Phylogenetic Distribution of CRISPR-Cas Systems in Antibiotic-Resistant Pseudomonas aeruginosa

Abstract

Pseudomonas aeruginosa is an antibiotic-refractory pathogen with a large genome and extensive genotypic diversity. Historically, P. aeruginosa has been a major model system for understanding the molecular mechanisms underlying type I clustered regularly interspaced short palindromic repeat (CRISPR) and CRISPR-associated protein (CRISPR-Cas)-based bacterial immune system function. However, little information on the phylogenetic distribution and potential role of these CRISPR-Cas systems in molding the P. aeruginosa accessory genome and antibiotic resistance elements is known. Computational approaches were used to identify and characterize CRISPR-Cas systems within 672 genomes, and in the process, we identified a previously unreported and putatively mobile type I-C P. aeruginosa CRISPR-Cas system. Furthermore, genomes harboring noninhibited type I-F and I-E CRISPR-Cas systems were on average ~300 kb smaller than those without a CRISPR-Cas system. In silico analysis demonstrated that the accessory genome (n = 22,036 genes) harbored the majority of identified CRISPR-Cas targets. We also assembled a global spacer library that aided the identification of difficult-to-characterize mobile genetic elements within next-generation sequencing (NGS) data and allowed CRISPR typing of a majority of P. aeruginosa strains. In summary, our analysis demonstrated that CRISPR-Cas systems play an important role in shaping the accessory genomes of globally distributed P. aeruginosa isolates.

Importance: P. aeruginosa is both an antibiotic-refractory pathogen and an important model system for type I CRISPR-Cas bacterial immune systems. By combining the genome sequences of 672 newly and previously sequenced genomes, we were able to provide a global view of the phylogenetic distribution, conservation, and potential targets of these systems. This analysis identified a new and putatively mobile P. aeruginosa CRISPR-Cas subtype, characterized the diverse distribution of known CRISPR-inhibiting genes, and provided a potential new use for CRISPR spacer libraries in accessory genome analysis. Our data demonstrated the importance of CRISPR-Cas systems in modulating the accessory genomes of globally distributed strains while also providing substantial data for subsequent genomic and experimental studies in multiple fields. Understanding why certain genotypes of P. aeruginosa are clinically prevalent and adept at horizontally acquiring virulence and antibiotic resistance elements is of major clinical and economic importance.

FIG 1

Genomic diversity of P. aeruginosa clinical isolates. (A) Phylogenetic tree of P. aeruginosa clinical isolates. Phylogenetically related ST groups with ≥7 members are shown in different colors. The prevalent lineages constitute 52% of all strains analyzed. ST groups that have only 1 member are delineated in grey, while black denotes phylogenetic lineages with 2 to 6 members. The phylogenetic locations of the common laboratory strains PAO1, PA14, and PA7 are marked in black for reference, while the 4 nonclinical isolates are labeled with small black asterisks. (B) Rarefaction curve analysis of the total number of genes within the P. aeruginosa pangenome as a function of the number of genomes sequenced. Extrapolation is shown with dotted lines. (C) Rarefaction curve analysis of the number of new genes identified as a function of the number of P. aeruginosa genomes sequenced. A distribution of values was obtained for each strain count by permuting the order of the strains 500 times. The median trend line and 95% confidence intervals are shown in blue.

FIG 2

Known and potential new MDRPA lineages. Phylogenetic tree illustrating the locations of ST groups harboring ≥5 MDR strains (colored branches), 2 to 4 MDR strains (black branches), or 0 or 1 MDR strain (grey branches). Black spikes protruding from the circle denote the locations of XDRPA strains.

