Overall view of KEGG pathways.

The overall view of the impact of DEG during lactation on KEGG pathway categories is shown in Figure 2. The pathways most impacted during lactation were related to the category ‘Metabolism’ (in particular the subcategory ‘Carbohydrate Metabolism’, ‘Lipid Metabolism’, ‘Amino Acid Metabolism’, ‘Glycan Metabolism’, and secondary metabolites, such as ‘Metabolism of Other Amino Acids’ and ‘Biosynthesis of Other Secondary Metabolites’, all of them strongly induced), ‘Environmental Information Processing’ (evidently induced), and ‘Organismal System’ (with ‘Immune System” and ‘Endocrine System’ induced and ‘Excretory System’ and ‘Sensory System’ inhibited during lactation). Interestingly, the ‘Metabolism’ category was induced overall at the onset of lactation (i.e., 1 vs. −15 d) and then inhibited at 240 vs. 120 d, when milk yield decreased (Figure 2 and Figure S1). The subcategories of pathways related to ‘Human Diseases’ probably have little to do with bovine mammary and will not be discussed in detail; however, the results indicated that these pathways were strongly impacted overall during lactation and inhibited as lactation progressed. The most inhibited subcategory was the ‘Immune System Diseases’ (Figure 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Overall view of the DIA results of the KEGG pathways with main categories and subcategories as provided by KEGG (http://www.genome.jp/kegg/pathway.html).

Reported are all times compared with −30 d, and the two consecutive times (1 vs. −15 d and 240 vs. 120 d) with the largest number of differentially expressed genes and overall impact on pathways. Blue horizontal bars denote the impact and the square on their right denote the direction of the impact (red = activation; green = inhibition). See companion paper for interpretation of the DIA output.

https://doi.org/10.1371/journal.pone.0033268.g002

In summary, the examination of KEGG pathway categories and subcategories indicated that during lactation the mammary gland increased overall metabolism, e.g. large increase in carbohydrate, lipid, and metabolism of other secondary molecules (e.g., glycans). Those events were coupled with a large increase in membrane transport and signaling and a strong participation of the immune and endocrine system. It was noted that pathways in the ‘Genetic information processing’ subcategory were slightly inhibited overall as lactation progressed. That was mostly due to an evident inhibition of ‘Translation’ and ‘Replication and Repair’ subcategories (Figure 2).

Most impacted KEGG pathways.

The detail of each KEGG pathway is reported in Additional file S1 (sheet ‘KEGG pathway’). The top impacted pathways as calculated and sorted by the mean impact from 15 to 120 vs. −30 d (with exclusion of ‘Human Diseases’ category) were ‘Caffeine metabolism’, ‘Galactose metabolism’ (i.e., lactose synthesis), ‘Glycosylphosphatidylinositol(GPI)-anchor biosynthesis’, and ‘PPAR signaling’. Some of those pathways were discussed in the companion paper. Among the top impacted pathways not discussed in the companion paper, the ‘Glycosylphosphatidylinositol (GPI)-anchor biosynthesis’, the “Sterol biosynthesis”, and the “ABC transporters” are novel and of biological interest.

The ‘Glycosylphosphatidylinositol (GPI)-anchor biosynthesis’ was the most impacted among the glycan biosynthesis-related pathways (Figure 2; details in Additional file S1). The pathway is involved in synthesis of GPI which anchors proteins in the membrane of the ER for transport to the Golgi and apical membranes [10]. The importance of the GPI-anchored proteins has not been evaluated in bovine mammary gland and although the concentration of GPI anchored proteins is high in human and pig milk, it is apparently undetectable in bovine milk [11]. GPI-anchors are pivotal for cell survival and affect many functions such as protein sorting in ER-Golgi trafficking, targeting of GPI-anchored proteins in polarized epithelial cells for apical export, cell-to-cell adhesion, signal transduction associated with cholesterol and sphingolipids in membrane rafts, and sorting in endocytic pathways [12]. Except for transport of folic acid through the internalization of its receptor, which is a known GPI-anchored protein, to date there are no known roles of GPI-anchored proteins in milk synthesis [12]. Thus, this finding is novel and more research on GPI-anchored proteins seems warranted.

