Association of Neonatal and Maternal Nasal Microbiome Among Neonates in the Intensive Care Unit

Abstract

The neonatal nasal microbiota may help protect neonates in the neonatal intensive care unit from pathogen colonization and infection. This preliminary study characterized the biodiversity of nasal microbiota comparing neonates in the neonatal intensive care unit and their mothers, highlighting the potential of strain sharing between mother–neonate pairs.

The microbiome plays critical roles in human health, including immune development and regulation, metabolism, and production of essential vitamins [1–3]. Additionally, the microbiome may protect against the colonization of pathogenic bacteria that predispose to infection through multiple mechanisms including antagonism, competitive exclusion, or immune modulation [4–6]. In the anterior nares, low biodiversity is associated with carriage of Staphylococcus aureus, an important healthcare-associated pathogen [7, 8]. In hospitalized neonates, low nasal microbiota diversity was associated with S aureus colonization and life-threatening bloodstream infections [9].

Numerous factors shape the early development of an infant's nasal microbiota, including antibiotic exposure, environmental microorganisms [10–13], and transmission from caregivers [14]. Microbiota transmission between healthy term infants and parents occurs in the gut, vaginal, and oral [15–17], but a weaker relationship has been reported in the nasal microbiota [13, 18]. Among neonates hospitalized in the neonatal intensive care unit (NICU), parents are a frequent source of S aureus exposure [19]. However, the strain-level sharing or similarity of nasal microbiota between neonates and parents in the NICU setting is poorly characterized despite the increased vulnerability of these infants to pathogen colonization and subsequent invasive infection. Thus, our study's objective was to compare the nasal microbiota of neonates hospitalized in the NICU and their mothers at the community, species, and strain levels.

METHODS

Population

We identified a convenience sample of mother–neonate pairs enrolled in the TREAT PARENTS clinical trial [19] and admitted to the Johns Hopkins Hospital or the Johns Hopkins Bayview Medical Center NICUs. Eligible pairs had stored anterior nasal samples collected at trial enrollment and neonates were tested weekly until NICU discharge or acquisition of S aureus. Neonates that acquired S aureus colonization were treated with intranasal mupirocin [20]. Samples were included from parents and neonates at time of trial enrollment and the subsequent week from neonates not treated with mupirocin.

Sample Collection, Storage, and Whole Genome Sequencing

Flocked swabs were used to collect samples from the anterior nares and stored at −80 °C before undergoing sequencing. We performed total DNA extraction, Nextera Flex library preparation, and sequencing as previously described [21–23]. Shotgun libraries targeted 40 million 2 × 150 bp Illumina Novaseq reads per sample. Positive (defined synthetic consortia) and negative extraction and DNA controls were included with each set of extractions and library preparations.

Metagenomics Processing

Demultiplexed Illumina reads were deduplicated using prinseq-lite (v0.20.4) [24] and trimmed with Trimmomatic (v0.39) [25], with parameters: seed mismatch = 2, palindrome clip threshold = 30, simple clip threshold = 10, and LEADING: 3, TRAILING: 3, SLIDINGWINDOW: 4:15, MINLEN: 36. Human reads were then removed with Bowtie2 (v2.2.9, very-sensitive mode) [26], mapping to the CHM13 (v2.0) human reference genome [27].

Species and Strain Level Profiling

Species-level community compositions were profiled using MetaPhlAn4.0 [28]. To estimate strain-level compositions, we used a reference-based method modified from Larson et al. (2022) [23]. Strain reference databases were generated by compiling all RefSeq genomes for Cutibacterium acnes and Staphylococcus epidermidis (as of 10/28/2020). Information for these genomes was available at https://github.com/ohlab/Strain_collection. Genomes were sequence-typed by submitting the genome sequence to pubMLST using the “mlst” software (https://github.com/tseemann/mlst) [29]. Reads were then mapped to the genome databases with Bowtie2 (v2.2.9) using k = 10 and “very sensitive” mode. We then used Pathoscope (v2.0.6) on the resulting SAM files using default parameters [30] for reassignment to nearest neighbor genomes. Finally, the relative abundance of each MLST were given by the sum of the relative abundances of those genomes that were assigned to that type.

