identified by RT-PCR followed by sequence analysis. We have used this technology to identify the epitopes of monoclonal antibodies (mAbs) and to map the epitopes targeted by a polyclonal antibody response [8–13]. Because displayed peptides are arrayed at high valency on VLPs, a format that is highly immunogenic, selected VLPs can also be directly employed as highly immunogenic vaccines.

Here, we asked whether direct immunization with a diverse pool of MS2 VLPs displaying sequences derived from a model target antigen could be used identify potential vaccine candidates, alleviating the need for pre-existing antibodies. Mouse IL-17A (mIL-17A) , a pro-inflammatory cytokine, was selected as a model antigen because it combines several features that make it well suited for evaluating a novel epitope discovery method. It is a small, structurally well-characterized protein supported by a repertoire of commercially available reagents, including recombinant protein and antibodies. In addition, as an endogenous self-protein, mIL-17A has not been subject to evolutionary pressure to evade adaptive immune responses, allowing epitope accessibility to be evaluated without the confounding effects of immune escape that are common among pathogen-derived antigens. Although immunological tolerance normally limits immune responses to self-antigens like mIL-17A, we have shown that the display of self-antigens at high density on the surface of VLPs can efficiently overcome tolerance mechanisms, allowing the efficient induction of anti-self antibody responses [14, 15].

We engineered a comprehensive library of VLPs displaying every possible linear 10-amino acid peptide (10-mer) derived from the 133-amino acid mature mIL-17A protein. We then used that library to immunize mice, hypothesizing that a subset of the mIL-17A peptide displaying VLPs would elicit anti-mIL-17A antibody responses that could recognize accessible epitopes on the native antigen. Following immunization, native IL-17A-reactive antibodies were purified and then used to select VLPs from the original VLP library to identify the specific peptides responsible for eliciting the anti-mIL-17A antibody response. These peptides were subsequently validated in immunization experiments, confirming their ability to elicit an anti-IL-17A immune response. This work provides proof-of-principle demonstration that a pooled-vaccine strategy is a potentially useful approach for identifying immunogenic epitopes in an antigen-agnostic manner.

2. MATERIALS AND METHODS

2.1. Production of mIL-17A MS2 plasmids and VLP libraries

The MS2 VLP expression plasmid, pDSP62, encoding a single-chain dimer version of the MS2 bacteriophage coat protein, was described previously [9]. To produce a library of plasmids that would allow the expression of MS2 VLPs displaying 10-amino acid sequences from mIL-17A, we designed an oligonucleotide library that consisted of 124 unique forward primers (synthesized by Integrated DNA Technologies; IDT). Forward primers were designed using the following format: 5’-CCCGTCGACAATGGC(XXX)₁₀GGCGACGTGACTGTCGCCCCA-3’, where (XXX)₁₀ represents nucleotides encoding every possible linear 10-amino acid sequence from mature mIL-17A. Each 10-mer sequence differed by two codons—one at the 5’ end and one at the 3’ end—to ensure comprehensive coverage. PCR was performed using the pool of forward primers, the reverse primer E3.2 (5’-CGGGCTTTGTTAGCAGCCGG-3’) and pDSP62 as the template, to produce amplicons encoding overlapping 10-amino acid IL-17A segments spanning the 133-amino acid mature form of mIL17-A. The pool of amplicons was cloned into pDSP62 by virtue of SalI and BamHI restriction sites and then the plasmid library was produced and amplified by electroporation of E. coli strain 10G (Lucigen) followed by amplification in liquid culture. To confirm the integrity of the library, plasmids were isolated from individual transformants and sequenced.

