THE JOURNAL OF B~OLOCICAL CHEMEXRY 0 1990 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 265, No. 26, Issue of September 15, pp. 15410-15417,lQQO Printed in U.S.A.

Biosynthesis of Lipid A in Escherichia coli ACYL CARRIER PROTEIN-DEPENDENT INCORPORATION OF LAURATE AND MYRISTATE*

(Received for publication, March 5, 1990)

Kathryn A. Brozek and Christian R. H. Raetz$ From the Department of Biochemistry, University of Wisconsin-Madison, Madison, Wisconsin 53706

In previous studies we described enzyme(s) from Escherichiu coli that transfer two %deoxy-D-manno- octulosonate (KDO) residues from two CMP-KDO mol- ecules to a tetraacyldisaccharide- 1 ,I’-&-phosphate precursor of lipid A, termed lipid IVA (Brozek, K. A., Hosaka, K., Robertson, A. D., and Raetz, C. R. H. (1989) J. Biol. Chem. 264,6956-6966). The product, designated (KD0)2-IVa, can be prepared in milligram quantities and/or radiolabeled with 32P at position 4’ of the IVA moiety. We now demonstrate the presence of enzymes in E. coli extracts that transfer laurate and/ or myristate residues from lauroyl or myristoyl-acyl carrier protein (ACP) to (KD0)2-IVA. Thioesters of coenzyme A are not substrates. The cytosolic fraction catalyzes rapid acylation with lauroyl-ACP, but not with myristoyl, R-3-hydroxymyristoyl, palmitoyl, or palmitoleoyl-ACP. The membrane fraction transfers both laurate and myristate to (KD0)2-IVA. Evidence for the enzymatic acylation of (KD0)2-IVA is provided by (a) conversion of [4 ‘-32P](KDO)z-IVA to more rap- idly migrating products in the presence of the appro- priate acyl-ACP, (b) incorporation of [ l-‘4C]laurate or [ l-‘4C]myristate into these metabolites in the presence of (KD0)2-IVA, (c) fast atom bombardment-mass spec- trometry, and (d) ‘H NMR spectroscopy. At protein concentrations less than 0.5 mg/ml, the acylation of (KD0)2-IV* by the cytoplasmic fraction is absolutely dependent upon the addition of exogenous acyl-ACP. These acyltransferases cannot utilize lipid IV* as a substrate, demonstrating that they possess novel KDO recognition domains. The unusual substrate specificity of these enzymes provides compelling evidence for their involvement in lipid A biosynthesis. Depending on the conditions it is possible to acylate (KD0)2-IVA with 1 or 2 lauroyl residues, with 1 or 2 myristoyl residues, or with 1 of each.

All glycerophospholipids of Escherichia coli consist of mo- lecular species differing in fatty acyl composition (Cronan and Rock, 1987; Raetz, 1986). Palmitate, palmitoleate, and cis-vaccenate are the predominant moieties esterified to sn- glycerol 3-phosphate backbone under ordinary circumstances, but cyclopropane derivatives of unsaturated fatty acids accu- mulate in stationary cells. Small amounts of myristate and laurate are also frequently encountered.

* This research was supported by National Institutes of Health Grant DK19551 (to C. R. H. R.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accord- ance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ To whom correspondence should be addressed: Dept. of Biochem- istry, Merck Sharp & Dohme Research Laboratories, P. 0. Box 2000, Rahway, NJ 07065.

Mutants defective in unsaturated fatty acid biosynthesis require exogenous unsaturated fatty acids for growth (Cronan and Rock, 1987; Raetz, 1986; Silbert, 1975). Many unnatural structures can satisfy this requirement, including trans-un- saturated, polyunsaturated, and branched-chain fatty acids. Despite their rapid uptake and efficient incorporation into glycerophospholipids, exogenous fatty acids cannot be elon- gated or further unsaturated by cells of E. coli (Silbert, 1975). Exogenous fatty acids are converted to coenzyme A thioesters, whereas endogenous fatty acids are synthesized as thioesters of acyl carrier protein (ACP)’ (Nunn, 1986). Elongation and unsaturation of fatty acids is dependent upon the ACP moiety (Silbert, 1975). In contrast, the sn-glycerol3-phosphate acyl- transferases can utilize thioesters of either coenzyme A or ACP (Green et al., 1981).

The lipid A moiety of E. coli lipopolysaccharide (LPS) (Fig. 1) is more homogeneous with respect to its fatty acyl compo- sition than are glycerophospholipids (Rietschel, 1984; Raetz, 1990). The acyltransferases that attach the O-linked and N- linked R-3-hydroxymyristate residues during the initial steps of lipid A biosynthesis (Fig. 1) have absolute specificity for ACP thioesters and considerable selectivity for R-3-hydroxy- myristate (Anderson and Raetz, 1987; Anderson et al., 1988). This observation is consistent with the fatty acid composition of E. coli lipid A (Rae& 1990) and the finding that exogenous, radiolabeled R-3-hydroxymyristate is not incorporated into lipid A by living cells, even though it is utilized as a carbon source (Harder et al., 1972; Osborn, 1979).

Enzymes capable of attaching laurate and myristate moie- ties to lipid A precursors (Fig. 1, dashed arrows near bottom) have not been reported previously. We now describe soluble and membrane-bound acyltransferase(s) in E. coli extracts that incorporate laurate and myristate, respectively. As in the case of the R-3-hydroxymyristoyl transferases (Fig. l), only the thioesters of ACP function as substrates. The observation that (KD0)2-IVA, but not lipid IV* (Fig. l), can accept laurate and myristate residues demonstrates that these enzymes pos- sess novel KDO recognition domains. The carbohydrate spec- ificity of these acyltransferases explains why KDO-deficient mutants of Salmonella typhimurium accumulate lipid IV* (Fig. 1) under nonpermissive conditions (Raetz et al., 1985; Rick et al., 1977).

EXPERIMENTAL PROCEDURES

Materials

[y-32P]ATP was a product of Amersham International. Nucleotide triphosphates and KDO were obtained from Sigma. Other items were purchased from the following companies: Triton X-100 (Research Products International, Elk Grove Village, IL); yeast extract and

‘The abbreviations used are: ACP, acyl carrier protein; LPS, lipopolysacccharide; KDO, 3-deoxy-D-manno-octulosonate; HEPES, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid.

