Improving the feed efficiency traits of livestock is of great significance, but it is not easy to estimate. Therefore, any measure that can effectively predict feed efficiency is meaningful for production. Although there is currently much work to study FE at the genetic level, few studies have linked metabolites to feed efficiency phenotypic traits. In this study, we analyzed and compared the metabolites in the feces of pigs in the high-FE and low-FE groups by LC-MC technology and interpretation tools, including WGCNA and Lasso regression. To the best of our knowledge, this is the first report combining these methods to study the metabonomic profile related to feed efficiency and related traits in DLY pigs.
At present, FCR and RFI are commonly used to evaluate FE traits, and it is believed that RFI can better represent feed efficiency [2, 3, 23], which is consistent with our WGCNA analysis results. In our data, we found that the use of powerful tools such as PCA and OPLS-DA were not sufficient to distinguish the different features between the high- and low-FE animals. There are many possible explanations for the unsatisfactory results of PCA and OPLS-DA, including but not limited to (1) the sampling process was carried out after the individual growth indicators were measured. When the pig reaches the slaughter weight (approximately 100 kg), its metabolic activity is often not as active as before, and the increase in weight has little effect on the growth performance of pigs after 100 kg [24]; (2) throughout the experiment, all test subjects were clinically healthy. Therefore, there is no physiological interference between the FE groups that could cause large metabolome differences. (3) The number of animal individuals in our study (25 individuals per group) may not have high statistical power, so more animal groups and more targeted experimental designs may be needed to evaluate feed efficiency in the future. Because the results of the PCA and OPLS-DA models were not ideal, we then adopted WGCNA analysis to select the modules and metabolites most closely related to RFI and FCR. After screening and annotation, we obtained nine metabolites in these models. Based on these metabolites, we identified four pathways from the KEGG database that were also significantly related to feed efficiency, including lipid metabolism (primary bile acid synthesis, linoleic acid metabolism), vitamin D, and glucose metabolism. Moreover, the Lasso regression model showed that all nine annotated metabolites contribute to feed efficiency.
The metabolite 22-OH-THC is a kind of bile alcohol, which is the end product of catabolism of cholestanoic acids [25–27]. Bile alcohol may be regarded as an intermediate and side product from the normal pathways in bile acid biosynthesis [28]. Notably, THC26 and DHCA were mainly involved in the biosynthesis of primary bile acids. The specific synthesis process is that cholesterol 7α-hydroxylase (CYP27A1) catalyzes the oxidation of steroid side chains to form THC26 or DHCA in the mitochondria of liver cells and then obtains the primary bile acid cholic acid (CD) or chenodeoxycholic acid (CDCA) under the catalysis of various enzymes [19, 29–32]. Interestingly, although the synthesis of bile acids is determined by a variety of cytochrome P450 enzymes (CYPs), both THC26 and DHCA are intermediate products catalyzed by CYP27A1 [33]. Bile acids start from the catabolism of cholesterol and are the final product of cholesterol catabolism; they play a critical role in food digestion and nutrient absorption, helping the absorption of lipids and fat-soluble vitamins in the intestine [32, 34–36]. After passing down the intestine with bile, approximately 95% of bile acids are reabsorbed in the terminal ileum and circulate back to the liver through the portal vein [29, 35, 37]. The performance of these functions of bile acid mainly depends on its enterohepatic circulation process, which is of great significance for nutrient absorption and distribution, metabolic regulation and homeostasis [29, 35, 37–39]. The results of metabolite network analysis showed that three metabolites related to bile acid synthesis were significantly negatively correlated with RFI traits, which means that they were positively correlated with feed efficiency. This result was consistent with our finding that these metabolites had higher levels in the high feed efficiency group. At present, little is known about the transport of bile acids and intermediates between different chambers, which may provide some references for understanding the important factors in the synthesis of bile acids.
In addition, (10S)-1α,19,25–trihydroxy-10,19-dihydrovitamin D3 and (22E)-1α-hydroxy–22,23-didehydrovitamin D3 belong to vitamin D, which is a steroid derivative [40, 41]. Vitamin D has various effects on lipid metabolism and immune system function through its effects on nuclear hormone receptors (such as vitamin D receptor and PPARγ) [42, 43]. Similarly, our results were consistent with the effect of vitamin D on lipid metabolism mediated through these receptors. Previous studies have shown that the CYP27A1 enzyme can catalyze the hydroxylation of compounds both in the biosynthesis of bile acids and the bioactivation of vitamin D3 [44–46]. The acidic pathway (or alternative pathway) of bile acid synthesis is initiated by CYP27A1, which is a mitochondrial cytochrome P450 enzyme widely distributed in most tissues and macrophages [29, 47]. CYP27A1 can not only catalyze the 25-hydroxylation of vitamin D3, which is required for the conversion of vitamin D3 into a functionally active form, but may also regulate cholesterol homeostasis by promoting the synthesis of bile acids or producing active oxysterols [46–49]. Although there are no current reports on the effect of adding this enzyme, this may warrant further concern. Additionally, metabolite C24:5n-6 was involved in the alpha linolenic acid and linoleic acid metabolism pathways [50]. Linoleic acid is the main dietary n-6 polyunsaturated fatty acid (PUFA), and livestock mainly obtain it from diets such as vegetable oil, soybeans, and corn [51]. Previous studies reported that higher n-6 PUFA intake can reduce liver fat in overweight individuals, improve liver metabolism, and regulate the balance between fatty acid oxidation and lipid synthesis [52, 53]. In the process of linoleic acid metabolism, linoleic acid is catalyzed by the rate-limiting enzyme fatty acid desaturase-2 (FADS2), and after a series of extensions, C24:5n-6 is produced by FADS2 catalyzed C24:4n-6 [54]. Our results showed that C24:5n-6 correlated negatively and significantly with RFI traits and was significantly higher in the high-FE group than in the low-FE group (p-value = 0.002). Notably, in the process of linoleic acid synthesis, there was no significant difference between the upstream metabolite linoleic acid in the high and low FE groups, while the downstream C24:5n-6 was extremely different in the two groups. Similarly, there is currently no research on adding FASD2 enzyme to feed, but the potential for it to lead to differences in the feed efficiency of livestock cannot be ignored. Based on this, these findings can also provide some references for further analysis of linoleic acid metabolism. Improving feed efficiency is a concern, but these two enzymes and potential metabolic markers deserve further evaluation and research to clarify their biological significance.