Identification of Oligosaccharides in Feces of Breast-fed Infants and Their Correlation with the Gut Microbial Community

Abstract

Glycans in breast milk are abundant and found as either free oligosaccharides or conjugated to proteins and lipids. Free human milk oligosaccharides (HMOs) function as prebiotics by stimulating the growth of beneficial bacteria while preventing the binding of harmful bacteria to intestinal epithelial cells. Bacteria have adapted to the glycan-rich environment of the gut by developing enzymes that catabolize glycans. The decrease in HMOs and the increase in glycan digestion products give indications of the active enzymes in the microbial population. In this study, we quantitated the disappearance of intact HMOs and characterized the glycan digestion products in the gut that are produced by the action of microbial enzymes on HMOs and glycoconjugates from breast milk. Oligosaccharides from fecal samples of exclusively breast-fed infants were extracted and profiled using nanoLC-MS. Intact HMOs were found in the fecal samples, additionally, other oligosaccharides were found corresponding to degraded HMOs and non-HMO based compounds. The latter compounds were fragments of N-glycans released through the cleavage of the linkage to the asparagine residue and through cleavage of the chitobiose core of the N-glycan. Marker gene sequencing of the fecal samples revealed bifidobacteria as the dominant inhabitants of the infant gastrointestinal tracts. A glycosidase from Bifidobacterium longum subsp. longum was then expressed to digest HMOs in vitro, which showed that the digested oligosaccharides in feces corresponded to the action of glycosidases on HMOs. Similar expression of endoglycosidases also showed that N-glycans were released by bacterial enzymes. Although bifidobacteria may dominate the gut, it is possible that specific minority species are also responsible for the major products observed in feces. Nonetheless, the enzymatic activity correlated well with the known glycosidases in the respective bacteria, suggesting a direct relationship between microbial abundances and catabolic activity.

Fig. 1.

Names and graphical representations of the different monosaccharides that compose human milk oligosaccharides (HMOs) and N-glycans. Representative branched and linear neutral and acidic HMOs and the different subclasses of N-glycans, along with their potential linkages are also depicted.

Fig. 2.

(A) Extracted compound chromatograms of a mother's breast milk oligosaccharide profile at day 24 postpartum compared with (B) her infant's fecal glycan profile at day 24 postnatal. The structure found in (A) at retention time 10.3 min is 2′FL, 13.8 min is LNT, and 14.4 min is LNnT with monosaccharides glucose (●), galactose (●), and fucose (▴), all having significantly lower (2′FL) or no abundance (LNT and LNnT) in (B). Sialylated structures found in (A) from 25–26 min show significant decrease in (B) at the same retention time.

Fig. 3.

Tandem mass spectra of oligosaccharides extracted from infant feces reveals they are of human milk oligosaccharide (HMO) and glycoprotein origins. Monosaccharides represented as glucose (●), galactose (●), mannose (●), N-acetylglucosamine (■), and fucose (▴). A, Doubly charged precursor ion with m/z 631.74 shows neutral losses of monosaccharides consistent with an intact N-glycan, including the reduced GlcNAc core. B, Precursor ion with m/z 1440.52 shows neutral losses of mono- and disaccharides consistent with an HMO structure. C, Precursor ion with m/z 913.35 has tandem mass spectrum pattern consistent with an HMO structure, including the lactose core.

Fig. 4.

Overlaid extracted ion chromatograms (EICs) of β-galactosidase BLNG_00015 digestion of a human milk oligosaccharide (HMO) pool (–) compared with an unreacted HMO pool (–). Monosaccharide composition of structures given as Hex_HexNAc_Fuc_Neu5Ac and represented as glucose (●), galactose (●), N-acetylglucosamine (■), and fucose (▴). A, EICs of doubly charged m/z 611.24 shows decreased abundance of specific fucosylated isomers in the reacted HMO pool (1: MFpLNH IV, 2: 4120a, 3: MFLNH III, 4: MFLNH I, 5: IFLNH III, 6: IFLNH I). B, EICs of m/z 1075.41 shows decreased abundance of specific undecorated isomers in the reacted HMO pool (7: LNH, 8: LNnH, 9: p-LNH). C, EICs of m/z 913.35 shows increased abundance of potential digestion products in the reacted HMO pool. D, EICs of m/z 751.29 shows increased abundance of the potential digestion product in the reacted HMO pool.

Fig. 5.

Oligosaccharide extracted in infant feces was determined to be a degraded N-glycan based on tandem mass spectra. Monosaccharides represented as mannose (●) and N-acetylglucosamine (GlcNAc) (■). Precursor ion with m/z 1034.37 shows neutral subsequent losses of hexose residues consistent with a high mannose type N-glycan. Only one GlcNAc present in the core is indicative of endoglycosidase activity on the glycoprotein this structure was derived from.

Fig. 6.

Changes in three infant fecal glycan and bifidobacteria profiles for 1, 3, 10, and/or 17 weeks postnatal. Bar graphs represent absolute abundances of the different glycan types (intact HMOs (■), digested HMOs (■), intact N-glycans (■), digested N-glycans (■), and indistinguishable glycans with overlapping compositions (■)), as well as relative abundances of bifidobacterial subspecies (B. breve (■), B. longum subsp. longum (■), B. longum subsp. infantis (■), and B. pseudocatenulatum (■)) in relation to the entire microbiota. Large decrease in intact HMOs for all infants from week 1 to week 17, along with an increase in digested HMOs in that same time frame. Infants 1004 and 1054 also showed an increase in digested N-glycans from week 1 to 17. Intact HMOs and bifidobacterial species display inverse relationship, with increase in relative abundance of bifidobacteria from week 1 to week 17 for infants 1004 and 1040. *Relative abundances of bifidobacteria were <1% at weeks 1 and 10 for infant 1004.