Mammary cells take up LCFA from albumin-bound fatty acids (NEFA) and lipoproteins. VLDL or chylomicrons are anchored to mammary endothelium by lipoprotein lipase (LPL), which then hydrolyzes TAG in the lipoprotein core to release FA 22. LPL has higher activity in mammary 22 compared with other tissues, probably due to its high mRNA abundance. The observed up-regulation in LPL mRNA as early as the onset of milk synthesis was remarkable (Figure 2) because mouse mammary tissue had only a 2-fold increase in LPL transcript between pregnancy and lactation. In the mouse, the increase was accompanied by 2-fold up-regulation of enzymatic activity 2324. In contrast with murine, bovine mammary LPL expression pattern was remarkably similar to the lactation curve (Additional file 1, Figure S1), which might be indicative of an important role of this gene in maintenance of milk synthesis.

Recent evidence points at VLDL receptor (VLDLR) as an essential component of LPL activity 25. VLDLR expression was up-regulated throughout lactation, particularly in the first mo post-partum (Figure 2). The mRNA abundance of LPL and VLDLR accounted for ~10% and ~0.1% of total genes measured (Table 1). Despite these differences, data suggest an important role for VLDLR in concert with LPL 25 in milk fat synthesis during lactation. Mammary VLDLR could act on chylomicrons or intestinal VLDL, which contain apo-B48 26. In general, our data are in agreement with previous work reporting higher efficiency of mammary TAG uptake from lipoproteins at the beginning of lactation 26. The pattern of mammary tissue expression of LPL during lactation was in accordance with the typical increase in blood LDL in dairy cows post-partum, which is an indirect index of VLDL utilization 27.

Passive diffusion of FA across membranes plays a minor role compared with protein-mediated FA uptake and the flip-flop mechanism 28. The main proteins involved in FA uptake in non-ruminant cells include fatty acid translocator FAT/CD36 (CD36) and fatty acid transport proteins (FATP or SLC27A) 28. CD36 mRNA in our study accounted for ~5% of total genes measured (Table 1) and had a large increase in expression (>8-fold) during lactation (Figure 2). This protein is believed to participate in the process of milk fat secretion 3 because of its presence in the milk fat globule membranes (MGFM) 29. Our data support an important role for CD36 in milk fat synthesis. Although a role for this gene in milk fat secretion cannot be excluded we believe that its involvement in FA import in bovine mammary cells is more important.

We previously showed that bovine mammary tissue expresses most of the known SLC27A isoforms, but only expression of SLC27A6 was up-regulated during the first mo of lactation suggesting a role in NEFA uptake 19. Up-regulation in expression of SLC27A6 and CD36, and the fact that their proteins co-localize in murine heart subcellular fractions, support the concept of cooperation between both proteins during FA uptake. CD36 also co-localizes with acyl-CoA synthetases (ACSL) and fatty acid binding proteins (FABP) 30. Clearly, FA uptake by bovine mammary cells is a complex and coordinated mechanism requiring evaluation of multiple genes/proteins.

Long-chain FA (LCFA) are esterified with CoA in the inner face of the plasma membrane prior to participating in metabolic pathways. FA activation occurs primarily via acyl-CoA synthetase long-chain family member isoforms (ACSL) 31. ACSL1 mRNA is predominant among ACSL isoforms in bovine mammary tissue 19, and it increased >4-fold at the onset of lactation suggesting this isoform is important for copious milk fat synthesis (Figure 2). Among enzymes involved in activation of short chain FA (SCFA), acyl-CoA synthetase short-chain family member 2 (ACSS2) had greater mRNA abundance and up-regulation in expression than ACSS1 (a.k.a. ACAS2L). mRNA abundance of ACSS1, ACSS2, and ACSL1 in each case was <1% of genes investigated (Table 1). Bovine ACSS isoforms have been isolated and characterized in tissues other than mammary 32. In the mouse, ACSS isoforms only have 43.8% amino acid similarity and are located in different cell compartments. ACSS2 (originally named AceCS1) is exclusively present in cytosol, while ACSS1 (originally named AceCS2) is primarily found in mitochondria 32. Both enzymes have high affinity for acetate, with ACSS2 showing greater affinity than ACSS1. The latter also has modest affinity for propionate 32.

