Liver plays a central role in supporting the anabolic capacity of the mammary gland. Net hepatic glucose production (3.1 kg/d) of mid-to-late lactating cows is able to meet glucose required for milk lactose synthesis and maintenance [23, 24]. In addition, liver plays dominant roles in determining the ultimate quantity and pattern of metabolites available or milk synthesis [24]. Metabolic function and, thus, energy metabolism of liver responds to a variety of environmental stimuli including fasting or level of feed intake [25], diet composition and productive (physiological) state [23]. Although a number of studies have been conducted to study effects of low-quality forage resources on lactation performance and rumen fermentation [2, 4, 7, 8], there are few data on the response by important organs such as the liver. Thus, we used transcriptomics and bioinformatics in an effort to better capture genome-wide transcriptional responses to feeding low-quality forage (CS) versus high-quality forage (MF).
Feeding CS Reduces Milk Performance
Consistent with a previous study where milk yield of cows fed more alfalfa than those fed corn stover (P = 0.07) decreased [2], in this study milk yield was lower with CS than MF (Table 1). In addition, milk protein content and yield, milk fat yield, and lactose yield were all decreased by CS compared with MF (Table 1). Clearly, a large portion of the decreased milk performance in this study by CS vs. MF was mainly attributed to the lower DMI of CS cows than MF cows [26]. The study of Zhu et al. (2013) showed that corn stover compared with alfalfa led to lower OM degradability in the rumen (53.2 vs. 47.8%, P = 0.01) [2], suggesting longer retention time of undegraded fiber. Thus, the lower DMI in CS vs. MF in this study was likely caused by excess bulk in the rumen.
Pathways in Liver were Extensively Inhibited in CS vs. MF
In this study, all categories and subcategories of the KEGG pathways in liver were overall inhibited to different extents by CS vs. MF (Fig. 1). Furthermore, among the top 20 impacted pathways, in liver tissue of corn stover (CS) compared to mixed forage (MF) uncovered by the DIA, most of the pathways were inhibited (Fig. 2). Data for inhibited pathways indicated an overall downregulated metabolism in liver of CS compared with MF cows, which agrees with results of Sun et al. (2015) in which ruminal fluid and serum metabolite concentrations decreased with a low-forage compared to high-forage diet [4]. Thus, together the data imply a decreased overall metabolism level when low-quality forage is fed.
Low-Quality Forage Inhibited Glycan Biosynthesis and Metabolism
As shown in Fig. 1, the subcategory “Glycan Biosynthesis and Metabolism” was the most impacted and was overall inhibited. Furthermore, among the top 20 most impacted pathways, approximately 25% were related to “Glycan Biosynthesis and Metabolism” with the pathway of “Glycosphingolipid biosynthesis – globo series” being the most impacted (Fig. 2). Glycans are simple or complex polymers composed of monosaccharides [27], and mediate a wide variety of biological processes including cell growth and differentiation, cell − cell communication, immune response, pathogen interaction, and intracellular signaling events [28]. At a molecular level, glycans are often the first points of contact between cells, and they function by facilitating a variety of interactions both in cis (on the same cell) and in trans (on different cells) [29]. Thus, the high perturbation of glycan biosynthesis and metabolism in this study suggests a potential effect of low-quality forage on hepatocyte communication or growth and differentiation, which was also validated by the results of DAVID and ClueGO where the biological process of “cell-cell adhesion” and “positive regulation of multicellular organism growth” were significantly enriched among the downregulated DEG (Fig. 3 and Fig. 4).
Among the top 20 impacted pathways, “Glycosphingolipid biosynthesis – globo series” and “Glycosphingolipid biosynthesis – lacto and neolacto series” were highly inhibited in CS vs. MF (Fig. 2). In addition, “Glycosphingolipid biosynthesis – ganglio series” was also highly impacted, but the change in direction of the DEG involved in the pathway was not consistent, which is embodied in the modest direction of the impact (Fig. 2). However, it was evident that glycosphingolipid biosynthesis metabolism was overall inhibited by CS vs. MF in this study. Glycosphingolipids (GSLs) comprise a heterogeneous group of membrane lipids formed by a ceramide backbone covalently linked to a glycan moiety [30], and are classified based on their carbohydrate structure into six major series in vertebrates including gangliosides, lacto-, neolacto-, muco-, isoglobo-, and globo-series GSL [31]. D'Angelo et al. (2013) compiled published data indicating that GSL could modulate various aspects of the biology of the cell including apoptosis, cell proliferation, endocytosis, intracellular transport, cell migration and senescence, and inflammation [30]. Zhang et al. (2004) concluded that specific changes in composition and metabolism of GSL occur during cell proliferation, cell cycle phases, brain development, differentiation, and neoplasia in various cell types [32]. In addition, GSL form “microdomains” or “rafts” within the cell membrane, which move within the fluid bilayer as platforms for the attachment of proteins during signal transduction and cell adhesion [32]. Therefore, in this study, the inhibited glycosphingolipid biosynthesis metabolism seems to offer further proof that the communication or growth and differentiation of hepatocytes was potentially inhibited by the low-quality forage diet. The significance of the perturbation at a deeper level could not be ascertained by the results of the present study.
