Báo cáo khoa học: Endotoxic activity and chemical structure of lipopolysaccharides from Chlamydia trachomatis serotypes E and L2 and Chlamydophila psittaci 6BC

doi:10.1046/j.1432-1033.2003.03392.x

Eur. J. Biochem. 270, 440–450 (2003) (cid:1) FEBS 2003

Endotoxic activity and chemical structure of lipopolysaccharides from Chlamydiatrachomatisserotypes E and L2 and Chlamydophilapsittaci6BC

Holger Heine, Sven Mu¨ ller-Loennies, Lore Brade, Buko Lindner and Helmut Brade

Research Center Borstel, Center for Medicine and Biosciences, Borstel, Germany

tested for its ability to induce proinflammatory cytokines in several readout systems in comparison to LPS from C. tra- chomatis serotype L2 and Chlamydophila psittaci strain 6BC as well as enterobacterial smooth and rough LPS and syn- thetic hexaacyl lipid A. The chlamydial LPS were at least 10 times less active than typical endotoxins; specificity of the activities was confirmed by inhibition with the LPS anta- gonist, B1233, or with monoclonal antibodies against chlamydial LPS. Like other LPS, the chlamydial LPS used toll-like receptor TLR4 for signalling, but unlike other LPS activation was strictly CD14-dependent.

Keywords: toll-like receptors; innate immunity; MALDI- TOF MS; ESI-FT-IR MS.

The lipopolysaccharide (LPS) of Chlamydia trachomatis serotype E was isolated from tissue culture-grown element- ary bodies and analyzed structurally by mass spectrometry and 1H, 13C and 31P nuclear magnetic resonance. The LPS is composed of the same pentasaccharide bisphosphate aKdo-(2–8)-aKdo-(2–4)-aKdo-(2–6)-bGlcN-4P-(1–6)- aGlcN-1P (Kdo is 3-deoxy-a-D-manno-oct-2-ulosonic acid) as reported for C. trachomatis serotype L2 [Rund, S., Lind- ner, B., Brade, H. and Holst, O. (1999) J. Biol. Chem. 274, 16819–16824]. The glucosamine disaccharide backbone is substituted with a complex mixture of fatty acids with ester or amide linkage whereby no ester-linked hydroxy fatty acids were found. The LPS was purified carefully (with contaminations by protein or nucleic acids below 0.3%) and

Chlamydia are obligatory intracellular bacteria [1] causing acute and chronic infections in animals and humans [2,3]. Little is known about the pathogenic mechanisms involved infections but chronic inflammation is in chlamydial observed in classical chlamydial diseases such as ocular trachoma, that is the world’s leading cause of preventable blindness, or chronic salpingitis, that is the major cause of secondary female infertility in developed nations. Athero- sclerosis is a chronic inflammatory disease of the arterial

walls which is presently considered to be associated with infection by Chlamydophila pneumoniae [4]. Members of the family Chlamydiaceae are Gram-negative bacteria contain- ing, in their outer membrane, a lipopolysaccharide (LPS) harbouring a surface-exposed, family specific epitope that has been well characterized [5]. It is well known that LPS of Gram-negative bacteria in general is one of the most potent stimulators of innate immunity and that the lipid A moiety of LPS is responsible for this activity [6,7]. Detailed studies on the structure–function relationships of lipid A have indicated that the number, type and distribution of fatty acids in lipid A determine whether it exhibits weak or strong agonist or antagonist activities [8]. Analytical data have shown that the fatty acids in LPS of Chlamydiae have acyl chains with up to 22 carbon atoms and also that 3-hydroxy fatty acids occur only with amide-linkage [9,10]. Studies on the biological activity of chlamydial LPS have shown that it possesses significant lower activity than enterobacterial LPS in terms of pyrogenicity or Schwartzman-reactivity in rabbits, lethality in galactosamine-sensitized mice, anti- complementary activity in guinea-pigs and induction of proinflammatory cytokines in human peripheral blood monocytes [10,11]. It was, however, mitogenic for mouse B-cells and activated mouse peritoneal macrophages, thus, producing prostaglandin E2 [10]. These reported data were obtained on LPS of unknown chemical structure and of ill- defined purity. Therefore, in this study we have compared the endotoxic activities of three chlamydial LPS of known structure and defined purity. Whereas the chemical struc- tures of the LPS from C. trachomatis serotype L2 and of Chl. psittaci 6BC have been reported previously [12,13], that of the LPS from C. trachomatis serotype E, which is the

Correspondence to H. Brade, Research Center Borstel, Center for Medicine and Biosciences, D-23845 Borstel, Germany. Fax: + 49 4537 188 419, Tel.: + 49 4537 188 474, E-mail: hbrade@fz-borstel.de Abbreviations: CSD, capillary skimmer dissociation; COSY, correlation spectroscopy; EB, elementary body; ESI-FT-ICR MS, electrospray ionization Fourier-transform ion cyclotron resonance mass spectrometry; FBS, fetal bovine serum; HEK, human epithelial kidney; HMBC, heteronuclear multiple bond correlation; HPAEC, high-performance anion-exchange chromatography; HSQC, hetero- nuclear single quantum coherence; Kdo, 3-deoxy-a-D-manno-oct-2- ulopyranosonic acid; LPS, lipopolysaccharide(s); mAb, monoclonal antibody; MALDI-TOF MS, matrix assisted laser desorption/ion- ization time-of-flight mass spectrometry; MNC, mononuclear cells; NOE, nuclear Overhauser effect; NOESY, nuclear Overhauser effect spectro-scopy; TLR, Toll-like receptor; TNF, tumor necrosis factor alpha; ROESY, rotating frame nuclear Overhauser effect spectro- scopy; TOCSY, total correlation spectroscopy. Note: H. H. and S. M. L. contributed equally to this study. (Received 11 July 2002, revised 11 November 2002, accepted 26 November 2002)

Structure and function of chlamydial lipopolysaccharide (Eur. J. Biochem. 270) 441

(cid:1) FEBS 2003

Mass spectrometry

most frequently occurring serotype in genital tract infec- tions, is reported here.

Materials and methods

LPS

De-O-acylated LPS was analyzed using an Electrospray Fourier-Transform Ion Cyclotron Resonance (ESI-FT- ICR) mass spectrometer (APEX II, Bruker Daltonics, Billerica, MA, USA) equipped with a 7 Tesla actively shielded magnet and an Apollo ion source. Capillary skimmer dissociation was induced by increasing the capil- lary exit voltage from )100 V to )350 V. Samples were dissolved at a concentration of (cid:1) 20 ngÆlL)1 in 2-propanol/ water/triethylamine in a ratio of 5 : 5 : 0.01 (v/v/v) and sprayed at a flow rate of 2 lLÆmin)1.

