Identification of Neutralizing Monoclonal Antibodies Targeting Novel Conformational Epitopes of the Porcine Epidemic Diarrhoea Virus Spike Protein

Since 2010, newly identified variants of porcine epidemic diarrhoea virus (PEDV) have caused high mortality in neonatal piglets which has devastated the swine industry. The spike (S) glycoprotein of PEDV contains multiple neutralizing epitopes and is a major target for PEDV neutralization and vaccine development. To understand the antigenicity of the new PEDV variant, we characterized the neutralizing epitopes of a new genotype 2b PEDV isolate from Taiwan, PEDV Pintung 52 (PEDV-PT), by the generation of neutralizing monoclonal antibodies (NmAbs). Two NmAbs, P4B-1, and E10E-1–10 that recognized the ectodomain of the full-length recombinant PEDV S protein and exhibited neutralizing ability against the PEDV-PT virus were selected. Recombinant truncated S proteins were used to identify the target sequences for the NmAbs and P4B-1 was shown to recognize the C-terminus of CO-26K equivalent epitope (COE) at amino acids (a.a.) 575–639 of the PEDV S. Interestingly, E10E-1–10 could recognize a novel neutralizing epitope at a.a. 435–485 within the S1A domain of the PEDV S protein, whose importance and function are yet to be determined. Moreover, both NmAbs could not bind to linearized S proteins, indicating that only conformational epitopes are recognized. This data could improve our understanding of the antigenic structures of the PEDV S protein and facilitate future development of novel epitope-based vaccines.

Determination of neutralizing antibodies. After purification and quantification of mAbs, the neutralizing ability of each mAb against the G2b Taiwan PEDV-PT strain was assessed using the virus neutralizing assay. The starting concentration for each mAb in the neutralizing assay was 20 μg/mL, with a two-fold serial dilution to obtain a final neutralizing concentration of 2.5 μg/mL. The mAbs, P4B-1, and E10E-1-10, possessed potent neutralizing ability that completely blocked the CPE of the Taiwan PEDV-PT strain under 5 μg/mL. On the contrary, the mAbs, R7H-2, and UNK-1, showed no neutralization ability against PEDV-PT even at 20 μg/mL.
ELISA results showing reactivity of mAbs to various truncated PEDV S proteins. To further confirm the reactivity of each mAb to the targeting epitopes, ELISAs were performed using plates coated with www.nature.com/scientificreports www.nature.com/scientificreports/ various truncated PEDV spike proteins. The PEDV NmAbs (P4B-1 and E10E-1-10) and the non-PEDV recognizing mAb (UNK-1) were all serially diluted to conduct the ELISA. As shown in Fig. 4, P4B-1 had a binding affinity with appropriate dilution effects toward the purified full-length PEDV spike protein, as well as PEDV S 1-639 , but had its affinity towards purified PEDV proteins, S 1-575 , S 1-509 , S 1-501 , S 1-485 and S 1-435 was weak. However, E10E-1-10, which targets a novel neutralizing epitope as shown with ICC staining, was also capable of binding to the full-length PEDV spike protein as well as truncated PEDV proteins, S 1-639 , S 1-575 , S 1-509 , S 1-501 , and S 1-485 with appropriate dilution effects. Similar to the results obtained by ICC staining, E10E-1-10 showed no binding ability  www.nature.com/scientificreports www.nature.com/scientificreports/ toward PEDV S 1-435 by ELISA and therefore, an NmAb targeting a novel epitope of the new variant of PEDV specifically at a.a. 435-485 in the S1 region was confirmed. The non-PEDV recognizing mAb, UNK-1 showed no binding affinity towards the various truncations of the spike protein and hence, it was used as the external control for the ELISA assay.