The DIA analysis of the ‘Steroid biosynthesis’ indicated an overall induction of this pathway (Additional file S1). In addition, the visualization of the pathway (Additional file S2) indicated an overall increase of cholesterol synthesis. Even though it has been demonstrated that the major part of cholesterol present in bovine milk originates from hepatic synthesis [13], and that the synthesis of cholesterol in mammary has been poorly studied in bovine, it has been shown that de novo cholesterol synthesis increases after parturition in rat mammary tissue [14]. Linked with cholesterol synthesis also is the ‘ABC transporters’ which was induced overall during lactation underscoring the importance of those transporters in lactating mammary gland (Additional file S1). Several ABC transporters seem to be involved in cholesterol transport and some are down-regulated during lactation in bovine mammary tissue [8], [15]. Others with still unknown function are highly abundant in bovine mammary tissue during lactation (e.g., ABCG2, [8]; see also companion paper).

Overall interpretation of KEGG pathways: Metabolism.

The pathways in the cell are highly interconnected and, for metabolic pathways, quite often the product of one pathway becomes the substrate for another. Here we attempt to provide an interpretative summary of the metabolism in mammary tissue during lactation based upon the results from the DIA analysis (see ‘KEGG Pathways’ sheet in Additional file S1).

The overall metabolism increased during lactation, as shown by the ‘Metabolism’ category (Figure 2). The overall visualization of the data (Additional file S1 and S2) suggested a channeling of glucose towards increasing the synthesis of lactose and decreasing use for energy (i.e., decrease of glycolysis and pyruvate metabolism). The DIA suggested that the ‘Lipid Metabolism’ subcategory was characterized by a large increase in synthesis of triacylglycerol (TAG) (through activation of ‘Biosynthesis of unsaturated fatty acids’, ‘Ether lipid metabolism’, ‘Glycerolipid metabolism’, and ‘Glycerophospholipid metabolism’ pathways, Additional file S1), synthesis of phospholipids (‘Glycerophospholipid metabolism’), cholesterol (‘Steroid biosynthesis’), leukotrienes (through ‘Arachidonic acid metabolism’, see Additional file S2), and sphingolipids (‘Sphingolipid metabolism’) and a reduction of ‘Lipid metabolism’ (i.e., lipid catabolism), ‘Steroid hormone biosynthesis’, and ‘Fatty acid elongation in mitochondria’ (Additional file S1). The suggested reduction of fatty acid metabolism together with an increase in synthesis of TAG indicated a preferential channeling of fatty acids taken up by the mammary tissue [16] towards synthesis of milk fat, including the components of cellular membranes.

Results also indicated an increase in catabolism of several amino acids (AA) (i.e., induction through lactation of ‘Arginine and proline metabolism’, ‘Histidine metabolism’, ‘Lysine degradation’, ‘Tryptophan metabolism’, and ‘Valine, leucine and isoleucine degradation’, Additional file S1). Many of the products of those pathways could be precursors of the TCA cycle pathway (Additional file S2) that the DIA suggested was strongly activated during lactation (Additional file S1). Considering that our data indicated a reduction of flux in the relative proportion of glucose and fatty acids for energy utilization and an increase in the proportion of AA import (see below and [7]) and degradation to potentially provide precursors for the TCA cycle, it appears that the AA are an important pool of energy in mammary tissue during lactation, particularly the class II essential AA [7] (i.e., Arg, Ile, Leu, Lys, and Val). These findings are supported by experiments conducted more than 30 years ago [17], [18]. The ‘Cysteine and methionine metabolism’ pathway was inhibited during lactation and induced at 240 vs. 120 d (Additional file S1, ‘KEGG Pathways’ sheet). This result suggests that the mammary gland spared Met for protein synthesis, which is supported by Met being one of the AA with lowest concentrations in blood but one with the highest extraction ratios by the mammary gland [19]. The synthesis of Lys, Phe, Tyr, and Thr was not impacted in any comparison.

