Design of multi-epitope vaccine based on G2c genotype PEDV antigen and its immunogenicity and protective effect in piglets

1 Introduction

PEDV was first described in the late 1970s as the etiological agent of porcine epidemic diarrhea (PED), a rapidly spreading enteric disease in swine (1). Clinically, PED is mainly characterized by severe diarrhea, vomiting, and dehydration, with the highest mortality typically observed in suckling piglets (2). Although pigs at different developmental stages can be infected, neonatal animals usually experience the greatest disease burden, which can translate into substantial losses for affected farms (3, 4). Transplacental transmission has been documented for systemic porcine pathogens such as classical swine fever virus (CSFV) (5–7), porcine circovirus type 2 (PCV2) (8), and porcine reproductive and respiratory syndrome virus (PRRSV) (9), whereas PEDV is an enteric pathogen transmitted predominantly through the fecal-oral route. Consequently, the protection of piglets is particularly important. In recent years, the global re-emergence and rapid diversification of highly virulent G2c variant strains have challenged the protective efficacy of conventional inactivated or live-attenuated vaccines, supporting continued development of new immunization strategies.

PEDV is classified as an enveloped, positive-sense, single-stranded RNA virus within the order Nidovirales, family Coronaviridae, and genus Alphacoronavirus. Its genome is approximately 28 kb and contains coding regions for the replicase polyprotein, spike (S), ORF3 accessory protein, envelope (E), membrane (M), and nucleocapsid (N) proteins, flanked by untranslated regions and a poly(A) tail. Functionally, the S protein is central to receptor binding and neutralizing-antibody induction, whereas the M and N proteins participate in virion assembly and contribute to antigenic and broader immune recognition.

Although experimental studies have identified compounds and natural products with anti-PEDV activity, no licensed antiviral drug is currently available for reliable field control of PEDV. Vaccination therefore remains the main practical measure for prevention. Its effectiveness, however, is challenged by rapid PEDV genetic variation and by insufficient mucosal protection from some existing vaccine formulations. Multi-epitope vaccine design provides a way to select conserved immunogenic regions rationally and to adjust antigen composition in a flexible manner (10, 11). Here, we developed and evaluated a genotype G2c PEDV multi-epitope vaccine candidate containing epitopes from the S, M and N proteins.

2 Materials and methods

2.1 Sequence collection and phylogenetic analysis

S1 gene sequences representing 50 PEDV strains with diverse geographic origins and virulence backgrounds were collected from NCBI in March 2023. The full genome of PEDV strain CH/HNLYL/2021 (ON960076.1) was also downloaded from NCBI in FASTA format. The reference strains included in this analysis are listed in Supplementary Table S1. The collected S1 sequences were aligned and analyzed phylogenetically in MEGA 11.0. ClustalW was used for S1-sequence alignment, and phylogenetic relationships were inferred with the neighbor-joining method using 1,000 bootstrap replicates.

2.2 Immunoinformatics-based epitope screening and MEV construction

Linear B-cell epitope candidates were screened from the PEDV S, M, and N proteins with BepiPred-2.0. This platform uses a random-forest framework trained on antibody–antigen structural epitope data (12). In the present analysis, a cutoff of 0.5 was applied, corresponding to a sensitivity of 0.58564 and specificity of 0.57158. The predicted peptides were then assessed for antigenicity using VaxiJen v2.0 and for sequence conservation using the IEDB-AR v2.22 epitope conservancy tool (13, 14). Only highly antigenic epitopes with complete sequence conservation were retained for downstream analysis.