Bovine ACSS1 activated >4-fold more ¹⁴C-acetate into CO₂ than lipid, suggesting it targets acetate towards oxidation 32. Human ACSS2 was shown to channel acetate towards FA synthesis 33. In our study, both ACSS1 and ACSS2 mRNA increased substantially during lactation (Figure 2). ACSS2 transcript pattern corresponded with bovine mammary acetyl-CoA production throughout lactation 34. Thus, its large increase at the onset of lactation along with the pattern of ACE during the first 60 d post-partum, suggest the protein encoded by this gene provides activated acetate for de novo FA synthesis. In addition to its use in FA synthesis, acetate is the chief carbon source for energy generation in mammary accounting for ~33% of total CO₂ produced by the tissue 35. Lower mRNA abundance and pattern of expression of ACSS1 throughout lactation is in agreement with acetate use for oxidation 2. Overall, ACSS isoforms expression reflected the need for activation of acetate in mammary tissue.

Free diffusion of LCFA into cells is too slow to account for the rapid transport and selective targeting towards specific organelles 36, thus, LCFA require specific transporters. Fatty acid binding protein (FABP) and acyl-CoA binding protein (ACBP or DBI) are the main intracellular FA transporters in non-ruminant cells 36. The former has high affinity for LCFA but also can bind acyl-CoA 3738. ACBP is the major intracellular transporter of acyl-CoA in several mammalian tissues 39. We previously observed the presence of mRNA of all FABP isoforms, except FABP2 mRNA, in bovine mammary tissue with greater abundance and up-regulation of FABP3 mRNA during lactation. Transcript of FABP4 and FABP5 also were up-regulated during lactation but were less abundant compared with FABP3 19. In the present study, FABP3 was the second most abundant transcript (~16%) among all measured, in accord with the large cytosolic content of its protein in mammary epithelium 38. The large mRNA abundance of this gene also was a consequence of the 80-fold up-regulation during lactation, whereas ACBP mRNA abundance was <0.2% among all genes and had a small increase (1.5-fold) during lactation (Table 1 and Figure 2). Low ACBP mRNA abundance agrees with protein abundance data in bovine mammary 40. Our results suggest a minor role of ACBP in bovine mammary lipid synthesis, also supported by murine data 23.

In addition to a trafficking role, FABP3 through binding of activated acyl-CoAs could buffer cells from negative effects of activated FA and prevent inhibition of ACACA and SCD (stearoyl-CoA desaturase), roles usually attributed to ACBP 39. A positive relationship between FABP and SCD has been demonstrated in chickens 41, indicating a coordinated function between both proteins in mammary tissue as also suggested previously from an evaluation of published data 6. Based on our longitudinal mRNA expression and fatty acid data we propose that an important function of FABP3 in bovine mammary is to provide FA for SCD. Large affinity of FABP4 for oleic acid and up-regulation of its mRNA during lactation in bovine mammary tissue 19, led us to propose that FABP3 provides stearoyl-CoA (or other substrates such as 16:0 and trans11-18:1) 38 to SCD which then releases oleic acid to FABP4. The FA are then available to other enzymes involved in TAG synthesis.

We observed a 30-fold increase in ABCG2 [ATP-binding cassette, sub-family G (WHITE), member 2] transcript during lactation (Figure 2). mRNA abundance of this gene accounted for ~9% of total genes measured (Table 1). ABCG2 is a member of the large ATP binding cassette family of membrane-spanning efflux pumps that actively extrude a wide range of xenobiotics 42. It is present in apical membrane of murine mammary alveolar epithelia and plays a role in active secretion of toxins into milk 42. It also is present in the milk fat globule membrane (MFGM) 29, probably in the external bilayer originating from plasma membrane, due to its apical membrane localization. The large ABCG2 mRNA abundance and up-regulation, both in lactating bovine and murine mammary tissue 1723, is biologically puzzling because the primary role of this transporter in other tissues is detoxification. Recently, it was reported that one amino acid substitution at position 581 (Tyr to Ser; Y581S) of ABCG2 resulted in decreased milk production but increased milk fat and protein concentration and yield 43. Those data clearly do not support a role of ABCG2 in synthesis or secretion of milk fat. What seems apparent based on current bovine data is that ABCG2 plays an essential role in secretion of "some" important milk constituent 42. Cholesterol transport was suggested 43 but it is not supported by the low amount 3 and pattern of cholesterol in bovine milk throughout lactation 44. The only demonstrated role of ABCG2 in secretion of a milk component is for riboflavin, an essential, but quantitatively marginal, nutrient for the neonate 17. Therefore, its large up-regulation is suggestive of other functions in milk synthesis besides riboflavin secretion.

