A Phosphoethanolamine Transferase Specific for the Outer 3-Deoxy-D-manno-octulosonic Acid Residue of Escherichia coli Lipopolysaccharide

Addition of a phosphoethanolamine (pEtN) moiety to the outer 3-deoxy-d-manno-octulosonic acid (Kdo) residue of lipopolysaccharide (LPS) in WBB06, a heptose-deficient Escherichia coli mutant, occurs when cells are grown in 5–50 mm CaCl2 (Kanipes, M. I., Lin, S., Cotter, R. J., and Raetz, C. R. H. (2001) J. Biol. Chem. 276, 1156–1163). A Ca2+-induced, membrane-bound enzyme was responsible for the transfer of the pEtN unit to the Kdo domain. We now report the identification of the gene encoding the pEtN transferase. E. coli yhjW was cloned and overexpressed, because it is homologous to a putative pEtN transferase implicated in the modification of the β-chain heptose residue of Neisseria meningitidis lipo-oligosaccharide (Mackinnon, F. G., Cox, A. D., Plested, J. S., Tang, C. M., Makepeace, K., Coull, P. A., Wright, J. C., Chalmers, R., Hood, D. W., Richards, J. C., and Moxon, E. R. (2002) Mol. Microbiol. 43, 931–943). In vitro assays with Kdo2-4′-[32P]lipid A as the acceptor showed that YhjW (renamed EptB) utilizes phosphatidylethanolamine in the presence of Ca2+ to transfer the pEtN group. Stoichiometric amounts of diacylglycerol were generated during the EptB-catalyzed transfer of pEtN to Kdo2-lipid A. EptB is an inner membrane protein of 574 amino acid residues with five predicted trans-membrane segments within its N-terminal region. An in-frame replacement of eptB with a kanamycin resistance cassette rendered E. coli WBB06 (but not wild-type W3110) hypersensitive to CaCl2 at 5 mm or higher. Ca2+ hypersensitivity was suppressed by excess Mg2+ in the medium or by restoring the LPS core of WBB06. The latter was achieved by reintroducing the waaC and waaF genes, which encode LPS heptosyl transferases I and II, respectively. Our data demonstrate that pEtN modification of the outer Kdo protected cells containing heptose-deficient LPS from damage by high concentrations of Ca2+. Based on its sequence similarity to EptA(PmrC), we propose that the active site of EptB faces the periplasmic surface of the inner membrane.

The envelope of Gram-negative bacteria consists of an inner membrane (1, 2), a peptidoglycan cell wall (3), and an outer membrane (4). The latter is an asymmetric bilayer with glyc-erophospholipids on its inner surface and lipopolysaccharide (LPS) 1 on the outside surface (5,6). LPS consists of three covalently linked portions (6,7) as follows: 1) the lipid A moiety, a glucosamine-based saccharolipid 2 that serves as the hydrophobic membrane anchor of LPS; 2) the core region, a nonrepeating oligosaccharide modified with phosphate-containing substituents; and 3) the O-antigen, a distal repeating oligosaccharide, which is absent in most strains of Escherichia coli K12 (6). The lipid A moiety and the 3-deoxy-D-manno-octulosonic acid (Kdo) residues of the core are essential for growth of E. coli and most other Gram-negative bacteria (6). Strains lacking all LPS sugars distal to Kdo are termed heptose-deficient or "deep rough" (6,8). These mutants are viable under laboratory conditions (9), but are hypersensitive to serum and antibiotics (4), and often show reduced virulence in animal models (7).
Under certain conditions, strains of E. coli and Salmonella synthesize LPS molecules modified with a phosphoethanolamine (pEtN) group at position 7 of the outer Kdo residue (9) (Fig. 1). Brabetz et al. (9) have reported that E. coli WBB06, which harbors a deletion spanning the heptosyl transferase genes waaC(rfaC) and waaF(rfaF), contains heptose-deficient LPS, modified with pEtN at position 7 of the outer Kdo sugar. Kanipes et al. (10) later demonstrated that pEtN addition to LPS in WBB06 is a consequence of the presence of Ca 2ϩ in the growth medium used by Brabetz et al. (9) and is unrelated to the deletion of the heptosyl transferase genes. A pEtN transferase activity is present in membranes of WBB06 grown in the presence of 5-50 mM Ca 2ϩ (10). The enzyme is stimulated by exogenous phosphatidylethanolamine (PE) and is selective for the outer Kdo residue of Kdo 2 -lipid A and related substrates (10). Addition of EDTA to the in vitro pEtN transferase assay was found to be inhibitory.
We have now identified the gene that encodes the Ca 2ϩinduced pEtN transferase. YhjW is one of six putative pEtN transferases present in E. coli that are homologous to a Neisseria meningitidis gene required for the modification of the lipo-oligosaccharide ␤-chain heptose with pEtN (11). We now demonstrate unambiguously that recombinant YhjW, renamed EptB, utilizes PE as its donor substrate in vitro, generates diacylglycerol as a by-product, and depends upon the presence of Ca 2ϩ in the assay system for activity. Expression of EptB can be re-engineered to be dependent upon the lac promoter, in which case induction of activity no longer requires the addition of Ca 2ϩ to the growth medium. Deletion of eptB in the heptosedeficient mutant WBB06 renders this strain strikingly hypersensitive to Ca 2ϩ at concentrations Ն5 mM, suggesting a function for pEtN modification of the Kdo region in the maintenance of envelope stability. Deletion of eptB in wild-type W3110 does not lead to significant Ca 2ϩ hypersensitivity, indicating that certain outer core sugars, when present, may provide a similar stabilizing effect.

