After the data processing, 633 metabolites were kept for further analysis: 226 from the positive-early dataset, 71 from the positive-late, 241 from the negative and 95 from the polar, covering a wide range of biochemical compounds (Supplementary Table S1). Previous studies performed on these lines suggested that the biological mechanisms that led to greater or lower IMF content in each line are not entirely the same (Zubiri-Gaitán et al., 2022b, 2023). Hence, in this study, a three-way approach was used to analyse the caecum metabolomics data: the PLS-DA to extract evidence regarding the pathways differentiating the lines, and the PLS-H and PLS-L to extract evidence regarding the specific pathways governing the IMF deposition within each line.
Microbial metabolic pathways affected by selection
The classification of the relevant metabolites according to their chemical nature and the metabolic pathways in which they are involved was performed based on the information available in the Human Metabolome Database and the Kyoto Encyclopaedia of Genes and Genomes database and is summarized in Figs. 2 and 3.
Lipids were the most abundant metabolites detected in all three analyses, although they had a greater relative importance in the PLS-L, representing over 50% of the relevant metabolites. A more complex microbial metabolism was related to IMF deposition in the H line, as evidenced by the greater number of microbial pathways detected with the PLS-H analysis. The results from the PLS-DA showed that the most representative pathways differentiating between lines were those of purines and pyrimidines (i.e., nucleotides), secondary bile acids, and of several amino acids, from which aromatic amino acids (AAA) and branched-chain amino acids (BCAA) stand out. Additionally, many relevant metabolites were food components (i.e., xenobiotics), dicarboxylate fatty acids, dipeptides, acyl carnitines, gamma glutamyl amino acids, and phosphatidylcholines (Fig. 3a). The results from the PLS-L (Fig. 3b) and PLS-H (Fig. 3c) suggested that the lysophospholipids content in the caecum is related to the IMF content in both lines. However, the caecum content of nucleotides’ metabolites, secondary bile acids, and xenobiotics seemed more closely related to the IMF content in the H line, while that of the long chain fatty acids seemed more closely related to the IMF content in the L line.
The PLS-DA identified greater abundance in the caecum of the H line of the aromatic amino acids (AAA) tryptophan (0.76 SD, P0 = 1) and tyrosine (1.16 SD, P0 = 1), together with tryptophan’s degradation products indole (1.02 SD, P0 = 1) and indoleacetate (0.49 SD, P0 = 0.96), while the 5-hydroxyindoleacetate (5-HIAA) was greater in the L line (-0.54 SD, P0 = 0.97). The tryptophan metabolism was also identified in the PLS-H, but not in the PLS-L. The previously mentioned indoleacetate was positively related to the IMF content in the H line (β > 0; P0 = 1), together with tryptophol (β > 0; P0 = 0.99), kynurenine (β > 0; P0 = 0.99) and anthranilate (β > 0; P0 = 0.97). Similar results regarding the kynurenine were found in the plasma metabolomic study performed on these lines (Zubiri-Gaitán et al., 2023).
The metabolism of BCAA (valine, leucine and isoleucine) was also affected by selection, evidencing differences between lines and linear relationship with IMF in both lines. The Nacetylisoleucine was more abundant in the caecum of the L line (-0.51 SD, P0 = 0.96), and it was also negatively related to the IMF content in the mentioned line (β < 0; P0 ≥ 0.99). Similar to what it was found in plasma (Zubiri-Gaitán et al., 2023), the isobutyrylcarnitine was greater in the H line (0.37 SD, P0 = 0.91) and, even though there was no evidence of differences between lines, the isovalerylglycine was negatively associated to IMF in the L line (β < 0; P0 = 0.98). Additionally, glycine, an amino acid whose availability has been shown to be related to the BCAA, was also negatively associated to IMF in the L line (β < 0; P0 = 0.88).
Also related to the metabolism of proteins and amino acids, most of the dipeptides and polypeptides found were more abundant in the caecum of the H line (from 0.41 to 0.89 SD, P0 ≥ 0.94), whereas the gamma-glutamyl amino acids were more abundant in the caecum of the L line (from − 0.55 to -0.62 SD, P0 ≥ 0.97). The gamma-glutamyl amino acids were also found in the PLSH and PLS-L, evidencing a negative relationship with the IMF content (β < 0; P0 ≥ 0.93), the same relationship previously found on the plasma of these lines (Zubiri-Gaitán et al., 2023). Additionally, 2hydroxybutyrate, a metabolite related to glutathione metabolism and consequently to gamma-glutamyl amino acids, was positively related to IMF in both H and L lines (β > 0; P0 ≥ 0.94).
The caecum of the L line showed greater abundance of numerous lipids, including mevalonate (-0.50 SD, P0 = 0.96), involved in the metabolism of cholesterol, 4-cholesten-3-one (-0.42 SD, P0 = 0.93), a cholesterol derivative formed in the gastrointestinal tract, sphingosines, phosphatidylethanolamines, phosphatidylcholines, and other metabolites related to phospholipid metabolisms (from − 0.36 to -0.72 SD, P0 ≥ 0.89). Several secondary bile acids were greater in the L line (from − 0.19 to -0.52 SD, P0 ≥ 0.76), similar to what it was found in the plasma of these lines (Zubiri-Gaitán et al., 2023), with the only exception of the 7-ketodeoxycholate, which was greater in the H line (0.82 SD, P0 = 1). Additionally, numerous secondary bile acids metabolites were positively related to the IMF content only in the H line (β > 0; P0 ≥ 0.95).
The lysophospholipids, which were not relevant to the discrimination between lines, were instead some of the most representative metabolites related to the IMF content in both lines (β > 0; P0 ≥ 0.94). Similarly, while there were no differences between lines neither in long-chain (LC) saturated fatty acids (LC-SFA) nor branched-chain fatty acids (BCFA), they were found to be positively related to IMF but only in the L line (β > 0; P0 ≥ 0.93).
Finally, the PLS-DA identified greater abundance in the caecum of the H line of purines and pyrimidines metabolites, including dinucleotides, nucleosides, and nucleotides (from 0.54 to 1.07 SD, P0 ≥ 0.97). The hypoxanthine, a product of purine catabolism, was also greater in the H line (0.56 SD, P0 = 0.98), while other products were greater in the L line like urate (-0.45 SD, P0 = 0.95), allantoin (-0.58 SD, P0 = 0.98), and dAMP (-0.40 SD, P0 = 0.92). Dinucleotides, nucleosides and nucleotides were also relevant results of the PLS-H analysis, and they all had a positive relationship with IMF (β > 0; P0 ≥ 0.94).