The mIL-17A VLP library was expressed and purified as described previously[13]. Briefly, the plasmid library was used to transform the E. coli expression strain C41(DE3) (Lucigen). Transformed colonies were amplified in liquid culture at 37°C and when the culture reached OD₆₀₀=0.9 protein expression was induced by adding isopropyl-β-D-1-thiogalactopyranoside (IPTG; ThermoFisher Scientific) to a final concentration of 0.4 mM, followed by overnight incubation at 37°C. Bacterial cultures were pelleted by centrifugation, resuspended in lysis buffer (100mM NaCl, 10mM EDTA, 50mM Tris pH8.5, 0.05% deoxycholate) and incubated on ice for 30 minutes. The solution was then subjected to sonication and the resulting lysate was treated with 20 μg/mL DNaseI (Sigma) in the presence of 2.5 mM MgCl₂ and 0.5mM CaCl₂ for one hour on ice. Following centrifugation, the supernatant was subjected to ammonium sulfate precipitation (at 70% saturation), the pellet was collected by centrifugation and then resuspended in column buffer (10mMTris-HCl, pH7.4, 100mM NaCl, 0.1mM MgSO₄, and 0.01mM EDTA). VLPs were then purified by size exclusion chromatography using a hand-packed Sepharose CL-4B column (Sigma-Aldrich). Fractions containing VLPs were confirmed by agarose gel electrophoresis and SDS-PAGE, were pooled and concentrated using ammonium sulfate precipitation, and then dialyzed against PBS (pH 7.4). Endotoxin was depleted from VLP preparations by performing three rounds of Triton X-114 phase separation, as described [16]. Individual recombinant MS2 VLPs were purified using the same protocol as described above.

MS2 VLPs displaying the L9 epitope from the Plasmodium falciparum circumsporozoite protein (CSP) were engineered as described previously [17] and purified as described above.

2.2. Conjugation of peptides to Qβ VLPs

Qβ VLPs were expressed and purified using the same methods that we used to prepare recombinant MS2 VLPs (described above and in [18]). To prepared conjugated Qβ VLPs, peptides representing selected mIL-17A epitopes were synthesized with a C-terminal linker sequence gly-gly-gly-cys. These included peptides representing mIL-17A amino acids 11-20 (PNTEAKDFLQ-GGGC), 21-30 (NVKVNLKVFN-GGGC), and 110-119 (TFRVEKMLVG-GGGC). Each peptide was conjugated separately to the exposed surface lysine residues on Qβ VLPs using the heterobifunctional amine-to-sulfhydryl crosslinker, succinyl 6-[(β-maleimidopropionamido)hexanoate] (SMPH; ThermoFisher Scientific). SMPH was incubated with Qβ VLPs at a molar ratio of 10:1 (SMPH:Qβ coat protein) for 1h at room temperature. Excess SMPH was removed using an Amicon Ultra-4 centrifugal unit with a 100kDa cutoff (Millipore). Peptides were individually added to Qβ VLPs at a molar ratio of 10:1 (peptide:Qβ coat protein) and incubated overnight at 4°C. Conjugation efficiency was measured by SDS-PAGE.

2.3. Immunizations

All animal research complied with and was approved by the Institutional Animal Care and Use Committee at the University of New Mexico School of Medicine (approved protocol #22-201289-HSC). 4- to 6-week-old female Balb/c mice (Jackson Laboratory) were used for all studies. Generally, mice received two intramuscular immunizations 3-4 weeks apart. Sera were collected two weeks after the second immunization. In some experiments sera were also collected at later timepoints to monitor the durability of antibody responses. Mice that were immunized with the mIL-17A VLP library received an initial dose of 5μg or 10μg of VLPs followed by a 10μg or 25μg boost. Mice that were immunized with the Spiked L9 VLP library received two doses of 10μg of the mIL-17A VLP library combined with 78ng of MS2 L9 VLPs. An additional group of control mice received two 5μg immunizations with wild-type MS2 VLPs.

Mice were also immunized with recombinant MS2 VLPs or conjugated Qβ VLPs displaying individual mIL-17A- or hIL-17A-derived peptides. For these experiments, groups of mice received two immunizations three weeks apart with 5μg Qβ or MS2 VLPs displaying mIL-17A peptides, or control wildtype VLPs. Sera was collected two weeks after the second immunization. The group of mice immunized with MS2 mIL-17A(11-20) received two additional doses of vaccine in combination with Cquim-MA adjuvant (2μg/dose; generously provided by ViroVax, LLC).

