Experimental determination of evolutionary barriers to horizontal gene transfer

Abstract

Background: Horizontal gene transfer, the acquisition of genes across species boundaries, is a major source of novel phenotypes that enables microbes to rapidly adapt to new environments. How the transferred gene alters the growth - fitness - of the new host affects the success of the horizontal gene transfer event and how rapidly the gene spreads in the population. Several selective barriers - factors that impact the fitness effect of the transferred gene - have been suggested to impede the likelihood of horizontal transmission, however experimental evidence is scarce. The objective of this study was to determine the fitness effects of orthologous genes transferred from Salmonella enterica serovar Typhimurium to Escherichia coli to identify the selective barriers using highly precise experimental measurements.

Results: We found that most gene transfers result in strong fitness costs. Previously identified evolutionary barriers - gene function and the number of protein-protein interactions - did not predict the fitness effects of transferred genes. In contrast, dosage sensitivity, gene length, and the intrinsic protein disorder significantly impact the likelihood of a successful horizontal transfer.

Conclusion: While computational approaches have been successful in describing long-term barriers to horizontal gene transfer, our experimental results identified previously underappreciated barriers that determine the fitness effects of newly transferred genes, and hence their short-term eco-evolutionary dynamics.

Keywords: Distribution of fitness effects; Dosage sensitivity; Evolutionary barriers; Gene length; Horizontal gene transfer.

Fig. 1

Schematic representation of the experimental design. a. Chromosomal modifications in the recipient strain, E. coli MG1655 att-λ::(tetR-Sp^R) att-p21::(CFP/YFP-Kn^R), for the transferred genes. Att-λ and att-p21correspond to the attachment sites of the phage λ and p21, respectively. TetR is the repressor protein controlling the expression of the transferred genes. Sp^R and Kn^R are the resistance genes for spectinomycin and kanamycin, respectively. P_N25 and P_λR are the constitutive promoters. See Methods section for details. b. Depiction of the competition assay. Blue cells with CFP represent the ‘mutant’ strain that carries the pZS*-HGT plasmid containing the introduced gene, whereas yellow cells with YFP represent the ‘wild type’ strain that carries the empty pZS*-HGT plasmid without the introduced gene. The plot illustrates an example where the fitness effect of the gene is deleterious, resulting in a decrease in the frequency of blue cells over time. Numbers inside the segments represents the frequency of the type of the cell with same color

Fig. 2

DFE of 44 newly transferred S. Typhimurium orthologs expressed in E. coli. On the x-axis, genes are sorted by their fitness effects (s). Error bars indicate the 95% CI of the selection coefficients from 32 replicate measurements for each gene. Empty circles represent the 5 genes with effects not significantly different from zero. Embedded plot gives the histogram representation of the same data

Fig. 3

Analyses of the selective barriers on HGT. The mean selective effects of 44 transferred S. Typhimurium genes expressed in E. coli background: a. divided into informational and operational genes based on functional categories; b. plotted against predicted number of PPIs; c. absolute deviation in %GC between orthologs; d. absolute deviation in codon usage between orthologs; e. gene length; f. number of disordered regions in the amino-acid sequences. All 5 factors given in a. to e. are used as explanatory variables in a multiple regression model (single p-values at the bottom right of each panel). e. We repeated the simple linear regression for the gene length as it was the only factor with a significant effect in the multiple regression. Black lines in e and f show the best fit for the simple linear regression between the two variables; gray dashed lines show the zero line

Fig. 4

Dosage sensitivity. Selection coefficients of newly transferred and deleterious genes from S. Typhimurium (white bars) expressed in E. coli background overlaid with those of their orthologs from E. coli (grey bars) expressed in E. coli background. On the x-axis genes are sorted according to their selection coefficients for the transfer of E. coli orthologs. Genes marked with asterisks show dosage sensitivity, as the fitness cost of the additional copies of an E. coli gene is the same or greater than the fitness cost of its S. Typhimurium ortholog

Fig. 5

Selection coefficients of 36 newly transferred and deleterious S. Typhimurium genes. Genes are plotted against a. gene length in nucleotides; b. number of disordered regions in their amino-acid sequence. Black data points represent the dosage sensitive genes, and grey data points with black frames represent the dosage insensitive genes. Lines are the simple linear regressions between the two variables with corresponding colors, gray dashed lines show the zero line