Abstract
The supernatants taken from Pseudomonas aeruginosa and Escherichia coli cultures in human sera or chemically defined M9 medium in the presence of ceftazidime (CAZ) contained high levels of endotoxin, while those taken from the same cultures in the presence of imipenem (IPM) yielded a very low level of endotoxin. The biological activities of endotoxin in the supernatants were compared with those of phenol water-extracted lipopolysaccharide (LPS). The endotoxin released from the organisms as a result of CAZ treatment (CAZ-released endotoxin) contained a large amount of protein. The protein, however, lacked endotoxic activity, since the endotoxin did not show any in vivo toxic effects in LPS-hyporesponsive C3H/HeJ mice sensitized with d-(+)-galactosamine (GalN) or any activation of C3H/HeJ mouse macrophages in vitro. The activities of CAZ- and IPM-released endotoxin (as assessed by a chromogenic Limulus test) were fundamentally the same as those of P. aeruginosa LPS, since their regression lines were parallel. The CAZ-released endotoxin was similar to purified LPS with respect to the following biological activities in LPS-responsive C3H/HeN mice and LPS-hyporesponsive C3H/HeJ mice: lethal toxicity in GalN-sensitized mice, in vitro induction of tumor necrosis factor- and NO production by macrophages, and mitogen-activated protein kinase activation in macrophages. The macrophage activation by CAZ-released endotoxin as well as LPS was mainly dependent on the presence of serum factor and CD14 antigen. Polymyxin B blocked the activity. These findings indicate that the endotoxic activity of CAZ-released endotoxin is due primarily to LPS (lipid A).
Endotoxin or bacterial lipopolysaccharide (LPS) causes various inflammatory symptoms and pathophysiological disorders, including fever, disseminated intravascular coagulation, multiple organ failure, and septic shock (30, 33). The septic shock induced by bacteremia caused by gram-negative bacteria is thought to be due to the massive release of endotoxin from infecting organisms by spontaneous release or bacterial lysis. Sometimes, patients treated with effective antibiotics may succumb to shock due to the endotoxin released from the killed bacteria (10, 36). LPS released from antibiotic-treated bacteria induces various proinflammatory activities that contribute to septic shock syndrome (4, 20).
Endotoxin or LPS is a component of the outer membrane of gram-negative bacteria, and cell wall-active antibiotics (such as β-lactams) are considered to be most responsible for the liberation of excess amounts of endotoxin. The amounts of endotoxin released from bacteria by antibiotics can vary depending on the bacterial strains, the types and efficacies of the antibiotics, the concentrations of the antibiotics, the length of time that the bacteria are exposed to the antibiotic (9, 14), and the presence of antibody and/or other serum constituents that can interact with endotoxin (18). Among the β-lactam antibiotics, the capabilities for releasing endotoxin vary greatly depending on the antibiotics used, and the difference can partly be explained by the binding affinities of the antibiotics to the different kinds of penicillin-binding proteins (PBPs). The binding to PBP inhibits the synthesis of the cell walls of bacteria and induces bacterial lysis. In Escherichia coli and Pseudomonas aeruginosa, inhibition of PBP 1 is associated with rapid killing and lysis, whereas inhibition of PBP 2 produces spherical cells that do not grow and inhibition of PBP 3 produces long filamentous cells (9, 14, 31, 42).
Endotoxin has many different pathophysiological activities both in vivo and in vitro (29). Its active component is LPS, a complex macromolecule that contains lipid A covalently linked to polysaccharide. Endotoxin or LPS is chemically extractable, and the hot phenol-water procedure (45) is an excellent means of extracting LPS containing extensive O-antigen repeating subunits (i.e., smooth strains). Most information on the bioactivity of endotoxin has been accumulated from experiments with purified LPS. However, only a few reports have dealt with comparative biological characterizations of the endotoxin released from organisms as a result of antibiotic treatment (antibiotic-released endotoxin) and chemically extracted LPS. In the present study, we therefore compared the biological activities of β-lactam antibiotic-released endotoxin with those of phenol-extracted LPS.
MATERIALS AND METHODS
Bacterial strains.
P. aeruginosa PAO1 and E. coli O55 were used throughout the experiments. The organisms were grown in heat-inactivated human serum obtained from healthy volunteers or in defined M9 medium (2) at 37°C for the appropriate lengths of time. The viable numbers of organisms were determined by quantitative cultivation on nutrient agar plates. The killed bacteria were prepared by exposing the bacteria to 0.05% glutaraldehyde in phosphate-buffered saline (PBS) at room temperature for 5 min. The organisms were washed four times with PBS containing 100 mM glycine and were resuspended in PBS.
Mice.
C3H/HeN and C3H/HeJ mice were bred and maintained under standard conditions in the Animal Facility of the Jichi Medical School. Female mice (age, 9 to 11 weeks) were used. Age-matched mice were used in individual experiments.
Infection.
The organisms were grown in M9 medium for 18 h and were then resuspended in saline. The number of bacteria in the suspension was adjusted optically. The bacterial suspension (0.2 ml/mouse) was injected intraperitoneally (i.p.) into the mice.
Antibiotics.
Ceftazidime (CAZ) was obtained from Glaxo Japan Co., Tokyo, Japan; ofloxacin (OFLX) and levoflaxacin (LVFX) were from Dai-ichi Pharmaceutical Co., Tokyo, Japan. Imipenem (IPM) was provided by Merck & Co., Rahway, N.J. Polymyxin B was purchased from Sigma Chemical Co., St. Louis, Mo. The antibiotics were dissolved in accordance with the manufacturers’ recommendations. Aliquots of the antibiotic solutions were stored at −80°C and were thawed immediately before use. During the study period, the antibiotics did not lose antimicrobial activity, as measured by the MICs for E. coli and P. aeruginosa.
Antibiotic-released endotoxin and phenol-extracted LPS.
The organisms were cultured in the presence or absence of antibiotic in heat-inactivated human serum or in M9 medium at 37°C for the indicated times. At the end of incubation, the culture supernatants were filtered through a membrane filter (Sterifil D-GV; pore size, 0.22 μm; Millipore Corp., Bedford, Mass.), and the filtrates were used as a source of endotoxin. Phenol-extracted E. coli O111:B4 LPS and P. aeruginosa (Fisher-Devlin immunotype 1; ATCC 27312) LPS were obtained from List Biological Laboratories, Inc., Campbell, Calif. Ra-chemotype LPS (Ra-LPS) was extracted with phenol-chloroform petroleum ether from Salmonella minnesota R60 and was obtained from List.
