HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 21, 18281-18290, May 24, 2002

This Article Abstract Full Text (PDF) All Versions of this Article: 277/21/18281    most recent M201057200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Sweet, C. R. Articles by Raetz, C. R. H. Search for Related Content PubMed PubMed Citation Articles by Sweet, C. R. Articles by Raetz, C. R. H. Social Bookmarking                      What's this?

Relaxed Acyl Chain Specificity of Bordetella UDP-N-acetylglucosamine Acyltransferases^*

Charles R. Sweet, Andrew Preston^§, Elinor Toland^§, Suzanne M. Ramirez^¶, Robert J. Cotter^¶, Duncan J. Maskell^§, and Christian R. H. Raetz

From the  Department of Biochemistry, Duke University Medical Center, Durham, North Carolina 27710, the ^§ Centre for Veterinary Science, Department of Clinical Veterinary Medicine, University of Cambridge, Cambridge CB3 0ES, United Kingdom, and the ^¶ Middle Atlantic Mass Spectrometry Facility, Department of Pharmacology and Molecular Sciences, The Johns Hopkins University, Baltimore, Maryland 21205

Received for publication, January 31, 2002, and in revised form, March 10, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Lipid A (endotoxin) is a major structural component of Gram-negative outer membranes. It also serves as the hydrophobic anchor of lipopolysaccharide and is a potent activator of the innate immune response. Lipid A molecules from the genus Bordetella are reported to exhibit unusual structural asymmetry with respect to the acyl chains at the 3- and 3'-positions. These acyl chains are attached by UDP-N-acetylglucosamine acyltransferase (LpxA). To determine the origin of the acyl variability, the single lpxA ortholog present in each of the genomes of Bordetella bronchiseptica (lpxA_Br), Bordetella parapertussis (lpxA_Pa), and Bordetella pertussis (lpxA_Pe) was cloned and expressed in Escherichia coli. In contrast to all LpxA proteins studied to date, LpxA_Br and LpxA_Pe display relaxed acyl chain length specificity in vitro, utilizing C₁₀OH-ACP, C₁₂OH-ACP, and C₁₄OH-ACP at similar rates. Furthermore, hybrid lipid A molecules synthesized at 42 °C by an E. coli lpxA mutant complemented with lpxA_Pe contain C₁₀OH, C₁₂OH, and C₁₄OH at both the 3- and 3'-positions, as determined by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. In contrast, LpxA from B. parapertussis did not display relaxed specificity but was selective for C₁₀OH-ACP. This study provides an enzymatic explanation for some of the unusual acyl chain variations found in Bordetella lipid A.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Lipopolysaccharide (LPS)¹ is the major glycolipid molecule present on the cell surface of Gram-negative bacteria. Early understanding of LPS has come from studies in the Enterobacteriaceae (1, 2). In these organisms the molecule consists of the following three domains: the lipid A, the core oligosaccharide, and the O-antigen polysaccharide. Lipid A, the hydrophobic anchor of LPS, is generally an acylated and phosphorylated disaccharide of glucosamine (reviewed by Rick and Raetz (3)). The structure, biosynthesis, and biological properties of LPS have been extensively studied in a wide variety of systems, although the molecule is best understood in Escherichia coli (reviewed by Raetz (4)). The minimal LPS structure required for full viability of E. coli under laboratory conditions is lipid A bearing two 2-keto-3-deoxy-D-manno-octulosonic acid (Kdo) residues (Fig. 1). In addition, Kdo₂-lipid A, or Re endotoxin, is a fully active agonist of the TLR-4 receptor-mediated inflammatory response (5-7).

View larger version (31K): [in this window] [in a new window]   Fig. 1.   The biosynthesis of E. coli lipid A. Nine constitutive enzymes compose the lipid A biosynthetic pathway in E. coli. The lipid A shown is the predominant structure. However, about one-third of E. coli lipid A is modified by the replacement of the 1-phosphate with 1-pyrophosphate (not shown). LpxA, the first enzyme in the pathway, acylates the glucosamine 3-position of UDP-GlcNAc, resulting in acyl incorporation at the 3- and 3'-positions of the mature lipid A structure.

The first step in Kdo₂-lipid A biosynthesis in E. coli (Fig. 1) is the transfer of an acyl chain to the glucosamine 3-position of UDP-N-acetylglucosamine (UDP-GlcNAc). This acylation is performed by UDP-GlcNAc acyltransferase, encoded by the gene lpxA (8, 9). This acylated UDP-GlcNAc is then de-acetylated at the 2-position by LpxC (10) before another primary hydroxyacyl chain is added at this position by LpxD (11). Next UMP is removed from a portion of the UDP-2,3-diacylglucosamine pool by LpxH (12), before the tetra-acylated glucosamine disaccharide is formed by LpxB (9). Kdo₂-lipid A biosynthesis is completed through 4'-phosphorylation by LpxK (13), addition of two Kdo residues via WaaA (formerly KdtA) (14), and addition of the secondary acyl chains by the late-acyltransferases LpxL (formerly HtrB) and LpxM (formerly MsbB) (15, 16).

