All experimental protocols were approved by the Animal Ethical and Welfare Committee (AEWC) of the Sichuan Agricultural University Academy of Sciences (approval no. 20180601). This study was conducted in accordance with the Chinese Guidelines for Animal Welfare. This study was conducted at the farm of the Sichuan Agricultural University (Ya'an, Sichuan Province, China). In this study, a total of 25 Yunnan semi-fine hair wether sheep, with similar BW (33.30 ± 1.77 kg), were randomly assigned to five groups. Each group of sheep was reared in a pen. The sheep in the five groups were fed diets with five metabolic energy levels as follows: 8.0, 8.6, 9.2, 9.8, and 10.4 MJ/kg. According to the standard energy requirement of Yunnan semi-fine hair sheep (BW: 33.3 kg and ADG: 80 g/day) (18), these five groups were distinguished as follows: low energy (LE; 86% energy requirement), medium–low energy (MLE; 93% energy requirement), medium energy (ME; 100% energy requirement), medium–high energy (MHE; 107% energy requirement), and high energy (HE; 114% energy requirement). The ingredients and chemical compositions of each diet are provided in Table 1.

The trial lasted 45 days, with the first 15 days being the preliminary period for the sheep to adapt to the diets and 30 days being the formal trial period. The sheep were fed twice daily at 8:00 AM and 6:00 PM and had free access to water. The criteria for euthanizing the sheep before the experiment included anesthesia and neck bleeding.

On the 30th day of the formal trial, three sheep from each group were selected to be weighed and anesthetized, which were eventually euthanized. Subsequently, the whole liver was separated and weighed. The left liver was collected and stored in 4% paraformaldehyde and PBS for liver section staining. The right liver was collected in a 30-ml Eppendorf (EP) tube for later measurement.

The data for the final BW and liver weight (LW) of sheep were recorded, and the liver index (LI) was calculated as LI = LW/BW (%).

Liver tissues were fixed with 10% formalin; they were embedded, sliced, and stained with Oil Red O. A microscopic imaging system was used to capture pictures. A total of three pictures were captured for each sample. Image-Pro Plus 6.0 software was used for analyzing the average optical density (AOD) of the positive results of Oil Red O staining.

The enzyme activity of alanine aminotransferase (ALT), aspartate aminotransferase (AST), malondialdehyde (MDA), protein carbonyl (PCO), superoxide dismutase (SOD), catalase (CAT), and glutathione peroxidase (GSH-Px) and the total antioxidant capacity (T-AOC) of liver samples were determined using enzyme-linked immunosorbent assay (ELISA) kits (19), purchased from Jiangsu Meimian Industrial Co., Ltd (Jiangsu, China).

Inflammation-related gene expression was determined using reverse-transcription polymerase chain reaction (RT-PCR). Total RNA was extracted from the liver tissues using the TRIzol RNA kit (Takara, Japan), catalog no. 15596026. Then its purity and content were measured. cDNA was obtained through reverse transcription of RNA samples with the TaKaRa Reverse Transcription Kit. Real-time PCR (qPCR) primers were designed using Primer 3 online software (Supplementary Table 1), and gene expression was determined using an ABI 7900 fluorescent quantitative PCR instrument using the SYBR (Takara) dye. The reaction conditions were as follows: 95°C for 30 s, followed by 40 cycles of amplification (95°C for 5 s and 60°C for 34 s) with melting phase set to 95°C for 15 s, 60°C for 1 min, and 95°C for 15 s. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as an internal reference gene, and its expression level was analyzed using the 2^−ΔΔCt method.

Samples for metabolites were extracted from liver tissues (20). LC-MS/MS analyses were performed using a Vanquish UHPLC system (Thermo Fisher Scientific) coupled with an Orbitrap Q Exactive HF-X mass spectrometer (Thermo Fisher Scientific) in both positive and negative modes. The raw data files generated by UHPLC-MS/MS were processed using the Compound Discoverer 3.1 (CD3.1, Thermo Fisher) to perform peak alignment, peak picking, and quantitation for each metabolite.

Principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) were performed at metaX (a flexible and comprehensive software for processing metabolomics data). Variable importance in the projection (VIP) is the importance of the variables to the model and describes the overall contribution of each variable to the differences between groups. We applied univariate analysis (t-test) to calculate the statistical significance (p-value). The metabolites with a VIP > 1, p-value < 0.05, and fold change (FC) ≥2 or ≤0.5 were considered to be differential metabolites.

As differential metabolites were selected, the metabolic pathways for these differential metabolites were obtained from the Kyoto Encyclopedia of Genes and Genomes database (http://www.genome.jp/kegg).

Statistical analyses except metabonomics were performed using the one-way ANOVA procedure of the SPSS statistical software (version 17.0, SPSS Inc., Chicago, IL, USA). Duncan's multiple-range test was used to compare the differences among five groups. The data were expressed as mean ± SEM. A p-value <0.05 was set for significant differences. A value of 0.05 < p < 0.10 was set for the trend of significant differences.

