Recombinant cytokine proteins were cloned and expressed in Pichia pastoris by Kingfisher Biotech, (Saint Paul, MN). At a contract facility, cytokine specific hybridomas were produced using BALB/c mice that were immunized subcutaneously twice at 4-week intervals (50 μg/dose) with swine IL-17A recombinant protein (rPoIL-17A; Kingfisher Biotech, Saint Paul, MN) or swine IFNγ recombinant protein (rPoIFNγ; Kingfisher Biotech, Saint Paul, MN). Once antibodies were detected in the serum, mice were injected with a final intravenous boost of rPoIL-17A or rPoIFNγ, and hybridoma fusion conducted (14). The primary hybridoma supernatants were screened for specificity by ELISA; supernatants positive for rPoIL-17A or rPoIFNγ, but negative for anti-carbohydrate reactivity, were cloned and expanded for mAb production and purification. A panel of 9 anti-PoIFNγ (αPoIFNγ) mAbs and 10 anti-PoIL-17A (αPoIL-17A) mAbs were selected for further characterization and validation by ELISA for specific binding, determinant reactivity, and intracellular staining ( Table 1 ).

Original hybridoma supernatants were screened by ELISA for antigen specificity with rPoIL-17A or rPoIFNγ, and for lack of reactivity with yeast carbohydrates using recombinant bovine IL-4 (rBoIL-4; Kingfisher Biotech, Saint Paul, MN). ELISA plates (ThermoFisher Fisher Scientific, Rochester, NY) were coated overnight with optimized amounts of recombinant proteins (1-2 μg/ml) in sodium carbonate/bicarbonate pH10 buffer. After washes, the plates were blocked with Phosphate Buffered Saline (PBS) with 1% Bovine Serum Albumin (BSA) (PBS-BSA) then washed before supernatants from different hybridoma clones were added for 1 hr. After washes, Peroxidase-Conjugated AffinePure Goat anti-mouse IgG (H+L) antibody (Jackson Immuno-Research Laboratories, West Grove, PA) was added and followed after 30 minutes with SureBlue Reserve Tetramethylbenzidine (TMB) Microwell Peroxidase Substrate (KPL, Gaithersburg, MD). Optical density (OD) at 650 nm was recorded using the VersaMax™ Tunable Microplate Reader (VWR, Radnor, PA). Only clones whose supernatants showed positive immunogen reactivity and no reactivity with rBoIL-4 protein (known high levels of carbohydrates) were selected for further expansion and screening for specificity.

Protein A purified αPoIL-17A or αPoIFNγ mAbs were biotinylated according to the manufacturer’s instructions using the EZ-Link™ Sulfo-NHS-LC-Biotin reagent (ThermoFisher Scientific, Waltham, MA). Purified αPoIL-17A or αPoIFNγ mAbs were incubated in ELISA plates precoated with purified rPoIL-17A or rPoIFNγ. The subsequent binding of biotin-labeled αPoIL-17A or αPoIFNγ mAbs was determined with Streptavidin-Horseradish Peroxidase conjugate (SAv-HRP) (Thermo Fisher Scientific, Waltham, MA). Percent inhibition of the binding of biotin-labeled mAb with a 100-fold excess of each purified (non-biotinylated) mAb was calculated. Antigenic determinants were assigned based on mAb cross-inhibition and binding to cross-species orthologs.

ELISAs were performed as noted in section ELISA Screening of Hybridoma Supernatants and mAb Characterization. Plates were coated with every available yeast expressed recombinant protein ortholog from different species (Kingfisher Biotech, Saint Paul, MN) and reactivity compared to that with rPoIL-17A or rPoIFNγ. Specifically, biotin-labeled anti-PoIL-17A mAbs were tested for binding to 13 rPoIL-17A orthologs from bovine, canine, mouse, dolphin, ovine, feline, zebrafish, human, caprine, rabbit, monkey, equine, and guinea pig; anti-PoIFNγ mAbs were tested against rPoIFNγ from the same species, except for guinea pig, which was replaced by murine. Any cross-species reaction whose ELISA OD was at equal or higher than 0.5 was considered positive.