Much has been learned about LpxA since its discovery in 1986 (9). The crystal structure of E. coli LpxA (LpxA_Ec) has been determined to 2.6 Å resolution (17), and its substrate binding and active sites have been partially characterized (18, 19). LpxA is functional as a homotrimer and has an unusual parallel -helical secondary structure. Based on site-directed mutagenesis, it has been proposed that the active site lies in a cleft between its subunits; His-125 is proposed to serve as a general base to facilitate the direct acyl transfer from the acyl-acyl carrier protein (ACP) to UDP-GlcNAc, forming an ester linkage to the sugar (19). In most bacteria, LpxA is highly specific with regard to its acyl-ACP substrate. LpxA_Ec is 50-fold more active with its native substrate, -OH myristoyl-ACP (C₁₄OH-ACP), than with either -OH palmitoyl-ACP (C₁₆OH-ACP) (20) or -OH lauroyl-ACP (C₁₂OH-ACP) (21); it is almost 3 orders of magnitude more active with C₁₄OH-ACP than with myristoyl-ACP (C₁₄-ACP) (22) or -OH decanoyl-ACP (C₁₀OH-ACP) (21).

The genus Bordetella contains several bacterial species, some of which are respiratory tract pathogens. Bordetella pertussis and Bordetella parapertussis cause whooping cough in humans (23-27), and B. parapertussis is also found in ovine species (28-30). Bordetella bronchiseptica infects many mammals and is commonly associated with atrophic rhinitis in pigs, snuffles in rabbits, and kennel cough in dogs (31-34). B. bronchiseptica has also been occasionally described as a respiratory tract pathogen in humans (35-37). All three pathogens are very closely related in terms of multilocus enzyme electrophoresis, DNA hybridization, and DNA sequence analyses (38, 39). The mechanistic basis for their different host ranges and pathogenicities are unknown but are likely to depend on differences in surface structures between the three pathogens.

Structural studies of the LPS molecules from the bordetellae indicate that they share many common structural features (40-42); however, their lipid A moieties have organism-specific acylation patterns (Fig. 2) apart from those from Bordetella hinzii and Bordetella trematum which are identical (43). B. bronchiseptica and B. hinzii are the only bordetellae in which hexa-acylated lipid A is reported (44, 45). The lipid A structures of other Bordetella species studied to date contain fewer acyl chains and are thus likely to have reduced endotoxin activity (46). In addition to interspecies variation, B. bronchiseptica lipid A structures vary between strains, with unusual fatty acyl substitutions on the diglucosamine backbone (45). Interestingly, all reported Bordetella lipid A structures exhibit unexpected acyl chain asymmetry between the 3- and 3'-positions. The generation of this asymmetry is not currently understood but might involve either multiple LpxA isoforms or a single LpxA with relaxed acyl-ACP specificity.

View larger version (32K): [in this window] [in a new window]   Fig. 2.   Reported structures of Bordetella lipid A. The proposed structures of the predominant lipid A molecules isolated from B. bronchiseptica NRCC 4650, 4170, and 4175 (45), B. parapertussis (59), and B. pertussis (58) are shown. These structures are unusual in that they are reported to contain different acyl chains at the 3- and 3'-positions. The E. coli structure is shown for comparison. Numbers at the bottom of the structure indicate the length of the acyl chains.

In this work, we demonstrate that a single lpxA gene is present in each of the genomes of B. pertussis, B. bronchiseptica, and B. parapertussis. The presence of a single lpxA gene in each organism despite the asymmetry of the primary ester-linked acyl residues suggested that these bacteria have a single UDP-GlcNAc O-acyltransferase of relaxed substrate specificity. We have overexpressed these genes in E. coli and have assayed extracts of these constructs to examine this hypothesis. B. pertussis LpxA (LpxA_Pe) and B. bronchiseptica LpxA (LpxA_Br) transfer C₁₀OH, C₁₂OH, and C₁₄OH acyl moieties from acyl-ACP to UDP-GlcNAc. However, both enzymes utilize C₁₀OH-ACP and C₁₄OH-ACP in 2-fold preference to C₁₂OH-ACP. In contrast, B. parapertussis LpxA (LpxA_Pa) displays a strong preference for C₁₀OH-ACP over all other acyl-ACPs tested.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Materials-- All growth media were purchased from Difco, Oxoid, or Mallinckrodt Chemical Works. Bacteria bearing plasmid DNA were selected using ampicillin at 100 µg/ml. Ampicillin, acyl carrier protein, fatty acids, and (RS)--hydroxylated fatty acids, glucosamine 1-phosphate, and standard chemicals were purchased from Sigma. PerkinElmer Life Sciences was the source for radiolabeled [-³²P]UTP. Silica Gel 60 thin layer plates (0.25 mm) were purchased from Merck. T4 ligase and preparative grade phenol/chloroform/isoamyl alcohol (25:24:1, v/v) were from Invitrogen. DNA restriction endonucleases and other modifying enzymes were bought from Roche Molecular Biochemicals or New England Biolabs. The LpxC inhibitor L-573,655 was provided by Dr. A. Patchett (Merck) (47). Plasmid DNA was purified with Qiagen spin columns and DNA fragments were gel-purified using the Qiagen Qiaquick Gel Extraction purification kit.

