Identification of Multiple Genes Encoding Membrane Proteins with Undecaprenyl Pyrophosphate Phosphatase (UppP) Activity in Escherichia coli*

The bacA gene product of Escherichia coli was recently purified to near homogeneity and identified as an undecaprenyl pyrophosphate phosphatase activity (El Ghachi, M., Bouhss, A., Blanot, D., and Mengin-Lecreulx, D. (2004) J. Biol. Chem. 279, 30106–30113). The enzyme function is to synthesize the carrier lipid undecaprenyl phosphate that is essential for the biosynthesis of peptidoglycan and other cell wall components. The inactivation of the chromosomal bacA gene was not lethal but led to a significant, but not total, depletion of undecaprenyl pyrophosphate phosphatase activity in E. coli membranes, suggesting that other(s) protein(s) should exist and account for the residual activity and viability of the mutant strain. Here we report that inactivation of two additional genes, ybjG and pgpB, is required to abolish growth of the bacA mutant strain. Overexpression of either of these genes, or of a fourth identified one, yeiU, is shown to result in bacitracin resistance and increased levels of undecaprenyl pyrophosphate phosphatase activity, as previously observed for bacA. A thermosensitive conditional triple mutant ΔbacA,ΔybjG,ΔpgpB in which the expression of bacA is impaired at 42 °C was constructed. This strain was shown to accumulate soluble peptidoglycan nucleotide precursors and to lyse when grown at the restrictive temperature, due to the depletion of the pool of undecaprenyl phosphate and consequent arrest of cell wall synthesis. This work provides evidence that two different classes of proteins exhibit undecaprenyl pyrophosphate phosphatase activity in E. coli and probably other bacterial species; they are the BacA enzyme and several members from a superfamily of phosphatases that, different from BacA, share in common a characteristic phosphatase sequence motif.

An essential carrier lipid, undecaprenyl phosphate (C 55 -P), 1 is required for the synthesis of various bacterial cell wall polymers such as peptidoglycan, lipopolysaccharides, and teichoic acids (1-5) (Scheme 1). It is synthesized as a pyrophosphate precursor (C 55 -PP) by the addition of eight isoprene units to farnesyl pyrophosphate, a reaction catalyzed by the well characterized cis-prenyl-pyrophosphate synthase UppS (6 -10). However, genes and enzymes involved in subsequent steps of C 55 -P synthesis and recycling still remained to be identified. Bacitracin is a dodecapeptide antibiotic known to specifically block this metabolism by forming a specific complex with C 55 -PP. As a result, cell wall biosynthesis is inhibited and cell lysis finally occurs (11)(12)(13)(14). Bacillus licheniformis strains that produce bacitracin are resistant to this antibiotic due to the presence of an appropriate ABC transporter efflux system (15,16). Several mutations leading to bacitracin resistance were identified in Escherichia coli and other Gram-negative bacteria. Interestingly, all these mutations were shown to block the synthesis of non-essential cell envelope polymers such as osmoregulated periplasmic glycans and capsule polysaccharides that also require the C 55 -P carrier lipid for their formation (17,18). The reduced in vivo requirements for C 55 -P resulting in higher availability of the carrier lipid for the synthesis of essential polymers (peptidoglycan, lipopolysaccharides) most probably explain the decreased mutant cell sensitivity to bacitracin. Not only mutations but also gene overexpression were earlier reported to be associated with bacitracin resistance (19,20). In particular, we recently demonstrated that one such E. coli gene, named bacA (19), encoded a membrane-bound C 55 -PP phosphatase activity (21). Overexpression of a C 55 -PP phosphatase activity might very likely accelerate the conversion of the pool of C 55 -PP, the bacitracin target, to C 55 -P, resulting in an increased resistance of the cell to this antibiotic.
