Human milk cells and lipids conserve numerous known and novel miRNAs, some of which are differentially expressed during lactation. Homo sapiens

Human milk (HM) is rich in miRNAs, which are thought to contribute to infant protection and development. We used deep sequencing to profile miRNAs in the cell and lipid fractions of HM obtained post-feeding from 10 lactating women in month 2, 4, and 6 postpartum. In both HM fractions, 1,195 mature known miRNAs were identified, which were positively associated with the cell (p=0.048) and lipid (p=0.010) content of HM. An additional 5,167 novel miRNA species were predicted, of which 235 were high-confidence miRNAs expressed in ≥3 samples with total reads of >20. HM cells contained more known miRNAs than HM lipids (1,136 and 835 respectively, p0.05), with great similarities in the profile of top 20 known miRNAs. These were largely similar also between the three lactation stages examined, as were the total miRNA concentration, and the number and overall expression of the known miRNAs commonly present in the two HM fractions (p>0.05). Yet, approximately a third of known miRNAs were differentially expressed during the first 6 months of lactation (p<0.05), with more pronounced miRNA upregulation seen in month 4. These findings indicate that although the total miRNA concentration of HM cells and lipids provided to the infant does not change in the first 6 months of lactation, the miRNA composition is somewhat altered, particularly in month 4 compared to months 2 and 6. This may reflect the remodeling of the gland in response to infant feeding patterns, which usually change after exclusive breastfeeding, and thus adaptation to infant needs. Overall design: Human milk samples were collected from 10 women longitudinally, at month 2, 4 and 6 postpartum. Total RNA and miRNA were extracted and quantified from cell and lipid fractions. Next-generation sequencing using Illumina HiSeq 2000 was performed on 45 samples (30 milk cells, and 15 lipid samples) to profile mature miRNAs in the breast milk cellular and lipid fractions longitudinally. Filtered (clean) reads were aligned to the latest version of miRBase 21.0. Also, mirdeep software was used to predict novel miRNAs. DESeq and Linear mixed effect models and ANOVA were used to compare miRNA expression levels between samples.