Development of Foot-and-Mouth Disease Virus-Neutralizing Monoclonal Antibodies Derived From Plasmablasts of Infected Cattle and Their Germline Gene Usage

Cattle are susceptible to foot-and-mouth disease virus (FMDV), and neutralizing antibodies are critical for protection against FMDV infection in this species. However, more information is needed on the host specific antigenic structure recognized by the FMDV-specific monoclonal antibodies (mAbs) and on the functional properties of the mAb that are produced in the natural host, cattle. Herein, we characterized 55 plasmablast-derived mAbs from three FMDV-infected cattle and obtained 28 FMDV-neutralizing antibodies by the single B cell antibody technique. The neutralizing mAbs (27/28) mainly recognized conformational epitopes that differ from the well-characterized immunodominant antigenic site 1 of FMDV as defined by murine mAbs. Of these FMDV-neutralizing mAbs, 13 mAbs showed intra-type broadly neutralizing activity against the three topotypes of FMDV serotype O (ME-SA, SEA, and Cathay topotypes). Moreover, all these intra-type broadly neutralizing antibodies competed with sera from FMDV infected or vaccinated cattle, which indicates their binding to native dominant epitopes, as revealed by a blocking ELISA. We further analyzed the germline V(D)J gene usage of the 55 FMDV-specific mAbs and found cattle IgG antibodies containing ultralong HCDR3 were exclusively restricted to usage of the germline gene segment VH 1-7*02. In addition, the restricted germline gene segments of VH 1-7*02 and VL1-47*01 or 1-52*01 pairing were observed in all IgG antibodies with ultralong HCDR3. Furthermore, antibodies with longer HCDR3 were more inclined to display FMDV-neutralizing activity. This study presents a novel method for screening FMDV-specific cattle mAbs which then provide the most useful tools for studying FMDV antigenic structure and variation.


INTRODUCTION
Foot-and-mouth disease (FMD) is a highly contagious and economically important viral disease of cloven-hoofed livestock, including cattle, sheep and pigs. The causative agent is footand-mouth disease virus (FMDV), an Aphthovirus in the Picornaviridae, and appears as seven serotypes (i.e., O, A, C, Asia1, SAT1, SAT2, and SAT3) and several topotypes, with uneven geographic distributions. FMDV serotype O has been a major threat to animal husbandry in recent years in China. Four lineages in the three topotypes of FMDV type O, namely, Cathay, Middle East-South Asia (ME-SA), and South-East Asia (SEA) topotypes, are introduced and currently circulating in China, which makes the situation rather complicated. Antigenic variation among these topotypes has been investigated in recent years (1,2). However, detailed differences in antigenic structure of these topotypes are still not delineated. Monoclonal antibodies (mAbs) recognizing neutralizing epitopes could provide important keys to the basis of this antigenic variation.
There is good evidence that humoral responses play a major role in protection against FMDV infection in natural hosts (3,4). As a natural host of FMDV, cattle have a distinct composition of immunoglobulin (Ig) repertoire compared to other vertebrates which display restricted lengths of the third heavy chain complement determining regions (HCDR3s) with an average of 12-16 amino acids (aa) in length (5). However, cattle produce antibodies containing HCDR3s with an average length of 26 aa, including an ultralong subset that can exceed 60 aa (6,7). The proportions of kappa (κ) and lambda (λ) light chains in cattle Ig are 5 and 95%, respectively, whereas those of rodents are 95 and 5%, respectively (8). These unique characters of Ig sequences make cattle a promising host for producing high avidity and broadly neutralizing antibodies (bnAbs), exemplified by the rapid elicitation of bnAbs to HIV by immunization of cattle; these bnAbs contained ultralong HCDR3s that were responsible for their serological breadth and potency (9). However, it is currently unknown whether the ultralong HCDR3s are responsible for their high avidity and broadly virus neutralization against FMDV.
Up to now, monoclonal antibodies (mAbs) selected from mouse hybridomas have been extensively used to investigate the antigenic profile of FMDV. As revealed by these mouse mAbs, five functionally-independent neutralizing antigen sites (3)(4)(5)(6)(7) have been identified on the capsid surface of FMDV serotype O. Site 1 is linear and trypsin sensitive, which encompasses the G-H loop and the C terminus of VP1, with critical residues at positions 144, 148 and 150, and 208 that affect antibody binding. However, other identified sites (i.e., sites 2-5) are all conformational and trypsin resistant. Site 2 is defined by mutations in the VP2 B-C or E-F loops, involving critical aa residues at positions 70-73, 75, 77, and 131. Critical residues at positions 43 and 44 in the VP1 B-C loop, and at position 58 in the VP3 B-B "knob" contribute to site 3 and site 4, respectively. Site 5 contains at least a functionally independent neutralizing epitope that involves a specific mutation at position 149 in the G-H loop of VP1, which is distinct from site 1 despite part of the G-H loop is encompassed (10)(11)(12)(13)(14). More recently, a new neutralizing epitope that involves the position 192 of VP2 at the 3-fold axis was reported (15). FMDV serotype O specific cattle mAbs selected from a mouse × cattle hetero-hybridoma were used to compare antigenic features defined by mouse mAbs, and these cattle mAbs recognize identical antigen sites 2, 3, and 4 but not antigen sites 1 and 5 (16). However, due to the inherent character of cattle-derived mAbs, the virus-neutralizing features of ultralong HCDR3-associated mAb need to be further surveyed using large numbers of mAbs derived from the natural host of cattle.
MAbs are produced from a single B-lymphocyte clone and can bind to a specific epitope. The first mAb was developed in the mouse by a hybridoma technique in 1975 (17). With the improvements in genetic sequencing and biomedical instrumentation, new methods for generating mAbs have been reported in recent years, including B cell immortalization (18, 19), phage display (19), and single B cell antibody technique (20). All these methods have different merits and were chose to develop mAbs based on different purposes and conditions. This study aimed to produce diverse mAbs in natural host by single B cell antibody technique, in order to explore the antigenic variation of FMDV and to shed light on FMDV hostspecific antigen site. The results show that diverse antibodies can be obtained by this single B cell antibody technique (procedures are showed in Figure 1), so providing abundant tools for exploration the antigenic variation of FMDV type O. Furthermore, rearrangements in antibody genes of FMDVspecific mAbs were analyzed, and restricted germline gene segment usage was observed in the ultralong HCDR3 containing antibodies. These data enrich our knowledge on the diversity of antibodies and their mechanism of production.