Sets of purified αPoIL-17A mAbs were tested to determine the optimal set of mAbs for quantitation of IL-17A. Capture mAbs were diluted in sodium carbonate buffer at optimal concentration (5 µg/µl) to coat ELISA plates (Thermo Fisher Scientific, Rochester, NY). Recombinant PoIL-17A (rPoIL-17A) (Kingfisher Biotech, Saint Paul, MN) was diluted serially from 0-100,000 pg/ml in PBS-BSA and added to wells for 1 hr to test for standard curve sensitivity. Biotinylated αPoIL-17A mAbs were added to each well at 0.1 µg/ml for 1 hr. After washing, SAv-HRP was added, and reactivity measured with SureBlue Reserve TMB Peroxidase Substrate. Once the best standard curves were established, reactivity with native PoIL-17A was tested using supernatants from stimulated peripheral blood mononuclear cells (PBMCs) and compared to media control.

PBMCs were separated from pig blood by density centrifugation using a Lymphocyte Separation Medium LymphoSepTM (MP Biomedicals, Solon, OH) (15) and used fresh or frozen in liquid nitrogen until thawed for in vitro cultures. Frozen and fresh PBMC from multiple pigs were used for the studies conducted at BARC and OSU. All cell cultures were conducted in blastogenic medium [850 ml RPMI 1640 medium (ThermoFisher Scientific, Waltham, MA), 100 ml fetal bovine serum (FBS) (HyClone, Logan, UT), 25 ml 1M HEPES pH7.3, 5 ml 2-mercaptoethanol, 10 ml (10000UI/10000µg) Penicillin-Streptomycin, 10 ml 200 mM L-glutamine]. All cultures were incubated at 37°C/5% CO₂.

Native PoIL-17A was prepared from PBMC cultured in 6 well plates at 4 x 10⁶ cells/well. Cells were cultured in blastogenic medium with phytohemagglutinin (PHA) at 10 μg/ml or phorbol myristate acetate and ionomycin (PMA/Iono) at 50 ng/ml and 500 ng/ml, respectively, to induce cytokine production. Cells were incubated for 24 or 48 hrs in a 37°C humidified CO₂ incubator, then harvested and centrifuged. Supernatants were collected, aliquoted, and stored at -20°C until use for ELISA assay and not refrozen after use.

Purified αPoIL-17A or αPoIFNγ mAbs were labeled with AF647 according to the manufacturer’s instructions using the Alexa Fluor^® 647 Protein Labeling Kit (ThermoFisher Scientific, Waltham, MA). For cell culture, frozen PBMCs were thawed and cultured overnight in blastogenic medium in 6 well plates at 4 x 10⁶ cells/well. Cells were stimulated for 5 hrs with BD Leukocyte Activation Cocktail (BD Biosciences, San Diego, CA), which is a ready-to-use polyclonal cell activation mixture containing PMA/Iono, and a protein transport inhibitor (Brefeldin A). For cell surface staining, Fc receptors were first blocked for 30 min at 4°C in complete Flow Cytometry Medium (FCM) [PBS-BSA with 20 mM of Sodium Azide, and 1% normal Rabbit Serum]. For dead/viable cell exclusion, the cells were stained with the fixable viability stain 520 (FVS) (BD Biosciences, San Diego, CA) in PBS for 7 min at 37°C and washed twice with PBS-BSA. Alternatively, the cells were stained with the VivaFix cell viability dye (Bio-Rad, Hercules, CA) in PBS for 30 min at room temperature, then washed twice with PBS-BSA.

