Human milk oligosaccharides induce acute yet reversible compositional changes in the gut microbiota of conventional mice linked to a reduction of butyrate levels

Abstract

Human Milk Oligosaccharides (HMOs) are glycans with prebiotic properties known to drive microbial selection in the infant gut, which in turn influences immune development and future health. Bifidobacteria are specialized in HMO degradation and frequently dominate the gut microbiota of breastfed infants. However, some species of Bacteroidaceae also degrade HMOs, which may prompt selection also of these species in the gut microbiota. To investigate to what extent specific HMOs affect the abundance of naturally occurring Bacteroidaceae species in a complex mammalian gut environment, we conducted a study in 40 female NMRI mice administered three structurally different HMOs, namely 6'sialyllactose (6'SL, n = 8), 3-fucosyllactose (3FL, n = 16), and Lacto-N-Tetraose (LNT, n = 8), through drinking water (5%). Compared to a control group receiving unsupplemented drinking water (n = 8), supplementation with each of the HMOs significantly increased both the absolute and relative abundance of Bacteroidaceae species in faecal samples and affected the overall microbial composition analyzed by 16s rRNA amplicon sequencing. The compositional differences were mainly attributed to an increase in the relative abundance of the genus Phocaeicola (formerly Bacteroides) and a concomitant decrease of the genus Lacrimispora (formerly Clostridium XIVa cluster). During a 1-week washout period performed specifically for the 3FL group, this effect was reversed. Short-chain fatty acid analysis of faecal water revealed a decrease in acetate, butyrate and isobutyrate levels in animals supplemented with 3FL, which may reflect the observed decrease in the Lacrimispora genus. This study highlights HMO-driven Bacteroidaceae selection in the gut environment, which may cause a reduction of butyrate-producing clostridia.

Keywords: HMO; bacteroides; butyrate; microbiome; microbiota; prebiotic.

Figure 1.

Experimental design and Bacteroidaceae culturing data in NMRI mice. (A) The experimental study design of the animal trial is shown indicating period of HMO supplementation and sampling times. (B) chemical structure of the three different HMOs, 6'SL, 3FL, and LNT included as study substrates. (C) Enumeration of CFUs from faecal samples obtained before, during, and after the HMO supplementation period is shown as mean values with error bars indicating standard deviations. The period of HMO supplementation in drinking water is highlighted as a shaded box. (D) Box-plots showing CFU counts of samples obtained from the cecum and colon at the end of the HMO supplementation period and after the 1-week washout period for the 3FL-WO group. Individual values are shown with whiskers highlighting minimum and maximum values. P-values were obtained by mixed-effects analysis followed by Dunnett's multiple comparisons tests between CTR and all HMO supplementation groups individually with *P < .05, **P < .01, ***P < .001, and ****P < .0001 or repeated-measures ANOVA followed by Dunnett's multiple comparisons tests comparing washout period time points to Day 8 CFU counts in the 3FL group with ^##P < .01 and ^###P < .001.

Figure 2.

Phylogenetic tree based on cultured strains, ASVs identified by 16S rRNA gene sequencing and reference type strains. The tree was generated by the ‘Neighbour-Joining’ method using the ‘Jukes-Cantor’ distance measures with bootstrap values shown (100 replicates). The bar shows phylogenetic distance.

Figure 3.

Effects of HMO supplementation on bacterial diversity and community composition. (A) Boxplots of total number of observed ASVs and (B) Shannon index based on 16S rRNA amplicon sequencing. Statistical significance between time-points was evaluated by paired t-tests within groups with *P < .05. (C) Profiling of bacterial composition in faecal samples obtained from individual animals indicates relative abundance at the class level. Grey lines indicate cocaging of animals. (D) Principal coordinate analysis based on Bray–Curtis dissimilarity on Day 0 and (E) Day 8/Day 15 coloured by group. The table shows results of pairwise comparisons performed by ANOSIM tests indicating R and P-values with significant differences highlighted in bold (P < .05).

Figure 4.

Effects of HMO supplementation on relative abundance of specific bacterial genera. (A) Boxplots showing the relative abundance of Phocaeicola, (B) Lacrimispora, (C) Bacteroides, and (D) Bifidobacterium based on 16S rRNA amplicon sequencing. Statistical significance between time-points was evaluated by paired t-tests within groups. *P < .05, **P < .01, ****P < .0001.

Figure 5.

Effects of HMO supplementation on faecal short-chain fatty acid concentrations. (A) Boxplots showing the concentration of acetate, (B) formate, (C) propionate, (D) butyrate, (E) valerate, and (F) isobutyrate. Values below detection level (LOD) are shown as the LOD. Statistical significance between groups was evaluated by Kruskal–Wallis tests followed by Dunn's multiple comparisons tests comparing to the CTR group. *P < .05 and **P < .01.

Figure 6.

Effects of HMO supplementation on colonic tissue gene expression. (A) Boxplots showing the relative gene expression of occludin, (B) Tjp1, (C) TNFα, and (D) Muc2. Statistical significance between all groups was evaluated by Kruskal–Wallis tests followed by Dunn's multiple comparisons tests comparing to the CTR group or Mann–Whitney test between the 6'SL, LNT, and 3FL groups aggregated vs. the CTR group. *P < .05 and **P < .01.

Figure 7.

Correlations between affected bacterial genera, alpha diversity and short-chain fatty acids in faecal samples. (A) Scatter dot plots of Lacrimispora vs. Phocaeicola relative abundances, (B) Phocaeicola/Lacrimispora (log₂) ratio vs. number of observed ASVs, (C) Phocaeicola/Lacrimispora (log₂) ratio vs. concentration of acetate and (D) Phocaeicola/Lacrimispora (log₂) ratio vs. concentration of butyrate. Associations were assessed by Spearman's rank correlation analysis with rho and P-values indicated.