NetMHCpan 4.1 was used to screen CTL-associated peptides for predicted binding to the selected swine leukocyte antigen class I (SLA-I) allele. Strong and weak binders were defined using percentile-rank thresholds of 0.5 and 2%, respectively (15). Candidate HTL epitopes were initially screened with NetMHCIIpan 4.0 against H-2-IAu, H-2-IEd, and H-2-IEk (15, 16). The five HTL epitopes incorporated into the original MEV construct were selected based on the initial MHC-II prediction together with sequence conservation, antigenicity, and predicted cytokine-inducing potential. To further assess their relevance to swine MHC-II, these five previously selected peptides were subsequently evaluated using MixMHC2pred v2.1 (17) against four SLA-DR heterodimers: SLA-DRA01:01 paired with SLA-DRB101:01, SLA-DRB102:01, SLA-DRB104:01, or SLA-DRB1*06:01. This SLA-DR analysis served as a retrospective assessment of the porcine MHC-II binding potential of the five HTL epitopes and did not alter the composition of the experimentally evaluated MEV construct. For each peptide–heterodimer pair, the percentile rank (%Rank) and best-scoring nine-residue core were recorded, with lower percentile ranks indicating stronger predicted binding.

To further characterize functional cytokine potential, IL-10Pred (18), IL-4Pred, IL-2Pred, and IFNepitope (19) were used to predict peptides associated with IL-10, IL-4, IL-2, and IFN-γ induction. Predictions were generated using the default settings of each tool.

The MEV sequence was assembled by arranging the selected CTL, HTL, and dominant B-cell epitopes with appropriate peptide linkers. Porcine β-defensin-1 was attached to the N terminus through an EAAAK linker; KK linkers separated adjacent linear B-cell epitopes; the helical spacer HEYGAEALERAG separated the CTL and HTL regions; GPGPG linkers separated adjacent HTL epitopes; and an RVRR linker preceded the C-terminal His tag (20). This arrangement was selected to maintain separation between functional regions and reduce junction-derived epitopes.

2.3 In silico characterization of the MEV construct

ANTIGENpro, VaxiJen v2.0, and AllerTOP v2.0 were applied to evaluate the predicted antigenicity and allergenicity of the vaccine construct (13, 21, 22). SOLpro was used to estimate solubility (23). Additional physicochemical parameters, including molecular weight, theoretical isoelectric point, instability index, aliphatic index, grand average of hydropathy, and estimated half-life, were calculated with ExPASy ProtParam.

Secondary-structure features of the final MEV were predicted using PSIPRED 4.0 and SOPMA (24, 25). These analyses were used to estimate the proportions of alpha-helix, extended strand, beta-turn, and random coil within the designed construct. The tertiary structure of the final vaccine sequence was generated using Robetta (26). The initial structural model was refined with GalaxyRefine (27), and model quality before and after refinement was assessed using Ramachandran plots from EBI-PDBSum (28). Potential disulfide engineering sites were further examined with Disulfide by Design 2.0 (29), while ProSA-web (30) and SAVES v6.0 were used to evaluate global and stereochemical model quality.

After structural refinement and validation, conformational B-cell epitopes were mapped on the MEV tertiary model using ElliPro (31). As an experimentally determined porcine TLR3 ectodomain structure was unavailable, molecular docking was performed between the MEV construct and the experimentally determined human TLR3 ectodomain (PDB ID: 2A0Z, chain A; 2.40 Å resolution) using PyDockWEB (32, 33). The docking complex was examined with LigPlot+ to identify hydrogen bonds and hydrophobic contacts at the binding interface. A porcine TLR3 ectodomain model (UniProt F1RZN4, residues 28–696) was generated with SWISS-MODEL using 2A0Z chain A as the structural template. Sequence identity, GMQE, QMEANDisCo, and QMEAN4 Z-score were used to assess similarity and model quality (34, 35). C-ImmSim was used to simulate immune responses induced by the final vaccine construct (36). Two immunization events were modeled 7 days apart at time steps 1 and 21 (1 time step = 8 h), while the remaining simulation parameters were kept at default settings.

To improve the predicted expression of the vaccine construct in Escherichia coli, codon optimization was performed using GenScript. The amino acid sequence was back-translated with Gene Infinity (37), and the optimized DNA sequence was evaluated in GenScript for codon adaptation index (CAI) and GC content.

