Neutralizing epitope mapping for deltacoronavirus receptor-binding domain and seroepidemiological survey across different animal species

ABSTRACT

Porcine deltacoronavirus (PDCoV) is an emerging enteric pathogen that poses a significant threat to the global swine industry and carries a potential public health risk. The development of neutralizing monoclonal antibodies (mAbs) is crucial for preventing and controlling PDCoV. However, research on neutralizing linear epitopes within the PDCoV receptor-binding domain (RBD) and the establishment of serological assays based on such mAbs remains limited. In this study, we generated two neutralizing mAbs, 1C12 and 3A11, targeting the PDCoV RBD. Epitope mapping revealed that 1C12 recognizes a linear epitope (³¹⁶DFGEARLD³²³), whereas 3A11 binds to a conformational epitope. Further residue analysis identified Arg321 as the most critical residue for binding to 1C12. Additionally, the integrity of the entire ³²¹RLD³²³ motif was found to be indispensable for the conformational recognition by mAb 3A11. Using these mAbs, we developed two blocking enzyme-linked immunosorbent assay (bELISA) methods (bELISA-1C12 and bELISA-3A11). Evaluation of 150 clinical samples demonstrated that bELISA-1C12 exhibited 93.3% concordance with a virus neutralization test (VNT), indicating higher diagnostic sensitivity than bELISA-3A11 (92.7%). A large-scale serosurvey of pig populations across ten Chinese provinces in 2025, conducted using bELISA-1C12, revealed a PDCoV seropositivity rate of 13.6%. Notably, when extending the serological screening to 21 additional animal species, we detected PDCoV antibodies in peafowl serum for the first time, suggesting possible exposure to the virus. This study identifies key antigenic epitopes on the PDCoV RBD and provides valuable tools for epidemiological surveillance and assessing the transmission risk of PDCoV.

Introduction

The zoonotic transmission of viruses from animal reservoirs to humans represents a significant threat to global public health [1–3]. Coronaviruses (CoVs), classified into four genera (Alphacoronavirus, Betacoronavirus, Gammacoronavirus, and Deltacoronavirus; α-CoV, β-CoV, γ-CoV, δ-CoV) based on phylogenetic traits [4], can infect a wide range of animals and humans, leading to respiratory and gastrointestinal diseases [5–7]. Among these, PDCoV is the only known δ-CoV that infects swine, alongside several α- and β-CoVs [8]. Since its initial identification in Hong Kong in 2012 [9] and its subsequent association with a diarrheal outbreak in U.S. swine in 2014 [10], PDCoV has been reported globally, becoming endemic in many pig populations including Thailand, Vietnam, and Lao PDR [11], China [12], Canada [13], South Korea [14], Japan [15], Peru [16], and Mexico [17]. Notably, PDCoV exhibits a broad host range, as it has been demonstrated to infect not only pigs and humans [7,10], but also chickens, calves, mice, ducks, geese and ferrets [18–22]. A critical study further revealed that the PDCoV receptor-binding domain (RBD) can interact with aminopeptidase N (APN) from eight out of seventeen tested species, including primates (human and rhesus macaque), carnivores (dog), rodents (mouse and rat), artiodactyls (pig and Arabian camel), and galliformes (chicken). It was further demonstrated that APN from all these binding species is functional in mediating cellular entry, underscoring a substantial cross-species transmission risk [8].

PDCoV is an enveloped virus with a single-stranded, positive-sense RNA genome of approximately 25 kb, encoding four major structural proteins: spike (S), envelope (E), membrane (M), and nucleocapsid (N) [4,23]. The S protein, a type I transmembrane glycoprotein, forms homotrimers that are cleaved into an N-terminal S1 subunit containing the RBD responsible for host cell attachment and cross-species transmission, and a C-terminal S2 subunit that mediates membrane fusion [24–26]. Given this essential role, the RBD represents an ideal target for developing therapeutic antibodies and serological diagnostics. This crucial function of the RBD is exemplified by neutralizing mAbs that block receptor interaction by binding to the “receptor-binding loop” at epitopes overlapping the APN binding site [25,27,28]. Most reported neutralizing antibodies recognize conformational epitopes [20,25,28–30]; yet, the linear B-cell epitopes within the RBD, which are crucial for designing peptide-based vaccines and immunodiagnostic reagents, remain largely unexplored.

Serological detection of PDCoV relies on various methods, including the indirect immunofluorescence assay (IFA) [31,32], the virus neutralization test (VNT) [4], the fluorescent focus neutralization test (FFNT) [4], and the fluorescent microsphere immunoassay (FMIA) [32]. The operational simplicity of the ELISA makes it favored for high-throughput testing. It has led to the development of multiple formats, such as indirect ELISAs using recombinant N [32,33], S1 [34,35], or M [36] protein, and blocking ELISAs (bELISAs) using recombinant S [37] or N [38] protein. It is noteworthy that the development of bELISAs for PDCoV antibody detection using neutralizing mAbs has been rarely reported.

Neutralizing antibody titers, a key component of the humoral immune response, are considered a reliable correlate of protection for evaluating vaccine efficacy [39,40]. However, despite this, the sensitivity of most existing PDCoV ELISAs remains benchmarked against IFA rather than virus neutralization assays [34,37,38]. Although our recent studies have systematically evaluated the correlation between indirect ELISAs using various structural proteins and neutralization assays [41], several critical questions regarding epitope-specific bELISAs remain unresolved. Specifically, it is unclear whether bELISAs based on neutralizing mAbs targeting distinct epitopes yield consistent antibody kinetic profiles and, more importantly, whether these profiles correlate with the accurate kinetics of neutralizing antibodies measured by VNT. Furthermore, the diagnostic concordance between such bELISAs and the VNT has not been systematically evaluated. Given the zoonotic threat of PDCoV, there is an urgent need to develop a broadly applicable blocking ELISA that is not limited by host species, which would greatly facilitate multispecies serological surveillance and control of outbreaks. Therefore, the development of bELISAs using neutralizing mAbs to detect antibodies targeting critical neutralizing epitopes holds significant value for accurately assessing vaccine-induced immunity and conducting effective seroepidemiological monitoring.

Here, we generated two mAbs targeting the RBD that demonstrated both neutralizing and blocking activities. Furthermore, we mapped the epitopes recognized by these mAbs and employed a trained prediction model, followed by experimental validation, to identify the key antigenic residues. Using these well-characterized mAbs, we subsequently developed two blocking ELISAs, respectively. We systematically evaluated these assays against a VNT using a diverse panel of samples, including clinical sera, colostrum, and serial samples from vaccinated pigs, with a focus on specificity, antibody kinetics, and diagnostic sensitivity. Finally, we selected the superior bELISA-1C12 for extensive serosurveillance across a broad range of animal species.

Materials and methods

Viruses, cells, and serum samples

Three PDCoV strains were utilized in this study. Strains GDSG10-2023 and SCNJ10-2021 were obtained from the National Research Center for Veterinary Medicine. Professor Yaowei Huang from South China Agricultural University kindly provided the PDCoV-HZYH-2019 strain [29].