The LW and LI of Yunnan semi-fine wool sheep are provided in Table 2. With an increase in dietary energy levels, LW and LI of the sheep were increased and subsequently stabilized (p < 0.05). There were no significant differences among HE, MHE, and HE groups.

The results of Oil Red O staining of the liver are provided in Figures 1A–E. Lipid droplets in hepatic cells were represented in an orange-red color (bar = 50 μm, × 200). The concentration of lipid droplets was higher in the LE, MLE, and HE groups than in the ME and MHE groups. The quantitative analysis of AOD values (Figure 1F) revealed that dietary energy levels had a significant effect on the AOD of liver lipid droplets (p < 0.05), and the AOD values of LE and HE groups were significantly higher than those of the ME and MHE groups.

The assessment of hepatic injury was based on transaminase activity, oxidative damage, and inflammation.

The assessment of transaminase activity in the liver is provided in Figure 2. The ALT enzyme activity in the liver decreased and subsequently increased with an increase in dietary energy levels (p < 0.05). The ALT enzyme activity of the ME group was significantly lower than that of the LE and HE groups (p < 0.05; Figure 2A). Dietary energy levels had no significant effect on the liver AST enzyme activity (p > 0.05; Figure 2B).

The assessment of oxidative injury in the liver is provided in Table 3. With an increase in dietary energy levels, the SOD and GSH-Px enzyme activities increased linearly (p < 0.05). MDA concentration in the ME and MHE groups was significantly lower than that in the remaining three groups (p < 0.05). PCO concentration in the ME group was significantly lower than that in the HE group (p < 0.05). CAT concentrations in the LE and ME groups were significantly higher than those in the remaining three groups (p < 0.05).

The assessment of the relative expression of inflammation-related genes in the liver is provided in Figure 3. With an increase in dietary energy levels, the expression of toll-like receptor-2 (TLR-2) and tumor necrosis factor-α (TNF-α) in the liver was downregulated and subsequently upregulated (p < 0.05; Figures 3F,H). The expression of interleukin-6 (IL-6) gene in the liver was upregulated in the MLE and HE groups, followed by LE, ME, and MHE groups (p < 0.05; Figure 3B). The dietary energy level had no significant effect on the expression of IL-6, nuclear factor-kappa B (NFKB), TLR-3, TLR-4, and nucleotide-binding oligomerization domain-containing protein 2 (NOD-2) in the liver (p > 0.05; Figures 3C–E,G).

As demonstrated in Supplementary Figure 1, the overlap degree of total ion chromatography (TIC) of quality control (QC) samples was significantly high in both positive (Supplementary Figure 1A) and negative (Supplementary Figure 1B) modes, indicating that the LC-MS/MS system was highly stable.

All liver samples were subjected to PCA as demonstrated in Figure 4. In the negative ion mode (Figure 4A), the five groups were separated from each other, and in the positive mode (Figure 4B), the five groups were separated more clearly.

PLS-DA was performed to analyze the differences among the groups. The LE, MLE, MHE, and HE groups were compared with the ME group for PLS-DA. The results of PLS-DA in the positive and negative modes are provided in Supplementary Figures 2, 3, respectively.

Compared with the ME group, 67, 51, 50, and 54 differential metabolites were detected in the LE, MLE, MHE, and HE groups, respectively, in the positive mode (Figure 5A). Similarly, 75, 26, 15, and 54 differential metabolites were detected in the LE, MLE, MHE, HE, and ME groups, respectively, in the negative mode (Figure 5B). The metabolites related to hepatic metabolism and health are listed in Table 4.

The metabolic pathway analysis by the KEGG database and literature showed that these differential metabolites were involved in various biochemical pathways, such as amino acid metabolism, glycolipid metabolism, bile acid metabolism, nucleotide metabolism, and energy metabolism. In order to visualize the correlation between these metabolites, the results were finally combined, and a metabolic network diagram was drawn (Figure 6).

We found that the MHE and HE groups exhibited a significantly higher concentration of glucose 6-phosphate than that exhibited by the ME group. The concentration of N-acetyl-d-galactosamine (a precursor of fructose 6-phosphate) was significantly lower in the LE group than in the ME group, which was consistent with previous studies, suggesting that the glycolysis pathway is enhanced in the presence of adequate energy substrates (28). The concentration of glucose 6-phosphate was also significantly lower in the MLE group than in the ME group, which may be due to inflammation. The proinflammatory cytokines IL-6, TNF-α, and IL-1β were found to enhance glycolysis (29), which was consistent with the assessment of inflammation-related gene expression.

The liver tissues of the LE, MLE, and HE groups exhibited more severe lipid deposition. Metabolomic results revealed that the concentration of hexanoic acid, as a free fatty acid (FFA), was significantly higher in the LE group than in the ME group, confirming that liver lipid deposition in the LE group may be caused by the reesterification of FFA in the liver owing to lipid mobilization, which results from insufficient energy intake. The assessment of acetoacetic acid, one of the intermediate metabolites of fatty acid oxidation in the liver (30), and acetic acid yielded similar results, suggesting that they supplied energy through fatty acid oxidation in the LE group. Owing to a sufficient energy intake in the HE group, the level of fatty acid oxidation in the liver should be lower (31), and lipid synthesis metabolism should be enhanced (32). As a result, lipid deposition was higher in the HE group than in the ME group. As amounts of very low-density lipoproteins (VLDLs) were restricted, their ability to export triglycerides is extremely limited. Therefore, sheep fed diets with both high and low energy levels are at a higher risk for developing fatty liver disease.