15410

by guest on August 12, 2019

http://ww

w.jbc.org/

Dow

nloaded from

Novel Acyltransferases in E. coli 15411

UDP-GlcNAc

FIG. 1. Biosynthesis of lipid A in E, coli. All of the enzymes shown have been characterized previously except for the ones that attach laurate and myristate (Raetz, 1990). Lipid IV, is the major precursor that accumulates in KDO-deficient mutants (Raetz et al., 1985).

tryptone (Difco Laboratories, Detroit, MI); Silica Gel-60 thin layer plates, 0.25 mm (E. Merck, Darmstadt, West Germany); silicic acid, BioSil A 200-400 mesh (Bio-Rad); bicinchoninnic acid (Pierce Chem- ical Co.).

Preparations

Lipid IVA (Raetz et cn., 1985), (KDO)s-IVA (Brozek et al., 1989), [4’-R’P]IVA (Brozek et al., 1987), and [4’-“‘P](KDO)z-IV* (Brozek et al., 1989) were prepared as described previously. Lauroyl-ACP, myr- istoyl-ACP, palmitoyl-ACP, palmitoleoyl-ACP, [ 1-“Cllauroyl-ACP, and [I-“C]myristoyl-ACP were prepared as previously described, using immobilized acyl-ACP synthase (Anderson and Raetz, 1987). Salmonella minnesota R59.5 LPS was kindly supplied by Dr. Kuni Takayama of the University of Wisconsin-Madison and was prepared by the chloroform/phenol/petroleum ether method (Galanos et al., 1969).

Bacterial Strains and Growth Conditions

All strains used are derivatives of E. coli K-12 and were obtained from the E. coli Genetic Stock Center, Yale University. E. coli DZle7 is an ampicillin-resistant strain with an rfa-I mutation and RcLPS. E. coli DZlf2 is derived from D21e7, carries an rfa-31 mutation in addition to rfa-I, and has ReLPS (Havekes et al., 1977). E. coli D31m4 is an ampillicin-resistant strain with the mutations rfa-229 and rfa- 230, and has ReLPS (Boman and Monner, 1975). S. typhimurium SA1377 (rfac), SL3600 (rfaD), and SLllOZ (rfaE) were obtained from the Salmonella Genetic Stock Center, University of Calgary. Cells were grown at 37 ‘C on LB broth containing 10 g of NaCI, 5 g of yeast extract, and 10 g of tryptone/liter (Miller, 1972).

Preparation of Cell Extracts

Cells were harvested by centrifugation when the optical density at 550 nm reached approximately 1.0, resuspended in 10 mM potassium phosphate, pH 7.5 (5-10 ml/liter culture), and disrupted using a French pressure cell at 18,000 p.s.i. To remove unbroken cells, the homogenate was centrifuged at 10,000 x g., for 15 min. This low speed supernatant was centrifuged at 150,000 x ga:8. for 90 min, yielding a high speed supernatant (the soluble fraction) and a pellet. The pellet was resuspended in 5-10 ml of 10 mM potassium phosphate, pH 7.5, and centrifuged again at 150,000 x g.” for PO min. The washed pellet was resuspended once again in 5-10 ml of 10 mM potassium phosphate (the membrane fraction). All steps were carried out at 4 “C, and the resulting samples were stored at -70 “C. Protein con- centrations were determined by the bicinchoninnic acid method using a bovine serum albumin standard (Smith et al., 1985).

Assay of (KDO),-IV, Metabolism

Method A-Concentrated soluble fractions were used in some experiments so that the amount of endogenous acyl or heptosyl donor(s) were high enough to yield reaction product(s). These reac- tions contained 50 mM HEPES, pH 7.5, 0.1% Triton X-100, 10 mM MgC12,5 mM ATP, 25 pM [4’-32P](KD0)2-IVA (l-2 X lo4 cpm/nmol), and lo-15 mg/ml E. coli-soluble fraction in a final volume of 20 ~1. The reaction mixtures were incubated at 30 “C for the indicated times in 500-~1 Eppendorf tubes. The reactions were stopped by spotting 5- ~1 samples directly onto sitica gel thin layer plates. The spots were dried with a cold air stream and developed with solvent A (chloroform, pyridine, 88% formic acid, water (30:70:16:10, v/v)) or with solvent B (chloroform, methanol, water, acetic acid (25:15:4:2, v/v)). The “lP- labeled starting material and products were located by autoradiogra- phy at -70 “C using an enhancing screen and quantitated by scraping the appropriate regions of the silica and counting in 10 ml of Biosafe II mixture.

Method B-In experiments to optimize the rate of (KDO)*-IVA acylation and to identify the possible acyl donors, exogenous rather than endogenous acyl donors were utilized. These reactions consisted of 50 mM HEPES, pH 7.5,0.1% Triton X-100,25 pM [4’-“‘P](KDO),- IVA, 25 PM acyl-ACP, and 0.3 mg/ml protein of a soluble fraction. In such dilute extracts, there was no detectable acylation of (KDO)l-IV,+ in the absence of added acyl donor, and there was no interfering heptosylation, which is dependent on the presence of M$+ and ATP. The reactions were processed and quantitated as above.

by guest on August 12, 2019

http://ww

w.jbc.org/

Dow

nloaded from

15412 Novel Acyltransferases in E. coli

Large Scale Preparation of Two Acyluted (KDO)z-ZVA Metabolites Generated in Concentrated Cytoplasmic Extracts

