Culture supernatants from Treponema maltophilum associated with periodontitis in humans and Treponema brennaborense found in a bovine cattle disease accompanied with cachexia caused a dose-dependent TNF-α synthesis in human monocytes increasing with culture time. This activity could be reduced significantly by blocking the CD14-part of the LPS receptor using the My 4 mAb and by polymyxin B. In the murine macrophage cell line RAW 264.7, Treponema culture supernatants induced TNF-α secretion in a LPS binding protein (LBP)-dependent fashion. To enrich for active compounds, supernatants were extracted with butanol, while whole cells were extracted using a phenol/water method resulting in recovery of material exhibiting a similar activity profile. An LPS-LBP binding competition assay revealed an interaction of the treponeme phenol/water extracts with LBP, while precipitation studies implied an affinity to polymyxin B and endotoxin neutralizing protein. Macrophages obtained from C3H/HeJ mice carrying a Toll-like receptor (TLR)-4 mutation were stimulated with treponeme extracts for NO release to assess the role of TLRs in cell activation. Furthermore, NF-κB translocation in TLR-2-negative Chinese hamster ovary (CHO) cells was studied. We found that phenol/water-extracts of the two strains use TLRs differently with T. brennaborense-stimulating cells in a TLR-4-dependent fashion, while T. maltophilum-mediated activation apparently involved TLR-2. These results indicate the presence of a novel class of glycolipids in Treponema initiating inflammatory responses involving LBP, CD14, and TLRs.
Early recognition of pathogenic microorganisms and subsequent initiation of an innate immune response is essential for the host to survive acute and chronic infections (1). Crucial elements of the reaction cascade initiated by Gram-negative bacteria have been elucidated by chemical and biological analysis of the major toxic cell wall component LPS. So far, the molecular mechanisms of host defense stimulation by Gram-positive bacteria and spirochetes have not been completely elucidated, and some of the results are yet controversial. Two compounds found in the cell wall of Gram-positive microorganisms, lipoteichoic acid (LTA)4 and peptidoglycan (PG), have been shown to induce inflammatory responses in host cells (2, 3, 4). There is growing evidence that these compounds use similar host receptor pathways as LPS, i.e., it has been reported that PG is also recognized by cellular CD14 (5, 6), and LTA from Bacillus subtilis interacts with LPS binding protein (LBP) (7). The structural basis for the immunostimulating properties of spirochetes has not yet been completely elucidated. The presence of LPS in spirochetes remains controversial; however, the analysis of the entire Treponema pallidum genome revealed the lack of LPS synthesis genes (8, 9, 10, 11). Lipoproteins of spirochetes clearly have been shown to induce proinflammatory responses in host cells involving CD14 (12, 13).
Treponemes are strictly anaerobic bacteria not easily maintained in culture. By molecular genetic analysis, several formerly “noncultivable” treponemes have been identified recently (14, 15, 16), some of which were associated with human periodontitis, a chronic inflammation of the periodontium causing severe costs to health care systems (17), or digital dermatitis, a disease commonly found in cattle (15). Two Treponema species were used in this study, Treponema brennaborense and Treponema maltophilum (15, 16, 18). Previous analyses of Treponema denticola, another putative periodontal pathogen, have shown that its cell wall contained a glycolipid chemically different from LPS (19). In this study, we focus on the interaction of treponeme glycolipids and host cells.
The host response to bacterial compounds is regulated and modulated by certain serum proteins and cellular receptor molecules (20, 21, 22). LBP and soluble CD14, both able to bind LPS, are present in serum in high quantities (23, 24). LBP is an acute-phase protein synthesized in the liver, the concentration of which rises dramatically during systemic infection and the acute-phase response (25, 26). It monomerizes LPS vesicles and transports LPS to the CD14 part of the cellular LPS receptor enabling inflammatory responses, such as TNF-α synthesis. LPS effects can be blocked by a range of inhibitors including polymyxin B, a polypeptide known to bind lipid A, the active moiety of LPS (27). LPS effects can also be blocked by the mAb My 4 directed against CD14 (21). In the last years, it has been shown convincingly that members of the Toll-like receptor (TLR) family are involved in the recognition of pathogens by a wide variety of host organisms (28). In Drosophila, Toll has been shown to be involved in antifungal responses (29), while a homologous protein, 18-wheeler, induces antibacterial responses (30). In vertebrates, strong evidence has been presented that TLR-4 recognizes LPS of Gram-negative bacteria (31, 32, 33), while TLR-2 recognizes PG of Gram-positive bacteria, as well as lipoproteins of mycobacteria or Borrelia (34, 35, 36, 37, 38, 39). Regarding LTA, results for an involvement of TLR-2 or -4 have been controversial (34, 35, 40).
Here we analyze the ability of two Treponema species isolated from a patient suffering from periodontitis and from a digital dermatitis lesion, respectively, to activate human monocytes, a murine macrophage cell line, macrophages obtained from C3H/HeJ mice, and Chinese hamster ovary (CHO) cells. First, we analyzed the involvement of the host LPS binding and receptor molecules LBP, CD14, as well as TLR-2 and -4. Second, a partial chemical purification was performed for structural analysis of the active compound. The results presented here should help to elucidate the role of spirochetes in chronic inflammatory reactions and to identify the mechanisms involved.
