Early-life skin microbiota in hospitalized preterm and full-term infants

Early-life skin microbiota in hospitalized preterm and full-term infants Background: The infant skin microbiota may serve as a reservoir of bacteria that contribute to neonatal infections and stimulate local and systemic immune development. The objectives of our study were to characterize the skin microbiota of preterm and full-term infants during their birth hospitalization and describe its relationship to the microbiota of other body sites and the hospital environment. Results: We conducted a cross-sectional study of 129 infants, including 40 preterm and 89 full-term infants. Samples were collected from five sites: the forehead and posterior auricular scalp (skin upper body); the periumbilical region, inguinal folds, and upper thighs (skin lower body); the oral cavity; the infant’s immediate environment; and stool. Staphylococcus, Streptococcus, Enterococcus, and enteric Gram-negative bacteria including Escherichia and Enterobacter dominated the skin microbiota. The preterm infant microbiota at multiple sites had lower alpha diversity and greater enrichment with Staphylococcus and Escherichia than the microbiota of comparable sites in full-term infants. The community structure was highly variable among individuals but differed significantly by body site, postnatal age, and gestational age. Source tracking indicated that each body site both contributed to and received microbiota from other body sites and the hospital environment. Conclusion: The skin microbiota of preterm and full-term infants varied across individuals, by body site, and by the infant’s developmental stage. The skin harbored many organisms that are common pathogens in hospitalized infants. Bacterial source tracking suggests that microbiota are commonly exchanged across body sites and the hospital environment as microbial communities mature in infancy. Keywords: Cutaneous, Microbiome, Neonate, Neonatal intensive care unit, Staphylococcus, Escherichia Background infections are caused by bacteria that are known to colonize After birth, the infant’s skin and mucosal surfaces are ex- the skin, such as Staphylococcus epidermidis [2]. Physical posed to a variety of maternal and environmental microbes and functional differences in the immature skin of preterm that may colonize the newborn. While our understanding infants may alter the resident microbiota relative to of the development of the fecal microbiome in infancy has full-term infants [3]. Furthermore, there are major expanded greatly over the past decade, acquisition and differences in the early-life exposures of preterm infants succession of the skin microbiota is less well-studied. The compared to full-term infants, including frequent treatment skin undergoes dynamic structural and functional changes with antibiotics, use of invasive lines and tubes, limited in infancy that may influence the development of the skin skin-to-skin contact with parents, and prolonged microbiome, including shifts in pH, water content, transe- hospitalization. Understanding the development of skin pidermal water loss, and sebum production [1]. The extent microbiota and its relationship to other body sites may be to which skin maturation, clinical factors, and environmen- of particular importance in this vulnerable population. tal exposures shape the neonatal skin microbiome is not The skin acts as a physical barrier and immunologic well understood. In preterm infants, many invasive interface to the external world including the local micro- biota. The resident microbiota and immune system provide * Correspondence: patrick.seed@northwestern.edu competitive exclusion of would-be pathogens. Interactions Department of Pediatrics, Northwestern University, 310 E. Superior, Morton between the infant’s developing immune system and the 4-685, Chicago, IL 60611, USA early-life microbiota stimulate immune development, Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Younge et al. Microbiome (2018) 6:98 Page 2 of 11 maturation, and tolerance. In germ-free mice, resident skin Research). We performed PCR to amplify the V4 region T cells exhibit attenuated cytokine responses in response to of the 16S rRNA gene with barcode-indexed 515F-806R inflammatory stimuli [4]. In conventional mice, microbial primers using previously described methods [7]. PCR colonization of the skin during a limited infant develop- reagents were pretreated with heat-labile shrimp DNase mental window leads to an influx of antigen-specific acti- to remove contaminating double-stranded DNA. The vated regulatory T cells into the skin and the development DNase was inactivated by heating at 65 °C for 10 min of tolerance [5]. Understanding community dynamics of before adding the genomic DNA template. PCR prod- the skin microbiota in early life may reveal strategies to ucts were visualized on a 1.5% agarose gel, and biologic protect against infections and the development of later and environmental samples without a visible gel band of diseases such as atopy [6]. the expected size were removed from further processing. The objective of our study was to characterize the skin Amplicons were pooled in equimolar concentrations, microbiota of preterm and full-term infants during their purified by gel extraction, and sequenced on the Illu- birth hospitalization. Second, we sought to determine mina MiSeq platform in two pools. Extraction controls the relationship of the skin microbiota with other body were processed in the same manner as samples and sites and the hospital environment. We hypothesized included in the sequencing run to control for potential that skin microbial diversity would vary by gestational sources of DNA contamination in the extraction kits or age, postnatal age, and body site, reflecting differences in buffers. environmental exposures and infant development. Further, we hypothesized that the skin microbiota of individual infants would share common features with Sequence processing other sites, suggesting an exchange of microbiota across We used the QIIME platform to demultiplex, filter, and body sites and the hospital environment during the merge paired ends of the sequences [8]. Sequences establishment of microbial colonization. sharing greater than 97% similarity were clustered into operational taxonomic units (OTUs). Taxonomy assign- Methods ments were made by aligning representative sequences Study cohort and sample collection for each OTU to the SILVA bacterial database. The We enrolled preterm (< 37 weeks’ gestational age) and distribution of reads, OTUs, and genera per sample by full-term (≥ 37 weeks’ gestational age) infants in the sample type is presented in Additional file 1: Table S1. neonatal intensive care unit (NICU) or newborn nursery We removed samples with < 100 reads and sparse OTUs during their birth hospitalization. Infants in the newborn that did not have counts of more than 10 in at least 10% nursery roomed in with their mothers. The study was of samples. We also removed OTUs with > 1% abun- approved by the Duke Institutional Review Board dance in the extraction control samples from the ana- (Pro00045553), and written informed consent was obtained lysis, as these OTUs were likely to originate from sample from parents. A single set of samples was obtained from preparation and reagents rather than the study infants each infant at the time of study enrollment. Sterile swabs (Additional file 2: Figure S1). However, we retained one were used to collect the samples in a consistent manner Staphylococcus OTU that was present in the extraction from three body sites: the forehead and posterior auricular controls (3% abundance), but was also the dominant scalp (skin upper body; n = 108); the periumbilical region, Staphylococcus OTU found in the biological samples. inguinal folds, and upper thighs (skin lower body; n = 110); We reasoned that laboratory contamination was unlikely and the oral cavity (n = 123). Stool samples were only to be the predominant source of the Staphylococcus collected if a fresh specimen was available at the time of OTU in the biological samples, given that the OTU sampling (n = 38). In a subset of the preterm and full-term accounted for a greater relative abundance of the micro- infants (n = 61), an additional sample was collected from biota within many biological sites than the extraction the infant’s immediate environment. For these samples, a controls, and in an inverse ratio with other dominant swab was rolled across commonly touched objects and contaminant OTUs. For example, the Staphylococcus immediately adjacent surfaces in the infant’s surroundings, OTU accounted for > 10% of total OTU abundance at including the bassinet or isolette (approximately 5 cm area the skin upper body site, while the most abundant OTU of internal surface and handles), the temperature probe, in the extraction control samples (genus: Caldinitratir- and vital sign monitor (approximately 2 cm area). All uptor; 17% abundance in extraction controls) accounted samples were stored at − 80 °C until further processing. for < 1% of total OTU abundance among the infant skin upper body samples. A median of 283 reads was Sample processing removed as contaminants per infant sample. Sequence Genomic DNA was extracted from swabs using counts were normalized using cumulative sums scaling bead-beating and commercial extraction kits (Zymo in the metagenomeSeq package [9]. Younge et al. Microbiome (2018) 6:98 Page 3 of 11 Statistical analysis 1000–1500 g at birth (very low birth weight). Compared Infant characteristics were described for preterm and to the full-term infants, the premature infants were more full-term infants. Wilcoxon rank-sum tests were used for likely to be multiples (twins or triplets; p < 0.001), to have comparison of continuous variables, and chi-square tests been born by cesarean section (p < 0.001), and to have were used for categorical variables. Analysis and received antibiotics (p <0.001; Table 1). Most of the visualization of the microbial sequencing data were premature infants (65%) were fed by a feeding tube (i.e., performed using R statistical software (version 3.2.2). orogastric, nasogastric, or gastrostomy tube) at the time of Alpha diversity and beta diversity measures were exam- sampling, with approximately half (48%) receiving mostly ined using functions within the Phyloseq package [10]. breast milk feeds. Major morbidities among the preterm We used adonis permutational multivariate analysis of infant cohort are presented in Additional file 3: Table S2. variance (PERMANOVA) of generalized UniFrac (alpha None of the infants had positive blood cultures during = 0.5) and Bray-Curtis distances with 999 permutations their hospitalization. to compare the microbiota community structure across body sites, with and without individual subjects included Microbiota composition and diversity by site and as a nested variable [11]. We applied PERMANOVA to gestational age evaluate the association between the skin microbiota A total of 440 infant samples were analyzed. The median and clinical characteristics, including gestational age number of samples collected per subject was 4 (IQR 3– (preterm vs. full term), postnatal age (< 3 vs. ≥ 3 days), 4). A total of 138 OTUs were included in the analysis diet (mostly human