|Home | About | Journals | Submit | Contact Us | Français|
The transglycosylation step of cell wall synthesis is a prime antibiotic target because it is essential and specific to bacteria. Two antibiotics, ramoplanin and moenomycin, target this step by binding to the substrate lipid II and the transglycosylase enzyme, respectively. Here, we compare the ramoplanin and moenomycin stimulons in the Gram-positive model organism Bacillus subtilis. Ramoplanin strongly induces the LiaRS two-component regulatory system, while moenomycin almost exclusively induces genes that are part of the regulon of the extracytoplasmic function (ECF) σ factor σM. Ramoplanin additionally induces the ytrABCDEF and ywoBCD operons, which are not part of a previously characterized antibiotic-responsive regulon. Cluster analysis reveals that these two operons are selectively induced by a subset of cell wall antibiotics that inhibit lipid II function or recycling. Repression of both operons requires YtrA, which recognizes an inverted repeat in front of its own operon and in front of ywoB. These results suggest that YtrA is an additional regulator of cell envelope stress responses.
The bacterial cell wall is a unique and vital molecular sieve. It provides the cell with structural strength and protects it from lysis due to high turgor pressures. This makes the cell wall an important target for many antimicrobial compounds (43). The major component of the cell wall is peptidoglycan (PG), an alternating polymer of the amino sugars N-acetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc). GlcNAc and MurNAc are synthesized inside the cell and incorporated, together with a pentapeptide side chain, into the lipid-linked PG precursor designated lipid II. After lipid II is translocated to the outside cell, the GlcNAc-MurNAc-pentapepide portion is inserted and cross-linked to the existing PG by the transglycosylase (TG) and transpeptidase (TP) activities of high-molecular-weight penicillin binding proteins (HMW PBPs).
The TG and TP reactions are both excellent drug targets, because they are essential, are easily accessible, and have no equivalent in eukaryotic cells. The TP reaction is specifically targeted by β-lactam antibiotics, which covalently modify the enzyme active site. The TG reaction is inhibited by several classes of antibiotics targeting either the substrate or the enzyme. Moenomycin (MOE) is a glycolipid that targets the active site of the TG enzyme (23). Among the TG inhibitors that bind lipid II, the glycopeptides vancomycin, teicoplanin, and ristocetin bind to the terminal d-Ala-d-Ala of the pentapeptide side chain (21, 30). The glycolipodepsipeptide ramoplanin sequesters lipid II at the interface between the extracellular environment and the bacterial membrane and binds the reducing end of the nascent glycan chain (9, 12, 13). The lantibiotic nisin recognizes the diphospho-sugar portion of lipid II (17). Bacitracin, a cyclic dodecylpeptide antibiotic, binds to the undecaprenyl pyrophosphate released after the TG reaction and thereby inhibits the recycling of the key undecaprenyl phosphate lipid carrier molecule (37, 38).
The exposure of bacteria to cell wall-active antibiotics often triggers a transcriptional stress response. The best characterized responses involve the activation of extracytoplasmic function (ECF) σ factors, general stress response σ factors, and two-component regulatory systems (TCS). Understanding how these systems individually and collectively mediate stress responses is critical for defining how bacteria sense and adapt to antibiotics. In Bacillus subtilis, there are seven ECF σ factors (σM, σW, σX, σY, σV, σZ, and YlaC), one general stress σ factor (σB), and five cell wall stress-related TCS (LiaRS, BceRS, PsdRS, YxdKJ, and YycFG) (20, 36).
In previous studies, we and others have explored the stress responses induced when B. subtilis is treated with a variety of cell envelope-active compounds. For example, in response to vancomycin, ~100 genes were induced within 3 min of exposure (5). Most of these induced genes are controlled by σWand σM. Similarly, treatment with bacitracin, nisin, and ramoplanin strongly induced LiaRS and its regulon liaIHGSF (14, 27, 28). Null mutations in the induced regulators or regulon members often, but not always, result in higher susceptibility to antibiotics. For example, inactivation of sigW results in a higher susceptibility to some cell envelope-active antibiotics, including vancomycin (26) and fosfomycin (3). The σW regulon is particularly important for protection against membrane-active compounds, due in part to σW-dependent remodeling of membrane composition (22). Conversely, in other cases, antibiotic-inducible genes do not confer an obvious protective effect. For example, no susceptibility changes were observed with the liaIH-null mutant when treated with either vancomycin or nisin, both of which strongly induce these genes (14, 44). This may be due, in part, to functional redundancy. Increases in antibiotic susceptibility are sometimes revealed only when multiple stress response pathways are inactivated. Well-characterized examples include strains deleted for either three (σX, σW, and σM) or all seven of the ECF σ factors of B. subtilis (24, 26).
