Staphylococcal diversity in atopic dermatitis from an individual to a global scale

Abstract

Atopic dermatitis (AD) is a multifactorial, chronic relapsing disease associated with genetic and environmental factors. Among skin microbes, Staphylococcus aureus and Staphylococcus epidermidis are associated with AD, but how genetic variability and staphylococcal strains shape the disease remains unclear. We investigated the skin microbiome of an AD cohort (n = 54) as part of a prospective natural history study using shotgun metagenomic and whole genome sequencing, which we analyzed alongside publicly available data (n = 473). AD status and global geographical regions exhibited associations with strains and genomic loci of S. aureus and S. epidermidis. In addition, antibiotic prescribing patterns and within-household transmission between siblings shaped colonizing strains. Comparative genomics determined that S. aureus AD strains were enriched in virulence factors, whereas S. epidermidis AD strains varied in genes involved in interspecies interactions and metabolism. In both species, staphylococcal interspecies genetic transfer shaped gene content. These findings reflect the staphylococcal genomic diversity and dynamics associated with AD.

Keywords: Staphylococcus; antibiotics; antimicrobial resistance; atopic dermatitis; mGWAS; metagenome; microbiome; mobilome; skin; strain sharing.

Figure 1.. Integrating shotgun metagenomic and whole genome sequencing allows for high-resolution investigation of the AD skin microbiome.

A, Diagram of study design. 44 moderate-to-severe AD children and 10 HC were sampled. Parallel samples underwent metagenomic sequencing and staphylococcal culturing (e.g. S. epidermidis, S. aureus, S. capitis) and sequencing. Data analyses incorporated all related publicly available datasets: 4,320 isolate genomes; 1,224 metagenomes; 473 subjects (n=298 AD, n=175 HC) B, Stacked barplot of relative abundances of Staphylococcus species in 38 AD flare and 10 HC skin samples. The tile plot depicts AD disease severity and disease signature: S. aureus/S. argenteus-predominant patients (SAP), S. epidermidis-predominant patients (SEP), other-Staphylococcus-predominant patients (OSP). Samples analyzed were derived from representative flare affected Ac (or affected Pc when Ac data was unavailable). C, Boxplot of total SCORAD in SAP and SEP patients. Welch two-sample t-test (P-value=0.0069). Boxes represent interquartile ranges (IQR); whiskers represent ±1.5 * IQR from each box’s quartiles. D, Accumulation curves for cultured and sequenced S. aureus and S. epidermidis isolates. Locally Weighted Scatterplot Smoothing with each line representing a different subject. E, Relative abundances of S. epidermidis and S. aureus sequence types (STs); patient-specific cultured isolates as reference genomes. A single representative for SAP (AD20) and SEP (AD17) strain profiles for the Ac is shown.

Figure 2.. Intra-ST genetic variability for S. aureus and S. epidermidis.

A, Phylogenetic tree of representative S. aureus STs in the global dataset. tSNE of S. aureus isolates based on gene content colored by ST illustrates genetic content is shared within an ST. B, Phylogenetic tree of representative S. epidermidis STs. tSNE of S. epidermidis isolates by gene content colored by ST also shows clustering by ST. To account for differences in the sampling frequency of subjects, in A-B, we selected a maximum of 3 representative isolates for each subject and ST C and D, Pairwise phylogenetic distance based on the number of SNPs versus pangenomic distance for all S. aureus C, and S. epidermidis D, isolates. The colors of the scatterplot and accumulation curves are defined in A, B. Circles in C, D represent comparisons between isolates derived from different subjects, while triangles represent comparisons between isolates derived from the same subject. Accumulation curves of number of accessory genes within an individual for E, S. aureus and F, S. epidermidis isolates for each ST cultured from an individual from the NIH dataset. Each line represents a different individual colored by ST. ST colors are shown in A-B. EggNOG annotation of genes that vary across an individual and ST for G, S. aureus and H, S. epidermidis STs colored based on the Cluster of Orthologous Groups (COG) categories. COG categories, represented by one-letter abbreviation, determined via eggNOG: C=energy production and conversion; D=cell cycle control, cell division, chromosome partitioning; E=amino acid transport and metabolism; F=nucleotide transport and metabolism; G=carbohydrate transport and metabolism; H=coenzyme transport and metabolism; I=lipid transport and metabolism; J=translation, ribosomal structure, and biogenesis; K=transcription; L=replication, recombination, and repair; M=cell wall/membrane/envelope biogenesis; N=cell motility; O=posttranslational modification, protein turnover, chaperones; P=inorganic ion transport and metabolism; Q=secondary metabolites biosynthesis, transport, and catabolism; R=general function prediction only; S=function unknown; T=signal transduction mechanisms; U=intracellular trafficking, secretion, and vesicular transport; V=defense mechanisms.