MALDI-TOF MS of LPS and free lipid A were performed on a Reflex III (Bruker-Daltonik, Bremen, Germany) in the linear TOF configuration applying an acceleration voltage of 20 kV. Lipid A samples were dispersed in methanol/water/triethylamine in a ratio of 5 : 5 : 0.01 (v/v/v) and mixed with a saturated matrix solution (2,4,6-trihydroxyacetophenon, in 0.1% trifluoro- acetic acid and acetonitrile in a ratio of 2 : 1 v/v). Aliquots of sample solution (0.5 lL) were deposited on the metallic sample holder, dried in a stream of air and analyzed immediately after evaporation. The spectra are the sum of at least 50 single laser shot experiments. External mass scale calibration was performed with E. coli F515 LPS of known structure.

NMR spectroscopy

C. trachomatis serotype E (ATCC) was grown in monolay- er cultures of mycoplasma-free L929 cells in multilayer trays (NUNC, 10 · 600 cm2) over 2 days. The cultures were killed by the addition of 0.5% (w/v) of phenol and the chlamydial elementary bodies (EB) were sedimented to- gether with the cell debris by centrifugation at 9000 g for 1 h and lyophilized yielding 8.29 g from 120 000 cm2 of infec- ted cells. LPS was obtained by the phenol/water extraction method, modified as follows. The sediment was suspended in 200 mL of 45% phenol, containing 2% N-lauroylsarco- sine sodium salt in water (w/v), heated at 68 (cid:3)C for 10 min and cooled on ice. After centrifugation (3000 g, 30 min), the water phase was removed and the phenol phase was extracted again with 100 mL of 2% aqueous N-lauroylsar- cosine sodium. After centrifugation and separation as before, the extraction of the phenol phase was repeated again. The water layers were combined, dialyzed against water and lyophilyzed (yield 1.55 g), then extracted twice with 35-mL portions of 90% aqueous phenol/chloroform/ light petroleum (boiling point 40–60 (cid:3)C) 2 : 5 : 8 (v/v/v). The organic solvents were evaporated and the LPS was precipitated from the phenol phase by drop-wise addition of water. The precipitated LPS was washed once with 80% aqueous phenol and with acetone (yield 54.6 mg), dissolved in water (10 mgÆmL)1) and precipitated with 10 vol. of ethanol/acetone 9 : 1 (v/v), dried, and dissolved in water (5 mgÆmL)1). Calcium chloride was added to a final concentration of 0.1 M, the precipitated LPS was washed, once with 10 mM HCl, once with water, and then neutral- ized with triethylamine and freeze dried (yield 52 mg). Quantification of Kdo, glucosamine, phosphorous and fatty acids was performed as described previously [13]. Contami- nation with protein or nucleic acid was determined by amino acid analysis and gas-liquid chromatography mass spectrometry of alditol acetates, respectively.

concentration was

LPS of C. trachomatis serotype L2 [12] and Chl. psittaci 6BC [13] were those described in the reference, LPS from Salmonella enterica serovar Friedenau, E. coli Re-mutant strain F515 and synthetic lipid A (compound 506) were used as controls. De-O-acylation and de-N-acylation were carried out as described [12]. The deacylated sample was desalted by gelfiltration on Sephadex G10 in NH4HCO3 and lyophilized three times.

Analytical HPAEC

The deacylated LPS from C. trachomatis E was investigated by one-dimensional 1H-NMR- (600 MHz), 13C-NMR- (150 MHz) and 31P-NMR-spectroscopy (243 MHz) and two-dimensional 1H,1H-COSY, 1H,1H-ROESY 1H,13C- HSQC, 1H,31P-HSQC with a Bruker DRX Avance spectro- meter equipped with a multinuclear z-gradient probehead. 1H) and 1,4-dioxane Acetone in D2O (2.225 p.p.m., (67.4 p.p.m., 13C) served as reference. For 31P-NMR, 85% phosphoric acid was used as reference (set to d 0 p.p.m.) All spectra were recorded at a temperature of 300 K and standard Bruker pulse programs were used in all experiments. Spectra were recorded from a solution of 1 mg sample in 0.5 mL D2O after addition of 5 lL of a 1 M NaOD stock solution (1 : 10 dilution in D2O of 40% NaOD, Merck). The final thus 10 mM NaOD. The addition of NaOD was necessary in order to achieve uniform signals in 31P-NMR spectra and a better resolution of deoxy protons belonging to Kdo-residues. The alkaline conditions led to upfield shifts of signals of H-2 belonging to GlcN-residues. For comparison, the same spectra and an additional 1H,13C-HMBC NMR spectrum were recorded under identical conditions on a sample of aKdo-(2–8)-aKdo-(2–4)-aKdo-(2–6)-bGlcN-4P-(1–6)- aGlcN-1P (pentasaccharide bisphosphate, PSBP) which was isolated from Salmonella enterica sv. Minnesota R595- 207 and purified by HPAEC as described previously [14]. An aliquot (10 mg) of this sample was dissolved in 500 lL D2O and 4% NaOD was added until the chemical shifts of GlcN H-2 signals had the same chemical shift as for the C. trachomatis E sample. Probably due to residual buffer from the gel filtration, 30 lL of 4% NaOD had to be added.

The deacylated LPS was analysed by analytical high- performance anion exchange chromatography (HPAEC) on a column of CarboPak PA1 (4 mm · 250 mm, Dionex) using the eluents (A) 0.1 M NaOH and (B) 1 M NaOAc in 0.1 M NaOH and a gradient from 30% to 70% of eluent B in 16 min at a flow-rate of 1 mLÆmin)1. The run was monitored by pulsed amperometric detection.

442 H. Heine et al. (Eur. J. Biochem. 270)

(cid:1) FEBS 2003

Stimulation of human mononuclear cells (MNC)

Polyfect (Qiagen) according to the manufacturers’ protocol. Expression plasmids containing Flag-tagged versions of human TLR2 and human TLR4 were a kind gift from P. Nelson, Fred Hutchinson Cancer Research Center, Seattle, USA and were subcloned into pREP9 (Invitrogen). The human MD-2 expression plasmid was a kind gift from K. Miyake, Institute of Medical Science, University of Tokyo, Tokyo, Japan. All plasmids were used at 400 ng per transfection, the total DNA content was kept constant at 800 ng per transfection using pCDNA3 (Invitrogen). After 24 h of transfection, cells were washed and stimulated for another 18 h. Finally, supernatants were collected and the IL-8 content was quantified using a commercial ELISA (Biosource).

Results

Human peripheral blood MNC (Ethical Committee, University of Luebeck) were isolated from heparinized blood of healthy adult donors by Ficoll-Isopaque density gradient centrifugation [15]. Isolated MNC were washed in NaCl/Pi and cultured in RPMI 1640 containing 10% human pooled serum, 100 UÆmL)1 penicillin and 100 lgÆmL)1 strep- tomycin (Biochrom, Berlin, Germany). Cells (1 · 106ÆmL)1) were stimulated in duplicates with various amounts of LPS or the synthetic lipopeptide, Pam3Cys-Ser-Lys4 (P3CSK4; EMC microcollections, Tu¨ bingen, Germany). In blocking experiments, cells were preincubated for 30 min with 10 lgÆmL)1 of the anti-CD14 Ig MEM-18 [16] or 10 lM of the LPS antagonist, B1233. After 18 h, supernatants were collected and the content of TNF was quantified by ELISA (a kind gift of H. Gallati, Intex AG, Muttenz, Switzerland).