Materials and Methods
Treponeme culture and processing of culture supernatants
Frozen stocks of T. brennaborense and T. maltophilum cells (300 μl, each stored at −80°C) were inoculated in 3 ml of culture medium (OMIZ-Pat) as described previously (16). Bacteria were cultured under anaerobic conditions (Anaerogen, Oxoid, Germany) at 37°C for 3–4 days. The cultures were then transferred to a larger volume of OMIZ-Pat (20–100 ml) and further incubated for 1–2 days. Viability of treponemes and possible presence of contaminating bacteria were assessed by dark field microscopy (400-fold magnification, BH2-RFCA microscope, Olympus, Hamburg, Germany). Sterility controls of the medium were performed by incubating OMIZ-Pat medium under aerobic and anaerobic conditions at 37°C for 1 wk. The pH value of the culture medium was measured repeatedly. Cultures were stopped at pH 6.0 and centrifuged at 12,000 × g at 4°C for 20 min. The supernatant was passed through 0.2-μm sterile filters (Schleicher & Schuell, Dassel, Germany). For some studies, culture supernatants were heat-inactivated at 100°C for 20 min and passed again through 0.2-μm sterile filters. OMIZ-Pat medium (16), treated similarly, was used as control.
Extraction of culture supernatants and whole Treponema cells
We used a modification of a published protocol for the extraction of LPS from Gram-negative cell walls using n-butanol (41). Briefly, filtered and heat-inactivated culture supernatants were mixed with an equal volume of n-butanol and incubated at 4°C for 1 h on a 180° shaker. Subsequently, the mixture was centrifuged at 26,000 × g at 4°C for 1 h, and the upper butanol phase was recovered. These steps were repeated once. Combined butanol phases were recentrifuged and lyophilized. For stimulation experiments, lyophilized extracts were dissolved in RPMI 1640 medium (Life Technologies, Eggenstein, Germany) to the original volume of supernatants before extraction. For silver stain analysis, butanol extracts corresponding to 1 ml of culture supernatant were dissolved in 50 μl of distilled water. OMIZ-Pat medium was processed similarly and used as control. For an extraction of whole treponeme cells, aqueous suspensions of treponeme cells were digested with RNase (Sigma, Deisenhofen, Germany), DNase (Merck, Darmstadt, Germany), and proteinase K (Merck). The suspensions were dialyzed and extracted using a hot phenol/water extraction method (42) or, after drying, a phenol/chloroform/petroleum ether (PCP) method (43). In brief, the phenol/water extraction was performed by mixing the cell suspension with an equal volume of 90% phenol and stirring at 68°C for 10 min. After cooling on ice, the mixture was centrifuged at 3000 × g for 10 min at 0°C, and the upper phase was collected. This procedure was repeated twice, and combined phases were dialyzed and lyophilized. PCP extraction was conducted with a mixture of 90% PCP in a volume ratio of 2.5:2.5:4. PCP extraction resulted in yields of 0.075% for T. brennaborense and 0.036% for T. maltophilum according to wet weight. By using the phenol/water method, higher yields were obtained (T. brennaborense, 0.42%; T. maltophilum, 0.47%), therefore, the latter material was used for additional experiments.
Stimulation of human monocytes and the murine macrophage cell line RAW 264.7
Peripheral blood samples obtained from healthy volunteers were mixed with 50 U/ml of heparin as anticoagulant and diluted 1:2 with RPMI 1640. Then 30 ml of the diluted blood was carefully layered on top of 15 ml Lymphoprep (Nycomed, Oslo, Norway) and centrifuged at 600 × g without brake at 21°C for 15 min. The intermediate phase was recovered, washed twice with RPMI 1640, and recentrifuged at 600 × g at 21°C for 5 min. To separate platelets, cells were further spun at 100 × g at 21°C for 15 min. Remaining cells were diluted in RPMI 1640 containing 5% human AB serum (Sigma) to a final concentration of 1 × 106 cells/ml. Cells were transferred to 96-well cell-culture plates (100 μl/well) and incubated at 37°C for 1.5 h, followed by two washing steps with RPMI 1640 to remove nonadherent cells. Remaining monocytes were stimulated with culture supernatants, butanol extracts of Treponema isolates, extracts of whole treponeme cells, Escherichia coli 0111:B4 LPS (Sigma), or PMA (Sigma) in the presence or absence of 5% human AB serum in a total volume of 100 μl. For certain experiments, polymyxin B (Sigma) at a concentration of 5 μg/ml was added directly before stimulation, as indicated. For selected experiments, monoclonal anti-CD14 Ab My 4 (Coulter, Hamburg, Germany) was incubated with the cells at a concentration of 5 μg/ml at 37°C for 20 min before addition of stimuli to block CD14. After 4 h, supernatants were harvested and viability of cells was assessed via trypan blue staining. Additionally, 5 × 104 RAW 264.7 cells per well (kindly provided by T. Blankenstein, Max-Delbrück-Centrum, Berlin, Germany) were cultured overnight in 96-well tissue culture plates using RPMI 1640 supplemented with 10% FCS. After repeated washing with RPMI 1640, stimulation was performed in the presence or absence of 1 μg/ml recombinant murine LBP (rmLBP) in a total volume of 100 μl. RAW 264.7 supernatants were harvested after 4 h of incubation, and cells were stained with trypan blue, ensuring integrity of cells.