The KEGG category ‘Metabolism of Other Amino Acids’ appeared to be relatively impacted and generally induced particularly during mid-lactation. Among those were ‘Beta-Alanine metabolism’, ‘Cyanoamino acid metabolism’, ‘Selenoamino acid metabolism’, ‘Glutathione metabolism’ and ‘Taurine and hypotaurine metabolism’ (Additional file S1 and S2). The large impact and induction of ‘Glutathione metabolism’ during lactation appears to confirm previous data [17], [20], and support a pivotal role of this process in AA availability to the mammary gland. The importance of the other pathways in milk synthesis needs to be addressed with additional investigations.

The DIA uncovered a large increase of TCA cycle activity and generation of energy in general, which is supported by activation of the ‘Oxidative phosphorylation’ and ‘Pentose and glucoronate interconversions’ pathways during lactation (Additional file S1). Changes in pathway related to AA suggested a likely increase in TCA intermediates, hence, TCA activity from AA degradation. Contrary to previous reports [21], [22], our data indicated that the ‘Pentose phosphate pathway’ was not highly impacted and not greatly activated during lactation (Additional file S1 and S2).

The importance of ‘Glycan Biosynthesis and Metabolism’ KEGG subcategory in mammary during lactation was evidenced in DIA by a large impact and induction of ‘GPI-anchor biosynthesis’ as discussed above, but also by induction of several other pathways involved in glycosaminoglycan biosynthesis, including ‘Chondroitin sulfate’, ‘Glycosphingolipid biosynthesis’ pathways, particularly ganglio- series, and ‘N-Glycan biosynthesis’ (see Additional file S1 for details). The importance of those pathways seemed to increase as lactation progressed until 120 vs. −30 d. The components synthesized by those pathways are essential for the structure of the extracellular matrix and the complex interactions which regulate cell-to-cell signaling, coordination of stromal-epithelial development, and the interaction with extracellular signaling molecules [23]. Although their quantity in bovine milk is low, the process of glycosphingolipid synthesis is of particular interest because of the current commercial focus in the production of these bioactive molecules due to their probiotic role in the human intestine [24]. In addition, gangliosides have an important role in membrane function such as modulating enzyme properties, cell signaling, cell adhesion, protein sorting, and formation of caveolae [25]. The globosides are present in bovine milk but at low concentrations [26], which our data appear to support (Additional file S1). A recent characterization of bovine milk oligosaccharides in colostrum and milk revealed a large decrease of their concentration during the transition from colostrum to mature milk as well as during the progression of lactation [27]. Our data, similar to recent transcriptomic data [28], suggested a potential overall increase in glycosphingolipids during lactation; however, the amount of those components decreases as lactation progresses [27].

Among the KEGG subcategory ‘Metabolism of Cofactors and Vitamins’ several pathways appeared to be impacted and largely induced during lactation, with ‘Caffeine metabolism’ being the most impacted among all KEGG pathways (Additional file S1). Most of those pathways were composed of only one gene that passed the selected thresholds (FDR and post-hoc P-value≤0.001 and ≥40% of the pathway covered by the microarray) rendering the interpretation of the data weak. The discussion of the ‘Caffeine metabolism’ was provided in the companion paper.

Overall interpretation of pathways: Genetic Information Processing.