Immune Response of Cattle Sequentially Infected With FMDV Serotype O
As revealed by the results of LPB-ELISA (Figure 2A), the humoral immune responses in the three cattle were detected on the fourth day and peaked on day 14 after initial infection with the FMDV O/Mya/98 (SEA topotype) strain. After a boost immunization with FMDV O/HN/CHA/93 (Cathay topotype), IgG titers increased rapidly and reached their highest levels on day 56. However, the third immunization with FMDV O/Tibet/99 (ME-SA topotype) failed to further promote increase in IgG level, and these high titers persisted for at least 4 months. The antibody titers against the four strains of the three topotypes of FMDV serotype O were detected by virusneutralizing test (VNT). A broad antibody spectrum against the three topotypes of FMDV serotype O were observed in the three cattle (Table S1), which was initiated from the first infection to the third immunization. To isolate the FMDV-specific plasmblasts, peripheral blood mononuclear cells (PBMCs) from cattle were generally collected after the third, depending on the antibody level. The cattle number and time points for isolation of PBMC are listed in Table S2, corresponding to each mAbs. FIGURE 1 | Workflow of cattle mAb production using singe B cell antibody techniques. Cattle were sequentially immunized with different FMDV serotype O strains, and then peripheral blood was collected to isolate PBMCs. The isolated PBMCs were stained with a panel of mouse anti-cattle CD21, anti-cattle IgM, and biotinylated antigen, followed by incubation with APC labeled anti-biotin secondary antibody. The single antigen-specific plasmablast was sorted into 96 well-plate with lysis buffer and immediately reverse-transcribed into cDNA. The paired VH and VL sequences was respectively amplified by nested PCR and then sequenced. The variable regions were codon optimized and cloned into pcDNA3.4 expression cassette containing cattle IgG constant region coding sequences. The recombinant mAbs were expressed in CHO-S and purified by affinity and further molecular sieve chromatography.

Distribution and Proportion of FMDV-Specific Plasmablasts in Infected Cattle
According to the FACS results shown in Figure 3, FMDVspecific plasmablasts were a minor population in the peripheral blood at a frequency of <0.1% PBMCs. At least 1 × 10 6 PBMCs were analyzed to check the distribution and proportion of FMDV-specific plasmablasts in infected cattle. The majority of FMDV-specific plasmablasts are distributed in the IgM + B cell population, including the majority population of IgM + CD21 + FMDV + B cells (0.0796% ± 0.0207%) and the minority population of IgM + CD21 − FMDV + B cells (0.0047% ± 0.0009%; n = 6, Figure 3B). Within the IgM − B cells, which might consist of either IgG + or IgD + B cells, the IgM − CD21 + FMDV + B and IgM − CD21 − FMDV + B cells accounted for about 0.0216 ± 0.0065% and 0.0081 ± 0.0010%, respectively (n = 6, Figure 3C). Owing to the phenotypic character of FMDV-specific plasmablasts, single IgM − CD21 +/− FMDV + B cells were sorted for the amplification of the IgG BCR variable gene and development of FMDVneutralizing mAbs.

