The N-terminal Subunit of the Porcine Deltacoronavirus Spike Recombinant Protein (S1) Does Not Serologically Cross-react with Other Porcine Coronaviruses

Porcine deltacoronavirus (PDCoV), belonging to family Coronaviridae and genus Deltacoronavirus, is a major enteric pathogen in swine. Accurate PDCoV diagnosis relying on laboratory testing and antibody detection is an important approach. This study evaluated the potential of the receptor-binding subunit of the PDCoV spike protein (S1), generated using a mammalian expression system, for specific antibody detection via indirect enzyme-linked immunosorbent assay (ELISA). Serum samples were collected at day post-inoculation (DPI) −7 to 42, from pigs (n = 83) experimentally inoculated with different porcine coronaviruses (PorCoV). The diagnostic sensitivity of the PDCoV S1-based ELISA was evaluated using serum samples (n = 72) from PDCoV-inoculated animals. The diagnostic specificity and potential cross-reactivity of the assay was evaluated on PorCoV-negative samples (n = 345) and samples collected from pigs experimentally inoculated with other PorCoVs (n = 472). The overall diagnostic performance, time of detection, and detection rate over time varied across different S/P cut-offs, estimated by Receiver Operating Characteristic (ROC) curve analysis. The higher detection rate in the PDCoV group was observed after DPI 21. An S/P cut-off of 0.25 provided 100% specificity with no serological cross-reactivity against other PorCoV. These results support the use of S1 protein-based ELISA for accurate detection of PDCoV infections, transference of maternal antibodies, or active surveillance.

PDCoV (originally "PorCoV HK15") was first detected and identified by sequencing during a surveillance study carried out in Hong Kong in 2012 [2], but it was not until 2014 that PDCoV emerged in the USA (Ohio and Indiana) associated with clinical cases of diarrhea, vomiting, and dehydration in suckling pigs [7,8]. Subsequently, it spread
This S1 recombinant protein was then used to develop an indirect (IgG) ELISA. The distribution of PDCoV S1-based ELISA IgG S/P values obtained for each inoculation group is shown in Figure 3. The ROC analysis performed on the cumulative data collected from PDCoV-positive and PDCoV-negative samples tested by the S1-based indirect ELISA showed that the diagnostic performance of the PDCoV S1-based ELISA was subjected to a specific cut-off ( Figure 4). The use of an ELISA S/P cutoff of 0.25 ensured a specificity of 100%, without serologic cross-reactivity against other PorCoV, while a cutoff of 0.10 maximized the diagnostic specificity. Thus, based on the ROC analysis, the detection rate of seropositive animals over time was estimated at different S/P cutoffs (0.10, 018, and 0.25) ( Table 1). The first positive animals were detected between DPI 7 (3/12), when using an S/P cutoff of 0.1, and DPI 10 using a cutoff of 0.18 (2/12) or 0.25 (3/12). The higher detection rate (12/12) was observed at DPI 42 with an S/P cutoff of 0.10 (Table 1). Variable seroconversion was observed across pigs regardless the cutoff. Only one pig over 12 (pig 9) was seronegative throughout the study when an S/P cutoff value of 0.25 was used for result interpretation. and dialyzed to phosphate-buffered saline pH 3.0 for next step Tobacco Etch Virus (TEV) cysteine protease enzyme digestion (b) Lane M: molecular weight protein marker; Lane 1: PDCoV S1 protein after TEV-cleavage (reduced; red arrow); Lane 2: PDCoV S1 protein after TEV-cleavage (non-reduced; white arrow); Lane 3: PDCoV S1 protein before TEV-cleavage (non-reduced); Lane 4: PDCoV S1 protein before TEV-cleavage (reduced; white arrow). Eluted fractions after TEV-cleavage were subjected to further purification by HisTrap TM FF and protein A affinity columns (GE Healthcare, Chicago, IL, USA) to separate the cleaved Fc tag (c) Lane 1: PDCoV S1-Fc fused protein; M: molecular weight protein marker; Lane 2: load sample (column equilibrated with 20 mM phosphate buffer, 500 mM NaCl, pH 7.4); Lane 3-4: flow-through; Lane 5-6: elution by 0.1 M glycine pH 2.5; Lane 7: elution by 0.1 M glycine pH 2.5 (non-reduced); Lane 8: elution by 0.5 M glycine pH 2.5. Lane 7 protein was subjected for next step purification through exclusion chromatography (Superdex ® 200; GE Healthcare), and selected fractions were dialyzed against phosphate buffered saline pH 7.4 as final product (d) M: molecular weight protein marker; Lane 1: PDCoV S1 protein (~55.8 kDa) (reduced); Lane 2: S1-PDCoV (non-reduced).
This S1 recombinant protein was then used to develop an indirect (IgG) ELISA. The distribution of PDCoV S1-based ELISA IgG S/P values obtained for each inoculation group is shown in Figure 3. The ROC analysis performed on the cumulative data collected from PDCoV-positive and PDCoV-negative samples tested by the S1-based indirect ELISA showed that the diagnostic performance of the PDCoV S1-based ELISA was subjected to a specific cut-off ( Figure 4). The use of an ELISA S/P cutoff of 0.25 ensured a specificity of 100%, without serologic cross-reactivity against other PorCoV, while a cutoff of 0.10 maximized the diagnostic specificity. Thus, based on the ROC analysis, the detection rate of seropositive animals over time was estimated at different S/P cutoffs (0.10, 018, and 0.25) ( Table 1). The first positive animals were detected between DPI 7 (3/12), when using an S/P cutoff of 0.1, and DPI 10 using a cutoff of 0.18 (2/12) or 0.25 (3/12). The higher detection rate (12/12) was observed at DPI 42 with an S/P cutoff of 0.10 (Table 1). Variable seroconversion was observed across pigs regardless the cutoff. Only one pig over 12 (pig 9) was seronegative throughout the study when an S/P cutoff value of 0.25 was used for result interpretation.