2.4 Recombinant expression, purification, and protein validation

The target sequence was synthesized by GenScript Biotech and inserted into pET-30a(+) for recombinant expression. The pET-30a(+)/MEV plasmid was validated by DNA sequencing and transformed into E. coli Rosetta2(DE3) cells. Recombinant MEV protein was purified by Ni²⁺ affinity chromatography. Recombinant MEV expression and solubility were assessed by 12% SDS-PAGE, and protein identity was confirmed by Western blotting. ECL chemiluminescence was used for membrane development, and signals were captured with Image Lab v4.0.1.

2.5 Piglet vaccination and challenge design

Twenty 5-week-old piglets were obtained from a commercial farm with no known history of PEDV outbreaks or PEDV vaccination. All animals were housed in temperature-controlled isolation rooms and maintained under the same feeding conditions. The experiment was approved by the Ethics Committee of Chengdu University of Traditional Chinese Medicine (approval number: 2024–049) and conducted in accordance with relevant guidelines and regulations. Piglets were randomly allocated to four groups (n = 5 per group) for an initial controlled evaluation of vaccine immunogenicity and protective efficacy. Piglets received the first immunization at 5 weeks of age and were challenged at 10 weeks of age. This schedule allowed completion of the two-dose immunization regimen and development of measurable adaptive immune responses before challenge. The post-weaning challenge model was used to assess vaccine efficacy using immunological, clinical, virological, and survival outcomes, with survival considered as one component of the overall efficacy evaluation. Vaccine doses were selected based on previous PEDV subunit-vaccine studies (38), and the experimental design is summarized in Table 1.

Table 1. Experimental design.

Dpv denotes days post vaccination, and IM denotes intramuscular injection.

Piglets in the control group received 2 mL of sterile PBS by intramuscular injection. Piglets in the vaccine groups received intramuscular administration of 100 μg peptide, 100 μg peptide formulated with ISA 71VG adjuvant (3.7 v/v), or 2 mL of commercial inactivated PEDV vaccine (inactivated genotype G2c strain; pre-inactivation virus titer ≥10⁷ TCID50). The commercial inactivated vaccine was included as a reference vaccine. At 35 dpv, each piglet was orally challenged with 2 × 10⁵ TCID50 of the highly virulent genotype G2c strain PEDV/CH/HN/NK1/2021. Before challenge, rectal swabs were collected from all piglets and tested negative for PEDV by real-time PCR. Clinical signs were monitored for 7 consecutive days after challenge. Fecal consistency was scored as 0, normal; 1, pasty; 2, semiliquid; and 3, liquid (39). Daily fecal samples were collected throughout the observation period for qRT-PCR detection of PEDV RNA. Piglets meeting any humane-endpoint criterion were immediately euthanized. Predefined humane endpoints included severe persistent diarrhea accompanied by marked dehydration, inability to stand or access feed and water, and pronounced loss of responsiveness. Spontaneous death and euthanasia after reaching a predefined PEDV-related humane endpoint were both treated as death events.

2.6 Antibody and virus-neutralization assays

Serum samples were obtained at 21, 28, and 35 dpv, and PEDV-specific IgG and IgA levels were determined by indirect ELISA.

PEDV neutralization assays were used to measure NAb titers against CV777, SCSZ-1, and PEDV/CH/HN/NK1. Vero cells were prepared at 2 × 10⁵ cells/well. Serum samples were serially diluted twofold at eight dilution points and tested in triplicate. Each diluted serum sample was mixed with an equal volume of virus suspension containing 2000 TCID50/mL and incubated at 37 °C for 96 h. Cytopathic effects were recorded, and neutralizing titers were calculated using the Reed–Muench method. Neutralizing-antibody titers were expressed as reciprocal 50% neutralization titers (NT50) and are presented without logarithmic transformation.

2.7 Cytokine detection

Serum IFN-γ, IL-10, IL-2, and IL-4 concentrations were quantified with antigen-capture ELISA kits according to the manufacturer’s protocol (Bio-Swamp Biotech Corporation, China). Each assay was performed in triplicate. Absorbance was measured at 450 nm, and OD450 values were converted to concentrations (pg/mL) using the standard curves supplied with the kits.

