Abstract
Recent research provides new insights into the early establishment of the infant gut microbiome, emphasizing the influence of breastfeeding on the development of gastrointestinal (GIT) microbiomes. In our study, we longitudinally examined the taxonomic and functional dynamics of the oral and GIT microbiomes of healthy infants (n=30) in their first year, focusing on the often over-looked aspects, the development of archaeal and anaerobic microbiomes.
Breastfed (BF) infants exhibit a more defined transitional phase in their oral microbiome compared to non-breastfed (NBF) infants, marked by a decrease in Streptococcus and the emergence of anaerobic genera such as Granulicatella. This phase, characterized by increased alpha diversity and significant changes in beta diversity, occurs earlier in NBF infants (months 1-3) than in BF infants (months 4-6), suggesting that breastfeeding supports later, more defined microbiome maturation.
We demonstrated the presence of archaea in the infant oral cavity and GIT microbiome from early infancy, with Methanobrevibacter being the predominant genus. Still, transient patterns show that no stable archaeome is formed. The GIT microbiome exhibited gradual development, with BF infants showing increased diversity and complexity between months 3 and 8, marked by anaerobic microbial networks. NBF infants displayed complex microbial co-occurrence patterns from the start. Those strong differences between BF and NBF infants GIT microbiomes are less pronounced on functional levels than on taxonomic level.
Overall, the infant microbiome differentiates and stabilizes over the first year, with breastfeeding playing a crucial role in shaping anaerobic microbial networks and overall microbiome maturation.
Introduction
The human microbiome is a complex ecosystem of microorganisms, undergoing substantial changes from birth to adulthood (1). Among the various microbiomes, the oral microbiome is one of the most intricate one, comprising over 700 identified species (2, 3). The oral cavity is a primary entry point for the colonization of both oral and gastrointestinal tract (GIT), making it an accessible site for assessing microbial communities. The unique community of microbes in the oral cavity is in fact very important and any disruption in early oral colonization and the establishment of a healthy oral microbiome is linked with several oral diseases including dental caries and periodontitis, which could start with the emergence of teeth (4), as well as increased susceptibility to systemic diseases such as cardiovascular disease, due to the presence of potential pro-inflammatory mediators present in periodontium (5).
The formation of the oral microbiome in early childhood is known to be influenced by both host and environmental factors, including genetics, delivery mode, antibiotic use during birth and early infancy, feeding mode, and the characteristics of the parental oral microbiome (6). However, the process of initial acquisition and development of this complex microbiome during infancy is not fully understood.
The oral cavity is constantly exposed to oxygen on its surfaces, yet it contains numerous anoxic environments that provide niche habitats and favorable conditions for anaerobic metabolism and microbial growth. These include biofilms, dental pockets, subgingival crevices and crypts of the tonsils (7). In general, facultative anaerobic Streptococcus is the predominant early colonizer of the infants’ oral cavity, favored by its ability to adhere to epithelial cells (8). By secreting extracellular polymers, it then paves the way for other microbes to emerge, such as Actinomyces spp. (9). The infants’ oral microbiome is less diverse compared to adults but becomes more complex within the first months, with mainly Streptococcus, Haemophilus, Neisseria and Veillonella colonizing (4, 8). Nevertheless, knowledge about colonization of non-bacterial microbiome members in the oral cavity is very scarce (4).
The oral cavity and gut are connected by the continuous flow of ingested food and saliva through the GIT. Despite this connection, they host distinct microbial communities within their unique microenvironments. Research has shown that these sites harbor locally adapted strains specific to their environments (7, 10, 11), and this segregation is thought to be through various environments including gastric barrier and antimicrobial bile acids within the duodenum. However, little is known about the parallel development of GIT and oral microbiome (12–14). This is particularly true regarding the radical shifts in the GIT due to the oxygen depletion and the unknown interaction of both environments during this time.
The human GIT in fact harbors the most versatile microbial community. In the initial aerobic phase immediately after birth, the GIT is populated by obligate aerobic or facultative anaerobic microbes which thrive in the presence of oxygen and are well-adapted to the aerobic environment of the newborn GIT (15–17). The shift to an anaerobic state is driven by oxygen depletion, caused by oxygen consumption by bacteria, colonocytes or non-biological chemical processes in the cecal contents (18). This step is an essential step in GIT maturation. As oxygen levels decrease, strictly anaerobic bacteria thrive, especially as Bifidobacterium species begin to dominate the GIT microbiome. These microbes are adapted to a milk-based diet, using the bifid shunt allowing for a fast growth at high lactose concentrations (19). Later, during weaning and introduction of complementary food, other microorganisms replace Bifidobacterium species as the dominant microbial group. Steward et al (20) defined three distinct phases of microbiome progression: a developmental phase at months three to 14, a transitional phase at months 15 to 30 and a stable phase at months 31-46. These changes are influenced by numerous factors, including birth mode, gestational age, host genetics, environmental factors and most importantly, feeding mode. The final maturation and stabilization of the GIT microbiome includes not only the settling of highly-oxygen sensitive bacteria, but also methanogenic archaea, which could be indicators for a mature microbiome situation (21).
Similar to fungi, archaea receive less attention regarding their role in the development of a healthy microbiome, although they are present in both the GIT and oral cavity, often in substantial numbers (22, 23). Few studies have recently shown the detection of archaeal signatures in young infants (23, 24).
Herein, we conducted a longitudinal study on a birth cohort (TRAMIC, https://clinicaltrials.gov/study/NCT04140747) of 30 Austrian infants to investigate the dynamics of aerobic and anaerobic bacteria and archaea in the oral cavity and GIT. The cohort included 15 vaginally delivered infants and 15 born via C-section. Daily up to monthly monitoring of the infants’ oral and GIT microbiomes was performed using shotgun metagenomic and amplicon sequencing. This allowed us to assess the parallel development of aerobic and anaerobic microbiomes in both sites, correlating these patterns with birth mode and infant nutrition.
By elucidating the colonization patterns and ecological dynamics of obligate anaerobes and archaea in both oral and stool environments, this study aims to provide insights into a more fine-tuned early development of the infant microbiome. Understanding the factors shaping microbial colonization during infancy is fundamental for deciphering the role of the microbiome in the course of life and may lead to new strategies to promote infant health and well-being.
Materials and Methods
Study design
A total of 32 mother-infant pairs were enrolled in the study during their visits to the Department of Gynecology at the state hospital Graz, Austria before delivery. These participants provided informed consent and obtained oral swabs and stool samples from their infants at various time intervals, commencing immediately after delivery. The primary objective of this pilot study was to investigate the anaerobic microbiome, with a specific focus on archaea, in the oral cavity and gastrointestinal tract of infants throughout their initial year of life.
Every pregnant woman included in the study was in good health, had not undergone antibiotic treatment within the past six months and was 18 years of age or older. Detailed inclusion and exclusion criteria can be found in our prior publication (25). Additionally, their infants were required to be healthy, full-term singletons without any anomalies.
Metadata from all women and infants are listed in the GitHub Repository https://github.com/CharlotteJNeumann/InfantDevelopmentTRAMIC.
In sum, two women opted to discontinue their participation during the study, resulting in 30 infants successfully completing the sample collection phase. Among them, 15 infants were delivered via C-section, while the remaining 15 were born vaginally.
Sample collection and processing
Oral swabs and stool samples were gathered from all 30 infants at various time points. Stool samples were collected three times during the initial days of life (S1 [first stool, day 1], S2 [days 2-3], S3 [days 3-5]). Oral samples were obtained twice during the first days of life (O1 [day 1], O2 [days 3-5]). Both sample types were collected monthly until the infants reached their first birthday (months 1 (M01) to 12 (M12)). The collection was performed either by the study nurse at the hospital or by the mothers themselves, following clear instructions on proper collection and storage procedures.
Stool samples were obtained using sterile collection tubes, while oral samples were collected from the buccal mucosa of the cheek using FLOQSwabs (Copan, Milan, Italy). Subsequently, all samples were refrigerated, transported to the laboratory on ice and stored at -80°C until further processing.
Genomic DNA was extracted from the oral swabs utilizing the QIAamp DNA Mini Kit (QIAGEN) with slight modifications: 500 µL of Lysis Buffer (sterile filtered, 20 mM Tris-HCl pH 8, 2 mM Na-EDTA, 1.2% Triton X-100) was added. To all samples, 50 µL of Lysozyme (10 mg/mL) and 6 µL of Mutanolysin (25 KU/mL) were added, followed by an incubation at 37°C for 1 hour. The resulting mixture was transferred to Lysing Matrix E tubes (MP Biomedicals) for mechanical lysis at 5,500 rpm for 30s twice using the MagNA Lyser Instrument (Roche, Mannheim, Germany). Following mechanical lysis, the samples were centrifuged at 10,000 x g for 2 minutes to separate the beads from the supernatant. Subsequently, DNA extraction was performed according to the provided instructions, with the elution of DNA in 60 µL of Elution Buffer.
