Metabolomic Profile of Personalized Donor Human Milk

Abstract

Human milk could be considered an active and complex mixture of beneficial bacteria and bioactive compounds. Since pasteurization drastically reduces the microbial content, we recently demonstrated that pasteurized donor human milk (DHM) could be inoculated with different percentages (10% and 30%) of mother's own milk (MOM) to restore the unique live microbiota, resulting in personalized milk (RM10 and RM30, respectively). Pasteurization affects not only the survival of the microbiota but also the concentration of proteins and metabolites, in this study, we performed a comparative metabolomic analysis of the RM10, RM30, MOM and DHM samples to evaluate the impact of microbial restoration on metabolite profiles, where metabolite profiles clustered into four well-defined groups. Comparative analyses of DHM and MOM metabolomes determined that over one thousand features were significantly different. In addition, significant changes in the metabolite concentrations were observed in MOM and RM30 samples after four hours of incubation, while the concentration of metabolites in DHM remained constant, indicating that these changes are related to the microbial expansion. In summary, our analyses indicate that the metabolite profiles of DHM are significantly different from that of MOM, and the profile of MOM may be partially restored in DHM through microbial expansion.

Keywords: metabolomics; microbiota; mother’s own milk.

Figure 1

Differentially abundant metabolites. (a,b) Heatmap of the statistically significant (p < 0.05) differential metabolites with fold change >2.0 (Top-80 features). One-way analysis of variance (ANOVA) was performed to compare the concentration of the features in each group. (a) Positive ion detection mode features (b) Negative ion detection mode features. (c–f) Principal Component Analysis plots (c) PC1 vs. PC2 and (e) PC1 vs. PC3 for Positive ion detection mode data (4794 metabolites). (d) PC1 vs. PC2 and (f) PC1 vs. PC3 for Negative ion detection mode (1837 features). The data was auto scaled before PCA was performed. Time points are indicated for samples analyzed before and after 4 h of incubation at 37 °C (T0 and T4, respectively). Color coding: red-DHM; green and blue-MOM; light blue and pink-RM10; yellow and purple-RM30 samples. The shaded ovals are the 95% data ellipses.

Figure 1

Differentially abundant metabolites. (a,b) Heatmap of the statistically significant (p < 0.05) differential metabolites with fold change >2.0 (Top-80 features). One-way analysis of variance (ANOVA) was performed to compare the concentration of the features in each group. (a) Positive ion detection mode features (b) Negative ion detection mode features. (c–f) Principal Component Analysis plots (c) PC1 vs. PC2 and (e) PC1 vs. PC3 for Positive ion detection mode data (4794 metabolites). (d) PC1 vs. PC2 and (f) PC1 vs. PC3 for Negative ion detection mode (1837 features). The data was auto scaled before PCA was performed. Time points are indicated for samples analyzed before and after 4 h of incubation at 37 °C (T0 and T4, respectively). Color coding: red-DHM; green and blue-MOM; light blue and pink-RM10; yellow and purple-RM30 samples. The shaded ovals are the 95% data ellipses.

Figure 2

Volcano plot representing changes in the concentration of features over time in MOM samples for both ion detection modes. A total of 81 features showing a statistically significant difference (p < 0.05) in relative intensity after 4 h of incubation (pink circles), with 69 showing elevated expression over time.

Figure 3

Molecular Networks predicted to be modulated by MOM and RM. The networks were obtained by analyzing the differentially expressed metabolites (Table S3) using IPA. Identified deregulated metabolites involved in the network are represented in green (decrease of concentration) and red (increase of concentration) colors. (a) Network 1: cellular compromise, lipid metabolism, small molecule biochemistry, (b) Network 2: cellular growth and proliferation, (c) Network 3: inflammatory response, and (d) Network 4: nucleic acid metabolism, amino acid metabolism, carbohydrate metabolism, molecular transport.

Figure 4

Identification of the main metabolic pathways that were determined to be impacted by microbial activity in MOM and RM samples using MetaboAnalyst 4.0. The metabolic pathways significantly impacted by fluctuations in concentration of the different features are listed as: (a) phenylalanine, tyrosine and tryptophan biosynthesis (p = 0.042), (b) taurine and hypotaurine metabolism (p = 0.028), (c) alanine, aspartate and glutamate metabolism (p = 0.002), (d) glycine, serine and threonine metabolism (p = 0.002), (e) tryptophan metabolism (p = 0.008), (f) ascorbate and aldarate metabolism (p = 0.028), and (g) arginine biosynthesis (p < 0.001).