ExpiCHO-S cells and corresponding expression medium, as well as the transfection reagent, were sourced from Thermo Fisher Scientific (Waltham, MA, USA). The cells were maintained at 32°C with 5% CO₂.

HEK-293T and LLC-PK1 cells were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM; Gibco, Langley, OK, USA) supplemented with 10% fetal bovine serum (FBS; Biological Industries, Beit HaEmek, Israel) at 37°C with 5% CO₂.

A panel of serum and colostrum samples was utilized in this study. The panel included 170 animal experiment serum samples (81 positive for PDCoV infection and 89 non-infected), 75 clinical pig sera, and 75 sow colostrum samples, which were provided by the National Research Center for Veterinary Medicine. To further validate the applicability of the developed blocking ELISAs for milk samples, five PDCoV antibody-positive colostrum samples were obtained from immunized sows, kindly provided by Dr. Bin Li of the Jiangsu Academy of Agricultural Sciences [39]. The neutralizing antibody status of these samples had been pre-determined.

Standard positive control sera against porcine circovirus 2 (PCV2), porcine circovirus 3 (PCV3), porcine reproductive and respiratory syndrome virus (PRRSV), pseudorabies virus (PRV), porcine rotavirus (PoRV), classical swine fever virus (CSFV), and porcine epidemic diarrhea virus (PEDV) were obtained from Beijing Sino-science Gene Technology Co. Ltd. (Beijing, China). African swine fever virus (ASFV) antibody-positive serum was acquired from the China Institute of Veterinary Drug Control (Beijing, China). Commercial ELISA kits for detecting PEDV-specific IgG and IgA antibodies were obtained from Luoyang Putai Biotechnology Co., Ltd. (Luoyang, China).

Clinical samples including pig serum samples (n = 2002), dog serum samples (n = 115), and serum samples from other species including peafowls, Kunming mice, bamboo rats, foxes, ferrets, tigers, rhinoceroses, alpacas, pangolins, horses, steppe eagles, lemur variegatus, masked palm civets, bears, pandas, yellow-throated martens, weasels, leopard cats, porcupines, and boars were provided by Luoyang Putai Biotechnology Co., Ltd., Yangzhou University, and National Research Center for Veterinary Medicine, respectively [42].

Antibodies

Monoclonal antibodies (mAbs) against PDCoV were generated using purified PDCoV GDSG10-2023 virus particles as the immunogen. Briefly, following established procedures [29,43], immunized five 6-week-old female BALB/c mice were bled every two weeks, and serum antibody titers against the PDCoV S protein were monitored using an indirect ELISA. The mouse exhibiting the highest titer was selected for splenocyte isolation and fusion with myeloma cells to generate hybridomas. The resulting hybridomas, which secreted specific anti-PDCoV antibodies, were subsequently subjected to three rounds of subcloning via limiting dilution. MAbs were produced in large quantities via the ascites method using five parous female BALB/c mice. The mice were primed with sterile liquid paraffin, and approximately 5 × 10⁶ hybridoma cells were injected intraperitoneally 7 days later. Ascitic fluid was harvested 7 − 10 days post-injection. MAbs were subsequently purified from ascites fluid using Protein A beads (TransGen Biotech, Beijing, China). The subclass and type of mAbs were determined using a mouse monoclonal antibody isotyping kit (Bai AoTong, Luoyang, China). Subsequently, the antibodies were conjugated with horseradish peroxidase (HRP) using a commercial kit (Biodragon, Suzhou, China). Both procedures were performed according to the manufacturer’s instructions.

Female BALB/c mice were obtained from Vital River Laboratory Animal Technology Co., Ltd. (Zhejiang, China). During the current study, mice received access to food and water ad libitum. The animal procedures in this study were approved by the Animal Care and Ethics Committee of Luoyang Putai Biotechnology Co., Ltd. (Approval No. PT20240003) and were conducted in accordance with the Guide for the Care and Use of Laboratory Animals. At the experimental endpoint, all immunized mice were humanely euthanized. Specifically, euthanasia was performed in accordance with the American Veterinary Medical Association (AVMA) Guidelines. Mice were first placed in a pre-filled chamber and exposed to a gradually increasing concentration of carbon dioxide (CO₂) at a displacement rate of 30 − 70% of the chamber volume per minute until deep anesthesia was achieved (as indicated by the loss of righting reflex and response to toe pinch). Then, cervical dislocation was performed promptly by a trained and skilled personnel. Death was confirmed by observing the cessation of respiration and the absence of a heartbeat.

The following commercial antibodies were used in this study: an anti – His monoclonal antibody (CWbio, Beijing, China); a fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse IgG (H + L), and an HRP-conjugated goat anti-mouse IgG (both from Invitrogen, Waltham, MA, USA).

Plasmid construction

The coding sequences for the S protein ectodomain and its S1 and receptor-binding domain (RBD) fragments were designed based on the reported PDCoV S protein structure [24,39]. The corresponding recombinant plasmids, with a signal peptide for secretory expression in the pcDNA3.1(+)-His vector, have been described previously [41].

To map the epitopes recognized by the mAbs on the RBD sequence, two sets of constructs were generated. First, a series of truncated RBD fragments was amplified with specific primers and cloned into the pEGFP-C1 vector. Second, RBD point mutants with substitutions at key amino acid residues were engineered using the pCDNA3.1-RBD plasmid as a template. All constructs were prepared for subsequent expression and analysis. All primers used for cloning are listed in Table S1.

Protein expression in ExpiCHO-S cells and purification

The recombinant S protein ectodomain, S1 subunit, and RBD were expressed and purified as previously described [41]. Briefly, the corresponding plasmids were transfected into ExpiCHO-S cells, and the supernatants were harvested after 10–12 days. The target proteins were purified from supernatants using Ni-NTA agarose resin (GenScript USA Inc., Piscataway, NJ, USA), followed by buffer exchange and concentration determination using a BCA assay kit (Invitrogen, Waltham, MA, USA).

SDS-PAGE and Western blot analysis

Purified proteins or cell lysates were separated on 10% SDS-PAGE gels and transferred to polyvinylidene fluoride (PVDF) membranes (Millipore, Darmstadt, Germany). The membranes were blocked with 5% bovine serum albumin (BSA; Solarbio, Beijing, China) for 2 hours at room temperature and then incubated with an anti-His monoclonal antibody (diluted 1: 2000) for 2 hours. After incubation with an HRP-conjugated goat anti-mouse IgG (H + L) secondary antibody (diluted 1: 10,000) for 1 hour at room temperature, the immunoreactive bands were visualized. Prestained protein markers (MP102, Vazyme Biotech Co., Ltd, Nanjing, China) were used as references for comparing immunoblot band positions.

Virus neutralization test

The neutralizing antibody titer against PDCoV was determined in LLC-PK1 cells using a fixed virus-diluted serum approach. Before the assay, serum and milk samples were heat-inactivated at 56°C for 30 minutes and 60°C for 45 minutes, respectively. Each sample was subjected to a two-fold serial dilution (from 1:2 to 1:256) in DMEM supplemented with 10 μg/mL trypsin (Invitrogen, Waltham, MA, USA), with four replicates for each dilution. An equal volume of PDCoV suspension containing 200 50% tissue culture infective dose (TCID₅₀) was added to the diluted samples. After gentle mixing, the virus-serum mixtures were incubated at 37°C under 5% CO₂ for 1 hour.