Estimation of endotoxic activity of antibiotic-released endotoxin.
The endotoxin obtained from bacteria grown in serum, but not from bacteria grown in M9 medium, was treated with an alkaline solution (41) to prevent the inhibition of the coagulation reaction of Limulus amoebocyte lysate (LAL). The endotoxic activities were estimated by a chromogenic endotoxin-specific assay with LAL coagulation enzyme (ES test; Seikagaku Corp., Tokyo, Japan), and endotoxin levels were calculated for comparison with the reference endotoxin levels. Linear regression analyses (32) were performed on the LAL coagulation data derived from dilutions of CAZ- and IPM-released endotoxin and compared with the data derived from phenol-water-extracted LPS. The endotoxin level in the M9 medium was 4.4 ± 0.4 pg/ml (0.013 ± 0.01 endotoxin units [EU]/ml).
Determination of endotoxin protein content.
The protein content of the endotoxins was determined by the DC Protein Assay (Bio-Rad Laboratories, Hercules, Calif.).
Lethality studies.
Mice were injected i.p. with 0.2 ml of the endotoxin or diluted phenol-water-extracted LPS and 0.5 ml of d-(+)-galactosamine (GalN) solution (18 mg/mouse). In other experiments, mice were injected i.p. with 0.2 ml of saline or antibiotics followed immediately by i.p. inoculation with 0.2 ml of the P. aeruginosa suspension. Mortality was scored at 48 h after the challenge, and the 50% lethal dose (LD50) was calculated.
Cell preparation.
Mouse peritoneal exudate cells were isolated by peritoneal lavage at 4 days after i.p. injection of 1.5 ml of sterile 3% Brewer thioglycolate medium (Difco Laboratories, Detroit, Mich.). The cells were washed with RPMI 1640 (Flow Laboratories, Irvine, Scotland), resuspended in RPMI 1640 containing 2% heat-inactivated fetal bovine serum (FBS; Summit Biotechnology Inc., Ft. Collins, Colo.), 4 mM l-glutamine, 100 U of penicillin per ml, and 100 μg of streptomycin per ml, and plated at 2 × 105 cells per well in 96-well plates. After 2 h of incubation at 37°C in 5% CO2–air, the cells were washed three times with serum-free RPMI 1640 to remove nonadherent cells. The adherent cells were used as macrophages.
Mouse macrophage-like J774.1 cells were kindly provided by T. Suzuki, University of Kansas Medical Center. An LPS-resistant and CD14-negative mutant cell line, the J7.DEF.3 cell line, was derived from J774.1 cells as described in a previous paper (16). These cells were grown in RPMI 1640 containing 8% FBS, 4 mM l-glutamine, 100 U of penicillin per ml, and 100 μg of streptomycin per ml at 37°C in 5% CO2–air. The cells were collected, washed once with culture medium, and plated in 96-well culture plates at 2 × 105 cells per well in 200 μl of RPMI 1640 containing 2% FBS.
Cell cultures and the culture supernatant.
The cells were cultured in the presence or absence of endotoxin or LPS in individual experiments for the times indicated in the figure legends. If necessary, the culture supernatants were collected at the end of incubation and were stored at −80°C until tumor necrosis factor (TNF), interleukin 6 (IL-6), and NO levels were determined.
TNF assay.
TNF-sensitive L929 cells were donated by M. J. Parmely of the University of Kansas Medical Center. TNF activity was determined by a functional cytotoxic assay with a TNF-sensitive cell line, L929, as described previously (16).
IL-6 assay.
IL-6-dependent hybridoma B13.29 cells were donated by W. P. Zeiljemaker of the University of Amsterdam. The cells were maintained in 8% FBS–RPMI 1640 supplemented with 50 μM mercaptoethanol and 20 U of IL-6 (Boehringer Mannheim GmbH, Mannheim, Germany) per ml. IL-6 activity was determined with B13.29 cells as described previously (19).
NO assay.
NO formation was measured as the stable end product nitrite (NO2−) in culture supernatants with the Griess reagent (12) as described previously (16).
Immunoblot analysis of phosphotyrosine and MAP kinases.
Macrophages at a density of 7 × 105 cells/60-mm plate were incubated for 15 min in 2% FBS–RPMI 1640 containing various stimulants. After stimulation, the cells were washed three times with PBS containing 1 mM Na3VO4 and were lysed for 30 min on ice with 200 μl of lysis buffer (20 mM Tris-HCl [pH 8.0], 137 mM NaCl, 2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 20 μM leupeptin, 1 mM Na3VO4, 10% glycerol, and 1% Triton X-100). The cell lysates were centrifuged (10,000 × g, for 10 min) to remove insoluble material. Aliquots of the cell lysate (20 μg protein) were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and were transferred onto polyvinylidene difluoride filters (Immobilon transfer membranes; Millipore Corp.). The membranes were probed with a monoclonal antibody (MAb; MAb PY-20) against phosphotyrosine (Transduction Laboratories, Lexington, Ky.) or an MAb (MAb Z033) against mitogen-activated protein (MAP) kinases 1 and 2 (Seikagaku Corp.), incubated with horseradish peroxidase-conjugated anti-immunoglobulin G antibodies (Amersham), and visualized with an enhanced chemiluminescence Western blotting detection system.
Statistical analysis.
Statistical significance was determined by Student’s t test for TNF, IL-6, and NO levels and the χ2 test for mortality.
RESULTS
Antibiotic-induced release of endotoxin from organisms cultured in human serum or M9 medium.
The MICs for P. aeruginosa PAO1 were 1 μg/ml for CAZ in M9 medium and in human serum, 0.5 μg/ml for IPM in M9 medium, 2 μg/ml for IPM in serum, 2.5 μg/ml for OFLX in serum, and 1 μg/ml for LVFX in serum. P. aeruginosa organisms were inoculated into serum (105 organisms/ml) and were cultured in the presence of 2× the MIC of CAZ, OFLX, LVFX, or IPM at 37°C for 8 h. At the end of the incubation the endotoxin levels in the culture supernatants were determined. The supernatants of the bacterial cultures treated in the presence of CAZ, OFLX, and LVFX contained approximately 300 ng of endotoxin per ml, while those treated in the presence of IPM contained less than 50 ng/ml.
