Accelerated evolution in networked metapopulations of Pseudomonas aeruginosa
- University of Ottawa, Canada
- McGill University, Canada
Figures
Figure 1
Design for de novo evolution experiment.
(A) Two topologies, star or well-mixed, constructed among four subpopulations. Arrows depict dispersal routes among subpopulations (circles). (B) Four combinations of mutation supply rates achieved by manipulating both effective population sizes of the subpopulations and the migration rates among the subpopulations. (C) Experimental evolution setup and subsequent assays performed.
Figure 2 with 1 supplement
Dynamics of adaptation in large metapopulations.
Fitness trajectories in large metapopulations connected by high and low migration rates (LPHM and LPLM). Increase in relative growth rate (upper panel) and relative carrying capacity (lower panel) in metapopulations propagated by either the star or the well-mixed topology with the LPHM (left panel) or LPLM (right panel) regime over the experimental time-period. Each point is the mean of eight replicate metapopulations for a particular day and network topology; error bars show 1 standard error of the mean (SE). Raw data from each replicate metapopulation shown in Figure 2—figure supplement 1. For large metapopulations (LPHM and LPLM), approximately 6.67 generations of growth happened per transfer.
Figure 2—figure supplement 1
Dynamics of adaptation in large metapopulations with individual replicate metapopulations.
Fitness trajectories in large metapopulations connected by high and low migration rates (LPHM and LPLM). Increase in relative growth rate (upper panel) and relative carrying capacity (lower panel) in metapopulations propagated by either the star or the well-mixed topology with the LPHM (left panel) or LPLM (right panel) regime over the experimental time-period. Each point is the mean of eight replicate metapopulations for a particular day and network topology; error bars show 1 standard error of the mean (SE). Raw data from each replicate metapopulation is shown as faded lines. For large metapopulations (LPHM and LPLM), approximately 6.67 generations of growth happened per transfer.
Figure 3 with 2 supplements
Dynamics of adaptation in small metapopulations.
Fitness trajectories in small metapopulations (SPHM and SPLM) connected by high and low migration rates. Increase in relative growth rate (upper panel) and relative carrying capacity (lower panel) in metapopulations propagated by either the star or the well-mixed topology with the SPHM (left vertical panel) or SPLM (right vertical panel) regime over the experimental time-period. Each point is the mean of eight replicate metapopulations for a particular day and network topology; error bars show 1 standard error of the mean (SE). Raw data from each replicate metapopulation is shown as faded lines in Figure 3—figure supplement 1. Small metapopulations (SPHM and SPLM) experienced approximately 13.28 generations of growth per transfer.
Figure 3—figure supplement 1
Dynamics of adaptation in small metapopulations with individual replicate metapopulations.
Fitness trajectories in small metapopulations (SPHM and SPLM) connected by high and low migration rates. Increase in relative growth rate (upper panel) and relative carrying capacity (lower panel) in metapopulations propagated by either the star or the well-mixed topology with the SPHM (left vertical panel) or SPLM (right vertical panel) regime over the experimental time-period. Each point is the mean of eight replicate metapopulations for a particular day and network topology; error bars show 1 standard error of the mean (SE). Raw data from each replicate metapopulation is shown as faded lines. Small metapopulations (SPHM and SPLM) experienced approximately 13.28 generations of growth per transfer.
Figure 3—figure supplement 2
Detection of outliers in the SPHM and SPLM dataset.
Visual detection of outliers (beyond 25 and 75 quartile range) in both SPHM and SPLM growth rate and carrying capacity dataset. Excluding the outliers from the statistical analyses did not change the significant main effect of network (p = 0.073 and p = 0.022) in explaining the variance of fold change r and K, respectively.
Figure 4 with 2 supplements
Number of mutational changes in each population size/network topology treatment combinations.
Shown are the mean number of mutational changes for each network topology treatment (star and well-mixed) under each population size (large and small). Error bars represent 1 standard error of the mean.