milk, mostly formula, or no feeds at following removal of sparse OTUs as well as 14 OTUs the time of sampling), delivery mode (vaginal vs. that were present in > 1% abundance in extraction cesarean delivery), and antibiotic use (any previous controls (Additional file 2: Figure S1). The dominant exposure), nested by sequencing run. Differences in rela- bacterial phyla within each site were Proteobacteria and tive abundance and the presence or absence of bacterial Firmicutes (Fig. 1a). At the genus level, taxa with the taxa between sample sites and gestational age groups greatest relative abundance in the skin microbiota were determined using the zero-inflated log-normal included Staphylococcus, Streptococcus, Haemophilus, mixture model (fitFeatureModel) in metagenomeSeq [9]. Enterococcus, and multiple genera in the family Entero- These comparisons were made at the level of bacterial bacteriaceae including Escherichia, Enterobacter, and genus or the lowest taxonomic classification for OTUs Serratia (Fig. 1b). Streptococcus was the most abundant that could not be assigned at the genus level. bacterial genus within the oral cavity. The dominant To investigate the relationship between body sites and the bacterial genera within the fecal samples included Acine- environment, we used a Bayesian microbial source-tracking tobacter, Escherichia, Haemophilus,and Enterobacter.Of model to estimate the proportion of microbiota within each note, the majority of these stool samples were collected site that originated from the other sites [12]. The model was in the first days of life [median (interquartile range) age first applied to intraindividual site-source pairs among 1(0–2) day] and therefore represent meconium, which complete cases (i.e., infants with no missing data), then is known to have a distinct microbiota compared to repeated with all subjects included in the model to examine infant feces collected at later time points [13]. interindividual site-source relationships. We used principal coordinate analysis (PCoA) of gener- alized UniFrac distances to examine the relationship of Results microbial communities across body sites (Fig. 2)[11]. Study cohort Microbiota community structure differed by site (adonis Samples were collected from a total of 129 infants, includ- PERMANOVA R =0.049; n = 379 samples), both when ing 89 full-term infants and 40 preterm infants (Table 1). comparing sites across all subjects (p =0.001) as well as Seventy-seven of the 89 full-term infants (87%) were when comparing sites nested within individual subjects (p healthy infants who roomed in with their mothers during = 0.001). However, there was a high degree of variation their birth hospitalization. The primary diagnoses for the between samples without distinct spatial separation by 12 full-term infants who were admitted to the NICU are body site. Differentiation between sites was greater among listed in Additional file 3: Table S2. A total of 85 (66%) of samples collected after the immediate postnatal period the infants were sampled in the immediate postnatal (i.e., postnatal age ≥ 3 days; Fig. 2). We observed similar period (< 3 days of age), while 44 (34%) were sampled at relationships between sites using the non-phylogenetic later time points. The median postnatal age at the time of Bray-Curtis distance metric (adonis R =0.043, p= 0.001; sampling was greater among the preterm infants than the Additional file 4: Figure S2A-B). Given that removal of full-term infants (p < 0.001; Table 1). Twenty-four (60%) contaminant OTUs may alter community composition, of the preterm infants weighed less than 1000 g at birth we also examined the relationship between body sites (extremely low birth weight), and 9 (23%) infants weighed including the contaminant OTUs in the analysis. Here Younge et al. Microbiome (2018) 6:98 Page 4 of 11 Table 1 Infant characteristics Preterm (N = 40) Term (N = 89) Baseline characteristics Birth weight (g), median (range) 845 (540–2508) 3365 (1820–4440) Gestational age (weeks), median (range) 27 (23–36) 39 (37–42) Female sex, n (%) 28 (70) 46 (52) Multiple gestation, n (%) 13 (33) 7 (8) Race, n (%) White 19 (48) 45 (51) Black or African American 20 (50) 28 (31) Asian 0 (0) 2 (2) Native Hawaiian or other Pacific Islander 0 (0) 1 (1) Unknown or not reported 1 (3) 13 (15) Hispanic or Latino, n (%) 0 (0) 5 (6) Mother hospital days prior to delivery, median (range) 3.5 (0–15) 0 (0–6) Labor prior to delivery, n (%) 21 (53) 62 (70) Prolonged rupture of membranes > 18 h, n (%) 6 (17) 14 (17) Cesarean section, n (%) 30 (86) 36 (43) Clinical factors at time of sampling Age at sampling (d), median (range) 42 (1–252) 1 (0–122) Location, n (%) Neonatal intensive care unit 40 (100) 12 (13) Mother’s room 0 (0) 77 (87) Type of bed at time of sampling, n (%) Open crib 18 (45) 87 (98) Warmer bed 1 (3) 2 (2) Isolette 21 (53) 0 (0) Diet, n (%) Mostly breast milk 19 (48) 56 (63) Mostly formula 15 (38) 33 (37) Any receipt of breast milk 32 (80) 63 (71) No feeds prior to sampling 6 (15) 0 (0) Primary feeding route, n (%) Breastfeeding 1 (3) 50 (56) Bottle 7 (18) 38 (43) Feeding tube 26 (65) 1 (1) No feeds prior to sampling 6 (15) 0 (0) Previous antibiotic exposure, n (%) 37 (93) 15 (17) again, we found that body site accounted for a minor samples was not attributable to any of the measured proportion of the variation between samples (R =0.056, clinical covariates, and between-site variation was p =0.001; Additional file 4: Figure S2C-D). notable within many of the individual subjects as well as The community structure of the skin microbiota (n = twin pairs (Additional file 5: Figure S3). 218 samples) differed by gestational age (adonis R = Next, we used zero-inflated log-normal mixture models 0.016, p = 0.018; generalized UniFrac distances) and to identify discriminatory bacterial genera between the postnatal age (R = 0.016, p = 0.024), but not by antibiotic skin and other body sites among all infants, regardless of exposure (p = 0.211), diet (p = 0.305), or delivery mode gestational age category. The skin upper body and lower (p = 0.089). Much of the variation in β-diversity between body sites differed only in the relative abundance of Younge et al. Microbiome (2018) 6:98 Page 5 of 11 Fig. 1 The relative abundance of bacterial genera at the level of phylum (a) and genus (b) for each body site. The lowest taxonomic classification is given for OTUs that were unable to be assigned a genus-level taxonomic classification Fig. 2 Principal coordinate analysis (PCoA) of generalized UniFrac distances. Each dot represents a sample and each color indicates a body site: stool (blue), skin upper body (green), skin lower body (red), and oral cavity (purple). The distribution of samples by body site is shown along the first and second axes of the PCoA plot. Along the first axis (PC1), the sample distribution differed significantly between the stool and skin upper body (p = 0.0048), the stool and oral cavity (p < 0.0001), skin upper body and skin lower body (p = 0.0049), skin upper body and oral cavity (p= 0.0049), and skin-lower body and oral cavity (p< 0.0001), but not between the stool and skin lower body (p = 0.1842; p values determined by pairwise Wilcoxon rank sum tests with Benjamini-Hochberg correction). Along the second axis (PC2), the oral samples differed from the skin upper body (p< 0.0001) and skin lower body (p< 0.0001), but other sites were not significantly different Younge et al. Microbiome (2018) 6:98 Page 6 of 11 Streptococcus, which was present in greater abundance in by delivery mode (C-section vs. vaginal delivery) at any the upper body site (Additional file 6: Table S3). The oral of the sites (data not shown). cavity had a significantly higher abundance of Streptococ- cus, Rothia,and Gemella than both the skin upper and Microbiota in the hospital environment lower body sites, and greater abundance of Neisseria and Environmental samples were obtained in a subset of the Haemophilus than the skin lower body. The stool con- infants (20 preterm infants and 41 full-term infants) to tained greater enrichment with Aeromonas, Enterobacter, determine the relationship between each infant’s skin Enterobacteriaceae (genus not classified), and an uncul- microbiota with the hospital environment. All of the tured bacterium of the class γ-Proteobacteria than the preterm infants and 6 (15%) of the full-term infants were skin upper body. In comparisons based on the presence or located in the NICU at the time of sampling; the absence of bacterial taxa, Corynebacteriaceae was less remaining 35 (85%) full-term infants were located in likely to be present in the stool than the skin upper body their mothers’ rooms. The environment was enriched in (OR 0.16, 95% CI 0.03–0.56, p =0.046). There were no adj many of the same genera found in the infant skin and taxa with significant differences in relative abundance other body sites, including Escherichia, Staphylococcus, between the stool and the skin lower body site. After the and Streptococcus (Additional file 8: Figure S4A). These immediate postnatal period (≥3days old; n = 77 samples), taxa dominated the environmental samples from both the skin contained a greater relative abundance of preterm and full-term infants. However, the preterm Staphylococcus (p <0.001), Veillonella (p =0.038), adj adj infant environmental microbiota had a greater relative Finegoldia (p = 0.007), and lower abundance of Neisseria adj abundance of several taxa, including members of the (p =0.038) and Enterobacter (p = 0.017) than in the adj adj Gammaproteobacteria class that were also more first days of life (n = 141 samples). abundant in the skin microbiota of the preterm infant We examined differences in the microbiota between compared to the full-term infant (Additional file 7: preterm infants and full-term infants (Fig. 3). The skin Table S4). Taxa with greater relative abundance in the microbiota of full-term infants (n = 147 samples) con- skin microbiota than the environment included Entero- tained a greater relative abundance of Neisseria, while coccus (p =0.034, skin upper; p = 0.003, skin lower), the preterm infants (n = 71 samples) had a greater abun- Streptococcus (p =0.008, skin upper), Bacteroides (p = dance of Staphylococcus, Bacillus, Escherichia, Entero- 0.039, skin upper; p = 0.026, skin lower), Anaerobacillus bacter, and other taxa within the Gammaproteobacteria (p = 0.031, skin lower), and Enterobacter (p= 0.026, skin class (Additional file 7: Table S4). The skin microbiota of lower). Median generalized UniFrac distances between the preterm infants was also more likely to contain bacteria infant microbiota and the environment were lower among within the Stenotrophomonas genus (OR 2.60; 95% CI preterm infants than full-term infants, but the difference 1.28–5.32; p = 0.037). The oral cavity of preterm adj was only statistically significant for the stool samples infants (n = 39 samples) had a greater abundance of (Additional file 8: Figure S4B). Stenotrophomonas, Lactococcus, and Enterobacter than the full-term infants (n = 84 samples). Within individual subjects, between-site generalized UniFrac distances Relationship between body sites and the hospital were significantly higher in the full-term infants than the environment preterm infants (Fig. 3b)[11]. We used bacterial