In E. coli and P. aeruginosa CAZ binds preferentially to PBP 3, while IPM binds with the highest affinity to PBP 2 (7, 33). The growth kinetics for P. aeruginosa and E. coli organisms in chemically defined M9 medium and endotoxin release in the presence or absence of CAZ or IPM are presented in Fig. 1. The numbers of CFU of the organisms cultured with 2× the MIC of CAZ or IPM decreased quickly and did not increase thereafter. However, the numbers of CFU of the organisms exposed to 0.5× the MICs of these antibiotics decreased initially but thereafter increased gradually (Fig. 1B and D). Bacterial growth in antibiotic-free M9 medium was accompanied by the spontaneous release of endotoxin from these organisms. Large amounts of endotoxin were released from the cultures with 0.5× and 2× the MIC of CAZ and 0.5× the MIC of IPM as well as from the control cultures, while small amounts of endotoxin were liberated from cultures with 2× the MIC of IPM (Fig. 1A and C). Microscopic observation indicated that the organisms cultured with 2× the MICs CAZ and IPM in M9 medium appeared as long filamentous structures and round structures, respectively (data not shown). These findings coincide with those of other investigators (7, 14), who showed that IPM kills these organisms but with lower levels endotoxin being released than the amount released after killing by CAZ.
FIG. 1.
Kinetics of bacterial growth and antibiotic-induced release of endotoxin from P. aeruginosa and E. coli incubated in synthetic M9 medium. The organisms were cultured in M9 medium in the presence or absence of antibiotics; 0.5× the MIC of CAZ (▵) or IPM (▴), 2× the MIC of CAZ (□) or IPM (▪), 4× the MIC of CAZ (⊠), and no antibiotic (○).
When these organisms were suspended in PBS (instead of M9 medium) in the presence or absence of 2× the MIC of CAZ and incubated for 8 to 25 h, the organisms could not proliferate, regardless of the presence or absence of CAZ (Fig. 2B and D), and the culture supernatants contained only traces of endotoxin (Fig. 2A and C). These findings indicate that endotoxin release does not occur when nutrients are not available for the bacteria.
FIG. 2.
Bacterial growth and antibiotic-induced release of endotoxin from P. aeruginosa and E. coli incubated in M9 medium or PBS. The organisms were cultured in M9 medium (○ and □) or PBS (• and ▪) in the presence of 2× the MIC of CAZ (□ and ▪) or with no drug (○ and •).
Protein content of CAZ-released endotoxin.
CAZ-released endotoxin contained not only LPS but also large amounts of protein. The protein contents in the supernatants of P. aeruginosa and E. coli cultures exposed to 2× the MIC of CAZ for 8 h in M9 medium were 11.8 and 15.3 mg per mg of LPS, respectively. The endotoxin spontaneously released from organisms cultured for 8 h in antibiotic-free M9 medium also contained a large amount of protein (7.7 mg per mg of LPS). The protein detected in the samples described above was therefore more likely derived from the organisms since synthetic M9 medium was used for culturing of the bacteria. On the other hand, phenol-water-extracted LPS preparations from P. aeruginosa and E. coli contained less than 5 and 195 μg of protein per mg of LPS, respectively.
Chromogenic LAL coagulation activities of various LPS preparations.
To examine whether the activities of CAZ- and IPM-released endotoxins were similar to those of phenol-extracted LPS, the activities of serially diluted samples were assessed by the ES test (an LPS-specific chromogenic LAL test [32]) and their regression lines were compared on a bilogarithmic scale (Fig. 3). Data obtained from the ES test exhibited good linearity relating endotoxin concentration and absorbance, with parallel lines obtained for concentration and absorbance. These findings suggest that these endotoxins (including spontaneously released endotoxin) have endotoxic activity or physiochemical properties that suggest that they have activities similar to those of phenol-extracted LPS (with regard to LAL activity). In other words, other endotoxin-like activities, such as protein-binding lipid A (26, 27), appear to be negligible in these endotoxin preparations.
FIG. 3.
Regression lines of the ES test with antibiotic-induced endotoxins. P. aeruginosa (closed symbols) or E. coli (open symbols) was cultured in M9 medium in the presence of 2× the MIC of CAZ (• and ○) or IPM (▴ and ▵) or with no drug (▪ and □) for 8 h. As controls, the activities of standard solutions containing known amounts of E. coli O111:B4 LPS (⊙) and P. aeruginosa LPS (⊠) were assessed.
Lethal toxicity of CAZ-released endotoxin in mice.
It is well-known that GalN-sensitized mice are markedly susceptible to the lethal effects of LPS (12). The relative lethalities of antibiotic-released endotoxin for LPS-responsive C3H/HeN mice and LPS-hyporesponsive C3H/HeJ mice treated with GalN were compared. GalN (18 mg/mouse administered i.p.)-treated mice were challenged i.p. with purified P. aeruginosa LPS or antibiotic-released endotoxin from P. aeruginosa cultured in synthetic M9 medium in the presence of 2× the MIC of CAZ for 8 or in the presence of 2× the MIC of CAZ for 24 h in M9 medium. The results are presented in Table 1. The LD50s of CAZ-released endotoxin from an 8-h culture for C3H/HeN and C3H/HeJ mice were 10 and >1,000 ng, respectively, while the LD50s of CAZ-released endotoxin from a 24-h culture were 65 and >1,000 ng, respectively. The LD50s of purified LPS for C3H/HeN and C3H/HeJ mice were 32 ng and >10,000 ng, respectively. These findings suggest that the lethal toxicity of CAZ-released endotoxin is similar to that of LPS in the LPS-sensitive strain of mice. These data also suggest that the CAZ-released endotoxin preparation contains little bioactive lipid A-associated protein, since it does not show any lethal effect in C3H/HeJ mice which are sensitive to protein-binding lipid A (26, 27), while they are more resistant to LPS.
TABLE 1.