Bacterial Strains and Plasmids-- The strains and plasmids used in this work are shown in Table I. B. pertussis BP536 is a streptomycin-resistant derivative of Tohama I (48). B. bronchiseptica RB50 has been described previously (49). B. parapertussis 12822a is a streptomycin-resistant derivative of 12822. Strain 12822 was isolated from an infant with persistent cough in Germany (a kind gift from U. Heininger). Whole genome sequencing of Tohama I, RB50, and 12822 is currently in progress (www.sanger.ac.uk). E. coli XL1-Blue (Stratagene) was used for cloning and maintenance of pUC plasmids. E. coli BL21(DE3)pLysS (Invitrogen) was used for overexpression of cloned genes. The plasmid pUC18 was used as a general cloning vector. pET21a (Stratagene) was used to overexpress lpxA genes. The E. coli conditional mutant RO138 (lpxA2 recA rpsL Tet^r) was provided by Dr. M. Anderson (Merck). RO138 is a recA derivative of SM101 (50, 51), and pBluescript II SK+ (Stratagene) was used to express the genes in R0138. Bordetella strains were grown on BG agar supplemented with 15% defibrillated horse blood and streptomycin (200 µg/ml). E. coli strains were grown on LB agar or in LB broth (pH 7.4).

PCR Cloning of LpxA Genes-- To prepare a genomic DNA template, plate-grown bacterial culture was resuspended in 0.5 ml of water, boiled in a water bath for 5 min, and centrifuged at top speed in a bench top microcentrifuge for 2 min. A 0.2-ml sample of supernatant was then recovered. PCR was performed using 1 µl of this supernatant. Each 50-µl PCR was composed of genomic DNA template, buffer as directed by the manufacturer, dNTPs (25 mM each), 20 ng of each primer, 5% (v/v) Me₂SO, 5 mM MgCl₂, and 2.5 units of polymerase. lpxA_Br was amplified using Pwo polymerase (Roche Molecular Biochemicals). lpxA_Pe and lpxA_Pa were amplified using Vent polymerase (New England Biolabs). The same primers were used for all three species as the sequences are identical in this region. The primers used were 5' TTTTTTCATATGTCAGGAAACATCCAT 3' (N terminus) and 5' TTTTTTAAGCTTTCATGGCCGGATGATG 3'. The N-terminal primer incorporated an NdeI site, and the C-terminal primer incorporated an HindIII site for subsequent cloning into pET21a. PCR was performed on a PerkinElmer Life Sciences 9600 automated cycler. A 5-min denaturing step at 94 °C was followed by 30 cycles of 94 (75 s), 55 (75 s), and 72 °C (120 s) with a final elongation step of 72 °C for 7 min. PCRs were treated with the Klenow fragment of DNA polymerase and T4 polynucleotide kinase with the addition of 1 mM ATP for 30 min at 37 °C to create blunt ends and to 5'-phosphorylate PCR products. These enzymes were then inactivated by heat denaturation at 65 °C for 30 min. PCRs were then analyzed by agarose gel electrophoresis prior to gel purification and cloning.

Purified products were cloned into pUC18 using the Ready-to-Go^TM system (Amersham Biosciences) and then subcloned into pET21a (Novagen) for protein expression. lpxA_Pe was also subcloned from pET21a into pBluescript II SK+ (Stratagene) for constitutive expression in E. coli. This pBluescript construct was transformed into RO138 for complementation and hybrid lipid A isolation. DNA was sequenced using an automated sequencer at the Department of Genetics, University of Cambridge, using universal M13 forward and reverse primers and the lpxA-specific primers: 5' GAGCAAAAGCGGTAGAAG 3'; 5' ACTGGTCATCGGCGACCG 3'; 5' GATCTTGGCGAACTGGTG 3'; and 5'GTCAATGTCGAAGGCTTG 3'. Additional DNA sequencing was performed by Duke University DNA Analysis Facility on an Applied Biosystems 377 instrument using Novagen T7 promoter and terminator primers. DNA sequence was analyzed using the GCG software package (Wisconsin Package version 10.0, Genetics Computer Group, Madison, WI).

Overexpression of Cloned Genes-- To screen for overexpressing clones, BL21 (DE3) pLysS cultures carrying the relevant pET construct were grown to A₆₀₀ = 0.4. Expression was induced by addition of isopropyl--D-thiogalactoside to a concentration of 1 mM and incubation at 37 °C for 4 h. Overexpression was confirmed by running a portion of the induced culture on 12% SDS-PAGE gels followed by Coomassie Blue staining (52). Non-induced cultures and cultures carrying pET vector alone acted as controls to demonstrate that the overexpressed proteins observed were dependent on induction and the presence of a cloned lpxA gene. To prepare cell lysates containing overexpressed Bordetella LpxA for use in assays, expression cultures were grown at 37 °C in 50 ml of LB broth containing 100 µg/ml ampicillin to an A₆₀₀ = 0.6, induced with isopropyl--D-thiogalactoside, and grown for a further 3 h. At the end of this period, the cultures were chilled on ice for 10 min and then harvested by centrifugation at 4000 × g at 4 °C for 10 min. The cell pellets were resuspended in 10 ml of ice-cold buffer (sterile 100 mM potassium phosphate at pH 7.0 with 200 mM NaCl and 20% glycerol) per 50 ml of original culture, and the cells were centrifuged again at 4000 × g at 4 °C for 10 min. Each washed cell pellet was resuspended in 2 ml of the same buffer and frozen at 20 °C overnight. The cells were then thawed on ice and broken by one passage through a French pressure cell at 14,000 pounds/square inch. The cell lysates were centrifuged at 10,000 × g for 20 min at 4 °C to remove cellular debris, and the supernatants (extracts) were collected and stored at 80 °C. Before use in the activity assays, the protein concentrations of these extracts were quantified using the Pierce bicinchoninic acid assay kit (53). The data were collected using a Molecular Devices SpectraMax 250 plate reader and analyzed relative to a bovine serum albumin standard with the software package Softmax Pro. The extracts were also characterized by SDS-PAGE (Bio-Rad mini-protean system) to assess the extent of LpxA overproduction.

