Human milk miRNAs primarily originate from the mammary gland resulting in unique miRNA profiles of fractionated milk

Abstract

Human milk (HM) contains regulatory biomolecules including miRNAs, the origin and functional significance of which are still undetermined. We used TaqMan OpenArrays to profile 681 mature miRNAs in HM cells and fat, and compared them with maternal peripheral blood mononuclear cells (PBMCs) and plasma, and bovine and soy infant formulae. HM cells and PBMCs (292 and 345 miRNAs, respectively) had higher miRNA content than HM fat and plasma (242 and 219 miRNAs, respectively) (p < 0.05). A strong association in miRNA profiles was found between HM cells and fat, whilst PBMCs and plasma were distinctly different to HM, displaying marked inter-individual variation. Considering the dominance of epithelial cells in mature milk of healthy women, these results suggest that HM miRNAs primarily originate from the mammary epithelium, whilst the maternal circulation may have a smaller contribution. Our findings demonstrate that unlike infant formulae, which contained very few human miRNA, HM is a rich source of lactation-specific miRNA, which could be used as biomarkers of the performance and health status of the lactating mammary gland. Given the recently identified stability, uptake and functionality of food- and milk-derived miRNA in vivo, HM miRNA are likely to contribute to infant protection and development.

Figure 1. RNA enriched in miRNA in HM cells and fat, and maternal PBMCs and plasma, and associations with HM components.

(A,B) RNA concentration of HM cells, PBMCs, HM fat, and plasma, obtained with NanoDrop 2000 (N) and the Bioanalyzer 2100 (B). (C,D) RNA integrity measured by the Bioanalyzer 2100, and RNA purity (260/280 ratio) using NanoDrop 2000 in all four sample groups. (E,F) Associations between total RNA eenriched in miRNA and HM cell content or maternal blood PBMC content using Bioanalyzer 2100 and Nanodrop 2000. (G) HM fat content (%) and RNA concentration of HM fat (ng). (H,I) Associations between HM volume or maternal blood volume with the total RNA enriched in miRNA. (J) Association between PBMC content of blood and HM cell content.

Figure 2

(A) Ranking for the classified Ct values across each sample run on the OpenArray platform. Reliable miRNAs had 8 ≤ Ct ≤ 29 and were detected in at least 4 samples per group, whilst unreliable miRNAs had markedly low Ct values (<5) or showed excessive variation between replicates. Undetermined miRNAs correspond to a failed assay or a rare miRNA species (Ct > 29). (B) Boxplot showing the number of miRNA species in all mothers and each sample group (n = 10 per sample group). (C) Heat map showing an overview of the Ct quality categories that have been assigned to the different miRNAs. The categories correspond to “Reliable”, “Undetermined” and “Unreliable” with the colour codes “green”, “red” and “yellow”, respectively.

Figure 3

(A) Principal Component Analysis (PCA) showing the distribution and clustering of the individual sample groups. The miRNA content in PBMCs form a tight cluster mainly in the top right of the graph, whilst the plasma samples show a broader distribution suggestive of high inter-individual biological variation. The HM cell and fat miRNAs are clustered together and share a similar space on the graph. (B) Correlation plot demonstrating the relationship between different sample groups. The regions of the graph that are shaded in hotter colours correspond to more similar profiles of miRNA presence and relative abundance. The HM cell and fat samples are all well correlated with each other, whilst a slight correlation between maternal PBMCs and plasma samples could be seen. Two plasma samples were markedly different to all other samples. (C) Differential expression analysis for the reliable miRNAs identified using Linear Models for Microarray Analysis (limma) in the R HTqPCR package. Results are presented in volcano plots, where the log-fold change is plotted on the x-axis and the −log10 (adjusted p value) on the y-axis. The six volcano plots demonstrate a comparison between the four sample groups (HM cells and fat, and maternal PBMCs and plasma) by graphic fold change versus significant (p < 0.05) to exhibit differences in miRNA expression between sample groups.

Figure 4. Shared reliable miRNAs (8 ≤ Ct ≤ 29 and present in at least 4 samples per group) between the four sample groups examined.

The bovine milk- and soy-based formulae are the results of a single assay and the observations are only illustrative (8 ≤ Ct ≤ 35). (A) Euler diagram showing overlapping reliable miRNA species between sample groups. (B) Euler diagram showing the number of reliable miRNA species in the HM cell and fat samples and their overlap with infant formulae. (C) Box plot showing high expression of plant-based miR-159a (16 replicates in each infant formula) in the two formulae tested.

Figure 5. Example of molecular pathway (epithelial-to-mesenchymal transition, EMT) controlled by some of the most highly expressed reliable miRNAs detected in HM (miR-200a/b/c/, miR-429, miR-141, and 205), and how they interact with other genes to regulate EMT.