P. azelaica HBP1 was used as wild-type strain and is able to completely mineralize 2-HBP (Kohler et al., 1988). P. azelaica strains were cultured at 30°C in liquid Pseudomonas Minimal Medium (PMM) (Gerhardt et al., 1981), or M9 minimal medium (Sambrook and Russell, 2001) amended with 5 mM sodium succinate and/or 2-HBP in different concentrations (Sigma-Aldrich, Switzerland). Tn5 mutants of strain HBP1 were cultured in the presence of 50 μg/mL of kanamycin (Km). To maintain the 2-HBP reporter plasmid pME6012_hbpR_gfp (see below) 10 μg/mL tetracycline (Tc) was added to the culture medium. As solid medium for P. azelaica strains we used Nutrient Agar (NA) or PMM agar (15 g/L, DIFCO BactoAgar; Brunschwig DB Difco, Basel, Switzerland) with either 5 mM sodium succinate or 2.7 mM 2-HBP, and with or without Km or Tc as indicated above. Escherichia coli DH5α (Sambrook and Russell, 2001) and S17-1/λpir (Simon et al., 1993) were cultured at 37°C in liquid Luria Bertani (LB) Broth or on LB agar; if required, under inclusion of the appropriate antibiotics.

Tn5 insertion mutants were generated by biparental matings between E. coli BW20767/pRL27 (Larsen et al., 2002) and P. azelaica HBP1 with selection on NA containing 50 μg/ml Km and 10 μg/ml chloramphenicol, to counterselect against the E. coli donor. About 10,000 mutant colonies, produced in eight independent mutagenesis experiments, were picked individually and tested by replica plating for growth on M9 minimal medium with 5 mM succinate vs. 2.7 mM 2-HBP as sole carbon source. Mutants which failed to develop colonies on 2-HBP after 3 days but were growing on succinate, were investigated further. To determine the transposon insertion site in these mutants, chromosomal DNA was extracted as described (Gamper et al., 1992), subjected to restriction with SacII, self-ligated and introduced into E. coli S17-1/λpir by electroporation, with selection for Km resistance. After isolation of the plasmid, the Tn5 insertion sites were determined by Sanger sequencing using the transposon-specific primer tnpRL17-1 (5′-AACAAGCCAGGGATGTAACG-3′) (Larsen et al., 2002). Sequences were mapped onto a draft P. azelaica genome (van der Meer, unpublished) using BLASTN and further interpreted according to database homologies of the predicted gene functions into which the insertions had occurred.

Total numbers of cells in culture were counted with a BD LSR Fortessa (BD Biosciences, Erembodegem, Belgium) equipped with 3-lasers (Blue = 488 nm, 50 mW; Red = 640 nm, 40 mW; UV = 355, 20 mW). Cells were hereto fixed with 4 g/L sodium azide (Sigma-Aldrich, Switzerland) for 1 h at 4°C and stained with 100-fold diluted SYBRGreen I solution (Invitrogen, Switzerland) for 15 min in the dark at room temperature. A fixed sample volume (200 μL) was aspirated by a High Throughput Screening device (BD Biosciences) at a flow rate of 1 μL/s. Culture samples exceeding 1000 events/s were further diluted in filtered PMM. Data were acquired in the FITC-channel and processed using the BD software DIVA (version 6.2, BD Biosciences). Side and forward scatter were set to 250 to reduce background noise.

Cell division rates were calculated from the increase of cell numbers in culture over time measured by FC, and converted to maximum specific growth rates (μ) by multiplying by ln2 (Czechowska et al., 2013).

