ABSTRACT
The nonmevalonate pathway is the sole pathway for isoprenoid biosynthesis in Burkholderia cenocepacia and is possibly a novel target for the development of antibacterial chemotherapy. The goals of the present study were to evaluate the essentiality of dxr, the second gene of the nonmevalonate pathway, in B. cenocepacia and to determine whether interfering with the nonmevalonate pathway increases susceptibility toward antibiotics. To this end, a rhamnose-inducible conditional dxr knockdown mutant of B. cenocepacia strain K56-2 (B. cenocepacia K56-2dxr) was constructed, using a plasmid which enables the delivery of a rhamnose-inducible promoter in the chromosome. Expression of dxr is essential for bacterial growth; the growth defect observed in the dxr mutant could be complemented by expressing dxr in trans under the control of a constitutive promoter, but not by providing 2-C-methyl-d-erythritol-4-phosphate, the reaction product of DXR (1-deoxy-d-xylulose 5-phosphate reductoisomerase). B. cenocepacia K56-2dxr showed markedly increased susceptibility to the β-lactam antibiotics aztreonam, ceftazidime, and cefotaxime, while susceptibility to other antibiotics was not (or was much less) affected; this increased susceptibility could also be complemented by in trans expression of dxr. A similarly increased susceptibility was observed when antibiotics were combined with FR900098, a known DXR inhibitor. Our data confirm that the nonmevalonate pathway is essential in B. cenocepacia and suggest that combining potent DXR inhibitors with selected β-lactam antibiotics is a useful strategy to combat B. cenocepacia infections.
INTRODUCTION
Isoprenoids are a diverse class of naturally occurring compounds with widely varied roles in physiological processes in all living organisms. Animals (including humans), fungi, Archaea, and some bacteria synthesize isoprenoids using the mevalonate pathway (1). An alternative metabolic pathway for the biosynthesis of isoprenoids, the 1-deoxy-d-xylulose-5-phosphate (DOXP) pathway or nonmevalonate pathway, was discovered in the late 1980s (2). This pathway starts with the condensation of d-glyceraldehyde-3-phosphate and pyruvate to DOXP, catalyzed by the enzyme DOXP-synthase (DXS; in Burkholderia cenocepacia strain J2315, encoded by BCAM0911). Next, DOXP is converted to 2-C-methyl-d-erythritol-4-phosphate (MEP) by DOXP-reductoisomerase (DXR; encoded by BCAL2085). Subsequently, 4-diphosphocytidyl-2-C-methyl-d-erythritol (CDP-ME) is formed by CDP-ME synthase (IspD; encoded by BCAL2016). CDP-ME is then phosphorylated by CDP-ME kinase (IspE; encoded by BCAL0802), forming CDP-ME-2-phosphate, which is subsequently converted to 2-C-methyl-d-erythritol-2,4-cyclodiphosphate (MER-cPP) by MER-cPP synthase (IspF, encoded by BCAL2015). This cyclic diphosphate molecule is the precursor of isopentenyl pyrophosphate (IPP) and dimethylallyl pyrophosphate (DMAPP) (3, 4). The rate-limiting step in the nonmevalonate pathway is the conversion of DOXP to MEP, which is catalyzed by DXR. Fosmidomycin, a natural phosphonic acid antibiotic, and FR900098, an N-acetyl derivative of fosmidomycin, are effective inhibitors of purified DXR from several bacterial species, but often have limited effects on whole bacterial cells (5, 6).
