Experimental Evolution of Escherichia coli Harboring an Ancient Translation Protein

Abstract

The ability to design synthetic genes and engineer biological systems at the genome scale opens new means by which to characterize phenotypic states and the responses of biological systems to perturbations. One emerging method involves inserting artificial genes into bacterial genomes and examining how the genome and its new genes adapt to each other. Here we report the development and implementation of a modified approach to this method, in which phylogenetically inferred genes are inserted into a microbial genome, and laboratory evolution is then used to examine the adaptive potential of the resulting hybrid genome. Specifically, we engineered an approximately 700-million-year-old inferred ancestral variant of tufB, an essential gene encoding elongation factor Tu, and inserted it in a modern Escherichia coli genome in place of the native tufB gene. While the ancient homolog was not lethal to the cell, it did cause a twofold decrease in organismal fitness, mainly due to reduced protein dosage. We subsequently evolved replicate hybrid bacterial populations for 2000 generations in the laboratory and examined the adaptive response via fitness assays, whole genome sequencing, proteomics, and biochemical assays. Hybrid lineages exhibit a general adaptive strategy in which the fitness cost of the ancient gene was ameliorated in part by upregulation of protein production. Our results suggest that an ancient-modern recombinant method may pave the way for the synthesis of organisms that exhibit ancient phenotypes, and that laboratory evolution of these organisms may prove useful in elucidating insights into historical adaptive processes.

Fig. 1

Sequence and structure analysis of EF-Tu. a Alignment of amino acid sequences of modern (AAC76364) and ancient EF-Tu from E. coli. Amino acid sequences were obtained from the NCBI database and aligned using Clustal Omega (Sievers et al. 2011). Figures were generated with the ESPript 3.0 server (Robert and Gouet 2014). Strictly conserved residues are shown in white. Partially conserved amino acids are boxed. Residues conserved in most of the members of one family are in red font. b The ribbon illustration of the EF-Tu adopted from the cryo-EM structure of E. coli ribosome–EF-Tu complex (PDB 5AFI). Domains I, II, and III are colored in slate, cyan, and wheat, respectively. The residues different in the ancient variant are shown with side chain (in red) and labeled accordingly. c Structure of EF-Tu–tRNA bound to the 70S ribosome in gray (PDB 5AFI) (Fischer et al. 2015) showing that the residues E250, Q252, S254, and I282 in domain II of EF-Tu were involved in the interaction with 70S ribosome. (Color figure online)

Fig. 2

a Fitness values of E. coli populations relative to the ancestral strain during adaptive evolution. Replacement of the endogenous EF-Tu gene with the reconstructed ancient EF-Tu allele significantly reduces the fitness of the ancient–modern hybrid relative to the original strain (dotted line). Hybrid population mean fitness rapidly improved during experimental evolution in minimal glucose medium (solid line). E. coli ΔtufA represents the bacteria that contain a single tufB gene. Fitness change in E. coli ΔtufA is shown as dashed line. Error bars show 95% confidence interval among six replicate populations for each system. b Total number of mutations over time. Total number of genomic mutations accumulated in laboratory evolved hybrid and non-hybrid populations over time. Each color represents a sequenced hybrid genome. Black lines represent the average total number of genomic changes relative to the ancestor in each sampled hybrid (solid) and non-hybrid (dashed) lineages over time. (Color figure online)

Fig. 3

Analysis of the mutations accumulated in the cis-regulatory region thrT/tufB. a The thrT/tufB promoter region in which five of six evolved hybrid populations were found to have accumulated mutations. b Allelic frequency of the mutations located in ancient EF-Tu gene’s promoter region per generation per population during laboratory evolution. c Relative abundance of ancient EF-Tu protein among evolved hybrid strains using the peak area quantification from MS proteomics data. Error bars obtained using ANOVA/t test. d Growth rates of an isogenic strain of E. coli REL606 lacking the tufA gene, as well as the unevolved ancient–modern hybrid E. coli, were evaluated in the presence of Anhydrotetracycline (ATC) inducer. Strains were induced with 500 mg/mL ATC in rich growth media for 3–4 h to achieve proper induction. Cells from these fresh induced cultures were inoculated in 96-well plates and grown at 37 °C with a starting OD₆₀₀ of ∼0.06 under respective ATC concentration. Doubling times were determined by fitting the exponential growth curves with an exponential function

Fig. 4

Sequence and structure analysis of NusA. a Alignment of amino acid sequences of modern and ancient NusA. Amino acid sequences were obtained from the NCBI database (AAC76203), the alignment and figure display were performed the same way as in Fig. 1. b Predicted model of E. coli NusA based on the crystal structure of NusA from Planctomyces limnophilus (4MTN) by the use of SWISS-MODEL (Biasini et al. 2014). The structure of the C-terminal domain demonstrates the deletion of nine residues (colored slate). c Structure prediction of the mutant NusA protein, harboring nine amino acid deletion in its C-terminal domain (CTD). d Wild-type NusA (colored wheat) and the mutant NusA with deletion of nine amino acids (colored salmon) alignment demonstrates the confirmational changes. e Fitness change after deletion of nusA gene from the ancestral and evolved bacterial genome. (Left) Bacterial constructs with isogenic nusA knockouts are generated and competed against the native E. coli bacteria for fitness measurement. (Right) The interactions between the native EF-Tu, ancient EF-Tu, and nusA variants are measured via in vitro isothermal calorimetry binding assays. (Color figure online)