Molecular determinants of Burkholderia pseudomallei BpeEF-OprC efflux pump expression

Abstract

Burkholderia pseudomallei, the cause of melioidosis, is intrinsically resistant to many antibiotics. Acquired multidrug resistance, including resistance to doxycycline and co-trimoxazole used for melioidosis eradication phase therapy, is mainly attributed to constitutive expression of the BpeEF-OprC efflux pump. Constitutive expression of this pump is caused by mutations affecting two highly similar LysR-type transcriptional regulators (LTTR), BpeT and BpeS, but their interaction with the regulatory region governing BpeEF-OprC expression has not yet been studied. The bpeE-bpeF-oprC genes are distally located in the llpE-bpeE-bpeF-oprC operon. The llpE gene encodes a putative lipase/esterase of unknown function. We show that in a bpeT mutant llpE is constitutively co-transcribed with bpeE-bpeF-oprC. As expected from previous studies with B. cenocepacia, deletion of llpE does not affect antibiotic efflux. Using transcriptional bpeE′-lacZ fusions, we demonstrate that the 188 bp bpeT-llpE intergenic region located between bpeT and the llpE-bpeE-bpeF-oprC operon contains regulatory elements needed for control of bpeT and llpE-bpeE-bpeF-oprC operon expression. By native polyacrylamide gel electrophoresis and electrophoretic mobility shift assays with purified recombinant BpeT and BpeS proteins, we show BpeT and BpeS form oligomers that share a 14 bp binding site overlapping the essential region required for llpE-bpeE-bpeF-oprC expression. The binding site contains the conserved T-N₁₁-A LTTR box motif involved in binding of LysR proteins, which in concert with two other possible LTTR boxes may mediate BpeT and BpeS regulation of BpeEF-OprC expression. These studies form the basis for further investigation of BpeEF-OprC expression and regulation at the molecular level by yet unknown external stimuli.

Introduction

Burkholderia pseudomallei is the etiologic agent of melioidosis [1, 2]. The bacteria is a significant cause of morbidity and mortality in its traditionally recognized endemic regions of Southeast Asia and Northern Australia, but is now known to have a much wider global distribution [3, 4]. Current estimates place global mortality from melioidosis exceeding that of dengue and leptospirosis [4]. This is of particular concern considering the lack of a vaccine to prevent B. pseudomallei infections, and the lengthy nature of treatment.

Clinicians typically administer bi-phasic therapy to overcome infections with B. pseudomallei [5, 6]. This involves up to 2 weeks of intravenous ceftazidime or a carbapenem in the intensive phase, and up to several months of oral co-trimoxazole during the eradication phase, or co-amoxiclav if complications or side effects arise [5, 6]. Even with an aggressive antibiotic treatment regimen therapy often fails, in part due to intrinsic and acquired resistance [7].

β-lactam resistance is seemingly confined to drug-specific mechanisms, such as increased expression [8], increased gene copy number [9], target mutation [10], or extension of substrate spectrum due to enzyme structural mutations [8, 11, 12].

Multidrug resistance (MDR) is caused by resistance nodulation cell division (RND) family efflux pump expression [13]. B. pseudomallei strains encode at least 10 RND pumps, three of which have been characterized in some detail [13]. AmrAB-OprA is constitutively expressed in most B. pseudomallei strains and responsible for the bacterium’s intrinsic resistance to aminoglycosides and macrolides [14]. Overexpression of AmrAB-OprA in amrR mutants causes doxycycline resistance [15]. Environmental or clinical strains susceptible to aminoglycosides either do not express this pump, contain pump component structural mutations, or completely lack the pump structural genes [16–18]. It was recently shown that clinical isolates exhibiting decreased meropenem susceptibility frequently contain mutations in amrR, encoding a TetR-family repressor of the amrAB-oprA operon [19]. BpeAB-OprB is expressed at detectable levels in wild-type strains and its expression is markedly increased in bpeR regulatory mutants where it causes a weak multidrug-resistant phenotype [20]. The clinical relevance of this pump is questionable because of the modest levels of drug resistance it causes [13, 20]. BpeEF-OprC extrudes a broad-spectrum of antibiotics that includes tetracyclines, chloramphenicol, fluoroquinolones and both components of the main-line eradication phase therapy; sulfamethoxazole and trimethoprim [13, 21–24]. Two LysR-type transcriptional regulators (LTTRs) that share 62 % identity at the protein level are involved in regulation of BpeEF-OprC expression [21, 23]. BpeT is encoded by bpeT, which is transcribed divergently from the llpE-bpeE-bpeF-oprC operon on chromosome II (Fig. 1) [23]. BpeS is encoded by bpeS, a gene distantly located on chromosome I (Fig. 1) [23]. The current model for regulation of llpE-bpeE-bpeF-oprC transcription predicts that in wild-type cells, the expression of this operon is regulated by BpeT and BpeS that are complexed with co-inducers (Fig. 1a). In previous studies, we have shown that select pump substrates alone can induce BpeEF-OprC expression; however, it is possible that these proteins interact with other environmental compounds [23]. Constitutive expression of BpeEF-OprC in clinical and laboratory-selected B. pseudomallei mutants is due to mutations in bpeT and bpeS. In this scenario, activation of llpE-bpeE-bpeF-oprC is predicted to be governed by binding of co-inducer independent BpeT and BpeS mutant proteins to the bpeT-llpE intergenic region (Fig. 1b) [21, 23].

