Characterization of the public transit air microbiome and resistome reveals geographical specificity

Abstract

Background: The public transit is a built environment with high occupant density across the globe, and identifying factors shaping public transit air microbiomes will help design strategies to minimize the transmission of pathogens. However, the majority of microbiome works dedicated to the public transit air are limited to amplicon sequencing, and our knowledge regarding the functional potentials and the repertoire of resistance genes (i.e. resistome) is limited. Furthermore, current air microbiome investigations on public transit systems are focused on single cities, and a multi-city assessment of the public transit air microbiome will allow a greater understanding of whether and how broad environmental, building, and anthropogenic factors shape the public transit air microbiome in an international scale. Therefore, in this study, the public transit air microbiomes and resistomes of six cities across three continents (Denver, Hong Kong, London, New York City, Oslo, Stockholm) were characterized.

Results: City was the sole factor associated with public transit air microbiome differences, with diverse taxa identified as drivers for geography-associated functional potentials, concomitant with geographical differences in species- and strain-level inferred growth profiles. Related bacterial strains differed among cities in genes encoding resistance, transposase, and other functions. Sourcetracking estimated that human skin, soil, and wastewater were major presumptive resistome sources of public transit air, and adjacent public transit surfaces may also be considered presumptive sources. Large proportions of detected resistance genes were co-located with mobile genetic elements including plasmids. Biosynthetic gene clusters and city-unique coding sequences were found in the metagenome-assembled genomes.

Conclusions: Overall, geographical specificity transcends multiple aspects of the public transit air microbiome, and future efforts on a global scale are warranted to increase our understanding of factors shaping the microbiome of this unique built environment.

Keywords: Air microbiology; Bioinformatics; High-throughput sequencing; Metagenomics; Microbial ecology; Microbiome.

Fig. 1

Effects of geography and related factors in driving public transit air microbiome. Colours represent each city: Denver (orange), Hong Kong (red), London (purple), New York (blue), Oslo (yellow), Stockholm (green). a Relative abundance of bacteria, fungi, virus, and archaea across cities. b Density plot of core species-level taxa (present in ≥ 75% of all samples). c and d Significant differences between c Shannon diversity index (Wald chi-square test p = 2.3 × 10⁻²⁶) and d normalized richness (Wald chi-square test p = 5.5 × 10⁻²⁵) of public transit air microbiomes were detected. Asterisks above horizontal bars indicate mixed model pairwise comparison significance following Tukey method p-value adjustment: *p < 0.05, **p < 0.01, ***p < 0.001. e Principal coordinates analysis plot of community composition based on Bray-Curtis dissimilarity of public transit air microbiomes grouped by city. The normal confidence ellipses indicate the confidence level at 95%

Fig. 2

Inferred species- and strain-level growth rates showed geographically specific profiles. GRiD and SMEG were respectively applied to infer the a species- and b strain-level growth rates. GRiD was shown for species-level taxa with indices detected in greater than 10% of samples in the dataset. Samples with coverage below the default threshold for each species could not have their growth rates inferred and are indicated as white spaces on the plots

Fig. 3

Strain-level geographical specificity in public transit air microbiome for bacteria C. acnes and M. luteus based on phylogenetic and phylogenomic analyses. a Percentages of non-polymorphic sites present within strains of C. acnes and M. luteus within metagenomes. b and c StrainPhlAn phylogenetic clustering of b C. acnes and c M. luteus. d and e Principal coordinates analysis plot of PanPhlAn phylogenomic gene content analysis of geography-based clustering based on Jaccard distances between strains within metagenomes. d C. acnes and e M. luteus genomes from different natural and built environments were included in the plot. f and g Geography-level KO biomarkers ranked by mean decrease in accuracy, with each KO colour coded by gene functional family (f), and the prevalence of the KO biomarkers in each city (light green and purple bars represent markers of C. acnes and M. luteus, respectively) (g)

Fig. 4

Geographical specificity in public transit air resistome. Heatmap of the top 30 AR protein families based on average reads per kilobase per million (RPKM) reads across metagenomes. Core AR protein families (those detected in ≥ 75% of the entire dataset) are indicated in red and asterisks

Fig. 5

Bayesian sourcetracking estimated public transit surface, human skin, and soil as major AR sources for public transit air resistome. Estimated proportions of resistome sources of different ecotypes in the public transit air microbiomes faceted by city (a) and by above- and underground stations within the Hong Kong public transit system (b)

Fig. 6

Public transit air resistome contained both chromosome- and plasmid-associated AR genes encoding multiple functional mechanisms of resistance to diverse antimicrobial classes. a Detection of AR genes and their genomic context (chromosomal or plasmid-based). b Histogram showing the number of contigs containing AR genes encoding genes conferring different mechanisms of resistance, faceted by genetic context in which the AR genes were detected. c Abundance data (in RPKM) of genes conferring resistances to different antibiotic classes detected across different cities and genetic contexts

Fig. 7.

MAGs within the public transit air microbiome contained a diverse collection of gene clusters encoding proteins involved in biosynthesis of secondary metabolites. MAGs with secondary metabolite BGCs. Species known to colonize the human skin, nasal, and urogenital tracts are indicated in red. Types of metabolites synthesized by BGCs in MAGs are indicated by filled tiles. The number of BGCs detected in MAGs, with bars coloured by type of metabolite