Preparation and stimulation of peritoneal elicited macrophages (PEM)
Peritoneal macrophages were isolated from C3H/HeJ or C3H/HeN mice (Charles River, Sulzbach, Germany), by thioglycollate elicidation. Female 7-wk-old mice were injected i.p. with 1.5 ml of 3% thioglycollate broth (Sifin, Berlin, Germany). After 3 days, mice were sacrificed and peritoneal macrophages were harvested by injection of 10 ml of ice-cold HBSS (Life Technologies) i.p. followed by aspiration. Cells were washed twice with RPMI 1640, and 2 × 105 cells were plated in 96-well tissue culture plates in RPMI 1640 containing 5% FCS. After 2 h, plates were washed twice with RPMI 1640 to remove nonadherent cells, and remaining cells were stimulated with treponeme phenol/water extracts or LPS for 24 h in RPMI 1640 containing 5% non-heat-inactivated FCS followed by NO detection as described below.
Quantitative detection of human and murine TNF-α and NO
Nunc MaxiSorp ELISA plates (Nunc, Roskilde, Denmark) were coated with 0.5 mg/ml of anti-human TNF (anti-hTNF) Ab (PharMingen, Heidelberg, Germany) in 100 mM NaHCO3, pH 8.3, and blocked with PBS containing 0.05% Tween 20 and 10% FCS. Cell supernatants and rhTNF standard (R&D Systems, Wiesbaden, Germany) in PBS containing 10% FCS were incubated at 4°C overnight. Bound hTNF was detected using a biotinylated mouse anti-hTNF Ab (PharMingen) at a concentration of 0.5 mg/ml. Subsequently, 1 μg/ml streptavidin peroxidase conjugate (Sigma) was added with ortho-phenylene-diphosphate (Sigma) as substrate. The detection limit of this assay was ∼10 pg/ml. For quantitation of murine TNF-α, MaxiSorp ELISA plates were coated with 3 μg/ml anti-murine TNF (anti-mTNF) Ab (PharMingen) in 100 mM Na3PO4, pH 6.0. Samples and rmTNF standard (R&D Systems) were incubated at room temperature for 3 h, followed by detection with a biotin-conjugated anti-mTNF-α Ab (PharMingen) and streptavidin-peroxidase with ortho-phenylene-diphosphate as substrate. The detection limit was ∼15 pg/ml. All in vitro TNF-α results were assessed statistically by the Student’s t test, and the inhibitory effects of polymyxin B and My 4 as well as the enhancing effects of LBP were highly significant (p < 0.001). NO2− accumulation in culture medium was assessed according to a published protocol (44). In brief, 100 μl of Griess reagent (Sigma) was added to 100 μl of culture medium in 96-well plates and measured in a microplate reader at 540 nm with a standard of NaNO2 diluted in RPMI 1640.
Estimation of NF-κB translocation
CHO cells transfected with human CD14 (CHO/CD14, generously provided by L. Hamann, Forschungszentrum Borstel, Germany) (45) were cultured overnight in six-well tissue culture plates at 4 × 105 cells per well with Ham’s nutrient medium F12 (PAA Laboratories, Linz, Austria) supplemented with 10% FCS and 400 μg/ml hygromycin B (Calbiochem, San Diego, CA). Before stimulation, cells were starved in FCS-free Ham’s medium for 3 h and incubated with LPS or treponeme extracts in the presence of 2% non-heat-inactivated FCS. After 1 h, cells were washed with ice-cold PBS containing 1 mM Na3VO4 and incubated in 150 μl of buffer A (1 mM Na3VO4, 10 mM HEPES, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM DTT, 0.5 mM PMSF, and 1 mM NaF). After 20 min, cells were harvested mechanically, transferred to 1.5-ml tubes, mixed with 25 μl Nonidet P-40, and centrifuged at 13,000 × g at 4°C for 1 min. Pellets were resuspended in 50 μl of buffer B (400 mM NaCl, 1 mM Na3VO4, 20 mM HEPES, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 1 mM PMSF, and 1 mM NaF), incubated for 30 min at 4°C, and spun at 13,000 × g at 4°C for 5 min. Supernatants containing nuclear proteins were collected, and nuclear extracts were analyzed by EMSA as described previously (46) using two synthetic oligonucleotides (Eurogentec, Seraing, Belgium) containing the NF-κB binding sequence of the murine Ig κ light chain gene enhancer.
Treponema culture supernatants, both native and butanol extracted, as well as extracts of whole cells were assayed for endotoxin contamination by using a chromogenic Limulus amoebocyte lysate (LAL) assay (LPS, Sinntal-Oberzell, Germany). The endotoxin content of the OMIZ-Pat culture medium was below 2 ng/ml regardless of whether treponemes were cultured or the medium was incubated without bacteria for control. Butanol extracts of the culture supernatants contained endotoxin of ∼25 pg/ml. The phenol/water extracts of T. maltophilum and T. brennaborense exhibited a LAL activity of 16.3 pg/μg corresponding to 0.16 endotoxin units (EU) and 35.2 pg/μg corresponding to 0.35 EU, respectively.