Membrane transport was studied using ethidium bromide (EB) as proxy. HBP1 wild-type cells or mutants were grown until mid exponential phase on 5 mM succinate, then sampled and stained with SYTO9 (5 μM, Invitrogen) plus EB (10 μM) for 10 min in the dark, either in absence or in the presence of potential membrane uncouplers. SYTO9 and EB fluorescence in individual cells were then immediately measured using FC, as above. As membrane uncouplers we used carbonyl cyanide-3-chlorophenyl-hydrazone (CCCP, at 15 μM), sodium azide (4 g/L), tetraphenylphosphonium (TPP, at 4 μM), valinomycin (at 1 μM), and ethylenediaminetetraacetic acid (EDTA, at 10 mM). All other compounds were obtained from Sigma-Aldrich (Switzerland). For EB efflux assays cells were grown until mid exponential phase on 5 mM succinate, stained with 10 μM EB for 10 min in the dark, centrifuged for 1 min at 13,000 g and resuspended in PMM without EB. The intensity of EB fluorescence was measured by FC after 0, 10, 20, 30, and 40 min.

To assess 2-HBP toxicity we exposed P. azelaica HBP1 and mutant cells sampled from exponentially growing culture on PMM with succinate, for 1 or 3 h to 100 μM, 500 μM, or 1 mM 2-HBP. Effects were compared to cells taken from the same culture, but exposed for 1 and 3 h at 30°C to CCCP, sodium azide and valinomycin (concentrations above). Samples were then stained immediately with the BacLight Live/Dead Kit according to the protocol provided by the manufacturer (Invitrogen). Similarly, the cellular membrane potential was measured by staining with 50 nM final concentration 3,3′-diethyloxacarbocyanine iodide (DiOC₂(3)), according to technical instructions of the supplier (Invitrogen). Fluorescence intensities were measured on a FACS Calibur flow cytometer (BD Biosciences) using FL1 (525 ± 15 nm) and FL3 (>650 nm) channels.

In order to test influx of 2-HBP into P. azelaica HBP1 and mutants we used a reporter plasmid that was previously developed for E. coli (Beggah et al., 2008). GFP production is here brought under control of the HbpR-and 2-HBP-dependent P_C-promoter (Jaspers et al., 2001). The vector of this reporter plasmid (pHBP269A0) was changed to that of pME6012 (Heeb et al., 2000), to make it better compatible with P. azelaica. Hereto, the hbpR-P_C::gfp gene fragment was recovered from pHBP269A0 by cutting with NheI and SalI, and ligated with pME6012 cut with NheI and XhoI. After ligation the plasmid was introduced into E. coli DH5α by heat shock transformation. Positive clones containing the new plasmid pME6012_hbpR_gfp were selected on LB agar containing 10 μg/mL Tc. The plasmid was isolated and verified by restriction enzyme digestion. Induction of GFP in E. coli (pME6012_hbpR_gfp) was verified by exposing the cells to 2.5 mM 2-HBP for 2 h (not shown). Purified pME6012_hbpR_gfp was then introduced by electroporation into P. azelaica HBP1 and the three selected Km-insertion mutants (mexA::Km, mexB::Km, oprM::Km).

Uptake of 2-HBP by strains HBP1, mexA::Km, mexB::Km, or oprM::Km was followed by measuring the GFP signal generated from the reporter plasmid pME6012_hbpR_gfp in the presence or absence of membrane uncouplers. Cells were hereto grown on PMM with 5 mM succinate until mid-exponential phase. Culture samples of 250 μL were transferred into wells of a 96-well black microtiter plate (Greiner Bio-One GmbH, Germany). Cells were incubated at 30°C with varying concentrations of 2-HBP (0, 2.4, 24, and 240 μM), upon which GFP fluorescence of the cultures was followed for 7 h. GFP fluorescence (at 520 nm) and culture turbidity (at 600 nm) were measured at regular time intervals in a FLUOstar Omega fluorimeter (BMG LABTECH SARL, France). GFP fluorescence values were normalized to the culture turbidity.

