Genomic analysis reveals that Pseudomonas aeruginosa virulence is combinatorial

Abstract

Background: Pseudomonas aeruginosa is a ubiquitous environmental bacterium and an important opportunistic human pathogen. Generally, the acquisition of genes in the form of pathogenicity islands distinguishes pathogenic isolates from nonpathogens. We therefore sequenced a highly virulent strain of P. aeruginosa, PA14, and compared it with a previously sequenced (and less pathogenic) strain, PAO1, to identify novel virulence genes.

Results: The PA14 and PAO1 genomes are remarkably similar, although PA14 has a slightly larger genome (6.5 megabses [Mb]) than does PAO1 (6.3 Mb). We identified 58 PA14 gene clusters that are absent in PAO1 to determine which of these genes, if any, contribute to its enhanced virulence in a Caenorhabditis elegans pathogenicity model. First, we tested 18 additional diverse strains in the C. elegans model and observed a wide range of pathogenic potential; however, genotyping these strains using a custom microarray showed that the presence of PA14 genes that are absent in PAO1 did not correlate with the virulence of these strains. Second, we utilized a full-genome nonredundant mutant library of PA14 to identify five genes (absent in PAO1) required for C. elegans killing. Surprisingly, although these five genes are present in many other P. aeruginosa strains, they do not correlate with virulence in C. elegans.

Conclusion: Genes required for pathogenicity in one strain of P. aeruginosa are neither required for nor predictive of virulence in other strains. We therefore propose that virulence in this organism is both multifactorial and combinatorial, the result of a pool of pathogenicity-related genes that interact in various combinations in different genetic backgrounds.

Figure 1

Circular map of the PA14 genome. The outermost circle represents the chromosomal location: major and minor ticks represent 500 and 100 kb increments, respectively. The origin and presumptive terminus of replication are indicated by green and red arrows, respectively. The locations and orientations of predicted genes are shown by rectangles in the next pair of circles; genes on the outer circle are transcribed on the plus strand and genes on the inner circle are transcribed on the minus strand. The genes are color-coded according to functional categories (see below). Blue arrow heads indicated the locations and relative orientations of four ribosomal RNA gene clusters; the published PAO1 sequence contains an inversion (gray arrow) with respect to PA14 resulting from a presumptive recombination event between two of the rRNA clusters. The innermost circle represents the GC content as calculated for non-overlapping 1 kb windows of the plus strand. A linear map and full annotations of each gene (including color codes for functional categories) are available at the Ausubel lab PA14 sequencing website [20] and the full sequence and annotations have been deposited in GenBank (GenBank: CP000438).

Figure 2

Chromosomal rearrangement in PAO1 repositions the replication terminus relative to the origin. (a) Schematic of PAO1 and PA14 chromosomes. The region with the same orientation in both strains is shown with a thick red line; a thin blue line represents the inverted region. Arrows represent the positions and orientations of the four ribosomal RNA clusters. PCR products designed with unique sequences flanking each rRNA cluster are indicated by numbers next to each arrow. PCR products 1 and 2 (purple numbers) are diagnostic for the PAO1 chromosome structure; PCR products 3 and 4 (black numbers) are diagnostic for the PA14 chromosome structure. The position of the presumptive terminus of replication in each strain is indicated by an orange triangle marked with the corresponding position along the chromosome (expressed as the percentage of the whole chromosome, starting from the origin of replication and moving in a clockwise direction). (b) Diagnostic long-range PCR spanning each ribosomal RNA repeat demonstrates the inversion in PAO1. PCR products corresponding to the numbers indicated in panel a were generated using genomic DNA from PAO1 (left panel) or PA14 (right panel). PCR products 1 and 2 were obtained only when using PAO1 genomic DNA (the weak background band for product 1 seen using PA14 as a template was also observed for 18 additional P. aeruginosa strains tested). PCR products 3 and 4 were obtained only when using PA14 genomic DNA. PCR products 5 and 6 were obtained from both strains. (c) GC skew analysis was performed using 1 kb windows, and the cumulative GC skew is shown on the y-axis as a function of chromosomal location (x-axis) for PA14 (black line) and PAO1 (purple line). The position of the peak indicates the likely position of the terminus of replication. For PA14 the peak is centered in the middle of the chromosome (at 49.2% of the genome, between coordinates 3,219,001 and 3,220,000), whereas the peak for PAO1 is offset with respect to the origin of replication as a result of the inversion (occurring at 38.8% of the genome, between coordinates 2,428,001 and 2,429,000). PCR, polymerase chain reaction.