EXPERIMENTAL PROCEDURES
Materials-32 P i and [␥-32 P]ATP were obtained from PerkinElmer Life Sciences. Silica Gel 60 (0.25 mm) TLC plates were from Merck. Chloroform, ammonium acetate, and sodium acetate were obtained from EM Science. Tryptone and yeast extract were from Difco. E. coli PE and other PE species were purchased from Avanti Polar Lipids. Phospholipase C from Bacillus cereus was purchased from Sigma, and 1-palmitoyl-2-[1-14 C]linoleoylglycero-3-phosphoethanolamine ([16: 0, 14 C-18:2]PE) (55 mCi/mmol) was from Amersham Biosciences. The bicinchoninic acid protein determination kit and Triton X-100 were from Pierce. All other chemicals were reagent grade and were purchased from either Sigma or Mallinckrodt.
Bacterial Strains-The bacterial strains used in this study are described in Table I. Typically, bacteria were grown at 37°C in LB medium, which consists of 10 g of NaCl, 10 g of tryptone, and 5 g of yeast extract per liter (12). When required for selection of plasmids, cells were grown in the presence of 100 g/ml ampicillin, 12 g/ml tetracycline, 25 g/ml chloramphenicol, or 30 g/ml kanamycin.
Molecular Biology Applications-Protocols for handling of DNA samples were those of Sambrook and Russell (13). Transformation-competent cells of E. coli were prepared by the method of Inoue et al. (14). When required, E. coli cells were prepared for electroporation by the method of Sambrook and Russell (13). Plasmids were prepared using the Qiagen spin prep kit. DNA fragments were isolated from agarose gels using the QIAquick gel extraction kit. Genomic DNA was isolated using the protocol for bacterial cultures in the Easy-DNA TM kit (Invitrogen). T4 DNA ligase (Invitrogen), restriction endonucleases (New England Biolabs), and shrimp alkaline phosphatase (U. S. Biochemical Corp.) were used according to the manufacturer's instructions. Doublestranded DNA sequencing was performed with an ABI Prism 377 instrument at the Duke University DNA Analysis Facility. Primers were purchased from MWG Biotec.
Construction of eptB Expression Vectors-The eptB (yhjW) gene of E. coli was cloned into pET28b (Novagen) behind the T7lac promoter. The predicted coding region for eptB was amplified by PCR from E. coli W3110 genomic DNA. The forward primer contained a clamp region, an NcoI site (underlined), and the eptB-coding region with its start codon (boldface). The reverse primer contained a clamp region, a BamHI site (underlined), and the coding region with its stop codon (boldface). Sequences of the forward and reverse primers were 5Ј-GCGCGCCCATG-GTCTTATCACCTGTTTGTCCA-3Ј and 5Ј-GCGCGCGGATCCTTAGT-TAGCCGCTGCCTC-3Ј, respectively. The PCR mixture contained 300 ng of genomic DNA as template, 0.2 g of each primer, 200 M each of dNTP, 100 mM Tris-HCl, pH 8.8, 35 mM MgCl 2 , 250 mM KCl, and 5 units of Pfu DNA polymerase (Stratagene) in a reaction volume of 0.1 ml. The reaction mixture was subjected to a 1-min denaturation at 94°C, followed by 25 cycles of 94°C for 1 min, 55°C for 1 min, 72°C for 2 min, and a final extension for 10 min at 72°C, using the PerkinElmer Life Sciences GeneAmp PCR System 2400. The PCR product and the pET28b vector were both digested with NcoI and BamHI, ligated together, and transformed into XL-1 Blue cells (Stratagene) for propagation of the desired plasmid, designated pEptB1. In some experiments, pEptB1 was used directly for EptB protein expression in strain C41(DE3) ( Table I).
FIG. 1. Proposed reaction catalyzed by EptB. The position of the pEtN substituent on the outer Kdo unit was determined by Brabetz et al. (9) for the pEtN-Kdo 2 -lipid A isolated from WBB06 grown in the presence of Ca 2ϩ . The product generated in vitro is proposed to have the same structure. The dependence of in vitro product formation on PE as the donor substrate is presented. Diacylglycerol is the proposed by-product.
The eptB gene was also moved from pEptB1 into pWSK29 (15), a lac-inducible, low copy expression vector. The XbaI/BamHI-digested fragment, consisting of the eptB gene as well as the pET28b-derived ribosome binding site (RBS), was ligated to the corresponding restriction sites of pWSK29. This plasmid, designated pEptB2, was then transformed into competent E. coli W3110 or electroporated into the heptose-deficient E. coli strain WBB06 (9).
