Association of Porcine Swine Leukocyte Antigen (SLA) Haplotypes with B- and T-Cell Immune Response to Foot-and-Mouth Disease Virus (FMDV) Peptides

Dendrimer peptides are promising vaccine candidates against the foot-and-mouth disease virus (FMDV). Several B-cell epitope (B2T) dendrimers, harboring a major FMDV antigenic B-cell site in VP1 protein, are covalently linked to heterotypic T-cell epitopes from 3A and/or 3D proteins, and elicited consistent levels of neutralizing antibodies and IFN-γ-producing cells in pigs. To address the contribution of the highly polymorphic nature of the porcine MHC (SLA, swine leukocyte antigen) on the immunogenicity of B2T dendrimers, low-resolution (Lr) haplotyping was performed. We looked for possible correlations between particular Lr haplotypes with neutralizing antibody and T-cell responses induced by B2T peptides. In this study, 63 pigs immunized with B2T dendrimers and 10 non-immunized (control) animals are analyzed. The results reveal a robust significant correlation between SLA class-II Lr haplotypes and the T-cell response. Similar correlations of T-cell response with SLA class-I Lr haplotypes, and between B-cell antibody response and SLA class-I and SLA class-II Lr haplotypes, were only found when the sample was reduced to animals with Lr haplotypes represented more than once. These results support the contribution of SLA class-II restricted T-cells to the magnitude of the T-cell response and to the antibody response evoked by the B2T dendrimers, being of potential value for peptide vaccine design against FMDV.


Introduction
The adaptive immune system comprises two arms-one responsible for the antibody (B-cell), the other for the cytotoxic (CTL) response-and both dependent on T-helper cells (Th cells). Presentation of T-cell epitopes in association with MHC class I and class II molecules is the initial and essential step in the activation of T CD4 + and CD8 + responses. Accordingly, an efficient vaccine must incorporate epitopes recognized by both B-and T-cells, ideally with the latter being widely recognized and presented by MHC alleles frequently occurring in individuals from natural host populations [1]. This requirement is particularly relevant for subunit vaccines harboring a limited number of T-cell epitopes for which the chances to be efficiently presented by all MHC haplotypes within a population can result in the existence of "non-or poor-responder" individuals [2,3].
Foot-and-mouth disease virus (FMDV) is the etiological agent of a devastating livestock disease of great concern worldwide [4,5]. The measures used to control the disease include extensive and frequent vaccination with whole virus inactivated vaccines and are, thus, of major economic importance [6]. Despite conferring FMD protection when adequately formulated, produced, and delivered, inactivated vaccines have limitations: (i) the requirement of a constant cold-chain (4 • C) to preserve the thermal stability of FMDV particles whose immunogenicity decreases significantly, due to its disassociation at higher temperatures into pentamers [7]; (ii) BSL-3 facilities are needed for the production of live viruses [7]; (iii) the problems associated with ensuring the DIVA (differentiating infected from vaccinated animals) condition of vaccines is met [7].
For these and other reasons, the development of alternative vaccines (such as those based on either the viral capsid protein, VP1 [8], or synthetic peptides) are equivalent to specific regions of the same protein have been pursued for decades. Bittle et al. showed that a peptide corresponding to the G-H loop in VP1 capsid protein-induced neutralizing antibodies (nAbs) in mice and protection in guinea pigs [9]. This peptide linearly juxtaposed to the C-terminal of VP1, previously reported to induce nAbs [10], was able to protect cattle from virus challenge [11]. Despite the vaccine potential of FMDV peptides, the main problem faced during decades was their weaker immunogenicity when compared with conventional, inactivated vaccines [12]. Advantages of peptide vaccines include: (i) safety, as a non-infectious material is required; (ii) DIVA condition; (iii) easy handling and storage (no cold chain is required); (iv) chemical stability; and (v) large affordable scale production. These potential benefits led to further attempts to improve their immunogenicity, such as the inclusion of promiscuous T-cell epitopes (recognized by different MHC molecules, capable of evoking adequate T-cell responses) [13]. Indeed, the improvement of epitopic vaccines by incorporation of T-cell epitopes has been reported for several virus models, such as human immunodeficiency virus (HIV) and hepatitis C virus (HCV) [14,15].
We previously reported the association between a polymorphism in bovine MHC (BoLA) class II molecules and both the immune response and protection induced by vaccines based on linear peptides corresponding to the VP1 G-H loop epitope, either alone or in combination with other B-cell epitopes and/or with a VP1 T-cell epitope identified in vaccinated cattle [13]. Protection showed a limited correlation with specific serum neutralizing activity and a higher correlation with the induction of T-cell responses [16]. Consistently, associations were observed between certain DRB alleles (DRB3.2 *1, 3, and 7) and increased levels of protection and the presence of others (DRB3.2 *12 and 18) and lack of protection [16]. These results supported that the gene polymorphisms in bovine class II MHC influenced the recognition of the individual epitopes, resulting in the animal-to-animal variation observed in both humoral and cellular immune responses [2]. In addition, several FMDV T-cell epitopes capable of binding defined swine leukocyte antigen (SLA) class I [17,18] and class II molecules [19] have been described. These results reinforced the need for adequate T-cell activation for efficient peptide-based protection.
The highly polymorphic nature of the porcine MHC, which codes for the swine leukocyte antigens (SLAs), allows for the presentation of a wide panel of antigenic peptides, and thus, influences disease resistance and vaccine responsiveness. As pathogen effects on SLA gene expression drive the regulation of swine immune responses, SLA-typed pigs are used in vaccine design (reviewed in [29,30]). Despite the ongoing domestication process involving selection for favorable traits and inbreeding, pigs still appear to maintain a high degree of SLA diversity as demonstrated by the presence of the 266 and 227 class I and class II alleles, respectively [29]. To date, there are 73 independent class I (SLA-1, -2, -3) and 51 class II (DRB1, DQB1, DQA) assigned haplotypes [29].
In the set of experiments in which the immunogenicity of B 2 T dendrimers was assessed, a total of 73 peptide-immunized and control animals were analyzed. In these experiments, the porcine T-cell epitopes selected to be part of the B 2 T dendrimeric vaccines evoked T-cell responses in all the 63 peptide-immunized pigs analyzed, being these responses high in most of the animals, which is in line with the high titers of neutralizing antibodies that, in average, they elicited.
In this study, we have typed these pigs at the SLA-I (SLA-1, SLA-3, SLA-2) and SLA-II (DRB-1, DQB-1, DQA) loci and analyzed the associations/correlations between particular SLA low-resolution haplotypes (Lr-Hp) with neutralizing antibody and T-cell responses induced by B 2 T peptides.