2.8 Viral RNA isolation and real-time PCR

Viral RNA was extracted from daily fecal samples with the Virus Genome DNA/RNA Extraction Kit (Dhelix, China) following the manufacturer’s protocol. PEDV N-gene RNA was detected by qRT-PCR as described previously (40). The raw Ct values obtained from qRT-PCR were used for analysis without normalization to a reference gene. Lower Ct values corresponded to higher levels of PEDV RNA.

2.9 Statistical methods

Data were analyzed in GraphPad Prism. Results are expressed as mean ± standard deviation (SD). Repeated-measures data, including antibody levels, cytokine concentrations, neutralizing-antibody titers, body-weight changes, and fecal qRT-PCR Ct values, were analyzed using a two-way mixed-effects model followed by Dunnett’s multiple-comparisons test against the PBS group. Diarrhea scores were compared using the Kruskal–Wallis test followed by Dunn’s multiple-comparisons test. Statistical significance was defined as *p < 0.05, **p < 0.01, and ***p < 0.001. Given the group size of five piglets, survival was summarized descriptively as exact counts and percentages, and no inferential between-group survival test was performed. The repeated-measures and nonparametric analyses described above were applied only to their corresponding outcomes.

3 Results

3.1 Phylogenetic background and MEV epitope design

S1-based phylogenetic analysis (Figure 1) resolved the 50 PEDV reference strains into three major lineages: S-INDEL, G1, and G2. The G1 lineage contained the classical G1a and G1b subgroups, whereas variant strains were distributed among G2a, G2b, and G2c within the G2 lineage. Most strains reported between 2010 and 2023 clustered with G2, consistent with the current predominance of variant PEDV strains. CH/HNLYL/2021 fell within the G2c branch and was closely related to recent Chinese epidemic isolates, supporting its use as a representative circulating field strain.

BepiPred 2.0 identified eight fully conserved linear B-cell epitopes from the PEDV S, M, and N proteins. The selected CTL epitope AVDGGDTAV was fully conserved, showed predicted binding to SLA-1*04:01, and had a VaxiJen score of 0.4298 (Table 2). Retrospective SLA-DR reassessment of the five HTL epitopes incorporated into the MEV revealed different predicted binding patterns across the tested porcine MHC-II heterodimers (Table 3). TTTIVYGRVGRSVNA showed %Rank ≤ 10 for all four SLA-DR heterodimers, followed by SIGYVPSQSGQVKIA (3/4), RDDLVAAVKDALKS (2/4), and QIRWRMRRGERIEQP (1/4). In contrast, EGVFWVAKEGAKTEP showed %Rank >10 for all four tested heterodimers. Overall, four of the five HTL epitopes showed predicted binding to at least one of the tested SLA-DR heterodimers, whereas EGVFWVAKEGAKTEP did not reach the %Rank ≤ 10 threshold within the four-heterodimer panel evaluated in this study. Predicted IFN-γ, IL-10, IL-2, and IL-4 induction and sequence conservation are summarized in Table 2.

Table 2. Summary of selected epitopes incorporated into the MEV construct.

P, predicted positive; N, predicted negative. H-2 class II alleles indicate the original MHC-II screening context used for HTL epitope selection. The five HTL epitopes were subsequently reassessed against porcine SLA-DR heterodimers, and the detailed results are presented in Table 3.

Table 3. Predicted binding of the five selected HTL epitopes to four porcine SLA-DR heterodimers.

SLA-DRA*01:01 was paired with each indicated SLA-DRB1 allele. Values are MixMHC2pred percentile ranks (%Rank), with lower values indicating stronger predicted binding. Coverage indicates the number of tested SLA-DR heterodimers with %Rank ≤ 10. A %Rank ≤ 10 cutoff was used to summarize predicted SLA-DR coverage.