Production of Plasmablast-Derived FMDV-Specific mAbs in Cattle
The pairs of γ heavy chain ( Figure S1A) and λ light chain ( Figure S1B) variable regions (designated as VH and VL, respectively) of cattle IgG were successfully amplified from sorted single plasmablasts by nested PCR and subsequently confirmed by DNA sequencing. The complete cattle IgG mAbs were successfully expressed in CHO-S cells and then purified, as shown by reduced ( Figure S1C) and non-reduced SDS-PAGE ( Figure S1D). A total of 55 FMDV-specific mAbs were produced from the three cattle. Among them, 23 out of 26 B-cell antibodies (89%) were obtained from #2334, 27 out of 37 B-cell antibodies (73%) from #1217, and 5 out of 6 B-cell antibodies (83%) from #0005 cattle (P = 0.8782; χ 2 test), as shown by staining of virusinfected cells by indirect immunofluorescent assay (IFA) and binding of the purified virus antigen by ELISA (Figures 4A,B, Table S2) and VNT. Of the 55 FMDV-specific mAbs, only four recognized denatured FMDV capsid proteins by western-blot (WB; Figure 4C). This result suggests that these FMDV-specific mAbs predominantly recognize the conformational epitopes of the viral capsid.
Of the 55 FMDV-specific antibodies, 28 FMDV-neutralizing mAbs were identified ( Figure 2B). No significant difference in the percentage of neutralizing mAb clones among the three cattle was observed, and 70% of FMDV-specific IgG plasmablasts were found in #2334, 33% were found in #1217 and 60% were found in #0005 (P = 0.3317, χ 2 test). However, a large proportion of the neutralizing clones from #2334 (8 out of 16; 50%) and #0005 cattle (2 out of 3; 67%) showed high-potency cross-reactivity with the three topotypes of FMDV serotype O.  7,14,21,28,35,42,49,56,63,132,169,223, and 225 after first infection. IgG titres were expressed as the reciprocal log10 of serum dilutions that yielded 50% of the absorbance value of the negative control wells in LPB-ELISA. IgG titres less than the sensitivity of the assay (0.9) were adjusted to 0.6 in the figure. (B) IgG plasmablasts from the 3 cattle were collected to produce mAbs, which were specific for FMDV in 23 out of 26 cells from #2334, 27 out of 37 cells from #1217, and 5 out of 6 cells from #0005.  Table 2). The 28 neutralizing mAbs obtained exhibited variable neutralizing breadth and potency against four representative FMDV strains, but all of them were able to neutralize O/Mya/98 virus. We believe that this results from using this virus as a bait antigen to bind FMDV-specific single B cells, of which, 13 mAbs (i.e., B57, B73, B77, B82, C4, C9, E46, E54, F18, F128, F145, F150, and F169) showed intra-type bnAbs activity as indicated by their ability to neutralize the four representative strains of FMDV serotype O.

