Diagnostic and Epitope Mapping Potential of Single-Chain Antibody Fragments Against Foot-and-Mouth Disease Virus Serotypes A, SAT1, and SAT3

Foot-and-mouth disease (FMD) affects cloven-hoofed domestic and wildlife animals and an outbreak can cause severe losses in milk production, reduction in meat production and death amongst young animals. Several parts of Asia, most of Africa, and the Middle East remain endemic, thus emphasis on improved FMD vaccines, diagnostic assays, and control measures are key research areas. FMD virus (FMDV) populations are quasispecies, which pose serious implications in vaccine design and efficacy where an effective vaccine should include multiple independent neutralizing epitopes to elicit an adequate immune response. Further investigation of the residues that comprise the antigenic determinants of the virus will allow the identification of mutations in outbreak strains that potentially lessen the efficacy of a vaccine. Additionally, of utmost importance in endemic regions, is the accurate diagnosis of FMDV infection for the control and eradication of the disease. To this end, a phage display library was explored to identify FMDV epitopes for recombinant vaccines and for the generation of reagents for improved diagnostic FMD enzyme-linked immunosorbent assays (ELISAs). A naïve semi-synthetic chicken single chain variable fragment (scFv) phage display library i.e., the Nkuku® library was used for bio-panning against FMD Southern-African Territories (SAT) 1, SAT3, and serotype A viruses. Biopanning yielded one unique scFv against SAT1, two for SAT3, and nine for A22. SAT1 and SAT3 specific scFvs were exploited as capturing and detecting reagents to develop an improved diagnostic ELISA for FMDV. The SAT1 soluble scFv showed potential as a detecting reagent in the liquid phase blocking ELISA (LPBE) as it reacted specifically with a panel of SAT1 viruses, albeit with different ELISA absorbance signals. The SAT1svFv1 had little or no change on its paratope when coated on polystyrene plates whilst the SAT3scFv's paratope may have changed. SAT1 and SAT3 soluble scFvs did not neutralize the SAT1 and SAT3 viruses; however, three of the nine A22 binders i.e., A22scFv1, A22scFv2, and A22scFv8 were able to neutralize A22 virus. Following the generation of virus escape mutants through successive virus passage under scFv pressure, FMDV epitopes were postulated i.e., RGD+3 and +4 positions respectively, proving the epitope mapping potential of scFvs.

