Identification of the Gene Encoding the Escherichia coli Lipid A 4′-Kinase

The genes for seven of nine enzymes needed for the biosynthesis of Kdo2-lipid A (Re endotoxin) inEscherichia coli have been reported. We have now identified a novel gene encoding the lipid A 4′-kinase (the sixth step of the pathway). The 4′-kinase transfers the γ-phosphate of ATP to the 4′-position of a tetraacyldisaccharide 1-phosphate intermediate (termed DS-1-P) to form tetraacyldisaccharide 1,4′-bis-phosphate (lipid IVA). The 4′-phosphate is required for the action of distal enzymes, such as Kdo transferase and also renders lipid A substructures active as endotoxin antagonists or mimetics. Lysates ofE. coli generated using individual λ clones from the ordered Kohara library were assayed for overproduction of 4′-kinase. Only one clone, [218]E1D1, which directed 2–2.5-fold overproduction, was identified. This construct contains 20 kilobase pairs of E. coli DNA from the vicinity of minute 21. Two genes related to the lipid A system map in this region: msbA, encoding a putative translocator, and kdsB, the structural gene for CMP-Kdo synthase. msbA forms an operon with a downstream, essential open reading frame of unknown function, designated orfE. orfE was cloned into a T7 expression system. Washed membranes from cells overexpressingorfE display ∼2000-fold higher specific activity of 4′-kinase than membranes from cells with vector alone. Membranes containing recombinant, overexpressed 4′-kinase (but not membranes with wild-type kinase levels) efficiently phosphorylate three DS-1-P analogs: 3-aza-DS-1-P, base-treated DS-1-P, and base-treated 3-aza-DS-1-P. A synthetic hexaacylated DS-1-P analog, compound 505, can also be phosphorylated by membranes from the overproducer, yielding [4′-32P] lipid A (endotoxin). The overexpressed lipid A 4′-kinase is very useful for making new 4′-phosphorylated lipid A analogs with potential utility as endotoxin mimetics or antagonists. We suggest that orfE is the structural gene for the 4′-kinase and that it be redesignated lpxK.

Identification of the 4Ј-kinase gene has been hampered because mutants lacking the 4Ј-kinase have not been identified (3). Presumably the 4Ј-kinase gene, like most other genes encoding enzymes of lipid A biosynthesis, is required for growth (3). Genetic screens for conditionally lethal mutants, such as that used for the identification of the Kdo transferase gene, kdtA (15), have not been developed. Attempts to purify the kinase to homogeneity have been thwarted by the protein's association with membranes and its instability in the presence of detergents (12,16).
The lipid A 4Ј-kinase can be used to make 4Ј-32 P-labeled lipid A precursors, such as [4Ј-32 P]lipid IV A and Kdo 2 -[4Ј-32 P]lipid IV A , for biochemical analyses of late pathway reactions (12, 16 -18). The 4Ј-kinase activity found in wild type E. coli membranes, however, is relatively inefficient and unstable, especially in the presence of low chemical concentrations of ATP (12,16). Only 0.1-0.5% incorporation of 32 P into [4Ј-32 P]lipid IV A is obtained when trace quantities of [␥-32 P]ATP are used together with the physiological substrate, DS-1-P (12, 16 -18). This low level of 32 P transfer makes it virtually impossible to use the 4Ј-kinase for phosphorylating DS-1-P analogs that are utilized less efficiently. Identification and overexpression of the 4Ј-kinase gene would allow investigation of DS-1-P analogs as substrates for the 4Ј-kinase, possibly facilitating the synthesis of interesting lipid IV A analogs with potential activity as endotoxin antagonists or mimetics (2,3). Studies of endotoxin uptake and metabolism (19 -21) would be simplified if [4Ј-32 P]lipid A or various analogs could be made with the 4Ј-kinase.
