Pre-school age is a critical period for the occurrence of recurrent wheeze (RW) which is a strong risk factor for asthma. Mucosal microbiota serves as a mediator between host immunity and the environment, and microbial dysbiosis was found in many immune-mediated diseases (Kozik and Huang, 2019). Nonetheless, the association between allergies and airway microbiome was underexplored due to limitations with low microbial biomass (Quince et al., 2017). Upper airway samples were widely used in pediatric studies owing to the difficulty in obtaining lower airway samples.

Upper airway dysbiosis is longitudinally related to recurrent wheezing and asthma. Seminal work from Bisgaard et al. (2007) first noticed that asymptomatic hypopharyngeal colonization of pathogenic bacteria including Streptococcus, Haemophilus, and Moraxella in neonates was related to asthma risk at 5 years old. Similarly, enrichment of Moraxella, Haemophilus, or Streptococcus, in the nasal microbiota from hospitalized bronchiolitis infants was reported being longitudinally related to increased risk for recurrent wheezing (Mansbach et al., 2020; Zhang et al., 2020). Nasal dominance of Moraxella and Staphylococcus in school-age asthmatic children increased exacerbation risk (McCauley et al., 2019). Nonetheless, there is limited evidence on airway microbiome during acute episodes of RW or asthma exacerbations, particularly in the pre-school age group. Fazlollahi et al. (2018) reported the asthma exacerbation in adult patients is associated with nasal enrichment of Prevotella, Alkanindiges, and Gardnerella. Robinson et al. (2019) conducted pioneering investigation on pre-school children referred for bronchoscope for severe RW, and reported that the dominance of Moraxella in bronchoalveolar lavage (BAL) is related to neutrophilic pulmonary inflammation. Nevertheless, BAL microbiome profile was not related to clinical wheezing phenotypes in this study, though a trend of episodic viral wheeze and Moraxella-profile was noticed.

Regarding the technological aspect, metagenomics studies are more powerful than 16S rRNA sequencing to delineate microbiome due to their higher resolution and functional profiling capability. A whole genome shotgun (WGS) study reported enriched Streptococcus pneumoniae and sphingolipid metabolism pathway in nasopharyngeal aspirate (NPA) to be associated with severe bronchiolitis in infants (Stewart et al., 2017).

We hypothesized that specific markers of nasopharyngeal microbiome (NPM) from pre-school wheezers at risk of developing asthma can discriminate them from non-wheeze inpatient and community controls (CCs). Besides, we hypothesized that human rhinovirus (HRV) infection might influence the association between NPM and pre-school wheezing. This study applied WGS to characterize the NPM in pre-school children hospitalized for RW who were positive for asthma predictive index (API), and explore the differences in NPM compositions and functional capabilities between RW cases and control groups and the possible effect of HRV infection on such associations between NPM and pre-school wheezing.

This WGS-based study revealed that M. catarrhalis (Proteobacteria phylum) and D. pigrum (Firmicutes phylum) were predominant species in nasopharyngeal samples of pre-school children. This was concordant with other studies on NPM among infants and children (Bogaert et al., 2011; Teo et al., 2015). We discovered that the NPM of RW with asthma predisposition was characterized by low biodiversity characterizing with enriched phylum Proteobacteria being the only representative taxa. Conversely, D. pigrum was overrepresented in CC. Adjusted for HRV status, D. pigrum was the only discriminative species across RW, IC, and CC groups. LEfSe analyses identified five discriminative MetaCyc pathways, including adenosine ribonucleotides de novo biosynthesis in RW and four bacterial pathways involved in purine and adenosine biosynthesis, and energy metabolism for CC group.

