SFTSV isolated from a human patient in Korea in 2013 was kindly provided by the Korea Centers for Disease Control and Prevention (KCDC) [25]. The SFTSV was grown in Vero E6 cells (American Type Culture Collection, USA) according to a previously described method [23].

For antigen production, the gene encoding the full-length NP protein of SFTSV strain LN3 (GenBank accession No. HQ141612) was expressed in a bacterial overexpression system. The S fragment of SFTSV was chemically synthesized using the Bioneer gene synthesis service (Bioneer, Korea). The 735 bp gene encoding N-protein was amplified using the following primers: SFTS-NP-1 CTCGGAATTCACATGTCA GAGTGGTCC and SFTS-NP-735 CTTCAAGCTTCAGGTT CCTGTAAGCAG. The PCR amplification was performed using a T3000 thermocycler (Biometra, Germany). The NP gene was cloned into pET-30a(+) (Invitrogen, USA) using the BamHI and XhoI restriction enzymes (New England Biolabs, UK), and subsequently transformed into Escherichia coli BL21 (DE3) (Yeastern Biotech, Taiwan) to express 6xHis-tagged fusion proteins. Following induction with 0.2 mM isopropyl β-D-1-thiogalactopyranoside (AMRESCO, USA) for 20 h at 25℃, the bacterial cell pellet was sonicated in chromatography buffer (20 mM sodium phosphate, 500 mM NaCl, 8 M Urea, and 20 mM imidazole, pH 7.4) and purified using Ni-NTA agarose (Qiagen, Germany) [22]. The affinity-purified protein was urea gradient dialyzed prior to use. The recombinant NP protein was solubilized in 8 M urea buffer, which was then exchanged for 150 mM Tris-HCl (iNtRon Biotechnology, Korea) for further use in subsequent experiments.

Animal experiments were conducted according to the protocol of the Institutional Animal Care and Use Committee of the Republic of Korea. Rabbit polyclonal antibody and mouse mAbs were generated after four immunizations with recombinant NP. Polyclonal antibodies were produced in rabbits after four injections of 500 µg of the NP antigen mixed with adjuvant at two-week intervals. The NP-specific mAbs were synthesized as previously described [2]. Briefly, eight mice were injected four times with 100 µg of recombinant NP. Following immunization, the mouse spleen cells were collected and fused with myeloma cells to generate mAbs. Among the mouse hybridoma cell lines, 14 clones were selected according to specific interactions with NP using indirect ELISA.

The cattle were immunized with an injection of formalin-inactivated SFTSV. Briefly, 60 mL of 10⁸ TCID₅₀/mL of the SFTSV was inactivated using 0.025% formalin solution for 72 h. The inactivated virus was concentrated fivefold via filtration using a 10K Ultracel (Millipore, Germany). The loss of viral infectivity was confirmed by culture in Vero cells. Two cattle aged 4 months were used to produce positive control sera against SFTSV. Approximately 2 mL of inactivated SFTSV was emulsified in gel-based adjuvant, and the cattle received intramuscular injections every 2 weeks, with a total of three 3 injections. Serum samples were collected 2 weeks after the last antigen injection. The serum used as a negative control in IFA and cELISA was collected from cattle prior to immunization.

IFA was performed as previously described, with slight modification [23]. Briefly, Vero E6 cells grown to 80% confluency in 150 cm² were infected with 2 mL of 1 × 10³ TCID₅₀/mL SFTSV. After 6 days, the infected cells (~3,000 cells per well) were spotted onto the microscope slides with reaction wells (Paul Marienfeld, Germany). The cells were immediately fixed with a methanol : acetone (1 : 1) solution for 30 min, after which non-specific signals were blocked with 5% horse serum in a total volume of 20 µL per well. The cells were incubated with mouse mAbs, α-NP specific polyclonal rabbit sera or field bovine serum samples for 1 h at room temperature, followed by detection using fluorescein isothiocyanate (FITC)-labeled secondary antibodies. The FITC-conjugated anti-mouse, rabbit, and bovine antibodies (KPL, USA) were diluted to a concentration of 2.5 µg/mL prior to use. The cells were observed using a Nikon TE-2000U fluorescence microscope (Nikon, Japan).

cELISA was performed as previously described, with slight modification [8]. Briefly, a Polysorp ELISA plate (Nunc, USA) was coated with 50, 100 or 200 ng/well of recombinant NP. The plate was then incubated for 16 hours at 4℃ to allow antigen binding. After washing with 0.5% Tween 20 (Sigma-Aldrich, USA) in 1× PBS to remove the unbound antigens, the plate was blocked with 5% skim milk and 1% horse serum in 1× PBS. For the serological test, 50 µL of diluted test sera and the same amount of HRP-conjugated mAb (1 : 500) were added to the antigen coated plate and incubated for 90 min at 37℃. The mAb was conjugated with HRP according to the instructions provided with the EZ-LinkTM Plus Activated Peroxidase kit (Thermo Fisher Scientific, USA). After washing the plate six times, 3,3',5,5'-tetramethylbenzidine peroxidase substrate (KPL) was added to develop the color. The reaction was terminated with 1 M sulfuric acid (Sigma-Aldrich). The optical density (OD) of the sample was calculated as follows: the OD at 450 nm was subtracted from the OD at 630 nm to adjust for the background absorbance. The OD values were converted to percent inhibition (PI) values using the following formula: PI =[1 – (OD test sample/OD negative control)] × 100.