The dephosphorylation of C 55 -PP to C 55 -P was expected to be an essential step in the metabolism of the carrier lipid. The fact that the bacA gene could be disrupted in the chromosome without apparent growth defect in E. coli, Staphylococcus aureus, and Streptococcus pneumoniae (21,22) was thus quite surprising. As no bacA homologous gene could be identified in the genome of these bacterial species, either the involvement of other phosphatase(s) or a metabolic bypass of this dephosphorylation step could be envisioned. As a residual 25% C 55 -PP phosphatase activity was still detected in the bacA mutant strain, other protein species might account for this activity. These other C 55 -PP phosphatases could be specific to C 55 -PP or * This work was supported in part by grants from the European Community (FP6, LSHM-CT-2003-503335, "COBRA" project) and from the Centre National de la Recherche Scientifique (UMR 8619). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Supported by Hoechst Marion Roussel AG. Present address: Unité des Yersinia, Institut Pasteur, 28  have a distinct function and act non-specifically on this substrate at a sufficient rate to sustain cell growth requirements. We have demonstrated here that multiple genes encoding proteins with C 55 -PP phosphatase activity exist in E. coli and that the inactivation of three of them is required to block cell wall synthesis and provoke cell lysis.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Growth Conditions-The E. coli strains DH5␣ (Invitrogen), JM83 (23), and C43(DE3) (Avidis-France) were used as hosts for plasmids (Table I). The plasmid vector pTrc99A was obtained from Amersham Biosciences, and the pMAK705 plasmid bearing a thermosensitive replicon was a kind gift from S. R. Kushner (24). The BW25113 strain and the pKD3, pKD4, pKD46, and pCP20 plasmids used for gene disruption experiments (25) were kindly provided by B. Wanner via the E. coli Genetic Stock Center (Yale University). The BW25113 ⌬bacA::Cam R strain was previously described (21), and other mutant strains constructed in this study are listed in Table I. 2YT medium (26) was used for growing cells, and growth was monitored at 600 nm with a Shimadzu UV-1601 spectrophotometer. For strains carrying drug resistance genes, ampicillin, kanamycin, and chloramphenicol were used at 100, 35, and 25 g⅐ml Ϫ1 , respectively.
General DNA Techniques and E. coli Cell Transformation-Polymerase chain reaction (PCR) amplification of genes from the E. coli chromosome was performed in a Thermocycler 60 apparatus (Bio-med) using the Expand-Fidelity polymerase (Roche Applied Science). The DNA fragments were purified using the Wizard PCR Preps DNA purification kit (Promega), and standard procedures for endonuclease digestions, agarose electrophoresis, ligation, and plasmid purification were used (27). E. coli cells were transformed with plasmid DNA by the method of Dagert and Ehrlich (28) or by electroporation.
Generation of Chromosomal Gene Deletions-The E. coli strain BW25113 ⌬bacA carrying a chromosomal deletion of the bacA gene was constructed recently (21) using the efficient method of Datsenko and Wanner (25). The same procedure was used to generate deletions of the ybjG, yeiU, pgpA, and pgpB genes. In each case, Inact1 and Inact2 oligonucleotides (Table II) were used for PCR amplification of either the chloramphenicol resistance gene from pKD3 or the kanamycin resistance gene from pKD4, flanked by sequences designed for specific disruption of the gene of interest. The PCR products were transformed by electroporation into BW25113(pKD46) cells expressing the phage lambda Red recombinase (25). Antibiotic-resistant clones were isolated, and the disruption of the bacA, ybjG, yeiU, pgpA, and pgpB genes in the chromosome was verified by PCR using BacA1 and BacA2, YbjG1 and YbjG2, YeiU1 and YeiU2, PgpA1 and PgpA2, and PgpB1 and PgpB3 as primers, respectively (Table II). Excision of the antibiotic resistance genes from the chromosome of the mutant strains was then obtained by transformation with the pCP20 plasmid expressing the Flp recombinase, as described previously (25). Antibiotic-sensitive clones were isolated, and the excision of the cassette was confirmed by PCR analysis as described above. Strains carrying multiple chromosomal gene deletions were generated by transduction of antibiotic resistance markers with phage P1 (26), followed by excision of the cassettes from the chromosome.
Construction of Plasmids-The plasmid pTrcBac1 allowing expression of the bacA gene under control of the isopropyl-␤-D-thiogalactopyranoside (IPTG)-dependent trc promoter has been previously described (21). Plasmids for expression of the ybjG, yeiU, and pgpB genes were similarly constructed as follows. The ybjG gene was amplified from the chromosome using oligonucleotides YbjG1 and YbjG3 (Table II), and the resulting fragment was treated with BspLU11I and HindIII and cloned between the compatible NcoI and HindIII sites of vector pTrc99A, generating pTrcYbjG. The yeiU gene was amplified from the chromosome using oligonucleotides YeiU1 and YeiU2; the resulting fragment was treated with BspHI and HindIII and cloned between the NcoI and HindIII sites of pTrc99A, generating pTrcYeiU. The pgpB gene was amplified from the chromosome using oligonucleotides PgpB2 and PgpB3; the resulting fragment was treated with BspHI and HindIII and cloned into the compatible NcoI and HindIII sites of vector pTrc99A, generating pTrcPgpB. A plasmid allowing conditional (thermosensitive) expression of bacA was constructed as follows: the bacA gene and its proximal region were amplified using oligonucleotides BacA1 and BacA3, and the resulting DNA fragment was cut with BamHI and SacI and cloned between the same sites of plasmid vector pMAK705 whose replication is thermosensitive (24), generating pMAKbacA Preparation of Crude Membrane Extracts-The different strains described in this work were grown in 2YT medium (0.5-liter cultures) at either 30, 37, or 42°C. For strains carrying plasmids, growth was performed in the presence of the corresponding antibiotic. When the optical density of the culture reached 0.6, 1 mM IPTG was added for induction of gene overexpression, and growth was continued for 3 h. In all cases, cells were harvested and washed with 40 ml of cold 20 mM potassium phosphate buffer, pH 7.2, containing 2 mM ␤-mercaptoethanol (buffer A). The cell pellet was suspended in 12 ml of the same buffer, and cells were disrupted by sonication in the cold (Bioblock Vibracell sonicator model 72412). The resulting suspension was centrifuged at 4°C for 30 min at 200,000 ϫ g. The pellet consisting of membranes and associated proteins was washed twice with buffer A and finally resuspended in buffer A supplemented with 150 mM NaCl, 1% n-dodecyl-␤-D-maltoside, and 20% glycerol.
