Evolutionary remodeling of global regulatory networks during long-term bacterial adaptation to human hosts

Abstract

The genetic basis of bacterial adaptation to a natural environment has been investigated in a highly successful Pseudomonas aeruginosa lineage (DK2) that evolved within the airways of patients with cystic fibrosis (CF) for more than 35 y. During evolution in the CF airways, the DK2 lineage underwent substantial phenotypic changes, which correlated with temporal fixation of specific mutations in the genes mucA (frame-shift), algT (substitution), rpoN (substitution), lasR (deletion), and rpoD (in-frame deletion), all encoding regulators of large gene networks. To clarify the consequences of these genetic changes, we moved the specific mutations, alone and in combination, to the genome of the reference strain PAO1. The phenotypes of the engineered PAO1 derivatives showed striking similarities with phenotypes observed among the DK2 isolates. The phenotypes observed in the DK2 isolates and PAO1 mutants were the results of individual, additive and epistatic effects of the regulatory mutations. The mutations fixed in the σ factor encoding genes algT, rpoN, and rpoD caused minor changes in σ factor activity, resulting in remodeling of the regulatory networks to facilitate generation of unexpected phenotypes. Our results suggest that adaptation to a highly selective environment, such as the CF airways, is a highly dynamic and complex process, which involves continuous optimization of existing regulatory networks to match the fluctuations in the environment.

Keywords: chronic infection; epistasis; evolution of regulatory networks; gene expression; microbial evolution.

Fig. 1.

Evolutionary relationship and dynamics of phenotypic changes in the DK2 lineage. (A) Phylogenetic tree showing the genetic relationship between isolates of the DK2 lineage. Symbols represent DK2 isolates sampled at different time points (indicated on the time line) from different patients with CF (indicated by symbol shape). (B) Heat map representing the dynamics of phenotypic changes in the DK2 lineage. Colors represent similarities/differences to PAO1 and the DK2 reference isolate CF333-2007, respectively, in terms of gene expression; red indicates an expression profile similar to that of PAO1 (WT) whereas light yellow indicates an evolved expression profile similar to that of isolate CF333-2007. The figure is based on genomic and gene expression data from ref. .

Fig. 2.

Phenotypic impact of four regulatory mutations. (A) Colony morphologies of PAO1-derived mutants constructed by stepwise introduction of the evolved alleles from the DK2 lineage (B) Colony morphologies of DK2 isolates without (CF114-1973) and with (CF30-1979) the four regulatory mutations. (C) Venn diagram showing the overlap between the expression profiles of CF30-1979 and the Q mutant. The numbers in brackets indicate the number of false positives expected by chance within each subset of data (Materials and Methods).

Fig. 3.

Interactions between evolved regulatory networks. (A) Four-way Venn diagrams show the overlap between expression profiles of the NS mutants and the Q mutant. Expression profiles were divided into subprofiles of up- and down-regulated genes. Gray shading indicates the number of overlapping expression profiles (intersection degree), whereas numbers refer to the numbers of genes within each intersection. The proportion of false positives to be expected by chance within the expression profiles are indicated according to the intersection degree. Numbers in brackets indicate the total number of genes within each (sub-) expression profile. (B) Cluster I: cluster profiles of genes oppositely expressed as a result of counteracting effects of different mutations. Cluster II: cluster of genes being significantly expressed in two different NS mutants but weakly expressed in the Q mutant. Cluster III: cluster of genes being positively expressed in Q-mutant despite neutral or negative expression of in the NS mutants. Black line indicates the mean expression of the genes in each cluster. Dashed line indicates the mean (baseline) expression level of WT (PAO1).

Fig. 4.

(A) Venn diagrams show size and overlap between native and evolved regulons of RpoN and AlgT, respectively, after correction for false positives. (B) Scatterplot shows regulatory response (fold changes) from the native and the evolved σ factors. Linear correlations are indicated by the squared value of R. Red data points indicate an opposite gene expression response between the native and the evolved network.

Fig. 5.

Epistatic effects cause increased tolerance toward tobramycin and ceftazidime. Shown are MIC values of tobramycin and ceftazidime for PAO1 and the PAO1-derived mutants. Double asterisks indicate significant differences (P < 0.01) in MIC relative to PAO1. “+ lasR” indicates that the mutant was complemented (in cis) with a WT copy of lasR. Error bars represent SDs from three experiments.

Fig. 6.

Phenotype switching in response to osmotic stress. Shown are changes in colony morphology of a mucA^DK2 algT^DK2 mutant and the WT (PAO1) during osmotic stress on agar plates (LB + 0.3 M NaCl). Both genotypes appear nonmucoid under normal growth conditions whereas only the mucA^DK2 algT^DK2 mutant becomes mucoid in response to osmotic stress. A time-lapse video is provided as Movie S1.

Fig. 7.

A mutation in rpoD results in constitutive mucoidy in the DK2 clone type. (A) PCA plot shows the expression profile of PAO1 (green), the nonmucoid DK2 isolates from patient CF333 (blue), and the constitutive mucoid DK2 isolates from CF333 (cyan). Error bars indicate SD from triplicate experiments. (B) Phylogenetic three showing the genetic relationship between DK2 isolates from patient CF333. (C) Schematic representation of the principal σ factor RpoD and the change caused by the 3-bp in-frame deletion in the region 3 domain (domains indicated by boxes). (D) Colony morphologies of PAO1 (WT) and the PAO1 derived mutants rpoD^DK2, mucA^DK2 algT^DK2, and mucA^DK2 algT^DK2 rpoD^DK2 grown on LB agar plates. Only the latter configuration of mutations gave rise to a constitutive mucoid phenotype similar to that of a mucA mutant.