Stool samples were processed utilizing the QIAamp DNA Stool Mini Kit (QIAGEN) with slight modifications: approximately 200 mg of stool was combined with 500 µL Inhibitex and homogenized. To the homogenized samples, 50 µL of Lysozyme (10 mg/mL) and 6 µL of Mutanolysin (25 KU/mL) were added and incubated at 37°C for 1 hour. Following the incubation, 500 µL Inhibitex was introduced to the samples, homogenized and transferred to Lysing Matrix E tubes (MP Biomedicals) for mechanical lysis at 6,500 rpm for 30s twice using the MagNA Lyser Instrument (Roche, Mannheim, Germany). Subsequent to the mechanical lysis process, the samples underwent a 5-minute incubation at 70°C, succeeded by a centrifugation step at 10,000 x g for 3 minutes to segregate the beads from the supernatant. The resulting supernatant was then transferred to 2 mL Eppendorf tubes and the remaining steps of the DNA extraction were conducted following the kit protocol. The elution of DNA was carried out using 200 µL of Elution Buffer.
Throughout the DNA extraction procedure, negative controls were incorporated and processed concurrently.
PCR amplification
The acquired genomic DNA was utilized to amplify the V4 region of the 16S rRNA gene employing Illumina-tagged primers, namely 515FB and 806RB (Table 1). To discern the archaeal communities within the samples, a nested PCR was conducted using the primer combination 344F-1041R/519F-Illu806R, as described previously (26). PCR reactions were executed in triplicates within a final volume of 25 µL, containing TAKARA Ex Taq® buffer with MgCl2 (10 X; Takara Bio Inc., Tokyo, Japan), primers at 200 nM, dNTP mix at 200 µM, TAKARA Ex Taq® Polymerase at 0.5 U, water (Lichrosolv®; Merck, Darmstadt, Germany) and DNA template (1-2 µL of genomic DNA) and pooled after amplification. The specific conditions for PCR amplification are outlined in Table 2.
Amplicon sequencing, bioinformatics and statistical analysis
The library preparation and sequencing of amplicons were conducted at the Core Facility Molecular Biology, Center for Medical Research, Medical University of Graz, Graz, Austria. Briefly, DNA concentrations were normalized using a SequalPrep™ normalization plate (Invitrogen) and each sample was uniquely indexed through an 8-cycle index PCR with a unique barcode sequence. Following the pooling of these indexed samples, a gel cut was performed to purify the products from the index PCR. Sequencing was executed using the Illumina MiSeq device along with the MS-102-3003 MiSeq® Reagent Kit v3-600 cycles (2×150 cycles). The generated 16S rRNA gene amplicon data are accessible in the European Nucleotide Archive under the study accession number PRJEB77729.
The analysis of the 16S rRNA gene amplicon data was performed using QIIME2 (27) 2021.1-12 following the previously outlined methodology (28). Quality filtering was performed with the DADA2 algorithm (29) which involved merging paired-end reads, truncation (−p-trunc-len-f 200 -p-trunc-len-r 150) and denoising for the generation of amplicon sequence variants (ASVs). Taxonomic classification (30) was based on the SILVA 138 database (31) and the resultant feature table and taxonomy file were used for subsequent analysis. Contaminating ASVs were identified and eliminated via decontam v 1.13 (32) in R (33), running iscontaminant in prevalence mode with varying thresholds (oral-bacteria: 0.3; stool-bacteria: 0.3; oral-archaea: 0.5; stool-archaea: 0.1). Following this, positive controls and negative controls were excluded from the datasets. Additionally, ASVs classified as chloroplast or mitochondria were removed as well as ASVs with < 1 read.
For normalization, different approaches were applied for the bacterial and archaeal datasets, taking into account their respective composition. SRS (scaling with ranked subsampling) normalization was run in QIIME2 (27) applying different cmin for the bacterial dataset (oral-bacteria: cmin = 8,400; stool-bacteria:cmin = 3,800). The archaeal datasets underwent TSS normalization (total sum normalization). The number of samples subjected to analysis and kept after normalization are listed in Suppl. Fig. 1.
Several plot types, including stacked bar plots and PCA plots, were generated using MicrobiomeExplorer (34) in R (33).
Differentially abundant taxa were defined by q2-aldex2(35–37) in QIIME2 (27). To display those taxa in boxplots (packages: ggplot2 (38), dplyr (39), reshape (40)(33), the data of relative abundance were first CLR transformed in R (33).
Alpha diversity numbers as well as beta diversity (PERMANOVA) were calculated with the microbiome package (41) in R (33) and plotted with ggplot2 (38) and dplyr (39).
Longitudinal linear mixed effect models were created with q2-longitudinal (42) in QIIME2 (27) with the option “linear-mixed-effects” for Shannon diversity and “first-distances” additionally for beta diversity.
Identification of oxygen requirements
The datasets of universal amplicon data were further investigated regarding the underlying type of respiration. This information had to be collected and entered manually. As resolution from amplicon sequencing is scarce on species level, the genus level was taken into account and physiology data were extracted from bacdive (https://bacdive.dsmz.de/). Therefore, type strain representatives were used, and the common denominator was chosen. We are aware of the problem that physiological data might differ between several species of the same genus, therefore we handle those data with great care and only as an approximation. In the category of respiration, we assigned three groups: obligate aerobe (listed as “obligate aerobes” and “aerobes”), facultative anaerobe (listed as “microaerophile”, “facultative aerobe”, “facultative anaerobe”) and obligate anaerobe (listed as “anaerobes” and “obligate anaerobes”).
Source tracking
Source Tracking was performed to depict the potential of single ASVs of the oral microbiome (source) to be transferred to the GIT microbiome (sink). Therefore, oral and stool datasets were first merged and then TSS normalized, once for the bacterial approach and once for the archaeal approach. Source Tracking was performed with SourceTracker2 (43) in QIIME2 (27). Rarefaction of source data (oral) and sink data (stool) and vice versa was performed as advised by SourceTracker2 (43) individually per time point. The rarefaction values are listed in a respective table on GitHub (URL: https://github.com/CharlotteJNeumann/InfantDevelopmentTRAMIC).
Additionally, using the “--per_sink_feature_assignments” option in SourceTracker2 (43) on TSS-normalized datasets, we could calculate the origin source of a single taxa. The counts were log-transformed for visualization.
Network calculations and visualization
To infer genus-level associations we employed SparCC (44) within the SCNIC tool v. 0.5 (Sparse Co-occurrence Network Investigation for Compositional data) (45). SparCC was run on default settings with 1,000 permutations and the multiple testing correction method set to ‘fdr bh’. Co-occurrence events were visualized in Cytoscape v.3.10.1 (46) where nodes represent taxa and edges represent co-occurrences according to the SparCC R values. Stress centrality and other network properties were calculated using Cytoscape. Files of stress centrality for single genera are provided on the GitHub repository (URL: https://github.com/CharlotteJNeumann/InfantDevelopmentTRAMIC).
Metagenomic data
Shotgun Metagenomic Sequencing
We performed shotgun metagenomic sequencing of a subset of infants for a few points (O2, S2, S3, M01, M06, M12). Sequencing libraries were generated with the TruSeq Nano DNA Library construction kit (Illumina, Eindhoven, the Netherlands) and sequenced on an Illumina NovaSeq 6000 platform (Illumina, Eindhoven, the Netherlands; Macrogen, Seoul, South Korea).
Metagenomic data processing
The raw reads were processed using the ATLAS v.2.18.0 workflow (47). There, quality control (PCR duplicates removal, quality trimming, host removal, common contaminant removal) was performed leading to QC reads which were then assembled into high-quality scaffolds using megahit. All parameters used for ATLAS are detailed in the config.yaml file, which is available in the GitHub repository (URL: https://github.com/CharlotteJNeumann/InfantDevelopmentTRAMIC). Genome binning was achieved with maxbin2 v. 2.2.7 (48), followed by quality assessment of genome bins with checkM v. 1.0.1 (49), bin refinement with DASTool v. 1.1.6 (50), dereplication with dRep v. 3.5.0 (51) and taxonomic classification of representative MAGs (metagenomic assembled genomes) with GTDB v 2.3.2 (52–54). Cutoffs for high quality MAGs were set as follows: completeness >90% and contamination <5%.