Figure 4—figure supplement 1
Average frequency of adaptive mutations in the metapopulations.
Average frequencies at which mutations in key adaptive genes present in the evolved metapopulations. The differences between the network treatments are not statistically significant. For a list of putatively adaptive genes, see annotations in Figure 6 in the manuscript.
Figure 4—figure supplement 2
Average frequency of mutations segregating in networks with different population sizes.
Shown are the mean frequencies of all mutations in each replicate (black filled circle, total six) for each network and population size combinations. The overall average and standard error are also shown.
Figure 5 with 1 supplement
Distribution of fitness effects among isolates in the selection medium.
Absolute fitness of evolved end-point isolates in the selection medium (LB supplemented with subinhibitory concentration of ciprofloxacin). Fitness effect distributions are shown as histograms of carrying capacity (K) for all mutants isolated from the large (LPHM and LPLM) metapopulations (A and B, respectively) or the small (SPHM and SPLM) metapopulations (C and D, respectively). Blue and red bars denote isolates collected from metapopulations propagated either by the star or the well-mixed topologies, respectively. Statistical significance determined at p < 0.05 by permutation K–S test (10,000 permutations). The vertical line on the left-hand side of each histogram is the mean K of the ancestor in the selection medium and the red dotted vertical lines are 1 standard error (SE) of the mean.
Figure 5—figure supplement 1
Empirical cumulative distribution function (ECDF) plots for observed absolute fitnesses of evolved clones.
(A) LPHM, (B) LPLM, (C) SPHM, and (D) SPLM, respectively. The ECDFs are shifted to the right for the small population size stars (C, D), consistent with large effect mutants being substituted in stars relative to well-mixed populations.
Figure 6 with 2 supplements
Genetic changes detected in the metapopulations (not showing intergenic mutations; full data file is available in Figure 6—figure supplement 1) after ~100 generations of evolution in the selection media (LB supplemented with subinhibitory concentration of ciprofloxacin).
From the top, the first and the second panels show large star and large well-mixed metapopulations, respectively. Similarly, the third and the fourth panels show small star and small well-mixed metapopulations, respectively. Mutation in genes highlighted with either red or green background denotes recurring canonical ciprofloxacin resistance genes and genes that are relevant for adaptation of P. aeruginosa to laboratory conditions, respectively. Size of the circles depicts observed frequencies of mutants in the metapopulations. More than one circle for a gene represents the presence of distinct genetic variants (alleles) in the same metapopulation.
Figure 6—figure supplement 1
Genetic changes detected in the metapopulations (showing all mutations above the frequency of 8%) after ~100 generations of evolution in the selection media (LB supplemented with subinhibitory concentration of ciprofloxacin).
All the other details are the same as in Figure 6.
Figure 6—figure supplement 2
Frequency spectra for all mutations.
Distribution of read frequencies for all mutations found across all evolved metapopulations – shown are the nonsynonymous, indels and intergenic mutations (blue) and synonymous (dark blue). The dashed vertical line represents the threshold of 8% which is the frequency of the majority of synonymous SNPs.
Author response image 1
Author response image 2
Author response image 3
Author response image 4
Author response image 5
Author response image 6
Tables
Table 1
Population-level parallelism.
Difference in population-level parallelism between the two topologies for each effective population size. Differences between the mean values for each metric (estimate) in the star and the well-mixed topologies are presented. Low dispersion and Jaccard index values, and high C-scores, correspond to high rates of parallelism. A positive value denotes the calculated metric for the star topology is higher than the well-mixed topology and vice versa. Significance (p < 0.05) is determined by a two-way ANOVA for dispersion and a two-way ANOVA followed by a permutation test (10,000 permutations) for Jaccard distance and C-score (Materials and methods).