Construction of waaC and waaC/waaF Expression Vectors-The waaC gene of E. coli (16) was cloned into pWSK29 behind the lac promoter to generate plasmid pWaaC. The coding region for waaC was amplified by PCR from E. coli W3110 genomic DNA. The forward primer contained a clamp region, an XbaI site (underlined), and an RBS (italicized) prior to the waaC start codon (boldface). The reverse primer contained a clamp region, a BamHI site (underlined), and the coding region with its stop codon (boldface). Sequences of the forward and reverse primers were 5Ј-GCGCGCTCTAGAAAGGAGATATAAT-GCGGGTTTTGATCGTTAAA-3Ј and 5Ј-GCGCGCGGATCCTTATAAT-GATGATAACTTTTC-3Ј, respectively. The PCR conditions were the same as those used to amplify eptB.
A second plasmid (pWaaCF), in which both waaC (16) and waaF (6,17) are under lac control in pWSK29, was constructed by amplifying the adjacent waaC and waaF genes from W3110 genomic DNA using a forward primer that contained a clamp region, an XbaI site (underlined), and an RBS (italicized) immediately in front of the waaF start codon (boldface). The sequence of the forward primer used is as follows: 5Ј-GCGCGCTCTAGAAAGGAGATATAATGAAAATACTGGT-GATCGGC-3Ј. The reverse primer was the same as that used to amplify waaC. Again, the PCR conditions were the same as those used to amplify eptB.
Preparation of Membranes for Assay of pEtN Transferase Activity-A single colony of WBB06 harboring either the pWSK29 vector or pEptB2 was used to inoculate 5 ml of LB broth (containing 100 g/ml ampicillin), which was then grown overnight at 37°C. Next, the overnight cultures were used to inoculate 200 ml of LB cultures (initial A 600 ϭ 0.02) containing 100 g/ml ampicillin and 1 mM IPTG. When the A 600 reached 1.0, cells were harvested by centrifugation at 4,000 ϫ g for 20 min (0 -4°C). Cells were resuspended in 4 ml of cold 50 mM Hepes, pH 7.5, and broken by a single passage through a French pressure cell at 10,000 pounds/square inch. Cellular debris was removed by centrifugation at 7000 ϫ g for 20 min. Membranes were prepared by ultracentrifugation at 100,000 ϫ g for 60 min. The membranes were resuspended in 8 ml of Hepes, pH 7.5, and the ultracentrifugation was repeated a second time at 100,000 ϫ g to remove contaminating cytosol. The final membrane pellet was stored in 750 l of 50 mM Hepes, pH 7.5, at Ϫ80°C. Protein concentrations were determined by the bicinchoninic acid method. In some experiments, membranes were prepared in the same manner from cultures of WBB06 grown either without CaCl 2 or in the presence of 5 or 50 mM CaCl 2 .
Solubilization of EptB-expressing Membranes-Membranes were prepared from IPTG-induced C41(DE3) cells, harboring either pET28b or pEptB1, as described above. Membranes at 5 mg/ml in 50 mM Hepes, pH 7.5, were solubilized with 2% Triton X-100 in a total volume of 750 l and rotated gently at 4°C for 2 h. The samples were subjected to ultracentrifugation at 55,000 rpm in a Beckman TLA 110 rotor for 1 h at 4°C in an Optima TX ultracentrifuge. The supernatants were collected, and protein concentrations were determined.
While the synthesis of the 4Ј-[ 32 P]lipid IV A was in progress, the components for the second step of the reaction were assembled in a separate 1.5-ml microcentrifuge tube. The reagents for Kdo addition to the 4Ј-[ 32 P]lipid IV A acceptor consisted of 2 M carrier lipid IV A , 0.1% Triton X-100, 10 mM CTP, 4 mM Kdo, 12.5 mM MgCl 2 , 0.03 units of purified E. coli CMP-Kdo synthase (24,25), and 0.65 g/ml purified E. coli Kdo transferase (26) in total volume of 50 l. Upon completion of the 4Ј-[ 32 P]lipid IV A step, the reaction components needed for the addition of the Kdo residues were added to the 4Ј-[ 32 P]lipid IV A -containing tube, and the mixture was incubated at room temperature for 30 min.