Burkholderia cepacia complex (Bcc) bacteria are highly resistant to most antibiotics and are most known for their ability to cause life-threatening respiratory tract infections in cystic fibrosis (CF) patients (7). Bcc bacteria, including B. cenocepacia, use the nonmevalonate pathway as sole pathway for isoprenoid biosynthesis. As this pathway is not present in human cells, it is a possible novel target for the development of antibacterial chemotherapy against Bcc infections (8). DXR (encoded by BCAL2085) was predicted to be essential in B. cenocepacia (9); a more recent study predicted that all five genes of the nonmevalonate pathway would be essential in B. cenocepacia J2315 (10). This makes the enzymes of the nonmevalonate pathway interesting candidates as drug targets. However, resistance against fosmidomycin, FR900098, and several other fosmidomycin derivatives was previously observed in a large collection of Bcc strains, and could partly be explained by the lack of a glycerol-3-phosphate transporter (GlpT) system, which is required for the import of fosmidomycin and FR900098 (8). However, Bcc species possess an alternative GlpT system, which can be induced by the addition of glucose-6-phosphate, leading to increased uptake of fosmidomycin. Unfortunately, a fosmidomycin resistance gene (fsr), which is involved in efflux of fosmidomycin and FR900098, is upregulated in treated cultures and is able to compensate for the increased fosmidomycin uptake in the presence of glucose-6-phosphate (8, 11). Isoprenoids are precursors for the biosynthesis of hopanoids, which are involved in regulation of membrane stability, permeability, and fluidity, and which contribute to the intrinsic resistance of Bcc bacteria against various antibiotics, including polymyxins, erythromycin, and chlorhexidine (11–13). Deletion of genes (BCAM2831 and BCAS0167) involved in hopanoid biosynthesis (squalene-hopene cyclases) resulted in the loss of swarming and swimming motility, suggesting that hopanoids play an important role in the physiology of B. cenocepacia (13). Several studies provide evidence that reducing the amount of hopanoids in the membrane of Bcc bacteria can increase susceptibility to antibiotics (11–15). For example, Malott et al. described resistance of Burkholderia multivorans isolates to fosmidomycin and colistin when used alone, but they observed antimicrobial synergy when both antibiotics were used in combination (11).
The goals of the present study were to determine whether dxr is truly essential in B. cenocepacia and to investigate whether interfering with the nonmevalonate pathway increased the susceptibility of B. cenocepacia to antibiotics.
RESULTS AND DISCUSSION
Construction of conditional mutants of B. cenocepacia K56-2.We constructed a B. cenocepacia K56-2 mutant (designated B. cenocepacia K56-2dxr), in which the expression of dxr (homolog of BCAL2085) is controlled by the rhamnose-inducible promoter Prha. The correct alignment of Prha with dxr was verified with Sanger sequencing (see additional data set 1 in the supplemental material). Quantitative real-time PCR (qPCR) was used to determine dxr expression in K56-2 and K56-2dxr under various conditions (Fig. 1). In the absence of rhamnose, dxr expression in K56-2dxr is very low (but not completely switched off), while in the presence of rhamnose, the mRNA expression level was similar to that observed in the wild-type strain.
Relative mRNA expression of dxr measured by qPCR. Expression levels shown are relative to that of the expression observed in B. cenocepacia K56-2. Error bars represent range; n = 2.
Differences in planktonic growth between B. cenocepacia K56-2 and K56-2dxr.Growth curves of K56-2 and K56-2dxr obtained in a previously described basal salts medium (BSM-O) (16) in the presence and absence of rhamnose are shown in Fig. 2. K56-2dxr shows a growth defect, confirming that DXR is essential in K56-2, as previously suggested (9, 10). Growth was completely inhibited in K56-2dxr if the precultures used for the growth curves were grown in the absence of rhamnose, but growth resumed after a delay. Consecutive batch cultures in medium without rhamnose, starting from a preculture grown in the presence of rhamnose, show that growth is only inhibited after several generations. The growth rate then recovers to near wild-type levels upon further cultivation, and remains at this level (see Fig. S1 in the supplemental material). This indicates that, after withdrawal of rhamnose, levels of dxr mRNA and/or DXR protein initially remain sufficient to sustain growth. However, after several generations the protein is depleted, leading to complete growth inhibition. This is followed by selection of suppressor mutants, which grow at rates nearly identical to that of the wild type. The suppressor mutant(s) are most likely “false suppressor mutants,” in which the rhamnose-inducible promoter and/or its regulators are mutated in such a way that the conditional dxr expression is converted to constitutive expression (17). Similar growth profiles were observed in other media (see Fig. S2 in the supplemental material), confirming that the phenotype is not specific for BSM-O medium. While we did not evaluate the effect of lack of dxr expression in vivo, it is worth mentioning that in a synthetic cystic fibrosis medium (18), a growth defect was also observed, suggesting that under the nutritional conditions encountered in the lungs of CF patients, dxr expression is important.