Subsequently, 100 μL of each mixture was inoculated onto LLC-PK1 cell monolayers in 96-well plates. The plates were incubated at 37°C with 5% CO₂ for 72 hours. Following incubation, the medium was aspirated, and the cells were washed three times with PBS. They were then fixed with 4% paraformaldehyde (Solarbio, Beijing, China) for 15 minutes at room temperature. IFA assessed viral neutralization with a mAb (1G12) provided by the National Research Center for Veterinary Medicine.

Indirect immunofluorescence assay (IFA)

Cells were fixed with 4% paraformaldehyde (Solarbio, Beijing, China) for 15 min and subsequently permeabilized using 0.5% Triton X-100 (Solarbio, Beijing, China) for 10 min. After blocking with 1% BSA in PBS for 1 hour, the cells were incubated with the primary antibody, a PDCoV-specific mAb 1G12 (diluted 1:1000), for 1 hour at 37°C. Following three washes with PBS, the cells were probed with a FITC-conjugated goat anti-mouse IgG (H + L) secondary antibody (diluted 1:1000) for 1 h at 37°C. Finally, fluorescence images were captured using an inverted fluorescence microscope.

Development of indirect ELISAs

To identify the target protein domain recognized by the mAbs, parallel indirect ELISAs were established separately using the purified recombinant PDCoV S, S1, and RBD proteins. Checkerboard titration was used to optimize assay conditions, with coating concentrations set at 0.1 μg/mL and 2 μg/mL, mAbs diluted at 1:50 and 1:100, and an HRP-conjugated goat anti-mouse secondary antibody diluted at 1:10,000. Coating and blocking conditions were performed at 4°C for 24 hours. Both primary and secondary antibody incubations were carried out at 37°C for 30 min, followed by 3,3′,5,5′-tetramethylbenzidine (TMB) substrate development at 37°C for 15 min. The reaction was stopped by incubation with 50 µl of 2 M H₂SO₄ at 37°C for 15 min. The absorbance was then measured at 450 nm using an ELISA microplate reader (BioTek, VT, USA). Based on the optimization results, the final selected conditions were a protein coating concentration of 2 μg/mL, mAb ascites dilution of 1:100, and secondary antibody dilution of 1:10,000, establishing an assay for the initial characterization of mAb specificity.

Development of PDCoV RBD-based blocking ELISAs

Following the confirmation that mAbs bound explicitly to the RBD domain by iELISA (Development of indirect ELISAs), we developed two blocking ELISAs (bELISAs) based on the RBD protein. The assay conditions were optimized through a checkerboard titration using four PDCoV antibody-positive and four negative serum samples. The final reaction conditions were selected based on the Negative-to-Positive (N/P) value, alongside a clear distinction between negative and positive controls. Key reaction conditions screened included antigen coating concentration (0.05, 0.1, and 0.2 μg/mL), blocking conditions (4°C for 24 h), serum dilution ratio (1:1), serum incubation time (30 min and 60 min at 37°C), dilution of HRP-conjugated mAb (1:10,000, 1:30,000, 1:50,000, and 1:80,000; 30 min at 37°C), and TMB development time (15 min at 37°C). The reaction was terminated with 50 µl of 2 M H₂SO₄. Absorbance at 450 nm (OD₄₅₀) was read on an ELISA microplate reader (BioTek, VT, USA).

Determination of the cut-off values

The cutoff values of the two blocking ELISAs were evaluated using a panel of 170 serum samples (81 from PDCoV-infected and 89 from non-infected pigs, Viruses, cells, and serum samples). For each sample, the percent inhibition (PI) was calculated. PDCoV antibody-negative swine serum served as the negative control. These PI values were then subjected to receiver operating characteristic (ROC) curve analysis in GraphPad Prism 8.0 (San Diego, CA, USA) to discriminate between positive and negative sera. The optimal cutoff value, along with its associated sensitivity and specificity, was determined from the ROC curve by maximizing Youden’s index [44,45].

Determination of the sensitivity, specificity, repeatability and reproducibility

To determine the analytical sensitivity, a PDCoV-positive control serum was evaluated across a series of two-fold dilutions (from 1:2 to 1:512) using the established blocking ELISA procedure.

To evaluate diagnostic sensitivity, a total of 75 clinical serum samples and 75 colostrum samples (see Viruses, cells, and serum samples) were analyzed by comparing the results of the developed blocking ELISA with the virus neutralization test.

To evaluate analytical specificity, we tested antisera against a range of swine pathogens (PCV2, PCV3, PRRSV, PRV, PoRV, CSFV, PEDV, and ASFV). The sample set included two positive sera for each pathogen, except for PEDV, for which twenty sera were used, as detailed in Viruses, cells, and serum samples.

To evaluate assay precision, both intra- and inter-assay variations of the PDCoV blocking ELISA were assessed using three antisera with varying antibody titers. For intra-assay precision, each sample was run in eight replicates within the same plate. Inter-assay precision was determined by testing eight replicates of each sample across three independently prepared plates. The precision of the assay was evaluated by calculating the coefficient of variation (CV) for each antiserum, using the formula CV = standard deviation (SD) / mean ($\bar{\times }$), where the mean ($\bar{\times }$) is derived from the OD₄₅₀ values.

Animal experiment

Five 4-week-old Large White piglets obtained from Henan Spring Agriculture Technology Co. Ltd. were immunized 7 days after arrival. Before the experiment, all piglets were confirmed to be seronegative for PDCoV-specific neutralizing antibodies.

The recombinant S protein subunit vaccine was prepared as previously described [39]. Briefly, the purified recombinant S protein was emulsified with GEL 01 adjuvant (SEPPIC, Paris, France) at a 1:1 ratio (v/v).

For immunization, piglets in the vaccine group received an intramuscular injection of the vaccine, containing 2 mL of a dose with 200 µg of recombinant S protein per piglet. Prime immunization was administered at 5 weeks of age, followed by two booster immunizations using the same dose at 3-week intervals. Blood samples were collected weekly from the jugular vein of each piglet for serum separation and subsequent analysis. The euthanasia procedure followed the AVMA Guidelines. A single rapid intravenous bolus of pentobarbital (150 mg/kg body weight; Merck, Darmstadt, Germany, 200 mg/mL solution) was delivered via an 18-G catheter placed in the auricular vein. The animal exhibited rapid loss of consciousness, becoming laterally recumbent within 10 − 15 seconds. This was followed by respiratory arrest within 60 − 90 seconds and cardiac arrest within 3 − 5 minutes. Death was confirmed by the absence of both pupillary light reflex and audible heartbeat upon auscultation for a minimum of 5 minutes. Two licensed veterinarians independently verified the cessation of all vital signs [41].

The animal study protocol was reviewed and approved by the Animal Care and Ethics Committee of the Luoyang Putai Biotechnology Co., Ltd. (Approval No. PT20240003). All animal experiments were performed in accordance with China’s Regulations on the Administration of Laboratory Animals (Ministry of Science and Technology) and the ARRIVE guidelines.

