The classroom microbiome and asthma morbidity in children attending three inner-city schools

Abstract

The classroom microbiome is different from the home microbiome. Higher classroom microbial diversity is associated with increased asthma symptoms. In this pilot study, a school-level integrated pest management intervention changed the classroom microbiome.

To the Editor

School endotoxin exposure contributes to asthma morbidity¹. Recent shotgun metagenomics studies suggest that microbial identity, rather than microbial toxin levels alone, drives impacts of microbial exposure on chronic diseases². In this pilot randomized controlled trial of school environmental interventions on indoor air quality and asthma symptoms, we sought to address three questions. First, is the school microbiome different from the home microbiome? Second, is the school microbial environment independently associated with asthma symptoms? Third, is the school microbiome modifiable?

We enrolled 25 children and performed a pilot randomized controlled trial of a combined integrated pest management (IPM) and high efficiency particulate air filter (HEPA) intervention in 21 classrooms from three schools. Asthma symptoms over a two-week period was assessed four times during the school year. Vacuumed dust was collected from homes once and schools three times for shotgun metagenomics sequencing (Figure S1). We inferred bacterial taxonomy at the species level, estimated microbial metabolic pathway abundances, used permutational analysis of variance (PERMANOVA)³ to test for differences in microbial community structure, and used MaAsLin⁴ to identify bacterial species associated with covariates. To test the hypothesis that school microbial diversity is associated with asthma symptoms, we used multivariate generalized additive mixed effects models (see detailed methods in Online Repository).

The study participants were on average 8.1 years old, from ethnic minority backgrounds, with 43.8% from households with annual incomes below the state poverty line (Table S1). Alpha diversity was significantly higher in classrooms compared to homes (Inverse Simpson β = 7.27 [1.21 – 13.34], p = 0.02) (Figure 1a). There were significant differences in microbial community structure between homes and classrooms (Figure 1b, PERMANOVA R² = 0.087, p = 0.001). In adjusted analyses using MaAsLin, 86 distinct bacterial species were differentially abundant between classrooms compared to homes (Figure S2, Table S2), including a lower relative abundance of multiple species of Bacteroides and Staphylococcus. The only metabolic pathway that was differentially abundant between classrooms compared to homes was the chondroitin sulfate degradation I (bacterial) pathway (classroom vs. homes β = −0.49, p < 0.001, FDR-corrected q < 0.001). When specifically evaluating the lipid A biosynthesis pathway, different bacterial species contributed depending on location (Table S3), with Enterobacter cloacae, Escherichia coli, and Klebsiella oxytoca contributing to this pathway in homes vs. Serratia marcescens in classrooms.

Figure 1. Classroom and home microbial community composition is different.

1a. Alpha diversity. Inverse Simpson index higher in schools compared to homes (β = β = 7.27 [1.21 – 13.34], p = 0.02). 1b. Beta diversity. Principal coordinates analysis of Bray-Curtis dissimilarity index demonstrates significant differences in microbial community composition between schools and homes (PERMANOVA R² = 0.087, p = 0.001).

In a multivariate model adjusting for age, gender, ethnicity, and season, higher classroom microbial diversity was associated with increased asthma symptoms (OR = 1.07 [1.00 – 1.14], p < 0.05)); whereas home microbial diversity was not (1.00 [1.00 – 1.00], p = 0.91, Table 1). IPM was associated with microbial community structure in the classrooms (PERMANOVA R² = 0.037, p = 0.004); HEPA-filtration was not (PERMANOVA R² = 0.016, p = 0.446). In adjusted analyses using MaAsLin, 13 bacterial species were associated with the IPM intervention (Table S4).

Table 1. Association between classroom microbial diversity and asthma symptoms^a.

Generalized additive mixed effects models were used to evaluate the association between both classroom and home microbial diversity and asthma symptoms, adjusting for gender, age, race, and season.

Covariate	Odds Ratio [95% CI]	p-value
Male vs. female	0.92 [0.29 – 2.98]	0.89
Age in years	1.33 [0.86 – 2.04]	0.20
Hispanic vs. black race	1.58 [0.42 – 5.9]	0.49
Other vs. black race	0.77 [0.15 – 4.02]	0.75
Classroom microbial diversityb	1.07 [1.00 – 1.14]	0.05c
Home microbial diversityb	1.00 [1.00 – 1.00 ]	0.91
Seasond	0.74 [0.38 – 1.47]	0.39

^aAsthma symptoms defined as presence or absence of any of the following asthma symptoms in the 14 days prior to symptom survey administration: (1) daytime wheezing, chest tightness, or cough; or (2) days on which the child had to slow down or discontinue play activities due to wheezing, chest tightness, or cough; or (3) nights with wheezing, chest tightness, or cough leading to disturbed sleep.

^bAlpha diversity measured using the inverse Simpson index

^cp = 0.0469

^dSeason modeled as penalized spline term on time

To our knowledge, this is the first study that compares the home vs. school microbial environment using shotgun metagenomics sequencing. Home and school indoor microbial communities are different, with higher classroom microbial diversity associated with increased asthma symptoms. An IPM intervention was associated with changes in the classroom microbiome. These findings suggest that it is important to account for microbial identity when defining healthy indoor environments, and that it may be possible for environmental interventions to alter the indoor microbial environment.

The home and school microbiome may differ due to construction, age of housing, presence of carpeting, or occupancy and composition of inhabitants. Most asthma studies assessing microbial diversity in the home have focused on asthma prevention rather than asthma symptom exacerbation. One recent cross-sectional home-based microbiome study found that higher home bacterial richness was associated with increased asthma severity in children⁵. While we observed that higher school microbial diversity was associated with increased asthma symptoms, we did not find the same association with home microbial diversity though home microbial diversity was assessed only once and exposure misclassification may have occurred. We assessed the classroom microbiome using vacuumed dust rather than air samples, which may explain why the HEPA air cleaner intervention was not associated with changes in measured classroom microbial community structure. This was a small pilot study, we were not adequately powered to address the health outcomes of this intervention, and our study cannot be taken to be representative of all schools in the Northeast United States. Our study did have several strengths. 1) It is based in a school setting, where children spend much of their day, and is therefore broadly relevant; future school-based interventions may be more feasible than targeting individual homes to create healthy indoor environments. 2) We used shotgun metagenomics sequencing rather than amplification and sequencing of marker genes such as the 16S rRNA gene, achieving higher taxonomic resolution, avoiding amplification bias⁶, and allowing for exploratory functional analysis of indoor microbial communities. 3) The study design is longitudinal, with randomized assignment of environmental interventions, allowing for causal interpretations. 4) We found no evidence of contamination in our negative controls.

There are several potential reasons why microbial identity is associated with asthma symptoms. A recent study showed that not all endotoxin is the same; endotoxin from Escherichia coli elicited a robust cytokine response in human peripheral blood mononuclear cells, while endotoxin from Bacteroides dorei inhibited the ability of Escherichia coli endotoxin to stimulate a cytokine response². In our study, different bacterial species contributed to lipid A biosynthesis in homes vs. schools. Environmental microbes may produce metabolites that are then inhaled leading to asthma symptoms⁷. Alternatively, a recent occupational study of adults working with laboratory mice found that a proportion of their nasal and skin microbiome could be traced to their immediate work environment⁸; it is possible that environmental microbes exert their effect by changing the human microbiome.

While studies to date have focused on manipulating the human microbiome for health, the environmental microbiome may likewise be manipulated for health. Future research should focus on defining what constitutes a healthy indoor microbiome, and the mechanisms by which environmental microbes affect asthma.