INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

As a major constituent of the outer membranes of infectious gram-negative bacteria (32), lipopolysaccharide (LPS) triggers many pathophysiological events. Most of the toxic properties of LPS can be attributed to the hydrophobic fatty-acylated disaccharide lipid A portion of the molecule (8). In both in vitro and in vivo systems, LPS-sensitive cells such as monocytes and macrophages can be stimulated by LPS or lipid A to release cytokines such as tumor necrosis factor alpha (TNF-), interleukin 1 (IL-1), IL-6, and other mediators. In vivo, release of pathophysiological concentrations of these cytokines and cellular mediators can lead to tissue damage. For this reason, a variety of therapeutic interventions have been undertaken to block the action of endotoxin and the sequelae of cellular activation by endotoxin. These approaches include administration of anti-lipid A antibodies, anticytokine antibodies, or soluble neutralizing receptors for cytokines or antagonism of cytokine receptors (1, 29, 35, 36).

For maximal therapeutic effect, it may be necessary to block the action of endotoxin at the primary event of receptor activation in order to block LPS-induced cellular activation and the subsequent release of the broad spectrum of potentially deleterious cytokines and cellular mediators that lead to septic shock. Although the mechanism of cellular activation by LPS has yet to be fully elucidated, several possible cell surface receptors for LPS have been proposed, and their mechanism of activation is being studied (16, 19, 23). Part of our understanding of cellular activation comes from the study of receptor antagonists. It has been reported that the LPSs from two nonenteric, nonpathogenic bacteria, Rhodobacter capsulatus and Rhodobacter sphaeroides, are poor agonists, and the lipid A's derived from these LPSs can antagonize more pathogenic LPSs (14, 17, 22, 28). These findings imply that certain lipid A derivatives can inhibit the acute inflammatory response to LPS and may be useful for treatment of LPS-induced shock or mortality. Although several reports have demonstrated a role for LPS antagonists, some inconsistencies in extrapolating in vitro studies to in vivo models still exist (5), possibly due to species differences in the animal model used (6) or the confounding presence of weak agonistic activity (24).

E5531 is a chemically synthesized lipid A analog that, while demonstrating potent antagonistic activity against LPS, possesses no intrinsic agonistic activity. We have previously reported that E5531 antagonizes LPS-induced cytokine production in a variety of in vitro assays (12). In this study, we further describe the in vitro and in vivo antagonism of LPS by E5531 in several models of LPS-mediated cellular activation, hepatotoxicity, and lethality.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Animals. Pathogen-free male C57BL/6 mice were obtained from SLC Inc. (Shizuoka, Japan) and Jackson Laboratories (Bar Harbor, Maine), and pathogen-free male BALB/c mice were obtained from Charles River Japan Inc. (Atsugi, Japan).

Reagents. E5531, 6-O-{2-deoxy-6-O-methyl-4-O-phosphono-3-O-[(R)-3-Z-dodec-5-enoyloxy decyl]-2-[3-oxo-tetradecanoylamido]--D-glucopyranosyl}-2-deoxy-3-O-(R)-3-hydroxydecyl-2-[3-oxo-tetradecanoylamido]-1-O-phosphono--D-glucopyranose tetrasodium salt, was synthesized at the Eisai Research Institute of Boston (Andover, Mass.). LPS from Escherichia coli (a phenol extract of serotype O111:B4) was obtained from Sigma Chemical Co. (St. Louis, Mo.) or from List Biological Inc., (Campbell, Calif.). Synthetic E. coli lipid A (serotype O111:B4; catalog no. LA-15-PP) was purchased from Daiichi Chemical Co., (Tokyo, Japan). D-(+)-Galactosamine HCl (GalN), peroxidase (type XII), and bovine serum albumin were purchased from Sigma Chemical Co. Recombinant murine TNF-, rabbit anti-mouse TNF- polyclonal antiserum, and a mouse TNF- enzyme-linked immunosorbent assay (ELISA) kit (Factor-Test mTNF-; Genzyme Corp., Boston, Mass.) were purchased from Genzyme Corp. (Cambridge, Mass.). N-Succinimidyl 4-(N-maleimidomethyl) cyclohexane-1-carboxylate (Pierce, Rockford, Ill.), o-phenylenediamine (Wako Pure Chemical Industries Ltd., Osaka, Japan), and other commercially available reagents were used for this study.