Serological survey by blocking ELISA

A comprehensive serological survey was conducted to investigate the epidemiology and host range of PDCoV in China. From 2025, a total of 2002 pig serum samples were collected from 175 farms located in 10 provinces for the monitoring of PDCoV. These samples were submitted to our laboratory for routine diagnostic testing and disease surveillance, and were selected for this study based on availability (convenience sampling).

Furthermore, a diverse serum panel was established to assess cross-species susceptibility. This panel included 115 dog sera (collected from four provinces between 2020 and 2023) and extended to numerous other species, including peafowls, Kunming mice, bamboo rats, foxes, ferrets, tigers, rhinoceroses, alpacas, pangolins, horses, steppe eagles, lemur variegatus, masked palm civets, bears, pandas, yellow-throated martens, weasels, leopard cats, porcupines, and boars (see Viruses, cells, and serum samples).

Establishment and evaluation of the B-Cell epitope prediction model

We developed a B-cell epitope (BCE) prediction model by fine-tuning the Transformer-based protein language model ESM-2 (esm2_t30_150M_UR50D). Experimentally validated BCEs and non-epitope sequences were curated from the Immune Epitope Database (IEDB) to construct positive and negative datasets, respectively, after removing redundant sequences. The dataset was randomly partitioned into a held-out test set (20% of the data), with the remaining 80% divided into a training set (72% of the total data) and a validation set (8% of the total data) in a 9:1 ratio. The model was fine-tuned for 20 epochs with a learning rate of 1e-5, and training was monitored using the validation set to prevent overfitting. The final model, designated BCE-Vir-Prediction, was evaluated on the independent test set. Performance was comprehensively assessed, with key metrics including Loss, Accuracy, Precision, Recall, F1-score, and the Area Under the Receiver Operating Characteristic Curve (AUROC).

For epitope mapping on viral protein sequences, a sliding window (length = 20) was applied to predict the BCE probability for each window. To generate a residue-level epitope propensity score, the prediction probabilities from all overlapping windows covering a given residue were aggregated by taking the maximum value. This process converted variable-length sequences into a fixed-length residue-wise probability profile, facilitating the identification and visualization of potential epitopic regions.

Homology analysis and epitope structure

To evaluate the conservation of the mapped epitope, we aligned its sequence with those from other PDCoV strains, deltacoronaviruses, and porcine coronaviruses using MEGA-12. All viral sequences were obtained from previously published studies [46,47].

To characterize the epitope structurally, we analyzed the published cryo-EM structure of the PDCoV S protein (PDB: 8r9w) using PyMOL (version 3.0.5).

Statistical analysis

All statistical analyses and graphical generation were conducted with GraphPad Prism software (version 8.0; San Diego, CA, USA). The PI value was calculated for each sample using the formula: = (1 – OD₄₅₀ value of sample serum/OD₄₅₀ value of negative sample) × 100%, with both controls included on every plate. Based on the normality outcome, the correlation between blocking ELISA PI value and virus neutralization titers was evaluated with the Pearson correlation coefficient (r). Statistical significance was defined as a p-value of less than 0.05.

Results

Expression, purification, and identification of recombinant PDCoV S, S1, and RBD proteins

The recombinant S, S1, and receptor-binding domain (RBD) proteins of PDCoV were expressed in a CHO cell system to ensure proper post-translational modifications, and their domain architectures are illustrated in Figure 1(A). Following cloning and expression, the proteins were purified and analyzed by SDS-PAGE, which revealed high-purity bands at the expected molecular weights (Figure 1(B)). Successful expression was further confirmed by Western blot using an anti-His tag antibody, which showed specific reactivity for each protein (Figure 1(C)).

Figure 1.

Analysis of recombinant PDCoV proteins expressed in CHO cells. (A) Schematic diagram of the constructed S protein fragments. (B, C) The purified proteins (s: ~180 kDa, lane 1; S1: ~70 kDa, lane 2; RBD: ~20 kDa, lane 3) were subjected to SDS-PAGE (b) and Western blot (C) analyses. A Western blot was performed using an anti-His tag antibody to confirm antigenicity. The “*” indicates the bands corresponding to the purified, glycosylated PDCoV S, S1, and RBD proteins. Molecular weight markers are shown in lane M.

Characterization of monoclonal antibodies

Initial characterization by IFA showed that both 1C12 and 3A11 specifically recognized three PDCoV strains, as indicated by intense cytoplasmic staining in infected LLC-PK1 cells (Figure 2(A)). To determine the nature of their epitopes, we performed Western blot analysis. Under denaturing conditions, mAb 1C12 recognized a linear epitope, whereas mAb 3A11 failed to bind, indicating its specificity for a conformational epitope (Figure 2(B)). ELISA further established that both mAbs bind within the RBD, as they reacted specifically with the full-length S protein, the S1 subunit, and the recombinant RBD itself (Figure 2(C)). This RBD-specific binding was corroborated by IFA using HEK293T cells expressing the RBDs from all three PDCoV strains, which were both strongly recognized by 1C12 and 3A11 (Figure 2(D)). An alignment of the RBD sequences highlighting amino acid variations among the strains is shown in Figure 2(E).

Figure 2.

Characterization of anti-PDCoV RBD monoclonal antibodies. (A) Immunofluorescence analysis of mAbs 1C12 and 3A11 reactivity. LLC-PK1 cells were infected with three distinct PDCoV strains, fixed, and stained with the respective monoclonal antibodies (mAbs). Scale bars, 50 μm. (B) Western blot analysis of lysates from PDCoV-infected LLC-PK1 cells with mAbs 1C12 and 3A11. molecular weight markers are shown in lane M. (C) ELISA showing the binding of mAbs 1C12 and 3A11 to the PDCoV full-length S trimer, S1 subunit, and RBD protein. (D) Immunofluorescence confirmation of mAb specificity to the RBD. HEK293T cells were transfected with plasmids expressing RBDs from three PDCoV strains, fixed at 24 hours post-transfection. Scale bars, 50 μm. (e) the RBD sequences of the three PDCoV strains used in this study were aligned using the DNAStar megalign software (version 7.1.0), with amino acid substitutions highlighted in red at positions 26, 97, and 109.

Isotype analysis identified both mAbs as IgG1 with kappa light chains (Table 1). Functionally, both antibodies showed potent neutralizing activity against three PDCoV strains (Table 2) and demonstrated blocking activity, indicating their potential for use in blocking ELISA development (Table 1).

Table 1.

Isotype identification and blocking activity of anti-PDCoV monoclonal antibodies.

Monoclonal antibody	Isotype	Light Chain	Blocking Activity
1C12	IgG1	Kappa	+
3A11	IgG1	Kappa	+

Note: + indicate positive.

Table 2.

Neutralizing antibody titers (log₂‑transformed, mean ± sd) of monoclonal antibodies against PDCoV strains.