| Parallelism estimate = (Star − Well-mixed) | ||||||
|---|---|---|---|---|---|---|
| Metric | ||||||
| Population size | Dispersion | Jaccard distance | C-score | |||
| Estimate | Significance | Estimate | Significance | Estimate | Significance | |
| Large | 0.0167 | 0.7153 | 0.0165 | 0.6569 | 0.0219 | 0.5356 |
| Small | –0.1265 | 0.0108 | –0.1514 | 0.0009 | 2.0156 | 0.0029 |
Author response table 1
Analysis of Deviance Table (Type II Wald chisquare tests).
| Response: fold_change_k | ||||
|---|---|---|---|---|
| Chisq | Df | Pr(>Chisq) | ||
| poly(Day, 2) | 620.6681 | 2 | < 2e-16 | *** |
| Treatment | 5.1939 | 1 | 0.02267 | * |
| Mig_rate | 0.4826 | 1 | 0.48726 | |
| poly(Day, 2):Treatment | 6.4263 | 2 | 0.04023 | * |
| poly(Day, 2):Mig_rate | 3.3172 | 2 | 0.19040 | |
| Treatment:Mig_rate | 0.3448 | 1 | 0.55709 | |
| poly(Day, 2):Treatment:Mig_rate | 1.4588 | 2 | 0.48219 |
-
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Author response table 2
Analysis of Deviance Table (Type II Wald chisquare tests).
| Response: fold_change_k | ||||
|---|---|---|---|---|
| Chisq | Df | Pr(>Chisq) | ||
| poly(Day, 2) | 416.6435 | 2 | < 2e-16 | *** |
| Treatment | 3.2106 | 1 | 0.07316 | |
| Mig_rate | 0.0912 | 1 | 0.76264 | |
| poly(Day, 2):Treatment | 1.4415 | 2 | 0.48639 | |
| poly(Day, 2):Mig_rate | 0.1835 | 2 | 0.91236 | |
| Treatment:Mig_rate | 0.0284 | 1 | 0.86612 | |
| poly(Day, 2):Treatment:Mig_rate | 2.0834 | 2 | 0.35285 |
-
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Author response table 3
Response: mut_count.
| Df | Sum Sq | Mean Sq | F value | Pr(>F) | |
|---|---|---|---|---|---|
| Pop_size | 1 | 130.73 | 130.731 | 2.6323 | 0.1212 |
| Treatment | 1 | 160.98 | 160.975 | 3.2412 | 0.0877 |
| Pop_size:Treatment | 1 | 0.57 | 0.573 | 0.0115 | 0.9156 |
| Residuals | 19 | 943.93 | 49.665 |
-
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Author response table 4
Pop_size = large.
| contrast estimate | SE | df | t ratio | p value |
|---|---|---|---|---|
| AMP – WM | -5.00 | 4.07 | -1.229 | 0.2341 |
Author response table 5
Pop_size = small.
| contrast estimate | SE | df | t ratio | p value |
|---|---|---|---|---|
| AMP – WM | 5.63 | 4.27 | -1.320 | 0.2025 |
Author response table 6
Analysis of Deviance Table (Type II Wald chisquare tests).
| Response: fold_change_k | ||||
|---|---|---|---|---|
| Chisq | Df | Pr(>Chisq) | ||
| poly(Day, 2) | 391.6967 | 2 | < 2.2e-16 | *** |
| Pop_size | 1.0132 | 1 | 0.3141308 | |
| Treatment | 0.8521 | 1 | 0.3559552 | |
| Mig_rate | 0.0032 | 1 | 0.9547399 | |
| poly(Day, 2):Pop_size | 18.1228 | 2 | 0.0001161 | *** |
| poly(Day, 2):Treatment | 1.2515 | 2 | 0.5348471 | |
| Pop_size:Treatment | 1.5562 | 1 | 0.2122273 | |
| poly(Day, 2):Mig_rate | 1.6337 | 2 | 0.4418171 | |
| Pop_size:Mig_rate | 0.7332 | 1 | 0.3918523 | |
| Treatment:Mig_rate | 0.0139 | 1 | 0.9062678 | |
| poly(Day, 2):Pop_size:Treatment | 0.9580 | 2 | 0.6194079 | |
| poly(Day, 2):Pop_size:Mig_rate | 0.5315 | 2 | 0.7666358 | |
| poly(Day, 2):Treatment:Mig_rate | 0.2008 | 2 | 0.9044948 | |
| Pop_size:Treatment:Mig_rate | 0.0384 | 1 | 0.8446889 | |
| poly(Day, 2):Pop_size:Treatment:Mig_rate | 0.2420 | 2 | 0.8860213 |
-
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Author response table 7
Analysis of Deviance Table (Type II Wald chisquare tests).
