The objective of this study was to investigate variations in milk protein composition induced by feed restriction, as well as the impact of the intensity of feed restriction on these variations. We first compared the milk protein composition in milk samples collected from the two trials before the feed restrictions were applied, in order to ensure that a comparison of both feed restriction trials was relevant. In term of concentrations, the major milk protein profile was similar between the trials, although α-LA was more concentrated under LM than SH trial. This difference in α-LA concentration was most likely due to a difference in the lactation stage, as the cows involved in the LM trial were around 77 DIM and those in SH were around 165 DIM. Similarly, the yields of major protein were higher under LM conditions than under SH, particularly with respect to α-LA, αs1-, αs2- and β-CN, this being linked to higher milk yields in the LM cows that were at peak lactation. Regarding proteomes, 345 proteins were identified during the trials, which was quite consistent with the milk proteomes published previously using LC-MS/MS. Indeed, among the 4654 proteins identified in the aggregation published by Delosière et al.6, 3288 were specific to colostrum and only 775 and 577 were identified during peak lactation and mid-lactation studies, respectively. Before the restriction period, 43 low-abundance proteins were exclusive to the SH trial and nine to the LM trial. Again, this difference was very likely due to a difference in lactation stage as the proteome changes during lactation, with some proteins being exclusive to each stage6. In our trials, these 52 exclusive proteins only accounted for 0.8% of the total protein counts prior to restriction periods. The pre-restriction milk proteomes of both trials were therefore very similar and it was possible to compare their modifications induced by feed restriction.
In both trials, the reduction in milk yield induced by feed restriction was concomitant with a decrease in the major milk protein yield. This effect on major milk proteins increased in line with the intensity of the restriction. During the SH trial, with an important negative energy balance (-42.3 MJ/d) and milk yield loss (-34%), all major milk proteins were quantitatively affected, whereas during the LM trial, with lower negative energy balance (-21.4 MJ/d) and milk yield loss (-9%), only casein quantities were affected. During the high intensity feed restriction, this reduction in yield lowered the concentrations of αs1, αs2- and β-CN. It appeared that the αs2-CN concentration was the most sensitive to feed restriction, as it was the most significantly affected during the SH trial (-25%) and tended to decrease under the LM trial. When studying corn versus grass diets, Vanbergue et al.3 only observed variations in milk concentrations of αs2-CN (-22%; p = 0.029) and β-CN (-20%; p = 0.014), which supports the hypothesis that αs2-CN is the most sensitive to feed variation, followed by β- and αs1-CN. Billa et al.10 also saw a reduction in CSN1S2 transcripts coding for αs2-CN in the mammary gland during the feed restriction period of the SH trial. This shows that a reduction in the αs2-CN concentration in milk is directly linked to a decrease in CSN1S2 gene expression in the mammary gland.
Proteomic analyses confirmed the decreased concentrations of α-LA, αs1 and αs2-CN during SH feed restriction, with αs2-CN being the most affected protein. This analysis also showed a significant effect of intense feed restriction on proteins involved in lipid metabolism, with 14 affected proteins involved in this metabolism in the SH trial. Among the seven proteins involved in this metabolism which displayed increased abundance during feed restriction, four involved in lipid transport and storage were found: apolipoproteins (A-I, A-IV and E) and perilipin. Moreover, the decreased abundance of CIDE-N domain-containing protein, a lipolysis inhibitor and storage activator, may have reflected increased lipid mobilization in adipose tissue, which is consistent with the increase in plasma NEFA concentrations observed in both trials. Lower concentrations of fatty acid synthase, which catalyzes the de novo biosynthesis of fatty acids, were observed in milk under SH conditions. This finding was in line with the reported decrease of FASN RNA in the cytosolic crescent of milk fat globules during 40% feed restriction over four days11. It was also consistent with the decrease in de novo synthesized fatty acids during the SH trial, reflected by the reduction in short chain fatty acid concentrations in milk8. A rise in the fat content of up to 13% was due to the uptake of long chain fatty acids from lipid mobilization, as indicated by the increase in plasma NEFA concentrations seen during both trials. Such adaptations of lipid metabolism in the context of a negative energy balance have already been well described1,12,13, and notably involved the downregulation of several mammary lipogenic genes during the first days of short-term feed restriction11. However, during our LM trial, concentrations of fatty acid synthase rose slightly after five days of feed restriction, suggesting that intense restriction is necessary for this shift in fatty acid metabolism to occur.
The modifications observed regarding on milk proteins and proteome were consecutive to changes to mammary metabolism, partly because of a reduction in nutrient uptake by the mammary gland during feed restriction, as shown previously by Guinard-Flament et al.14. Indeed, these authors showed that feed restriction reduced mammary blood flow alongside reductions in mammary nutrient and dioxygen uptakes during a -30% DMI feed restriction14. Nevertheless, under SH conditions in our study, mammary metabolism appeared to partially compensate for decreased nutrient uptake by increasing carbohydrate catabolism and lipid transport. Indeed, among the 12 proteins involved in carbohydrate metabolism that were more abundant in milk during the SH trial, seven are involved in glycolysis (hexokinase, glucose-6-phosphate isomerase, fructose-bisphosphate aldolase, glyceraldehyde-3-phosphate dehydrogenase, phosphoglycerate kinase 1, alpha-enolase and pyruvate kinase), one is involved in the pentose phosphate pathway (6-phosphogluconate dehydrogenase, decarboxylating), which is the parallel pathway to glycolysis, and glycogen phosphorylase catalyzes the rate-limiting step in glycogenolysis that produces substrate for both the glycolysis and pentose phosphate pathways. This increased abundance of proteins involved in carbohydrate degradation in milk may reflect the high level of energy required to maintain mammary gland metabolism in a lactating cow.