The amount and pattern of cholesterol secretion into bovine milk agrees with the pattern of ABCA1 (ATP-binding cassette, sub-family A, member 1) transcript we observed (Figure 2). ABCA1 mRNA accounted for <1% of all genes measured (Table 1). This gene also has low expression in a number of other bovine tissues, and bovine mammary in particular 45. ABCA1 is crucial for efflux of cholesterol from cells 46. Taken together, data suggest a minor role of ABCA1 in bovine mammary cholesterol flux.

Production of SCFA and palmitate from acetate is under control of ACACA, considered the rate-limiting step in de novo FA synthesis 2. In subsequent steps, both, acetyl-CoA and butyryl-CoA (mostly from plasma β-hydroxybutyrate) are primers for the cytosolic multifunctional protein fatty acid synthase (FASN) 16. The major product of FASN is palmitate but in ruminants the enzyme also produces SCFA 16. In the present study, ACACA mRNA abundance accounted for <1% whereas FASN accounted for 7% of total genes measured (Table 1). However, ACACA mRNA had greater up-regulation during lactation compared with FASN (Figure 3). These data are consistent with activity values of the two enzymes from pregnancy through lactation in dairy cows 34. Despite differences in magnitude, expression patterns among both genes were similar (r = 0.90; P < 0.01; see Excel File in Additional file 2), confirming previous findings summarized in a recent review article 6. In fact, this was the case for several genes involved in TAG synthesis and transport. Clearly, bovine mammary lipid synthesis requires coordinate expression of several genes for TAG synthesis and secretion. This point was stressed in studies of rabbit mammary lipid metabolism several decades ago 47. However, the scope of enzymes studied previously was relatively small compared with our study.

Only a fraction of FA taken up by the mammary gland is unsaturated owing to extensive ruminal biohydrogenation. The primary enzyme involved in monounsaturated FA synthesis is stearoyl-CoA desaturase (SCD), which introduces a double bond in the Δ⁹ position of myristoyl-, palmitoyl-, and stearoyl-CoA, primarily 48. To date, two SCD isoforms have been identified and characterized in bovine: SCD1 and SCD5. SCD1 was first characterized in bovine adipose 49 and until the discovery of SCD5, it was believed to be the only SCD present in this species. Thus, the bovine gene is simply referred to as SCD. Bovine SCD5 has been identified and characterized only recently 50 and it is expressed almost exclusively in brain. The nucleotide sequence for a bovine SCD6 homolog A (S. cerevisiae) has been deposited at NCBI 51, but no characterization of this isoform has been performed to date. SCD1 has 33.5% and 31.3% global nucleotide alignment identity 52 with SCD5 and SCD6, respectively. The primer pair used in the present study (Additional file 1, Table S2) is specific only for SCD1. The SCD mRNA abundance was the highest (23%; Table 1) among all genes measured. The large SCD mRNA abundance, relative to other classical lipogenic genes (e.g., ACACA, FASN), and the >40-fold up-regulation during lactation (Figure 3) agrees with the suggestion by Kinsella, based on lactating mammary SCD activity 53, that it plays a crucial role in TAG synthesis.

All indexes of "apparent" desaturase activity increased during lactation except for stearic acid and palmitic acid desaturation. The latter index had a peak at the beginning of lactation followed by a dramatic decrease (Figure 1, bottom panel). The pattern of SCD mRNA was not significantly correlated with any of the Δ⁹ desaturase indexes (Additional file 2), and was nearly opposite to the overall Δ⁹ desaturase index (Table 2). A lack of correlation between desaturase indexes and SCD desaturase activity was found previously in bovine intramuscular fat 54, which along with our findings suggests that use of indexes is inappropriate for inferring SCD gene expression/activity at least when considering the lactation cycle 4. Previous single-time point studies observed a positive correlation between SCD mRNA in goat mammary tissue and milk fat oleic acid 55 or desaturase indexes 56. In our study, relative % mRNA abundance of SCD (23%; Table 1) was related to the total amount of Δ⁹ FA (14:1c9 + 16:1c9 + 18:1c9 + cis9, trans11-18:1), which accounted for 19% of total milk FA (molar proportion; Additional file 1, Table S6).

Overall, our results support a central function of SCD in milk fat synthesis as previously demonstrated by intravenous infusions of sterculic acid, an inhibitor of SCD activity 57. This point is supported, to some extent, by the recent discovery that both SCD and DGAT1 polymorphisms affect saturation level of milk fat and largely explain genetic variance on the desaturase indexes. The effect, however, was greater for SCD and primarily on medium/long-chain FA (from 10- to 16-carbon FA). Polymorphism in DGAT1 explained less variability but had a greater effect on long-chain FA (18-carbon) 58.