The total reaction mixture was then spotted onto a 10 ϫ 20-cm TLC plate and dried under a cold air stream. The plate was developed in the solvent system chloroform, pyridine, 88% formic acid, water (30:70:16: 10, v/v). Following chromatography, the plate was dried under a cold air stream and exposed to x-ray film for 20 s to locate the Kdo 2 -4Ј-[ 32 P]lipid A. The region of the silica plate containing the product was removed by scraping, transferred to a thick-walled glass tube, and resuspended in 4 ml of an acidic single-phase Bligh/Dyer mixture (29,30) consisting of chloroform, methanol, 0.1 M HCl (1:2:0.8, v/v). The suspension was vigorously mixed with the aid of a vortex and subjected to sonic irradiation in a bath for ϳ30 s. The silica particles were removed with a clinical centrifuge set at top speed for 10 min. The supernatant, con- taining the 32 P-labeled lipid, was removed, and the extraction of the silica was repeated, first with 4 ml and then with 8 ml of the acidic single-phase Bligh/Dyer mixture. The extracts were pooled, passed through a 4-ml glass-wool column to remove remaining silica particles, and placed into glass tubes equipped with Teflon-lined caps. The solution was converted to a two-phase Bligh/Dyer mixture, consisting of chloroform, methanol, 0.1 M HCl (2:2:1.8, v/v). The phases were separated in a clinical centrifuge. The lower phases were transferred to new glass tubes. The upper phases was extracted a second time by the addition of fresh, pre-equilibrated acidic lower phases. The lower phases were pooled, neutralized by addition of pyridine (1 drop per 2-ml lower phase), and dried under a stream of N 2 . The Kdo 2 -4Ј-[ 32 P]lipid A was resuspended in 25 mM Tris-HCl, pH 7.8, containing 1 mM EDTA, 1 mM EGTA, and 0.1% Triton X-100 and then stored at Ϫ20°C. The amount of Kdo 2 -4Ј-[ 32 P]lipid A recovered was typically greater than 50 Ci. Nonradioactive carrier Kdo 2 -lipid A was prepared from WBB06 in milligram quantities, as described previously (31). Assay Conditions for Detecting pEtN Transferase Activity-The EptB transferase was assayed under optimized conditions in a 15-l reaction mixture containing 50 mM Hepes, pH 7.5, 0.1% Triton X-100, 1 mM calcium chloride (CaCl 2 ), 1.25 mM dithiothreitol (DTT), ϳ0.6 mM E. coli PE, and 10 M Kdo 2 -4Ј-[ 32 P]lipid A (50,000 cpm/nmol). Washed membranes were employed as the enzyme. Assay mixtures were incubated at 30°C for the indicated times, and 4-l portions were spotted onto Silica Gel 60 TLC plates to stop the reactions. Substrate and product(s) were separated using the solvent chloroform/methanol/water/acetic acid (25:15:4:4, v/v). Following chromatography, the plates were dried and analyzed using a Amersham Biosciences PhosphorImager (STORM 840), equipped with ImageQuant software.
Assay for Kdo 2 -Lipid A-dependent Diacylglycerol Formation-EptBcatalyzed diacylglycerol formation was assayed in a 15-l reaction mixture containing 50 mM Hepes, pH 7.5, 0.1% Triton X-100, 1 mM Separation of Inner and Outer Membranes-Membranes of W3110 harboring pEptB2 were separated by isopycnic sucrose gradient centrifugation. First, 120-ml LB cultures of W3110 harboring pEptB2 were grown with 1 mM IPTG until A 600 ϭ 1.0. The cells were harvested as above, and membranes were prepared by the method of Osborn and Munson (34). The membranes were applied to a seven-step gradient (35) and subjected to ultracentrifugation at 35,000 rpm in a Beckman SW40.1 rotor for 18 h at 4°C. Each fraction (0.5 ml) was assayed for NADH oxidase (inner membrane marker), phospholipase A (outer membrane marker) (36), and protein (37). Each fraction was also assayed for pEtN transferase activity.
In-frame Replacement of eptB with a Kanamycin Resistance Cassette in E. coli WBB06 -To create an in-frame replacement of eptB with the kanamycin resistance cassette (kan), WBB06 was transformed by electroporation with pKD46, an arabinose-inducible, -RED helper plasmid (38). PCR was utilized to construct a linear piece of DNA containing kan, flanked on the 5Ј end by an RBS plus 42 bp of chromosomal DNA upstream of eptB, and flanked 3Ј by 42 bp of chromosomal DNA located downstream of eptB. The forward and re-verse primers are as follows: 5Ј-CCGCACTTTTTCCCTGCCGGGCCT-GAAAAGCCACTAAGCAGG -AAGGAGATATAATGAGCCATATTCAA-CGGGAA-3Ј and 5Ј-TAGCAAAATGCCTTTTGATCGGCGAGAAAGTC-AGCAGGCCGCTTAGAAAAACTCATCGAGCAT-3Ј. The portion of the kanamycin cassette is underlined and the engineered RBS is in italics. The kanamycin resistance gene (Tn903) from plasmid pWKS130 (15) served as the template. PCR conditions used were the same as described above for the amplification of the eptB gene from genomic DNA. The PCR product was resolved on a 1.0% agarose gel and purified with the QIAquick gel kit. The product was electroporated into WBB06/pKD46, which had been grown at 30°C with 1 mM L-(ϩ)arabinose. After growth for 2 h at 37°C, the cells were plated on LB agar containing kanamycin (20 g/ml) and incubated overnight at 37°C. Kanamycin-resistant colonies were re-purified on LB kanamycin plates, which were incubated overnight at 37°C to cure the strain of pKD46. The eptB::kan replacement on the chromosome of WBB06 was verified by PCR using external primers located 95 bp upstream (5Ј-G-CACACTCTTTCCCCACACTTTTTCC-3Ј) and 118 bp downstream (5Ј-CCTCCGACCCCTTCGTCCCGAACGAAG-3Ј) of eptB. A single PCR product was resolved on a 1% agarose gel, purified with the QIAquick gel extraction kit, and sequenced using the same primers to confirm the replacement. The resulting strain, designated WBB06eptB::kan, was made electrocompetent by the method of Sambrook and Russell (13) and then transformed with the following pWSK29-derived plasmids: pWSK29 (vector control), pEptB2, pWaaC, or pWaaCF.