2.4. Quantitating antibody responses

Specific antibody responses were measured by ELISA using recombinant mIL-17A (SinoBiological), recombinant CSP (generously provided by Gabriel Gutierrez [19]), the mIL-17A VLP library, or wild-type MS2 VLPs as the target antigen. Immulon 2 ELISA plates (ThermoFisher Scientific) were coated with 250ng of target antigen in 50μL PBS and incubated overnight at 4°C. Wells were blocked with PBS-0.5% non-fat dry milk for 1h at room temperature. Pooled sera or sera isolated from individual immunized mice were serially diluted in PBS-0.5% milk and applied to wells overnight at 4°C or for 2h at room temperature. For evaluating the ELISA reactivity of purified anti-mIL-17A IgG, antibody was diluted 1:10. Reactivity was measured by the addition of HRP-labeled goat anti-mouse IgG (Jackson Immunoresearch), diluted 1:4,000 in PBS-0.5% milk, and detected by the addition of TMB substrate (Thermo Scientific). Reactions were stopped using 1% HCl and the optical density of wells was measured at 450nm (OD₄₅₀).

2.5. Purification of mIL-17A specific antibodies

Sera from mice immunized with the mIL-17A VLP library were collected two weeks following the second immunization and pooled. 20μL of sera were mixed with 10μg of biotinylated IL-17A (Kactusbio) in a final volume of 300μL and incubated at room temperature for 2h with gentle shaking. Following this incubation, the sera/biotin-mIL-17A mixture was incubated with 50μL of streptavidin-conjugated magnetic beads (ThermoFisher Scientific) for 1h at room temperature with gentle agitation. Beads were captured by using a magnet and then were washed 3 times with 500μL PBS. Antibodies were eluted off the beads by adding 30μL of 0.1M glycine (pH 2.7) and incubating for 5 min at room temperature. The eluate was then brought to neutral pH by adding 3μL of 1M Tris (pH 9). To confirm that this protocol specifically isolated mIL-17A-reactive antibodies, and not antibodies that were reactive with the MS2 VLP platform, purified IgG was diluted 1:10 and evaluated for binding to the mIL-17A VLP library or wild-type MS2 VLPs by ELISA, using the methods described above.

2.6. Affinity selection of VLPs using mIL-17A specific antibodies

Affinity selection was performed as was previously described [13, 20, 21] with a few modifications. Eluted IL-17A specific IgG (500ng) purified from immune sera (see above) was incubated with 40μg of the IL-17A VLP library in a total volume of 100μL in PBS overnight at 4°C with rotation. Antibody-VLP complexes were then captured using 10μL of Dynabeads Protein G (ThermoFisher Scientific) by incubating for 1h at 4°C with rotation. After incubation, the mixture was briefly spun, placed on a magnetic stand for 2min, and the unbound VLP-containing supernatant was discarded. The VLP-antibody-protein G bead pellet was washed six times with 200μL PBS containing 0.5% Tween, followed by two washes with 200μL PBS. Antibody-bound VLPs were then eluted by adding 50μL 0.1 M glycine (pH 2.7) and incubating for 5 min at RT. The final eluate was then immediately brought to neutral pH by adding 5μL 1M Tris (pH 9).

2.7. Identification of affinity-selected VLPs using RT-PCR

Nucleic acid was recovered from selected VLPs by reverse transcription followed by PCR (RT-PCR). Reverse transcription was performed using the SuperScript First-Strand Synthesis kit (Invitrogen) using ~20% of the affinity-selected VLPs and the E2 primer (5′-TCAGCGGTGGCAGCAGCCAA-3′). Approximately 10% of the resulting cDNA was then amplified by PCR using forward primer 62up (5′-CTATGCAGGGGTTGTTGAAG-3′) and the reverse primer E3.2 (5′-CGGGCTTTGTTAGCAGCCGG-3′) and High Fidelity Platinum Taq polymerase (ThermoFisher Scientific). The PCR product was then re-cloned into the expression vector pDSP62 and individual clones were sequenced.