Lethal toxicity of CAZ-released endotoxin in GalN-sensitized C3H/HeN and C3H/HeJ micea
| P. aeruginosa endotoxinb | Dose (ng/mouse) | No. of survivors/total no. of mice (%)
|
|
|---|---|---|---|
| C3H/HeN mice | C3H/HeJ mice | ||
| LPS | 10,000 | —c | 4/4 (100) |
| 100 | 0/4 (0) | — | |
| 10 | 4/4 (100) | — | |
| CAZ-endotoxin-8 h | 1,000 | — | 4/4 (100) |
| 100 | 0/4 (0) | 4/4 (100) | |
| 10 | 4/8 (50) | — | |
| 1 | 4/4 (100) | — | |
| CAZ-endotoxin-24 h | 1,000 | 0/4 (0) | 4/4 (100) |
| 100 | 3/8 (38) | — | |
| 10 | 4/4 (100) | — | |
C3H/HeN and C3H/HeJ mice were injected i.p. with endotoxin preparations at the indicated doses and with 18 mg of GalN.
CAZ-endotoxin-8 h and CAZ-endotoxin-24 h, endotoxins from bacteria exposed to 2× the MIC of CAZ for 8 and 24 h, respectively.
—, not determined.
Production of TNF, IL-6, and NO by macrophages stimulated with CAZ-released endotoxin.
Abundant current evidence supports the existence of host-derived proinflammatory effector molecules produced by LPS-activated macrophages and other host cells. These molecules are responsible for the detectable systemic inflammatory responses observed after the administration of LPS to humans and experimental animals. TNF, IL-6, and NO are thought to be the three molecules most responsible for the pathophysiological manifestations of septic shock. We therefore examined the in vitro production of TNF, IL-6, and NO by peritoneal macrophages of LPS-responsive C3H/HeN or LPS-hyporesponsive C3H/HeJ mice in response to exposure to CAZ-released P. aeruginosa endotoxin (Fig. 4 and 5). C3H/HeN mouse macrophages cultured in the presence of serum produced TNF, IL-6, and NO when they were stimulated with low doses of CAZ-released endotoxin as well as LPS (Fig. 4). In the absence of serum, however, a relatively large dose of endotoxin was required to induce the production of similar cytokine levels. These findings suggest that both CAZ-released endotoxin and LPS activate C3H/HeN mouse macrophages and that this activation depends predominantly on serum components, such as LPS-binding protein (LBP) (38) and septin (46). On the other hand, neither CAZ-released P. aeruginosa endotoxin (data not shown), CAZ-released E. coli endotoxin (data not shown) nor LPS (Fig. 5) induced the effective production of TNF and NO by LPS-hyporesponsive C3H/HeJ mouse macrophages. These findings also suggest that the endotoxic activity of the protein-lipid A complex in CAZ-released endotoxin is negligible.
FIG. 4.
Serum dependency of TNF, IL-6, and NO production by C3H/HeN macrophages of C3H/HeN mice stimulated with phenol-water-extracted LPS or antibiotic-released endotoxin from P. aeruginosa (P.a.). Thioglycolate-elicited macrophages were cultured in the presence (•) or absence (○) of 2% FBS with LPS or CAZ-released endotoxin that had been prepared from bacteria exposed to 2× the MIC of CAZ for 8 h. Culture supernatants were collected at 4 h for the TNF assay (A and B), 16 h for the IL-6 assay (C and D), and 48 h for the NO assay (E and F). Each point represents the mean ± standard error of the mean. The data represent the results of one of two or three experiments with similar results.
FIG. 5.
TNF, IL-6, and NO production by LPS-responsive C3H/HeN mouse macrophages and LPS-hyporesponsive C3H/HeJ mouse macrophages stimulated with antibiotic-released endotoxin. Thioglycolate-elicited macrophages from C3H/HeN mice (•) or C3H/HeJ mice (○) were cultured with P. aeruginosa (P.a.) LPS (A, C, and E) or endotoxin from bacteria exposed to 2× the MIC of CAZ for 8 h (B, D, and F) in the presence of 2% FBS. The culture supernatants were collected at 4 h for the TNF assay (A and B), 16 h for the IL-6 assay (C and D), and 48 h for the NO assay (E and F), and the activities were assessed. Each point represents the mean ± standard error of the mean. The data represent the results of one of two experiments with similar results.
Polymyxin B neutralizes the activity of LPS by binding to the lipid A moiety of LPS (15, 27). The ability of CAZ-released endotoxin to induce TNF production by C3H/HeN mouse macrophages was blocked by polymyxin B in a dose-dependent fashion (Table 2). These findings therefore suggest that lipid A is the active molecule released by CAZ.
TABLE 2.
Inhibition of antibiotic-released endotoxin activity by polymyxin Ba
| P. aeruginosa endotoxin | Dose of polymyxin B (ng/ml) | TNF activity (U/ml [mean ± SEM]) |
|---|---|---|
| LPS | 0 | 211 ± 10 |
| 5 | 65 ± 5 | |
| 50 | 11 ± 1 | |
| 500 | <10 | |
| CAZ-released endotoxin | 0 | 533 ± 72 |
| 5 | 58 ± 5 | |
| 50 | <10 | |
| 500 | <10 |
LPS (10 ng/ml) or endotoxin (endotoxin from bacteria exposed to 2× the MIC of CAZ for 8 h; 10 ng/ml) solutions were incubated with various doses of polymyxin B. The solutions were added to C3H/HeN mouse macrophage cultures 30 min later. The macrophages were incubated for 4 h, and the TNF activity in the culture supernatants was determined by a cytotoxicity assay. The data represent the results of one of two experiments with similar results.
Activation of protein kinases in macrophages by CAZ-released endotoxin.
Protein phosphorylation was found to be the initial step of macrophage activation by LPS (3, 39, 43). By immunoblot analysis the activation of phosphotyrosine kinases was detected in C3H/HeN mouse macrophages but not in C3H/HeJ mouse macrophages. When C3H/HeN mouse macrophages were stimulated with CAZ-released Pseudomonas endotoxin as well as Pseudomonas or Salmonella LPS, proteins of approximately 42 kDa were phosphorylated (Fig. 6A). The phosphorylation, however, was not obvious when the macrophages were stimulated with killed P. aeruginosa organisms. LPS stimulates the tyrosine phosphorylation of the MAP kinases p44, p42, and p41 (1). The activation of MAP kinases 1 and 2 in response to CAZ-released endotoxin was detected in C3H/HeN mouse macrophages but not C3H/HeJ mouse macrophages (Fig. 6B).
FIG. 6.