A 15-ml reaction mixture consisted of 67 mM HEPES, pH 7.5, 0.1% Triton X-100, 10 mM MgC12, 6.7 mM CTP, 2.7 mM kfi0, 226 milliunits of CMP-KDO svnthase (Brozek et al.. 1989). 105 vM 14’- ‘*P]lipid IVA (6300 cpm/nmol), and 7 mg/ml protein’of a’soluble fraction from E. coli D21f2. The first six components were mixed and incubated for 10 min at room temperature to generate CMP-KDO. To prevent its precipitation with magnesium, the 14’-32Pllipid IV* was-first prepared as-a 400~pM stock solution in 0.06% Triton X-100 to form mixed micelles. Followine the addition of the 14’-3*Pllinid IVJTriton X-100 mixture and the D21t2. extract to the’CMP-KbO generating system, the reaction was incubated at 30 “C with gentle shaking for 18 h. To extract the products from the reaction mixture, 75 ml of methanol, 37.5 ml of chloroform, 15 ml of H20, and 0.8 ml of concentrated HCl were added to form a single-phase Bligh and Dyer mixture (Bligh and Dyer, 1959). The entire mixture was shaken, poured into 150-ml Corex glass bottles, and centrifuged at 1200 X g,, for 15 min at 20 “C. The clear supernatant was removed, and the precipitate was washed with a mixture consisting of 19 ml of chloro- form, 37 ml of methanol, and 15 ml of 0.2 M HCl. After mixing and centrifuging as above, the clear supernatants were combined and added to a flask containing 56 ml of H20 and 56 ml of chloroform, generating a two-phase Bligh and Dyer system. The mixture was shaken, poured into Corex bottles, and centrifuged at 1200 X g,, for 15 min to separate the phases. The lower phases (112 ml) were removed, and the upper phases were washed with 56 ml of acidic pre- equilibrated lower phase and recentrifuged. The lower phases were combined, and 9.3 ml of distilled pyridine was added. The solvents were removed by rotary evaporation. The residue was dissolved in 8 ml of solvent C (chloroform, pyridine, methanol, 88% formic acid, water (50:60:2.5:15:3, v/v)) and applied to a BioSil A column (-1 x 10 cm) equilibrated in solvent C. The column was washed with 50 ml of solvent C, 30 ml of a 1:l (v/v) mixture of solvent C and solvent D (chloroform, pyridine, methanol, 88% formic acid, water (45:65:2:15:5 (v/v)), and 25 ml of solvent D. Fractions of approximately 2 ml were collected. Next, 5-~1 samples were spotted on a silica gel plate and developed in solvent A. Fractions containing the putative monoac- ylated (37-44) and diacylated (19-30) products (see below), as judged by autoradiography of the thin layer plate, were pooled and diluted with 32 and 48 ml of chloroform/methanol (955, v/v), respectively. Each mixture was loaded onto a separate 2 ml BioSil A column equilibrated in chloroform/methanol (925, v/v), and the columns were then washed with 15 ml of chloroform/methanol (95:5, v/v). The desired compounds were each eluted with a mixture of 2.5 ml of chloroform, 5 ml of methanol, and 2 ml of 100 mM ammonium acetate, adjusted to pH 1.5 with HCl. To each eluate was added 2.5 ml of chloroform and 2.5 ml of HrO. The lower phases were removed, and the upper phases were washed with 5 ml of acidic pre-equilibrated lower phase. The lower phases for each sample were pooled, mixed with 0.5 ml of distilled nvridine. and dried under a stream of NI. Approximately 0.2 mg of monoacylated and 0.5 mg of diacylated (KDO)P-IV~ were recovered. Traces of solvent were removed with a vacuum pump.

Mild Acid Hydrolysis Conditions

A small sample of lipid IV* or lauroyl-(KDO)2-IVA was dispersed in 0.9 ml of 0.1 M HCl bv bath sonication. After 60 min at 100 “C!. 1 ml each of chloroform and methanol were added to generate a two- phase Bligh and Dyer mixture (Bligh and Dyer, 1959). The lower phases were collected and dried under nitrogen.

(KDOkIVa Is Converted to Several Different Products by Soluble Fractions of E. coli and S. typhimurium-We have previously identified a key intermediate in the biosynthesis of lipopolysaccharide containing 2 KDO residues, designated (KDO)z-IV* (Brozek et al., 1989). Since the minimal LPS sufficient to support cell growth has this structure plus the two normal fatty acids, laurate and myristate (Fig. l), we examined the possibility that (KDO)z-IV* might be a sub- strate for these acylations, as well as for other modifications, such as heptose addition or phosphorylation. To investigate these questions, [4’-32P](KD0)2-IVA was incubated with con-

centrated soluble fractions of E. coli or S. typhimurium (Fig. 2). Depending on the extract, the (KDO)*-IV* was rapidly converted with time to a complex array of more polar and/or more hydrophobic compounds (Fig. 2). The formation of (KDO)P-IVA metabolites was not observed when the extract protein concentration was below 0.5 mg/ml, presumably be- cause endogenous acyl and heptosyl donors were diluted out (see below).

In the course of our studies, we observed that all available heptose-deficient strains of E. coli and S. typhimurium (Fig. 2, lanes 2-6) failed to form the more polar metabolites and only made the two more hydrophobic ones. This result sug- gests that the polar metabolites are products of heptose ad- dition. Omission of ATP from the reaction mixture similarly inhibited the formation of the more polar metabolites in extracts of heptose-containing strains, but ATP was not re- quired for the synthesis of the two more hydrophobic com- pounds (data not shown).

Lauroyl-ACP Stimulates Formation of the More Hydropho- bic Metabolites of fKDO)n-IV*-To investigate whether the hydrophobic metabolites of (KDO)*-IV* reflected the addition of laurate or myristate, we diluted the E. coli-soluble fraction to the point where no reaction was observed (0.3 mg/ml), presumably due to lack of sufficient endogenous acyl donors. Lauroyl or myristoyl derivatives of coenzyme A and acyl carrier protein were then added back, but only lauroyl-ACP was capable of stimulating significant formation of a more rapidly migrating product (Fig. 3, lane 2 in panel A). If the reaction was allowed to proceed for a longer time, a second more hydrophobic product was also formed in the presence of lauroyl or myristoyl ACP (lanes 2 and 3 in panel B), but only a trace of the first product was formed in the presence of lauroyl coenzyme A (data not shown). Although the soluble

8

-(KDO)2IVA

123456 7

FIG. 2. Metabolism of (KD0)2-IVa by concentrated cyto- plasmic fractions of various strains of E. coli and S. typhi- murium. A 50-pl reaction, containing 50 mM HEPES, pH 7.5, 10 mM MgCl*, 0.1% Triton X-100, 5 mM ATP, 25 pM [4’-32P](KD0)2- IV* (10,300 cpm/nmol), and 10 mg/ml cytoplasmic protein was in- cubated at 30 “C for 30 min, as in Method A of “Experimental Procedures.” A 5-~1 portion was withdrawn from each reaction and spotted onto a thin layer chromatography plate, which was developed in solvent A. Autoradiography was carried out for 15 h at - 70 “C using an enhancing screen. Lone 1, no extract; 2, E. coli D31m4; 3, E. coli D21f2; 4, S. typhimurium SA1377; 5, S. typhimurium SL3600; 6, S. typhimurium SL1102; and 7, E. coli D21e7.