Monoclonal antibody	Neutralizing antibody titers (log2)
	PDCoV GDSG10-2023	PDCoV HZYH-2019	PDCoV SCNJ10-2021
	ׯ ± SD	ׯ ± SD	ׯ ± SD
1C12	7.01 ± 0.23	10.24 ± 0.25	6.48 ± 0.22
3A11	5.96 ± 0.42	6.62 ± 0.34	2.88 ± 0.54

Neutralizing epitope mapping of mAbs

To map the neutralizing epitope of mAb 1C12, we generated a panel of truncated RBD mutants and expressed them in HEK293T cells (Figure 3).

Figure 3.

Mapping of the linear B-cell epitope recognized by mAb 1C12. (A) Initial epitope mapping. HEK293T cells expressing GFP-fused full-length RBD (aa 300–419, which defines the RBD domain of the PDCoV S protein) or the indicated truncations were analyzed by Western blot using mAbs 1C12 and 3A11. (B) Fine mapping of the epitope. Western blot analysis using mAb 1C12 on cells expressing a series of truncated RBD fragments centered on the aa 300–349 region. (C) Identification of the core epitope. The amino acid sequence of the minimal epitope is shown. The minimal core epitope region is delineated between the two blue dashed lines. Molecular weight markers are shown in lane M. Red and green arrows indicate peptide fragments that exhibited positive and negative binding to the mAb, respectively.

Western blot analysis of three overlapping fragments showed that mAb 1C12 explicitly bound to the full-length RBD and the N-terminal fragment (amino acid (aa) 300–349), but not to the middle or C-terminal fragments. This result localized the linear epitope to the 300–349 aa region. As expected for a conformation-specific antibody, mAb 3A11 failed to bind any of the truncated mutants (Figure 3(A)).

Subsequent fine mapping using two additional series of truncated peptides progressively narrowed the epitope: first to a critical 20 amino acid region (aa 310–329; Figure 3(B)), and finally to an 8 amino acid core segment (aa 316–323; Figure 3(C)). Based on these results, we conclusively identified ³¹⁶DFGEARLD³²³ as the minimal linear B-cell epitope recognized by mAb 1C12.

Identification of the key amino acid residues

To identify the key amino acids influencing the binding affinity of mAb 1C12, we developed a machine learning-based model for predicting B-cell epitopes. This model, BCE-Vir-Prediction, demonstrated outstanding performance on an independent test set (AUROC: 0.969, F1-score: 0.929; Table S2), indicating high reliability for epitope mapping.

Application of the model to the RBD sequence generated a comprehensive residue-wise epitope propensity profile (Supplementary Dataset 2). Analysis of this profile identified the tri-residue motif ³²¹RLD³²³ as a peak region with exceptionally high prediction scores (a representative visualization is shown in Figure 4(A)). To experimentally validate the critical role of this motif, we generated a series of RBD mutants with individual alanine substitutions at R321, L322, and D323, as well as a triple alanine mutant (Figure 4(B)). Western blot analysis confirmed the motif’s essential role, as binding was completely abolished by both the R321A substitution and the triple alanine mutation. Furthermore, individual alanine substitutions at L322 and D323 drastically reduced binding, with the L322A mutation having a more pronounced effect than D323A (Figure 4(B)).

Figure 4.

Mapping the core binding residues of the mAb 1C12 linear B-cell epitope. (A) Schematic representation of the residue-wise epitope propensity profile across the PDCoV RBD, predicted by the BCE-Vir-prediction model. The red and blue bars together indicate the previously mapped linear epitope (aa 316–323). The red segment within this region contains the tri-residue motif (R321, L322, D323) identified as the peak in the prediction profile. (B) Western blot analysis of mAb 1C12 binding to purified wild-type RBD protein and a panel of alanine-substitution mutants (R321A, L322A, D323A, and RLD321–323AAA). Molecular weight markers are shown in lane M. Red and green arrows indicate peptide fragments that exhibited positive and negative binding to the mAb, respectively. (C) IFA of HEK293T cells transfected with plasmids encoding wild-type or alanine-substitution mutant RBD proteins. Cells were probed with an anti-His tag antibody, mAb 1C12, or the conformation-specific mAb 3A11. Scale bars, 50 μm.

As shown in Figure 4(C), IFA provided further confirmation of these results. Although an anti-His antibody confirmed successful expression of all His-tagged RBD constructs (wild-type and mutants), mAb 1C12 produced a starkly different binding pattern. It yielded strong fluorescence only for the wild-type RBD, with no detectable signal for any of the alanine substitution mutants (R321A, L322A, D323A, or the triple RLD321–323AAA). In a parallel assay, the conformation-specific mAb 3A11 also failed to bind any of the mutants, indicating that these mutations disrupt its conformational epitope as well.

Character analysis of the neutralizing epitope

We assessed the sequence conservation of the identified neutralizing epitope (³¹⁶DFGEARLD³²³) by aligning 27 PDCoV strains from nine countries. Notably, this epitope displayed 100% amino acid identity across all analyzed strains (Figure 5(A)).

Figure 5.

Conservation and structural localization of the linear B-cell epitope recognized by mAb 1C12. (A) Sequence alignment of the epitope region from 27 PDCoV strains across nine countries. (B) Analysis of epitope conservation across PDCoV strains from different hosts. The percentage values represent sequence identity compared to the PDCoV GDSG10-2023 strain. (C) Sequence comparison of the epitope with other porcine coronaviruses (PEDV, TGEV, SADS-CoV, PRCV, PHEV). the percentage values indicate amino acid identity with the PDCoV epitope. (D) Localization of the neutralizing epitope (red) on the S protein structure.

We further investigated its conservation across a broader host range, including strains from humans, mammals, and birds. The epitope sequence was identical (100%) among PDCoV GDSG10-2023 and all analyzed mammalian strains (including human, ferret badger CFBCoV, and leopard cat ALCCoV) (Figure 5(B)). In contrast, sequence similarity with avian strains was substantially lower (0–37.5%) (Figure 5(B)). Comparisons with other porcine coronaviruses also revealed low conservation (12.5–25%), which confirms the high specificity of this epitope for PDCoV (Figure 5(C)).

Structural analysis showed that the neutralizing epitope is fully exposed on the surface of the S protein, indicating that it is readily accessible for antibody binding (Figure 5(D)).

Establishment, optimization, and cut-off value determination of the blocking ELISA

We developed two blocking ELISAs (bELISAs) based on the neutralizing mAbs 1C12 and 3A11, designated bELISA-1C12 and bELISA-3A11, respectively. Checkerboard titration was performed to establish optimal working conditions that yielded a high negative/positive (N/P) ratio.

Checkerboard titration defined the optimal assay conditions. Specifically, both assays used a coating antigen concentration of 0.1 μg/mL and a serum dilution of 1:1. The optimal working dilutions for the HRP-conjugated mAbs were determined to be 1:50,000 for bELISA-1C12 and 1:30,000 for bELISA-3A11 (Figure 6(A,B)). Additionally, the optimal incubation times were set at 60 minutes for serum and 30 minutes for the HRP-conjugated mAbs (Figure 6(C)). The optimized conditions established for serum analysis were identically applied to the colostrum samples.

Figure 6.