The KEGG category ‘Genetic Information Processing’ was not highly impacted overall during lactation except at mid-to-late lactation when the impact increased and an overall inhibition was apparent as lactation progressed (Figure 2). Among the pathways in this category the only ones induced during lactation were ‘RNA polymerase’, ‘Aminoacyl-tRNA biosynthesis’ (particularly during the first 60 d post-partum), ‘Protein export’, and ‘Protein processing in endoplasmic reticulum’ (Additional file S1 and S2).The KEGG pathways ‘Ribosome’, ‘Ribosome biogenesis in eukaryotes’, ‘RNA transport’, ‘RNA degradation’, ‘Sulfur relay system’ (particularly the tRNA thiolation for which a function is still unknown in eukaryotes [29]) and all the pathways involved in nucleotide replication and repair subcategory were clearly inhibited during lactation (Additional file S1). The pattern of ‘Ribosome’ had a strong induction and impact at −15 vs. −30 d (i.e., 2 weeks prior parturition), a minimal impact at the beginning of lactation and a consistent increase in impact and inhibition as lactation progressed and ended relative to −30 d (Additional file S1). This latter observation, while consistent with observations in mouse mammary tissue during lactation [6], does not seem to support the apparent increase in protein synthesis inferred by transcriptome analysis suggested by Finucane et al. [9]. The above observations together with the slightly greater number of down-regulated genes (Figure 1) and the significant increase in total amount of RNA during lactation [30], likely due to the large increase in expression of few genes (e.g. caseins and other milk proteins), appear to support the existence of “translational competition” within the mammary gland during lactation [7].

Overall, the data indicated that during lactation the mammary tissue increased processing and export of proteins from the ER and, despite the decrease of the protein synthesis machinery, it also increased AA charging to tRNA and the export of proteins. Those observations appear to be supported by the overall increase in milk protein synthesis and secretion by the mammary gland [31], [32]. As discussed below, the data also indicated a decrease in overall cell cycle (e.g., inhibition of ‘DNA replication’, see Additional file S1), which supports previous observations [9].

Overall interpretation of pathways: Environmental Information Processing.

This category of KEGG pathways was highly impacted and activated overall during lactation (Figure 2), suggesting an general increase in intra- and extra-cellular signaling during lactation. Among the pathways in this category the most impacted and induced during lactation were ‘Hedgehog signaling’, ‘Jak-STAT signaling’, and ‘TGF-beta signaling’; whereas, ‘Cell adhesion molecules (CAMs)’, ‘mTOR signaling pathway’, and ‘Notch signaling pathway’ were inhibited (Additional file S1 and S2).

The Hedgehog and TGF-beta signaling pathways (together with ‘Wnt signaling’) appear to have a role in mammary stem cell self-renewal and, when deregulated, mammary cancer [33]. The Hedgehog signaling pathway has been previously reported to be induced during lactation in rodents and with an essential, albeit unknown, role in lactation [34]. Our data supported those previous findings. The importance of TGF-beta in mammary gland is still unclear but a consistent increase in its concentration in bovine milk has been demonstrated with a likely beneficial biological effect on human intestinal cells [35]. In addition, the TGF-beta pathway appears to have a negative role on mammary cell proliferation and a positive role on apoptosis [36]. The TGF-beta signaling through Smad is known to play a pivotal role in the cell cycle [37].

The Jak-STAT signaling is essential for the induction of milk protein expression in mammary tissue of non-ruminants, but in ruminants and particularly in the bovine, expression of milk proteins appears to require a basic activity of this pathway but is not directly modulated by the change in expression of Jak and/or Stat genes, rather, it seems to be modulated by down-stream effectors of the pathway (e.g., ELF5) [7].

The inhibition of Notch signaling pathway was a novel finding. This pathway is known to be involved in cell development and physical organization of cell into tissues, cell proliferation and apoptosis [38] and in mammary cell lines it has been observed that artificial activation of Notch signaling inhibit mammary branching and lactation [39].

The cell adhesion molecules are important in maintenance of structural architecture of the mammary epithelia [40]. Thus, the apparent inhibition of the ‘Cell adhesion molecules’ KEGG pathway is not clear; however, when the pathway was evaluated in larger details (Additional file S2) it became clear that the inhibition was mostly caused by adhesion molecules related to the immune system, but there was a clear induction of the tight junction molecules (see below).