Preparation of murine peritoneal macrophages. Peritoneal macrophages were isolated from mice treated intraperitoneally with 2 mg of a cell wall preparation from heat-killed M. bovis BCG (Ribi Immunochem Research Inc., Hamilton, Mont.) in 200 µl of pyrogen-free 0.9% saline. Three to 6 days later, the animals were killed with CO₂ gas and macrophages were isolated by aseptic lavage with ice-cold RPMI 1640 supplemented with penicillin (80 U/ml), streptomycin (100 µg/ml), 1 mM sodium pyruvate (Sigma Chemical Co.), 2% heat-inactivated fetal bovine serum (JRH Biosciences, Lenexa, Kans.), and 10 U of heparin per ml. The harvested cells were centrifuged and the pellet was resuspended with 1 ml of erythrocyte lysing buffer (Sigma Chemical Co.), followed by the addition of 40 ml of serum-free RPMI 1640 containing the antibiotics specified above. After three washes by centrifugation, the final cell pellet was diluted to a concentration of 2 × 10⁶ cells/ml in RPMI 1640 containing 10% fetal bovine serum and was allowed to adhere to 24-well plastic tissue culture plates (Falcon Primaria; Becton Dickinson, Lincoln Park, N.J.). After 3 h at 37°C in 5% CO₂, the nonadherent cells were removed from the plate by washing twice with serum-free RPMI 1640.

In vitro analysis of E5531 in mouse peritoneal macrophages. Analysis of inhibition of LPS-induced release of TNF- by E5531 was performed with cultures of mouse peritoneal macrophages in RPMI 1640 supplemented with 10% fetal bovine serum. E5531 was added to achieve the various concentrations, followed by the addition of LPS at a final concentration of 10 ng/ml. After incubation for 3 h at 37°C, the cultures were clarified by centrifugation (2,000 × g for 5 min at 4°C), and the resulting supernatants were stored at 80°C until the assay for TNF- was performed.

Analysis of LPS binding to peritoneal macrophages. Adherent macrophages (2 × 10⁶ cells/well) in 24-well plastic tissue culture plates were washed twice with 1 ml of prewarmed (37°C) serum-free RPMI 1640, followed by the addition of 200 µl of medium supplemented with 2% fetal bovine serum, 25 µl of inhibitor in RPMI 1640 containing 2% fetal bovine serum, and finally, 25 µl of [¹²⁵I]ASD-LPS (~60 ng of 4.1 µCi/µg). After incubation for 1 h at 37°C in 5% CO₂, the wells were washed three times with 1.0 ml of 50 mM Tris buffer (pH 7.4) containing 150 mM NaCl and 2 mg of bovine serum albumin per ml. After solubilization of the cells with 1 ml of 0.1 N NaOH, 950 µl of each sample was analyzed for ¹²⁵I with an autogamma counter (Gamma 5500B; Beckman Instruments, Fullerton, Calif.). Nonspecific binding was evaluated by the addition of 1 mg of unlabelled LPS per ml with the [¹²⁵I]ASD-LPS. Specific binding was calculated by subtracting the nonspecific binding from the total binding. These binding studies used 2% fetal calf serum, which we have found to be optimal for LPS binding to mouse macrophages. At lower serum concentrations, binding is reduced, presumably due to decreased concentrations of LPS binding protein. At higher concentrations (10% or more), some inhibition of binding is seen, likely due to interference with other factors in serum.

Analysis of plasma TNF- increases and mortality induced by LPS in BCG-primed mice. For BCG priming, 0.2 ml of a 10-mg/ml BCG suspension was intravenously (i.v.) injected into the tail veins of C57BL/6 mice (ages, 5 to 6 weeks), and the primed mice were used for experiments 10 to 14 days later (30). To examine the effect of E5531 on TNF- production, 0.4 ml of a pyrogen-free 5% glucose solution containing the various amounts of LPS and E5531 was i.v. injected into the tail vein. One hour later, mice were anesthetized with ether and 30 µl of blood was drawn from the retro-orbital vein and placed in a heparinized hematocrit tube. Plasma TNF- levels were determined by ELISA.