Although induction of ECF σ factors and activation of TCS can account for much of the observed cell envelope stress response, the regulatory pathways controlling some strongly induced genes have not been defined. One example is the ytrABCDEF operon (referred to here as the ytrA operon), whose expression is strongly induced by vancomycin (5) and bacitracin (27). Previously, the ytrA operon was proposed to encode a putative ABC transport system repressed by YtrA and involved in utilization of acetoin, a secreted metabolite resulting from carbon overflow metabolism that accumulates in stationary-phase cultures (45). This assignment was based on a slight decrease in the rate of acetoin reutilization in a ytrA operon deletion. However, the ytrA operon was not induced by acetoin (45), and subsequent studies indicate that acetoin catabolism is determined by the carbon metabolite-repressed and acetoin-inducible acoABCL operon (1, 35, 41). Conversely, induction of the ytrA operon was noted in previous global analyses of antibiotic stress responses, and ytrA was even proposed as a reporter for glycopeptide antibiotics (18).
Here, we compared the stimulons of ramoplanin (RAM) and moenomycin (MOE) in B. subtilis. Both antibiotics target the TG reaction, but they trigger distinct stress responses: RAM strongly induced the LiaRS TCS as well as the ytrA and ywoB operons, while MOE almost exclusively induced the σM regulon. We further demonstrate that YtrA binds as a repressor to inverted repeats in the regulatory regions of both the ytrA and ywoB operons and is required for induction in response to antibiotic stress.
B. subtilis strains used are derivatives of either W168 (trpC2) or CU1065 (trpC2 attSPβ). Escherichia coli strain DH5α was used for standard cloning procedures. Bacteria were grown in Luria-Bertani (LB) medium at 37°C with vigorous shaking. Antibiotics were added to the growth medium when appropriate: 100 μg/ml ampicillin and 34 μg/ml chloramphenicol for E. coli and 1 μg/ml erythromycin plus 25 μg/ml lincomycin (macrolide-lincomycin-streptogramin B [MLS] resistance), 10 μg/ml chloramphenicol, 100 μg/ml spectinomycin, and 10 μg/ml kanamycin for B. subtilis. Optical density at 600 nm (OD600) readings were taken on a Spectronic 21 (Milton Roy) spectrophotometer.
Cultures were grown to mid-log phase (OD600 of 0.4) and split into two flasks. One culture was treated with an antibiotic (5 μg/ml RAM for 10 min or 1 μg/ml MOE for 15 min), and the other was an untreated control. RAM was obtained from Stefano Donadio (Vicuron) as previously reported (28), and 5 μg/ml induces the liaI operon strongly but does not lead to cell lysis over this time frame (28). MOE was obtained from Biovet (Peshtera, Bulgaria), and the amount used is 10-fold below the MIC but is sufficient to inhibit growth of wild-type (WT) cells, as indicated by an increased lag phase. RNA isolation and microarray analysis was performed as previously described for RAM (27) and for MOE (11). Each microarray was performed three times with biological triplicates. The fold induction values were calculated by using the signal intensity values of treated samples divided by those of untreated samples.
