INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Lipopolysaccharides (LPSs), the endotoxins of gram-negative bacteria, stimulate various cell types to release mediators, including cytokines (1, 28, 38). The main sources of inflammatory cytokines are the monocytes/macrophages. At low concentrations, LPS leads to a modest release of mediators which exhibit positive immunstimulatory effects (2, 29, 31, 39). High amounts of LPS, however, induce an extensive production of mediators leading to pathophysiological manifestation of sepsis, such as fever, tachycardia, tachypnea, leukopenia, and hypotension (32). LPS of wild-type (S-form) strains consists of a polysaccharide and a lipid part (41), termed lipid A. The polysaccharide part comprises the O-specific chain and the core oligosaccharide, forming an oligomer consisting of up to 50 oligosaccharide units and specific for a bacterial serotype. The core region, like the lipid compound, is structurally more conserved. Lipid A represents the toxic principle of LPS (13). Bacterial strains which lack the O-specific chain or parts of the core region express rough (R)-form LPS. R mutants are grouped depending on the biosynthetic defects of the bacteria from which the LPS is derived as Ra to Re chemotypes (Fig. 1). Although the mechanisms of activation of monocytes by LPS are not clear in all details, it is accepted that LPS binds to a specific membrane receptor, the CD14 molecule (42), and that binding of LPS to this receptor is catalyzed by an acute-phase serum protein, i.e., LPS binding protein (34). There is evidence that binding of LPS is followed by stimulation of the cell and LPS internalization. According to present knowledge, stimulation of cytokine production in monocytes/macrophages represents a receptor-mediated process, which does not require internalization (16). Internalization, however, may represent a process necessary for the degradation of endotoxin and its elimination from the circulation. Various studies have been performed dealing with the structure-activity relationship of different LPS or lipid A structures in their capacity to bind and stimulate monocytes/macrophages. However, a comparison of different partial structures of LPS concerning their internalization subsequent to binding has not been performed. In this study, we have therefore investigated the binding and internalization properties of several rough LPS chemotypes and LPS partial structures by using human monocytes as target cells in comparison with functional parameters, i.e., tumor necrosis factor alpha (TNF-) release and oxidative burst.

View larger version (28K): [in this window] [in a new window]   FIG. 1.   Schematic structures of the compounds used. GlcN, glucosamine; Kdo, 3-deoxy-D-manno-octulopyranosonic acid; Hep, L,D-heptopyranose.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

LPS, lipid A, and compound 406. The schematic structure of the compounds used in this study is shown in Fig. 1. LPSs of the Escherichia coli Re mutant (strain F515), the Salmonella minnesota Rd1 mutant (strain R7), and the E. coli Ra mutant (strain K-12) were obtained by extraction with phenol-chloroform-petroleum ether (14). Free lipid A was prepared by acetate buffer treatment (1 M, 100°C, 1.5 h) of E. coli Re LPS, subsequent dialysis, and conversion to the triethylammonium salt form. The synthetic tetraacylated bisphosphate precursor of E. coli lipid A (also known as compound 406, lipid IV_a, or LA-14-PP) was synthesized as described elsewhere (19). LPS, lipid A, and partial structures were solubilized in pyrogen-free distilled water and stored in aliquots at 1 mg/ml and 4°C.

Antibodies. The anti-lipid A monoclonal antibody (MAb) (A6) was used as a serum-free hybridoma supernatant (24). A6 recognizes natural lipid A as well as compound 406. MAb A25 recognizes the 3-deoxy-D-manno-octulopyranosonic acid region on the E. coli Re LPS (5). MAb S32-32 recognizes S. minnesota R7 Rd1 LPS (35), and E. coli K-12 Ra LPS is recognized by MAb S31-21 (6). The anti-CD14 MAb IOM 2 was from Immunotech (Marseille, France). Dichlorotriazinyl-amino-fluorescein (DTAF)-conjugated goat anti-mouse immunoglobulin G was purchased from Dianova (Hamburg, Germany). Isotype control antibodies were purchased from Sigma (Deisenhofen, Germany).