Growth of WBB060-derived Strains in the Presence of Divalent Cations-Strains were grown overnight on LB broth at 37°C and used to inoculate 50 ml of pre-warmed LB broth to an A 600 ϭ 0.02. Growth was allowed to continue with shaking at 250 rpm (37°C). When A 600 reached ϳ0.2, various divalent cations were added, as indicated below (time ϭ 0). Growth was allowed to continue at 37°C. Whenever the A 600 reached 0.3-0.4, the cultures were diluted 10-fold into 50 ml of prewarmed LB, containing the relevant divalent cations, in order to keep the cells in log phase.
Extraction and Purification of LPS from E. coli WBB06 -To determine whether the LPS molecules in WBB06 are covalently modified with pEtN when eptB is expressed under lac control from pWSK29 in the absence of added Ca 2ϩ , 1-liter cultures of WBB06 harboring either pWSK29 or pEptB2 were grown in the presence of 100 g/ml ampicillin and 1 mM IPTG until A 600 ϭ 1.0. The LPS was extracted and purified, as described previously (10).
In-frame Replacement of eptB with a Kanamycin Resistance Cassette in E. coli W3110 -The eptB gene was also replaced with kan on the chromosome of E. coli W3110. The purified PCR product used to create the replacement of eptB on the chromosome of WBB06 (see above) was utilized to construct W3110eptb::kan. First, the PCR product was electroporated into the -RED strain, DY330 (39). After growth for 2 h at 37°C, the cells were plated on LB agar containing kanamycin (20 g/ml) and incubated overnight at 37°C. Kanamycin-resistant colonies were re-purified on LB kanamycin plates. One of the kanamycin-resistant colonies was designated DY330eptB::kan. A P1 vir bacteriophage lysate of donor strain DY330eptB::kan was made as described by Miller (12) and used to tranduce stationary cells of W3110. The transduction mixture was plated onto LB agar, containing 20 g/ml kanamycin and 5 mM sodium citrate, and then incubated at 37°C overnight. Kanamycin-resistant colonies were re-purified with selection, and the eptB::kan replacement on the chromosome of W3110 was verified by PCR using the same primers used to confirm the eptB::kan replacement on the chromosome of WBB06 (see above). The PCR product was resolved on a 1% agarose gel, purified with the QIAquick gel extraction kit, and sequenced to confirm the replacement.
Mass Spectrometry of Kdo 2 -Lipid A Samples-Spectra were acquired in the negative-ion linear mode using a Kratos (Manchester, UK) analytical matrix-assisted laser desorption ionization/time of flight (MALDI/TOF) mass spectrometer with a 337-nm nitrogen laser, a 20-kV extraction voltage, and time-delayed extraction. Each spectrum was the average of 50 shots. The lipid A samples were prepared for MALDI/TOF analysis by depositing 0.3 l of the lipid sample, dissolved in chloroform/methanol (4:1, v/v), followed by 0.3 l of a saturated solution of 6-aza-2-thiothymine in 50% acetonitrile and 10% tribasic ammonium citrate (9:1, v/v) as the matrix. The samples were dried at room temperature before the spectra were acquired.

RESULTS
pEtN Transferase Activity in Membranes of Ca 2ϩ -treated WBB06 Versus EptB Overexpressing WBB06 -Membranes of the heptose-deficient mutant WBB06, grown in LB broth con-taining 5-50 mM CaCl 2 , exhibit pEtN transferase activity in vitro with the tetra-acylated LPS precursor Kdo 2 -4Ј-[ 32 P]lipid IV A as the acceptor substrate (10). The pEtN transferase activity was also observed when hexa-acylated Kdo 2 -4Ј-[ 32 P]lipid A was employed as the lipid acceptor (Fig. 2, lanes 3 and 4). To determine whether the pEtN transferase activity seen in Ca 2ϩtreated WBB06 is because of EptB, we examined membranes of EptB-overexpressing cells. As shown in Fig. 2, lane 6, the R f of the product generated by membranes of WBB06/pEptB2 was the same as that produced by membranes of Ca 2ϩ -treated WBB06 (Fig. 2, lanes 3 and 4). However, much more product was formed with WBB06/pEptB2 membranes. No reaction was seen with the vector control WBB06/pWSK29 (Fig. 2, lane 5). The band labeled x in Fig. 2 reflects the formation of the lipid A 1-diphosphate variant (40,41), which is generated from an as yet unidentified membrane-bound phosphate donor. Lipid A 1-diphosphate formation is partially suppressed by the presence of 1 mM Ca 2ϩ in the assay system, and it is not observed when EptB-overexpressing membranes are diluted (see below).
PE and Ca 2ϩ Are Required for EptB Activity in Vitro-As shown in Fig. 3A, the production of pEtN-Kdo 2 -4Ј-[ 32 P]lipid A is dependent upon the addition of E. coli PE to the assay mixture. In order to demonstrate unequivocally the dependence of in vitro product formation on exogenous PE, the concentration of membrane protein utilized in the assay system had to be 0.025 mg/ml (or lower) to dilute out endogenous PE present in the EptB-overexpressing membranes.