Susceptibility tests to antibiotics/chemicals were conducted using disk diffusion assays and MIC tests. Disk diffusion assays were performed with Mueller-Hinton agar as previously described (24). We used BBL Sensi-Disc susceptibility test disks (BD; azithromycin, cefoperazone, ceftriaxone, meropenem, oxacillin, piperacillin, amoxicillin-clavulanic acid, and isoniazid) and prepared disks made with fresh stocks and Whatman paper disks (7-mm diameter) (aztreonam, 30 μg; cefuroxime, 30 μg; d-cycloserine, 300 μg; bacitracin, 150 μg; fosfomycin, 250 μg; vancomycin, 30 μg; RAM, 50 μg; MOE, 7.5 μg; tunicamycin, 50 μg; lysozyme, 500 μg; mutanolysin, 500 μg; daptomycin, 3 μg; polymyxin B, 250 μg; defensin HNP-1, 1.25 μg; defensin HNP-2, 1.25 μg; nalidixic acid, 250 μg; novobiocin, 250 μg; ciprofloxacin, 30 μg; ofloxacin, 150 μg). MIC was determined by diluting overnight cultures 1:100, growing them to an OD600 of 0.4, and diluting them to 5 × 105 CFU/ml in microtiter plates with a total inoculum of 200 μl. Growth was measured spectrophotometrically (OD600) using a Bioscreen incubator (Growth Curves USA, Piscataway, NJ) at 37°C with vigorous shaking. The absorbance was recorded every 20 min for 24 h. Inhibition was defined as a final OD600 of <0.05 (at the 24-h time point). With MOE, MIC measurements are impractical since higher concentrations of antibiotic lead to a lengthened lag phase but do not prevent growth. After 24 h, the wild-type strain was able to grow even with the addition of 2,000 μg/ml MOE. Here, to compare different mutants, we defined the MIC for MOE as the lowest concentration that inhibits cell growth for 10 h (OD < 0.05).
Gene deletions were created by replacing the coding region with an antibiotic resistance cassette. Long flanking homology PCR (LFH-PCR) followed by DNA transformation was performed as described previously (27) with the following changes: flanking fragments were amplified using iProof DNA polymerase (Bio-Rad), and the flanking fragments and antibiotic resistance marker were joined using the Expand long-template PCR system (Roche). A detailed protocol is available at http://www.micro.cornell.edu/cals/micro/research/labs/helmann-lab/supplements.cfm. The ytrA-null mutant was created using the pMAD shuttle vector as described previously (2, 25). The primers were 3676, 3677, 3678, and 3679 (see Table S1 in the supplemental material). The ytrA deletion was verified by PCR amplification followed by DNA sequencing (Cornell DNA sequencing facility).
DNA fragments containing the ytrA and ywoB regulatory regions were PCR amplified (primers 3700 and 3701 for PytrA; 3734 and 3735 for PywoB). The products were digested with EcoRI and BamHI and cloned into pDG1661 containing a promoterless lacZ gene (10), resulting in plasmids pLS33 and pLS34, respectively. The sequences of the inserts were verified by DNA sequencing (Cornell DNA sequencing facility). B. subtilis strains were transformed with the ScaI-linearized plasmids, which integrate into the amyE locus. For quantitative measurements of β-galactosidase activity, cells were grown in LB medium at 37°C with vigorous shaking, and samples were collected at different OD600 points. To test promoter induction, cells were grown in LB to an OD600 of ~0.4, and then the culture was split into aliquots, which were challenged with vancomycin (2 μg/ml final), bacitracin (20 μg/ml final), or RAM (10 μg/ml final). The cultures were returned to 37°C, and samples were collected after 30 min. β-Galactosidase activity was measured according to the method of Miller (29), except that cells were lysed by the addition of lysozyme (20 μg/ml, final) and a 30-min incubation at 37°C.
An ectopic integration of a PytrA-lacZ fusion was constructed based on the vector pAC6 (39). Promoter fragments of increasing lengths were generated by PCR, using primers 1007 to 1008 (PytrA forward primers for −289 and −245 fragments, respectively) and primer 1011 (PytrA reverse primer). Standard cloning techniques were applied (34). The insert was verified by DNA sequencing. The constructed pAC6-derived plasmids (Table 1) were linearized with ScaI and used to transform B. subtilis with chloramphenicol selection, resulting in strains TMB547 and TMB548.
For quantitative measurements of β-galactosidase activity, cells of the reporter strains were grown in LB medium at 37°C with agitation until they reached an OD600 of ~0.45. The culture was split, and vancomycin (final concentration of 1 μg/ml) was added to one half, leaving the other half untreated (uninduced control). Both cultures were incubated for 30 min at 37°C. Cell pellets were resuspended in 1 ml of working buffer (20 mM β-mercaptoethanol, 60 mM Na2HPO4, 40 mM NaH2PO4, 10 mM KCl, 1 mM MgSO4, pH 7.0) and assayed for β-galactosidase activity as described elsewhere, with normalization to cell density (29).