by guest on August 12, 2019

http://ww

w.jbc.org/

Dow

nloaded from

Novel Acyltransferases in E. coli

-SF * w*

-SF +FFA

-- c 2ndACYLATION *- - IStACYLATION

I-f - ;; -(KD0)2-IVA d

-0

12 345 12345

A B FIG. 3. Dependence of the formation of more hydrophobic

metabolites of (KDO),-IV* on added acyl-ACP. The experiment was carried out with dilute cytoplasmic extract as described in Method B under “Experimental Procedures” with 0.3 mg/ml cytoplasmic nrotein of E. coli D21e7. Panel A. 30 min of incubation: Panel B. 4 h bf incubation. Panel A: lane 1, no added acyl donor; 2,25 pM lauroyl- ACP; 3, 25 pM myristoyl-ACP; 4, 25 jtM lauroyl-coenzyme A; and 5, 25 pM myristoyl-coenzyme A. Panel B: lane I, no acyl donor; 2, 25 pM lauroyl-ACP; 3,50 pM lauroyl-ACP; 4, 25 WM myristoyl-ACP; and 5, 50 PM myristoyl-ACP.

fraction was capable of forming both acylated derivatives of (KDO)P-IVA efficiently in the presence of lauroyl-ACP, the addition of myristoyl-ACP supported very little (KDO)z-IV* acylation (lanes 4 and 5 in panel B). One factor contributing to the poor myristoylation of (KDO)r-IVA by the soluble fraction may be hydrolysis of the myristoyl-ACP, which oc- curred to a significantly greater extent than the hydrolysis of lauroyl-ACP (data not shown).

Incorporation of Radiolabeled Laurate and My&ate into the Products-To demonstrate that actual acyl transfer was taking place in the soluble fraction and that acyl-ACP was not merely acting as a regulator, [1-‘%]acyl-ACP was em- ployed in a reaction with non-radioactive (KDO)*-IVn (Fig. 4, lanes I-4). In the presence of (KD0)2-IVA, the [l-‘%I-acyl- ACP is converted to two new products (lanes 2 and 3). These migrate with the same R* as the compounds generated from [4’-32P](KDO)P-IVA in the presence of lauroyl or myristoyl- ACP (Fig. 3, panel B).

The membrane fraction of the extract was also tested for its ability to transfer laurate or myristate from acyl-ACP to (KDO)r-IV, (Fig. 4, lanes 5-8). The membranes catalyzed the transfer of both laurate and myristate (lanes 6 and 7). There was substantially more myristoyl transfer by membranes than by the soluble fraction (lanes 3 and 7) but about the same amount of lauroyl transfer (lanes 2 and 6). Interestingly, the membranes in the presence of myristoyl-ACP generated a greater proportion of the second acylation product (lane 7), suggesting that the second acyltransferase transfers myristate more rapidly than the first under these conditions.

Without [‘%]acyl-ACP, it is difficult to investigate the ability of membranes to metabolize (KD0)2-IVA. First, there is very little endogenous lauroyl- or myristoyl-ACP in the membrane fraction as prepared in this study. Second, mem-

P A n - ACYL-ACP

-0

12345676

FIG. 4. Transfer of [1-“Cllauroyl and [ 1-W]myristoyl moieties from [l- W]acyl-ACPs to (KDO)z-IVA. Experimental conditions were identical to those given in Fig. 3, except that 25 fiM nonradioactive (KD0)2-IV, was substituted for the [4’:32P] (KDdh- IVn. and 25 YM ll-“Clacvl-ACP (58 vCilrrmo1) was substituted for the non-radioactive a&-ACP. At 60 min; 5-~1 samples were spotted on thin layer chromatography plates and developed in solvent B. Lanes 1, 4, 5, and 8 are reactions without (KDO)*-IV*. Reactions Z-4 contained 0.3 mg/ml E. coli DZle7-soluble fraction, and reaction 5- 8 contained 0.3 mg/ml E. coli D21e7 membrane fraction. Reactions corresponding to lanes 1, 2, 5, and 6 contained [l-‘4C]lauroyl-ACP, and 3, 4, 7, and 8 contained [l-W] myristoyl ACP. Autoradiography was carried out overnight. 0, origin; SF, solvent front; FFA, free fatty acid.

branes can generate yet another more hydrophobic metabolite of [4’-32P](KD0)2-IVA that is distinct from both products discussed here (data not shown). This substance is likely to be a palmitoylated derivative of (KDO)z-IV*, generated by the same system responsible for the conversion of lipid X to lipid Y (Brozek et al., 1987). The only way to assay membranes for lauroyl or myristoyl transfer without interference by the palmitoyl transferase is with a dilute preparation of mem- branes and exogenous [‘4C]acyl-ACP.

Whether endogenous or exogenous acyl donors are em- ployed (Figs. 2-4), two acylated products are always generated from (KDO)?-IVA. Contrary to what one would expect, based on the accepted structure of E. coli lipid A (Fig. l), lauroyl- ACP and myristoyl-ACP are not both necessary together to generate the second acylation product. This result suggests that species with two identical acyl chains attached to (KD0)2-IVA are being generated in uitro. In mature lipid A isolated from E. coli, the predominant species is thought to be substituted almost exclusively with one laurate on the N- linked hydroxymyristate and one myristate on the O-linked hydroxymyristate of the non-reducing GlcN residue (Fig. l), although minor dilauroyl and dimyristoyl species cannot be excluded. Perhaps, the lack of absolute specificity in the in vitro acylation of (KD0)2-IVA is due to the loss of some regulatory mechanism present in the intact cell. It is also uncertain how many distinct acyltransferases are involved in the reactions demonstrated in Figs. 2-4. Purification of the acyltransferases will be necessary to address this question.

Characterization of the Soluble Lauroyl Transferase-Since the observed activities are somewhat complex, we chose to further characterize the soluble activity that transfers the first lauroyl group to (KD0)2-IVA. Although the activity is recovered in the soluble fraction of a cell-free extract, the enzyme requires detergent for optimal activity (Table I), presumably because the substrates must be dispersed in a mixed micellar state. To test the specificity of the reaction for acyl chain length, several acyl-ACPs were employed using

by guest on August 12, 2019

http://ww

w.jbc.org/

Dow

nloaded from

15414 Novel Acyltransferases in E. coli

TABLE I Requirements for the formation of monolauroyl-(KD0)2-IVA

Assays in duplicate were carried out using the cytoplasmic fraction and Method B, except that various components were omitted, as indicated.