The lack of correlation between temporal desaturase indexes (i.e., apparent SCD activity) and SCD mRNA expression is not surprising due to the many factors that likely play a role in determining milk FA output. An important factor to consider is the selective uptake of stearic acid from blood VLDL by the mammary gland 59. Mammary tissue relies heavily on utilization of VLDL-TAG during lactation 27. Another key aspect is the high concentration of oleic acid, both in plasma at mid-lactation 60 and the NEFA pool at early lactation 61. Oleic acid is the predominant FA in ER membranes 62 and Δ⁹ desaturase is essential for their functional maintenance 48. FABP3 expression and activity also could play a role in preferential channelling of palmitic and stearic acid for desaturation. Bovine FABP3 has a high affinity for both stearic and palmitic acid 3863.

Synthesis of very-long-chain FA is carried out by fatty acid desaturase 1 (FADS1) and 2 (FADS2), which add double bonds at the Δ⁵ and Δ⁶ position of PUFA. Arachidonic acid (20:4n-6), eicosapentaenoic acid (20:5n-3), and docosahexaenoic acid (22:6n-3) are synthesized via FADS1 and FADS2 6465. FADS1 and FADS2 mRNA abundance was <1% of total genes measured (Table 1). The greater relative mRNA abundance of FADS1 compared with FADS2 is similar to the one reported in rat mammary 65. Relative mRNA abundance in bovine mammary tissue was in concordance with the amount of product of both enzymes, which accounted only for 0.12% and 0.03% of total milk FA (Additional file 1, Table S6). FADS1 mRNA expression increased 18-fold by d 60 postpartum, whereas FADS2 mRNA increased only 3-fold (Figure 3). Expression pattern of FADS1 mRNA did not agree with the observed Δ⁵ desaturase index, whereas FADS2 mRNA had a similar pattern compared with its index of activity (Additional file 1, Figure S2). In practical terms, it appears that FADS1 mRNA abundance/activity might be more amenable to dietary manipulation in order to increase omega-3/omega-6 ratio in milk fat.

Discrete steps in the pathway of TAG synthesis 66 have been discerned in classical lipogenic tissues (e.g., liver, adipose; 67) and, despite lack of functional studies of mammary lipin, the same steps likely apply to the mammary gland. Expression of GPAM (glycerol-3-phosphate acyltransferase, mitochondrial), AGPAT6 (1-acylglycerol-3-phosphate O-acyltransferase 6), DGAT1 (diacylglycerol acyltransferase 1), and LPIN1 (lipin 1) mRNA accounted for >2%, >1%, ~0.1%, and ~0.1%, respectively, of total transcripts measured (Table 1). DGAT2 mRNA expression was nearly undetectable. Despite these differences, we observed that LPIN1 mRNA was up-regulated during lactation by 20-fold (Figure 3). The more abundant GPAM and AGPAT6 mRNA increased by 10- and 15-fold by d 60 post-partum (Figure 3). GPAM expression agrees with the greater enzyme activity in mammary gland during lactation in non-ruminants (e.g., 47), and confirms its crucial role in TAG synthesis 16. AGPAT6 and LPIN1 are the major isoforms within each gene family in bovine mammary tissue 19. When the former was knocked out in lactating mice, they failed to synthesize milk fat 68.

Recent characterization of AGPAT6 showed that this gene is in fact the ortholog of human GPAT4 (microsomal glycerol-3-phosphate acyltransferase 4) 69. The product of GPAT4 did not have AGPAT activity but instead a clear glycerol-3-phosphate acyltransferase activity 69. However, authors failed to demonstrate an increase in TAG synthesis after overexpression of GPAT4. Despite these results in the mouse, our data support an important role of AGPAT6 in bovine mammary and very likely in TAG synthesis, as previously discussed 19. Our data seem to suggest that AGPAT6 has AGPAT activity. This is inferred by the lower mRNA abundance and temporal increase in the transcript of other AGPAT in bovine mammary tissue 19, as well as the large up-regulation of GPAM expression. It seems unlikely that bovine mammary tissue would require a larger number of enzymes with GPAT activity relative to other downstream FA acylating enzymes.

A potential function of LPIN1 in regulation of transcription of other genes involved in milk fat synthesis cannot be excluded. Recently, it was demonstrated that LPIN1 is essential for PPARα 70 activation but it also interacts with PPARγ 71. In addition, LPIN1 is a target of insulin-stimulated phosphorylation through mTOR 72, which in turn seems to promote microsomal vs. cytosolic localization of the protein. It could be possible that in bovine mammary tissue insulin signalling through INSR (insulin receptor) and IRS1 (insulin receptor substrate-1) as well as mTOR (FRAP1), all of which had a significant increase in mRNA expression (1.5-4-fold) through lactation 8, induces LPIN1 phosphorylation and localization to ER for DAG synthesis and TAG formation.