Statistical Analysis

Shannon index was computed using the “diversity” function in the “vegan” package (v2.5.7) [31]. Bray-Curtis dissimilarity was computed using the “vegdist” function from the “vegan” package. Wilcoxon signed-rank tests were conducted using the “wilcox.test” function. Spearman's correlation coefficients were estimated using the “cor.test” function. Proportion of shared MLSTs was computed only with those MLSTs constituting at least 5% of the population and was otherwise identical to the Jaccard index. Proportion of shared MLSTs was implemented using a custom function. Subgroup analyses accounted for a priori identified subgroups of birth weight (<1500 g or ≥1500 g), delivery mode (vaginal vs cesarean section) and antibiotic exposure (any antibiotics exposure before the sample collection). P values < .05 were considered statistically significant. R version 4.1.3 was used in the analyses described previoiusly [32].

Patient Consent Statement

The study was approved by the Johns Hopkins Medicine institutional review board to use stored samples that were collected with written informed consent.

Data Availability

All raw sequences have been deposited in PRJNA1108688.

RESULTS

Fifteen neonate-mother pairs had available baseline samples, and 8 neonates had an available sample 1 week later and had not been treated with intranasal mupirocin (see Supplementary Figure 1 and Supplementary Table 1). Among the 15 neonates, the median age at time of sample collection was 9 days (interquartile range: 7–17.5 days), 9 (60%) were born weighing less than 1500 g, 9 (60%) were born via cesarean section, and 10 (66.7%) were administered antibiotics before baseline sample collection.

Neonates had lower alpha diversity in the nasal microbiota compared to mothers at baseline (Wilcoxon signed-rank test, P = .008) (Figure 1A). There was a moderate positive correlation between alpha diversity of the nasal microbiome samples at baseline from neonate–mother pairs (Spearman's rho = 0.4, P = .13). (Figure 1B) When comparing species-level similarities between neonates at baseline and their moms as opposed to unrelated moms, the median of Bray-Curtis dissimilarity of neonate-mother pairs was slightly lower than that of neonate–unrelated mother pairs (Wilcoxon rank-sum test P = .42) (Figure 2A). Comparison of strain-level relative abundance, which was approximated using dissimilarity of MLST groups, between neonate–mother pairs and neonate–unrelated mother found a nonstatistically significant lower median of Bray-Curtis dissimilarity for C acnes (Wilcoxon rank-sum test P = .34) in neonate–mother pairs and a similar median dissimilarity for S epidermidis (Wilcoxon rank-sum test P = .88) between the 2 groups (Figure 2B, 2C). The distribution of the proportions of shared strains is similar between neonate–mother pairs and neonate–unrelated mother in C acnes (Wilcoxon rank-sum test P = .49) and S epidermidis (Wilcoxon rank-sum test P = .44) (Supplementary Figures 2A, 2B).

Figure 1.

Comparison of nasal microbiome between paired mother and neonates. (A ) Boxplot of Shannon index of nasal microbiome of mother and neonates (Wilcoxon signed-rank test, P = .008). (B ) Scatterplot of Shannon index of paired mother–neonate pairs (Spearman's rho = 0.4, P = .13).

Figure 2.

Comparison of nasal microbiota between paired neonate–mother pairs and neonate–unrelated mother pairs. (A ) Boxplot of Bray-Curtis dissimilarities estimated by species-level relative abundance of paired neonate–mother pairs (same) and neonate–unrelated mother pairs (different) (Wilcoxon rank-sum test P = .42). (B ) Boxplot of Bray-Curtis dissimilarities estimated by S epidermitis MLST relative abundance of paired neonate–mother pairs and neonate–unrelated mother pairs (Wilcoxon rank-sum test P = .34). (C ) Boxplot of Bray-Curtis dissimilarities estimated by Cutibacterium acnes MLST relative abundance of paired neonate–mother pairs and neonate–unrelated mother pairs (Wilcoxon rank sum test P = .88).