Based on the dominant epitopes identified through immunoinformatics screening, a chimeric MEV was constructed with 8 B-cell epitopes, 1 CTL epitope, and 5 HTL epitopes. PBD-1 was attached to the N terminus through an EAAAK linker. KK linkers separated adjacent B-cell epitopes, HEYGAEALERAG separated the CTL and HTL regions, GPGPG linkers separated adjacent HTL epitopes, and an RVRR linker preceded the C-terminal His tag. The construct organization is shown in Figure 2A.

3.2 In silico characterization and structural evaluation of the MEV

ANTIGENpro and VaxiJen v2.0 yielded antigenicity scores of 0.716959 and 0.6860, indicating that the construct had predicted antigenic potential. AllerTOP v2.0 classified the MEV as non-allergenic, and SOLpro predicted a solubility probability of 0.769879, which exceeded the standard cutoff. ProtParam analysis showed that the construct comprised 285 amino acids, with a molecular weight of 30.92 kDa, theoretical pI of 10.36, instability index of 28.96, aliphatic index of 58.84, and GRAVY value of −0.830 (Supplementary Table S2). Together, these parameters support the stability and hydrophilicity of the designed recombinant antigen.

PSIPRED and SOPMA predicted that the MEV contained 23.16% alpha-helix, 24.21% extended strand, 11.23% beta-turn, and 41.40% random coil. Solvent-accessibility analysis indicated that most residues were exposed or moderately exposed, whereas only a small fraction was buried or assigned to disordered regions. This distribution suggests that the construct has favorable antigen accessibility. The optimal MEV tertiary model was generated by Robetta and visualized with UCSF ChimeraX (Figure 2B).

Following GalaxyRefine optimization, the refined model showed improved geometry relative to the initial model (Supplementary Figure S3). The proportion of residues located in the most favored regions of the Ramachandran plot increased from 84.6 to 86.4% after refinement. ProSA-web produced a z-score of −6.37, and the predominantly negative energy profile indicated an acceptable global fold. SAVES v6.0 further supported model quality, with 80.70% of residues showing an average 3D-1D score ≥ 0.1 and an overall quality factor of 87.68. These results indicated that the refined tertiary model was suitable for subsequent epitope mapping and docking analysis. ElliPro identified 6 discontinuous B-cell epitopes on the MEV tertiary model. These epitopes contained 3 to 60 residues and had scores ranging from 0.628 to 0.779 (Supplementary Figure S4; Supplementary Table S3). Because each score exceeded the 0.5 threshold, the construct was predicted to have favorable conformational antigenicity.

Docking of the MEV with the experimentally determined human TLR3 ectodomain generated 101 candidate models. Model 528 ranked first and was selected for interface analysis. Hydrogen bonds were predicted with Glu146, His283, His284, and Arg22, while hydrophobic contacts involved Arg41, His285, Arg43, Lys44, Arg48, Gly20, Met142, Gly42, Gln84, Gly36, and Lys182 (Figure 3). For comparison with the porcine receptor, a porcine TLR3 ectodomain model was generated using human TLR3 (2A0Z) as the template. The porcine sequence shared 83.11% identity with the human template, and the model had a GMQE of 0.649, QMEANDisCo of 0.805 ± 0.053, and QMEAN4 Z-score of −1.008. These results support conservation of the overall TLR3 ectodomain architecture between the two species.

3.3 Immune-response simulation

C-ImmSim predicted a stronger secondary immune response than the primary response after vaccination. As antigen levels declined, immunoglobulin levels increased, including IgG1 + IgG2, IgM, and combined IgG + IgM. The simulation also indicated sustained B-cell activation, robust helper and cytotoxic T-cell responses, coordinated activation of natural killer cells, dendritic cells, and macrophages, and increased IFN-γ and IL-2. Overall, the predicted immune profile suggests that the MEV can induce both humoral and cellular immunity (Figure 4).

3.4 Recombinant expression and validation of the MEV protein

The vaccine construct was reverse-translated into an 855-nucleotide cDNA sequence using Gene Infinity. For E. coli expression, GenScript predicted a CAI value of 1.00 and GC content of 61.78%, indicating favorable codon usage and nucleotide composition for the selected host.

