Modulation of Bifidobacterium by HD5 during weaning is associated with high abundance in later life

Abstract

Background: Bifidobacterium colonization of the intestine is believed to have beneficial effects on our health from infancy throughout life. However, how particular members of the genus Bifidobacterium colonize the neonatal intestine and whether early-life bifidobacterial colonization affects establishment of Bifidobacterium-rich microbiota in later life remain unanswered. α-Defensin secreted from small intestinal Paneth cells elicits selective bactericidal activities that efficiently kill pathogens while hardly affecting commensals including Bifidobacterium in vitro, thus contributing to intestinal microbiota regulation.

Methods: One hundred forty-eight fecal samples were serially obtained from 33 children from postnatal 3-5 days to 3 years old, conducting a longitudinal cohort study of mothers and children living in Iwamizawa city, Hokkaido, Japan (SMILE Iwamizawa study). Microbiota composition and secretory level of α-defensin, human defensin 5 (HD5), were assessed to investigate the relationship between HD5 and Bifidobacterium colonization.

Results: We show that HD5 is associated with colonization of Bifidobacterium in early life from pre-weaning to weaning periods. Furthermore, high relative abundance of Bifidobacterium in the weaning period, which positively correlates with HD5 secretion, is associated with the establishment of Bifidobacterium-rich microbiota at 3 years old, when the intestinal microbiota matures.

Conclusions: This study suggests the importance of the weaning period in establishing long-lasting homeostasis interwoven with the host innate immunity and Bifidobacterium in the intestinal microbiota.

Fig. 1. The intestinal microbiota of children maturates along with the development.

a Comparison of an α-diversity index, observed features between mothers at 4–5 m (n = 28) and children at each time point (n = 22 at 3–5 d; n = 17 at 1 m; n = 21 at 4–5 m; n = 27 at 8–9 m; n = 28 at 1.5 y; n = 33 at 3 y). b Stacked bar chart of relative abundance of the intestinal microbiota at the phylum level. The error bars represent the means ± SD. Statistical significance between mothers and children at each age was evaluated by one-way ANOVA followed by Dunnett’s multiple comparison test in (a). p < 0.05 was considered statistically significant.

Fig. 2. High relative abundance of Bifidobacterium in children at the weaning period is associated with the formation of a Bifidobacterium-rich microbiota at 3 y at the genus level.

a Transition of the relative abundance of the Bifidobacterium genus in children. b Correlation analysis of the relative abundance of the Bifidobacterium genus between children at 3 y and those of each age. The dashed lines in (b) represent the 95% confidence interval range. Statistical significance among children at each age was evaluated by one-way ANOVA followed by Tukey’s multiple comparison test in (a) and Spearman’s rank correlation coefficient test in (b).

Fig. 3. High relative abundance of Bifidobacterium in children at the weaning period is associated with the formation of Bifidobacterium-rich microbiota at 3 y at the species level.

a Stacked bar chart of the relative abundance of Bifidobacterium species in children at each age. b Correlation matrix between the relative abundance of total Bifidobacterium in children at 3 y and each Bifidobacterium species in children at 8–9 m, 1.5 y, and the weaning period (combined data set of 8–9 m and 1.5 y). Statistical significance was evaluated by Spearman’s rank correlation coefficient test in (b). In (b), * in each cell indicates statistically significant correlation (p < 0.05).

Fig. 4. High relative abundance of Bifidobacterium in children at 3 y is associated with low BMI and compositional differences in the intestinal microbiota.

a Correlation analysis between anthropological data and the relative abundance of Bifidobacterium in children at 3 y. b Comparison of the relative abundance of Bifidobacterium between the LowBMI (n = 7; BMI ≤ 33.3rd percentile in age- and sex-matched Japanese children), MidBMI (n = 11; 33.4–66.7th) and HighBMI (n = 14; >66.7th) groups in children at 3 y. c Comparison of the relative abundance of Bifidobacterium between the LowBifido (n = 16) and HighBifido (n = 17) groups, to which the children are assigned in ascending order of the relative abundance of Bifidobacterium at 3 y. d Differentially abundant genera between the LowBifido and HighBifido groups identified by LefSe analysis. The error bars represent the mean ± SD. Dashed lines in (a) represent the 95% confidence interval range. Statistical significance was evaluated by Spearman’s rank correlation coefficient test in (a), one-way ANOVA followed by Tukey’s multiple comparison test in (b), and the Mann-Whitney’s U test in (c). In (d), taxa with an |LDA score| > 2.0 were considered differentially abundant between the groups.

Fig. 5. High HD5 concentration in feces is associated with a high relative abundance of Bifidobacterium of children in early life.

a Transition of fecal HD5 concentration in children along with their development. b Correlation matrix between the relative abundance of each Bifidobacterium species and the fecal HD5 concentration in children at each age, all period (combined data set of all time points), the pre-weaning period (3–5 d, 1 m, and 4–5 m), and the weaning period. Error bars represent the means ± SD in (a). In (b), * indicates statistically significant correlation (p < 0.05). Statistical significance was evaluated by one-way ANOVA followed by Tukey’s multiple comparison test among children at each age in (a) and Spearman’s rank correlation coefficient test in (b).

Fig. 6. HD5 shows only minimal in vitro bactericidal activities against commensal bacteria at the sub-μM concentrations.

In vitro bactericidal activities of HD5 against Escherichia coli (E. coli) ML35, Staphylococcus aureus (S. aureus) ATCC27217, Bacteroides fragilis (B. fragilis) JCM11019, Bifidobacterium breve (B. breve) JCM1192, Bifidobacterium longum (B. longum) ATCC15707, and Lactobacillus casei (L. casei) ATCC393. Five individual experiments were conducted, and the bacterial survival rate at each concentration was calculated relative to 0 μM concentration of each bacterium. The error bars represent the means ± SE. Statistical significance was evaluated by one-way ANOVA followed by Dunnett’s multiple comparison test against 0 μM.