source tracking to explore the We compared alpha diversity, as measured by the predicted sources of microbiota within each body site Shannon diversity index, between body sites (Fig. 3c). (Fig. 4). First, we applied the source-tracking model Alpha diversity was significantly lower among oral using only intra-individual site-source pairs. The pre- samples than the stool (p = 0.014), skin upper body (p < dicted sources of microbiota for each site varied between 0.0001), and skin lower body (p< 0.0001; Fig. 3c). Both individual subjects, but the majority of the microbiota skin sites (upper and lower bodies) had significantly within each site was attributable to the infant’s other lower alpha diversity among preterm infants than body sites (Fig. 4a). The skin microbiota appeared to full-term infants (p = 0.030 and p = 0.017, respectively). both receive microbiota from and contribute to the Alpha diversity within the stool and oral cavity were not microbiota of other body sites (Fig. 4a, c). The hospital significantly different between gestational age groups. environment was the predicted source for approximately The diversity of the fecal microbiota was lower among one quarter of the microbiota at each body site. The samples collected at postnatal age > 2 days than samples source-tracker model was then reapplied to investigate collected in the first days of life (median 1.89 vs. 3.12, p site-source relationships between infants. In this model, = 0.037), but there were no significant differences in the vast majority of the microbiota within each site diversity by postnatal age within the skin or oral could not be attributed to a known source (Fig. 4b). microbiota. Shannon indices did not differ significantly These findings suggest that the infant microbiota is Younge et al. Microbiome (2018) 6:98 Page 7 of 11 Fig. 3 Comparison of the preterm and full-term infant microbiota across body sites. a The mean proportion (per sample) of the top OTUs within each body site in preterm and full-term infants. b Intra-individual generalized UniFrac distances between body sites in preterm and full-term infants. Between-site distances were greater in full-term infants than preterm infants (median 0.75 vs 0.70, p = 0.006). c Shannon diversity across body sites and gestational age groups. Alpha diversity was significantly lower among oral samples than the stool, skin upper body, and skin lower body (p values determined by pairwise Wilcoxon rank sum tests with Benjamini-Hochberg correction). Shannon diversity did not differ significantly across the other body sites. Alpha diversity was lower in the skin among preterm infants compared to the full-term infants. *p < 0.05, **p < 0.01 more closely related to their own environment and other body and skin lower body sites, the relative abundance of body sites than to the microbiota of other infants. bacteria within the Proteobacteria phylum was higher in the skin lower body site, potentially reflecting fecal Discussion contamination of the skin. The high abundance of In this study, we characterized the skin microbiota of potential pathogens in this skin region is worth noting hospitalized preterm and full-term infants and described given the frequent use of umbilical and femoral central its relationship to other body sites and the hospital envir- vascular catheters in critically ill infants. Our bacterial onment. The skin microbiota varied between individuals source-tracking model indicated that the skin microbiota and by gestational age, postnatal age, and body region. It both acquires and contributes microbiota to other body was enriched in typical skin-associated bacteria such as sites, suggesting that body sites can serve as bacterial Streptococcus and Staphylococcus but also in many taxa reservoirs for one another in infancy. that are typically associated with the gut microbiome, Studies in adults have shown that the microbiota is including Escherichia, Enterobacter,and Enterococcus. highly differentiated in structure and function across Many of the abundant taxa in the skin microbiota are body sites and skin surfaces [14, 15]. In our study, we common causes of late-onset sepsis in preterm infants [2]. found differences in microbiota community structure, While many of same genera were present in skin upper alpha diversity, and relative abundances of bacterial taxa Younge et al. Microbiome (2018) 6:98 Page 8 of 11 Fig. 4 Source tracking of microbiota across body sites. The mean proportion of microbiota within each site (“sink”) attributable to each source are shown among intraindividual (a) and interindividual (b) sink-source pairs. Intraindividual relationships between sites are further depicted (c), with the weight of arrows between sites showing the relative contribution of each source between sites. In general, however, discrimination the living environment alters the development of the skin between body sites was relatively weak. Similar findings microbiota [21]. The open landscape and pro-tolerogenic have been reported in other preterm and full-term immune bias of the neonate may make the skin more neonatal cohorts, with greater distinction between sites susceptible to invasion by environmental microbes than in occurring in early infancy [16–19]. We found that site later life, but the timing of this window of susceptibility differentiation was greater for samples collected after the and the specific host and microbial community factors first two postnatal days, suggesting the rapid develop- that confer colonization resistance to environmental ment of niche selection. The dynamic progression from microbes are unclear. In adults, skin microbial communi- nonspecific colonization with a common inoculum to ties are largely stable over time and microbial community the formation of site-specific microbial communities is niches appear to be maintained primarily by growth of likely driven, in part, by concurrent physical, chemical, indigenous strains rather than the acquisition of new and immunologic changes in the neonatal period and strains from the environment [14]. In our study, we found early infancy. A recent study illustrated that in two substantial overlap between the infant’s skin, gut, and oral preterm infants, identical strains colonized the infant’s microbiota and the hospital environment. While we oral cavity, skin, and gut, but demonstrated differential cannot fully determine the directionality of transfer of growth rates by site [20]. microbiota between the infant and the environment from The extent to which environmental acquisition of our cross-sectional data, a substantial proportion of microbes contributes to the development of skin microbial microbiota in the infant skin and other sites were attrib- communities is not entirely understood. A recent study uted to the hospital environment in our source tracking demonstrated that the skin microbiota differs between model. In addition to the physical environment, infants children living in rural and urban environments, particu- acquire microbiota from their mothers through delivery, larly during early childhood (ages 1–4), suggesting that skin-to-skin contact, and breast milk feeding [22]. We did Younge et al. Microbiome (2018) 6:98 Page 9 of 11 not collect samples from mothers and thus could not no association between neonatal sepsis and the skin determine the relative contribution of maternal and microbiome, but the study was limited by small sample size environmental sources of microbiota in our study. Critic- (n = 12) and inconsistent timing of sample collection ally ill infants with prolonged hospital stays may acquire a between subjects relative to the onset of sepsis [26]. greater proportion of their initial microbiota from the The study reported herein has limitations. The hospital environment than healthy infants, given their lack cross-sectional nature of our study design limited our of physical contact with their mothers and the often ability to delineate maturational changes and interac- delayed introduction of human milk feedings. Further tions between body sites and the environment over time, study is needed to understand the acquisition and persist- and to explore the relationships between the skin ence of the environmental microbiota in these infants and microbiota and relevant clinical outcomes. The median its potential effects on subsequent maturation of the postnatal age at the time of sampling was lower in the microbiome and clinical outcomes. full-term infants than the preterm infants. Differences in The influence of delivery mode on the neonatal micro- the time of sampling combined with the cross-sectional biome has been a major area of interest. Several studies study design may have confounded the comparisons have suggested a difference in the microbiota of infants between gestational age and postnatal age groups. We who are born by vaginal delivery versus those born by lacked the genetic resolution to be able to determine C-section [18, 23–25]. A recent study by Chu et al. exam- strain variation within taxa across sites and individuals. ined the influence of delivery mode on infant microbial Further, we did not evaluate the functional capacity of communities in a cohort of infants with a mean gesta- the microbiota, limiting our ability to say whether the tional age of 38 ± 2.5 weeks [17]. They identified modest compositional and structural differences we observed differences in the microbiota of infants by delivery mode corresponded to functional differences in microbial among 157 infants sampled at the time of birth, but there communities. Future longitudinal studies directed at elu- were no appreciable differences in microbiota community cidating the interactions between the infant microbiota structure or diversity among 60 infants who had repeat and environmental sources through metagenomics may sampling at age 4–6 weeks. Further, there were no notable provide a more comprehensive understanding of micro- differences in microbial community function by delivery biota assembly in infancy. mode in a subset of infants who were studied by whole-genome shotgun sequencing. In the current study, Conclusions mode of delivery did not appear to have a strong influence In conclusion, the skin microbiota was highly variable on the infant microbiota. We did not see significant differ- across individuals in this large cohort of hospitalized ences in alpha or beta diversity between infants born by full-term and preterm infants. The skin microbiota C-section and those born by vaginal delivery. differed across stages of infant development, shared com- We found that the skin microbiota of preterm infants dif- monalities with the developing microbial communities at fered from that of the full-term infants, with greater enrich- other body sites, and was predicted to be, in part, shaped ment of Staphylococcus and several taxa that are typically by microbiota acquired from the hospital environment. associated with the fecal microbiota, such as Escherichia. Despite sampling at later time points, UniFrac distances Additional files between sites were modestly lower in preterm infants than full-term infants, suggesting less site differentiation. Alpha Additional file 1: Table S1. Raw sequencing reads, OTUs, and genera diversity was also lower in the skin of preterm infants, per sample