Strain HB5485 (CU1065 containing an in-frame deletion of ytrA) was grown in LB (37°C with vigorous shaking) to mid-exponential stage (OD600 of ~0.4). RNA was isolated using the RNeasy minikit (Qiagen), followed by DNase treatment with TURBO DNA-free (Ambion) and ethanol precipitation overnight. The RNA was dissolved in RNase-free water and quantified using a NanoDrop spectrophotometer (NanoDrop Technologies, Inc., Wilmington, DE). Two micrograms of total RNA was used with the 5′ random amplification of cDNA ends (RACE) kit (Invitrogen) according to the manufacturer's specifications (primers 4155 and 4156). Resulting DNA fragments were sequenced at the Cornell DNA sequencing facility.
Hierarchical clustering for whole-genome microarray data sets was performed using Cluster 3.0 (6). Transcriptome data sets were derived from results from our lab and other published studies (7, 19, 27). The data sets represent the following treatment conditions and times: 16 μg/ml amoxicillin, 10 min; 4 μg/ml penicillin G, 10 min; 0.25 μg/ml cephalexin, 10 min; 1 μg/ml cefotaxime, 10 min; 0.5 μg/ml cefoxitin, 10 min; 16 μg/ml d-cycloserine, 10 min; 0.25 μg/ml oxacillin, 10 min; 256 μg/ml fosfomycin, 10 min; 100 μg/ml bacitracin, 5 min; 0.5 μg/ml ristocetin, 10 min; 2 μg/ml vancomycin, 10 min; 5 μg/ml RAM, 10 min; 1 μg/ml MOE, 20 min; 64 μg/ml polymyxin B, 10 min; 1 μg/ml daptomycin, 20 min; 0.03 μg/ml gramicidin A, 10 min; 0.125 μg/ml monensin, 10 min; 0.008 μg/ml nigericin, 10 min; 64 μg/ml Triton X-114, 10 min; 0.25 μg/ml novobiocin, 10 min; 0.5 μg/ml norfloxacin, 10 min; 8 μg/ml nalidixic acid, 10 min; 0.5 μg/ml ciprofloxacin, 10 min; 4 μg/ml chloramphenicol, 10 min; 128 μg/ml spectinomycin, 10 min; 64 μg/ml puromycin, 10 min; and 0.5 μg/ml tetracycline, 10 min. The log10 ratios of fold changes comparing the treated versus untreated samples of B. subtilis wild-type lab strains 168 or CU1065 were used in the cluster analysis. The result was visualized using TreeView (8).
YtrA was purified using E. coli strain HE5322, BL21(DE3) pLysS (Novagen) containing pLS36 (a pET11a derivative, which places ytrA under the control of a T7lac promoter). A 1-liter culture was grown from a fresh colony of HE5322 in LB broth containing 0.5% glucose, 100 μg/ml ampicillin, and 34 μg/ml chloramphenicol to enhance plasmid stability. At an OD600 of 0.4, isopropyl-β-d-thiogalactopyranoside (IPTG; Gold BioTechnology) was added to 1 mM, and growth continued for 2 h. Cell pellets were resuspended in 10 ml resuspension buffer (40 mM Tris-HCl [pH 8.0], 1 mM EDTA, 5% [vol/vol] glycerol, and 1 mM dithiothreitol [DTT]) to which 1 tablet of Complete Mini EDTA-free protease inhibitor cocktail (Roche) was added. The cell suspension was disrupted by freeze-thawing followed by pulsed sonication and clarified by 15-min centrifugation at 16,000 × g. The resulting supernatant was applied to a heparin column (HiTrap Heparin HP; 10 ml; GE Healthcare). Bound proteins were eluted with an NaCl gradient (0.2 to 1 M) in elution buffer (20 mM Tris-HCl [pH 8.0], 1 mM EDTA, 5% [vol/vol] glycerol, and 1 mM DTT) and YtrA-containing fractions identified by Coomassie staining after 15% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Pooled YtrA was then loaded onto a Bio-Rad Bio-Scale Mini Macro-Prep High S cartridge, and proteins were eluted with a 0 to 1.0 M NaCl gradient in elution buffer (20 mM Tris-HCl [pH 7.0], 1 mM EDTA, 5% [vol/vol] glycerol, and 1 mM DTT). YtrA-containing fractions were pooled and dialyzed into storage buffer (20 mM Tris-HCl [pH 8.0], 1 mM EDTA, 1 mM DTT, and 50% glycerol) for storage at −20°C.