Fig. 1.

Model for regulation of expression of the B. pseudomallei llpE-bpeE-bpeF-oprC operon. The bpeE, bpeF and oprC genes (light grey) are located on chromosome II and encode structural components of the BpeEF-OprC efflux pump. The bpeT regulatory gene (white) is located adjacent to and transcribed divergently from the llpE-bpeE-bpeF-oprC genes. The bpeS regulatory gene (dark grey) is located distantly on chromosome I [23]. BpeT and BpeS code for closely related LysR-type transcriptional regulators (LTTRs). The model predicts that the bpeE, bpeF and oprC genes are co-transcribed with llpE. The llpE gene (dotted) is conserved in pathogenic and non-pathogenic Burkholderia species and annotated as a lipase/esterase [48]. (a) Regulation of llpE-bpeE-bpeF-oprC expression in the wild-type. Published data suggest that transcription of the llpE-bpeE-bpeF-oprC operon is activated by BpeT and BpeS protein/co-inducer complexes. Co-inducers can be select pump substrates, possibly antibiotics, and/or hitherto unidentified compounds. (b) Regulation of llpE-bpeE-bpeF-oprC expression in regulatory mutants. In bpeS and bpeT regulatory mutants, the llpE-bpeE-bpeF-oprC operon is constitutively expressed in a co-inducer independent fashion. Such mutations mostly affect the BpeT and BpeS co-inducer binding domains, but in one instance a mutation affecting the BpeS helix-turn-helix DNA binding domain also led to constitutive llpE-bpeE-bpeF-oprC transcription.

CeoAB-OpcM, the orthologues of BpeEF-OprC, were first described in B. cenocepacia and shown to constitute an efflux system for chloramphenicol, ciprofloxacin and trimethoprim [25, 26]. These studies also showed a transcriptional arrangement similar to that proposed for bpeEF-oprC (Fig. 1): (1) the presence of llpE, a unique gene encoding a putative lipase/esterase of the alpha/beta hydrolase superfamily preceding the efflux system structural genes, but not required for efflux of known pump substrates as shown by complementation analysis; and (2) a divergently transcribed gene, ceoR, encoding an uncharacterized LTTR.

In this study, we built on recent knowledge about regulatory mutants and proteins to lay the foundation for future in-depth molecular studies of the regulation of BpeEF-OprC by: (1) re-examining the llpE-bpeE-bpeF-oprC transcriptional organization; (2) identifying DNA sequences in the bpeT-llpE intergenic region governing BpeT- and BpeS-mediated BpeEF-OprC expression; and (3) gaining some basic knowledge about the BpeT and BpeS LTTR proteins.

Methods

Bacterial strains and growth

Escherichia coli strains used in this study were DH5α (ThermoFisher Scientific, Waltham, MA), NEB5α (New England Biolabs, Ipswich, MA) and XL10-Gold (Agilent, Santa Clara, CA) for routine cloning purposes and RHO3 [27] for conjugal transfer of plasmids into B. pseudomallei. B. pseudomallei strains used were select agent-excluded Bp82 [28] (www.selectagents.gov/SelectAgentsandToxinsExclusions.html) and its derivatives, and the virulent strain 1026b [21] and its derivative Bp282 [24]. B. pseudomallei strains utilized in this study are listed in Table 1. Bacterial strains were grown in Luria broth (LB) Lennox or agar (5 g l⁻¹ NaCl) at 37 °C unless otherwise stated. E. coli RHO3 strains used for conjugation were grown in media containing 400 µg ml⁻¹ diaminopimelic acid (DAP). All B. pseudomallei Bp82 derivatives were grown either in Lennox-LB liquid media or on agar plates supplemented with adenine to final concentrations of 80 and 40 µg ml⁻¹, respectively. Growth media were supplemented with antibiotics at the following concentrations: ampicillin, 100 µg ml⁻¹ (E. coli); chloramphenicol, 35 µg ml⁻¹ (E. coli); kanamycin, 35 µg ml⁻¹ (E. coli and AmrAB-OprA deficient B. pseudomallei) and 1000 µg ml⁻¹ (B. pseudomallei AmrA⁺B⁺-OprA⁺); gentamicin, 15 µg ml⁻¹ (E. coli and AmrAB-OprA deficient B. pseudomallei) and 500 µg ml⁻¹ (B. pseudomallei AmrA⁺B⁺-OprA⁺). All procedures with select agent-excluded strain Bp82 and its derivatives were performed at BSL-2 with Institutional Biosafety Committee approval. Virulent B. pseudomallei strains were manipulated in select agent-approved BSL-3 facilities in the Rocky Mountain Regional Biosafety Laboratory at Colorado State University or at the University of Florida using approved select agent-compliant procedures and protocols.

Table 1. B. pseudomallei strains used in this study.