The addition of pEtN to the outer Kdo residue of LPS in living cells of WBB06 was shown previously to be dependent upon the inclusion of 5-50 mM CaCl 2 in the growth medium (10). However, it was not entirely clear if Ca 2ϩ was also a necessary component of the in vitro assay system. Addition of EDTA to the assay inhibits pEtN transferase activity (10). To determine whether or not Ca 2ϩ is required for pEtN transfer to Kdo 2 -4Ј-[ 32 P]lipid A, diluted membranes of WBB06 overexpressing EptB were assayed in the presence or absence of 1 mM CaCl 2 . Ca 2ϩ greatly stimulated pEtN transferase activity in vitro when 0.025 mg/ml of membranes from EptB-overexpressing cells were employed as the enzyme source (Fig. 3B). The optimal Ca 2ϩ concentration in the in vitro system was 1 mM (data not shown); higher concentrations of Ca 2ϩ were inhibitory. As noted above, the formation of the lipid A 1-diphosphate (x in Fig. 3) was suppressed by Ca 2ϩ . All subsequent assays of EptB therefore included both 0.6 mM PE and 1 mM Ca 2ϩ , unless otherwise indicated. Production of pEtN-Kdo 2 -lipid A by overexpressed EptB was linear for almost 60 min under the optimized conditions (Fig. 4, A and B), and it was linear with membrane protein up to 0.2 mg/ml at 30 min (data not shown).
EptB Selectivity for Ca 2ϩ Ions-The induction of pEtN transferase in membranes of WBB06 was strictly dependent upon the addition of Ն5 mM Ca 2ϩ to the growth medium; other common divalent cations were not effective (10). To determine whether other divalent cations could replace Ca 2ϩ in vitro, we tested several other ions at 1 mM. Ca 2ϩ stimulated the greatest conversion of Kdo 2 -4Ј-[ 32 P]lipid A to pEtN-Kdo 2 -4Ј-[ 32 P]lipid A. Sr 2ϩ also caused a slight stimulation of pEtN transferase activity, but Mg 2ϩ and Ba 2ϩ were inactive (data not shown).  (Fig. 6B, lane 13) (43). By comparing the radioactive products generated by hydrolysis of pEtN-Kdo 2 -4Ј-32 P-lipid A to those of Kdo 2 -4Ј-[ 32 P]lipid A (Fig. 7), we showed that EptB adds pEtN mainly to the outer Kdo residue. Hydrolysis of pEtN-Kdo 2 -4Ј-[ 32 P]lipid A (Fig. 7B) Fig. 1 (9), remains to be established.

Efficacy of PE Molecular Species as Substrates for EptB-
Inner Membrane Localization of EptB-Membranes of wildtype E. coli W3110 overexpressing EptB were subjected to isopycnic sucrose gradient centrifugation to separate inner and outer membranes. The EptB activity was found almost entirely within the inner membrane (Fig. 8). As shown in Fig. 8B, the pEtN transferase closely followed the activity of the inner membrane marker NADH oxidase (peak activity at fraction 14). As shown in Fig. 8A, we also determined the membrane protein concentration of each fraction and assayed the outer membrane marker phospholipase A (peak activity in fraction 3).
MALDI/TOF Mass Spectrometry of LPS Isolated from WBB06/pEptB2-LPS of WBB06 was shown previously to contain a pEtN substituent attached to its outer Kdo residue when the cells were grown in the presence of 5-50 mM CaCl 2 (10). To confirm that EptB expressed under lac control catalyzes the addition of pEtN to Kdo 2 -lipid A in living cells, we isolated LPS from WBB06 harboring either pEptB2 or the vector control pWSK29, grown with IPTG, but in the absence of added CaCl 2 . The LPS was analyzed by MALDI/TOF mass spectrometry in both the negative and positive ion modes. As shown in Fig. 9A, negative ion MALDI/TOF mass spectrometry revealed that DEAE-cellulose purified LPS from WBB06/pWSK29 consisted mostly Kdo 2 -lipid A ([M Ϫ H] Ϫ at m/z ϭ 2236.4 atomic mass units). The spectrum also shows the presence of some free lipid A (m/z ϭ 1796.6 atomic mass units), which arises by loss of the Kdo residues during MALDI/TOF analysis. The negative-mode MALDI/TOF spectrum of LPS purified from WBB06/pEptB2 showed the same ions as described above, but in addition also contained a third species at m/z ϭ 2360.0 atomic mass units, interpreted as [M Ϫ H] Ϫ of pEtN-Kdo 2 -lipid A (Fig. 9B). Based upon the mass spectrometry and TLC analysis of LPS isolated from WBB06 harboring pEptB2 (data not shown), we estimate that about one-third of the LPS molecules are modified with the pEtN unit when EptB is expressed from the lac promoter.