Growth of B. cenocepacia K56-2 wild type and the K562dxr conditional mutant in permissive and nonpermissive conditions. Top panels: planktonic growth. Solid squares, K56-2dxr; open circles, K56-2 wild type; orange, BSM-O with 0.5% glucose; blue, BSM-O with 0.5% rhamnose. Absence of rhamnose inhibits growth of K56-2dxr, but growth resumes after a delay (left). Addition of rhamnose enables growth of the conditional mutant to near wild-type levels. Growth inhibition in the presence of glucose is less pronounced if precultures are grown in the presence of rhamnose (right). Growth curves are representative of three biological replicates, performed in 96-well round bottom microtiter plates. Bottom panel: colony formation on agar plates. A, K56-2 + pSCrhaB2; B, K56-2dxr. Cultures grown in BSM-O with 200 mg/liter trimethoprim without addition of sugars were normalized to an optical density at 590 nm (OD590) of 1.0, and then a 1:10 dilution series was performed. Aliquots of 15 μl of the diluted cultures were spotted onto BSM-O (with 200 mg/liter trimethoprim) with 0.5% rhamnose (left) or 0.5% glucose (right).
The growth defect in K56-2dxr could be complemented by expression of dxr in trans from a constitutive promoter (Fig. 3). However, growth could not be restored by adding MEP, the reaction product of DXR (highest concentration tested, 5 mM) (Fig. 4); this is likely due to a lack of uptake of MEP.
Expression of dxr in trans complements the growth defect observed in B. cenocepacia K56-2dxr. Growth medium, LBB with 200 mg/liter trimethoprim. Orange, 0.5% glucose; blue, 0.5% rhamnose. pDA-dxr expresses dxr constitutively. Growth curves represent the average of 4 biological replicates, performed in 96-well round bottom microtiter plates.
Growth of B. cenocepacia K56-2dxr in the presence of 2-C-methyl-d-erythritol 4-phosphate (MEP). Growth medium, BSM-O. Solid squares, K56-2dxr; open circles, K56-2 wild type; orange, 0.5% glucose; blue, 0.5% rhamnose; yellow, 5 mM MEP.
Differences in in vitro antimicrobial susceptibility.MICs of fosmidomycin and FR900098 were >500 μg/ml for B. cenocepacia K56-2 and K56-2dxr, irrespective of whether rhamnose was present. This is in agreement with concentrations previously reported for B. cenocepacia strain LMG 16656 (8). Subsequently, we determined the MIC for a range of antibiotics, for B. cenocepacia K56-2, B. cenocepacia K56-2 in the presence of subinhibitory concentration of FR900098 (250 μM), B. cenocepacia K56-2dxr, B. cenocepacia K56-2dxr in the presence of rhamnose, and for the complemented strain (Table 1). Deletion of dxr resulted in an increased susceptibility toward all β-lactam antibiotics tested, with the most pronounced differences for aztreonam (16-fold difference), the cephalosporin ceftazidime (16-fold), and cefotaxime (32-fold). Although differences in MIC were also observed for the carbapenems tested, these were lower (2-fold for meropenem and 4-fold for imipenem). For the other antibiotics, difference in MIC between the wild type and the dxr mutant were either small (bacitracin, chloramphenicol, ciprofloxacin, tetracycline, and rifampin) or no difference was observed (erythromycin and tobramycin). For the wild-type strain, the addition of rhamnose did not change the MIC of any of the antibiotics tested; the addition of rhamnose to the dxr mutant returned all MICs to wild-type levels, indicating that the reduced expression of dxr was the cause for the MIC changes (Table 1). Expressing dxr in trans under the control of a constitutive promoter returned MIC levels of antibiotics (with the exception of bacitracin) for the dxr mutant back to wild-type levels. Interestingly, when the DXR inhibitor FR900098 was added along with the antibiotics, MICs for the β-lactam antibiotics were (much) lower and in the same range as that observed for the dxr mutant (Table 1). Compared to FR900098, the potentiating effect of fosmidomycin was less pronounced (see Fig. S3 in the suplemental material), which is in line with the previously observed lower antimicrobial activity of the latter (8).
