Francisella tularensis: FupA mutation contributes to fluoroquinolone resistance by increasing vesicle secretion and biofilm formation

Abstract

Francisella tularensis is the causative agent in tularemia for which the high prevalence of treatment failure and relapse is a major concern. Directed-evolution experiments revealed that acquisition of fluoroquinolone (FQ) resistance was linked to factors in addition to mutations in DNA gyrase. Here, using F. tularensis live vaccine strain (LVS) as a model, we demonstrated that FupA/B (Fer-Utilization Protein) expression is linked to FQ susceptibility, and that the virulent strain F. tularensis subsp. tularensis SCHU S4 deleted for the homologous FupA protein exhibited even higher FQ resistance. In addition to an increased FQ minimal inhibitory concentration, LVSΔfupA/B displayed tolerance toward bactericidal compounds including ciprofloxacin and gentamicin. Interestingly, the FupA/B deletion was found to promote increased secretion of outer membrane vesicles (OMVs). Mass spectrometry-based quantitative proteomic characterization of vesicles from LVS and LVS∆fupA/B identified 801 proteins, including a subset of 23 proteins exhibiting differential abundance between both strains which may therefore contribute to the reduced antibiotic susceptibility of the FupA/B-deleted strain. We also demonstrated that OMVs are key structural elements of LVSΔfupA/B biofilms providing protection against FQ. These results provide a new basis for understanding and tackling antibiotic resistance and/or persistence of Francisella and other pathogenic members of the Thiotrichales class.

Keywords: OMVs; antibiotics; biofilms; fluoroquinolones.

Figure 1.

Maximum likelihood tree of DUF3573-containing proteins in Thiotrichales (62 sequences, 303 amino acid positions used). The 62 sequences were selected among the 1,214 homologs identified in the 384 Thiotrichales proteomes available at the NCBI, by keeping sequences from one representative strain per species or subspecies in the case of Francisella. The scale bar represents the average number of substitutions per site. Numbers at nodes correspond to ultrafast approximated bootstrap values. For clarity, values lower than 75% are not shown. The five protein subfamilies corresponding to FupA, FupB, FslE, FmvA, and FmvB are shown in green, blue, orange, pink, and yellow. The five sequences from Francisella tularensis subsp. tularensis SCHU S4 and the four sequences from F. tularensis subsp. holarctica LVS are shown in purple and red, respectively. Noteworthy, in the latter as in other LVS strains, FupA and FupB are fused. Regions corresponding to FupA and FupB in the LVS strain were analyzed separately to infer this tree. As expected, they group within subfamilies corresponding to FupA and FupB, respectively.

Figure 2.

Restoration of ciprofloxacin resistance phenotype in fupA/B-complemented mutants. (a) Left panel: FupA/B expression was evaluated by western-blot and using anti-IglC as positive control on whole lysates of LVS mutants resulting from directed-evolution experiments [4] showing that protein expression in P12V3 is restored upon gene trans-complementation using the plasmid pMP828. Right panel: The same strains were assayed for ciprofloxacin susceptibility, measured as MIC values. (b) Same experiments performed on LVS and LVSΔfupA/B strains. (c) Ciprofloxacin-susceptibility of F. tularensis SCHU S4 as well as ΔfupA, ΔfupB and the double ΔfupA/ΔfupB deletion mutants. MIC were determined from triplicate cultures from three different experiments. **** P < 0.0001.

Figure 3.

Killing curves of LVS and LVSΔfupA/B. Exponential growth phase LVS (black circles) or LVSΔfupA/B (white circles) were exposed to 25x the MIC of (a) ciprofloxacin (0.5 and 1.6 mg/L respectively), (b) gentamicin (6.25 mg/L) or (c) doxycycline (6.25 mg/L) and the CFU were determined by plating. This graph is representative of 3 independent experiments.

Figure 4.