Protective effect of E5531 on LPS-induced hepatic injury in P. acnes-primed or GalN-sensitized mice. BALB/c mice (age, 8 weeks) were primed 7 days prior to use by i.v. injection of 100 µl of 20 mg of P. acnes per ml in pyrogen-free 0.9% saline into the tail vein. Hepatitis was induced by LPS in primed mice by i.v. injection of 0.4 ml of pyrogen-free 5% glucose solution containing 1 µg of LPS mixed with the various amounts of E5531. Alternatively, mice were sensitized with GalN as reported previously (10, 20). The tail veins of male BALB/c mice were injected with 0.3 ml of pyrogen-free 5% glucose containing 20 mg of GalN (pH 7.0) and 1.0 ng of LPS mixed with 0, 10, 30, 100, or 300 ng of E5531. Additional mice were given GalN alone. Twenty-four hours after injection, mice were anesthetized with ether and approximately 0.5 ml of blood was drawn from the abdominal aorta and placed in tubes containing heparin, plasma was prepared, and L-alanine aminotransferase (ALT; EC 2.6.1.2) activities were determined by standard spectrophotometric methods (11) with an latrozyme TA-LQ Kit (Iatron Laboratories Inc., Tokyo, Japan).

View larger version (15K): [in this window] [in a new window]   FIG. 1.   Inhibition of LPS-mediated induction of TNF- release in primary cultures of mouse peritoneal macrophages. E5531 was added to cultures of mouse peritoneal macrophages at the indicated concentrations, immediately followed by the addition of 10 ng of LPS per ml. After a 3-h incubation at 37°C, supernatants were assayed for TNF- as described in the Materials and Methods section. Each point and error bar represent the mean and standard error of the mean obtained for three separate cultures in a single experiment and represents three experiments.

Analysis of TNF-. TNF- was measured by an ELISA with anti-mouse TNF- polyclonal antiserum purified by affinity chromatography (protein A-Sepharose [MAPS-II; Bio-Rad, Richmond, Calif.]). Fab' fragments were generated as described previously (31) and conjugated to peroxidase with N-succinimidyl 4-(N-maleimidomethyl) cyclohexane-1-carboxylate.

Statistical analysis. Results are expressed as the mean ± standard error of the mean. For in vitro studies, the concentration of E5531 that inhibited 50% of the LPS-induced increases in TNF- concentrations (the 50% inhibitory concentration [IC₅₀]) was calculated by a log-linear interpolation from the two points that span the 50% value. In the studies with animals, statistically significant differences between the control and treated groups were determined by Student's t test or a one-way analysis of variance with Fisher's protected least significant difference procedure (25). In lethality studies, the statistically significant differences between the control group and the treated groups were determined by ² test. Differences with P values of less than 0.05 were considered statistically significant. The 50% effective doses (ED₅₀s) of E5531 were calculated by a nonlinear least-squares method.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Inhibition of LPS-mediated release of TNF- in murine peritoneal macrophages. Unstimulated primary cultures of mouse peritoneal macrophages produced no measurable TNF-. However, 3 h of incubation with LPS (10 ng/ml) triggered the release of 1,973 ± 120 pg of TNF- per ml (n = 3). As shown in Fig. 1, E5531 inhibited this LPS-triggered TNF- release at concentrations of 0.1 to 100 nM in a dose-dependent manner (IC₅₀ = 2.2 nM), with complete inhibition observed at concentrations greater than 100 nM. These results indicate that E5531 is a potent antagonist of LPS-mediated activation of murine macrophages.

Antagonistic effect of E5531 on LPS-induced plasma TNF- increases in mice. In vivo antagonism of LPS by E5531 was determined by analysis of the ability of E5531 to inhibit LPS-induced plasma TNF- increases in BCG-primed mice. In preliminary experiments, it was determined that peak plasma TNF- levels were observed 1 h after LPS administration (data not shown). As shown in Table 1, LPS doses of greater than 0.1 µg induced dose-dependent increases in plasma TNF- levels. E5531 inhibited this response, and the degree of inhibition was inversely correlated to the amount of LPS used as an agonist. Thus, 3 µg of E5531 completely inhibited the LPS response with 0.1 µg of LPS, but inhibition by 3 µg of E5531 was reduced to 50% when mice were challenged with three times as much LPS (0.3 µg). These results indicate that the efficacy of E5531 is related to the challenge dose of LPS used. E5531 alone, even at 100 µg/mouse, did not elicit detectable amounts of TNF- in plasma (less than 0.4 ng/ml), indicating that it lacks agonistic activity in BCG-primed mice (data not shown).