by sample type. (DOCX 13 kb) which may render the skin microbiota more susceptible to Additional file 2: Figure S1. Contaminant OTUs identified in extraction control samples. A. Relative abundance of bacterial taxa in extraction invasion by pathogens. However, there was overlap between control samples. B. OTUs with greater than 1% relative abundance in many of the full-term and preterm infant samples. The lack extraction controls. These OTUs were excluded from subsequent analyses of strong differentiation between gestational age groups as they were presumed to be contaminants, except the highlighted Staphylococcus OTU that was found to be the dominant Staphylococcus may indicate that the susceptibility of preterm infants to OTU in the biological samples. C. Relative abundance of the contaminant infection is primarily driven by differences in host biology, OTUs (in aggregate) that were excluded from subsequent analyses within including immune function and barrier integrity, rather each sample site. The contaminant OTUs contributed to a minority of the total OTU abundance in each of the sample sites. OTU = operational than by differences in skin colonization. Our data are taxonomic unit. (PPTX 264 kb) limited in that none of the infants developed bloodstream Additional file 3: Table S2. Diagnoses and morbidities among infants infections in our cohort and we did not examine the full admitted to the neonatal intensive care unit. (DOCX 12 kb) genetic potential of the microbiota. It is possible that the Additional file 4: Figure S2. Principal coordinates analysis (PCoA) of microbiota of preterm infants harbored more virulent samples across body sites. PCoA of infant samples excluding contaminant OTUs (A, B) or including contaminant OTUs (C, D). Similar relationships bacterial strains than the full-term infants, despite sharing between body sites are seen using generalized UniFrac distances (A, C) many of thesameOTUs. Arecentlypublished studyfound Younge et al. Microbiome (2018) 6:98 Page 10 of 11 Received: 17 August 2017 Accepted: 18 May 2018 and Bray-Curtis distances (B, D). In panel C, the first and second axes are rotated to keep the orientation of samples consistent with the other panels, but it should be noted that the vertical axis accounts for the majority of the variation between samples in this panel. (PPTX 717 kb) References Additional file 5: Figure S3. Skin and oral microbiota of twin pairs. A. 1. Oranges T, Dini V, Romanelli M. Skin physiology of the neonate and infant: Characteristics of the five twin pairs that were included in the study are clinical implications. Adv Wound Care (New Rochelle). 2015;4(10):587–95. shown, including the gestational age (preterm or full-term), the postnatal 2. Stoll BJ, Hansen N, Fanaroff AA, Wright LL, Carlo WA, Ehrenkranz RA, age (in days) at the time of sample collection, and the percentage of oral Lemons JA, Donovan EF, Stark AR, Tyson JE, et al. Late-onset sepsis in very and skin OTUs that were shared between the infants in each twin pair. low birth weight neonates: the experience of the NICHD Neonatal Research The relative abundance of the top bacterial genera within the skin and Network. Pediatrics. 2002;110(2 Pt 1):285–91. oral microbiomes are shown for the individual infants (Twin A, Twin B) 3. Eichenfield LF, Frieden IJ, Mathes EF, Zaenglein AL. Neonatal and infant within each twin pair (Twin Pairs 1–5). B. Principal coordinates analysis of dermatology. 3rd ed. London: Elsevier Saunders; 2015. 1 online resource samples from the twins based on generalized UniFrac distances. The twin (xiii, 553 pages). pairs (1–5) are grouped by color. (PPTX 363 kb) 4. Naik S, Bouladoux N, Wilhelm C, Molloy MJ, Salcedo R, Kastenmuller W, Additional file 6: Table S3. Bacterial taxa with differences in Deming C, Quinones M, Koo L, Conlan S, et al. Compartmentalized control abundance between body sites. (DOCX 12 kb) of skin immunity by resident commensals. Science. 2012;337(6098):1115–9. 5. Scharschmidt TC, Vasquez KS, Truong HA, Gearty SV, Pauli ML, Nosbaum A, Additional file 7: Table S4. Bacterial taxa with differences in Gratz IK, Otto M, Moon JJ, Liese J, et al. A wave of regulatory T cells into abundance between gestational age groups. (DOCX 13 kb) neonatal skin mediates tolerance to commensal microbes. Immunity. 2015; Additional file 8: Figure S4. The environmental microbiota of preterm 43(5):1011–21. and full-term infants. A. Relative abundance of the top genera in the 6. Kennedy EA, Connolly J, Hourihane JO, Fallon PG, McLean WH, Murray D, Jo hospital environment. B. Generalized UniFrac distances between infant JH, Segre JA, Kong HH, Irvine AD. Skin microbiome before development of body sites and their corresponding environmental samples. Median atopic dermatitis: early colonization with commensal staphylococci at 2 distances were lower among preterm infants. *p < 0.05. (PPTX 118 kb) months is associated with a lower risk of atopic dermatitis at 1 year. J Allergy Clin Immunol. 2017;139(1):166–72. 7. Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Huntley J, Fierer N, Abbreviations Owens SM, Betley J, Fraser L, Bauer M, et al. Ultra-high-throughput microbial NICU: Neonatal intensive care unit; OTU: Operational taxonomic unit community analysis on the Illumina HiSeq and MiSeq platforms. ISME J. 2012;6(8):1621–4. Acknowledgements 8. Kuczynski J, Stombaugh J, Walters WA, Gonzalez A, Caporaso JG, Knight R. We thank the families who participated in the study. Using QIIME to analyze 16S rRNA gene sequences from microbial communities. Curr Protoc Microbiol. 2012;Chapter 1:Unit 1E.5. 9. Paulson JN, Stine OC, Bravo HC, Pop M. Differential abundance analysis for Funding microbial marker-gene surveys. Nat Methods. 2013;10(12):1200–2. The study authors received support from the Hartwell Foundation, the 10. McMurdie PJ, Holmes S. Phyloseq: an R package for reproducible National Institutes of Health (K12 HD043494-14 [NY], R01GM108494 [PS, DB]), interactive analysis and graphics of microbiome census data. PLoS One. the Duke School of Nursing Center for Nursing Research, and the Duke Jean 2013;8(4):e61217. and George Brumley, Jr. Neonatal Perinatal Research Institute. The funding 11. Chen J, Bittinger K, Charlson ES, Hoffmann C, Lewis J, Wu GD, Collman RG, bodies had no role in the design of the study, analysis, interpretation of data, Bushman FD, Li H. Associating microbiome composition with or writing of the manuscript. environmental covariates using generalized UniFrac distances. Bioinformatics. 2012;28(16):2106–13. 12. Knights D, Kuczynski J, Charlson ES, Zaneveld J, Mozer MC, Collman RG, Availability of data and materials Bushman FD, Knight R, Kelley ST. Bayesian community-wide culture- The dataset supporting the conclusions of this article has been deposited in independent microbial source tracking. Nat Methods. 2011;8(9):761–3. the NCBI Sequence Read Archive (SRA) under BioProject PRJNA451534. 13. Moles L, Gomez M, Heilig H, Bustos G, Fuentes S, de Vos W, Fernandez L, Rodriguez JM, Jimenez E. Bacterial diversity in meconium of preterm Authors’ contributions neonates and evolution of their fecal microbiota during the first month of NY, PS, and DB conceptualized and designed the study. NY, FA, and PS life. PLoS One. 2013;8(6):e66986. performed the experiments. NY and PS analyzed the data. All authors 14. Oh J, Byrd AL, Deming C, Conlan S, Program NCS, Kong HH, Segre JA. participated in drafting (NY) or revising (PS, DB, FA) the manuscript and Biogeography and individuality shape function in the human skin approved the final manuscript as submitted. metagenome. Nature. 2014;514(7520):59–64. 15. Costello EK, Lauber CL, Hamady M, Fierer N, Gordon JI, Knight R. Bacterial community variation in human body habitats across space and time. Ethics approval and consent to participate Science. 2009;326(5960):1694–7. The study was approved by the Duke Institutional Review Board 16. Costello EK, Carlisle EM, Bik EM, Morowitz MJ, Relman DA. Microbiome (Pro00045553), and written informed consent was obtained from parents. assembly across multiple body sites in low-birthweight infants. MBio. 2013; 4(6):e00782-00713. Competing interests 17. Chu DM, Ma J, Prince AL, Antony KM, Seferovic MD, Aagaard KM. The authors declare that they have no competing interests. Maturation of the infant microbiome community structure and function across multiple body sites and in relation to mode of delivery. Nat Med. 2017;23(3):314–26. Publisher’sNote 18. Dominguez-Bello MG, Costello EK, Contreras M, Magris M, Hidalgo G, Fierer Springer Nature remains neutral with regard to jurisdictional claims in N, Knight R. Delivery mode shapes the acquisition and structure of the published maps and institutional affiliations. initial microbiota across multiple body habitats in newborns. Proc Natl Acad Sci U S A. 2010;107(26):11971–5. Author details 19. Pammi M, O'Brien JL, Ajami NJ, Wong MC, Versalovic J, Petrosino JF. 1 2 Department of Pediatrics, Duke University, Durham, NC, USA. Department Development of the cutaneous microbiome in the preterm infant: a of Pediatrics, Northwestern University, 310 E. Superior, Morton 4-685, prospective longitudinal study. PLoS One. 2017;12(4):e0176669. Chicago, IL 60611, USA. Duke University School of Nursing, Durham, NC, 20. Olm MR, Brown CT, Brooks B, Firek B, Baker R, Burstein D, Soenjoyo K, USA. Thomas BC, Morowitz M, Banfield JF. Identical bacterial populations colonize Younge et al. Microbiome (2018) 6:98 Page 11 of 11 premature infant gut, skin, and oral microbiomes and exhibit different in situ growth rates. Genome Res. 2017;27(4):601–12. 21. Lehtimaki J, Karkman A, Laatikainen T, Paalanen L, von Hertzen L, Haahtela T, Hanski I, Ruokolainen L. Patterns in the skin microbiota differ in children and teenagers between rural and urban environments. Sci Rep. 2017;7:45651. 22. Pannaraj PS, Li F, Cerini C, Bender JM, Yang S, Rollie A, Adisetiyo H, Zabih S, Lincez PJ, Bittinger K, et al. Association between breast milk bacterial communities and establishment and development of the infant gut microbiome. JAMA Pediatr. 2017;171(7):647–654. 23. Bokulich NA, Chung J, Battaglia T, Henderson N, Jay M, Li H, DL A, Wu F, Perez-Perez GI, Chen Y, et al. Antibiotics, birth mode, and diet shape microbiome maturation during early life. Sci Transl Med. 2016; 8(343):343ra382. 24. Biasucci G, Rubini M, Riboni S, Morelli L, Bessi E, Retetangos C. Mode of delivery affects the bacterial community in the newborn gut. Early Hum Dev. 2010;86(Suppl 1):13–5. 25. Hill CJ,Lynch DB,Murphy K,UlaszewskaM,Jeffery IB,O'Shea CA, Watkins C, Dempsey E, Mattivi F, Tuohy K, et al. Evolution of gut microbiota composition from birth to 24 weeks in the INFANTMET cohort. Microbiome. 2017;5(1):4. 26. Salava A, Aho V, Lybeck E, Pereira P, Paulin L, Nupponen I, Ranki A, Auvinen P, Andersson S, Lauerma A. Loss of cutaneous microbial diversity during first 3 weeks of life in very low birthweight infants. Exp Dermatol. 2017;26(10):861–867. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Microbiome Springer Journals

Early-life skin microbiota in hospitalized preterm and full-term infants

Free
11 pages
Loading next page...
 
/lp/springer_journal/early-life-skin-microbiota-in-hospitalized-preterm-and-full-term-dn2uOFHNDQ
Publisher
BioMed Central
Copyright
Copyright © 2018 by The Author(s).