MIC values of antibiotics belonging to different classes for B. cenocepacia K56-2, B. cenocepacia K562dxr, and the complemented mutanta
The effect of reduced dxr expression or of adding FR900098 was most pronounced for aztreonam, ceftazidime, and cefotaxime. Aztreonam and cefotaxime both target penicillin-binding protein (PBP) 3 in Escherichia coli (19, 20). In Burkholderia pseudomallei, the deletion of BSSP1219, the gene encoding a PBP3, results in ceftazidime resistance, suggesting that ceftazidime targets PBP3; a homolog of BSSP1219 is present in B. cenocepacia (21). PBP3 is involved in peptidoglycan synthesis, as it is part of the peptidoglycan synthesis core of the divisome, and it modulates the interaction of FtsW with lipid II (22). Lipid II [undecaprenyl-pyrophosphoryl-MurNAc-(pentapeptide)-GlcNAc] is the precursor for peptidoglycan synthesis, and the lipid undecaprenyl phosphate (C55-P) is synthesized from C5 isopentenyl units (23), which are in turn produced by the nonmevalonate pathway (24). Interestingly, reduced dxr expression and the addition of the DXR inhibitor FR900098 also reduced the MIC for bacitracin (by 2- and 4-fold, respectively). Bacitracin inhibits cell wall formation by binding to C55-isoprenyl pyrophosphate (C55-IPP), which is responsible for transporting peptidoglycan precursors across the bacterial cell membrane (25). Our data suggest that reducing dxr expression in B. cenocepacia K56-2 or adding FR900098 (partially) inhibits cytoplasmic steps of peptidoglycan biosynthesis, leading to lower levels of C55-P and C55-IPP and increased susceptibility to antibiotics that target interaction partners of C55-P and C55-IPP. Carbapenems, such as meropenem and imipenem, target PBP2 (26). In contrast to PBP3 (which is part of the divisome), PBP2 is part of the elongasome (responsible for elongation of rod-shaped cells) and interacts with several other proteins in the Rod system (27). The different targets of the β-lactam antibiotics tested (PBP2 versus PBP3) could explain the different effects of reduced dxr expression and the potentiating activity of FR900098. The genome of B. cenocepacia strain J2315 (very similar to that of strain K56-2) encodes two PBP3s (BCAM1362 and BCAL3468 genes, 49% similarity); this likely explains the strain's extreme resistance to aztreonam (MIC, >2,048 μg/ml), ceftazidime (MIC, 128 μg/ml), and cefotaxime (MIC, 2,048 μg/ml) observed in B. cenocepacia (in contrast, for Pseudomonas aeruginosa, EUCAST clinical breakpoints for aztreonam and ceftazidime are 16 and 8 μg/ml, respectively).
Conclusions.Our data confirm that dxr, and by extension the nonmevalonate pathway, is essential in B. cenocepacia and is a potentially interesting drug target. Interfering with this pathway, either by reducing the expression of dxr (coding for the key enzyme DXR) or by inhibiting DXR activity using the acetylated fosmidomycin derivative FR900098, drastically reduced the MIC of certain β-lactam antibiotics. This was most obvious for β-lactam antibiotics that target interaction partners of C55-P and C55-IPP, explaining the link between reduced isoprenoid synthesis and resistance to these antibiotics. High concentrations of aztreonam (up to 985 mg/kg sputum) can be obtained after inhalation therapy (28) and these concentrations are above the MIC that is obtained in the wild type in combination with FR900098. It will be interesting to explore whether the combination of carefully selected β-lactam antibiotics and DXR inhibitors could provide an alternative strategy to combat B. cenocepacia infections.
