Maternal diet alters human milk oligosaccharide composition with implications for the milk metagenome

Healthy lactating women organized into two cohorts: Glu/Gal cohort (n=7; 8–11 weeks postpartum; obese) and Carb/Fat cohort (n=7; 9–12 weeks postpartum; lean). Inpatient dietary control with paired diet periods and washouts; milk samples collected for human milk oligosaccharide (HMO) and milk microbiome analyses; secretor status evaluated and non-secretors removed from metagenomic analyses to avoid confounding.

Glu/Gal cohort: isocaloric, isonitrogenous drink containing either glucose or galactose as the sole carbohydrate source; the sugar drink provided ~60% of daily energy requirement; protein supplement Beneprotein and safflower oil (essential fatty acids) added to the drink to provide an additional 10% of energy; total energy from macronutrients ~70% (83% carbohydrate, 10% protein, 7% fat); consumed every 3 hours for 30–57 hours; after a 1–2 week washout, the alternate sugar was provided. Carb/Fat cohort: eight days on each diet, isocaloric and isonitrogenous; carbohydrate diet (60% carbohydrate, 25% fat, 15% protein) or high-fat diet (30% carbohydrate, 55% fat, 15% protein); caloric content set at 1.3 × basal metabolic rate; meals provided as packed meals with snacks and consumed by the participants during the inpatient period.

Maternal diet during lactation acutely alters HMO composition and shifts the milk microbiome’s functional gene content without major changes to taxonomic composition. In the Glu/Gal cohort, total HMO-bound fucose decreased with glucose versus galactose (p = 0.03); some HMOs increased up to ~50% with galactose and ~30% with high-fat diet. HMO-bound fucose concentration correlated with the relative abundance of bacterial fucosidase, notably in Streptococcus spp. (S. mitis, S. pneumoniae, S. oralis) carrying fucosidase genes; in vitro, fucosylated HMOs (2′FL and LNFP I) promoted growth of S. mitis and S. oralis with fucosidase; S. parasanguinis and S. cristatus showed no growth advantage. No consistent association between HMO-bound sialic acid and sialidase abundance was observed in the Glu/Gal cohort. In the Carb/Fat cohort, taxonomic composition remained largely unchanged, but metagenomic pathways (including amino acid biosynthesis) differed by diet. The findings indicate that maternal diet can shape milk HMO profiles and the milk microbiome’s functional capacity, with potential implications for infant gut colonization and maternal health (e.g., mastitis risk); longer-term studies are needed to confirm health outcomes.

Small sample size (n=7 per cohort) limiting statistical power and multivariate analyses; short intervention durations; non-secretor removal from microbiome analyses reduces generalizability; potential low-biomass milk microbiome contamination despite controls; some HMO data gaps due to sample volume; inability to assess gene expression or protein activity from metagenomics; no direct infant health outcomes measured.