The reduction of ‘mTOR signaling pathway’ was unexpected given that mTOR seems to play a role in milk protein synthesis. Previous data from our laboratory and from others ([7] and references therein) clearly indicated that mTOR plays an important role in milk protein synthesis; however, at the transcriptomics level the regulation was evident only for few of the mTOR pathway-related genes [7]. The present data appear to support our previous conclusion [7] that the process of translation is basically inhibited (via inhibition of mTOR signaling) and the inhibition is overridden by insulin signaling. In the present analysis insulin signaling was overall slightly inhibited but highly-impacted by lactation (Additional file S1). In the previous analysis [7] we measured few genes involved in insulin signaling and the data indicated enhanced sensitivity of the mammary gland to insulin via the increases in expression of the receptor and few down-stream signaling molecules. Most of those molecules were also found to be up-regulated with the microarray analysis and the overall inhibition was due, for the most part, to a decrease in expression of genes involved in non-mTOR related pathways (Additional file S2).

Overall interpretation of pathways: Cellular Processes.

This category of pathways also was highly activated during lactation, particularly between 60 and 120 d compared to −30 d, with an overall induction of subcategory ‘Transport and Catabolism’ and ‘Cell Communication’ and an general inhibition of ‘Cell Growth and Death’ (Figure 2). The most impacted and activated pathways in this category during lactation were ‘Peroxisome’ and ‘Tight junctions’; ‘Phagosome’, ‘p53 signaling’, and ‘Gap junctions” appeared to be inhibited during the late lactation, while ‘Cell cycle’ was clearly inhibited during the whole lactation (Additional file S1).

Peroxisomes have been reported to be present in human mammary tissue and their activity is decreased in breast cancer [41]; however, we are unaware of published studies on peroxisomes in bovine mammary tissue. Our data indicated an overall induction of peroxisome activity and proliferation during lactation; particularly induced were peroxisome biogenesis, the prime metabolism (⇑ XDH), and the antioxidant activity (⇑ SOD), but an evident decrease of fatty acid oxidation was suggested as well (Additional file S2).

The importance of junctions in mammary tissue integrity has been understood for some time [42]. All the junctions appeared to have been induced overall during lactation (Additional file S1 and S2), but particularly the ‘Tight junctions’ and the ‘Adherens junctions’ were induced. Among all junctions, the tight junction has been known to be the most important during lactation, when it blocks the paracellular route allowing the highly-controlled secretion of milk components by the epithelial cells [43].

The inhibition of ‘Cell cycle’ coupled with the low impact of ‘Apoptosis’ (Additional file S1 and S2) indicated a reduction of proliferation but also a stasis from the point of view of number of cells in the mammary gland during lactation, which is in accordance with previous bovine mammary transcriptomics data in early lactation [9]. Measurements of bovine mammary parenchymal tissue apoptosis and proliferation at both the end of pregnancy and lactation have been performed [44], [45]. Data from those work indicated that apoptosis was higher at 48 days prior to parturition, decreased 2 weeks prior to parturition, increased 2 weeks after parturition where it reached a peak, and decreased thereafter with a numerical increase at mid-to-late lactation (90–120 d) [44], [45]. Those work also uncovered that proliferation of mammary cells decreases significantly from 2 weeks prior to parturition and remains low until late lactation, with a slight, albeit significant increase from 2 weeks postpartum to mid-lactation [45]. Our data appeared to be in accord with those previous reports (Additional file S1), indicating a stabilization of cell numbers at parturition and during most of lactation. The DIA results indicated that both apoptosis and proliferation resumed slightly between 120 and 240 d (Additional file S1), suggesting a decrease of cell number when milk yield declined, in accordance with previous data [46].

Overall interpretation of pathways: Organismal System.

The ‘Organismal System’ category of KEGG pathways also was highly impacted (Figure 2). An apparent large impact and induced activity was observed for “Endocrine System’, ‘Immune System’, ‘Nervous System’, and ‘Circulatory System’ subcategories and an important impact and inhibition was observed for ‘Excretory System’ and ‘Sensory System’ subcategories (Figure 2). Among the pathways included in those subcategories the ‘Antigen processing and presentation’, the ‘Proximal tubule bicarbonate reclamation’, and the ‘PPAR signaling pathway’ were the most impacted. The former were clearly inhibited, while ‘PPAR signaling’ was clearly activated (Additional file S1).