3. RESULTS

3.1. Generating a mIL-17A VLP library

MS2 VLPs can be engineered to display single defined epitopes [7, 22], random amino acid sequences [9–11], or defined sets of peptide epitopes [20, 21]. Because the goal of this study was to identify potentially immunogenic epitopes from mIL-17A, we engineered an MS2 VLP antigen fragment library that displayed every potential 10 amino acid peptide (124 total) from the sequence of the mature form of mIL-17A (Figure 1; steps 1-5). A 10-mer library were selected because ~90% of random 10-mer peptides inserted into the MS2 coat protein are compatible with VLP assembly [6, 9] and because most linear B cell epitopes are shorter than 10 amino acids long [23]. A custom primer set was designed to include E. coli codon-optimized sequences corresponding to mIL-17A sequences, flanked by a 5’ sequence that included a unique restriction site and a 3’ sequence that would allow the primer to anneal to the MS2 coat protein coding region downstream of the insertion site. This primer pool was used as a forward primer in a polymerase chain reaction (PCR) to amplify mIL-17A containing sequences. The bulk PCR product was then cloned into the MS2 VLP expression vector to generate a mIL-17A VLP plasmid library. As a preliminary validation of this plasmid library, we randomly selected five plasmids for sequence analysis. Each of these five clones contained a unique mIL-17A derived sequence representing amino acids 3-12, 6-15, 33-42, 68-77, and 122-131. The plasmid library was then used to transform E. coli to produce the library of mIL-17A VLPs.

Figure 1. Experimental design.

1 A library of forward primers was designed to encode a series of overlapping 10-mer peptide inserts covering the entire mature mIL-17A sequence (aa 1–133). (2-3) These inserts were amplified by PCR, digested, and cloned into the MS2 single-chain dimer expression vector (pDSP62). (4–5) The resulting plasmid library was transformed into E. coli to produce a diverse MS2 VLP library displaying mIL-17A derived peptides, which was then used to (6) immunize mice. (7) mIL-17A specific antibodies were enriched using biotinylated mIL-17A bound to streptavidin tetramer magnetic beads, and isolated using magnetic separation. Enriched antibodies were then used to capture VLPs displaying mIL-17A peptides. (8) After washing away unbound VLPs, the bound VLPs were collected, and their encapsulated RNA was extracted, reverse-transcribed, and (9) sequenced to identify the selected peptide epitopes. (10) Selected VLPs were then individually expressed.

3.2. Individual components of the library can elicit antibody responses at low doses

The library was predicted to include a maximum of 124 different VLPs displaying every possible linear 10-amino acid from mIL-17A. A low (250ng) dose of recombinant MS2 VLPs can elicit antibody responses [7], but we have not previously shown that immunization with a complex VLP library can elicit antibodies against individual components of the library. To evaluate this, we spiked the mIL-17A VLP library with a small amount of an unrelated VLP, MS2 L9 VLPs, which displays an epitope derived from the Plasmodium falciparum sporozoite surface antigen circumsporozoite protein (CSP) [17]. Because a 10μg dose of the mIL-17A VLP library is calculated to contain ~78 ng of each individual VLP, we spiked this library with an additional 78ng of MS2 L9 VLPs. In mice, immunization with two doses of the L9-Spiked VLP library elicited anti-CSP IgG responses that were markedly higher than in mice immunized with the unspiked library (Figure 2a), indicating that an individual recombinant MS2 VLP is immunogenic when administered both at a low dose and as a component of a large pool of VLPs. Interestingly, anti-CSP antibody responses were still detectable 8 months after the initial immunization (Figure 2a, open circles). Thus, immunization with a single recombinant VLP present at a nanogram dose as part of a complex library was capable of eliciting durable antibody responses. These results indicate that the highly repetitive nature of the VLP surface can compensate for the low abundance of individual library members and demonstrate that individual VLPs are immunogenic even within highly complex mixtures.

Figure 2. Immunogenicity of spiked and unspiked mIL-17A VLP libraries.

A Groups of Balb/c mice (n=3) received two intramuscular immunizations (at weeks 0 and 3) with 10μg mIL-17A VLPs. One group of mice was vaccinated with mIL-17A VLPs that were spiked with 78ng of MS2 L9 VLPs (L9-Spiked VLP library). Anti-CSP IgG responses were measured by ELISA using sera collected two weeks after the second immunization or eight months following vaccination. Results show mean ELISA data, error bars denote standard error of the mean (SEM). B Anti-mIL-17A antibodies in mice immunized with mIL-17A VLPs or, as a control, wildtype MS2 VLPs. Two different dosing regimens were evaluated – mice either received a 5μg prime followed by a 10μg boost (closed pink circles; n=3) or a 10μg prime followed by a 25μg boost (open pink circles; n=3). Sera was collected from mice two weeks following the second immunization and then pooled. Anti-mIL-17A IgG responses in each pool were measured by ELISA.