MATERIALS AND METHODS
Bacterial strains, plasmids, and growth conditions.Bacterial strains and plasmids used in this study are listed in Table 2. E. coli and B. cenocepacia strains were grown in Luria-Bertani agar or broth (LBA or LBB) or in more defined and phosphate-buffered media, BSM-O or M9-O (16). BSM-O is based on a basal salt medium (18.6 mM K2HPO4, 7.2 mM NaH2PO4, 37.4 mM NH4Cl, 0.1 g/liter nitrilotriacetic acid, 0.2 g/liter MgSO4·7 H2O, 0.012 g/liter FeSO4·7 H2O, 0.003 g/liter MnSO4·H2O, 0.003 g/liter ZnSO4·7 H2O, and 0.001 g/liter CoSO4·7 H2O) (29) and supplemented with yeast extract (5 g/liter), Casamino Acids (2 g/liter), and 0.5% (wt/vol) glycerol (16). M9-O is based on M9 (42 mM Na2HPO4, 22 mM KH2PO4, 8 mM NaCl, and 10 mM NH4Cl) and was supplemented with the same organic substrates as those for BSM-O. Media were supplemented with 0.5% (wt/vol) rhamnose or glucose, 50 μg/ml gentamicin (Gen), 40 μg/ml kanamycin (Kan), trimethoprim (Tmp, 50 or 200 μg/ml), or tetracycline (Tet; 20 or 200 μg/ml) (all from Sigma-Aldrich) as required. Liquid cultures were grown in a shaking incubator at 37°C.
Bacterial strains and plasmids
Construction of conditional dxr mutant.A conditional knockdown mutant was constructed by replacing the native promoter of the target gene dxr (BCAL2085 in B. cenocepacia J2315) with the plasmid-borne rhamnose-inducible promoter Prha (16, 30), using homologous recombination. Primers are listed in Table 3. The genome contigs of B. cenocepacia K56-2 were obtained from the Burkholderia Genome Database (http://www.burkholderia.com/; whole genome shotgun sequence of Burkholderia cenocepacia K56-2Valvano, RefSeq NZ_ALJA02000015). Primers DXR F1/DXR R2 were designed to amplify a 283-nucleotide (nt) fragment spanning the 5′ region of dxr. The fragment was digested with restriction enzymes XbaI and NdeI (Promega) and ligated into digested pSC200 downstream of Prha, resulting in plasmid pSC200dxr. The recombinant plasmid was transformed into E. coli DH5α λpir with the heat shock method using LBA supplemented with 50 μg/ml Tmp and 0.5% glucose as selective agar. Exconjugants were screened for presence of insert by PCR using primers rhaF/200-MCS-R. The recombinant plasmid was transferred into B. cenocepacia K56-2 by triparental mating (16). Overnight liquid cultures of B. cenocepacia K56-2, E. coli HB101 (pRK2013), and E. coli donor strain were mixed in equal volumes, and 300 μl of the mix was spotted onto LBA with 0.5% rhamnose and incubated at 37°C for 6 h. Cells were washed off the agar with physiological saline and plated on LBA containing 50 μg/ml Gen (to select against E. coli), 200 μg/ml Tmp and 0.5% rhamnose. Plates were incubated at 37°C for 48 h, and exconjugants were screened for correct insertion of the plasmid by PCR using primers rhaF/DXR R4. A successful exconjugant (here referred to as K56-2dxr) was subcultured on LBA plus 200 μg/ml Tmp and 0.5% rhamnose, and the respective PCR product was Sanger sequenced (GATC Biotech, Additional data set 1).
Primers
To allow growth experiments on LBB with Tmp, B. cenocepacia K56-2 wild type was transformed with a replicative plasmid (pSCrhaB2) carrying the dhfr gene (31).
Construction of B. cenocepacia K56-2dxr expressing dxr in trans.To complement the dxr mutant, a modified version of pDAI-SceI, a plasmid used for constitutive I-SceI (homing endonuclease) expression from the constitutive P1 promoter for dhfrIIb (32), was used. I-SceI was excised using NdeI and XbaI and the plasmid was gel cleaned. The linear plasmid was then either blunted and self-ligated, giving rise to the vector control plasmid pDA−, or ligated with the dxr gene, amplified from the B. cenocepacia K56-2 genome using primers dxr-exprF and dxr-exprR and restricted with NdeI and XbaI, giving rise to pDA-dxr. Correct insertion of dxr was screened by PCR using primers pDA-F and rha1R. Plasmids pDA− and pDA-dxr were transformed into K56-2dxr by triparental mating, using LBA with 200 Tet, 200 Tmp, 50 Gen, and 0.5% rhamnose as selective medium. dxr on pDA-dxr of a successful exconjugant of K56-2dxr was Sanger-sequenced (GATC Biotech, Additional data set 1).