To identify the gene encoding the lipid A 4Ј-kinase, we employed the approach used by Clementz et al. (22,23) to find the gene for the Kdo-dependent lauroyltransferase of the lipid A pathway. Individual lysates generated from the ordered Kohara library (22)(23)(24) were assayed for overproduction of 4Јkinase activity. Using this procedure, a clone was identified that spans 20 kilobase pairs near minute 21 of the E. coli chromosome and is capable of directing slight (2-2.5-fold) overproduction of 4Ј-kinase activity. Subcloning of a candidate open reading frame present on this hybrid bacteriophage revealed that an essential gene of unknown function, designated orfE (25), resulted in ϳ2000-fold overproduction of 4Ј-kinase when expressed behind a T7 promoter. orfE is the downstream gene in an operon that also includes msbA, a possible translocator for nascent LPS in the inner membrane (25,26). We believe that orfE is the structural gene for the 4Ј-kinase and should be redesignated lpxK. orfE does not display significant sequence homology to other known kinases. The overexpressed 4Ј-kinase is very useful for the efficient synthesis of labeled precursors and novel lipid A analogs.

EXPERIMENTAL PROCEDURES
Materials-32 P i and [␥-32 P]ATP were obtained from NEN Life Science Products, 0.25-mm glass-backed Silica Gel 60 thin layer chromatography plates from Merck, yeast extract and tryptone from Difco, restriction enzymes from New England Biolabs, T4 DNA Ligase from Life Technologies, Inc., and Pfu DNA polymerase from Stratagene. Solvents for thin layer chromatography were reagent grade from Malinckrodt, and solvents for lipid preparations were high pressure liquid chromatography grade (Aldrich).
Bacterial Strains and Growth Conditions- Table I lists the strains used in this study. Cells were cultured at 37°C in Luria Broth (LB) consisting of 10 g of NaCl, 5 g of yeast extract, and 10 g of tryptone per liter (27). Antibiotics were added, when required, at 50 g/ml for ampicillin, 12 g/ml for tetracycline, and 30 g/ml for chloramphenicol.
DNA Techniques-E. coli W3110 chromosomal DNA was isolated as described by Ausubel et al. (28). Minipreparations of plasmid DNA were made using the Promega Wizard minipurification system. Large scale preparations of plasmid DNA were made using the 5Ј-3Ј Bigger Prep kit. Polymerase chain reactions were optimized using the Stratagene Optiprime Kit. DNA fragments were isolated from agarose gels using the Qiagen Qiaex II gel extraction kit. Restriction enzymes and T4 ligase were used according to the manufacturer's directions. Transformation of E. coli with plasmid DNAs was done using salt-competent cells (29).
Kohara Library Preparation and Screen for 4Ј-Kinase Activity-Fresh lysates of the Kohara library were made following the method of Clementz et al. with slight modifications (23). The host E. coli strain, W3110, was grown overnight at 37°C in LB medium, supplemented with 0.2% maltose and 10 mM MgSO 4 . The culture was diluted 1:1 with 10 mM CaCl 2 , 10 mM MgCl 2 . The lysates used by Clementz et al. (23) were diluted 1:100 and 1:1000 in SM buffer (5.8 g of NaCl, 2 g of MgSO 4 , 50 ml of 1 M Tris, pH 7.5, per liter). Using 96-well microtiter plates, 5 l of the individual diluted lysates and 10 l of the diluted host cell suspension were mixed and incubated at 37°C for 15 min. LB medium supplemented with 10 mM MgSO 4 (150 l) was added to each well, and incubation continued at 37°C. After 4 h, the A 600 of each well was measured using a Molecular Dynamics Spectramax 250 microplate reader. When the cell suspension had cleared to an A 600 less than 0.1, it was considered lysed and was transferred to a fresh microtiter plate at 4°C. Lysis was evaluated every hour until 8 h after infection. The lysates originating from the 1:1000 dilution of the originals were chosen for assay. Any hybrid bacteriophages that did not yield fresh suitable lysates with the 1:1000 dilutions of the original stock were generally obtained from the 1:100 dilutions of the original stock. The final lysates were stored at Ϫ80°C overnight. The 4Ј-kinase activity of each lysate was assayed in a 10-l reaction mixture containing 5 l of lysate, 100 M DS-1-32 P (1000 cpm/nmol), 1 mg/ml cardiolipin, 50 mM Tris, pH 8.5, 5 mM ATP, 1% Nonidet P-40, and 5 mM MgCl 2 . After incubation at 30°C for 60 min, the reaction was stopped by spotting 5 l onto a Silica Gel 60 TLC plate. The plates were developed in chloforom/methanol/water/ acetic acid (25:15:4:2, v/v/v/v), dried, and exposed to a Molecular Dynamics PhosphorImager screen. Conversion of DS-1-32 P to [1-32 P]lipid IV A was quantified using ImageQuant software (Molecular Dynamics).