PCR fragments containing either PytrA (274 bp; primers 3700 and 3701), PywoB (192 bp; primers 3734 and 3735), or a nonspecific control (PyoeB; 106 bp) were end labeled with T4 polynucleotide kinase (PNK; New England BioLabs) and [γ-32P]ATP (6,000 Ci [222 TBq]/mmol; PerkinElmer). Increasing amounts of YtrA (0 to 160 nM) were incubated at room temperature for 15 min with the labeled promoter fragments in 10 μl binding buffer (20 mM Tris-HCl [pH 8.0], 5% [vol/vol] glycerol, 50 mM NaCl, 50 μg/ml bovine serum albumin [BSA], and 5 μg/ml sheared salmon sperm DNA). Samples were loaded on a 6% polyacrylamide gel prepared and run in 90 mM Tris-borate buffer with 2 mM EDTA. The gel was dried and imaged on a Storm 840 PhosphorImage scanner (Molecular Dynamics) after overnight exposure to a PhosphorImage screen.
Primers 4223 and 4226 were end labeled with T4 PNK and [γ-32P]ATP and used with unlabeled primers 4224 and 4225 to generate PCR-amplified 444-bp PytrA and 329-bp PywoB fragments. Binding reactions (50 μl) contained 1× binding buffer (40 mM Tris-HCl [pH 8.0], 10 mM MgCl2, 5 mM DTT, 50 mM NaCl, 5 μg/ml BSA, and 5 μg/ml sheared salmon sperm DNA), labeled DNA, and purified YtrA (0 to 8 nM). Reaction mixtures were incubated at room temperature for 15 min and then digested by the addition of RQ1 DNase I (Promega) with 5 mM CaCl2 for 3 min. Seven-hundred microliters of stop solution (650 μl of 100% ethanol, 50 μl of 3 M sodium acetate, 10 μl of 0.5-mg/ml yeast RNA) was added followed by precipitation at −20°C. Samples were recovered by microcentrifugation for 30 min, washed with 750 μl of 70% ethanol, and centrifuged briefly. Samples were resuspended in formamide loading buffer, boiled for 5 min at 95°C, and loaded onto a 6% sequencing gel. The G+A ladder was generated by incubating (25 min at 37°C) ~100,000 cpm of DNA with 2 μl of 1 M formic acid (pH 2.0) and 2 μl of 1-mg/ml sheared salmon sperm DNA in 20 μl. The reaction mixture was placed on ice, 150 μl piperidine solution (135 μl H2O, 15 μl stock piperidine) was added, and incubation continued at 95°C for 30 min, followed by 5 min on ice. One milliliter of n-butanol was added and mixed by vortexing, and then the mixture was centrifuged for 2 min, and supernatant was carefully removed. The pellet was washed first with 150 μl of 1% SDS and 1 ml of n-butanol and then with 0.5 ml of n-butanol. The pellet was then dried under vacuum for 10 min, suspended in 20 μl formamide loading buffer, and stored at −20°C. The gel was dried and imaged on a Storm 840 PhosphorImage scanner (Molecular Dynamics) after overnight exposure of a PhosphorImage screen.
The complete sets of raw and normalized data for each of the triplicate DNA microarray experiments involving B. subtilis treated with MOE and RAM are available in the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/) under accession numbers GSE30001 and GSE30002, respectively.