Undecaprenyl Pyrophosphate Phosphatase Assay-The assay was performed in a reaction mixture (20 l) containing 100 mM Tris-HCl buffer, pH 7.5, 10 mM MgCl 2 , 3.9 mM n-dodecyl-␤-D-maltoside, 5 M [ 14 C]C 55 -PP (2,035 Bq), and enzyme. The reaction mixture was incubated for 30 min at 37°C, and the reaction was stopped by heating at 100°C for 1 min. As described previously (21), the substrate (C 55 -PP) and reaction product (C 55 -P) were separated by TLC on precoated plates of silica gel 60 (Merck) using diisobutyl ketone/acetic acid/water (8:5:1, v/v/v) as a mobile phase (Rf values of C 55 -PP and C 55 -P were 0.36 and 0.5, respectively). The radioactive spots were located and quantified with a radioactivity scanner (model Multi-Tracermaster LB285; Berthold-France). One unit of enzyme activity corresponds to one nanomole of product formed/min.
Pool Levels of Peptidoglycan Precursors-Cells of the wild-type strain BW25113 and of the thermosensitive mutant strain BWTsbacA (0.8liter cultures) were grown exponentially at 30°C in 2YT medium. At the appropriate cell density (ϳ10 7 cells/ml), the temperature of the culture was either maintained at 30°C or increased to 42°C. Incubation was continued until the first effects on the growth of the mutant strain at 42°C were observed, ϳ2-3 h later. At this time, cells were rapidly chilled to 0°C and harvested in the cold. The analytical procedures used for the extraction, isolation, and quantitation of the peptidoglycan nucleotide precursors were as previously described (29 -31).
Isolation and Quantification of Peptidoglycan-Exponential phase cells of the wild-type BW25113 and mutant BWTsbacA strains were grown at 30°C, or first at 30 and then 42°C as described above. Harvested cells were washed with a cold 0.85% NaCl solution and then resuspended with vigorous stirring in 20 ml of a hot (95-100°C) aqueous 4% sodium dodecyl sulfate solution for 30 min. After standing overnight at room temperature, suspensions were centrifuged for 20 min at 200,000 ϫ g, and the pellets were washed several times with water. Final suspensions made in 2.5 ml of water were homogenized by brief sonication. Aliquots were hydrolyzed and analyzed as described previously (32). The peptidoglycan content was expressed in terms of its characteristic and specific components, MurNAc, GlcNAc, and diaminopimelic acid (32).
Bacitracin Sensitivity-100 l of appropriate dilutions of overnight cultures (10 2 -10 5 cells) were spread onto 2YT plates containing bacitracin at concentrations ranging between 30 and 200 units⅐ml Ϫ1 . For induction of gene overexpression from the pTrc99A-derivative plasmids, IPTG was added at a final concentration of 1 mM. Colonies observed after overnight incubation at 37°C were counted, and results were expressed as the percentage of survival as compared with untreated cultures.
Protein Monitoring-Protein concentrations were determined by using either the Bradford procedure (33) or the QuantiPro BCA assay kit (Sigma) with bovine serum albumin as the standard.
Chemicals-DNA restriction and modification enzymes were obtained from New England Biolabs, and oligonucleotides were from MWG-Biotech. DNA purification kits were from Promega. C 55 -OH, C 55 -P, and C 55 -PP were provided by the Institute of Biochemistry and Biophysics of the Polish Academy of Sciences. Pure BacA protein, [ 14 C]C 55 -PP (1.5-2.2 GBq⅐mmol Ϫ1 ), and UDP-MurNAc peptides were prepared as described previously (21,29). n-Dodecyl-␤-D-maltoside was from Anatrace, and antibiotics and reagents were from Sigma.