The total protein content decreased during both trials but only by 4% in SH, despite the lower concentrations of some major milk proteins. Lacy-Hulbert et al.15, who had observed an increased total protein content in milk (+ 8%) during intense feed restriction (-50% of DMI for 26 days) hypothesized that feed restriction tended to concentrate serum-derived proteins in milk. Indeed, 43 proteins identified in the SH milk samples during feed restriction had not been present before restriction, and among the seven proteins with increased concentrations in milk under both the LM and SH conditions, five are normally present in plasma and two were not found in milk before restriction (ceruloplasmin, apolipoprotein A-IV, alpha-1B-glycoprotein, angiotensinogen and serotransferrin). This increase in plasma protein concentrations in milk may reflect a loss of mammary epithelial barrier integrity, which could play a role in reducing milk production during feed restriction. This had already been suggested by Herve et al.9 who observed an elevated rate of mammary epithelial cell exfoliation under LM trial, as well as an increased Na+ concentration in milk, and by Stumpf et al.16 who saw an increased permeability of mammary cell tight junctions during short and intense feed restriction (-50% DMI for seven days). Under our SH conditions, we observed an elevation of lactotransferrin concentrations in milk, an increase that is known to happen during the first days of the dry period17 when involution starts and the epithelial barrier loses its integrity. This increased permeability of the epithelial barrier is coupled with increased leucocyte infiltration of the mammary gland, as shown by higher milk somatic cell count in the LM trial9 and during other feed restriction experiments15,18−20, and an upregulation of immune genes, as observed in the mammary tissue during involution21. Moreover, 12 of the 13 proteins involved in positive regulation of immune system processes were more abundant in milk during the feed restriction period under SH conditions, suggesting a similar immune system upregulation. Variations in milk of the immune system related protein were confirmed for five of them, with similar changes to their transcript levels in the mammary gland10, and in particular C3, which plays a central role in activation of the complement system. However, among the 33 proteins involved in protein metabolism, 12 over-abundant proteins have a protease inhibition function (α-1-antiproteinase, α-2-macroglobulin, antithrombin-III, factor XIIa inhibitor, inter-α-inhibitor heavy chain H4, leukocyte elastase inhibitor and serpins A3-2, A3-3, A3-6, A3-7, B4 and G1) and four are involved in the inhibition of complement activation, inflammation and cell death (chitinase-3-like protein 1, clusterin, factor XIIa inhibitor and heat shock protein HSP 90-α). These results therefore suggest a greater regulation of the immune system in the mammary gland during feed restriction.
These adaptations, which are reminiscent of some of those observed during early involution, remained reversible during these restriction trials, as both milk yield and composition recovered after a return to ad libitum feeding. Delosière et al.6 proposed some milk proteins exclusive to early lactation as biomarkers of negative energy balance, and none of these were found in milk during the negative energy balance induced by feed restriction later in lactation. Nevertheless, some proteins are affected by both moderate and high intensity feed restrictions: alpha-enolase, ceruloplasmin, apolipoprotein A-IV, alpha-1B-glycoprotein, angiotensinogen and serotransferrin.
Alpha-enolase is an enzyme present in all tissues that catalyzes the interconversion of 2-phosphoglycerate to phosphoenolpyruvate; its upregulation indicates an enhancement of glycolysis and has also been observed during ketosis22, a common metabolic disease induced by a negative energy balance. The five other proteins affected by both moderate and intense feed restriction were mainly found secreted in plasma. Apolipoprotein A-IV is primarily synthesized in the small intestine; this lipid-binding protein is involved in numerous physiological processes such as lipid metabolism and glucose homeostasis23. Apolipoprotein A-IV upregulation in the bovine mammary gland has been described during inflammation challenges where its anti-inflammatory activities may balance the immune response24. Ceruloplasmin, alpha-1B-glycoprotein, serotransferrin and angiotensinogen are mainly expressed in the liver. Ceruloplasmin is a copper-binding glycoprotein with antioxidant and cytoprotective activities. Increased concentrations of cerulaplasmin in bovine milk have been described during subclinical and clinical mastitis25 and may indicate inflammation. Alpha-1B-glycoprotein is a glycoprotein of unknown function. In the cow, its serum level seems to increase during various stresses such as tuberculosis26, high-altitude hypoxia27 or mastitis28. Serotransferrin, an iron binding transport glycoprotein, is seen at high concentrations in milk during early lactation, and then fall rapidly over time. Mastitis events can also increase serotransferrin concentrations in milk through changes to the mammary gland epithelium29. Angiotensinogen is the precursor of angiotensin. In dairy cows it has been shown that ketosis may alter the metabolism of angiotensinogen to angiotensin30.