Conditions % Conversion in 30 min

Complete system 28.0 -Triton X-100 1.8 -Lauroyl-ACP 0.0 -E. coli cytoplasmic fraction 0.0

TABLE II Acylation of (KDO)p-IV, by cytoplasmic extracts with various acyl-

ACPs Assays were carried out using cytoplasmic extracts and Method B,

except that different acyl-ACPs were used, as indicated. Acyl-ACP % Conversion

Lauroyl 27.0 Myristoyl 3.8 R-3-Hydroxymyristoyl 2.1 Palmitoyl 0.7 Palmitoleovl 0.0

0 30 60 90 120

TIME (min)

--SF

-0

12

FIG. 5. Addition of laurate to (KD0)2-IVA but not to lipid IVA. The assay was conducted using Method B (0.3 mg/ml protein from E. coli D2le7-soluble fraction and 25 pM lauroyl-ACP). Reac- tions were stopped after 1 h, and 5-~1 samples were analyzed by thin layer chromatography in solvent A. Autoradiography was overnight at -70 “C with an enhancing screen. Even after a 12-h reaction (not shown), no conversion was observed with lipid IVA.

the standard assay conditions. As shown in Table II, the enzyme has a relatively high specificity for laurate, but a small amount of product is always observed in the presence of myristoyl-ACP. Whether this is due to the presence of a distinct myristoyl transferase or to a lack of absolute specific- ity of the lauroyl transferase is uncertain. Importantly, lipid IV* is not an acceptor of a lauroyl group under the conditions used (Fig. 5), demonstrating the need for KDO addition prior to the addition of laurate.

The enzymatic formation of the first acylated product is linear for about 30 min when 25 pM (KDO)z-IV* and 25 FM lauroyl-ACP are used as substrates, but the reaction does not go to completion (Fig. 6A). The dependence of reaction prod- uct on soluble protein concentration (B) is approximately linear. If higher concentrations of lauroyl-ACP are used, significant amounts of the second acylation product are

0 0.1 0.2 0.3 0.4 0.5

PROTEIN, mg,ml

FIG. 6. Dependence of laurate addition to (KDO)z-IVA upon time and protein concentration. Panel A, time course of the reaction. Assays contained 0.1 mg/ml (lower line) or 0.3 mg/ml (upper line) protein from the soluble fraction of E. coli D21e7 and 25 pM lauroyl-ACP. At the indicated times, 5-~1 samples were removed and spotted on thin layer plates, which were developed in Solvent A. Autoradiography was carried out overnight at -70 “C with an en- hancing screen. The remaining (KDO),-IV* and monoacylated prod- uct formed were quantitated as described in Method A. The percent conversion for each reaction was calculated and plotted. Panel B, Dependence of (KDO)z-IV* acylation on protein concentration. Each 20-~1 reaction contained 50 mM HEPES, pH 7.5, 0.1% Triton X-100, 25 pM [4’-3ZP](KD0)2-IVA (20,000 cpm/nmol), 25 pM lauroyl-ACP, and 0.05-0.5 mg/ml protein from E. co/i DZle7-soluble fraction. After 20 min at 30 “C, 5-~1 samples were withdrawn and processed as in panel A.

formed. We have not yet found a way to inhibit the second acylation while maximizing the rate of the first.

Preliminary experiments show that the apparent K, values for both substrates are below 10 pM, indicating that the lauroyl transferase is saturated in the concentration range (25-50 pM) in which we are working (data not shown). Increasing the concentration of acyl-ACP to 75 pM is somewhat inhibi- tory.

Fast Atom Bombardment-Mass Spectrometry of Enzymati- tally Acylated (KDO)n-IVa-To further characterize the two products, a large scale reaction employing a concentrated soluble fraction from E. coli and endogenous acyl donors was prepared as described under “Experimental Procedures.” Fast atom bombardment-mass spectrometry in the negative mode of the putative monoacylated product yielded a single major peak at 2027.9, as expected for [M - HI- of lauroyl-(KDO)n- IV*. The putative diacylated (KDO)*-IV* generated two prominent peaks of approximately equal intensity at 2208.5 and 2237.5, which may be interpreted as [M - HI- of dilau- royl-(KDO)n-IVA and lauroyl-myristoyl-(KDO)p-IVA, respec-

by guest on August 12, 2019

http://ww

w.jbc.org/

Dow

nloaded from

Novel Acyltransferases in E. coli 15415

tively. These results are consistent with the radiochemical experiments shown in Figs. 2-4, but, as noted above, are not entirely consistent with the fatty acyl composition of mature lipid A isolated from E. coli (Fig. 1).

The mass spectra also confirm that laurate is utilized pref- erentially during the first acylation when endogenous acyl donors are employed, consistent with the kinetic selectivity of the first acylation for laurate under dilute conditions with exogenous acyl donors (Table II).

‘H NMR Spectroscopy of the Diacylated Product-The en- zymatic synthesis and isolation of the above samples for mass spectral analysis provided enough material for anlaysis by ‘H NMR. A 0.5-mg sample of the diacylated product was pre- pared as described under “Experimental Procedures” and subjected to ‘H NMR at 500 MHz (Fig. 7, panel C). For comparison, samples of (KD0)2-IVA and the well-defined LPS from S. minnesota R595 (ReLPS) (Christian et al., 1984; Brozek et al., 1989) were prepared as pyridinium salts and analyzed in an identical manner (Fig. 7, panels A and B, respectively). The features suggestive of acyloxyacyl residues are present in the diacylated product derived from (KDOh?- IVA in vitro, as in R595 LPS, but not in (KDO)*-IV*. These include 1) a large peak in the a-methylene region at 2.2-2.3 ppm, indicative of the a-methylenes of the newly added nor- mal fatty acids, 2) a large peak at 1.6 ppm, representing the /3-methylenes of the newly incorporated normal fatty acids, and 3) the additional peaks at 5.1-5.3 ppm, not present in (KD0)2-IVA, presumed to arise from two protons on the /3- carbons of two R-3-hydroxymyristoyl moieties that have been esterified with laurate and/or myristate.