Relative mRNA abundance of DGAT1 was 17-fold greater compared with DGAT2 (Table 1) and had modest up-regulation particularly in early lactation (Figure 3). The temporal pattern in expression of this gene was similar to butyrate yield (Additional file 1, Table S7). DGAT1 acylates the sn-3 position of DAG and most butyrate in milk TAG is found here 44. This protein has high affinity for butyryl-CoA and even higher for palmitoyl-CoA 73. However, DGAT1 might favour use of butyrate for the sn-3 position of DAG as indicated by the larger affinity of AGPAT for LCFA in mammary 74 along with the preferential incorporation of palmitic acid in the sn-1 and sn-2 position 3.

The pattern of DGAT1 expression was unexpected because it is considered a QTL 75 for milk production traits, and is essential for murine mammary gland development and milk synthesis 76. Data from the present study (i.e., fold-change and mRNA abundance) suggest that DGAT1, compared with other genes involved in TAG synthesis, is of minor importance in the overall process of milk fat synthesis. The temporal decrease or lack of increase in expression of Dgat1 in mammary tissue of FVB mouse 23 provides additional support. We do not, however, believe our findings contradict previous functional studies 77, demonstrating a pivotal role for DGAT1 in increasing milk TAG. The fact remains that DGAT1 is one of many proteins composing the TAG synthesis pathway 67. A lack in functionality of any gene in this pathway can likely reduce the efficiency of TAG synthesis. Protein expression and functional studies during the entire lactation should be conducted to clarify the importance of DGAT, and others, in mammary lipid synthesis.

Milk fat globules are formed in the ER membrane via incorporation of newly-formed TAG, transported to the apical membrane, and eventually released 3. Well-defined proteins involved in these processes in mammary include butyrophilin (BTN1A1), xanthine dehydrogenase (XDH), and adipophilin (ADFP) 31529. Relative mRNA abundance among these genes confirms the large amount of the respective protein product found in MFGM 3. ADFP mRNA abundance (~10%) relative to other genes measured was almost twice that of BTN1A1 (~5%) (Table 1). Expression of BTN1A1, XDH, and ADFP increased during lactation and averaged 14-, 8-, and 3-fold by d 60 postpartum, respectively (Figure 3). The larger increase in expression of BTN1A1 seems to support a more crucial role for this gene, as recently suggested 78, in milk fat secretion compared with XDH and ADFP. However, despite the greater increase in mRNA expression for BTN1A1 than ADFP at 60 d post-partum relative to pre-partum, the larger overall abundance of ADFP transcript resulted in similar mRNA abundance for both genes at 60 d (data not shown). The similar proportion in relative mRNA of BTN1A1 and ADFP is in accordance with their protein abundance in mammary tissue 3. Our results highlight the limitations of reporting gene expression data exclusively as n-fold change. Relative mRNA abundance also needs to be considered. The precise mechanism of milk fat secretion is still debated (e.g. 1578) but our data confirmed a role for BTN1A1, XDH, and ADFP. The similar pattern of expression and large correlation observed (r ≥ 0.92, P < 0.01; Supplemental Excel file) are indicative of a concerted action among the 3 genes and, thus, provides support for the tripartite model of murine milk lipid secretion 15.

Perilipins are a family of proteins localized in the periphery of intracellular lipid droplets and are essential for droplet formation as well as lipolysis in adipose tissue 79. ADFP and the perilipin gene (PLIN) are both part of the perilipin family. Relative PLIN mRNA abundance was ~0.01% of total genes measured (Table 1) and its expression was only numerically up-regulated (~3-fold by d 60) during lactation. Our data do not support a significant role for PLIN in mammary lipid droplet formation. Functional studies could clarify the involvement, if any, of PLIN in this process.

