Resistance to Bacitracin in Bacillus subtilis: Unexpected Requirement of the BceAB ABC Transporter in the Control of Expression of Its Own Structural Genes

Abstract

The Bacillus subtilis BceAB ABC transporter involved in a defense mechanism against bacitracin is composed of a membrane-spanning domain and a nucleotide-binding domain. Induction of the structural bceAB genes requires the BceR response regulator and the BceS histidine kinase of a signal transduction system. However, despite the presence of such a transduction system and of bacitracin, no transcription from an unaltered bceA promoter is observed in cells lacking the BceAB transporter. Expression in trans of the BceAB transporter in these bceAB cells restores the transcription from the bceA promoter. Cells possessing a mutated nucleotide-binding domain of the transporter are also no longer able to trigger transcription from the bceA promoter in the presence of bacitracin, although the mutated ABC transporter is still bound to the membrane. In these cells, expression of the bceA promoter can no longer be detected, indicating that the ABC transporter not only must be present in the cell membrane, but also must be expressed in a native form for the induction of the bceAB genes. Several hypotheses are discussed to explain the simultaneous need for bacitracin, a native signal transduction system, and an active BceAB ABC transporter to trigger transcription from the bceA promoter.

Bacitracin is an antibiotic composed of a complex mixture of branched cyclic dodecylpeptides synthesized by some species of bacilli. In the presence of divalent metal ions needed for its biological activity, bacitracin binds to undecaprenyl pyrophosphate (UPP) (33), leading to the arrest of bacterial peptidoglycan biosynthesis. Indeed, during this cellular process, undecaprenyl phosphate (UP) serves as a lipid carrier and is essential for the synthesis of many cell wall polymers. In the case of peptidoglycan biosynthesis, UP is responsible for the translocation of peptidoglycan building blocks from the cytosol to the external side of the cytoplasmic membrane. Transfer of these precursors to the extremity of the nascent peptidoglycan chain releases UPP, which is then dephosphorylated by UPP phosphatases, regenerating the UP. By preventing this recycling step, the binding of bacitracin to UPP reduces the amount of UP available for peptidoglycan precursor translocation, thus impeding further peptidoglycan synthesis and eventually leading to cell lysis.

Due to the presence of an outer membrane, gram-negative bacteria, such as Escherichia coli, are not very susceptible to bacitracin, and various UPP phosphatases ensure a high level of resistance (6). UPP phosphatases of the PAP2 family, such as BcrC in Bacillus subtilis (formerly YwoA) (3), are found in some gram-positive bacteria, in which they compete with bacitracin for UPP, thus conferring a certain level of protection on the cells (4, 23). The bacitracin producer Bacillus licheniformis ATCC 10716 possess two PAP2 family members, BcrC and a BcrC-like protein (YP_080959) (21, 27, 38). However, in addition to the UPP phosphatases, some gram-positive bacteria also possess an ABC transporter(s), which is an even more efficient bacitracin protection system. Such is the case for B. subtilis, Streptococcus mutans, Enterococcus faecalis, and B. licheniformis ATCC 10716 (4, 17, 19, 21, 22, 36). However, the mechanism by which these transporters mediate bacitracin resistance remains unknown. According to the ABCDB database (http://www-abcdb.biotoul.fr/) (28), these ABC transporters fall into two different families. BcrAB from B. licheniformis and BcrAB from E. faecalis belong to family 7, and both the elements controlling the expression of their structural genes and the way these elements operate differ (21). Family 9 contains BceAB from B. subtilis, YtsCD from B. licheniformis, and MbrAB from S. mutans, each genetically and functionally linked to a signal transduction system including a histidine kinase and a response regulator (4, 22, 36, 38).

BceAB from B. subtilis is composed of a nucleotide-binding domain (NBD) (BceA) and a membrane-spanning domain (BceB). The bceA and bceB cognate structural genes constitute a transcriptional unit whose expression is under the control of the BceRS signal transduction system (4, 22). In a classical view of the phenomenon, after detecting the presence of bacitracin, the BceS histidine kinase is thought to autophosphorylate and to activate the BceR response regulator by transphosphorylation (32). Once activated, the phosphorylated regulator can modulate the expression of its main target genes, including the bceAB operon. However, unlike most histidine kinases, BceS does not possess an extracytoplasmic input domain. Classified in the intramembrane-sensing histidine kinase family (18), it is supposed to sense a signal at the level of the membrane or within it, and it has been postulated that BceS may detect either the bacitracin-UPP complex directly or, alternatively, a perturbation of the cell envelope structure (19).