Foot-and-mouth disease (FMD) affects cloven-hoofed domestic and wildlife animals and an outbreak can cause severe losses in milk production, reduction in meat production and death amongst young animals. Several parts of Asia, most of Africa, and the Middle East remain endemic, thus emphasis on improved FMD vaccines, diagnostic assays, and control measures are key research areas. FMD virus (FMDV) populations are quasispecies, which pose serious implications in vaccine design and efficacy where an effective vaccine should include multiple independent neutralizing epitopes to elicit an adequate immune response. Further investigation of the residues that comprise the antigenic determinants of the virus will allow the identification of mutations in outbreak strains that potentially lessen the efficacy of a vaccine. Additionally, of utmost importance in endemic regions, is the accurate diagnosis of FMDV infection for the control and eradication of the disease. To this end, a phage display library was explored to identify FMDV epitopes for recombinant vaccines and for the generation of reagents for improved diagnostic FMD enzyme-linked immunosorbent assays (ELISAs). A naïve semi-synthetic chicken single chain variable fragment (scFv) phage display library i.e., the Nkuku ® library was used for bio-panning against FMD Southern-African Territories (SAT) 1, SAT3, and serotype A viruses. Biopanning yielded one unique scFv against SAT1, two for SAT3, and nine for A22. SAT1 and SAT3 specific scFvs were exploited as capturing and detecting reagents to develop an improved diagnostic ELISA for FMDV. The SAT1 soluble scFv showed potential as a detecting reagent in the liquid phase blocking ELISA (LPBE) as it reacted specifically with a panel of SAT1 viruses, albeit with different ELISA absorbance signals. The SAT1svFv1 had little or no change on its paratope when coated on polystyrene plates whilst the SAT3scFv's paratope may have changed. SAT1 and SAT3 soluble scFvs did not neutralize the SAT1 and SAT3 viruses; however, three INTRODUCTION Diseases caused by RNA viruses are often difficult to control because of the high mutation rate and the continual emergence of novel genetic and antigenic variants that allow escape from immunity (1). Antigenic differences between viruses play a role in whether the immunity induced by one virus is effective against another. Foot-and-mouth disease (FMD) virus (FMDV), a singlestranded, positive-sense RNA virus, and the prototype member of the Aphthovirus genus in the family Picornaviridae (2), is an example of an antigenically variable pathogen with the ability to evade the immune system (3)(4)(5). Of the seven clinically indistinguishable FMDV serotypes, viruses belonging to the three Southern African Territories (SAT) serotypes display appreciably greater genomic and antigenic variation (6).
Two key research focus areas for enhanced FMD control are improved vaccines that offer a broad immunogenic response and improved specific diagnostic assays (7). However, the high antigenic diversity that exists within the FMDV serotypes hinders FMD control by vaccination, as vaccination against one serotype does not confer protection against another and may only be partially effective against some subtypes within the same serotype (8). This poses serious implications in vaccine design and efficacy where an effective vaccine should include multiple independent epitopes to elicit an immune response (9). The humoral immune response has generally been accepted as the most important factor in conferring vaccine-induced protection against FMD, as a strong correlation has been reported between the levels of virus-neutralizing antibody produced after vaccination and subsequent protection of cattle, one of the main target species for vaccination (10)(11)(12)(13). To develop more effective vaccines or peptide vaccines, numerous FMDV studies have been undertaken to identify these neutralizing antigenic sites in more detail (14). Neutralizing antigenic sites have been identified for serotype A (15)(16)(17), O (18)(19)(20)(21), C (22), Asia-1 (23), and SAT2 (19,24,25). However, information regarding the antigenic determinants of SAT serotypes, which are confined geographically to Africa, is scarce (26). Mapped SAT2 epitopes include: (i) βG-βH loop of VP1; (ii) residue 210 in the C-terminus of VP1; (iii) VP1 84-86, 109-111, VP2 71,72,133,134; and (iv) VP1 159, [133][134][148][149][150]24,25,27,28). Four independent antigenic determinants were identified for SAT1 viruses i.e., (i) two occurring in the βG-βH loop of VP1; (ii) two simultaneous residues one in VP3 (position 135 or 71 or 76) and one in VP1 (position 179 or 181); (iii) a conformation dependant site Abbreviations: FMD, foot-and-mouth disease; FMDV, FMD virus; scFv, singlechain variable fragment; MAb, monoclonal antibody. within VP1 position 181 and VP2 72; and (iv) VP1 position 111 (24). To date, no neutralizing sites have been determined for viruses of the SAT3 serotype. It has been shown that the majority of FMDV-neutralizing antibodies are directed against conformational epitopes located on the β-barrel connecting loops, especially the highly mobile βG-βH loop in VP1 (15,18,26,29,30). Therefore, knowledge of the amino acid residues that comprise the antigenic determinants of FMDV, and those that function as protective epitopes in particular, will greatly improve our understanding of virus neutralization in vivo (12,26,31).
Diagnostic assays hampered by the lack of specificity caused by polyclonal capture and detection antibodies highlighted the need for more specific tests. Monoclonal antibodies are highly specific reagents and are being used for a variety of research and diagnostic purposes within the FMD field and their pivotal role in all aspects of FMD research is now clear. However, traditional monoclonal antibodies, produced using hybridoma technology, and used in diagnostics have several limitations such as its high cost, time-consuming production, and the expertise required (32)(33)(34).
The development of large combinatorial antibody libraries based on antibody genes expressed and displayed on phages have revolutionized the selection and isolation of unique antibodies to an antigen and aided in the development of recombinant reagents for ELISA (35). A key advantage of phage display of antibody fragments is that the generation of specific singlechain variable fragment (scFv) or antigen binding fragment (Fab) to a particular antigen can be completed within a few weeks compared to hybridomas taking months. Antibody libraries can be either immune (from immunized donors) or naïve (from non-immunized donors). An immune library will have an antibody range that is highly enriched for antibodies generated in response to a particular immunogen, whereas, the naïve library can advantageously be used for an unlimited array of immunogens (36).
Phage display libraries have been used with success to map epitopes for FMDV for serotype O (37,38) and the SAT2 serotype (25). The Nkuku R phage-display library, which is a large semisynthetic library of recombinant filamentous bacteriophages displaying scFv's derived from combinatorial pairings of chicken variable heavy and light chains, was used for this study (36). This naïve library has been utilized to generate a variety of antibodies against antigens such as the bluetongue virus, African horse sickness virus, echovirus 1, coxsackievirus B3, FMDV of the SAT2 serotype as well as a mycobacterial 16 kDa antigen (25,36,(39)(40)(41)(42). These studies prove that this library is sufficiently diverse for the recognition of a variety of different haptens, proteins, bacteria, and viruses.
In this study, phage display technology was used to obtain specific scFvs from panning with FMDV serotype A, SAT1, and SAT3 viruses. This is a novel study as recombinant monoclonal antibodies (scFvs) have not been isolated for FMDV serotype A, SAT1, and SAT3. The scFvs resulting from the biopanning were investigated in virus neutralization assays in the pursuit of epitope identification and for their prospective use as FMDV diagnostic reagents in an ELISA.
Selection of scFvs Against SAT1/KNP/196/91, SAT3/KNP/10/90, and A22 Selection of virus-specific scFvs from the Nkuku R phage display library was performed as described by van Wyngaardt et al. (36) and Opperman et al. (25). Briefly, 2-ml immunotubes (Nunc R Maxisorp), after being coated overnight with purified virus (30 µg/ml) were blocked with 1 × PBS containing 2% (w/v) milk powder (Elite) and incubated with the Nkuku R library phage particles (10 12 -10 13 transducing particles). Exponentially growing Escherichia coli TG1 cells (Stratagene, USA) were infected with eluted phage-displayed scFvs that had bound to the specific viruses before plating on TYE plates (15 g/l agar, 8 g/l NaCl, 10 g/l tryptone, 5 g/l yeast extract) supplemented with 2% (w/v) glucose and 100 µg/ml ampicillin. Subsequent to overnight incubation, the bacteria were collected and the phagemids rescued by the addition of M13KO7 helper phage. Infected bacterial cells were incubated overnight in 2 × TY medium (16 g/l tryptone, 10 g/l yeast extract, 5 g NaCl/l) containing 100 µg/ml ampicillin and 25 µg/ml kanamycin. Phages displaying scFvs were precipitated from the cell-free culture supernatant with one-fifth of the original culture volume of 20% (w/v) PEG-8000 in NaCl and were then suspended in 1 × PBS for use in the next selection round. A total of three such selection rounds were performed. The input and output phages from each selection round was titered to monitor enrichment. The outputs of each consecutive selection round was tested in a polyclonal ELISA. Single clones from the third selection round was tested