Bp82 and derivatives
Strain	Description	Source
Bp82	1026b ∆purM select agent-excluded strain	[28]
Bp82.270	Bp82 bpeTS280P	[23]
Bp82.284	Bp82 bpeSP29S	[23]
Bp82.285	Bp82 bpeSK267T	[23]
Bp82.419	Bp82.270 ∆llpE	This study

Bp282 and derivatives
Strain	Description	Source
Bp282	Δ(amrAB-oprA) Δ(bpeAB-oprB) bpeTS280P	[24]
Bp769	Bp282 :: mini-Tn7T-Gm-lacZ*	This study
Bp770	Bp282 :: mini-Tn7T-Gm-bpeE′-lacZ 188 bp IR†	This study
Bp771	Bp282 :: mini-Tn7T-Gm-bpeE′-lacZ IR ∆88	This study
Bp773	Bp282 :: mini-Tn7T-Gm-bpeE′-lacZ IR Δ112	This study
Bp774	Bp282 :: mini-Tn7T-Gm-bpeE′-lacZ IR Δ142	This study
Bp816	Bp282 :: mini-Tn7T-Gm-bpeE′-lacZ IR Δ188	This study
Bp817	Bp282 :: mini-Tn7T-Gm-bpeE′-lacZ IR Δ126	This study
Bp825	Bp282 :: mini-Tn7T-Gm-bpeE′-lacZ IR Δ53	This study
Bp826	Bp282 :: mini-Tn7T-Gm-bpeE′-lacZ IR Δ14	This study

*Strains Bp770-Bp826 harbour the mini-Tn7 insertions at the glmS2-associated attTn7. The insertion in Bp769 is located at the glmS3-associated attTn7.

†The fusion constructs also contain the entire 855 bp llpE gene, the 49 bp llpE-bpeE intergenic region and the first 178 bp of bpeE (see Fig. 3).

Construction of an llpE deletion mutant

For construction of a marker-less non-polar Bp82 ΔllpE mutant a 530 bp upstream fragment was PCR-amplified using Q5 DNA polymerase (New England Biolabs) using primers P3212 and P3213 (primers were purchased from Integrated DNA Technologies, Coralville, IA and are listed in Table S1, available in the online version of this article) and Bp82.270 [23] genomic DNA purified using the Wizard Genomic DNA purification kit (Promega) as a template. A 630 bp downstream fragment was similarly amplified using P3214 and P3215. These fragments were assembled with NotI and XhoI digested pEXKm5 [27] DNA using the NEBuilder High-Fidelity DNA Assembly System (New England Biolabs). The assembly mixture was transformed into DH5α selecting for kanamycin resistance. Plasmid DNA was extracted using the Sigma-Aldrich (St. Louis, MO) GeneJet Mini-prep kit according to the supplied protocol for Gram-negative bacteria or the Macherey-Nagel (Bethlehem, PA) NucleoSpin Plasmid kit. Plasmids with correct inserts were identified by restriction enzyme digestion and DNA sequencing. The resulting pPS3398 (plasmids used in this study are listed in Table S2) was transformed into RHO3 and after transfer to Bp82.270 a ΔllpE mutant was isolated using sacB counter-selection as described previously [27]. The presence of ΔllpE in the resulting Bp82.419 was verified by PCR and DNA sequencing.

Antimicrobial susceptibility testing

The trimethoprim susceptibility of Bp82 and its derivatives was assessed by determining MICs using the Etest method according to the manufacturer’s instructions (AB bioMérieux, Marcy l′Etoile, France) and as previously described [22]. The MICs of chloramphenicol, ceftazidime, ciprofloxacin and doxycycline (all from Sigma-Aldrich) were determined by the standard broth microdilution (BMD) method, following Clinical and Laboratory Standards Institute guidelines [29]. All MIC tests were incubated at 37 °C for 16 to 20 h prior to visual inspection. To support growth of Bp82 and its derivatives, cation-adjusted Mueller-Hinton broth and agar were supplemented with 80 and 40 µg ml⁻¹ adenine, respectively.

RT-PCR and RT-qPCR

Bp82 (wild-type) and Bp82.270 (bpeT_S280P) cells were grown at 37 ˚C in LB medium to the mid-log phase (OD_600nm=0.6 to 0.8). Total RNA was isolated using the RNeasy Protect Bacteria mini kit (Qiagen, Valencia, CA) and cDNA synthesis was performed as previously described [20, 22]. For reverse transcription PCR (RT-PCR), a fragment bridging the llpE-bpeE intergenic region was amplified using Q5 high-fidelity polymerase according to the manufacturer's recommendations (NEB, Ipswich, MA) using primers P1933 and P1934 (Table S1). Reverse transcription-quantitative PCR (RT-qPCR) was performed as previously described [20] to assess expression levels of bpeF and llpE mRNAs. Then, 23S rRNA was used as the housekeeping control. The primer sets used were Bp23S_F and Bp23S_R for 23S rRNA [30], bpeF-Fw and bpeF-Rv for bpeF and llpE_Fw and llpE_Rv for llpE (Table S1).