Substrate Preparation and in Vitro Assay Conditions-- The [-³²P]UDP-GlcNAc and all acyl-ACP substrates were prepared as described previously (21), with the addition, as necessary, of an octyl-Sepharose column step to remove residual ACP from the acyl-ACP preparations. The purity of these acyl donors was verified by urea gel electrophoresis (data not shown). The LpxA assay follows the conversion of [-³²P]UDP-GlcNAc to [-³²P]UDP-(3-O-acyl)-GlcNAc based on a mobility shift during TLC on a silica plate (21, 54). Each 10-µl reaction contained 40 mM HEPES, pH 8.0, 1 mg/ml bovine serum albumin, 0.2 mg/ml L-573,655, 10 µM of one or more acyl donors (as indicated), 10 µM UDP-GlcNAc, [-³²P]UDP-GlcNAc (2 × 10⁵ dpm/reaction), and an appropriate amount of cell-free lysate extract. L-573,655 prevents the further metabolism of the LpxA reaction product by inhibition of LpxC (47). The assays were carried out at 30 °C, and time points were analyzed by spotting 1-µl portions of each reaction mixture onto a silica gel TLC plate. The plate was developed in chloroform/methanol/water/acetic acid (25:15:4:2, v/v) and analyzed by PhosphorImager (Molecular Dynamics Storm 840 system), using Molecular Dynamics ImageQuant 1.2 for Macintosh. Data were analyzed statistically using Microsoft Excel 2001 and plotted by Abelbeck Kaleidagraph 3.0.5.

Purification and Analysis of Hybrid Lipid A Species-- For structural studies, hybrid lipid A species were isolated and partially purified from an E. coli lpxA2 mutant complemented with LpxA_Pe. This strain, RO138/pCS561, was grown for 16 h (to stationary phase) at 42 °C in 80 ml of LB with 100 µg/ml ampicillin and 12 µg/ml tetracycline. Extraction, hydrolysis at pH 4.5, and lipid A purification by DE52 chromatography were carried out according to techniques described previously (22).

Mass Spectrometry of Hybrid Lipid A Species-- Spectra were acquired in the negative ion and positive ion linear modes by using a Kratos Analytical (Manchester, UK) matrix-assisted laser desorption/ionization time of flight (MALDI/TOF) mass spectrometer, equipped with a 337 nm nitrogen laser, a 20-kV extraction voltage, and time-delayed extraction. Each spectrum was the average of 50 shots. The matrix was a mixture of saturated 5-aza-2-thiothymine in 50% aqueous acetonitrile and 10% tribasic ammonium citrate (9:1 v/v). The lipid A samples were prepared by solubilizing the lipid in 4:1 v/v chloroform/methanol and then depositing 0.3 µl of the sample, followed by 0.3 µl of matrix, which was allowed to dry at room temperature prior to mass analysis. E. coli lipid A (Sigma) was used as an external standard.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Identification and Cloning of Bordetella LpxA Genes-- To identify Bordetella lpxA genes, genome sequence data were searched for similarities to E. coli LpxA, using tBLASTn run on the Sanger Centre blast server. This identified two open reading frames with similarity to E. coli lpxA in each of B. pertussis, B. bronchiseptica, and B. parapertussis. Further BLAST searches of GenBank^TM databases with these open reading frames revealed one to be the Bordetella lpxA gene (Fig. 3) and the other to be lpxD, the UDP-3-O-((3R)-hydroxymyristoyl)-glucosamine N-acyltransferase (data not shown). Both open reading frames were present in the same sequence contig in all three species. As in many other Gram-negative species, these genes are arranged as follows: lpxD, fabZ (encoding (R)-3-hydroxymyristoyl-ACP dehydratase), lpxA, and lpxB (55), within macromolecular biosynthesis operon II (56) of the bordetellae.

View larger version (71K): [in this window] [in a new window]   Fig. 3.   Sequences of Bordetella LpxAs and comparison to LpxAEc. This sequence pileup shows the sequence of the LpxABr, LpxAPa, and LpxAPe, as well as their homology to each other and to LpxAEc. The Bordetella protein sequences are almost identical, differing at only two residues (shown in bold). One of these residues, at position 173 in the LpxAEc sequence, has been shown to be a chain length determinant (18). This residue is altered in LpxAPa from glycine to a serine. This suggests that LpxAPa will have specificity for short acyl donors. The proposed catalytic residue, histidine 125 (19), is conserved in all species.

The predicted amino acid sequences of the three Bordetella LpxAs are 53% identical and 67% similar to that of E. coli lpxA and are nearly identical to each other. At position 170, LpxA_Pa contains serine, whereas the other two contain glycine; at position 229, LpxA_Pe contains glycine, whereas the other two contain alanine. Thus LpxA_Pe and LpxA_Br differ from one another at a single residue; LpxA_Br and LpxA_Pa differ from each other at a single residue, and LpxA_Pe and LpxA_Pa differ from each other at two residues.