Subject
Biomedicine; Medical Microbiology; Bioinformatics; Microbial Ecology; Microbiology; Microbial Genetics and Genomics; Virology
eISSN
2049-2618
D.O.I.
10.1186/s40168-018-0486-4
Publisher site
See Article on Publisher Site

Abstract

Background: The infant skin microbiota may serve as a reservoir of bacteria that contribute to neonatal infections and stimulate local and systemic immune development. The objectives of our study were to characterize the skin microbiota of preterm and full-term infants during their birth hospitalization and describe its relationship to the microbiota of other body sites and the hospital environment. Results: We conducted a cross-sectional study of 129 infants, including 40 preterm and 89 full-term infants. Samples were collected from five sites: the forehead and posterior auricular scalp (skin upper body); the periumbilical region, inguinal folds, and upper thighs (skin lower body); the oral cavity; the infant’s immediate environment; and stool. Staphylococcus, Streptococcus, Enterococcus, and enteric Gram-negative bacteria including Escherichia and Enterobacter dominated the skin microbiota. The preterm infant microbiota at multiple sites had lower alpha diversity and greater enrichment with Staphylococcus and Escherichia than the microbiota of comparable sites in full-term infants. The community structure was highly variable among individuals but differed significantly by body site, postnatal age, and gestational age. Source tracking indicated that each body site both contributed to and received microbiota from other body sites and the hospital environment. Conclusion: The skin microbiota of preterm and full-term infants varied across individuals, by body site, and by the infant’s developmental stage. The skin harbored many organisms that are common pathogens in hospitalized infants. Bacterial source tracking suggests that microbiota are commonly exchanged across body sites and the hospital environment as microbial communities mature in infancy. Keywords: Cutaneous, Microbiome, Neonate, Neonatal intensive care unit, Staphylococcus, Escherichia Background infections are caused by bacteria that are known to colonize After birth, the infant’s skin and mucosal surfaces are ex- the skin, such as Staphylococcus epidermidis [2]. Physical posed to a variety of maternal and environmental microbes and functional differences in the immature skin of preterm that may colonize the newborn. While our understanding infants may alter the resident microbiota relative to of the development of the fecal microbiome in infancy has full-term infants [3]. Furthermore, there are major expanded greatly over the past decade, acquisition and differences in the early-life exposures of preterm infants succession of the skin microbiota is less well-studied. The compared to full-term infants, including frequent treatment skin undergoes dynamic structural and functional changes with antibiotics, use of invasive lines and tubes, limited in infancy that may influence the development of the skin skin-to-skin contact with parents, and prolonged microbiome, including shifts in pH, water content, transe- hospitalization. Understanding the development of skin pidermal water loss, and sebum production [1]. The extent microbiota and its relationship to other body sites may be to which skin maturation, clinical factors, and environmen- of particular importance in this vulnerable population. tal exposures shape the neonatal skin microbiome is not The skin acts as a physical barrier and immunologic well understood. In preterm infants, many invasive interface to the external world including the local micro- biota. The resident microbiota and immune system provide * Correspondence: patrick.seed@northwestern.edu competitive exclusion of would-be pathogens. Interactions Department of Pediatrics, Northwestern University, 310 E. Superior, Morton between the infant’s developing immune system and the 4-685, Chicago, IL 60611, USA early-life microbiota stimulate immune development, Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Younge et al. Microbiome (2018) 6:98 Page 2 of 11 maturation, and tolerance. In germ-free mice, resident skin Research). We performed PCR to amplify the V4 region T cells exhibit attenuated cytokine responses in response to of the 16S rRNA gene with barcode-indexed 515F-806R inflammatory stimuli [4]. In conventional mice, microbial primers using previously described methods [7]. PCR colonization of the skin during a limited infant develop- reagents were pretreated with heat-labile shrimp DNase mental window leads to an influx of antigen-specific acti- to remove contaminating double-stranded DNA. The vated regulatory T cells into the skin and the development DNase was inactivated by heating at 65 °C for 10 min of tolerance [5]. Understanding community dynamics of before adding the genomic DNA template. PCR prod- the skin microbiota in early life may reveal strategies to ucts were visualized on a 1.5% agarose gel, and biologic protect against infections and the development of later and environmental samples without a visible gel band of diseases such as atopy [6]. the expected size were removed from further processing. The objective of our study was to characterize the skin Amplicons were pooled in equimolar concentrations, microbiota of preterm and full-term infants during their purified by gel extraction, and sequenced on the Illu- birth hospitalization. Second, we sought to determine mina MiSeq platform in two pools. Extraction controls the relationship of the skin microbiota with other body were processed in the same manner as samples and sites and the hospital environment. We hypothesized included in the sequencing run to control for potential that skin microbial diversity would vary by gestational sources of DNA contamination in the extraction kits or age, postnatal age, and body site, reflecting differences in buffers. environmental exposures and infant development. Further, we hypothesized that the skin microbiota of individual infants would share common features with Sequence processing other sites, suggesting an exchange of microbiota across We used the QIIME platform to demultiplex, filter, and body sites and the hospital environment during the merge paired ends of the sequences [8]. Sequences establishment of microbial colonization. sharing greater than 97% similarity were clustered into operational taxonomic units (OTUs). Taxonomy assign- Methods ments were made by aligning representative sequences Study cohort and sample collection for each OTU to the SILVA bacterial database. The We enrolled preterm (< 37 weeks’ gestational age) and distribution of reads, OTUs, and genera per sample by full-term (≥ 37 weeks’ gestational age) infants in the sample type is presented in Additional file 1: Table S1. neonatal intensive care unit (NICU) or newborn nursery We removed samples with < 100 reads and sparse OTUs during their birth hospitalization. Infants in the newborn that did not have counts of more than 10 in at least 10% nursery roomed in with their mothers. The study was of samples. We also removed OTUs with > 1% abun- approved by the Duke Institutional Review Board dance in the extraction control samples from the ana- (Pro00045553), and written informed consent was obtained lysis, as these OTUs were likely to originate from sample from parents. A single set of samples was obtained from preparation and reagents rather than the study infants each infant at the time of study enrollment. Sterile swabs (Additional file 2: Figure S1). However, we retained one were used to collect the samples in a consistent manner Staphylococcus OTU that was present in the extraction from three body sites: the forehead and posterior auricular controls (3% abundance), but was also the dominant scalp (skin upper body; n = 108); the periumbilical region, Staphylococcus OTU found in the biological samples. inguinal folds, and upper thighs (skin lower body; n = 110); We reasoned that laboratory contamination was unlikely and the oral cavity (n = 123). Stool samples were only to be the predominant source of the Staphylococcus collected if a fresh specimen was available at the time of OTU in the biological samples, given that the OTU sampling (n = 38). In a subset of the preterm and full-term accounted for a greater relative abundance of the micro- infants (n = 61), an additional sample was collected from biota within many biological sites than the extraction the infant’s immediate environment. For these samples, a controls, and in an inverse ratio with other dominant swab was rolled across commonly touched objects and contaminant OTUs. For example, the Staphylococcus immediately adjacent surfaces in the infant’s surroundings, OTU accounted for > 10% of total OTU abundance at including the bassinet or isolette (approximately 5 cm area the skin upper body site, while the most abundant OTU of internal surface and handles), the temperature probe, in the extraction control samples (genus: Caldinitratir- and vital sign monitor (approximately 2 cm area). All uptor; 17% abundance in extraction controls) accounted samples were stored at − 80 °C until further processing. for < 1% of total OTU abundance among the infant skin upper body samples. A median of 283 reads was Sample processing removed as contaminants per infant sample. Sequence Genomic DNA was extracted from swabs using counts were normalized using cumulative sums scaling bead-beating and commercial extraction kits (Zymo in the metagenomeSeq package [9]. Younge et al. Microbiome (2018) 6:98 Page 3 of 11 Statistical analysis 1000–1500 g at birth (very low birth weight). Compared Infant characteristics were described for preterm and to the full-term infants, the premature infants were more full-term infants. Wilcoxon rank-sum tests were used for likely to be multiples (twins or triplets; p < 0.001), to have comparison of continuous variables, and chi-square tests been born by cesarean section (p < 0.001), and to have were used for categorical variables. Analysis and received antibiotics (p <0.001; Table 1). Most of the visualization of the microbial sequencing data were premature infants (65%) were fed by a feeding tube (i.e., performed using R statistical software (version 3.2.2). orogastric, nasogastric, or gastrostomy tube) at the time of Alpha diversity and beta diversity measures were exam- sampling, with approximately half (48%) receiving mostly ined using functions within the Phyloseq package [10]. breast milk feeds. Major morbidities among the preterm We used adonis permutational multivariate analysis of infant cohort are presented in Additional file 3: Table S2. variance (PERMANOVA) of generalized UniFrac (alpha None of the infants had positive blood cultures during = 0.5) and Bray-Curtis distances with 999 permutations their hospitalization. to compare the microbiota community structure across body sites, with and without individual subjects included Microbiota composition and diversity by site and as a nested variable [11]. We applied PERMANOVA to gestational age evaluate the association between the skin microbiota A total of 440 infant samples were analyzed. The median and clinical characteristics, including gestational age number of samples collected per subject was 4 (IQR 