Metagenomic data could only be obtained for nine oral samples in total. Therefore, no further analyses on metagenomic oral data were possible. Strain tracking was performed in inStrain v. 1.5.7 (55) in ATLAS (47) with the following cutoffs: percent_genome_compare: ≥50% and popANI: ≥ 99.999% as indicated in the documentation of inStrain (56). Functional annotations were also run within the ATLAS pipeline (47). First, Prodigal v.2.6.3 (57) was applied for gene prediction and linclust (58) to cluster redundant genes (minid = 0.9 and coverage = 0.9) (58). The quantification of gene abundance per sample was performed using the combine_gene_coverages function via the BBmap suite v.39.01-1 (59). Employing eggnog-mapper (v.2.0.1) (60, 61) on the EggNOG database 5.0, taxonomic and functional annotations were assigned. KEGG annotations were extracted (62–64) and read counts were implemented and analyzed in R, following https://github.com/metagenomeatlas/Tutorial/blob/master/R/Analyze_genecatalog.Rmd. Annotated gene counts were normalized (size factor normalization) and tested for differential expression between BF and NBF infants using DESeq2 (64).
Read-centric metagenome analysis
Species’ relative abundances were determined using Kraken2/Bracken (65, 66). Initially, Kraken2 v.2.1.2 (65) was employed to profile the quality-filtered reads from ATLAS v.2.18.0 (47) against the Unified Human Gastrointestinal Genome (UHGG v.2.0.1) (67) database of bacterial and archaeal genomes. Subsequently, Bracken v.2.7. (66) was used with default settings to analyze the Kraken2 output and calculate the relative abundance of bacterial and archaeal species. The resulting report files were merged to generate an abundance table of microbial species for further analysis.
Additional tools used in the manuscript
ChatGPT.com and deepl.com were used for language checks, but not for interpreting the data. An overview of the available data is displayed in two figures: Suppl. Fig. 1 is following the STORM guideline and was created with drawio.com (URL: https://drawio.com). Suppl. Fig. 2 displays the data available per sample and individual.
Ethics statement
This study has been registered on clinicaltrials.gov (NCT04140747). The samples were collected under the ethical approval number 28-524 ex15/16 by the respective local ethics committees, the Ethics Committee at the Medical University of Graz, Graz, Austria and in adherence to the principles outlined in the Declaration of Helsinki.
Reproducibility
We conducted a prospective pilot study whereas sample size was not predetermined beforehand. Randomization and blinding of the investigators were not foreseen in the chosen study setup. A full study flow chart is provided in Suppl. Fig. 1 and 2. Participants 13 and 17 were excluded from the study due to incompleteness. Overall, the study is considered to be only partially reproducible, as the data are dependent on the study cohort, which was only sampled once within this study, and sampling of cohorts at the same time window cannot be repeated. However, starting from the raw sequencing data, the analysis is fully reproducible, and all required data, scripts, and details are provided.
Results
Overview on the study population and sample description
Infancy is a dynamic period for microbiome development, with the first 1,000 days of life being the most critical period (68). We highlight the dynamics of microbiome composition and co-occurrence patterns in oral and GIT microbiomes in the first year of life and their transmission patterns, with a focus on the dynamics of anaerobic microorganisms.
Oral and stool samples of 30 infants born either spontaneous (n = 15) or via C-section (CS) (n = 15) were collected at different time points (Suppl. Fig. 2). Stool samples were initially collected at three time points (tps) (S1 [first stool, meconium, day 1], S2 [meconium, days 2-3] and S3 [days 3-5]), while oral samples were obtained at two-time intervals (O1 [day 1, prior to feeding, immediately after delivery], O2 [days 3-5]). Both sample types were collected monthly until reaching the age of 1 year (months M01 to M12). The characteristics (covariates) of the study groups did not significantly differ with regard to the mode of delivery (Chi-square test, p > 0.5) except for gestational age, which is significantly lower in infants born via CS (Chi-square test, p < 0.001). The metadata of the studied cohort can be found in Table 3 and Fig. 1.
Graphical abstract of the study including recruitment, sample collection and genomic data received.
Breastfeeding is considered the most significant microbiome covariate within the first year of life (20). We could also observe that in our dataset with breastfeeding showing that the feeding type significantly impacts four and one timepoint for oral and GIT samples, respectively, (PERMANOVA; oral: p < 0.05 for four tps; stool p < 0.05 for one tp) but only 1 timepoint for both sample types (PERMANOVA; oral: p < 0.05 for one tp; stool p < 0.05 for one tp). Based on this observation, we mainly focus on the feeding types and their impact on the anaerobic microbiome in the oral cavity and GIT and their transitional phase.
The oral cavity and GIT are rapidly exposed to strict anaerobes
Samples taken right after birth (labeled “O1” and “S1”) and within the first days of life (labeled “S2”, “S3” and “O2”) showed that newborns get colonized rapidly by various microbes. The first obligate anaerobic bacteria detected in the oral cavity and GIT are Rothia (oral), Streptococcus, Staphylococcus (both oral and GIT), Bifidobacterium and Enterococcus (GIT). Interestingly, next to bacteria, also archaeal signatures could be detected in those early samples (Suppl. Fig. 3). Archaeal diversity was higher at those early-stage samples with Methanobrevibacter, Methanobacterium, Methanosphaera and Methanocorpusculum (all obligate anaerobes) being present next to unclassified Woesearchaeales (oral and GIT) and unclassified Nitrososphaeraceae (GIT). For the latter two, the oxygen requirements are unknown, as these taxa were not classified deep enough. At M01, Methanobrevibacter was predominant amongst Archaea in the oral cavity whereas in the GIT. As expected, samples collected at the very early stages showed different microbial profiles compared to the ones collected at M01, revealing a shift from S1/O1 to M01 (Suppl. Fig. 3). It is assumed that the first samples taken immediately after birth do not reflect the inhabitant microbial community, but rather a microbial contamination given the sterile environment in the womb (69). However, although the microbial ecosystem is not fully functional at this time stage, microbial colonization can already start due to exposure to the environment and subsequent oral-GIT transmission. Still, the main analyses drawn out in this paper will be focusing on samples collected at M01 and later, when microorganisms have started to establish.
Staphylococcus and Streptococcus are early but transient colonizers of the oral cavity
In the first months of life, the human skin (parents, family members) is an important source of microbial influx from the environment (8, 70). This is underlined by our data, showing high relative abundances of Staphylococcus (facultatively anaerobic) representing a taxon that is mainly skin-(and mucosa) associated (Fig. 2a)). We did not find significant differences in the relative abundance of Staphylococcus between BF and NBF infants (Aldex2, all tps, p > 0.05, Suppl. Fig. 4), indicating a general substantial transfer from skin to the oral cavity, independent from feeding mode.
To assess the connectivity and co-occurrence of microbes, we built networks for each time point by forming modules in SCNIC at the genus level (Fig. 2b, Suppl. Fig. 5). From M03 on (Suppl. Fig. 5), Staphylococcus has a very minor relative abundance and appears only sporadically in the co-occurrence networks with low centrality compared to other players (Fig. 2b) and Suppl. Fig. 5 for the complete networks; stress centrality BF: M02: 4, M04: 8, M06: 4; M07: 32; NBF: M01: 150, M02: 28, M05: 82, M10: 44, M12: 28), indicating its transient colonization in the oral cavity of infants in early life.
Especially in the first months of life, the infant’s early oral microbiome is predominated by facultatively anaerobic Streptococcus (Fig. 2a). Interestingly, the centrality of Streptococcus in microbial networks is surprisingly low, although the abundance is very high (>60% relative abundance) (Fig. 2b). This indicates that even if a microbe is very abundant, this does not necessarily mean that it is an important player in the networks formed by the microbial community. It appears that streptococci do not interact with other microbes on a large scale, but rather rely on themselves and act independently. Streptococci, who are mainly involved in carbohydrate metabolism, are considered pioneer species that lead the assembly of a complex oral microbiome (71). The dominance of Streptococcus is higher in BF infants, reflected by both relative abundance (Aldex2, M04, M05, M06 & M09: p < 0.05, q > 0.05, Suppl. Fig. 4; Fig. 2a) as well as network centrality (Fig. 2b and Suppl. Fig. 5, tp M04 and M06 as an example: BF: M04 = 48, M06= 18; NBF M04= 128, M06= 88). Streptococcus shows a decrease in relative abundance starting from M05 onwards (Fig. 2a; Aldex2, p > 0.05 at any tp pairwise comparison).
