The HicA toxin from Burkholderia pseudomallei has a role in persister cell formation

Abstract

TA (toxin-antitoxin) systems are widely distributed amongst bacteria and are associated with the formation of antibiotic tolerant (persister) cells that may have involvement in chronic and recurrent disease. We show that overexpression of the Burkholderia pseudomallei HicA toxin causes growth arrest and increases the number of persister cells tolerant to ciprofloxacin or ceftazidime. Furthermore, our data show that persistence towards ciprofloxacin or ceftazidime can be differentially modulated depending on the level of induction of HicA expression. Deleting the hicAB locus from B. pseudomallei K96243 significantly reduced persister cell frequencies following exposure to ciprofloxacin, but not ceftazidime. The structure of HicA(H24A) was solved by NMR and forms a dsRBD-like (dsRNA-binding domain-like) fold, composed of a triple-stranded β-sheet, with two helices packed against one face. The surface of the protein is highly positively charged indicative of an RNA-binding protein and His24 and Gly22 were functionality important residues. This is the first study demonstrating a role for the HicAB system in bacterial persistence and the first structure of a HicA protein that has been experimentally characterized.

Figure 1. Killing of B. pseudomallei with 100×MIC (200 μg/ml) ciprofloxacin or ceftazidime over 30 h

(a) Stationary phase cultures (D₅₉₀=3.0), (b) exponential phase (D₅₉₀=0.2). The data are shown as the means±S.E.M. for at least two biological repeats.

Figure 2. Culturability of a ΔhicAB mutant of B. pseudomallei containing the plasmid cloned hicA gene

Expression of hicA was induced with 0.2% rhamnose or repressed with 0.2% glucose. At the times indicted, samples were plated on to LB agar and colonies enumerated. The data are shown as the means±S.E.M. for three biological repeats. *P<0.05, following an unpaired Student's t test as determined using the Holm–Sidak method for multiple comparisons.

Figure 3. Persister cell formation in E. coli MG1655 encoding the hicA toxin cloned into pBAD/His and the hicB antitoxin cloned into pME6032

Standardized exponential phase cultures were incubated with arabinose or glucose to induce or repress hicA expression respectively, before treating with antibiotic. (a) hicA was induced with 0.2% arabinose for 3 h and then incubated with 100×MIC ciprofloxacin or ceftazidime and CFU tracked over 30 h. (b) hicA was repressed with 0.2% glucose and induced with a range of arabinose concentrations (0.002–0.2%) for 3 h before treatment with 100×MIC ciprofloxacin for 24 h. (c) As described for (b) except treatment was with 100×MIC ceftazidime. In both (b) and (c), persister frequencies were calculated as CFU numbers post-antibiotic treatment divided by CFU numbers pre-antibiotic treatment. The data are shown as the means±S.E.M. for three biological repeats. ***P<0.001, as determined using one-way ANOVA with Tukey's post test. ND, no detectable colonies.

Figure 4. Effect of deletion or overexpression of hicA on persister cell formation in B. pseudomallei K96243

(a and b) Stationary phase B. pseudomallei K96243 or B. pseudomallei K96243 ΔhicAB treated with 100×MIC (200 μg/ml) ciprofloxacin (a) or ceftazidime (b) respectively for 24 h. The data are shown as the means±S.E.M for at least three biological repeats. *P<0.05, as determined using a Wilcoxon matched-pairs signed rank test. (c and d) Cultures of B. pseudomallei K96243 ΔhicAB with pSCrhaB3 cloned hicA were grown to early exponential phase before repressing or inducing expression of hicA with 0.2% glucose or 0.2% rhamnose for 4 h. Standardized cultures were then exposed to 100×MIC ciprofloxacin (c) or ceftazidime (d) respectively for 24 h. Persister frequencies were calculated as CFU numbers post-antibiotic treatment divided by CFU numbers pre-antibiotic treatment. The data are shown as the means±S.E.M. for at least three biological repeats. ***P<0.001, as determined a Student's t test. ND, no detectable colonies; wt, wild-type.

Figure 5. Graphical representation of the amino acid residues in 75 homologous HicA sequences following sequence alignment

The amino acid sequence of B. pseudomallei K96243 HicA. The asterisks (*) and crosses (×) indicate >90% and >80% amino acid conservation respectively. Residues that were chosen for mutagenesis are underlined. Sequences were aligned using Clustal Omega [47] and the graphic generated using Web logo [48].

Figure 6. Functional analysis of site-directed hicA mutants

(a) Fold change in the number of culturable E. coli MG1655 cells harbouring pBAD/His or pBAD/His encoding wild-type hicA or mutant alleles. Cultures were grown to early exponential phase before inducing expression of hicA or mutants with 0.2% arabinose for 2 h. The data are shown as the means±S.E.M. for three biological repeats. **P<0.01, ***P<0.001 and ****P<0.0001, as determined using one-way ANOVA and Dunnett's post-test, comparing the means with HicA. (b) Western blot of E. coli-expressing H24A (lanes 1 and 2) or G22C (lanes 3 and 4) HicA variants under glucose-repressed (lanes 1 or 3) or arabinose-induced (lane 2 and 4) conditions. The reactive band size is 13 kDa. The molecular mass (M) is indicated in kDa on the left-hand side. (c and d) Persister cell formation by E. coli MG1655 expressing wild-type HicA or HicA mutants after 100×MIC antibiotic treatment. Cultures were grown to early exponential phase before inducing expression of the toxin or toxin mutants. Standardized cultures were incubated with 100×MIC of ciprofloxacin (c) or ceftazidime (d) for 24 h in a 24-well plate at 37°C. After washing, cells were plated on to LB agar. Persister frequency was calculated as CFU counts post antibiotic treatment divided by CFU counts pre-treatment. The data are shown as the means±S.E.M. for three biological repeats. ***P<0.001, as determined using a one-way ANOVA with a Dunett's post-test, comparing the means with HicA. ND, no detectable colonies.

Figure 7. Structure analysis of HicA

(a) Structure of the HicA(H24A) protein with the positions of the five residues that were selected for mutagenesis coloured blue. The N- and C-termini and secondary structure elements are indicated. (b) Ensemble of ten calculated structures of HicA(H24A). Disordered residues (as calculated by PSVS 1.4 [34]) are coloured grey. (c) Structure of HicA(H24A) with side chains of conserved residues shown as sticks (blue and yellow); yellow residues contribute to the hydrophobic core, as do Val³⁶ and Val³⁸ and Phe²⁷, indicated as red sticks. Residues 6–2 have been omitted from all panels, as they are completely unstructured. (d) Charge distribution of the surface of HicA(H24A). Regions of positive, negative and no charge are coloured blue, red and white respectively. (e) Superimposition of HicA(H24A) (red) with TTHA1913 from T. thermophilus (blue; PDB code 1WHZ).