Immunodominant Antigenic Profile of FMDV Serotype O Specified in the Natural Host Was Revealed by Intra-Type bnAbs
To shed light on the antigenic profile of FMDV serotype O specified in the natural host, these 28 cattle-derived neutralizing mAbs were checked by WB for the differentiation of the linearneutralizing antigen site (site 1) and conformational antigen sites (sites 2-5) bound by these mAbs. The WB results showed that only mAb B66 (1/28) reacted with the bait antigen O/Mya/98 strain ( Figure 4C) and recognized the linear C terminal in the VP1 of FMDV (Figures 4D,E). The remaining mAbs (27/28) were all negative by WB but obviously reacted with the FMDV antigen in IFA and ELISA (Figures 4A,B, Table S2). This result indicated that the immunodominant antigen site in cattle was dominated by non-linear epitopes rather than the linear epitope in antigen site 1 that was defined by conventional mousederived mAbs. For the surveying of the dominance of the antigen sites bound by 13 intra-type bnAbs, a panel of 20 serum samples from experimentally infected and vaccinated cattle was used in blocking ELISA for the testing of competitiveness with the 13 intra-type bnAbs. The results showed that the 13 intra-type bnAbs competed with all the tested serum samples, with a mean titer of >1:128, suggesting that these intra-type bnAbs all recognized dominant epitopes located in the antigenic structure of FMDV serotype O ( Figure 5). By analyzing the titer frequencies of each bnAb corresponding to the tested serum samples, relatively high values were observed for bnAbs C4 (652 ± 71) and B77 (655 ± 243). Given the minor standard error of mean (SEM), the antigen site recognized by bnAb C4 might be the most immunodominant as this site elicited a relatively high and even antibody titer among different individuals.
Ultralong HCDR3 Appeared in FMDV Serotype O-Neutralizing and Non-neutralizing IgG Antibodies, but Longer HCDR3 Biased for the Neutralizing IgG Antibodies The sequences of variable domains of the 55 FMDV-specific mAbs were compared and analyzed. Every mAb was unique and harbored somatic mutations. The VH was indeed variable in length, ranging from 113 to 170 aa. VL was relatively similar in length ( Figure 6A). Six complementary-determining regions (CDRs) from the VH and VL sequences of the 55 FMDV-specific mAbs were identified using the IMGT numbering scheme. We found that the diversity of VH was mainly caused by amino acid replacement and length alteration in HCDR3, which were 7-64 aa in length, whereas the other CDRs in the VH and VL of different clones were similar in length, although amino acid replacements appeared in these regions (Figures 6A,B).
Notably, the proportion of ultralong HCDRs with more than 45 aa was 26% ( Figure 6C; 14/55) and significant difference in HCDR3 length was observed between the neutralizing and nonneutralizing mAbs at statistical level ( Figure 6D; P < 0.05). As revealed by mAb F145 with the ultralong HCDR3, which showed most potent neutralizing activity against all four representative FMDV strains, suggesting a relationship between HCDR3 length and the neutralizing breadth and potency against FMDV. These results suggest that cattle antibodies with longer HCDR3 are more inclined to display FMDV-neutralizing capacity and also suggest an involvement of ultralong HCDR3 in recognizing diverse virus antigens.
Restricted Germline Gene Segments of V H 1-7 * 02 and V λ 1-47 * 01 (or 1-52 * 01) Pairing Were Observed in All Cattle IgG Antibodies With Ultralong HCDR3s The usage of cattle IgG germline gene segments, the junction sequences and the somatic hypermutation counts of the FMDVspecific mAbs are listed in Table 3. All these antibody sequences exhibited somatic hypermutations and higher non-synonymous than synonymous mutations, as the observed amino acid rate was higher than the nucleotide mutation rate in both VH and VL sequences. At the genomic level, cattle Ig germline V(D)J gene rearrangements produced 37 unique V H -D H -J H recombinations in the γ heavy chain and nine unique V λ -J λ recombinations in the λ light chain (Figure 6K) (Figures 6I,J). Notably, all the cattle IgG antibodies that contained ultralong HCDR3 (total 14) were restricted to using the V H 1-7 * 02 gene segment. Moreover, γ heavy chain containing The result corresponding to each tested mAb was expressed as the serum dilution that yielded 50% of the absorbance OD 450 value of the PBS control wells.     the germline gene segment of V H 1-7 * 02 was exclusively paired with the λ light chain containing germline gene segments of V λ 1-47 * 01 or 1-52 * 01, and this was observed in all the cattle IgG antibodies with ultralong HCDR3 (Figure 7A). In contrast, the paring with V λ 1-47 * 01 or 1-52 * 01 containing light chain was not solely restricted to γ heavy chain with the V H 1-7 * 02 gene segment, as other V H gene segments, such as V H 1-10 * 01/2, V H 1-30 * 01/2, and V H 1-21 * 01 (V H 1-33 * 01), are also involved in the pairing, accounting for 11, 6, and 6%, respectively ( Figure 7B). We further analyzed the usage of D H gene segment in the 14 HCDR3 (150-192nt) sequences from nine FMDV-neutralizing mAbs and five non-neutralizing mAbs. As shown in Figure 6G, eight D H gene segments, namely, D H 8-2 * 01, D H 3-1 * 01, D H 6-2 * 01, D H 7-3 * 01, D H 5-2 * 01, D H 6-3 * 01, D H 4-1 * 01, and D H 1-2 * 01, were used in the ultralong HCDR3s. Given that these eight D H gene segments were only 13-23 nt in length that matched with their ultralong HCDR3s, the non-templated nucleotideaddition (also known as N-nucleotides) thus mainly contributed to the appearance of the ultralong HCDR3.

DISCUSSION
In this study, fully cattle-derived FMDV-neutralizing mAbs were successfully produced and characterized by a method combining the FACS isolation of antigen-specific single B cell, RT-PCR amplification of IgG antibody genes from single B cell and expression of antibody in CHO cells. This method is simple, convenient and feasible, and avoided complex hybridoma fusion and screening approaches. It thus represents an acceptable alternative for the development of diverse mAbs from a natural host.
The single B cell antibody approach has several unique advantages: (1) it harbors potential in isolating reactive functional mAbs against conformational determinants that can only be produced in natural hosts; (2) it facilitates the rapid screening of the correct clone efficiently and thus represents an approach that reduces time-and labor-intensive efforts and (3) it allows the screening of bnAbs that represent important tools in developing universal antiviral pharmaceuticals. To date, the generation of a mAb from plasmablasts was mainly described in humans (21), mice (22), rats (23), and rabbits (24). Recently, a pioneering study reported the possibility of obtaining mAbs from cattle plasmablasts by using this method and a model antigen derived from HIV (9). Our study used FMDV as the antigen and confirmed that the method of mAb production in cattle is feasible and efficient.
Generation of a diverse mAb repertoire is of great importance for different host species to respond specifically to countless antigenic challenges. Several mechanisms, including V-D-J recombination, somatic gene conversion and templated or untemplated somatic hypermutation, are fundamental to mAb repertoire diversification in mammals. Compared with human and mouse chromosomes, cattle chromosome 21 has a single IGH genetic locus family, which is inherently less variable and consists of 12 functional V H gene segments, 23 D H gene segments, and four functional J H gene segments (http://www.imgt.org/ IMGTveterinary/). In the obtained 55 mAbs sequences, all the heavy chain sequences corresponding to V-D-J germline genes displayed variable degrees of mutation at the nucleotide and amino acid level, and different lengths of N-nucleotides insertion at the V-D junction and/or D-J junction positions were observed in these mAbs. These results indicated that the sorted FMDVspecific plasmablasts showed somatic hypermutation, and all were mature B cells.
Sixteen functional D H genes had been reported from 23 D H gene segments in cattle and were divided into nine subgroups based on the length and identity of the coding sequence (25). It had been speculated that all members of D H 2 and D H 5 subgroups were pseudogenes because of divergent recombination signal sequences (25). However, we isolated one and five correct coding sequences in the D H 2 and D H 5 subgroups, respectively ( Figure 6F), which confirmed that D H 2 and D H 5 subgroups were functional. Cattle Ig have an ultralong HCDR3, which is thought to compensate for the limited diversity of V-D-J recombination (7). The proportion of mAbs with ultralong HCDR3s was slightly higher than a previous observation (∼9%) in bovine B cell hybridomas derived from mitogen-activated peripheral blood B cells (26). This discrepancy could result from the different methods used for isolation of the single B cell in cattle.
In this study, we observed the common phenomenon of ultralong HCDR3 appearing in the V H sequence that was associated with the V H 1-7 * 02 germline gene segment. This observation further confirmed previous descriptions (27,28). A previous report indicated that restricted V H and V λ pairing appeared in ultralong HCDR3 containing IgM antibodies derived from fetal and adult cattle B cells without antigen stimulation (29). Our study extends these results by finding that restricted V H and V λ pairing also exists in cattle IgG antibodies after the antigen-stimulated µ heavy chain (IgM) isotype switching. In previous studies, long HCDR3s greatly contributed to antigen binding activity in several potent human bnAbs against HIV and influenza virus (30,31). The characteristics of ultralong HCDR3 showed by anti-FMDV bnAbs in this study were consistent with those of bovine-derived bnAbs against HIV (9).
However, in this study, no clear evidence showed that the neutralization breadth and potency of FMDV-neutralizing mAbs are related to ultralong HCDR3 in cattle, although four intra-type bnAbs (E46, E54, F145, and F150) indeed displayed an ultralong HCDR3. Thus, selecting more FMDV strains to check the neutralization breadth and potency of these clones displaying ultralong HCDR3 could further determine the role of ultralong HCDR3 in the immune response of cattle against FMDV infection.
The screening of diverse neutralization mAbs will provide abundant tools for defining the antigenic structures of FMDV type O. Antigenic profiling as defined by mouse mAbs supported the existence of five neutralizing antigen sites on FMDV serotype O. Cattle-derived FMDV mAbs were previously developed by a mouse × cattle (lymph-derived B cell) hetero-hybridoma technique, and eight neutralizing mAbs were found to recognize identical antigen sites 2, 3 and 4 but not antigen sites 1 and 5 (16). Our study showed that most of the obtained cattle mAbs (27/28) bound to the conformational epitopes outside of antigen site 1. This may represent a significant feature of humoral immunity of cattle against FMDV. We need to further map the exact antigen sites recognized by these mAbs. However, the present data can only confirm that the immunodominant epitopes of FMDV serotype O responded by natural hosts are conformational epitopes rather than linear epitopes. Thus, comparing the data from different laboratories might provide an approach for improving understanding of the immunodominant antigenic structure of FMDV in natural hosts.
Due to the error-prone transcription of RNA-dependent RNA polymerase, and the high RNA recombination rate in vivo, the continuing circulation of FMDV in the field resulted in the emergence of many antigenic and genetic variants (32,33). Comparative analysis of the epitopes revealed by these strainspecific mAbs can explain the antigenic variation of FMDV serotype O among ME-SA, SEA and Cathay topotypes. Currently, vaccination is the cheapest method for disease control in FMDVendemic countries. However, new antigenic variants emerge periodically and likely contributed to vaccine failure. These intratype bnAbs provide abundant tools for the identification of conserved epitopes and thus potentially provide references for the design of a universal vaccine against FMDV.
Apart from the resolution of FMDV antigenic structure, cattle-derived mAbs are endowed with a wide array of potential applications. These applications can be the serotyping of virus antigens with improved specificity, measurement of neutralizing antibody titers by competitive or blocking ELISA based on bnAbs, and quantification of virus antigens in inactivated vaccines as a quality control measure. Due to the natural features of these mAbs derived from cattle, it will afford diagnostic methods good performance with improved specificity and sensitivity.
In summary, we produced diverse FMDV-specific mAbs from cattle by single B cell antibody technique and mapped the corresponding cattle germline V, D, and J gene segments. The diversity of FMDV-specific mAbs was mainly arisen from somatic mutations, N-nucleotides insertion and germline V(D)J genes recombination. The restricted usage of germline gene segment in the ultralong HCDR3 containing antibodies were also observed in these FMDV-specific mAbs. The obtained FMDV-neutralizing mAbs from cattle predominately recognized conformational epitopes on the capsid surface of FMDV serotype O. This study provided a novel method for preparing FMDV-specific mAbs repertoire from the natural host, and these mAbs will be fundamental tools for exploring protective antigenic structure and variation.

Identification of FMDV-Specific Plasmablasts in Infected Cattle
PBMCs were isolated from the heparinized blood samples of the three cattle with HISTOPAQUE 1.083 (Sigma-Aldrich, USA) according to the manufacturer's instructions, and then used in the identification of FMDV-specific plasmablasts. Highly purified FMDV O/Mya/98 (FMDV) inactivated 146S antigen was biotinylated with EZ-Link TM NHS-LC-Biotin reagent (Thermo Fisher Scientific, USA) according to the manufacturer's instructions, and the resulting biotin-FMDV 146S in combination with anti-biotin APC were used for the staining of FMDV-specific plasmablasts (34). For staining, freshly isolated PBMCs were first stained with biotin-FMDV 146S, anti-bovine CD21-PE (Bio-Rad, USA) and anti-bovine IgM-FITC (Bio-Rad, USA, labeled with FITC in-house) for 30 min at 4 • C in PBS buffer containing 2 mM EDTA and 0.5% BSA. Then, a second step antibody, mouse anti-biotin APC (Miltenyi Biotec, Germany), was added and incubated for 20 min at 4 • C. The parallel staining of PBMCs that lacked biotin-FMDV 146S was used as fluorescence minus one (FMO) control. These stained samples were immediately analyzed by flow cytometry and one million PBMCs were acquired for counting the proportion of FMDV-specific plasmablasts.

Single-Cell Sorting of FMDV-Specific Plasmablasts From Cattle PBMCs Using Flow Cytometry
Cattle PBMCs stained as above description were sorted by flow cytometry (BD FACS Aria II, USA) using a 100 µm nozzle. Cells were sorted for FMDV 146S-APC + /IgM-FITC − /CD21-PE +/− events. Before the sorting of a single cell, 10 µl/well of single cell lysis buffer containing 1 µl of DNase (ThermoFisher Scientific, USA) was added into full skirt 96-well plates (Brand, Germany), then the targeted single cells were sorted at one cell per well by a BD FACSAria Fusion TM cell sorter. Immediately, the 10 µl/well of SuperScript TM VILO TM Master Mix (Thermo Fisher Scientific, USA) containing random primers were added for the synthesis of cDNA. After 60 min incubation at 42 • C, the reaction was terminated at 85 • C for 5 min. The cDNA templates were stored in −20 • C for subsequent PCR amplification.

