A longitudinal study of the development of the saliva microbiome in infants 2 days to 5 years compared to the microbiome in adolescents

Abstract

Understanding oral microbiota programming attracts increasing interest due to its importance for oral health and potential associations with systemic diseases. Here the oral microbiota was longitudinally characterized in children from 2 days (n = 206) to 5 years of age and in young adults (n = 175) by sequencing of the v3-v4 region of the 16S rRNA gene from saliva extracted DNA. Alpha diversity increased by age, with 2-day- and 3-month-old infants in one sub-group, and 18-month- and 3-year-old children in another. Firmicutes decreased up to 3 years of age, whereas Proteobacteria, Actinobacteria, Bacteroidetes and Fusobacteria abundances increased. Abiotrophia, Actinomyces, Capnocytophaga, Corynebacterium, Fusobacterium, Kingella, Leptotrichia, Neisseria and Porphyromonas appeared from 18-months of age. This was paralleled by expansions in the core microbiome that continued up to adulthood. The age-related microbiota transformation was paralleled by functional alterations, e.g., changed metabolic pathways that reflected e.g., breastfeeding and increasing proportions of anaerobic species. Oral microbiotas differed by feeding mode and weakly by mode of delivery, but not gender, pacifier use or cleaning method or probiotic intake. The study shows that the saliva microbiota is diverse 2 days after birth and under transformation up to 5 years of age and beyond, with fluctuations possibly reflecting age-related environmental influences.

Figure 1

Saliva microbiota alpha- and beta diversity from 2-days to 18 years of age. Amplicon sequence variant (ASV) based diversities built on results from 924 saliva samples collected at ages 2-days, 3- and 18-months and 3-, 5- and 18-years. Figures a-c show rarefaction curves (alpha diversity) based on (a) observed amplicon sequence variants (ASVs), (b) Faith’s phylogenetic diversity and (c) Shannon index at varying sequencing depths. Figures (d,e) show PCoA plots illustrating beta-diversity based on (d) Jaccard diversity, and (e) Unweighted Unifrac distance for the same ASVs. The first component explained nearly 14% based on the Jaccard diversity of the individual variation and 34% based on the Unweighted Unifrac distance.

Figure 2

Saliva microbiota transformation from 2 days to 18 year of age. Taxa detection prevalence (% carriers) at the (a) phylum and (b) genus level according to the eHOMD database assignments at 98.5% identity and at least 2 ASVs per taxa for 924 saliva samples collected at the ages 2 days, 3 and 18-months and 3-, 5- and 18-years. The Venn diagram (c) illustrates taxa overlaps with an attached list of the 35 taxa identified at all sampling occasions, i.e., from 2 days to 5 years, in at least three children.

Figure 3

Saliva bacterial species driving age group separations in 2-day- to 5-year-old children. Pairwise comparisons of saliva-swab samples collected when the children were 2 days, 3- and 18-months, and 3- and 5-years of age (n = 749) were made in PLS regression models. The left panel shows PLS score plots with each dot representing a child when comparing (a) 2 days versus 3 months, (b) 3 months versus 18 months, (c) 18 months versus 3 years, and (d) 3 years versus 5 years old children. The scores plot scale results from the projection of the data onto the principal components, with t[1] and t[2] representing the new created variables summarizing the x-variables for component 1 and 2, respectively. The bar graphs to the right show the top genera and species, i.e., VIP values > 2.0, driving the separation of the individuals in each age group comparison.

Figure 4

Saliva microbiota diversity by gender, feeding- and birth-mode, pacifier habits, and exposure to probiotic bacteria. Figure (a) shows the scree plot from the Principal Coordinates Analysis (PCoA) eigen-vector analysis of 924 saliva-sample-matrix, i.e. samples collected in 2 days to 18 year old participants. Figures b - h are the Jaccard PCoA plots for (b) gender, (c) feeding mode in 3-month-old infants, (d) birth mode in 2-days-old infants, (e) use of a pacifier or not versus in 3 months (encircled symbols) and 18 months old children, (f) parental cleaning mode of the pacifier if the child used it at 18 months of age (the question was introduced in the 18-month questionnaire), (g) use of probiotic drops in 3-month-old infants (red dots), and (h) use of other probiotic products in from 3- (red dots) and 5 year old children (pink dots).

Figure 5

Saliva microbiome characterization for children with 5 year repeated samplings from birth to pre-school age. The results are based on 71 children with saliva swab samples collected when the children were 2-days, 3- and 18-months, and 3- and 5-years of age. None had any antibiotic exposure in the preceding 3-month period. Figure (a) shows the beta diversity from the Jaccard dissimilarity PCoA plot (PERMANOVA p < 0.001 among and between age groups). Figure (b,c) are violin plots including box plots for (b) the Jaccard beta-diversity differences, and (c) the Shannon index for alpha diversity. Figures (d,e) show taxa proportions at the phylum and genus level by age, and (f) the phylogenetic tree for bacterial species detected among the 71 children and with color-coded indications of taxa found in 75% (open circles) or 95% (filled circles) of the children at different ages.

Figure 6

Longitudinal transformation of species detection in the saliva microbiota from birth to pre-school age. Scatter-line plots illustrating fluctuations of species prevalences, i.e., % of the 71 children with presence of the species in the saliva-swab sample, at the five repeated sampling occasions. Species presence was defined according to criteria described in the methods section. (a) the 3 bacterial species with virtually unchanged detection over the 5-year period; (b,c) species with decreasing detection; (d–i) a selection of species with increasing detection prevalence over the 5 year longitudinal study period.

Figure 7

Predicted functional potential from the 16S rRNA gene information. By using Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt2) within QIIME2, the identified 16S rRNA genomes were linked to pathways by ortholog annotation (KEGG orthologues, KO) to evaluate transformations in estimated functions in the saliva microbiome of 71 children with saliva swab samples from all five visits and free from antibiotics in the preceding 3-month period. (a) Bray-Curtis dissimilarity PCoA plot of participants based on their predicted abundance of KO orthologues (overall PERMANOVA, FDR, p = 0.0001 and between-age-group PERMANOVA, FDR, p < 0.009). (b) Proportions of major estimated functions at each age group. (c–j) Violin plots including box plots in a selected panel of functions illustrating changes between the five monitored ages (for all functions, overall PERMANOVA, FDR, p < 0.001).