Isolation of cells. Peripheral blood mononuclear cells (PBMC) were obtained from healthy volunteers by density gradient centrifugation (4) on Ficoll-Isopaque (Pharmacia, Freiburg, Germany). After repeated washing in Hanks balanced salt solution (Biochrom, Berlin, Germany), monocytes were isolated by counterflow elutriation with the JE-6B elutriator system (Beckman Instruments, Palo Alto, Calif.) as described previously (17). The cell preparations were >94% monocytes as determined by immunofluorescence staining with the anti-CD14 MAb IOM 2.

Binding assay. PBMC (10⁶) were resuspended in 500 µl of RPMI 1640 supplemented with 10% (vol/vol) human serum and treated with LPS, free lipid A, or compound 406 at the indicated concentrations for 1 h at 4°C. After three washing steps (150 × g for 10 min), the cells were resuspended in 100 µl of ice-cold azide-phosphate-buffered saline (PBS) containing the respective anti-LPS MAbs and incubated at 4°C for 20 min. Prior to use, the optimal concentration of each antibody was determined. After being washed with 500 µl of ice-cold azide-PBS, the cells were incubated with the secondary DTAF-conjugated antibody in 100 µl of ice-cold azide-PBS for 20 min at 4°C. Following a further washing step with 500 µl of ice-cold azide-PBS, cells were resuspended in 400 µl of azide-PBS. The binding of the antibodies was determined by flow cytometry analysis with a Cytofluorograf cytometer (System 50H; Ortho Diagnostic Systems, Inc., Westwood, Mass.). For analysis, monocytes were gated, and the percentage of positive cells and the mean fluorescence intensity (MFI) were determined. Unlabeled cells were used as a control. Results for one of three independent experiments with different donors and similar results are shown. The results are expressed as MFI or as number of positive gated cells. The LPS binding to monocytes is not affected by lymphocytes (unpublished observations and data not shown).

Internalization assay. For the internalization assay, 15 × 10⁶ PBMC (for immunocytochemistry, 15 × 10⁶ monocytes) were pulsed with 30 ng of LPS or LPS partial structure per ml in 5 ml of RPMI 1640 supplemented with 10% human serum for 1 h at 4°C. The cells were washed three times to remove unbound compounds and then further incubated in medium for 1 to 6 h, one portion at 4°C and another at 37°C. Afterwards, the cells were labeled with specific antibodies to detect the membrane-bound compound as described for the binding assay for flow cytometry or prepared for immunocytochemistry. In the latter case, cytospin preparations were fixed in acetone for 15 min, followed by fixation in chloroform. After 15 min, the cytospin preparations were incubated with the specific MAb for 30 min and immunostained according to the alkaline phosphatase-anti-alkaline phosphatase method with new fuchsin development (10). Slides were counterstained with hematoxylin and mounted. All immunostained samples were controlled by the development of alkaline phosphatase alone to exclude staining due to endogenous enzyme activity and the use of murine primary control MAb. All control assays consistently yielded negative results and are not mentioned further. For flow cytometry analysis, the results are presented as the percent binding of the specific antibodies (MFI after pulsing with compound × 100/MFI after further incubation = percent binding). Means ± standard deviations of three independent experiments with different donors are shown.

TNF- release. Cell cultures were performed in 1 ml of RPMI 1640 medium (Biochrom) supplemented with 10% heat-inactivated fetal calf serum (Bioconcept, Umkirch, Germany) and 2 mM L-glutamine (Biochrom). PBMC (4 × 10⁶) were cultured in Nunclon polystyrene six-well plates (Nunc, Roskilde, Denmark) in the presence or absence of 30 ng of different LPS chemotypes per ml for 1 to 6 h at 37°C in a humidified atmosphere with 5% CO₂. The supernatants were harvested to determine TNF- secretion. The concentration of TNF- in supernatants was determined with enzyme-linked immunosorbent assay reagents, kindly provided by H. Gallati (Hoffmann-La Roche, Basel, Switzerland). The assay was carried out as recommended by the manufacturer and as described by Gallati (15). Means of three independent experiments with different donors are shown. The standard deviation in each experiment was less than 10%.

Determination of LPS-induced chemiluminescence (CL). For CL measurements, human PBMC were adjusted to 4 × 10⁵/ml in CL medium (Dulbecco modified Eagle medium for CL; Boehringer-Mannheim, Mannheim, Germany) supplemented with 10% heat-inactivated fetal calf serum, put into polystyrene tubes at 300 µl/tube, and incubated for 1 h at 37°C. Afterwards, samples were put into the six-channel measuring device (Biolumat LB 9505; Bertold, Wildbad, Germany), in which the temperature of the measuring chambers was 37°C. Five minutes prior to CL measurement, 10 µl of luminol (2 mg of 5-amino-2,3-dihydro-1,4-phthalazindione per ml; Boehringer-Mannheim) was added to each sample as a CL-mediating compound. CL was induced by adding 10 µl of a stock solution of each compound, resulting in a final concentration of 30 ng/ml. Results of a typical experiment are shown.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Dose dependence of the binding of LPS and LPS partial structures to human monocytes. PBMC were treated with increasing doses (0.3 ng/ml to 1 µg/ml) of compound 406, lipid A, Re LPS, Rd1 LPS, or Ra LPS for 1 h at 4°C in the presence of human serum. Membrane-bound compounds were then detected by the use of MAbs in a Cytofluorograf cytometer. Figure 2A shows the percentage of monocytes positive for LPS or lipid A, whereas Fig. 2B indicates the MFI of the gated monocytes, corresponding to the amount of bound material. The results show a dose-dependent and saturable binding of each compound. Binding was half saturated at 10 to 30 ng/ml in all cases. Differences in absolute amount of intensity of each compound are due to different affinities of MAbs used.

View larger version (25K): [in this window] [in a new window]   FIG. 2.   Binding of LPS and LPS partial structures to PBMC. Human PBMC were incubated with various doses of compound 406 and different LPS types for 1 h at 4°C. Membrane-bound compounds were detected with specific MAbs and analyzed by flow cytometry.

Internalization of Rd1 LPS. In a preliminary experiment, we examined the internalization rate of Rd1 LPS by human monocytes. First, the cells were treated for 1 h with Rd1 LPS (30 ng/ml), washed extensively, and then incubated at 4 or 37°C for different periods. Finally, the fraction of Rd1 LPS, being still present on the membrane and not internalized, was detected by anti-Rd1 LPS MAb. Figure 3A shows the results of this experiment. At 4°C, the amount of membrane-bound LPS is relatively constant. Over the 6 h of incubation, only a 20% decrease was observed. In contrast, for the cells incubated at 37°C the amount of LPS on the surface decreased with time; after about 3 h, only half of LPS was detectable, and no LPS was detectable after 6 h of incubation. To exclude the possibility of enzymatic digestion of LPS at the cell surface, the experiment was repeated in the presence of 3 µM cytochalasin D (Sigma), which is known as a strong inhibitor of internalization (30). If the decrease in staining at 37°C is caused by enzymatic degradation of LPS at the cell surface, the staining should also disappear at 37°C in the presence of cytochalasin D. Figure 3B shows that this is not the case and that Rd1 LPS is still detectable at the cell surface under these conditions. In order to confirm that the Rd1 LPS was, in fact, internalized and not shed or present in a critical position not accessible to the MAb, internalization was directly visualized by immunocytochemical methods. In contrast to the staining for flow cytometry, staining by immunocytochemical methods detects both membrane-bound and internalized LPS. Figure 4 shows that, in contrast to flow cytometric analysis, the presence of LPS is still detectable within monocytes by immunocytochemical methods after 6 h of incubation at 37°C, indicating the absence of LPS on the surface of the cell but the presence of immunoreactive LPS within the cell. Furthermore, the LPS concentration in the supernatant was found to be unchanged as determined by Limulus amoebocyte lysate assay (data not shown). Taken together, these results show that LPS is not released from cells but quantitatively internalized.

View larger version (13K): [in this window] [in a new window]   FIG. 3.   (A) Internalization of Rd1 LPS. PBMC were pulsed with 30 ng of Rd1 LPS per ml. After several washing steps, the cells were further incubated at 4 or 37°C and at the indicated times stained with specific MAbs and analyzed by flow cytometry. (B) Internalization of Rd1 LPS in the presence of cytochalasin D. PBMC were pulsed with 30 ng of Rd1 LPS per ml in the presence or absence of cytochalasin D. After several washing steps, the cells were further incubated at 37°C with or without cytochalasin D and at the indicated times stained with specific MAbs and analyzed by flow cytometry.

View larger version (130K): [in this window] [in a new window]   FIG. 4.   Immunocytochemical detection of Rd1 LPS. Monocytes were pulsed with 30 ng of Rd1 LPS per ml. After several washing steps, the cells were stained after 6 h of further incubation at 37°C.

Internalization of free lipid A and compound 406. To obtain comparable results with cells from different donors, we used the internalization rate of lipid A as a standard and determined uptake of all other partial structures relative to the internalization kinetics of lipid A. Figure 5 shows the internalization kinetics of compound 406 as well as of lipid A. It was found that lipid A and compound 406 were internalized with similar kinetics. After about 4 to 5 h of incubation, neither lipid A nor compound 406 was detectable on the cellular surface.

View larger version (20K): [in this window] [in a new window]   FIG. 5.   Internalization of compound 406 and free lipid A. PBMC were pulsed with 30 ng of precursor Ia or lipid A per ml. After several washing steps, the cells were further incubated at 37°C and at the indicated times stained with specific MAbs and analyzed by flow cytometry.

Internalization of LPS. Re LPS was analyzed in the same way as lipid A and was found to exhibit a lower internalization rate than lipid A (Fig. 6). The half-time of internalization of lipid A was 3.5 h, and after 6 h, internalization was complete. In contrast, the half-time of internalization of Re LPS was at 5 h, and at the end of the experiment (after 6 h) 30% of the LPS was still detectable.

View larger version (20K): [in this window] [in a new window]   FIG. 6.   Internalization of Re LPS in comparison to that of lipid A. PBMC were pulsed with 30 ng of Re LPS or lipid A per ml. After several washing steps, the cells were further incubated at 37°C and at the indicated times stained with specific MAbs and analyzed by flow cytometry.

View larger version (18K): [in this window] [in a new window]   FIG. 7.   Internalization of Rd1 LPS in comparison to that of lipid A. PBMC were pulsed with 30 ng of Rd1 LPS or lipid A per ml. After several washing steps, the cells were further incubated at 37°C and at the indicated times stained with specific MAbs and analyzed by flow cytometry.

View larger version (19K): [in this window] [in a new window]   FIG. 8.   Internalization of Ra LPS in comparison to that of lipid A. PBMC were pulsed with 30 ng of Ra LPS or lipid A per ml. After several washing steps, the cells were further incubated at 37°C and at the indicated times stained with specific MAbs and analyzed by flow cytometry.

TNF- release by human PBMC. Human PBMC were incubated for 1 to 6 h in the presence of fetal calf serum with 30 ng of LPS or LPS partial structures per ml. The content of TNF- in the supernatants was analyzed by an enzyme-linked immunosorbent assay. Means of three independent experiments with cells of different donors are shown in Fig. 9. As expected, compound 406 did not cause TNF- production by PBMC, but all other compounds exhibited stimulatory capabilities. The capacity to stimulate TNF- was expressed most by Ra LPS followed by Rd1 LPS and Re LPS. Lipid A was the less active preparation. The kinetics of cytokine induction were very similar for Ra LPS and Rd1 LPS, TNF- release reaching a maximum after 5 h of incubation. The kinetics of the TNF- release after stimulation with lipid A and Re LPS, however, seemed to be somewhat faster, reaching a maximum response as soon as 3 h of incubation. This ranking was found in all of the three experiments performed.

View larger version (21K): [in this window] [in a new window]   FIG. 9.   TNF- release by human PBMC after challenge by LPS and partial structures. PBMC were incubated with or without 30 ng of preparations per ml. At the times indicated, supernatants were collected and TNF- content was determined.

Respiratory burst-derived superoxide-anion release. PBMC (1.2 × 10⁵) were incubated with 30 ng of LPS or LPS partial structures per ml, and luminol-dependent (myeloperoxidase-mediated) CL was measured. As shown in Fig. 10, all preparations, except compound 406, induced a respiratory burst. This activity was best expressed by Ra LPS, followed by Rd1 LPS. Both lipid A and Re LPS showed lower activity in their capacity to induce an oxidative burst. The superoxide-anion release started at about 5 min after stimulation and had its half-maximum at about 10 to 15 min after stimulation with Ra or Rd1 LPS or at about 20 min after stimulation with Re LPS or lipid A. 

View larger version (26K): [in this window] [in a new window]   FIG. 10.   CL induced by LPS and partial structures. PBMC were incubated with 30 ng of each preparation per ml. The time course of luminol-enhanced CL was detected.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Interaction of LPS with monocytes/macrophages involves LPS binding to the cells via CD14, internalization of LPS, and cell activation. Many studies have focused on the mechanism of LPS binding and LPS-mediated cellular activation, whereas few reports deal with the process of internalization. LPS internalization by various cell types has been reported elsewhere (18, 23, 27, 36, 37). Bona (3) described the internalization of LPS by human macrophages, and Kang et al. (21) investigated the distribution of LPS in human monocytes. LPS internalization has also been shown previously for human granulocytes (40). Internalization seems to be a process which is not necessary for activation of the cells but is of great importance for clearance of LPS from the circulation. The significance of the chemotype for LPS bioactivity was demonstrated in various systems including complement activation (11), binding to serum (37) or membrane proteins (8), interaction with Limulus-derived factors (25), interactions with antibiotics (33), physicochemical properties (7), plasma clearance (9), and induction of cytokine production in monocytes/macrophages (12, 20, 26). Taken together, these data suggest that lipid A activity is markedly modulated by the glycosyl region.