The serine in position 170 of LpxA_Pa is particularly important, as it is analogous to the highly conserved residue Gly-173 in LpxA_Ec, which is involved in regulation of acyl chain length (18). In Pseudomonas aeruginosa LpxA (LpxA_Ps), this glycine is replaced by methionine, resulting in a C₁₀OH-ACP-specific enzyme. Furthermore, LpxA_Ec G173M is specific for C₁₀OH-ACP, and LpxA_Ps M169G selectively incorporates C₁₄OH-ACP (18). The presence of serine in this position in LpxA_Pa predicts that this enzyme will utilize short acyl chain substrates in preference to long ones, as there appears to be a reciprocal relationship between the size of the amino acid at this position and the length of the acyl chain substrate that can be accommodated in the active site cleft.

To investigate the substrate specificity of these enzymes, the B. bronchiseptica, B. pertussis, and B. parapertussis lpxA genes were cloned into pUC18. Three sequenced constructs were designated pAP1, pAP5, and pAP6, respectively. The genes were then subcloned from pUC18 into the expression vector pET21a, and verified clones in this vector were labeled pAP10, pAP50, and pAP60, respectively. These constructs were then transformed into E. coli BL21 DE3 (pLysS) to allow expression of the lpxA genes from the T7 promoter of pET21a. A clone that gave high level, inducible expression of a protein matching the predicted molecular weight of Bordetella LpxA was obtained for each of the three LpxA enzymes (data not shown).

To study the ability of Bordetella LpxA to complement an E. coli conditional mutant, lpxA_Pe was subcloned into pBluescript II SK+. The positive clone used herein was designated pCS561. This plasmid construct was transformed into RO138 for in vivo studies.

In Vitro Acyl Transfer Assays of the Three LpxAs-- Cell-free extracts of the BL21 cultures were prepared and assayed for the ability to acylate UDP-GlcNAc, using pools of acyl-ACPs or -OH acyl-ACPs. Each pool consisted of 10 µM each of 10-, 12-, 14-, and 16-carbon chain length acyl ACPs. All three LpxA proteins have high catalytic activity with at least one of the eight acyl donors tested (Fig. 4). Two of them, LpxA_Br and LpxA_Pe, have significant catalytic activity with three of the acyl donors. This is the first report of an LpxA capable of efficiently acylating UDP-GlcNAc with more than one acyl chain.

View larger version (66K): [in this window] [in a new window]   Fig. 4.   Broad acyl chain specificity of two Bordetella LpxA proteins. Extracts of BL21 (DE3) pLysS bearing the Bordetella lpxA genes were assayed at 0.02 mg/ml with a mixture of acyl donors at 30 °C for 8 min in a 10-µl reaction, as described under "Experimental Procedures." Conversion of [-32P]UDP-GlcNAc to [-32P]UDP-3-O-(-OH acyl)-GlcNAc was followed by differential migration on a Silica Gel 60 TLC plate, as indicated by the ladder of product spots (A-C). These three products contain -OH myristate, -OH laurate, and -OH decanoate, respectively. Reactions spotted in lanes 1-4 contained extracts of pAP10 (LpxABr), pAP50 (LpxAPe), pAP60 (LpxAPa), or pET21a, respectively, in the E. coli expression strain BL21 (DE3) pLysS. Lane 5 shows the substrate control with no extract present. These extracts displayed no activity when this assay was performed using normal (non-hydroxy) acyl-ACP donors.

To determine the acyl specificity of all three Bordetella LpxAs, a data set was gathered for each enzyme with each acyl donor (Fig. 5). All of the data reported below were derived from several independent assays over a range of extract concentrations (5-100 µg/ml). Error is expressed as standard deviation in these experiments.

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Quantification of Bordetella LpxA-specific activities with individual acyl-ACPs. Extracts of BL21 (DE3) pLysS containing pAP10, pAP50, or pAP60 were assayed with each active acyl donor. All assays were in the linear range, with respect to time and extract concentrations. Reaction rates were calculated from linear regressions of percent conversion as a function of time (pmol/min), and these rates were then used to calculate enzymatic specific activity (pmol/min/mg of extract). Specific activity, S.D., and n values are given in the text.

Acyl Selectivity of Extracts Containing Overexpressed LpxA_Br and LpxA_Pe-- The specific activity of extracts containing LpxA_Br is 1100 ± 200 (pmol/min)/mg of total cell-free extract protein with C₁₀OH-ACP (n = 6), 470 ± 67 (pmol/min)/mg with C₁₂OH-ACP (n = 6), 1000 ± 110 (pmol/min)/mg with C₁₄OH-ACP (n = 6), and 7.5 ± 0.19 (pmol/min)/mg with palmitoyl-ACP (n = 2). Likewise, the specific activity of LpxA from extracts expressing LpxA_Pe is 870 ± 150 (pmol/min)/mg with C₁₀OH-ACP (n = 6), 350 ± 45 (pmol/min)/mg with C₁₂OH-ACP (n = 6), 790 ± 86 (pmol/min)/mg with C₁₄OH-ACP (n = 5), and 5.9 ± 0.72 (pmol/min)/mg with C₁₆OH-ACP (n = 2). The endogenous LpxA_Ec background activity with C₁₄OH-ACP (20 (pmol/min)/mg) is negligible in comparison to that of extracts bearing LpxA_Br and LpxA_Pe.