β-hydroxybutyrate (BHBA) is the major ketone body produced in bovine species under most circumstances. Bovine mammary gland takes up large amounts of BHBA from blood 80. Previous studies, focused mostly on de novo synthesis and oxidation of FA, concluded that use of BHBA (as 4-carbon units) by mammary cells is primarily for de novo FA synthesis. A minor portion is used as energy source through the TCA cycle, leaving unaccounted the fate of a substantial portion of the BHBA taken up 80. BDH1 (3-hydroxybutyrate dehydrogenase, type 1) and OXCT1 (3-oxoacid CoA transferase 1) catalyze the initial and committed steps of BHBA utilization in mitochondria 81. mRNA abundance for BDH1 and OXCT1 accounted for <0.1% of genes measured (Table 1). The large increase in expression and relative mRNA abundance during lactation (Figure 3; Table 1) correspond with their enzymatic activity level in lactating rat mammary tissue 82. Based on these data, and a previous microarray study 8, we propose that the major fate of BHBA in bovine mammary is the synthesis of citrate (Additional file 1, Figure S3 for model and details).

A large body of evidence supports the suggestion that SREBP1 (sterol regulatory element-binding protein 1) is pivotal in the regulation of milk fat synthesis in mouse 7 and cow 83. SREBP1 and 2 reside as inactive precursors in the ER membrane and are transported to the Golgi for proteolytic cleavage (i.e. activation) prior to entering the nucleus and activation of sterol responsive element-containing genes (e.g., ACACA, FASN). The transport step to the Golgi is blocked by sterols via the sterol-sensing protein SCAP (SREBP cleavage activating protein). SCAP is essential for the movement of SREBP isofoms from the ER to the Golgi, essentially acting as gate keeper for movement of inactive SREBP1 and 2 84. Insulin induced gene (INSIG) 1 and 2 are proteins that interact with SCAP in an oxysterol-dependent and independent fashion and regulate the responsiveness of SREBP1 and 2 processing via SCAP, thus altering rates of lipogenesis 84.

Expression of SREBP genes (SREBF1 and SREBF2), and thyroid hormone responsive SPOT14 (THRSP) averaged ~2-fold by day 60 postpartum (Figure 4) but relative mRNA abundance was ~0.13% for SREBF1 and SREBF2, and only 0.01% of total genes measured for THRSP (Table 1). There are two isoforms of SREBP1 (a and c) that can be expressed at different levels in tissues. The two isoforms differ by only 84 nucleotides at the first exon 85 and each appears to be specific in the control of transcription of genes involved in cholesterol (isoform a) or TAG synthesis (isofom c) 86. The primer pair used in our study is unable to differentiate between these because it amplifies mRNA at the 14^th exon.

INSIG1, INSIG2, and SCAP mRNA abundance accounted for ~0.4%, ~0.1%, and ~0.1%, respectively, of total genes studied (Table 1). These genes had increased mRNA expression during lactation, ranging from ~1.5-fold for SCAP to ~12-fold for INSIG1 at or close to peak lactation (Figure 4). It is apparent from our comprehensive analysis of SREBF1, in concert with its co-factors (SCAP, INSIG1 and 2), that the complete activation program of SREBP is up-regulated in the bovine mammary gland during lactation. More INSIG1 mRNA was expressed in mammary tissue than SCAP or SREBF1 and 2 (Table 1), and its level of up-regulation reached 10-12-fold between peak through mid-lactation (Figure 4). If this pattern of INSIG1 mRNA expression extended to the protein level, more SREBP might be retained in the ER via SCAP-INSIG1 binding (i.e., rendered inactive) 84.

INSIG1 binds oxysterols (not cholesterol) specifically and this specificity is directly correlated with the ability of these compounds to inhibit SREBP cleavage 84. Data on oxysterol concentration/synthesis in bovine mammary tissue are not readily available but considering that mammary tissue uptake and synthesis of cholesterol is small we can infer low presence of these compounds. In this regard, it is interesting that genes coding for oxysterol binding proteins were up-regulated during lactation (discussed below). Cleavage inhibition of SREBP also can occur in the absence of sterols, particularly when the ratio of INSIG1/SCAP increases as in the present study 84. Therefore, it appears that an increase of INSIG1 alone can be enough to block SREBP cleavage. In fact, decreased SREBP activity as a consequence of increased INSIG1 transcript has been observed in liver when INSIG1 is overexpressed 87 or during high fat diet-induced INSIG1 up-regulation 88. Based on previous results and our data we postulate that mammary lipid synthesis cannot rely solely on transcriptional regulation via SREBP1 which is probably inhibited or controlled by INSIG1.

