The objectives of this study were first to produce a well annotated and an easy to use DNA chip to analyze the immune response in pig and second to validate its relevance by investigating transcriptome modifications in PBMCs stimulated with LPS or PMA/ionomycin for 24 hours. The same seven biological replicates from seven distinct animals were used for transcriptome analysis, qRT-PCR and ELISA validation, and another set of seven animals was used for validation by FACS analysis. Reproducibility of the results was good.
Relevance of the SLA-RI/NRSP8-13K chip
DNA chips targeting immunity have been reported for human, mouse and a few domestic species including cow [21, 22], chicken [23, 24] and to a lesser extent pig with a unique report of a nylon membrane comprising less than 100 genes . Designing dedicated chips may be criticized because it is contradictory to the global approach that underlies a transcriptome study. Since no genome-wide expression array exists for pig and since efficient tools are required to study immunity and resistance to disease, we have constructed a generic array enriched in immunity genes. We combined a well-annotated oligonucleotide set referred to as the NRSP8-13K set that partially covers the pig genome  to a set of oligonucleotides referred to as the SLA-RI set that targets all annotated loci within the SLA complex and immunity genes outside the SLA complex. Here, we report that after LPS stimulation, among 258 differentially expressed genes (403 probes, see Table 2), 61 were common to both the generic and the SLA-RI sets and 84 were present only in the SLA-RI set. Similarly, after PMA/ionomycin stimulation, among 2689 differentially expressed genes, 353 were present in both sets and 424 were present only in the SLA-RI set. The SLA-RI set was highly informative for the analyses reported here. The SLA-RI set may be merged with any other generic set and it is anticipated that the number of overlapping probes between sets should increase as a function of the genome coverage in the next generation expression arrays. The SLA-RI/NRSP8-13K chip was shown to be suitable to identify immunity and disease-related biological pathways and functions as well as to construct relevant gene networks. Validation of differential expression was carried out for several genes at the RNA level by qRT-PCR and at the protein level by ELISA tests or FACS analysis. The results show significant correlations between mRNA and protein expression levels, confirming the accuracy of the chip annotation. DNA chips for expression studies are currently replaced by sequence-based transcriptome using the NGS technology, suggesting that the design of genome-wide DNA chips could be skipped and that sequencing could be used directly for transcriptome analysis. However, concentrating all efforts on the NGS technology might hamper the analysis of numerous animals and samples as required for eQTL studies and genetic genomics . We are quite convinced that the NGS technology and well-annotated DNA chips will remain complementary for a while in domestic species. The SLA-RI/NRSP8-13K chip reported here represents an accurately annotated chip dedicated to the pig immune system and will provide a valuable tool for diagnostics and research.
Choice of the in vitro models to study immune response activation
PMA, also known as 12-O-tetradecanoylphorbol-13-acetate (TPA), is a potent tumor promoter often used in biomedical research . Ionomycin is an ionophore produced by Streptomyces conglobatus. PMA in conjunction with ionomycin is known to activate T and B cells and has been used in numerous immune-related studies [25, 28, 29]. LPS is a major structural component of the outer membrane of gram-negative bacteria and is a well-referenced PAMP. LPS stimulation of mammalian cells occurs through a series of interactions with proteins including LPS binding protein, CD14, MD-2 and TLR4 . LPS is one of the best studied immunostimulatory components of bacteria and can induce systemic inflammation and sepsis if excessive signals occur . LPS-stimulation mimics a bacterial infection and has been extensively used to study innate immune response [31, 32]. Two recent studies in pig have reported transcriptome modifications in mesenteric lymph node or spleen after infection by Salmonella enterica serovar Choleraesuis (S. Choleraesuis)  and Haemophilus parasuis (H. parasuis) , respectively. S. Choleraesuis and H. parasuis are both gram-negative bacteria. Our results on LPS-stimulation reveal that many genes already identified after in vivo infection by S. Choleraesuis and H. parasuis are up- or down-regulated confirming that in vitro LPS activation of PBMCs is a good model to study innate immune response to infection with gram negative bacteria in pig.