View this table: [in this window] [in a new window]   TABLE 1.   Effect of E5531 on TNF- production induced by different doses of LPS in BCG-primed mice

Antagonism of LPS-induced lethality in BCG-primed mice. The i.v. administration of 6 µg of LPS to BCG-primed mice resulted in 90% lethality within 16 h (Fig. 2). No further deaths occurred in this group during a further 20 h of observation. Administration of E5531 along with the LPS challenge protected them from death in a dose-dependent manner, with complete protection achieved with doses of 30 and 100 µg. Compared to the results for the controls, the protection provided by E5531 was statistically significant at doses of greater than 3 µg/animal.

View larger version (17K): [in this window] [in a new window]   FIG. 2.   E5531 suppression of LPS-induced mortality in BCG-primed mice. BCG-primed mice (10 animals per group) were i.v. injected with 6 µg of LPS in 0.4 ml of 5% glucose only () or along with 3 µg (), 10 µg (), 30 µg (), or 100 µg () of E5531, and mortality was assessed at the indicated times. *, P < 0.05; **, P < 0.01 versus control group (2 test; n=10). The control was 5% glucose solution plus LPS.

Effect of E5531 on LPS-induced hepatic injury in P. acnes-primed or GalN-sensitized mice. When P. acnes-primed mice were dosed i.v. with LPS, plasma ALT values increased more than 14-fold compared to those in mice receiving vehicle only, indicating that LPS injured the livers of these animals (Table 2).

E5531 inhibits [¹²⁵I]ASD-LPS binding to mouse macrophages in vitro. Analysis of binding of [¹²⁵I]ASD-LPS to primary cultures of murine macrophages indicates that binding is both specific and saturable. As shown in Fig. 3, E5531 inhibited the specific binding of [¹²⁵I]ASD-LPS in a dose-dependent manner, with an IC₅₀ of 0.77 µg/ml (0.5 µM), and greater than 80% inhibition was observed with 10 µg of E5531 per ml. LPS and lipid A were also tested for their ability to inhibit [¹²⁵I]ASD-LPS binding. Specific binding of [¹²⁵I]ASD-LPS was inhibited >90% by 100 µg of E. coli LPS per ml, with an IC₅₀ of 0.69 µg/ml. Lipid A inhibited specific LPS binding with an IC₅₀ of 11.2 µg/ml (5.6 µM). These results indicate that cell surface LPS receptors have a greater affinity for E5531 than for lipid A. 

View larger version (24K): [in this window] [in a new window]   FIG. 3.   Inhibition of [125I]ASD-LPS binding to mouse peritoneal macrophages by E5531. Mouse macrophages were incubated for 1 h with 60 ng of 4.1 µCi of [125I]ASD-LPS per µg and the indicated concentration of E5531 (), lipid A (), or LPS () in RPMI 1640 containing 2% fetal bovine serum. The cells were washed, followed by the addition of 0.1 N NaOH to solubilize the radioactivity. Aliquots of the mixture were analyzed with a gamma counter. Specific binding of [125I]ASD-LPS was determined by subtracting the value for nonspecific binding from the total binding. Nonspecific binding was evaluated separately by adding 1 mg of unlabelled LPS per ml to the incubation mixture. Each point represents the mean value of specific binding for duplicate cultures, and the results are expressed as counts per minute per culture well.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Lipid X, a monosaccharide biosynthetic precursor of lipid A, has been reported to protect mice and sheep against the acute lethal toxicity of LPS (21). However, lipid X has proven to be a poor LPS antagonist in vivo (5), and difficulties in obtaining purely antagonistic preparations of lipid X have yielded confusing results (2, 15). More promisingly, nontoxic LPS or lipid A's obtained from nonpathogenic bacteria such as R. capsulatus and R. sphaeroides, as well as the disaccharide biosynthetic precursors of LPS such as lipid IV_A, have been shown to be more potent inhibitors of the LPS-induced production of TNF- in murine RAW 264.7 macrophage-like tumor cells in vitro (14, 17, 18) and in mice (17, 22). As reported previously, E5531 is a novel synthetic analogue of the lipid A from R. capsulatus (4, 12). In vitro, E5531 inhibited the LPS-induced release of cytokines such as TNF-, IL-1, and IL-6 from human monocytes and human whole blood (4). In addition, E5531 potently inhibited the binding of LPS to human monocytes and monocyte-derived macrophages. In this study, similar inhibition of LPS-induced TNF- release and inhibition of LPS binding was observed in mouse macrophages. The ability of 100 and 1,000 nM E5531 to completely inhibit induction of TNF- by LPS also implies that E5531 has no intrinsic LPS-like agonistic activity, and in fact, no significant increases in TNF- concentrations were detected in cultures treated with concentrations of E5531 up to 10 or 100 µM. Thus, unlike lipid IV_A (9) and bacterium-derived lipid A from R. sphaeroides (24), E5531 was devoid of agonistic activity in human and mouse systems, even when it was tested at high concentrations.