Optimization of assay conditions and determination of diagnostic cutoff values for the blocking ELISAs. (A, B) Checkerboard titration for optimizing the coating antigen concentration, and working dilution of HRP-conjugated mAbs for bELISA-1C12 (a) and bELISA-3A11 (b). Data are presented as heatmaps of N/P values. (C) Optimization of incubation times for serum and HRP-conjugated mAbs. (D, E) Assessment of diagnostic performance and determination of the cut-off value for bELISA-1C12 (d) and bELISA-3A11 (e). For each assay, the left panel displays the ROC curve, along with the AUC. The right panel shows the PI values for positive and negative serum samples, with the optimal cut-off (dashed line) determined by maximizing Youden’s index. Student’s t-test, *p < 0.05.

Diagnostic cutoff values for the two optimized bELISAs were established using a panel of 170 swine serum samples (81 virus neutralization test (VNT)-confirmed positive and 89 negative). Receiver operating characteristic (ROC) curve analysis demonstrated excellent diagnostic accuracy for both assays, with an area under the curve (AUC) of 0.9892 (95% confidence interval (CI): 0.9731–1.000; p < 0.0001) for bELISA-1C12 and 0.9909 (95% CI: 0.9824–0.9994; p < 0.0001) for bELISA-3A11 (Figure 6(D,E), left panel). Using GraphPad Prism 8.0, the optimal cutoff value was determined by maximizing Youden’s index and was set at 31.45% PI for bELISA-1C12 and 35.00% PI for bELISA-3A11 (Figure 6(D,E), right panel).

Analytical specificity and sensitivity

The analytical specificity of the bELISAs was evaluated using sera negative for PDCoV but positive for other pathogens. No cross-reactivity was observed with sera positive for PEDV (n = 20) (Figure 7(A)) or with sera positive for other common swine pathogens (PCV3, PCV2, PRRSV, PRV, PoRV, ASFV, CSFV) (Figure 7(B)). Collectively, these results confirm that both bELISAs exhibit high analytical specificity for detecting PDCoV antibodies without cross-reactivity to other swine pathogens. These results demonstrate that both bELISA-1C12 and bELISA-3A11 exhibit high specificity for detecting PDCoV antibodies.

Figure 7.

Evaluation of the analytical specificity and sensitivity of bELISA-1C12 and bELISA-3A11. (A) Specificity assessment against PEDV-positive sera. The PEDV-specific IgG/IgA S/P ratios (commercial ELISA, left axis) are shown alongside the corresponding PI values for the bElisas (right axis). (B) Specificity assessment against sera positive for other common swine pathogens (PCV2, PCV3, PRRSV, PRV, PoRV, CSFV, ASFV). The dashed line indicates the assay cut-off value. (C) Analytical sensitivity determination. Serial two-fold dilutions of a high-titer PDCoV-positive serum were tested. The dashed line indicates the assay cutoff value used to determine the endpoint titer.

The analytical sensitivity, evaluated using serial dilutions of a high-titer PDCoV-positive serum, showed that both bELISAs could detect specific antibodies at an endpoint dilution of 1:256, as defined by the established cutoff value (Figure 7(C)).

Evaluation of the repeatability and reproducibility

The precision of the bELISA-1C12 and bELISA-3A11 was evaluated by determining intra- and inter-assay coefficients of variation (CVs). As shown in Table 3, both assays demonstrated high precision, with intra-assay CVs of 4.7–8.6% and inter-assay CVs of 6.6–9.7% for bELISA-1C12, and intra-assay CVs of 3.9–8.8% and inter-assay CVs of 5.1–8.9% for bELISA-3A11. The low CV values observed for both assays indicate excellent repeatability and reproducibility, confirming their reliability for the serological detection of PDCoV antibodies.

Table 3.

Intra- and inter-assay precision of the developed bElisas.

PDCoV Antisera	Intra-Assay						Inter-Assay
	bELISA-1C12			bELISA-3A11			bELISA-1C12			bELISA-3A11
	ׯ	SD	CV	ׯ	SD	CV	ׯ	SD	CV	ׯ	SD	CV
Strongly positive	0.251	0.0216	8.6%	0.322	0.0283	8.8%	0.294	0.0285	9.7%	0.359	0.0320	8.9%
Moderately positive	0.579	0.0400	6.9%	0.541	0.0211	3.9%	0.538	0.0355	6.6%	0.485	0.0403	8.3%
Weakly positive	0.911	0.0428	4.7%	0.877	0.0553	6.3%	0.852	0.0673	7.9%	0.916	0.0467	5.1%

Diagnostic sensitivity and antibody kinetics against VNT benchmark

To evaluate the diagnostic sensitivity and characterize the kinetics of the antibody response, we analyzed serial serum samples from pigs immunized with a recombinant PDCoV S protein subunit vaccine (Figure 8(A)). The VNT was performed on all serum samples to provide a reference standard for evaluating the antibody response (Figure 8(B)).

Figure 8.

Comparative analysis of antibody kinetics in pigs immunized with a PDCoV subunit vaccine. (A) Schematic of the immunization and serum collection timeline. (B) Kinetics of the virus-neutralizing antibody response. (C, D) Kinetics of the antibody response measured by bELISA-1C12 (C) and bELISA-3A11 (d). The dashed horizontal lines indicate the respective assay cutoff values.

The antibody kinetics profiles, as determined by both bELISAs, were highly concordant, showing a sustained increase post-immunization (Figure 8(C,D)). In contrast, the neutralizing antibody response exhibited a distinct biphasic pattern, characterized by an initial rise and decline after the primary immunizations, a second increase following the third immunization, and a final decrease (Figure 8(B)). These results indicate that the bELISAs detect a steadily rising antibody profile, which differs temporally from the fluctuating dynamics of the neutralizing antibody response.

Neutralizing antibodies were detectable in all five pigs as early as two weeks after the primary immunization (Figure 8(B)). At this two-week time point, bELISA-1C12 detected seroconversion in four of the five pigs, with the remaining animal (#357) seroconverting one week later (Figure 8(C)). In contrast, bELISA-3A11 detected seroconversion at two weeks in only one pig (#492); the other four animals seroconverted at week three (Figure 8(D)). These results indicate that bELISA-1C12 exhibits a higher diagnostic sensitivity for early seroconversion than bELISA-3A11.

We evaluated the diagnostic performance of both bELISAs against a VNT benchmark using 150 clinical samples (Table 4). Compared to the VNT benchmark, bELISA-1C12 exhibited a total concordance of 93.3%, with 96.1% (73/76) positive agreement and 90.5% (67/74) negative agreement. bELISA-3A11 showed a comparable positive agreement of 96.1%; however, its negative agreement was lower (89.2%, 66/74), resulting in a slightly lower total concordance of 92.7%. These results indicate that bELISA-1C12 has a marginally superior overall diagnostic agreement. Detailed data are provided in Table S3.

Table 4.

diagnostic performance of the bElisas compared to the virus neutralization test (VNT) for detection of PDCoV antibodies in clinical samples (n = 150).

