Evolution of a Plasmid Regulatory Circuit Ameliorates Plasmid Fitness Cost

Abstract

Plasmids play a major role in rapid adaptation of bacteria by facilitating horizontal transfer of diverse genes, most notably those conferring antibiotic resistance. While most plasmids that replicate in a broad range of bacteria also persist well in diverse hosts, there are exceptions that are poorly understood. We investigated why a broad-host range plasmid, pBP136, originally found in clinical Bordetella pertussis isolates, quickly became extinct in laboratory Escherichia coli populations. Through experimental evolution we found that inactivation of a previously uncharacterized plasmid gene, upf31, drastically improved plasmid maintenance in E. coli. This gene inactivation resulted in decreased transcription of the global plasmid regulators (korA, korB, and korC) and numerous genes in their regulons. It also caused transcriptional changes in many chromosomal genes primarily related to metabolism. In silico analyses suggested that the change in plasmid transcriptome may be initiated by Upf31 interacting with the plasmid regulator KorB. Expression of upf31 in trans negatively affected persistence of pBP136Δupf31 as well as the closely related archetypal IncP-1β plasmid R751, which is stable in E. coli and natively encodes a truncated upf31 allele. Our results demonstrate that while the upf31 allele in pBP136 might advantageously modulate gene expression in its original host, B. pertussis, it has harmful effects in E. coli. Thus, evolution of a single plasmid gene can change the range of hosts in which that plasmid persists, due to effects on the regulation of plasmid gene transcription.

Figure 1:. Plasmid evolution rapidly improved pBP136Km persistence.

(a) Persistence of the ancestral plasmid pBP136Km (blue) and three populations (red, green, magenta) inoculated with clones that were isolated on Day 5 of the ancestral plasmid assay (see dotted ellipse and arrow). Clones isolated on Day 5 had greatly improved persistence compared to the ancestor. Lighter shades indicate 95% confidence interval. (b) Genomic map of pBP136Km in evolved Day 5 clones revealed deletions in the accessory region (green) between the trb-tra operons (light and dark blue, respectively).

Figure 2:. upf31 inactivation alone explained improved persistence in E. coli hosts.

(a) The poor persistence of plasmid pBP136Km (blue lines) in E. coli strains K-12 MG1655, DH5α, and JM109 (square, circle, and diamond respectively) was markedly improved after an internal 168-bp deletion in plasmid gene upf31 (red lines). Lighter shades indicate 95% confidence interval. (b) A map of pBP136KmΔupf31 showing the 168-bp deletion (red) within the coding sequence of upf31.

Figure 1:. The presence of upf31 encoded on pBP136Km lowered host fitness.

(a) Bacterial growth in batch culture was slower (blue) when carrying pBP136Km with ancestral upf31 compared to the evolved plasmid pBP136KmΔupf31 with inactivated upf31 (red). Lighter shades indicate 95% confidence interval. (Two sample independent T-test of max growth rates, n=10, p-value= 2.35x10⁻³) (b) With the fitness of plasmid-free K-12 normalized to 1 (dashed line), the relative fitness (w) of E. coli with ancestral pBP136Km (in blue) was 0.51, i.e., a ~49% cost (Two sample independent T-test, n=5, p-value=1.70x10⁻⁶). The relative fitness of evolved genotype K-12 (pBP136KmΔupf31) to plasmid-free K-12 was statistically indistinguishable (Two sample independent T-test, n=5, p-value=0.40). Box is interquartile range, green triangle is mean, orange line is median, ***=p ≤ 0.001.

Figure 2:. Upf31 requires the presence of pBP136Km to significantly reduce K-12 growth rate.

(a) In the absence of plasmid pBP136Km there was no significant difference in maximum growth rate between E. coli K-12 expressing control sacB (purple) and upf31 (brown) (Two sample independent T-test, n=10, p-value=0.46). (b) K-12 (pBP136KmΔupf31) with upf31 expressed in trans (orange) showed a lower maximum growth rate than K-12 (pBP136KmΔupf31) with control sacB expressed in trans (green) (Two sample independent T-test, n=10, p-value=4.23x10⁻³). Box is interquartile range, green triangle is mean, orange line is median, n.s. = p > 0.05, **=p ≤ 0.01.

Figure 3:. Expression of global plasmid regulators and their associated operons was increased in the presence of ancestral upf31 compared to the deletion mutant.

(a) Genetic map of pBP136Km with operons and operator sites annotated for global regulators KorA, KorB, KorC, and TrbA. Asterisks denote differentially expressed plasmid genes in the presence of upf31. (b) Table of plasmid-encoded genes with at least 2-fold differential expression (black asterisks in panel a) in the presence of upf31. Note expression of the ancestral upf31 is ~16-fold lower than that of the truncated upf31 knockout.

Figure 4:. AlphaFold2 modeling reveals the potential interaction between pBP136 Upf31 and KorB is mediated by the C-terminal dimerization domain.

(a) AlphaFold2 modelling of the pBP136 Upf31 homodimer (light and dark blue) with the pBP136 KorB homodimer (red and orange). (b) Superimposed AlphaFold2 models of pBP136 Upf31 (dark blue) and R751 Upf31 (magenta) highlighting the additional C-terminal dimerization domain of pBP136 Upf31. (c) Overlay of the AlphaFold2 model of pBP136 Upf31 homodimer (dark blue) with the crystal structure of the RK2 KorA homodimer (PDB: 5CM3, DNA duplex removed). Basic residues in the dimerization domain are highlighted in magenta or yellow. KorA Y84 is highlighted in deep red, Upf31 W198 is highlighted in cyan.