Induction of protein tyrosine phosphorylation and tyrosine phosphorylation of MAP kinases in LPS-responsive C3H/HeN mouse macrophages and LPS-hyporesponsive C3H/HeJ mouse macrophages stimulated with CAZ-released endotoxin from an 8-h culture. Macrophages from C3H/HeN mice (lanes 1 to 7) or C3H/HeJ mice (lanes 8 to 14) were stimulated for 15 min with the following stimulants: RPMI 1640 (lanes 1 and 8), M9 medium (lanes 2 and 9), endotoxin from P. aeruginosa exposed to 2× the MIC of CAZ for 8 h (290 ng/ml; lanes 3 and 10), endotoxin from E. coli exposed to 2× the MIC of CAZ for 8 h (390 ng/ml; lanes 4 and 11), P. aeruginosa LPS (270 ng/ml; lanes 5 and 12), glutaraldehyde-killed P. aeruginosa (5 × 105/ml; lanes 6 and 13), and phenol-chloroform-petroleum ether-extracted Ra-LPS (1 μg/ml) plus mouse recombinant gamma interferon (20 U; Shionogi Pharmaceutical Co., Osaka, Japan) (lanes 7 and 14). Macrophage expression of protein tyrosine phosphorylation and tyrosine phosphorylation of MAP kinases 1 and 2 was detected by Western blot analysis with MAb PY-20 (A) and MAb Z033 (B), respectively. p-MAP1 and pMAP2, tyrosine-phosphorylated MAP kinases 1 and 2, respectively.
CD14 dependency on macrophage activation by CAZ-released endotoxin.
The response to LPS confirmed the important role of the cell surface form of CD14 (mCD14) (47). However, a pathway of LPS-induced cell activation that does not require CD14 has been suggested (16). The CD14 dependency of CAZ-released P. aeruginosa endotoxin was examined with a CD14-positive murine macrophage-like cell line (J774.1 cells) and its LPS-resistant mutant (J7.DEF3 cells) (18, 19). J7.DEF3 cells do not express CD14 antigen (unpublished observation). Parental J774.1 cells in the presence of FBS produced TNF in response to low doses of both CAZ-released endotoxin and LPS, while the mutant J7.DEF3 cells produced TNF only when stimulated with a higher dose of the endotoxin or LPS (Fig. 7). Under serum-free conditions, both parental and mutant strains manifested only minimal responses to even very high doses of the CAZ-released endotoxin or LPS. These findings indicate that both CAZ-released endotoxin and LPS are mainly recognized by the receptors on macrophages after they are complexed with a serum factor(s), presumably LBP or CD14 antigen.
FIG. 7.
TNF production by J774.1 cells and J7.DEF3 cells stimulated with LPS or endotoxin. J774.1 cells (•) and J7.DEF3. cells (○) were incubated with P. aeruginosa (P.a.) LPS or endotoxin (endotoxin from bacteria exposed to 2× the MIC of CAZ for 8 h) in the presence (A and B) or absence (C and D) of 2% FBS for 4 h. The data represent the results of one of two experiments with similar results.
Lethal toxicity of in vivo antibiotic-released endotoxin.
CAZ, but not IPM, induced the release of relatively large amounts of endotoxin from organisms in vitro (Fig. 1). Table 1 illustrates the relative lethalities of CAZ-released endotoxin and purified LPS and shows that maximum release of endotoxin has occurred within 24 h of CAZ treatment in C3H/HeN mice. If CAZ treatment induces the in vivo release of massive amounts of endotoxin from the infecting organisms, the treatment may have some harmful effects on the host. C3H/HeN and C3H/HeJ mice were inoculated with P. aeruginosa organisms (3 × 106 and 6 × 106 CFU/mouse i.p., respectively), treated with CAZ (20 mg/kg) or IPM (20 mg/kg), and then sensitized with GalN. As indicated in Table 3, approximately one-half of the mice in the control groups that had been infected but that had not received antibiotic treatment died within 48 h. All C3H/HeN mice treated with CAZ died within 48 h, while those mice treated with IPM survived. On the other hand, 44% of the LPS-hyporesponsive C3H/HeJ mice survived after treatment with CAZ. No C3H/HeJ mice treated with IPM died. These results suggest that the endotoxin released by antibiotics may play a role in the death of the hosts.
TABLE 3.
Effects of IPM and CAZ on lethality for GalN-sensitized C3H/HeN and C3H/HeJ mice infected with P. aeruginosaa
| Mouse strain | Treatment (dose [mg/kg]) | No. of survivors/total no. of mice (%) |
|---|---|---|
| C3H/HeN | Saline | 5/10 (50) |
| IPM (20) | 8/8 (100) | |
| CAZ (20) | 0/10 (0) | |
| C3H/HeJ | Saline | 9/16 (56) |
| IPM (20) | 16/16 (100) | |
| CAZ (20) | 7/16 (44) |
Groups of C3H/HeN and C3H/HeJ mice were injected i.p. with CAZ (20 mg/kg), IPM (20 mg/kg), or saline and were immediately inoculated i.p. with P. aeruginosa organisms (3 × 106 and 6 × 106 CFU/mouse, respectively); the mice were injected i.p. with 18 mg of GalN 105 min later.
DISCUSSION
Recently, many studies have pointed out that the sudden release of endotoxin from gram-negative bacteria killed by antibiotics may have harmful effects on the hosts (for a review, see reference 7). However, the amounts of antibiotic-liberated endotoxin seem to depend at least partly on the PBP binding affinity of these cell wall-active antibiotics: IPM (which binds most strongly to PBP 2 of the organisms) is known to induce the release of less endotoxin than the amount induced by other β-lactam antibiotics, which bind mainly to PBP 3 (7, 14). The amount of endotoxin released by P. aeruginosa organisms cultured in human serum in the presence of IPM, which strongly binds to PBP 2, was comparatively less than that released in the presence of other antibiotics (such as CAZ, OFLX, and LVFX). The spontaneous release of endotoxin without antibiotic treatment was also observed when P. aeruginosa or E. coli organisms were incubated in chemically defined M9 medium (Fig. 1). Endotoxin release from bacteria must be required for energy consumption, since no spontaneous release was observed when the organisms were suspended in carbon source- and nitrogen source-free PBS, even if CAZ was present (Fig. 2) in the PBS. The antibacterial effects of IMP seemed to be similar to those of CAZ (Fig. 1B and D). However, typical differences in the amount of endotoxin release induced by CAZ and IPM were seen when the organisms were cultured in the presence of growth-inhibiting drug doses (2× the MIC) in M9 medium (Fig. 1A and C). CAZ-treated organisms may continue to synthesize the cell wall and LPS incompletely, resulting in death; on the other hand, IPM-treated organisms may cease these syntheses quickly, resulting in the release of less endotoxin (7).