E5531 inhibits the interaction of [¹²⁵I]ASD-LPS with the cell surface of mouse macrophages, with an IC₅₀ of 0.77 µg/ml (0.50 µM) for ~240 ng of LPS per ml. This makes E5531 a more potent inhibitor than E. coli lipid A (IC₅₀ = 11.2 µg/ml or 5.6 µM) but a less potent inhibitor than E. coli LPS, which has an estimated IC₅₀ of approximately 23 nM (0.69 µg/ml; using an estimated average molecular mass of 30 kDa). By using this estimated molar basis, LPS inhibits binding of [¹²⁵I]ASD-LPS more potently than lipid A which is the toxicophore and presumably the determinant of LPS binding to cells. This discrepancy in affinity is likely due to a physicochemical characteristic of the more hydrophobic lipid A that limits its solubility or ability to form monomers or small aggregates. At this time, we are unsure if E5531 possesses greater solubility than E. coli lipid A or a higher affinity for the LPS receptor. However, either conclusion supports the idea that E5531 may exert its effect at a cell surface receptor(s) where measurable binding of LPS occurs. Finally, while we have not addressed the role of LPS binding protein (LBP) in these studies, we have found LPS binding to be immeasureable without some serum (2% is optimal in this system) or partially purified LBP from rabbit serum (ion-exchange-purified LBP). Use of partially purified LBP in a similar binding assay with a mouse macrophage cell line (RAW 264.7 cells) yielded results similar to those presented in Fig. 3.

In order to prove the efficacies of LPS antagonists in rodents, a variety of animal models have been developed. In all cases, steps had to be taken to overcome the relative insensitivity of rodents to endotoxin compared to the sensitivity of humans (3, 26, 30). To this end, we have used sensitization or priming with BCG or P. acnes as well as GalN (7, 10, 20, 27, 30). Priming or sensitization of mice with BCG or P. acnes increases their sensitivity to LPS and causes accumulation of activated macrophages in the reticuloendothelial system (30). As described in Table 1, BCG-primed mice generated high levels of TNF- in response to endotoxin. This response enabled us to examine the in vivo antagonistic activity of E5531 against LPS using the levels of TNF- in plasma as an index. Consistent with the idea that E5531 is a competitive antagonist of LPS, the effective dose of E5531 was affected by the dose of LPS used; administration of higher doses of LPS required higher doses of antagonist. Because it is well accepted that TNF- is a key mediator of the pathological action of endotoxin and of the morbidity caused by endotoxin, suppression of LPS-induced TNF- generation by E5531 in vivo encouraged investigation of the effect of E5531 on LPS-induced lethality and hepatic injury. As expected from results measuring the reduction in the level of TNF-, E5531 also suppressed lethality caused by LPS in our BCG-primed murine model in a dose-dependent manner (Fig. 2).

Administration of LPS induces hepatic injury in the P. acnes-primed murine model. By using the plasma ALT level as an indicator of hepatotoxicity, E5531 antagonized this LPS-mediated effect with an ED₅₀ of about 10 µg/animal. In this model, as well as in mice sensitized with GalN, hepatotoxicity has been attributed to the generation of TNF- (10, 20). Therefore, it is likely that the suppressive effect of E5531 on the hepatic injury may be due to attenuation of LPS-induced cytokine generation (including the generation of TNF-).

In the LPS-induced hepatitis model with mice sensitized with GalN, doses of LPS as low as 1 ng/mouse caused liver injury. In this case, the effective dose of E5531 needed to suppress this injury was as low as 30 ng/mouse, again suggesting that the dose of E5531 required for antagonistic efficacy was related to the LPS dose in vivo.

In conclusion, E5531 inhibits LPS-mediated cellular activation and LPS binding in murine macrophages in vitro. In vivo, E5531 suppresses the LPS response and hepatic injury in mice in an apparently competitive manner. These results suggest that the lipid A derivative suppresses LPS responses through blockade of an LPS receptor both in vitro and in vivo.