Figure 3

ORF-by-ORF alignments of PA14 and PAO1. Each predicted ORF in PA14 was compared using BLAST with all annotated PAO1 ORFs, and the best match was indicated as a single data point with the chromosomal locations of the match in each genome. Reciprocal BLAST searches were also performed (using individual PAO1 ORFs against the complete set of PA14 ORFs). Matches in the forward direction are indicated by red diamonds, and matches in the reverse orientation are indicated by blue diamonds. Genes that fall outside of the diagonals (non-colinear genes) are a combination of translocations and gene pairs in which a true ortholog is missing in one strain but a similar gene exists elsewhere in its genome and the two genes have been selected as reciprocal best BLAST matches. Genes in one genome that fail to have a BLAST match counterpart in the other genome are represented on the x-axis for PA14-specific genes (green diamonds) and on the y-axis for PAO1-specific genes (pink diamonds). The GC contents for each genome are represented next to the two axes using a light-blue graph; the positions that correspond to GC contents of 30%, 50%, and 70% are indicated. BLAST, basic local alignment search tool; ORF, open reading frame.

Figure 4

Poor correlation between PA14-pathogenicity genes and virulence in other strains. (a) C. elegans survival curves in the presence of 19 P. aeruginosa strains and OP50 (an E. coli control). The names of each strain tested are sorted according to the rank order of virulence, from most virulent at the top to the least virulent at the bottom, as determined by examining the time required to kill 50% of the nematodes. Black brackets indicate strains with indistinguishable virulence. Strain names are followed by the strain source and color-coded by strain source, as shown in panel b. Strain CF27 is not shown in this dataset; however, a similar experiment places its rank order virulence between strains E2 and S36004 (indicated by a black arrow). The same relative rank orders were obtained in two additional experiments. (b) Dendrogram representing the relatedness of 20 P. aeruginosa strains based on the presence or absence of genes as assayed by genomic DNA hybridizations to a custom microarray. Hierarchical clustering analysis was performed using the city-block distance metric. The name of each strain is shown, along with the source of the strain (UTI, urinary tract infection; CF, cystic fibrosis respiratory infection; env, environmental isolate), and the rank order virulence of each strain as determined in panel a (1, most virulent; 20, least virulent). Strains with indistinguishable virulence were given a tied rank order (also see Table 2). (c) Presence or absence of PA14 virulence genes in additional isolates. Data for each strain tested is presented in columns. Strains are arranged in order from left to right in order of decreasing virulence, in the same order as shown in panel a; column headers refer to the 20 strain numbers used in Table 2. Columns with no gaps between them represent groups of strains with indistinguishable virulence (strains 4 and 5; strains 8, 9 and 10; and strains 14, 15 and 16). Genes assayed are represented as rows, with adjacent rows representing ORFs present within a given gene cluster (PA14 region names shown on the left are abbreviated to remove the 'PA14' prefix). Each gene is described as present (blue), absent (yellow), or indeterminate (red). The positions of mutations resulting in reduced virulence in PA14 are indicated by lines to the right; numbers in black refer to entries in Table 3, and red numbers in parenthesis indicate the Spearman's rank correlation coefficients between presence or absence of the gene in other strains and their rank order virulence ratings.