Figure panel on the oral microbiome a) Stacked bar chart showing the relative abundance of the 10 most common bacterial genera in the oral cavity per time point (months M01 to M12). Data are shown separately for breastfed (BF) and non-breastfed (NBF) infants. The oxygen requirement of the respective genera is highlighted by colored stars: pink: obligate anaerobes, yellow: facultative anaerobes, petrol: obligate aerobes. b) Networks on oral samples of BF and NBF infants of selected time points (months M01, M03, M06, M09 and M12). Font size indicates stress centrality, colors indicate oxygen requirement: pink: obligate anaerobes, yellow: facultative anaerobe, petrol: obligate aerobes. c) Pairwise beta diversity comparisons (PERMANOVA) from months M01 and M12 to all other timepoints to depict the transition phase. Gray and shaded green colors indicate q-values; data are separated for breastfed (BF) and non-breastfed (NBF) infants. d) Shannon diversity of oral samples depicted for breastfed (BF =light green) and non-breastfed (NBF = lavender) infants with asterisks indicating significant difference (q - values) between those two groups. Significant q - values between tps in BF infants are indicated with light green asterisks.
Distinct transitional phases of the oral microbiome in BF infants
Taking several analyses into account (alpha diversity, beta diversity, Aldex2, networks), we were able to outline a time frame in which the oral microbiome changes the most and which thus represents a transition phase of the oral microbiome towards a more mature microbial community.
When the beta diversity of all oral samples from BF infants was compared with the first (M01) and last time points (M12), it was found that the samples from M03 to M06 differed significantly from these reference points (PERMANOVA; Fig. 2c). Interestingly, this effect was way less pronounced in NBF infants, where only the first four months are significantly different to M12 (Fig. 2c), indicating a less defined maturation period in this group.
These observations go hand in hand with patterns we observed in alpha diversity and help us to further define the transitional phase of the oral microbiome. This phase seems to take place overall weaker and more gradual in NBF than in BF infants. Alpha diversity was shown to be in general significantly increasing within the first year of life (Shannon Fig. 2d and evenness and richness, Suppl. Fig. 6) (Shannon, longitudinal linear mixed-effect model (LME), p < 0.001, Suppl. Fig. 7) and higher for NBF infants than for BF infants (LME, p = 0.002, Suppl. Fig. 7). For NBF infants’ alpha diversity was more rapidly increasing within the first four months of life (M01-M04) (Fig. 2d). Highest differences in alpha diversity between BF and NBF infants could be observed at M03-M06 and M09 (Shannon diversity, t-test: M03 q = 0.00044, M04 q = 0.0013, M05 q = 0.0079, M06 q = 0.0081, M09 q = 0.0096; evenness, t-test: M03 q = 0.0056, M04 q = 0.00052, M05 q = 0.029, M06 q = 0.0098, M09 q = 0.0096; richness, Wilcoxon: M05 q = 0.0055). For BF infants, a significant increase in both Shannon diversity and richness could be observed from M05 to M06 (Wilcoxon, q = 0.0055, Fig. 2d and Suppl. Fig. 6).
As already pointed out above, the decrease of Streptococcus is also taking place earlier in NBF than in BF infants, supporting the observation of two different dynamics patterns of microbiome maturation in BF and NBF infants. It appears that the transitional phase of the oral microbiome of BF infants takes place later (M03-M06), whereas the microbiome of NBF maturates earlier (M01-M03).
During this transitional phase, in contrast to Staphylococcus and Streptococcus which were declining, several genera increased in relative abundances. A distinction can be made between genera such as Gemella, Rothia and Haemophilus that were already abundant in the first months (all facultative anaerobes) and genera such as Granulicatella (obligate anaerobe), Neisseria, Veillonella (obligate anaerobe), Alloprevotella (obligate anaerobe) and Leptotrichia which were newly introduced (Fig. 2a). During the transition phase on BF infants (M03-M06) Streptococcus did not form any co-occurrence connections with those “new” genera Neisseria, Alloprevotella, or Leptotrichia at all, except with Neisseria at M12 (Fig. 2b). After M07, Alloprevotella and Leptotrichia began to co-occur indirectly with Streptococcus, primarily with Gemella serving as the connecting node. This suggests that Gemella may mediate the integration of co-occurrence between Alloprevotella - Streptococcus and Leptotrichia - Streptococcus. This integration could exemplify how the introduction of a new microbe (Alloprevotella/Leptotrichia) into the community may be facilitated by an existing microbe (Gemella), contributing to community maturation. Interestingly, niche-sharing between Streptococcus and the genera that were already fairly abundant at the beginning, Gemella and Rothia, can be found overall. But co-occurrence of Streptococcus with Haemophilus could exclusively be observed in NBF infants. Additionally, Streptococcus showed intensive co-occurrence connections with Granulicatella and Veillonella.
Those “new” bacterial genera showed a lagged increase in relative abundance in BF infants between mainly M04 and M05 (Suppl. Fig. 4; Aldex2, Granulicatella p = 0.003, Neisseria p = 0.012, Veillonella p = 0.003, Leptotrichia p = 0.002). This supports our findings of a lagged transitional phase between BF and NBF infants (can also be seen in Fig. 2a).
Breastmilk maintains simplicity of oral microbial network structures
Starting around M07, after the transitional phase of the BF infants’ oral microbiome has concluded, the microbiomes of BF and NBF infants became more similar in terms of alpha diversity, beta diversity and differentially abundant taxa, with fewer significant differences observed on genus level. This coincides with the time when solid food typically constitutes a large portion of the infants’ diet, suggesting that solid food acts as a levelling factor for the oral microbiomes of BF and NBF infants.
A notable difference between the oral microbiomes of BF and NBF infants was the overall organization of the microbial networks (Fig. 2b). From the beginning, the microbial networks in NBF infants were very complex, with multiple nodes (microbial genera) and edges (co-occurrences between genera). Several genera exhibited high stress centrality, meaning they co-occurred with multiple genera and are thus were central in the network. This structure changed only slightly over time. In contrast, the oral microbial network of BF infants was much less complex, with fewer microbial players sharing a similar niche. These differences in complexity especially pronounced in the first months until about M06. This could be attributed to the different nutrient compositions of BM and formula milk, with BM being digested very efficiently, leading to a simplified microbial community.
After M06, the networks in BF infants became more complex, with higher stress centrality of individual microbes (from < 50 until M06 to 800 at M11, Suppl. Fig. 8) and more microbial members co-occurring (18 nodes at M06 to 31 nodes at M12). This likely reflects the introduction of solid food, which provokes a new mode of microbial (inter-)action. However, despite solid food becoming a major component of the diet of a one-year-old child, the administration of BM, even in smaller proportions, still seemed to influence the GIT microbiome. Even at M12, clear differences between the microbial networks of BF and NBF infants were still evident (Fig. 2b).
Neisseria and its key role in the thriving of obligate anaerobes in the oral cavity
Within the first year of life, the relative abundance of facultative anaerobes decreased from 96% to 76%, with a corresponding increase in obligate anaerobes and obligate aerobes appeared (Suppl. Fig. 9). NBF infants exhibited a higher load in obligate anaerobes in their oral cavity compared to BF infants (Suppl. Fig. 9). Despite the high oxygen exposure in the oral cavity, various microenvironments and mechanisms, such as biofilm formation, create suitable niches for obligate anaerobic microbes.
A key player in the infant’s oral microbiome is Neisseria, an obligate aerobe that plays a major role in biofilm-based oral microbiome networks. In fact, Neisseria can protect obligate anaerobes from oxygen (72), likely by consuming it through respiration.
In our networks, Neisseria was present at all tps, primarily co-occurring with obligate anaerobes (e.g. Porphyromonas, Fusobacterium, Lachnoanaerobaculum) and facultative anaerobes (e.g. Haemophilus, Streptobacillus, Leptotrichia), but not with other obligate aerobes such as Bergeyella (Fig. 2b). Interactions between Streptococcus species and Veillonella are typically found during the early stages of oral biofilm formation (73). However, an exception was observed in the networks of NBF infants, where directed nodes between Neisseria and obligate aerobic genera such as Flavobacterium (M02) and Bergeyella (M07), were seen. Neisseria was more abundant in NBF infants (Aldex2 at M04 - M07, M09, M11 Suppl. Fig. 4) and showed high centrality in the networks (Fig. 2b), indicating its prominent role in the oral microbiome of NBF infants.