Construction of mini-Tn7-lacZ reporter fusion-containing strains

Fragments of varying sizes of the bpeT-llpE intergenic region were amplified by PCR from 1026b genomic DNA using primers listed in Table S1. To obtain a read out for the bpeE-bpeF-oprC portion of the llpE-bpeE-bpeF-oprC operon, the fragments contained various portions of the bpeT-llpE intergenic region, the entire 855 bp llpE gene, the 49 bp llpE-bpeE intergenic region and the first 178 bp of bpeE. Most fragments were cloned into pGEM-T Easy (Promega, Madison, WI) and sequenced, before being subcloned into pTZ120 [31] and then pUC18T-mini-Tn7T-Gm-lacZ [32]. After confirming the presence of the correct sequences, the mini-Tn7-lacZ elements containing either the 5′ deletion fragments, or the wild-type intergenic region (IR) or the mini-Tn7-lacZ empty vector control were introduced into bpeT_S280P strain Bp282, by electroporation along with helper plasmid pTNS3 [33]. Transformants were selected on Lennox-LB agar plates with 15 µg ml⁻¹ of gentamicin and 50 µg ml⁻¹ of X-Gal. Transformants that were gentamicin resistant and blue contained the mini-Tn7-lacZ elements inserted at one of three possible glmS-associated attTn7 sites. Insertion at these sites was verified by PCR as described previously [33]. Only isolates with single mini-Tn7-lacZ insertions were further studied. All insertions except the wild-type control were at the glmS2-associated attTn7 site.

β-galactosidase assays

β-galactosidase (β-gal) assays were performed as described previously [34]. Mid-log phase cells (OD_600nm=0.6–0.8) were permeabilized with chloroform and SDS, and hydrolysis of o-nitrophenyl-β-d-galactopyranoside (Sigma-Aldrich) was measured. β-gal activity is expressed in Miller units. The transcriptional activity of each sample was normalized to the baseline activity produced by both the background strain and again by the β-gal activity produced by the strain harbouring an empty mini-Tn7-lacZ element. One-way ANOVA with Dunnet’s post test were performed for each sample against the full-length control using GraphPad Prism V6 to identify significant loss of transcriptional activity.

Protein expression and purification

The bpeT or bpeS wild-type or mutant coding sequences were PCR amplified from genomic B. pseudomallei DNA or plasmid DNA as detailed in Table S2 and ~1 kb fragments containing wild-type or mutant bpeT and bpeS fragments were cloned into pGEM-T Easy following the manufacturer’s protocol. The constructs were confirmed by DNA sequencing. Subcloning into pET-21b (EMD Millipore, Burlington, MA) vectors on NdeI+HindIII (for bpeT) or NdeI+EcoRI (for bpeS) fragments fused the bpeT and bpeS coding sequences in frame to a carboxy-terminal hexahistine tag encoded by pETR-21b. After construct verification by DNA sequencing, correct constructs were transformed into strain BL21-CodonPlus (DE3)-RP (Agilent Technologies) to facilitate protein expression from genes of GC-rich organisms. Protein purification was scaled to 150–500 ml cultures using these isolates. All samples were induced by the addition of 1 mM IPTG (Gold Biotechnology, St. Louis, MO) starting at OD_600nm=0.8–1.0 for up to 4 h or an OD_600nm=1.5–2.0. Cells were harvested and proteins were extracted using an adapted protocol first developed elsewhere [35]. Cleared lysate was applied to a column packed with Ni NTA (ThermoFisher Scientific, Waltham, MA) slurry. BpeS or BpeT was eluted in fractions of 40–100 mM imidazole in buffer PCW [8 mM NaH₂PO₄, 286 mM NaCl, 1.4 mM KH₂PO₄, 2.6 mM KCl, 0.1 % sarkosyl (pH 7.4)]. After SDS-polyacrylamide gel analysis, fractions with the highest concentrations of the proteins of interest were combined and dialysed against 175 mM Tris-HCl (pH 7.2) or PBS [137 mM NaCl, 2.7 mM KCl, 8 mM Na₂HPO₄ and 2 mM KH₂PO₄(pH 7.4)]. Proteins were stored in dialysis buffer at −20 °C until use.

Native gel electrophoresis

In total, 500 ng to 1 µg of protein per sample was mixed 1 : 2 with native sample buffer containing 62.5 mM Tris-HCl, pH 6.8, 0.01 % bromophenol blue and 40 % glycerol. All samples and 5 µl of NativeMark Standard (ThermoFisher Scientific) diluted 1 : 20 in sample buffer were loaded onto a 4–15 % Mini-Protean TGX gel (Bio-Rad, Hercules, CA) and electrophoresed in buffer containing 25 mM Tris-HCl, pH 8.3, 192 mM glycine without SDS. Gels were run at 130 V for approximately 1 h before being fixed in a solution containing 30 % ethanol and 10 % acetic acid for 30 min. The gels were washed with ultrapure water, and stained with silver stain (ThermoPierce, Rockford, IL) according to the manufacturer’s protocol for native polyacrylamide gel application.