Endotoxin-associated protein is intimately associated with LPS (11, 26, 27) and has potent and unique biological activities. For example, endotoxin-associated protein is a powerful mitogen for C3H/HeJ mouse cells and human lymphocytes, which are hyporesponsive to LPS (24, 40). The antibiotic-released endotoxin prepared from chemically defined M9 medium contained a large amount of protein. However, the protein in CAZ-released endotoxin lacked the activity of the endotoxin protein, since the CAZ-endotoxin preparation had no toxic effect on GalN-sensitized C3H/HeJ mice, had no mitogenic effect on the B lymphocytes of C3H/HeJ mice, and did not induce cytokine or NO production by the macrophages of C3H/HeJ mice. Alternatively, the high concentration of nonactive protein associated with CAZ-released endotoxin may hinder the endotoxic action of the lipid A-associated protein in the preparation.
The regression lines of CAZ- and IPM-released endotoxin in an LAL test (ES test) paralleled those of purified P. aeruginosa LPS and E. coli LPS, suggesting that these antibiotic-released endotoxins possess endotoxic properties the same as (or at least very similar to) those of chemically purified LPS. The findings that polymyxin B can neutralize the activity (Table 2) support this notion. In other words, the activity of antibiotic-released endotoxin is due almost completely to the LPS contained in the endotoxin preparation.
It is well-established that the soluble and surface forms of the CD14 molecule and serum factors, such as LBP (38) and septin (46), participate in macrophage activation by LPS (47). When the LPS-LBP complex binds to mCD14 on the surfaces of macrophages, they become activated to produce large amounts of active molecules (such as TNF, IL-6, and NO [29]). On the other hand, it has been suggested that other CD14-independent pathways exist (5, 6, 15, 16, 23). Macrophage activation by CAZ-released endotoxin as well as LPS depends mainly on the serum factor(s) and CD14 molecule, since NO, TNF, and IL-6 production by macrophages of C3H/HeN mice is increased in the presence of serum (Fig. 4), and CD14− J7.DEF3 cells require relatively larger doses of endotoxin or LPS to produce TNF amounts equivalent to those produced by CD14+ J774.1 cells (Fig. 7).
Binding of LPS to mCD14 on macrophages induces the phosphorylation of several proteins in the cells, and the phosphorylation is thought to be the initial step in intracellular signaling (1, 13, 21, 39, 44). MAP kinases or extracellular signal-regulated kinase, members of the tyrosine kinase family, rapidly phosphorylate the tyrosine residues of some proteins. These phosphorylated proteins, in turn, activate the cells and regulate the intracellular signaling pathways (34). LPS activates MAP kinases 1 and 2 in monocytes and macrophages (1, 8, 13, 21, 25, 34, 37, 43, 44). CAZ-released P. aeruginosa endotoxin also activated MAP kinases 1 and 2 of the C3H/HeN mouse macrophages (Fig. 6). However, neither the endotoxin nor the LPS could stimulate the kinases of C3H/HeJ mouse macrophages. Furthermore, glutaraldehyde-killed P. aeruginosa organisms could hardly stimulate MAP kinases of the macrophages of C3H/HeN or C3H/HeJ mice. The treatment of organisms with glutaraldehyde may destroy the activity of endotoxin or may hinder cell wall digestion and the release of endotoxin by macrophages.
The importance of the concept that antibiotic-released endotoxin contributes to the pathogenesis of experimental sepsis caused by gram-negative organisms has been documented (7, 28). In the present study, all findings from the in vitro model with M9 medium support the idea that the endotoxic activity of antibiotic-released endotoxin is mainly due to LPS or lipid A but not protein. Nevertheless, the endotoxin preparations contain large amounts of protein that are released or produced by the organisms. However, when the production of TNF, IL-6, or NO in the serum-containing system is compared between cells exposed to phenol-water-extracted LPS and cells exposed to CAZ-released endotoxin, the levels of all three of these reactants are higher sooner in the cells exposed to the CAZ-released endotoxin (Fig. 4). This suggests that the greater effect of the endotoxin may be due to the production of a more potent LPS or to the presence of a larger amount of protein in the endotoxin preparation. Moreover, in infected mice in the in vivo model (Table 3), some other factor(s), in addition to the antibiotic-released endotoxin, seems to participate in the death of the animal. The observation was based on the fact that the survival rate of LPS-hyporesponsive C3H/HeJ mice infected with P. aeruginosa was not improved by CAZ treatment, although the infecting organisms were effectively killed by CAZ (data not shown). The factor(s) contributing to death may include the release of endotoxic protein from the organisms and components of the host’s body (such as serum and phagocytic cells). Alternatively, this may be due to the large amounts of endotoxin released, and these large amounts of endotoxin could somehow cause an increased bacterial growth rate (17, 22, 35). It would reduce the protective efficacy of an antibiotic due to the increased bacterial load with reduced assistance from the host. This would subsequently require the use of increased amounts of agents such as CAZ to kill the bacteria. Our current studies are designed to investigate these hypotheses.
ACKNOWLEDGMENTS
This study was supported by grants from the Ministry of Education, Science and Culture, Japan (grant 08670315 to M.N. and grant 08457090 to T.K.); the Waksman Foundation of Japan Inc., Tokyo, Japan (to M.N.); the Banyu Pharmaceutical Co., Tokyo, Japan (to M.N. and T.K.); and Seikagaku Corp., Tokyo, Japan (to M.N. and T.K.).
