BcrC from Bacillus subtilis Acts as an Undecaprenyl Pyrophosphate Phosphatase in Bacitracin Resistance*

Overexpression of the BcrCBs protein, formerly called YwoA, in Escherichia coli or in Bacillus subtilis allows these bacteria to stand higher concentrations of bacitracin. It was suggested that BcrCBs was a membrane-spanning domain of an ATP binding cassette (ABC) transporter involved in bacitracin resistance. However, we hypothesized that this protein has an undecaprenyl pyrophosphate (UPP) phosphatase activity able to compete with bacitracin for UPP. We found that overexpression of a recombinant His6-BcrCBs protein in E. coli (i) increased the resistance of the cells to bacitracin and (ii) increased UPP phosphatase activity in membrane preparations by 600-fold. We solubilized and prepared an electrophoretically pure protein exhibiting a strong UPP phosphatase activity. BcrCBs, which belongs to the type 2 phosphatidic acid phosphatase (PAP2) phosphatase superfamily (PF01569), differs totally from the already known BacA UPP phosphatase from E. coli, a member of the PF02673 family of the Protein family (Pfam) database. Thus, BcrCBs and its orthologs form a new class of proteins within the PAP2 phosphatase superfamily, and likely all of them share a UPP phosphatase activity.

Bacitracin, a mixture of related cyclic polypeptides, is a powerful antibiotic that strongly binds as a complex with metallic ions to undecaprenyl pyrophosphate (UPP), 1 (1) thus preventing its dephosphorylation into undecaprenyl phosphate (UP) (2). UP is a lipid carrier that is essential for the synthesis of many cell wall polymers and, more specially, for peptidoglycan biosynthesis. By sequestrating UPP and preventing its dephosphorylation back into UP, bacitracin weakens UP loading or reloading with peptidoglycan precursors. Because translocation of these precursors to the external side of the membrane is drastically or totally impaired, further biosynthesis of the con-stituents of the cell envelope is stopped, eventually resulting in cell death.
Bacitracin is produced by several species of Bacillus such as Bacillus licheniformis and Bacillus subtilis. Both in the B. licheniformis bacitracin-producing strains and in the B. subtilis naturally resistant non-producing strains, resistance to the antibiotic is ensured by an efficient specific ABC export system (3)(4)(5)(6). Simultaneously with others (4 -6), we identified a second bacitracin resistance system in B. subtilis 168. This bacterium has at least two different and independent bacitracin resistance systems, both belonging to the bacitracin regulon (5). The first, an ABC exporter (family 9 (7)), is encoded by the bceAB Bs operon, the expression of which is strongly induced by bacitracin via a classical two-component system, BceRS Bs (5,6,8). After the addition of the antibiotic, the response regulator triggers the transcription of the bce-AB Bs genes, probably strongly increasing the ABC protein level in the membrane. The putative transporter of ABC type, BceAB Bs , causes most of the resistance of the cells to bacitracin. When it is lacking, the IC 50 of the cells decreases from 251 to 6 M (6). The second system, BcrC Bs (YwoA), is similar to BcrC Bl from B. licheniformis (3) and to BcrC Ec from Escherichia coli (9), both of which are involved in bacitracin resistance. BcrC Bs is encoded by the bcrC Bs (ywoA) structural gene and also participates, but to a lesser extent than BceAB Bs , in the B. subtilis resistance to bacitracin. When BcrC Bs is lacking, the IC 50 of the cells decreases from 251 to 31 M (6). The expression of the bcrC Bs gene is also controlled by bacitracin but does not depend on the BceRS two-component system (6). Although the molecular mechanism of induction by bacitracin is not clear, it depends on M, X, and both W and its cognate anti-, RsiW (4,10).
BcrC Bl was formerly classified as a putative membranespanning domain of an ABC transporter (3). Accordingly, BcrC orthologs, which all have at least four putative transmembrane domains, have been annotated in databases as putative membrane proteins. Their predicted functions go from unknown to bacitracin transporter. All, however, also belong to the PF01569 Pfam family of PAP2 phosphatases, which contains more than 600 proteins. Some members of this family, but no BcrC orthologs, have a demonstrated phosphatase activity. This is the case, for instance, for PhoC from Morganella morganii (11) and Zymomonas mobilis (12), PhoN from Salmonella typhimurium (13), PgpB from E. coli (14), LpxE from Rhizobium leguminosarum (15), and two DOLPP1 orthologs from Saccharomyces cerevisiae and Mus musculus (16,17). We then hypothesized (6) that such might also be the case for the BcrC Bs protein and that its true function would be to dephosphorylate UPP. To test these hypotheses, we purified and characterized the BcrC Bs protein from B. subtilis. We showed that BcrC Bs has UPP phosphatase activity and that it is probably not involved in an ABC transport system.