Electrophoresis and silver staining of treponeme extracts
Stacking gels (5%) and separating gels (15, 16, and 20%, respectively) were prepared without SDS. Prestained and unstained low molecular mass markers ranging from 3 to 43 kDa (Life Technologies), 30 μl of culture supernatants, 30 μl of butanol extracts, and 30 μl of LPS solutions derived from E. coli 0111:B4 or Salmonella minnesota Re 595 LPS (Sigma) were boiled in sample buffer (2 ml 1 M Tris, 4 ml 1 M DTT, 800 mg SDS, 40 mg bromophenol blue, and 4 g glycerol ad 10 ml H2O) for 5 min, loaded onto the gels, and submitted to electrophoresis. Gels were stained with the silver stain plus kit (Bio-Rad, Munich, Germany) according to the manufacturers instructions. In addition to the original protocol, gels were oxidized with 0.7% periodic acid after fixation (47). For some experiments, glycolipids were hydrolyzed and/or dephosphorylated as explained below.
Chemical analysis of the phenol/water-extracted cell wall fractions of T. maltophilum and T. brennaborense
Phosphate was determined according to Lowry (73), and 3-deoxy-d-manno-octulosonic acid (Kdo) was estimated by the thiobarbituric acid method (43). Amino acids were identified as their phenyl isothiocyanate derivatives by reversed-phase HPLC using a Waters PICO-TAG system (Waters, Eschborn, Germany) under conditions described previously (48). Amino sugars were analyzed using HPLC (49). Gas-liquid chromatography (GLC) and combined gas-liquid chromatography/mass spectrometry (GLC-MS) were applied for the analysis of neutral sugar alditol acetates (50) and fatty acid methyl esters liberated after strong methanolysis (2 M HCl/MeOH, 120°C, 24 h) and extraction with chloroform. GLC was performed on a model 3700 Varian gas chromatograph (Varian Associates, Palo Alto, CA), and GLC-MS was performed on a Hewlett Packard 5989A instrument equipped with a gas chromatograph (model 5890 Series II, Hewlett-Packard, Palo Alto, CA) operating under identical conditions as for GLC. For structural analysis of treponeme glycolipids, 3.8 mg of T. maltophilum and 4.9 mg of T. brennaborense phenol/water extracts were dephosphorylated in 100 μl HF (48% by volume) at 4°C for 24 h in a sealed Teflon tube. Samples were extensively dialyzed (cut-off at 10–14 kDa) against water and lyophilized. The yields obtained were 1.3 mg for T. maltophilum and 1.8 mg for T. brennaborense. Dephosphorylated glycolipids were peracetylated in 0.8 ml pyridine/acetic acid anhydride (5:3 by volume, 85°C, 30 min) and subjected to GLC-MS analysis. For analysis of the glycosyl part, peracetalyted glycolipids were further purified on a silica gel column (3.5 × 1.5 cm, Kieselgel 60, Merck, 230–400 mesh), eluted with a stepwise gradient of increasing amounts of ethanol in toluene (1–50% by volume). For GLC-MS analysis, fatty acids were released from the glycolipid by alkaline hydrolysis (50 μl 0.5 M NaOH, 1 h at 65°C), and the product was permethylated (51). GLC-MS analysis of the permethylated and deacylated glycolipids was performed using a gradient of 150°C (3 min) to 330°C at 10°/min.
Binding of phenol/water extracts to rmLBP, endotoxin neutralizing protein (ENP), and polymyxin B
Binding of phenol/water extracts to mLBP was investigated with a slightly modified competition assay published elsewhere (52). Briefly, MaxiSorp ELISA plates were coated with E. coli 0111:B4 LPS. Free protein binding sites were blocked by incubation with 10 mg/ml BSA in 150 mM NaCl, 50 mM HEPES, pH 7.4, at 37°C for 30 min. Washing and dilution steps were performed with blocking buffer containing 1 mg/ml BSA. E. coli 0111:B4 LPS and S. minnesota Re 595 LPS, and phenol/water extracts of T. brennaborense and T. maltophilum were assayed for their ability to bind to 100 ng/ml mLBP by inhibiting binding of LBP to LPS-coated plates. LPS-bound LBP was detected by a polyclonal rabbit-anti-mLBP Ab and incubated with goat anti-rabbit IgG-Ab, conjugated with HRP (Biogenes, Berlin, Germany). Ortho-phenylene-diphosphate was used as a substrate. E. coli 0111:B4 LPS and treponeme phenol/water extracts were precipitated by polymyxin B-coupled Agarose beads (Sigma) or Sepharose beads conjugated with ENP (Associates of Cape Cod, Falmouth, MA), respectively. LPS and extracts in a volume of 500 μl at a final concentration of 10 μg/ml were mixed with 50 μl of polymyxin B beads or with ENP beads, respectively, as recommended by the manufacturer. Samples were incubated at 4°C for 24 h using a 180° shaker. After centrifugation at 3000 × g at 4°C for 10 min, supernatants were collected and loaded onto 15% SDS-PAGE gels, followed by silver staining as described above. Control samples were treated accordingly, however, without addition of any beads. Murine LBP was expressed in a baculovirus system and purified as described (26).