Acyl Selectivity of Extracts Containing Overexpressed LpxA_Pa-- Unlike LpxA from the other two Bordetella species, LpxA_Pa only shows in vitro activity with two acyl donors. The specific activity of this enzyme extract with C₁₀OH-ACP is 590 ± 100 (pmol/min)/mg (n = 5) and with C₁₂OH-ACP it is 15 ± 3.4 (pmol/min)/mg (n = 2). The minor activity with C₁₂OH-ACP is similar to the activity of LpxA_Br, LpxA_Pe, and LpxA_Ec with C₁₆OH-ACP and is most likely too low to be physiologically relevant. LpxA_Pa extracts have no activity with C₁₄OH-ACP over endogenous E. coli activity.

None of the Bordetella LpxA proteins display any measurable activity (limit of detection <1 (pmol/min)/mg) with any non-hydroxylated (normal) acyl donor, even at high (0.5 mg/ml) extract concentrations and long (1 h) time points.

Complementation of RO138 with lpxA_Pe and Isolation of Hybrid Lipid A Species-- To study possible intracellular modulation of Bordetella LpxA activity, LpxA_Pe was used to complement the E. coli lpxA2 conditional mutant, RO138. This mutant, LpxA_Ec G189S, is temperature-sensitive for both lipid A biosynthesis and viability. RO138 complemented with pCS561 displayed wild-type colony morphology and growth rate on LB plates and in liquid LB medium at both the permissive temperature (30 °C) and the non-permissive temperature (42 °C) (data not shown).

Based on the in vitro assay results, the hybrid lipid A structures formed by the complemented mutants were expected to contain C₁₀OH, C₁₂OH, and C₁₄OH acyl chains at the 3- and/or 3'-positions. It was possible, however, that the E. coli lipid A biosynthetic apparatus might generate asymmetric structures or exclude one or more of these acyl chains. To identify the acyl chains being incorporated by LpxA_Pe, the hybrid lipid A pool was isolated from this strain when grown at 42 °C and analyzed by thin layer chromatography. Similar to wild-type lipid A (4), the lipid A profile of this strain shows two fractions that differ in charge. The predominant fraction is hexa-acyl lipid A 1-monophosphate (as shown in Figs. 1 and 2), whereas the less abundant fraction is hexa-acyl lipid A 1-pyrophosphate (not shown).

Negative Ion MALDI/TOF Mass Spectrometry of Lipid A Species in R0138/pCS561-- The negative mode spectrum of the purified lipid A 1-monophosphate pool shows five molecular ions [M  H], at m/z 1798.0, 1770.7, 1742.4, 1714.0, and 1686.4 (Fig. 6 and Table II). These peaks are attributed to wild-type E. coli lipid A (M_r 1798.4), and a series of variants showing the loss of 2, 4, 6, or 8 methylene units, respectively. These data suggest that the complemented mutant synthesizes lipid A molecules with C₁₄OH/C₁₄OH, C₁₂OH/C₁₄OH, C₁₂OH/C₁₂OH, and/or C₁₀OH/C₁₄OH, C₁₀OH/C₁₂OH, and C₁₀OH/C₁₀OH acyl chains, respectively, in the primary ester-linked (3- and 3')-positions (Fig. 7).

View larger version (14K): [in this window] [in a new window]   Fig. 6.   Negative ion MALDI/TOF mass spectrometry of hybrid lipid A species found in RO138/pCS561. The hybrid lipid A 1,4'-bis-phosphates were purified and analyzed as described under "Experimental Procedures." The expanded region shows the five primary [M  H] ions. Predicted and observed [M  H] ion masses are compared in Table II. The deduced hybrid lipid A structure is shown in Fig. 7.

View larger version (20K): [in this window] [in a new window]   Fig. 7.   Proposed structures of the hybrid lipid A 1,4'-bis-phosphates synthesized by RO138/pCS561. This hybrid lipid A mixture bears -OH myristate, -OH laurate, or -OH decanoate at the 3- and 3'-positions. This combinatorial heterogeneity of chain length leads to the five primary masses seen in the MALDI/TOF spectra, which presumably represent six hybrid lipid A species, as described further under "Results." Numbers at the bottom of the structure indicate the length of the acyl chains.

Positive Ion MALDI/TOF Mass Spectrometry of Lipid A Species in RO138/pCS561-- To characterize these lipid A structures further, positive mode MALDI/TOF spectrometry was also performed. In this experiment, the three main peak clusters seen in the spectra are interpreted as the [M + Na]⁺ series, the B1⁺ fragment series, and the B2⁺ fragment series (Fig. 8 and Table II) (57). The B1⁺ fragment arises from the distal (non-reducing) sugar moiety after cleavage of the glycosidic linkage between sugars. The B2⁺ fragment arises from loss of the 1-phosphate moiety from the anomeric carbon. The [M + Na]⁺ peaks are seen at m/z 1822.1, 1792.5, 1764.1, 1736.2, and 1707.6. The corresponding B2⁺ peaks have values of m/z 1700.0, 1671.8, 1644.3, 1616.4, and 1588.5 (Fig. 9A and Table II). Like the negative mode data, both of these sets of peaks strongly support the existence of a set of at least five molecular species, with C₁₀OH, C₁₂OH, or C₁₄OH acyl chains attached at the 3- and 3'-positions in a combinatorial manner (see above). Furthermore, the positive mode spectrum of this fraction displays three B1⁺ ions, with values of m/z 1087.4, 1060.0, and 1031.5 (Fig. 9B and Table II), corresponding to B1⁺ ions bearing C₁₀OH, C₁₂OH, or C₁₄OH acyl chains. These data confirm that all three expected acyl chain lengths are incorporated into the distal sugar, which means that these three, and only these three, acyl chains must also be present at the proximal sugar, in order to arrive at the five primary ion masses observed (Figs. 6-9).