MATERIALS AND METHODS
Bacterial Strains, Plasmids, and Growth Conditions-The bacterial strains and plasmids are listed in Table I. E. coli and B. subtilis strains were grown in Luria-Bertani broth medium. The following antibiotics were used: ampicillin at 50 g/ml in E. coli and kanamycin, erythromycin, and spectinomycin at 20, 0.3, and 100 g/ml, respectively, in B. subtilis. All antibiotics were from Sigma.
General Molecular Biology Techniques-Unless otherwise stated, all molecular biology procedures were carried out as described in Sambrook and Russell (18). DNA-modifying enzymes were used as recommended by the manufacturer (New England Biolabs). DNA fragments were purified using either a Microcon-30 (Millipore) or the Qiaquick nucleotide removal kit (Qiagen). Cloning of DNA was done either in E. coli DH5␣ or in E. coli C41(DE3) strain. PCR amplifications were done in a 50-l final volume, using Yellow Star polymerase (Eurogentec) as recommended by the manufacturer. All oligonucleotides are listed in Table II. The interrupted bcrC Bs mutant was constructed following the modified long flanking homology-PCR method (5,19) using the bcrC1, bcrC2, bcrC3, bcrC4, 2rspec2_inv, and 2rspec1_inv oligonucleotides (Table II).
Cloning of the bcrC Bs Coding Sequence into the pDG148-Stu or the pET22-Pml Plasmid-The entire bcrC Bs coding sequence was amplified by PCR from the B. subtilis chromosome using the BcrC(pdg)ATG and BcrC(pdg)stop oligonucleotides. The amplified DNA fragment was introduced into the pDG148-Stu plasmid (20). After transformation of the DH5␣ strain with the resulting mixture, plasmid carrying the DNA fragment was used to transform B. subtilis (wild type or mutant) or E. coli C41(DE3) strain.
A 779-bp DNA fragment encompassing the entire bcrC Bs coding sequence but lacking the start codon was amplified by PCR from the B. subtilis chromosome using the BcrC-dir and BcrC-rev oligonucleotides. It was then cloned into the pET22-Pml plasmid, a modified version of pET22bϩ (Novagen), as described previously (21). The resulting mixture was used to transform either the DH5␣ or the C41(DE3) strain. In each case, the sequence inserted in the recombinant plasmid was checked by DNA sequencing.
Preparation of Membrane Fraction and Purification of the BcrC Bs Protein-E. coli C41(DE3) cells carrying the pET22-Pml-bcrC Bs plasmid were grown at 37°C in LB medium supplemented with ampicillin (1-liter culture). When the absorbance at 600 nm reached 0.6, isopropyl-␤-D-thiogalactopyranoside (IPTG) was added at a final concentration of 1 mM, and growth was continued for 4 h. Cells were harvested by low speed centrifugation, washed twice with 10 ml of cold 50 mM tris(hydroxymethyl)aminomethane hydrochloride buffer, pH 8, supplemented with 5 mM MgCl 2 , 1 mM dithiothreitol, and 1 mM phenylmethylsulfonyl fluoride. The cell pellet was suspended in 10 ml of the same buffer, and the cells were lysed, in the cold, by two successive passages through a French pressure cell (16,000 p.s.i.). Cold 500 mM EDTA (500 l), pH 8, was added to the resulting suspension and submitted to a 30-min centrifugation at 15,000 ϫ g at 4°C. Phenylmethylsulfonyl fluoride (1 mM final concentration) was added to the resulting supernatant (crude extract) and then submitted to a 90-min centrifugation at 100,000 ϫ g at 4°C. The pellet was suspended and solubilized by n-dodecyl-␤-Dmaltoside (DM) (0.9%, w/v) following all the steps described by El Ghachi et al. (22). The BcrC Bs purification involved a final affinity step with Ni 2ϩ -nitrilotriacetate agarose (Ni 2ϩ -NTA agarose, from Qiagen) using the procedure recommended by the manufacturer.
