Parallel evolutionary paths to produce more than one Pseudomonas aeruginosa biofilm phenotype

Abstract

Studying parallel evolution of similar traits in independent within-species lineages provides an opportunity to address evolutionary predictability of molecular changes underlying adaptation. In this study, we monitored biofilm forming capabilities, motility, and virulence phenotypes of a plethora of phylogenetically diverse clinical isolates of the opportunistic pathogen Pseudomonas aeruginosa. We also recorded biofilm-specific and planktonic transcriptional responses. We found that P. aeruginosa isolates could be stratified based on the production of distinct organismal traits. Three major biofilm phenotypes, which shared motility and virulence phenotypes, were produced repeatedly in several isolates, indicating that the phenotypes evolved via parallel or convergent evolution. Of note, while we found a restricted general response to the biofilm environment, the individual groups of biofilm phenotypes reproduced biofilm transcriptional profiles that included the expression of well-known biofilm features, such as surface adhesive structures and extracellular matrix components. Our results provide insights into distinct ways to make a biofilm and indicate that genetic adaptations can modulate multiple pathways for biofilm development that are followed by several independent clinical isolates. Uncovering core regulatory pathways that drive biofilm-associated growth and tolerance towards environmental stressors promises to give clues to host and environmental interactions and could provide useful targets for new clinical interventions.

Keywords: Biofilms; Next-generation sequencing.

Fig. 1. Biofilms of clinical P. aeruginosa isolates fall into three major clusters independent of their phylogenetic background.

a Despite a large structural diversity in biofilms of 414 clinical isolates, groups of strains that share structural characteristics were identified by visual inspection of biofilm microscopy images. Biofilms were grown for 48 h in a microtiter plate-based in vitro biofilm assay; images were acquired using confocal laser scanning microscopy (CLSM) following live/dead staining. Living cells are displayed in green (Syto9); dead cells in red (propidum iodide: PI). 3D reconstructions were generated with the Imaris Software. The scale bar represents 50 µm. b Representative biofilm images of selected clinical isolates show exemplarily the structural characteristics of the three major biofilm clusters, of which each contains 59 strains (cluster A), 38 strains (cluster B), and 47 strains (cluster C), respectively. c The phylogenetic relationship of 33 representative clinical isolates is displayed in a phylogenetic tree based on 3524 genes that are present in the DNA sequences of 414 clinical isolates and 5 reference strains. The color code of the strain names represents the affiliation to a certain biofilm cluster: Red—cluster A; green—cluster B; blue—cluster C. Reference strains are displayed in black. The proportion of PAO1-like strains is highlighted in light gray; PA14-like strains are highlighted in dark gray. d Crystal violet quantification was performed for 33 representative clinical strains to assess air–liquid biofilm formation on a PVC surface after 24 h. Each datapoint represents one clinical isolate. Statistical significance was calculated using Tukey’s HSD (honest significant difference) following analysis of variance (ANOVA) and is displayed as *p < 0.05.

Fig. 2. Transcriptional profiles recorded for biofilm-associated conditions exhibit a higher diversity than those recorded for planktonic conditions.

a The multidimensional scaling plot (MDS) of transcriptional profiles of 77 clinical strains shows a higher divergence in biofilm conditions (BF; filled circles) as compared to planktonic conditions (PL; triangles). b Pairwise measurements of Euclidian distances and c the Pearson’s distance (Pearson correlation coefficient subtracted from 1 to describe the variance) between the samples of the two culture conditions (biofilm and planktonic) are depicted. Significance of the Wilcoxon’s rank sum-test is displayed as ****p < 0.0001. Boxplot elements are: center line—median; box limits—upper and lower quartiles; whiskers—1.5× interquartile range; points—outliers. d Clinical isolates show a broad range in the number of differentially expressed genes (biofilm versus planktonic). The number of core biofilm transcriptome genes present in each strain is displayed in dark red (upregulated in biofilms) and dark blue (downregulated in biofilms).

Fig. 3. Biofilm clusters exhibit distinct transcriptional signatures in biofilm growth conditions.

Biofilm (BF) but not planktonic (PL) transcriptional profiles show a grouping according to the biofilm structure. Ninety-five percent confidence intervals are displayed by ellipses. Each datapoint represents the transcriptional profile of one clinical isolate in a certain growth condition (BF: filled circles; PL: triangles). The color code represents the affiliation to a certain biofilm cluster: red—cluster A; green—cluster B; blue—cluster C.

Fig. 4. Upregulated genes in biofilm growth in comparison to planktonic growth.

a The Venn diagram depicts commonly upregulated genes (70 genes) among all three biofilm clusters as well as cluster-specific regulated genes (A: 388; B: 33; C: 85 genes). b The GO term enrichment analysis of upregulated genes shows biological functions that are exclusively regulated in a certain biofilm cluster or shared by two or all three structural groups. The red color gradient represents the value of the enrichment factor. c A significant higher pyoverdine production was observed in biofilms (48 h) compared to planktonic cultures (24 h), as shown exemplarily for 33 clinical isolates. d Pyoverdine production in biofilm cultures was enhanced in all three biofilm clusters, independent of structural characteristics. Statistical significance was calculated with the Student’s t test and is displayed as *p < 0.05. Each datapoint represents one individual clinical isolate. Error bars represent the standard deviation.

Fig. 5. Biofilm cluster-specific isolates differ in their production of various virulence factors.

a In vivo virulence using the Galleria mellonella model, b in vitro cytotoxicity on A549 epithelial cells, c intracellular second messenger c-di-GMP levels, d swimming motility, e twitching motility, f swarming motility, g pyocyanin production, h elastase secretion, and i protease production of selected isolates belonging to the three different biofilm clusters are depicted. Each dot represents one individual clinical isolate. Levels of statistical significance were calculated using Tukey’s HSD (honest significant difference) following analysis of variance (ANOVA) and are displayed as ****p < 0.0001, **p < 0.01, or *p < 0.05.