To better understand the induction process of the bceAB genes, we used a strain carrying a bceA::lacZ transcriptional fusion due to a pMUTIN plasmid insertion (37). Unexpectedly, we observed that the bceA::pMUTIN mutant was unable to express the lacZ gene upon addition of bacitracin. This paper reports the surprising observation that the BceAB ABC transporter is required, together with the BceRS transduction system, for B. subtilis to trigger transcription from its own promoter in the presence of bacitracin in the medium.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

The bacterial strains and plasmids used in this study are listed in Table 1 and Table S1 in the supplemental material, respectively. All complementations in trans were done using the pDG148-Stu plasmid (which is referred to as pDG below) as a vector, following the procedure described previously (11). In each case, the sequence of the recombinant plasmid was checked by DNA sequencing. The BFS82 and BFS83 strains were obtained from the Japanese/European Consortium for B. subtilis Genome Function Analysis (14). Strain BSGY005 (22) was kindly provided by Kazuo Kobayashi. E. coli and B. subtilis were grown in Luria broth (LB) at 37°C with aeration. Unless otherwise stated, isopropyl-β-d-thiogalactopyranoside (IPTG) was present throughout growth at a 1 mM final concentration and β-galactosidase activity was determined in cells collected 1 hour after bacitracin addition at mid-exponential growth phase.

TABLE 1.

E. coli and B. subtilis strains used in this study

Strain	Relevant characteristics	Source or reference
E. coli
DH5α	λ− φ80dlacZΔM15 Δ(lacZYA-argF)U169 recA1 endA1 hsdR17(rK− mK−) supE44 thi-1 gyrA relA1	Invitrogen
C41DE3	F−ompT hsdSB(rB− mB−) gal dcm (DE3)	Avidis S. A., France
B. subtilis
Wild type 168	trpC2	1
BSGY005	168 amyE::bceAp::lacZ; Cmr	22
BFS82	168 bceA::pMUTIN::lacZ; Emr	Micado web site
BFS83	168 bceB::pMUTIN::lacZ; Emr	Micado web site
BSmrs92	BFS82/pDGbceR; Emr Kmr	This study
BSmrs171	BFS82/pDGbceA; Emr Kmr	This study
BSmrs197	BSGY005/pDGbcrC; Emr Kmr	This study
BSmrs230	BSGY005/pDG; Cmr Emr	This study
BSmrs231	BSGY005/pDGhis6-bceA(E169A); Cmr Kmr	This study
BSmrs232	BSGY005/pDGhis6-bceA; Cmr Kmr	This study
BSmrs241	BFS83/pDGbceAB; Emr Kmr	This study
BSmrs246	BFS82/pDGhis6-bceA; Emr Kmr	This study
BSmrs247	BFS82/pDGhis6-bceA(E169A); Emr Kmr	This study

Recombinant strains were grown in medium containing antibiotics at the following concentrations: ampicillin (50 μg ml⁻¹) for E. coli and erythromycin (0.3 μg ml⁻¹), kanamycin (20 μg ml⁻¹), tetracycline (10 μg ml⁻¹), and chloramphenicol (5 μg ml⁻¹) for B. subtilis. All antibiotics were from Sigma-Aldrich.

General molecular biology techniques.

Unless otherwise stated, all molecular biology procedures were carried out as described by Sambrook and Russell (30). DNA fragments were purified using either a Microcon-30 (Millipore) or the Qiaquick nucleotide removal kit (Qiagen). Cloning of DNA was done in either the E. coli DH5α or C41DE3 strain. PCR amplifications were done in a 50-μl final volume, using Expand high-fidelity PCR (Roche Diagnostics) as recommended by the manufacturer. Plasmid purifications were done using either a plasmid Midi kit or a plasmid Mini kit from Qiagen.

All oligonucleotides used in this study are listed in Table S2 in the supplemental material.

mRNA preparation, cDNA synthesis, and quantitative PCR.

All mRNA preparation, cDNA synthesis, and quantitative PCR procedures were done as previously described (12).