In the ‘Immune System’ subcategory during lactation besides an activation and large impact of ‘Complement and coagulation cascades’ and ‘Hematopoietic cell lineage’ and inhibition of the ‘Antigen processing and presentation’, all the other immune-related pathways were modestly impacted and activated overall. This was most evident in mid-to-late lactation (i.e., after 60 d; see Additional file S1). For the ‘Antigen processing and presentation’ the major histocompatibility complex (MHC) class I was particularly inhibited during the whole lactation (Additional file S1 and S2). From the beginning of lactation to 120 d compared to −30 d, the overall analysis indicated a consistent increase in impact and induction of pathways related to the innate immune system (e.g., ‘Toll-like receptor signaling’, ‘NOD-like receptor signaling’, ‘RIG-I-like receptor signaling’, and ‘Cytosolic DNA-sensing’ pathways). Also pathways related to immune cells (e.g., ‘B cell receptor signaling’, ‘Leukocyte transendothelial migration’, ‘Natural killer cell mediated cytotoxicity’, and ‘T cell receptor signaling’ pathways) were induced and impacted to the same degree during lactation (Additional file S1). Overall, the data indicated a consistent increase of amount/activity of immune cells in mammary tissue until the decline of milk synthesis; however, there was a concomitant large decrease of MHC indicating the absence of hyper-activation of the immune system.

Among the pathways composing the ‘Endocrine System’, the ‘PPAR signaling’ and ‘GnRH signaling’ pathways were among the most impacted and induced during lactation. In contrast, ‘Adipocytokine signaling’ was moderately impacted and mostly inhibited during lactation and ‘Insulin signaling’ appeared slightly inhibited overall during lactation (Additional file S1).

The apparent induction of ‘GnRH signaling’ is novel. The gonadotropin-releasing hormone receptor (GnRHR) is typically expressed in the anterior pituitary to regulate, under stimulus of the GnRH, the production and release of the gonadotropins, luteinizing and follicle-stimulating hormones. The GnRHR, however, has also been found in extrapituitary tissues including mammary gland [47]. The function of GnRHR in extrapituitary tissues is not entirely clear but it appears to play a role in reducing cell proliferation in breast cancer cells [47]. Interestingly, our data indicated that the expression of genes coding for the beta polypeptide of both luteinizing and follicle-stimulating hormones (LHB and FSHB) was significantly ⇑ during lactation (see companion paper and Additional file S2). An increase in expression of genes coding for those hormones in mammary tissue during lactation also was reported in microarray data from mouse [4], [5]. This finding is novel and, to our knowledge, no extrapituitary protein expression of those genes has been reported previously in bovine mammary tissue. Many hormones have been detected in milk but not LH or FSH [48]. We have not verified such expression using qPCR or Western blot; thus, further investigation is required to verify these findings.

Due to specific functions that are not pertinent to the mammary, the pathways in others subcategories under ‘Organismal System’ are probably not applicable in toto to mammary gland (e.g., ‘Digestive System’, ‘Excretory System’); however, a similar function can be envisaged in mammary tissue for several DEG. In particular, a few observations can be made by using Figure 2 and the Additional file S1 and S2. A few pathways among those subcategories were highly impacted and evidently induced by the DEG, including ‘Bile secretion’, ‘Fat digestion and absorption’, and ‘Collecting duct acid secretion’. Others were highly impacted but evidently inhibited during lactation such as ‘Endocrine and other factor-regulated calcium reabsorption’ and ‘Proximal tubule bicarbonate reclamation’. All of those pathways had in common a large number of genes coding for proteins involved in ion transport, particularly the ATP pumps for the exchange of Na/K and Ca and Cl export; in addition, ‘Bile secretion’ was composed of genes associated with steroid transport. The overall data analysis indicated a large degree of transport of Na, K, Ca, and Cl. It is difficult to ascertain if those genes code for proteins involved in export or import of those ions inside the epithelial cells (see below for additional results and discussion).

The subcategory ‘Nervous System’ was clearly impacted and induced during lactation (Figure 2). Overall, the data indicated that the nervous system was highly active during lactation. The importance of the nervous system in the mammary gland is well-known mainly for its role in milk ejection [49] but also for its likely control of blood flow [50], both of which are essential for lactation.

Novel findings among most impacted Gene Ontology terms.

The detailed results of the GO analysis are reported in Additional file S1 (“GO Biological process”, “GO Cellular component”, and “GO Molecular function” sheets). We have summarized the data using REVIGO [51] for the interpretation of the main impacted GO terms during lactation (see also Additional file S4). The TreeMap summarizing the direction of the impact in the most impacted categories of terms for the GO-BP are reported in Figure 3 (activated) and Figure 4 (inhibited). Treemap view of the impact for the GO-BP is available in Figure S3 and scatterplot view of the same results is reported in Figure S4 and Figure S5. Results from REVIGO for the GO-CC are shown in Figure S6 and Figure S7 and for GO-MF are reported in Figure S8 and Figure S9 and discussed in Additional file S3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. TreeMap visualization obtained by REVIGO analysis (see Materials and Methods for details) of the summary of GO Biological process deemed by DIA to be overall activated during full lactation (from 15 to 120 d) compared to −30 d.

Similar colors denote semantic similarity and dimension of the area is proportional to the overall direction of the impact, as calculated by the DIA. Not all the terms are reported due to space constraints (see Additional file S4 for details).

https://doi.org/10.1371/journal.pone.0033268.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. TreeMap visualization obtained by REVIGO analysis (see Materials and Methods for details) of the summary of GO Biological process deemed by DIA to be overall inhibited during full lactation (from 15 to 120 d) compared to −30 d.

Similar colors denote semantic similarity and dimension of the area is proportional to the overall direction of the impact, as calculated by the DIA. Not all the terms are reported due to space constraints (see Additional file S4 for details).

https://doi.org/10.1371/journal.pone.0033268.g004

Among novel findings in GO we observed an induction of several terms related to regulation of inflammation (e.g., ‘Negative regulation of inflammatory response to antigenic stimulus’), transport of several components including amino acids (in particular aspartate and glutamate), phosphate, copper, long-chain fatty acids (LCFA), sodium, and calcium (particularly exocytosis of calcium); proliferation and angiogenesis (e.g., ‘Regulation of mammary gland epithelial cell proliferation’, ‘Angiogenesis involved in wound healing’), Golgi and endoplasmic reticulum (e.g., ‘Golgi organization’, ‘Rough endoplasmic reticulum’), oxidoreductase activity, telomeric region, and epigenetic regulation (i.e,. ‘PgC protein complex’, involved in epigenetic repression of gene expression [52]).Among the novel and most relevant impacted and inhibited terms were several related to immune function (e.g., ‘Positive regulation of leukocyte chemotaxis and monocyte activation’, MHC I), tissue remodeling (including neurons, e.g., ‘Regulation of epidermis differentiation’, ‘Regulation of glial cell differentiation’), tissue sensory capacity including sensing bacteria (e.g., ‘Response to mechanical stimulus’, ‘Detection of bacteria’), chromatin components (e.g., ‘Euchromatin’, ‘Chromocenter’) particularly related to epigenetic modification (e.g., ‘Histone deacetylase complex’), transports of several components such as chloride, arginine, and ammonia, and modification and metabolism of several RNA types (e.g., ‘tRNA modification’, ‘snRNA metabolism’).

Most of the results from GO analysis confirmed previous results from KEGG analysis (see above and Additional file S3) but also provided new insights. For instance, the increased regulation of proliferation suggested that the previously reported decrease in mammary proliferation as lactation begins [45] appeared to be actively regulated at the transcriptional level. The increase in angiogenesis, which also included an increase in ‘Blood vessel endothelial cell migration’ (Additional file S1, ‘GO Biological process’ sheet), suggested an important role of angiogenesis during lactation in bovine mammary gland with a continuous formation of blood vessels. The blood flow across the mammary gland increases dramatically at the onset of lactation [53] with a consistent increase during the first 3 months of lactation, particularly for multiparous cows [54]. It has been well-established that there is a positive correlation between blood flow and milk yield [53]; however, this appears to be true only for cows with less than 3 lactations [54].