RESULTS
Identification of Putative C 55 -PP Phosphatase Genes-The gene encoding C 55 -PP phosphatase activity was expected to be an essential gene, as the function of this enzyme is to generate the carrier lipid C 55 -P that is absolutely required for the biosynthesis of peptidoglycan and other essential cell wall polymers (Scheme 1). The observation that a bacA (uppP) null mutant did not exhibit particular morphological changes or growth defects suggested that other proteins might exhibit such an activity in vivo (21). However, no bacA homologue was identified in the chromosome of E. coli and other bacterial species, namely S. aureus and S. pneumoniae, in which this gene was also proved to be non-essential (21,22). Other putative genes for this activity were therefore searched for in bacterial genomes based essentially on sequence similarity with already known phosphatase activities. Phosphatases of unknown function that had been previously characterized to some extent biochemically were also considered as potential candidates.
A membrane-bound pyrophosphatase of rather broad substrate specificity was earlier purified from Sulfolobus acidocaldarius (34). The bacitracin sensitivity of S. acidocaldarius and the catalytic properties of this enzyme made the authors speculate that it could be a dolichol-PP phosphatase. At that time, only the N-terminal amino acid sequence of the protein was determined, which allowed us to identify the corresponding gene in the S. solfataricus sequenced genome. The product of this gene was predicted to be a highly hydrophobic protein of 220 amino acid residues exhibiting sequence similarity with, and characteristic signature sequences of, members of the phosphatidic acid-phosphatase family. A BLAST search further showed that it presented the highest similarity (ϳ30% residue identity) with proteins BcrC from B. licheniformis, YbjG from E. coli, and YwoA from Bacillus subtilis. Interestingly, the BcrC protein whose function is yet unknown was identified as one of the three components of the ABC transporter system responsible for the self protection of B. licheniformis against the antibiotic it produces, bacitracin (15,16). Furthermore, overexpression of the ybjG gene in E. coli and ywoA gene in B. subtilis was recently reported to increase the resistance to bacitracin of these respective bacterial species (20,(35)(36)(37). Other E. coli genes encoding members of the phosphatidic acid-phosphatase family were yeiU, of unknown function, and pgpB, encoding phosphatidylglycerolphosphate phosphatase (38). It was previously shown that two genes, pgpA and pgpB, could sustain phosphatidylglycerolphosphate phosphatase activity in E. coli (38,39); however, their inactivation did not result in the loss of viability but only in some thermosensitivity of growth. As shown in Fig. 1, sequences identical or quite similar to the common conserved phosphatase motif KX 6 RP-(X 12-54 )-PSGH-(X 31-54 )-SRX 5 HX 3 D, previously identified by Stuckey and Carman (40) and Neuwald (41), could effectively be found in the PgpB, YeiU, and YbjG proteins as well as in the B. subtilis YwoA and B. licheniformis BcrC proteins. Interestingly, this con- a Restriction sites (in bold) introduced in oligonucleotides are indicated in parentheses.
served phosphatase motif was not detected in the sequence of BacA, and the absence of significant sequence homology between BacA and the above mentioned proteins clearly indicated that they belonged to two different classes of proteins. All of them were predicted to be integral membrane proteins having five (YeiU), six (YbjG, PgpB), and eight (BacA) transmembrane segments, respectively. Inactivation of All Three Genes, bacA, ybjG, and pgpB, Is Lethal-An E. coli strain with the chromosomal bacA gene disrupted was recently constructed and shown to be viable (21). Membranes from wild-type and ⌬bacA::Cam R strains were prepared, and enzymatic assays showed a 75% decrease of C 55 -PP phosphatase activity in the mutant strain (21). This result confirmed the identification of the bacA gene product as a C 55 -PP phosphatase and revealed its important, but not exclusive, contribution to the total activity detected in a wild-type strain. Following the efficient mutagenesis procedure of Datsenko and Wanner (25), other putative genes for this function, ybjG, yeiU, pgpB, and pgpA, were inactivated on the chromosome. Strains with either of these four genes deleted and replaced by an antibiotic resistance cartridge were obtained, indicating that none of them was essential for growth, as previously observed for bacA (21). Strains carrying multiple gene deletions were then constructed by successive rounds of transduction by phage P1 of the integrated antibiotic resistance markers, followed by excision of the latter markers from the chromosome by using the FLP recombinase. Although most of the different combinations of multiple mutations were generated without any difficulty (Table I), those involving inactiva-tion of the triplet of genes bacA, ybjG, and pgpB were never obtained, suggesting a lethal effect of this particular triple mutation. This finding prompted us to construct a triple mutant strain exhibiting conditional expression of bacA. For this purpose, the bacA gene and its proximal region were cloned into pMAK705, a low copy number plasmid whose expression is impaired at 42°C. This plasmid, pMAKbacA, or the control vector pMAK705 were transformed into the BW25113 ⌬bacA,⌬ybjG double mutant strain, and transformants were selected on chloramphenicol plates at 30°C. In each case, one of these transformants was used as a recipient strain for transduction by P1 of the ⌬pgpB::Kan R marker. Many clones were isolated on kanamycin plates at 30°C when the recipient strain contained the pMAKbacA plasmid, but none was obtained when the recipient strain contained the pMAK705 vector. PCR experiments confirmed that the kanamycin-resistant clones thus isolated effectively had all three genes, bacA, ybjG, and pgpB, inactivated in the chromosome. At the restrictive temperature for plasmid replication (42°C), these clones failed to grow on 2YT plates, indicating that the triple mutant ⌬bacA,⌬ybjG,⌬pgpB::Kan R (pMAKbacA), named BWTsbacA, was not viable in the absence of the plasmid carrying the wild-type bacA gene. This finding explained why first attempts to generate this triple mutant by phage P1 transduction were unsuccessful.