Binding affinity of NmAbs to linearized epitopes. To further determine whether the binding of neutralizing epitopes require conformational integrity, immunodot blotting was performed by denaturing the various truncated PEDV spike proteins. The conformational proteins as well as the denatured proteins of S 1-435 , S 1-485 , S 1-501 , S 1-509 , S 1-575 , S 1-639 , and the full-length spike of PEDV-PT were dotted onto the membrane. As shown in Fig. 5(a-c), P4B-1 was able to detect the conformational S 1-639 and the full-length spike protein on the membrane but was unable to detect the proteins upstream of a.a. 575 of spike, consistent with the findings from the ELISA and ICC staining. The other NmAb, E10E-1-10, could bind to S 1-485 , S 1-501 , S 1-509 , S 1-575 , S 1-639 and the full-length spike proteins in conformational structures. The R7H-2, a spike-recognizing but non-neutralizing mAb, was only capable of binding to the conformational full-length spike protein. P4B-1, E10E-1-10, and R7H-2 were unable to stain the denatured target proteins on the membrane, indicating the loss of binding affinity after protein linearization. Moreover, to ensure all the proteins were equally dotted on the membrane, the results of mAb-stripping and re-probing with anti-V5 tag antibodies are shown in Fig. 5(d-f).

Modelling of PEDV S protein.
A molecular model of recombinant PEDV S ectodomain protein was established using SWISS Model and re-edited using UCSF ChimeraX molecular modelling system. As shown in Fig. 6, the monomer of PEDV S protein was divided into S1 (a.a. 1-735, in light-blue) and S2 (a.a. 736-1378, in grey). The novel neutralizing epitope (a.a. 435-485 of the S protein), characterized by the E10E-1-10 NmAb labelled in yellow in Fig. 6(a), was inferred to be at the C terminal region of the S1 A domain. The two mutations, V 441 I, and mAb S 1-435 S 1-485 S 1-501 S 1-509 S 1-575 S 1-639 S  www.nature.com/scientificreports www.nature.com/scientificreports/ S 477 A that were different from the CV777 strain were highlighted in red sphere; and the one mutation, E 459 Q that was different from the Taiwan historic PEDV HC070225-S strain was highlighted with a green sphere. On the other hand, the region of the neutralizing epitope (a.a. 575-639 of the S protein; labelled in yellow) characterized by the P4B-1 NmAb was deduced at the C-terminal region of the S1 B domain in Fig. 6(b). The a.a. mutation (G 596 S) different from the CV777 strain was highlighted with a purple sphere; and the a.a. mutations, D 608 E and L 615 F, different from the Taiwan historic PEDV strain HC070225-S were highlighted with pink spheres. The mutation, Q 636 E that was different from both CV777 and Taiwan historic PEDV strain HC070225 was highlighted with an orange sphere. The six truncated as well as full-length spike protein of PEDV-PT were coated into separate wells and probed with two-fold serially diluted mAbs, P4B-1 and E10E-1-10. The non-PEDV recognizing mAb, UNK-1, was added as the external control for ELISA under the same dilution conditions. The two-fold serially diluted anti-V5 tag antibody was used as the standard to show the dilution effect of the binding affinity assay. The data is represented as the S/P ratio. The wells incubated with only secondary antibodies and the wells incubated with anti-V5 tag antibody at 1,250 ng/mL were used as negative and positive controls for S/P ratio respectively. The results of the ELISA for the various truncations of the spike proteins are shown as follows: www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
Since 2010, outbreaks of infections caused by new PEDV variants have been correlated to viral mutations in critical neutralizing epitopes of the spike protein 8,20 . Although several sequences of the new variant PEDV spike genes have been decoded 6,21 and multiple domain architectures of the PEDV S protein have been predicted by three-dimensional structures of alphacoronaviruses 16 , the effects of genetic mutations on the antigenicity of PEDV remains poorly understood. A thorough evaluation of the neutralizing epitopes and antigenicity of the new PEDV variants is therefore needed. In the present study, we generated a panel of mAbs using PEDV-PT viral particles. By selecting mAbs that exhibited high binding affinity to full-length spike ectodomain-expressing HEK293 cells, we successfully identified two representative NmAb against PEDV-PT. The targeting epitopes identified by these two NmAbs were located at a.a. 575-639 and a.a. 435-485 of the PEDV spike protein, respectively, as shown by ICC staining and ELISA. Further, none of these NmAbs were able to bind to the linearized forms of  The recombinant ectodomain of S protein of PEDV-PT strain was modelled using