Positive (interpreted as the B 2 ϩ ion of pEtN-Kdo 2 -lipid A) (44). These ions were absent in the positive-mode spectrum of the LPS purified from WBB06/pWSK29. Ca 2ϩ Sensitivity of WBB06eptB::kan and Its Reversal by Excess Mg 2ϩ -To study the function of EptB, we constructed an in-frame replacement of the chromosomal copy of eptB in WBB06 with a kanamycin resistance cassette. The parental strain WBB06 grew normally on LB medium in the presence or absence of added Ca 2ϩ (Fig. 10A). The growth rate of WBB06eptB::kan was almost the same as that of WBB06 in the absence of Ca 2ϩ (Fig. 10A), but addition of 5 mM (or higher) Ca 2ϩ during mid-log phase (Fig. 10A, arrow) rapidly inhibited its subsequent growth. Neither Mg 2ϩ (Fig. 10B) nor Sr 2ϩ (not shown) had any effect on the growth of WBB06 (not shown) or WBB06eptB::kan. However, inclusion of 50 mM MgCl 2 in the medium prevented the Ca 2ϩ -induced growth arrest of WBB06eptB::kan (Fig. 10B).
Suppression of Ca 2ϩ Sensitivity by Restoration of Either EptB or the Full Core Domain to WBB06 -To exclude the possibility that the Ca 2ϩ hypersensitivity was because of a polar effect arising from the replacement of the eptB gene with kan, WBB06eptB::kan was transformed with pEptB2. This construct grew normally in Ca 2ϩ -containing medium (Fig. 10C), whereas the vector control was Ca 2ϩ -sensitive.
Cultures of WBB06eptB::kan harboring either pWaaC or pWaaCF were challenged with 5 mM CaCl 2 to determine whether restoration of a part (or all) of the LPS core could eliminate the Ca 2ϩ sensitivity. As shown in Fig. 10C, supplying waaC in trans, which results in the incorporation of a single heptose residue (6,8), did not alleviate the Ca 2ϩ hypersensitivity. However, supplying both heptosyl transferase genes (waaC and waaF) in trans, which fully restores the LPS core (6,8), allowed the eptB knock-out strain to grow in the presence of 5 mM Ca 2ϩ . These findings are consistent with the additional observation (data not shown) that deletion of eptB in wild-type E. coli W3110 does not result in Ca 2ϩ sensitivity. DISCUSSION Phosphoethanolamine residues are commonly found as substituents of LPS (6 -8) and other surface glycoconjugates in  2 ). B, cultures of WBB06eptB::kan were grown to A 600 ϭ 0.2 at 37°C, and then 5 mM CaCl 2 or 50 mM MgCl 2 was added singly or in combination (time ϭ 0), as indicated. Cells were diluted 10-fold whenever A 600 reached 0.3-0.4. The plotted A 600 represents the cumulative growth yield corrected for dilution. Open squares, WBB06eptB::kan (no added CaCl 2 ); open circles, WBB06eptB::kan (ϩ50 mM MgCl 2 ); closed squares, WBB06eptB::kan (ϩ5 mM CaCl 2 ); closed circles, WBB06eptB::kan (ϩ5 mM CaCl 2 and 50 mM MgCl 2 ). C, WBB06eptB::kan was transformed with the following plasmids: pWSK29 (vector), pEptB2, pWaaC, or pWaaCF. Cells were grown to A 600 ϭ 0.2 at 37°C, and then 5 mM CaCl 2 was added (time ϭ 0), as indicated. Cells were diluted 10-fold whenever A 600 reached 0.3-0.4. The plotted A 600 represents the cumulative growth yield corrected for dilution. Open squares, WBB06eptB::kan/pWSK29; closed squares, WBB06eptB::kan/pEptB2; open circles, WBB06eptB::kan/pWaaC; closed circles, WBB06eptB::kan/pWaaCF. Gram-negative bacteria (45). Very little is known about the enzymes that generate the pEtN-modified molecules present in the cell envelope or about the biological function of pEtN modifications. In order to explore these issues, it is necessary to identify the genes encoding the relevant pEtN transferases and to construct mutants lacking them. In the present work we have found that the eptB gene, formerly yhjW, encodes the Ca 2ϩ -induced pEtN transferase known to modify the outer Kdo residue of E. coli LPS (9, 10). We have constructed a mutant in which eptB is deleted, and we found that it is extremely sensitive to added Ca 2ϩ ions, provided that it is also lacking the heptose residues of the LPS core domain.
The following strategy was used to find eptB. Mackinnon et al. (11) first reported the lpt-3 gene of N. meningitidis, which is required for the modification of the lipo-oligosaccharide core of that organism with a specific pEtN residue. They found the lpt-3 gene by screening random mutants with an antibody directed against pEtN-modified lipo-oligosaccharide. However, an in vitro enzymatic assay was not developed to prove that the protein encoded by lpt-3, NMB2010, is an enzyme (11). The N. meningitidis protein, NMB2010, has six significant orthologs in E. coli, five of which are predicted to be inner membrane proteins, containing four or five putative trans-membrane helices near their N termini. We therefore cloned all six of these E. coli NMB2010 orthologs. We found that only yhjW expressed behind the lac promoter on pWSK29-directed massive in vitro overexpression of the Kdo-selective pEtN transferase (10) with Kdo 2 -lipid A as the acceptor substrate (Fig. 2). In this setting, enzyme induction was no longer Ca 2ϩ -dependent. The subcellular fractionation study shown in Fig. 8 confirmed the localization of EptB to the inner E. coli membrane.