Pigs and Experimental Design
The immune response to B 2 T dendrimer peptides was assessed in 9 to 12 weeks-old White cross-bred Landrace female pigs (Agropardal SA breed) for experiments 1, 3, 4, and 5 and TOPIGS 20TM breed for experiment 2 ( Table 2). Pigs were randomly assigned to different animal groups and immunized by intramuscular injection with 2 mL of Montanide ISA 50V2 emulsion (Seppic, Puteaux, France) with the B 2 T dendrimer peptide as the schedule shown in Table 2. Non-vaccinated pigs inoculated with PBS with Montanide ISA 50V2 were included as controls. FMDV neutralizing antibodies to O/UKG/11/01 expressed as virus neutralization titers (VNT) and the frequency of in vitro IFN-γ-producing specific T-cells were determined as described [24]. For the virus neutralization test, the numbers represent end-point titers, calculated as the reciprocal of the final serum dilution that neutralized 100 TCID 50 of homologous FMDV in 50% of the wells. The frequency of peptide-specific T cells in the responding population determined by IFN-γ-ELISPOT is expressed as the mean number
The pigs included in this study belonged to five independent experiments, for which references are provided.

SLA Typing by PCR-SSP
SLA polymorphisms were analyzed using DNA extracted from PBMC of immunized animals. Therefore, genomic DNA was isolated from porcine PBMC using a DNeasy Blood and Tissue Kit (Qiagen, Düsseldorf, Germany) according to the manufacturer's instructions. SLA class I (SLA-I) and SLA class II (SLA-II) low-resolution haplotypes (Lr-Hp) were identified by a PCR-based typing assay (PCR-SSP) to define the animals' MHC background on the allele-group level. SLA typing was performed by PCR with the complete set of typing primers specific for the allele groups of three SLA class I loci, SLA-1, SLA-2, and SLA-3; and three SLA class II loci, DRB1, DQB1, and DQA [32,33].
The criteria and nomenclature used for SLA-I and SLA-II haplotyping were based on those proposed by the SLA Nomenclature Committee [34] and reviewed in [29]. Interpretation of the results was based on the presence of allele-specific PCR products of the expected size in each lane. Low-resolution SLA class I and class II haplotypes were deduced based on the comparison with previously published haplotypes [34] and reviewed in [29] and unpublished breed or farm-specific haplotypes (S.E. Hammer, C.-S. Ho, unpublished data).