The observed up-regulation of INSIG1 during lactation apparently does not make teleological sense because mammary tissue requires substantial up-regulation of genes involved in de novo lipid synthesis (Figure 1; Table 2). It is striking, however, that INSIG1 mRNA was at its peak (d 120) when the cows likely were in positive energy balance, i.e. high blood insulin/glucagon, and the ratio of de novo-synthesized FA to imported FA was maximal (Figure 1). INSIG1 expression was positively correlated with the ratio of synthesized/imported FA (r = 0.38, P < 0.05; Additional file 2). Paradoxically, these data suggest some involvement of INSIG1 in inducing FA synthesis. This suggestion is supported by recent data 83, where a decrease of INSIG1 during experimentally-induced milk fat depression in lactating cows was observed.

Several factors could drive the marked increase in INSIG1 mRNA during lactation. For example, up-regulation of INSIG1 expression might be a consequence of SREBF isoform mRNA up-regulation and increased activity (i.e., induction of gene expression) of the corresponding proteins. SREBP1a and SREBP2 directly regulate INSIG1 gene expression 84. Given the lipogenic capacity of mammary tissue, it is more likely that SREBP1c is the more abundant isoform. Thus, INSIG1 up-regulation in bovine mammary tissue could be under control of SREBF2 (Figure 4). Another reason for marked INSIG1 mRNA up-regulation might be its very short half-life 84, or as a necessary mechanism to sense low mammary cholesterol levels in order to regulate de novo FA synthesis.

Our data support a need of INSIG1 in controlling the induction of gene expression by SREBF isoforms. Therefore, INSIG1 could play a central role in orchestrating lipid metabolism (i.e., TAG vs. cholesterol) in bovine mammary tissue during lactation. In support of this, it previously has been suggested that high levels of INSIG1 create a situation in which low levels of endogenous sterols can trigger SCAP binding to INSIG1 without the necessity for exogenous sterols 84. The "brake" effect of INSIG1 on TAG accumulation has been clearly demonstrated in mouse adipose tissue 88, when overexpression of INSIG1 in 3T3-L1 cells led to a decrease in mRNA abundance of lipogenic genes (e.g. Srebp1c, Acaca, Chrebp, Pparg). Our data, however, do not support a similar effect of INSIG1 on expression of mammary lipogenic genes in ruminant.

SREBF2 expression was up-regulated to a greater extent than SREBF1 expression during lactation (Figure 4). The reason for up-regulation of SREBF2 expression, and the similar mRNA abundance compared with SREBF1, is not apparent because this gene is thought to be involved primarily in cholesterol biosynthesis 85 and the amount of cholesterol in milk is low 89. Cholesterol is almost exclusively synthesized de novo in bovine mammary tissue 90, thus SREBF2 expression also might be necessary to meet cholesterol requirements for MFGM formation 3. In support of this we observed, via microarray analysis, that expression of several genes involved in cholesterol synthesis was significantly up-regulated (1.5-2-fold compare to -30 d) during lactation 8. An additional function of SREBP2 is to induce mRNA expression of genes involved in FA synthesis 85. It could be possible that SREBF2 mRNA up-regulation during lactation might compensate for the potential inhibition of INSIG1 on both SREBP.

Genes involved in FA transport such as LPL, CD36, and ACSL1 are peroxisome proliferator-activated receptor gamma (PPARγ) target genes 91. In the lactating mouse, expression of LPL and ACSL1 was up-regulated significantly despite the fact that expression of PPARγ was down-regulated 23. However, it has recently been demonstrated that changes in abundance of adipocytes at several stages of pregnancy in murine mammary tissue (i.e., high in early pregnancy vs. low in late pregnancy) could account for the decrease in PPARγ gene (PPARG) mRNA abundance 92. In bovine mammary, PPARG mRNA accounted only for 0.01% of total genes measured (Table 1) but was consistently up-regulated during lactation (Figure 4). The low mRNA abundance of "adipocyte-specific" genes (e.g., DGAT2, PPARG, ACBP, and PLIN), particularly at the end of pregnancy (i.e. -15 d), clearly indicates that biopsied cow mammary tissue contained low amounts of adipocytes. Thus, our longitudinal data on PPARG expression should represent that of epithelial cells. With this premise and despite the low mRNA abundance, up-regulation of PPARG mRNA during lactation suggests a potential role of this nuclear receptor in milk fat synthesis. A recent study with PPARγ-knockout mice indicated that absence of PPARG increased utilization of FA for synthesis of inflammatory lipids due to reduced TAG synthesis 93. A role of PPARG in regulating the entire bovine milk fat synthesis machinery also is supported by recent results from our laboratory where treatment of MacT cells (immortalized bovine mammary epithelial cells) with rosiglitazone, a specific PPARγ agonist, resulted in coordinated up-regulation in expression of genes involved in FA import (e.g., CD36), de novo FA synthesis (e.g., ACACA, FASN, SREBF1), and TAG synthesis (e.g., LPIN1, SCD) 94.

