Advertisement

A rapid, small-scale procedure for the structural characterization of lipid A applied to Citrobacter and Bordetella strains: discovery of a new structural element

  • Alina Tirsoaga
    Footnotes
    Affiliations
    Equipe Endotoxines, Unité Mixte de Recherche 8619 du Centre National de la Recherche Scientifique, Institut de biochimie et biophysique moléculaire et cellulaire, Université de Paris-Sud, Orsay, France
    Search for articles by this author
  • Asmaa El Hamidi
    Affiliations
    Equipe Endotoxines, Unité Mixte de Recherche 8619 du Centre National de la Recherche Scientifique, Institut de biochimie et biophysique moléculaire et cellulaire, Université de Paris-Sud, Orsay, France
    Search for articles by this author
  • Malcolm B. Perry
    Affiliations
    Institute for Biological Sciences, National Research Council of Canada, Ottawa, Canada
    Search for articles by this author
  • Martine Caroff
    Affiliations
    Equipe Endotoxines, Unité Mixte de Recherche 8619 du Centre National de la Recherche Scientifique, Institut de biochimie et biophysique moléculaire et cellulaire, Université de Paris-Sud, Orsay, France
    Search for articles by this author
  • Alexey Novikov
    Correspondence
    To whom correspondence should be addressed
    Affiliations
    Equipe Endotoxines, Unité Mixte de Recherche 8619 du Centre National de la Recherche Scientifique, Institut de biochimie et biophysique moléculaire et cellulaire, Université de Paris-Sud, Orsay, France
    Search for articles by this author
  • Author Footnotes
    1 Present address of A. Tirsoaga: Department of Physical Chemistry, University of Bucharest, Bucharest, Romania.
Open AccessPublished:August 16, 2007DOI:https://doi.org/10.1194/jlr.M700193-JLR200
      Endotoxins [lipopolysaccharides (LPSs)] are part of the outer cell membrane of Gram-negative bacteria. Their biological activities are associated mainly with the lipid component (lipid A) and even more specifically with discrete aspects of their fine structure. The need for a rapid and small-scale analysis of lipid A motivated us to develop a procedure that combines direct isolation of lipids A from bacterial cells with sequential release of their ester-linked fatty acids by a mild alkali treatment followed by MALDI-MS analysis. This method avoids the multiple-step LPS extraction procedure and lipid A isolation. The whole process can be performed in a working day and applied to lyophilized bacterial samples as small as 1 mg. We illustrate the method by applying it to the analysis of lipids A of three species of Citrobacter that were found to be identical. On the other hand, when applied to two batches of Bordetella bronchiseptica strain 4650, it highlighted the presence, in one of them, of hitherto unreported hexosamine residues substituting the lipid A phosphate groups, possibly a new camouflage opportunity to escape a host defense system.