GenScript Biotech synthesized the target sequence and inserted it into pET-30a(+) to generate pET-30a(+)/MEV, which was verified by DNA sequencing (Figure 5A). The recombinant plasmid was introduced into E. coli Rosetta2(DE3) cells, and the expressed rotein was purified using Ni²⁺ affinity chromatography.

Recombinant MEV expression and solubility were evaluated by SDS-PAGE (Figure 5B), and protein identity was confirmed by Western blotting (Figure 5C). SDS-PAGE showed a clear band at approximately 32 kDa, consistent with the expected size of the construct. Western blotting with an anti-His monoclonal antibody confirmed the recombinant protein, and predominant detection in the soluble fraction indicated successful soluble expression in E. coli.

3.5 Antibody responses

Serum IgG and IgA were measured by ELISA to evaluate humoral immunogenicity. At 21, 28, and 35 dpv, all vaccinated groups showed markedly higher PEDV-specific IgG levels than the PBS group (Figure 6A). The SMN peptide–ISA 71VG and commercial inactivated-vaccine groups showed relatively high IgG responses, while the non-adjuvanted SMN peptide group also displayed a sustained increase over time.

Serum IgA followed a comparable trend. Relative to PBS controls, piglets receiving SMN peptide, SMN peptide–ISA 71VG, or the commercial inactivated vaccine had clearly increased serum IgA levels from 21 to 35 dpv (Figure 6B). The SMN peptide–ISA 71VG group showed numerically higher serum IgA levels at several time points, consistent with an enhanced humoral response after adjuvant formulation.

3.6 Virus-neutralization analysis

The neutralizing activity elicited by the SMN peptide vaccine was assessed against three PEDV strains: CV777, SCSZ-1, and CH-HN-NK1.

Virus neutralization analysis detected reciprocal NT50 responses in all vaccinated groups, whereas the PBS group remained negative (Figures 6C–E). Against the G1a strain CV777, both peptide-immunized groups showed higher numerical titers than the commercial inactivated-vaccine group. For the homologous G2c strain CH-HN-NK1, the SMN peptide–ISA 71VG group showed the highest numerical response, followed by the non-adjuvanted SMN peptide group. In contrast, the commercial inactivated vaccine showed the highest numerical titers against the G2b strain SCSZ-1. Overall, the peptide vaccine induced neutralizing activity against all three tested PEDV isolates.

3.7 Cytokine-response analysis

To characterize systemic cytokine responses, serum IL-2, IL-4, IL-10, and IFN-γ were measured by ELISA. Cytokine concentrations were elevated in all vaccinated groups compared with the PBS group across the tested time points (Figure 7). The simultaneous increases in IFN-γ and IL-2, together with IL-4 and IL-10, indicate that vaccination induced a mixed Th1- and Th2-related cytokine response. The SMN peptide–ISA 71VG group showed numerically higher concentrations for most measured cytokines.

3.8 Clinical outcomes after PEDV challenge

Protective efficacy was evaluated by orally challenging the four groups at 35 dpv with 2 × 10⁵ TCID50 of the highly virulent genotype G2c PEDV strain PEDV/CH/HN/NK1/2021 (Figure 8). Before challenge, all piglets were clinically healthy and tested negative for PEDV by real-time PCR. After challenge, the PBS group developed disease earlier and more severely than the vaccinated groups, whereas disease onset was delayed in the SMN peptide and SMN peptide–ISA 71VG groups. During the 7-day observation period, all piglets survived in the SMN peptide group (5/5) and the SMN peptide–ISA 71VG group (5/5), compared with three of five piglets in the commercial inactivated-vaccine group and none of the five piglets in the PBS group. The SMN peptide–ISA 71VG group showed relatively stable daily weight gain, while PBS controls exhibited the greatest weight loss. Fecal Ct values varied during the observation period. Compared with the PBS group, vaccinated groups showed higher Ct values at several time points, corresponding to lower levels of detectable PEDV RNA during these periods. Together, the clinical, survival, and virological findings show that both peptide formulations improved outcomes following G2c PEDV challenge.

