Sexual recombination and increased mutation rate expedite evolution of Escherichia coli in varied fitness landscapes

Abstract

Sexual recombination and mutation rate are theorized to play different roles in adaptive evolution depending on the fitness landscape; however, direct experimental support is limited. Here we examine how these factors affect the rate of adaptation utilizing a "genderless" strain of Escherichia coli capable of continuous in situ sexual recombination. The results show that the populations with increased mutation rate, and capable of sexual recombination, outperform all the other populations. We further characterize two sexual and two asexual populations with increased mutation rate and observe maintenance of beneficial mutations in the sexual populations through mutational sweeps. Furthermore, we experimentally identify the molecular signature of a mating event within the sexual population that combines two beneficial mutations to generate a fitter progeny; this evidence suggests that the recombination event partially alleviates clonal interference. We present additional data suggesting that stochasticity plays an important role in the combinations of mutations observed.

Fig. 1

Theoretical impacts of mutation rate and sexual recombination on population structures. a Mutationally limited fitness landscape, where most mutations fix in a single sweep with very little inter-clonal competition. b Increased mutation rate over (a), more competition and faster evolution observed, but some mutations are lost due to inter-clonal competition. c Adding sexual recombination to (a), due to few available beneficial mutations, no strong influence on evolution is observed. d Both sexual recombination and increase in mutation rate from (a), more rapid evolution is observed and fewer beneficial mutations are lost due to clonal interference

Fig. 2

Fitness improvements during ALE. Black: asexual ara− populations. Blue: asexual ara+ populations. Turquoise: genderless ara− populations. Magenta: genderless ara+ populations. a Chloramphenicol resistance vs. generations in G9+CM evolution. b Trimethoprim resistance vs. generations in G9+TM evolution. c Growth rate vs. generations in G9 evolution. The dotted line in a and b represent the chosen threshold antibiotic resistance level

Fig. 3

Correlating changes in population average fitness with dynamics of mutations in AIG1 and GIG1. a Population fitness data for population AIG1 (asexual ara+). b Population fitness distribution at Time = 1, 2, and 3 for AIG1. c Population-level sequencing data for observed mutations at Time = 1, 2, and 3 for AIG1. d Population fitness of GIG1 (genderless ara+). e Population fitness distribution at Time = 1, 2, and 3 for GIG1. f Population-level sequencing data for observed mutations at Time = 1, 2, and 3 for GIG1. See Supplementary Fig. 2 for correlation between population fitness and average isolate fitness

Fig. 4

Dynamics of select mutations for population GIG1. a Changes in mutation frequencies in the population over time. Error bars are s.d. (n = 3 technical replicates). b Genotype distribution of 100 random isolates in the final population tested with allele specific PCR. Dark gray box with a “+”: mutated genotype. White box with a “−”: wild-type genotype

Fig. 5

The relative fitness improvement over the wild type of each reconstructed mutant. Dark gray box with a “+”: mutated genotype. White box with a “−”: wild-type genotype. Error bars are s.e.m. (n = 36 biological replicates for single mutant and wild type and n = 18 biological replicates for double and triple mutants)

Fig. 6

Epistatic interactions between mutations. Each bar depicts the epistatic contribution to fitness improvement for each reconstructed genotype with multiple mutations, calculated relative to the sum of component single mutations. Dark gray box with a “+”: mutated genotype. White box with a “−”: wild-type genotype. Error bars are s.e.m.