Isolation and characterization of chlamydial LPS

CHO/CD14 cells

The CHO/CD14 reporter cell line, clone 3E10, is a stably transfected CD14-positive CHO cell line that expresses inducible membrane CD25 (Tac antigen) under transcrip- tional control of a partial human E-selectin promoter (pELAM.Tac) which contains an essential nuclear factor, NF-jB binding site. CHO cells were grown in Ham’s F12 medium containing 10% fetal bovine serum and 1% penicillin/streptomycin at 37 (cid:3)C in a humidified 5% CO2 environment. Medium was supplemented with 400 U of hygromycin BÆmL)1.

LPS from C. trachomatis serotype E was obtained from tissue culture grown bacteria by a modified phenol/water/ detergent extraction, purified by phenol/chloroform/light petroleum-extraction and ethanol precipitation and conver- ted into the uniform triethylamine salt form. The total yield of LPS was approximately 400 ngÆcm)2 of infected cell monolayer. Compositional analyses indicated that the LPS was composed of GlcN (703 nmolÆmg)1), Kdo (981 nmolÆmg)1), phosphorus (791 nmol mg)1) and fatty acids (1800 nmolÆmg)1) in the molar ratio of 2 : 2.8 : 2.2 : 5.1. Analysis for ribose and amino acids indicated that the contamination with nucleic acids or protein was below 0.3%. Fatty acid analysis (Table 1) indicated that the major fatty acids were C14:0, iso-C15:0, C18:0, C20:0 and 3-OH-C20:0.

MALDI-MS and ESI-MS analysis

(10 lM) or mAb against

Flow cytometry analysis of NF-jB activity Cells were plated at a density of 2.5 · 105 per well in 24-well plates. The following day, the cells were stimulated for 18 h in Ham’s F12 medium containing 10% fetal bovine serum (total volume of 0.3 mL per well). Subsequently, the cells were harvested with trypsin/EDTA and expression of CD25 with FITC-labelled-antihuman CD25 was analyzed by flow cytometry (DAKO, Germany). In blocking experiments, B1233 chlamydial LPS (20 lgÆmL)1) were added 30 min prior to the stimulation with LPS or rIL-1b. The preparation of mAb against chlamydial LPS was described elsewhere [17]. Mab S25-2 and S25-23 recognize, as a minimal structure, the oligosac- charides aKdo-(2–8)-aKdo and aKdo-(2–8)-aKdo-(2–4)- aKdo, respectively, representing family specific epitopes of chlamydial LPS. They were used as immunopurified stock solutions of 1 mgÆmL)1.

Activation of transiently transfected HEK/HEK-CD14 cells

HEK293T cells were stably transfected with an expression vector for human CD14 (pCDNA3huCD14, a kind gift of D. Golenbock, Worcester, MA, USA) or with the vector without insert as control. Stable transfected cell lines (HEK- vec and HEK-CD14) were isolated and cultured in DMEM containing 10% fetal bovine serum, 1% penicillin/strepto- mycin and 400 lgÆmL)1, G418 (Invitrogen). For transient transfection, cells were plated at a density of 2 · 105ÆmL)1 in 24-well plates in complete medium without G418. The following day, cells were transiently transfected using

As a first step in the structural analysis, LPS and de-O- acylated LPS of C. trachomatis serotype E was analyzed by MALDI-TOF and ESI-FT-ICR mass spectrometry. The spectra were compared with corresponding spectra of LPS samples obtained from C. trachomatis serotype L2 and Chl. psittaci strain 6BC, the chemical structures of which are already known [12,13]. The negative ion MALDI-TOF mass spectrum of the native LPS (Fig. 1) exhibited a complex pattern of several groups of ion peaks representing molecular ions (centered around m/z 2510, 2300 and 2005) and of laser-induced fragment ions [18] representing the lipid A moiety of LPS (centered around m/z 1850, 1641, and 1347). The mass difference of D m/z 660 between the respective groups of ions of the complete LPS and lipid A indicated that the core oligosaccharide was built up by three Kdo residues. An enlargement of the spectrum of the lipid A region is depicted in Fig. 2A in comparison to mass spectra of lipid A isolated from C. trachomatis L2 and Chl. psittaci 6BC (Fig. 2B–C), showing nearly identical groups of ion peaks for C. trachomatis E and L2 that could be assigned to penta-, tetra- and tri-acylated lipid A species, differing by the mass of C14 : 0 (D m/z 210) and C20 : 0 (D m/z 294) fatty acid. The molecular ion peaks within each group differed by D m/z ± 14 and indicated the biological heterogeneity in the type of fatty acid residues in accordance

Structure and function of chlamydial lipopolysaccharide (Eur. J. Biochem. 270) 443

(cid:1) FEBS 2003

Table 1. Fatty acid analysis of LPS from C. trachomatis serotype E.

Amount present in LPS (nmolÆmg)1) Fatty acid

Nonhydroxylated

C14:0 iso-C15:0 anteiso-C15:0 anteiso-C16:0 C16:0a C18:0 iso-C19:0 anteiso-C19:0 C20:0 C20:1b iso-C21:0 anteiso-C21:0 C22:0 310 161 37 5 83 141 17 40 202 19 16 17 9

Hydroxylated C20:1c C21:1c 3-OH-C14:0 3-OH-C18:0 anteiso-3-OH-C20:0 3-OH-C20:0a iso-3-OH-C21:0 anteiso-3-OH-C21:0 3-OH-C22:0 Total 83 14 3 33 7 491 35 55 22 1800

a Iso- and anteiso-form not separated. b Unsaturated fatty acid with unknown position of double bond. c D2-Unsaturated fatty acid; assumed derivative of the corresponding 3-hydroxy fatty acids by a-elimination.

to the variety of different fatty acids as detected by chemical analysis (Table 1). In accordance to published data, the peak at m/z 1850 corresponded to a bisphosphorylated GlcN disaccharide carrying five fatty acid residues, two (3-OH)-C20:0 and one C14:0, C15:0 and C20:0 each, with a calculated molecular mass of 1850.6 Da. The mass spec- trum obtained from LPS of Chl. psittaci showed a similar complex pattern of penta-, tetra-, and tri-acylated lipid A species, however, shifted by D m/z + 42 (corresponding to three CH2 groups) indicating that the lipid A moiety carried fatty acids with longer chain length. The negative ion CSD ESI-FT-ICR mass spectrum of de-O-acylated LPS from

