(Received for publication, April 28, 1997, and in revised form, June 20, 1997)
From the Department of Biochemistry, Duke University Medical Center, Durham, North Carolina 27710
The genes for seven of nine enzymes needed for
the biosynthesis of Kdo2-lipid A (Re endotoxin) in
Escherichia 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
-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 of
E. coli generated using individual
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 overexpressing orfE 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.
Lipopolysaccharide
(LPS)1 is the major
glycolipid of the outer membrane of Gram-negative bacteria (1-5).
Lipid A, or endotoxin, is the hydrophobic anchor of LPS (1-5), and it
is a potent immunostimulant. It also appears to be responsible for many
of the features of septic shock that can accompany severe Gram-negative
infections (1-5). Lipid A is a disaccharide of glucosamine that is
phosphorylated at the 1- and 4-positions and is acylated with
R-3-hydroxymyristate at the 2-, 3-, 2
-, and 3
-positions
(1-5). In Escherichia coli, two additional fatty acyl
chains are also esterified to the 2
- and
3
-R-3-hydroxymyristate groups to form acyloxyacyl units
(Fig. 1) (1-5).
Lipid A biosynthesis begins with the acyl-ACP dependent acylation of
UDP-N-acetylglucosamine (6-10). Nine enzymes are required for the complete synthesis of Kdo2-lipid A (Fig. 1) (2, 3, 11). Seven of the nine structural genes coding for the enzymes of lipid
A biosynthesis in E. coli have been identified (3). However,
the lipid A 4-kinase gene has remained elusive (3). The 4
-kinase
catalyzes the sixth step of the pathway (Fig. 1) (12). It
phosphorylates the 4
-position of a tetraacyldisaccharide 1-phosphate
intermediate (termed DS-1-P) to form tetraacyldisaccharide 1,4
-bis-phosphate, also known as lipid IVA
(Fig. 1) (12-14).
Identification of the 4-kinase gene has been hampered because mutants
lacking the 4
-kinase have not been identified (3). Presumably the
4
-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 4-kinase can be used to make 4
-32P-labeled
lipid A precursors, such as [4
-32P]lipid IVA
and Kdo2-[4
-32P]lipid IVA, for
biochemical analyses of late pathway reactions (12, 16-18). The
4
-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 32P into
[4
-32P]lipid IVA is obtained when trace
quantities of [
-32P]ATP are used together with the
physiological substrate, DS-1-P (12, 16-18). This low level of
32P transfer makes it virtually impossible to use the
4
-kinase for phosphorylating DS-1-P analogs that are utilized less
efficiently. Identification and overexpression of the 4
-kinase gene
would allow investigation of DS-1-P analogs as substrates for the
4
-kinase, possibly facilitating the synthesis of interesting lipid
IVA analogs with potential activity as endotoxin
antagonists or mimetics (2, 3). Studies of endotoxin uptake and
metabolism (19-21) would be simplified if [4
-32P]lipid
A or various analogs could be made with the 4
-kinase.
To identify the gene encoding the lipid A 4-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
(22-24) were assayed for overproduction of 4
-kinase 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 4
-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 ~2000-fold
overproduction of 4
-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 4
-kinase and should be redesignated lpxK.
orfE does not display significant sequence homology to other
known kinases. The overexpressed 4
-kinase is very useful for the
efficient synthesis of labeled precursors and novel lipid A
analogs.
32Pi and
[-32P]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).
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 µg/ml for ampicillin, 12 µg/ml for tetracycline, and 30 µg/ml for chloramphenicol.
|
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 5-3
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).
DS-1-32P was made according to
Radika and Raetz (30). Milligram quantities DS-1-P were enzymatically
synthesized from UDP-2,3-diacylglucosamine and 2,3-diacylglucosamine
1-phosphate (lipid X) using a partially purified E. coli
disaccharide synthase preparation (30). 3-aza-DS-1-P was made in the
same manner except that 3-aza-lipid X (31) was used in place of lipid
X. Base-treated DS-1-P was made by treating 2 mg of DS-1-P for 30 min
with 0.2 M NaOH in 1 ml of chloroform/methanol (2:1, v/v).
The mixture was diluted 10-fold with chloroform/methanol (95:5) and
loaded onto a 5-ml silica column equilibrated with 50 ml of
chloroform/methanol (95:5). The column was washed with 25 ml of each of
the following ratios (v/v) of chloroform/methanol: 95:5, 90:10, 70:30,
1:1, 30:70, and 20:80. Thirty 5-ml fractions were collected. The
resolved hydroxy fatty acids and the deacylated DS-1-P were detected by
spotting 5 µl of each fraction onto a thin layer chromatography
plate, developing the plate in chloroform/methanol/water/acetic acid
(25:15:4:2, v/v/v/v), and charring with sulfuric acid. The hydroxy
fatty acids eluted with the 70:30 (v/v) solvent mixture. The
base-treated, deacylated DS-1-P eluted with the 1:1 solvent ratio. The
relevant fractions were pooled, and the solvent was removed by rotary
evaporation. The base-treated, deacylated 3-aza-DS-1-P was prepared in
the same manner. The elution profile for this compound was the same as
for the base-treated, deacylated DS-1-P. For use as substrates in
4-kinase assays, all lipid substrates were dispersed in 50 mM HEPES, pH 7.4, by sonic irradiation for 2 min.
