The low-quality forage diets (CS and RS) largely inhibited the activity of AA metabolism but improved the urea cycle in dairy cows, based on DMEs (
Figure 4). The current findings of decreased tyrosine, valine, phenylalanine and pyruvic acid were consistent with the dietary differences and the improved MUN concentration in cows fed RS compared to AH [
3]. For MUN content, arterial phenylalanine was shown to be highly positively correlated, but arterial 4-hydroxyproline and phenylpropanoate were negatively correlated (
Figure 4). Free blood-derived 4-hydroxyproline is generated from the degradation of collagens or other proteins containing 4-hydroxylprolyl residues [
14]. Phenylpropanoate is the indirect catabolite of phenylalanine [
15]. Thus, the positive correlation between these two metabolites and MUN might confirm that protein or AA degradation in cows would result in increased MUN in milk. In addition, these findings confirmed restricted AA and glucose metabolism after cows were fed low-quality forage diets (CS and RS), as shown in our previous studies [
5,
16]. Sun et al [
8] found that cows fed a low-quality CS diet showed changes in the glycine, serine, and threonine metabolism, tyrosine metabolism, and phenylalanine metabolism pathways, which are similar to the results of current study. The important pathways based on identified DMEs between AH and cereal straw diets were mainly related to phenylalanine, tyrosine, and tryptophan biosynthesis; phenylalanine metabolism; and valine, leucine and isoleucine biosynthesis. Phenylalanine, tyrosine, and tryptophan biosynthesis and valine, leucine and isoleucine biosynthesis do not exist in mammary glands [
17], but this process may indicate the limited AA anabolism and AA redistribution in the liver of cows fed cereal straws. The DMEs of phenylalanine, tyrosine, and valine enriched in these metabolic pathways decreased after cows were fed cereal straw diets, indicating that a shortage of these AAs was the limiting factor in the lactating cows fed cereal straws. These AAs are the key milk protein precursors in dairy cows [
18], and their shortage in arterial supply would then result in decreased mammary uptake [
10]. Thus, the decreased levels of these AAs in arterial plasma might result in restricted milk protein synthesis. In addition, the shortage of valine in venous plasma was consistent with our previous study using quantitative analysis of AA [
5]. In the current study, the arterial glutamate concentration was decreased after cows were fed CS and RS compared to AH, but it did not decrease in the vein. This finding is different from the results of our previous study in which we found the similar arterial glutamate concentration between the three diets [
5]. Glutamate is the main AA precursor and primarily acts as a nitrogen donor for other AAs [
19]. A high demand for glutamate for protein synthesis has been reported in the mammary gland [
20,
21], which is able to generate a large arteriovenous difference in glutamate [
21]. The increased arterial glutamate along with decreased essential AAs concentrations in CS-fed cows confirmed the shortage of essential AAs, which then required more glutamate uptake by the mammary gland for the transaminase process [
10]. Thus, our results confirmed the function of glutamate as an AA precursor and a nitrogen donor.
The lower abundance of arterial pyruvic acid in RS group might reflect the lack of energy supply of the RS diet compared to the AH diet, which is consistent with restricted gluconeogenesis in the liver and glucose metabolism in the mammary glands [
16]. In addition, arterial phenylpropanoate in cows fed CS or RS was much greater than that in cows fed AH, and prostaglandin A2 was only found in the RS diet. Phenylpropanoate is a decomposed metabolite of phenylalanine; thus, the greater arterial phenylpropanoate in cows fed CS and RS may indicate the shortage of energy or other nonessential AAs from CS and RS diets that require the decomposition of phenylalanine [
22]. Prostaglandin A is derived from arachidonic acid and has an anti-proliferative function by regulating the expression of apoptosis genes [
23]. Thus, the generation of prostaglandin A in cows fed RS may restrict mammary cell proliferation, which would decrease the number of mammary cells [
24]. In addition, in mammals, prostaglandins play important roles in several physiological and pathological processes. For example, they have been shown to induce chemokines resulting in infiltration of inflammatory cells, such as eosinophils, neutrophils, and macrophages [
25], which are important for consequent adaptive immunity [
26]. Therefore, the greater prostaglandin A secretion in cows fed RS might indicate that the mammary glands of RS group cows might have had a certain inflammatory response. A previous study also reported that urea at a high concentration could stimulate the over-secretion of prostaglandin E2 in dairy cows [
27]. Thus, the greater abundance of prostaglandin A in the arteries of cows fed RS might be attributed to the lower nitrogen utilization efficiency, which was revealed in our previous study [
3]. However, we could not explain why cows fed with CS diet did not have prostaglandin A2. More studies will be conducted in the future to determine the underlying mechanism.
A much greater number of DMEs were found in the artery than the vein in the comparisons between AH and cereal straw diets, indicating the important role of arterial blood in the study of nutrient metabolism in the mammary glands. Previous studies have reported that there are certain differences between arteries and veins with regard to metabolic activity [
28]. The vein is always selected as sampling item in dairy cows’ studies [
29]. However, the artery is more sensitive to reflect the AA metabolism in dairy cows than vein [
10]. Thus, the results based on arterial and venous metabolomics from this study confirmed that the artery is very effective in identifying dietary differences. On the other hand, most of the DMEs identified in our study showed a higher q-value above 0.05. This might be mainly due to the insufficient number of N = X cows used in each group. Thus, more replicates should be included in future studies in analyzing plasma metabolome. In addition, another limitation of our study is that the blood was sampled in the morning before feeding, which may have affected the metabolites obtained in our study. Therefore, further study should address the effects of sampling time on the changes of plasma metabolome in dairy cows.