| Response: fold_change_k | ||||
|---|---|---|---|---|
| Chisq | Df | Pr(>Chisq) | ||
| poly(Day, 2) | 476.1801 | 2 | < 2.2e-16 | *** |
| Pop_size | 7.9828 | 1 | 0.004722 | ** |
| Treatment | 4.7775 | 1 | 0.028833 | * |
| Mig_rate | 0.1931 | 1 | 0.660310 | |
| poly(Day, 2):Pop_size | 39.6929 | 2 | 2.403e-09 | *** |
| poly(Day, 2):Treatment | 0.4113 | 2 | 0.814130 | |
| Pop_size:Treatment | 1.9869 | 1 | 0.158663 | |
| poly(Day, 2):Mig_rate | 0.1203 | 2 | 0.941628 | |
| Pop_size:Mig_rate | 0.2862 | 1 | 0.592641 | |
| Treatment:Mig_rate | 0.1344 | 1 | 0.713942 | |
| poly(Day, 2):Pop_size:Treatment | 3.6815 | 2 | 0.158698 | |
| poly(Day, 2):Pop_size:Mig_rate | 0.6942 | 2 | 0.706739 | |
| poly(Day, 2):Treatment:Mig_rate | 0.0364 | 2 | 0.981947 | |
| Pop_size:Treatment:Mig_rate | 0.3081 | 1 | 0.578840 | |
| poly(Day, 2):Pop_size:Treatment:Mig_rate | 0.6055 | 2 | 0.738798 |
-
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Author response table 8
Small population size, r.
| contrast estimate | SE | df | T ratio | P value | |
|---|---|---|---|---|---|
| AMP – WM | 0.0787 | 0.0398 | 47.7 | 1.978 | 0.0538 |
-
Results are averaged over the levels of: Mig_rate
-
Degrees-of-freedom method: kenward-roger
Author response table 9
eff_size(emm_T, σ = σ(lmer.r.small.pop), edf = edf).
| contrast effect.size | SE | df | Lower CL | Upper CL | |
|---|---|---|---|---|---|
| AMP – WM | 0.929 | 0.479 | 47.7 | 0.0349 | 1.89 |
-
Results are averaged over the levels of: Mig_rate σ used for effect sizes: 0.08469
-
Degrees-of-freedom method: inherited from kenward-roger when re-gridding
-
Confidence level used: 0.95
Author response table 10
Small population size, K.
| contrast estimate | SE | df | t.ratio | p.value | |
|---|---|---|---|---|---|
| AMP – WM | 0.709 | 0.206 | 51.2 | 3.441 | 0.0012 |
-
Results are averaged over the levels of: Mig_rate
-
Degrees-of-freedom method: kenward-roger
Author response table 11
| contrast effect.size | SE | df | lower.CL | upper.CL | |
|---|---|---|---|---|---|
| AMP – WM | 1.53 | 0.468 | 51.2 | 0.586 | 2.47 |
-
Results are averaged over the levels of: Mig_rate σ used for effect sizes: 0.4649
-
Degrees-of-freedom method: inherited from kenward-roger when re-gridding Confidence level used: 0.95
Additional files
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Accelerated evolution in networked metapopulations of Pseudomonas aeruginosa
eLife 15:e107189.
https://doi.org/10.7554/eLife.107189