In the present study, two approaches were applied for collecting nasopharyngeal samples from hospitalized and community subjects. For hospitalized patients, NPAs were obtained as standard practice for pathogen detection upon admission (Leung et al., 2010). On the other hand, FNPSs were collected from CC subjects as such approach (versus NPA) obviates the need for negative-pressure facilities during the procedure (Chan et al., 2008). In practice, Tunsjø et al. (2015) reported that these two sample types did not differ in virus detection among children. Previous studies have also indicated that FNPS samples represented a less invasive and more reliable option than NPA in terms of the sensitivity and specificity for respiratory virus detection especially for children below 5 years old (Abu-Diab et al., 2008; Lambert et al., 2008). These findings supported our use of FNPS in CC subjects and NPA for inpatients and would not introduce any substantial difference in nasopharyngeal microbial communities from the methodological point of view.

Our study reported that hospitalized children with RW (and positive S-API) or URTI had lower microbial biodiversity in upper airway than CC. Specifically, pre-school children with RW had higher abundance of Proteobacteria while the latter group possessed higher abundance of phylum Actinobacteria and genus Dolosigranulum in their nasopharynx. These observations were consistent with previous studies for wheeze and asthma that the asthmatic upper airway had lower biodiversity but more abundant Proteobacteria (Depner et al., 2017; Fazlollahi et al., 2018). Particularly, Moraxella is a genus under Proteobacteria that was repeatedly reported to be related to asthma and wheeze. Children hospitalized for severe bronchiolitis who carried more Moraxella in upper airway had increased risk for RW at later childhood (Mansbach et al., 2020; Zhang et al., 2020). Similar observations between nasal Moraxella and asthma exacerbation were found in schoolchildren (McCauley et al., 2019; Zhou et al., 2019). McCauley et al. (2019) reported nasal domination of Moraxella spp. was related with type-2 inflammation in nasal epithelium. All our RW cases were positive for S-API which reflected familial and personal history of allergic diseases. Further investigation is warranted to link the host inflammatory profile to NPM dysbiosis during acute wheezing episodes.

Moreover, Robinson et al. reported that pre-school episodic viral wheezers had Moraxella-dominating lung microbiome during the stable phase. Their results also suggested that Moraxella-enriched microbial profile was related to local but not systemic neutrophilic inflammation (Robinson et al., 2019). Our study extended the relationship between Proteobacteria and wheezing illnesses into pre-school age group. Additionally, our finding supported that NPM dysbiosis shared similar patterns with that found in BAL from severe pre-school wheezers. This observation was consistent with a previous study that upper airway microbiota in young children was generally representative for lung disease-related microbial patterns in the lower airway (Marsh et al., 2016).

On the contrary, Actinobacteria and Dolosigranulum might protect against asthma (Biesbroek et al., 2014; Ta et al., 2018). Both taxa were normal upper airway commensals that colonized infants’ nasopharyngeal niche within the first month of life (Bosch et al., 2017; Man et al., 2017). The predominance of Dolosigranulum in the airway was associated with lower risk for asthma and viral respiratory diseases (Bosch et al., 2017; Teo et al., 2018). Dolosigranulum and Corynebacterium (Actinobacteria) also played an important role in respiratory health (Laufer et al., 2011; Kozik and Huang, 2019). Our study extended such relationship into the pre-school age group. Our results suggested that RW-related dysbiosis in the upper airway was present prior to the onset of asthma among pre-school children. This finding may provide an investigative approach for asthma prediction in young children.

The ability of WGS sequencing to taxonomically identify microbes down to the species level (e.g., M. catarrhalis in RW and D. pigrum in CC) is one of its strengths over 16S rRNA sequencing. Our metagenomic data identified that D. pigrum was the only species that remained group-discriminating after adjustment for HRV status. This bacterium was first described as a new lineage of lactic acid bacterium (Aguirre et al., 1993), and its whole-genome sequence was announced in 2017 (Mukhopadhyay et al., 2017). Recent studies revealed an inverse correlation in RAs between D. pigrum and Staphylococcus aureus, a pathobiont in the respiratory tract (Liu et al., 2015; Escapa et al., 2018). Furthermore, D. pigrum was reported to inhibit the in vitro growth of S. aureus (Brugger et al., 2020) and decrease the production of epithelial pro-inflammatory cytokines induced by S. aureus which maintain epithelial barrier integrity against disruptive interleukin-4 (De Boeck et al., 2017).

