The Predominant Oral Microbiota Is Acquired Early in an Organized Pattern

Abstract

The human oral cavity is sterile prior to birth, and we have limited knowledge of how complex oral communities are assembled. To examine bacterial acquisition and community assembly over the first year of life, oral samples from a cohort of nine infants and their mothers were collected, and bacterial community composition was studied by 16S rRNA gene sequencing. Exogenous species including skin and environmental bacteria were present initially, but were quickly replaced by a small, shared microbial community of species common to all infants and adults. Subsequent ordered microbial succession and the formation of increasingly complex communities was observed. By one year of age oral microbial community composition converged to a profile that was remarkably similar among children. The introduction of new nutrient sources, but not tooth eruption, was associated with increasing complexity. Infants had fewer species than mothers, mostly accounted for by the lack of certain anaerobes, and showing that the acquisition and assembly of oral microbial communities continues past infancy. When relative abundance was considered, a shared set of species accounted for the majority of the microbial community at all ages, indicating that the dominant structure of the oral microbiome establishes early, and suggesting that it persists throughout life.

Figure 1

Oral microbial communities in mothers, infants, and toddlers. (A) Venn diagram and relative abundance boxplots for species detected in mothers, infants and toddlers are shown. Number of species detected for each Venn intersection are shown within the diagram, and species are listed for each intersection. For each species, boxplots show square root-adjusted relative abundances for infants (green), toddlers (blue) and mothers (pink). Only species that met empirical probability thresholds were included. (B) The contribution to total microbial relative abundance of those 13 species present at all ages are shown in boxplots.

Figure 2

Comparison of microbial community composition in infants, toddlers and mothers. (A) Overall composition differed among infants, toddlers and mothers, as shown in a Bray Curtis Multidimensional Scaling plot where centroids are shown as square points. Multiple samples from a single individual were combined within each age range. A PERMANOVA analysis showed overall differences between the groups while post hoc PERMANOVAs showed significant differences between mothers and either of the baby groups (mothers:0–2 month p = 0.0039; mothers:10–12 month p = 0.078). (B) Shannon diversity and (C) richness varied significantly among age groups (by Kruskal-Wallis), and increased with increasing age (by LME). Bars indicate significant post hoc comparisons (Mann Whitney test) with the two asterisks showing p values between 0.01 and 0.001. The mean number of species detected per infant from 0–2 mo of age was 30, from 10–12 months it was 33, and in mothers it was 83 (95% CI = 14–47, 15–50, 64–102, respectively).

Figure 3

Similarity of microbial community composition among children over the 1^st year of life converged. Using the first dimension of a non-metric Multidimensional Scaling (MDS) based on total microbial community composition, the distance to the centroid was plotted against time and analyzed using a linear mixed effects model (p < 0.003).

Figure 4

Species were acquired in an ordered sequence. A heatmap based on prevalence of all species found in ≥50% of subjects is shown. Monthly samples were collapsed into 3-month intervals for each child, and rarefied as described in methods. Species are sorted by decreasing overall prevalence, and the stepped heavy line demarcates the age at which they were detected in ≥75% of children.

Figure 5

Introduction of solid food but not tooth eruption had a significant effect on microbial community composition. Time of tooth eruption was recorded, and samples taken immediately pre- and post- tooth eruption and pre- and post- introduction of solid foods were included in the analyses. On average infants began solid foods at 6.8 ± 0.9 months of age and tooth eruption at 8.7 ± 2.2 months. Bray Curtis Multidimensional Scaling plots with group centroids plotted as square points and box plots for richness and diversity are shown (analyzed by Wilcoxon Signed Rank Test).