Whole genome sequencing distinguishes skin colonizing from infection-associated Cutibacterium acnes isolates

Abstract

Introduction: Cutibacterium acnes can both be a helpful colonizer of the human skin as well as the causative agent of acne and purulent infections. Until today, it is a moot point whether there are C. acnes strains exclusively devoted to be part of the skin microbiome and others, that carry special features enabling them to cause disease. So far, the search for the molecular background of such diverse behavior has led to inconsistent results.

Methods: In the present study, we prospectively collected C. acnes strains from 27 infected persons and 18 healthy controls employing rigid selection criteria to ensure their role as infectious agent or colonizer. The genome sequences from these strains were obtained and carefully controlled for quality.

Results: Deduced traditional phylotyping assigned almost all superficial isolates to type IA1, while the clinical strains were evenly distributed between types IA1, IB, and II. Single locus sequence typing (SLST) showed a predominance of A1 type for the control strains, whereas 56% of the clinical isolates belonged to types A1, H1 and K8. Pangenome analysis from all the present strains and 30 published genomes indicated the presence of an open pangenome. Except for three isolates, the colonizing strains clustered in clades separate from the majority of clinical strains, while 4 clinical strains clustered with the control strains. Identical results were obtained by a single nucleotide polymorphism (SNP) analysis. However, there were no significant differences in virulence gene contents in both groups.

Discussion: Genome-wide association studies (GWAS) from both the pangenome and SNP data consistently showed genomic differences between both groups located in metabolic pathway and DNA repair genes. Thus, the different behavior of colonizing and infectious C. acnes strains could be due to special metabolic capacities or flexibilities rather than specific virulence traits.

Keywords: Cutibacterium (Propionibacterium) acnes; GWAS - genome-wide association study; SLST; SNP; pangenome analyses.

Figure 1

Pangenome- and SNP- based cladograms. Trees were built with FastTree (Price et al., 2010), with the GTR substitution model for nucleotide sequences (-gtr -nt) from the alignment of core pangenome genes (A), or from SNPs obtained upon the alignment of all novel isolates against the NCBI GenBank designated C. acnes reference genome HL096PA1. (B). Colors in (A) stand for genome isolate groups, e.g. orange for novel genomes isolated from bone and joint, deep seated tissues or systemic, black for superficial anatomical novel isolates, and blue for known complete C. acnes genomes; while in (B) colors stand for bone and joint (orange), systemic (blue), or deep tissue (green) isolates, and black for those isolates that could not be classified as any from the former categories. The type strain ATCC 6919 is indicated by an arrow.

Figure 2

Pangenome- and SNP- based genome wide associations. Manhattan plots from the genome-wide associations (GWAS) for clinical phenotypes in the novel C. acnes isolates. GWAS for (A) pangenome data, and (B) single nucleotide polymorphisms (SNPs). Labels for the pangenes (A) are shown for those with -log(adjusted p-value) > 3 and with loci fully annotated, while labels for the genes corresponding to SNPs (B) are shown for those with -log(adjusted p-value) > 2. The x-axis shows the genomic positions (in base pairs) for the pangenes and SNPs along the reference genome; and the y-axis is the minus logarithm for the association (adjusted p-value).