The relative mRNA abundance for the PPAR gamma coactivators, PPARGC1A and PPARGC1B, was 0.04% and 0.01% of total genes measured (Table 1). Whereas expression of PPARGC1A was substantially up-regulated through d 120 post-partum (~11-fold), expression of PPARGC1B was consistently down-regulated during the entire lactation (Figure 4). Differences in relative mRNA abundance, large temporal up-regulation of PPARGC1A mRNA, and down-regulation of PPARGC1B transcript, suggest an important role of the former in bovine milk fat synthesis. The importance of PPARGC1A in the overall process of mammary lipid synthesis likely is more related to its well-defined role in regulating mitochondrial biogenesis and energy metabolism 95. Up-regulation of PPARGC1A during lactation agrees with reported increases in numbers and turn-over rate of mitochondria in lactating mammary tissue 21. Furthermore, our combined data (Table 1, Figure 4) also point to a concerted action of PPARGC1A and INSIG1. This last observation is captivating based on previous observations in murine 96 and bovine mammary epithelial cells 94 where INSIG1 was demonstrated to be a PPARγ responsive gene, suggesting that PPARG in mammary tissue could serve as regulator of SREBP activity.

Ceramide, which is involved in cell signaling, cell cycle, and regulation of protein transport from ER to Golgi, is one of the most studied sphingolipids in nature 9798. Other sphingolipids with signaling roles include sphingosine (Sph) and sphingosine-1-phosphate (S1P) 98. Sphingomyelin synthesis from ceramide is important for milk quality because this compound is considered a functional food component 16. Although minor compared with TAG, sphingolipids are the third most important lipid component 99 in bovine milk fat. MFGM formation relies on sphingolipid and cholesterol availability 100, thus coordinated synthesis of both compounds is pivotal to milk lipid droplet formation/secretion. Sphingolipids are involved in lipid synthesis regulation through their action on SREBP 101. Mammary tissue synthesizes sphingolipids de novo 99 from palmitoyl-CoA, leading to ceramide formation and incorporation into sphingomyelin 99. Thus, palmitic acid used for ceramide synthesis in mammary appears a required step and also might represent a regulatory point for FA synthesis because ceramides can inhibit this process by blocking the activity of AKT/PKB 102. Regulation of FA synthesis by sphingolipids in mammary tissue has never been investigated. Thus, we selected genes crucial in ceramide synthesis/degradation, as well as enzymes involved in Sph and S1P synthesis (Additional file 1, Figure S5) to explore further their role in lipid synthesis regulation.

In accordance with the minor concentration of sphingolipids in milk, genes in this pathway had low mRNA abundance ranging from 0.05% (N-acylsphingosine amidohydrolase-like or ASAHL) to 0.61% (LAG1 homolog, ceramide synthase 2 or LASS2) of total genes examined (Table 1). LASS2 was the most abundant among sphingolipid-related genes and the only one with >2-fold up-regulation during lactation (Figure 5). LASS isoforms are orthologues of the yeast Longevity-assurance gene. The enzyme is localized in the ER 97 and isoforms appear to have specific tissue distribution, suggesting they perform "specialized" functions 1997. LASS2 had peak expression at 60 d postpartum, in agreement with previous data on milk sphingolipid concentration 99. Combined data on genes associated with ceramide synthesis suggest an increase in synthesis coupled with decreased degradation throughout lactation (Figure 5 and Additional file 1, Figure S5). An increase in mammary ceramide synthesis might potentially serve as a positive signal for proteins involved in lipid synthesis through activation of SREBP1 as suggested previously 101. Increased sphingolipid synthesis during lactation also could affect availability of cholesterol for MFGM, and might explain the inverse pattern between milk cholesterol ester and sphingolipid 99.

Transport of ceramide from ER to Golgi is achieved by oxysterol binding proteins (OSBP), which also act as sterol sensors whose function is to integrate cellular sterol status with sphingomyelin metabolism 103. A novel role for OSBP in regulation of lipid synthesis was demonstrated when overexpression of OSBP led to increases in TAG synthesis in mouse liver and concomitant up-regulation of SREBF1 and INSIG1 104. mRNA abundance of OSBP and related genes (OSBPL2, OSBPL10) in mammary was comparable with that of SREBF1 and 2, and their expression was up-regulated ~1.5-fold during lactation (Figure 5) suggesting a functional role, likely involving the regulation of SREBF1 and the coordination of sphingolipid and cholesterol synthesis.