C. trachomatis E (Fig. 3A) revealed singly charged molecu- lar ions in the region around m/z 1779, prominent fragment ions of the de-O-acylated lipid A moiety around m/z 1119.75, and ions of minor intensity originating from the consecutive loss of three Kdo residues. Further ions observed in the spectrum originated from the loss of phosphate or from cation adduct ions. Comparison with Fig. 3B and C demonstrated that the structure of de-O-acylated LPS from C. trachomatis L2 and E were nearly identical, whereas the LPS of Chl. psittaci exhibits a second set of ions representing molecular species with four Kdo residues (m/z 2000). The enlargements of the ion region representing ions of lipid A fragments (right side) showed patterns of isotopic peaks with mass differences of D m/z ± 14 and )2 corresponding to chain length differences and to unsaturated fatty acids, respectively. The most abundant ion peak of de-O-acylated lipid A of C. trachomatis L2 and E at m/z 1119.75 represen- ted a bisphosphorylated GlcN disaccharide with two amide- linked (3-OH)-C20:0 fatty acids (calculated monoisotopic mass of 1120.75). Thus, the intense peaks at m/z 1123.75 and 1137.74 of de-O-acylated lipid A of Chl. psittaci indicated that in this species (3-OH)-C21 : 0 was the most abundant amide-linked fatty acid.

Analytical HPAEC and NMR spectroscopy

Fig. 2. Negative ion MALDI-TOF mass spectra of lipid A isolated from C. trachomatis serotype E (A) and L2 (B) and from Chl. psittaci 6BC (C).

The LPS of C. trachomatis serotype E was deacylated and the resulting phosphorylated carbohydrate backbone

Fig. 1. Negative ion MALDI-TOF mass spectrum of LPS from C. trachomatis serotype E.

444 H. Heine et al. (Eur. J. Biochem. 270)

(cid:1) FEBS 2003

Fig. 3. Negative ion CSD ESI-FT-ICR mass spectra of de-O-acylated LPS from C. trachomatis serotype E (A) and L2 (B) and from Chl. psittaci 6BC (C). On the right side the enlargements of the mass region of de-O-acylated lipid A are shown.

desalted by gelfiltration. In analytical HPAEC one major oligosaccharide as well as three minor components were observed (Fig. 4). Comparison with the HPAEC profile of a

mixture of compounds obtained from deacylation of S. enterica sv. minnesota R595-207 indicated that these were aKdo-(2–8)-aKdo-(2–4)-aKdo-(2–6)-bGlcN-4P-(1–6)- aGlcN-1P (PSBP), pentasaccharide 1-monophosphate (PS 1-MP), tetrasaccharide bisphosphate (TSBP) which lacks the terminal Kdo, and tetrasaccharide 1-monophosphate (TS 1-MP). One-dimensional 1H- and 31P-, and two-dimen- (1H,1H-DQF-COSY) and heteronuclear sional homo- (1H,13C-,1H,31P-HMQC) NMR spectroscopy (Figs 5 and 6) and assignment of all signals (Table 2) revealed that the carbohydrate structure of the major component of C. tra- chomatis E LPS consisted of three Kdo- and two GlcpN- residues. The 1H-NMR spectrum was almost identical to a spectrum of PSBP obtained from S. enterica sv. Minne- sota R595-207. All 1H-detected spectra contained three pairs of deoxy-protons (cid:1) 2 p.p.m. with a characteristic shift of pyranosidic a-Kdo-residues. The resonance frequencies and 3J1,2 coupling constants of signals belonging to the anomeric protons of GlcpN-residues revealed that one was a- and the other was b-configured. These residues

Fig. 4. Analytical HPAEC of deacylated LPS obtained from C. tra- chomatis serotype E on CarboPak PA1 (3 250 mm, Dionex) using the eluents (A) 0.1 M NaOH and (B) 1 M NaOAc in 0.1 M NaOH and a gradient from 30% to 70% of eluent B in 16 min at a flow-rate of 1 mLÆmin)1.

Structure and function of chlamydial lipopolysaccharide (Eur. J. Biochem. 270) 445

(cid:1) FEBS 2003

the C-5 carbon whereas carbon 4 resonates at lower field in comparison to Kdo which is unsubstituted at this position [19]. Because Kdo lacks an anomeric proton, the substitution pattern cannot be identified by direct interresidual ROE across the glycosidic linkage. However, protons H-8a and H-8b experience a significant shielding upon substitution with Kdo at the 8-position which then both resonate at approximately 3.6 p.p.m. vs. 3.7 and 3.9 p.p.m. when unsubstituted [19]. In accordance with published data, the substitution of the 8-position of Kdo residue D by the third Kdo (E) was also indicated by the downfield shifts of protons H-6 and H-7 of residue D [12]. Two phosphate groups were identified by 31P-NMR spectroscopy and the 31P,1H-HSQC (Fig. 6) spectrum revealed that they were located at positions A1 (31P signal at d 3.64 p.p.m) and B4 (31P signal at d 4.99 p.p.m) and thus belonged to the lipid A backbone.

Fig. 5. 1H,13C-HSQC-NMR spectrum of deac-ylated LPS from C. trachomatis serotype E in 10 mM NaOD.

In order to verify the assignment, the NMR spectra were compared with the spectra of PSBP obtained in larger amounts from S. enterica sv. Minnesota R595-207 [14] which were measured under the same conditions and unequivocally assigned by 1H,13C-HSQC, 1H,13C-HMBC, 1H,1H-COSY and 1H,1H-ROESY. The spectra were, apart from minor differences, identical to the C. trachomatis E sample.

Taken together, the chemical shift data and ROE contacts revealed that the carbohydrate structure of the major component of deacylated LPS from C. trachomatis E is aKdo-(2–8)-aKdo-(2–4)-aKdo-(2–6)-bGlcN-4P-(1–6)- aGlcN-1P and thus identical to the one found in C. tra- chomatis L2. HPAEC analysis revealed that small amounts of PS 1-MP, TSBP and TS 1-MP were present in the oligosaccharide mixture which gave rise to NMR-signals with low intensity. Due to the low relative amounts and the similarity of the compounds (overlapping signals) we have not attempted to assign their NMR chemical shifts.

Biological activity of chlamydial LPS in human MNC

We then investigated the biological activity of the chlamy- dial LPS preparations in comparison with a smooth LPS from Salmonella, a rough E. coli LPS of the Re-type and a

represented the lipid A backbone and an Overhauser enhancement from B1 to A6a indicated their 1–6-linkage. Because the signals of deoxy-protons were not well-resolved and the one-dimensional 31P-NMR spectrum did not show any signals indicating several ionic states we have performed all measurements in 10 mM NaOD. Under these conditions, 31P signals were sharp and proton signals were more dispersed. However, the signal of the anomeric carbon of the b-GlcpN was absent from the spectrum under these conditions and could not be assigned. The severe overlap of signals belonging to H-4 and H-5 of all Kdo-residues nevertheless made the unequivocal assignment of spin systems impossible by 1H,1H-COSY. However, they were identified by the ROE contact between the equatorial deoxy proton of residue C and proton 6 of the second Kdo residue (D). This ROE is characteristic of an a2–4-linked Kdo- disaccharide [19]. This substitution also leads to a b-shift of

Fig. 6. 1H,31P-HSQC-NMR spectrum of deacylated LPS from C. trachomatis serotype E in 10 mM NaOD.