Target mAb	1C12/ 3A11/ VNT(+)	1C12/ 3A11/ VNT(–)	1C12/ 3A11(+)/ VNT(+)	1C12/ 3A11(–)/ VNT(–)	1C12/ 3A11(–)/ VNT(+)	1C12/ 3A11(+)/ VNT(–)	Coincidence
1C12	80 (53.3%)	70 (46.7%)	73 (96.1%)	67 (90.5%)	3	7	93.3%
3A11	81 (54.0%)	69 (46.0%)	73 (96.1%)	66 (89.2%)	3	8	92.7%
VNT	76 (50.7%)	74 (49.3%)	/	/	/	/	/

Note: +: positive; –: negative.

Furthermore, we assessed the correlation between the percent inhibition (PI) values of both bELISAs and neutralizing antibody titers using three independent sample sets. These included 40 clinical sera, 75 colostrum samples, and serial sera from the immunization study. A strong positive correlation was observed in all sample types (Figure 9(A-F)), with Pearson’s analysis confirming statistically significant relationships (p < 0.0001). The correlation coefficients (r) exceeded 0.92 for immunized pig sera (Figure 9(A,D)), 0.86 for clinical sera (Figure 9(B,E)), and 0.71 for colostrum samples (Figure 9(C,F)). All individual data are available in Table S3.

Figure 9.

Correlation between bELISA results and neutralizing antibody titers. Pearson correlation analysis was performed to assess the relationship between PI values from (A – C) bELISA-1C12 and (D – F) bELISA-3A11, and virus neutralization titers in immune serum, clinical serum and colostrum samples. Each graph shows the correlation coefficient (r) and the corresponding p-value.

Based on a comprehensive evaluation that integrates clinical performance (Table 4), antibody kinetic profiles (Figure 8(A-D)), and correlation analyses (Figure 9(A-F)), these findings establish the bELISA-1C12 assay as a reliable tool for PDCoV serological testing.

Multispecies serological survey for PDCoV using blocking ELISA

To delineate the epidemiology of PDCoV in China, we conducted a multispecies serosurvey using the highly sensitive and specific bELISA-1C12 assay. Within the swine population, a total of 2002 serum samples were collected from 28 cities across 10 provinces and autonomous regions in 2025. The overall seropositivity rate for PDCoV antibodies was 13.6% (273/2002), as summarized in Table 5. These porcine samples originated from 175 swine farms (Table 6), among which 53 tested positive, yielding a farm-level positivity rate of 30.3%. Geographically, the majority of porcine samples were collected from farms in Henan Province, which accounted for 32.4% (47/145) of the total, indicating that the prevalence rate of serum in pigs remains high.

Table 5.

Serological survey of PDCoV antibodies in multiple species.

No.	Animal	Province	Collection time	PI readout	Status	Ratio (%)
1	Swine	10 Provinces of mainland Chinaa	2025	−33.4% − 94.2%	273/2002 Positive	13.6
2	Dog	Henan/Hubei/Sichuan/Tianjin	2020–2023	−37.1% − 20.5%	115/115 Negative	0
3	Peafowl	Chongqing	2020	−12.7%/ −10.0% / 35.3%/ 36.4%	2/4 Positive	50
4	Kunming mice	–	2020	−13.8% − 24.2%	6/6 Negative	0
5	Bamboo rat	Guangxi	2020	−16.8% − 7.0%	7/7 Negative	0
6	Fox	–	2019	−16.0% − 5.8%	16/16 Negative	0
7	Ferret	Sichuan	2019	−9.0% − 9.6%	7/7 Negative	0
8	Tiger	Liaoning/Jilin	2015/2018	−39.0% – −4%	6/6 Negative	0
9	Rhinoceros	Guangxi	2018	−18.7% – −11.5%	5/5 Negative	0
10	Alpaca	Shanxi	2018/2019	−23.6% − 23.2%	10/10 Negative	0
11	Pangolin	Guangxi	2020	−28.1% − 12.2%	15/15 Negative	0
12	Horse	Jilin/Hunan	2022	−33.7% − 15.7%	7/7 Negative	0
13	Steppe eagle	Chongqing	2020	5.4%	1/1 Negative	0
14	Lemur variegatus	Shanghai	2009	−11.9%	1/1 Negative	0
15	Masked palm civet	Sichuan	2019	−6.7% − 18.3%	14/14 Negative	0
16	Bear	Sichuan/Chongqing	–	−9.7% − 15.6%	6/6 Negative	0
17	Panda	Shaanxi/Sichuan	–	−10.2% − 16.5%	8/8 Negative	0
18	Yellow-throated Marten	Sichuan	–	−15.4% – −11.8%	2/2 Negative	0
19	Weasel	Sichuan	–	−6.8% – −3.8%	2/2 Negative	0
20	Leopard Cat	Shaanxi	–	−15.5% – −4.3%	3/3 Negative	0
21	Porcupine	Sichuan	–	−4.3%	1/1 Negative	0
22	Boar	Shaanxi	–	−12.4%	1/1 Negative	0

Note: ^a 10 Provinces of mainland China: Anhui/Fujian/Guizhou/Henan/Hubei/Inner Mongolia/Shanxi/Shaanxi/Sichuan/Zhejiang.

Table 6.

Serologic analysis of PDCoV in swine farms in 10 provinces of mainland China.

Province	a Positive	a Negative	a Total	Ratio (%)
Anhui	0	1	1	0.0%
Fujian	0	1	1	0.0%
Guizhou	0	4	4	0.0%
Henan	47	98	145	32.4%
Hubei	0	1	1	0.0%
Inner Mongolia	2	8	10	20.0%
Shanxi	1	6	7	14.3%
Shaanxi	1	1	2	50.0%
Sichuan	1	1	2	50.0%
Zhejiang	1	1	2	50.0%
Total	53	122	175	30.3%

Note: ^a means the Number of pig farms.

To explore the potential host range of PDCoV, we applied the bELISA-1C12 assay to screen sera from 21 animal species, in addition to swine. Notably, we report the first detection of PDCoV antibodies in peafowl, providing serological evidence of cross-species exposure. The complete results of this multispecies screening are detailed in Table 5.

Discussion

PDCoV poses a significant zoonotic threat to the swine industry and public health [4,7]. However, critical gaps remain, particularly in identifying neutralizing linear epitopes within the receptor-binding domain (RBD) and in applying neutralizing monoclonal antibodies (mAbs) to develop serological assays for multispecies surveillance. To bridge these gaps, we generated two neutralizing mAbs against the RBD, defined their critical epitopes, and established corresponding blocking ELISAs. In a direct comparison benchmarked against the virus neutralization test (VNT), bELISA-1C12 demonstrated superior performance over bELISA-3A11, establishing it as a reliable serological tool. Consequently, we employed bELISA-1C12 for extensive multispecies surveillance, which successfully revealed an expanded host range for PDCoV.