Inconsistent with the above 4 pathways, “Glycosaminoglycan biosynthesis - keratan sulfate” was highly activated by CS vs. MF (Fig. 2). Keratan sulfate (KS) is one of the glycosaminoglycans (GAG), occurring as keratan sulfate proteoglycans on the cell surface and in the extracellular matrix [33]. Pomin (2015) concluded that GAG displays anti-inflammatory functions by activating leukocyte rolling along the endothelial surface of inflamed sites and also regulating chemokine action on leukocyte guidance, migration and activation [34]. The study of Vailati-Riboni et al. (2016) in transition cows demonstrated that feeding at 125% of nutrient requirements activated hepatic GAG synthesis pathways in under-conditioned cows, while it inhibited it in optimally-conditioned cows [35]. Thus, it was suggested that overfeeding for fatter cows may decrease the synthesis of anti-inflammatory compounds and consequently induce some detrimental effects [35]. Taken together, previous and present data suggest activation of “Glycosaminoglycan biosynthesis - keratan sulfate” in response to feeding CS as an anti-inflammatory response. This idea was also validated by DAVID analysis where “complement activation” and “Complement and coagulation cascades” were significantly enriched with up-DEG (Fig. 3). However, the underlying mechanisms could not be ascertained from results of the present study.
The biosynthesis of KS is often markedly altered in response to metabolic, pathologic, or developmental changes in tissues [36]. Davies et al. (1999) suggested that the expression of keratan sulfate is down-regulated in migrating corneal endothelial cells, while abundance on the cell surface returns when cells cease migration [37]. Thus, this suggests that KS has an anti-migration character. However, the anti-adhesive properties of KS were previously reviewed by Caterson and Melrose (2018) and Funderburgh (2000) [36, 38]. Therefore, the exact function of KS as it relates to cell-cell adhesion in hepatocytes is difficult to ascertain with the available data. In the present study, the paradoxical effect of CS vs. MF on cell adhesion was also highlighted by results of DAVID, where both “cell-cell adhesive” and “negative regulation of cell-matrix adhesion” were significantly enriched by the down- and up-regulated DEG, while “cell adhesive” was significantly enriched among the up-regulated DEG (Fig. 3).
Low-Quality Forage Inhibits Amino Acid Metabolism
Metabolism of amino acids was overall inhibited by low-quality forage (Fig. 1). Among the top 20 impacted pathways, “Arginine biosynthesis”, “Selenoamino acid metabolism”, “beta-Alanine metabolism”, and “Tryptophan metabolism” were all inhibited (Fig. 2). Similar results were also revealed by ClueGO where “amide biosynthesis process” and “positive regulation of proteasomal protein catabolic” were significantly enriched by downregulated DEG (Fig. 4). Sun et al. (2015) studied metabolite profiles from four biofluids (rumen fluid, milk, serum, and urine) of cows fed different forage resources using metabolomics, with 55, 8, 28, and 31 significantly different metabolites identified in the rumen fluid, milk, serum, and urine, respectively [4]. These metabolites were involved in glycine, serine, and threonine metabolism; tyrosine metabolism; and phenylalanine metabolism [4]. Sun et al. (2016) in a subsequent urine metabolomics analysis demonstrated that Tyr metabolism and Phe, Tyr and Try biosynthesis pathways had the most variation when corn stover replaced alfalfa hay [39]. The study of Wang et al. (2018) showed that cows fed CS had lower absorbable Leu in the duodenum, which suggested this diet led to shortage of microbial Leu [8]. Sun et al. (2015) demonstrated that, under different quality forage resources, the concentrations of Phe and Tyr in rumen fluid exhibited lower fold-change values (0.54 and 1.19, respectively) than those in the serum (1.01 and 1.34, respectively), which implied that Phe and Tyr may be utilized more in the liver of cows fed high-quality forage than compared with low-quality forage [4]. Therefore, we speculate that the inhibition of amino acid metabolism by CS vs. MF in this study was suggestive of an inhibited amino acid utilization in liver in the cows fed low-quality forage diet.