Monoclonal Phage ELISA
Following the third round of panning, monoclonal phage antibodies were screened by randomly selecting individual clones from the titration plates and inoculating it into a 96-well tissueculture plate (Nunc R ) containing 2 x TY medium supplemented with 100 µg/ml ampicillin and 2% (w/v) glucose. The bacteria were grown overnight with shaking at 30 • C. Using a 96-well inoculation device (Sigma-Aldrich: Cat. No. R-2508), bacterial cells were transferred from the overnight plate to a second plate containing 150 µl of fresh medium per well-followed by incubation for 2.5 h at 37 • C with shaking. Subsequently, 50 µl of medium that contained 2 × 10 9 of the M13K07 helper phage was added to each well and the plate incubated for 30 min at 37 • C without shaking. Thereafter, plates were centrifuged at 600 × g for 10 min, the supernatant fractions were removed and replaced with 150 µl of 2 × TY medium containing 100 µg/ml ampicillin and 25 µg/ml kanamycin prior to incubation overnight at 30 • C with shaking. Following centrifugation at 600 × g for 10 min to pellet bacterial cells, the supernatant fractions, which contained the phage-displayed scFvs were removed and mixed 1:1 with 1 × PBS containing 4% (w/v) casein and 0.2% (v/v) Tween-20 prior to undergoing ELISA testing as described for the polyclonal phage ELISA above.

Monoclonal Soluble scFv ELISA
The monoclonal soluble scFv ELISA has been described by van Wyngaardt et al. (36) and Opperman et al. (25). The method is similar to the monoclonal phage ELISA described above, except, instead of rescuing phages with the M13K07 helper phage, soluble scFvs were induced by adding 2 × TY containing 100 µg/ml ampicillin and 3 mM IPTG. The anti-c-Myc MAb 9E10, expressed from the murine hybridoma Mycl-9E10 (Sigma-Aldrich) and the polyclonal rabbit anti-mouse IgG conjugated to horseradish peroxidase (P0260; Dako) detected the secreted soluble scFvs. The ELISA colormetric reaction was performed as described above.

Large Scale Expression and Purification of Soluble scFvs
Glycerol stocks (150 µl) of selected phage clones were inoculated in 90 ml of 2 × TY medium (16 g/L tryptone, 10 g/L yeast extract, 5 g/L NaCl) containing 100 µg/ml ampicillin and 2% (w/v) glucose and incubated overnight at 30 • C with shaking. A 1:10 dilution of the overnight culture was prepared in 800 ml of fresh 2 × TY medium containing 100 µg/ml ampicillin and 2% (w/v) glucose and incubation, with shaking, continued for a further 8 h after which the bacterial cells were pelleted at 4,000 × g for 30 min. All traces of glucose-containing 2 × TY media was removed and the bacterial pellet resuspended in 1 L of 2 × TY media containing 100 µg/ml ampicillin and 1 mM IPTG and incubated overnight at 30 • C with shaking. The expressed soluble scFv was harvested by pelleting the bacterial cells at 4,000 × g for 30 min. The bacterial pellets were resuspended in 50 ml of TSA buffer (0.05 M Tris, 0.1 M NaCl, 0.02% NaN 3 , 0.02% sodium azide; pH 8.0) and treated with 0.01% (v/v) of 100 mg/ml lysozyme for 30 min at 30 • C. Freshly prepared 200 mM phenylmethylsulfonyl fluoride [0.01% (v/v)] in isopropanol was added to the bacterial suspension and the suspension mixed by inverting. The bacterial suspension was sonicated for 3 min (30 s pulses with 30 s pauses), where after, the bacterial pellet was collected by centrifugation at 15,000 × g for 30 min and the clear lysate filtered through 0.8 and 0.45 µm filters, respectively. The anti-myc-Sepharose column coupled with 143 mg of 9E10 Mab (prepared by Janine Frischmuth, the National Bioproducts Institute, Biotechnology division, Pinetown, South Africa) was washed with TSA buffer before the clear lysate (containing soluble scFvs) was loaded through the column via a peristaltic pump. The column was washed with TSA buffer until the spectrophotometer reading at absorbance A 280nm fell below 0.3. Soluble scFvs were eluted from the column with elution buffer (0.1 M glycine, 0.14 M NaCl, pH 2.2) and fractions collected. Peak fractions were pooled and the scFvs dialyzed (Sigma-Aldrich, dialysis tubing cellulose membrane, 10 mm flat width), for 48 h at 4 • C, in 2 L PBS pH 7.4.

Binding Specificity of Soluble scFvs
The specificity of the soluble scFvs was tested with an ELISA essentially performed as described for the monoclonal phage ELISA. ELISA plates were coated in duplicate with 30 µg/ml of purified SAT1/KNP/196/91, SAT2/ZIM/7/83, SAT3/KNP/10/90, A22, or A24 viruses as well as with BHK-21 cell extract, 2% (w/v) sucrose and 1 × PBS containing 2% (w/v) milk powder as negative controls. Virus dilutions containing ∼500, 50, and 5 infectious particles were prepared in the appropriate cell medium (RPMI for IB-RS-2 cells and GMEM for BHK-21 cells) and were applied in triplicate wells across a microtiter plate and diluted two-fold down the plate. Virus was incubated for 1 h at 37 • C in an atmosphere of 5% CO 2 after purified scFvs at a concentration between 0.03 and 0.23 mg/ml were added neat to appropriate wells. A control plate without soluble scFvs was included. BHK-21 and IB-RS-2 cells supplemented with 1% (v/v) FCS and antibiotics (virus growth medium, VGM), with a cell count of 0.3 × 10 6 cells/ml for both cell lines, were subsequently added to the respective microtiter plates. Incubation of microtiter plates then occurred for 72 h at 37 • C and fixation and staining with a methylene blue-formaldehyde stain to allow for inspection of the cytopathic effect, which was scored as a measure of neutralization.

Neutralization Assays and Generation of Virus Escape Mutants
To generate virus neutralization escape mutants, the viruses (SAT1/KNP/196/91, SAT3/KNP/10/90, and A22) were passaged under scFv pressure as described by Crowther et al. (19) and Opperman et al. (25). Equal volumes of ca. 25 infectious virus particles were diluted two-fold in GMEM or RPMI medium on a microtiter plate before being mixed with an equal volume of the respective purified scFv (neat), followed by 30 min incubation at 37 • C. The virus-scFv complexes were added to either BHK-21 or IB-RS-2 monolayer cells and incubated for 1 h at 37 • C. Following incubation, the virus-scFv complexes were removed. Monolayers were washed twice with GMEM or RPMI (Sigma-Aldrich) medium before VGM containing a 1:50 dilution of purified scFvs at a concentration between 0.6 and 4.6 µg/ml were added. All scFvs were tested. Each virus was subjected to four consecutive passages under scFv pressure.

Characterization of Virus Escape Mutants
Virus escape mutants were then characterized by sequencing the Leader-P1-2A coding region (ca. 3 kb) of the virus genome. RNA extraction was performed using the QIAmp viral RNA kit (Qiagen). cDNA was synthesized with SuperScript III first strand synthesis kit (Invitrogen), using the genome specific primer WDA 5 ′ -GAAGGGCCCAGGGTTGGACTC-3 ′ .
The Leader-P1-2A coding region of the escape viruses were amplified using the Expand Long Template PCR System (Roche) and forward primer NCR1 5 ′ -TACCAAGCGACACTCGGGATCT-3 ′ and reverse primer WDA 5 ′ -GAAGGGCCCAGGGTTGGACTC-3 ′ . Briefly, each 50 µl PCR reaction mixture consisted of 3 µl of the first strand cDNA reaction mixture, 0.3 µM of each oligonucleotide, 2.5 U of Expand Long Template DNA polymerase, 1 × Expand buffer, 0.75 mM MgCl 2 and 2 µM of each dNTP. Using a thermocycler (GeneAmp 9700, Applied Biosystems) after initial denaturation at 94 • C of 2 min, the reactions were subjected to 35 cycles of 94 • C for 20 s, 55 • C for 20 s and 68 • C for 4 min with a final cycle of 68 • C for 7 min to complete the synthesis of all strands.
To determine the nucleotide sequence of the gel-purified amplicons, 0.16 µM of the appropriate oligonucleotide ( Table 2) and the ABI PRISM Big Dye Terminator Cycling Ready Reaction kit v3.0 (Applied Biosystems) was utilized. The extension products were resolved on an ABI PRISM TM 3100 automated sequencer (Applied Biosystems) and sequences analyzed using the BioEdit v.7.0.9 (45) and Sequencher v5.4.6 (Gene Codes Corporation, Ann Arbor, MI, USA) software of the ca. 2.2-kb P1-coding region.

Investigation of the SAT1 and SAT3 Soluble scFvs as Capturing Antibodies in an Indirect ELISA
The SAT1 and SAT3 soluble scFvs were tested in an indirect ELISA against a panel of viruses ( Table 1) to determine whether it can be used as capturing antibodies in routine testing of suspected FMDV cases. The PEG concentrated viruses were titrated in a liquid phase blocking ELISA [LPBE; (46)] to determine the optimal dilution where an absorbance value at 450 nm (A 450nm ) of ca. one was obtained. The virus dilution of 1:8 was chosen for the scFv ELISA as this was the highest dilution where an A 450nm ∼1 (with standard deviation of 0.25, observed after the ELISA colorimetric reaction) was obtained for the viruses tested. The purified, neat scFvs (SAT1scFv1 0.03 mg/ml, SAT3scFv1 0.039 mg/ml, and SAT3scFv2 0.09 mg/ml) were used to coat 96-well Maxisorp immunoplates (Nunc R ) overnight at 4 • C, following which, ELISA plates were washed The large semi-synthetic naïve Nkuku R phage display library based on chicken immunoglobulin genes, was panned by exposing the recombinant antibody repertoire to SDG purified virions of the FMD SAT1/KNP/196/91, SAT3/KNP/10/90, and A22 viruses. After three consecutive biopanning rounds, polyclonal phage displayed scFvs were tested in a polyclonal ELISA to evaluate enrichment (Figure 1) for each of the three biopannings to the specific viruses. The phage outputs from the three consecutive selection rounds were tested and an aliquot of the library prior to panning was included as a non-enriched control (Figure 1) (Figures 1A-C).
Phage displayed scFvs specific for FMDV were identified after a helper phage rescue of single bacterial colonies from the third round of panning and tested in ELISA. Both the phage displayed and soluble scFv formats were tested and the results are summarized in Table 3. A total of 94 clones each for SAT1/KNP/196/91 and SAT3/KNP/10/90 and 188 clones for A22 were screened. Twenty clones expressed phage-displayed scFvs specific to SAT1/KNP/196/91 with ELISA signals more than two-fold greater than that of the negative control. Furthermore, of these, seven clones secreted soluble scFvs that bound to SAT1/KNP/196/91. Sequencing of the seven clones revealed one unique binder for SAT1/KNP/196/91, designated SAT1scFv1 (Table 4), which had seven identical clones. Analysis of the 94 clones for SAT3/KNP/10/90 revealed three clones expressing phage-displayed scFvs specific to SAT3/KNP/10/90, all of which secreted soluble scFvs that bound to SAT3/KNP/10/90. Sequencing of these clones indicated two unique binders designated SAT3scFv1 and SAT3scFv2 ( Table 4). In addition, of the 188 clones for A22, 25 clones expressed phage-displayed scFvs specific to A22, whilst 9 clones secreted soluble scFvs that bound to A22 and sequencing revealed nine unique binders for A22 designated A22scFv1 to A22scFv9 (Table 4). Interestingly, sequencing results revealed that SAT3scFv2 and A22scFv6 had an identical sequence in all three of the CDR's for the heavy and light chains and are essentially the same binder (Table 4). However, it must be noted that these biopannings were executed independently of each other and at different times and thus the possibility of cross-contamination is ruled out. The result inferred that the SAT3scFv2 and A22scFv6 binders recognize a conserved amino acid (aa) motif on Phage-displayed scFvs that bound to SAT1/KNP/196/91, SAT3/KNP/10/90, and A22 were eluted and enrichment of virus specific phage-displayed scFvs (black bars) was determined by a polyclonal phage ELISA of the outputs for three consecutive biopannings. The unpanned aliquot of the Nkuku ® phage display library was a non-enriched control (selection round 0). The negative control used was 2% milk powder (labeled 2% MP, gray bars). The data are averages ±SD of three repeats.
both SAT3/KNP/10/90 and A22 viruses. The soluble scFvs were subsequently successfully purified by means of affinity chromatography and further characterized.

Binding Specificity of Soluble scFvs to FMDV
The specificity of the virus-specific soluble scFvs were determined by measuring their ability to bind to purified, complete 146S virions of viruses from serotypes A, SAT1, SAT2, and SAT3 in an ELISA. Negative controls were BHK-21 cell extract, 2% sucrose, and 2% milk powder (Figure 2). SAT1scFv1 phage-displayed binder cross-reacted with A22, SAT2/ZIM/7/83, and SAT3/KNP/10/90, however, the soluble scFv format did not exhibit any cross-reactivity (Figure 2). SAT3scFv1 phagedisplayed and soluble scFv cross-reacted with A22 and SAT1/KNP/196/91 whilst there was borderline cross-reactivity with SAT2/ZIM/7/83 observed for the phage-displayed scFv (Figure 2). The amino acid sequence of the heavy and light chain complementary determining regions (CDRs) of SAT3scFv2 and A22scFv6 are identical and the cross-reactivity between SAT3scFv2 and A22 as well as A22scFv6 and SAT3/KNP/10/90 was observed with both the phage-displayed and soluble scFv (Figure 2). SAT3scFv2 phage-displayed scFv also cross-reacted with SAT1/KNP/196/91, whereas the soluble scFv did not crossreact (Figure 2). A22scFv5 soluble scFv showed cross-reactivity to A24 virus and for the phage-displayed scFv, borderline cross-reactivity was observed (Figure 2). There was no crossreactivity observed with the A22scFv1, A22scFv2, A22scFv3, A22scFv4, A22scFv7, A22scFv8, and A22scFv9 binders with SAT1/KNP/196/91, SAT2/ZIM/7/83, SAT3/KNP/10/90, or A24 viruses. No cross-reactivity was observed with the reagents used for virus propagation and purification. The specificity investigations of the scFvs showed that the phage-displayed scFvs exhibited more prominent cross-reactivity when compared to the soluble scFvs. Due to this, further investigations for this study was continued with the soluble scFvs.  Table 5. FMDV A22 was serially passaged in the presence of soluble A22scFv1, A22scFv2, and A22scFv8 to select viruses from the A22 quasispecies population that escape neutralization by the soluble scFvs. Thus, the A22 viruses that escaped neutralization by soluble scFvs A22scFv1, A22scFv2, and A22scFv8 were designated scFv resistant virus (SRV) 1, 2, and 3, respectively. Following four consecutive passages under scFv pressure, the P1 nucleotide, and aa sequences were determined for SRV1,   SRV2, and SRV3. Comparative analysis of the aa sequences of the SRVs compared to the parental A22 sequence indicated that SRV1 exhibited one aa substitution i.e., from a non-polar proline (Pro) to a polar serine (Ser) change at VP1 aa position 149 i.e., RGD+3 (Pro149 Ser) (Figure 3). Interestingly SRV2 had no aa changes occurring in the P1 region, but SRV3 exhibited a single aa substitution of a leucine (Leu) to a phenylalanine (Phe) at position 150 of VP1 i.e., RGD+4 (Leu150 Phe) (Figure 3). The aa substitutions for SRV1 and SRV3 occurred in the surface exposed and structurally flexible VP1 βG-βH loop, downstream of the RGD sequence. The A22 virus was neutralized by three soluble scFvs and SRVs for A22, which showed a potential binding site for two of the scFvs in the GH-loop of the VP1 protein.