446 H. Heine et al. (Eur. J. Biochem. 270)

(cid:1) FEBS 2003

Table 2. 1H- and 13C-NMR chemical shift data of aKdo-(2–8)-aKdo-(2–4)-aKdo-(2–6)-bGlcN-4P-(1–6)-aGlcN-1P (PSBP) obtained from C. tra- chomatis E (compound 1) and Salmonella enterica sv. Minnesota R595-207 (compound 2) measured in 10 mM NaOD. ND, not determined.

Chemical shift of 1H (top) and 13C (bottom) in p.p.m.

Residue Compound 1 2 3ax 3eq 4 5 6a 6b 7 8a 8b

1 Afi6-aGlcN 1P

2 5.399 94.89 5.396 94.91 2.698 55.90 2.703 55.84 3.619 73.93 3.618 73.84 – – – – 3.489 70.28 3.464 70.38 4.074 71.75 4.074 71.77 3.788 69.65 3.757 69.77 4.254 – 4.243 – – – – – – – – – – – – –

1 Bfi6-bGlcN 4P

2 4.423 nd 4.425 103.40 2.689 56.66 2.670 56.67 3.583 76.44 3.582 76.52 – – – – 3.660 73.51 3.646 73.52 3.655 74.80 3.651 74.83 3.478 63.35 3.446 63.37 3.752 – 3.755 – – – – – – – – – – – – –

1 Cfi4-aKdo

2 – nd – 175.23 – nd – 99.9 1.897 33.96 1.892 34.02 2.062 – 2.056 – 4.068 70.77 4.047 70.53 4.121 65.59 4.105 65.64 3.744 71.71 3.731 71.64 – – – – 3.904 70.27 3.904 70.41 3.724 63.56 3.702 63.70 3.889 – 3.887 –

1 Dfi8-aKdo

2 – nd – 175.93 – nd – 101.1 1.834 34.96 1.822 34.86 2.139 – 2.132 – 4.089 66.24 4.089 66.26 4.137 67.50 4.104 67.43 3.877 72.01 3.848 71.70 – – – – 4.247 70.66 4.178 71.05 3.547 63.23 3.535 63.22 3.612 – 3.608 –

E aKdo 1

activation of MNC. However, both C. trachomatis prepa- rations reached the same stimulatory activity as the other LPS preparations when higher amounts were used (100 ngÆmL)1). In contrast, Chl. psittaci LPS was about 10-fold less active than those from C. trachomatis.

synthetic E. coli lipid A. As can be seen in Fig. 7, the LPS of both C. trachomatis serotypes had almost identical biologi- cal activity in terms of stimulation of TNF release from human MNC. In comparison to the other LPS chemotypes, however, they exhibited substantially less activity with lower amounts; a minimum of 10 ngÆmL)1 was required for the

Activation of human MNC by chlamydial LPS can be inhibited by anti-CD14 Ig and LPS antagonists

The primary binding receptor for LPS is the glycosylphos- phatidylinositol (GPI)-linked CD14 molecule [20]. At low LPS concentrations, blocking of CD14 completely abro- gated stimulation of human MNC. We investigated the remaining stimulatory activity of all LPS preparations after preincubation with mAb MEM-18 against CD14 or the synthetic LPS antagonist, B1233 (Table 3); both com- pounds blocked completely the activation of cells by all LPS tested.

Activation of NF-jB in CHO/CD14 cells

2 – nd – 175.57 – nd – 100.2 1.788 34.48 1.779 34.48 2.039 – 2.032 – 4.105 66.20 4.105 66.23 4.100 66.80 4.098 66.81 3.692 71.84 3.675 71.84 – – – – 3.954 69.98 3.934 69.92 3.752 63.35 3.735 63.46 3.933 – 3.942 –

In the next set of experiments, a different readout system was used which was based on the observation that CHO cells become hypersensitive to LPS upon transfection with CD14 [21]. Such transfected cells are currently used widely to analyze LPS signalling pathways. We used CHO/CD14 reporter cells expressing the human CD25 antigen upon activation of NF-jB [22] to study the activation pathways of chlamydial LPS. Both C. trachomatis serotypes as well as Chl. psittaci were able to induce the translocation of NF-jB in this system (Fig. 8). Again, chlamydial LPS was less active than the other LPS chemotypes. In comparison to the

Fig. 7. Relative TNF inducing capacity of different LPS chemotypes in human MNC. Human MNC were stimulated with increasing concen- trations of LPS chemotypes S. enterica sv. Friedenau (solid circle), E. coli F515 (solid square), synthetic hexaacylated compound 506 (solid triangle), C. trachomatis L2 (open square), C. trachomatis E (open diamond) or Chl. psittaci (open circle) for 18 h. Subsequently, TNF content of supernatants was determined by ELISA as described in Materials and methods. The data are calculated from three different donors and the TNF value obtained by stimulation with 10 ngÆmL)1 LPS from S. enterica sv. Friedenau was set to 100%. Confidence values were below 15%.

Structure and function of chlamydial lipopolysaccharide (Eur. J. Biochem. 270) 447

(cid:1) FEBS 2003

Fig. 9. Inhibition of LPS-induced translocation of NF-jB by monoclo- nal antibodies against chlamydial LPS. CHO/CD14 cells were stably transfected with a reporter plasmid containing an NF-jB-responsive promoter driving the expression of human CD25. After preincubation with 20 lgÆmL)1 of different mAb against chlamydial-LPS for 30 min, cells were stimulated with the indicated amounts [lgÆmL)1] of C tra- chomatis E and S. enterica sv. Friedenau LPS. After 18 h, expression of CD25 was determined with PE-anti-CD25 mAb by FACS analysis. Expression of CD25 is given as geometrical fluorescence intensity (mean ± SD). One representative out of three experiments is shown.

Kdo(2–8)Kdo(2–4)Kdo trisaccharide of chlamydial LPS or its 2–8-linked disaccharide portion, respectively, were tested, and both were able to block completely the activation of CHO reporter cells by LPS (1 lgÆmL)1) of C. trachomatis serotype E, whereas the activation by LPS of S. enterica sv. Friedenau was not affected (Fig. 9).

TLR4/MD-2 and CD14 are required for the activation of cells by chlamydial LPS

stimulation of human MNC, higher LPS concentrations had to be used under these conditions. Nevertheless, in these cells the LPS antagonist, B1233, was able to completely block activation.