Obtaining His₆-bceA and His₆-bceA(E169A) genes. (i) bceA cloning into the pET22-Pml plasmid.

An 831-bp DNA fragment encompassing the entire bceA coding sequence but lacking the start codon was obtained by PCR using bceA-dir and bceA-rev as primers (see Table S2 in the supplemental material) and B. subtilis genomic DNA as a template. The fragment was cloned into the pET22-Pml plasmid as described previously (13). The recombinant plasmid pET22-Pml-bceA was used to transform E. coli strain DH5α. The sequence of the entire insert was checked.

(ii) Site-directed mutagenesis of the pET22-Pml-bceA plasmid.

Mutagenesis was performed on the pET22-Pml-bceA plasmid as described previously (2) using the bceA-mut1 and bceA-mut2 primers (see Table S2 in the supplemental material), leading to the plasmid pET22-Pml-bceA_E169A. The mixture was treated with DpnI (New England Biolabs) to eliminate the native pET22-Pml-bceA plasmid and then used to transform E. coli strain DH5α. After plasmid purification, the DNA sequence of the insert was checked for the presence of the mutation.

(iii) Cloning of His₆-bceA and His₆-bceA(E169A) into the pDG plasmid.

The entire His₆-bceA or His₆-bceA(E169A) gene was amplified by PCR from the corresponding plasmid pET22-Pml-bceA or pET22-Pml-bceA(E169A) using bceA(pdg)atg and bceA(pdg)stop as primers (see Table S2 in the supplemental material) and introduced into pDG as described above. After purification of the recombinant plasmids, the entire sequence of each insert was checked.

Inhibition of the bacitracin response by reserpin.

Strain BSGY005 was grown in the appropriate medium to an optical density at 600 nm (OD₆₀₀) of 0.3. Then, 50 μg/ml of bacitracin was added and the culture was immediately split. A solution of reserpin (Sigma-Aldrich) in ethanol was added to one of the split cultures to reach a 40 μM final reserpin concentration. The same volume of ethanol alone was added to the other culture. After 45 min of culture at 37°C under agitation, the bacteria were harvested and the β-galactosidase activity was measured. We observed no bacterial lysis after 1 hour of incubation in the presence of bacitracin and/or reserpin.

Cell lysate and membrane preparation.

For cell lysate and membrane preparation, we followed the procedure described by Bernard et al. (3).

β-Galactosidase assay.

One milliliter of cell culture was centrifuged, and the pellets were suspended in 1 ml of Z buffer (20). A volume, V (expressed in milliliters), was diluted with Z buffer to a 980-μl final volume. The mixture was incubated for 15 min at 37°C after the addition of 10 μl of lysosyme (10 mg/ml). We then added 10 μl of 10% Triton X-100 and incubated the resulting mixture at 28°C. The assay was started with the addition of 200 μl of orthonitrophenylgalactoside (Sigma-Aldrich) solution (4 mg/ml) and stopped by the addition of 500 μl of 1 M Na₂CO₃, and the OD was measured at 420 nm. β-Galactosidase units are expressed according to the following equation: 1 unit = 1,000/4.8 × OD₄₂₀ × 1/t × 1.7/V × 1/OD₆₀₀, where t represents the time of enzymatic reaction (in minutes) and OD₆₀₀ reflects the cell density just before assay.

RESULTS

The BceAB ABC transporter is needed for the induction of the bceAB operon by bacitracin in B. subtilis.