Data Preparation and Statistical Analyses
Before conducting the analysis, the two haplotypes (determined for both SLA-I and SLA-II, respectively) were combined and converted to a text string in which the order in the combination was irrelevant (X + Y ≥ XY) and (Y + X ≥ XY). On the other hand, since both T-cell responses and antibody titers were continuous variables, they were discretized in 3 (0: <50; 1: >50 and <90; 2: >90 and <140; 3: >140) and 5 (0: <0.0; 1: >0.9 and <1.5; 2: >1.5 and <1.8; 3: >1.8 and <2.4; 4: >2.4 and <3; 5: >3) groups, respectively. Due to the variability in the T-cell response (measured in IFN-γ ELISPOT assays), discretization in the three selected groups was chosen as it distinguishes between low, medium, and high responder animals. The five groups in which the VNTs were discretized corresponded to the titers from each of the ten serum dilutions analyzed (two-fold dilutions starting at 1 4 ). Each value of discretization (from 0-5) included titers from two consecutive serum dilutions. As a robustness analysis, in addition to the proposed discretization, we have conducted alternative correlation tests with different categorizations using a ±4% variations of the proposed limits.
Since SLA-I, SLA-II, and SLA I+SLA II Lr haplotypes data can be considered nominal variables, a non-parametric statistical test, the Chi-square (χ 2 ) correlation analysis was used [35]. In addition, the Cramer's value (Cv) was determined; it does not provide estimates of significance, but measures the strength of the association among variables [36]. This value ranges from 0 to 1, with those closest to one denoting a strong influence among studied variables. To complete the statistical analysis, two additional tests were used: Maximum likelihood ratio chi-square (LR χ 2 ) and Fisher's exact test, more suitable for too small sample size, and analyzing paired variables with less 20-dimension, respectively. In all cases, the null hypothesis was rejected with a confidence greater than 95% (α = 0.05).

Ethics Statement
Experiments 1, 3, 4, and 5: Experimental procedures were conducted in accordance with protocols approved by the CSIC Committees on Ethical and Animal Welfare and by the National Committee on Ethics and Animal Welfare (PROEX 034/15). Experiment 2: The study was approved (no. 2013121) by the Central Veterinary Institute (CVI) animal experiment ethical review committee in compliance with Dutch law.

SLA-I and SLA-II Allele-Groups and Haplotypes Identified in the Studied Pigs
A total of 63 samples from pigs immunized with one or two doses of different FMDV dendrimer peptides and 10 samples from control non-immunized animals were available from five previous independent experiments in which pigs were immunized with B 2 T dendrimer peptides (Tables 1  and 2). These samples were subjected to SLA class I and class II typing, and the allele-groups and the low-resolution haplotypes identified are presented in Table S1.
2). These samples were subjected to SLA class I and class II typing, and the allele-groups and the lowresolution haplotypes identified are presented in Table S1.

Correlations between SLA-I and SLA-II Haplotypes and Immune Response to Dendrimer Peptides
This set of data was used to seek for possible correlations between SLA-I, SLA-II, and SLAI+SLAII Lr haplotypes and the two immune parameters analyzed. To this end, the individual values of VNT (henceforth antibody response), as well as of frequencies of IFN-γ-releasing activated T-cells (henceforth T-cell response) were grouped in five or three categories, respectively (Table S2). In addition, the SLA-I and SLA-II Lr haplotypes defined by the different allele combinations were identified and termed alphabetically (Table 3). Statistical analyses were performed in order to find possible correlations between haplotypes an immune response, and only those correlations giving positive values with χ 2 and LR χ 2 tests (α < 0.05), were considered as significant. Indeed, the only robust significant correlation found for the 63 B 2 T immunized pigs (group 0) was noticed between SLA-II Lr haplotype and the T-cell response (Figure 2A), which was statistically significant when using the χ 2 (α = 0.015) and LR χ 2 (α = 0.038) tests, and with a Cramer's value of Cv = 0.66, which indicates a strong association among variables (Table 4). In addition, all correlation tests conducted with alternative discretization limits (±4%) were also statistically significant for SLA-II Lr haplotype and the T-cell response, supporting the evidence of having a strong and robust association among these variables. A trend for the association of SLA-I Lr haplotypes and the cellular response was also observed, although it was only statistically significant when using the LR χ 2 test (α = 0.046) ( Figure 2C, Table 4). On the other hand, no significant correlations were observed with any of the tests used among SLA-I or SLA-II Lr haplotypes and antibody responses ( Figure 2B,D, Table 4), although a trend in correlation was observed for the combination of SLA-I and SLA-II Lr haplotypes, only supported by a statistically significant χ 2 test (χ 2 α = 0.042, LR χ 2 α = 0.914) with a high Cramer's value, Cv = 0.81 (Table 4). Table 3. SLA-I and SLA-II Low-resolution Haplotype notation and occurrence.  Table S3. Antibody and cellular response values as in Table S2 and Figure S1.  Table S3. Antibody and cellular response values as in Table S2 and Figure S1.