REFERENCES
- 1.Arditi M, Zhou J, Torres M, Durden D L, Stins M, Kim K S. Lipopolysaccharide stimulates the tyrosine phosphorylation of mitogen-activated protein kinases p44, p42, and p41 in vascular endothelial cells in a soluble CD14-dependent manner. Role of protein tyrosine phosphorylation in lipopolysaccharide-induced stimulation of endothelial cells. J Immunol. 1995;155:3994–4003. [PubMed] [Google Scholar]
- 2.Atlas R M. M9 medium. In: Parks L C, editor. Handbook of microbiological media. Vol. 529. Boca Raton, Fla: CRC Press, Inc.; 1993. [Google Scholar]
- 3.Beaty C D, Franklin T L, Uehara Y, Wilson C B. Lipopolysaccharide induced cytokine production in human monocytes: role of tyrosine phosphorylation in transmembrane signal transduction. Eur J Immunol. 1994;24:1278–1284. doi: 10.1002/eji.1830240606. [DOI] [PubMed] [Google Scholar]
- 4.Bucklin S E, Morrison D C. Differences in therapeutic efficacy among cell wall-active antibiotics in a mouse model of gram-negative sepsis. J Infect Dis. 1995;172:1519–1527. doi: 10.1093/infdis/172.6.1519. [DOI] [PubMed] [Google Scholar]
- 5.Chen T-Y, Bright S W, Pace J L, Russell S W, Morrison D C. Induction of macrophage-mediated tumor cytotoxicity by a hamster monoclonal antibody with specificity for a lipopolysaccharide receptor. J Immunol. 1990;145:8–12. [PubMed] [Google Scholar]
- 6.Cohen L, Haziot A, Shen D R, Lin X, Dia C, Harper R, Silver J, Goyert S M. CD14-independent responses to LPS require a serum factor that is absent from neonates. J Immunol. 1995;155:5337–5342. [PubMed] [Google Scholar]
- 7.Dofferhoff A S M, Buys J. Dose- and type-dependent antibiotic-induced endotoxin release. In: Faist E, editor. Differential release and impact of antibiotic-induced endotoxin. New York, N.Y: Raven Press; 1995. pp. 11–19. [Google Scholar]
- 8.Dong Z, Qi X, Fidler I J. Tyrosine phosphorylation of mitogen-activated protein kinases is necessary for activation of murine macrophages by natural and synthetic bacterial products. J Exp Med. 1993;177:1071–1077. doi: 10.1084/jem.177.4.1071. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Eng R H K, Smith S M, Fan-Havard P, Ogbara T. Effect of antibiotics on endotoxin release from gram-negative bacteria. Diagn Microbiol Infect Dis. 1993;35:185–189. doi: 10.1016/0732-8893(93)90109-k. [DOI] [PubMed] [Google Scholar]
- 10.Faist E, editor. Differential release and impact of antibiotic-induced endotoxin. New York, N.Y: Raven Press; 1995. [Google Scholar]
- 11.Goodman G W, Sultzer B M. Characterization of the chemical and physical properties of a novel B-lymphocyte activator, endotoxin protein. Infect Immun. 1979;24:685–696. doi: 10.1128/iai.24.3.685-696.1979. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Green L C, Wagner D A, Glogowski J, Skipper P L, Wishnok J S, Tannenbaum S R. Analysis of nitrate, nitrite, and [15N]nitrate in biological fluids. Anal Biochem. 1982;126:131–138. doi: 10.1016/0003-2697(82)90118-x. [DOI] [PubMed] [Google Scholar]
- 13.Han J, Lee J-D, Tobias P S, Ulevitch R S. Endotoxin induces rapid protein tyrosine phosphorylation in 70Z/3 cells expressing CD14. J Biol Chem. 1993;268:25009–25014. [PubMed] [Google Scholar]
- 14.Jackson J J, Kropp H. Differences in mode of action of β-lactum antibiotics influence morphology, LPS release and in vivo antibiotic efficacy. J Endotoxin Res. 1996;3:201–218. [Google Scholar]
- 15.Kirikae F, Kirikae T, Qureshi N, Takayama K, Morrison D C, Nakano M. CD14 is not involved in Rhodobacter sphaeroides diphosphoryl lipid A inhibition of tumor necrosis factor alpha and nitric oxide induction by taxol in murine macrophages. Infect Immun. 1995;63:486–497. doi: 10.1128/iai.63.2.486-497.1995. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Kirikae T, Schade F U, Kirikae F, Rietschel E T, Morrison D C. Isolation of a macrophage-like cell line defective in binding of lipopolysaccharide. Influence of serum and lipopolysaccharide chain length on macrophage activation. J Immunol. 1993;151:2742–2752. [PubMed] [Google Scholar]
- 17.Koch T, Duncker H P, Axt R, Schiefer H G, van Ackern K, Neuhof H. Alterations of bacterial clearance induced by endotoxin and tumor necrosis factor. Infect Immun. 1993;61:3143–3148. doi: 10.1128/iai.61.8.3143-3148.1993. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Lamp K C, Rybak M J, McGrath B J, Summers K K. Influence of antibiotic and E5 monoclonal immunoglobulin M interactions on endotoxin release from Escherichia coli and Pseudomonas aeruginosa. Antimicrob Agents Chemother. 1996;40:247–252. doi: 10.1128/aac.40.1.247. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Lansdorp P M, Aarden L A, Calafat J, Zeiljemaker W P. A growth-factor dependent B-cell hybridoma. Curr Top Microbiol Immunol. 1986;132:105–113. doi: 10.1007/978-3-642-71562-4_14. [DOI] [PubMed] [Google Scholar]
- 20.Leeson M C, Fujihara Y, Morrison D C. Evidence for lipopolysaccharide as the predominant proinflammatory mediator in supernatants of antibiotic-treated bacteria. Infect Immun. 1994;62:4975–4980. doi: 10.1128/iai.62.11.4975-4980.1994. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Liu M K, Herra-Velit P, Brownsey R W, Reiner N E. CD14-dependent activation of protein kinase C and mitogen-activated protein kinases (p42 and p44) in human monocytes treated with bacterial lipopolysaccharide. J Immunol. 1994;153:2642–2652. [PubMed] [Google Scholar]