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 MgSO4. The culture was
diluted 1:1 with 10 mM CaCl2, 10 mM
MgCl2. 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 MgSO4, 50 ml of 1 M Tris, pH 7.5, per
liter). Using 96-well microtiter plates, 5 µl of the individual
diluted lysates and 10 µl of the diluted host cell suspension were
mixed and incubated at 37 °C for 15 min. LB medium supplemented with
10 mM MgSO4 (150 µl) was added to each well, and incubation continued at 37 °C. After 4 h, the
A600 of each well was measured using a Molecular
Dynamics Spectramax 250 microplate reader. When the cell suspension had
cleared to an A600 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
80 °C overnight. The 4
-kinase
activity of each lysate was assayed in a 10-µl reaction mixture
containing 5 µl of lysate, 100 µM DS-1-32P
(1000 cpm/nmol), 1 mg/ml cardiolipin, 50 mM Tris, pH 8.5, 5 mM ATP, 1% Nonidet P-40, and 5 mM
MgCl2. After incubation at 30 °C for 60 min, the
reaction was stopped by spotting 5 µl 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-1-32P to
[1-32P]lipid IVA was quantified using
ImageQuant software (Molecular Dynamics).
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: 5-GTTTGGCATATGATCGAAAAAATCTGG-3
and
5
-ATTCATGGATCCATCAATCGAACGCTG-3
. 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.
pJK2 was transformed
into BLR(DE3)pLysS cells and grown at 37 °C in 2 liters of LB. When
the cultures reached an A600 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 × 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 × 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 × 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
80 °C. The membrane-free cytosol was also stored in aliquots at
80 °C. The protein
concentration was determined using the Bio-Rad protein assay kit with
bovine serum albumin as a standard.
Two methods for analyzing
4-kinase activity of various protein fractions were employed. The
first (method 1) utilizes DS-1-32P as the labeled
substrate. Typically, 100 µM DS-1-32P (1000 cpm/nmol), 1 mg/ml cardiolipin, 50 mM Tris, pH 8.5, 5 mM ATP, 1% Nonidet P-40, and 5 mM
MgCl2 are mixed with 0.5-500 µg/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 [
-32P]ATP (~ 8 × 106 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 µM and only nonradioactive DS-1-P (final concentration of
100 µM) 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).
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
enzymatic 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
4-kinase. The 4
-kinase activity was assayed in the lysates using
method 1 (DS-1-32P 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 4
-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 2-2.5-fold 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. To control for this
variation among lysates, the plaque-forming units (pfu) for lysates
derived from clones [216]13E3, [217]6D12, [218]E1D1, and
[320]15G10 were determined (Fig. 3).
Matched lysates were then made by infecting E. coli W3110
with 4 × 10
3 pfu. After 7-8 h, lysis occurred in
each case, and the lysates were again assayed for 4
-kinase activity as
before. The result is shown in Fig. 4.
The lysate of
[218]E1D1 persisted in having 2-2.5-fold
overproduction of 4
-kinase activity, compared with controls. This
finding led us to investigate further the genes on
[218]E1D1.
Genes Present on [218]E1D1
The clone [218]E1D1
contains a 20-kilobase pair fragment of the E. coli genome
spanning minutes 20.8-21.3 (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 4
-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.
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 4
-kinase activity using
DS-1-32P as the phosphate acceptor. The result of this
assay is shown in Fig. 5. The 4
-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 4
-kinase activity. Table
II shows the specific activities of the
4
-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 4-kinase activity is also associated with the membranes,
consistent with the hypothesis that orfE encodes the enzyme
(Table II).
Data base searches identified only two open reading frames of unknown
function from other Gram-negative bacteria with significant homology to
orfE. The predicted amino acid sequence of an open reading
frame identified in the Hemophilus influenzae Rd genome (39)
is 70.2% similar and 48.4% identical to the predicted amino acid
sequence of E. coli orfE. The valB gene (40) from
Francisella novicida encodes a protein that is 66.8%
similar and 41.4% identical to orfE. This strongly suggests
that the H. influenzae open reading frame and F. novicida valB may also be genes encoding lipid A 4-kinase
variants. An alignment of the three protein sequences is shown in 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 4-kinase is a useful tool for
making 32P-labeled substrates for the biochemical analysis
of the enzymes catalyzing the late steps of the lipid A pathway
(16-18, 23). 32P-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 4
-kinase, several DS-1-P
analogs were analyzed as 4
-kinase substrates (Fig.