Quantitative and qualitative analysis of OMVs in F. tularensis LVS. (a) Nanosight enumeration of OMVs purified from F. tularensis LVS (black columns), ΔfupA/B mutant (dotted columns) and ΔfupA/B mutant complemented with fupA/B (hatched columns). Bacteria grown in MHM at 37°C under shaking were collected either at the exponential or stationary growth phase, as indicated. Because no significant difference was shown between bacteria in exponential phase of growth, the complemented mutant was not considered under such conditions. Data are expressed as mean ± SEM of 4 different experiments. ****P < 0.0001. (b) Nanosight sizing of OMVs purified from stationary growth phase bacteria. (c) LPS and protein profiles of whole bacterial lysates and of purified OMVs. Left and middle panels: whole bacterial extracts (5 µg) and OMVs samples (1.5 µg) from (1) LVS, (2) ΔfupA/B or (3) ΔfupA/B + fupA/B strains were separated on 4-20% gradient gel before transfer and immunostaining with the anti-LPS antibody. Right panel: Silver staining of OMV proteins from (1) LVS, (2) ΔfupA/B (1.5 µg) separated on 12% SDS-PAGE.

Figure 5.

Proteomic characterization of OMV proteomes from wild-type F. tularensis LVS and the ΔfupA/B mutant. (a) Localization of proteins from total proteome (2190 proteins) and purified OMVs (801 identified proteins) as predicted by PSORTb. Over-representation of OMVs proteins compared to total proteome in each category was tested using Fisher’s exact test (**P < 0.01, ***P < 0.001, ****P < 0.0001). (b) Predicted localization of proteins identified in OMVs from wild-type LVS strain and ΔfupA/B mutant. Each category is represented by the summed abundances (iBAQ values) of the contributing proteins. (c) Volcano plot representing the -log₁₀(P-value) plotted against the log₂(fold change) for each quantified OMV protein. One protein from the mutant strain (blue dot) and twenty-two proteins from the wild-type strain (green dots) were found to be statistically differentially enriched.

Figure 6.

Biofilm formation by F. tularensis LVS. (a) Crystal violet staining was performed to assess the biofilm formation by F. tularensis LVS (black columns), LVSΔfupA/B (dotted columns) and LVSΔfupA/B complemented with fupA/B (hatched columns) grown for 72 h in 96 well plates at 37°C without shaking. (b) The biofilm formation was evaluated after 72 h incubation of F. tularensis LVS supplemented with different amount of freshly purified LVSΔfupA/B OMVs numerated using the Nanosight Instrument and added in wells containing 2 × 10⁸ bacteria/ml at a final ratio bacteria:vesicles of 1:1–1:20. Data are expressed as mean ± SEM of 3 different experiments. ****P < 0.0001, ***P < 0.001; *P < 0.05. (c) Biofilm formation was also examined using confocal laser-scanning microscopy from F. tularensis LVS and (d, e) LVSΔfupA/B stained with FM®1-43X dye. Bar scales 10 µm (c, d) and 2 µm (e). Arrows indicate the presence of OMVs.

Figure 7.

Cofocal laser-scanning microscopy images of F. tularensis LVSΔfupA/B. Biofilms-embedded bacteria stained with (a) FM®1-43FX, (b) ConA-FITC, (c) merge. (d-f) 3D reconstruction of a, b and c. Each channel of the raw data (i. e, xyz files) were deconvoluted using the “iterative Deconvolve 3D” plugin (ImageJ software). The UCSF ChimeraX software [39] was used for 3D reconstruction of processed images.

Figure 8.

Comparative bactericidal effect of ciprofloxacin against planktonic and biofilm populations of F. tularensis LVSΔfupA/B. The metabolic activity of planktonic (a) or biofilm bacteria (c) exposed for 24 h to increasing concentrations of ciprofloxacin was assessed by resazurin-reduction assay. The bacterial replication was monitored by OD_600nm value for planktonic cells (b) and by CFU counting for biofilm (d). Data represent the mean ± SEM of three different experiments performed in triplicate. *P < 0.05.