In this study, the three intermediate metabolites of the tricarboxylic acid cycle, namely, citric acid, malic acid, and adenylosuccinic acid (one of the precursors of ferredoxin), exhibited significantly higher concentrations in the LE group than that exhibited by the ME group, which is inconsistent with Wu's study (30). Combined with lipid metabolism, this may be due to the higher concentration of acetyl coenzyme generated by FFA β-oxidation in the LE and MLE groups.

Nicotinamide adenine dinucleotide (NAD+), guanosine diphosphate (GDP), and adenosine diphosphate (ADP) are involved in energy metabolism (33). The concentrations of NAD+, GDP, and ADP were significantly higher in the HE group than those in the ME group, which indicates that the sheep in the HE group were more likely to exhibit metabolic injury compared with those in the ME group owing to a high energy intake. High dietary energy levels significantly enhanced the levels of substrate energy metabolism and bio-oxidation (34, 35). In addition, it has been demonstrated that some metabolic intermediates produced during the tricarboxylic acid cycle, such as citric acid, are inflammatory signals (36), suggesting that there is a greater possibility of inflammation in the LE group than in the ME group, which is consistent with the assessment of inflammation-related gene expression.

Nucleotides are important components of cells; they are not only involved in the synthesis of genetic material but also play a significant role in energy metabolism, function regulation, and immunity (37). In addition, nucleotides can particularly affect the growth, structure, morphology, and function of the liver. Researchers have demonstrated that exogenous nucleotides or endogenous nucleosides promote the growth of liver cells (38). Uridine, a component of uridine monophosphate (UMP), is naturally produced by the liver. In this study, uridine concentration in the LE, MLE, HE, and MHE groups was significantly lower than that in the ME group, indicating that the liver of the ME group exhibited a better growth potential, which is consistent with the assessment of liver growth provided in Table 1. Furthermore, uridine phosphorylase disrupts hepatic pyrimidine nucleotide metabolism by expressing or inhibiting dihydroorotate dehydrogenase, leading to liver steatosis. Uridine supplementation can inhibit liver steatosis caused by dihydroorotate dehydrogenase (39). The results were consistent with the trend of hepatic steatosis scores provided in Table 2, indicating that the degree of hepatic steatosis may be minimum in the ME group.

Bile acid is an important component of bile, which is important for digestion and lipid metabolism (40). In recent years, researchers have demonstrated that bile acids affect and regulate physiological processes such as glucolipid metabolism and inflammatory reaction by activating downstream signals through their receptors (41). It has been reported that primary and some secondary bile acids can inhibit the release of TNF, suggesting that bile acids exert an anti-inflammatory effect (42). Therefore, the LE and HE groups were more likely to exhibit an inflammatory reaction, which is consistent with the assessment of inflammation-related gene expression.

Owing to different dietary energy levels, in addition to the above-mentioned liver metabolites, other differential metabolites are related to hepatic health as well, including glutathione, ergothioneine, p-cresol, betaine, and cortisol.

Glutathione is a tripeptide of glutamic acid, cysteine, and glycine, which contains γ-amide bonds and sulfhydryl groups. It is involved in converting harmful toxic substances into harmless substances (43, 44), thereby maintaining the normal immune functions of an organism (45). In this study, glutathione concentration was significantly lower in the MLE group than that in the ME group, suggesting that the ME group exhibited better immune function. The results were partially consistent with those of antioxidant analysis because glutathione is a component of GSH-Px.

Ergothioneine is a natural antioxidant with anti-inflammatory and cytoprotective effects (46, 47), which is distributed in certain tissues and organs of mammals. Melville reported that ergothioneine is also found in cereal plants (48). In this study, ergothioneine concentration in the ME group was significantly higher than that in the LE and MLE groups, which may be owing to differences in the intake of maize. However, compared with the ME group, ergothioneine concentration did not increase in the MHE and HE groups, suggesting that its uptake was limited. However, further investigation is required. Owing to the natural antioxidant function of ergothioneine, the antioxidant activity in the ME group was better than that in the LE and MLE groups, which is partially consistent with the results of oxidative stress analysis.

Oxoadipate is metabolized by lysine through the zymocin and pipecolic acid pathways. If lysine, tryptophan, and hydroxylysine are metabolized incorrectly, oxoadipate production is greatly increased (49). In this study, oxoadipate concentration was significantly higher in the LE and HE groups than that in the LME and MHE groups, indicating that low or high dietary energy levels may cause disorders in the lysine, tryptophan, and hydroxylysine metabolism. This is not conducive to liver health.