Strains and growth conditions
We used Pseudomonas aeruginosa wildtype strain PAO1 (ATCC 15692) and a pyoverdine-negative mutant, both constitutively expressing GFP (PAO1-gfp, PAO1-ΔpvdD-gfp), as positive and negative controls for pyoverdine production, respectively. We further used PAO1-pvdS_gene and PAO1-pvdS_prom, two mutants with strongly reduced pyoverdine production, that evolved de novo from PAO1-gfp during experimental evolution in iron-limited media (2.5 gL−1 BactoPeptone, 3 gL−1 NaCl, 5 mgL−1 Cholesterol, 25 mM MES buffer pH = 6.0, 1 mM MgSO4, 1 mM CaCl2, 200 μM 2,2′-Bipyridyl (Granato ET, Ziegenhain C, Marvig RL & Kümmerli R, unpublished)). PAO1-pvdS_gene carries a non-synonymous point mutation (G > C) in the pvdS gene that leads to an amino acid change (Met135Ile). PAO1-pvdS_prom carries a point mutation (G > T) in the consensus sequence of the −35 element in the promoter region upstream of pvdS. Both mutants constitutively express GFP. Throughout this publication, the two mutants are referred to as “pvdS_gene” and “pvdS_prom”.
For overnight pre-culturing, we used Luria Bertani (LB) medium, and incubated the bacteria under shaking conditions (190–200 rpm) for 16–18 h. All experiments in this study were conducted at 37 °C. Optical density (OD) of pre-cultures was determined at a wavelength of 600 nm in a spectrophotometer. We induced strongly iron-limiting growth conditions by using casamino acids (CAA) medium (5 gL−1 casamino acids; 1.18 gL−1 K2HPO4*3H2O; 0.25 gL−1 MgSO4*7H2O) supplemented with 25 mM HEPES and 400 μM of the iron chelator 2,2′-Bipyridyl. All chemicals were purchased from Sigma-Aldrich, Switzerland.
For conditions with medium or high iron availability, we further added FeCl3 at final concentrations of 1 μM or 40 μM, respectively. These levels of iron supplementation have previously been shown to either reduce pyoverdine production to intermediate levels (1 μM FeCl3) or to completely stall pyoverdine synthesis (40 μM FeCl3) [14]. Furthermore, competition experiments between PAO1 cooperators and their cheating isogenic knock-out mutant (PAO1 ΔpvdD ΔpchEF), deficient for siderophore production, revealed that cheats could only invade without the supplementation of extra iron [13].
We manipulated the spatial structure of the environment by growing bacteria either in liquid medium under shaking conditions (180 rpm; unstructured environment) or in viscous medium containing 0.1% agar under static conditions (structured environment). Competition experiments between PAO1 and its knock-out cheat, previously conduced in our laboratory, showed that cheats experienced a significant relative fitness advantage under well-mixed, but not under more viscous conditions [13].
Ancestral growth and pyoverdine kinetics
To measure growth and pyoverdine production kinetics of all strains in iron-limited media prior to experimental evolution, we washed bacterial pre-cultures twice with sterile NaCl (0.85%), adjusted OD600 to 1.0, and diluted 10−4 into 200 μL of iron-limited CAA (Bipyridyl 400 μM) per well in a 96-well plate. The plate was then incubated in a Tecan Infinite M-200 plate reader (Tecan Group Ltd., Switzerland) for 24 h, and OD600 and pyoverdine-specific fluorescence (emission 400 nm, excitation 460 nm) were measured every 15 min.
Experimental evolution
We conducted experimental evolution with pvdS_gene and pvdS_prom as starting points. We let each strain evolve independently under six different experimental treatments in a full-factorial design: 2 spatial structures (unstructured vs. structured) × 3 iron availabilities (low vs. medium vs. high iron availability) in three replicate independent lines (Fig. 2). At the start of the experimental evolution, overnight cultures of both clones were washed twice with NaCl (0.85%), adjusted to an OD600 of 1.0 and diluted 1:1000 into 200 μL of nutrient medium in 96-well plates. Plates were wrapped with parafilm, incubated for 24 h and subsequently diluted 1:1000 in fresh nutrient medium. We repeated this cycle for 20 consecutive transfers, allowing for approximately 200 generations of bacterial evolution (Fig. 2). At the end of the experiment, we prepared freezer stocks for each evolved population (n = 36) by mixing 100 μL of bacterial culture with 100 μL of sterile glycerol (85%). Samples were stored at −80 °C.