Archaeal signatures are rare and probably transient in the oral microbiome of the first year of life
A high-resolution (taxonomy), highly specific nested PCR approach was used to detect the archaeal taxonomic diversity. The procedure was successful for 224 out of 415 oral samples which gave a high-quality amplicon result (see also overview Suppl. Fig. 2).
Methanobrevibacter was indicated to be the dominant archaeal player in the oral niche (Fig. 3a). All infants harbored archaeal signatures in their oral cavity in at least one tp (Fig. 3b). The sporadic loss and pop-up of these archaeal signatures in the oral cavity underlines our hypothesis that archaea are transient and dependent on environmental input, and we could not define any longitudinal development pattern.
Besides Methanobrevibacter, unclassified Woesearchaeales could be detected in several infants and time points, followed by Methanobacterium, unclassified Nitrososphaeraceae (probably stemming from human skin, (22)) and Candidatus Nitrosotenuis.
Given that a nested PCR approach is unfavorable for drawing conclusions about abundance, the statistics for the archaeal genera were only performed for their presence/absence using Fisher’s t-test. No significant differences could be found for feeding type or birth mode at any time point. As such, we conclude, that young infants do not carry a stable oral archaeome.
Figure panel on the oral archaeome a) Stacked bar chart showing the relative abundance of the 10 most common archaeal genera in the oral cavity per time point (months M01 to M12); data are shown separately for breastfed (BF) and non-breastfed (NBF) infants. b) Beeswarm plot on absence/ presence of archaeal signatures in the oral microbiome per infant and time point (months M01 to M12); time points at which the infants were breastfed are underlaid with gray and infants are sorted by their mode of delivery.
The influence of the oral microbiome on the GIT microbiome decreases within the first year
To evaluate the potential of the oral microbiome as a source of microbes transferred to the GIT over the first year of life, we conducted source tracking analyses. Overall, the oral microbiome contributed minimally to the GIT microbiome with highest contribution at tp M01 (mean 18,27% probability) gradually decreasing over time (mean 7.63% probability at M12) (Fig. 4a).
We also calculated the origin source of individual taxa. An overview of the top30 ASVs (Fig. 4b) showed that at M01, various ASVs found in the GIT were derived from the oral cavity (see below). This could be attributed to the GIT’s limited and unstable colonization by microbes at this early stage, making it more susceptible to influence from the oral microbiome. Additionally, the gastric barrier may not be fully developed at this stage, allowing more microbial transmission from the oral cavity to the GIT.
The main representatives of this early transmission were Bifidobacterium, Staphylococcus and Streptococcus. Streptococcus is the primary genus transferred from oral cavity to the GIT, with one dominant ASV being constantly transmitted over the first year (Fig. 4). In contrast, one Haemophilus ASV (ASV06) gained source tracking potential starting from M07. Bifidobacterium showed a notable peak at tp M07 and M08 (Fig. 4b).
It is notable that the genera of ASVs tracked from the oral cavity to the GIT generally play central roles in microbial networks or exhibit high abundance. Running source tracking in reverse mode (from GIT (source) to oral (sink)), indicated a number of ASVs shared between oral cavity and GIT: Bifidobacterium ASV01, Haemophilus ASV06 and ASV16, Lactobacillus ASV26, Rothia ASV15 and ASV24, Staphylococcus ASV07 and ASV27, Streptococcus ASV02 - ASV05 and Veillonella ASV18 (Suppl. Fig. 10 and Suppl. Fig. 11).
Source tracking was further performed for the archaeal dataset (nested PCR approach, based on presence/absence). It was found that the oral microbiome cannot be considered a potent source for the GIT archaeome, as only in three samples a minimal contribution 0.1% (i29_M02), 0.3% (i23_M06) and 0.3% (i05_M12) was detected.
Source Tracking from oral to GIT a) Source Tracking probability of bacterial taxa being transferred from oral and unknown sources to the GIT as a sink, depicted per time point (months M01 to M12). b) Source Tracking of specific bacterial taxa being transferred from oral source to GIT sink. The top30 ASVs are depicted per time point (months M01 to M12), the length of the bars indicates the log-transformed counts of a taxa. The oxygen requirement of the respective ASV is highlighted by colored stars: pink: obligate anaerobes, yellow: facultative anaerobes, petrol: obligate aerobes.
The GIT microbiome develops steadier throughout the first year of life than the oral microbiome
Similar to the oral microbiome, stool samples from BF infants exhibited a distinct but prolonged transition period from M03 to M08 (PERMANOVA, Fig. 5a). Again, no such obvious time frame was observed for NBF infants, highlighting fewer differences in the GIT microbiome composition between the start and end points of comparison. This is further illustrated by generally lower Bray-Curtis distances in NBF infants compared to BF infants (Suppl. Fig. 12) This pattern is also reflected by our microbial networks and alpha diversity. Similar to the oral microbiome, Shannon diversity, evenness and richness of the GIT microbiome increased over time within the first year of life (LME for Shannon, p = 0.004, Suppl. Fig. 13), with a more pronounced increase in BF infants compared to NBF infants (Shannon diversity, LME p < 0.001). However, these changes were not signicantly different between individual tps (Fig. 5b and Suppl. Fig. 14). In general, alpha diversity and mainly Shannon diversity was again slightly higher in NBF infants than in BF infants (Suppl. Fig. 13, t-test: M04: q = 0.029 M05: q = 0.044; richness, Wilcoxon: M03 q = 0.019, M05: q = 0.021, Suppl. Fig. 14).
Figure panel on the GIT microbiome a) Pairwise beta diversity comparisons (PERMANOVA) from months M01 and M12 to all other timepoints to depict the transition phase. Gray and shaded green colors indicate q-values; data are separated for breastfed (BF) and non-breastfed (NBF) infants. b) Shannon diversity of stool samples depicted for breastfed (BF =light green) and non-breastfed (NBF = lavender) infants with asterisks indicating significant difference (q - values) between those two groups. No significant q - values between tps within one group (BF or NBF). c) Networks on stool samples of BF and NBF infants of selected time points (months M01, M03, M06, M09 and M12) Font size indicates stress centrality, colors indicate oxygen requirement: pink: obligate anaerobes, yellow: facultative anaerobe, petrol: obligate aerobes. d) Stacked bar chart showing the relative abundance of the ten most common bacterial genera in the GIT per time point (months M01-M12). Data are shown separately for breastfed (BF) and non-breastfed (NBF) infants. The oxygen requirement of the respective genera is highlighted by colored stars: pink: obligate anaerobes, yellow: facultative anaerobes, petrol: obligate aerobes.
The GIT microbiome stabilizes by establishing complex anaerobic microbial communities
Within the first year of life, a very complex microbial network was established (BF M012: nodes n = 45, edges n = 78, average number of neighbors n = 3.467; NBF M12: nodes n = 38, edges n = 56, average number of neighbors n = 3.056). In the first months, especially for BF infants, only few bacterial genera were co-occuring and stress centrality in general was low. (Suppl. Fig. 15). As alpha diversity increased over time, the number of genera included in networks also grew. The microbial networks in the NBF infants were more complex from the early months with a higher number of nodes and edges, indicating more microbial interactions compared to BF infants (Fig. 5c).
The GIT microbiome was predominated by various obligate anaerobic microbes at all tps and their relative abundances were constantly increasing within the first year of life (Suppl. Fig. 16, from ∼50% in M01 to ∼70% in M12). Initially, the human GIT contains little amounts of oxygen which is gradually consumed by microbial activity. Indeed, in the first months, some facultatively anaerobic bacteria were still detectable, and played central roles in the bacterial networks (Fig. 5c and all networks in Suppl. Fig. 17, Suppl. Fig. 16), including taxa of the genera Escherichia-Shigella, Rothia, Haemophilus, Staphylococcus, Enterococcus, Lactobacillus and Gemella. This was particularly evident in BF infants, with Haemophilus in particular showing high stress centrality. In contrast, hardly any obligate aerobes could be found in the GIT, correlating with very low oxygen levels after initial oxygen consumption (Suppl. Fig. 16).
The most prominent obligate anaerobe in the GIT was Bifidobacterium (Fig. 5d), consistently representing about 30% (relative abundance) at all tps. Similar to Streptococcus in the oral microbiome, Bifidobacterium was, beyond its predominant abundance, not harboring a central role in the network of the GIT microbiome, yielding only low stress centrality except for few tps (BF: M08; NBF: M01, M02). This may be due to Bifidobacterium’s unique metabolic ability to metabolize human milk oligosaccharides (HMOs), which limits its niche overlap with other microbes.