Mild Acid Hydrolysis of (KDO),-IV* and (Lauroyl)-(KDO& IV,-Although the radiochemical experiments, the mass spectra, and the NMR data demonstrate the presence of esterified laurate and myristate in the monoacylated and diacylated products, it remains to be shown precisely where these modifications have taken place. To rule out esterifica-

C

1 h ,:, iw bJ

50 10 10 10 10 PPM

FIG. 7. ‘H NMR spectroscopy of (KD0)2-IVA, S. minnesota R595-ReLPS, and the diacylated derivative of (KDO),-WA. Panel A, (KDO),-IVA; panel B, S. minnesota R595-ReLPS; panel C, diacylated enzymatic product (see text).

tion of the lauroyl group to KDO, a small sample of lauroyl- [4’-“2P](KD0)2-IVA was treated with 0.1 M HCl at 100 “C for 60 min, and the reaction products were analyzed by thin layer chromatography (Fig. 8). This treatment removes the ano- merit phosphate as well as the KDO residues (Qureshi et al., 1982). If the lauroyl group were linked to KDO, the acid hydrolysis products of lauroyl-[4’-“2P](KD0)2-IVA would mi- grate the same as those obtained upon acid hydrolysis of [4’- “‘PIlipid IV*, which are often a doublet (Fig. 8A) (Galloway and Raetz, 1990). The major hydrolysis product of the lauroyl- [4’-“2P](KD0)2-IVA migrates faster than the product of [4’- 32P]lipid IV* (B). This result confirms that the lauroyl group is attached to the lipid A disaccharide, not to KDO.

It is likely that the lauroyl and myristoyl groups incorpo- rated under our conditions are part of acyloxyacyl units. First, the only free hydroxyl on the lipid A disaccharide is at the 4- position of the reducing glucosamine residue, but acyl- ation of this position is sterically unfavorable. Second, NMR analysis of the diacylated product (Fig. 7C) rules out the formation of acylated phosphates, in which case the additional protons observed between 5.1 and 5.3 ppm would not be present. Consequently, the hydroxyl groups of R-3-hydroxy- myristoyl residues are the only ones available for acylation with laurate and myristate. Since tri- and tetraacylated de- riatives of (KD0)2-IV~ are not generated (Figs. 2-4), we presume that the laurate and myristate have been properly incorporated into the non-reducing GlcN residue of lipid A (Fig. 1).

Acylated (KDO)a-IVas as Acceptors of Heptose Residues- (KD0)2-IVA can be converted to more polar metabolites (Fig. 2, lane 7) in extracts made from heptose-containing cells, presumably reflecting the addition of heptose residues. To compare the ability of (KDO)*-IVA and its acylated metabo- lites to accept heptose, a mixture of mono- and dilauroyl- (KDO)*-IV* was generated in uitro (Fig. 9, lane I), after which crude ADP-heptose (Kontrohr and Kocsis, 1986) was added to stimulate the formation of the putative heptose-containing products (Fig. 9). In this experiment, ADP-D-glycero-D- manno-heptose was used. Mono- and dilauroyl-(KDO)n-IV* were much better substrates for heptosylation than was (KDO)*-IVA (Fig. 9). After 2 h, all of the preformed, acylated (KDO)*-IV+, (lanes 1-3) had been converted to more polar

A B tSF

+ + LAUR(KDO)ilVA

t0 1 2 1 2

FIG. 8. Acid hydrolysis of [4’-32P]lipid IVA and [4’-‘*P]lau- royl-(KDO)z-IVA. Samples were treated and prepared as explained under “Experimental Procedures,” dissolved in chloroform/methanol (2:1), loaded on a thin layer plate, and developed in solvent A. The plate was exposed to x-ray film overnight at -70 “C with an enhancing screen. In this experiment, the lipid IVA was not completely hydro- lyzed.

by guest on August 12, 2019

http://ww

w.jbc.org/

Dow

nloaded from

15416 Novel Acyltransferases in E. coli

1234567

FIG. 9. Acylated (KD0)2-IVAs compared with (KDO)*-IVA as acceptors of heptose residues. A reaction mixture was prepared as described in Method B and incubated for 2 h to generate both acylation products, using 25 pM lauroyl-ACP as the acyl donor. A parallel reaction with no added lauroyl-ACP was incubated as a control. At time 0, 5-~1 portions of the first and second incubations were spotted on a thin layer plate (lanes 1 and 4, respectively). Immediately afterward, 200 pM ADP-D-glycero-D-manno-heptose (Kontrohr and Kocsis, 1986) was added to each reaction mixture, and the incubation was continued at 30 “C. At 30 min (lanes 2 and 5) and 2 h (lanes 3 and 6), 5-~1 samples of both mixtures were sampled and analyzed. The plate was developed with solvent A and autoradi- ographed overnight.

products, but relatively little (KDO)*-IV* was metabolized (lanes 4-6) under otherwise identical conditions.

DISCUSSION

Previous studies of lipid A biosynthesis in E. coli and S. typhimurium have shown that an underacylated precursor of lipid A, known as lipid IV, (Fig. l), accumulates under non- permissive conditions in mutants defective in KDO biosyn- thesis. This important observation strongly suggests that KDO transfer to lipid IVA precedes the attachment of laurate and myristate to nascent lipid A molecules (Fig. 1) in wild- type cells.

We have now discovered a novel set of E. coli enzymes, consisting minimally of a cytoplasmic lauroyl transferase and a membrane-associated myristoyl transferase, the unusual substrate specificity of which accounts for the behavior of KDO-deficient mutants. Our recent development of efficient enzymatic preparations of (KDO)?-IV* enabled us to search for these activities. In particular, the cytoplasmic lauroyl transferase that adds the first lauroyl moiety to (KDO)s-IVA has the following properties. (a) It recognizes (KDO)*-IV* but not lipid IV* (Fig. 5). (b) It is extremely specific for laurate (Table II), consistent with the fatty acid composition of lipid A. (c) It functions with lauroyl-ACP but not with lauroyl- coenzyme A (Fig. 3A), consistent with the observation that exogenous laurate is not incorporated into lipid A by intact cells (Osborn, 1979). Given the fact that two lauroyl moieties are incorporated into (KDO)*-IV, with prolonged incubation (Figs. 2-4), it is possible that the monolauroyl-(KD0)2-IVA generated in vitro (Fig. 3A) is actually a mixture of isomers. Resolution of this issue will require additional spectroscopic and structural analyses to determine the location(s) of the lauroyl group in the product. Purification of the cytosolic and membrane-bound acyltransferases will also be necessary to ascertain how many distinct enzymes are involved in (KD0)2- IVA acylation.