Neutralization and Escape Mutant Investigations of the Identified scFvs
The three CDRs of the H-chain of A22scFv1 and A22scFv8 are identical while the L-chain of A22scFv1 and A22scFv8 displayed different sequences ( Table 4). Soluble scFvs for SAT1/KNP/196/91 and SAT3/KNP/10/90 viruses did not neutralize the respective viruses. Thus, the SAT1 and SAT3 scFvs were investigated for its potential use as diagnostic reagents in an ELISA.

SAT Virus-Specific scFvs as a FMDV Capturing Antibody in a Sandwich ELISA
To determine whether the SAT1 and SAT3 soluble scFv's retain the correct conformation to act as capturing reagents in a diagnostic sandwich ELISA, the soluble scFvs were coated directly onto maxisorp immunoplates. The antigen-binding activity of the immobilized soluble SATscFv1 was tested against a panel of PEG concentrated SAT1 (n = 11) viruses whilst soluble SAT3scFv1 and SAT3scFv2 was tested against a panel of SAT3 (n = 12) viruses ( Table 1).
Overall, the ELISA results using the scFvs as capturing reagents indicate that the soluble SAT1scFv1 was able to successfully bind to the polystyrene ELISA plate and react to viruses within the SAT1 serotype. The SAT3scFv1 and SAT3scFv2 FIGURE 4 | A sandwich ELISA with soluble scFvs as capturing antibodies. The SAT1scFv1 was tested with a panel of PEG concentrated SAT1 viruses (A) and SAT3scFv1 and SAT3scFv2 was tested with the SAT3 viruses (B). For the negative control (neg), 2% milk powder was included in the assays replacing the soluble scFvs coating the plate. The data are means ± SD of two independent experiments. An ELISA signal more than two-fold that of the negative control A 450nm was considered a positive result and the cut-off is indicated by a red line. exhibited no or borderline reactivity with the SAT3 viruses tested. The low signals may be attributed to the conformational changes of the scFvs when binding to the ELISA plate. Furthermore, aa differences in the viral proteins may result in the variable ELISA signals observed.

SAT Virus-Specific scFvs as a FMDV Detecting Antibody in an ELISA
The soluble scFvs, SAT1scFv1, SAT3scFv1, and SAT3scFv2, were also applied as detecting antibodies in a sandwich ELISA. FMD virus was captured by polyclonal rabbit antiserum and the soluble scFvs was used to detect the 146S virus particles using the panel in Table 1. The standard diagnostic sandwich ELISA used for antigen detection (46) was performed concurrently as a comparison of the scFv ELISA performance.
Results showed that the diagnostic antigen detection ELISA was able to detect all viruses tested and produced positive ELISA signals (Figures 5A,B). The ELISA assay using SAT1scFv1 as a detecting antibody revealed two characteristic reactivity profiles against the panel of SAT1 viruses, i.e., (i) A 450nm >1.4 was observed for SAT1/SAR/9/03, SAT1/KNP/196/91, and SAT1/NAM/272/98 viruses and (ii) A 450nm absorbance values of ≥0.4, ≤0.82 for the remaining eight SAT1 viruses in the panel (Figure 5A).
The soluble SAT3scFv1 and SAT3scFv2 showed low absorbance signals for the SAT3 viruses tested (Figure 5B). FIGURE 5 | Sandwich ELISA using soluble SAT1scFv1, SAT3scFv1, and SAT3scFv2 as detecting reagents. The standard diagnostic sandwich ELISA used for FMD antigen detection (46) [used as a positive control (SAT1 diag or SAT3 diag)] was adapted, where the detection antibody was replaced with a soluble SAT scFv. The ELISA was executed to determine the detecting potential of soluble SAT1scFv1 to PEG concentrated SAT1 viruses (A) and for SAT3scFv1 and SAT3scFv2 to PEG concentrated SAT3 serotype viruses (B). For the negative control, 2% milk powder was included in the assays replacing the virus component. The negative control ELISA background signal was deducted when plotting the ELISA A 450nm result. The data are means ± SD of two independent experiments.
Absorbance signals (A 450nm ) of 0.76 and 0.84 were observed for SAT3/KNP/10/90 and SAT3/KNP/6/08, respectively when SAT3scFv1 was the detecting antibody in the ELISA. All other A 450nm values were below 0.26 for the SAT3 viruses tested with SAT3scFv1 and SAT3scFv2 ( Figure 5B).
Taken together, the results indicate that the two SAT3 specific scFvs showed poor potential as a detecting ELISA antibody whilst conversely, SAT1scFv1 showed good potential for incorporation as a detecting antibody in a diagnostic ELISA.

DISCUSSION
Serotype-specific serological tests for FMD detect antibodies against the structural proteins that are elicited by either vaccination or infection (48). Both the solid-phase competition (SPCE) and LPBE ELISAs for SAT1, SAT2, and SAT3 are OIE recognized and well established assays. Considering the high genetic diversity of the SAT-type viruses (2,6,26,49), there is a continuous need for improvement of these assays as crossreactivity has been noted with the SAT types using this assay. Additionally, for FMD vaccine matching where the antigenic variability of field virus strains is measured against current vaccine strains, the virus neutralization assay is utilized. This assay, however, is laborious and can cause a delay in decision making regarding FMD control measures. To address this shortcoming, utilizing MAbs or virus-specific scFv's for FMDV that recognize virus exposed antigenic epitopes in an ELISA format where results can be obtained timeously and accurately, was explored. These scFv's or small recombinant MAbs can be produced in large quantities (50) and can also be beneficial in predicting epitopes, which can in turn be used in the design of improved FMD chimeric vaccines containing various antigenic sites that can elicit a wide immunological response or protection in vaccinated animals. To this end, phage display technology was explored.
The Nkuku R phage-display library has previously been panned on FMDV to obtain unique scFv binders for serotype SAT2 (25). Opperman et al. (25) has shown that one SAT2specific soluble scFv neutralized SAT2/ZIM/7/83. This scFv interacts with a novel epitope at residue position 159 of VP1 and was applied in a scFv-based ELISA assay. We broadened this study by including FMDV serotype SAT1, SAT3, and A. Using naïve phage display libraries allows for the selection of recombinant antibodies where unique paratopes bind to exposed and complementary parts of the immobilized antigen. Thus, the possibility of obtaining antigen-specific binders would depend on the presence and accessibility of suitable surface-exposed structures of the antigen, which in this case, was the FMD virion. Another factor to consider when using phage display technology, is that the quality and the size of the naïve library plays an important role in the success of phage display (51), as paratopes that are not present within the library cannot be isolated.
In this study, the biopanning process with SAT1/KNP/196/91, SAT3/KNP/10/90, and A22 viruses resulted in unique FMDVspecific scFv binders i.e., one for SAT1, two for SAT3, and nine for serotype A. The nine A22 binders attained in this study is unique. Although none of the SAT scFv binders showed neutralization capability, three of the nine A22 binders exhibited neutralization. Of the three A22 neutralizing binders, two binders i.e., A22scFv1 and A22scFv8 had the same heavy chain sequences and only differed in the light chain region sequences. This characteristic is of interest as Hamers-Casterman et al. (52) showed that the V H play a more important role in the binding of antibody fragments to antigens than V L. In addition, investigations by Williamson and Matthews (53) showed that three neutralizing scFvs against pertussis toxin all had the same heavy chain sequences and were related. Thus, to obtain more neutralizing scFvs, one could modify only the heavy chain of non-neutralizing scFvs of differing V Ls to be the same as their neutralizing counterparts.
For the generation of escape mutants for FMDV serotype A using the neutralizing scFvs, no aa changes were observed in the P1 region for SRV2, which may be because the mutation occurred in the minor population and due to Sanger sequencing, it was not detected. Additionally, the aa substitutions for SRV1 and SRV3 occurred in the surface exposed and structurally flexible VP1 βG-βH loop, downstream of the RGD sequence. The position is identical to the identified FMDV serotype A antigenic site I and furthermore, the βG-βH loop residues 140-160 have been shown to play an important role in antigenicity in most FMDV serotypes (15-17, 19, 54-57). Additionally, changes at the conserved aa leucine residue 150 has also been shown i.e., L P or L R on escape mutants pressured by using soluble integrins (58,59). SRV1 and SRV3 alone did not solve the binding footprint of the A22scFv1 and A22scFv8 binders, respectively. It is reasonable to expect different binding footprints on the virion for the two scFvs even though they have a common binding site at VP1 aa position 149/150. For SRV1, there was a Pro to Ser aa change at VP1 position 149 (RGD +3 position) and for SRV3 the aa change occurred at VP1 position 150, RGD+4 i.e., from a Leu to a Phe. The residues succeeding the RGD motif are important for receptor recognition (58,60) and the RGD is flanked on both sides by hypervariable sequences, which delivers a domain that is capable of adopting different conformations. Both residues are highly conserved in serotype A viruses and the substitution of amino acid residues at position 149 or 150 of MAR-viruses have been described for A 10 , A 12, and A 24 (15,16,57). Opperman et al. (25) showed with scFv neutralization investigations of FMDV SAT2/ZIM/7/83, an aa change at the base of the GH loop i.e., VP1 position 159 where there was an Arg to His change. Furthermore, a synthetic peptide ELISA confirmed VP1 aa 159 as an important residue in the epitope to which the SAT2 scFv binds (25). These investigations lead us to postulate that the epitope site for the A22scFv1 and A22scFv8 binding involves the VP1 aa position 149 and 150, respectively which is part of the antigenic sequence GDLGSLA for serotype A viruses (15). However, future investigations with SRV1 and SRV3, will be to derive a synthetic peptide from the predicted epitopic site and to confirm results with a synthetic peptide blocking ELISA.
A unique finding from this study is the result of two soluble, non-neutralizing scFvs, each from different FMDV serotypes but having the same heavy and light chain sequences i.e., SAT3scFv2 and A22scFv6. It is postulated that a common epitope between SAT3/KNP/10/90 and A22 resulted in the same soluble scFv from the antibody repertoire of the Nkuku R library. Neither of the two scFv's were able to neutralize the respective viruses and thus it was impossible to identify a binding site for the two scFv's on the viral capsid. Competing monoclonal antibody studies or capsid protein peptide libraries will be used to confirm the common or cross-reacting epitope between A22 and SAT3.
