In this study, we analyzed the global gene expression changes in the bovine mammary gland that occurred during the acute phase and chronic stage of experimental E. coli mastitis in dairy cows in early lactation. Tissue samples were collected with a biopsy pistol, which made it possible to sample the same E. coli-infected cows in vivo at two different time periods.
The acute phase was verified by the presence of high E. coli CFU/ml in milk, fever, increased SCC and the presence of acute phase proteins MAA and SAA in milk and blood, respectively. Marked changes in gene expression were detected in the acute stage at 24 h p.i. In agreement with results from other related studies the up-regulated genes were related to the immune response, with the majority of genes involved in the induction and regulation of the local inflammatory response and the APR [9, 16]. Among the top 10 KEGG pathways associated with these genes were pathways involved in chemokine signalling, Toll-like receptor signalling, leukocyte transendothelial migration, cytokine-cytokine receptor interaction, cell adhesion and Fc gamma R- mediated phagocytosis. This finding indicates the importance of recruitment and activation of macrophages and neutrophils to sites of infection during acute E. coli mastitis [17, 18].
In the acute phase, the group of down-regulated genes in the mammary gland was mainly associated with the lipid metabolism and fatty acid pathways. It is well known that cows with E. coli mastitis experience decreased milk production [19, 20]; however, the affected molecular mechanisms related to fat synthesis in those cows has been undefined . The results presented in this paper, in conjunction with the results of Mitterhuemer et al. , clearly demonstrate that E. coli infection of the mammary gland reduces the expression of genes involved in fat metabolism. However, further research is needed to verify whether these genes actually influence the total amount of fat secretion in the milk or instead influence the fat percentage in the milk.
In the chronic, subclinical stage examined at 192 h p.i., all cows were found to have cleared the E. coli bacteria and re-established normal body temperatures. Yet, the increased SCC and increased concentrations of MAA in milk were maintained. At this stage, gene expression had returned to control levels, with the notable exception of the genes encoding for the acute phase protein haptoglobin (HP) and the chemokine CXCL5. Both genes were found to be up-regulated at the APR and chronic stage response. CXCL5 expression on the surface of eosinophils facilitates their ability to recruit and activate CXC receptor 2 (CXCR2)-bearing cells, such as neutrophils, to the site of inflammation. In this manner, CXCL5 can influence tissue remodelling , such as would be expected in recovering udder tissue. It is likely that more up-regulated genes associated with tissue modelling could have been identified if the samples had been collected earlier. However, as the biopsy procedure itself causes tissue damage and bleeding, there is a limit to how often a biopsy can be collected without influencing the gene expression results.
By comparing our study's results with those from two recent array studies using post-mortem tissues from E. coli infected cows [14, 15], we identified a particularly strong dataset of the statistically significant genes and genetic pathways present in the mammary tissue during the acute phase at 24 h. All three studies were conducted in the Holstein Friesian breed, but differed with regard to E. coli strain and dose, cows' lactation number and stage, and the sampling procedure. In our study, a low dose (20-40 CFU/quarter) model was established in primiparous dairy cows in early lactation, and tissue samples were collected in vivo using biopsies at time points 24 h and 192 h p.i. in the lobuli-alveoli area. Rinaldi et al.  used a medium dose (400 CFU/quarter) in a two-step inoculation model in two quarters, with a 12 h time-delay in multiparous cows in late lactation; these tissue samples were collected post-mortem from euthanized cows at 12 h and 24 h p.i. from four different portions of the udder. Mitterhuemer et al.  also used a medium dose model (500 CFU/quarter), but relied on mid-lactation cows and collected tissue samples post-mortem at 6 h and 24 h p.i. from a less well-defined area of the udder. Furthermore, the gene analysis platforms used were different between each of the studies. All of the factors mentioned above can influence disease kinetics and gene expression levels in the mammary gland at 24 h p.i. This fact may explain why each study found a unique set of differentially expressed genes, as shown in the non-overlapping areas of the Venn-diagram [Additional file 4: Supplemental Figure S1].
