Nasopharyngeal Microbiome in Respiratory Syncytial Virus Resembles Profile Associated with Increased Childhood Asthma Risk

To the Editor:

Early-life acute respiratory infection (ARI) with respiratory syncytial virus (RSV) has been strongly associated with the development of childhood asthma (1), but the pathways underlying this association are poorly understood. We hypothesized that RSV alters the infant nasopharyngeal microbiome in a way that may help explain how RSV contributes to asthma development (1, 2). As a first step to address this question, we characterized the nasopharyngeal microbiome of healthy and RSV-infected infants using bacterial 16S ribosomal RNA (rRNA) gene sequencing.

We compared the nasopharyngeal microbiome of 33 infants assessed during a well-child visit with that of 99 infants assessed during confirmed RSV ARI. These 132 infants were enrolled as part of a large (N = 1,952) prospective cohort of previously healthy infants recruited near birth and with routine respiratory illness surveillance during their first winter season (3). Eligible infants were born between June and December, so they were on average ≤6 months of age during their first winter viral season. Thus, an RSV infection during this time likely represents their first exposure to this virus. The Institutional Review Board of Vanderbilt University approved this study, and parents provided informed consent.

The nasopharyngeal microbiome was sampled using a dry filter paper in healthy infants and a nasal wash in infants with RSV, as previously described (3, 4). The diagnosis of RSV was made by reverse transcriptase–polymerase chain reaction. We have also previously reported the methods used to characterize the nasopharyngeal microbiome (4). In brief, after cell lysis and extraction of the microbial genomic DNA from the samples, the V1-V3 region of 16S rRNA was amplified using 27F/534R primers and pyrosequencing was performed on a 454 sequencer. A mothur-based automated annotation pipeline, YAP, was used to perform initial processing of the 16S rRNA datasets (5). Operational taxonomic units (OTUs) were clustered at 97% sequence identity. After quality filtering, 1,231,758 reads were retained (range, 424–20,987 reads per sample). For statistical analyses, only samples with sequence counts of more than 2,000 sequences per sample were retained, and these were rarified to 2,047 sequences. Statistical analyses were done with MGSAT, using R (6).

The median age for healthy infants and for those with RSV was 5 (interquartile range, 2–9) and 22 (interquartile range, 13–27) weeks, respectively (P = 0.001). There were no other significant differences between groups in sex, race or ethnicity, gestational age, birth weight, mode of delivery, exposure to antibiotics in utero or after birth, breastfeeding, maternal smoking, maternal asthma, or type of insurance (data not shown).

The most abundant genera in samples collected among healthy infants included Streptococcus, Corynebacterium, Staphylococcus, and Dolosigranulum, whereas the most abundant genera during acute RSV infection were Streptococcus, Moraxella, Corynebacterium, and Haemophilus (Figure 1). Other than Dolosigranulum (base mean = 456.7; log₂ fold change = −0.9321; q-value = 0.36), the abundance of these dominant genera was significantly different between healthy infants and those with RSV (Figure 2). Staphylococcus (base mean = 265.1; log₂ fold change = −2.48; q-value = 2.14 × 10⁻³) and Corynebacterium (base mean = 1,694; log₂ fold change = −2.31; q-value = 1.39 × 10⁻³) abundance was higher in healthy infants. In contrast, Haemophilus (base mean = 1,384; log₂ fold change = 8.83; q-value = 1.96 × 10⁻²⁷), Moraxella (base mean = 2,075; log₂ fold change = 4.77; q-value = 4.55 × 10⁻⁷), and Streptococcus (base mean = 2,991; log₂ fold change = 2.04; q-value = 7.15 × 10⁻⁵) abundances were higher during RSV infection.

Figure 1.

Nasopharyngeal microbiome composition of healthy infants (left) and infants during acute respiratory syncytial virus infection (RSV; right). Stacked bar graph showing genus-level taxonomic composition for each individual nasopharyngeal sample, expressed as a proportion of reads of infants in each group. All samples are shown in this figure, regardless of sequence count. The relative abundance of the six most prevalent genera is shown; all other genera are not shown in this figure. Within each grouping, samples are sorted by predominant genera within that sample. Infants with acute RSV infection had a nasopharyngeal microbiome dominated by Streptococcus, Haemophilus, or Moraxella, whereas Haemophilus and Moraxella were observed at very low frequencies in healthy infants.