In addition, another strength of WGS is its capability to identify all kingdoms of microbes that use DNA as their genetic materials while 16S rRNA profiling could only reconstruct phylogeny of bacteria and archaea. This WGS study uncovered six viral families including Siphoviridae, Herpesviridae, Anelloviridae, Papillomaviridae, Polyomaviridae, and Retroviridae from nasopharyngeal samples that accounted for nearly 10% of non-human reads. Among them, Herpesviridae, Polyomaviridae, and Anelloviridae were reported in children with community-acquired pneumonia (Li Y. et al., 2019). We found porcine type C oncovirus of the Retroviridae family to be more abundant in CC (Supplementary Figure 3), and existing data suggested low risk of in vivo infection of this virus in human (Łopata et al., 2018).

Another remarkable advantage of WGS related to its ability to characterize the functional potential of a metagenome. Accumulating evidence indicated that mucosal microbiome impacted allergy and asthma development through perturbation of the host immunity (Kemter and Nagler, 2019). We found a higher functional capacity for purine nucleotide de novo biosynthesis (PWY-7229, PWY-6126, and PWY-6125) in CC subjects (Table 2), suggesting nasopharyngeal microbes with greater proliferation potential could contribute to immune homeostasis in healthy individuals. Nevertheless, these discriminative metabolic pathways were present in only less than one quarter of all CC samples (Supplementary Figures 7A–C). Hence, the immunological implications of overrepresentation of these pathways required further investigation integrating metatranscriptomics and metabolomics. In this regard, Stewart et al. (2017) interrogated the microbiome and metabolome in NPA and explored their relationship to bronchiolitis in infants. In school-age children with asthma, serum metabolomic biomarkers were linked to oropharyngeal microbiome to illustrate the interaction between microbes and host immunity (Chiu et al., 2020). Similar exploration leveraging multi-omics approaches are warranted to implement in pre-school children with recurrent wheezing, which would further our understanding on how functionally aberrant airway microbiome modulates host immune response during this critical phase and leads to later asthma.

As discussed above, D. pigrum has been shown to inhibit in vitro growth of S. aureus (Brugger et al., 2020). Among the nine D. pigrum-involved pathways, PWY-7219 represented the only discriminative function enriched in RW by LEfSe analysis (and HRV status as a subclass grouping factor). Interestingly, this pathway was mainly contributed by M. catarrhalis in RW but primarily by D. pigrum in CC (Supplementary Figure 7D). S-adenosyl-L-methionine cycle I (PWY-6151) responsible for amino acid metabolism was another D. pigrum-involved pathway. By providing methyl in living cells (Caspi et al., 2017), this pathway was important in the biosynthesis of multiple antibiotics including neomycin B (Kudo et al., 2014) and anthracycline (Grocholski et al., 2015), which might partially explain inhibition of S. aureus growth by D. pigrum in co-culture. Moreover, some asthmatic children present with chronic cough (Morice et al., 2014). Of interest, lung-derived ATP interacted with P2X receptor (evolutionarily conserved ATP-gated ion channels) to stimulate hypersensitive cough through peripheral nerves (Turner and Birring, 2019). Endogenous ATP could be produced locally in response to tissue stresses such as inflammation or epithelial damage (Ford and Undem, 2013). Our metagenomic data revealed that the NPM from CC group was enriched for purine and adenosine biosynthesis. Whether this finding may be related to ATP-P2X pathway require further investigation.

