Untargeted Human Milk Metabolomics

Human milk (HM) is considered the gold standard for infant nutrition. HM contains macro- and micronutrients, as well as a range of bioactive compounds (hormones, growth factors, cell debris, etc.). The analysis of the complex and dynamic composition of HM has been a permanent challenge for researchers. The use of novel, cutting-edge techniques involving different metabolomics platforms has permitted to expand knowledge on the variable composition of HM. Here, the state-of-the-art in untargeted metabolomic studies of HM, with emphasis on sampling, extraction and analysis steps is presented.

human milk;metabolome;sampling;extraction;liquid chromatography–mass spectrometry

1. Introduction

Human milk (HM) has been markedly established as the optimal way of providing infants with the necessary nutrients and bioactive factors for their early development. Many health associations and organisms, including World Health Organization, recommend exclusive breastfeeding for the first six months of life [1]. Health benefits of HM for infants include reduced mortality and morbidity, including sepsis, respiratory diseases, otitis media, gastroenteritis, and urinary tract infections, among others [2]. In addition, studies reporting on long-term benefits of HM consumption such as lower risk of suffering from type 1 diabetes and inflammatory bowel disease or overweight in adulthood emerged [3]. HM may also be associated with a slightly improved neurological outcome as cohort studies report [4], especially in preterm infants [5], although potential confounders must be accounted for [6].
HM composition is dynamic and influenced by several factors including genetics, gestational and infant’s age, circadian rhythm, maternal nutrition, or ethnicity. It provides a series of nutrients such as lipids, proteins, carbohydrates, and vitamins, jointly with a number of bioactive factors that contribute to several physiological activities in the newborn infant as well as to short- and long-term outcomes [7][8][7,8]. Living cells including stem cells, hormones, growth factors, enzymes, microbiota, and even genetic material are part of this vast array of HM components with impact in early development, particularly the immune system [9]. In addition, HM appears to be one of the richest sources of microRNAs [10]. On the other hand, because of the maternal environmental exposure and lifestyle, the presence of some contaminants such as persistent organic pollutants or pharmacologically active substances in HM has been described [11][12][11,12].
Due to its complex composition, the analysis of HM is not straightforward. While the advent of “omics” approaches has offered valuable insights into the composition of this unique biofluid, untargeted metabolomic and lipidomic studies have only recently been applied to HM [13]. The comprehensive study of the HM metabolome, which includes the intermediate and end products of metabolism, can shed light on maternal status or phenotype [14][15][14,15]. The generation, analysis, and integration of large and complex data sets obtained in metabolomic studies go hand in hand with the following challenges: (i) the intrinsic complexity of the sample: a rich variety of jointly present, structurally heterogeneous compounds at concentrations that strongly vary covering several orders of magnitude; (ii) pre-analytical steps related to sampling, storage, and pre-processing (e.g., extraction, clean-up); and (iii) the diversity of platforms currently available including nuclear magnetic resonance (NMR), as well as gas chromatography (GC), liquid chromatography (LC), and capillary electrophoresis (CE) coupled to mass spectrometry (MS). The analysis of the HM metabolome has been approached employing a variety of extraction and analytical techniques to respond to a spectrum of clinically relevant questions. Several studies have compared HM metabolome with formula milk [13][16][17][18][19][20][13,16,17,18,19,20] or with milk from other mammalian species including monkey [21], donkey [17], and cow [18], whereas others have made efforts in defining the metabolome of preterm milk [13][16][22][23][24][25][26][13,16,22,23,24,25,26] and the evaluation of the HM metabolome during the course of lactation [15][23][27][28][29][30][15,23,27,28,29,30]. Furthermore, the influence of maternal diet [14][15][31][14,15,31], phenotype [14][32][14,32], obesity [30], or atopy status [33], as well as geographical location [33][34][33,34], time of the day [29][35][29,35], chemotherapy [36], or preeclampsia during pregnancy [31] on the HM metabolome have been reported.
2. Metabolite Extraction from HM
HM is a biofluid characterized by a dynamically varying composition according to several factors including lactation time, time of the day, throughout each feed, maternal status, and the environmental exposure. Although compositional variations have been mainly studied regarding the protein content of HM [37][42], changes of other compound classes such as fat or vitamins have been also reported [38][39][43,44]. Considering the intrinsic variability of HM, the complexity of obtaining representative HM samples is not negligible. Sources of variation related to sample manipulation and compositional variation can be minimized using standard operational procedures (SOPs). SOPs are fundamental to maintain quality assurance (QA) and quality control (QC) process and facilitate repeatable and reproducible research within and across laboratories. However, biologically meaningful results across studies will only be obtained if several key factors during the sample collection process are successfully controlled. This is of special importance in untargeted approaches, where the interpretation of results is especially challenging, and confounding factors introduced by a non-exhaustive sampling protocol can be wrongly attributed to differences between subjects of a studied population. Conversely, biologically meaningful information can be missed or remain unnoticed due to unwanted bias introduced during sample collection. For metabolite extraction from HM, an array of methods has been reported. An overview of the employed approaches is shown in Figure 13. The selection of the extraction method is conditioned by the study objective and the subsequent analysis method. As in other untargeted metabolomics workflows, for HM metabolomics, the selected sample preparation approach should enable a high degree of metabolome coverage while making the sample matrix compatible with the analytical platform. Other considerations might include the available amount of sample volume and the use of one sample extraction procedure for subsequent analysis by multiple, complementary analytical platforms [13][27][28][13,27,28].
Figure 13.
 Sample preparation approaches employed in human milk (HM) metabolomics.
Liquid-liquid extraction (LLE) is the classical extraction method employed in metabolomics and lipidomics. This method, developed by Folch et al. [40][57] in 1957, uses a chloroform-methanol mixture (2:1, v/v), which results in two differentiate phases: an upper phase containing polar metabolites and a lower phase containing nonpolar metabolites. Subsequently, in 1959 Bligh and Dyer [41][58] developed a modified method using a miscible chloroform-methanol-water mixture and later separated into two phases by adding chloroform or water. Both approaches enable the separation of polar and nonpolar metabolites, thus, allowing the analysis of a wide range of metabolites and making them compatible with several analytical platforms. While the use of Bligh and Dyer LLE is widely extended for HM metabolomics studies (see Table 1) [13][16][17][18][19][24][25][29][32][13,16,17,18,19,24,25,29,32], only Andreas et al. [28] used a modified Folch extraction protocol for processing HM samples.
Table 1.
 Sample preparation steps and platforms employed in untargeted analysis of HM metabolome.