Stratified analysis illustrated the impact of birth weight, delivery mode, and prior antibiotic exposure on the association of the neonatal and maternal microbiomes. The median of Bray-Curtis dissimilarities between baseline nasal microbiome of mothers and neonates delivered vaginally or with birth weight <1500 g were smaller than neonates delivered via cesarean section or with birth weight ≥1500 g (Supplementary Figures 3A, 3B). The baseline nasal microbiome of mothers and neonates with and without prior antibiotics exposure were similar (Supplementary Figure 3C).

Among the 8 neonates who contributed nasal samples at baseline and 1 week later, the alpha diversity of neonates’ nasal microbiome at baseline and 1 week later were similar (Wilcoxon singed-rank test P = .80) (Supplementary Figure 4). Comparing neonates’ nasal microbiome to their mothers, the maternal baseline and neonate's microbiome 1 week later had a larger median of Bray-Curtis dissimilarity compared the neonatal and maternal microbiome at baseline (Supplementary Figure 5).

DISCUSSION

Hospitalized neonates acquire their microbiota from parents, healthcare workers and from the physical environment. Our observational study included a convenience sample of mother–neonate nasal microbiota pairs and was limited in sample size and power to make statistical inferences. However, as the first observational study using whole genome sequencing to characterize the transmission of nasal microbiota between moms and neonates in the NICU, we identified trends of interest that can inform and power larger observational studies to characterize transmission and infection risk as well as interventional studies targeting neonatal nasal microbiota as an infection prevention strategy in the NICU.

As hypothesized, our data demonstrate that a neonate's nasal microbiota shortly after birth was less diverse than the mother's nasal microbiome. Similar to other observational studies on healthy neonates [13, 18], we found only a moderate correlation in alpha diversity between neonatal and maternal nasal microbiota. Like prior studies showing that neonates in the NICU acquire S aureus colonization from their parents [19], our data also suggested a trend that strains of dominant nasal bacteria such as C acnes but not S epidermidis, may be likely transmitted from mothers to their neonates. We only examined 2 keystone commensal nasal microbiota, C acnes and S epidermidis, which had sufficient sequencing depth to reconstruct strain similarity.

We also observed trends in clinical characteristics of neonates including delivery mode, birthweight, and antibiotics exposure, suggesting that these factors might impact the similarity between neonates’ and mothers’ nasal microbiota. We did not measure environmental sources or exposures to other caregivers, which may influence the nasal microbiome in neonates to become less similar to their mother's nasal microbiome over time, another trend observed here [33].

A less diverse microbiota together with immature mucosal immunity may make neonates in the NICU more vulnerable to colonization of pathogens such as S aureus. A current strategy to prevent S aureus abundance and subsequent infection include includes nasal decolonization; however, this strategy is beneficial but insufficient to prevent S aureus disease in hospitalized neonates [34–36]. Strategies to modulate gut dysbiosis in neonates have included prebiotics, probiotics, and even maternal microbiota transplant [37–39]. In the absence of nasal probiotics, our findings should be confirmed in a larger neonatal study to support or refute the hypothesis that seeding of nasal microbiota from healthy mothers could potentially increase the biodiversity of neonatal nasal microbiota and prevent pathogen colonization in neonates.

Supplementary Data

Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. We thank all the families and staff who contributed to the TREAT PARENTS trial. This work was supported in part by the National Institute of Health (K24AI141580 to A.M., and R01AR078634, R01AR083742, and 2U19AI142733 to J.O.), a Johns Hopkins Children's Center Innovation Grant, the Jackson Laboratory Shared Services (Genome Technologies at The Jackson Laboratory for Genomic Medicine for support with sample processing and sequencing), and by a Basic Cancer Center Core Grant from the National Cancer Institute (CA034196 to J.O.).

Author contributions. A.M., S.X., and J.O. conceived and designed the study. S.D., R.C., and W.Z. collected or generated data. W.Z. and S.X. performed data processing and analysis. S.X. and A.M. drafted the manuscript. All authors reviewed and approved of the final manuscript.