8). DS-1-P is the physiological substrate
for the 4
-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 ester-linked
hydroxymyristate moieties. The structures of the resulting compounds,
designated base-treated DS-1-P and base-treated 3-aza-DS-1-P, are also
shown.
Each lipid analog was tested as a 4-kinase substrate using membranes
containing recombinant, overexpressed 4
-kinase in conjunction with the
method 2 assay conditions (0.6 µM
[
-32P]ATP as the phosphate donor and 100 µM lipid acceptor). DS-1-P becomes phosphorylated to form
[4
-32P]lipid IVA by BR7 membranes
(containing wild-type kinase levels), but only with a yield of ~0.5%
(Fig. 9, lane 2). When
BLR(DE3)pLysS/pJK2 membranes are used (Fig. 9, lane 7) more
than 50% of the 32P is incorporated into
[4
-32P]lipid IVA. The more rapidly migrating
product is most likely palmitoylated lipid IVA 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 4
-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).
Enzymatic Synthesis of [4
The
results using the DS-1-P analogs show that the 4-kinase can
efficiently phosphorylate disaccharides with two, three, or four acyl
chains. However, some of the most important lipid A-like molecules that
display either endotoxin agonist or antagonist activity contain five or
six acyl (or alkyl) chains (3, 41-44). To address whether the
recombinant, overexpressed 4
-kinase would be useful for making
32P-endotoxin agonists or antagonists, we attempted to
phosphorylate compound 505, a synthetic
hexaacyldisaccharide-1-phosphate (45-47). The 4
-phosphorylation of
this compound yields the major molecular species that constitutes
E. coli K-12 lipid A (endotoxin) (48), as shown in Fig.
10. Compound 505 was tested in a
4
-kinase assay system using membranes from BR7 and
BLR(DE3)pLysS/pJK2. The results are shown in Fig.
11. When BR7 membranes are used with
DS-1-P, a small amount of [4
-32P]lipid IVA
product (representing ~ 0.5% of the input
[
-32P]ATP) is formed (Fig. 11, lane 2). Use
of BLR(DE3)pLysS/pJK2 membranes leads to the formation of a large
amount of lipid IVA (Fig. 11, lane 5). If
compound 505 is assayed with BR7 membranes, no detectable product
([4
-32P]lipid A) is formed (Fig. 11, lane 3).
However, with BLR(DE3)pLysS/pJK2 membranes, about 5% of the input
32P from [
-32P]ATP is incorporated into
[4
-32P]lipid A (Fig. 11, lane 6). This novel,
enzymatically labeled product has the same migration as lipid A
isolated by pH 4.5 hydrolysis (48) from wild-type E. coli
cells (data not shown). In summary, when greatly overexpressed, the
4
-kinase is capable of phosphorylating glucosamine disaccharides that
are either more or less acylated than the native substrate, DS-1-P.
Other Properties of the 4
Consistent with previous
results (12), lipid X (3, 49) and UDP-diacylglucosamine (3, 50) were
not substrates for the 4-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 4-phosphate from the lipid A precursor, Kdo2-lipid IVA (51). Solubilized CE3
membranes2 were used to make
4
-dephosphorylated Kdo2-lipid IVA. The
4
-dephosphorylated Kdo2-lipid IVA is not a
substrate for the overexpressed 4
-kinase (data not shown). The Kdo
disaccharide may interfere with the presentation of the 4
-OH of the
glucosamine disaccharide to the kinase.
In the present study, we have identified the gene encoding the
lipid A 4-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 4-kinase will
require a reexamination of the msbA knockout phenotype. In
this knockout construct, the 4
-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 4
-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 4-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 4
-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 4-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 [4
-32P]lipid
IVA (16, 17, 51, 55). Since it was difficult to prepare
[4
-32P]lipid IVA in yields greater than
0.5% relative to the input [
-32P]ATP (because of the
instability of the kinase at low ATP concentrations), no other
substrates derived from [4
-32P]lipid IVA,
such as [4
-32P]Kdo2-lipid IVA,
could be generated in high radiochemical yields. Consequently,
enzymatic studies of the late reactions (Fig. 1) have been limited.
Because the overexpression of the 4
-kinase is so great (~2000-fold)
in membranes from the LpxK overproducer (Table I), we can now prepare
[4
-32P]lipid IVA from
[
-32P]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
32P-labeled form.
Overexpression of the 4-kinase has also been shown to be useful for
the enzymatic synthesis of 4
-32P-labeled analogs of lipid
IVA, including [4
-32P]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 4-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 4
-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 4-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
4
-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 4
-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 4
-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
4
-phosphorylated derivatives of existing DS-1-P analogs (31, 61, 62)
for endotoxin antagonist or agonist activity.
Easy access to [4-32P]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, 63-65).
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 the elusive signaling receptor that
distinguishes endotoxin agonists from antagonists (3, 63, 64).
Radioactive lipid A analogs, such as [4
-32P]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.
We thank Dr. D. Golenbock and Dr. S. Kusumoto for providing compound 505.