As mutants of each of the components of the bce system are much more sensitive to bacitracin (50% inhibitory concentration [IC₅₀] = 6 μg/ml) than the parental strain (IC₅₀ = 350 μg/ml) (4), for all the bce::pMUTIN mutants constructed by the Japanese/European Consortium for B. subtilis Genome Function Analysis (14; Micado database [http://genome.jouy.inra.fr/cgi-bin/micado/index.cgi]), a bacitracin concentration of 4 μg/ml was routinely used. It is worth noting that this bacitracin concentration is sufficient to allow bceAB induction in a wild-type background. Indeed, in an amyE::bceAp::lacZ strain, β-galactosidase activities of 60 and 130 units were detected using 1 μg/ml and 10 μg/ml of bacitracin, respectively, whereas almost no β-galactosidase (<1 unit) was detected in the absence of bacitracin. Note that upon integration of pMUTIN, the affected gene is interrupted, a lacZ transcriptional fusion is generated, and downstream genes belonging to the same multicistronic unit are placed under the control of the inducible Pspac promoter carried by pMUTIN (37). When the bceA::pMUTIN mutant (strain BFS82) was grown in the presence of either bacitracin alone or bacitracin plus IPTG to induce the expression of the bceB gene, no β-galactosidase activity could be detected (data not shown). The same negative result was observed when the bceB::pMUTIN strain (BFS83) was grown in the presence of bacitracin. In many signal transduction systems, overexpression of the response regulator mimics the presence of the inducer (24, 31). When plasmid pDGbceR carrying the bceR response regulator structural gene under the control of IPTG was introduced into the bceA::pMUTIN strain, almost no β-galactosidase activity was obtained in cells grown in the absence of IPTG, whereas 300 ± 25 units of β-galactosidase was found when bceR expression was induced in the presence of IPTG, indicating that the bceA promoter was functional and that it had not been affected by the pMUTIN insertion. Unless otherwise stated, in all experiments β-galactosidase activities were measured 1 hour after bacitracin addition and IPTG was present throughout growth.

As the only known defect of strains BFS82 and BFS83 is the lack of the BceAB transporter, a complementation experiment was performed to confirm that this absence was indeed responsible for the lack of bceA::lacZ fusion expression. A plasmid bearing the bceA gene under the control of IPTG was introduced into strain BFS82, and the bceA::pMUTIN/pDGbceA cells (BFS171) were grown in the presence of bacitracin with or without IPTG. In these experiments, it is worth noting that IPTG induces the expression of both bceA (carried by the pDGbceA plasmid) and bceB (placed under the control of the inducible Pspac promoter carried by pMUTIN). Almost no β-galactosidase activity was detected in the bceA::pMUTIN/pDGbceA cells grown in the absence of IPTG, whereas a significant level was noted in its presence. Indeed, upon the addition of bacitracin, 6 ± 1 units and 24 ± 5 units of β-galactosidase activity were observed after 20 min and 60 min of incubation, respectively (Fig. 1). Similar results were obtained using the bceB::pMUTIN/pDGbceAB strain when it was grown in the presence of bacitracin and IPTG to induce the expression of bceAB (data not shown).

FIG. 1.

β-Galactosidase specific activity of a bceA strain complemented or not complemented by bceA in trans. A bceA::pMutin strain carrying the plasmid pDG-bceA was inoculated into LB medium supplemented with (▪) or without (□) IPTG. Bacitracin (4 μg/ml) was added at mid-exponential growth phase, taken as time zero. Cells were harvested at the indicated times. The results are given as the mean values from three experiments plus standard deviations.

Strain BSGY005, which contains a bceAp::lacZ fusion at the amyE locus, was used to follow the response to bacitracin in a BceAB BceRS wild-type context (22). We first checked that this amyE::bceAp::lacZ strain was able to respond at low bacitracin concentrations (see the results above), indicating that the bceAp promoter is perfectly functional when located at that ectopic position.

Inhibition of the BceAB transporter affects the response to bacitracin.

In an earlier paper, we showed that the bacitracin resistance due to the BceAB ABC transporter is drastically affected by the plant alkaloid reserpin, a strong inhibitor of efflux systems (4). We tested the effect of 40 μM reserpin (a sublethal concentration) on the response of the amyE::bceAp::lacZ strain to bacitracin. For cells grown with 50 μg/ml of bacitracin, β-galactosidase activity was almost 60% lower than for cells grown without reserpin (71 ± 11 versus 160 ± 20 units), suggesting that the BceAB transporter must possess its full transport capacity to allow the cells to respond to bacitracin.

Mutation of the NBD subunit (BceA) of the BceAB transporter eliminates the response to bacitracin.