The present study deals with the relationship among structure, activity, and internalization of various LPS chemotypes and partial structures. Since most of the procedures used to determine LPS internalization do not allow a distinction between membrane-bound and internalized LPS, we have developed a new method for the determination of binding and the internalization rate of LPS. Following preincubation at 4°C and several washing steps, cells are further incubated at 37°C and the disappearance of membrane-bound LPS is determined by staining with MAbs. This method was used to investigate possible interdependences between LPS internalization and cellular activation. While incubation with LPS at 4°C did not result in considerable reduction of surface LPS, an incubation of the cells at 37°C resulted in a drastic disappearance of LPS from the membrane. This process can be blocked by cytochalasin D, demonstrating that the disappearance of the staining is not due to enzymatic degradation of LPS at the cell surface. In the flow cytometry assay, which detects only the membrane-bound LPS, after 6 h of incubation at 37°C no staining is detectable. In the immunocytochemical assay, which detects the internalized LPS also, clear staining is visible, demonstrating that LPS is internalized at this time point. Therefore, the loss of membrane-bound LPS reflects the internalization of LPS by the cells. In contrast, the LPS concentration of the culture supernatant was found to be unchanged as determined by the Limulus amoebocyte lysate assay (data not shown). With two concentrations of lipid A, 30 ng/ml and 1 µg/ml, the same internalization rates were determined, indicating that there is no influence of the concentration on the time kinetics of internalization of lipid A (data not shown). The results presented in this paper show that LPS chemotypes can be arranged in a ranking of the kinetics of internalization as follows: Ra LPS > Rd1 LPS > Re LPS. This suggests an inverse correlation between the internalization rate of the LPS chemotypes and the hydrophobicity of these compounds which is likely to decrease with the length of the sugar chain (7). However, lipid A and compound 406 both show a different behavior. The internalization rate of lipid A and compound 406 is comparable to that of Rd1 LPS, whereas they are not comparable in hydrophobicity. Furthermore, compound 406 possesses less fatty acids than does lipid A. However, they show only small differences in internalization kinetics. We also tested lipid A from S. minnesota, which differs from E. coli lipid A in the lipid structure (additional fatty acids), to investigate the influence of the lipid part on the internalization rate. Lipid A from S. minnesota does not differ from E. coli in internalization kinetics (data not shown). These data demonstrate that for internalization of LPS and partial structures the sugar part is of great importance. Modifications in the lipid part seem to be without influence on the internalization rate. Investigations of structure-activity relationships of LPS during the induction of cytokines have revealed a direct correlation between polysaccharide chain length (hydrophobicity) and cytokine induction (12, 26). This finding may suggest that for internalization as well as for cytokine induction both the physicochemical structure and the hydrophobicity of LPS are of importance. We, therefore, compared the internalization rates of different LPS chemotypes with two functional parameters, TNF- production and the release of oxygen species. Although we found significant differences in the capacities of the LPS chemotypes to induce TNF- or an oxidative burst, the kinetics of the preparations were the same for the different LPS chemotypes. Furthermore, the oxidative burst first appears a few minutes after stimulation and is ended after 1 h, kinetics which are not matched by the internalization process. It thus can be excluded that cellular activation by LPS is modulated by the internalization process.

The greatest advantage of the new method is the use of native compounds, which avoids the false reaction of chemically modified LPS such as fluorescein isothiocyanate-LPS. Furthermore, this system allows the investigation of a variety of different LPS and lipid A structures. It should be noted that the membrane-bound LPS or lipid A is recognized by the MAb used. This indicates that the acyl chains of the lipid A component are membrane associated, whereas the core oligosaccharide as well as the lipid A backbone disaccharide is exposed for recognition by the antibodies. All MAbs used are known to interact with the hydrophilic regions of LPS and lipid A. These data also show that membrane-bound LPS is immunoreactive with anti-LPS MAb on living cells.

Kitchens and Munford (22) compared the internalization rates of LPS in different aggregate sizes by fluorescence quenching and a protease-protecting assay. They found markedly greater internalization rates in THP-1 cells than those with our assay employing monocytes. After a few minutes, the cell-bound LPS was found to be proteinase K resistant or could not be quenched by antifluorescein MAb. Our results demonstrate, however, that LPS bound to monocytes is available for binding to anti-LPS MAb for a longer time, depending on the chemotype used. This indicates that LPS after binding to the cell surface rapidly changes its location but is still detectable with anti-LPS MAbs on the surface. Whether LPS is bound to a second proteinase K-resistant membrane receptor or is intercalated into the phospholipid bilayer remains to be investigated.

Taking the results together, we were able to demonstrate that the kinetics of LPS internalization vary among the different chemotype structures and that for total internalization of LPS up to 6 h of incubation is necessary. However, a structure-effect relationship exists between the composition of sugars and the internalization rate of LPS or lipid A. The internalization rate of compound 406 is similar to that of lipid A. It therefore can be concluded that the lack of biologic activity of compound 406 is not caused by a slower or a missing uptake of compound 406. A correlation between internalization and activation of the cells was not found. The biophysicochemical background for the different behaviors of LPS and lipid A during internalization remains to be investigated.