Indeed, LPS and PMA/ionomycin stimulations were chosen because they are widely used as gold standard in vitro models to measure cytokines released in the medium by PBMCs in many species. A unique time point was studied and we are aware that all the results reported here correspond to this unique time point i.e. 24 hours after stimulation. It has been reported that time points earlier than 24 hours are more relevant to decipher the onset of the response to stimulus as shown in kinetics studies in cow , pig , mouse  or human . Moreover, kinetics studies have revealed that many genes return to their basal expression level by 48 hours of stimulation, suggesting that homeostasis is restored at that time [22, 25]. In this report, we were interested in studying the PBMC transcriptome at the time when cytokines released in the medium are efficiently measured. Our results provide many candidate genes to test for kinetics studies and ongoing complementary studies focus on this topic.
Significant positive correlations have been reported between transcriptomes of total PBMCs and purified monocytes stimulated with LPS, suggesting that for studies focussing on the most differentially expressed genes, separating and analysing cell subpopulations may be unnecessary . Therefore, the results reported here correspond to the most striking transcriptome modifications during immune response activation and may miss some subtle changes that occur in each cell subtype. Identifying transcriptome modifications occurring in each cell subtype is a major objective to better decipher immune response. However, transcriptomic signatures of blood or total PBMCs are of high interest in clinical research and most studies relate to total PBMCs in pig .
Specific transcriptome modifications after LPS stimulation
Almost half of the transcriptome modifications due to LPS stimulation are related to Disease and Disorder biological function (see Figure 3). Most of the up-regulated genes relate to inflammation and innate immune response, as expected. SAA1 and pro-inflammatory chemokines IL8, CCL2, CXCL5, CXCL3, CXCL2 and CCL8 belong to the top-ten most up-regulated genes (see Table 3), SAA1 being the most up-regulated gene with a 27-fold change by comparison to mock-stimulated PBMCs. SAA1 encodes the major acute-phase protein Serum Amyloid A (SAA), the precise role of which is still unclear despite reports suggesting a key role in the establishment and maintenance of inflammation notably as an antiapoptotic agent for neutrophils  and as an opsonin that would facilitate phagocytosis of gram-negative bacteria . SAA1 was also found as the most up-regulated gene in spleen seven days after infection by H. parasuis . The chemokines IL8, CXCL5, CXCL3 and CXCL2 have chemotaxis for neutrophils whereas the chemokines CCL2 and CCL8 have a broader chemotaxis spectrum specific for T, dendritic and NK cells as well as monocytes and basophils . Up-regulation of IL8 has already been reported in pig PBMCs [39, 40] and amnion  after bacterial infection. In human, stimulation of PBMCs with LPS induces the secretion of CCL2 , CXCL3 and CXCL2 . CXCL5 is up-regulated in LPS-challenged bovine mammary epithelial cells . All these results confirm the essential role of chemokines in chemoattraction and cell guidance to the site of infection during bacterial infection. IL1 has been reported to activate chemokine production . In our study, we found that IL1 was moderately up-regulated after 24 hours of stimulation and that it occupies a central position in the LPS-related network 2 providing a global image of inflammation activation (Figure 4B).
We have also found other strongly up-regulated genes after LPS stimulation including SOD2, S100A9 and S100A12. S100A9 and S100A12 are members of the S100 family, which encodes proteins containing two EF-hand calcium-binding motifs and are involved in the regulation of a number of cellular processes such as cell cycle progression and differentiation. In human, S100A9 has been reported to be up-regulated in LPS-stimulated bronchial epithelial cells , suggesting that this gene has a role in innate immune defense. SOD2 (superoxide dismutase 2) is a member of the iron/manganese superoxide dismutase family and it has been shown to be up-regulated in dendritic cells after LPS stimulation . These results suggest that the calcium pathways as well as oxidative processes are strongly affected in LPS stimulated PBMCs. Interestingly, the six genes S100A9, CXCL5, S100A12, IL8 CXCL2 and SOD2 were also found strongly up-regulated in mesenteric lymph nodes of pigs infected by S. Choleraesuis  and S100A9, S100A12 and SOD2 were also up-regulated in spleen after H. parasuis infection . Overall these results confirm a predominant role of common genes in the innate immune response of pig to gram-negative bacterial infections.