Persistent bacterial bronchitis (PBB) is another leading cause of chronic cough in young children. Differentiating these two conditions is important as the treatment plan and prognosis are distinct. Kazachkov et al. (2018) investigated microbiome of BAL samples from pre-school children with chronic cough. They found that asthmatic BAL was enriched for Rothia, Gemellaceae, and Granulicatella, while that of PBB was characterized by increased abundance for Prevotella. Another study reported bronchial dysbiosis in pre-school PBB to be associated with enrichment of Haemophilus and Neisseria when compared to healthy controls (Cuthbertson et al., 2017). Similarly, Bao et al. (2018) extended this finding to infants by showing that those with PBB had enriched Haemophilus and Bacteroides but decreased Lactococcus and Lactobacillus.

Some studies suggested that upper airway samples, such as throat swabs, demonstrated representative microbial communities for lower airways as assessed by BAL or bronchial brushes (Ahmed et al., 2018). This finding supported the use of upper airway samples for studying microbiome in different respiratory diseases in children as it would be too invasive to obtain BAL samples by bronchoscopy in most cases. Studies using upper airway surrogates are warranted to unveil potential microbial signals for early recognition of asthma-predisposed signals among children with chronic cough.

Of interest, this study identified pathways (PWY-7383, PWY-5690, and PWY-7357) specific to fungi among the control subjects. This supported the hypothesis that fungi in NPM may play a role in the pathogenesis of recurrent wheezing (Li E. et al., 2019). Fujimura et al. (2016) reported higher RAs of specific fungi (Candida and Rhodotorula) in the gut pose a high risk of childhood asthma. Fungal dysbiosis characteristic of Pneumocystis overrepresentation in the lung was associated with severe asthma in children (Barcik et al., 2020). More studies focusing on airway fungiome are needed to deepen our understanding of the mechanisms through which fungi influence respiratory health.

There are several limitations in this study. Firstly, we could not analyze the effect of HRV on RW as all our RW cases were HRV positive. Our local findings found that the large majority of hospitalized children with RW and asthma exacerbation had HRV infection (Leung et al., 2010; Mak et al., 2011). However, we were still able to adjust for HRV by having HRV-negative subjects in the IC and CC groups. Secondly, this was a cross-sectional study on pre-school recurrent wheezers which attempted to apply NPM to predict later development of asthma. The results will need to be validated in prospective cohorts that followed RW group into later childhood. In addition, WGS is able to reveal the metabolic potentials embedded in the microbial genes. The integration of multi-omics information allows us to explore the mechanisms through which microbial dysbiosis affects respiratory health, but in vitro and animal experiments are required to confirm the functionality of these microbes. Besides, NPM analyses in this study was based on NPA from hospitalized children in the RW and IC groups and FNPS from children in the CC group. There has not been any study that compared microbiome compositions in these upper airway samples, so it is unknown if such might introduce bias in the bacterial community. A previous local study revealed that these two sampling methods had comparable power to detect respiratory viruses by molecular assays (Chan et al., 2008). Further comparative studies are required to analyze upper airway microbiota in NPA and FNPS samples.

We identified M. catarrhalis and D. pigrum to be the predominant species in nasopharynx of pre-school children. Hospitalized children with RW and positive for S-API had lower nasopharyngeal biodiversity than CC, while Proteobacteria was the only representative taxa in RW. On the other hand, D. pigrum was overrepresented in CC. Adjusted for HRV status, D. pigrum and five MetaCyc pathways were discriminative across groups. These microbial biomarkers provide insights into RW pathogenesis in pre-school children.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

The studies involving human participants were reviewed and approved by the Joint Chinese University of Hong Kong – New Territories East Cluster Clinical Research Ethics Committee. Written informed consent to participate in this study was provided by the participants’ legal guardian/next of kin.

TL obtained research funding, designed this study, collected clinical data, and drafted this manuscript. YS collected clinical data, performed laboratory experiments, analyzed laboratory and WGS data, and drafted this manuscript. JH and JK conducted bioinformatics analysis, interpreted WGS data, and contributed to the manuscript. ST supervised bioinformatic data analysis and results interpretation. HW and MW contributed as statistical consultation. MT contributed to laboratory experiments. KT and RC conducted experiments for HRV detection. AL and GW analyzed clinical data. All authors reviewed and approved the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.