SWISS model and edited using UCSF ChimeraX. The S protein was present as monomer and the S1 and S2 regions of PEDV were displayed in light-blue and grey, respectively. The conformational neutralizing epitope regions characterized by the mAbs are given as follows: (a) E10E-1-10 mAb labelled in yellow; two mutations (V 441 I and S 477 A) on the epitope different from the CV777 strain were highlighted in red sphere and one mutation (E 459 Q) on the epitope different from the Taiwan historic PEDV strain (HC070225-S) was highlighted in green sphere. (b) P4B-1 mAb was labelled in yellow; the G 596 S mutation on the epitope different from the CV777 strain was highlighted in a purple sphere; mutations, D 608 E and L 615 F, on the epitope different from the Taiwan historic PEDV strain (HC070225-S) were highlighted in pink sphere. The orange sphere represented the common mutation (Q 636 E) noted on the epitope of PEDV-PT S protein that was different from both of CV777 and Taiwan historic PEDV (HC070225-S) strains. (2019) 9:2529 | https://doi.org/10.1038/s41598-019-39844-5 www.nature.com/scientificreports www.nature.com/scientificreports/ the truncated spike proteins, indicating that they recognize the integral epitopes with appropriate conformation. Taken together, we discovered a novel conformational neutralizing epitope in the S1 A domain at a.a. 435-485, whose function has not been clearly determined in coronaviruses. In addition, we also discovered a conformational neutralizing epitope close to the C-terminus of the COE domain, at a.a. 575-639.
The neutralizing epitopes of the historic strains of PEDV on the S protein that have been previously published include the CO-26K equivalent epitope (COE epitope; a.a. 499-638 in the Brl strain; a.a. 501-640 in the PEDV-PT strain) 12 , 2C10 epitope (a.a. 1368-1374 in the CV777 strain; a.a. 1373-1379 in the PEDV-PT strain) 13 , and S1D (a.a. 636-789 in the CV777 strain; a.a. 640-794 in the PEDV-PT strain) 14 . In case of the new variants of PEDV, the domain B (a.a. 510-640 in the PEDV GDU and the PEDV-PT strain) and domain 0 in the S1 region (a.a. 1-219 in the PEDV GDU and the PEDV-PT strain) are demonstrated to have neutralizing epitopes 17 . Furthermore, Okda et al. also discovered several linear epitopes at the N-terminus of S2 18 . Previously, the identification of neutralizing epitopes of the historic PEDVs led to the proposal that several single nucleotide polymorphisms (SNPs), such as three serine amino acid substitutions in the COE epitope and two serine amino acid substitutions in the S1D epitope, may be related to newly-occurring global outbreaks 6,22 . In the present study, the NmAb, P4B-1, was able to recognize the epitope within a.a. 575-639 on the S protein, which overlaps with the C-terminus of the COE epitope, echoing the findings of previous studies 12 . As the S1 B domain consists of the main COE neutralizing epitope of PEDV and is the receptor binding domain for the majority of the coronaviruses, its importance has been emphasized in many studies 11,23,24 . Li et al. have demonstrated that the binding affinity of neutralizing antibodies could be dramatically altered by exchanging one amino acid on the S protein of PEDV 17 . Hence, mutations in S1 B domain, especially in neutralizing epitopes of the PEDV S protein, are speculated to represent the evolution of viral escape from antibodies, like the one evoked by CV777-based vaccine immunization of a historic PEDV strain infection 4,25 . Interestingly, the other NmAb, E10E-1-10, presented a high binding affinity toward a.a. 435-485 of the S protein that precludes the COE epitope and locates in the S1 A domain has also been identified in our study. To our knowledge, the function of the S1 A domain of the PEDV S protein has not been well studied. The S1 A domain of human coronavirus HCoV-NL63 is speculated to interact with heparan sulfate proteoglycans on the host cells to mediate the virus anchoring and infection 26 . The N-terminus of S1 A domain of bovine coronavirus has been shown to have two sugar binding loops which can recognize the carbohydrate receptor on the host cells 27 . Moreover, heparan sulfate has also been demonstrated as an attachment factor for PEDV 28 . However, as the actual topography of the PEDV trimeric S protein remains to be solved and the function of S1 A domain is yet to be determined, the actual interaction of the S1 A domain targeting mAb with these sugars or the heparan sulfate proteoglycans-binding domains on the PEDV S glycoprotein remains to be further evaluated. Visualization of the topology and interaction of PEDV S glycoprotein with these neutralizing antibodies for guiding future immunogen and therapeutics design is an important future goal.