Interestingly, this was not the case in the GIT of NBF infants which did not receive breastmilk containing HMOs. In these infants, Bifidobacterium likely relied more syntrophic interaction with other microbes. In BF infants, Bifidobacterium mainly co-occurred with Escherichia-Shigella and Enterococcus, whereas in NBF infants it associated with a wider range of partners. Contrary to expectations, Bifidobacterium did not have a higher relative abundance in BF than in NBF infants’ GIT microbiome (Fig. 5d).
Strain tracking of MAGs revealed several Bifidobacterium strains in several infants (Fig. 6). Even though MAGs of overall seven Bifidobacterium species could be detected (Suppl. Fig. 18), only three of them, including B. adolescentis, B. longum and B. pseudocatenulatum, were trackable in just one infant at two consecutive tps (Fig. 6). Bifidobacterium longum, on the other hand, could be tracked in four infants between S3 and M01 (Fig. 6).
Comparing BF and NBF infants, we observed only a few bacterial genera that were significantly differentially abundant between the two groups (Suppl. Fig. 19). Lactobacillus was lower in relative abundance in NBF infants in months M08 to M11 (Aldex2, M08 p = 0.013, M09 p = 0.004, M10 p = 0.006, M11 p = 0.001). Intestinibacter was significantly differentially abundant, showing higher relative abundance in M03, M04 and M08 in NBF infants (Aldex2, M03 p = 0.015, M04 p < 0.001, M08 p = 0.011). This strong differential abundance of Intestinibacter at several time points is interesting, as Intestinibacter did not occur to be prominent in any other analysis. Interestingly, between M05 and M07, no genera were differentially abundant.
Species for which strains could be tracked within one or more infants (number represented by dot size) over time (number of time points represented by the intensity of green color). Stroke style indicates if the strain could be tracked at consecutive timepoints (line) or non-consecutive time points (dashed). The oxygen requirement of the respective strain is highlighted by colored stars: pink: obligate anaerobes, yellow: facultative anaerobes, petrol: obligate aerobes.
Persistent colonization is sparse in the first year of life
A subset of samples was subjected to shotgun sequencing, resulting in the assembly of 133 high-quality MAGs (metagenome-assembled genomes, completeness >90%, contamination <5%), derived from 65 samples from 21 infants. An overview of the samples is provided in Suppl. Fig. 1. Using these data, strains of several bacterial species, in addition to Bifidobacterium, were tracked in several infants at various tps (Fig. 6). Surprisingly, only a few strains could be detected in an infant across more than two or three consecutive tps, which would typically indicate persistent colonization of the lower GIT by that strain. Persistent colonization was only observed for few species, including Aeromonas caviae, three Bifidobacterium species, Blautia A wexlerae, Faecalibacterium prausnitzii_D, several Streptococcus species, e.g. Streptococcus parasanguinis_E and Staphylococcus species, e.g. Staphylococcus lugdunensis, Rothia, and Veillonella.
Also, the archaeal profile did not reveal a steady colonization. Only 134 out of 442 stool samples and 224 out of 415 oral samples gave a high-quality amplicon output (See also overview Suppl. Fig. 2). Archaeal signals were only detected in S3 for infants i20 and i21, but not from M01 to M12. Archaeal presence in the lower GIT was confirmed early in life, but a highly variable and transient pattern was observed both between infants and over time (Fig. 7a and 7b).
Methanobrevibacter was the most predominant archaeal genus in the GIT similar to the oral microbiome. Some infants exhibited several archaeal genera at various tps (e.g. i24 and i26), while others showed only one genus at single tps (e.g. i07 and i09). Statistical analysis using Fisher’s t-test revealed no significant differences in archaeal presence/absence based on feeding type or birth mode. However, However, Methanobrevibacter and Methanosphaera were more common in NBF infants, whereas BF infants displayed a more diverse archaeal pattern, with higher relative abundances of unclassified Nitrososphaeraceae and Candidatus Nitrosocosmicus, possibly due to mouth-to-skin contact during breastfeeding (Fig. 7a). Kraken/Bracken of metagenomic sequences identified five archaeal species: Methanobrevibacter_A_sp900766745, Methanobrevibacter_A_smithii, Methanobrevibacter_A_woesei, Methanosphaera_cuniculi and Methanosphaera_sp900322125, with Methanobrevibacter_A_sp900766745 being the most predominant (Suppl. Fig. 21). All 21 infants with metagenomic data, showed archaeal signatures in their GIT across 41 samples, but at very low relative abundances (<0.07%). The highest archaeal abundances were observed at M12, indicating an increase of archaea in the first year of life. We could show that infants already have archaeal signatures in their upper and lower GIT in the first months of life but colonization takes place late or even after M12.
Figure panel on the GIT archaeome a) Stacked bar chart showing the relative abundance of the 10 most common archaeal genera in the GIT per time point (months M01 to M12); data are shown separately for breastfed (BF) and non-breastfed (NBF) infants. b) Beeswarm plot on absence/ presence of archaeal signatures in the GIT microbiome per infant and time point (months M01 to M12); time points at which the infants were breastfed are underlaid with gray and infants are sorted by their mode of delivery; for infants i20 and i21 no archaeal data could be obtained at any time point.
Differences between BF and NBF infants GIT microbiomes are less pronounced on functional levels
Comparative, functional analyses were performed on metagenome stool samples of tps M01, M06 and M12 (Fig. 8). The very high numbers of functions that were significantly differentially abundant between the tps (DeSeq2, M01-M06 n = 320 with q < 0.05, M06-M12 n = 2,218 with q < 0.05, M01-M12 n = 2,901 with q < 0.05) indicate a very high dynamic of microbial potentials in the first year of life. Picking only those functions with the top20 highest log fold change, we could see that in M01, the functional potential of the GIT microbiome was mainly characterized by growth. This was mainly reflected by basic metabolic pathways being characteristic and a high number of genes that are responsible for metabolism, especially energy metabolism (e.g. oxidative phosphorylation) and carbohydrate metabolism. At M06 in comparison, genes for proteins of secretion systems involved in signaling and cellular processes were overrepresented. Examples for this were transporters and signaling proteins. The aging microbiome again shifted towards metabolism when the infants were one year old (M12), but with more complex pathways covering the food chain down to e.g. methane. Notably, antimicrobial resistance genes could already be found at M06 and M12.
When comparing the functional potential of the GIT microbiome of BF and NBF infants at one tp, a slighter difference was observed (DeSeq2, M01 n = 85 with q < 0.05, M06 n = 19 with q < 0.05, M12 n = 195 with q < 0.05; Suppl. Fig. 20) than it was seen between tps. Additionally, differences between BF and NBF infants GIT microbiomes are less pronounced on functional levels than on taxonomic level.
At M01, no gene was significantly differentially abundant for NBF infants (DeSeq2, all q > 0.05), meaning that all those 85 genes are exclusively associated with BF infants. As most of those functions are also somehow connected with metabolism, the GIT microbiome of infants that receive breast milk could offer a higher number of genes that are needed to metabolize this very complex food. At M06, five genes were exclusive for NBF infants: ABC transporters and proteins for genetic information processing or signaling and cellular processes. In contrast to this, proteins of secretion systems, metabolism and signal transduction were assigned to BF infants. In the GIT microbiome of infants of one year of age (M12), the nature of genes (higher hierarchical levels) that were assigned either to BF or NBF infants are very alike, even though on the lowest hierarchical level they differences were observed. In can be concluded that the GIT microbiome is fulfilling the same grand functions, but with different taxon-dependent genetic inventory.
Hierarchical KEGG functional annotation of genes of highest top30 log fold change rates of pairwise comparison of i) months M01 (orange) and M06 (green) and ii) M06 (green) and M12 (purple); negative log fold change: red, higher in a) M01 (vs. M06) or b) M06 (vs. M12); positive log fold change: blue, higher in a) M06 (vs. M01) or b) M12 (vs. M06); significant q-value indicated by asterisks and base mean indicated by bar charts, colored by the respective time point; abbreviation env. info. processing = environmental information processing.
Discussion
In infants, initial microbiome development is influenced by factors such as mode of delivery and feeding type (74). Throughout the first year of life, additional factors - including the introduction of solid foods, teething, and increased mobility - shape the microbiome’s structure (20, 74, 75). These changes expose infants to new microbes and create diverse environments and conditions for microbial growth, facilitating the establishment of obligate anaerobes with tense networks. Our study provides valuable insights into the early development and transition of the oral microbiome highlighting differences between breastfed (BF) and non-breastfed (NBF) infants. We could determine a distinct time frame in which the oral microbiome transitions most and showed that this time frame is lagged between BF and NBF infants.
