Distribution of fitness effects of cross-species transformation reveals potential for fast adaptive evolution

Abstract

Bacterial transformation, a common mechanism of horizontal gene transfer, can speed up adaptive evolution. How its costs and benefits depend on the growth environment is poorly understood. Here, we characterize the distributions of fitness effects (DFE) of transformation in different conditions and test whether they predict in which condition transformation is beneficial. To determine the DFEs, we generate hybrid libraries between the recipient Bacillus subtilis and different donor species and measure the selection coefficient of each hybrid strain. In complex medium, the donor Bacillus vallismortis confers larger fitness effects than the more closely related donor Bacillus spizizenii. For both donors, the DFEs show strong effect beneficial transfers, indicating potential for fast adaptive evolution. While some transfers of B. vallismortis DNA show pleiotropic effects, various transfers are beneficial only under a single growth condition, indicating that the recipient can benefit from a variety of donor genes to adapt to varying growth conditions. We scrutinize the predictive value of the DFEs by laboratory evolution under different growth conditions and show that the DFEs correctly predict the condition at which transformation confers a benefit. We conclude that transformation has a strong potential for speeding up adaptation to varying environments by profiting from a gene pool shared between closely related species.

Fig. 1. Library preparation.

A The BVAL random segment replacement library was generated by transforming the recipient with genomic DNA of B. vallismortis for 2 h, plating, and picking of single colonies each consisting of a monoclonal hybrid. B The library consists of hybrids between donor and recipient. Random segments carry the donor alleles of core genes, deletions, and insertions from the donor’s accessory genome.

Fig. 2. Distribution of fitness effects of BVAL and BVAL_single libraries.

DFE was determined by competition against gfp reporter strain (Bs175) during exponential phase in complex medium for 4 h. Grey: Control DFE obtained from competition experiments between recipient (Bs166) and gfp reporter strain (grey). A Distribution of selection coefficients s of the BVAL library. Arrows denote outliers (large effect transfers) from the distribution derived at a significance level of α = 0.05 after Bonferroni correction. Numbers denote respective strains of BVAL. B Mean selection coefficients of core distributions excluding outliers <s^s>. Error bars: confidence intervals obtained from bootstrap analysis. C Standard deviation of selection coefficients of core distributions excluding outliers σ^se. Error bars: confidence intervals obtained from bootstrap analysis. D Distribution of selection coefficients s of the BVAL_single library.

Fig. 3. Distribution of fitness effects of BSPIZ library.

DFE was determined by competition against gfp reporter strain (Bs175) during exponential phase in complex medium for 4 h. Grey: Control DFE obtained from competition experiments between recipient (Bs166) and gfp reporter strain (grey). A Distribution of selection coefficients s. Arrows denote outliers (large effect transfers) from the distribution derived at a significance level of α = 0.05 after Bonferroni correction. Numbers denote respective strains of BSPIZ. B Mean selection coefficients of core distributions excluding outliers <s^se>. Error bars: confidence intervals obtained from bootstrap analysis. C Standard deviation of selection coefficients of core distributions excluding outliers σ^se. Error bars: confidence intervals obtained from bootstrap analysis.

Fig. 4. Fitness effects depend on growth conditions.

Scatter plots of selection coefficients of BVAL determined under different growth conditions plotted against the selection coefficients in complex medium at 37 °C °C as shown in Fig. 2A. Red circles: outliers from DFE under different conditions, black diamonds: outliers from DFE in complex medium at 37 °C, grey: strains with small effect transfers. A Complex medium including the lag phase at 37 °C. B Complex medium excluding the lag phase at 42 °C. C Defined medium excluding lag phase at 37 °C. D Defined medium with glycerol as only carbon source excluding lag phase at 37 °C.

Fig. 5. Fitness effects of transformation in evolving populations.

A Sketch of the evolution experiment. 88 batches of the recipient (REC, Bs166) were transformed with donor DNA prior to the evolution experiment, creating a population consisting of different hybrids within a single batch. During laboratory evolution, the OD was monitored, and the batches were diluted repeatedly to maintain exponential growth. After ~450 generations (5 days in CM and 12.5 days in DM), a single clone was picked from each batch, creating a library of evolved strains (BVALevoCM in CM and BVALevoDM in DM). The same procedure was repeated using the untransformed recipient, creating RECevoCM and RECevoDM. The selection coefficients of the libraries were determined by competition against the reporter Bs175. B, E Distribution of selection coefficients in B complex medium and E defined medium. Grey: control (Bs166) without evolution, black: evolved recipient strains, pink: evolved hybrid strains. C, F Mean selection coefficients and D, G standard deviation in C, D complex medium and F, G defined medium. Error bars: confidence intervals obtained from bootstrap analysis. Shaded area: confidence level of evolved recipient data.

Fig. 6. Hotspots of mutations, indels, and orthologous recombination.

A hotspot was defined as a gene that showed sequence changes in at least two of the evolved strains. De novo INDEL: small insertion or deletion of bases within the gene, de novo SNP: single nucleotide polymorphism in the gene that does not arise from orthologous replacement with a donor segment, Recombination: recipient gene replaced by donor gene through orthologous recombination.