The acyl chain composition of E. coli and S. typhimurium lipid A is reasonably well-documented, consisting of four R- 3-hydroxymyristates, one laurate, and one myristate (Tak- ayama et al., 1983a; Raetz, 1990). Appreciable amounts of

subfractions containing four R-3-hydroxymyristates and two laurates or two myristates have not been reported. Given the properties of the (KD0)2-IVA acyltransferases that we have discovered, however, it is conceivable that dilauroyl and di- myristoyl species might yet be found, for instance, at extremes of growth temperature. The apparent lack of absolute speci- ficity in the enzymatic acylation of (KDO),-IVA might be compensated for in the cell by substrate availability or com- partmentalization.

The proposed minimal structure of lipopolysaccharide thought to be required for the growth of E. coli is shown at the bottom of Fig. 1 (Raetz, 1990). Included in this structure are the lauroyl and myristoyl substituents, although physio- logical evidence for their essentiality is weak. As yet, no mutants specifically defective in the enzymatic addition of laurate and/or myristate to (KDO)*-IV* have been isolated (Raetz, 1990). A temperature-sensitive mutant that produces a lipopolysaccharide deficient in laurate and myristate at nonpermissive temperatures has recently been reported (Leh- mann and Benninghoff, 1988). It is unclear whether the genetic defect in this mutant is directly related to the enzymes that we have discovered, or to some other “nonspecific” mem- brane function affecting fatty acid metabolism, as suggested by Lehmann and Benninghof. It will be of great interest to study the phenotypes of mutants defective in the acylation of (KDO),-IV* and to clone the relevant genes.

We previously demonstrated that the two acyltransferases that generate UDP-2,3-diacylglucosamine from UDP-GlcNAc and R-3-hydroxymyristoyl-ACP (Fig. 1) are also absolutely dependent upon the ACP moiety (Anderson and Raetz, 1987; Anderson et al., 1988). Apparently, coenzyme A derivatives have no function whatever in the biosynthesis of lipid A. This finding explains why exogenous radiolabeled R-3-hydroxy- myristate, like laurate and myristate, is not incorporated by living cells into lipid A (Harder et al., 1972; Walenga and Osborn, 1980b). Uptake of exogenous fatty acids leads exclu- sively to the formation of coenzyme A thioesters (Nunn, 1986). Consequently, the existing fatty acid auxotrophs of E. coli (Nunn, 1986; Raetz, 1986) are not likely to be of use for the modification of the fatty acyl composition of lipid A, although this issue has not been examined directly. Novel genetic and biochemical strategies will have to be devised to study the effects of lipid A modification on outer membrane biogenesis and cell growth.

An unexpected finding in the present study is the existence of a significant pool of lauroyl- and myristoyl-ACP (or some- thing that can substitute for them) in cytoplasmic extracts of E. coli and S. typhimurium (Fig. 2, lanes 2-6). We estimate that a solution of 10 mg/ml cytoplasmic protein contains approximately 10 pM lauroyl-ACP. This explains the exten- sive conversion of 25 FM (KDO)p-IVA to more hydrophobic products in the absence of added acyl donors under appropri- ate conditions (Fig. 2, lanes 2-6). In previous studies of UDP- GlcNAc acyltransferase, no endogenous pool of R-3-hydrox- ymyristoyl-ACP was detected (Anderson and Raetz, 1987). Direct measurements of the composition of the small acyl- ACP pool in E. coli have not revealed a preponderance of lauroyl- or myristoyl-ACP (Rock and Jackowski, 1982). The possibility that a novel acyl donor is involved in the acylation of (KD0)2-IV~, perhaps in addition to or in conjunction with acyl-ACP, cannot be excluded until the extracts have been fractionated.

As shown in Fig. 9, cytoplasmic extracts can convert (KDO)*-IV* to more polar products, thought to contain hep- tose (lanes 4-6). This observation is consistent with previous studies of cerulenin-treated cells, in which it was noted that

by guest on August 12, 2019

http://ww

w.jbc.org/

Dow

nloaded from

Novel Acyltransferases in E. coli 15417

laurate and myristate addition to lipid A (which is blocked preferentially in u&o by cerulenin treatment) is not required for the extension of the LPS core with heptose, glucose, and galactose (Walenga and Osborn, 198Ob). However, mono- and dilauroyl-(KDO)2-IVA are much better acceptors of heptose than is (KD0)2-IVA (Fig. 9, lanes I-3). This may explain why LPS molecules in living cells generally possess hexaacylated lipid A moieties (Raetz, 1990).

In a previous report, we described a membrane-bound pal- mitoyltransferase that catalyzes the conversion of lipid X to lipid Y (Brozek et al., 1987). The acyl donor can be any glycerophospholipid bearing palmitate at the sn-1 position, and the palmitate is transferred to the hydroxyl of the amide- linked R-3-hydroxymyristate of lipid X. Although the en- zymes described here resemble the palmitoyl transferase in their ability to generate an acyloxyacyl unit, they differ in their use of acyl-ACP as the acyl donor. The function of the palmitoyl transferase remains unknown, since very little pal- mitate is associated with E. coli lipid A under ordinary circum- stances (Raetz, 1990).

The functions of the acyloxyacyl moieties of lipid A also remain unknown. They are not required for LPS export to the outer membrane (Osborn, 1979). However, their presence in the lipid A of all Gram negatives examined so far hints at some general function. Perhaps, a physiologically important change in physical state takes place once laurate and/or myristate are added. Re-endotoxin supposedly forms long tubular micelles in water (Hayter et al., 1987), but studies in our laboratory indicate this is not the case with lipid IVA, which forms small vesicles.* The physical state of (KD0)2- IV* in aqueous solution has not been investigated.

The presence of acyloxyacyl residues enhances certain ef- fects of lipid A on animal cells, making it more toxic to mammals as compared with lipid A precursors lacking these substituents (Galanos et al., 1984). In human neutrophils and macrophages, there are enzymes that remove laurate and myristate from lipid A, presumably to detoxify it (Munford and Hall, 1986). It will be interesting to compare the catalytic specificity and three-dimensional structures of the eucaryotic deacylases that remove laurate and myristate from lipid A to the (KDO)*-IV*-specific acyltransferases of E. coli.