Physiological Effects of a BacA Depletion in the Triple Mutant Background-As the pMAKbacA plasmid bears a thermosensitive replicon, the effects of the specific inactivation of the bacA gene were observed by shifting exponentially growing cells of BWTsbacA and BW25113 from 30°C to 42°C. Both strains showed an identical growth rate and morphology when grown at 30°C. However, after 2-3 h at 42°C, mutant cells stopped growing and finally lysed after prolonged incubation at the restrictive temperature, as judged by a net decrease of turbidity of the culture (Fig. 2). Observation by phase contrast microscopy showed that the mutant cells had progressively changed from normal to greatly enlarged rods or ovoids following the temperature upshift, whereas the morphology of the parental BW25113 strain was unaltered (data not shown). The lytic phenotype clearly suggested the inhibition of the synthesis of some essential cell envelope component, such as peptidoglycan. This was perfectly consistent with the previously established function of BacA to dephosphorylate C 55 -PP, thereby providing the essential C 55 -P carrier lipid required for cell wall synthesis. The fact that growth inhibition only occurred after a few hours is explained by the time required for the plasmid and the preexisting BacA enzyme molecules to be progressively diluted and cured from the cells. This relatively short delay, namely a few SCHEME 1. Biosynthesis of undecaprenyl phosphate (C 55 -P) and its use for cell wall peptidoglycan synthesis. The C 55 -PP precursor synthesized by the UppS synthetase is dephosphorylated by the UppP enzyme(s) into C 55 -P. The MraY and MurG enzymes catalyze the successive transfers of the MurNAc-pentapeptide (M-pep) and GlcNAc (G) motifs from the peptidoglycan nucleotide precursors onto the C 55 -P lipid, generating the lipid I and lipid II intermediates, respectively. The lipid II is then translocated to the outer side of the membrane where polymerization reactions catalyzed by the penicillin-binding proteins (PBPs) occur. At the end of this process, the carrier lipid is released in C 55 -PP form and should be dephosphorylated before being reused for de novo peptidoglycan (or other cell wall components) synthesis. It is at present not known whether the regeneration of C 55 -P occurs at the outer side of the membrane or only after translocation (flippase?) of the C 55 -PP to the inner side. (40). Numbers indicate the lengths of amino acid sequences preceding and following the phosphatase motifs in the protein sequences. The total number of amino acids and molecular mass of each protein are indicated in parentheses.

FIG. 1. Phosphatase motifs detected in the amino acid sequences of demonstrated (E. coli PgpB, YbjG, and YeiU proteins) or putative (B. subtilis YwoA and B. licheniformis BcrC proteins) C 55 -PP phosphatase activities, as compared with the consensus sequence KX 6 RP-(X 12-54 )-PSGH-(X 31-54 )-SRX 5 HX 3 D previously reported by Stuckey and Carman
generation times for the bacteria, suggests that this enzyme is probably not present in great excess in cells.