A large group of DE genes were common between the three studies and were determined to be unaffected by E. coli strain, dose, cow factors, sampling method and gene expression analysis ([Additional file 4: Supplemental Figure S1], [Additional file 5: Supplemental Table S4]). The gene sets obtained from the analysis of the common 248 DE genes were mainly related to the immune response and are associated with the response to E. coli mastitis [14, 15] (Table 3). However, in all three studies two of the 15 pathways were of interest because they are principally related to intracellular infections (e.g. viruses, protozoa and bacteria).
The first pathway, the natural killer cell-mediated cytotoxicity pathway (KEGG04650), ranked fifth on our list (Table 1) and third in the comparison between the E. coli gene expression studies (Table 3). The NK cells are innate lymphocytes, which function to eliminate all aberrant cells which have completely lost, or express insufficient amounts of, MHC class I molecules; virally-infected cells or tumorigenic cells are examples of such targets [23, 24]. Moreover, NK cells are major cytokine producers and participate in the onset and shaping of the adaptive immune response. So far, bovine NK cells have been investigated in relation to intracellular infections with protozoa and mycobacteria . The mechanisms underlying NK cells role(s) in the defence against extracellular bacteria, such as E. coli, remains a question of interest. NK cell lyses of damaged apoptotic and necrotic cells is one possibility. Furthermore, NK cells might be involved in a response to E. coli via the dendritic cells (DC) . The main role of the DC is processing antigens and presenting antigens to other cells, namely the antigen presenting cells (APC). In this manner, the NK cells may play a role in maintaining homeostasis of the immune response during bacterial infections by regulating the APC amount in the secondary lymphoid organs . In [Additional file 5: Supplemental Table S4], CD69, a DC-induced activation marker of NK, as well as IL2RA and IL2RG, which are part of the IL2 receptor complex on lymphocytes and NK cells, are found in the natural killer cell-mediated cytotoxicity pathway. This suggests a possible function of NK cells in the immune defence process directed against E. coli in the mammary gland. To our knowledge, NK and DC have not been identified in healthy or inflamed mammary gland tissues from dairy cows; and, it remains to be shown if and how they are recruited from the blood compartment or local lymph nodes to the mammary gland.
The second interesting pathway, the Rig-I-like receptor signalling pathway (KEGG04622), ranked sixth on our list. Cells of the innate immune system, such as macrophages and DC, express a number of pattern-recognition receptors (PRR) that specifically recognize unique pathogen-associated molecular patterns (PAMP). In this study, as well as in others, TLR2 and TLR4 are characterized as being significantly influenced by E. coli mastitis. Comparison of the three related studies also provided evidence for involvement of molecules functioning in the RIG-I-like receptor pathway [Additional file 5: Supplemental Table S4]. The gene-encoded products include receptors such as RIG-I, Mda5, and LGP2, which are cytoplasmatic proteins that recognize viral DNA and RNA . Following recognition of viral-associated molecular patterns, the RIG-I-like receptors stimulate the synthesis of multiple cytokines including the group of interferon type 1 cytokines and pro-inflammatory cytokines that influence protein synthesis, growth arrest and apoptosis. Furthermore, the RIG-I-like receptors enhance DC maturation, NK cell activation, antibody production and differentiation of cytotoxic T-lymphocytes, which in turn facilitate the adaptive immune response . Whether the RIG-I-like receptors are up-regulated directly from contact with E. coli components in the mammary gland or indirectly via cross-induction through other TLRs is yet unknown.
The exact role of the natural killer cell-mediated cytotoxicity pathway (KEGG04650) and the Rig-I-like receptor signalling pathway (KEGG04622) in the response to E. coli infection of the bovine udder is still unresolved. The fact that identical pathway-associated genes were similarly regulated during the APR in three independent studies indicates that further investigation of these pathways and their related cells will provide significant insights into the pathogenesis of E. coli mastitis. Furthermore, this work sets stage for additional work to determine whether patterns of gene expression associated with udder infections with Gram positive bacteria e.g. S. aureus and S. uberis differs from that of Gram negative infection and whether the timing of gene changes are similar.