Pan-genome and comparative genome analyses of propionibacterium acnes reveal its genomic diversity in the healthy and diseased human skin microbiome

Abstract

Propionibacterium acnes constitutes a major part of the skin microbiome and contributes to human health. However, it has also been implicated as a pathogenic factor in several diseases, including acne, one of the most common skin diseases. Its pathogenic role, however, remains elusive. To better understand the genetic landscape and diversity of the organism and its role in human health and disease, we performed a comparative genome analysis of 82 P. acnes strains, 69 of which were sequenced by our group. This collection covers all known P. acnes lineages, including types IA, IB, II, and III. Our analysis demonstrated that although the P. acnes pan-genome is open, it is relatively small and expands slowly. The core regions, shared by all the sequenced genomes, accounted for 88% of the average genome. Comparative genome analysis showed that within each lineage, the strains isolated from the same individuals were more closely related than the ones isolated from different individuals, suggesting that clonal expansions occurred within each individual microbiome. We also identified the genetic elements specific to each lineage. Differences in harboring these elements may explain the phenotypic and functional differences of P. acnes in functioning as a commensal in healthy skin and as a pathogen in diseases. Our findings of the differences among P. acnes strains at the genome level underscore the importance of identifying the human microbiome variations at the strain level in understanding its association with diseases and provide insight into novel and personalized therapeutic approaches for P. acnes-related diseases.

Importance: Propionibacterium acnes is a major human skin bacterium. It plays an important role in maintaining skin health. However, it has also been hypothesized to be a pathogenic factor in several diseases, including acne, a common skin disease affecting 85% of teenagers. To understand whether different strains have different virulent properties and thus play different roles in health and diseases, we compared the genomes of 82 P. acnes strains, most of which were isolated from acne or healthy skin. We identified lineage-specific genetic elements that may explain the phenotypic and functional differences of P. acnes as a commensal in health and as a pathogen in diseases. By analyzing a large number of sequenced strains, we provided an improved understanding of the genetic landscape and diversity of the organism at the strain level and at the molecular level that can be further applied in the development of new and personalized therapies.

Figures

FIG 1

P. acnes pan-genome. (A) A power law regression for new genes (n) discovered with the addition of new genome sequences (N). (B) A power law regression for total genes (n) accumulated with the addition of new genome sequences (N). Circles are the medians of n for 200 simulations. Error bars indicate the standard deviations for the 200 simulations.

FIG 2

A phylogenetic tree of 82 P. acnes strains and the distance matrix among the strains. A phylogenetic tree of 82 P. acnes strains was constructed based on the 123,223 SNPs in the core regions (2.20 Mb). The distances between strains were calculated as nucleotide substitution rates at all SNP sites, colored according to the scale bar. The strains from the same individuals (SSIs) belonging to the same lineages are marked with “+.”

FIG 3

SNP distribution in the core regions. (A) SNP frequencies (percentages of polymorphic sites) of the genes in the core regions. (B) K-S statistics for genes that had higher SNP frequencies with more than two standard deviations (SD). (C) Nonsynonymous mutation frequencies of the genes in the core regions. (D) K-S statistics for genes that had higher nonsynonymous mutation frequencies with more than 2 standard deviations.

FIG 4

P. acnes strains within each lineage share unique noncore genomic regions. Rows represent 82 P. acnes genomes, and columns represent 314 noncore regions that are longer than 500 bp. The genomes and the noncore regions were clustered based on similarity. The width of each block plotted is not proportional to the genomic length of each noncore region. The presence of a noncore region is colored in yellow, and the absence is colored in blue. The color schemes used for RT and clades are the same as in Fig. 2.