The implementation of surveillance and management programmes has done much to reduce the incidence and prevalence of BTB over the past number of decades; however, M. bovis infection remains an important livestock disease worldwide. This is due, in part, to well-documented limitations of the currently available diagnostics tests (such as the SICTT and IFN-γ tests) leading to a failure to detect all infected animals [43, 44].
In recent years, research has shifted from a focus on protein-based diagnostics to functional genomics technologies that interrogate the host transcriptome in response to M. bovis infection. In particular, microarray technologies coupled with the rapid development of more sophisticated bovine genome resources has enabled high-resolution analyses of the genes and cellular pathways governing the host response to infection with M. bovis [23, 25, 45–47]. In the present study, we have compared the transcriptomes of PBL from non-infected control animals with actively-infected BTB animals using a high-density genome-wide bovine microarray platform.
Modulation of the host PBL transcriptome in response to M. bovis infection was evident from the large number of DE genes between the two experimental groups. Statistical analyses of the microarray data identified a total of 2,757 DE genes. Of these, 1,281 (46%) showed increased expression and 1,476 (54%) displayed decreased expression in the BTB group compared to the control animals. It is important to note, however, that the differences in cell subpopulations observed between the M. bovis-infected and control animals (Figure 1) may have contributed to the gene expression changes detected between the two experimental groups. Also, the haematology analyser results only provided a general description of the PBL cell subsets and do not provide information concerning T lymphocyte subsets in the infected and control animal groups. In addition, the cell subset results presented here differ from previous work performed by us , most likely due to the different cell sample types examined.
Analysis of the DE genes using IPA provided information regarding the immunobiology of active BTB. The highest ranking functional category identified using IPA was inflammatory response and the highest ranking subcategory within this category-affects immune response-revealed a marked bias in the number of genes displaying a decrease in relative expression (64.5%) compared to those showing an increase in relative expression (35.5%) in PBL from the BTB group.
Previous work by our research group demonstrated that suppression of host gene expression was associated with active M. bovis infection in cattle . This earlier work involved the comparison of RNA isolated from PBMC of M. bovis-infected and control animals using an immuno-specific bovine cDNA microarray (BOTL-5). The results presented here, based on the analysis of PBL-derived RNA using a genome-wide microarray, lend further support to our previous study. Indeed, published investigations by other workers suggest that transcriptional suppression is a common feature of mycobacterial infection in mammals [48–50].
Further inspection of the inflammatory response functional category in IPA identified several genes that were previously reported to be differentially expressed in cattle and other mammalian species infected with mycobacterial pathogens [23, 40, 42, 51, 52]. For example, microarray analysis showed that TLR2 and TLR4 displayed contrasting expression patterns between PBL from the two groups: TLR2 showed decreased expression and TLR4 showed increased expression relative to the control animal group. We have previously observed decreased expression of TLR2 in PBMC from actively infected BTB cattle using the immuno-specific BOTL-5 cDNA microarray; however, contrary to the results of the present study, TLR4 also showed decreased expression with the BOTL-5 cDNA microarray in actively infected animals .
The gene expression results obtained by Meade and colleagues using PBMC from M. bovis-infected and control non-infected animals were also used to identify a panel of 15 genes predictive of disease status . Four of these genes were found to be similarly differentially expressed in the current study: UNC84B (now SUN2), GAN, SFPQ and NRP1. Four other of the 15 genes identified previously (TBK1, 28S [now RN28S1], GPR98 and an anonymous BOTL clone [BOTL0100013_F01]) were not present on the Affymetrix® GeneChip® Bovine Genome Array. However, the seven remaining genes (NCOR1, PPP2R5B, UCP2, ZDHHC19, NFKB1, NRM and FGFR1) were not differentially expressed in the PBL samples from M. bovis-infected and control non-infected animals used for the present study. This discordance may be due to a number of factors, including: the blood cell sample types used (PBL versus PBMC); differences in sensitivity between the two types of microarray (the single-colour in situ-synthesised Affymetrix® GeneChip® versus a dual-colour spotted cDNA array [53, 54]); and the requirement for more stringent control of the FDR with larger numbers of genes (24,072 probe sets versus 1,391 duplicate spot features).
The role of TLR molecules in the recognition of mycobacterial PAMPs is well established [11, 38, 39, 41, 42, 55–57]. TLR2 and TLR4 activation signals are linked to the interleukin-1 receptor-associated protein kinases (IRAKs) through the adaptor molecule, myeloid differentiation primary response protein 88 (MYD88), which triggers a downstream protein signalling cascade involving tumour necrosis factor receptor-associated factor 6 (TRAF6) and mitogen-activated protein kinases (MAPKs) [58, 59]. This cascade culminates in the expression of many NF-κB-inducible genes, including CCL2, IL1B, IL12, IL18 and TNF, causing natural killer (NK) and T cells to release IFN-γ and TNF-α, which ultimately results in granuloma formation .
In the present study, several TLR-mediated proinflammatory cytokines and signalling molecules were differentially expressed in the BTB group compared to the non-infected control animals. These included CCL2 (increased), CXCR4 (increased), CXCL5 (increased), IL1A (increased), IL8 (increased), IL18 (decreased), IRAK4 (decreased), MAPK6 (increased), MAPK13 (decreased), MAPK14 (decreased) and MYD88 (decreased). This was also supported by canonical pathway analysis using IPA, which identified TLR signalling as a molecular pathway affected by M. bovis infection.