We confirmed that the effects of the triple mutation on bacterial growth were correlated with the depletion of C 55 -PP phosphatase activity. Indeed, BWTsbacA cells contained about 5-fold less C 55 -PP phosphatase activity than BW25113 cells (0.06 and 0.3 units/mg of proteins, respectively) after growth for 2-3 h at 42°C. An accumulation of C 55 -PP and a concomitant depletion of C 55 -P is expected to occur after growth of the mutant at the restrictive temperature. A depletion of the pool of C 55 -P should theoretically result in the accumulation of nucleotide precursors used for the synthesis of the different cell envelope components. As shown in Table III, both wild-type and mutant strains contained similar levels of the two main nucleotide peptidoglycan precursors, UDP-GlcNAc and UDP-MurNAc-pentapeptide, when grown at 30°C. However, a significant accumulation of the pool of UDP-MurNAc-pentapeptide, the last cytoplasmic peptidoglycan precursor and substrate of the C 55 -P-dependent MraY translocase activity (Scheme 1), was observed in mutant cells, but not in wild-type cells, following a temperature shift to 42°C (Table III). This was clearly correlated to the arrest of synthesis of BacA and depletion of the pool of C 55 -P. UDP-GlcNAc also somewhat accumulated after growth at 42°C, but this was similarly observed in both types of cells and consequently not due to the mutation. It should be noted that UDP-GlcNAc is involved in other pathways and is in particular the substrate of the C 55 -P-dependent WecA enzyme involved in O-antigen synthesis. The fact that UDP-GlcNAc did not accumulate to a greater extent in mutant cells was likely explained by its unrestricted consumption for the synthesis of the other cytoplasmic peptidoglycan precursors. Furthermore, we showed that the peptidoglycan content of the triple mutant cells was decreased by ϳ20% after growth at the restrictive temperature (Table III), demonstrating that the lytic phenotype was effectively due to the inhibition of the biosynthesis of at least this essential cell envelope component.
Overexpression of ybjG, yeiU, pgpB, and bacA Genes Results in Increased C 55 -PP Phosphatase Activity-The overexpression of the E. coli bacA gene from the pTrcBac1 plasmid, a pTrc99A derivative, was previously shown to be associated with a 280fold increase of C 55 -PP phosphatase activity in membranes (21). Cloning and IPTG-induced expression from the same vector of the ybjG, yeiU, and pgpB genes also led to significant increases of the latter activity, by ϳ5-, 10-, and 70-fold, respec-tively. This result clearly confirmed the participation of these different proteins in the total C 55 -PP phosphatase activity detected in E. coli membranes. As the levels of overproduction of these different proteins could not be accurately estimated, these data should yet only be considered as apparent and qualitative. Only the purification to homogeneity of these different membrane proteins will enable us to precisely determine and compare their respective specific activities. Interestingly, although the yeiU gene was shown here to encode a protein with C 55 -PP phosphatase activity, the inactivation of this gene was not required to create the conditional lethal mutant strain BWTsbacA. This result suggests that the C 55 -PP phosphatase activity originating from the expression of the chromosomal yeiU gene is not sufficient to sustain the specific requirements for cell envelope biosynthesis and growth.
Overexpression of ybjG, pgpB, and bacA Genes Results in Increased Bacitracin Resistance-As compared with Gram-positive species, E. coli and other Gram-negative species are not very sensitive to bacitracin, most likely because of the presence of the additional outer membrane barrier and limited uptake (20 -22). For this reason, only limited variations were expected in sensitivity to this antibiotic of the different strains generated in this work. Growth of the wild-type strain BW25113 was unaffected for concentrations of bacitracin up to 90 units/ml; concentrations of 120 and 160 units/ml led to 5 and 0% survival, respectively. The different ybjG, yeiU, pgpB, and bacA single mutants and the ⌬bacA,⌬pgpB double mutant did not show modified sensitivity to the antibiotic. The two other double mutants tested, ⌬bacA,⌬ybjG and ⌬pgpB,⌬ybjG, appeared slightly more susceptible than the parental strain because they resisted only up to 60 units of bacitracin/ml. Further introduction of the ⌬yeiU mutation in the double mutant ⌬bacA,⌬pgpB decreased its level of resistance from 90 to 60 units/ml. The effects of the overexpression of these different genes from the pTrc99A plasmid vector were also investigated. In the absence of IPTG, cells harboring either the plasmid vector or its derivatives carrying the genes ybjG, yeiU, pgpB, or bacA behaved similarly (5% survival at 120 units/ml). However, induction with 1 mM IPTG allowed cells overexpressing the genes ybjG, pgpB, or bacA to resist up to 200 units of bacitracin/ml (100% survival was observed in all cases at 160 units/ml, and 17, 20, and 70% survival was observed at 200 units/ml, respectively). In the case of the pTrcYeiU plasmid, an arrest of growth immediately followed the addition of IPTG, precluding analyses of the sensibility of the corresponding strain to the antibiotic. The correlation existing between the levels of C 55 -PP phosphatase activity and the cell sensitivity to bacitracin was thus clearly confirmed and validated with the products of the three genes whose co-inactivation was required to affect E. coli cell viability, namely bacA, ybjG, and pgpB.