Breastfeeding is recognized as a significant factor influencing the GIT but also oral microbiome (76). Our results confirm that breastfeeding notably impacts microbiome composition at several time points for both oral and GIT samples. Specifically, breastfed (BF) infants exhibit a more defined transitional phase in their oral microbiome compared to non-breastfed (NBF) infants. This transitional phase is marked by a decrease in Streptococcus and the emergence of new genera such as Granulicatella, Neisseria, Veillonella, Alloprevotella, and Leptotrichia. It is also characterized by increased alpha diversity of the colonizing species and significant changes in the microbial community as indicated by beta diversity. As this transitional phase occurs earlier in NBF infants (months 1-3) than in BF infants (months 4-6), we can infer that breastfeeding supports a later, but more defined, maturation of the oral microbiome. By month 7, after the BF infants’ transitional phase has ended, the microbiomes of both groups become more similar in terms of alpha and beta diversity, as well as differentially abundant taxa. This convergence is likely influenced by the introduction of solid food, which acts as a leveling factor between BF and NBF infants’ microbiomes. This aspect was already discussed before (77, 78) but still, complete cessation of BM rather than introduction of solid food is the major driver for aligning the microbiomes (79).
Microbial network complexity also differs significantly between BF and NBF infants. NBF infants have more complex networks from the first months, with multiple genera exhibiting high stress centrality and consistent network structure over time. In contrast, the microbial networks in BF infants are less complex, with fewer genera sharing similar niches. This difference in complexity is most distinct in the first six months and can be attributed to the differing nature of breast milk and formula milk. Factors listed by Xiao et al 2020 (13) include transmission of bacteria through breast milk (BM) (80, 81), various milk components influencing the attachment of bacteria to the oral cavity (82) and utilization of different carbohydrates, like HMOs (83, 84). After six months, the microbial network of BF infants becomes more complex, likely reflecting the introduction of solid food and the subsequent increased microbial colonization and co-occurrences for the improved nutrient degradation.
In the first months of life, the human skin significantly contributes to the microbial influx into the oral cavity (8, 70). This is evidenced by the high relative abundances of Staphylococcus, a skin- and mucosa-associated facultative anaerobe. Our data showed no significant differences in Staphylococcus abundance between BF and NBF infants, suggesting substantial skin-to-oral transfer independent of feeding mode. However, from month three onwards, Staphylococcus presence diminished and appeared only sporadically in the co-occurrence networks with low centrality, indicating its transient colonization in the oral cavity during early life.
In addition to the human skin, transmission from other individuals as well as environmental exposures can also be considered as the origin of the normal microbiome of the oral cavity. In fact, the predominant component of the oral microbiome, e.g, Streptococcus is transmitted through these routes (13). This facultatively anaerobic bacterial taxon, is known for its role in carbohydrate metabolism and is in fact considered as a pioneer species in oral microbiome assembly (71). Despite its high relative abundance especially in the first three months (>60% in both BF and NBF infants), Streptococcus exhibited low network centrality, suggesting its inferior co-occurrence with other microbes and its independent functionality. Interestingly, this bacterial taxon showed a higher dominance in BF infants, as reflected by both its relative abundance and network centrality. Streptococcus decreases in abundance, with new microbial members emerging, marking a transitional phase in the oral microbiome.
Microbes from the oral cavity are constantly swallowed and transitioned through the GIT. Source tracking analyses showed that the contribution of the oral microbiome to the GIT microbiome was overall modest and even decreased over time. Key genera such as Bifidobacterium, Staphylococcus, and Streptococcus were identified as being transferred from the oral cavity to the GIT. The presence of these genera in both microbiomes highlights the interconnectedness of the oral and GIT microbiomes in early life. However, it is believed that the similarity of oral and GIT microbiome decreases over time due to the development and maturation of gastric barrier, including gastric pH, motility and enzyme production by one year of age (24, 85–87). Wernroth et at (88) already highlighted that some OTUs are shared between saliva and fecal samples. They showed that one Veillonella OTU is mainly shared and that the similarity between saliva and stool decreased over time (88).
It should be mentioned, that our methods did not allow for a distinction of living and dead microorganisms, and we cannot make definitive statements about colonization status in the early months of development, as microbial signatures might remain detectable throughout the GIT, although the microorganisms from the oral cavity might have died during passage.
The detection of archaea in the infant GIT as early as the first months of life provides new insights into the microbial ecology of the infant GIT. Methanobrevibacter is the predominant archaeal genus, with its abundance increasing over time. The presence of other archaeal genera, however, particularly skin-associated ones in BF infants, indicates a more diverse archaeal community possibly influenced by close contact during breastfeeding. Although archaea are detected at low relative abundance, their presence becomes more pronounced by month 12 (M12), indicating a gradual establishment in the GIT microbiome.
Our findings support findings from other studies that infants under one year already carry archaea in their intestinal tracts (23, 24). Archaea found in human colostrum and breast milk (77) suggest vertical transfer from mother to child during breastfeeding. Other potential sources include cow milk, dairy products (78), and the archaeomes of other humans, exposing both BF and NBF infants to archaeal sources.
The sporadic presence and absence of archaeal signatures over time support the continuous transition of archaea from the environment into and through the infant’s intestinal tract. More frequent detection of archaea in oral samples compared to stool samples implies that the input of archaea exceeds their successful colonization in the lower intestinal tract.
Next to strictly anaerobic archaea like Methanobrevibacter, also anaerobic bacteria played a massive role in the stabilization of microbiomes. Veillonella, and Alloprevotella appeared in the oral cavity and GIT microbiomes, indicating early colonization by anaerobic bacteria. Neisseria, an obligate aerobe, plays a crucial role in facilitating the survival of anaerobes in the oral cavity by creating microenvironments with lower oxygen levels (XXX).
Interestingly, despite the increasing relative abundance of Bifidobacterium over time (which is expected due to the decreasing levels of oxygen), this bacterial taxon does not play a central role in microbial networks, especially in BF infants. This could be due to its unique metabolic niche of HMO conversion, highlighting the metabolic specialization and niche partitioning within the infant GIT microbiome. Notably, the strain tracking of Bifidobacterium in several infants over time, and therefore its persistence, highlights their colonizing potential and possible role in maintaining GIT health and stability. Other strains that we could track were on the one hand Blautia_A wexlerae (obl. anaerobe) and Faecalibacterium prausnitzii_D (obl. anaerobe), both highly beneficial bacterium with anti-inflammatory properties (89) that are known to maintain GIT health by aiding in the production of short-chain fatty acids (SCFAs) (90). Further, positive tracking events could be observed for Streptococcus parasanguinis_E (obl. anaerobe) and Veillonella_A seminalis (obl. Anaerobe), which are both involved in early colonization in the oral cavity but can also be found in the GIT (73, 91). Veillonella parvula_A (obl. anaerobe), which like other Veillonella species, plays a role in maintaining a balanced GIT microbiome (91) could also be tracked over time. On the other hand, we see also bacterial species with not such a clear role, like Staphylococcus lugdunensis (fac. anaerobe), a skin commensal whose presence in the GIT is less understood (92) and can cause severe infections, especially in hospital (93) and Aeromonas caviae (fac. anaerobe), an opportunistic pathogen that is rare in healthy infants (94). Staphylococcus lugdunensis could be tracked even in several infants.
While the GIT microbiomes of BF and NBF infants differed in composition and complexity, the functional potential of these microbiomes is rather influenced by age than by feeding mode as indicated by the functional analysis of GIT metagenomes. Our results indicated a dynamic shift in gene abundance over time, with significant differences between months, but overall functional redundancy across BF and NBF infants.
Conclusion
Our findings underscore the dynamic nature of the microbiome during infancy and the significant impact of breastfeeding on microbial development throughout the entire digestive tract. We could show that the oral and GIT microbiomes of breastfed infants undergo distinct phases of increased dynamics within the first year of life. In contrast, the microbiomes of non-breastfed infants are more mature from the first months, leading to a steadier development without distinct transitional phases in the first year. Additionally, we found that archaeal signatures are present in infants under one year of age, but they do not form a stable archaeome. While the oral microbiome initially influenced the GIT microbiome during infancy, the GIT microbiome gradually stabilized and differentiated over the first year of life. This transition was marked by a decreasing influence of the oral microbiome on the GIT microbiome composition, suggesting a maturation of the GIT microbial community independent of early oral influences. These insights provide valuable insights into the often-overlooked aspects shaping the infant microbiome development.