Discussion
In the absence of commercial vaccines, the detection of PDCoV antibodies indicates that an animal is or was previously infected or received passive antibodies through lactation. Different immunoassays for the detection of PDCoV antibodies have been described [31][32][33][34][35][36][37]. However, pigs are exposed to different coronaviruses, and the potential antibody crossreactivity may contribute to false-positive results, particularly when using conservative antigens for assay development [38][39][40]. Therefore, any immunoassays would have limited utility in the field unless the absence of potential cross-reactivity against other porcine coronaviruses commonly circulating in commercial swine herds is ruled out.
In a previous study focused on PEDV, we demonstrated that among different structural proteins evaluated, the N-terminal portion of the S protein (S1) not only contains major antigenic sites, but it is also virus-specific, showing no serological cross-reactivity with other porcine coronaviruses [39]. Immunoassays based on a more conservative protein (e.g., structural protein N) [23,31,33,38,41] are more susceptible to potential cross-reactivity with other coronaviruses. Therefore, among the different PDCoV structural and nonstructural proteins, the present study evaluated the suitability of the S1 recombinant protein for specific detection of PDCoV antibodies used on an indirect ELISA platform.
As with other coronaviruses, PDCoV S protein is a glycosylated structural protein that consists of two subunits, the N-terminal S1 receptor-binding globular hydrophilic head and the C-terminal S2 membrane-fusion stalk, which is more conserved and highly hydrophobic [42]. Here, the S1 subunit was produced using a mammalian expression system and purified following a stepwise purification approach that combined affinity chromatography techniques, aided by the addition of both IgG-Fc and His tags, and molecular exclusion purification techniques. This strategy allowed for the generation of a recombinant protein in its native structure, including the appropriate post-translational modifications required for a glycosylated protein to be functional, e.g., proper antigen recognition, and receptor binding. Although the aminopeptidase N (APN) was initially proposed as a cellular receptor for PDCoV [43,44], further evidence suggested the existence of additional unknown receptors for PDCoV [45]. Different studies identified major APN receptor-binding domains in the PDCoV S1 protein that are key targets of neutralizing antibodies [46,47]. The neutralizing activity was partially attributed to the blockage of sugarbinding activity, highlighting the importance of glycosylation and native conformation of the S1 protein [47].
Previous reports described the development of PDCoV ELISAs, based on the S1 protein [32,36], but none of them adequately addressed the potential cross-reactivity with other PorCoVs. Thus, the main goal of this study was to assess the presence or absence of antibody cross-reactivity against anti-sera obtained from pigs individually infected under experimental conditions with different porcine coronaviruses currently circulating in most commercial herds. As expected, the diagnostic sensitivity and specificity, and the analytical specificity were directly associated with the specific cut-off used to interpret the ELISA results. No serological cross-reactivity was detected by ELISA when using an S/P cut-off value of 0.25, providing 100% specificity, though this had an impact on the overall time of detection and overall diagnostic sensitivity. Although all pigs were inoculated at the same time and with the same inoculum, both the time and the pattern of seroconversion varied across pigs. Contrarily, an S/P cutoff of 0.10 maximized the detection rate over time and overall diagnostic sensitivity but compromised the diagnostic specificity. In any case, the selection of an appropriate cutoff should be linked to the purpose of the testing and status of the herd (e.g., diagnosis vs. surveillance).
Previous studies in poultry [48], calves [49], mice [47], and even humans [50] evidenced the potential of PDCoV for interspecies and cross-species transmission. The PDCoV S1based indirect ELISA described herein could be easily adapted for antibody detection in different animal species simply by switching the labelled secondary antibody, i.e., speciesspecific antibody or non-species-specific protein A or G for primary binding antibodies. This could help in future serological investigations of the potential circulation of PDCoV across species.
This study concluded that, despite the complexity of the Coronaviridae family, coronavirus-specific serological tools could be developed when the right antigen target (i.e., S1) and the testing platform are appropriately selected and designed.