Inhibition of cell activation by monoclonal antibodies against chlamydial LPS

CHO/CD14 reporter cells were also used to analyze the inhibitory capacity of mAb against chlamydial LPS. Two different mAbs, S25-23 and S25-2, directed against the

Next, we investigated which receptors were involved in the activation of cells by chlamydial LPS. In the first set of experiments, HEK293 cells or stable transfected HEK/ CD14 cells were transiently transfected with either Toll-like receptor (TLR)2 or TLR4/MD-2 and stimulated with synthetic lipopeptide or LPS from S. enterica sv. Friedenau or C. trachomatis serotype E. As can be seen in Table 4, both LPS activated cells exclusively via TLR4/MD-2 only,

Fig. 8. Inhibition of LPS-induced translocation of NF-jB by the LPS antagonist, B1233. CHO/CD14 cells were stably transfected with a reporter plasmid containing a NF-jB-responsive promoter driving the expression of human CD25. After preincubation with or without 10 lM of the lipid A antagonist B1233 for 30 min, cells were stimulated with indicated amounts [lgÆmL)1] of LPS from C. trachomatis E (E), C. trachomatis L2 (L2), Chl. psittaci (6BC), E. coli F515 (Re), synthetic hexa-acylated compound 506 (506), S. enterica sv. Friedenau (SeF), or recombinant IL-1b [UÆmL)1]. After 18 h, expression of CD25 was determined with PE-anti-CD25 mAb by FACS analysis. Expression of CD25 is given as the geometrical fluorescence intensity (mean ± SD). One representative out of three experiments is shown.

Table 3. Inhibition of LPS-induced TNF release in human MNC by the lipid A antagonist B1233 and the anti-CD14 mAb MEM-18. Human MNC were preincubated either with the lipid A antagonist, B1233 (10 lM) or the anti-CD14-mAb MEM-18 (10 lgÆmL)1) and then stimulated with the indicated amounts of inducers.

Amount of TNF (pg/mL) obtained after activation in the presence of:

Inducer Concentration (ngÆmL)1) Lipid A antagonist B1233 Anti-CD14-mAb MEM-18 No inhibitor

– 100 100 100 100 10 336 ± 123 a 1914 ± 25 2553 ± 372 900 ± 166 1951 ± 276 2270 ± 106 300 ± 2 69 ± 48 235 ± 11 191 ± 146 98 ± 47 114 ± 47 159 ± 23 194 ± 19 424 ± 48 176 ± 13 197 ± 21 230 ± 25

a Data are calculated from two independent experiments.

None C. trachomatis E C. trachomatis L2 Chl. psittaci E. coli F515 Compound 506 S. enterica sv. Friedenau 10 3138 ± 51 149 ± 72 183 ± 24

448 H. Heine et al. (Eur. J. Biochem. 270)

(cid:1) FEBS 2003

Table 4. Comparison of IL-8 release of differently transfected HEK293 cells. HEK-vec or HEK-CD14 cells were transiently transfected with indicated plasmids as described in (cid:2)Materials and methods(cid:3) and stimulated with synthetic lipopeptide Pam3CysSK4, LPS from S. enterica serovar Friedenau or C. trachomatis E or 10 ngÆmL)1 recombinant TNF.

Relative IL-8 release (%)a

Inducer Concentration (ngÆmL)1) HEK-CD14 + TLR2 HEK-vec + TLR4-MD2 HEK-CD14 + TLR4-MD2 HEK-vec + TLR2

– – None Pam3CysSK4 –b 78 ± 17 92 ± 25 – – –

a Amount of IL-8 release induced by 10 ngÆmL)1 recombinant TNF was set to 100%. Data are calculated from two independent experi- ments. b Below 2%.

whereas the activation of cells by the synthetic lipopeptide Pam3CSK4 required the expression of TLR2. In addition, the results of these experiments suggested strongly that chlamydial LPS may be dependent on the expression of CD14, as C. trachomatis E LPS was unable to stimulate IL-8 release from HEK control cells transfected with TLR4/ MD-2. As expected, S. enterica serovar Friedenau LPS activated both cell lines upon transfection with TLR4/ MD-2, with higher IL-8 release in the CD14 expressing HEK cells (Table 5).

Finally, we tested whether the activity of all investigated chlamydial LPS was dependent on the presence of CD14. Surprisingly, both C. trachomatis serotypes as well as Chl. psittaci LPS showed no stimulatory capacity unless HEK cells expressing CD14 were used, even up to a concentration of 1 lgÆmL)1 (Table 5). In contrast, HEK cells could be activated by LPS of E. coli F515, synthetic lipid A and S. enterica LPS at a concentration as low as 1–10 ngÆmL)1. However, the expression of CD14 enhanced the sensitivity towards these LPS chemotypes, in particular at low concentrations.

S. enterica sv. Friedenau C. trachomatis E 100 1000 100 1000 100 1000 – – – – 84 ± 21 104 ± 21 – – – – 44 ± 14 29 ± 5 3 ± 2 3 ± 2 – – – 78 ± 8 72 ± 9 35 ± 15 32 ± 7

Discussion

Members of the Gram-negative bacterial family Chlamydi- aceae cause diseases in man such as ocular trachoma and infections of the genitourinary tract of men and women. It is characteristic of all chlamydial infections that they can persist for many years without clinical symptoms but active chronic inflammation is also observed where it is unclear whether living bacteria are essential for the inflammation or whether bacterial components can sustain the inflammatory process. It is definitely known from ocular trachoma that no living bacteria are found in the late stage of the disease that is characterized by pannus formation and vascularization of the cornea. There is good evidence that chlamydial heat shock proteins are involved in the immunopathology of trachoma [23,24]. The clinical course and the histopathology observed in occluding salpingitis is very similar to that of trachoma and there is an ongoing discussion on the putative role of Chl. pneumoniae in the development of atheroscler- osis which is also a chronic inflammation of the arterial walls.

One of the most potent bacterial inducers of proinflam- matory host responses is LPS that is known to be a constituent of the chlamydial cell wall [25]. It was first recognized as the major surface antigen of all chlamydiae and is still an important diagnostic marker. The formerly called genus-specific epitope (according to taxonomical changes now called family-specific epitope) is a Kdo trisaccharide of the sequence Kdo(2–8)Kdo(2–4)Kdo which represents the saccharide portion of chlamydial LPS [26]. The epitope and the antibodies directed against it have been characterized; this subject has been reviewed recently [5]. Although we have a detailed understanding of how endotoxins act on immune cells in general, little is known about the endotoxic activity of chlamydial LPS. The main reason for this is the fact that the extensive purification of LPS and its analytical control require relatively large quantities. One should keep in mind that even minor contamination with proteins, lipoproteins or nucleic acids may be relevant in the sensitive assays available today. Nevertheless, there is good evidence from our group [10] and others [11] that chlamydial LPS is less active than typical enterobacterial LPS in classical endotoxin assays. When we decided to investigate the biological activity of chlamydial LPS in more detail we wanted to base this study on preparations whose structures had been characterized fully. We have published the structures of C. trachomatis serotype L2 [12] and of Chl. psittaci 6BC [13] earlier as these chlamydial strains are fast growing. As the distribution of serotype E is worldwide and the most frequently occurring serotype in genital C. trachomatis infections we decided to include this strain in our study and, therefore, first had to determine its structure. We were able to isolate more than 50 mg of LPS from this serotype, a quantity which allowed extensive purification and structural analyses. Using 500 lg of LPS each, we determined that contamination by protein or nucleic acid was below 0.3%. Similarly purified LPS from E. coli Re-mutants, where up to 20 mg can be used to measure putative contaminants, have shown that these are below 0.02%. As seen from the NMR and MS data, the structure of LPS from serotype E is very similar to that of serotype L2 with two GlcN and three Kdo residues, two phosphates and a heterogeneous acylation pattern. The