Construction of pJK2 Bearing orfE under the Control of a T7 Promoter-The gene encoded by the open reading frame orfE was cloned into pET3a cloning vector (Novagen). orfE was amplified by polymerase chain reaction of E. coli W3110 genomic DNA using Pfu DNA polymerase (according to the manufacturer's specifications) and the following primers: 5Ј-GTTTGGCATATGATCGAAAAAATCTGG-3Ј and 5Ј-AT-TCATGGATCCATCAATCGAACGCTG-3Ј. The first primer introduces a NdeI site at the start codon of orfE, and the second primer introduces a BamHI site downstream of the stop codon. The polymerase chain reaction product was digested with NdeI and BamHI and ligated into a similarly cut pET3a vector. A portion of the ligation reaction was transformed into E. coli SURE cells (Stratagene, La Jolla, CA), and colonies resistant to ampicillin were selected. Plasmid DNA was isolated from ampicillin-resistant clones and was digested with BamHI and NdeI to identify those constructs that contained the desired 1-kilobase pair insert. One correct plasmid was designated pJK2.
Expression of the orfE Gene Product-pJK2 was transformed into BLR(DE3)pLysS cells and grown at 37°C in 2 liters of LB. When the cultures reached an A 600 of 0.6, isopropyl-1-thio-␤-D-galactopyranoside was added (final concentration of 1 mM) to induce expression of the orfE gene product. After 3 h of induction, the cells were collected by centrifugation at 10,000 ϫ g for 15 min at 4°C, washed with 1 liter of 50 mM HEPES, pH 7.5, and resuspended in 30 ml of the wash buffer. Cells were broken in a cold French pressure cell at 20,000 p.s.i., and unbroken cells were removed by centrifugation at 3,500 ϫ g to form the cell-free extract. The membrane and soluble fractions were isolated by centrifugation of the entire cell-free extract at 150,000 ϫ g for 60 min. After centrifugation, the soluble fraction was removed to a fresh tube, and the membrane pellet was resuspended in 50 ml of 50 mM HEPES, pH 7.5. The soluble fraction and the resuspended membranes were both centrifuged a second time. The final membrane pellet was resuspended by homogenization in 10 ml of the HEPES buffer and stored frozen in aliquots at Ϫ80°C. The membrane-free cytosol was also stored in aliquots at Ϫ80°C. The protein concentration was determined using the Bio-Rad protein assay kit with bovine serum albumin as a standard.
Assays for 4Ј-Kinase Activity-Two methods for analyzing 4Ј-kinase activity of various protein fractions were employed. The first (method 1) utilizes DS-1-32 P as the labeled substrate. Typically, 100 M DS-1-32 P (1000 cpm/nmol), 1 mg/ml cardiolipin, 50 mM Tris, pH 8.5, 5 mM ATP, 1% Nonidet P-40, and 5 mM MgCl 2 are mixed with 0.5-500 g/ml protein fraction and incubated at 30°C for various times. Reactions were stopped by spotting a portion of the reaction onto a Silica Gel 60 thin layer chromatography plate. Plates were developed in chloroform/ methanol/water/acetic acid (25:15:4:2, v/v/v/v) and analyzed as described above. Method 2 (which is not intended for the quantitative determination of specific activities) utilizes [␥-32 P]ATP (ϳ 8 ϫ 10 6 cpm/nmol) as the labeled substrate. The reaction conditions are exactly the same as for method 1, except that the ATP concentration is lowered to 0.6 M and only nonradioactive DS-1-P (final concentration of 100 M) is added. The reactions were stopped as described above, and plates were developed in chloroform/pyridine/formic acid/water (30:70:16:10, v/v/v/v).

Screening for Overproduction of Lipid A 4Ј-Kinase in Kohara
Library Lysates-Kohara et al. (24) have generated a library of 3400 mapped hybrid bacteriophage clones that covers the E. coli genome. A subset of this library containing 476 clones is available that covers 99% of the genome with some overlap between the clones (23, 32). Clementz et al. showed that enzy-matic activity could be detected in E. coli lysates produced by these hybrid clones (23). Activities of several enzymes involved in LPS biosynthesis were detected, and lysates generated from the clones containing the gene coding for the enzymes of interest displayed 5-10-fold overproduction of the activities (23). Assay of each individual lysate for overproduction of the lauroyltransferase (Fig. 1) led to the identification of htrB as the structural gene for that enzyme (23).