From the specificity analysis, the SAT1 and SAT3 soluble scFvs from this study bound to complete 146S virions of the virus used for biopanning i.e., SAT1/KNP/196/91 and SAT3/KNP/10/90, respectively. However, the phage displayed scFv formats did show cross-reactivity across the FMD serotype viruses tested. The SAT1 phage displayed scFv was found to bind to complete 146S virions of SAT2/ZIM/7/83 and SAT3/KNP/10/90 and similarly, the SAT3 phage-displayed scFvs were found to bind SAT1/KNP/196/91 and SAT2/ZIM/7/83 146S virions. Crossreactivity of scFvs is not uncommon as in a study by Toth et al. (61), which obtained scFv clones against the potato leafroll virus where 7 clones did not cross-react with other luteoviruses whilst 4 clones did. Additionally, Toth et al. (61) proved that the cross-reacting scFvs are directed against continuous epitopes that are present on the coat proteins of certain related luteoviruses whereas the scFvs that did not show cross-reactivity, bound to discontinuous or conformation-dependant epitopes that are specific to potato leafroll virus. A major continuous FMDV epitope is located in the GH loop spanning VP1 residues around positions 140-160 (62,63). Thus, the cross-reacting phagedisplayed SAT1 and SAT3 scFvs from this study may possibly be recognizing continuous epitopes of the SAT serotype viruses and should be investigated further in this regard.
The SAT soluble scFvs, which had reduced cross-reactivity compared to the phage displayed scFvs, were further investigated for their possible use as diagnostic reagents in an ELISA format as a FMDV capturing and a detecting reagent. For both ELISA formats, the one SAT1 soluble scFv was able to produce high reactivity to the various SAT1 PEG concentrated viruses tested implying that there was little or no effect on its paratope when coated on polystyrene plates and was thus a good capturing reagent. However, the two SAT3 scFvs produced A 450nm signals just above the positive cut-off absorbance as a FMDV capturing reagent and a detecting reagent, which is not acceptable as a positive result for a diagnostic ELISA. It is vital for a capturing antibody in an ELISA assay to be efficiently immobilized onto the ELISA plate such that it is able to retain both the antibody conformation and the antigenbinding activity (64). Furthermore, important factors such as surface charge, hydrophobicity, co-adsorption of or exchange with surfactants, and other proteins play a role in determining stability and specificity of absorbed antibodies in ELISA assays (65,66). These factors may have played a role to reduce the performance of the scFv ELISAs in this study. For conventional ELISAs where complete monoclonal or polyclonal antibodies are used, immobilization onto the ELISA plates occur via physical adsorption (66)(67)(68)(69). Conversely, immobilization of small antibody fragments such as scFvs onto plastic surfaces causes unfavorable conformational changes to occur (66). Essentially, a hydrophobic interaction occurs in the linked V H and V L regions that forms a paratope resulting in a conformational change of the antigen-binding domain, which in turn results in decreased antigen-binding activity (66,70,71).
The analytical specificity of the SAT scFvs for the capturing and detecting SAT1 and SAT3 ELISAs showed that the soluble scFvs were specific for the respective serotype viruses tested and no cross-reactivity was observed. Additionally, the analytical sensitivity of both the capturing and detecting ELISAs for SAT1 and SAT3 scFvs against the SAT1 and SAT3 viruses, respectively, was found to detect all PEG concentrated viruses at a 1:8 dilution tested albeit the low A 450nm signals for the SAT3 scFv ELISAs. The SAT1scFv1 shows promise as a good detecting and immunocapture reagent due to the high reactivity for the SAT1 viruses tested. Studies have shown that when scFv fragments are utilized as a soluble protein and are not within the phage display system, low expression levels, or a low inherent affinity can occur (36). Also, the monomeric scFv fragments can have moderate binding affinities when binding to a large multivalent antigen like FMDV, which is in contrast to greater binding affinities that can be achieved by the multivalent display on the phage (72)(73)(74). To overcome the low SAT3 scFv ELISA signals, random mutations can be introduced in the gene coding for the scFv and the length of the linker within the scFv, increasing the bacterial expression of the scFvs and thus increasing the ELISA signal (41). Another approach is to stabilize the scFv to other proteins whilst retaining functionality. For example, fusion or linking to other proteins such as constant light chain domain (75), leucine zipper dimerization domain (76), Fc fragment (CH2 and CH3 domains) of mouse IgG1 (77), and alkaline phosphatase (78). Such approaches may be used in future studies to enhance the diagnostic potential of the SAT1 and SAT3 scFvs from this study.
This study has been beneficial to gain unique recombinant antibodies against FMDV SAT1, SAT3, and A serotype viruses. Although the SAT scFvs did not neutralize FMDV, the potential was shown as ELISA reagents, especially for SAT1scFv1. Further investigation and validations of SAT1scFv1 will be continued to improve ELISA absorbance signal.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/supplementary material.

AUTHOR CONTRIBUTIONS
MC: majority of the experimental work, troubleshooting, result and data analysis, and manuscript composition and editing. PO: scFv expertise and training, experimental work, troubleshooting, result and data analysis, manuscript editing, and research grant acquisition. LR: serotype A experimental work, troubleshooting, result and data analysis, and manuscript editing. JFe and WW: scFv expertise advisor, Nkuku library, and manuscript editing. JFr: scFv column development, scFv purification training and expertise/advisor, and manuscript editing. ER: research grant acquisition, PI on grant at PIADC, project advisor, and manuscript editing. FM: research grant acquisition, PI on grant at ARC, project advisor, scFv expertise, and manuscript editing. All authors contributed to the article and approved the submitted version.