Presented at the Neonatal Society 2012 Autumn Meeting.
Andreas NJ, Garcia-Perez I, Jeffries S, Gale C, Hyde MJ, Holmes E, Modi N
Section of Neonatal Medicine, Chelsea & Westminster Campus, Imperial College London, UK
Background: Breast milk composition is known to vary widely, both temporally (each day, and throughout lactation) and between individual mothers. How maternal characteristics determine breast milk composition is poorly understood. We are currently using a range of high throughput technologies to extract maximal compositional data on breast milk in order to study the variations in milk micronutrient content. 1H NMR provides a robust and repeatable way of determining the metabolites present in breast milk. This metabonomic approach, of untargeted profiling of low molecular weight metabolites, in order to characterise breast milk is novel in the field of breast milk research. In this initial report we set out to describe the method development phase of metabolomic characterisation of human breast milk and some preliminary findings on metabolites identified as changing over the time course of lactation.
Methods: The study had Research Ethics approval (10/H0713/5) and informed consent was obtained from the mothers. Human milk samples (5ml) were obtained from 64 mothers of term infants, between 2 to 80 days postpartum. Samples were stored at -80ºC until analysis. Milk samples were Folch extracted: 400μl of milk was added to a mixture of 2ml chloroform:methanol (2:1 v/v) and vortexed, following which 600μl of demineralised water was added. The sample was then centrifuged (13000RPM, 10mins). The upper (aqueous phase) and the lower (lipid phase) was pipetted off and speed vacuumed to dryness. The aqueous phase was reconstituted in 600μl D2O phosphate buffer (1); the lipid phase in 600μl deuterated chloroform. These samples were then transferred to 5mm capillary tubes and 1H NMR spectra were acquired using a Bruker DRX 600 MHz spectrometer operating at 300 K, using a 1-dimensional NMR pulse sequence – NOESY (nuclear overhauser effect spectroscopy). In total 64 aqueous fractions and 59 lipid fractions were analysed. The resulting 1H NMR spectra were digitalized and imported into Matlab for phasing and baseline-correction. Spectra were normalized to the total area and data was scaled using unit variance scaling. Multivariate data analysis was carried out using SIMCA-P+ 12.0.1. Data was analysed by Principal Component Analysis (PCA) to establish if there was any clustering of the data in relation to the time point of milk collection. Then data was modelled using PLS-DA and OPLS-DA in order to compare different time points to one another.
Results: In the aqueous phase of the milk, metabolites which were found to cause separation of the groups in relation to time of collection post-partum in the OPLS-DA model were identified. Metabolites identified were lactose, glutamine and glycerophosphocholine which were all increased in intensity compared to the median spectra, with increasing time of lactation. Furthermore, citrate and various N-Acetyl carbohydrates caused separation in the OPLS-DA model, decreasing in intensity as lactation proceeded. Compared to the aqueous fraction, there was little change in the lipid metabolite profile over time, as analysed to date.
Conclusion: Results to date have agreed with previously reported literature on the changes in breast milk composition, for example, the increase in glycerphosphocholine content with increasing time of lactatio (2). There appears to be less variation in the hydrophobic phase (mainly lipid) in comparison to the aqueous metabolites in the OPLS-DA models. In order to study the changes in lipid composition of milk, more suitable techniques may include GC-MS, which can resolve more detailed information such as the fatty acid composition of the milk. 1H NMR appears to be a promising technique for the investigation of global milk composition.
Corresponding author: firstname.lastname@example.org
1. Beckonert, O., et al. Nat Protoc, 2007. 2(11): p. 2692-703.
2. Ilcol, Y.O., et al. J Nutr Biochem, 2005. 16(8): p. 489-99.