Figure 3.. Associating genomic loci and strains of S. aureus and S. epidermidis with disease severity.

A, Phylogenetic tree of 75 ST5 S. aureus isolates with a heatmap of select genes with frameshift mutations in S. aureus in two AD siblings: AD20 (SAP) and AD23 (SEP). B, Phylogenetic tree of 83 ST5 S. epidermidis isolates with heatmap of select genes with nonsynonymous mutations in S. epidermidis in two siblings: AD23 (SEP) and healthy sibling HC10. C, Manhattan plot associating significant unitigs in S. aureus involving 102 genes with higher disease severity (total SCORAD). Representative significant hits are annotated. D, Manhattan plot of the GWAS of S. epidermidis isolates associated unitigs involving 307 genes with high S. aureus:S. epidermidis relative abundance ratio, determined via corresponding metagenomes. Representative significant hits are annotated. GWAS hits with P-value below the significance limit and a mean allele frequency >0.05 are shown in blue, with light blue signifying a β (effect size) <0 and dark blue signifying a β>0. Annotations in A-D were obtained via Prokka.

Figure 4.. Strains predominant on the skin are influenced by host adaptation and antibiotic usage.

A, PCoA of KEGG pathways detected in isolates colored by species (e.g. S. aureus, S. argenteus, and known skin commensals S. epidermidis and S. capitis). B, Top 15 differentially enriched KEGG pathways in different S. aureus STs (top), and in different S. epidermidis STs (bottom). Representative STs of available data from ≥2 subjects are shown. C, Proportion of reads mapping to AMR genes in NIH metagenomes for affected body sites (Ac/Pc) compared to typically unaffected site (Vf) during baseline, flare, and post-flare. Boxes represent interquartile ranges (IQR); whiskers represent ±1.5 * IQR from each box’s quartiles. D, Proportion of AMR reads subdivided by antibiotic class. E, Proportion of S. aureus isolates from different countries resistant to different classes of antibiotics, e.g. fusidic acid resistance. F, Proportion of S.epidermidis isolates resistant to different antibiotic classes, e.g. mupirocin- (57%) and trimethoprim-resistance (100%).

Figure 5.. Comparative analyses of S. aureus and S. epidermidis isolates reveal shared mobilomes between different Staphylococcus species.

A, Distribution of mecA-positive S. aureus isolates across different countries, shown as the hashed proportion of subjects. B, Distribution of mecA-positive S. aureus isolates and S. epidermidis isolates across the same individuals. C, Alignment of mecA-positive regions of a S. aureus and S. epidermidis isolate derived from the same subject. D, Concatenated gene alignment tree of Staphylococcus phages from skin isolates with tip labels and heatmap showing host species of phages and the predominant genetic cargo of the phages, respectively E, CRISPR host analysis of viral clusters with multi-species host range. F, Network of the 14,298 plasmids identified in more than one isolate, clustered by identity. Only one plasmid from each plasmid cluster and subject is shown. G, Alignment of S. aureus plasmid from AD25 against a plasmid from a S. epidermidis isolate from the same subject.