3– (preterm vs. full term), postnatal age (< 3 vs. ≥ 3 days), 4). A total of 138 OTUs were included in the analysis diet (mostly human milk, mostly formula, or no feeds at following removal of sparse OTUs as well as 14 OTUs the time of sampling), delivery mode (vaginal vs. that were present in > 1% abundance in extraction cesarean delivery), and antibiotic use (any previous controls (Additional file 2: Figure S1). The dominant exposure), nested by sequencing run. Differences in rela- bacterial phyla within each site were Proteobacteria and tive abundance and the presence or absence of bacterial Firmicutes (Fig. 1a). At the genus level, taxa with the taxa between sample sites and gestational age groups greatest relative abundance in the skin microbiota were determined using the zero-inflated log-normal included Staphylococcus, Streptococcus, Haemophilus, mixture model (fitFeatureModel) in metagenomeSeq [9]. Enterococcus, and multiple genera in the family Entero- These comparisons were made at the level of bacterial bacteriaceae including Escherichia, Enterobacter, and genus or the lowest taxonomic classification for OTUs Serratia (Fig. 1b). Streptococcus was the most abundant that could not be assigned at the genus level. bacterial genus within the oral cavity. The dominant To investigate the relationship between body sites and the bacterial genera within the fecal samples included Acine- environment, we used a Bayesian microbial source-tracking tobacter, Escherichia, Haemophilus,and Enterobacter.Of model to estimate the proportion of microbiota within each note, the majority of these stool samples were collected site that originated from the other sites [12]. The model was in the first days of life [median (interquartile range) age first applied to intraindividual site-source pairs among 1(0–2) day] and therefore represent meconium, which complete cases (i.e., infants with no missing data), then is known to have a distinct microbiota compared to repeated with all subjects included in the model to examine infant feces collected at later time points [13]. interindividual site-source relationships. We used principal coordinate analysis (PCoA) of gener- alized UniFrac distances to examine the relationship of Results microbial communities across body sites (Fig. 2)[11]. Study cohort Microbiota community structure differed by site (adonis Samples were collected from a total of 129 infants, includ- PERMANOVA R =0.049; n = 379 samples), both when ing 89 full-term infants and 40 preterm infants (Table 1). comparing sites across all subjects (p =0.001) as well as Seventy-seven of the 89 full-term infants (87%) were when comparing sites nested within individual subjects (p healthy infants who roomed in with their mothers during = 0.001). However, there was a high degree of variation their birth hospitalization. The primary diagnoses for the between samples without distinct spatial separation by 12 full-term infants who were admitted to the NICU are body site. Differentiation between sites was greater among listed in Additional file 3: Table S2. A total of 85 (66%) of samples collected after the immediate postnatal period the infants were sampled in the immediate postnatal (i.e., postnatal age ≥ 3 days; Fig. 2). We observed similar period (< 3 days of age), while 44 (34%) were sampled at relationships between sites using the non-phylogenetic later time points. The median postnatal age at the time of Bray-Curtis distance metric (adonis R =0.043, p= 0.001; sampling was greater among the preterm infants than the Additional file 4: Figure S2A-B). Given that removal of full-term infants (p < 0.001; Table 1). Twenty-four (60%) contaminant OTUs may alter community composition, of the preterm infants weighed less than 1000 g at birth we also examined the relationship between body sites (extremely low birth weight), and 9 (23%) infants weighed including the contaminant OTUs in the analysis. Here Younge et al. Microbiome (2018) 6:98 Page 4 of 11 Table 1 Infant characteristics Preterm (N = 40) Term (N = 89) Baseline characteristics Birth weight (g), median (range) 845 (540–2508) 3365 (1820–4440) Gestational age (weeks), median (range) 27 (23–36) 39 (37–42) Female sex, n (%) 28 (70) 46 (52) Multiple gestation, n (%) 13 (33) 7 (8) Race, n (%) White 19 (48) 45 (51) Black or African American 20 (50) 28 (31) Asian 0 (0) 2 (2) Native Hawaiian or other Pacific Islander 0 (0) 1 (1) Unknown or not reported 1 (3) 13 (15) Hispanic or Latino, n (%) 0 (0) 5 (6) Mother hospital days prior to delivery, median (range) 3.5 (0–15) 0 (0–6) Labor prior to delivery, n (%) 21 (53) 62 (70) Prolonged rupture of membranes > 18 h, n (%) 6 (17) 14 (17) Cesarean section, n (%) 30 (86) 36 (43) Clinical factors at time of sampling Age at sampling (d), median (range) 42 (1–252) 1 (0–122) Location, n (%) Neonatal intensive care unit 40 (100) 12 (13) Mother’s room 0 (0) 77 (87) Type of bed at time of sampling, n (%) Open crib 18 (45) 87 (98) Warmer bed 1 (3) 2 (2) Isolette 21 (53) 0 (0) Diet, n (%) Mostly breast milk 19 (48) 56 (63) Mostly formula 15 (38) 33 (37) Any receipt of breast milk 32 (80) 63 (71) No feeds prior to sampling 6 (15) 0 (0) Primary feeding route, n (%) Breastfeeding 1 (3) 50 (56) Bottle 7 (18) 38 (43) Feeding tube 26 (65) 1 (1) No feeds prior to sampling 6 (15) 0 (0) Previous antibiotic exposure, n (%) 37 (93) 15 (17) again, we found that body site accounted for a minor samples was not attributable to any of the measured proportion of the variation between samples (R =0.056, clinical covariates, and between-site variation was p =0.001; Additional file 4: Figure S2C-D). notable within many of the individual subjects as well as The community structure of the skin microbiota (n = twin pairs (Additional file 5: Figure S3). 218 samples) differed by gestational age (adonis R = Next, we used zero-inflated log-normal mixture models 0.016, p = 0.018; generalized UniFrac distances) and to identify discriminatory bacterial genera between the postnatal age (R = 0.016, p = 0.024), but not by antibiotic skin and other body sites among all infants, regardless of exposure (p = 0.211), diet (p = 0.305), or delivery mode gestational age category. The skin upper body and lower (p = 0.089). Much of the variation in β-diversity between body sites differed only in the relative abundance of Younge et al. Microbiome (2018) 6:98 Page 5 of 11 Fig. 1 The relative abundance of bacterial genera at the level of phylum (a) and genus (b) for each body site. The lowest taxonomic classification is given for OTUs that were unable to be assigned a genus-level taxonomic classification Fig. 2 Principal coordinate analysis (PCoA) of generalized UniFrac distances. Each dot represents a sample and each color indicates a body site: stool (blue), skin upper body (green), skin lower body (red), and oral cavity (purple). The distribution of samples by body site is shown along the first and second axes of the PCoA plot. Along the first axis (PC1), the sample distribution differed significantly between the stool and skin upper body (p = 0.0048), the stool and oral cavity (p < 0.0001), skin upper body and skin lower body (p = 0.0049), skin upper body and oral cavity (p= 0.0049), and skin-lower body and oral cavity (p< 0.0001), but not between the stool and skin lower body (p = 0.1842; p values determined by pairwise Wilcoxon rank sum tests with Benjamini-Hochberg correction). Along the second axis (PC2), the oral samples differed from the skin upper body (p< 0.0001) and skin lower body (p< 0.0001), but other sites were not significantly different Younge et al. Microbiome (2018) 6:98 Page 6 of 11 Streptococcus, which was present in greater abundance in by delivery mode (C-section vs. vaginal delivery) at any the upper body site (Additional file 6: Table S3). The oral of the sites (data not shown). cavity had a significantly higher abundance of Streptococ- cus, Rothia,and Gemella than both the skin upper and Microbiota in the hospital environment lower body sites, and greater abundance of Neisseria and Environmental samples were obtained in a subset of the Haemophilus than the skin lower body. The stool con- infants (20 preterm infants and 41 full-term infants) to tained greater enrichment with Aeromonas, Enterobacter, determine the relationship between each infant’s skin Enterobacteriaceae (genus not classified), and an uncul- microbiota with the hospital environment. All of the tured bacterium of the class γ-Proteobacteria than the preterm infants and 6 (15%) of the full-term infants were skin upper body. In comparisons based on the presence or located in the NICU at the time of sampling; the absence of bacterial taxa, Corynebacteriaceae was less remaining 35 (85%) full-term infants were located in likely to be present in the stool than the skin upper body their mothers’ rooms. The environment was enriched in (OR 0.16, 95% CI 0.03–0.56, p =0.046). There were no adj many of the same genera found in the infant skin and taxa with significant differences in relative abundance other body sites, including Escherichia, Staphylococcus, between the stool and the skin lower body site. After the and Streptococcus (Additional file 8: Figure S4A). These immediate postnatal period (≥3days old; n = 77 samples), taxa dominated the environmental samples from both the skin contained a greater relative abundance of preterm and full-term infants. However, the preterm Staphylococcus (p <0.001), Veillonella (p =0.038), adj adj infant environmental microbiota had a greater relative Finegoldia (p = 0.007), and lower abundance of Neisseria adj abundance of several taxa, including members of the (p =0.038) and Enterobacter (p = 0.017) than in the adj adj Gammaproteobacteria class that were also more first days of life (n = 141 samples). abundant in the skin microbiota of the preterm infant We examined differences in the microbiota between compared to the full-term infant (Additional file 7: preterm infants and full-term infants (Fig. 3). The skin Table S4). Taxa with greater relative abundance in the microbiota of full-term infants (n = 147 samples) con- skin microbiota than the environment included Entero- tained a greater relative abundance of Neisseria, while coccus (p =0.034, skin upper; p = 0.003, skin lower), the preterm infants (n = 71 samples) had a greater abun- Streptococcus (p =0.008, skin upper), Bacteroides (p = dance of Staphylococcus, Bacillus, Escherichia, Entero- 0.039, skin upper; p = 0.026, skin lower), Anaerobacillus bacter, and other taxa within the Gammaproteobacteria (p = 0.031, skin lower), and Enterobacter (p= 0.026, skin class (Additional file 7: Table S4). The skin microbiota of lower). Median generalized UniFrac distances between the preterm infants was also more likely to contain bacteria infant microbiota and the environment were lower among within the Stenotrophomonas genus (OR 2.60; 95% CI preterm infants than full-term infants, but the difference 1.28–5.32; p = 0.037). The oral cavity of preterm adj was only statistically significant for the stool samples infants (n = 39 samples) had a greater abundance of (Additional file 8: Figure S4B). Stenotrophomonas, Lactococcus, and Enterobacter than the full-term infants (n = 84 samples). Within individual subjects, between-site generalized UniFrac distances