Bovine field serum samples (n = 416) were collected from breeding cattle in the southern part of Korea (Gyeongsang province) from September 2013 to August 2014.

cELISA was conducted using antisera against different viruses: bovine viral diarrhea (BVD) type 1 virus neutralization positive control serum, BVD type 2 virus neutralization positive control serum (National Veterinary Services Laboratories, USA), bovine coronavirus, bovine rotavirus, Bovine leukemia virus and Akabane virus (93FMX) (generated in the Animal and Plant Quarantine Agency, Korea).

The positive and negative groups were compared by one-way ANOVA and the Mann-Whitney rank sum test, with p values ≤ 0.001 indicating a significant difference between groups (SYSTAT ver. 4.0; Systat Software, USA). Bee swarm and box plots indicating the distributions of negative or positive groups were generated using an online application [14]. Other graphs and receiver operating characteristic (ROC) [27] curves were generated using the Excel graph function.

The gene for the recombinant NP was amplified from the commercially synthesized S fragment of SFTSV (GenBank accession No. HQ141612) and cloned into a protein expression vector. A 34 kDa recombinant protein band was well maintained after all purification procedures (Fig. 1). This recombinant NP was used as the antigen to generate polyclonal and mAbs. NP-specific polyclonal antibody was obtained in rabbits after four antigen immunizations, and the titer of the antibody was tested by IFA and Western blotting (Fig. 2). NP-specific mAbs were produced from the spleen hybridoma cells of immunized mice, and the three clones (6D55, 8F31, and 10G7) showed strong interactions in the IFA and Western blot tests; therefore, these clones were used as competitive antibodies to develop the cELISA. A clear band corresponding to the size of NP, which appeared only after SFTSV infection, was observed upon Western blot analysis with all generated antibodies (panel A in Fig. 2). Moreover, IFA using these antibodies showed intense cytoplasmic staining, specifically in SFTSV-infected Vero cells (panel B in Fig. 2).

SFTSV-positive cattle sera were generated from two cattle that were injected three times with inactivated SFTSV. Serially diluted positive or negative serum samples were tested to determine the optimized dilution rate to distinguish SFTSV-positive serum samples from negative samples in IFA. The maximum dilution rate of serum for the IFA at which a non-specific background signal was not observed in the negative control was 1/80 (Fig. 3).

ELISA plates coated with different concentrations of antigen were used to determine the optimal coating antigen concentration (Supplementary Fig. 1). Among the mAb clones screened based on IgG responses in the IFA and Western blot tests, 10G7 was selected as the competitive antibody for cELISA because this mAb showed stable reactivity after the conjugation of horseradish peroxidase (HRP) protein. The 1 : 500 diluted mAb-peroxidase conjugate yielded the maximum percent inhibition rate in the cELISA (Supplementary Fig. 1).

The immunized cattle serum samples were tested for the cELISA. Both 1 : 2 and 1 : 5 dilutions of the positive serum samples showed higher PI values than the negative serum. However, a 1 : 5 dilution of the positive control sera yielded a large difference in the optical density (OD) between the positive and negative serum samples (Fig. 4). The repeatability of this finding was confirmed by three independent experiments.

The specificity of the SFTSV cELISA was examined using antisera against different types of viruses from cattle. The maximum PI value of laboratory-immunized bovine serum in the cELISA was 93%, while other antisera against different viruses, such as BVD type 1 and BVD type 2, bovine coronavirus, bovine rotavirus, Bovine leukemia virus and Akabane virus (93FMX), yielded PI values of – 18 to 11% (Supplementary Fig. 2). These findings demonstrate that the cELISA is highly specific for the detection of antibodies against SFTSV.

A total of 416 bovine serum samples and two positive control sera were tested with the IFA to distinguish negative serum samples from SFTSV. The results showed that 97.4% (407/418) of the bovine serum samples were negative, while 2.6% (11/418) of the bovine serum samples were positive according to the IFA. cELISA was performed using the SFTS-negative group samples, and the mean inhibition percent was 11.3% with a standard deviation of 12.7%. Therefore, we determined the cut-off value to be 49.5% (mean ± 3SD) (Fig. 5). The cELISA and IFA were compared using this cut-off value, and 98.1% (410/418) consistency was observed between results (Table 1). The accuracy of cELISA was demonstrated with a plot of the distribution of PI values obtained from the IFA-negative or positive cattle group (Fig. 6).

Because official standard methods for the antibody detection of SFTSV have not been determined, the cut-off criteria determination for IFA requires further optimization. Therefore, ROC curve was employed to determine the sensitivity and specificity and define the optimal cut-off value for IFA. The ROC curves of four different IFA cut-off values were compared, and the results showed that a 1/80 dilution of the sample strongly supports the result of cELISA. For this dilution, the area under the curve (AUC) was 0.91. While a 1/128 dilution rate as an IFA cut-off point maximizes the specificity and sensitivity compared with other dilution rates, this condition is too strict to detect positive samples above the strict threshold in cELISA. The other IFA cut-off conditions (1/16 and 1/64) were also excluded because they showed moderate AUC values (Fig. 7).

Using a 1/80 dilution as cut-off for the IFA and a PI of 49.5% as a cut-off for cELISA, 0.95% (4/418) of the bovine serum samples were identified as positive, while 97.1% (406/418) of the serum samples were negative.