- 22.Luo G, Niesel D W, Shaban R A, Grimm E A, Klimpel G R. Tumor necrosis factor alpha binding to bacteria: evidence for a high-affinity receptor and alteration of bacterial virulence properties. Infect Immun. 1993;61:830–835. doi: 10.1128/iai.61.3.830-835.1993. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Lynn W A, Liu Y, Golenbock D T. Neither CD14 nor serum is absolutely necessary for activation of mononuclear phagocytes by bacterial lipopolysaccharide. Infect Immun. 1993;61:4452–4461. doi: 10.1128/iai.61.10.4452-4461.1993. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Mangan D, Wahl S M, Sultzer B M, Mergenhagen S T. Stimulation of human monocytes by endotoxin-associated protein: inhibition of programmed cell death. Infect Immun. 1992;60:1684–1686. doi: 10.1128/iai.60.4.1684-1686.1992. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Manthey C L, Brandes M E, Perra P Y, Vogel S N. Taxol increases steady-state levels of LPS-inducible genes and protein-tyrosine phosphorylation in murine macrophages. J Immunol. 1992;149:2459–2465. [PubMed] [Google Scholar]
- 26.Manthey C L, Vogel S N. Elimination of trace endotoxin protein from rough chemotype LPS. J Endotoxin Res. 1994;1:84–91. [Google Scholar]
- 27.Morrison D C, Betz S J, Jacobs D M. Isolation of a lipid A bound polypeptide responsible for ’LPS-initiated’ mitogenesis of C3H/HeJ spleen cells. J Exp Med. 1976;144:840–846. doi: 10.1084/jem.144.3.840. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Morrison D C, Bucklin S E. Endotoxemia, bacteremia, and the pathogenesis of gram-negative sepsis. In: Faist E, editor. Differential release and impact of antibiotic-induced endotoxin. New York, N.Y: Raven Press; 1995. pp. 37–46. [Google Scholar]
- 29.Morrison D C, Danner R L, Dinarello C A, Munford R S, Natanson C, Pollack M, Spitzer J J, Ulevitch R J, Vogel S N, McSweegan E. Bacterial endotoxins and pathogenesis of gram-negative infections: current status and future direction. J Endotoxin Res. 1994;1:71–83. [Google Scholar]
- 30.Morrison D C, Ryan J L. Endotoxins and disease mechanisms. Annu Rev Med. 1987;38:417–432. doi: 10.1146/annurev.me.38.020187.002221. [DOI] [PubMed] [Google Scholar]
- 31.Neu H C. Relation of structural properties of β-lactum antibiotics to antibacterial activity. Am J Med. 1985;79:2–13. doi: 10.1016/0002-9343(85)90254-2. [DOI] [PubMed] [Google Scholar]
- 32.Obayashi T, Tamura H, Tanaka S, Ohki M, Takahashi S, Kawai T. Endotoxin-inactivating activity in normal and pathological human blood samples. Infect Immun. 1986;53:294–297. doi: 10.1128/iai.53.2.294-297.1986. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Parillo J E, Parker M M, Natanson C, Suffredini A F, Danner R L, Cunnion R E, Ognibene F P. Septic shock in humans. Advances in the understanding of pathogenesis, cardiovascular dysfunction, and therapy. Ann Intern Med. 1990;113:227–241. doi: 10.7326/0003-4819-113-3-227. [DOI] [PubMed] [Google Scholar]
- 34.Pelech S L, Sanghera J S. MAP kinases: charting the regulatory pathways. Science. 1992;257:1355–1356. doi: 10.1126/science.1382311. [DOI] [PubMed] [Google Scholar]
- 35.Porat R, Clark B D, Wolff S M, Diarello C A. Enhancement of growth of virulent strains of Escherichia coli by interleukin-1. Science. 1991;254:430–432. doi: 10.1126/science.1833820. [DOI] [PubMed] [Google Scholar]
- 36.Prins J M. Antibiotic induced release of endotoxin—clinical data and human studies. J Endotoxin Res. 1996;3:269–273. [Google Scholar]
- 37.Reimann T, Buscher D, Hipskind R A, Krautwald S, Lohmann-Matthes M-L, Baccarini M. Lipopolysaccharide induces activation of the Raf-1/MAP kinase pathway: a putative role for Raf-1 in the induction of the IL-1β and TNF-α genes. J Immunol. 1994;153:5740–5749. [PubMed] [Google Scholar]
- 38.Schumann R R, Leong S R, Flaggs G W, Gray P W, Wright S D, Mathison J C, Tobias P S, Ulevitch R J. Structure and function of lipopolysaccharide binding protein. Science. 1990;249:1429–1431. doi: 10.1126/science.2402637. [DOI] [PubMed] [Google Scholar]
- 39.Shinomiya H, Hirata H, Nakano M. Purification and characterization of the 65-kDa protein phosphorylated in murine macrophages by stimulation with bacterial lipopolysaccharide. J Immunol. 1991;146:3617–3625. [PubMed] [Google Scholar]
- 40.Sulter B M, Castagna R. The activation of C3H/HeJ cell by certain type of lipopolysaccharides. Adv Exp Biol. 1990;256:149–157. doi: 10.1007/978-1-4757-5140-6_12. [DOI] [PubMed] [Google Scholar]
- 41.Tamura H, Arimoto Y, Tanaka S, Yoshida M, Obayashi T, Kawai T. Automated kinetic assay endotoxin and (1→3)-β-d-glucan in human blood. Clin Chem Acta. 1994;226:109–112. doi: 10.1016/0009-8981(94)90110-4. [DOI] [PubMed] [Google Scholar]
- 42.Tuomanen E, Gilbert K, Tomasz A. Modulation of bacteriolysis by cooperative effects of penicillin-binding proteins 1a and 3 in Escherichia coli. Antimicrob Agents Chemother. 1986;30:659–663. doi: 10.1128/aac.30.5.659. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Weinstein S, Sanghera J S, Lemke K, DeFranco A L, Pelech S L. Bacterial lipopolysaccharide induces tyrosine phosphorylation and activation of mitogen-activated protein kinases in macrophages. J Biol Chem. 1992;267:14955–14962. [PubMed] [Google Scholar]
- 44.Weinstein S L, June C H, Defranco A L. Lipopolysaccharide-induced protein tyrosine phosphorylation in human macrophages is mediated by CD14. J Immunol. 1993;151:3829–3838. [PubMed] [Google Scholar]
- 45.Westphal O, Lüderitz O, Bister F. Über die Extraktion von Bakterien mit Phenol/Wasser. Naturforschung. 1952;7b:148–155. [Google Scholar]
- 46.Wright S D, Ramos R A, Patel M, Miller D S. Septin: a factor in plasma that opsonizes lipopolysaccharide-bearing particles for recognition by CD 14 on phagocytes. J Exp Med. 1992;176:719–727. doi: 10.1084/jem.176.3.719. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Wright S D, Ramos R A, Tobias P S, Ulevitch R J, Mathison J C. CD14, a receptor for complexes of lipopolysaccharide (LPS) and LPS binding protein. Science. 1990;249:1431–1433. doi: 10.1126/science.1698311. [DOI] [PubMed] [Google Scholar]