3.3. Immunization with a library of mIL-17A VLPs elicits weak anti-mIL-17A antibody responses

Next, we determined whether the mIL-17A VLP library could elicit anti-mIL-17A antibody responses in mice. Because we were unsure of the effects of dose on immunogenicity, we utilized two different dosing schedules. One group received an initial (prime) dose of 5μg of VLPs followed by a 10μg boost three weeks later. A second group received a prime of 10μg followed by a 25μg boost. Two weeks following the second immunization, sera from immunized mice were collected, pooled, and anti-mIL-17A antibody responses were measured by ELISA using native, recombinant homodimeric mIL-17A as a target antigen. As is shown in Figure 2b, both the high-dose and intermediate-dose vaccination regimens elicited low-titer anti-mIL-17A IgG antibody responses. Immunization with wild-type MS2 VLPs failed to elicit anti-mIL-17A antibodies. Thus, these data suggest that a subset of VLPs in the library elicited antibodies capable of binding to native mIL-17A.

3.4. Affinity selection of VLPs utilizing vaccine-induced mIL-17A antibodies

Based on the weak anti-mIL-17A responses generated, we hypothesized that only a subset of VLPs in the mIL-17A VLP library elicited mIL-17A-reactive antibodies. To identify these VLPs, mIL-17A-reactive IgG were purified from the sera of immunized mice by using streptavidin magnetic beads that were coated with biotinylated mIL-17A (Figure 1; steps 6-7). To ensure that we did not inadvertently purify antibodies that were reactive against the MS2 VLP platform, eluted antibodies were tested for reactivity to the mIL-17A VLP library or to wild-type MS2 VLPs by ELISA. As is shown in Figure 3a, purified IgG was strongly reactive to the mIL-17A VLP library, but not to wild-type MS2 VLPs.

Figure 3. VLP affinity selection using purified mIL-17A-reactive IgG.

A mIL-17A-reactive IgG were purified from pooled sera of immunized mice, diluted 1:10, and then tested for reactivity to the original mIL-17A VLP library or to wild-type MS2 VLPs by ELISA. B The sequence of mIL-17A with the location of selected mIL-17A epitopes displayed in green, red, or blue. C Each selected epitope is represented by different colors on the structure of the mIL-17A monomer, as predicted by Alphafold (UniProt #: Q62386). Two views of the protein are illustrated.

Although a library of overlapping synthetic mIL-17A peptides could be employed to map the epitopes targeted by anti-mIL-17A antibodies, we took advantage of the preexisting VLP library and our ability to perform affinity selections by directly screening recombinant mIL-17A MS2 VLPs. Purified anti-mIL-17A IgG were used to affinity select VLPs from the original mIL-17A MS2 VLP library. Briefly, purified IgG was bound to an ELISA plate and then the mIL-17A MS2 VLP library was added to the wells. After an incubation period followed by multiple washes to remove unbound VLPs, only the anti-mIL-17A IgG-bound VLPs remained and were eluted. Because MS2 VLPs encapsidate their coding mRNA [6, 9], mIL17-A sequences displayed on selected VLPs could be identified by reverse transcription and PCR (RT-PCR) followed by re-cloning of selected sequences into the expression vector pDSP62. After cloning, ten individual clones were analyzed by Sanger sequencing (Figure 1; steps 8-10). Figures 3b and 3c show the identity of selected peptides (highlighted on the sequence of mature mIL-17A) and their location on the alphafold-predicted [24] structure of the monomeric form of mIL-17A (UniProt: Q62386). Eight of the selectants encoded the same sequence (PNTEAKDFLQ; mature mIL-17A amino acids 11-20, highlighted in green). This epitope is largely located in the disordered N-terminal region of the molecule. The other two selected sequences mapped to amino acids 21-30 (NVKVNLKVFN, red), which encompasses strand 0 of the protein and contains amino acids that are known to contact the receptor IL-17RA, and to amino acids 110-119 (TFRVEKMLVG, blue), which are located on central strand 4 of the protein. Each of these epitopes is predicted to be exposed on the surface of the mIL-17A homodimer [25].