These results suggest that genes encoding TLR-mediated signalling pathway molecules have a role in governing the host response to BTB and may also serve as targets for immuno-subversion by M. bovis. For example, genes encoding several innate immune receptors and chemokines (such as TLR4, CD83, CCL2, CXCR4, CXCL5, IL1A and IL8)—several of which participate in the initiation of a T cell response during infection [61–64]—showed increased relative expression in the BTB animal group. Transcriptional profiles suggesting initiation of a T cell response are supported by the comparative analysis of the PBL cell populations in the two animal groups; a significant increase in the mean number of lymphocytes and a significant decrease in the mean number of monocytes were observed in the BTB group relative to the control animals. This difference in the PBL cell composition may represent recruitment of host cytotoxic lymphocytes for the destruction of infected monocytes in the control of M. bovis infection [3, 65].
It is important to note, however, that the observed decreased expression of host PRR genes (such as TLR2) and the genes encoding their associated adaptor and signalling pathway molecules (such as MYD88, IRAK4, MAPK13 and MAPK14) may indicate that the adaptive response in BTB animals is inferior due to the repression of these innate immune genes. Indeed, previous work has proposed that mycobacterial antigens, such as the early secreted antigenic target protein 6 (ESAT-6) protein, attenuates the host innate immune response by inhibiting MYD88-IRAK4 binding, thus causing suppression of NF-κB-induced transcription of upstream genes required for T cell response initiation . These workers also demonstrated that activation of v-akt murine thymoma viral oncogene homolog kinases (AKTs) is necessary to prevent MYD88-IRAK4 complex formation. Notably, the AKT2 gene displayed increased relative expression (+1.22-fold) in the BTB animal group in the present study.
Repression of host innate immune genes that elicit an adaptive response to M. bovis infection is further supported by the analysis of genes belonging to the interferon signalling pathway, which has been shown to have a role in human tuberculosis [37, 67–71]. IFN-γ is secreted by NK cells and CD4+ T cells upon activation by IL-12 produced by infected macrophages. IFN-γ recruits additional macrophages to the site of infection while also providing the stimulus for activating microbicidal functions in infected macrophages [14, 71–73]. IFN-γ also induces MHC class II gene expression in infected macrophages by signalling through its receptor (IFN-γ-receptor) [74–76]. This stimulates the JAK-STAT pathway, resulting in induction of transcriptional activators of MHC class genes, such as the MHC class II transactivator gene (CIITA). Mycobacterial antigen presentation via MHC class II molecules is critical for the recruitment of additional CD4+ T cells and the formation and maintenance of granulomas .
The results from the current study support a role for interferon signalling pathways during M. bovis infection. The genes encoding interferon (alpha, beta and omega) receptor 2 (IFNAR2); interferon gamma receptor 2 (IFNGR2); interferon-induced protein with tetratricopeptide repeats 2 (IFIT2); interferon-induced protein with tetratricopeptide repeats 5 (IFIT5); interferon-induced transmembrane protein 3 (IFITM3); protein tyrosine phosphatase, non-receptor type 2 (PTPN2); and signal transducer and activator of transcription 2 (STAT2) displayed differential expression in the BTB animals based on the microarray and/or real time qRT-PCR analyses.
These findings suggest that, in addition to the targeting of TLR-mediated signalling pathways, M. bovis may also target genes involved in the IFN-signalling pathway, resulting in an attenuated T cell response that enables mycobacterial survival and disease progression. It is tempting to speculate that suppression of IFN-signalling in response to M. bovis infection may result in the impairment of the antigen presenting process required for adaptive immunity; however, further work is required to investigate this possibility. Notably, the gene encoding IFN-γ (IFNG) was not differentially expressed in the current study, despite the BTB animals having tested positive for increased IFN-γ based on the BoviGAM® assay. However, it is important to note that unlike the blood samples used for the BoviGAM® assay in the current work, the PBL fraction from which the RNA was derived in this study was not stimulated with protein purified derivative of tuberculin (PPD), which is required to elicit IFN-γ secretion . In addition, contrary to previous results obtained by Meade and colleagues  we did not detect differential expression of the TNF gene between the M. bovis-infected and control animals examined here. The most likely explanation for this apparent discrepancy is the different cell sample types used for gene expression analyses (PBL versus PBMC).
IPA canonical pathway analysis identified a number of DE genes which, to our knowledge, have not previously been reported to be involved in the host response to tuberculosis in cattle or other mammalian species. These included CTLA4 and TLR3. TLR3 encodes an intracellular PRR involved in the recognition of viral-derived nucleic acids . In the present study, reduced relative TLR3 expression in BTB animals (-1.55-fold) may suggest some hitherto unknown role for this PRR in intracellular mycobacterial infection. In support, we have observed significant differential expression of TLR3 in bovine monocyte-derived macrophages (MDM) stimulated in vitro with M. bovis when compared to non-stimulated control MDM (unpublished data). CTLA4 encodes an inhibitor of the T cell-mediated response [78, 79] and this gene displayed increased relative expression (+3.20-fold, P ≤ 0.05) in the M. bovis infected animals in the present study. The observed increased relative expression of CTLA4 may reflect a mechanism of immuno-modulation used by M. bovis to subvert a host T cell response.
Finally, hierarchical clustering analysis was performed here using a total of 5,388 genes that passed the informative probe filtering criteria. This analysis unambiguously differentiated animals on the basis of their disease status. This result suggests that genome-wide expression profiling of PBL from BTB animals can be used to enable the identification of suitable transcriptional markers for the detection of infected animals within herds and augment current surveillance strategies in countries where control programmes have been implemented [21, 22]. However, further work using PBL samples from additional animals infected with M. bovis and other microbial pathogens will be required to identify and validate robust M. bovis-specific transcriptional signatures of infection.