      Abbreviations

      C16
      hexadecanoic acid
      C12
      dodecanoic acid
      C14
      tetradecanoic acid
      C14-OH
      hydroxytetradecanoic acid
      GlcN
      glucosamine
      HF
      hydrofluoric acid
      LPS
      lipopolysaccharide
      Endotoxins are lipopolysaccharides (LPSs), major components of the external membrane of Gram-negative bacteria. They may cause several pathophysiological symptoms, such as fever, septic shock, and death, but they are also able to elicit beneficial activities, such as the production of tumor necrosis factor, adjuvant, and radioprotection effects (
      • Raetz C.R.
      • Whitfield C.
      Lipopolysaccharide endotoxins.
      ,
      • Caroff M.
      • Karibian D.
      Structure of bacterial lipopolysaccharides.
      ).
      LPS molecular architecture has three regions: a hydrophobic moiety, called lipid A, a core oligosaccharide, and a serospecific O-polysaccharide composed of repeating oligosaccharide units. Lipid A is embedded in the external bacterial membrane together with phospholipids and proteins. It is responsible for the major toxic and beneficial properties characteristic of bacterial endotoxins (
      • Caroff M.
      • Karibian D.
      Structure of bacterial lipopolysaccharides.
      ,
      • Alexander C.
      • Rietschel E.T.
      Bacterial lipopolysaccharides and innate immunity.
      ).
      Lipid A structure generally consists of a diglucosamine backbone substituted with varying numbers (usually four to seven) of ester- or amide-linked fatty acids. In most cases, phosphates, with and without other substituents, are linked to carbons at the C-1 and C-4′ positions of the lipid A disaccharide unit (
      • Caroff M.
      • Karibian D.
      Structure of bacterial lipopolysaccharides.
      ,
      • Alexander C.
      • Rietschel E.T.
      Bacterial lipopolysaccharides and innate immunity.
      ,
      • Trent M.S.
      Biosynthesis, transport, and modification of lipid A.
      ). These and the number and chain lengths of fatty acids are highly important for the toxic effects of lipids A (
      • Zahringer U.
      • Salvetzki R.
      • Wagner F.
      • Lindner B.
      • Ulmer A.J.
      Structural and biological characterisation of a novel tetra-acyl lipid A from Escherichia coli F515 lipopolysaccharide acting as endotoxin antagonist in human monocytes.
      ). The addition of a single fatty acid can be responsible for an increase or decrease in bacterial virulence (
      • Preston A.
      • Maxim E.
      • Toland E.
      • Pishko E.J.
      • Harvill E.T.
      • Caroff M.
      • Maskell D.J.
      Bordetella bronchiseptica PagP is a Bvg-regulated lipid A palmitoyl transferase that is required for persistent colonization of the mouse respiratory tract.
      ).
      Lipid A classical structural analysis is a rather long and complicated process that includes the following main stages: LPS extraction from the bacteria, LPS purification, LPS acid hydrolysis to split the molecule into its hydrophobic and hydrophilic moieties, and lipid A extraction. This is followed by its characterization by different methods: MS, TLC, and the identification and localization of fatty acids, phosphate groups, and other substituents, if any, on the glucosamine backbone.
      Endotoxins can be isolated from Gram-negative bacteria by a variety of different methods (
      • Caroff M.
      • Karibian D.
      Structure of bacterial lipopolysaccharides.
      ). Long and strong hydrolytic conditions, which are occasionally required to cleave the lipid A-polysaccharide bond, result in partial dephosphorylation and O-deacylation of lipid A (
      • Karibian D.
      • Deprun C.
      • Caroff M.
      Use of plasma desorption mass spectrometry in structural analysis of endotoxins: effects on lipid A of different acid treatments.
      ). Such modifications strongly diminish the biological activities of the molecule. Milder hydrolysis conditions, such as pH 4.4–4.5 in sodium acetate buffer, were shown to be efficient for lipid A liberation (
      • Rosner M.R.
      • Tang J.
      • Barzilay I.
      • Khorana H.G.
      Structure of the lipopolysaccharide from an Escherichia coli heptose-less mutant. I. Chemical degradations and identification of products.
      ) and were usually improved by the addition of SDS when the hydrolysis kinetics were too slow or ineffective (
      • Caroff M.
      • Tacken A.
      • Szabo L.
      Detergent-accelerated hydrolysis of bacterial endotoxins and determination of the anomeric configuration of the glycosyl phosphate present in the “isolated lipid A” fragment of the Bordetella pertussis endotoxin.
      ). These hydrolytic processes are the conventional first steps for lipid A analysis after LPS isolation from the bacteria. An interesting approach using SDS-promoted hydrolysis of intact bacteria has been reported (
      • Zhou Z.
      • Lin S.
      • Cotter R.J.
      • Raetz C.R.
      Lipid A modifications characteristic of Salmonella typhimurium are induced by NH4VO3 in Escherichia coli K12. Detection of 4-amino-4-deoxy-L-arabinose, phosphoethanolamine and palmitate.
      ). Other new conditions for a quick extraction of lipid A directly from bacterial cells were developed recently (
      • El Hamidi A.
      • Tirsoaga A.
      • Novikov A.
      • Hussein A.
      • Caroff M.
      Microextraction of bacterial lipid A: easy and rapid method for mass spectrometric characterization.
      ).
      We demonstrated previously that, because of steric hindrance, fatty acid ester linkages could be differentially hydrolyzed by alkaline treatment. Sequential alkaline deesterification conditions in combination with mass spectrometry can reveal the substitution positions of primary ester-linked fatty acids on the glycose residues as well as secondary ester-linked positions on the hydroxyl groups of other fatty acids, called acyloxyacyl acids. In mild conditions, the acyloxyacyl ester could be released without splitting the secondary ester linkage between the two fatty acids and characterized by GC-MS. These conditions have been used to determine the structures of many lipid A preparations, such as those of Bordetella (
      • Caroff M.
      • Brisson J.R.
      • Martin A.
      • Karibian D.
      Structure of the Bordetella pertussis 1414 endotoxin.
      ,
      • Zarrouk H.
      • Karibian D.
      • Bodie S.
      • Perry M.B.
      • Richards J.C.
      • Caroff M.
      Structural characterization of the lipids A of three Bordetella bronchiseptica strains: variability of fatty acid substitution.
      ,
      • Aussel L.
      • Brisson J.R.
      • Perry M.B.
      • Caroff M.
      Structure of the lipid A of Bordetella hinzii ATCC 51730.
      ,
      • Caroff M.
      • Aussel L.
      • Zarrouk H.
      • Martin A.
      • Richards J.C.
      • Therisod H.
      • Perry M.B.
      • Karibian D.
      Structural variability and originality of the Bordetella endotoxins.
      ), Helicobacter (
      • Therisod H.
      • Monteiro M.A.
      • Perry M.B.
      • Caroff M.
      Helicobacter mustelae lipid A structure differs from that of Helicobacter pylori.
      ), and Yersinia (
      • Aussel L.
      • Therisod H.
      • Karibian D.
      • Perry M.B.
      • Bruneteau M.
      • Caroff M.
      Novel variation of lipid A structures in strains of different Yersinia species.
      ). We also demonstrated the importance of such a sequential release in the case of Yersinia lipid A structure previously erroneously described as being identical to that of Escherichia coli. In this special case, two pairs of fatty acids substituted different glycose positions (C-2′ and C-3′), leading to the same total molecular weight but introducing different structures. Therefore, it is recommended to consider this point with any lipid A having a molecular mass similar to that of any other well-known structure. Our first demonstration of selective conditions for the release of ester-linked fatty acids in lipid A was done in a 1 day step-wise use of alkali reagents (
      • Aussel L.
      • Therisod H.
      • Karibian D.
      • Perry M.B.
      • Bruneteau M.
      • Caroff M.
      Novel variation of lipid A structures in strains of different Yersinia species.
      ): 10–15 min in 0.2 M NaOH for primary esters, and 1 h in hydrazine at 37°C for the secondary ester linkages. The two-step de-O-acylation strategy was also used by others (
      • Silipo A.
      • Lanzetta R.
      • Amoresano A.
      • Parrilli M.
      • Molinaro A.
      Ammonium hydroxide hydrolysis: a valuable support in the MALDI-TOF mass spectrometry analysis of lipid A fatty acid distribution.
      ) on lipids A isolated from LPS by SDS-promoted mild hydrolysis (
      • Caroff M.
      • Tacken A.
      • Szabo L.
      Detergent-accelerated hydrolysis of bacterial endotoxins and determination of the anomeric configuration of the glycosyl phosphate present in the “isolated lipid A” fragment of the Bordetella pertussis endotoxin.
      ). Ammonium hydroxide is frequently used in de-O-acylation, but it requires hydrolyses too long for our purpose.
      When the use of hydrazine was restricted for security reasons, we established new conditions, described here, that are also better adapted to our small lipid A samples isolated directly from bacteria.
      Citrobacter belongs to the Enterobacteriaceae group, and 11 species are known at this time. For the present study, three species were selected. They are C. freundii and C. sedlakii, two human pathogens afflicting particularly neonates, the elderly, and immunocompromised patients, and C. rodentium, a strict pathogen for mice (
      • Vinogradov E.
      • Conlan J.W.
      • Perry M.B.
      Serological cross-reaction between the lipopolysaccharide O-polysaccharaide antigens of Escherichia coli O157:H7 and strains of Citrobacter freundii and Citrobacter sedlakii.
      ,
      • Chart H.
      • Willshaw G.A.
      • Cheasty T.
      • Rowe B.
      Structure and antigenic properties of Citrobacter freundii lipopolysaccharides.
      ,
      • Kocharova N.A.
      • Knirel Y.A.
      • Kholodkova E.V.
      • Stanislavsky E.S.
      Structure of the O-specific polysaccharide chain of Citrobacter freundii O28,1c lipopolysaccharide.
      ,
      • MacLean L.L.
      • Perry M.B.
      Structural studies on the O-polysaccharide of the lipopolysaccharide produced by Citrobacter rodentium (ATCC 51459).
      ). Apart from the latter, all of the Citrobacter species are opportunistic pathogens, especially in nosocomial infections.
      The Bordetella genus contains nine species. The lipid A structures of seven of them have been described (15, our unpublished results). These structures were notable for their peculiarly high variability among species and even strains. In B. bronchiseptica, lipid A structural variability has been attributed to relaxed enzyme specificity (
      • Preston A.
      • Maxim E.
      • Toland E.
      • Pishko E.J.
      • Harvill E.T.
      • Caroff M.
      • Maskell D.J.
      Bordetella bronchiseptica PagP is a Bvg-regulated lipid A palmitoyl transferase that is required for persistent colonization of the mouse respiratory tract.
      ,
      • Zarrouk H.
      • Karibian D.
      • Bodie S.
      • Perry M.B.
      • Richards J.C.
      • Caroff M.
      Structural characterization of the lipids A of three Bordetella bronchiseptica strains: variability of fatty acid substitution.
      ,
      • Caroff M.
      • Aussel L.
      • Zarrouk H.
      • Martin A.
      • Richards J.C.
      • Therisod H.
      • Perry M.B.
      • Karibian D.
      Structural variability and originality of the Bordetella endotoxins.
      ). The lipids A of two human pathogens, B. pertussis and B. parapertussis, have been associated with their hypoacylation and short-chain fatty acids causing reduced endotoxicity (
      • Cavaillon J.M.
      • Haeffner-Cavaillon N.
      Characterization of the induction of human interleukin-1 by endotoxins.
      ,
      • Laude-Sharp M.
      • Haeffner-Cavaillon N.
      • Caroff M.
      • Lantreibecq F.
      • Pucineri C.
      • Kazatchine M.D.
      Dissociation between the interleukin 1-inducing capacity and limulus reactivity of lipopolysaccahrides from Gram-negative bacteria.
      ). Because of this high structural variability, the Bordetella strains presented a very suitable model for testing the new method.
      Here, we characterize three Citrobacter and two Bordetella lipids A by a new procedure involving direct extraction from cells. This method is especially convenient when only small amounts of bacteria, LPS, or lipid A are available. When applied to Bordetella strains as a routine test, the method led us to discover a new original lipid A structural element, increasing the LPS structural and biosynthetic originality and perhaps giving the bacteria a camouflage strategy to escape a host defense system.