For intracellular staining, after staining for dead/viable cell exclusion as noted above, Fc receptors were blocked for 30 min at 4°C in complete FCM. The cells were then stained with PE-conjugated αPoCD3 mAb (BD Biosciences, San Diego, CA) for 30 min at 4°C in normal FCM, fixed for 30 min at 4°C in Fixation & Permeabilization Buffer (BD Biosciences, San Diego, CA), washed twice with 1x Permeabilization Buffer (BD Biosciences, San Diego, CA), and labeled with AF647-conjugated αPoIL-17A or αPoIFNγ mAb ( Table 1 ) for 30 min at 4°C in 1x Permeabilization Buffer. The cells were washed twice with 1x Permeabilization Buffer and re-suspended in normal FCM. For flow cytometric analyses, data on labeled cells were acquired either on an Accuri C6 or an Accuri C6 Plus flow cytometer (BD Biosciences, San Diego, CA) and analyzed using FlowJo Software, version 10.7.1 (BD Biosciences, San Jose, CA), gating on live lymphocytes and live T cells.

Single best sequence homologs for each immunogen were retrieved from NCBI protein BLAST searches (https://blast.ncbi.nlm.nih.gov/Blast.cgi), performed restricting hits to a single sequence and using all other default settings. Corresponding swine reference amino acid sequences were BLAST queries. Hits corresponding to species of interest were aligned using the NCBI COBALT tool (https://www.ncbi.nlm.nih.gov/tools/cobalt/cobalt.cgi) with default settings.

ELISA data were analyzed using Microsoft Excel program (Microsoft Software, Redmond, WA). The mean ODs of duplicates were plotted using GraphPad Prism 5 software (GraphPad Software, La Jolla, CA). Any cross-species reaction whose optical density (OD) was equal to, or higher than, 0.5 was considered positive. The lower and upper sensitivity limits of the sandwich ELISA were determined as the first data point which detects reactivity above background level, and the upper sensitivity limit was determined as the apex point on the standard curve. Flow Cytometry data were analyzed using FlowJo Software version 10.7.1 (BD Biosciences, San Jose, CA), gating on live lymphocytes and for some analyses on live CD3+ T cells.

To define individual antigenic determinants that are recognized by each of the 10 αPoIL-17A mAbs, a competition ELISA was used to measure the ability of excess unlabeled mAb to inhibit the binding of each biotin-labeled mAb to the target rPoIL-17A. Shown in Table 1 is the complete list of antibodies used in these studies. Table 2 notes percent inhibition of the binding of biotin-labeled mAbs by a 100-fold excess of the unlabeled mAbs. Several biotin-labeled mAbs (αPoIL-17A-2.4, -2.5, -2.6, and -2.8) were inhibited by self and most other αPoIL-17A mAbs, whereas a few others (αPoIL-17A-1.1, -1.2) were inhibited by no or a few other mAbs.

Most αPoIL-17A mAbs showed varying levels of cross-reactivity with orthologous rIL-17A proteins ( Figure 1 ). Only one mAb (αPoIL-17A-1.2) did not cross-react with any orthologs we tested. Based on the cross-reactivity patterns shown in Figure 1 , and the inhibition patterns shown in Table 2 , eight antigenic determinants (A-H) were proposed for the 10 αPoIL-17A mAbs: Group A: Represented by αPoIL-17A-1.1, this mAb cross-reacted with human, dolphin, and monkey rIL-17A proteins and was not inhibited by any of the other mAbs. Group B: Represented by αPoIL-17A-1.2, this mAb did not cross-react with any orthologous rIL-17A proteins we tested. Group C: Represented by αPoIL-17A-2.1, -2.3, and -2.6, these mAbs cross-reacted with human, dolphin, equine, guinea pig, bovine, and ovine rIL-17A and had similar cross-inhibition patterns. Group D: Represented by αPoIL-17A-2.4, this mAb cross-reacted with human, dolphin, monkey, equine, guinea pig, bovine and ovine rIL-17A. Group E: Represented by αPoIL-17A-2.5, this mAb cross-reacted with only equine IL-17A. Group F: Represented by αPoIL-17A-2.8, this mAb cross-reacted with dolphin, guinea pig, human, equine, and ovine rIL-17A. Group G: Represented by αPoIL-17A-2.9, this mAb cross-reacted with monkey and weakly with feline rIL-17A. Group H: Represented by αPoIL-17A-2.10, this mAb cross-reacted with dolphin, human, monkey, equine, caprine, ovine, bovine, rabbit, guinea pig, and weakly with feline rIL-17A. No cross-reactivity was found for any αPoIL-17A mAb with canine, mouse, or zebrafish rIL-17A.