Association of SLA-I and SLA-II Haplotypes with High Immune Responses
Next, we searched for SLA Lr haplotypes associated with high immune responses. Shared SLA Lr haplotypes among animals eliciting higher antibody response (values of 5 and 4 in Table S2) and cellular response (value of 3 in Table S2) are shown in Table 3 Table S3, antibody response values as in Table S2 and Figure S1. Finally, in group 2, statistical significance only observed in χ 2 tests, suggested a faint correlation between antibody response and SLA-II Lr haplotypes (χ 2 α = 0.050, LR χ 2 α = 0.264, Fisher's exact α = 0.074, Cv = 0.62) and in the combination of SLA-I and SLA-II Lr haplotypes (χ 2 α = 0.035, LR χ 2 α = 0.416, Fisher's exact α = 0.807, Cv = 0.8) (Table 4).

Association of SLA-I and SLA-II Haplotypes with High Immune Responses
Next, we searched for SLA Lr haplotypes associated with high immune responses. Shared SLA Lr haplotypes among animals eliciting higher antibody response (values of 5 and 4 in Table S2) and cellular response (value of 3 in Table S2) are shown in Table 3 The allele-group combinations mainly associated with low antibody responses (value of 1 in Table S2) were Lr-Hp 37.0. (SLA-I) and Lr-Hp 0.15b and 0.24 (SLA-II). Interestingly, while Lr-0.15b is one of the most frequently represented haplotypes and it also appeared in the group of high antibody responders, Lr-0.24, although less frequent, is present only in animals with low antibody responses ( Table 5). Table 5. Shared allele-groups and Low-resolution Haplotypes among the T-cell high responding animals and antibody high and low responder animals.