Bacitracin Sensitivity Assay-The concentration of bacitracin leading to 50% of growth inhibition (IC 50 ) was determined using the microtiter tray assay described by Ohki et al. (4).
UPP Phosphatase Assay-UPP phosphatase activity of biological samples was determined as described previously (22).
Bioinformatic Analysis-Transmembrane domains in proteins were searched through the TMHMM program (23). Protein Blast (24) searches on B. subtilis were done on the SubtiList data base (25). Membership to a particular Pfam family (26) was defined using the Pfam database.

Reserpine Does Not Affect BcrC Bs
Functioning-The plant alkaloid reserpine is a powerful inhibitor of both mammalian and Gram-positive bacterial efflux systems, and its inhibitory effect on bacteria ABC transporters is well documented (27,28). Accordingly, the addition of reserpine (at sublethal concentration) during bacterial growth should drastically decrease the bacitracin resistance of the cells if this resistance is mostly due to an ABC transporter. If it is not, almost no effect should be observed. As shown in Fig. 1, the addition of reserpine to a culture of the B. subtilis 168 parental strain resulted in a 16-fold decrease in the cellular IC 50 for bacitracin, indicating that at least one system is strongly inhibited. Similarly, when the BcrC Bs Ϫ BSmrs173 mutant in which only the putative ABC transporter BceAB Bs system is functional was grown in the presence of reserpine, the IC 50 of the cells was again dramatically decreased (18-fold). Inversely, reserpine had only a slight effect (less than 2-fold) on the IC 50 of the BceAB Bs Ϫ BFS82 mutant in which the resistance to bacitracin depends only on the BcrC Bs system. Together, these results indicated that reserpine (i) inhibits the functioning of the BceAB Bs involved in B. subtilis resistance but (ii) does not affect that of BcrC Bs , making it very unlikely for this protein to be involved as a membrane-spanning domain partner in an ABC transporter.
Overexpression of BcrC Bs Allows B. subtilis and E. coli to Stand Higher Concentrations of Bacitracin-When the The bcrC Bs coding sequence was then cloned into a modified version of the pET22-Pml expression vector (21) immediately downstream from the His 6 tag coding sequence carried by the vector, leading to plasmid pET22-Pml-bcrC Bs . In this construction, expression of the gene fusion encoding a His 6 -BcrC Bs hybrid protein is controlled by the addition of IPTG. After growth in the presence of IPTG of the C41(DE3)/pET22-Pml-bcrC Bs strain, the IC 50 of the cells for bacitracin increased from 1.4 to 5.6 mM, indicating that this hybrid protein is active (data not shown).
Overexpression of BcrC Bs Increases UPP Phosphatase Activity-The E. coli C41(DE3) cells harboring plasmid pET22-Pml-bcrC Bs were grown without IPTG to be used as control cells or with 1 mM IPTG to overproduce the His 6 -BcrC Bs protein for further studies. Crude extracts and membrane fractions prepared from these cells, as indicated under "Materials and Methods," were analyzed by SDS-PAGE. Coomassie Blue stain-ing of the gel (Fig. 3, lanes 1-3) showed a broad band corresponding to a protein species of about 22 kDa, the size expected for the His 6 -BcrC Bs hybrid protein, in the membrane fraction of the cells in which the expression of the gene had been induced by IPTG. Scanning of the stained gel indicated that this band represented almost 41% of all membrane proteins. In contrast, no such band was detected in the soluble fraction of the same cells or in soluble fractions of the cells grown without IPTG   (data not shown). There was, however, a faint band at the same position in the crude extract of non-induced cells, probably due to a leakage of the promoter controlling the T7 polymerase expression (Fig. 3, lane 1).
