Untargeted metabolomics study of mature human milk from women with and without gestational diabetes mellitus

Highlights

• MS-based metabolomics to unveil GDM-associated changes in human milk metabolome.
• A total of 268 DEMs associated with GDM were identified.
• Insulin resistance and poor glucose metabolism persist in the post-partum period.
• GDM-associated human milk metabolome may affect optimal nutrition for infants.

Abstract

Gestational diabetes mellitus (GDM) is a prevalent metabolic disorder during pregnancy that alters the metabolites in human milk. Integrated Gas Chromatography-Mass Spectrometry (GC–MS) and Liquid Chromatography-Mass Spectrometry (LC-MS) were employed for comprehensive identification and comparison of metabolites in mature human milk (MHM) from women with and without GDM. A total of 268 differentially expressed metabolites (DEMs) were identified. Among these, linoleic acid, arachidonic acid, 9R-HODE and L-glutamic acid were significantly elevated and 12,13-DHOME was significantly decreased in MHM of women with GDM. These metabolites are significantly enriched in linoleic acid metabolism, fatty acid biosynthesis, galactose metabolism and ABC transporters pathways. Disorders in these metabolic pathways are associated with insulin resistance and poor glucose metabolism indicating these conditions may persist postpartum.

2. Materials and methods

2.1. Materials

Methanol, acetonitrile and formic acid were purchased from Fisher Scientific (Hampton, NH, USA). L-2-chlorophenylalanine was purchased from Shanghai HC Biotech Co., Ltd. (Shanghai, China). Methoxylamine hydrochloride pyridine solution was purchased from Shanghai Macklin Biochemical Co., Ltd. (Shanghai, China). BSTFA was purchased from TCI (Shanghai) Development Co., Ltd. (Shanghai, China). N-hexane was purchased from CNW Technologies GmbH (Dusseldorf, Germany). Methyl octanoate (C8:0) and methyl palmitate (C16:0) were purchased from Dr. Ehrenstorfer GmbH (Augsburg, Germany). Methyl nonanoate (C9:0), methyl decanoate (C10:0), methyl laurate (C12:0), methyl myristate (C14:0), methyl stearate (C18:0), methyl eicosanoate (C20:0), methyl docosanoate (C22:0) and lignoceric acid methyl ester (C24:0) were purchased from Nu-Chek Prep, INC (Minnesota, USA).

2.2. Mature human milk collection

Mature human milk (3–6 months postpartum) was collected from twenty women (GDM mothers: n = 10, non-GDM mothers: n = 10) living in Ningbo, China. The complete breast expression samples were collected using the electronic breast pump, stored in milk storage bags and immediately frozen at −80 °C. All the analysis was conducted within 2 months after milk collection. Milk samples from GDM or non-GDM women were randomly allocated into eight groups to minimize the influence of individual and dietary variations (3 samples per group, each sample can only be allocated into three groups maximum). The demographic information of these mothers and their newborns is presented in Table S1. All women were healthy with no pre-existing or ongoing diseases (e.g. mastitis and breast cancer) and did not engage in any lifestyle behaviour that may affect the growth and development of infants during pregnancy and lactation period (e.g., smoking, alcohol and drug abuse). GDM was diagnosed according to the standard of the International Association of Diabetes and Pregnancy Study Groups (IADPSG) guidelines as follows: fasting plasma glucose (FPG) ≥ 5.1 mmol/L, FPG 1 h post-load glucose ≥10.0 mmol/L or FPG 2 h post-load glucose ≥8.5 mmol/L (Diabetes, I. A. o. and Panel, P. S. G. C, 2010). All participants had signed informed consent and this research was approved by the Medical Ethics Committee of Ningbo University (number: NBU-2021-142; Ningbo, Zhejiang, China).

2.3. GC–MS analysis of human milk metabolome

2.3.1. Mature human milk metabolite extraction and derivatization

Mature human milk samples were thawed at room temperature. The milk samples (150 μL) were mixed with methanol-acetonitrile (450 μL, V: V = 2:1) solution containing 2 μg/mL of L-2-chlorophenylalanine (internal standard) for 1 min. Following that, the mixture was ultrasonicated in an ice water bath for 10 min, and stored at −40 °C for 30 min to precipitate proteins thoroughly. The mixture was then centrifuged at 4 °C at 12000 rpm for 10 min. The supernatant (150 μL) was collected and dried using a centrifugal freeze dryer (Taicang Huamei Biochemical Instrument Factory, Jiangsu, China). Methoxylamine hydrochloride pyridine solution (80 μL, 15 mg/mL) was added to the freeze-dried samples to block carbonyl groups (such as α-ketonic acids and sugars) and reduce the generation of derivative byproducts. The oximation process was performed in a shaking incubator (Shanghai Lichen Bangxi Instrument Technology Co., Ltd., Shanghai, China) at 37 °C for 60 min. Finally, 50 μL of BSTFA (with 1% TMCS, introducing silyl groups), 20 μL n-hexane and 10 μL of internal standard (including C8, C9, C10, C12, C14, C16, C18, C20, C22, C24) were added into the mixture and was derivatized at 70 °C for 60 min. The samples were left at room temperature for 30 min before GC–MS analysis. Equal amounts of the extracted solutions for all samples were mixed as a quality control sample.