We employed the same approach to identify the gene for the lipid A 4Ј-kinase. The 4Ј-kinase activity was assayed in the lysates using method 1 (DS-1-32 P and 5 mM ATP). Under these conditions, product formation was linear with respect to time and protein concentration (data not shown), there were no side products, and the results were reproducible for a given lysate.
Fresh lysates of W3110 were prepared and assayed for 4Ј-kinase activity in six sets of 80. Fig. 2 shows the assay results for one set, hybrid clones [201]4H7 to [280]22E3. No single lysate in the collection gave the 5-10-fold overproduction seen with other enzymes of lipid A biosynthesis (23). However, there were several lysates with significantly higher activity than their neighboring lysates (Fig. 2). To choose lysates for further analysis, the mean and standard deviation values for each set of 80 were calculated. Fifteen clones, the activity of which surpassed the mean by more than two standard deviations, were reassayed (data not shown). One lysate, [218]E1D1 (marked by an asterisk in Fig. 2) consistently displayed 2-2.5fold more kinase activity than the other lysates.
Differences in lysis time could account for some of the variation of the activities seen in the lysates. The original lysates used to make the library were not generated from a fixed titer.  (33). Two genes in this region, msbA and kdsB, are related to the lipopolysaccharide system (33). kdsB encodes the CMP-Kdo synthase (3,34,35), and msbA encodes a putative LPS transporter (Fig. 3) (25,26) with homology to mammalian Mdr proteins. msbA was first identified by Karow and Georgopoulos (25,26) as a multicopy suppressor of htrB (36 -38), the gene encoding the Kdo-dependent lauroyltransferase ( Fig. 1) (22,23). msbA forms an operon with an essential downstream open reading frame of unknown function, orfE. (25). Insertion of an ⍀-chloramphenicol resistance cassette into the msbA gene blocks transcription of both msbA and orfE (25). Complementation of this msbA/orfE knockout only occurs with hybrid plasmids encoding both msbA and orfE, supporting the view that both genes are essential (25). As shown in Fig. 3, only about half of the msbA coding region is on clone [218]E1D1. In this clone, orfE is missing its native msbA promoter, and expression of this gene would be from read-through of genes. Given the relatively low overproduction of the 4Ј-kinase activity found in lysates generated with [218]E1D1 and the indication that orfE does not have its own endogenous promoter, we constructed a plasmid to overexpress orfE using the T7 RNA polymerase system.
Massive Overexpression of 4Ј-Kinase Activity on a Hybrid Plasmid Bearing orfE-The gene encoding orfE was cloned behind the T7 promoter of pET3a to form pJK2. Plasmid pJK2 was transformed into BLR(DE3)pLysS cells, an E. coli strain that carries the T7 RNA polymerase as a lysogen (Table I). The expression of T7 RNA polymerase is induced with isopropyl-1-thio-␤-D-galactopyranoside and leads to the expression of genes from the T7 promoter. Washed membranes from BR7, an E. coli strain deficient for diglyceride kinase (12,16), BLR(DE3)pLysS/pJK2, and BLR(DE3)pLysS/pET3a were assayed for 4Ј-kinase activity using DS-1-32 P as the phosphate acceptor. The result of this assay is shown in Fig. 5. The 4Ј-kinase activity was highly overexpressed in cells with pJK2 versus strain BR7 or cells with pET3a vector alone (Fig. 5, lanes  3 and 4 versus lanes 1, 2, and 5). When assayed at a protein dilution in which product formation is linear with respect to time, overexpression of orfE led to several thousand-fold overproduction of 4Ј-kinase activity. Table II shows the specific activities of the 4Ј-kinase in cell-free extracts, membrane-free cytosols (subjected to two ultracentrifugations), and washed membranes. orfE encodes a 328-amino acid protein with a predicted molecular mass of 36 kDa (25). Analysis of protein fractions from BLR(DE3)pLysS/pJK2 cells by SDS-polyacrylamide gel electrophoresis shows an overexpressed protein that is not present in protein fractions from BLR(DE3)pLysS/pET3a cells (Fig. 6). The overexpressed protein migrates with the molecular mass predicted from the sequence of orfE and is associated with the membranes (Fig. 6, lane 8). This is consistent with the hydropathy profile of orfE, which predicts 1 or 2 transmembrane helices in the N-terminal region of the protein. Like the protein, the 4Ј-kinase activity is also associated with the membranes, consistent with the hypothesis that orfE encodes the enzyme (Table II) Fig. 7. orfE and its homologues do not display significant sequence similarity to any other type of kinase, including those involved in carbohydrate, lipid, nucleic acid, or protein phosphorylation.