Induction of TNF-α in myeloid cells by treponeme culture supernatants in the presence of serum, polymyxin B, mAb My 4, or rmLBP
Treponema culture supernatants induced TNF-α in freshly isolated human monocytes. This activity increased with culture time reaching a maximum at day 3 (Fig. 1). OMIZ-Pat culture medium alone, incubated with monocytes for the same period of time, failed to induce any detectable amounts of TNF-α (data not shown). For the following experiments, bacteria were cultured for 3 days. Cultures were monitored by pH measurement (6.0) to guarantee similar growth conditions. Both viability and motility of treponemes were assessed by dark field microscopy. To compare the activity of the treponeme supernatants with LPS, we performed experiments with monocytes in the presence and absence of serum. Cytokine induction caused by treponeme culture supernatants increased significantly in the presence of 5% human serum (Fig. 2,A). As compared with T. brennaborense, serum-independent stimulation was significantly stronger for T. maltophilum culture supernatants (Fig. 2,A). Both polymyxin B and the inhibitory monoclonal anti-CD14 Ab My 4 were able to significantly reduce cytokine levels induced by both treponeme cultures. However, the effect was more pronounced for LPS (Fig. 2,B). The cytokine-inducing activity of T. brennaborense and T. maltophilum culture supernatants was inhibited in the presence of polymyxin B or My 4 mAb at least by 50%. In contrast, polymyxin B and My 4 did not influence cytokine induction caused by PMA, a phorbol ester causing cytokine induction by activating protein kinase C directly without receptor interaction (53) (data not shown). To investigate LBP effects on cytokine induction caused by treponemes, purified rmLBP and the murine macrophage cell line RAW 264.7 were used. In RAW 264.7 cells, TNF-α-induction caused by both treponeme culture supernatants was significantly increased by addition of mLBP (Fig. 2 C).
Induction of cytokines by butanol extracts of culture supernatants and phenol/water extracts of whole cells in the presence of serum, polymyxin B, My 4, and rmLBP
Based upon the notion that the stimulatory activity found within the supernatants shared characteristics with LPS, we purified the compounds from culture supernatants as well as from whole cells using extraction methods commonly used for LPS. Supernatants were treated with butanol, while whole cells were subjected to phenol/water or the PCP extraction. Yields obtained from whole cells with the phenol/water method were clearly higher; therefore, this method was used in the following experiments. To identify possible LPS contamination during the preparation, a mock extraction including all media and chemicals used during the procedure, was performed. TNF-α induction in human monocytes caused by butanol extracted Treponema supernatants was clearly reduced by both polymyxin B and My 4 (Fig. 3,A). In RAW 264.7, an LBP-dependent cytokine induction was observed (Fig. 3,B). In contrast to the culture supernatants, no TNF-α was induced in the absence of LBP. Similarly, phenol/water extracts of whole treponeme cells revealed a serum-dependent cell-stimulating capacity in human monocytes (data not shown). However, to achieve a TNF-α release equivalent to that caused by LPS, the concentrations of the extracts had to be increased by 1000-fold. Addition of polymyxin B and mAb My 4 led to a significant decrease of cytokine levels (Fig. 3,C). The cytokine induction in RAW 264.7 cells was greatly increased by the addition of rmLBP (Fig. 3 D). In case of T. maltophilum, it was LBP dependent. In all experiments, the mock extracts did not cause cytokine induction.
Induction of NO synthesis by treponeme phenol/water extracts in PEM derived from C3H/HeJ and C3H/HeN mice
The role of TLR-4 in treponeme-mediated cell stimulation was analyzed using PEM derived from LPS hyporesponsive C3H/HeJ mice, a strain bearing a dominant negative mutation in the gene encoding TLR-4 (31). We isolated PEM from C3H/HeJ mice, as well as from the control C3H/HeN strain normally responsive to LPS. Cells were stimulated with increasing amounts of LPS and treponeme phenol/water extracts, followed by measurement of NO. LPS exhibited a significantly stronger stimulatory activity toward C3H/HeN as compared with C3H/HeJ PEM (Fig. 4). While the mock extract failed to stimulate cells, phenol/water extracts derived from T. brennaborense revealed a stimulation pattern comparable to LPS, leading to a significantly weaker NO release by C3H/HeJ macrophages as compared with C3H/HeN cells. In contrast, extracts derived from T. maltophilum led to a comparable NO production in PEM of both strains, suggesting a less important role of TLR-4.
Translocation of NF-κB in CHO cells by treponeme phenol/water extracts
To elucidate the role of TLR-2 in treponeme-mediated signaling, we investigated CHO cells. These cells carry a mutation for TLR-2 leading to a defective receptor expression (54). CHO cells transfected with human CD14 inducing responsiveness to LPS (CHO/CD14) were stimulated with treponeme phenol/water extracts as well as with LPS. LPS induced a strong translocation of NF-κB in CHO/CD14 cells as shown in an EMSA (Fig. 5). T. brennaborense phenol/water extracts induced a translocation of NF-κB at concentrations of 1 μg/ml comparable to the LPS effect. In contrast, T. maltophilum-derived extracts, at 1 μg/ml, failed to induce NF-κB translocation, indicating an involvement of TLR-2.