Both RAM and MOE target the TG activity required for PG synthesis, with the former binding to the substrate (lipid II) and the latter to the enzyme (HMW PBP) active site (12). Previously, we noted that RAM was a strong inducer of the LiaRS TCS as judged by monitoring transcription of the LiaR-activated liaI promoter (28). Maximal induction (>400-fold) was obtained with 5 μg/ml RAM, while cell lysis was observed after 1 to 2 h with concentrations of 20 μg/ml and higher. In parallel with these earlier studies, we therefore prepared RNA from cells treated for 10 min with 5 μg/ml RAM for transcriptome analysis using DNA microarrays. Since RAM binds to lipid II and thereby inhibits the TG reaction of peptidoglycan synthesis, we anticipated that the resulting stress responses might be a composite response reflecting the results of both lipid II sequestration and TG inhibition. Conversely, since MOE is a TG inhibitor that does not interact with lipid II (and instead binds to the enzyme active site), we reasoned that a comparison of the transcriptomes elicited by MOE and RAM might provide insights into those aspects of cell envelope stress due specifically to TG inhibition and those due to other effects resulting from lipid II sequestration.
Both RAM and MOE induce the expression of a large set of genes, including many belonging to known cell envelope stress-related regulons (Fig. 1). However, the stimulons induced upon antibiotic treatment differ dramatically. RAM, but not MOE, leads to a strong induction of LiaRS TCS regulon members, including the key liaI and liaH genes (Fig. 1A; see also Table S2 in the supplemental material). LiaRS was originally named for lipid II-interacting antibiotics, because it was strongly upregulated after treatment with the lipid II-targeting antibiotics vancomycin and bacitracin (28). However, induction of LiaRS is not restricted to lipid II-interacting antibiotics, as evidenced by the strong induction by the membrane-active compound daptomycin (11). Although RAM induces the LiaRS regulon, this does not appear to be adaptive: a ΔliaIH mutant is unaffected in sensitivity to RAM.
Both RAM and MOE induce a large number of genes assigned previously to regulons controlled by ECF σ factors and, in particular, σM (Fig. 1B; see also Table S2 in the supplemental material). The σM regulon comprises ~60 genes, most of them involved in cell envelope synthesis, cell division and shape control, and DNA repair (7). σM also contributes to transcriptional activation of the Spx transcription factor, which can account for the induction of Spx-regulated genes in response to cell envelope stress as shown previously (7). The σM regulon is known to be induced by cell wall-active antibiotics like vancomycin and bacitracin, as well as a variety of other stresses (5, 7, 20, 40). A sigM-null mutant was highly sensitive to MOE, with a MIC of 0.4 μg/ml, compared to 10 μg/ml for the WT, but was unaffected in susceptibility to RAM. The σM-dependent genes responsible for resistance to MOE are unknown, but we do note that σM activates transcription of at least one HMW PBP (PBP1a), which might help to overcome MOE inhibition (7). Conversely, since RAM interacts with the lipid II substrate, the role of σM in increasing expression of HMW PBPs may not affect resistance.
Since the regulation and regulons of LiaRS and σM have been studied intensively (7, 44), we focused our attention on two operons of unknown regulation (ytrABCDEF and ywoBCD) that were preferentially induced by RAM but not MOE. Induction of these operons has also been noted previously in studies with two other substrate-binding antibiotics, vancomycin (5) and bacitracin (27).
To further define the specific class of antibiotics that induce the ytrA and ywoB operons, we performed a hierarchical cluster analysis using transcriptional profiles resulting from treatment of B. subtilis with RAM, MOE, daptomycin (11), bacitracin (27), vancomycin (7), and 22 other antibiotics as reported by Hutter et al. (19) (Fig. 2). Genes in the ytrA and ywoB operons clustered tightly, demonstrating induction by bacitracin, vancomycin, ristocetin, and RAM but weak if any induction by other cell wall-targeting antibiotics (such as MOE, β-lactams, d-cycloserine, and fosfomycin) or by antibiotics with different modes of action (such as ciprofloxacin and tetracycline). We propose that the key difference between the RAM and MOE stimulons is due to the fact that the former, but not the latter, binds directly to lipid II. Compounds that bind lipid II, including many glycopeptides, not only inhibit PG synthesis but also affect wall teichoic acid synthesis. This may account for the broader set of induced genes, which includes the ytrA and ywoB operons. We note that the ytrA and ywoB operons are induced specifically by lipid II-binding compounds (Fig. 2 and and3),3), whereas the LiaRS TCS appears to also respond to at least some compounds that more generally perturb membrane functions (e.g., daptomycin  and Triton X-114 ) (Fig. 3). Indeed, the ytrA and ywoB genes have been previously proposed to serve as reporters for compounds with a glycopeptide mode of action (18).
