Shared mechanisms of enhanced plasmid maintenance and antibiotic tolerance mediated by the VapBC toxin:antitoxin system

Abstract

Toxin:antitoxin (TA) systems are widespread in bacteria and were first identified as plasmid addiction systems that kill bacteria lacking a TA-encoding plasmid following cell division. TA systems have also been implicated in bacterial persistence and antibiotic tolerance, which can be precursors of antibiotic resistance. Here, we identified a clinical isolate of Shigella sonnei (CS14) with a remarkably stable pINV virulence plasmid; pINV is usually frequently lost from S. sonnei, but plasmid loss was not detected from CS14. We found that the plasmid in CS14 is stabilized by a single nucleotide polymorphism (SNP) in its vapBC TA system. VapBC TA systems are the most common Type II TA system in bacteria, and consist of a VapB antitoxin and VapC PIN domain-containing toxin. The plasmid stabilizing SNP leads to a Q12L substitution in the DNA-binding domain of VapB, which reduces VapBC binding to its own promoter, impairing vapBC autorepression. However, VapB^L12C mediates high-level plasmid stabilization because VapB^L12 is more prone to degradation by Lon than wild-type VapB; this liberates VapC to efficiently kill bacteria that no longer contain a plasmid. Of note, mutations that confer tolerance to antibiotics in Escherichia coli also map to the DNA-binding domain of VapBC encoded by the chromosomally integrated F plasmid. We demonstrate that the tolerance mutations also enhance plasmid stabilization by the same mechanism as VapB^L12. Our findings highlight the links between plasmid maintenance and antibiotic tolerance, both of which can promote the development of antimicrobial resistance.

Importance: Our work addresses two processes, the maintenance of plasmids and antibiotic tolerance; both contribute to the development of antimicrobial resistance in bacteria that cause human disease. Here, we found a single nucleotide change in the vapBC toxin:antitoxin system that stabilizes the large virulence plasmid of Shigella sonnei. The mutation is in the vapB antitoxin gene and makes the antitoxin more likely to be degraded, releasing the VapC toxin to efficiently kill cells without the plasmid (and thus unable to produce more antitoxin as an antidote). We found that vapBC mutations in E. coli that lead to antibiotic tolerance (a precursor to resistance) also operate by the same mechanism (i.e., generating VapB that is prone to cleavage); free VapC during tolerance will arrest bacterial growth and prevent susceptibility to antibiotics. This work shows the mechanistic links between plasmid maintenance and tolerance, and has applications in biotech and in the design and evaluation of vaccines against shigellosis.

Keywords: Shigella; TA systems; VapBC; antibiotic tolerance; plasmid stability.

Fig 1

Stabilization of S. sonnei CS14 pINV is mediated by a single amino acid substitution in VapB. (A) Emergence of avirulent CR^- colonies at 21°C during liquid growth of S. sonnei 53G and S. sonnei CS14 (n = 2; error bars, SD). (B) The effect of vapB^Q12 and vapB^L12 in S. sonnei 53G and S. sonnei CS14 on pINV loss. pSTAB2 loss in S. flexneri M90T and S. sonnei 53G lacking pINV at 37°C (C) or 21°C (D). vapBC alleles are indicated. Results are from nine colonies from ≥3 independent experiments. Statistical significance determined using Kruskal–Wallis and two-way ANOVA: ****, P < 0.0001; ***, P < 0.001; **, P < 0.01; ns, P > 0.05.

Fig 2

VapB^L12C has impaired binding to its promoter. (A) Schematic of the autorepressed vapBC bicistronic operon, including the promoter containing two operator sites, OS1 and OS2 (each a pair of dotted arrows), that are bound by VapBC and overlap the −10 and −35 sequences of the promoter. (B) Structure of S. sonnei 53G VapBC hetero-octamer (VapC, purple; VapB, gold; PDB ID: 6SD6). The VapB dimers (VapB1 with VapB2, and VapB3 with VapB4) are indicated, with a close-up of the hydrogen bonds between the side chain of residue VapB1^Q12 and the backbone of VapB2^L22. (C) EMSA demonstrating interactions between VapB^Q12C or VapB^L12C and the native S. sonnei 53G vapBC promoter (227 bp, pVapBC), or (D) the 150 bp constitutive promoter, J23101; a 68 bp intergenic sequence was included as a control. (E) The effect of autoregulation on pSTAB2 loss in S. flexneri M90T or S. sonnei 53G lacking pINV. Loss of pSTAB2 expressing different vapBC alleles (indicated) under the control of the native vapBC promoter (red circles) or the constitutive promoter J23101 (blue circles); n = 9 colonies from ≥3 independent experiments. Statistical significance calculated using one-way ANOVA and Kruskal–Wallis. Plasmid loss with the native vs constitutive promoter for all vapBC alleles was not significant.

Fig 3

Enhanced degradation of highly efficient VapBCs by Lon protease. (A) Stability of pSTAB2 expressing vapBC alleles as indicated in S. flexneri M90T or S. flexneri M90T Δlon (n = 9 colonies from ≥3 independent experiments). Statistical significance calculated by two-way ANOVA of S. flexneri M90T vs S. flexneri M90TΔlon carrying the same vector: **, P = 0.0008, ****, P < 0.0001. (B) In vitro proteolysis performed at 37°C with hetero-octameric VapBC as indicated with S. sonnei in the presence of ATP, Lon, polyphosphate (Pp), vapBC promoter DNA (pDNA), or control DNA (conDNA). Representative result from three independent experiments.

Fig 4

Lon cleavage sites in VapB. (A) Lon cleaves VapB next to residues F51 and F60 located in the C-terminal arm of VapB (backbone with Phe side chains) that inhibits VapC (purple, space filling). Lon cleavage sites on the C-terminal side of F51 and F60 are indicated by dashed lines. (B) The effect of introducing point mutations F6A, F51A, and F60A into VapB^Q12C or VapB^L12C on pSTAB2 loss in S. flexneri. Each strain was tested in three independent experiments.

Fig 5

Antibiotic tolerance mutations in E. coli vapB affect post-segregational killing. (A) Loss from S. sonnei of pSTAB2 with different vapBC alleles expressed under the native vapBC promoter or a constitutive promoter (J23101). Each strain was tested in three independent experiments (except VapB^{T3N+A13P+L16R}C with a constitutive promoter, which was tested once). Plasmid loss is reduced with all tolerance mutations compared with VapB^Q12C (VapB^WTC) with the native vs constitutive promoter was not significant (ns, P > 0.05) for all vapBC alleles. (B) pSTAB2 loss from S. flexneri M90T with (gray circles) or without lon (green circles). VapB^T3NC, VapB^T3N+L7PC, VapB^V5EC, and VapB^L7PC reduce plasmid loss in a Lon-dependent manner, while the absence of Lon does not have a significant impact for VapB^A13PC, VapB^L16RC, VapB^V20GC, and VapB^{T3N+A13P+L16R}C. Each strain was tested in three independent experiments, and statistical significance was calculated using Kruskal–Wallis: ****, P < 0.0001; ***, P = 0.0004; **, P = 0.0092.

Fig 6

Impact of vapB tolerance mutations on the VapBC structure. (A) The structure of hetero-octameric VapBC encoded on the E. coli F plasmid (top and side view) and (B) S. sonnei pINV (top view; PDB: 6SD6) (10). The hetero-octameric structure and overall folds are conserved between this and the structure of F plasmid VapBC. (Cα rmsd 4.2–4.7 Å over 132 atoms for each VapC; Cα rmsd of 103–107 Å over 68 atoms for each VapB; Cα rmsd 6.2 Å over 264 atoms for the half complex.) (C) Close-up view of the VapB DNA-binding domain (box in panel A). The DNA-binding domain of VapB has been aligned on top of DNA (blue) from the VapBC:DNA structure (PDB: 6IFM) (34). Structures solved with residues altered by tolerance mutations are shown in magenta.