Single-Cell PCR Amplification and Cloning of Variable Region Genes of IgG
The amplification of antibody variable region genes was performed by nested PCR using primers for cattle IgG gamma and lambda chains (Ig γ and Ig λ; Table 1). Given the λ chain accounting for 95% of cattle light chains, only the λ light chain was amplified by PCR. Two nested PCR amplifications were run per well for the λ light and γ heavy chains. The first PCR reaction was set up in 25 µl volume with 3 µl cDNA template and outer primer pairs ( Table 1) using HotStar Taq Master Mix (Qiagen, Germany). The first round PCR begin with an initial hot-start at 95 • C for 5 min, followed by 35 cycles of 95 • C for 30 s (denaturation), 55-60 • C for 50 s (annealing temperature listed in Table 1) and 72 • C for 1 min (elongation), and ended with a final 5 min elongation at 72 • C. The second nested PCR was performed as described above, and reaction was set up in a total volume of 50 µl with 5 µl of the first round PCR product as the template and inner primer pairs ( Table 1). The final PCR product was Sanger sequenced. The paired heavy and light chain variable domain genes were synthesized by GenScript Inc. (www.genscript.com) with codon optimisation for expression in CHO cells, then cloned into cattle antibody expression vectors (CH-pcDNA3.4 and CL-pcDNA3.4 prepared in house; see Supplemental File 2), respectively, which contained cattle IgG2 heavy and light chain constant regions (CH and CL) with a tandem Myc and His tags at their respective C-terminus. These cloned antibody-expressing plasmids were proliferated in E. coli and then extracted and purified using an EndoFree Maxi plasmid kit (TIANGEN, China).

Analysis of the Germline V(D)J Gene Segments Recombination in Cattle IgG Heavy and Light Chains
The usage of cattle Ig germline V H , D H and J H and V λ and J λ gene segments and the length of CDRs in heavy and light chains were analyzed by IMGT/V-QUEST program (Version 3.5.13) in the internet server (http://www.imgt.org/IMGT_vquest/input) (35,36). This freely available program describes the mutations in VH and VL, hot spot positions in the closest germline V gene and insertions and deletions in the submitted sequences by alignment with the germline Ig gene and allele sequences of the IMGT reference directory (37), as well as the definition of CDRs based on IMGT unique numbering scheme (38).

Full Cattle-Derived Antibody Production and Purification
The full cattle-derived antibodies were expressed in ExpiCHO-S TM cells (Invitrogen, USA) following the instruction of the manufacturer. Briefly, the 2-3 × 10 5 cells/ml of CHO cells were suspended in 25 ml of serum free media in 125 mL flasks that were fixed in an orbital shaker platform at a speed of 125 rpm with a 25 mm shaking diameter. The cells were cultured at 37 • C with ≥80% relative humidity and 8% CO 2 . After 3 days of culture, CHO cells at concentration of 6 × 10 6 cells/ml were collected and resuspended in fresh media for transfection. The antibody expressing plasmids for light chain and heavy chain at a ratio of 3:2 were co-transfected into ExpiCHO-S TM cells with an ExpiFectamine TM CHO transfection kit (Invitrogen, USA), then the enhancer and feed component were added at 18 h post-transfection according to the standard protocol. Antibody-containing supernatants were harvested 10 days after transfection. The expressed mAbs were firstly purified by a HisTrap TM excel column, and the obtained elution was concentrated using a 100 kDa ultrafiltration tube and then further purified by size exclusion chromatography using Superdex 200 increase 10/300 column in an AKTA plus protein purification system (GE Life Sciences, Piscataway, NJ, USA). The purity and size of each full-length cattle IgG2 mAb were assessed by reducing and non-reducing SDS-PAGE. The concentrations of the final obtained IgG2 mAbs was determined by measuring their corresponding absorption values at a wavelength of 280 nm (A280).

Virus Neutralizing Test
The cattle-derived antibodies were titrated for viral neutralizing activity against four representative strains from the three topotypes of FMDV serotype O by using a micro-neutralization assay as previously described (39). The four strains of FMDV were O/Mya/98 (SEA topotype), O/HN/CHA/93 (Cathay topotype), O/Tibet/99 (PanAsia lineage in ME-SA topotype), and O/XJ/CHA/2017 (India 2001 lineage in ME-SA topotype, GenBank No. MF461724). Briefly, antibody samples were 2-fold serially diluted in 96-well cell culture plates in a total volume of 50 µl, and then 100 TCID 50 of FMDV in 50 µl of culture media was added to each well. After incubation for 1 h at 37 • C, ∼5 × 10 4 BHK-21 cells in 100 µl media were added to each well as indicators of residual infectivity. Normal cell wells, and 10, 100, and 1,000 TCID 50 virus control wells were used in each plate. The plates were incubated at 37 • C under 5% CO 2 conditions for 72 h before fixing and staining. The endpoint titers were calculated as the reciprocal of the last serum dilution to neutralize 100 TCID 50 FMDV in 50% of the wells. Final antibody titer was expressed as IC 50 calculated according to previously published method (9).