The ATP-binding site of an ABC transporter NBD subunit often contains a conserved glutaminyl residue that is important for the ATPase activity (25). This residue was changed in BceA, and His₆-tagged derivatives (mutated or not) were expressed in bceA::pMUTIN cells using the pDG plasmid. The resulting cells were grown in the presence of bacitracin with or without IPTG. In all cultures, IPTG was added at the indicated times before bacitracin, itself always added in the mid-exponential growth phase. Cells were harvested 45 min after the addition of bacitracin, and the β-galactosidase activity was determined. A faint β-galactosidase activity (<1 unit) could be detected in bceA::pMUTIN/pDGhis₆-bceA cells grown without IPTG, while a 28-fold increase of activity was recorded in cells grown with IPTG, indicating that the His₆-BceA protein was fully active (Fig. 2). When the above-mentioned conserved glutaminyl residue was replaced by an alaninyl residue, almost no β-galactosidase activity was detected in the bceA::pMUTIN/pDGhis₆-bceA(E169A) cells grown with IPTG despite the presence of bacitracin (Fig. 2). As this negative result could be explained either by the absence of the mutant protein due to increased proteolysis or by its wrong localization resulting from the glutaminyl replacement by alaninyl, its production was followed using an anti-His₆ tag antibody. As indicated in Fig. 3, neither the His₆-BceA nor the His₆-BceA(E169A) protein was detected when IPTG was omitted during bacterial growth, whereas both proteins with the expected molecular mass (ca. 28 kDa) were present when IPTG was added. It is worth noting that no such signal was found in bceA::pMUTIN/pDG cells grown in the presence of IPTG (data not shown). Subcellular fractionations of the crude extracts indicated that these proteins are very likely localized in the membrane fraction (see Fig. S1 in the supplemental material).

FIG. 2.

Effects of BceA, His-BceA, and His-BceA(E169A) overexpression in a bceA strain. The bceA::pMutin cells carrying plasmid pDGbceA (bars without outlines), pDGhis₆-bceA (bars with thick outlines), or pDGhis₆-bceA(E169A) (bars with thin outlines) were inoculated into LB medium supplemented (light gray) or not (dark gray) with IPTG. Bacitracin (4 μg/ml) was added at the mid-exponential growth phase, taken as time zero. Cells were harvested at the indicated times, and β-galactosidase activities were measured. The results are given as the mean values from three experiments, and standard deviations are indicated.

FIG. 3.

Levels of His-BceA and His-BceA(E169A) proteins in B. subtilis strains. (A) Sodium dodecyl sulfate-polyacrylamide gel electrophoresis of crude lysates obtained from bceA::pMutin strains carrying either the pDGhis₆-bceA (lanes 1 and 2) or pDGhis₆-bceA(E169A) (lanes 3 and 4) plasmid. The cells were grown for 4 h at 37°C under agitation with (lanes 2 and 4) or without (lanes 1 and 3) IPTG (1 mM). After being harvested, they were broken by two passages through a French press (16,000 lb/in²). Fractions containing the same amount of proteins were subjected to electrophoresis on a 12.5% acrylamide gel. L, ladder of molecular mass standards corresponding to carbonic anhydrase (33 kDa), β-lactoglobulin (24 kDa), and lysozyme (20 kDa). (B) Immunoblot of the corresponding gel probed with a mouse anti-His antibody. A second antibody (rabbit anti-mouse immunoglobulin G) coupled with peroxidase) was used to reveal the blot with the GE Healthcare ECL Plus Western blotting detection system. All products were from GE Healthcare. The arrows indicate the expected molecular masses (28 kDa) of the proteins.

A competition between the native BceA and the His₆-BceA(E169A) mutated hybrid proteins was then done in which the pDGhis₆-bceA(E169A) plasmid was introduced into the amyE::bceAp::lacZ strain. A series of experiments were conducted in which IPTG was added at different times relative to bacitracin, which was added in the mid-exponential growth phase. Cells were collected 45 min after the addition of bacitracin, and β-galactosidase activity was determined (Fig. 4A). In fact, the earlier the addition of IPTG, the lower the activity; the maximum effect corresponded to an almost complete lack of β-galactosidase detection being observed when IPTG was present in both the preculture and the culture (Fig. 4A, time ∞). Using these conditions, the β-galactosidase activities in amyE::bceAp::lacZ cells bearing the pDG plasmid and either the pDGhis₆-bceA or pDGhis₆-bceA(E169A) plasmid were compared. As shown in Fig. 4B, similar and significant levels of activity were obtained in cells containing the pDG plasmid and in cells expressing the His₆-BceA protein. On the other hand, a very low level of β-galactosidase activity was recorded in bacteria expressing the His₆-BceA(E169A) protein. These results clearly indicated that (i) the His₆ tag was not responsible for the observed decrease in β-galactosidase activity, (ii) overproduction of the tagged BceA protein did not affect β-galactosidase synthesis, and (iii) the His₆-BceA(E169A) protein seemed able to compete with native BceA, likely by interacting with the native BceB subunit, thus titrating the latter to give an inactive ABC transporter.