The identification of this ϳ22-kDa protein species with the His 6 -BcrC Bs protein was confirmed by Western blot detection using antibodies directed against the His 6 tag. As expected, no signal could be detected in the other soluble fractions (see above), indicating that the His 6 -BcrC Bs hybrid protein is totally inserted in the membrane (data not shown). Although this protein seemed to be present in the membrane of the noninduced control cells, it was not detected by the antibody, probably because it has a low affinity for the His tag. UPP phosphatase activity was then determined as described previously (22) in membrane fractions originating from control cells and from cells overproducing the His 6 -BcrC Bs protein. The membrane fraction from the non-induced C41(DE3)/pET22-Pml-bcrC Bs cells already has UPP phosphatase activity, which reaches 13.5 Ϯ 0.5 nmol of UPP transformed per minute and per mg of protein (nmol min Ϫ1 mg Ϫ1 ). This level was higher than that found (0.3 nmol min Ϫ1 mg Ϫ1) in a membrane preparation from the C43(DE3) E. coli strain containing the plasmid without insert, and it probably reflects a leakage of the T7 polymerase promoter that controls the expression of the inserted gene in the pET22-Pml plasmid.
As expected, however, a much higher level of UPP phosphatase activity, up to 198 Ϯ 9 nmol min Ϫ1 mg Ϫ1 (a 660-fold increase), was observed in the membrane fraction originating from the C41(DE3)/pET22-Pml-bcrC Bs cells in which expression of the tagged BcrC Bs protein had been induced. This observation strongly supported the hypothesis that the BcrC Bs protein has UPP phosphatase activity. Accounting for the very low basal level of UPP phosphatase activity in E. coli cell membranes (0.3 nmol min Ϫ1 mg Ϫ1 ), His 6 -BcrC Bs corresponds to almost all the activity detected in the membrane preparation coming from the induced cells. In this fraction in which the His 6 -BcrC Bs protein represents 41 Ϯ 1% of all proteins, the specific activity of the pure membrane-embedded His 6 -BcrC Bs protein can be estimated to be greater than 450 nmol min Ϫ1 mg Ϫ1 .
Purified BcrC Bs Protein Displays UPP Phosphatase Activity-The detergent DM is efficient for extracting the BacA Ec protein from membranes of E. coli cells (22). Accordingly, the membrane fractions originating from the C41(DE3)/pET22-Pml-bcrC Bs cells overproducing or not the His 6 -BcrC Bs protein were subjected to four successive extraction treatments with DM, leading to fractions DM1-4. SDS-PAGE analysis of each fraction followed by Coomassie Blue staining and scanning of the gel indicated that up to 93% of the membrane proteins were recovered in fractions DM1 and DM2 (Fig. 3, lanes 4 and 5). The His 6 -BcrC Bs protein was observed in the DM1 and DM2 fractions but not in the DM3 and DM4 fractions. Scanning of the Coomassie Blue stained gel indicated that the His 6 -BcrC Bs protein, if alone in the 22-kDa band, represented 49 Ϯ 1% of the proteins of fraction DM1. The UPP phosphatase activity of this fraction reached 230 Ϯ 18 nmol min Ϫ1 mg Ϫ1 (Table III). Thus, because the estimated specific activities of the His 6 -BcrC Bs protein before and after DM extraction were similar, the protein is apparently not damaged during this extraction step. 6.7 mg of proteins (supposedly containing 3.3 mg of the His 6 -BcrC Bs protein, Table III) from the DM1 fraction was subjected to Ni 2ϩ -NTA agarose purification. Each fraction (flow-through, washes, and elutions with imidazole) was then analyzed for its protein content by SDS-PAGE analysis followed by Coomassie Blue staining (Fig. 4) and Western blot analysis (data not shown). 4.7 mg of proteins (supposedly containing 1.9 mg of the His 6 -BcrC Bs protein, Table III) was recovered in the flow-through, washes, and 10 mM imidazole pooled fraction. Thus, almost 58% of the His 6 -BcrC Bs protein was not retained on the Ni 2ϩ -NTA agarose column, explaining the very poor yield of His 6 -BcrC Bs purification. This might be due to a low affinity of the His 6 -BcrC protein for Ni 2ϩ -NTA in the presence of n-dodecyl-␤-D-maltoside. The His 6 -BcrC Bs protein was detected in the 30, 60, and 80 mM imidazole fractions (0.2, 0.32, and 0.56 mg, respectively; Table III). Unfortunately, all these fractions contained some contaminant proteins as judged on SDS-PAGE gel (Fig. 4). Only the last elution fraction done with 300 mM imidazole showed a single band corresponding to the His 6 -BcrC Bs protein (Fig. 4) and allowed us to recover about 0.12 mg of protein. Determination of its UPP phosphatase activity indicated that it has a specific activity of 730 Ϯ 60 nmol min Ϫ1 mg Ϫ1 (Table III).