We performed a BLAST search on porcine IL-17A proteins to see whether amino-acid sequences that are shared with several orthologous proteins we tested, may correlate with the cross-species reactivity patterns we observed in this study ( Supplementary Figure 1 ). Next to each sequence name is a percentage identity value with respect to the aligned region. For porcine IL-17A, the closest ortholog is bovine (83% identity), then ovine, caprine, and dolphin, all of which with 81% identity. Despite the highest sequence identity mAb reactivity with the rBoIL-17A ortholog was relatively low; higher reactivity was found with equine and human rIL-17A with 74% and 72% sequence identity, respectively, to porcine IL-17A.

As with IL-17A, porcine IFNγ antigenic determinants recognized by each of the nine αPoIFNγ mAbs were defined by competition ELISA, testing 100-fold excess unlabeled αPoIFNγ mAbs capacity to competitively inhibit binding of biotin-labeled αPoIFNγ mAbs to rPoIFNγ. As shown in Table 3 , for αPoIFNγ mAbs, half of the unlabeled αPoIFNγ mAbs inhibited the binding of self biotin-labeled mAbs (αPoIFNγ-1.1, -17.1, -21.3, -23.2). The other inhibition patterns were heterogeneous, ranging from none, weak to medium, or strong. There were also some mAbs that failed to inhibit self (αPoIFNγ-35.1 and -45.2) ( Table 3 ).

We used a direct ELISA to assess the ability of the αPoIFNγ mAbs to react with orthologous rIFNγ proteins. Fourteen orthologous rIFNγ proteins from different species were tested. Based on the binding patterns ( Figure 2 ), four determinants (A-E) representing potential antigenic determinants emerged ( Table 3 ). For group A, one mAb (αPoIFNγ-1.1) was the only mAb that unexpectedly cross-reacted with zebrafish rIFNγ protein. Most of the αPoIFNγ mAbs showed a strong cross-reactivity with orthologous rIFNγ proteins from dolphin, bovine, caprine, and ovine. Three αPoIFNγ mAbs (αPoIFNγ-23.2, -24.1, -35.1) also cross-reacted weakly with human and equine rIFNγ and were assigned as Group B. Group C: Represented by αPoIFNγ-17.1 and -21.3, these mAbs cross-reacted with canine, dolphin, bovine, ovine and caprine rIFNγ and had strong cross inhibition patterns. Group D: Represented by αPoIFNγ-27.3; -34.2 and -45.2, these mAbs cross-reacted with bovine, ovine, caprine, dolphin, and canine rIFNγ. No αPoIFNγ mAb cross-reactivity was detected with rabbit, chicken, or mouse rIFNγ, and only weak reactivity was seen with feline and monkey rIFNγ.

Shown in Supplemental Figure 2 are multiple sequence alignments against porcine IFNγ. Next to each sequence name is a percentage identity value with respect to the aligned region. For porcine IFNγ, the closest sequence homologies are with dolphin (80.7% identity), then bovine, ovine, and caprine, all of which with 77-78% identity. This sequence homology may explain why most of our new mAbs cross-reacted with bovine, ovine, caprine, equine, canine, feline, dolphin, and human IFNγ. However, it cannot explain the unique cross-reactivity of one mAb, αPoIFNγ-1.1, with zebrafish. Indeed, zebrafish IFNγ protein was the least identical to porcine IFNγ protein, with only 25% identity.