Epitopes for antibody recognition can generally be divided into two main classes: linear and conformational forms. Linear epitopes are formed by a continuous sequence of amino acid in a protein, while conformational epitopes consist of amino acid sequences that are discontinuous in the protein but are brought together upon three-dimensional protein folding. It has been demonstrated that 90% of B cell-recognizing epitopes are conformational epitopes that result from the antigen internalizing process and special antigen-recognizing ability 29,30 . Many conformational epitope-targeting antibodies, including neutralizing antibodies, are particularly difficult to determine because the antibody-antigen complex is formed solely in the native structure of the protein. Over the past several decades, with the use of phage-display peptide probing, the neutralizing epitope of CV777, named 2C10 has been identified 13 . Using ELISA or Pepscan assays, several linear neutralizing epitopes have been discovered on the S2 glycoprotein subunit of new PEDV variants 18 . Additionally, two B cell epitopes, named SS2 and SS6, in the region of the S1D neutralizing epitope were evaluated using a combination of phage-display peptide probing and Pepscan 14,15 . However, these epitope mapping approaches focus mainly on linear epitopes 31 . Using ICC staining, ELISA, and denatured immunodot blotting, all of the B cell-recognizing epitopes identified in the present study were found to be conformational epitopes. Thus, these affinity-binding assays could serve as valuable platforms for studying the conformational epitopes of NmAbs. In conclusion, the NmAbs and various truncated S proteins constructed generated in the present study could contribute to a better understanding of the antigenicity and immunogenicity of highly virulent PEDVs, especially in revealing the antigenic role of S1 A domain, and the pathogenesis of immune escape that leads to an outbreak of PEDV. Moreover, our study may provide important fundamental information for the future development of novel epitope-based vaccines.

Virus passage and purification. PEDV was isolated and passaged in Vero cells (American Type Culture
Collection (ATCC) No. CRL-1586) as previously described 32 . After freezing and thawing, the harvested viral supernatant was centrifuged at 3000 rpm for 30 min to remove cell debris and filtered through a 0.22 µm ultrafiltration cup (Corning, New York, USA). The viral supernatant was then added onto a 20% sucrose (Sigma, Missouri, USA) cushion, purified by ultracentrifugation at 75,000 × g for 2.5 h. The viral pellet was re-suspended in phosphate buffered saline (PBS) (Gibco, Gaithersburg, USA), and then applied to a 20-60% sucrose-TNE (20 mM Tris-HCl (pH 7) (Sigma), 100 mM NaCl, 2 mM EDTA (Sigma)) gradient, and centrifuged at 75,000 × g for 2.5 h in an Optima ™ L-100XP preparative ultracentrifuge using an Avanti J-25 rotor (Beckman Coulter, Sykesville, USA). Purified virions were diluted in TNE buffer, pelleted by centrifugation at 75,000 × g for 1.5 h to remove the sucrose and then, re-suspended in TNE buffer. www.nature.com/scientificreports www.nature.com/scientificreports/ mAb production. Three BALB/c mice were intramuscularly (IM) immunized with 20 μg purified PEDV viral particles mixed with 100 μL complete Freund's adjuvant (Sigma). After two weeks, two IM booster injections were administered using 20 μg purified PEDV viral particles with 100 μL Incomplete Freund's adjuvant (Sigma) at intervals of 3 weeks. Three days before sacrifice, mice were immunized with 20 μg purified PEDV viral particles in PBS (Gibco) via intrasplenic (IS) injection. Serum antibody titres at each immunization were monitored using a complete PEDV viral particle ELISA and the mouse with the highest titre was sacrificed for hybridoma preparations.