Structure and function of chlamydial lipopolysaccharide (Eur. J. Biochem. 270) 449

(cid:1) FEBS 2003

Table 5. Comparison of LPS-induced IL-8 release of HEK-vec and HEK-CD14 cells. HEK-vec or HEK-CD14 cells were transiently transfected with indicated plasmids as described in (cid:2)Materials and methods(cid:3) and stimulated with different amounts of indicated LPS chemotypes or 10 ngÆmL)1 recombinant TNF.

Relative IL-8 release (%)a

HEK-vec HEK-CD14 Conc (ngÆmL)1) Inducer

Control C. trachomatis E

– 0.1 1 10 100 1000 –b – – – 3 ± 2 3 ± 2 – – – 14 ± 3 41 ± 9 39 ± 1

than smooth LPS by a factor of 100 whereas, LPS from Chl. psittaci was of lower activity than the two other chlamydial LPS. Nevertheless, we could show that the observed activities were specific for LPS and not induced by other chlamydial components. Thus, we could inhibit the TNF release in human MNC by MEM-18, a monoclonal antibody against CD14 and we could block the activity with the LPS antagonist, B1233, that is a synthetic lipid A-like molecule (Table 3). The inhibition of chlamydial LPS activity by B1233 was also determined in a second readout system in which CD14-transfected CHO cells express, in response to LPS-induced NF-jB translocation, cell surface- exposed CD25 that can be quantified by flow cytometry (Fig. 8). In the same readout system, we showed that the activity of C. trachomatis serotype E LPS could be inhibited with monoclonal antibodies which recognize carbohydrate epitopes occurring exclusively in chlamydial LPS (Fig. 9); the same antibodies had no effect on the activity of LPS from S. enterica.

C. trachomatis L2

0.1 1 10 100 1000 – – – – 3 ± 2 – – – 20 ± 3 36 ± 9

Chl. psittaci

0.1 1 10 100 1000 – – – – – – – 2 ± 1 19 ± 1 35 ± 1

E. coli F515

0.1 1 10 100 1000 – 7 ± 1 32 ± 1 48 ± 17 58 ± 21 18 ± 2 43 ± 12 42 ± 4 45 ± 4 49 ± 6

Compound 506

0.1 1 10 100 1000 – – 9 ± 1 26 ± 7 48 ± 18 – 10 ± 2 41 ± 1 56 ± 6 56 ± 7

Finally, we investigated for the C. trachomatis serotype E LPS whether TLR2 or TLR4 were involved in the signalling pathway. As shown in Table 4, signalling occurred only in cells expressing TLR4 and not in those expressing TLR2. To our surprise, the chlamydial LPS was only active in cells coexpressing CD14 and TLR4, unlike LPS from S. enterica which was able, although to a lesser extent, to activate cells that did not express CD14. This experiment was then repeated including the other two chlamydial LPS an E. coli rough-mutant LPS (Re-LPS) and synthetic lipid A. The three typical endotoxins smooth and rough LPS and lipid A were able to stimulate TLR4-expressing HEK cells in the absence of CD14 whereas none of the chlamydial LPS showed activity in absence of CD14 (Table 5). This striking dissimilarity can not be explained simply by the apparent difference in the biological activity, otherwise the highest concentration of chlamydial LPS should have shown at least some activity. We speculate that for chlamydial LPS CD14 is required for an efficient transfer to the signalling receptor TLR4.

S. enterica sv. Friedenau

a Amount of IL-8 release induced by 10 ngÆmL)1 recombinant TNF was set to 100%. Data are calculated from two independent experiments. b Below 2%.

the biovar trachoma group, however,

In summary, we have determined the chemical structure of the LPS from another serotype of C. trachomatis that was necessary as our work on serotype L2 could not be assumed to be representative for the whole genus Chla- mydia and the disease caused by the lymphogranuloma venerum group (serotype L1–L3) is very different from genital tract infections caused by the serotypes D through K, among which serotype E is the most abundant. Some in data are available on the serotype F [27], however, that study the acylation pattern of one fraction obtained after extensive degradation was investigated. It would be the serotypes A–C which worth investigating one of these represent strains show – at least in our hands – very poor growth in tissue culture.

fatty acid composition is also very similar to that reported for serotype L2 [9,12] and F [27]. It seems that the LPS of different serotypes of C. trachomatis show only minor variations in the acylation pattern which we have observed also between different batches of LPS from the same serotype (data not shown). The acylation pattern is, however, different from that of Chl. psittaci 6BC where fatty acids with longer chains are found [13].

The fact that chlamydial LPS are endotoxically less active than typical LPS, e.g. enterobacterial LPS does not exclude its participation in the pathogenesis of chronic inflammation in chlamydial infections. We know that endotoxins are a major cause of multiorgan failure and lethal shock in severe systemic Gram-negative infection but very little is known about the role of endotoxins in local, particularly chronic infections like those caused by chlamydiae (e.g. trachoma,

The biological activity of all three chlamydial LPS was tested in a variety of readout systems in comparison to typical endotoxins. In human peripheral MNC, TNF production could be induced with LPS concentrations > 10 ngÆmL)1 (Fig. 7). LPS from C. trachomatis sero- type E and L2 were of similar activity and both less active

0.1 1 10 100 1000 – 6 ± 2 30 ± 12 56 ± 25 35 ± 11 41 ± 1 71 ± 3 62 ± 9 72 ± 6 68 ± 3

450 H. Heine et al. (Eur. J. Biochem. 270)

(cid:1) FEBS 2003

salpingitis, respiratory tract infections or atherosclerosis). The microenvironment of an inflammatory focus may favour the interaction of LPS with its target cells leading to local concentrations that also make endotoxins of low activity significant promoters of inflammatory processes.

14. Holst, O., Thomas-Oates, J.E. & Brade, H. (1994) Preparation and structural analysis of oligosaccharide monophosphates obtained from the lipopolysaccharide of recombinant strains of Salmonella minnesota and Escherichia coli expressing the genus- specific epitope of Chlamydia lipopolysaccharide. Eur. J. Biochem. 222, 183–194.