View larger version (20K): [in this window] [in a new window]   Fig. 8.   Positive ion MALDI/TOF mass spectrometry of hybrid lipid A species found in RO138/pCS561. This spectrum displays clusters of peaks in the expected B1+, B2+, and [M + Na]+ regions. The structures of the B1+ and B2+ cleavage fragments are shown.

View larger version (19K): [in this window] [in a new window]   Fig. 9.   Positive mode MALDI/TOF mass spectrometry expansions. A, the B1+ region of the spectrum shows three B1+ ion masses, as noted. B, the B2+/[M + Na]+ region of the spectrum displays two sets of five ion masses, as noted. All of the predicted and observed masses are compared in Table II. The deduced hybrid lipid A structure is shown in Fig. 7.

MALDI/TOF analysis of the lipid A 1-pyrophosphate fraction was complicated by low sample mass and contamination by lipid A 1-monophosphate. However, these spectra (not shown) are consistent with the proposed acylation pattern (Fig. 7).

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

An interesting feature of Bordetella lipid A is that the acyl chains present at the 3- and 3'-positions are reported to differ (58) (Fig. 2). This is in contrast to most other lipid A molecules, in which the acyl chains present at these positions are identical, as they are derived from a single enzyme, LpxA. The asymmetric nature of the acyl chains at the 3- and 3'-positions of Bordetella lipid A structures suggested two hypotheses as follows: either these bacteria have two LpxA proteins with different acyl donor specificities, or a single LpxA that has reduced specificity for its acyl-ACP substrates. Either scenario is in contrast to previous studies that demonstrate other bacteria, including E. coli (20) Neisseria meningitidis (21), P. aeruginosa (18), Chlamydia trachomatis (22), and Helicobacter pylori² have a single lpxA gene, which encodes a protein of high specificity for a single acyl-ACP substrate.

DNA sequencing of the B. pertussis genome is now complete (www.sanger.ac.uk/Projects/B_pertussis/). Those of B. bronchiseptica and B. parapertussis are nearing completion. In all three genome sequences, a single lpxA gene was identified, arguing against the existence of multiple alleles. The data presented here now demonstrate that the second scenario, in which a single LpxA has relaxed substrate specificity, is indeed correct. These results, however, do not entirely support the reported lipid A structures (43).

In both individual and multiple substrate assays, B. bronchiseptica and B. pertussis LpxA have significant activity with three acyl donors, C₁₀OH-ACP, C₁₂OH-ACP, and C₁₄OH-ACP. Both enzymes show 2-fold preferences for C₁₀OH-ACP and C₁₄OH-ACP over C₁₂OH-ACP. In the case of B. pertussis, this activity correlates well with the reported structure, which contains both C₁₀OH and C₁₄OH acyl chains at glucosamine 3-positions (58). However, some additional level of acyl control must act upon LpxA in vivo in B. pertussis, as C₁₂OH is not incorporated at the 3- or 3'-position in B. pertussis lipid A.

Similarly, in B. bronchiseptica, only C₁₂OH and C₁₄OH are reported to be present in the 3- and 3'-positions of the lipid A (58). The ability of the enzyme to incorporate C₁₀OH in vitro suggests that in this organism as well there must be some additional aspect of acyl substrate control, which is not suitably re-created in the assay. For example, it is possible that in B. bronchiseptica, levels of C₁₀OH-ACP are low compared with C₁₂OH-ACP and C₁₄OH-ACP and that this excludes incorporation of C₁₀OH into lipid A. A similar pool-size effect could limit C₁₂OH-ACP usage in B. pertussis. It is unlikely that the unexpected utilization of additional acyl ACPs by these enzymes is purely an artifact of the in vitro assay design or kinetics, as the mass spectrometry data from complemented E. coli lpxA2 show that LpxA_Pe also incorporates C₁₂OH-ACP efficiently in E. coli. In both the in vitro assay and the complementation however, the Bordetella LpxAs must use E. coli ACP. Although this has not been a problem in previous studies of other LpxAs (18, 21, 22), it is possible that in this case the heterologous nature of these studies causes altered substrate presentation to LpxA, altering the substrate specificity.

There is also a discrepancy between the reported structure of B. parapertussis and the assays results. In vitro, LpxA_Pa prefers C₁₀OH-ACP 40-fold over C₁₂OH-ACP and does not utilize C₁₄OH-ACP or any non-hydroxylated (normal) acyl donors. This is in agreement with the presence of serine at the putative chain length determining position, which predicts that LpxA_Pa should only utilize short chain length acyl donors (18). It has been reported, however, that B. parapertussis contains a non-hydroxylated C₁₆ acyl chain at the 3-position (59). Furthermore, one strain of B. bronchiseptica has been reported to bear a non-hydroxylated C₁₂ acyl chain at the 3-position (58), whereas no C₁₂-ACP utilization was observed in the in vitro assay with LpxA_Br. It is possible that these differences between the structural data (Fig. 2) and enzymatic data (Fig. 5) represent real strain differences in these species; B. pertussis 536, B. bronchiseptica RB50, and B. parapertussis 12822a may have different acylation patterns from those strains for which structural data was determined (B. bronchiseptica NRCC 4170, 4175, and 4650; B. pertussis 1414, A100, and L84; and B. parapertussis ATCC 15989 and 15311). This hypothesis is currently under investigation in collaboration with Dr. M. Caroff of the Université de Paris-Sud, Orsay, France. It may also prove fruitful to examine the structure of these lipids with an independent method, such as NMR, which would provide more information about the nature and location of the acyl substituents.