SDS-PAGE silver stain analysis of native and butanol-extracted culture supernatants and whole cell phenol/water extracts
To identify and further characterize the active components in pure or butanol-extracted treponeme culture supernatants, the material was analyzed on 15% polyacrylamide gels that were silver stained subsequently. Visible bands appeared in the range of about 4 and 6 kDa in butanol-extracted supernatants from T. maltophilum and T. brennaborense, respectively (Fig. 6,A). The low molecular material reflected rough LPS derived from S. minnesota Re 595 consisting only of lipid A and the inner core region with a size of 2.5 kDa (lane 1). The relative concentration present in the butanol extract of T. brennaborense revealed that this material contained about twice as much of the low molecular compound as the T. maltophilum extract. Digestion of both butanol extracts with pronase failed to change the profile of bands observed in SDS-PAGE, suggesting a nonproteinaceous nature of the immunostimulatory compound (data not shown). Phenol/water extracts of both strains were analyzed by silver-stained SDS-PAGE, revealing striking differences (Fig. 6 B). The material obtained from T. brennaborense displayed a ladder-like pattern similar to that of smooth LPS presumably containing numerous repeating carbohydrate units. In contrast, T. maltophilum extracts exhibited few repeating units of larger molecular size. For both butanol-extracted culture supernatants and phenol/water extracts of whole cells, the size of their smallest units was similar.
Compositional and structural analysis of the phenol/water-extracted cell wall fractions of T. maltophilum and T. brennaborense
The results of a chemical analysis of phenol/water extracts from T. maltophilum and T. brennaborense are shown in Table I. Most of the components in both fractions could be analyzed and quantified (56% (w/w) for T. maltophilum and 71% for T. brennaborense). As expected from SDS-PAGE analysis, the amount of total fatty acids in the smaller glycolipid of T. maltophilum was significantly higher (8.3%, w/w) as compared with the high molecular T. brennaborense glycolipid (2.8%). LPS-characteristic β-hydroxylated fatty acids, as well as Kdo and heptose, were completely lacking in both strains. In T. maltophilum, galactosamine was identified to be the main sugar component (13%), whereas in T. brennaborense, glucose was identified as the dominating sugar (50.8%). In both preparations, a characteristically high amount of phosphate could be identified (5–10%). Besides traces of contaminating residual amino acids, alanine was the only amino acid identified in T. maltophilum, whereas T. brennaborense completely lacked amino acids.
|Componenta .||Concentration .||.||.||.|
|.||T. maltophilum .||.||T. brennaborense .||.|
|.||nmol/mg .||% (w/w) .||nmol/mg .||% (w/w) .|
|Polar head groups|
|Componenta .||Concentration .||.||.||.|
|.||T. maltophilum .||.||T. brennaborense .||.|
|.||nmol/mg .||% (w/w) .||nmol/mg .||% (w/w) .|
|Polar head groups|
Chemical analysis was performed employing GLC and HPLC as described in Materials and Methods.
The amino-dideoxy-hexose could not be quantified due to the lack of reference compound.
After alkaline hydrolysis of both glycolipids by treatment with KOH, the material could not be visualized by silver staining, indicating the loss of ester-bound fatty acids (Fig. 7, A and B). To cleave phosphate-interlinked sugar chains, glycolipids were dephosphorylated by HF, resulting in material of ∼1 kDa size, which also could not be stained after KOH treatment (Fig. 7, A and B). This result suggests the presence of diacylglycerol as lipid anchor in both glycolipids and excludes sphingosine and steroids. For identification of the lipid anchor in the glycolipid of T. maltophilum, extracts were peracetylated after dephosphorylation with HF. GLC-MS analysis (CI mode) of the peracetylated glycolipids from T. maltophilum revealed one major set of peaks (tR ∼17 min) expressing pseudomolecular ions ([M+NH4]+) of m/z = 572 and 586, respectively. Their molecular masses (Mr = 554 and 568 kDa, respectively) are consistent with monoacetylated diacyl-glycerol carrying two tetradecanoic acid residues (14:0) and one pentadecanoic acid (15:0), respectively (Fig. 7,C). However, attempts to identify a similar lipid anchor in T. brennaborense by using the same approach were unsuccessful, most likely due to the small proportion of the lipid in the whole molecule (see Table I).
For identification of the glycosyl part within the glycolipid, the peracetylated glycolipids were further purified by silica gel chromatography, and fatty acids were released by alkaline hydrolysis, followed by permethylation of the remaining material. GLC-MS analysis of T. maltophilum, revealed two glycosyl derivatives with retention times of 24.9 and 28.7 min expressing in the CI-mode pseudomolecular ions [M+NH4]+ with m/z = 764 (Mr = 746 kDa) and [M+H+]+ m/z = 788 (Mr = 787 kDa). These could be assigned to a permethylated trisaccharide glyceride with three hexoses (Hex3-Gro), and a trisaccharide consisting of two hexoses and one hexosamine (Hex2-HexNAc-Gro) (Fig. 7,D). Both permethylated glycosylglycerides were present in ∼1:9 proportion. The electron impact-mass spectrometry of the major permethylated glycosylglyceride showed diagnostic fragments derived from the reducing part of the molecule (m/z = 307, 557) as well as those from the nonreducing part (m/z = 464, 668), thus allowing the sequence of the trisaccharide to be assigned to Hex-HexNAc-Hex-Gro (Fig. 7 D). Despite several attempts, the structure of the glycosyl part of the putative glycolipid isolated from T. brennaborense could not be revealed.