      MATERIALS AND METHODS

      Bacterial strains

      Bacterial strains used (all from the National research council of Canada, Ottawa, Canada) were C. sedlakii, C. freundii (ATCC 51541), C. rodentium (
      • Vinogradov E.
      • Conlan J.W.
      • Perry M.B.
      Serological cross-reaction between the lipopolysaccharide O-polysaccharaide antigens of Escherichia coli O157:H7 and strains of Citrobacter freundii and Citrobacter sedlakii.
      ,
      • Chart H.
      • Willshaw G.A.
      • Cheasty T.
      • Rowe B.
      Structure and antigenic properties of Citrobacter freundii lipopolysaccharides.
      ,
      • Kocharova N.A.
      • Knirel Y.A.
      • Kholodkova E.V.
      • Stanislavsky E.S.
      Structure of the O-specific polysaccharide chain of Citrobacter freundii O28,1c lipopolysaccharide.
      ,
      • MacLean L.L.
      • Perry M.B.
      Structural studies on the O-polysaccharide of the lipopolysaccharide produced by Citrobacter rodentium (ATCC 51459).
      ), E. coli (strain 0119), and B. bronchiseptica (NRCC 4650).

      Bacterial growth conditions

      Cells were grown as described (
      • Zarrouk H.
      • Karibian D.
      • Bodie S.
      • Perry M.B.
      • Richards J.C.
      • Caroff M.
      Structural characterization of the lipids A of three Bordetella bronchiseptica strains: variability of fatty acid substitution.
      ,
      • Vinogradov E.
      • Conlan J.W.
      • Perry M.B.
      Serological cross-reaction between the lipopolysaccharide O-polysaccharaide antigens of Escherichia coli O157:H7 and strains of Citrobacter freundii and Citrobacter sedlakii.
      ,
      • Chart H.
      • Willshaw G.A.
      • Cheasty T.
      • Rowe B.
      Structure and antigenic properties of Citrobacter freundii lipopolysaccharides.
      ,
      • Kocharova N.A.
      • Knirel Y.A.
      • Kholodkova E.V.
      • Stanislavsky E.S.
      Structure of the O-specific polysaccharide chain of Citrobacter freundii O28,1c lipopolysaccharide.
      ,
      • MacLean L.L.
      • Perry M.B.
      Structural studies on the O-polysaccharide of the lipopolysaccharide produced by Citrobacter rodentium (ATCC 51459).
      ) Briefly, Citrobacter cells were grown to late exponential phase in a brain-heart infusion (Difco) at 37°C under constant aeration in a New Brunswick 25 liter fermenter.
      B. bronchiseptica cells (two batches) were grown in 70 liter fermenters using a 3.7% brain-heart infusion containing 5% horse serum at 37°C and 200 rpm with aeration for 18 h (
      • Zarrouk H.
      • Karibian D.
      • Bodie S.
      • Perry M.B.
      • Richards J.C.
      • Caroff M.
      Structural characterization of the lipids A of three Bordetella bronchiseptica strains: variability of fatty acid substitution.
      ). All cells were killed with phenol (1% final concentration) before harvesting.

      LPS extraction conditions

      The wet bacteria were washed with 1% saline and were extracted by stirring with 50% aqueous phenol at 65°C and collected by centrifugation for 15 min (
      • Westphal O.
      • Jann K.
      Bacterial lipopolysaccharide. Extraction with phenol-water and further application of the procedure.
      ). The cooled extract was diluted with water (2 volumes), insoluble material was removed by centrifugation, and the cleared extract was dialyzed under tap water until free from phenol. The lyophilized retentate was dissolved in 0.02 M sodium acetate (pH 7), sequentially treated with RNase, DNase, and proteinase K, and cleared by centrifugation (105,000 g, 12 h, 4°C), and the precipitated LPS gel was dissolved in water and lyophilized (
      • MacLean L.L.
      • Perry M.B.
      Structural studies on the O-polysaccharide of the lipopolysaccharide produced by Citrobacter rodentium (ATCC 51459).
      ).
      Lipid A isolation from whole cells was described in detail previously (
      • El Hamidi A.
      • Tirsoaga A.
      • Novikov A.
      • Hussein A.
      • Caroff M.
      Microextraction of bacterial lipid A: easy and rapid method for mass spectrometric characterization.
      ). Briefly, lyophilized bacterial cells (10 mg) were suspended in 400 μl of isobutyric acid-1 M ammonium hydroxide mixture (5:3, v/v) and were kept for 2 h at 100°C in a screw-cap test tube under magnetic stirring. The suspension was cooled in ice water and centrifuged (2,000 g for 15 min). The supernatant was diluted with the same volume of water and lyophilized. The lyophilized sample was then twice washed with 400 μl of methanol and centrifuged (2,000 g for 15 min). Finally, the lipid A was extracted from the pellet in 100 to 200 μl of a mixture of chloroform, methanol, and water (3:1.5:0.25, v/v/v). For 1 mg samples, 100 μl of the solvent mixtures was used at each step.

      Sequential liberation of ester-linked fatty acids by mild alkali treatment

      This method was first developed with a relatively homogeneous E. coli lipid A. The following reagents were tested to define convenient two-step liberation of ester-linked fatty acids: methylamine, dimethylamine, ethylamine, diethylamine, triethylamine, and ammonium hydroxide, at various concentrations. Different temperatures (37, 50, and 60°C) were also tested and followed by kinetics, and the products were monitored by TLC and MALDI-MS.
      The following conditions were then selected for the first-step liberation of primary ester-linked fatty acids. Lipid A (50 μg) was suspended (1 mg/ml) in 35% ammonium hydroxide and stirred for 5 h at 50°C. To liberate the secondary ester-linked fatty acids, the resulting lipid A (50 μg) was suspended in 50 μl of 41% methylamine and stirred for 5 h at 37°C. The resulting samples were dried under a stream of nitrogen, and the residues were taken up in a mixture of chloroform, methanol, and water (3:1.5:0.25, v/v/v) and followed by TLC and MALDI-MS analyses.
      Methylamine and ethylamine were similarly efficient at liberating ester-linked fatty acids. We chose to use the amine with the smallest alkyl moiety. In addition, diethylamine produced some degradation products. This was not surprising, because secondary amines are known to be more strongly basic.
      Micro quantities of lipid A isolated directly from the bacterial cells were used for deesterification. Fifty to 100 μl aliquots of the chloroform-methanol-water extracts were transferred to Eppendorf® tubes and dried with a stream of nitrogen before treatment. To define volumes of the solutions necessary for the treatment, the total mass of the isolated lipid A was estimated to be ∼1% of the initial mass of the bacterial sample.

      Hydrolysis procedures used to liberate lipid A

      Acetic acid hydrolysis

      LPS was suspended in 2% acetic acid (5 mg/ml) and kept for 2 h at 100°C under stirring. Acid was removed under vacuum, and the residue, suspended in water (5 mg/ml), was ultracentrifuged (45 min, 300,000 g, 4°C). The pellet containing lipid A was lyophilized, and lipid A was extracted with chloroform-methanol-water extraction mixture (3:1.5:0.25, v/v/v).

      SDS-promoted hydrolysis

      LPS was dispersed at a concentration of 5 mg/ml in 20 mM sodium acetate-acetic acid buffer (pH 4.5) containing 1% SDS and hydrolyzed at 100°C for 1 h. After removal of SDS with acidified ethanol, lipid A was isolated as described previously (
      • Caroff M.
      • Tacken A.
      • Szabo L.
      Detergent-accelerated hydrolysis of bacterial endotoxins and determination of the anomeric configuration of the glycosyl phosphate present in the “isolated lipid A” fragment of the Bordetella pertussis endotoxin.
      ).

      Hydrolysis procedures used for liberation of the lipid A glycosidic phosphate

      Hydrochloric acid hydrolysis

      LPS or lipid A was suspended in 0.1 M HCl at a concentration of 5 mg/ml and kept for 15 mn at 100°C under stirring (
      • Karibian D.
      • Deprun C.
      • Caroff M.
      Use of plasma desorption mass spectrometry in structural analysis of endotoxins: effects on lipid A of different acid treatments.
      ). The acid was neutralized with a 0.1 M NaOH solution and ultracentrifuged. The pellet containing the dephosphorylated lipid A was lyophilized, and the lipid A was extracted as above.

      Hydrofluoric acid treatment

      LPS or lipid A was suspended at 5 mg/ml in hydrofluoric acid (HF) and kept at 4°C under stirring for 48 h. After solvent removal under a stream of nitrogen under a hood, the residue was taken up in water and lyophilized before extraction.

      MALDI-MS

      Analyses were performed on a PerSeptive Voyager-DE STR time-of-flight mass spectrometer (Applied Biosystems) at the Institut de biochimie et biophysique moléculaire et cellulaire, Université de Paris Sud. The analysis of the small lipid A samples used was done in linear mode with delayed extraction. Both negative- and positive-ion spectra were recorded. The ion-accelerating voltage was set at 20 kV. Dihydroxybenzoic acid (Sigma Chemical Co., St. Louis, MO) was used as a matrix. A few microliters of lipid A solution (1 μg/μl) in the extraction mixture was desalted with a few grains of ion-exchange resin [Dowex 50W-X8 (H+)], either in an Eppendorf® tube or for small samples in a single surface droplet on Parafilm®. A 1 μl aliquot of the solution was deposited on the target and covered with the same volume of the matrix dissolved at 10 mg/ml in the same solvent. Different analyte-matrix ratios were tested when necessary. B. pertussis or E. coli highly purified lipids A were used as external standards for mass calibration.

      Thin-layer chromatography

      Chromatography was performed on aluminum-backed silica TLC plates (Merck), and compounds were visualized by charring at 145°C after spraying with 10% sulfuric acid in ethanol. Mixtures of isobutyric acid and 1 M ammonium hydroxide were used for migration of oligosaccharides (3:5, v/v) and LPSs (5:3, v/v) (
      • Caroff M.G.
      • Karibian D.
      Several uses for isobutyric acid-ammonium hydroxide solvent in endotoxin analysis.
      ). The solvent used for lipid A migration was a mixture of chloroform, methanol, water, and triethylamine (3:1.5:0.25: 0.1, v/v/v/v) (
      • Caroff M.
      • Tacken A.
      • Szabo L.
      Detergent-accelerated hydrolysis of bacterial endotoxins and determination of the anomeric configuration of the glycosyl phosphate present in the “isolated lipid A” fragment of the Bordetella pertussis endotoxin.
      ).

      RESULTS AND DISCUSSION

      The need for a rapid method for analyzing lipid A structures on small bacterial samples initiated our search for new lipid A isolation methods (
      • El Hamidi A.
      • Tirsoaga A.
      • Novikov A.
      • Hussein A.
      • Caroff M.
      Microextraction of bacterial lipid A: easy and rapid method for mass spectrometric characterization.
      ). In the present work, we applied selective mild alkaline treatments sequentially liberating fatty acids as well as acid-dephosphorylating treatments to micro quantities of Citrobacter and Bordetella lipids A. The latter, which were isolated by microhydrolysis of bacteria, could thus be further characterized in 1 day experiments, as shown schematically in Fig. 1 .
      Figure thumbnail gr1
      Fig. 1.Schematic representation of the analytical steps used for the analysis of lipid A isolated after hydrolysis of freeze-dried bacteria. The latter were hydrolyzed in a mixture of isobutyric acid and 1 M ammonium hydroxide (5:3) for 2 h at 100°C (
      • El Hamidi A.
      • Tirsoaga A.
      • Novikov A.
      • Hussein A.
      • Caroff M.
      Microextraction of bacterial lipid A: easy and rapid method for mass spectrometric characterization.
      ). The dotted arrow indicates alkaline treatment for 5 h at 50°C with NH4OH, leading to partial de-O-acylation, and the black arrows indicate complete de-O-acylation with methylamine for 5 h at 37°C. Liberation of acyloxyacyl fatty acids and free fatty acids is indicated.

      Full-scale analysis of Citrobacter lipids A

      Comparison of direct microhydrolysis of bacteria with conventional hydrolytic methods

      We compared the lipid A preparations obtained by direct microhydrolysis of whole Citrobacter bacterial cells with those obtained by conventional hydrolytic methods applied to phenol-extracted endotoxins of Citrobacter. Figure 2 shows negative-ion MALDI mass spectra of C. sedlakii lipid A obtained by direct microhydrolysis of bacteria (Fig. 2A), mild SDS-promoted pH 4.5 hydrolysis of the LPS, 1 h at 100°C (Fig. 2B), 2% acetic acid hydrolysis of the LPS, 2 h at 100°C (Fig. 2C), 0.1 M HCl hydrolysis of the LPS, 10 min at 100°C (Fig. 2D), and HF treatment (48 h) of lipid A isolated by direct microhydrolysis of the bacteria (Fig. 2E).
      Figure thumbnail gr2
      Fig. 2.Negative-ion MALDI mass spectra of C. sedlakii lipid A obtained by various methods. A: Hydrolysis of bacterial cells in a mixture of isobutyric acid and 1 M ammonium hydroxide (5:3, v/v) for 2 h at 100°C. B: SDS-promoted hydrolysis at pH 4.5 for 1 h at 100°C. C: Hydrolysis of lipopolysaccharide (LPS) in 2% acetic acid for 2 h at 100°C. D: Hydrolysis of LPS in 0.1 M HCl for 10 min at 100°C. E: A 48% hydrofluoric acid (HF) lipid A hydrolysis at 4°C for 48 h. −P, minus phosphate.
      Interestingly, the direct microhydrolysis performed on bacteria (Fig. 2A) shows peaks for the four main molecular species having four to seven fatty acid structures as well preserved as, if not better than, those obtained by the SDS-promoted mild hydrolysis (Fig. 2B). The quality of the spectrum is comparable with a good signal-to-noise ratio for molecular ion peaks and a negligible level of dephosphorylation. The latter point is a great advantage for good structural and biological practice. Few lipid A preparations can be obtained without dephosphorylation, especially of the phosphate in glycosidic linkage on glucosamine (GlcN) I. Although dephosphorylation has to be avoided for biological activity determination of the native lipid A molecules, complete or incomplete dephosphorylation can be useful in other biological experiments or for structural modifications (
      • Caroff M.
      • Karibian D.
      Structure of bacterial lipopolysaccharides.
      ). For example, it was early shown with B. pertussis lipid A that cleavage of the glycosidic phosphate leads to a nontoxic and nonpyrogenic lipid A molecular species (
      • Caroff M.
      • Cavaillon J.M.
      • Fitting C.
      • Haeffner-Cavaillon N.
      Inability of pyrogenic, purified Bordetella pertussis lipid A to induce interleukin-1 release by human monocytes.
      ).
      It was shown previously that α-glycosidic phosphate groups substituting N-acyl glucosamine could be distinguished from their β-anomer by their stability at pH 4.5 (
      • Caroff M.
      • Tacken A.
      • Szabo L.
      Detergent-accelerated hydrolysis of bacterial endotoxins and determination of the anomeric configuration of the glycosyl phosphate present in the “isolated lipid A” fragment of the Bordetella pertussis endotoxin.
      ). As shown by comparing synthetic α- and β-derivatives, even the mildest conditions at pH 4.5 would liberate 100% of a β-derivative (
      • Caroff M.
      • Tacken A.
      • Szabo L.
      Detergent-accelerated hydrolysis of bacterial endotoxins and determination of the anomeric configuration of the glycosyl phosphate present in the “isolated lipid A” fragment of the Bordetella pertussis endotoxin.
      ). The resistance of the glycosidic phosphate as observed in these conditions (Fig. 2B) allows us to conclude that an α-glycosidic phosphate group is present in Citrobacter lipids A, as in most lipid A structures. This α-glycosidic phosphate anomery was first demonstrated in B. pertussis (
      • Caroff M.
      • Tacken A.
      • Szabo L.
      Detergent-accelerated hydrolysis of bacterial endotoxins and determination of the anomeric configuration of the glycosyl phosphate present in the “isolated lipid A” fragment of the Bordetella pertussis endotoxin.
      ), and the procedure is useful for samples that are too small for NMR analysis.
      If some dephosphorylation has taken place during hydrolysis, the mass spectrum shows satellite peaks 80 Da lower than those of nonmodified lipid A species. Partial monodephosphorylation was observed after 2% acetic acid hydrolysis for 2 h at 100°C (Fig. 2C). The penta-acylated molecular species appears as a “doublet” because of hydroxylation of some of the tetradecanoic acid (C14). The monodephosphorylation is almost complete after 10 min at 100°C with 0.1 M HCl. (Fig. 2D). HF treatment leads to monodephosphorylation or bisdephosphorylation, depending on steric hindrance (
      • Karibian D.
      • Deprun C.
      • Caroff M.
      Use of plasma desorption mass spectrometry in structural analysis of endotoxins: effects on lipid A of different acid treatments.
      ). When position C-3′ is substituted with fatty acid(s), the HF treatment does not release the phosphate group at C-4′ but only that at C-1. This is illustrated in the spectrum shown in Fig. 2E by the presence of the monodephosphorylated molecular peaks remaining even after 48 h of HF treatment. The bisdephosphorylation would have drastically changed the ionization of the molecules, leading to the complete disappearance of the molecular ion peaks in the negative-ion mode. Therefore, the degree of dephosphorylation depends not only on the hydrolytic conditions used but also on the degree of substitution of the diglucosamine backbone. This allowed us to conclude that position C-3′ of GlcN II was substituted in the molecular species present in Citrobacter lipid A.
      The spectra obtained for the other two Citrobacter lipid A isolates were similar (data not shown) and at first glance were comparable to those of E. coli or Salmonella lipid A spectra, with the presence of molecular species having four (4FA) to seven (7FA) fatty acids. However, this comparison was not sufficient to describe their structures accurately, as we demonstrated previously for Yersinia lipid A (
      • Aussel L.
      • Therisod H.
      • Karibian D.
      • Perry M.B.
      • Bruneteau M.
      • Caroff M.
      Novel variation of lipid A structures in strains of different Yersinia species.
      ). The analyses of de-O-acylated lipids A as well as of lipid A fragmentation patterns in the positive-ion mode were required for this purpose.