The fact that at least two Bordetella LpxAs utilize more than one acyl substrate presents another conundrum, namely the asymmetric structural distribution of the varied acyl chains in the bordetellae. In B. pertussis, for example, UDP-GlcNAc is acylated with both C₁₀OH and C₁₄OH, yet all of the C₁₀OH is reported to be present on the proximal sugar of the molecule and all of the C₁₄OH on the distal sugar. This suggests that LpxH and/or LpxB are capable of discriminating between different acyl chains at the 3-position of UDP-2,3-diacylglucosamine (Fig. 1) and that these substrate specificities result in the asymmetric distribution observed in vivo. This effect has been demonstrated previously in E. coli when the lipid A biosynthesis pathway is presented with more than one UDP-2,3-diacylglucosamine (UDP-DAG) species (22). When E. coli lpxA2 was complemented with C. trachomatis lpxA, growth conditions could be manipulated such that the complemented mutant would produce 3-O-C₁₄OH-UDP-DAG and 3-O-C₁₄-UDP-DAG simultaneously. Under these conditions, the E. coli lipid A biosynthetic machinery always placed the fully hydroxylated species (25% of total UDP-DAG pool) at the distal end of the molecule, where it would then be available for the addition of the sixth acyl chain by LpxM (Fig. 1).

Unlike the hypothesized mechanism in the bordetellae, however, E. coli LpxH and/or LpxB are not capable of discriminating chain length; all three acyl chains incorporated into the hybrid lipid A of RO138/pCS561 were found in roughly equal proportions at both the 3- and 3'-positions. There is evidence, however, for asymmetric acyl distribution at the 3- and 3'-positions of the lipid A of Shigella sonnei, Porphorymonas gingivalis, and Rhodobacter capsulatis (60), which lends additional credence to the general hypothesis that downstream enzymes in the pathway are capable of generating asymmetric distribution. At this time, little is known about the substrate specificity of these enzymes in any system, or the mechanism by which they effect this control.

The biological roles and consequences of these unusual asymmetric acylation patterns of Bordetella lipid A are also unclear. The bordetellae are regarded as non-systemic pathogens, which do not usually produce fever following infection. However, Bordetella LPS displays in vitro many of the properties typical of other endotoxins that have identical acyl chains at the 3- and 3'-positions (42, 61, 62), suggesting that identical chains at these positions are not required for typical endotoxic stimulation of the inflammatory response. Studies have also demonstrated B. pertussis LPS to have less potent endotoxin activity than E. coli LPS in some assays, such as induction of interleukin-6 expression from human blood monocytes (63). Thus, it is possible that the short acyl chain present in, and/or asymmetric acylation of, B. pertussis lipid A may modulate certain activities of the LPS. However, it has been widely established that penta-acyl lipid A structures, such as the one found in B. pertussis, have greatly reduced endotoxicity and may be antagonists in some species, including humans. Therefore, it is possible that the asymmetric acyl structure has some other role in Bordetella biology.

Although there are many similarities in the pathogenesis of B. pertussis, B. bronchiseptica, and B. parapertussis, there are differences in host specificity, severity of disease, and length of carriage of the bacteria (64). It is tempting to speculate that LPS structure may contribute to these differences, but this remains to be investigated. Further study of Bordetella lipid A structure and biosynthesis may facilitate a greater understanding of the role of lipid A in the pathology of Bordetella, as well as the biology of Gram-negative bacteria in general. These studies may also illuminate the biological function of the curious acyl asymmetry of Bordetella as well as the mechanism by which it is generated. Furthermore, continued study of the substrate specificity of these LpxAs may yield better understanding of its molecular mechanism, which remains poorly characterized.

	FOOTNOTES

^* This work was supported in part by National Institutes of Health Grants GM51310 (to C. R. H. R.) and GM54882 (to R. J. C.), by National Institutes of Health Training Grant GM08558 in Biological Chemistry to Duke University (to C. R. S.), and by Wellcome Trust (UK) Program Grant 054588 (to D. J. M.).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

To whom correspondence should be addressed. Tel.: 919-684-5326; Fax: 919-684-8885; E-mail: raetz@biochem.duke.edu.

Published, JBC Papers in Press, March 11, 2002, DOI 10.1074/jbc.M201057200

² C. R. Sweet, unpublished data.

	ABBREVIATIONS

The abbreviations used are: LPS, lipopolysaccharide; ACP, acyl carrier protein; MALDI/TOF, matrix-assisted laser desorption/ionization time-of-flight mass spectrometry; Kdo, 2-keto-3-deoxy-D-manno-octulosonic acid; UDP-DAG, UDP-2,3-diacylglucosamine.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES