The higBA- Type Toxin-Antitoxin System in IncC Plasmids Is a Mobilizable Ciprofloxacin-Inducible System

Abstract

A putative type II toxin-antitoxin (TA) module almost exclusively associated with conjugative IncC plasmids is homologous to the higBA family of TA systems found in chromosomes and plasmids of several species of bacteria. Despite the clinical significance and strong association with high-profile antimicrobial resistance (AMR) genes, the TA system of IncC plasmids remains largely uncharacterized. In this study, we present evidence that IncC plasmids encode a bona fide HigB-like toxin that strongly inhibits bacterial growth and results in cell elongation in Escherichia coli. IncC HigB toxin acts as a ribosome-dependent endoribonuclease that significantly reduces the transcript abundance of a subset of adenine-rich mRNA transcripts. A glycine residue at amino acid position 64 is highly conserved in HigB toxins from different bacterial species, and its replacement with valine (G64V) abolishes the toxicity and the mRNA cleavage activity of the IncC HigB toxin. The IncC plasmid higBA TA system functions as an effective addiction module that maintains plasmid stability in an antibiotic-free environment. This higBA addiction module is the only TA system that we identified in the IncC backbone and appears essential for the stable maintenance of IncC plasmids. We also observed that exposure to subinhibitory concentrations of ciprofloxacin, a DNA-damaging fluoroquinolone antibiotic, results in elevated higBA expression, which raises interesting questions about its regulatory mechanisms. A better understanding of this higBA-type TA module potentially allows for its subversion as part of an AMR eradication strategy. IMPORTANCE Toxin-antitoxin (TA) systems play vital roles in maintaining plasmids in bacteria. Plasmids with incompatibility group C are large plasmids that disseminate via conjugation and carry high-profile antibiotic resistance genes. We present experimental evidence that IncC plasmids carry a TA system that functions as an effective addiction module and maintains plasmid stability in an antibiotic-free environment. The toxin of IncC plasmids acts as an endoribonuclease that targets a subset of mRNA transcripts. Overexpressing the IncC toxin gene strongly inhibits bacterial growth and results in cell elongation in Escherichia coli hosts. We also identify a conserved amino acid residue in the toxin protein that is essential for its toxicity and show that the expression of this TA system is activated by a DNA-damaging antibiotic, ciprofloxacin. This mobile TA system may contribute to managing bacterial stress associated with DNA-damaging antibiotics.

Keywords: Enterobacteriaceae; antibiotic resistance; plasmids; toxin-antitoxin systems.

FIG 1

Phylogenetic analysis, amino acid sequence alignments, and prediction of tertiary structures for IncC HigB toxin and HigA antitoxin proteins. (A and B) Phylogenetic trees for HigA/RelB antitoxins (A) and HigB/RelE toxins (B) from 7 bacterial species were constructed by the maximum-likelihood method, with a bootstrap value of 500. The percentage of trees in which the associated taxa clustered together is shown next to the branches. The homologs for HigB and HigA found in IncC plasmids are indicated by the black circles. The TA systems from the different bacterial species are represented using the following abbreviations: Vc, Vibrio cholerae; Cb, Coxiella burnetii; Pv, Proteus vulgaris; Ec, E. coli; Ab, Acinetobacter baumannii; Mtb, Mycobacterium tuberculosis; and Cc, Caulobacter crescentus. (C) Amino acid sequence alignments of the putative antitoxin and toxin of the IncC plasmid with the HigA antitoxin and HigB toxin of A. baumannii plasmid pAB120. Identical amino acids are shaded in black. (D) Modeled tertiary structures for the putative antitoxin and toxin of the IncC plasmid and their homologs found in A. baumannii plasmid pAB120. The structural differences between the toxin proteins are marked with black circles.

FIG 2

Overexpression of the IncC higB-like toxin gene strongly inhibits growth, while coinduction of the higA-like antitoxin gene restores the growth rate of the E. coli J53 host. (A) The slow growth of the E. coli J53 host strain in which variants 1 (right panel, orange line) and 2 (right panel, green line) of the higB-like toxin gene were overexpressed in trans compared to their expression in the uninduced reference groups for the same strains (left panel) demonstrates the strong deleterious effects of the putative toxin on host growth. (B) The E. coli J53 host strain in which both the higB-like toxin gene and the higA-like antitoxin gene were overexpressed in trans (green line) grew faster than the strain in which only the higB-like gene was overexpressed (orange line), suggesting that the higA-like gene codes for the cognate antitoxin gene of the HigB-like toxin.

FIG 3

The IncC plasmid TAS functions as an effective addiction module that promotes plasmid stability in an antibiotic-free growth environment. E. coli J53-gfpuv strains that carried the low-copy-number pACYC184 vectors with and without the higBA-like operons from IncC plasmid were propagated in antibiotic-free LB broth for 70 generations. The J53-gfpuv/pACYC184 control lineage (blue line) gradually lost plasmids during the passaging. In contrast, nearly full retention of the pACYC184-higBA_v1 (orange line) and pACYC184-higBA_v2 (green line) plasmids was observed under the same experimental conditions in J53-gfpuv lineages that carried those plasmids.

FIG 4

Amino acid residue G64 in the coding region of the HigB-like toxin is conserved and essential for its toxicity. (A) The inactivating mutation G64V in the coding region of the higB-like toxin gene fully offsets the deleterious effects of overexpressing the toxin gene on J53 host strain growth. (B) Alignment of the amino acid sequences of the HigB toxins. It was found that G64 (highlighted in cyan in the alignment and in the consensus sequence) of IncC HigB is well conserved in the other HigB toxins found in the plasmids and chromosomes of different species. (C) The change in modeled protein tertiary structures of the HigB toxin of the IncC plasmid after the G64V mutation is indicated by the black arrows.

FIG 5

The HigB-like toxin of IncC plasmids depletes the transcript abundance of a subset of translated adenine-rich RNA substrates. (A) Overexpression of the higB toxin gene in the J53 host strain in trans strongly reduces the expression levels of both lpp and ompA relative to those in the uninduced groups for the same J53/pBAD33-higB strain (**, P < 0.01). The nonsynonymous mutation G64V in the higB toxin gene abolishes the downregulation of these genes. (B) Overexpression of the higB toxin gene in the J53 host strain in trans strongly reduces the relative transcript abundance of the mRNA-like domain of the transfer mRNA ssrA compared to that of the untranslated tRNA-like domain of the same gene (***, P < 0.001). Overexpressing higB that contains the G64V mutation has negligible effects on the relative transcript abundances of both domains of ssrA.

FIG 6

Overexpression of the IncC HigB toxin resulted in cell elongation of the E. coli J53-gfpuv host strain. (A and B) Overexpression of the wild-type higB toxin gene resulted in elongation of the J53-gfpuv host cells compared to those of the uninduced control. (C and D) Overexpressing the higB toxin gene that contains the G64V mutation significantly reduced the extent of cell elongation. (E) Unopposed expression of wild-type higB resulted in a significant increase in average cell length (***, P < 0.001). Although higB G64V overexpression also caused a statistically significant increase in cell length (*, P = 0.0317), the extent of cell elongation was greatly reduced. (F) HigB-induced cell elongation shows similarity to phenotypic changes observed in J53-gfpuv treated with ciprofloxacin.

FIG 7

Overexpressing the IncC HigB toxin does not trigger the SOS response. (A) Whereas ciprofloxacin treatment upregulated lexA and recA in the J53/pBAD33 strain, which is a classic gene expression signature for the activation of the SOS response, overexpression of the wild-type higB-like toxin gene did not result in upregulation of lexA and recA relative to their expression in the untreated and uninduced J53/pBAD33 reference group. (B and C) Overexpression of the wild-type higB-like toxin gene resulted in a DAPI nucleus-staining pattern different from that of ciprofloxacin-treated J53-gfpuv. When higB is overexpressed in trans, DAPI staining can be observed throughout the cytoplasm (Fig. 8B). In ciprofloxacin-treated J53-gfpuv, DAPI staining revealed multiple distinct nuclei within the same elongated cells (Fig. 8C), which differs from the staining pattern observed in Fig. 8B.

FIG 8

Expression of the higBA-like operon of IncC plasmids increases in response to ciprofloxacin treatment. (A) J53-gfpuv host strains with two different higBA operon-containing plasmids, pACYC184-higBA_v1 and pEc158ΔMDR-tetA, were treated with a sub-MIC of ciprofloxacin (0.02 μg/ml). An increase in the expression levels of lexA, recA, higB, and higA were observed in the ciprofloxacin-treated groups relative to those in the untreated reference groups for the same strains. (B) The promoter region of higBA of IncC plasmids contains a putative noncanonical SOS box that shares 65% and 60% homology with the canonical SOS box in E. coli and the SOS box of dinG, respectively. The positions of the two highly conserved trimers in the canonical sequence for SOS boxes in E. coli are indicated by the blue lines.