      Negative-ion MALDI-MS analysis of untreated and de-O-acylated lipid A samples

      Figure 3 presents negative-ion MALDI mass spectra of the C. freundii nonmodified lipid A (Fig. 3A, A′) as well as the same lipid A after ammonium hydroxide (Fig. 3B, B′) and methylamine (Fig. 3C, C′) treatments. The spectra of the lipid A preparations obtained both by direct microhydrolysis of bacteria (Fig. 3A, B, C) and by SDS acetate pH 4.5 hydrolysis of phenol-water-extracted LPS (Fig. 3A′, B′, C′) were compared. Again, it can be appreciated that the qualities of the spectra obtained for the two preparations are comparable and the same peaks were observed in both cases.
      Figure thumbnail gr3
      Fig. 3.Negative-ion MALDI mass spectra of untreated and de-O-acylated C. freundii lipid A isolated by direct bacterial cell microhydrolysis (A, B, C) or by the SDS-promoted hydrolysis of LPS (A′, B′, C′). The spectra in A and A′ were obtained from untreated lipid A samples; the spectra in B and B′ were from lipid A samples treated with ammonium hydroxide at 50°C for 5 h; the spectra in C and C′ were from lipid A samples treated with methylamine at 37°C for 5 h. +Na, sodiated peaks.
      Negative-ion mass spectra of the nonmodified C. freundii lipid A samples gave a rough idea of the degree of molecular heterogeneity in the preparations (Fig. 3A, A′). Signals corresponding to lipid A [M-H] ions of hepta-acyl (m/z 2,036), hexa-acyl (m/z 1,797), penta-acyl (m/z 1,587 and 1,571), and tetra-acyl (m/z 1,361) molecular species were observed. It is well known that these peaks correspond to molecular species naturally present in lipid A samples and not to MS fragmentation in the negative-ion mode. They can result from different levels of biosynthetic steps and from late enzymatic modification of the structure (
      • Raetz C.R.
      • Whitfield C.
      Lipopolysaccharide endotoxins.
      ). Mass differences between the adjacent molecular ion peaks gave a first idea of the fatty acid components [i.e., hexadecanoic acid (C16) (238 units), hydroxytetradecanoic acid (C14-OH; 226 units), and C14 (210 units)]. The only difference observed relative to the E. coli lipid A spectrum was the presence of the second penta-acyl molecular species with the mass of 1,588 units. Small peaks of monodephosphorylated molecular species produced during hydrolysis were also observed at 80 units below those of the nonmodified molecular ions.
      Negative-ion mass spectra of the de-O-acylated lipid A samples (Fig. 3B, B′, C, C′) gave additional information about the fatty acid content in the lipid A molecules. After primary ester-linked fatty acid liberation by NH4OH de-O-acylation treatment, two main molecular ion peaks were observed in the negative-ion MALDI mass spectra at m/z 1,135 and 1,373. They correspond to triacylated and tetra-acylated molecules, respectively. The difference of 238 units between these peaks corresponds to the mass of a C16 fatty acid. Comparison of the de-O-acylated mass spectra with those of the nonmodified lipid A (Fig. 3A, A′) allowed us to conclude that NH4OH treatment liberated one (C14-OH) from the tetra-acyl, two (C14-OH plus C14 or two C14-OH) from the penta-acyl, and three (one C14 and two C14-OH) ester-linked fatty acids from the hexa-acyl molecular species. All of these molecular ion species translated to a main basic molecular ion species appearing at m/z 1,135 and probably corresponding to the lipid A backbone carrying a dodecanoic acid (C12) fatty acid, which had to be in secondary linkage, and two amide-linked C14-OH. The hepta-acyl molecular ion species liberated three ester-linked fatty acids (one C14 and two C14-OH) translating to the molecular species appearing at m/z 1,373, corresponding to the substitution of the second amide-linked C14-OH by a C16 fatty acid.
      As one can see, the acyloxyacyl ester linkage was not cleaved or was incompletely cleaved during the NH4OH treatment. The acyloxyacyl cleavage requires stronger or extended hydrolytic conditions, such as lipid A treatment for 5 h at 37°C with methylamine (Fig. 3C, C′). In these conditions, secondary ester-linked fatty acids were also liberated and the major molecular ion species, after liberation of C12 and C16, appeared at m/z 952 (Fig. 3C, C′). Minor peaks were observed at m/z 872 corresponding to a monodephosphorylated molecular ion and at m/z 974 corresponding to sodiated molecular ions. The major peak observed at m/z 952 could be attributed to a molecular species corresponding to the bisphosphoryl GlcN disaccharide substituted by two C14-OH. Their stability in the given conditions allows us to deduce their location on the two amide groups. To establish the distribution of the ester-linked fatty acids on the glucosamine backbone, analysis of the lipid A fragmentation patterns in the positive-ion mode gives the necessary information.

      Positive-ion mode MALDI-MS analysis of nonmodified lipid A samples

      Positive-ion MALDI mass spectra of the C. freundii nonmodified lipid A isolated by direct microhydrolysis of bacteria and by SDS acetate pH 4.5 hydrolysis of phenol-water-extracted LPS were compared (data not shown). Apart from signals corresponding to multisodiated quasimolecular ions [M+Nan-Hn-1]+ (m/z 1,821, 1,843, and 1,865), a series of fragment ions was observed. A prominent signal was detected at m/z 1,086 (
      • Karibian D.
      • Brunelle A.
      • Aussel L.
      • Caroff M.
      252Cf-plasma desorption mass spectrometry of unmodified lipid A: fragmentation patterns and localization of fatty acids.
      ) obtained only in the positive-ion mode and derived from the intact protonated GlcN II part of the lipid A molecule, with its phosphate ester and four fatty ester substituents but not the oxygen atom linkage to GlcN I. The m/z 886 signal indicated the loss of the C12 by this fragment ion in the form of neutral acid (1086-200). It could be concluded that GlcN-II bears two C14-OH, one C14, and one C12. The latter two had to be carried by the hydroxylated fatty acids at C-2′ and C-3′.

      Citrobacter lipid A molecular structures

      The totality of the data obtained allows us to propose the following structures for the lipid A molecules synthesized by the Citrobacter bacteria (Fig. 4). For tetra-acyl lipid A (1361.7 units), the bisphosphorylated diglucosamine backbone is substituted with C14OH fatty acids at the amide position of GlcN I and at the amide and C-3′ ester positions of GlcN II. The GlcN II amide-linked C14OH is substituted with a C12 fatty acid. Localization of one of the C14OH fatty acids at the C-3′ position is justified by the observed resistance to the HF treatment, caused by steric hindrance, of the phosphate group at the C-4′ position. Penta-acyl lipid A (1572.1 and 1588.1 units) has two different structures both derived from the tetra-acyl lipid A structure by the addition of a C14 in acyloxyacyl linkage at C-3′ or by the addition of a C14OH at C-3. Both of these substitutions are present in the hexa-acyl lipid A structure (1798.5 units). Hepta-acyl lipid A (2036.9 units) carries a C16 fatty acid esterifying the C14OH at the GlcN I amide position.
      Figure thumbnail gr4
      Fig. 4.Citrobacter lipid A molecular structures.

      Negative-ion MALDI-MS analysis of native and de-O-acylated lipids A of B. bronchiseptica strain 4650 batches 1 and 2

      Lipids A of two different batches of strain B. bronchiseptica 4650, grown in similar conditions, were compared by MALDI-MS. The spectrum in Fig. 5A was heterogeneous and included tetra- to hexa-acylated molecular species (at m/z 1,361, 1,587, and 1,825), as reported previously (
      • Zarrouk H.
      • Karibian D.
      • Bodie S.
      • Perry M.B.
      • Richards J.C.
      • Caroff M.
      Structural characterization of the lipids A of three Bordetella bronchiseptica strains: variability of fatty acid substitution.
      ). Some monodephosphorylated (-P) molecular species were observed. In Fig. 5B, corresponding to batch 2 lipid A, extra peaks appeared at the same values plus 161 units (m/z 1,522, 1,748, and 1,986). Other peaks have two times 161 units; they appeared at m/z 1,909 and 2,147. After methylamine treatment of the lipids A, the spectrum corresponding to batch 1 showed only one peak at m/z 952, attributed to a molecular ion species corresponding to the bisphosphoryl GlcN disaccharide substituted by two C14OH units. The corresponding batch 2 spectrum presented two other major peaks at m/z 1,114 and 1,275, in addition to the peak at m/z 952, and corresponding respectively to the addition of one and two hexosamine units esterifying the phosphate groups. As shown in Fig. 5B, peaks at (-P) carry only one hexosamine, confirming their position if necessary.
      Figure thumbnail gr5
      Fig. 5.Comparison by negative-ion MALDI-MS of two batches of B. bronchiseptica strain 4650 lipids A. A: Batch 1, untreated. A′: Batch 1, de-O-acylated. B: Batch 2, untreated. B′: Batch 2, de-O-acylated.