When comparing our purification procedure to that used for BacA Ec (see "Materials and Methods"), it appeared that the additional low speed centrifugation step greatly facilitates the action of the mild detergent used to extract the proteins from membranes. The elimination of large debris and unbroken cells before the high speed centrifugation step led to a membrane preparation devoid of jamming material and thus more susceptible to the DM treatment and also allowed a better extraction of all membrane-embedded proteins, with 80% of them extracted in the first DM fraction. BcrC Bs Protein Is Not Active on Bacitracin-bound UPP-The UPP phosphatase activity of the purified His 6 -BcrC Bs protein was determined in the presence of various concentrations of bacitracin and was compared with that of the native membrane-embedded enzyme. For that purpose, we used aliquots of both preparations giving the same percentage of substrate transformation. As seen in Fig. 5, similar results were obtained with both preparations, the fraction of dephosphorylated substrate remaining almost unaffected as long as the concentration of the antibiotic was lower than the initial concentration of the substrate. When both concentrations were of the same order, the activity started to decrease in parallel with the increase in bacitracin concentration. According to the association constant (10 Ϫ6 M) between UPP and bacitracin (1) and assuming that there is no interaction between bacitracin and the enzyme, our results indicated that the enzyme, native or solubilized, has no access to bacitracin-bound UPP and acts only on free UPP.
Research of Other Putative UPP Phosphatases in B. subtilis-As the B. subtilis bcrC Bs deleted mutant could still grow, we looked within the protein sequences of the bacterium for other proteins able to synthesize or recycle UPP. First, we focused on possible B. subtilis ortholog(s) of BacA Ec UPP phosphatase (22). In the Pfam data base, BacA Ec belongs to the PF02673 family, whereas BcrC Bs and BcrC Bl belong to the PF01569 family. Using BacA Ec to scan all the putative proteins from B. subtilis, only one hypothetical protein, YubB Bs , showed a high score after Blast screening. A yubB Bs mutant, obtained by insertion of the pMUTIN plasmid (29), was constructed by the B. subtilis functional analysis consortium (30). This suggested that YubB Bs might not be essential in B. subtilis, although Cao and Helmann (10) did not succeed in obtaining a yubB Bs deletion mutant. It is worth noting that deletion mutants have been reported for the bacA orthologous genes in E. coli (22), Staphylococcus aureus, and Streptococcus pneumoniae (31). We constructed the yubB Bs bcrC Bs double mutant, which grew perfectly in LB medium and presented the same sensitivity to bacitracin as the single bcrC Bs mutant (data not shown). These results suggested that (i) YubB Bs does not contribute significantly to bacitracin resistance in B. subtilis and (ii) at least one other protein might be required for UP and peptidoglycan synthesis in this double mutant. DISCUSSION BLAST analysis indicated that BcrC Bs presents the most similarities with proteins from Gram-positive bacteria such as Bacillus anthracis, Bacillus thurinfrengis, Bacillus cereus, and Clostridium acetobutilicum and from Gram-negative bacteria such as Methanosarcina masei, Salmonella typhi, and E. coli, all predicted to belong to the PF01569 Pfam family of PAP2 phosphatases. Because bacitracin prevents the dephosphorylation of UPP into UP, we had hypothesized (6) that the BcrC Bs protein might be an intrinsic membrane phosphatase with UPP phosphatase activity that competed with bacitracin for UPP, rather than the membrane-spanning domain partner of an ABC transporter.
A first indication of the non-inference of BcrC Bs in a transport system is provided by the plant alkaloid reserpine, a strong inhibitor of efflux systems. Reserpine has only a marginal effect on the bacitracin resistance due to the BcrC Bs protein, but it drastically affects that due to the putative transporter BceAB Bs . The inhibitory effect of reserpine on BceAB Bs suggested that it participates in active transport, but this is unlikely for BcrC Bs because reserpine does not inhibit this protein. This latter point prompted us to check whether BcrC Bs has UPP phosphatase activity.
