Fig 1.
Bacteria integrate stimuli from the environment and decide whether to make biofilms or to move using the c-di-GMP network.
A: Bow-tie architecture of c-di-GMP signaling network: c-di-GMP is synthesized by diguanylate cyclase (DGC) proteins with GGDEF domains such as WspR, DipA, and SadC, and degraded by phosphodiesterases (PDE) proteins with EAL or HD-GYP domains such as BifA, and SadR. The DGCs and PDEs could sense stimuli—such as chemoattractants which could be a signal for motility, or mechanical contact with surfaces which could be a signal for biofilm formation—and change intracellular c-di-GMP levels in response; c-di-GMP effectors—such as c-di-GMP binding proteins and riboswitch RNAs—then sense c-di-GMP levels and control phenotype outputs such as biofilm formation, motility, virulence and cell division. B: At low levels of c-di-GMP the bacteria express flagella genes and go into motile mode. C: At high levels of c-di-GMP the bacteria repress flagella genes, express biofilm genes and go into biofilm mode.
Fig 2.
Phenotypic diversity in 28 P. aeruginosa isolates from acutely infected cancer patients at MSKCC explained by many small-effect alleles in c-di-GMP network.
A: Bulk c-di-GMP levels collected from bacterial colonies, including for the laboratory strain PA14. B: Biofilm levels measured in microtiter plates using the crystal-violet assay. C: Motility measured as swarm area after 16 h of incubation. D: Phylogenetic tree reconstructed from 88,347 genetic variants identified in core genes, including PA14 and two other laboratory strains PAO1 and PA7. Numbers shown represent the number of open-reading frames (ORFs) identified with c-di-GMP related motifs: GGDEF domain for synthesizing c-di-GMP, EAL for degrading c-di-GMP, and effector for sensing c-di-GMP. Some ORFs encode both GGDEF and EAL domains. E: Explaining diversity in c-di-GMP, biofilm and swarming required many alleles of small-effect in c-di-GMP genes identified within the 28 genomes. Model selection using LASSO revealed that a model that explains 85% of the phenotypic deviance requires including at least 21 genetic variants in c-di-GMP related genes. E’ shows a detail of LASSO model selection, which increases the tuning parameter λ and selects variants to include in the model. F: Each of the 21 genetic variants by itself explains 27% or less of the phenotypic variance, even in the best model selected by LASSO. The analysis supports that the phenotypic diversity observed among clinical isolates is due to small-effect alleles.
Fig 3.
Bow-tie model of biochemical reactions in the c-di-GMP network explains mutants of PA14 evolved in the laboratory.
A: Diagram of the bow-tie model showing the α and β coefficients for sensor and effector modules. B-D: A water tank diagram explains how the relative values of c-di-GMP and the effector setpoint lead to the specialist phenotypes for mutants (C,D) evolved from a generalist laboratory strain (PA14) in laboratory experiments (B).
Fig 4.
Specialist strains produced by strong selection in laboratory evolution have large-effect alleles in c-di-GMP network.
A: Bulk c-di-GMP levels measured for evolved mutants, collected from bacterial colonies. B: Diagram of drip flow biofilm reactor used in biofilm selection. C: Biofilm levels quantified by the crystal violet assay. D: Production of extracellular polymers required for biofilm formation, measured using the Congo-red assay. E: Expression of the gene fliC required for flagella synthesis, measured as GFP expressed by the reporter fusion PfliC-GFP. The data of three evolved mutants fleN*dipA*, fleN*dipA** and fleN*wspF* in B-E are statistically different from ancestral strain fleN* (P<0.05). F: Phylogenetic representation of the mutants evolved in laboratory experiments showing the tradeoff between biofilm and swarming.
Fig 5.
Mathematical model reveals how the c-di-GMP network fitness depends on the strength of selection and on the number of sensors.
A: The environment history was modeled as a succession of n binary environments E, 0-black or 1-white, corresponding respectively to motility- or biofilm-favoring environments. Stimuli (X) were generated from each environment by introducing noise to the original signal; the expressed phenotypes Y were calculated from the β’s and the matrix X; the fitness of the network in each environment is the agreement between the expressed phenotype and the favored phenotype in that environment; the fitness across the n environments is the geometric mean fitness. B: The fittest network was calculated using logistic regression algorithm. and
are the fitting parameters of the unbiased network for infinite history. C: The fittest network was presented to a new environment and a new set of stimuli and we calculated the expressed phenotype, as well as the fitness in that new environment. D: The fitness in changing environments depended strongly on n, the number of environments that tuned the c-di-GMP network during strain evolution. Strong selection selected for networks adapted to recent environment (small n) favoring specialists; weak selection provided the opportunity to learn from a long history of fluctuating environments (large n), favoring generalists. E: The fitness achieved by a c-di-GMP network depends on the number of sensory modules (m) and the length of evolutionary history (n, where small n corresponds to strong selection and large n corresponds to weak selection). Networks with too many sensors (m > n/2) performed well in the past but poorly in the future. The curves presented in D-E were obtained from numerical simulations of the scheme described in A-B-C (Logistic regression over a m × n matrix followed by the estimation of the fitness of the network on one new environment; 1000 independent simulations per conditions m, n; η = 0.6). Arithmetic mean was used to average these simulation results.
Fig 6.
Experimental tests reveal new mutations that regain hyperswarming to a biofilm specialist.
A: The wspF* biofilm specialist which has a repeat-insert in the wspF gene, initially cannot swarm but regains swarming by losing the repeat-insert when in swarming selection. B: An engineered fleN*ΔwspF strain also regains swarming despite lacking the wspF gene entirely. Survival analysis reveals that this mutant takes significantly longer than the wspF* to start swarming, but does so eventually. C: A spontaneous mutant in wspA regained swarming in the fleN*ΔwspF background. D-F: Diagram explaining how Wsp mutations enable switching between extremes of biofilm and swarming. When WspA senses an attachment signal, it transduces the signal to other Wsp proteins that phosphorylate protein WspR, which then produces c-di-GMP and the cells form biofilm (D). When WspF gains the insertion mutation, it fails to demethylate. WspR therefore is hyper-phosphorylated even in the absence of an attachment signal. (E). A ΔwspF mutant phenocopies wspF*. However, a spontaneous mutation in wspA enables cell to swarm. This mutation impairs biofilm formation even when the cells are placed under biofilm forming condition (F). G: Compilation of mutations identified from the mass swarming selection experiment started with the fleN*ΔwspF strain that revealed 43 new Wsp-disabling mutations.