Electrophoretic mobility shift assays

Electrophoretic mobility shift assay (EMSA) was used to assess binding of BpeT and BpeS to sites within the bpeT-llpE IR. Fragments containing various segments of the intergenic region were PCR amplified. These DNA fragments were excised and purified from a 1 % agarose gel using a Gen-Elute kit (Sigma-Aldrich) following the manufacturer’s protocol. Overall, 2× binding buffer (375 mM KCl, 100 mM Tris pH 8.5, 10 mM DTT, 10 mM MgCl₂, 2 mM EDTA, 10 % glycerol) was mixed with 10–100 ng of probe DNA, 30 µg µl⁻¹ BSA and 1.25 ng µl⁻¹ of poly(dI:dC) (ThermoFisher Scientific). From 150 nM–1.5 µM recombinant protein was added and reactions were incubated at room temperature for 20 min. Samples were loaded onto a pre-equilibrated (1 h, 100 V, 1×TBE) 5 % TBE polyacrylamide gel (Bio-Rad, Hercules, CA) and run at 120–130 V for approximately 1.5 h in 1× TBE buffer. Gels were stained with SYBR Gold (Molecular Probes, Eugene, OR) before visualization on a Bio-Rad Chemidoc (Hercules, CA).

Results and discussion

Refined transcriptional and functional studies on llpE

Although llpE and efflux pump gene co-transcription was previously shown in B. cenocepacia using RNA isolated from a wild-type strain and RT-PCR [25] we decided to take advantage of a bpeT_S280P regulatory mutant that constitutively expresses BpeEF-OprC [23] to show co-transcription and co-regulation of llpE and bpeE in this mutant compared to the wild-type using RT-PCR and RT-qPCR. Using cDNA derived from the bpeT_S280P strain Bp82.270 or genomic DNA isolated from its isogenic parental strain Bp82 as the template, PCR amplification using primers that hybridize to regions flanking the 49 bp llpE-bpeE intergenic region (Fig. 2a) generated the same 384 bp fragment as assessed by agarose gel electrophoresis and sequencing (Fig. 2b). RT-qPCR analysis indicated that bpeF and llpE mRNA levels were ~70-fold higher in the bpeT_S280P strain Bp82.270 when compared to Bp82 (Fig. 2c). Previous studies with plasmids that harboured the B. cenocepacia ceoA-ceoB-opcM genes with or without the llpE gene indicated that LlpE was not required for efflux of known pump substrates in B. cenocepacia [25]. Here, we assessed LlpE’s role in efflux by conducting antibiotic susceptibility tests with isogenic llpE⁺ wild-type and a non-polar ΔllpE mutant that constitutively express BpeEF-OprC. These studies showed that MICs for pump substrates were the same in both strains (Table 2). In summary, our studies corroborate previous findings made with B. cenocepacia, i.e. that llpE forms an operon with bpeE, bpeF and oprC and that its product is not required for BpeEF-OprC-mediated antibiotic efflux. In addition, our studies confirm that the genetic determinants governing the expression of the llpE and the bpeE-bpeF-oprC genes are the same. Regulatory sequences are likely located within the 188 bp bpeT-llpE IR since the ΔllpE mutant contains a complete deletion from start to stop codon and the 49 bp llpE-bpeE region contains no recognizable promoter sequences. The data also lend credence to using bpeE or bpeF as an indicator of llpE-bpeE-bpeF-oprC transcription as has been done in this study for transcriptional fusion analysis or customarily for RT-qPCR (this study and [22, 23, 30]).

Fig. 2.

Co-expression of llpE and bpeF. (a) Details of the B. pseudomallei bpeT-llpE-bpeE-bpeF-oprC region. Intergenic regions and their extents are indicated in base pairs. The approximate positions of primers used to determine co-transcription of llpE-bpeE are shown as numbered black arrows. For details on genes, see Fig. 1. (b) The llpE-bpeE-bpeF-oprC genes form an operon. cDNA synthesized from Bp82.270 total RNA was used as the template to determine llpE-bpeE co-transcription with primers P1933 and P1934. Lane 1, Hi-Lo DNA ladder (Minnesota Molecular, Minneapolis MN); lane 2, cDNA template minus reverse transcriptase (RT) (negative control); lane 3, Bp82 genomic DNA template (positive control); lane 4, cDNA template. (c) The llpE and bpeF genes are expressed at similar levels and overexpressed in a regulatory mutant. Cells of Bp82 (wild-type) and Bp82.270 (bpeT_S280P) were grown to mid-log phase in LB medium and total RNA was isolated. The llpE (grey bars) and bpeF (black bars) and mRNA levels were determined by RT-qPCR. The level of expression of llpE and bpeF is shown relative to strain Bp82. Error bars indicate the error of the mean between three biological replicates.

Table 2. LlpE is not required for BpeEF-OprC-mediated antibiotic efflux.

Strain	Relevant genotype	MIC (μg ml−1)*
		Ceftazidime	Chloramphenicol	Ciprofloxacin	Doxycycline	Trimethoprim
Bp82	Prototype†	2	16	2	0.5	0.75
Bp82.270	bpeTS280P‡	2	1024	16	8	>32
Bp82.419	bpeTS280P ΔllpE	2	1024	16	8	>32

*MICs were determined by broth microdilution, except trimethoprim for which Etest was employed.

†Bp82 and its derivatives are ΔpurM.

‡Strains containing bpeT_S280P constitutively express BpeEF-OprC.