RNA extraction and cDNA synthesis.Cells of wild-type and mutant B. cenocepacia K56-2 were cultured in either M9-O or LBB. Both media were supplemented with 0.5% rhamnose or glucose as required. After 24 h incubation at 37°C, cells were collected by centrifugation and RNA was extracted using the RiboPure bacteria kit (Ambion) according to the manufacturer's instructions. RNA concentrations were determined using a BioDrop μlite (Isogen Life Science), and 500 ng of each RNA sample was used to synthesize cDNA using the qScript cDNA synthesis kit (QuantaBio).
Quantitative real-time PCR (qPCR).qPCR experiments were performed on a Bio-Rad CFX96 real-time PCR detection system using the Perfecta SYBR green Fastmix (QuantaBio), according to the manufacturer's instructions. The PCR cycling protocol included an initial denaturation at 95°C for 3 min, followed by 40 amplification cycles consisting of 15 s at 95°C and 1 min at 60°C; a melting curve analysis was performed at the end of each run. All PCR data were normalized to the mean expression value of three reference genes (BCAS0175, BCAM2784, and BCAL2694) that were stably expressed under all conditions (33).
Growth curves.Bacterial cultures were grown overnight in LBB supplemented with 200 μg/ml Tmp and 0.5% rhamnose as required and subsequently diluted to a final inoculum of 5 × 105 CFU/ml in various growth media, including LBB, BSM-O, M9-O, M9 with 0.5% glucose, tryptic soy broth (TSB), Mueller-Hinton broth (MHB), and synthetic cystic fibrosis medium (SCFM; [18]). 2-C-Methyl-d-erythritol 4-phosphate lithium salt (MEP) was purchased from Sigma-Aldrich and added to a final concentration of 5 mM. Glucose and rhamnose were added to a final concentration of 0.5% as required. A total of 200 μl of each bacterial suspension was added in duplicate to a flat-bottomed 96-well microtiter plate (SPL Life Sciences), and bacterial growth was monitored by measuring absorbance at 590 nm in an Envision multilabel plate reader (PerkinElmer, Waltham, MA), every 30 min for 48 to 72 h at 37°C, until the stationary phase was reached. Growth curves for consecutive reinoculated batch cultures were performed in a Cell Growth Quantifier (Aquila Biolabs).
Evaluation of in vitro antimicrobial susceptibility.Antimicrobial susceptibility was investigated by determination of the MIC according to the EUCAST broth microdilution guidelines (34). Antibiotics tested included bacitracin, cefotaxime, ceftazidime, chloramphenicol, ciprofloxacin, erythromycin, imipenem, rifampin, tetracycline, and FR900098 (Sigma-Aldrich), meropenem (Hospira Benelux), tobramycin, aztreonam (TCI Europe), and fosmidomycin (Invitrogen). MICs were determined in triplicate using flat-bottomed 96-well microtiter plates, with concentrations adapted to the high antimicrobial resistance of B. cenocepacia. Planktonic bacterial cultures were grown overnight in the appropriate medium and added to the 96-well plates at a final inoculum of 5 × 105 CFU/ml. Plates were incubated for 24 h at 37°C, and absorbance was measured at 590 nm using an Envision plate reader. Initial tests for the potentiating effect of fosmidomycin and FR900098 were performed in MHB, with concentrations of potentiator ranging from 4 to 500 μM. Further MIC analyses were then conducted with FR900098 at a concentration of 250 μM, because this showed the highest effect on MIC values, and in LBB, because the complemented mutants did not grow in MHB. Neither Tmp nor Tet was added to the MIC tests.
ACKNOWLEDGMENT
The authors acknowledge funding support by the Research Foundation—Flanders (FWO). The funder had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
FOOTNOTES
- Received 22 December 2017.
- Returned for modification 23 January 2018.
- Accepted 7 February 2018.
- Accepted manuscript posted online 12 February 2018.
Supplemental material for this article may be found at https://doi.org/10.1128/AAC.02607-17.
- Copyright © 2018 American Society for Microbiology.