Experimental Samples of Known Porcine Coronavirus Infection Status
The animal study was approved by the Iowa State University Office for Responsible Research and the Institutional Animal Care and Use Committee (IACUC #12-17-8658-S). A conventional wean-to-finish farm with no history of porcine coronavirus infections was selected as the source to procure the pigs used in this study. Seven-week-old pigs (n = 83) that tested negative for PDCoV, PEDV, TGEV, PRCV, and PHEV by quantitative reverse transcription PCR (RT-qPCR) [23,[51][52][53] and serologic [immunofluorescence assay (IFA) or enzyme-linked immunosorbent assay (ELISA)] methods [23,39,40,52,54] were randomized into seven inoculation groups allocated to separate rooms. Detailed information about specific virus strains, virus titers, inoculum, inoculation routes, and number of pigs per group of inoculation is presented in Table 2. The pigs were closely observed twice daily for clinical signs throughout the study. Serum samples were collected from each group on day post-inoculation (DPI) -7, 0, 3, 7, 10, 14, 17, 21, 28, 35, and 42. All pigs were euthanized at DPI 42 by penetrative captive bolt (Accles and Shelvoke, Ltd., Sutton Coldfield, UK).
Serum samples (single well) and internal positive and negative controls (duplicate) were diluted to 1/100 in PBS pH 7.4 solution containing 50% goat serum (Gibco ® , Thermo Fisher Scientific), and added to the PDCoV S1 protein-coated plates. The plates were incubated for 1 h at 37 • C, then washed three times with 350 µL per well of PBST. Next, 100 µL of HRP-labelled goat anti-pig IgG (Fc) (Bethyl Laboratories Inc., Montgomery, TX, USA) at 1:50,000 was added to each well, followed by a 30 min incubation at 37 • C. After an additional washing step with PBST, a 100 µL tetramethylbenzidine-hydrogen peroxide (TMB) substrate solution (SurModics IVD, Inc., Eden Prairie, MN, USA) was added to each well and incubated for 5 min at room temperature in the dark. The reaction was stopped by adding 100 µL of stop solution (SurModics IVD, Inc.) per well, and the optical density (OD 450 ) at 450 nanometers was measured using an ELISA plate reader (SoftMax Pro 7; Molecular Devices, San Jose, CA, USA) operated with SoftMax Pro7 software (Molecular Devices). Antibody responses were expressed as the sample-to-positive (S/P) ratios: S/P ratio = (sampleOD 450 − negativecontrolmeanOD 450 ) (positivecontrolmeanOD 450 − negative control meanOD 450 )

Data Analysis
The ELISA S/P serum IgG responses across inoculation groups were plotted using GraphPad Prism ® (GraphPad Software Inc., La Jolla, CA, USA). The R software package pROC (https://cran.r-project.org/web/packages/pROC/index.html, accessed on 24 July 2022) [56] was used to perform a Receiver Operating Characteristic (ROC) analysis on experimental samples of known PorCoV infection status to estimate the diagnostic performance of the PDCoV S1-based indirect ELISA. The non-parametric DeLong method was used to estimate the 95% confidence intervals (CIs) for the area under curve (AUC) [57]. Before performing the analyses, the S/P data were normalized by a 1/3 power transformation; diagnostic sensitivity and specificity were derived from the ROC analyses for specific assay cut-offs.
The diagnostic sensitivity was estimated on sera from PDCoV-inoculated pigs collected after 14 DPI (n = 72). The diagnostic specificity and potential cross-reactivity of the PDCoV S1-based ELISA was assessed on serum samples from PorCoV-negative pigs (n = 333) and pigs inoculated with PEDV, TGEV, PRCV, and PHEV between DPI 7 to 42 (n = 472).
Typical methods for estimating confidence intervals of proportions are based on binomial distribution and do not account for the correlated structure of longitudinal data, i.e., repeated observations from the same animals over time. Therefore, a method for deriving CI for correlated, normally distributed data was utilized [39]. In brief, the correlation of the data was taken into account by fitting normalized data into a linear mixed model.
In Equation (1), Y ij is the j th observation for the i th subject, µ is the overall mean for samples classified as PDCoV antibody-negative, γ i is the random effect of the i th subject, τ is the fixed effect indicating the mean difference between PDCoV antibody-negative and positive groups; s ij is the disease status of the j th observation for the i th subject; and ij is the random error of the j th observation for the i th subject. The variances from equation 1 were used to calculate the 95% CIs for ROC-derived diagnostic sensitivity and diagnostic specificity estimates. Logit transformation was used to prevent the estimated intervals from exceeding the range of probability, i.e., (0, 1).