Specific transcriptome modifications after PMA/ionomycin stimulation
By examining the top-ten most up-regulated genes (Table 3) and the representation of KEGG pathways (see Table 5 and Figure 5B), we have found that after PMA/ionomycin stimulation, T cells are more activated than B cells. Three of the most up-regulated genes are IL2 (fold change close to 10), CD69 (fold change = 6.9) and TNFRSF9 (fold change = 6.6), which are related to T cell activation. B cell markers such as CD40 and INHBA are also up-regulated (Additional file 9 : SLA_RI_Table_S9.xls) but with fold changes of 1.6 and 5, respectively. It has been reported that transcription of the cytokine IL2 is the main consequence of T cell activation and that IL2 is produced by T helper cells harboring a Th1 cytokine profile . IL2 is essential for the generation and regulation of immune response. Binding of IL2 activates the Ras/MAPK, JAK/Stat and PI 3-kinase/Akt signaling module pathways (IPA, Ingenuity). IL2 signals through the IL2 receptor and in this study, we have found that the interleukin 2 receptor alpha (IL2RA) is also up-regulated after PMA/ionomycin stimulation as previously reported . The IL-2/IL-2R interaction stimulates the growth, differentiation and survival of antigen-selected cytotoxic T cells . The activation of T lymphocytes, both in vivo and in vitro, induces expression of CD69 that has been reported as the earliest inducible cell surface glycoprotein acquired during lymphoid activation. This molecule is involved in lymphocyte proliferation and functions as a signal-transmitting receptor in lymphocytes, natural killer cells and platelets . TNFRSF9 also known as 4-1BB is a member of the TNF-receptor superfamily and is a CD4+ T cell marker that regulates CD28 co-stimulation to promote Th1 cell response. The expression of this receptor is induced by lymphocyte activation and is involved in T cell division and expansion . In agreement with these findings, IL12B that is known to trigger Th1 response is also found up-regulated with a limited fold change of 1.5. Globally, our results show a strong up-regulation of cytokines and genes related to Th1 response, suggesting a more pronounced activation of the Th1 response compared to a Th2 response after PMA/ionomycin stimulation for 24 hours.
Strikingly, a very strong down-regulation of the THBS1 gene was observed after PMA/ionomycin stimulation for 24 hours with a reduced expression of 66 fold change by comparison to mock-stimulation (Table 3). This gene is also down-regulated in porcine aortic endothelial cells treated with PMA  and in bovine PBMCs stimulated with Concanavalin A . Interestingly, THBS1 down-regulation has been shown to become stronger with time after Concanavalin A stimulation in bovine PBMCs , suggesting a persistent role of this gene repression during immune activation and a delayed response or no response for returning to pre-induction levels. THBS1 encodes an adhesive glycoprotein that mediates cell-to-cell and cell-to-matrix interactions. THBS1 has been described with many diverse functions that may relate to its structure and consequently to its ability to bind to matrix proteins, cell surface receptors or other molecules including cytokines . THBS1 is known as a potent natural inhibitor of angiogenesis and endothelial cell migration . THBS1 has been shown to be regulated by DNA methylation and to be a target for a transcription repression induced by the protein arginine methyltransferase 6 (PRMT6). Our work suggests that the strong repression of THBS1 observed in pig PBMCs may be due to methylation and that the PRMT6 gene may have a role in this repression. Interestingly, CD47, which encodes a membrane protein that is a receptor for the C-terminal cell binding domain of THBS1  was also found slightly down-regulated (fold change of -1.4). Recent findings suggest that THBS1 contributes to the vascular system regulation by acting via its receptor CD47 to inhibit nitric oxide signaling . Our findings suggest a major role of THBS1 repression in T/B cell activation upon stimulation with PMA/ionomycin, by enhancing the ability of cells to proliferate and migrate. Whether this role is connected to CD47 or to other receptors has to be further investigated.
Our study confirms an up-regulation of the pro-inflammatory cytokine IL8 but has not found an over-expression of IFNG as previously reported by Ledger et al. .