FIG. 4.

Effects of the overproduction of His-BceA and His-BceA(E169A) on the response to bacitracin. Bacteria were harvested 45 min after bacitracin addition, and β-galactosidase specific activities were determined. The results are given as the mean values from three experiments, and standard deviations are indicated. (A) The amyE::bceAp::lacZ (BSGY005) strain carrying the pDGhis₆-bceA(E169A) plasmid was grown on LB medium until mid-exponential phase. Overproduction of His-BceA(E169A) was either uninduced or induced by adding IPTG, which was added at the indicated times before the addition of bacitracin at 4 μg/ml. ∞ indicates that the cells were always in contact with IPTG (during preculture and culture). The results are expressed as the percentages of the β-galactosidase response obtained in the absence of IPTG (35 ± 5 units). (B) BSGY005 cells carrying the empty plasmid, the pDGhis₆-bceA plasmid, or the pDGhis₆-bceA(E169A) plasmid were grown in medium supplemented with IPTG (1 mM) during both preculture and culture. When the cultures reached mid-exponential phase, bacitracin was added (4-μg/ml final concentration) to the culture media. The results are expressed as the percentages of the β-galactosidase response obtained in the BSGY005/pDG strain grown in the presence of IPTG (20 ± 2 units). The error bars indicate standard deviations.

bceB transcription levels needed for response to bacitracin.

In the bceA::pMUTIN/pDGbceA cells, the chromosomal bceB gene is under the control of the IPTG-inducible Pspac promoter of pMUTIN while the pDG-borne bceA gene is under the control of an IPTG-inducible promoter carried by the plasmid. Accordingly, one would expect both genes to be induced by IPTG efficiently enough for the BceAB ABC transporter to be produced and for cells to be resistant to bacitracin. However, this was not the case, since they were as sensitive to the antibiotic as the uncomplemented bceA::pMUTIN cells. As the bceA gene was carried on a multicopy plasmid while a unique copy of bceB was present on the chromosome, we suspected that the level of bceB transcripts was limiting for BceB production. Using a real-time quantitative PCR technique, the levels of bceB transcripts were measured in different strains and/or under different growth conditions. Cells were grown and collected for mRNA preparation as described in Materials and Methods. The bceB transcript level of parental cells grown in LB medium without bacitracin was taken as 1 ± 0.02 arbitrary units (AU). Under the same conditions, the bceB transcript level in the bceA::pMUTIN/pDGbceA cells was only 0.02 ± 0.002 AU. However, when these cells were grown in the presence of IPTG, the bceB transcript level increased 15-fold, reaching 0.3 ± 0.005 AU. Although threefold lower than that of the noninduced wild-type cells, this level of bceB transcripts was sufficient to produce enough BceB subunits to associate with the BceA subunits produced from the pDGbceA plasmid. This yielded enough ABC transporter to trigger a limited but detectable transcription of the chromosomal bceAp::lacZ fusion (formed upon insertion of pMUTIN) in the presence of bacitracin. Indeed, in the presence of both IPTG and bacitracin, the β-galactosidase activity observed in the bceA::pMUTIN/pDGbceA cells reached 25 units, whereas no detectable activity could be found in the bceA::pMUTIN cells (Fig. 1).

Modulating UPP/UP recycling modulates the bacitracin response.

In the amyE::bceAp::lacZ/pDGbcrC strain, overexpression of the BcrC UPP phosphatase upon IPTG induction enhanced the IC₅₀ from 350 to 420 μg/ml. This strain and amyE::bceAp::lacZ/pDG were grown under different conditions, and the β-galactosidase activities were recorded. When these strains were grown in the presence of IPTG but without bacitracin, a very low level of β-galactosidase activity was observed (<1 unit) (Table 2). However, when growth occurred in the presence of IPTG and 50 μg/ml bacitracin, a high level of β-galactosidase (102 ± 4 units) was found in cells from the amyE::bceAp::lacZ/pDG strain, whereas an almost 70% decrease in β-galactosidase production was observed in cells of the amyE::bceAp::lacZ/pDGbcrC strain (Table 2). As BcrC overproduction, as well as that of other UPP-phosphatases, was proposed to reduce the UPP pool of the cells (6), this result very likely points to an important role of UPP in triggering transcription from the bceA promoter in response to bacitracin.