Relationship between body sites and the hospital were significantly higher in the full-term infants than the environment preterm infants (Fig. 3b)[11]. We used bacterial source tracking to explore the We compared alpha diversity, as measured by the predicted sources of microbiota within each body site Shannon diversity index, between body sites (Fig. 3c). (Fig. 4). First, we applied the source-tracking model Alpha diversity was significantly lower among oral using only intra-individual site-source pairs. The pre- samples than the stool (p = 0.014), skin upper body (p < dicted sources of microbiota for each site varied between 0.0001), and skin lower body (p< 0.0001; Fig. 3c). Both individual subjects, but the majority of the microbiota skin sites (upper and lower bodies) had significantly within each site was attributable to the infant’s other lower alpha diversity among preterm infants than body sites (Fig. 4a). The skin microbiota appeared to full-term infants (p = 0.030 and p = 0.017, respectively). both receive microbiota from and contribute to the Alpha diversity within the stool and oral cavity were not microbiota of other body sites (Fig. 4a, c). The hospital significantly different between gestational age groups. environment was the predicted source for approximately The diversity of the fecal microbiota was lower among one quarter of the microbiota at each body site. The samples collected at postnatal age > 2 days than samples source-tracker model was then reapplied to investigate collected in the first days of life (median 1.89 vs. 3.12, p site-source relationships between infants. In this model, = 0.037), but there were no significant differences in the vast majority of the microbiota within each site diversity by postnatal age within the skin or oral could not be attributed to a known source (Fig. 4b). microbiota. Shannon indices did not differ significantly These findings suggest that the infant microbiota is Younge et al. Microbiome (2018) 6:98 Page 7 of 11 Fig. 3 Comparison of the preterm and full-term infant microbiota across body sites. a The mean proportion (per sample) of the top OTUs within each body site in preterm and full-term infants. b Intra-individual generalized UniFrac distances between body sites in preterm and full-term infants. Between-site distances were greater in full-term infants than preterm infants (median 0.75 vs 0.70, p = 0.006). c Shannon diversity across body sites and gestational age groups. Alpha diversity was significantly lower among oral samples than the stool, skin upper body, and skin lower body (p values determined by pairwise Wilcoxon rank sum tests with Benjamini-Hochberg correction). Shannon diversity did not differ significantly across the other body sites. Alpha diversity was lower in the skin among preterm infants compared to the full-term infants. *p < 0.05, **p < 0.01 more closely related to their own environment and other body and skin lower body sites, the relative abundance of body sites than to the microbiota of other infants. bacteria within the Proteobacteria phylum was higher in the skin lower body site, potentially reflecting fecal Discussion contamination of the skin. The high abundance of In this study, we characterized the skin microbiota of potential pathogens in this skin region is worth noting hospitalized preterm and full-term infants and described given the frequent use of umbilical and femoral central its relationship to other body sites and the hospital envir- vascular catheters in critically ill infants. Our bacterial onment. The skin microbiota varied between individuals source-tracking model indicated that the skin microbiota and by gestational age, postnatal age, and body region. It both acquires and contributes microbiota to other body was enriched in typical skin-associated bacteria such as sites, suggesting that body sites can serve as bacterial Streptococcus and Staphylococcus but also in many taxa reservoirs for one another in infancy. that are typically associated with the gut microbiome, Studies in adults have shown that the microbiota is including Escherichia, Enterobacter,and Enterococcus. highly differentiated in structure and function across Many of the abundant taxa in the skin microbiota are body sites and skin surfaces [14, 15]. In our study, we common causes of late-onset sepsis in preterm infants [2]. found differences in microbiota community structure, While many of same genera were present in skin upper alpha diversity, and relative abundances of bacterial taxa Younge et al. Microbiome (2018) 6:98 Page 8 of 11 Fig. 4 Source tracking of microbiota across body sites. The mean proportion of microbiota within each site (“sink”) attributable to each source are shown among intraindividual (a) and interindividual (b) sink-source pairs. Intraindividual relationships between sites are further depicted (c), with the weight of arrows between sites showing the relative contribution of each source between sites. In general, however, discrimination the living environment alters the development of the skin between body sites was relatively weak. Similar findings microbiota [21]. The open landscape and pro-tolerogenic have been reported in other preterm and full-term immune bias of the neonate may make the skin more neonatal cohorts, with greater distinction between sites susceptible to invasion by environmental microbes than in occurring in early infancy [16–19]. We found that site later life, but the timing of this window of susceptibility differentiation was greater for samples collected after the and the specific host and microbial community factors first two postnatal days, suggesting the rapid develop- that confer colonization resistance to environmental ment of niche selection. The dynamic progression from microbes are unclear. In adults, skin microbial communi- nonspecific colonization with a common inoculum to ties are largely stable over time and microbial community the formation of site-specific microbial communities is niches appear to be maintained primarily by growth of likely driven, in part, by concurrent physical, chemical, indigenous strains rather than the acquisition of new and immunologic changes in the neonatal period and strains from the environment [14]. In our study, we found early infancy. A recent study illustrated that in two substantial overlap between the infant’s skin, gut, and oral preterm infants, identical strains colonized the infant’s microbiota and the hospital environment. While we oral cavity, skin, and gut, but demonstrated differential cannot fully determine the directionality of transfer of growth rates by site [20]. microbiota between the infant and the environment from The extent to which environmental acquisition of our cross-sectional data, a substantial proportion of microbes contributes to the development of skin microbial microbiota in the infant skin and other sites were attrib- communities is not entirely understood. A recent study uted to the hospital environment in our source tracking demonstrated that the skin microbiota differs between model. In addition to the physical environment, infants children living in rural and urban environments, particu- acquire microbiota from their mothers through delivery, larly during early childhood (ages 1–4), suggesting that skin-to-skin contact, and breast milk feeding [22]. We did Younge et al. Microbiome (2018) 6:98 Page 9 of 11 not collect samples from mothers and thus could not no association between neonatal sepsis and the skin determine the relative contribution of maternal and microbiome, but the study was limited by small sample size environmental sources of microbiota in our study. Critic- (n = 12) and inconsistent timing of sample collection ally ill infants with prolonged hospital stays may acquire a between subjects relative to the onset of sepsis [26]. greater proportion of their initial microbiota from the The study reported herein has limitations. The hospital environment than healthy infants, given their lack cross-sectional nature of our study design limited our of physical contact with their mothers and the often ability to delineate maturational changes and interac- delayed introduction of human milk feedings. Further tions between body sites and the environment over time, study is needed to understand the acquisition and persist- and to explore the relationships between the skin ence of the environmental microbiota in these infants and microbiota and relevant clinical outcomes. The median its potential effects on subsequent maturation of the postnatal age at the time of sampling was lower in the microbiome and clinical outcomes. full-term infants than the preterm infants. Differences in The influence of delivery mode on the neonatal micro- the time of sampling combined with the cross-sectional biome has been a major area of interest. Several studies study design may have confounded the comparisons have suggested a difference in the microbiota of infants between gestational age and postnatal age groups. We who are born by vaginal delivery versus those born by lacked the genetic resolution to be able to determine C-section [18, 23–25]. A recent study by Chu et al. exam- strain variation within taxa across sites and individuals. ined the influence of delivery mode on infant microbial Further, we did not evaluate the functional capacity of communities in a cohort of infants with a mean gesta- the microbiota, limiting our ability to say whether the tional age of 38 ± 2.5 weeks [17]. They identified modest compositional and structural differences we observed differences in the microbiota of infants by delivery mode corresponded to functional differences in microbial among 157 infants sampled at the time of birth, but there communities. Future longitudinal studies directed at elu- were no appreciable differences in microbiota community cidating the interactions between the infant microbiota structure or diversity among 60 infants who had repeat and environmental sources through metagenomics may sampling at age 4–6 weeks. Further, there were no notable provide a more comprehensive understanding of micro- differences in microbial community function by delivery biota assembly in infancy. mode in a subset of infants who were studied by whole-genome shotgun sequencing. In the current study, Conclusions mode of delivery did not appear to have a strong influence In conclusion, the skin microbiota was highly variable on the infant microbiota. We did not see significant differ- across individuals in this large cohort of hospitalized ences in alpha or beta diversity between infants born by full-term and preterm infants. The skin microbiota C-section and those born by vaginal delivery. differed across stages of infant development, shared com- We found that the skin microbiota of preterm infants dif- monalities with the developing microbial communities at fered from that of the full-term infants, with greater enrich- other body sites, and was predicted to be, in part, shaped ment of Staphylococcus and several taxa that are typically by microbiota acquired from the hospital environment. associated with the fecal microbiota, such as Escherichia. Despite sampling at later time points, UniFrac distances Additional files between sites were modestly lower in preterm