The DIA results of our microarray data clearly indicated an inhibition of the immune activation in the mammary gland. This finding, also supported by the results from the DIA analysis of KEGG pathways (see above), is quite striking. In modern high-producing cows, the importance of the immune response in mammary gland is pivotal due to the need for preventing/fighting pathogen-causing mastitis [55]. Our data highlighted several unique aspects of mammary tissue:

• The large inhibition of immune-related functions, particularly MHC class I, together with the predominance of epithelial cells in our core biopsies (see companion paper) indicated that the mammary gland is actively involved in the immune system and supports the suggestion of the mammary gland being an evolutionary product of the innate immune system [56];
• The mammary gland during lactation has an apparent increase in activity/preparation of the immune system, particularly the innate immune system (see above and Additional file S1, sheet ‘KEGG pathway’), but there was a decrease in chemotaxis and activity of immune cells. The infiltration of immune cells, particularly macrophages, into mammary tissue is essential for the initiation and resolution of an inflammatory response [57]. Our data suggested increased accumulation of macrophages or immune cells in general as suggested by the ‘Hematopoietic cell lineage’ (Additional file S1 and S2) and specific immune cell markers (Figure S10). Those data were indicative of a more pronounced increase in macrophage infiltration between 1 through 60 d when it reached a peak, followed by a subsequent decrease. The increase in number of macrophages in mammary tissue during the surge in milk production was similar to observations reported for the rat [58]. In contrast, neutrophil infiltration seemed to increase through the end of lactation (Figure S10);
• Another striking finding was the consistent decrease of the major histocompatibility complex (MHC), which is similar to microarray data of mouse mammary gland during lactation [4] despite the fact that the MHC class II in mouse is not expressed in epithelial cells [59] as it is in bovine [60]. The MHC class I is present in all nucleated cells and it is the mechanism whereby all cells participate in the immune defense, presenting antigens from bacteria and virus to CD8+ T-lymphocytes [61]. An association between Class I alleles and mastitis traits has been reported [62]. The mRNA expression of bovine mammary MHC components was increased due to intramammary challenge with S. Uberis strain O140J [63], probably due to the epithelial response, as previously demonstrated in vitro [60].

Overall, those data confirmed the lack of mastitis in the mammary tissue used in the present investigation. The reason for the decrease of the MHC pathway, particularly class I, is not readily apparent because the mammary gland is an easily accessible site of bacterial entry which in turn requires a rapid immune response. Under such scenario we propose two hypotheses (see also [64]): 1) genetic selection of cattle for high milk production, which has brought about an increase in mastitis incidence [65], might have inadvertently reduced the expression of the MHC during lactation resulting in lower sensitivity to bacteria and a consequent decrease in energy expenditure for immune responses; 2) the MHC is a vesicle-dependent process which uses ER-Golgi networks [61] as do milk components; thus, the MHC can be considered a competitor of the vesicle-transport system. The MHC inhibition can be considered a way for the mammocytes to spare resources.

The GO results indicated an inhibition of the active components of the epigenetic modification and increase of long term epigenetic stabilizer (e.g., PcG protein complex, [52]). Those data highlighted the seemingly crucial importance of epigenetic regulation in bovine mammary, thus, support recent findings [66]. In particular, the data suggested that bovine mammary once the euchromatic status is established tends to inhibit any epigenetic modification as a means to maintain a consistent transcriptome until milk synthesis declines. This idea is supported by the number of DEG reported in Figure 1, where very few DEG were observed in the comparison between consecutive time points from 15 to 120 d postpartum, but a large number of DEG was observed when milk yield declined (between 120 and 240 d).