      Conclusion

      Lipid A structural elucidation is usually done by following a complex series of chemical and physical experiments described by different authors (
      • Raetz C.R.
      • Whitfield C.
      Lipopolysaccharide endotoxins.
      ,
      • Caroff M.
      • Karibian D.
      Structure of bacterial lipopolysaccharides.
      ,
      • Alexander C.
      • Rietschel E.T.
      Bacterial lipopolysaccharides and innate immunity.
      ). The procedure usually starts after LPS extraction and acidic hydrolysis. The method presented here allows a rapid and effective determination of the major characteristics of lipid A structure. It is thus convenient for following the reproducibility of bacterial LPS extractions and for identifying possible late enzymatic structural modifications causing variability with important effects on biological activities. PagP palmitoylation of lipid A (
      • Preston A.
      • Maxim E.
      • Toland E.
      • Pishko E.J.
      • Harvill E.T.
      • Caroff M.
      • Maskell D.J.
      Bordetella bronchiseptica PagP is a Bvg-regulated lipid A palmitoyl transferase that is required for persistent colonization of the mouse respiratory tract.
      ,
      • Guo L.
      • Lim K.B.
      • Poduje C.M.
      • Daniel M.
      • Gunn J.S.
      • Hackett M.
      • Miller S.I.
      Lipid A acylation and bacterial resistance against vertebrate antimicrobial peptides.
      ,
      • Bishop R.E.
      • Gibbons H.S.
      • Guina T.
      • Trent M.S.
      • Miller S.I.
      • Raetz C.R.
      Transfer of palmitate from phospholipids to lipid A in outer membranes of Gram-negative bacteria.
      ) is one such modification that leads to decreased signaling through TLR-4 (
      • Tanamoto K.
      • Azumi S.
      Salmonella-type heptaacylated lipid A is inactive and acts as an antagonist of lipopolysaccharide action on human line cells.
      ) and increased bacterial resistance to cationic peptides. PagL enzyme was described as a deacylase acting at the C-3 position (
      • Preston A.
      • Maxim E.
      • Toland E.
      • Pishko E.J.
      • Harvill E.T.
      • Caroff M.
      • Maskell D.J.
      Bordetella bronchiseptica PagP is a Bvg-regulated lipid A palmitoyl transferase that is required for persistent colonization of the mouse respiratory tract.
      ). The best known example of phosphate group substitution was observed in Salmonella minnesota Re 595 strain, with phosphorylethanolamine attached to the glycosidic phosphate and 4-amino-4-deoxy-l-arabinose on the phosphate at C-4′ of GlcN II. The presence of these positively charged substituents is known to increase the resistance of the bacteria to cationic antibiotics (
      • Gunn J.S.
      Bacterial modification of LPS and resistance to antimicrobial peptides.
      ,
      • Vaara M.
      Agents that increase the permeability of the outer membrane.
      ). The presence of a galactosamine at position C-1 in Francisella tularensis lipid A (
      • Phillips N.J.
      • Schilling B.
      • McLendon M.K.
      • Apicella M.A.
      • Gibson B.W.
      Novel modification of lipid A of Francisella tularensis.
      ) is another example of phosphate substitution.
      Because of antigenic similarities, it is often difficult to differentiate the Citrobacter from Salmonella or Escherichia species. However, they can be differentiated by their growth characteristics. The lipid A structures of three different Citrobacter species described here were identical and very similar to the well-known E. coli lipid A structure. The only difference from the E. coli structure was observed at the level of the penta-acyl molecular species with molecular mass 1,588 units, carrying four C14-OH and one C12, less commonly observed than the molecular species of mass 1572 units, carrying three C14-OH, one C14, and one C12 (Fig. 4). It is known that the presence of different molecular species in a lipid A preparation leads to quantitative differences in biological activities, and some molecular species have been shown to act as antagonists (
      • Zahringer U.
      • Salvetzki R.
      • Wagner F.
      • Lindner B.
      • Ulmer A.J.
      Structural and biological characterisation of a novel tetra-acyl lipid A from Escherichia coli F515 lipopolysaccharide acting as endotoxin antagonist in human monocytes.
      ,
      • Tanamoto K.
      • Azumi S.
      Salmonella-type heptaacylated lipid A is inactive and acts as an antagonist of lipopolysaccharide action on human line cells.
      ) for the other structures in the preparation (
      • Trent M.S.
      Biosynthesis, transport, and modification of lipid A.
      ). The presence of this molecular species in higher amounts could affect biological activities.
      The procedure described here was found to be particularly useful for comparing B. bronchiseptica lipids A isolated from various strains and batches available in this laboratory. The discovery of two hexosamine residues substituting for the phosphate groups in one batch of B. bronchiseptica NRCC 4650 lipid A could represent a possibility for the bacteria to escape the host defense system by neutralizing the negative charges of the phosphate group(s). Other batches tested previously did not have this hexosamine (
      • Zarrouk H.
      • Karibian D.
      • Bodie S.
      • Perry M.B.
      • Richards J.C.
      • Caroff M.
      Structural characterization of the lipids A of three Bordetella bronchiseptica strains: variability of fatty acid substitution.
      ). As the same source strain and identical growth conditions were used, we consider that the process is regulated in a late biosynthetic step in only one of the tested preparations. Further experimental evidence and genetic data will be required to explain this structural feature.
      This is the first time that hexosamine substituents were found at positions C-1 and C-4′ in a Bordetella lipid A. Other Bordetella species and strains will be examined by the procedure described above that allows rapid screening of lipids A.

