Daily occupational exposure in swine farm alters human skin microbiota and antibiotic resistome

Abstract

Antimicrobial resistance (AMR) is a major threat to global public health, and antibiotic resistance genes (ARGs) are widely distributed across humans, animals, and environment. Farming environments are emerging as a key research area for ARGs and antibiotic resistant bacteria (ARB). While the skin is an important reservoir of ARGs and ARB, transmission mechanisms between farming environments and human skin remain unclear. Previous studies confirmed that swine farm environmental exposures alter skin microbiome, but the timeline of these changes is ill defined. To improve understanding of these changes and to determine the specific time, we designed a cohort study of swine farm workers and students through collected skin and environmental samples to explore the impact of daily occupational exposure in swine farm on human skin microbiome. Results indicated that exposure to livestock-associated environments where microorganisms are richer than school environment can reshape the human skin microbiome and antibiotic resistome. Exposure of 5 h was sufficient to modify the microbiome and ARG structure in workers' skin by enriching microorganisms and ARGs. These changes were preserved once formed. Further analysis indicated that ARGs carried by host microorganisms may transfer between the environment with workers' skin and have the potential to expand to the general population using farm workers as an ARG vector. These results raised concerns about potential transmission of ARGs to the broader community. Therefore, it is necessary to take corresponding intervention measures in the production process to reduce the possibility of ARGs and ARB transmission.

Keywords: antibiotic resistome; metagenomic sequencing; occupational exposure; skin microbiota; swine farm.

Figure 1

Changes in skin microbiota diversity following environmental exposure. (A) Experimental design schematic: Farmers, eleven swine farm workers; Students, eleven university students; T‐1 pre‐exposure, T0 skin cleaning, T1 5 h postexposure, and T2 10 h postexposure. (B) Changes in microbial diversity (Shannon diversity index) of samples from groups Fh and Ns of two cohorts (Farmers and Students) at T‐1, T1, and T2. p values marked in red mean are less than 0.05. (C) Nonmetric multidimensional scaling (NMDS) analysis of the microbial community structures in group Fh (forehead) and group Ns (nose) of two cohorts based on Bray–Curtis dissimilarity at T‐1 (red), T1 (blue), and T2 (green).

Figure 2

The variation of workers' skin microbiota composition across time points. (A) Ternary plot showing taxa abundance changes in samples from Fh and Ns during T‐1, T1, and T2. Dot size indicates the average abundance of the genus in the sample, and dot color indicates their corresponding phylum. (B) Microbial co‐occurrence networks on genus level demonstrated reduced inter‐genera correlation after exposure. (C) LEfSe analysis for characteristic microbial genera at T‐1 and T1. Only results with Linear discriminant analysis (LDA) score (log ₁₀) > 3 are shown. (D) Stacked area plots showing relative abundance of microbes at the phylum, genus, and species resolutions. The label on the X‐axis was omitted is sample name.

Figure 3

Transmission of microbes from the swine farm environment to workers' skin. (A) Volcano plots showing alteration of microbiota composition on species level of forehead and nasal vestibular skin after 5 h of occupational exposure with the x axis denoting log₂(Fold Change) and the y‐axis denoting −log₁₀ (p value). Significantly increased species at T1 are red and significantly decreased species are blue. Dashed vertical and horizontal lines reflect the filtering criteria (absolute fold‐change (FC)  ≥  1.0 and p < 0.05). (B) Venn diagram quantifying shared and unique microbial species between dust samples and samples from three time points of forehead and nasal vestibular skin. (C) Source prediction of skin microbiota during T‐1, T1, and T2. (D) Stacked area plot showing the top 15 relative abundant species in dust. Nine of the top fifteen species in relative abundance were the significantly changed species we previously identified. The label on the x axis was omitted is sample name. (E) Phylogenetic tree of Corynebacterium xerosis at strain‐level using StrainPhlAn3. The reference genome of Corynebacterium xerosis is from Corynebacterium xerosis ASM364124v1. Bootstrap support values are indicated by the color of legend.

Figure 4

Antibiotic resistome structure was influenced by occupational exposure. (A) Sum of forehead skin antibiotic resistance gene (ARG) abundance across T‐1, T1 and T2. Boxes show the distribution of workers' samples (n = 10/11 biologically independent samples per time point) (boxes show medians/quartiles; error bars extend to the most extreme values within 1.5 interquartile ranges). p values in red are less than 0.05. (B) Comparison of relative abundance of ARG types in forehead skin. Statistics were conducted by the Student's T‐test and Benjamini–Hochberg FDR correction. (C) Volcano plots showing the alteration of distribution of ARG subtypes in forehead skin after 5 h of occupational exposure. The log₂FoldChange were used to illustrate the variation of ARGs at T1 compared with T‐1. The red/blue dots represented the ARGs significantly increased/decreased in T1 compared to T‐1. Dots marked with text represent the common ARGs from groups Fh and Ns enriched at T1. (D) Sankey plot showing correlation between the ARG composition of forehead skin samples across time points and the ARG composition of environmental samples. The height of the rectangles and the depth of color all indicate the Spearman correlation between human skin microbime and swine farm environment microbiome. (E) Procrustes analysis connecting the microbiomes and resistomes of microbiota in forehead skin and environmental samples.

Figure 5

ARGs were transferred via microbes among swine farm environment and human skin. (A) Network analysis of co‐occurrence between changed ARGs and microbes transmitted from environment (p < 0.05, absolute correlation coefficient >0.80). (B) Phylogenetic assignment of metagenome‐assembled genomes (MAGs). Organisms are colored based on phyla. (C) Network analysis of co‐occurrence patterns between major MAGs (changed microbes) and ARG subtypes. Nodes are colored according to species. Node sizes are proportional (average weighted) to the number of connections. The width of the curves represented the abundance of ARGs carried by their hosts. (D) Genomic structure patterns of mobile ARG patterns present in both environment and human samples. Four ARGs AAC(6’)_Ie_APH(2”)_Ia, sul1, poxtA, and tet(X) are illustrated and aligned for each pattern, ARGs are shaded in grey.