The bceRSAB operon also clusters with the ytrA and ywoB operons (Fig. 2). The bceRSAB operon encodes the BceRS TCS and an ABC-type transporter (BceAB). The BceRS regulon is induced by and confers resistance against bacitracin and the lantibiotics mersacidin and actagardine and the defensin plectasin (31, 36). Since the bceRSAB operon is adjacent to the ytrA operon, we considered the possibility that these might both be regulated by the BceRS TCS. However, the expression profiles of the ytrA and ywoB operons differ from that noted for the bceR operon (Fig. 3), and these operons lack a recognizable binding site for BceR.
The ytrABCDEF operon has been reported to be transcribed from a σA-dependent promoter sequence (PytrA; TTGACt-17nt-TATtgT [consensus bases in uppercase]) (15) 244 bp upstream of the ytrA start codon (45). However, the relationship between YtrA and the regulation of this operon during antibiotic stress has not been defined. Here, we show that YtrA is required for the repression and antibiotic responsiveness of PytrA (Fig. 4A). A transcriptional PytrA-lacZ fusion is repressed in the WT but fully derepressed in either a ytrA in-frame deletion mutant or an allelic replacement mutant lacking the entire ytrA operon (ytrABCDEF::kan) (Fig. 4A). In contrast, deletion of the ywoB operon did not affect regulation (data not shown).
The similar regulatory responses of the ytrA and ywoB operons (Fig. 3) suggested that they may both be controlled by YtrA. Support for this notion is provided by analysis of a ywoB-lacZ transcriptional fusion which, like ytrA (Fig. 4A), is antibiotic inducible in the wild type but fully derepressed in strains lacking YtrA (Fig. 4B). The transcriptional start site of ywoB was mapped to 33 bp upstream of the start codon using 5′ RACE-PCR, corresponding to a σA-dependent promoter (TTGACA-17nt-TActgT [consensus bases in uppercase]) (15) (Fig. 5B).
To determine if PytrA is the only promoter for the ytrA operon, and if this promoter responds to antibiotic stress, we analyzed a series of promoter truncation mutants (Fig. 5A). The results indicate that PytrA is both necessary and sufficient for vancomycin-responsive transcriptional regulation: a truncation at position −289 is active and inducible (from 4.6 ± 0.1 to 66.7 ± 12.3 Miller units with vancomycin treatment), while a truncation at position −245 relative to the ytrA translational start codon is expressed at only background levels (1.6 ± 0.1 and 1.4 ± 0.1 Miller units without and with vancomycin, respectively). There is no evidence for additional promoter activity immediately upstream of either ytrA or ytrB.
During the course of these studies, we noted that there is an additional small open reading frame (ORF) encoding a 45-amino-acid hypothetical protein immediately upstream of ytrA (Fig. 5A). This coding sequence is preceded by a good ribosome-binding site and has been annotated as a candidate gene in several recently sequenced B. subtilis genomes. We therefore propose the designation ytrG for this newly annotated gene. Expression studies using a translational ytrG-lacZ fusion indicate that this coding sequence is translated and is, as expected, antibiotic inducible (data not shown). The function of the short YtrG peptide, which has significant amino acid similarity with the first transmembrane segment of the YtrF protein, is presently unknown.
We have also explored the possibility that the ytrA operon might be regulated by σW, since a candidate σW consensus sequence was previously observed 144 bp upstream of the start codon of ytrA (4) overlapping the initiation codon of ytrG (Fig. 5A). However, promoter-lacZ fusion analysis showed that this promoter was inactive even when σW was overexpresssed in an rsiW background lacking the cognate anti-sigma factor (data not shown). We conclude that σW is not involved in antibiotic induction of the ytrA operon.