3.5. Immunogenicity of affinity-selected individual mIL-17A MS2 VLPs

To evaluate the immunogenicity of the selected mIL-17A MS2 VLPs, each of the three VLPs were expressed, purified, and used to vaccinate mice. Groups of mice received two immunizations with 5μg of VLPs, without exogenous adjuvant. Because the VLPs displaying mIL-17A amino acids 110-119 [MS2 mIL-17A (110-119)] were challenging to purify, only two mice received this vaccine. Two weeks after the final boost, sera were collected and anti-mIL17A antibody responses were detected by ELISA. As shown in Figure 4a, each of the selected mIL-17A MS2 VLPs elicited mIL-17A specific antibody responses, whereas wild-type MS2 VLPs did not. Anti-mIL-17A antibody levels elicited by MS2 mIL-17A (11-20) (green) and MS2 mIL-17A (110-119) (blue) were weak, whereas MS2 mIL-17A (21-30) (red) elicited stronger responses. To investigate whether combining VLPs with a potent adjuvant could increase antibody responses, mice that received the one of VLPs that elicited weaker responses [MS2 mIL-17A (11-20) VLPs] received two additional boosts of vaccine in combination with the adjuvant Cquim-MA, a dual TLR7/8 adjuvant which we have used previously to improve antibody responses to bacteriophage VLP-based vaccines[18]. As is shown in Figure 4b, booster immunizations with Cquim-MA adjuvanted MS2 mIL-17A (11-20) VLPs resulted in stronger antibody responses against mIL-17A. Taken together, these data show that affinity-selected individual VLPs can elicit anti-mIL-17A antibody responses.

Figure 4. Antibody responses in mice immunized with affinity-selected individual mIL-17A epitope-displaying MS2 VLPs.

A Groups of mice (n=2-5) received two intramuscular doses (at weeks 0 and 3). Two weeks after the second dose sera were collected and anti-mIL-17A antibody responses were measured by ELISA. B Mice that were immunized with MS2 mIL-17A(11-20) received two additional booster doses, adjuvanted with Cquim-MA, at weeks 40 and 44 post-prime. Sera was collected at four weeks following the final boost and anti-mIL-17A antibody levels were evaluated by ELISA and compared to the response after two unadjuvanted doses. Graphs show mean ELISA values at each dilution, error bars represent standard error of the mean (SEM).

3.6. Conjugation of selected epitopes to Qβ VLPs elicit strong anti-mIL-17A responses

One of the major challenges in designing epitope-based vaccines is displaying the target epitope in a context that mimics its structure within the native protein. The epitopes that were identified by affinity selection map to flexible regions of mIL-17A (11-20 & 21-30) or reside within a beta sheet (110-119) (Figure 3C). In contrast, on MS2 VLPs, peptides are displayed in a surface-exposed beta-hairpin structure, suggesting that this loop-like context may not be optimal for eliciting mIL-17A binding antibodies to these specific epitopes. As an alternative approach, peptides representing the three selected epitopes were synthesized and chemically conjugated at high valency to the surface of Qβ bacteriophage VLPs (Figure S1), which presents peptides in unconstrained context. Each peptide was engineered to include a C-terminal gly-gly-gly-cys linker sequence, and then conjugated to the surface of Qβ VLPs by using a chemical crosslinker with sulfhydryl and amine reactive groups [26]. Mice received two immunizations with 5 μg of conjugated Qβ VLPs or wild-type Qβ VLPs, and antibody responses against native mIL-17A were measured by ELISA. As is shown in Figure 5, Qβ VLPs displaying peptides representing mIL-17A (11-20) and (21-30), but not mIL-17A (110-119), elicited high-titer antibodies that bound to native mIL-17A protein. Thus, although the selection scheme failed to directly identify vaccines that elicited strong antibody responses, it was successful at identifying peptide epitopes capable of eliciting strong target-specific antibody responses when displayed on a VLP in a different structural context.

Figure 5. Antibody responses in mice immunized with Qβ VLPs displaying selected mIL-17A peptides.

Groups of mice (n=3-5) received two intramuscular doses (at weeks 0 and 3) of VLPs. Two weeks after the second dose, sera were collected, and mIL-17A antibody responses were measured by ELISA. The graph shows mean ELISA values at each dilution, with error bars representing SEM.