Co-Expression Network Analysis
In the co-expression network, degree represents the number of connections of a node in a network and betweenness centrality is the number of times that a path passes through the node, which represents the influence this node exerts over other nodes and their potential interactions in the network [41]. Thus, both degree and betweenness centrality are measures of the function of a node in network connectivity [22]. As shown in Fig. 5, the co-expression network revealed 7 genes (FAM210A, SLC26A6, FBXW5, EIF6, ZSCAN10, FPGS, ARMCX2) with higher degree and betweenness centrality (ranking in top 7, Additional file 5) than others, indicating a more critical role played by them in the network.
Among the 7 genes, FAM210A (a mitochondrial gene) which had the highest degree has a crucial role in regulating bone structure and function [42]. SLC26A6 belongs to the solute carrier 26 family, and encodes a protein involved in transporting chloride, oxalate, sulfate and bicarbonate [43–47]. Thus, the inhibited expression of SLC26A6 indicated a decreased transporting ability of chloride, oxalate, sulfate and bicarbonate in liver of CS cows (Fig. 5). FBXW5 is a the TSC2 binding receptor of CUL4 E3 ligase complex [48]. Hu et al. (2008) demonstrated that FBW5 (FBXW5) promotes ubiquitination of tumor suppressor TSC2 by DDB1-CUL4-ROC1 ligase, and depletion of FBW5 stabilizes TSC2 [49]. Ha et al. (2014) demonstrated that intracellular accumulation of TSC2 inhibits the activity of mTOR and increase autophagy [48]. Thus, in the present study, the downregulated FBXW5 seems to imply an intracellular accumulation of TSC2 and consequently an increase in autophagy in liver of CS vs. MF. EIF6 is eukaryotic translation initiation factor. Depletion of eIF6 (using specific siRNA-mediated knockdown) in Mz-ChA-2 and TFK-1 cell lines inhibit cell proliferation and induced apoptosis [50], while EIF6 over-expression increases the motility and invasiveness of cancer cells [51]. However, in this study, “positive regulation of apoptotic process” was significantly enriched by downregulated DEG implying that apoptosis was not induced by low-quality forage diet (Fig. 4). In addition, “protein folding” was significantly enriched by downregulated DEG (Fig. 4) and the pathway “Aminoacyl-tRNA biosynthesis” was highly inhibited by CS vs. MF. Thus, combined with the inhibited EIF6, the results of the present study suggested an inhibited protein synthesis in liver of cows fed low-quality forage.
FPGS is a gene encoding the folylpolyglutamate synthetase enzyme. This enzyme has a central role in establishing and maintaining both cytosolic and mitochondrial folylpolyglutamate concentrations and, therefore, is essential for folate homeostasis and the survival of proliferating cells [52, 53]. Folate plays an essential role in nucleotide biosynthesis and biological methylation reactions as a methyl donor [54, 55]. Consistent with the downregulation of FPGS (Fig. 5), in this study, “Folate biosynthesis” was also inhibited by CS vs. MF (Fig. 2). Taken together, the results of the present study suggested a decrease in folate homeostasis in cows fed low-quality forage. Therefore, the downregulation of “nucleotide-excision repair” may be a downstream cascade reaction due to decreased folate homeostasis (Fig. 3). In addition, the downregulated ZSCAN10 and ARMCX2 with high degree and betweenness centrality were also unraveled by co-expression network analysis (Fig. 5), but the significance of the perturbation was unclear.
Annotation information analysis for the genes within the co-expression network was performed using ClueGO and was shown in Fig. 6. A potential explanation for the perturbation of all these biological processes is beyond the scope of the present study. However, it is noteworthy that almost all these biological processes are energy-requiring except for “pyruvate metabolic process”, which is highly associated with energy metabolism. Furthermore, “pyruvate metabolic process” was also significantly enriched by the whole downregulated DEG in CS vs. MF (Fig. 4), indicating that energy metabolism in liver was inhibited by low-quality forage. Taken together, the inhibited energy metabolism unraveled by this study was suggestive of a central role in the whole metabolic perturbation in liver. The DMI of CS cows was indeed 4 kg/d lower than the MF cows, which may account at least in part for the decreased energy metabolism in liver in CS vs. MF.