Analysis of Substrate Specificity and Generation of Novel Analogs with the Overproduced Kinase-The 4Ј-kinase is a useful tool for making 32 P-labeled substrates for the biochemical analysis of the enzymes catalyzing the late steps of the lipid A pathway (16 -18, 23). 32 P-Labeled lipid A precursors and substructures are also useful for studying the interactions of lipid A-like molecules with mammalian cells (19,20). To demonstrate the synthetic utility of the overexpressed 4Ј-kinase, several DS-1-P analogs were analyzed as 4Ј-kinase sub- strates (Fig. 8). DS-1-P is the physiological substrate for the 4Ј-kinase (3,12). 3-aza-DS-1-P has an amide-linked hydroxymyristate group at the 3-position instead of an ester-linked group (Fig. 8, NH indicated in boldface type). Mild alkaline hydrolysis of these compounds results in removal of the esterlinked hydroxymyristate moieties. The structures of the resulting compounds, designated base-treated DS-1-P and basetreated 3-aza-DS-1-P, are also shown.
Each lipid analog was tested as a 4Ј-kinase substrate using membranes containing recombinant, overexpressed 4Ј-kinase in conjunction with the method 2 assay conditions (0.6 M [␥-32 P]ATP as the phosphate donor and 100 M lipid acceptor). DS-1-P becomes phosphorylated to form [4Ј-32 P]lipid IV A by BR7 membranes (containing wild-type kinase levels), but only with a yield of ϳ0.5% (Fig. 9, lane 2). When BLR(DE3)pLysS/ pJK2 membranes are used (Fig. 9, lane 7) more than 50% of the 32 P is incorporated into [4Ј-32 P]lipid IV A . The more rapidly migrating product is most likely palmitoylated lipid IV A formed by the endogenous palmitoyltransferase present in the membranes (55). The 3-aza-DS-1-P, the base-treated DS-1-P, and the base-treated 3-aza-DS-1-P are also well utilized substrates for the recombinant, overexpressed 4Ј-kinase. Product a is formed efficiently from 3-aza-DS-1-P in the presence of BLR(DE3)pLysS/pJK2 membranes, but not BR7 membranes (Fig. 9, lane 8 versus lane 3). Products b and c are formed from base-treated DS-1-P and base-treated 3-aza-DS-1-P, respectively. In each case, the reaction is 100 -1000-fold more effective with membranes from BLR(DE3)pLysS/pJK2 membranes than with membranes from BR7 (Fig. 9, lanes 9 and 10 versus  lanes 4 and 5).
Other Properties of the 4Ј-Kinase-Consistent with previous results (12), lipid X (3, 49) and UDP-diacylglucosamine (3,50) were not substrates for the 4Ј-kinase, even when the enzyme was highly overexpressed. The enzyme apparently has a strong preference for glucosamine disaccharides.
Membranes from Rhizobium etli strain CE3 contain an unusual phosphatase that removes the 4Ј-phosphate from the lipid A precursor, Kdo 2 -lipid IV A (51). Solubilized CE3 membranes 2 were used to make 4Ј-dephosphorylated Kdo 2 -lipid IV A . The 4Ј-dephosphorylated Kdo 2 -lipid IV A is not a substrate for the overexpressed 4Ј-kinase (data not shown). The Kdo disaccharide may interfere with the presentation of the 4Ј-OH of the glucosamine disaccharide to the kinase.

DISCUSSION
In the present study, we have identified the gene encoding the lipid A 4Ј-kinase (12) as orfE, a previously sequenced open reading frame found near minute 21 in E. coli (25). Until the present work, the function of orfE was unknown. orfE is an essential downstream gene in an operon that also includes msbA, a recently discovered gene encoding a possible inner membrane lipid A translocator with homology to mammalian Mdr proteins (25,26,33). In two organisms other than E. coli, the msbA and orfE genes are similarly grouped together (39,40). The msbA/orfE operon represents a distinct new cluster of genes involved in the processing of LPS. Other LPS-related gene clusters in E. coli include the lpxA/B regions for lipid A biosynthesis, the rfa (waa) operons for core glycosylation and the rfb (wba) region for O-antigen assembly (3,35,52,53).