TABLE 2.

Measurement of β-galactosidase specific activity in response to bacitracin in the amyE::bceAp::lacZ BSGY005 strain expressing BcrC or not^a

Strain	Bacitracina	IPTGa	β-Galactosidase unitsb
BSGY005/pDG	−	+	0.62 ± 0.4
	+	+	102 ± 4
BSGY005/pDGbcrC	−	+	0.38 ± 0.1
	+	+	31.5 ± 8

^a+, present; −, absent.

^bBacteria were grown in LB medium to mid-exponential phase. Bacitracin (50 μg/ml) and IPTG (1 mM) were added at the indicated concentrations, and the bacteria were incubated for 45 min at 37°C with agitation. Bacteria were collected, and β-galactosidase activity was determined. The results are given as mean values from three experiments ± standard deviations.

DISCUSSION

Various authors have reported that the level of transcripts from the B. subtilis bceA promoter is tremendously enhanced upon the addition of bacitracin and that this increase is mediated through the BceRS signal transduction system (4, 19, 22). Using a bceA::lacZ fusion, we accumulated the following information concerning transcription from the bceA promoter: (i) when one partner of the BceAB transporter is missing, almost no β-galactosidase activity can be detected in the presence of bacitracin; (ii) the expression of the bceA::lacZ fusion is restored upon expression of the BceAB transporter in trans, and (iii) a nonfunctional His₆-BceA(E169A) mutant protein competes with the native BceA NBD subunit, resulting in a drastic decrease in β-galactosidase activity. Finally, as supported by the small but detectable level of bceA and bceB transcripts in noninduced wild-type cells, enough BceAB transporter molecules are already present in the membrane to play their role in the induction process (12). Altogether, these data demonstrate that in B. subtilis growing in the presence of bacitracin, transcription from the bceA promoter is entirely dependent not only on the BceRS components of the signal transduction system, but also on the native BceAB ABC transporter.

In cells devoid of the BceAB ABC transporter, at least three hypotheses can explain the lack of transcription from the bceA promoter upon addition of bacitracin. In the first hypothesis, the stimulus is detected by the sensor but the BceR response regulator remains unphosphorylated in the absence of the BceAB transporter. This could result either from the dephosphorylation of phosphorylated BceR (BceR∼P) by a phosphatase or from a BceR status that would impede its transphosphorylation by BceS∼P. One can then postulate that the ABC transporter inactivates the putative phosphatase when interacting with it or that BceR can only be transphosphorylated when interacting with the ABC transporter. Interactions of regulators and transporters have been described in several instances (5, 16, 35), but never with a native ABC transporter. Only once was an interaction with an NBD subunit reported. Indeed, MalK, the NBD of the MalKEFG maltodextrin ABC transporter from E. coli, was shown to negatively control the status of the MalT regulator (10, 29). In the second hypothesis, the stimulus does exist in the absence of the transporter but is not detected by the sensor. The lack of detection of the existing stimulus by the BceS histidine kinase in the absence of the BceAB ABC transporter could be explained either by supposing that the sensor needs a direct contact with the BceAB transporter to be active or that the transporter presents the stimulus to the sensor. Alternatively, in the third hypothesis, the stimulus might not exist at all in the absence of the ABC transporter. This situation is encountered for β-lactam resistance induction in the gram-negative bacterium Enterobacter cloacae. Indeed, the AmpG transporter involved in recycling of muropeptides was shown to be essential for high-level expression of the AmpC β-lactamase in response to cefotaxime addition (15). Once imported into the cell, the muropeptides are enzymatically modified, giving rise to 1,6-anhydro-MurNAc tripeptide and UDP-MurNAc pentapeptide (9). These peptides act as effectors, modulating the AmpR regulator activity for β-lactamase induction.