      References

        • Raetz C.R.
        • Whitfield C.
        Lipopolysaccharide endotoxins.
        Annu. Rev. Biochem. 2002; 71: 635-700
        • Caroff M.
        • Karibian D.
        Structure of bacterial lipopolysaccharides.
        Carbohydr. Res. 2003; 338: 2431-2447
        • Alexander C.
        • Rietschel E.T.
        Bacterial lipopolysaccharides and innate immunity.
        J. Endotoxin Res. 2001; 7: 167-202
        • Trent M.S.
        Biosynthesis, transport, and modification of lipid A.
        Biochem. Cell Biol. 2004; 82: 71-86
        • Zahringer U.
        • Salvetzki R.
        • Wagner F.
        • Lindner B.
        • Ulmer A.J.
        Structural and biological characterisation of a novel tetra-acyl lipid A from Escherichia coli F515 lipopolysaccharide acting as endotoxin antagonist in human monocytes.
        J. Endotoxin Res. 2001; 7: 133-146
        • Preston A.
        • Maxim E.
        • Toland E.
        • Pishko E.J.
        • Harvill E.T.
        • Caroff M.
        • Maskell D.J.
        Bordetella bronchiseptica PagP is a Bvg-regulated lipid A palmitoyl transferase that is required for persistent colonization of the mouse respiratory tract.
        Mol. Microbiol. 2003; 48: 725-736
        • Karibian D.
        • Deprun C.
        • Caroff M.
        Use of plasma desorption mass spectrometry in structural analysis of endotoxins: effects on lipid A of different acid treatments.
        Prog. Clin. Biol. Res. 1995; 392: 103-111
        • Rosner M.R.
        • Tang J.
        • Barzilay I.
        • Khorana H.G.
        Structure of the lipopolysaccharide from an Escherichia coli heptose-less mutant. I. Chemical degradations and identification of products.
        J. Biol. Chem. 1979; 254: 5906-5917
        • Caroff M.
        • Tacken A.
        • Szabo L.
        Detergent-accelerated hydrolysis of bacterial endotoxins and determination of the anomeric configuration of the glycosyl phosphate present in the “isolated lipid A” fragment of the Bordetella pertussis endotoxin.
        Carbohydr. Res. 1988; 175: 273-282
        • Zhou Z.
        • Lin S.
        • Cotter R.J.
        • Raetz C.R.
        Lipid A modifications characteristic of Salmonella typhimurium are induced by NH4VO3 in Escherichia coli K12. Detection of 4-amino-4-deoxy-L-arabinose, phosphoethanolamine and palmitate.
        J. Biol. Chem. 1999; 274: 18503-18514
        • El Hamidi A.
        • Tirsoaga A.
        • Novikov A.
        • Hussein A.
        • Caroff M.
        Microextraction of bacterial lipid A: easy and rapid method for mass spectrometric characterization.
        J. Lipid Res. 2005; 46: 1773-1778
        • Caroff M.
        • Brisson J.R.
        • Martin A.
        • Karibian D.
        Structure of the Bordetella pertussis 1414 endotoxin.
        FEBS Lett. 2000; 477: 8-14
        • Zarrouk H.
        • Karibian D.
        • Bodie S.
        • Perry M.B.
        • Richards J.C.
        • Caroff M.
        Structural characterization of the lipids A of three Bordetella bronchiseptica strains: variability of fatty acid substitution.
        J. Bacteriol. 1997; 179: 3756-3760
        • Aussel L.
        • Brisson J.R.
        • Perry M.B.
        • Caroff M.
        Structure of the lipid A of Bordetella hinzii ATCC 51730.
        Rapid Commun. Mass Spectrom. 2000; 14: 595-599
        • Caroff M.
        • Aussel L.
        • Zarrouk H.
        • Martin A.
        • Richards J.C.
        • Therisod H.
        • Perry M.B.
        • Karibian D.
        Structural variability and originality of the Bordetella endotoxins.
        J. Endotoxin Res. 2001; 7: 63-68
        • Therisod H.
        • Monteiro M.A.
        • Perry M.B.
        • Caroff M.
        Helicobacter mustelae lipid A structure differs from that of Helicobacter pylori.
        FEBS Lett. 2001; 499: 1-5
        • Aussel L.
        • Therisod H.
        • Karibian D.
        • Perry M.B.
        • Bruneteau M.
        • Caroff M.
        Novel variation of lipid A structures in strains of different Yersinia species.
        FEBS Lett. 2000; 465: 87-92
        • Silipo A.
        • Lanzetta R.
        • Amoresano A.
        • Parrilli M.
        • Molinaro A.
        Ammonium hydroxide hydrolysis: a valuable support in the MALDI-TOF mass spectrometry analysis of lipid A fatty acid distribution.
        J. Lipid Res. 2002; 43: 2188-2195
        • Vinogradov E.
        • Conlan J.W.
        • Perry M.B.
        Serological cross-reaction between the lipopolysaccharide O-polysaccharaide antigens of Escherichia coli O157:H7 and strains of Citrobacter freundii and Citrobacter sedlakii.
        FEMS Microbiol. Lett. 2000; 190: 157-161
        • Chart H.
        • Willshaw G.A.
        • Cheasty T.
        • Rowe B.
        Structure and antigenic properties of Citrobacter freundii lipopolysaccharides.
        J. Appl. Bacteriol. 1993; 74: 583-587
        • Kocharova N.A.
        • Knirel Y.A.
        • Kholodkova E.V.
        • Stanislavsky E.S.
        Structure of the O-specific polysaccharide chain of Citrobacter freundii O28,1c lipopolysaccharide.
        Carbohydr. Res. 1995; 279: 327-330
        • MacLean L.L.
        • Perry M.B.
        Structural studies on the O-polysaccharide of the lipopolysaccharide produced by Citrobacter rodentium (ATCC 51459).
        Eur. J. Biochem. 2001; 268: 5740-5746
        • Cavaillon J.M.
        • Haeffner-Cavaillon N.
        Characterization of the induction of human interleukin-1 by endotoxins.
        in: Paubert-Braquet M. Lipid Mediators in the Immunology of Shock. NATO Asi Series. Plenum Press, New York1987: 395-407
        • Laude-Sharp M.
        • Haeffner-Cavaillon N.
        • Caroff M.
        • Lantreibecq F.
        • Pucineri C.
        • Kazatchine M.D.
        Dissociation between the interleukin 1-inducing capacity and limulus reactivity of lipopolysaccahrides from Gram-negative bacteria.
        Cytokine. 1990; 2: 253-258
        • Westphal O.
        • Jann K.
        Bacterial lipopolysaccharide. Extraction with phenol-water and further application of the procedure.
        Methods Carbohydr. Chem. 1965; 5: 83-91
        • Caroff M.G.
        • Karibian D.
        Several uses for isobutyric acid-ammonium hydroxide solvent in endotoxin analysis.
        Appl. Environ. Microbiol. 1990; 56: 1957-1959
        • Caroff M.
        • Cavaillon J.M.
        • Fitting C.
        • Haeffner-Cavaillon N.
        Inability of pyrogenic, purified Bordetella pertussis lipid A to induce interleukin-1 release by human monocytes.
        Infect. Immun. 1986; 54: 465-471
        • Karibian D.
        • Brunelle A.
        • Aussel L.
        • Caroff M.
        252Cf-plasma desorption mass spectrometry of unmodified lipid A: fragmentation patterns and localization of fatty acids.
        Rapid Commun. Mass Spectrom. 1999; 13: 2252-2259
        • Guo L.
        • Lim K.B.
        • Poduje C.M.
        • Daniel M.
        • Gunn J.S.
        • Hackett M.
        • Miller S.I.
        Lipid A acylation and bacterial resistance against vertebrate antimicrobial peptides.
        Cell. 1998; 95: 189-198
        • Bishop R.E.
        • Gibbons H.S.
        • Guina T.
        • Trent M.S.
        • Miller S.I.
        • Raetz C.R.
        Transfer of palmitate from phospholipids to lipid A in outer membranes of Gram-negative bacteria.
        EMBO J. 2000; 19: 5071-5080
        • Tanamoto K.
        • Azumi S.
        Salmonella-type heptaacylated lipid A is inactive and acts as an antagonist of lipopolysaccharide action on human line cells.
        J. Immunol. 2000; 164: 3149-3156
        • Gunn J.S.
        Bacterial modification of LPS and resistance to antimicrobial peptides.
        J. Endotoxin Res. 2001; 7: 57-62
        • Vaara M.
        Agents that increase the permeability of the outer membrane.
        Microbiol. Rev. 1992; 56: 395-411
        • Phillips N.J.
        • Schilling B.
        • McLendon M.K.
        • Apicella M.A.
        • Gibson B.W.
        Novel modification of lipid A of Francisella tularensis.
        Infect. Immun. 2004; 72: 5340-5348