Detection of Zaire ebolavirus in swine: Assay development and optimization

Ebolaviruses (family Filoviridae , order Mononegavirales ) cause often fatal, haemorrhagic fever in primates including humans. Pigs have been identi ﬁ ed as a species susceptible to Reston ebolavirus (RESTV) infection, with indicated transmission to humans in the Philippines; however, their role during Ebola outbreaks in Africa needs to be clari ﬁ ed. To perform surveillance studies, detection of ebolavirus requires a prerequisite validation of viral RNA and antibody detection methods in swine samples. These diagnostic tests also need to be suitable for deployment to low-level containment laboratories. In this study, we developed a set of tests for detection of antibodies against Zaire ebolavirus (EBOV) in swine. Recombinant EBOV nucleoprotein was produced using a baculovirus expression system for indirect ELISA development. Evaluation of this assay was performed using laboratory and ﬁ eld samples, achieving a diagnostic speci ﬁ city of 99%. Importantly, the indirect ELISA was able to detect antibodies to EBOV at 7 dpi, 3 days earlier than virus neutralization tests (VNT). The format of the VNT in this work was modi ﬁ ed to a microtitre plaque reduction neutralization assay (miPRNT) complemented with immunostaining to provide a more rapid and highly speci ﬁ c assay. Finally, a con ﬁ rmatory immunoblot assay was generated to supplement the indirect ELISA results. to

clinical disease in experimentally infected pigs, while displaying the ability to transmit to in-contact pigs or indirectly to non-human primates (Kobinger et al., 2011;Weingartl et al., 2012;Nfon et al., 2013). Thus, in the aftermath of the unprecedented outbreak of EBOV in West Africa, it is important to elucidate a potential role of swine in the transmission chain of ebolaviruses (Atherstone, Smith, Ochungo, Roesel, & Grace, 2015). To support serosurveillance of swine in Africa, diagnostic assays evaluated using swine samples need to be available. Due to biosafety (BSL4) requirements for work with live EBOV, preference should be given to assays which inactivate the viruses in early steps and use recombinant reagents, further limiting the need for live virus work.
Immunohistochemistry is not used in human diagnostics during outbreaks to detect the ebolaviruses; however, it can be used in retrospective studies focusing on pathogenesis (Martines, Ng, Greer, Rollin, & Zaki, 2015). In contrast, immunohistochemistry is a widely accepted approach in veterinary diagnostics with the advantage of being performed outside CL4 conditions following fixation of the tissues in formalin. Immunohistochemical detection of ebolaviruses was used in several pathogenesis studies in multiple animal species, including cynomolgus macaques, guinea pigs, bats from the genus Rousettus, in addition to swine (Jones et al., 2015;Wong et al., 2015;Weingartl et al., 2012;Kobinger et al., 2011).
Ebolavirus titration and virus neutralization assays are assessed using three approaches under BSL4 conditions to quantify virus replication: TCID 50 , plaque-forming units using neutral red overlay (Shurtleff et al., 2016 ) and fluorescent foci units. Drawbacks to these procedures include relatively long incubation times for the first two methods, and in the case of TCID 50 , assays may be difficult to read. The third method, detection of fluorescent foci, uses GFPlabelled recombinant virus providing improved visualization; however, high containment is still required. To alleviate the need for BSL4 conditions, an alternative option is to use a pseudotyped virus such as VSV-ebolavirus G protein, provided access to a high-resolution fluorescent microscope is available (Ebihara et al., 2007;Cap et al., 2016;Takada et al., 1997). An additional drawback is the time required to develop multiple recombinant or chimeric viruses which may be necessary, due to lack of cross-reactivity between the recombinant pseudotyped recombinant virus and antibodies developed against the circulating virus.
The preferred assay for detection of antibodies for BSL4 agents is an ELISA which employs recombinant antigen. Serum samples are the only component for this test potentially containing ebolavirus, thereby requiring inactivation. An established method widely used in the field is heat inactivation of filoviruses in serum samples at 60°C for 1 hr (Mitchell & McCormick, 1984). More recently, other inactivation methods have been developed, but in general require more sophisticated equipment (Cap et al., 2016). Experimental ELISAs using recombinant antigens have been developed for human diagnostic testing in both human and non-human primate samples (Ikegami et al., 2003;Nakayama et al., 2010). In addition, some of these tests have also been adapted for use in the field (Vu et al., 2016;Nidom et al., 2012). A number of reagents intended to develop ELISAs for the detection of ebolavirus antibodies have recently become commercially available to advance future test development. As with any new test, it is critical these tests are validated or at least evaluated for use in the target species. The aim of this work was to develop and evaluate serological tests suitable for the detection of antibodies to EBOV infections in swine for diagnostic purposes.