The RBD within the S1 subunit is essential for PDCoV entry, as it mediates viral attachment to the aminopeptidase N (APN) host receptor. Consequently, blocking this interaction represents a validated strategy for inhibiting viral infection [29]. In this study, we generated two neutralizing mAbs targeting the RBD, specifically 1C12 and 3A11. To elucidate their mechanisms of action and inform the design of vaccines and diagnostics [48], we precisely mapped their binding epitopes. Our results show that 1C12 recognizes a linear epitope, ³¹⁶DFGEARLD³²³. Within this epitope, Arg321 (corresponding to the APN-binding residue Arg322 reported in prior structural studies) and Phe317 (aligned with the functionally critical Phe318) are both essential for receptor engagement [20,24]. This observation indicates that the 1C12 epitope spatially overlaps with the receptor-binding site. Notably, although 1C12 recognizes a linear epitope, its binding site substantially overlaps with the conformational epitope of the broad-neutralizing antibody PD41. This spatial overlap suggests that 1C12 functions through an analogous mechanism of direct competitive inhibition by sterically hindering APN binding. In contrast, 3A11 targets a conformational epitope centered on the ³²¹RLD³²³ motif. Its mode of action resembles that of the conformational antibody PD33, as 3A11 binding is abolished by mutations that disrupt local RBD folding [28]. Although the epitopes are recognized by neutralizing antibodies, its intrinsic neutralizing capacity as an independent peptide and its precise mechanism of blocking viral entry (e.g. competition with APN) warrant further investigation through peptide competition assays and structural biology studies. Recent studies have further elucidated the complexity of PDCoV entry. A genome-wide CRISPR/Cas9 screen identified C16orf62 as a host factor that regulates surface expression of APN, thereby affecting PDCoV binding and internalization [49]. Additionally, the natural compound cepharanthine has been shown to inhibit PDCoV infection by competing with APN binding [50].

To facilitate the screening of key residues within the epitope, we developed a B-cell linear epitope prediction model based on the Transformer architecture. This end-to-end deep learning model demonstrated strong performance on a held-out test set (AUROC: 0.969, F1-score: 0.929) and was instrumental in identifying and validating the ³²¹RLD³²³ motif targeted by the neutralizing antibodies 1C12 and 3A11. Our approach more effectively captures complex sequence features than traditional feature-based predictors like BepiPred-2.0 [51], aligning with the methodology of advanced frameworks like LBCE-BERT [52]. Nevertheless, as a data-driven model, its performance is inherently constrained by the quality and scale of available training data, a common limitation in the field [51]. Future efforts will therefore focus on integrating structural information to enhance interpretability and expanding the training datasets to improve generalization.

Monoclonal antibodies targeting particular epitopes are essential for developing robust serological detection tools [53]. In this study, we identified two such epitopes: the linear epitope ³¹⁶DFGEARLD³²³ (recognized by mAb 1C12) and the core conformational motif ³²¹RLD³²³ (targeted by mAb 3A11). Both epitopes are highly conserved across diverse PDCoV strains yet exhibit negligible sequence similarity with the corresponding regions of other porcine alpha- and beta-coronaviruses, highlighting their exceptional specificity for PDCoV. This high specificity, combined with the PDCoV cross-species transmission potential, makes these epitopes and their corresponding mAbs ideal candidates for developing serological assays that are independent of the host species.

Previous studies have reported blocking ELISA based on the PDCoV N protein [38] or competitive ELISA (cELISA) using the S protein [37]. Although these methods partially overcome the limitations associated with species-specific secondary antibodies, the mAbs used in these assays lack verified neutralizing activity. In this study, we developed two blocking ELISAs based on neutralizing mAbs that target the well-defined epitopes of 1C12 and 3A11. Systematic evaluation demonstrated that both assays exhibit reasonable specificity and sensitivity, providing practical tools for multispecies serological surveillance of PDCoV. A limitation of this study is that matrix normalization was not applied for the multispecies sera. Because our objective was qualitative screening rather than precise quantitative comparison across species, all samples were tested under uniform conditions (1:1 dilution) using swine negative serum as the reference. We acknowledge that matrix effects could influence PI values. Additionally, the specificity evaluation did not include antisera against other porcine coronaviruses such as transmissible gastroenteritis virus (TGEV), swine acute diarrhea syndrome-coronavirus (SADS-CoV), or porcine hemagglutinating encephalomyelitis virus (PHEV). Given that the RBD sequences of PDCoV share low homology with these coronaviruses, the likelihood of cross-reactivity is expected to be minimal. Nevertheless, future studies incorporating these controls would further validate the specificity of the assay.

In the analysis of antibody kinetics from vaccinated sera, bELISA-1C12 detected seroconversion earlier than bELISA-3A11, underscoring its superior sensitivity for early diagnosis. This observed difference in performance may be attributed to the distinct epitopes targeted by the two assays, suggesting that epitope characteristics can be a critical factor in shaping assay performance. Although both bELISAs showed a consistent increasing trend in antibody levels, their kinetic profiles differed from those of the neutralizing antibodies measured by the VNT. This difference indicates that although both assays target neutralizing epitopes, the antibody pools they detect likely comprise a mixture of neutralizing and non-neutralizing antibodies. Previous studies have shown that IgA-based indirect ELISAs using S1 or RBD produce kinetic trends more consistent with VNT [41], suggesting that such methods may be more indicative for immune efficacy evaluation.

Our bELISAs demonstrated a strong correlation between percentage inhibition (PI) and neutralizing antibody titers in both immune and clinical samples, with correlation coefficients (r-values) surpassing those of an S1-IgA indirect ELISA [41], and comparable to a previously reported nanobody-based cELISA (r = 0.86) [37]. Although bELISAs showed a strong correlation between PI values and neutralizing antibody titers in immune and clinical samples, the overall kinetic curves of antibody dynamics they produced differed from those measured by the VNT. This discrepancy demonstrates that a strong correlation coefficient does not equate to concordant dynamic monitoring.

PDCoV has a broad host range, and the interaction between its S protein and APN receptors from various species has been extensively studied [8,24,54,55]. Beyond APN, heparan sulfate (HS) has been identified as an essential attachment factor that facilitates PDCoV entry, with HS biosynthesis enzymes (SLC35B2, EXT1, NDST1) being critical for infection [56]. Based on its superior performance in analytical sensitivity evaluations, we selected bELISA-1C12 for extensive serological screening across 22 animal species. This screening detected PDCoV-specific antibodies in pigs and, notably, in peafowl. This finding not only expands the known host range of PDCoV but also provides critical serological evidence supporting the hypothesis that birds may be the natural and ancestral hosts of this virus [9,55].

Funding Statement

This work was supported by the National Key Research and Development Program of China [2023YFD1800500], Postgraduate Research & Practice Innovation Program of Jiangsu Province [SJCX25_2370], Taishan Industrial Leading Talent Project, the 111 Project D18007, and the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The experimental dataset supporting this study is openly available in the ScienceDB repository at https://doi.org/10.57760/sciencedb.35197 [57] under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.

The source code for the BCE-Vir-Prediction model is available on GitHub at https://github.com/JackKuo666/BCE-Vir-Prediction under the MIT License, and archived on Zenodo at https://doi.org/10.5281/zenodo.18300838 [58] is distributed under the Creative Commons Zero (CC0) License.

Ethics Statement

The animal study was approved by the Luoyang Putai Biotechnology Co., Ltd. with Permit No. PT20240003. The study was conducted in accordance with the local legislation and institutional requirements. We have adhered to ARRIVE guidelines in this study.

Supplemental data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21505594.2026.2668161