Overexpression of the other five E. coli orthologs of NMB2010, using the lac promoter of pWSK29, led to the identification of EptA/PmrC (46), the enzyme that modifies the phosphate groups of lipid A with pEtN residues in polymyxinresistant mutants (47,48). The active site of PmrC/EptA has been shown by gene fusion experiments to face the periplasm (48). However, PmrC/EptA has not been purified or characterized enzymatically (48), and our EptA in vitro assay (46) was reported only in abstract form. The full characterization of EptA as an enzyme will be described elsewhere.
In addition to being induced selectively by Ca 2ϩ , EptB transferase activity is dependent upon the presence of 1 mM Ca 2ϩ in vitro (Fig. 3). EptB joins a growing list of inner (46, 49 -50) and outer (20, 51) membrane-bound enzymes that modify LPS; however, EptB is the first of these to exhibit a requirement for Ca 2ϩ . The exact role that Ca 2ϩ plays in the transfer of pEtN from PE to the outer Kdo moiety of LPS remains to be determined. The well characterized outer membrane phospholipase A of E. coli exhibits a similar Ca 2ϩ requirement, which in that case may play a role in dimer formation (52). We found that Sr 2ϩ could partially stimulate the pEtN transferase activity of EptB in our in vitro system. Studies with the phospholipase A likewise revealed that Sr 2ϩ could substitute for Ca 2ϩ to some degree.
Our studies show that EptB utilizes several different PE molecular species (Fig. 7) provided they contain at least one double bond. EptB may have evolved to utilize the most abundant PE species normally present in the E. coli envelope (2). We have also confirmed that recombinant EptB modifies the outer Kdo residue selectively (Fig. 7) but cannot transfer pEtN to the inner Kdo unit or to the lipid A moiety at an appreciable rate. The results of the pH 4.5 hydrolysis of pEtN-Kdo 2 -4Ј-[ 32 P]lipid A are definitive with regard to the outer Kdo (Fig. 7), but they do not establish the proposed localization of the pEtN unit at position 7 (Fig. 1).
The four additional EptB orthologs (besides EptA) encoded within the E. coli genome are likely to include the enzyme responsible for the addition of the pEtN moiety to the outer heptose residue of LPS (6). To date, no in vitro assays with defined acceptor substrates have been described for this heptose modification. Another one of the five EptB orthologs might be required for the transfer of pEtN from PE to a sub-set of the periplasmic membrane-derived oligosaccharides, the biosynthesis of which is induced at low osmolarity (45,53).
EptB generates stoichiometric diacylglycerol as a by-product during the transfer of the pEtN unit from PE to Kdo 2 -lipid A (Fig. 6), consistent with the scheme proposed in Fig. 1. This novel source of diacylglycerol, which is likely also generated by the other members of the EptB family, explains why the diacylglycerol that accumulates in E. coli dgk mutants is not solely due to membrane-derived oligosaccharide biosynthesis (32). In principle, the EptB reaction should be reversible and share some common mechanistic features with the better characterized eucaryotic phosphoethanolamine transferases that generate PE from CDP-ethanolamine and diacylglycerol (54). However, the EptB family of proteins displays no obvious sequence similarity to the eucaryotic phosphoethanolamine transferases, which generally contain seven predicted trans-membrane segments. Iterative analysis of certain EptB orthologs with the Psi-Blast algorithm does suggest a very distant relationship to the phosphoethanolamine transferases that participate in the assembly of the phosphatidylinositol-linked glycans in eucaryotic cells (55).
The Ca 2ϩ -sensitive phenotype of the heptose-deficient mutant, lacking EptB, is very intriguing, as it suggests a possible function for the pEtN modification of the outer Kdo residue. Based on the observed Ca 2ϩ sensitivity of WBB06eptB::kan (Fig. 10), we propose that pEtN modification of the outer Kdo moiety of LPS is critical for tolerance of elevated levels of Ca 2ϩ in heptose-deficient E. coli. It may be that modification of the outer Kdo residue with pEtN renders the outer membrane less permeable to Ca 2ϩ . A reduction in Ca 2ϩ permeability may help maintain the very low level of intracellular Ca 2ϩ (ϳ0.1 M or less) normally present in E. coli (56). A dramatic increase in the intracellular Ca 2ϩ concentration might be toxic. The fact that excess Mg 2ϩ can block the growth inhibition produced by Ca 2ϩ (Fig. 10B) suggests that excess Mg 2ϩ may prevent Ca 2ϩ from binding to a critical site that destabilizes the cell envelope, preventing the influx of Ca 2ϩ . The observation (Fig. 10C) that eptB mutants are not Ca 2ϩ -sensitive in strains with a complete core suggests that the pEtN units that are normally present on the outer heptose residue of the core may substitute for the Ca 2ϩ -inducible pEtN unit attached to the Kdo region. Further genetic and biochemical characterization of the Ca 2ϩ -sensitive phenotype of WBB06eptB::kan should provide insights into the function of this modification.