The orfE gene is cotranscriptionally expressed with msbA (25). The msbA gene was discovered by Karow and Georgopoulos (25) as a multicopy suppressor of the temperature sensitive phenotype of insertion mutations in the htrB gene. Clementz et al. (22,23) first identified HtrB as an acyltransferase of the lipid A pathway (Fig. 1). The role of msbA in LPS transport has been inferred from its genetic interaction with htrB (25), by the finding that htrB mutants accumulate LPS in their inner membranes (26), and by the sequence similarity of msbA to the ABC family of transporters (25). The mechanism of LPS transport from the inner leaflet of the inner membrane to the outer leaflet of the outer membrane is not well understood (3,54).
The identification of orfE as the lipid A 4Ј-kinase will require a reexamination of the msbA knockout phenotype. In this knockout construct, the 4Ј-kinase activity is deleted as well. While the lethality of this knockout is clearly due to a requirement for both msbA and orfE, the accumulation of LPS in the inner membranes of such constructs (26) must be reevaluated in light of the fact that lipid A biosynthesis is also being severely compromised by the lack of the 4Ј-kinase. A construct in which only msbA is inactivated (or the introduction of a hybrid plasmid expressing orfE into the msbA/orfE knockout) will be necessary to properly evaluate msbA's role in the cell. Last, it has not escaped our attention that OrfE has two potential membrane spanning domains and could itself be a component of the export machinery.
The 4Ј-kinase appears to be unique in that it shares no obvious sequence homology with other kinases. This may be so because phosphorylation of glucopyranosides at the 4-hydroxyl group is a relatively uncommon event in biology. Given the massive overproduction (Table I) of 4Ј-kinase observed when orfE is expressed behind a T7 promoter, it is likely that orfE is the structural gene for the kinase. We therefore suggest the new name lpxK.
Following the first description of the 4Ј-kinase in 1987 (12), washed membranes from cells containing wild type levels of kinase have been used to prepare substrates useful for studying the late reactions of the lipid A pathway (Fig. 1), such as [4Ј-32 P]lipid IV A (16,17,51,55). Since it was difficult to prepare [4Ј-32 P]lipid IV A in yields greater than 0.5% relative to the input [␥-32 P]ATP (because of the instability of the kinase at low ATP concentrations), no other substrates derived from [4Ј-32 P]lipid IV A , such as [4Ј-32 P]Kdo 2 -lipid IV A , could be generated in high radiochemical yields. Consequently, enzymatic studies of the late reactions ( Fig. 1) have been limited. Because the overexpression of the 4Ј-kinase is so great (ϳ2000-fold) in membranes from the LpxK overproducer (Table I), we can now prepare [4Ј-32 P]lipid IV A from [␥-32 P]ATP and DS-1-P with at least 50% yields (Fig. 9). This development will greatly facilitate the preparation of the distal intermediates of the lipid A pathway (Fig. 1) in 32 P-labeled form.
Overexpression of the 4Ј-kinase has also been shown to be useful for the enzymatic synthesis of 4Ј-32 P-labeled analogs of lipid IV A , including [4Ј-32 P]lipid A (Figs. 9 and 11). Using the DS-1-P analogs shown in Fig. 8 and compound 505 shown in Fig. 10, we have demonstrated that membranes containing the overexpressed kinase can be used to phosphorylate DS-1-P analogs containing two, three, four, or six fatty acyl chains. These analogs are not phosphorylated to any appreciable extent when wild-type membranes are employed as the enzyme source ( Figs. 9 and 11).
In the future, these phosphorylations should proceed even more efficiently when the 4Ј-kinase is available in homogeneous form and its kinetic properties have been characterized. It will be important to stabilize the solubilized kinase to facilitate purification and assay. From the results in Fig. 6, we anticipate that about a 20-fold purification will be required to achieve homogeneity. A histidine tagged variant of LpxK could facilitate the development of a rapid purification. The N-terminal sequence of the purified, overproduced 4Ј-kinase will be analyzed to verify that it is indeed the protein encoded by orfE.