The bpeT-llpE IR contains the regions necessary for transcription of llpE-bpeE-bpeF-oprC

Transcriptional lacZ gene fusions carrying nested deletions in the 188 bp bpeT-llpE IR were constructed and used to approximate the location of sequences required for llpE-bpeE-bpeF-oprC operon expression. We utilized bpeE′-lacZ fusion constructs since the overall goal was to study BpeEF-OprC efflux pump expression. The various portions of the bpeT-llpE IR contained in the bpeE′-lacZ fusion constructs are shown in Fig. 3(a). The gene fusions contained on a mini-Tn7 element were integrated in single-copy into the genome of strain Bp282, which expresses bpeE′-lacZ constitutively driven by the bpeT_S280P mutation, and β-gal activities were measured (Fig. 3b). Cells containing constructs with the entire 188 bp IR, or constructs lacking the first 14 or 53 bp of the IR expressed significant levels of β-gal activity. In contrast, no β-gal activity was detectable in cells with constructs where the first 88 bp or more of the IR were deleted. Interestingly, cells containing a fusion with the entire IR expressed significantly lower levels of β-gal activity, when compared to cells with constructs that had either 14 or 53 bp deleted from the 5′ end of the IR, which includes the putative T-N_11-A LTTR box in this region (see Fig. 7). This infers that the deleted region has an inhibitory effect on llpE-bpeE-bpeF-oprC transcription. In summary, aside from providing a glimpse into likely bpeT autoregulation these experiments suggested a critical region located between nt 54 and 88 of the IR that is required for llpE-bpeE-bpeF-oprC expression. Since we failed to obtain the llpE-bpeE-bpeF-oprC operon transcriptional start site(s) using either fluorescent primer extension or S1 nuclease mapping methods, and promoter prediction algorithms fail to identify any obvious promoter sequences, these data did not provide insight into whether the critical region contains a promoter or activator binding region or both. Investigations using transcriptional lacZ fusions to map critical regions required for bpeT expression were largely unsuccessful, although conventional transcript mapping technologies identified two disparate bpeT transcriptional start sites at nucleotide 45 (primer extension) and between nt 88–90 (S1 nuclease) [36]. Further studies focused on exploring regulatory proteins of llpE-bpeE-bpeF-oprC operon expression.

Fig. 3.

Use of transcriptional lacZ fusions for identification of sequences within the bpeT-llpE intergenic region (IR) required for llpE-bpeE-bpeF-oprC operon transcription. (a) Nucleotide sequence of the 188 bp bpeT-llpE IR. Arrows above underlined ATG sequences indicate the initiation codons of bpeT (left arrow) and llpE (right arrow), respectively. Right-angled arrows mark the wild-type IR and sequentially shorter IR segments (numbered relative to the first 5′ nucleotide of the IR region) that are fused to the lacZ gene including the entire 855 bp llpE gene, the 49 bp llpE-bpeE intergenic region and the first 178 bp of bpeE to generate bpeE’-lacZ transcriptional fusions. Triangles indicate the region of critical sequences between nt 54 and 88 required for bpeE′-lacZ expression. Numbered asterisks indicate the positions of significant nt discussed elsewhere in the text. (b) β-galactosidase (β-gal) activities of bpeE′-lacZ fusions containing nested IR deletions. The gene fusions containing the IR segments shown in (a) carried on mini-Tn7 elements were integrated into the genome of BpeT_S280P strain Bp282 that constitutively expresses the llpE-bpeE-bpeF-oprC operon. β-gal expression in all strains was normalized to an empty vector lacZ reporter integrant strain and compared to a wild-type control consisting of the full 188 bp IR upstream of the entire llpE gene, the llpE-bpeE intergenic region and the first 178 bp of bpeE fused to lacZ. β-gal activities were determined and are expressed in Miller units+/−sd. Data is shown as means of two biological replicates in technical triplicate, with error bars representing one standard deviation. Statistical analysis was performed by one-way ANOVA and Dunnett’s multiple comparison using GraphPadPrism. ****=P<0.0001, WT compared to ∆88 through ∆188; ***=P<0.001, WT compared to ∆14 and ∆53.

Fig. 7.

Regulatory sequences within the bpeT-llpE intergenic region required for llpE-bpeE-bpeF-oprC transcription. The bpeT-llpE IR extends from nt 1 and 188 and separates the initiation codons of the divergently transcribed bpeT (left thin arrow) and llpE (right thin arrow) genes, respectively. SD indicates putative bpeT and llpE ribosome-binding sites. Triangles indicate sequences required for expression of a bpeE′-lacZ transcriptional fusion. Sequences located between 1 and 53 exhibited an inhibitory effect on llpE-bpeE-bpeF-oprC transcription. Sequences located between nt 74 and 89 are required for BpeT and BpeS binding. This region contains a putative T-N_11-A LTTR box (boxed sequence) and inverted repeats that are marked with thick horizontal arrows. Two additional putative LTTR box motifs may be present in the IR, one overlapping the putative bpeT SD and another located downstream of the critical sequences required for llpE-bpeE-bpeF-oprC transcription.