4 Discussion

This study combined immunoinformatics-guided antigen design with piglet-based validation to develop a multi-epitope vaccine candidate targeting a PEDV G2c strain. The construct incorporated conserved epitopes from the S, M, and N proteins, together with N-terminal PBD-1 as an immunostimulatory component. The inclusion of epitopes from multiple structural proteins was intended to broaden antigenic coverage and combine humoral and T-cell-related targets within a single recombinant antigen. The final construct showed favorable predicted antigenicity, solubility, and structural properties, supporting its further experimental evaluation.

The retrospective SLA-DR analysis provided additional information on the predicted porcine MHC-II relevance of the five HTL epitopes incorporated into the construct. TTTIVYGRVGRSVNA showed predicted binding across all four tested SLA-DR heterodimers, while SIGYVPSQSGQVKIA, RDDLVAAVKDALKS, and QIRWRMRRGERIEQP showed allele-dependent binding patterns. In contrast, EGVFWVAKEGAKTEP did not meet the %Rank ≤ 10 criterion for any of the four tested heterodimers. The CTL epitope AVDGGDTAV was selected by considering its complete sequence conservation and predicted binding to SLA-1*04:01 together with its VaxiJen score of 0.4298. These results support considering MHC binding, sequence conservation, and antigenicity together during epitope evaluation. SLA-based prediction tools such as PigMatrix provide complementary approaches for assessing swine T-cell epitopes (41). Because of SLA diversity and the limited four-heterodimer panel examined here, and because the experimental piglets were not SLA genotyped, broader SLA-II coverage and experimental peptide–SLA-II binding should be evaluated in future studies (42).

Because an experimentally determined porcine TLR3 ectodomain structure was unavailable, the human TLR3 ectodomain structure (PDB ID: 2A0Z) was used as a structural reference and as the template for construction of a porcine TLR3 homology model. Docking with 2A0Z suggested a potential MEV–TLR3 interaction surface. The 83.11% human–porcine sequence identity and the overall similarity of the porcine homology model provided structural support for this analysis. However, species-specific differences at the local binding interface may affect the predicted interaction. Therefore, binding to and activation of porcine TLR3 require further experimental validation.

Vaccination induced PEDV-specific serum IgG and IgA, increased serum cytokine concentrations, and elicited neutralizing activity against all three PEDV isolates tested. The simultaneous increases in IFN-γ and IL-2 together with IL-4 and IL-10 indicate a mixed Th1- and Th2-related systemic cytokine response. These findings support the systemic immunogenicity of the MEV formulations. Because IgA was measured in serum, however, the present results do not directly demonstrate intestinal mucosal immunity. Further studies examining intestinal IgA, other mucosal responses, and antigen-specific T-cell responses would help clarify the mechanisms contributing to protection against PEDV infection.

The challenge was performed at 10 weeks of age to allow completion of the two-dose immunization regimen and development of measurable adaptive immunity before challenge. Thus, this post-weaning model was intended to compare vaccine-associated immunological, clinical, virological, and survival outcomes rather than to reproduce the high mortality observed in neonatal PEDV infection. The study included five piglets per group and used a single G2c challenge strain; therefore, the magnitude and reproducibility of protection should be confirmed in larger studies. Further work should also examine long-term immunity, optimal dosing, mucosal responses, and protection against genetically diverse PEDV field strains. Overall, the present findings support further development of the multi-epitope vaccine strategy for PEDV.

Generative AI statement

The author(s) declared that Generative AI was used in the creation of this manuscript. During this work, the author used ChatGPT-5.5 in order to improve language only. After using the tool, the author reviewed and edited the content and assumes full responsibility for the manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fvets.2026.1942409/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Abel Gutiérrez-Ortega, CONACYT Centro de Investigación y Asistencia en Tecnología y Diseño del Estado de Jalisco (CIATEJ), Mexico

Reviewed by: Md Aminul Islam, Harvard University, United States

Subrat Kumar Swain, Siksha O Anusandhan University, India