Down-regulation of MHC mediated antigen presentation pathways after both stimulations
A strong down-regulation of MHC class II or MHC class I and II molecules was observed after LPS or PMA/ionomycin stimulation, respectively. Classical class II molecules are involved in antigen presentation to CD4+ T cells whereas classical class I genes have a double function of antigen presentation to CD8+ T cells and regulation of natural killer cell cytotoxicity by interacting with NK receptors such as NKG2D . In pig, down-regulation of MHC genes has been reported in vivo in the spleen of animals infected by H. parasuis  and in vitro in PK15 cells infected by the pseudorabies virus  and in PBMCs stimulated with PMA/ionomycin . In human, such a repression has also been reported in PBMCs infected by bacterial LPS and diverse killed bacteria . Our results show that the repression program includes classical (DR and DQ series) and non classical (DM series) SLA class II genes after LPS and PMA/ionomycin stimulation as reported in human PBMCs. In addition, classical class I genes corresponding to SLA-3 and likely to SLA-1 and SLA-2 are also repressed together with the non classical genes SLA-6 and SLA-7  that map to the SLA complex on chromosome 7 and CD1 that maps to chromosome 4 thus outside of the MHC locus. Strikingly, in our study, after PMA/ionomycin stimulation, biological networks connect the down-regulation of MHC class I molecules to a significant increase in transcription of numerous heat shock proteins known to act as chaperones as well as in transcription of all genes involved in the cascade of peptide processing before loading to the MHC molecule binding groove.
Induction of MHC class I expression is mainly transcriptional and promoters of class I genes contain IFN-stimulated response elements (ISRE) that bind factors of the IFN regulatory factor (IRF) family. Therefore, expression of IRFs influences transcription of class I genes. In our study, IRF1 and IRF8 are found up-regulated after PMA/ionomycin stimulation in contrast to IRF2 and IRF5 that are repressed. IRF8-mediated inhibition of antigen presentation by dendritic cells in the tumor microenvironment has been described in human  but our results are not in concordance with a possible role of IRF8 in MHC class I repression since the repression in peptide presentation by class I molecules was linked with a down-regulation of IRF8 together with a down-regulation of the peptide processing cascade . In contrast, the down-regulation of IRF1 is in agreement with a possible role of this gene in inhibiting transcription of MHC class I genes .
Comparison of transcriptomic signatures specific to LPS and PMA/ionomycin stimulations
In this study, about ten times more genes are found differentially expressed after PMA/ionomycin than after LPS stimulation. This might be related to the fact that LPS targets monocytes and macrophages  expressing CD14 and that PMA/ionomycin have a much wider spectrum of target cells. However, it cannot be ruled out that the significant difference in the number of differentially expressed genes according to stimulation is due to variations in the dynamics of the response. The onset of response may occur much earlier for LPS than for PMA/ionomycin. As a counterpart, the return to basal levels of gene transcription may also occur earlier after LPS stimulation, providing a possible hypothesis for a reduced number of differentially expressed genes after 24 hours stimulation. Additional studies are required to specifically address this question.
Specific and common features in transcriptome modifications were identified for both stimulations at 24 hours. Strikingly, the most significant similarly regulated genes after both stimulations are found down-regulated and many specific genes appear to be up-regulated.
Hierarchical clustering of genes found differentially expressed in both stimulation conditions provided a clear image of genes that were regulated either in the same direction or in opposite directions according to stimulation. In that respect, clusters C2, C4 and C7 (Figure 2) are the most informative to compare signatures and target possible markers that might be regulated in opposite directions according to stimulation. THBS1 (cluster C2), SAA1, chemokines CCL2, CXCL5 and CXCL6 (cluster C4) as well as IL1 receptor, immunoglobulins and LTB (cluster C7) provide a limited subset of genes that are specifically up-regulated after LPS stimulation and down-regulated after PMA/ionomycin stimulation. Similarly, cluster C7 including genes such as the chemokine CXCL10 and IRF8 provides a reservoir of genes specifically up-regulated after PMA/ionomycin stimulation and down-regulated after LPS stimulation.
Role of non-coding transcripts
After LPS and PMA/ionomycin stimulation, quite a high number of probes corresponding to annotated transcripts in the anti-sense orientation are expressed. It is likely that the expressed antisense probes correspond to either new non annotated transcripts or to antisense transcripts from annotated genes. Interestingly, few of the anti-sense probes are differentially expressed after PMA/ionomycin stimulation suggesting a role in immune response activation that has to be further explored. These preliminary results on the expression of non-coding transcripts mapping to the SLA complex corresponds to a pilot study that would be worth extending to the whole genome.