In addition to the possibility of making radioactive analogs and intermediates efficiently, our demonstration of the enzymatic generation of 4Ј-phosphorylated lipid A analogs, using the overproduced kinase, has potential pharmaceutical implications. It may be possible to design novel enzymatic processes for the large scale 4Ј-phosphorylation of lipid A-like molecules. Such compounds are of great interest because certain molecules of this kind, like E5531 (43), have activity as endotoxin antagonists and are potentially useful for the therapy of endotoxin shock (56,57). For biological activity either as agonists or antagonists, lipid A-like molecules appear to require the presence of the 4Ј-phosphate (4,19,41,46). While E5531 has been synthesized entirely by chemical methods (43), it may yet be possible to design enzyme-based processes for large scale production. Indeed, there is an extensive literature on the use of the recombinant LpxB (3, 30, 58 -60) (the disaccharide synthase that functions just before LpxK in lipid A biosynthesis) for the preparation of diverse DS-1-P analogs (31,61,62). None of these compounds could ever be phosphorylated previously, because the cloned, overexpressed 4Ј-kinase was not available, and chemical methods were not applicable (47). The 3-aza-DS-1-P (Fig. 8) is an example of such an LpxB-generated analog (31). It would be of considerable interest to examine novel 4Ј-phosphorylated derivatives of existing DS-1-P analogs (31,61,62) for endotoxin antagonist or agonist activity.
Easy access to [4Ј-32 P]lipid A (Figs. 10 and 11) and its analogs should be useful for the identification and characterization of various lipid A binding proteins in animal cell membranes (20,(63)(64)(65). For instance, lipid A-like molecules bind to surface proteins, including CD14 (63,64,66) and the scavenger receptor (20,67). Recombinant LpxK may be useful in the preparation of defined, highly radioactive lipid A analogs bearing photoaffinity probes. Such probes may reveal additional, perhaps minor membrane proteins that interact with endotoxin, such as FIG. 9. Effective phosphorylation of DS-1-P and DS-1-P analogs with recombinant, overproduced 4-kinase. Washed membranes from strain BR7 and BLR(DE3)pLysS/pJK2 were used, as indicated, in 4Ј-kinase assays containing 100 M of the indicated DS-1-P analog, 0.6 M [␥-32 P]ATP, 50 mM Tris, pH 8.5, 5 mM MgCl 2 , 1% Nonidet P-40, 1 mg/ml cardiolipin, and 0.5 mg/ml washed membranes. After a 10-min incubation at 30°C, 5 l of the reaction was spotted onto a Silica Gel 60 TLC plate and developed in chloroform/pyridine/formic acid/ water (30:70:16:10, v/v/v/v). The plate was dried, exposed to a Phospho-rImager screen, and visualized using ImageQuant software. Arrows indicate the products of the reactions. If the native substrate DS-1-P is present in the reaction, [4Ј-32 P]lipid IV A is formed. When 3-aza-DS-1-P, base-treated DS-1-P, or base-treated 3-aza-DS-1-P is present, the 4Јphosphorylated product a, b, or c is formed, respectively. The migration of each of these products is slower than the nonphosphorylated substrate analog (not shown), consistent with the phosphate incorporation. As in Fig. 5, BLR/pJK2 is an abbreviation for BLR(DE3)pLysS/pJK2, and NE indicates the no enzyme control. Formation of 4Ј-phosphorylated products is 100 -1000-fold more effective with membranes from BLR/pJK2 than with BR7 membranes. Minor 32 P-containing lipids generated by membranes of BLR(DE3)pLysS/pJK2 in the absence of any acceptor substrate arise by the action of diglyceride kinase on endogenous glycerophospholipids (see especially lane 6). The diglyceride kinase is inactivated by mutation in strain BR7 (16) .   FIG. 10. Proposed reaction catalyzed by the 4-kinase with compound 505  the elusive signaling receptor that distinguishes endotoxin agonists from antagonists (3,63,64). Radioactive lipid A analogs, such as [4Ј-32 P]E5531, would also be very useful for in depth studies of lipid A uptake, metabolism, and distribution in animals. For studies of individual cells, the synthesis of fluorescent lipid A analogs might be helpful. Given that all of the genes encoding the enzymes of the system from LpxB to MsbB are now cloned (3), important synthetic opportunities remain to be explored.