| Viruses
Virus stocks of EBOV-Kikwit and EBOV Gueckedou were prepared in Vero E6 cells (ATCC) in T75 flasks (Corning). The flasks were frozen after 9 days of incubation at 37°C in 5% CO 2 atmosphere and thawed, and the supernatants were clarified by centrifugation at 2000 g for 15 min, and the aliquots stored at À150°C.

| Serum utilized for assay development
Swine samples generated in previous experimental infections of pigs with 10 6 TCID 50 of EBOV-Kikwit were used for the development and evaluation of all assays described in this report (Kobinger et al., 2011;Weingartl et al., 2012;Nfon et al., 2013). The 264 negative control sera used for assay development were obtained either from an archived collection of experimental samples at the National Centre for Foreign Animal Disease (NCFAD) or submitted to the NCFAD for laboratory diagnosis. Field sera (total of 51) were collected in the summer of 2014 in Guinea Forresti ere, which borders Sierra Leone.

| Plaque immunostaining
Immunostaining was performed in 96-well plates on cells fixed with formalin for 24 hr. Formalin and overlay were removed from plates and captured for proper disposal. The plates were then washed twice with Milli-Q water and incubated in 0.3% Triton X-100 (Sigma-Aldrich) for 10 min at room temperature. The cells were washed two times with 100 ll/well of Milli-Q water and twice with Tris-buffered saline -0.05% Tween-20 (TBST; M edimabs). Plates were then incubated with 50 ll/well of monoclonal antibody against ZEBOV VP40 (cat#:EVP407-M, Alpha Diagnostics) diluted 1/4000 in TBST for 1 hr at 37°C on a shaker set at 650 rpm and washed twice with 100 ll/well of TBST. EnVision+ System-HRP (horseradish peroxidase-labelled polymer conjugated with goat anti-mouse antibodies) was used to detect the primary mouse monoclonal antibody (cat#:K4007; Dako). Plates were incubated with the EnVision system for 30 min at room temperature on a shaker set at 650 rpm, washed twice with TBST and twice with Milli-Q water, followed by 5-min incubation with DAB (3,3 0 -diaminobenzidine; Dako) at room temperature (one drop DAB/ml of diluent). The plates were washed a further two times with 100 ll/well of Milli-Q water and allowed to dry.
It is recommended to read the plaques with a magnifying glass or under a low-power microscope.

| Microtitre plaque reduction neutralization assay (miPRNT)
All sera were heat inactivated for 1 hr at 56°C, diluted serially 2fold in DMEM, and individual aliquots were incubated with 100 PFU of ZEBOV at 37°C/5% CO 2 for 1 hr. Vero E6 cell monolayers in 96-well plates were washed once with DMEM before the addition of 100 ll/well of the virus/serum sample mix. This inoculum was allowed to adsorb on the cells for 1 hr at 37°C/5% CO 2 and then replaced with 200 ll/well of a CMC overlay. The plates were kept at 37°C/5% CO 2 for 4 days in BSL4 when they were fixed with 10% formalin for 24 hr. The plates were immunostained as described above following the removal from BSL4. Antibody titres were determined based on 90% inhibition of viral plaques compared to virus controls. Serum was assayed in duplicate for each sample.

| Antigen preparation
The full-length Ebola-Kikwit nucleoprotein (NP; KC242799.1) gene was synthesized by Genescript TM into pAB-bee TM -FH vector (AB vector). This pAB-bee TM -FH-NP construct was transfected into Sf9 cells with Profold TM -ER1 baculovirus according to the supplier (AB vector) and incubated for 4 days at 28°C. Cells were pelleted at maximum speed (Eppendorf 5417R) for 5 min, and the baculovirus containing supernatant was transferred to a new tube. Baculovirus titrations were performed by infecting new semi-confluent monolayers for 1 hr, followed by the addition of insect overlay (Sf-900 Medium 1.39 with 4% agarose, Gibco). Plaques were picked based on green fluorescence intensity (Olympus DP80 microscope with X-cite Q series 120) after 5 days of incubation at 28°C and transferred to fresh insect media establishing the P 0 generation of Ebola-NP-producing baculovirus (newly purified P 0 plaques were subsequently used to infect new Sf9 cells producing a P 1 generation). Protein expression by Western blot was verified in five separate baculovirus isolations. Cell monolayers were harvested for both supernatant and cell pellets, to test for antigen expression by Western blot using an Anti-His HRP-conjugated antibody (EMD Millipore). Baculovirus plaque isolates exhibiting robust Ebola-NP protein expression were amplified a further two rounds in liquid suspension as follows: first, the P 2 generation was amplified by infecting 5 9 10 5 Sf9 cells in 30 mls of insect cell media with P 1 generation at a dilution of 1:1000.
The flask was incubated at 28°C with shaking (120 rpm) for 7 days; baculovirus Ebola-NP-producing cells were harvested by centrifugation at 930 g for 20 min. A P 3 generation of baculovirus Ebola-NP was generated by infecting Sf9 cells in suspension again, set-up at 5 9 10 5 cells per ml in 100 ml of insect media with P 2 generation baculovirus at a dilution of 0.001. After 7 days at 28°C with shaking, cells were centrifuged at 930 g for 20 min. Overexpression of Ebola-NP antigen was established by infecting TINI HiFi insect cells 1:2 with P 3 generation baculovirus containing pAB-bee TM -FH-NP. Following 5 days incubation at 28°C, supernatants were harvested and stored, while cell pellets were lysed by adding I-PER buffer (ThermoFisher Scientific I-PER Insect Cell F I G U R E 2 Optimization of antigen coating and serum titration. Panel A; Serum from EBOV infected or uninfected swine was titrated across various antigen concentrations to define optimal plate coating for ELISA assay. A selection of negative ■ or , false positive ; weak (7 dpi) ▲; 21 dpi •; or 28 dpi ♦ sera was utilized. (In all cases, serum was diluted to 1:100). Panel B; the same set of sera from panel A, now presented as sera from EBOV infected ♦ or uninfected swine was serial diluted to determine the optimal dilution for ELISA assay. Neutralizing activity was determined for serum diluted less than or up to 1:100 3 | RESULTS

| Virus plaque assay and virus neutralization assay
Classical Ebola-Zaire virus isolation performed in Vero E6 cells requires 10 days of incubation to obtain discernible cytopathic effect (CPE). In general, routine diagnostic assays based on CPE formation are not optimal due to subjectivity in reading the CPE (i.e., TCID 50 ) and the time required to perform the assay. A microtitre immunostained plaque assay (miPRNT) was developed to provide simultaneous quantification and confirmation of the presence of EBOV, with a turn-around time of 4-5 days (Figure 1). Although the assay was developed for the purpose of virus titration, it was also used for virus isolation from tissue homogenates or fluid samples, as well as in virus neutralization assays. The miPRNT was evaluated on archived samples and confirmed by previously reported neutralizing antibody titres (Kobinger et al., 2011;Weingartl et al., 2012;Nfon et al., 2013).  (Table 1), which were also confirmed as positive by immunoblot. Furthermore, all infected animals starting at 10 dpi were detected by indirect ELISA, with confirmation by both immunoblot and miPRNT (VNT data shown in bold, Table 1). The sera from infected animals tested positive until the last collection day at 28 dpi, although there appeared to be a decline in OD values when  Figure 4). The three samples with OD 405 above 0.5 also tested negative by virus neutralization.

| Indirect IgG ELISA and Immunoblot
All the field sera from Guinea were negative for antibodies against EBOV-NP protein by indirect ELISA. When those were included in the negative sample set, the diagnostic specificity reached 99%. Further, background OD 405 values were lower for the Guinea field sera compared to NCFAD serum ( Figure S2), indicating reduced cross-reactivity from the sera tested in this study.

| DISCUSSION
To date, there are no validated assays for the detection of EBOV in swine. Considering the potential role swine may play in the transmission chain of ebolaviruses, it is therefore essential to have the capability to detect both virus and antibodies in this species. Here, we report the development and evaluation of two serological assays for the detection of antibodies in swine inoculated with EBOV: a microtitre immunostained plaque reduction neutralization test (miPRNT) and an IgG detecting indirect ELISA. Samples from experimentally infected animals were used in the evaluation of these assays. Successful infection of animals was previously confirmed by virus isolation and by real-time RT-PCR (Kobinger et al., 2011;Weingartl et al., 2012;Nfon et al., 2013).
An immunostaining approach for visualization of plaques in the miPRNT was adopted from previously developed and reported immunohistochemical staining of tissues. This approach mitigated the extended time requirements and the subjective bias employed in the interpretation of results when using traditional virus neutralization assays (Kobinger et al., 2011;Weingartl et al., 2012;Nfon et al., 2013).
Although employing immunostaining for virus detection by plaque isolation or virus neutralization plaque reduction assay did not eliminate the need for BSL4 containment, introduction of immunostaining provided several advantages. First, the plates were fixed relatively early post-infection compared to classical plaque based assays, shortening the incubation time from 9-10 days to 4 days. Secondly, the plaques are easily discernible by the naked eye or at low magnification, and for detection of virus by plaque isolation, the use of specific primary antibody for immunostaining offers species confirmation of an ebolavirus in question. Moreover, immunostaining is also readily modified for detection of antibodies against different ebolavirus species using specific antibodies for plaque staining. This eliminates the need to develop new recombinant viruses required for virus neutralization assays in low containment.  (Nakayama et al., 2010). As such, the use of NP may provide an opportunity for indirect ELISA use against different Ebola species. Limited work has been targeted to diagnostic assay development in livestock, and therefore, it is unknown how the assay will perform when comparing different ebolavirus sera. However, preliminary results from experimentally challenged swine with Sudan virus suggest there is cross-reactivity lending the potential for broader diagnostic application.
Based on experimental inoculations, it appears that EBOVinfected animals develop neutralizing antibodies starting at 10 dpi, which is in agreement with the delayed development of antibodies reported for RESTV (Marsh et al., 2011). In comparison, the ELISA was able to detect all infected animals at 7 dpi. Due to the restraints of keeping animals in high containment for long periods of time, the duration of antibody detection in swine following EBOV infection is unknown. There appears to be a decrease in the anti-NP antibodies at 28 dpi compared to 21 dpi, and it cannot be excluded that in the field after several months, the indirect ELISA may not be able to detect animals infected prior to this time point.
The titres of virus neutralization antibodies on the other hand appeared to be maintained. Therefore, it would be important at least in the initial serosurveys to ensure that the samples are analysed by ELISA with confirmation by virus neutralization assay (miPRNT) and immunoblot.
Early detection of infected pigs is critical for preventing transmission to other livestock or possibly humans. In the light of the recent West African outbreak, the potential threat EBOV represents in endemic countries or if introduced into North America is significant.
The ELISA assay developed in this work, combined with virus detection in oral-nasal fluids by real-time RT-PCR, provides a viable option for rapid outbreak detection. RT-PCR detection of EBOV oral-nasal shedding was detected as early as 1 or 2 days post infection and up to 7 dpi, while ELISA detected the anti-NP antibodies as early as 7 dpi in all animals. Thus, all infected animals would be detected very early post-infection by either approach. It should be noted that based on preliminary unpublished data, it appears specific tests need to be developed and validated for each Ebola species.

ACKNOWLEDG EMENTS
We would like to thank both Matthew Suderman and Mathieu Pinette for their assistance with antigen production and immunoblot analysis and Dr. Sandra Diederich for critical review of the manuscript.