Genome analysis of a transmissible lineage of pseudomonas aeruginosa reveals pathoadaptive mutations and distinct evolutionary paths of hypermutators

Abstract

Genome sequencing of bacterial pathogens has advanced our understanding of their evolution, epidemiology, and response to antibiotic therapy. However, we still have only a limited knowledge of the molecular changes in in vivo evolving bacterial populations in relation to long-term, chronic infections. For example, it remains unclear what genes are mutated to facilitate the establishment of long-term existence in the human host environment, and in which way acquisition of a hypermutator phenotype with enhanced rates of spontaneous mutations influences the evolutionary trajectory of the pathogen. Here we perform a retrospective study of the DK2 clone type of P. aeruginosa isolated from Danish patients suffering from cystic fibrosis (CF), and analyze the genomes of 55 bacterial isolates collected from 21 infected individuals over 38 years. Our phylogenetic analysis of 8,530 mutations in the DK2 genomes shows that the ancestral DK2 clone type spread among CF patients through several independent transmission events. Subsequent to transmission, sub-lineages evolved independently for years in separate hosts, creating a unique possibility to study parallel evolution and identification of genes targeted by mutations to optimize pathogen fitness (pathoadaptive mutations). These genes were related to antibiotic resistance, the cell envelope, or regulatory functions, and we find that the prevalence of pathoadaptive mutations correlates with evolutionary success of co-evolving sub-lineages. The long-term co-existence of both normal and hypermutator populations enabled comparative investigations of the mutation dynamics in homopolymeric sequences in which hypermutators are particularly prone to mutations. We find a positive exponential correlation between the length of the homopolymer and its likelihood to acquire mutations and identify two homopolymer-containing genes preferentially mutated in hypermutators. This homopolymer facilitated differential mutagenesis provides a novel genome-wide perspective on the different evolutionary trajectories of hypermutators, which may help explain their emergence in CF infections.

Figure 1. Patient origin and sampling time of genome sequenced P. aeruginosa DK2 isolates.

The collection of 55 P. aeruginosa isolates of the DK2 clone type was sampled from 21 different CF patients over 38 years. Bacterial isolates are indicated by symbols, and if multiple isolates were sampled the same year from a patient, they are represented by stacked symbols. The isolates are named from the patient from whom they were isolated, and their isolation year (e.g. isolate CF173-1991).

Figure 2. Maximum-parsimonious reconstruction of the phylogeny of the P. aeruginosa DK2 clones.

The phylogenetic tree is based on 7,326 SNPs identified from whole-genome sequencing, and lengths of branches are proportional to the number of mutations. Outlier isolate CF510-2006 (not shown) was used as an outgroup to determine the root of the tree. Branches leading into mutS, mutL, or, mutY hypermutable isolates are named as indicated by italic letters. Statistics on mutations accumulated in the specific branches are summarized in Table 1. Circles labeled A1, A2, B1, B2, C1, and C2, respectively, denotes the position of the first and last genotype of each of the DK2 sub-lineages A, B, and C which were observed to have infected patient CF173.

Figure 3. Total number of SNPs accumulated in each DK2 isolate.

The number of SNPs accumulated in each of the isolates since their most recent common ancestor (MRCA) is plotted against isolate sampling year.

Figure 4. Mutation rates of homopolymers.

Rates of mutation of homopolymers of different sizes are shown for seven DK2 sub-lineages evolving with a mutS/mutL DNA MMR-deficiency. The rates were calculated as the number of observed indels per homopolymer per mutS/mutL MMR-deficiency caused SNP (see Materials and Methods).

Figure 5. Bayesian phylogenetic reconstruction and divergence date estimates of the P. aeruginosa DK2 clones.

Bayesian statistics were used to estimate the divergence times of predicted ancestors. The tree was based on 736 unique SNPs identified from whole-genome sequencing. Circles labeled A1, A2, B1, B2, C1, and C2, respectively, denotes the position of the first and last genotype of each of the DK2 sub-lineages A, B, and C which were observed to have infected patient CF173. The position of CF173-1991 (A2) is approximated from an equivalent Bayesian phylogenetic analysis including the hypermutator isolates (Figure S3).

Figure 6. Increased pressures of selection for mutations in the top most mutated genes.

Measures of the selection pressures were plotted for genes acquiring ≥X mutations during the evolution of the DK2 lineage. Plot A shows the dN/dS ratio, and plot B shows the ratio of indels relative to silent SNPs.

Figure 7. Pathoadaptive genes.

Genes identified from parallel evolution to be involved in host adaptation. Colors of squares denotes if the gene was mutated relative to the MRCA of the DK2 clones. Only genes mutated in any of the isolates from CF173 are shown (see full list of pathoadaptive genes in Table S2). The presence of mutations is shown for the first and last genotypes of each of the DK2 sub-lineages A, B, and C, which were observed to have infected patient CF173. The total sum of mutations observed within each of the genotypes is indicated at the top. Genes are grouped by function. Details about specific mutations and their fixation in the DK2 isolates are given in Table S5, Table S6, and Figure S4.