Acknowledgements

15. Bo¨ yum, A. (1968) Isolation of mononuclear cells and granulocytes from human blood. Isolation of monuclear cells by one centrifu- gation, and of granulocytes by combining centrifugation and sedimentation at 1 g. Scand. J. Clin. Laboratory Invest Suppl. 97, 77–89.

We thank U. Agge, S. Cohrs, R. Engel, I. Goroncy, H. Lu¨ thje and V. Susott for technical assistance and S. Kusumoto (Osaka/Japan) for lipid A (compound 506) and L. Hawkins (North Andover, MA USA) for compound B1233 and for reading the manuscript. This work was supported by the Deutsche Forschungsgemeinschaft grant SFB 470 C1 (to L. B) and grant LI-448 (to B. L). 16. Bazil, V., Horejsi, V., Baudys, M., Kristofova, H., Strominger, J.L., Kostka, W. & Hilgert, I. (1986) Biochemical characterization of a soluble form of the 53-kDa monocyte surface antigen. Eur. J. Immunol. 16, 1583–1589.

References

1. Moulder, J.W. (1991) Interaction of Chlamydiae and host cells 17. Fu, Y., Baumann, M., Kosma, P., Brade, L. & Brade, H. (1992) A synthetic glycoconjugate representing the genus-specific epitope of chlamydial lipopolysaccharide exhibits the same specificity as its natural counterpart. Infect. Immun. 60, 1314–1321. in vitro. Microbiol. Rev. 55, 143–190. 2. Schachter, J. & Caldwell, H.D. (1980) Chlamydiae. Annu. Rev. Microbiol. 34, 285–309. 18. Lindner, B. (2000) Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry of lipopolysaccharides. Methods Mol. Biol. 145, 311–325. 3. Storz, J. (1971) Chlamydia and Chlamydia-Induced Diseases. Charles C. Thomas, Springfield, Illinois, USA.

4. Grayston, J.T. (2000) Background and current knowledge of Chlamydia pneumoniae and atherosclerosis. J. Infect. Dis. 181, S402–S410. 19. Bock, K., Thomsen, J.U., Kosma, P., Christian, R., Holst, O. & Brade, H. (1992) A nuclear magnetic resonance spectroscopic investigation of Kdo-containing oligosaccharides related to the genus-specific epitope of Chlamydia lipopolysaccharides. Carbohydr. Res. 229, 213–224.

5. Brade, H. (1999) Chlamydial lipopolysaccharide. In Endotoxin in Health and Disease (Brade, H., Opal, S.M., Vogel, S.N. & Morrison, D.C., eds), pp. 229–242. Marcel Dekker Inc., New York, USA/Basel, Switzerland. 20. Wright, S.D., Ramos, R.A., Tobias, P.S., Ulevitch, R.J. & Mathison, J.C. (1990) CD14, a receptor for complexes of lipopolysaccharide (LPS) and LPS binding protein. Science 249, 1431–1433.

6. Rietschel, E.Th, Kirikae, T., Schade, F.U., Mamat, U., Schmidt, G., Loppnow, H., Ulmer, A.J., Za¨ hringer, U., Seydel, U., Di Padova, F., Schreier, M. & Brade, H. (1994) Bacterial endotoxin: molecular relationships of structure to activity and function. FASEB J. 8, 217–225. 7. Raetz, C.R. & Whitfield, C. (2002) Lipopolysaccharide endotoxins. Annu. Rev. Biochem. 71, 635–700. 21. Golenbock, D.T., Liu, Y., Millham, F.H., Freeman, M.W. & Zoeller, R.A. (1993) Surface expression of human CD14 in Chi- nese hamster ovary fibroblasts imparts macrophage-like respon- siveness to bacterial endotoxin. J. Biol. Chem. 268, 22055–22059. 22. Delude, R.L., Yoshimura, A., Ingalls, R.R. & Golenbock, D.T. (1998) Construction of a lipopolysaccharide reporter cell line and its use in identifying mutants defective in endotoxin, but not TNF- alpha, signal transduction. J. Immunol. 161, 3001–3009.

23. Watkins, N.G., Hadlow, W.J., Moos, A.B. & Caldwell, H.D. (1986) Ocular delayed hypersensitivity: a pathogenetic mechanism of chlamydial-conjunctivitis in guinea pigs. Proc. Natl Acad. Sci. USA 83, 7480–7484.

8. Hauschildt, S., Brabetz, W., Schromm, A.B., Hamann, L., Zabel, P., Rietschel, E.T. & Mu¨ ller-Loennies, S. (2000) Structure and Activity of Endotoxins. In Bacterial Protein Toxins (Aktories, K. & Just, I., eds), pp. 619–667. Springer, Heidelberg, Germany. 9. Nurminen, M., Rietschel, E.T. & Brade, H. (1985) Chemical characterization of Chlamydia trachomatis lipopolysaccharide. Infect. Immun. 48, 573–575. 24. Morrison, R.P., Lyng, K. & Caldwell, H.D. (1989) Chlamydial disease pathogenesis. Ocular hypersensitivity elicited by a genus- specific 57-kD protein. J. Exp. Med. 169, 663–675.

10. Brade, L., Schramek, S., Schade, U. & Brade, H. (1986) Chemical, biological, and immunochemical properties of the Chlamydia psittaci lipopolysaccharide. Infect. Immun. 54, 568–574.

11. Ingalls, R.R., Rice, P.A., Qureshi, N., Takayama, K., Lin, J.S. & Golenbock, D.T. (1995) The inflammatory cytokine response to Chlamydia trachomatis infection is endotoxin mediated. Infect. Immun. 63, 3125–3130.

12. Rund, S., Lindner, B., Brade, H. & Holst, O. (1999) Structural analysis of the lipopolysaccharide from Chlamydia trachomatis serotype L2. J. Biol. Chem. 274, 16819–16824.

25. Nurminen, M., Leinonen, M., Saikku, P. & Ma¨ kela¨ , P.H. (1983) The genus specific antigen of Chlamydia: resemblance to the lipopolysaccharide of enteric bacteria. Science 220, 1279–1281. 26. Brade, H., Brade, L. & Nano, F.E. (1987) Chemical and ser- ological investigations on the genus-specific lipopolysaccharide epitope of Chlamydia. Proc. Natl Acad. Sci. USA 84, 2508–2512. 27. Qureshi, N., Kaltashov, I., Walker, K., Doroshenko, V., Cotter, R.J., Takayama, K., Sievert, T.R., Rice, P.A., Lin, J.S. & Golenbock, D.T. (1997) Structure of the monophosphoryl lipid A moiety obtained from the lipopolysaccharide of Chlamydia trachomatis. J. Biol. Chem. 272, 10594–10600. 13. Rund, S., Lindner, B., Brade, H. & Holst, O. (2000) Structural analysis of the lipopolysaccharide from Chlamydophila psittaci strain 6BC. Eur. J. Biochem. 267, 5717–5726.