Bacterial and fungal communities in indoor aerosols from two Kuwaiti hospitals

Abstract

The airborne transmission of COVID-19 has drawn immense attention to bioaerosols. The topic is highly relevant in the indoor hospital environment where vulnerable patients are treated and healthcare workers are exposed to various pathogenic and non-pathogenic microbes. Knowledge of the microbial communities in such settings will enable precautionary measures to prevent any hospital-mediated outbreak and better assess occupational exposure of the healthcare workers. This study presents a baseline of the bacterial and fungal population of two major hospitals in Kuwait dealing with COVID patients, and in a non-hospital setting through targeted amplicon sequencing. The predominant bacteria of bioaerosols were Variovorax (9.44%), Parvibaculum (8.27%), Pseudonocardia (8.04%), Taonella (5.74%), Arthrospira (4.58%), Comamonas (3.84%), Methylibium (3.13%), Sphingobium (4.46%), Zoogloea (2.20%), and Sphingopyxis (2.56%). ESKAPEE pathogens, such as Pseudomonas, Acinetobacter, Staphylococcus, Enterococcus, and Escherichia, were also found in lower abundances. The fungi were represented by Wilcoxinia rehmii (64.38%), Aspergillus ruber (9.11%), Penicillium desertorum (3.89%), Leptobacillium leptobactrum (3.20%), Humicola grisea (2.99%), Ganoderma sichuanense (1.42%), Malassezia restricta (0.74%), Heterophoma sylvatica (0.49%), Fusarium proliferatum (0.46%), and Saccharomyces cerevisiae (0.23%). Some common and unique operational taxonomic units (OTUs) of bacteria and fungi were also recorded at each site; this inter-site variability shows that exhaled air can be a source of this variation. The alpha-diversity indices suggested variance in species richness and abundance in hospitals than in non-hospital sites. The community structure of bacteria varied spatially (ANOSIM r ² = 0.181-0.243; p < 0.05) between the hospital and non-hospital sites, whereas fungi were more or less homogenous. Key taxa specific to the hospitals were Defluvicoccales, fungi, Ganodermataceae, Heterophoma, and H. sylvatica compared to Actinobacteria, Leptobacillium, L. leptobacillium, and Cordycipitaceae at the non-hospital site (LefSe, FDR q ≤ 0.05). The hospital/non-hospital MD index > 1 indicated shifts in the microbial communities of indoor air in hospitals. These findings highlight the need for regular surveillance of indoor hospital environments to prevent future outbreaks.

Keywords: bacteria; bioaerosol; exhaled air; fungi; virus.

FIGURE 1

(A) Taxonomic composition of bacteria in bioaerosols. The Wilcoxon rank test (p < 0.05) was applied to median abundances of bacterial OTUs to create the differential abundance tree. The nodes represent the bacterial taxon. The size of the node denoted the OTU counts, and the color represents its presence in a number of samples. A key to the OTU count and sample prevalence is provided in the right-hand side corner. Box plots showing RA of (B) top 10 dominant bacterial genera and (C) ESKAPE pathogens. The RA are plotted on the Y-axis and the corresponding genera on the X-axis. Each box represents the inter-quartile range (25–75%), upper and lower whiskers (10–90%), and dashed lines are the mean RA%. Black dots represent the RA of the individual sample.

FIGURE 2

(A) Taxonomic composition of fungi in bioaerosols. The Wilcoxon rank test (p < 0.05) was applied to median abundances of fungal OTUs to create the differential abundance tree. The nodes represent the fungal taxon. The size of the node denoted the OTU counts and the color represents its presence in a number of samples. A key to the OTU count and sample prevalence is provided in the right-hand side corner. (B) Box plots showing RA of top 10 dominant fungal species. The RA values are plotted on the Y-axis and the corresponding genera on the X-axis. Each box represents the inter-quartile range (25–75%), upper and lower whiskers (10–90%), and dashed lines are the mean RA%. Black dots represent the RA of the individual sample.

FIGURE 3

Variation in RA% of bacterial and fungal taxa in hospitalized (H1 and H2), and non-hospitalized (G2) settings. (A) Variations in bacterial phyla; (B) variations in fungal phyla; (C) variations in bacterial genera; and (D) variations in fungal genera. For all the bar plots, the groups are plotted on the x-axis, and relative abundances (RA) are shown on the y-axis. The color key to each taxon is presented on the top of each bar graph. OTU Clustering analysis: (E) Common and unique bacterial OTUs; (F) common and unique fungal OTUs. Differential abundance analysis (G) between bacterial taxa of H1 and H2; (H) between bacterial taxa of H1 and G2; (I) between bacterial taxa of H2 and G2; (J) between fungal taxa of H1 and H2; (K) between fungal taxa of H1 and G2; and (L) between fungal taxa of H2 and G2. The Wilcoxon rank test (p < 0.05) was applied to median abundances of OTUs to create the differential abundance tree.

FIGURE 4

Bar charts showing the variations of RA within each sub-location of H1, H2, and G2. (A) Bacterial genera, (B) fungal species, (C) RA of ESKAPEE genera, and (D) bar chart showing variations in RA of ESKAPEE genera. The RA is plotted on the y-axis and the corresponding taxa on the x-axis.

FIGURE 5

Box plots with alpha-diversity indices: (A–F) for bacterial species and (G–L) for fungal species. The boxes represent the range of alpha diversity within each location. Corresponding alpha-diversity indices (observed, Ace, Chao1, Shannon’s, Simpson, and Fisher) are plotted on Y-axis. All the comparisons were done employing Tukey’s and Wilcox’s tests at p ≤ 0.05.

FIGURE 6

Beta-diversity analysis of microbial communities of indoor aerosols of hospitalized and non-hospitalized settings in Kuwait. (A) PCA plot of bacterial communities, (B) PCA plot of fungal communities, (C) ANOSIM analysis of bacterial community, (D) ANOSIM analysis of fungal communities, (E) LDA analysis of bacterial taxa, and (F) LDA analysis of fungal taxa. LDA scores are plotted on the x-axis on a logarithmic scale. The corresponding taxa are shown on the Y-axis. The LEfSe algorithm applying the non-parametric factorial Kruskal–Wallis (KW) sum-rank at an FDR (q) ≤ 0.05 followed by the linear discriminant analysis (LDA) was employed to select significantly differential taxa.

FIGURE 7

Network analysis and functional prediction: (A) Network lattice of bacterial genera, (B) network lattice of fungal species, (C) functional prediction of bacterial 16s rRNA genes, and (D) major KEGG pathways in bacterial genera.