Discussion
Incorporation of epitopes efficiently recognized by both B-and T-cells, ideally with the latter being widely recognized and presented by MHC alleles frequently represented in host populations, is a crucial issue for developing efficient new FMD vaccines based on virus subunits, such as peptide vaccines.
Here, we have analyzed three SLA class I loci and three SLA class II loci to search for possible correlations between individual allele-groups and Lr haplotypes with the outcome of the antibody and the cellular responses elicited by B 2 T, which had induced high levels of neutralizing antibodies, as well as evoked significant T-cell responses, supporting the requirement of specific T-cell help for the induction of FMDV neutralizing antibodies in this species.
A cohort of 63 pigs, immunized with FMDV B 2 T dendrimers, and 10 non-immunized (control) animals, was analyzed. In these animals 23 SLA class allele-groups, and 23 SLA class II allele-groups at the three SLA loci for both SLA gene clusters were found. The found allele-groups contribute to 15 SLA-I and 16 SLA-II Lr-Hp, reflecting the presumed high level of SLA polymorphism in farmed pigs. For example, among a studied cohort of 518 European farmed pigs, 51 SLA-I and 36 SLA-II Lr-Hp have been described (S.E. Hammer, unpublished data).
The association found between the polymorphism in SLA molecules and the neutralizing antibodies and T-cell responses induced by FMD dendrimer peptide vaccines in swine is similar to that reported in cattle between polymorphism in bovine BoLA class II molecules elicited by linear FMDV peptides in cattle [2]. Nevertheless, while for linear peptides, the only significant correlation was observed between certain BoLA class II allele groups and lack of protective response in cattle, our results point to a robust significant correlation in the 63 immunized pigs (group 0) analyzed between SLA-II Lr haplotypes and the T-cell response.
Studies on the correlation of SLA polymorphisms and phenotypic traits have to face the need for a large number of individuals to be able to obtain significant results. In this scenario, the selection of the statistical analyses and the criteria for significance followed turns essential. Thus, we adopted a strict criterion to consider a correlation as statistically robust being significant when using Pearson Chi-square (χ 2 ) test-which adjusts to correlation analyses using categorical values and multiple groups-, the maximum likelihood ratio chi-square (LR χ 2 ) test, and when feasible, the Fisher's exact test. In addition, Cramer's value (Cv) has been calculated, giving robustness to the correlation study as it indicates the degree of correlation between analyzed variables.
In our complete cohort (73 pigs), the proportion of Lr haplotypes found in a single animal was 36% for SLA-I, and 40% for SLA-II, which limited the significance of the comparisons. Indeed, that was reflected in the fact that the correlation between SLA-II and the T-cell response was the only scored as significant with the tests used. When animals with haplotypes represented were only removed once from the analysis, several significant correlations were observed, entrenching the value of the analysis. Overall, these results support the contribution of SLA-II restricted T-cells to the magnitude of the specific SLA-II restricted T-cells evoked by the B 2 T dendrimers in pigs.
On the other hand, full significant correlations with B-cell antibody response were only observed in group 3 (comprising the combination of SLA-I and SLA-II Lr haplotypes represented in the cohort), with both SLA-II and SLA-II+SLA-I giving full significant correlations.
The correlation of T-cell response with SLA-I haplotype was not as evident as with SLA-II, since robust correlations were only observed when the cohort was reduced to animals with SLA-I Lr haplotypes represented more than once. These results are consistent with the kind of T-cell epitopes included in the dendrimers, as they were previously identified as swine T helper epitopes, mainly recognized by SLA class II molecules [21,23]. No robust significant correlation between antibody response and SLA-I Lr haplotypes was found in any of the groups, which is consistent with a limited implication of class I-restricted responses in B-cell activation.
The implication of single Lr haplotypes was also noticed in the magnitude of the specific responses elicited in immunized pigs. Strong involvement in the T-cell response was observed for Lr-Hp 22.0, 59.0, and 37.0 (SLA-I) and Lr-Hp 0.15b and 0.27 (SLA-II). Likewise, the allele-groups associated with high antibody titers were Lr-Hp 59.0 and 22.0 (SLA-I) and Lr-Hp 0.15b and 0.27 (SLA-II). Interestingly, one of the SLA-II Lr haplotypes, Lr-0.24, was found mostly associated with low antibody responses.
SLA-I Lr-Hp 59.0, 22.0, and 37.0 are the most frequently represented in animals with high humoral and cellular responses, but Lr-37.0 is also highly represented in animals with low antibody titers. As Lr-Hp 37.0 and 59.0 share the allele group SLA-3*05XX, the other non-shared allele groups could be responsible for these differences. So, SLA-1*08XX, and 11XX, SLA-306XX and SLA-216:02 and *12XX allele groups could be associated with higher immune responses.
The most frequent SLA-II Lr haplotype is 0.15b, represented at a high rate in animals with high immune responses, while SLA-II Lr haplotype 0.27 is only present in animals with high antibody and cellular response. Two of the allele groups belonging to this haplotype (DRB1*09XX and DQB1*09XX) could also be related to high immune responses. The third allele group (DQA*04XX), which is also present in Lr haplotype 0.21, is represented in animals with high antibody titers, but not high T-cell responses, and is mainly associated with a high humoral response.
SLA-II haplotype Lr-0.24, which was represented in only eight of the immunized pigs, was not found among the high antibody responder animals, suggesting that its presence might lead to lower antibody responses. As Lr-Hp 0.15b and 0.24 haplotypes share the same allele groups for the DQB1, and DQA loci, a possible association of DRB1*07XX allele group, only present in Lr-Hp 0.24, with a low humoral response, could be found out.
The correlations found could bear potential, as SLA-I Lr-Hp Lr-59.0 and Lr-22.0, as well as Lr-0.15b, are highly abundant Lr haplotypes in European farmed pigs [S.E. Hammer, unpublished data].
Despite part of the pigs were challenged with FMDV (experiments 2 and 3), attempts to correlate protection and SLA Lr haplotypes were not possible, due to the limited sample size. Nevertheless, the results presented here and previous results with B 2 T dendrimer peptides in which no SLA typing was available [22,26] show a high level of protection and support a clear trend between high neutralizing antibody titers and protection against FMDV challenge.

Conclusions
This study of the contribution of the polymorphism of the porcine MHC on the immunogenicity of B 2 T dendrimers FMDV displaying B-and T-cell FMDV epitopes supports the contribution of SLA-II restricted T-cells to the magnitude of the T-cell response, and slightly less significantly, to the antibody response evoked by the B 2 T dendrimers. These results are of potential value for FMD peptide vaccine design.
Supplementary Materials: The following are available online at http://www.mdpi.com/2076-393X/8/3/513/s1. Figure S1: Antibody and cellular responses of 73 pigs to dendrimeric peptides. Table S1: SLA-I and SLA-II allele-groups and Low-resolution Haplotypes in the cohort of 73 studied pigs; Table S2: SLA-I and SLA-II Low-resolution Haplotypes of 73 pigs and their antibody and cellular responses to dendrimeric peptides; Table S3: Groups of pigs with Low-resolution Haplotypes (Lr-Hp) represented more than once.