Replica plating of some 10,000 P. azelaica transposon mutants from eight independently created mutant libraries resulted in ~1% of colonies growing on succinate but unable to grow with 2.7 mM (460 mg/L) 2-HBP as a sole carbon source. The transposon insertion and flanking DNA sequences of 95 mutants unable to grow on 2-HBP were cloned and the insertion sites determined by sequence analysis and BLASTN comparison (Table S1, Data File S1). We found that 28 insertions had occurred at different positions in a gene cluster homologous to mexAB-oprM (9 insertions occurred in mexA, 12 in mexB, and 7 in oprM), specifying a multidrug resistance efflux pump (Figure 1). Percent amino acid similarities across the whole length between the P. azelaica and P. aeruginosa MexAB-OprM systems approached 87% (MexA), 96% (MexB), 89% (OprM), and 69% (MexR). Interestingly, no transposon insertions that caused failure to grow on 2-HBP were recovered in mexR (Figure 1), a presumed regulatory gene for mexAB-oprM expression. A further 6 had inserted into a cluster of six genes, encoding a putative ABC transport system providing resistance to organic compounds with over 90% amino acid similarity to Cluster of Orthologous Groups (COG) PA4452-4455 (Figure 1B). Insertions from the remaining mutants had occurred throughout the P. azelaica genome but in no further cases were found multiple times in the same gene. Intriguingly, no transposon mutants were recovered with insertions in the known hbpR-CAD genes for 2-HBP metabolism (Jaspers et al., 2001).

In order to better understand the effects of 2-HBP on the growth of wild-type HBP1 and three transposon mutants, one in each of the mexAB or oprM genes, we re-examined culture growth on 2-HBP in low concentrations between 0 and 100 μM (Figure 2). Population growth at these low concentrations was measured by FC counting of fixed and stained cells. Interestingly, although the three mutants did not grow on 2.7 mM 2-HBP, they did grow to some extent on 2-HBP concentrations up to 25 μM (Figures 2B–D), indicating they were unable to cope with toxicity exerted by high 2-HBP concentrations. Calculated maximum specific growth rates even at low 2-HBP concentrations were lower for all transposon mutants compared to HBP1 wild-type, with HBP1 mexB::Km and oprM::Km being the most severely affected, followed by HBP1 mexA::Km (Figure 2E).

In order to understand the possible mechanism of toxicity of 2-HBP to strain HBP1 we exposed cells from mid-exponential phase on sodium succinate to 100 μM, 500 μM, and 1 mM of 2-HBP for 1 and 3 h. Cells exposed to 100 or 500 μM stained after 1 and 3 h with SYTO9 plus EB or SYTO plus propidium iodide (PI; part of the Live/Dead stain) did not show any increase in EB fluorescence or in the proportion of injured/dead cells, compared to non-exposed cells (Figures 3, S1). In contrast, 1 h exposure to 1 mM 2-HBP resulted in a significant increase of the proportion of injured/dead cells, as well as a strong decrease in both EB, SYTO9, and PI fluorescence. This decrease is similar as what is observed in stationary phase cells and suggests a complete metabolic arrest of cells (not shown). Some recovery of EB, SYTO9, and PI signal in the live cell proportion occurred 3 h after exposure to 1 mM 2-HBP, suggesting that cells started to divide again (Figures 3, S1).

An increased red fluorescence intensity was observed of cells exposed to 500 μM and 1 mM 2-HBP stained with DiOC₂(3) compared to the non-exposed cells, both after 1 h (Figure 4A) and 3 h (Figure 4B), suggesting that such cells have increased depolarization of their membranes as a consequence of being exposed to 2-HBP. The effect caused by 1 mM 2-HBP is comparable to the one observed when cells are exposed to 4 g per L of sodium azide (Figure S2). In both cases an increase of red fluorescence reflecting DiOC₂(3) intracellular aggregation due to membrane potential loss is observed (Joux and Lebaron, 2000), suggesting that 2-HBP toxicity is exerted at the level of uncoupling. These data therefore indicated that also wild-type HBP1 experiences acute 2-HBP toxicity at higher concentrations but somehow can cope with this and resume growth. From the absence of growth on 2-HBP by the mexAB-oprM mutants we hypothesized that the MexAB-OprM system would be responsible for mitigating 2-HBP toxicity in the wild-type.