BceAB belongs to an ABC transporter family containing more than 500 members, which are all predicted to be exporters, since no substrate binding protein has ever been associated with any of them (28; http://www-abcdb.biotoul.fr/). If we speculate that the BceAB export capacity is involved in the constitution of the stimulus detected by the BceS sensor, what might be its transported substrate?

An antagonist neutralizing the bacitracin in the external medium might be a good candidate, supposing that the true inducer will be the antagonist-bacitracin complex. However, such a compound was not found in the supernatant of an S. mutans culture (36) and our preliminary results do not favor this hypothesis. As UPP is the membrane target of bacitracin (33) and as bacitracin is required but is not sufficient for induction, we propose that the UPP-bacitracin complex is the transported substrate. Both BcrC UPP phosphatase and bacitracin compete for UPP in the cells. Accordingly, the increased resistance of B. subtilis to bacitracin upon overexpression of BcrC (4, 23) might be due to the rapid conversion of UPP to UP, which decreases the internal pool of UPP, as proposed previously (6). Thus, the threefold reduction of β-galactosidase activity we observed in the amyE::bceAp::lacZ cells overproducing BcrC supports the notion that UPP plays a central role in stimulus generation.

To account for the very short extracytoplasmic loop (three residues) of the BceS histidine kinase, Mascher and collaborators had already proposed that this protein directly senses the UPP-bacitracin complex through an interaction with its transmembrane helixes (19). However, if this were sufficient to trigger transcription from the bceA promoter in the presence of bacitracin, there should be no need for the ABC transporter. The fact that no activity at all was detected in a ΔbceAB amyE::bceAp::lacZ strain excludes any direct stimulation of BceS by the UPP-bacitracin complex.

The site of UPP dephosphorylation, the mechanism of UP recycling, and the cellular location of bacitracin are still open questions. As UPP, the bacitracin target, is a membrane component, it is legitimate to suppose that the UPP-bacitracin complex is located within the membrane. Assuming that this complex is able to freely diffuse from one side of the membrane to the other, it might be equally distributed between the inner and outer leaflets. In this context, BceAB would work as a flippase (8), accumulating the UPP-bacitracin complex in the outer leaflet of the bacterial membrane and thus creating an asymmetric distribution of the complex (Fig. 5) that might be detected by the BceS sensor. In this scenario, the low basal level of BceAB is sufficient to generate a membrane asymmetry detected by BceS, thus triggering the bceAB gene expression that leads to a chain reaction with a very high final level of BceAB ABC transporter production. In this context, UPP-bacitracin will accumulate in the outer leaflet, reaching a very high local concentration. Taking into account the weak binding constant (1 × 10⁻⁶ M⁻¹) for the interaction between bacitracin A and UPP (34), part of the bacitracin might be released into the external medium. As a result of this event cascade, bacitracin would be pumped out of the bacterial membrane and the cells would be protected from its action.

FIG. 5.

UPP-bacitracin flipping and membrane asymmetry. Bacitracin (○) interacts with UPP (black stems) to form the UPP-bacitracin complex (circles on stems). Assuming that the complexes are able to freely diffuse from the outer to the inner leaflet (and vice versa), they might be equally distributed between the two leaflets (left part of the diagram). When the BceAB ABC transporter flips the UPP-bacitracin complex from the inner to the outer leaflet of the membrane (right part of the diagram), a membrane asymmetry is created. One can imagine that this asymmetry leads to repositioning of the two transmembrane domains (white rectangles) of the BceS histidine kinase, inducing the autophosphorylation of its transmitter domain (black rectangle). Thus, the response regulator can be activated after transferring the phosphate group (P in circle) to the receiver domain.

In accounting for the very special regulation of bceAB of B. subtilis, it is tempting to speculate that YtsAB transporter of B. licheniformis or MbrAB transporter of S. mutans is also required for the expression of their cognate structural genes in response to bacitracin. Can this feature be generalized to other members of the genetically and functionally linked signal transduction systems/ABC transporters (family 9) found in the phylum Firmicutes (12, 13, 18)? We are currently investigating whether this is true for YtsABCD from B. licheniformis and MbrABCD of S. mutans, using bacitracin as an inducer, as well as for YvcPQRS and YxdJKLM of B. subtilis, which have recently been shown to be induced by enduracidin, a peptide antibiotic (7), and LL-37, a human antimicrobial peptide (26), respectively.