The native BcrC Bs protein was overproduced in B. subtilis or E. coli, and a His 6 -BcrC Bs tagged protein was overproduced in E. coli. In each case, a strong stimulation of the resistance of the cells to bacitracin was observed, indicating that the respective proteins are functional in both bacteria. After several purification steps, the His 6 -BcrC Bs protein was observed in the membrane fraction but not in the soluble fraction of the ultracentrifugation step, which confirmed that BcrC Bs is an intrinsic membrane protein.
As the membrane fraction enriched in the His 6 -BcrC Bs protein had significantly increased UPP phosphatase activity, the protein was purified by the same protocol as that for BacA Ec of E. coli, the only protein known to have UPP phosphatase activity (22). The electrophoretically pure extracted BcrC Bs protein had a significant UPP phosphatase specific activity of the same order of magnitude as that of BacA Ec (2200 nmol min Ϫ1 mg Ϫ1 ) (22).
The specific activity of the electrophoretically pure extracted BcrC Bs protein did not differ significantly from that of the membrane-embedded BcrC Bs protein (730 versus 450 nmol min Ϫ1 mg Ϫ1 ), and no protein had access to the bacitracinbound UPP. That the protein presents the same enzymatic Phosphatase activity was tested as described under "Materials and Methods" on [ 14 C]UPP used at a 2.7 M final concentration. Bacitracin was premixed to the substrate before adding the biological samples to test. Different concentrations of bacitracin were used to obtain the indicated UPP/ bacitracin concentration ratios. After 1 h of incubation at 37°C, samples were subjected to thin layer chromatography as described previously (22 characteristics in different environments suggests that it has not been dramatically affected by the extraction procedure. One can then hypothesize that the quaternary structure of the solubilized form may not be very different from that of the membrane form, at least in the vicinity of the active center. This consideration is of interest for any structure/function study of the BcrC Bs UPP phosphatase. Although BcrC Bs and BacA Ec belong to completely different Pfam families, their specific activities are highly similar. Thus, for the first time, the enzymatic characterization of a BcrC Bs protein has been performed and, as we had suggested in an earlier publication (6), the BcrC Bs protein has UPP phosphatase activity and is probably not involved in the formation of an ABC transporter. It is then tempting to postulate that this will be true for all BcrC orthologs and that they all have UPP phosphatase activity.
Only 24 members of the PAP2 superfamily have been biochemically characterized; all others are considered putative or hypothetical proteins. All 24 characterized proteins but two display phosphatase activity. Interestingly, the UPP phosphatase activity of the BcrC Bs protein enlarges the spectrum of the phosphatases of the PAP2 protein superfamily known to contain acid and neutral phosphatases (32,33), lipid A 1-phosphatase (15), phosphotyrosyl phosphatase (34), glucose-6-phosphatase (35), phosphatidylglycerophosphate phosphatase (14), dolichylpyrophosphate phosphatase (16), lipid phosphate phosphohydrolase (36), diacylglycerol pyrophosphate phosphatase (37), and sphingosin-1-phosphate phosphatase (38) (only one reference is given for each type of substrate). Closer examination in the UniProt data base of the two proteins of the PAP2 superfamily, Q6U5Q3 and Q6WB33, referred to as signal peptidases, reveals that they bear, in addition to the PAP2 domain, a Peptidase_A8 domain (PF01252) responsible for signal peptidase activity (39). It is then probable that all members of the PAP2 superfamily have phosphatase activity in addition to other enzymatic activity or not.
In B. subtilis, BcrC Bs participates in bacitracin resistance concomitantly with the BceAB system, and their structural genes belong to the bacitracin regulon (5). The present work showed that BcrC Bs acts as a UPP phosphatase, thus i) withdrawing the bacitracin target and ii) participating in UP metabolism. As pointed out in the Introduction, UP is a carrier lipid essential for the synthesis of many cell wall polymers, and the bacterium requires a minimal level of global UPP phosphatase for growth. That no growth defect could be observed in the B. subtilis bcrC Bs mutant indicates that the BcrC Bs enzyme is dispensable and that other protein(s) fulfill the need for UPP phosphatase. The YubB Bs protein could be a candidate since it is orthologous to BacA Ec . The yubB Bs mutant being viable without visible growth defect, the YubB Bs protein also seems to be dispensable. Since a yubB Bs bcrC Bs double mutant is also viable without visible growth defect, we are currently looking for another protein able to ensure the viability of this B. subtilis double mutant.