In order to understand how the mexAB-oprM encoded system in P. azelaica would function to abolish 2-HBP toxicity, we used a previously developed EB equilibrium assay (Czechowska and van der Meer, 2012). This assay measures EB fluorescence in live cells exposed or not to membrane inhibitors under the premise that EB influx is spontaneous and efflux is energy dependent. Our hypothesis was that the MexAB-OprM system is an energy-dependent efflux system in strain HBP1 capable of expulsing EB. Interestingly, the staining patterns of the three mutants in all three treatments were similar but very distinct from the one of HBP1 wild-type (Figure 5 shows the results for wild-type and the mexA::Km mutant). In particular the SYTO9 fluorescence intensity was lower and independent of the treatment. EB fluorescence decreased in wild-type cells exposed to CCCP compared to the control, but increased after exposure to sodium azide and EDTA. Addition of TPP had no visible effect (Figure 5A). Under the assumption that EB influx would remain constant this suggested that efflux energy was interrupted by sodium azide, but not by TPP or CCCP. In contrast, EB fluorescence of the mutants was already 10 times higher, and SYTO9 fluorescence 10 times lower than wild-type in non-exposed and TPP-exposed cells. Addition of CCCP and sodium azide caused a slight decrease of EB fluorescence in mutant cells compared to the control. EDTA caused a further increase of EB fluorescence intensity in mutant cells (Figure 5B). Assuming that EB influx rates remain constant, the fact that higher equilibrium EB levels occurred in mutant cells and cellular inhibitors only exerted slight effects then suggests that EB efflux was impaired. This would be in agreement with the hypothesis that the mexAB-oprM system of P. azelaica HBP1 encodes a multi-drug efflux pump similar as in P. aeruginosa (Li et al., 1998).

To test impaired efflux directly, cells were preloaded with EB and examined for their ability to efflux the dye over time (Figure 6). EB fluorescence decreased in HBP1 wild-type cells over time with an apparent zeroth order constant of −1.15 1/min. In contrast, all mutants started with a 10-fold higher EB load, which in about 50% of those cells decreased to the same level as in the wild-type after 10 min. However, even after 40 min a large proportion of mutant cells were still carrying high loads of EB, which was most pronounced for the mexB::Km and oprM::Km mutants (Figure 6). This indicated that indeed EB-efflux was impaired in the mutants, and suggested that 2-HBP efflux could be disrupted as well.

In order to measure the effect of the mexA, mexB, or oprM interruptions in P. azelaica HBP1 on 2-HBP transport (here: the combined effect of in- and efflux), we used an indirect assay since radio-active 2-HBP was not available for direct measurement of uptake or efflux. The derivative assay is based on the intracellular measurement of 2-HBP by the HbpR protein, which controls expression of the GFP reporter protein from the P_c-promoter (Jaspers et al., 2000; Beggah et al., 2008). Previous results from such a 2-HBP-responsive E. coli bioreporter indicated that 2-HBP likely enters the cell by spontaneous partitioning and diffusion through the cellular membranes (Beggah et al., 2008). GFP fluorescence is thus an indirect (delayed) measure of the intracellular 2-HBP concentration. Wild-type P. azelaica and the three mutant strains were equipped with a plasmid carrying the hbpR gene under control of its own promoter and the GFP gene under control of the HbpR-dependent and 2-HBP inducible P_c-promoter (Jaspers et al., 2000; Beggah et al., 2008). Indeed, GFP expression was induced in a 2-HBP-dependent manner in wild-type HBP1 (Figure 7A). Interestingly, at 2-HBP concentrations of 0.029, 0.29, and 0.87 mM GFP levels increased in the first period after induction but then leveled off. This suggested that the cells had metabolized all the 2-HBP and no further induction occurred. At higher 2-HBP outside concentrations the cells use longer to metabolize all 2-HBP and, consequently, the GFP induction process continues (Figure 7A). In contrast, the mutant strains were much more sensitive to 2-HBP than wild-type HBP1 (Figure 7B). GFP expression was already induced at 2.4 μM 2-HBP, indicating that it enters the cells but is not expelled as effectively as in the wild-type (Figure 7B). GFP values saturated at 2.4 and 24 μM 2-HBP concentration but not at 0.24 mM (Figure 7B), further suggesting that low 2-HBP concentrations can be metabolized by the mutant cells, but that metabolism stops at higher (0.24 mM) 2-HBP concentrations.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.