infants than full-term infants, suggesting less site differentiation. Alpha Additional file 1: Table S1. Raw sequencing reads, OTUs, and genera diversity was also lower in the skin of preterm infants, per sample by sample type. (DOCX 13 kb) which may render the skin microbiota more susceptible to Additional file 2: Figure S1. Contaminant OTUs identified in extraction control samples. A. Relative abundance of bacterial taxa in extraction invasion by pathogens. However, there was overlap between control samples. B. OTUs with greater than 1% relative abundance in many of the full-term and preterm infant samples. The lack extraction controls. These OTUs were excluded from subsequent analyses of strong differentiation between gestational age groups as they were presumed to be contaminants, except the highlighted Staphylococcus OTU that was found to be the dominant Staphylococcus may indicate that the susceptibility of preterm infants to OTU in the biological samples. C. Relative abundance of the contaminant infection is primarily driven by differences in host biology, OTUs (in aggregate) that were excluded from subsequent analyses within including immune function and barrier integrity, rather each sample site. The contaminant OTUs contributed to a minority of the total OTU abundance in each of the sample sites. OTU = operational than by differences in skin colonization. Our data are taxonomic unit. (PPTX 264 kb) limited in that none of the infants developed bloodstream Additional file 3: Table S2. Diagnoses and morbidities among infants infections in our cohort and we did not examine the full admitted to the neonatal intensive care unit. (DOCX 12 kb) genetic potential of the microbiota. It is possible that the Additional file 4: Figure S2. Principal coordinates analysis (PCoA) of microbiota of preterm infants harbored more virulent samples across body sites. PCoA of infant samples excluding contaminant OTUs (A, B) or including contaminant OTUs (C, D). Similar relationships bacterial strains than the full-term infants, despite sharing between body sites are seen using generalized UniFrac distances (A, C) many of thesameOTUs. Arecentlypublished studyfound Younge et al. Microbiome (2018) 6:98 Page 10 of 11 Received: 17 August 2017 Accepted: 18 May 2018 and Bray-Curtis distances (B, D). In panel C, the first and second axes are rotated to keep the orientation of samples consistent with the other panels, but it should be noted that the vertical axis accounts for the majority of the variation between samples in this panel. (PPTX 717 kb) References Additional file 5: Figure S3. Skin and oral microbiota of twin pairs. A. 1. Oranges T, Dini V, Romanelli M. Skin physiology of the neonate and infant: Characteristics of the five twin pairs that were included in the study are clinical implications. Adv Wound Care (New Rochelle). 2015;4(10):587–95. shown, including the gestational age (preterm or full-term), the postnatal 2. Stoll BJ, Hansen N, Fanaroff AA, Wright LL, Carlo WA, Ehrenkranz RA, age (in days) at the time of sample collection, and the percentage of oral Lemons JA, Donovan EF, Stark AR, Tyson JE, et al. Late-onset sepsis in very and skin OTUs that were shared between the infants in each twin pair. low birth weight neonates: the experience of the NICHD Neonatal Research The relative abundance of the top bacterial genera within the skin and Network. Pediatrics. 2002;110(2 Pt 1):285–91. oral microbiomes are shown for the individual infants (Twin A, Twin B) 3. Eichenfield LF, Frieden IJ, Mathes EF, Zaenglein AL. Neonatal and infant within each twin pair (Twin Pairs 1–5). B. Principal coordinates analysis of dermatology. 3rd ed. London: Elsevier Saunders; 2015. 1 online resource samples from the twins based on generalized UniFrac distances. The twin (xiii, 553 pages). pairs (1–5) are grouped by color. (PPTX 363 kb) 4. Naik S, Bouladoux N, Wilhelm C, Molloy MJ, Salcedo R, Kastenmuller W, Additional file 6: Table S3. Bacterial taxa with differences in Deming C, Quinones M, Koo L, Conlan S, et al. Compartmentalized control abundance between body sites. (DOCX 12 kb) of skin immunity by resident commensals. Science. 2012;337(6098):1115–9. 5. Scharschmidt TC, Vasquez KS, Truong HA, Gearty SV, Pauli ML, Nosbaum A, Additional file 7: Table S4. Bacterial taxa with differences in Gratz IK, Otto M, Moon JJ, Liese J, et al. A wave of regulatory T cells into abundance between gestational age groups. (DOCX 13 kb) neonatal skin mediates tolerance to commensal microbes. Immunity. 2015; Additional file 8: Figure S4. The environmental microbiota of preterm 43(5):1011–21. and full-term infants. A. Relative abundance of the top genera in the 6. Kennedy EA, Connolly J, Hourihane JO, Fallon PG, McLean WH, Murray D, Jo hospital environment. B. Generalized UniFrac distances between infant JH, Segre JA, Kong HH, Irvine AD. Skin microbiome before development of body sites and their corresponding environmental samples. Median atopic dermatitis: early colonization with commensal staphylococci at 2 distances were lower among preterm infants. *p < 0.05. (PPTX 118 kb) months is associated with a lower risk of atopic dermatitis at 1 year. J Allergy Clin Immunol. 2017;139(1):166–72. 7. Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Huntley J, Fierer N, Abbreviations Owens SM, Betley J, Fraser L, Bauer M, et al. Ultra-high-throughput microbial NICU: Neonatal intensive care unit; OTU: Operational taxonomic unit community analysis on the Illumina HiSeq and MiSeq platforms. ISME J. 2012;6(8):1621–4. Acknowledgements 8. Kuczynski J, Stombaugh J, Walters WA, Gonzalez A, Caporaso JG, Knight R. We thank the families who participated in the study. Using QIIME to analyze 16S rRNA gene sequences from microbial communities. Curr Protoc Microbiol. 2012;Chapter 1:Unit 1E.5. 9. Paulson JN, Stine OC, Bravo HC, Pop M. Differential abundance analysis for Funding microbial marker-gene surveys. Nat Methods. 2013;10(12):1200–2. The study authors received support from the Hartwell Foundation, the 10. McMurdie PJ, Holmes S. Phyloseq: an R package for reproducible National Institutes of Health (K12 HD043494-14 [NY], R01GM108494 [PS, DB]), interactive analysis and graphics of microbiome census data. PLoS One. the Duke School of Nursing Center for Nursing Research, and the Duke Jean 2013;8(4):e61217. and George Brumley, Jr. Neonatal Perinatal Research Institute. The funding 11. Chen J, Bittinger K, Charlson ES, Hoffmann C, Lewis J, Wu GD, Collman RG, bodies had no role in the design of the study, analysis, interpretation of data, Bushman FD, Li H. Associating microbiome composition with or writing of the manuscript. environmental covariates using generalized UniFrac distances. Bioinformatics. 2012;28(16):2106–13. 12. Knights D, Kuczynski J, Charlson ES, Zaneveld J, Mozer MC, Collman RG, Availability of data and materials Bushman FD, Knight R, Kelley ST. Bayesian community-wide culture- The dataset supporting the conclusions of this article has been deposited in independent microbial source tracking. Nat Methods. 2011;8(9):761–3. the NCBI Sequence Read Archive (SRA) under BioProject PRJNA451534. 13. Moles L, Gomez M, Heilig H, Bustos G, Fuentes S, de Vos W, Fernandez L, Rodriguez JM, Jimenez E. Bacterial diversity in meconium of preterm Authors’ contributions neonates and evolution of their fecal microbiota during the first month of NY, PS, and DB conceptualized and designed the study. NY, FA, and PS life. PLoS One. 2013;8(6):e66986. performed the experiments. NY and PS analyzed the data. All authors 14. Oh J, Byrd AL, Deming C, Conlan S, Program NCS, Kong HH, Segre JA. participated in drafting (NY) or revising (PS, DB, FA) the manuscript and Biogeography and individuality shape function in the human skin approved the final manuscript as submitted. metagenome. Nature. 2014;514(7520):59–64. 15. Costello EK, Lauber CL, Hamady M, Fierer N, Gordon JI, Knight R. Bacterial community variation in human body habitats across space and time. Ethics approval and consent to participate Science. 2009;326(5960):1694–7. The study was approved by the Duke Institutional Review Board 16. Costello EK, Carlisle EM, Bik EM, Morowitz MJ, Relman DA. Microbiome (Pro00045553), and written informed consent was obtained from parents. assembly across multiple body sites in low-birthweight infants. MBio. 2013; 4(6):e00782-00713. Competing interests 17. Chu DM, Ma J, Prince AL, Antony KM, Seferovic MD, Aagaard KM. The authors declare that they have no competing interests. Maturation of the infant microbiome community structure and function across multiple body sites and in relation to mode of delivery. Nat Med. 2017;23(3):314–26. Publisher’sNote 18. Dominguez-Bello MG, Costello EK, Contreras M, Magris M, Hidalgo G, Fierer Springer Nature remains neutral with regard to jurisdictional claims in N, Knight R. Delivery mode shapes the acquisition and structure of the published maps and institutional affiliations. initial microbiota across multiple body habitats in newborns. Proc Natl Acad Sci U S A. 2010;107(26):11971–5. Author details 19. Pammi M, O'Brien JL, Ajami NJ, Wong MC, Versalovic J, Petrosino JF. 1 2 Department of Pediatrics, Duke University, Durham, NC, USA. Department Development of the cutaneous microbiome in the preterm infant: a of Pediatrics, Northwestern University, 310 E. Superior, Morton 4-685, prospective longitudinal study. PLoS One. 2017;12(4):e0176669. Chicago, IL 60611, USA. Duke University School of Nursing, Durham, NC, 20. Olm MR, Brown CT, Brooks B, Firek B, Baker R, Burstein D, Soenjoyo K, USA. Thomas BC, Morowitz M, Banfield JF. Identical bacterial populations colonize Younge et al. Microbiome (2018) 6:98 Page 11 of 11 premature infant gut, skin, and oral microbiomes and exhibit different in situ growth rates. Genome Res. 2017;27(4):601–12. 21. Lehtimaki J, Karkman A, Laatikainen T, Paalanen L, von Hertzen L, Haahtela T, Hanski I, Ruokolainen L. Patterns in the skin microbiota differ in children and teenagers between rural and urban environments. Sci Rep. 2017;7:45651. 22. Pannaraj PS, Li F, Cerini C, Bender JM, Yang S, Rollie A, Adisetiyo H, Zabih S, Lincez PJ, Bittinger K, et al. Association between breast milk bacterial communities and establishment and development of the infant gut microbiome. JAMA Pediatr. 2017;171(7):647–654. 23. Bokulich NA, Chung J, Battaglia T, Henderson N, Jay M, Li H, DL A, Wu F, Perez-Perez GI, Chen Y, et al. Antibiotics, birth mode, and diet shape microbiome maturation during early life. Sci Transl Med. 2016; 8(343):343ra382. 24. Biasucci G, Rubini M, Riboni S, Morelli L, Bessi E, Retetangos C. Mode of delivery affects the bacterial community in the newborn gut. Early Hum Dev. 2010;86(Suppl 1):13–5. 25. Hill CJ,Lynch DB,Murphy K,UlaszewskaM,Jeffery IB,O'Shea CA, Watkins C, Dempsey E, Mattivi F, Tuohy K, et al. Evolution of gut microbiota composition from birth to 24 weeks in the INFANTMET cohort. Microbiome. 2017;5(1):4. 26. Salava A, Aho V, Lybeck E, Pereira P, Paulin L, Nupponen I, Ranki A, Auvinen P, Andersson S, Lauerma A. Loss of cutaneous microbial diversity during first 3 weeks of life in very low birthweight infants. Exp Dermatol. 2017;26(10):861–867.

Journal

MicrobiomeSpringer Journals

Published: May 31, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off