YtrA is the founding member of the YtrA subfamily in the GntR superfamily (33). GntR family proteins generally contain an N-terminal helix-turn-helix (HTH) DNA binding domain and a variable C-terminal domain which tends to be shorter in YtrA subfamily proteins (33). Yoshida et al. identified a 31-bp inverted repeat overlapping PytrA as a possible binding site for YtrA (Fig. 5A). The ywoB promoter contains an identical inverted repeat sequence (Fig. 5B).
We used EMSA and DNase I footprinting to determine if YtrA binds directly to the inverted repeat sequences overlapping the PytrA and PywoB promoter regions. Purified YtrA (5 nM) shifted DNA fragments containing the promoter/operator regions of both ytrA and ywoB (Fig. 6) but did not bind to a control DNA fragment. Interestingly, increasing the concentration of YtrA (20 nM for PytrA and 40 nM for PywoB) resulted in a supershift of the promoter fragments, suggesting the possibility of multiple binding sites. In DNase I footprinting assays, complete protection was achieved with the addition of 2 nM YtrA protein at both promoters (Fig. 7). Thus, YtrA binds specifically and with high affinity to these operator sites.
Comparison of the ytrA and ywoB operator sites reveals perfect identity within all four 6-bp half-sites of each inverted repeat (Fig. 5C). Indeed, these two sites are the only occurrence of this sequence in the B. subtilis genome with this spacing. Although it is likely that these two hexamers are necessary for YtrA binding, they may not be sufficient: these two operators also have more extended elements of identity (Fig. 5C) that may also be important for protein-DNA interaction.
The ytrA operon encodes a typical ABC-type transporter system. YtrB and YtrE are putative ATPases containing nucleotide-binding domains (NBD), whereas YtrC and YtrD contain the membrane-spanning domains (MSD) and presumably comprise the permease. It is unknown whether YtrBCDEF acts as one integrated or two separate ABC transport systems (32, 45). Although initially suggested to encode an uptake system for acetoin, based on a modest decrease in acetoin utilization in a mutant strain (45), more recent studies indicate that acetoin catabolism is determined by the acetoin-inducible acoABCL operon (35, 41). Moreover, acetoin does not induce ytrA expression (45).The ywoBCD operon encodes a protein of unknown function, a cysteine hydrolase, and an MFS family major facilitator transporter, respectively.
GntR-type regulators regulate antibiotic production and morphogenesis in Streptomyces coelicolor (16), as well as the expression of efflux pumps affecting quinolone and β-lactam resistance in Staphylococcus aureus (42). We therefore hypothesized that the ytrA- and ywoB-encoded transporters might be involved in antibiotic resistance. However, despite extensive susceptibility testing (see Materials and Methods for the full list of tested compounds and conditions), we failed to identify an antibiotic sensitivity phenotype linked to null mutations of the ytrA operon (which constitutively expresses the ywoB operon), the ywoB operon, or both.
In conclusion, we here demonstrate that the two TG inhibitors RAM and MOE trigger distinct cell envelope stress responses in B. subtilis. Although both RAM and MOE induce the σM regulon, RAM additionally induces several other stress responses, including the LiaRS TCS and, as shown here, the YtrA regulon. We hypothesize that this broader stress response reflects the fact that compounds that bind lipid II not only inhibit PG synthesis but also impair wall teichoic acid synthesis by sequestration of the common undecaprenyl lipid carrier. Using in vivo and in vitro assays, we demonstrated that YtrA acts as a repressor of both the ytrA and ywoB operons, and we identified the corresponding operator sites. The pattern of induction of the ytrA and ywoB operons by antibiotics suggests that YtrA controls a subset of antibiotic stress response genes. How YtrA senses antibiotic stress, and how the ytrA and ywoB operons contribute to cell fitness, remains unclear.
We thank S. Donadio (Vicuron) for the generous gift of purified ramoplanin samples. We thank Shawn MacLellan and Ahmed Gaballa for assistance with protein purification and Julia Zoller for plasmid construction. We also thank Alan D. Grossman and Catherine A. Lee (MIT) for performing the microarray hybridization for the RAM-treated RNA.
This work was supported by a grant from the National Institutes of Health (GM-047446 to J.D.H.) and from the Deutsche Forschungsgemeinschaft (MA2837/3-1 to T.M.).
§Supplemental material for this article may be found at http://jb.asm.org/.
Published ahead of print on 19 August 2011.