Figure 2.

Average relative abundance of genera in nasopharyngeal samples of healthy infants and infants with acute respiratory syncytial virus (RSV) infection. Within each sample, counts were normalized to simple proportions. The relative abundance of the 20 most prevalent genera is shown; all other genera are not shown in this figure.

The taxonomic composition of the nasopharyngeal microbiome also differed between health and RSV at both the genus (P < 0.001; R² = 0.08) and the OTU (P < 0.001; R² = 0.05) level when using the Bray-Curtis dissimilarity index. In addition, there was a significant decrease in richness at both the genus and OTU levels during RSV, using the abundance-based estimates Chao1, observed species counts, and inverted Shannon index Hill number N₁ when applying a linear model (P < 0.05 for all estimates).

To control for the potential confounding effect of age, we further compared a subset of healthy infants (n = 21) who overlapped in age with children with acute RSV infection (n = 37) by restricting the analysis to infants between 3 and 20 weeks of age. The median age for these healthy infants and for those with RSV was 9 (interquartile range, 6–14) and 10 (interquartile range, 8–13) weeks, respectively (P = 0.7032). In this exploratory analysis, we found similar results to our main analysis in the larger dataset with regard to the differences in genus abundance and richness. The nasal microbiome significantly differed by the Bray-Curtis index at both the genus and OTU levels, and richness remained significantly lower during acute RSV infection (P < 0.05 for all estimates). Furthermore, Moraxella, Haemophilus, and Streptococcus remained higher during acute RSV infection, whereas Staphylococcus and Corynebacterium remained lower during this illness (data not shown). These findings suggest that the nasopharyngeal microbiome changes we observed are not a result of the age difference between groups.

This study adds to the small but increasing literature on the association of early-life ARIs and the nasopharyngeal microbiome in children. To our knowledge, this is the first study to directly compare the nasal microbiome of healthy infants with the nasal microbiome during acute RSV infection in early life, using bacterial 16S rRNA gene sequencing. Prior studies using conventional culture techniques have also shown an increase in Haemophilus, Moraxella, and/or Streptococcus in the nasopharynx of infants with RSV when compared with healthy controls (7, 8). More recently, another study using 16S rRNA profiling in infants at high risk for atopy also found that these bacteria were more frequent in infants with ARIs, even after adjusting for viral detection (9). In another study using 16S rRNA sequencing, Moraxella-dominated profiles in infancy were associated with a lower rate of parental report of ARIs; however, most of these were upper ARIs, and viral detection was not done (10), which may explain the discrepant findings. Furthermore, there were several Moraxella OTUs noted in this study, and RSV could have distinct effects on different species within this genus. Taken together, these studies provide evidence for physiologic linkage between respiratory viruses and compositional shifts in the nasopharyngeal microbiome.

Interestingly, the predominant genera found in infants with RSV have been previously associated with an increased risk for childhood asthma (2). Thus, RSV may in part increase the risk for asthma through alterations in the composition of the nasopharyngeal microbiome.

One limitation to our findings is that the methods to collect the nasal samples were different between the two groups. We have recently shown that nasal dry filter papers, similar to nasal washes, can capture a wide variety of expected nasopharyngeal taxa (4); however, it is possible that certain bacterial genera could be under- or overrepresented by one of these methods when compared with the other. In addition, the region sampled by nasal filters is likely to be more proximal than that sampled by nasal washes. Thus, our findings merit replication in larger longitudinal cohorts. It is also possible that the microbial changes associated with acute RSV infection are not exclusive to this respiratory virus but are a result of the local effects of viral ARIs in general; however, limited evidence suggests that compositional shifts in the airway microbiome during childhood ARIs are virus specific (11). In spite of these limitations, our study lays the groundwork for future studies to examine the interactions among RSV infection, the infant’s nasopharyngeal microbiome, and how RSV may contribute to the development of childhood asthma.