2.3.2. GC–MS analysis

The derivatized milk metabolites were analyzed using an Agilent 8890B gas chromatography system coupled to an Agilent 5977B MSD system (Agilent Technologies Inc., CA, USA). The derivative milk metabolites were separated using a DB-5MS capillary column (30 m × 0.25 mm × 0.25 μm, Agilent J & W Scientific, Folsom, CA, USA). The helium flow rate (> 99.999%) was 1.0 mL/min. The injector temperature and injection volume were set to 260 °C and 1 μL, respectively. The GC conditions were set as follows: The initial oven temperature was set at 60 °C and maintained for 0.5 min. Then, the temperature was increased to 125 °C at a rate of 8 °C/min, to 210 °C at a rate of 8 °C/min, to 270 °C at a rate of 15 °C/min, to 305 °C at a rate of 20 °C/min, and finally held at 305 °C for 5 min. The MS conditions refer to previous research (Zhang et al., 2024). The GC–MS data were processed according to the methods of Wang et al. (2023). The GC–MS data were processed using MS-DIAL software for peak detection, peak recognition, MS2Dec deconvolution, characterization, peak alignment, wave filtering and missing value interpolation. The metabolites were identified using the LUG database.

2.4. LC-MS analysis of human milk metabolome

2.4.1. Mature human milk metabolite extraction

The milk samples (150 μL) were mixed with methanol-acetonitrile (450 μL, V: V = 2:1) solution containing 2 μg/mL of L-2-chlorophenylalanine (internal standard) for 1 min. After that, the mixture was placed at −40 °C for 2 h and centrifuged at 13000 rpm and 4 °C for 10 min. The supernatant (150 μL) was filtered through a 0.22 μm microfilter and transferred to an LC vial and stored at −80 °C until LC-MS analysis. Equal amounts of the extracted solutions for all samples were mixed as a quality control sample.

2.4.2. LC-MS analysis

A Dionex U3000 UHPLC coupled to Q-Exactive plus quadrupole orbitrap mass spectrometer equipped with a heated electrospray ionization source (Thermo Fisher Scientific, Waltham, MA, USA) was used for metabolic profiling analysis in both ESI positive and ESI negative ion modes (a type of MS scanning mode). ACQUITY UPLC HSS T₃ column (100 mm × 2.1 mm, 1.8 μm, Waters, Milford, Massachusetts, USA) was used in both positive and negative modes. The binary gradient elution system consisted of strongly polar water (solvent A, containing 0.1% formic acid, v/v) and weakly polar acetonitrile (solvent B), which could effectively shorten the analysis cycle, improve separation ability, reduce tailing and increase sensitivity. Gradient separations refer to previous research (Ren et al., 2023). The flow rate and column temperature were 0.35 mL/min and 45 °C, respectively. All the samples were kept at 4 °C during the analysis. The injection volume was 5 μL. The LC-MS data were processed according to the method by Zhang, Tan, et al. (2023). The raw LC-MS data were processed using MS-DIAL and Progenesis QI V2.3 for baseline filtering, peak identification, integral, retention time correction, peak alignment and normalization. The metabolites were identified using the Human Metabolome Database (HMDB), Lipidmaps (V2.3), Metlin, EMDB, PMDB and our in-house databases.

2.5. Bioinformatic analysis

Metaboanalyst 5.0 and the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway software database (KEGG PATHWAY Database (genome.jp) were used for metabolic pathway analysis. MetaMapp and Cytoscape 3.9.1 were used for pathway mapping.

2.6. Statistical analysis

The GC–MS and LC-MS matrix were imported into R, principal component analysis (PCA) and orthogonal partial least-squares-discriminant analysis (OPLS-DA) were performed. t-test and fold-change values were used to screen DEMs between GDM and non-GDM groups. p < 0.05 (t-test) and Variable Important in Projection (VIP) > 1.0 were considered as significance thresholds. Graphpad prism was used to draw a Venn diagram, volcano plot and heat map. Cytoscape was used to draw an interaction network diagram.

4. Conclusion

Untargeted metabolomic analysis was performed using integrated GC–MS and LC-MS platforms to unveil the differential expressed metabolites in MHM of GDM and non-GDM women. A total of 268 DEMs were identified with a low abundance of 12,13-DHOME and a high abundance of linoleic acid, arachidonic acid, 9R-HODE and L-glutamic acid in MHM of GDM indicating disorder in linoleic acid metabolism, fatty acid biosynthesis, galactose metabolism and ABC transporters pathways. Abnormal levels of these metabolites showed poor glucose metabolism may persist for a longer period postpartum period. Mitigation strategies need to be developed to improve glucose metabolism, prevent transition to type 2 diabetes in GDM women and ensure healthy and optimal growth in infants. This study was limited in terms of sample sizes. However, findings from this study will pave the way for larger prospective cohort study on GDM physiopathology and the health of GDM mothers and babies.