Acknowledgments-We thank Dr. Joanne Williamson and Dr. Matt Anderson of Merck Sharp & Dohme for preparing the acyl- ACPs, and Dr. T. Kontrohr of the University of P&s, Hungary, for providing us with partially purified ADP-D-glycero-D-mann-heptose. Mass spectrometry was performed by Dr. Richard Caprioli of the University of Texas Medical Center at Houston. NMR experiments were conducted at the National Magnetic Resonance Facility at Madison, Department of Biochemistry, University of Wisconsin.

REFERENCES

Anderson, M. S., Bulawa, C. E., and Raetz, C. R. H. (1985) J. Biol. Chem. 260, 1553615541

Anderson, M. S., and Raetz, C. R. H. (1987) J. Biol. Chem. 262, 5159-5169

Anderson, M. S., Robertson, A. D., Macher, I., and Raetz, C. R. H. (1988) Biochemistry 27, 1908-1917

*M. Hofer, R. Y. Hampton, and C. R. H. Raetz, unpublished results.

Bliah, E. G.. and Dver, J. J. (1959) Can. J. Biochem. Phvsiol. 37, 911-918 -

Boman, H. S., and Monner, D. A. (1975) J. Bacterial. 121, 455-464 Brozek, K. A., Bulawa. C. E.. and Raetz. C. R. H. (1987) J. Biol.

Cheri 262;5170-5179 Brozek, K. A., Hosaka, K., Robertson, A. D., and Raetz, C. R. H.

(1989) J. Biol. Chem. 264. 6956-6966 Christian, R., Schulz, G., Waldstatten, P., and Unger, F. M. (1984)

Tetrahedron Lett. 25,3433-3436 Cronan, J. E., Jr., and Rock, C. 0. (1987) in Escherichia coli and

Salmonella typhimurium, Vol., I (Neidhardt, F. C., ed.) pp. 474- 497, ASM Publications, Washington, D. C.

Galanos, C., Lehmann, V.. Luderitz, 0.. Rietschel. E.-Th., Westnhal. O., Brad& H., Brade, L., Freudenberg, M. A., HansenLHagge, T.; Liideritz. T.. McKenzie. G.. Schade. U.. Strittmater. W.. Tanamoto. K., Zahringer, U., Imbto; M., Ybshimura, H., Yamamoto, M.: Shimamoto, T., and Shiba, T. (1984) Eur. J. Biochem. 140, 221- 221

Galanos, C., Luderitz, O., and Westphal, 0. (1969) Eur. J. Biochem. 9, 245-249

Galloway, S. M., and Raetz, C. R. H. (1990) J. Biol. Chem. 265, 6394-6402

Green, P. R., Merrill, A. H., Jr., and Bell, R. M. (1981) J. Biol. Chem. 256,11151-11159

Harder, M. E., Beacham, I. R., Cronan, J. E., Jr., Beacham, K., Honegger, K. L., and Silbert, D. E. (1972) Proc. Natl. Acad. Sci. U. S. A. 69,3105-3109

Havekes, L., Tommassen, J., Hoekstra, W., and Lugtenberg, B. (1977) J. Bacterial. 129. l-8

Hayter, J. B., Rivera, M., and McGroarty, E. J. (1987) J. Biol. Chem. 262,5100-5105

Kontrohr, T., and Kocsis, B. (1986) J. Chromatogr. 354, 417-423 Lehmann, V., and Benninghoff, B. (1988) Eur. J. Biochem. 178, -

459-464 Miller, J. R. (1972) Experiments in Molecular Genetics, Cold Spring

Harbor Laboratory, Cold Spring Harbor, NY Munford, R. S., and Hall, C. L. (1986) Science 234, 203-205 Osborn. M. J. (1979) in Bacterial Outer Membranes (Inouve. M.. ed)

pp. 15-34, Wiley, New York “ , ? ,

Qureshi, N., Takayama, K., Heller, D., and Fenselau, C. (1983) J. Biol. Chem. 258,12947-1‘2951

Qureshi, N., Takayama, K., and Ribi, E. (1982) J. Bioi. Chem. 257, 11808-11815

Raetz, C. R. H. (1986) Annu. Reu. Genet. 20,253-295 Raetz, C. R. H. (1990) Annu. Reu. Biochem. 59, 129-170 Raetz, C. R. H., Purcell, S., Meyer, M. V., Qureshi, N., and Takayama,

K. (1985) J. Biol. Chem. 260, 16080-16088 Rick, P. D., Fung, L. W.-M., Ho, C., and Osborn, M. J. (1977) J. Biol.

Chem. 252,4904-4912 Rietschel, E. T. (ed)(l984) Handbook of Endotoxin: Chemistry of

Endotoxins, Vol. 1, p. 419, Elsevier Science Publishers, Amsterdam Rock, C. O., and Jackowski, S. (1982) J. Biol. Chem. 257, 10759-

10765 Seydel, U., Lindner, B., Wollenweber, H.-W., and Rietschel, E. T.

(1984) Eur. J. Biochem. 145.505-509 Silbert, D. F. (1975) Annu. Reu. Biochem. 44,315-339 Smith, P. K., Krohn, R. I., Hermanson. G. T.. Mallia. A. K.. Gartner.

F. H., Provenzano, M. D., Fujimoto; E. K:, Goeke, N. M., Olson; B. J., and Klenk. D. C. (1985) Anal. Biochem. 150. 76-85

Takayama, K., Qureshi, N., and Mascagni, P. (1983a) J. Biol. Chem. 258,12801-12803

Takayama, K., Qureshi, N., Mascagni. P., Anderson. L.. and Raetz. C. k. H. (1983b) J. Biol. Chem. 258; 14245-14252’

Takayama, K., Qureshi, N., Mascagni, P., Nashed, M. A., Anderson, L., and Raetz, C. R. H. (1983c) J. Biol. Chem. 258,7379-7385

Walenga, R. W., and Osborn, M. J. (1980a) J. Biol. Chem. 255,4252- 42.56

Walenga, R. W., and Osborn, M. J. (198Ob) J. Biol. Chem. 255, 4257-4263

by guest on August 12, 2019

http://ww

w.jbc.org/

Dow

nloaded from

K A Brozek and C R Raetzincorporation of laurate and myristate.

Biosynthesis of lipid A in Escherichia coli. Acyl carrier protein-dependent

1990, 265:15410-15417.J. Biol. Chem. 

  http://www.jbc.org/content/265/26/15410Access the most updated version of this article at

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/265/26/15410.full.html#ref-list-1

This article cites 0 references, 0 of which can be accessed free at

by guest on August 12, 2019

http://ww

w.jbc.org/

Dow

nloaded from