Binding of treponeme phenol/water extracts to mLBP, ENP, and polymyxin B
To analyze the potential interaction of Treponema glycolipids with LBP, we performed competition assays with rmLBP (Fig. 8,A). These studies revealed that the glycolipids are able to compete with LPS-LBP binding. Both extracts displayed a comparable affinity, while LPS derived from S. minnesota Re 595 and LPS from E. coli 0111:B4 exhibited a stronger activity. Next, we investigated possible interactions of Treponema extracts by precipitation with polymyxin B and beads coated with ENP, an endotoxin binding protein used in the Limulus assay. These experiments indicate that the glycolipids studied interact with both, polymyxin B, and ENP, as precipitation led to a marked reduction of bands shown in SDS-PAGE analysis (Fig. 8 B). While for ENP no differences were observed comparing LPS and the treponeme extracts, the effect was less pronounced for polymyxin B, especially regarding T. maltophilum interaction.
The interaction of microorganisms with the host defense system is a hallmark during infection, deciding over survival of the microbe and/or development of an inflammation. Bacteria exhibit certain features enabling the host to discriminate them as “foreign” and to mount responses leading to elimination of the pathogen. There is a number of bacteria causing chronic inflammations leading to tissue destruction and severe, although not acute life-threatening, diseases. Treponema, generally known for their lack of antigenicity, induce a moderate innate immune response leading to chronic rather than acute, often systemic, infections. However, a distinct local inflammatory reaction pattern can be observed in many spirochete infections, i.e., the primary local stage of Lyme disease. Local release of cytokines causing tissue inflammation is likely, and macrophages have been proposed to play a major role during this process (55). Spirochetes may also directly stimulate resident defense cells for mediator release, although clear molecular mechanisms for this process have to be defined.
Immunostimulatory elements of bacteria interact with soluble and cell-bound receptor molecules of the host organism, a key element of the host’s repertoire to modulate an inflammatory reaction. As we have shown recently, the acute-phase response to a systemic infection leading to elevated levels of the hepatic acute-phase protein LBP can greatly modulate the hosts response to a systemic challenge with LPS (26). It is likely that LBP serves for toxic Treponema cell wall products as well as a modulator in vivo and may be able to reduce or enhance the inflammatory reaction. Recent results by others and us provide evidence that LBP interacts not only with LPS, but also with other bacterial products, i.e., LTA (7). Furthermore, our results showing an involvement of the CD14 part of the LPS receptor are in agreement with CD14 acting as a pattern recognition receptor (56, 57).
The immunostimulatory treponeme cell wall compounds described here are apparently released by live bacteria or after cell death. We present evidence that the compounds retained from the supernatants correspond to the glycolipids extracted from whole cells regarding size and biological characteristics. After cell death, bacteria release immunostimulating particles such as LPS in Gram-negative bacteria, causing a strong inflammatory response potentially leading to septic shock and subsequent death of the host (58, 59). In Gram-positive bacteria, elements like PG and LTA of the outer cell wall, released after cell disintegration often following antibiotic treatment, also stimulate cytokine release in host cells (3, 60). However, for spirochetes, the predominant inflammatory active element of their cell wall has not yet been clearly identified. For some spirochetes like T. hyodysenteriae and T. innocens, the presence of LPS-like molecules has been described, while for others it has been clearly ruled out (10). Borrelia burgdorferi fails to contain LPS in the cell wall (8, 9); however, it contains PG (61) and a set of outer membrane proteins eliciting inflammatory responses in immune cells (62, 63, 64). It is likely that these proteins are identical with lipoproteins described to induce TNF-α-synthesis in human monocytes (12, 65). T. pallidum possesses a number of TNF-α-inducing membrane proteins, and for this spirochete the presence of LPS was definitively ruled out after the completion of the whole genome sequence (8, 11, 66). Our observations provide evidence that treponemes contain a glycolipid-like material within their membranes that is chemically different from LPS while exhibiting comparable biological characteristics including, for one of our isolates, involvement of TLR-4.
Analysis of the role of TLRs is important in light of the recent paradigm of Gram-negative bacteria using TLR-4 via LPS and other bacteria stimulating cells via TLR-2. Recently, two immunostimulatory fractions isolated from the cell walls of T. denticola have been compared, lipoproteins on one hand and lipooligosaccharides, comparable to the glycolipids isolated in our study, on the other hand (67). Both fractions were able to induce cytokines and NO in host cells of normal and C3H/HeJ mice, suggesting a TLR-4-independent activity, which is in line with our results for T. maltophilum. Lipoproteins isolated from Borrelia recently also have been found to stimulate host cells via TLR-2 similar to lipoproteins isolated from mycobacteria (36, 37, 39). According to our results, active cell wall compounds from genetically closely related spirochetes stimulate cells via different TLRs. However, chemical analysis revealed differences in the composition of the glycolipids isolated. Thus structural differences may explain the differential use of pattern recognition receptors. While the T. brennaborense glycolipid contained significantly more carbohydrates and revealed the presence of a high number of small “repeating units” in silver gel analysis, T. maltophilum glycolipids displayed a small number of larger “repeats.” Recently, the TLR-2/TLR-4 paradigm was questioned by two other studies showing involvement of TLR-4 in non-LPS-mediated cell stimulation. Viable mycobacteria, in contrast to isolated lipoarabinomannan, stimulated CHO cells overexpressing both TLR-2 or TLR-4 (68). A recent study comparing the TLR-2 and the TLR-4 knockout mouse provided evidence that LTA from Gram-positive bacteria also stimulate macrophages via TLR-4 (40). Furthermore, this group compared different types of LPS leading to a different degree of use of members of the TLR family (69).
We describe here the molecules involved in the reaction pattern of myelo-monocytic host cells to contact with cell wall components of recently identified spirochetes. Certain features of this interaction, i.e., involvement of LBP, CD14, and the use of TLR-4 by T. brennaborense, as well as the inhibitory effect of polymyxin B, resemble the cell stimulation pattern induced by LPS of Gram-negative bacteria. However, for polymyxin B, it recently has been shown that it interacts with numerous structures including phospholipids (70). Furthermore, the Treponema glycolipids described here bound to ENP, a protein usually considered to bind specifically to LPS. However, it is known that agents other than LPS cross-react in the Limulus assay, potentially due to similar physical properties (71, 72). Because the glycolipids described in this study interact with a range of other LPS-binding structures, it is likely that the discrete LAL activity observed is caused by the extracted compounds themselves and not by contaminating LPS. Furthermore, precipitation studies using ENP revealed a specific affinity of the treponeme extracts to this protein.
Our chemical analysis suggests a glycolipid structure in T. maltophilum and T. brennaborense differing significantly from that of LPS. This is based on the absence of structural components characteristic for LPS, such as heptose, Kdo, and β-hydroxy fatty acids. In contrast, Treponema glycolipids displayed LTA-like elements such as sugar, high phosphate, and alanine similar to that previously identified in T. denticola (19). This similarity was further supported by isolation and analysis of the dephosphorylated glycosyl part of the repeating units, being a hexasaccharide in T. maltophilum and a glucan in T. brennaborense (data not shown). Moreover, in T. maltophilum we identified two glycolipids composed of Hex3Gro and Hex-HexN-Hex-Gro (Fig. 7,D). GLC-MS analysis of the lipid anchor revealed two monoacetylated diacylglycerols, the predominant one containing two tetradecanoic acids (14:0) (Fig. 7,C). Our interpretation that T. brennaborense contains a glycolipid of similar structure is based on results obtained from SDS-PAGE (Fig. 7, A and B) and from TLC analysis (data not shown).
Taken together our chemical results indicate that T. maltophilum and T. brennaborense both exhibit a glycolipid consisting of a diacylglycerol-lipid anchor, a core region, in the case of T. maltophilum consisting of three sugars, and carbohydrate repeating units (Fig. 7 E). As indicated by silver stain analysis, T. maltophilum exhibits a low number of large repeating units, each being composed of ∼20–30 sugars, while T. brennaborense contains a high number of small repeating units, each being composed of ∼5 sugars. Like in T. denticola (19), these glycolipids share structural characteristics with LTA and apparently represent the major membrane component. The differences in chemical composition between the two strains are significant considering the close genetic relatedness of both strains (15) and may be the cause for the different interactions with TLRs. A more detailed chemical analysis will be performed in our laboratories to further support this interpretation.
Our data complement the list of bacterial cell wall components recognized by TLRs and CD14, explaining results by others showing a CD14 involvement for spirochete-mediated host cell stimulation (13, 64). The differential use of TLRs by the treponeme glycolipids may help in understanding basic mechanisms of innate immunity caused by spirochetes as well as other microorganisms.
We acknowledge the excellent technical assistance of Nicole Siegemund, Fränzi Creutzburg, Marco Kachler, and Cyndi Hefenbrock (Charité), as well as Hermann Moll, and Ursula Schombel (Forschungszentrum Borstel).
This work was supported in part by grants given by the Deutsche Forschungsgemeinschaft (Schu 828/1-5, to R.R.S.) and the Bundesministerium für Bildung und Forschung (01 KI 94750, to R.R.S.; 01 KI 9471/9 and 01 KI 9851/0, to U.Z.; and 01 KI 9318, to U.B.G.). K.S.M. was supported by the Boehringer-Ingelheim Stiftung.
Abbreviations used in this paper: LTA, lipoteichoic acid; CHO, Chinese hamster ovary; ENP, endotoxin neutralizing protein; Kdo, 3-deoxy-d-manno-octulosonic acid; LAL, Limulus amoebocyte lysate; LBP, LPS binding protein; OMIZ-Pat, Treponema culture medium; PCP, phenol/chlorophorm/petroleum ether; PEM, peritoneal elicited macrophages; PG, peptidoglycan; TLR, Toll-like receptor; h, human; m, murine; GLC, gas-liquid chromatography; MS, mass spectrometry.