We established an IL-17A sandwich ELISA using rPoIL-17A standard curves and testing several sets of mAbs based on their different determinant reactivities. As shown in Figure 3A , three of the best performing mAb pairs were compared with αPoIL-17A-2.6 mAb as the capture mAb, and αPoIL-17A-1.1 mAb as the detection mAb (2.6/1.1 pair) proving to be the most sensitive pair at detecting rPoIL-17A with a lower sensitivity limit of 100-300 pg/ml and an upper limit of 3,000 pg/ml. The reverse mAb pair (1.1/2.6) and a different pair (1.2/1/1) were less sensitive with a lower limit of 300 pg/ml and an upper limit of 10,000 pg/ml. We did not test our newly developed anti-IFNγ mAbs for development of a sandwich ELISA because such assays are already commercially available from both R&D Systems and BD Biosciences.

Subsequent experiments were performed using the αPoIL-17A-2.6/-1.1 pair. We investigated the ELISA sensitivity for rPoIL-17A detection in PBS+BSA versus pig serum. As shown in Figure 3B , the lower sensitivity limit for the PBS+BSA curve was at 300 pg/ml, while the lower limit was around 1,000 pg/ml in the samples containing pig serum at 10% and 20%. There was an increasingly higher background with increasing concentrations of pig serum. This may make it more difficult to use this assay with samples containing >20% pig serum, because the PBS+BSA standard curve had better background levels. Thus, it is recommended to use samples containing no more than 20% pig serum to maintain the sensitivity of the assay.

Next, we needed to affirm the reactivity of the assay with native porcine IL-17A. As shown in Figure 3C , the αPoIL-17A-2.6/1.1 mAb pair was sensitive for detecting IL-17A in supernatants from cultured pig PBMC. This was evidenced by the low background levels in the cell culture medium only and the supernatant from unstimulated cells (BM) compared to the detection of native porcine IL-17A in supernatants from PHA or PMA/Iono-stimulated PBMC. More IL-17A was expressed in supernatants from PBMC stimulated with PHA or PMA/Iono for 48 hrs compared to supernatants from PBMC stimulated with PHA or PMA/Iono for 24 hrs.

We tested the ability of newly produced mAbs to specifically bind porcine IL-17A intracellularly. Several of the tested αPoIL-17A mAbs stained well and our analyses showed that PMA/ionomycin stimulation induced significant expression of IL-17A by CD3+ T cells. As shown in Figure 4 , gating on live lymphocytes, the tested αPoIL-17A mAbs showed a clear population of T cells expressing IL-17A. Repeated assays revealed αPoIL-17A-2.5, -2.6, and -2.10 to be the most reliable mAbs. The mean frequency of IL-17A expressing CD3+ T cells from PBMCs was 1.33 ± 0.36%. Using αHuIL-17-SCPL1362 mAb as a positive control, our data confirmed a previous report that it cross-reacted with pig cells, staining 0.64 ± 0.19% CD3+ T cells after PMA/Iono stimulation (16). As shown in Figure 4 three of our mAbs, αPoIL-17A-2.5, -2.6, and -2.10 stained 1.01, 1.19 and 1.03% PMA/ionomycin stimulated cells, respectively. One mAb, αPoIL-17A-2.8, only stained 0.69% of those same cells. We also compared our mAbs with αBoIL-17A-2A mAb, which we confirmed was cross-reactive with swine IL-17A (3) (data are not shown).

Four mAbs (αPoIFNγ-1.1, 17.1, 34.2, and 35.1) labeled with AF647 were compared to αPoIFNγ P2G10 (BD Bioscience, San Diego, CA) labeled with AF647 for their ability to bind to native IFNγ in PBMC stimulated for 5 hrs with BD leukocytes activation cocktail. With the gate set on live lymphocytes and CD3+ T cells, we found unusually high frequencies of IFNγ positive cells with one mAb (anti-PoIFNγ-1.1; Supplementary Figure 3C ), which were not plausible, likely due to cell shift upon staining with the AF647-conjugated mAbs. Overall, none of the tested new anti-IFNγ mAbs worked in intracellular staining ( Supplementary Figure 3 ).