FIG 3

Distribution of CRISPR-Cas systems in P. aeruginosa. (A) Locations of CRISPR-Cas subtypes and anti-CRISPR-Cas genes among the 672 genomes analyzed. Purple, yellow, and red markers denote the locations of I-F, I-E, and I-C subtypes, respectively. Strains encoding CRISPR-Cas-inhibitory proteins are denoted in violet for type I-F and orange for type I-E. The locations and colors of major STs are highlighted identically to their depiction in Fig. 1A to aid in orientation. Degenerate type I-F systems lacking cas genes are denoted by short purple bars. (B) Diagram of the conserved gene content and location of each CRISPR-Cas subtype. Colored diamonds denote subtype-specific CRISPR repeats, while multicolored squares denote unique spacer content. (C) Heatmap illustrating that CRISPR array spacer content associates with phylogeny. The presence of every unique spacer sequence in each CRISPR-positive strain is denoted by blue squares. All strains are grouped by ST along the y axis, as denoted by various colors. The 3 most prevalent STs harboring each CRISPR subtype are shown.

FIG 4

In silico CRISPR target prediction. CRISPR arrays were computationally identified using PILER-CR and subsequently oriented uniformly. Wavy lines denote individual P. aeruginosa genomes. Sequences containing type I-C CRISPR arrays are shown with red diamond repeats, while those with I-E or I-F harbor yellow and purple diamonds, respectively. Conserved repeats of all three subtypes (I-C, I-E, and I-F) were computationally extracted, leaving only spacer content (rainbow-colored boxes), which provides the sequence specificity required for CRISPR-Cas targeting. Removal of spacer sequences with greater than 90% identity to other spacers provides a nonredundant spacer library. Unique spacer sequences are shown as different colors across the top of the image. Black squares indicate the presence of that spacer in a given strain. Spacer complementarity to known viral and plasmid databases indicates substantial CRISPR targeting; however, the majority of putative spacer targets were found in the accessory genomes of P. aeruginosa strains analyzed in this study. CRISPR-Cas target sequences (spacers) reside largely in the nonannotated accessory genomes of P. aeruginosa strains analyzed in this study. The pie chart denotes the percentage of unique spacers with complementarity to a given type of target sequence. Spacer sequences can be used to highlight previously uncharacterized accessory elements that are likely to be mobile genetic elements. Purple, red, and yellow squares represent the sites of complementarity for nonredundant type I-F, I-C, and I-F spacer sequences, respectively. φ1, φ2, and φ3 represent prophages predominantly found in the highly MDR and clinically prevalent ST111 P. aeruginosa lineage, while ICE1 represents a pKLC102-like element that is highly prevalent in ST111 strains.

FIG 5

Correlations between CRISPR-Cas subtypes and antibiotic resistance. (A) Clinical isolates containing CRISPR-Cas systems have, on average, smaller genomes than those lacking the same systems. Red, yellow, and purple dots represent the genome sizes of individual strains harboring the respective CRISPR-Cas system. Blue and grey dots denote the genome sizes of strains either lacking CRISPR-Cas systems or predicted to have CRISPR-Cas systems that are inactivated by anti-CRISPR-Cas genes, respectively. Box plots illustrate the distribution of genome sizes in each category. The upper and lower “hinges” of each box correspond to the first and third quartiles of the genome sizes for the clinical isolates in each category. Upper and lower whiskers extend from the given hinge to the highest or lowest value that is within 1.5* IQR, where IQR is the inter-quartile range, or distance between the first and third quartiles. The Wilcoxon rank-sum test and t test were used to compare the distributions of sizes between pairs, and multiple-hypothesis-adjusted P values are provided. (B) Locations of spacer target sequences in previously sequenced plasmids and pathogenicity islands. The colors correspond to the CRISPR-Cas subtypes depicted in panel A. (C) Correlation of antibiotic susceptibility with the presence or absence of a given CRISPR-Cas subtype. (D) Correlation of antibiotic resistance genes with the presence or absence of a given CRISPR-Cas subtype. (C and D) Data are corrected for the presence of anti-CRISPR-Cas genes. The P values are from the Wilcoxon rank-sum test. The false discovery rate (FDR) adjustment of the P values was performed using the Benjamini-Hochberg method with the function p.adjust in R.