TABLE III
Peptidoglycan content and pool levels of its main precursors in the parental and thermosensitive mutant strain Cells were grown exponentially in 2YT medium at 30°C, or first at 30°C and then for 2-3 h at 42°C (the time at which the growth rate of the mutant strain began to decrease (see Fig. 2). Cells were harvested, and peptidoglycan and its precursors were extracted and quantified as described under "Experimental Procedures." The peptidoglycan content is expressed in terms of its diaminopimelic acid content. In bacteria, the C 55 -P carrier lipid is required for the synthesis of a variety of cell wall polymers such as peptidoglycan, lipopolysaccharides, and teichoic acids (1)(2)(3)(4)(5), but curiously, its metabolism has been only partially elucidated to date. A cis-prenyl pyrophosphate synthase, UppS, first catalyzes the formation of its precursor, C 55 -PP, by successive additions of eight isoprene units to farnesyl pyrophosphate. This enzyme has been characterized in detail, both biochemically and structurally (6 -10). The next step consists of the dephosphorylation of C 55 -PP into C 55 -P. Bacitracin, a cyclic polypeptide antibiotic, is known to inhibit this reaction through sequestration of the substrate C 55 -PP, thereby provoking an arrest of cell wall synthesis and finally cell lysis (11)(12)(13)(14). The first gene encoding a C 55 -PP phosphatase was only recently identified (21). This gene had initially been named bacA by Cain et al. (19) because its overexpression resulted in bacitracin resistance, and these authors had hypothesized that it could encode a C 55 -OH kinase activity. In fact, the recent purification to homogeneity and characterization of the BacA protein allowed us to unambiguously show that it exhibited C 55 -PP phosphatase activity, but not C 55 -OH kinase activity. We therefore proposed to rename this gene uppP for Undecaprenyl PyroPhosphate Phosphatase (21).
Considering the indispensable function of the bacA gene product, which is to generate the essential C 55 -P lipid, the observation that this gene could be deleted from the chromosome without apparent effect on cell morphology and cell growth was clearly unexpected. In fact, we observed that the C 55 -PP phosphatase activity was not completely abolished but reduced by 75% in a bacA mutant (21). This finding validated the newly established function of BacA and suggested that other protein species might be involved to catalyze such an activity and sustain the in vivo requirements for C 55 -P molecules of cell wall synthesis. It could be hypothesized that such additional phosphatase(s) either are specific to C 55 -PP or could have a different cellular function and act non-specifically on this substrate. Several genes encoding putative candidate proteins for this function were identified in the genome of E. coli. All of these proteins were predicted to be integral membrane proteins and contained a characteristic phosphatase signature (40,41) in their amino acid sequences (see Fig. 1). Two of these genes, ybjG and yeiU, had unknown functions to date, but the pgpB gene was already known to encode one of the two main phosphatidylglycerolphosphate phosphatases of E. coli (38,39). PgpA, the other phosphatidylglycerolphosphate phosphatase, did not exhibit the same phosphatase motif. All these different genes were disrupted individually in the chromosome without apparent effect on cell growth, and the construction of all combinations of double to quadruple mutants allowed us to demonstrate that the co-inactivation of genes bacA, ybjG, and pgpB was lethal. A thermosensitive conditional triple mutant strain was generated in which bacA expression was blocked at the restrictive temperature. Its lytic thermosensitive phenotype, which was clearly correlated with an inhibition of cell wall peptidoglycan synthesis, confirmed the involvement of these three proteins in the synthesis of the essential C 55 -P carrier lipid.
As observed previously with bacA, the overexpression of genes ybjG and pgpB led to an increased level of C 55 -PP phosphatase activity in E. coli cells, confirming that the products of these two genes effectively catalyze such an activity. This was in agreement with the finding that not only BacA, but also YbjG and PgpB proteins, participated in the total activity detected in wild-type cells and could efficiently complement a bacA mutation. The presence of an intact chromosomal copy of only one of the bacA, ybjG, and pgpB genes was sufficient to allow cell growth. Although the YeiU protein also appeared to exhibit C 55 -PP phosphatase activity, the inactivation experiments clearly showed that this gene alone was not able to sustain the cell growth requirements under the conditions tested. The purification and characterization of these different proteins is now required to determine and compare their respective specific activities and substrate specificity. This will help to discern phosphatase activities specifically involved in C 55 -P metabolism from non-or less specific phosphatases of distinct metabolic function that can also use C 55 -PP as a substrate. This also raises the question of how many specific C 55 -PP phosphatases are expected to be involved in this process. C 55 -PP is initially synthesized at the inner side of the cytoplasmic membrane, but it is also released at the outer side of the membrane during late polymerization steps of peptidoglycan biosynthesis (Scheme 1). Whether the dephosphorylation of C 55 -PP occurs on both membrane sides or only on one side remains to be elucidated. Depending on the model considered, the involvement of two phosphatases with catalytic sites orientated toward the cytoplasm and the periplasm or of a single phosphatase could be envisaged. In both cases, a transbilayer movement of the carrier lipid should occur that could be mediated by a putative associated flippase.