BpeT and BpeS are oligomeric LTTR proteins

To study protein interactions with the bpeT-llpE IR, BpeT-His₆ and BpeS-His₆, as well as their His₆-tagged mutant derivatives BpeT_S280P, BpeS_P29S and BpeS_K267T that cause constitutive BpeEF-OprC expression [23], were expressed in E. coli and purified to near homogeneity (Fig. S1). The apparent molecular masses of the His₆-tagged BpeT and BpeS proteins were ~38 000–39 000 Da. These values are consistent with calculated masses of 38 924 and 38 934 Da for the monomeric 348 amino acid BpeT-His₆ and BpeT_S280P-His₆ proteins, and 37 601, 37 574 and 37 591 Da for the monomeric 343 amino acid BpeS-His₆, BpeS_P29S-His₆ and BpeS_K267T-His₆ proteins, respectively.

LTTR proteins are generally active as homo-tetramers [37]. To qualitatively assess BpeT and BpeS multimer formation we determined the native size of purified BpeT-His₆ and BpeS-His₆ wild-type and mutant proteins (Figs 4 and S1a). In the wild-type BpeT preparation a single protein band with an extrapolated molecular mass of 109 kDa was visible, possibly correlating to the formation of a protein trimer using the calculated 39 kDa weight of BpeT-His₆ monomer. With the BpeT_S280P-His₆ mutant protein the 109 kDa band was not detected, but complexes with extrapolated masses of 218 and 353 kDa were detected, suggesting hexamer and nonamer complexes. In the BpeS-His₆ wild-type preparation, protein bands with extrapolated masses of 108, 220, 329 and 642 kDa were present. These oligomers would correspond to protein complexes containing 3, 6, 9 and 18 H₆-tagged monomers using the calculated 37.6 kDa weight of the BpeS-His₆ protein monomers. The same pattern was observed with BpeS_K267T-His₆ mutant although the complex sizes were slightly different (98.5, 220, 344 and 642 kDa). The overall complex size distribution is very similar to what we observed for BpeT_S280P. No monomers were detectable in any of the BpeT and BpeS preparations by native PAGE although proteins of the expected monomer sizes were detectable when analysed by SDS-PAGE (Fig. S1b). A band with a mass of >15 kDa was detected in BpeT and BpeS preparations. This is likely a degradation product as a corresponding protein was not detected in the same preparations analysed by SDS-PAGE (Fig. S1b).

Fig. 4.

Native polyacrylamide electrophoresis. 500 ng of wild-type BpeT-His₆ (a) and BpeS-His₆ (b) proteins were analysed by electrophoresis on a 4–15 % native polyacrylamide gel and detected by silver staining. Molecular mass was determined by comparison to NativeMark standard proteins (open circles; ThermoFisher Scientific) whose sizes are 720, 480, 242, 146 and 66 kilodaltons (kDa). Observed BpeT-His₆ and BpeS-His₆ complexes are indicated by arrows and marked with filled squares and triangles, respectively. Numbers indicate observed sizes in kDa.

Native sizes of both wild-type and mutant His₆-tagged BpeT and BpeS were also determined by size exclusion chromatography (data presented in [36]). For BpeT samples, two peaks were detected, one with a calculated partition coefficient correlating to a molecular weight of 66 kDa, consistent with the formation of a dimer. The second peak had a calculated molecular mass of >15 kDa, which is likely the same degradation product detected by native PAGE. Two peaks with estimated molecular masses of 34 and 202 kDa were detected in BpeS samples. These likely correspond to monomers and hexamers.

While these results show that BpeT and BpeS in their native form exist as oligomers, our data indicate homo-trimers in solution, in contrast to previous findings that most bacterial LTTRs form homo-tetramers [37, 38]. However, atypical organization of LTTR oligomers is not unknown; hexameric oligomers of LysR protein HsdR have been identified in regulation of steroid metabolism by Comomonas testosteroni [39]. As has been shown with other LTTRs, we encountered insolubility and precipitation difficulties with the full-length versions of these proteins. The poor solubility and stability of LTTR proteins seem to be associated with the oligomerization of their DNA-binding domains in solution. Removal of the DNA-binding region [40, 41] or addition of target DNA [42] have been shown to stabilize and increase solubility of the proteins. Poor solubility leads to ready oligomerization, which likely explains the high-order oligomerization observed in this study.

Wild-type and mutant BpeT and BpeS bind to the bpeT-llpE IR

EMSAs were employed to assess binding of purified His₆ tagged wild-type and mutant proteins to the 188 bp bpeT-llpE IR. These assays showed that both wild-type and mutant BpeT-His₆ and BpeS-His₆ mutant proteins bound to the IR (Figs 5 and S2). EMSA and defined fragments of the IR were then used to locate binding sites for BpeT and BpeS in the IR. These fragments were designed to correlate to 5′ deletion sequences present in transcriptional lacZ fusions used to locate critical regions for llpE-bpeE-bpeF-oprC expression. With both BpeT (Fig. 6a) and BpeS (Fig. 6b), no band shift was observed with fragments containing bp 1–73 and 89–188. This suggests that BpeT and BpeS preferentially bind to the 14 bp sequence encompassed by nt 74–88 (Fig. 7).

Fig. 5.