Isolation of single clones
To check whether evolved clones showed altered pyoverdine production levels compared to the ancestral pvdS_gene and pvdS_prom strains, we isolated a total of 720 evolved clones (20 per replicate and treatment). Specifically, we regrew evolved bacterial populations from freezer stocks in 5 mL LB medium for 16–18 h (180 rpm) and subsequently adjusted them to OD600 = 1.0. Then, 200 μL of 10−6 and 10−7 dilutions were spread on large LB agar plates (diameter 150 mm), which we incubated at 37 °C for 18–20 h. We then randomly picked twenty colonies for each of the 36 evolved populations, and immediately processed the clones for the pyoverdine measurement assay (see below).
Screen for evolved pyoverdine production levels
For each of the 720 evolved clones, we transferred a small amount of material from the agar plate directly into 200 μL of CAA + Bipyridyl (400 μM) in individual wells on a 96-well plate. We incubated plates with clones originating either from unstructured environments or structured environments for 24 h under shaken (180 rpm) or static conditions, respectively. Following incubation, we measured OD600 and pyoverdine-specific fluorescence (emission 400 nm, excitation 460 nm) in the Tecan Infinite M-200 plate reader as a single endpoint measurement. As controls, we included in three-fold replication on each plate: the high-producing PAO1 wildtype (positive control); the pyoverdine knockout mutant PAO1-ΔpvdD-gfp (negative control); the two low-producing mutants pvdS_gene and pvdS_prom; and blank growth medium. To preserve all tested clones for future experiments, we mixed 100 μL of bacterial culture with 100 μL of sterile glycerol (85%) for storage at −80 °C.
Confirmation of evolved pyoverdine phenotypes
Based on the screen above, we identified 34 clones with an altered pyoverdine production level (Additional file 1: Table S1). Specifically, we found five clones that seem to have restored pyoverdine production by roughly 50% (i.e. in terms of the difference between the low-producing ancestor cheat and the high-producing wildtype) and 29 clones that seem to produce less than 33% of pyoverdine compared to their ancestral pyoverdine low-producers (either pvdS_gene or pvdS_prom). We subjected these clones to an in-depth repeated screening of their pyoverdine phenotype. In addition, we selected two random clones per treatment (n = 24), from different evolved populations, that displayed no change in their production levels (compared to pvdS_gene or pvdS_prom). One clone had to be excluded due to contamination, so that the final sample size for this group of clones was n = 23. For all of these evolved clones (n = 57), we re-measured their pyoverdine production level in three-fold replication using the same protocol and controls as described above.
Sequencing of pvdS promoter and coding region
Since the ancestral low-producing strains (pvdS_gene or pvdS_prom) had mutations in the pvdS gene or its promoter, we were wondering whether the altered phenotypes observed in the evolved clones were based on reversion or additional mutations in this genetic region. To address this question, we PCR amplified and sequenced the pvdS gene and the upstream region containing the promoter sequence of all 57 evolved clones screened above. PCR mixtures consisted of 2 μl of a 10 μM solution of each primer, pvdS_fw (5′-GACGCATGACTGCAACATT-3′) and pvdS_rev (5′-CCTTCGATTTTCGCCACA-3′), 25 μl Quick-Load Taq 2X Master Mix (New England Biolabs), 1 μl of DMSO, and 20 μl of sterile Milli-Q water. We added bacterial biomass from glycerol stocks to the PCR mixture distributed in 96-well PCR plates. Plates were sealed with an adhesive film. We used the following PCR conditions: denaturation at 95 °C for 10 min; 30 cycles of amplification (1 min denaturation at 95 °C, 1 min primer annealing at 56 °C, and 1 min primer extension at 72 °C); final elongation at 72 °C for 5 min. The PCR products were purified and commercially sequenced using the pvdS_fw primer. While sequencing worked well for 51 clones, it failed for two clones, and resulted in partial sequences for six clones (Additional file 1: Table S1).
Statistical analysis
All statistical analyses were performed using R 3.2.2 [45]. We tested for treatment differences in the frequency of non- or low-producing strains using Fisher’s exact test and corrected for multiple testing using the Bonferroni correction. To compare pyoverdine production of evolved clones to that of the low-producing ancestors, we used one-way analyses of variance (ANOVA) and corrected for multiple testing using Tukey’s HSD (honest significant difference) test.