Funding details
This research was funded in whole or in part by the Austrian Science Fund (FWF) [Grant-DOI 10.55776/KLI784 and 10.55776/DOC31]. For open access purposes, the author has applieda CC BY public copyright license to any author accepted manuscript version arising from this submission. The study was financially supported by the City of Graz (to M.R.P. and C.N.) and the Austrian Commission for UNESCO and L’ORÉAL with the L’OREAL Fellowship for Women in Science (to M-.R.P.). R.M. and M-.R.P. was trained in the Doctoral Program MolMed, T.S. was trained in the Doctoral Program RespImmun and C.N. was trained in the PhD Program DP-iDP at the Medical University of Graz.
Disclosure statement
No potential competing interest was reported by the author(s).
Data availability statement
Data, tables and scripts that support our findings are openly available in our GitHub Repository https://github.com/CharlotteJNeumann/InfantDevelopmentTRAMIC.
Data deposition
The generated 16S rRNA gene amplicon data are accessible in the European Nucleotide Archive under the study accession number PRJEB77729.
Supplementary Figures
A detailed overview of the number of samples, reads and ASVs obtained from recruiting over sample processing to data obtained.
An overview of data obtained for “universal” and archaeal amplicon sequencing approach of oral and stool samples. For each infant (rows) and time point (columns) it is indicated if samples could not be collected (gray), were lost during data processing (pink) or could successfully be obtained (green)
PC plots and stacked bar plots of early life samples of stool (S1, S2, S3) and oral samples (=1, O2) in comparison to the respective samples taken at month M01; data are separated by “universal” and archaeal amplicon sequencing approach and sample type (stool and oral); in each stacked bar plot the top 10 abundant archaeal or bacterial genera are presented.
Differentially abundant bacterial genera in the oral microbiome per time point (months M01 to M12) between breastfed (BF = light green) and non-breastfed (NBF = lavender) infants, performed by Aldex2. Selection was based on the significant p-value of Aldex2 (p < 0.05); significance asterisks indicate BH (Benjamini Hochberg) corrected p-values (q < 0.05). The oxygen requirement of the respective genera is highlighted by colored stars: pink: obligate anaerobes, yellow: facultative anaerobes, petrol: obligate aerobes.
Network on oral samples of BF and NBF infants. Per time point (months M01 to M12), one network is depicted. Font size indicates stress centrality, colors indicate oxygen requirement: pink: obligate anaerobes, yellow: facultative anaerobe, petrol: obligate aerobes.
Alpha diversity of oral samples depicted for breastfed (BF =light green) and non-breastfed (NBF = lavender) infants with asterisks indicating significant difference (q - values) between those two groups. a) Shannon diversity, b) evenness, c) richness. Significant q - values between tps in BF infants are indicated with light green asterisks.
Longitudinal linear mixed effect (LME) for Shannon diversity of oral samples per time point (months M01 to M12); data are separated for breastfed (BF = light green) and non-breastfed (NBF) infants.
Stress centrality in the oral microbiome networks of top 10 bacterial genera per time point (months M01 to M12); data are shown separately for a) breastfed (BF) and b) non-breastfed (NBF) infants.
Stacked bar plots of relative abundances of oral bacterial taxa per time point (months M01 to M12), colored by their oxygen requirements: pink: obligate anaerobes, yellow: facultative anaerobes, petrol: obligate aerobes, grey: not assigned; data are separated for breastfed (BF) and non-breastfed (NBF) infants.
Source Tracking probability of bacterial taxa being transferred from gut and unknown sources to oral as a sink, depicted per time point (months M01 to M12).
Source Tracking of specific bacterial taxa being transferred from gut source to oral sink. The top30 ASVs are depicted per time point (months M01 to M12), the length of the bars indicates the log-transformed counts of a taxa. The oxygen requirement of the respective ASV is highlighted by colored stars: pink: obligate anaerobes, yellow: facultative anaerobes, petrol: obligate aerobes.
Longitudinal linear mixed effect (LME) on oral and stool data displaying Bray Curtis distance within one infant from one time point (tp) to the consecutive tp, with M01 as reference; colors indicate breastfed (BF = light green) or non-breastfed (NBF = lavender) infants.
Longitudinal linear mixed effect (LME) for Shannon diversity of stool samples per time point (months M01 to M12); data are separated for breastfed (BF = light green) and non-breastfed (NBF) infants.
Alpha diversity of stool samples depicted for breastfed (BF =light green) and non-breastfed (NBF = lavender) infants with asterisks indicating significant difference (q - values) between those two groups. a) Shannon diversity, b) evenness, c) richness. No significant q - values between tps within one group (BF or NBF).
Stress centrality in the gut microbiome networks of top 10 bacterial genera per time point (months M01 to M12); data are shown separately for a) breastfed (BF) and b) non-breastfed (NBF) infants.
Stacked bar plots of relative abundances of bacterial taxa of the gut per time point (months M01 to M12), colored by their oxygen requirements: pink: obligate anaerobes, yellow: facultative anaerobes, petrol: obligate aerobes, gray: not assigned; data are separated for breastfed (BF) and non-breastfed (NBF) infants.
Network on stool samples of breastfed (BF) and non-breastfed (NBF) infants. Per time point, one network is depicted. Font size indicates stress centrality, colors indicate oxygen requirement: pink: obligate anaerobes, yellow: facultative anaerobe, petrol: obligate aerobes.
Beeswarm plot on Bifidobacterium species in stool samples for metagenomics sequencing for time points S2, S3 and months M01, M06 and M12; time points at which the infants were breastfed are underlaid with gray and infants are sorted by their mode of delivery.
Differentially abundant bacterial genera in the gut microbiome per time point between breastfed (BF = light green) and non-breastfed (NBF = lavender) infants, performed by Aldex2. Selection was based on the significant p-values of Aldex2 (p < 0.05); significance asterisks indicate BH (Benjamini Hochberg) corrected p-values (q < 0.05). The oxygen requirement of the respective genera is highlighted by colored stars: pink: obligate anaerobes, yellow: facultative anaerobes, petrol: obligate aerobes.
Hierarchical KEGG functional annotation of genes of highest top30 log fold change rates between breastfed (BF = light green) and non-breastfed (NBF = lavender) infants per time point (months a) M01, b) M06 and c) M12): negative log fold change: red, higher in BF infants; positive log fold change: blue, higher in NBF infants; significant q-value indicated by asterisks and base mean indicated by bar charts colored by respective group; abbreviation env. info. processing = environmental information processing.
Suppl. Fig. 21: Beeswarm plot on Methanobrevibacter species in stool samples for metagenomics sequencing for time points S2, S3 and months M01, M06 and M12; time points at which the infants were breastfed are underlaid with gray and infants are sorted by their mode of delivery.
Acknowledgments
The authors thank the Medical University of Graz for the computational resources of the MedBioNode and the Life Science Compute Cluster (LiSC) operated by the Computational Systems Biology group at the University of Vienna. We thank the Medical University of Graz ZMF Galaxy Team: Core Facility Computational Bioanalytics, Medical University of Graz, funded by the Austrian Federal Ministry of Education, Science, and Research, Hochschulraum-Strukturmittel 2016 grant as part of BioTechMed Gral 2016 grant as part of BioTechMed Graz. We thank the Department of Obstetrics and Gynecology of Medical University of Graz for the sample collection, namely Bettina Amtmann and Petra Winkler. A special thanks goes to the participants of this study for providing samples and information.
Conceptualization was done by the following: M-.R.P., E.J-.K., and C.M-.E.
Methodology was done by the following: C.J.N., M-.R.P., T.S., C.K., A.M., and C.M-.E.
Formal analysis was done by the following: C.J.N.
Investigation was done by the following: C.J.N., M-.R.P., and P.Y.W.
Writing—original draft was done by the following: C.J.N., R.M., P.Y.W. and C.M-.E.
Writing—review and editing was done by the following: C.J.N., R.M., M-.R.P., P.Y.W., T.K., V.H., T.S., P.M., A.M., C.K., E.J-.K., and C.M-.E.
Visualization was done by the following: C.J.N.
Supervision was done by the following: C.M-.E. and E.J-.K.
Project administration was done by the following: C.J.N-, M-R.P.
Funding acquisition was done by the following: C.J.N., M-.R.P., C.M-.E., and E.J-.K.
References
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.
- 87.↵
- 88.↵
- 89.↵
- 90.↵
- 91.↵
- 92.↵
- 93.↵
- 94.↵