Hybridoma preparation. Splenocytes were isolated from the mice immunized with purified PEDV particles. After gentle washing with brief centrifugation, splenocytes were fused with SP2 myeloma cells at a cell ratio of approximately 10:1 using 50% polyethylene glycol (Sigma). Hybridomas were seeded onto 96-well culture plates in RPMI-1640 medium supplemented with 20% foetal bovine serum (Gibco), 100 mg/mL streptomycin, and 100 IU/mL penicillin (Sigma), and incubated overnight at 37 °C in a humidified incubator with 5% CO 2 . After incubation, approximately 50% medium was removed from each well, and a selective HAT RPMI-1640 medium (HAT-RPMI) (Sigma) was added to achieve a final concentration of 20% foetal bovine serum (Gibco), 100 mg/mL streptomycin, 100 IU/mL penicillin, 100 mM hypoxanthine (Sigma), 400 mM aminopterin (Sigma), and 16 mM thymidine (Sigma). Wells containing growing hybridoma cells were screened for antibody production by ICC staining using PEDV-infected Vero cells or HEK293 cells (ATCC CRL-1573 ™ ) expressing the full-length PEDV S protein. Positive clones were isolated for limiting dilution and incubated in selective HT RPMI-1640 medium without aminopterin. After two limiting dilutions, the supernatant from each line was further tested for anti-PEDV S-specific antibodies by ICC staining using PEDV-infected Vero cells or HEK293 cells expressing the full-length recombinant PEDV S protein.
mAb purification and quantification. To purify mAbs from cultured supernatants, Pierce ™ Protein L Magnetic Beads (Thermo Fisher Scientific, Waltham, USA), which selectively bind to mouse immunoglobulin were utilized following the manufacturer's instructions. The beads were mixed with 40 mL supernatants of each mAb and incubated for 1 h after which the antibody-bound beads were collected using a magnetic stand. After washing thrice wash buffer (Tris-buffered saline (TBS), 0.05% Tween-20 detergent), the mAbs were eluted with 60 μL elution buffer (0.1 M glycine, pH 2.0) for 10 min and then, alkalized PBS buffer (pH 8.5) was added for neutralization. The concentrations of each purified mAb were determined by the Pierce ™ BCA Protein Assay Kit (Thermo Fisher Scientific).
Purification and western blot detection of truncated spike proteins. Large-scale purification of the truncated S proteins of PEDV-PT was performed as previously described 19 . The protein expressing HEK-293 cells were harvested, resuspended, and cultured in FreeStyle 293 expression medium (Gibco) for one week. After supernatant collection and removal of cell debris through a 0.22 μm filter, the proteins were banded with HisPur Cobalt Resin (Thermo Fisher Scientific) following the manufacturer's protocol. The purified proteins were subsequently concentrated with a 30 kDa Vivaspin ® 20 (GE Healthcare Life Sciences, NY, USA), cOmplete ™ EDTA-free protease inhibitor cocktail (Roche Molecular Biochemicals, Quebec, Canada) was added and the concentrations were measured using the Pierce ™ BCA protein assay kit. The purified proteins were denatured in a buffer containing NuPAGE ® LDS sample buffer and NuPAGE ® reducing agent, and heated for 5 min at 95 °C. Using Bio-Rad Mini-PROTEIN ® electrophoresis system(Bio-Rad, Hercules, CA, USA), the samples were separated using 10% sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) cast in a gradient T-Pro EZ Gel Solution (T-Pro Biotechnology, Taiwan) according to the manufacturer's protocol. Following this, proteins were wet-blotted onto a polyvinylidene difluoride (PVDF) membranes (Bio-Rad)blocked with 5% skim milk in Tris-buffered saline with 0.05% Tween 20 (TBS-T) buffer at room temperature (RT) for 1 h. Membranes were incubated with a 1:5000 diluted anti-V5 tag antibody for 1 h at RT. After washing with TBS-T, horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG secondary antibody (1:10,000 dilution in blocking buffer; Jackson ImmunoResearch Laboratories, Philadelphia, USA) was added and incubated at RT for 1 h, the signals were visualized using Clarity ™ Western ECL Blotting Substrates (Bio-Rad) and detected with a ChemiDoc ™ XRS+ Imaging System (Bio-Rad).