4. Discussion

Identifying the immunogenic regions of a protein remains a major challenge in the development of epitope-based vaccines. In this study, we developed a new strategy that utilizes the immune system to identify peptide epitopes capable of eliciting antibodies that recognize a native target protein. Using mIL-17A as a model antigen, we generated a comprehensive library of MS2 VLPs displaying every possible linear 10-mer derived from the mature protein, immunized mice with the pooled library, and then used affinity-purified anti-mIL-17A antibodies to recover the VLPs responsible for inducing the response. This approach identified three distinct peptide epitopes, each predicted to be accessible on the surface of the native protein. All three recombinant VLPs could elicit antibodies that recognized native mIL-17A, although responses were weak. Nevertheless, these findings provide proof-of-principle that pooled VLP immunization can be used as a strategy for epitope discovery and vaccine identification.

One novel aspect of this study is the use of a VLP library as an immunogen and as an approach, through affinity selection, to identify target epitopes. Although a synthetic peptide library could also be employed for epitope identification, there are a few advantages to performing selections on the VLP library. VLPs display target peptides in a particular conformation that may better preserve local conformations of surface-exposed epitopes than unconstrained synthetic peptides. In addition, because each library member is itself an immunogenic VLP, peptides identified through screening can be advanced directly for immunogenicity testing. Lastly, use of the library avoids the added expense of peptide synthesis.

The weak antibody responses that we observed may reflect the extent to which the displayed peptide adopts a conformation resembling that found in native mIL-17A. Interestingly, when selected epitopes were displayed as unconstrained peptides on Qβ VLPs, rather than within loops on recombinant MS2 VLPs, the two epitopes that mapped to flexible regions of mIL-17A elicited strong antibody responses. Thus, additional optimization of epitope presentation may be required following epitope discovery.

This study has several limitations. First, the library was composed exclusively of linear 10-mer peptides and therefore could not identify conformational or discontinuous epitopes. As a result, potentially important antibody targets that depend on higher-order protein structure would not be represented. Second, epitopes that contain post-translational modifications were not represented in our VLP library. IL-17A is glycosylated at N₄₆; this may explain why we did not select any VLPs that contained sequences near this glycosylation site. Third, because anti-mIL-17A responses generated by library immunization were relatively weak, it is possible that additional immunogenic epitopes were present within the library but were not recovered during affinity selection. This may be a particular problem if a target has a strongly immunodominant epitope. We have previously used deep sequencing-based analysis of affinity selected populations [11, 21, 27]; this analysis tool would likely provide a more comprehensive view of the epitope repertoire identified by this approach. Fourth, this approach could potentially be improved by displaying mIL-17A on VLPs in a different structural context. We have previously shown that diverse peptides can be displayed in an unconstrained structural context at the N-terminus of the MS2 coat protein dimer [7, 22]. Including an N-terminal mIL-17A epitope MS2 VLP library in the immunization and selection phases could potentially expand the utility of this approach. Finally, while the selected vaccines elicited antibodies that bound native mIL-17A, the selection scheme is unable to identify epitopes that induce functional antibodies. Although the functional properties of the mIL-17A antibodies induced in this study were not evaluated, in parallel studies we engineered Qβ VLPs displaying homologous peptides from human IL-17A (hIL-17A). These VLPs elicited anti-hIL-17A reactive antibodies, but these sera failed to block receptor binding. Thus, this vaccine identification scheme must be paired with in vitro and/or in vivo functional studies.

5. Conclusion

The results presented here demonstrate a new framework for epitope discovery and vaccine development. Rather than relying on structural information, computational prediction, or pre-existing monoclonal antibodies, this approach leverages the immune response generated by pooled VLP immunization to identify peptide epitopes capable of eliciting target-reactive antibodies. The ability to move directly from epitope identification to vaccine evaluation is particularly attractive because the selected peptides are already displayed on a highly immunogenic vaccine platform. Although additional optimization may be necessary to maximize immunogenicity, the overall strategy provides a potentially generalizable method for identifying epitope-based vaccine candidates against antigens for which protective or therapeutically relevant epitopes have not yet been defined.

HIGHLIGHTS.

• A library of VLPs displaying every possible 10 amino acid peptide from IL-17A was engineered and used to immunize mice.

• Immunization with the library elicited IL-17A-binding antibodies, which were used to select VLPs from the starting library.

• Immunization with individual VLPs displaying selected epitopes elicited IL-17A antibodies.

Data availability

Data will be made available on request.