Indirect Immunofluorescence Assay
BHK-21 cells in 24-well plates were infected with FMDV and then fixed in precooled methanol-acetone solution at a ratio of 1:1 (v/v) for 15 min at 4 • C before appearance of cell lesions. After rinsing in PBS buffer, 200 µl/well cattle-derived antibodies at a concentration of 5 µg/ml were added and incubated for 1 h at 37 • C. After washing with PBS buffer, FITC-rabbit anti-cattle IgG (Thermo-Fisher, USA) at a 1:1,000 dilution were added and incubated in the dark for 1 h at 37 • C. Plates were washed three times with PBS and observed under an EVOS R FL Imaging System (Life Technology, USA).

Enzyme-Linked Immunosorbent Assay
Liquid-phase blocking ELISA (LPB-ELISA) was used in the evaluation of antibody titers against FMDV after vaccination of cattle by a commercial LPB-ELISA kit from Lanzhou Veterinary Research Institute (LVRI; Lanzhou, Gansu, China) and performed according to the manufacturer's instructions (40).
Indirect ELISA was used in the assessment of the reactivity of cattle-derived antibodies with the FMDV 146S antigen. In indirect ELISA experiments, 200 ng/well of viral antigen was coated in 96-well plates overnight at room temperature. The plates were then washed three times with PBST (PBS buffer plus 0.05% Tween) and blocked with 1% gelatin in PBS at 37 • C for 2 h. After three washes, the cattle-derived antibodies at a concentration of 1-10 µg/ml were added and incubated at 37 • C for 1 h. The plates were washed three times with PBST, and then HRP-conjugated anti-His tag antibody (Genscript, China) at a dilution of 1:5,000 was added to the wells. The plates were then incubated at 37 • C for 30 min and washed three times with PBST. Color was developed by adding 50 µl of TMB substrate (Pierce, Life Technology) for 10 min at room temperature. The process was stopped by adding equal volumes of 1 M H 2 SO 4 . Optical density at 405 nm (OD 450 ) was measured on a microplate reader (BioRad).
Whether the expressed mAb can bind to a native and dominant epitope of virus particles by blocking with sera derived from FMDV infected or vaccinated cattle were determined by blocking ELISA. Briefly, the tested cattle mAbs were coated in a 96-well plate overnight, and the serum samples (PBS as control) were 2-fold serially diluted and incubated with a suitable concentration of FMDV 146S antigen for 30 min at room temperature and then transferred to an ELISA plate that was coated with mAbs for 1 h incubation at room temperature. After five washes with PBST, one HRP conjugated non-neutralizing mAb (HRP-B81) binding to FMDV capsid protein was added and incubated for 1 h. HRP-B81 antibody was used as an indicative antibody to show the capture of the virus capsid protein. After 5 washes in PBST, the color was developed and the reaction was stopped by the same procedure as previously described in indirect ELISA. The value corresponding to each tested mAb was expressed as the serum dilution that yielded 50% of the OD 450 value of the PBS control. This value can in some extent reflect the dominance of the antigen site recognized by the tested mAb.

Western-Blot
FMDV 146S antigen or GST fusion protein were subjected to 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The protein bands in the gel were electrotransferred to a methanol-activated PVDF membrane followed by blocking with 5% non-fat milk in PBS overnight at 4 • C. The membrane was probed with cattle-derived mAbs diluted to 2 µg/ml in PBS at 37 • C for 1 h, followed by incubation with HRPconjugated anti-His tag antibody that was diluted 1:4,000 at 37 • C for 30 min. After thorough washing with PBST, the membrane was incubated with enhanced chemiluminescence reagent solution (Thermo Fisher Scientific) for 1 min and then exposed to X-ray film.

DATA AVAILABILITY STATEMENT
The sequences of all the cattle-derived monoclonal antibodies have been deposited in GenBank (Accession Numbers MN612644-MN612753).

ETHICS STATEMENT
All cattle experiments were performed in a biosafety level 3 laboratory at Lanzhou Veterinary Research Institute (LVRI), Chinese Academy of Agricultural Sciences (CAAS). The study was approved by the Animal Ethics Committee of CAAS. The cattle were acclimated for at least 1 week before experimentation and humanely bred during the experiment. All cattle used in this study were euthanized at the end of the experiment.