As observed previously with bacA (19,21), the overexpression of the different genes identified here was correlated to an increased resistance to bacitracin. As compared with Grampositive bacterial species, E. coli is not very susceptible to this antibiotic, probably owing to the presence of the outer membrane and limited uptake. However, an increase of resistance was observed when cells were transformed with plasmids carrying the ybjG and pgpB genes. It had earlier been reported that the overexpression of the E. coli ybjG gene was associated with bacitracin resistance, but the C 55 -PP activity of its product was not suspected at that time (20). Interestingly, two very close homologues of the ybjG gene product (ϳ32% identity) are encoded by the bcrC gene of B. licheniformis, the bacitracin-producing strain, and the ywoA gene of B. subtilis. The bcrC gene belongs to the bcrABC cluster of genes involved in the natural resistance of the strain to the antibiotic it produces (15,16). It was demonstrated to have an essential role in the resistance process, but its function has never been elucidated (16). Similarly, the function of the B. subtilis ywoA gene product is unknown, but its overexpression was also shown to result in bacitracin resistance (35)(36)(37), and a hypothesis was recently made that this gene could encode a C 55 -PP phosphatase activity (36). As both BcrC and YwoA proteins are predicted to be integral membrane proteins and carry the characteristic phosphatase signature, our results with YbjG clearly suggest that these two proteins should also display C 55 -PP phosphatase activity.
Our data prove the involvement of more than one C 55 -PP phosphatase in E. coli. It is not clear whether this situation will be similarly observed in other bacterial species. In this search, we identified genes of unknown function as well as genes with already assigned related function, all of them encoding putative or validated phosphatases. The need to inactivate several genes to block C 55 -P synthesis does not indicate that all this machinery was dedicated to this function but only reveals that several enzymes can do the job at a rate sufficient to sustain growth of E. coli. Most probably, the number of C 55 -PP phosphatase genes identified in a given bacterial species will be dependent on the specific activity and, more importantly, substrate specificity of the encoded enzymes. The PgpB phosphatidylglycerolphosphate phosphatase is here demonstrated to catalyze dephosphorylation of C 55 -PP. This enzyme has a broad substrate specificity, as it was earlier shown to also act on phosphatidic and lysophosphatidic acid (38,39) as well as on diacylglycerol pyrophosphate (42). This raises the question of its exact primary physiological function. The other identified phosphatidylglycerolphosphate phosphatase, PgpA, is apparently more specific (38,39). The fact that the pgpA gene is intact in the thermosensitive triple mutant strain suggests that its product might not exhibit significant C 55 -PP phosphatase activity, if any. It should be noted that the amino acid sequence of PgpA does not contain the characteristic phosphatase motif or exhibit any obvious homology with that of PgpB, indicating that these two proteins are in fact quite different. As observed in E. coli, the bacA gene was demonstrated to be dispensable in S. aureus and S. pneumoniae (22), suggesting the involvement of additional C 55 -PP phosphatases also in these bacterial species. The absence of BacA in the latter pathogenic species was not, however, without any effect because a hypersensitivity to bacitracin and a reduced virulence in a mouse model of infection were observed. A decrease of the C 55 -PP phosphatase activity was consistent with the bacitracin hypersensitivity, and the reduced virulence of the mutant was likely because of a slowdown of the synthesis of some cell envelope components.
Our data indicate that two different classes of proteins can catalyze the reaction of dephosphorylation of C 55 -PP into C 55 -P in E. coli and probably other bacterial species: the BacA enzyme, on one hand, and several members from a family of phosphatases, on the other hand, all of which appear to be integral membrane proteins. The number of the latter could probably vary from one bacterial species to another, depending on their substrate specificity and ability to catalyze the formation of the carrier lipid at a rate compatible with cell viability. This essential step of C 55 -P metabolism was expected to be an interesting target in a search for new antibiotics. The present demonstration that different classes of enzymes participate in this process could make this search more problematic. The purification and further characterization of these different UppP enzymes will help to elucidate their respective physiological role in C 55 -P metabolism.