DNA binding of purified wild-type and mutant BpeT-His₆ and BpeS-His₆ proteins. A DNA fragment containing the 188 bp bpeT-llpE intergenic region (IR) was incubated with or without the indicated BpeT-His₆ (0.2 µM), BpeS-His₆ (0.2 µM), BpeS_K267T-His₆ (0.8 µM) and BpeS_P29S-His₆ (0.15 µM, L and 1.5 µM, H) proteins. BpeT_S280P-His₆ also shifted the IR fragments indistinguishable from BpeT-His₆ (not shown). After electrophoresis, DNA was stained with SYBR Gold. The arrow indicates shifted fragments.

Fig. 6.

Approximate localization of the BpeT and BpeS binding sites in the bpeT-llpE intergenic region (IR). For electrophoretic mobility shift assays DNA fragments containing different regions of the bpeT-llpE IR were incubated with (+) or without (–) 1 µM BpeT-His₆ (a) and BpeS-His₆ (b). After electrophoresis DNA was stained with SYBR Gold. The entire IR extends from nt 1–188 (sequences are numbered relative to the first 5′ nucleotide of the IR). Arrows indicate a detectable shift and asterisks mark weak shifted fragments. In (a) fragments used for EMSA contained the indicated IR segments on 208, 282, 295, 439 and 473 bp fragments containing IR-flanking bpeT or llpE DNA. In (b) the unshifted DNA fragments contained the same IR segments (plus the additional 54–188) as those shown in (a). The unshifted DNA fragments were arbitrarily aligned to show that BpeS binds to and shifts the same IR segments as BpeT does.

Taken together several conclusions can be drawn from these results. First, BpeT and BpeS bind to the same site in the bpeT-llpE IR. Second, mutant BpeT and BpeS proteins bind to the IR. As the carboxy-terminus of these proteins contains both oligomerization domains and co-inducer binding regions [37, 38, 43], and no distinct changes in multimerization could be observed in mutant forms of BpeT or BpeS, these amino acid substitutions may instead promote a co-inducer independent state. This leads to constitutive BpeEF-OprC pump expression observed in both clinical and laboratory isolates [23]. We do not yet understand why the BpeS_P29S mutant protein leads to constitutive BpeEF-OprC expression. Although the amino acid substitution lies with the helix-turn-helix domain, it does not impede protein-DNA binding as the mutant protein binds to the IR (Fig. 5) and does so in a concentration-dependent manner (Fig. S2). Constitutive expression is thus not due to disabling a BpeS repressor function, but likely due to co-inducer independent activation of bpeEF-oprC expression.

Conclusion

This study represents the initial identification of regulatory sequences needed for BpeEF-OprC expression in response to yet unknown external stimuli. Despite the fact that our studies were hampered by poor BpeT and BpeS solubility, we were able to arrive at several conclusions. The regulatory sequences required for control of bpeT and llpE-bpeE-bpeF-oprC expression are located within the small (188 bp) bpeT-llpE IR. Sequences that inhibit expression of llpE-bpeE-bpeF-oprC transcription are located upstream of bpeT within the first 53 bp of the IR. The essential sequence identified for llpE-bpeE-bpeF-oprC transcription overlap the 14 bp sequence needed for both BpeS and BpeT binding (Fig. 7). It contains the canonical T-N_11-A LTTR box involved in DNA binding of LysR-type proteins [44]. Together with other possible LTTR boxes present in the IR, the box identified in the BpeT and BpeS binding region may promote the DNA bending function conserved by the LysR regulatory protein family (reviewed in [37]). LTTR proteins in this family typically have multiple binding sites within a regulatory region, but with EMSA we were only able to identify one binding site and attempts at DNaseI footprinting were unsuccessful. It is also interesting to note the overlap of the BpeS and BpeT binding sites and the sequences required for llpE-bpeE-bpeF-oprC transcription. This sequence may contain a promoter for transcription of llpE-bpeE-bpeF-oprC and as such, would explain both the overlap of BpeT and BpeS binding sites, and loss of transcriptional activity caused by its absence. Binding of two LTTRs at the same sites also poses many unresolved questions such as mechanisms of regulation of LTTR expression, including autoregulation [37, 39], formation of BpeT and BpeS hetero-oligomers as suggested for other LTTRs [45], and external stimuli involved in BpeT and BpeS mediated BpeEF-OprC expression. One such environmental cue may be low iron levels in the host, demonstrated by B. cenocepacia CeoAB-OpcM induction by iron starvation and salicylate [25, 26]. This scenario is, however, unlikely in B. pseudomallei. First, the llpE-bpeE-bpeF-oprC operon genes were not identified in previous global expression studies as being upregulated in cells grown in iron-depleted medium [46, 47]. Second, our experiments showed that llpE-bpeE-bpeF-oprC operon expression is not induced, but rather repressed in the presence of salicylate (Fig. S3b). As previously shown, select antibiotics that are pump substrates alone can induce BpeEF-OprC expression [23], e.g. doxycycline (Fig. S3a). Further studies are needed to identify whether other environmental signals or perhaps other factors impact regulation of llpE-bpeEF-oprC expression and subsequent efflux pump function.