Potent TLR4-dependent cell activation by Gram-negative bacterial endotoxins depends on sequential endotoxin-protein and protein-protein interactions with LPS-binding protein, CD14, myeloid differentiation protein 2 (MD-2), and TLR4. Previous studies have suggested that reduced agonist potency of underacylated endotoxins (i.e., tetra- or penta- vs hexa-acylated) is determined by post-CD14 interactions. To better define the molecular basis of the differences in agonist potency of endotoxins differing in fatty acid acylation, we compared endotoxins (lipooligosaccharides (LOS)) from hexa-acylated wild-type (wt), penta-acylated mutant msbB meningococcal strains as well as tetra-acylated LOS generated by treatment of wt LOS with the deacylating enzyme, acyloxyacylhydrolase. To facilitate assay of endotoxin:protein and endotoxin:cell interactions, the endotoxins were purified after metabolic labeling with [3H]- or [14C]acetate. All LOS species tested formed monomeric complexes with MD-2 in an LPS-binding protein- and CD14-dependent manner with similar efficiency. However, msbB LOS:MD-2 and acyloxyacylhydrolase-treated LOS:MD-2 were at least 10-fold less potent in inducing TLR4-dependent cell activation than wt LOS:MD-2 and partially antagonized the action of wt LOS:MD-2. These findings suggest that underacylated endotoxins produce decreased TLR4-dependent cell activation by altering the interaction of the endotoxin:MD-2 complex with TLR4 in a way that reduces receptor activation. Differences in potency among these endotoxin species is determined not by different aggregate properties, but by different properties of monomeric endotoxin:MD-2 complexes.

Invading Gram-negative bacteria typically provoke rapid mobilization of host responses through the ability of the innate immune system to detect even minute numbers of bacteria. Recognition of and response to Gram-negative bacterial endotoxin play important roles in this process and are mediated swiftly and sensitively through activation of TLR4 (1, 2). Although initiation of host defense responses at the outset of Gram-negative bacterial invasion is essential, it is equally important for the host immune system to regulate the duration and intensity of these responses to endotoxin. Modulation of host responses is necessary to avoid overexuberant and sustained host inflammatory responses that could provoke severe pathological consequences, even death (3). The work from this laboratory as well as a number of others has contributed to establishing the endotoxin:protein interactions that result in TLR4-dependent activation and subsequent cellular responses (1, 2, 4, 5, 6, 7, 8). Efficient and maximal response to endotoxin requires sequential interactions involving LPS-binding protein (LBP),3 CD14, myeloid differentiation protein 2 (MD-2), and TLR4 (1, 2, 4, 5, 6, 8, 9, 10, 11). LBP binds endotoxin aggregates (agg) and endotoxin-rich bacterial outer membranes, facilitating extraction of endotoxin by CD14 and formation of a monomeric endotoxin:CD14 complex that activates cells containing endogenous MD-2 and TLR4 (7, 8). Endotoxin is readily transferred from an isolated endotoxin:CD14 complex to soluble, rMD-2 to form a stable monomeric endotoxin:MD-2 complex that can be isolated by gel filtration chromatography (12). This isolated endotoxin:MD-2 complex activates cells in a TLR4-dependent manner at picomolar concentrations, consistent with a system that can initiate potent proinflammatory responses in response to small numbers of invading bacteria. Although the ligand that initiates physiological responses by provoking activation of TLR4 has been shown to be endotoxin:MD-2, the mechanism by which activation through TLR4 occurs has still not been clearly defined (12, 13).

The lipid A portion of endotoxin consists of diphosphorylated β-1,6-linked d-glucosamine disaccharide linked via amide or ester linkages to 3-hydroxy fatty acids that are further substituted by nonhydroxylated fatty acids. This region represents the key component of endotoxin that determines the potency of its agonist properties (14, 15, 16, 17, 18). The fatty acids of the lipid A portion of endotoxin, specifically features such as number, length, symmetry, and saturation, have been shown to be a major determinant of the potency of endotoxin in eliciting TLR4-dependent host responses (19, 20, 21). It has been demonstrated that underacylated endotoxins exhibit markedly reduced ability to provoke endotoxic responses and can inhibit in a dose-dependent manner the strong endotoxic response elicited by hexa-acylated endotoxins (15, 16, 17, 18). Among the underacylated endotoxins that have been studied are naturally occurring underacylated endotoxins, mutant bacterial strains that are unable to produce necessary acyltransferases, synthetic analogues of lipid A, and endotoxins that have been deacylated by the enzyme acyloxyacylhydrolase (AOAH). Although these underacylated variants of lipid A have been shown to have reduced agonist and/or antagonistic activities, they appear to interact with CD14 in a manner similar to that observed for hexa-acylated endotoxins (22, 23). Because the ability to activate TLR4 effectively is a reflection of the bioactivity of the endotoxin, the reduced potency of these endotoxins is probably determined by a molecular event that involves interactions with MD-2 and/or TLR4.

To further define the molecular basis of endotoxic potency, we used endotoxins with variations in the fatty acid content of the lipid A structure of Neisseria meningitidis serotype B (NMB lipooligosaccharides (LOS)). The wild-type LOS (LOSwt) lipid A from NMB contains a symmetrical distribution of fatty acids with secondary laurate substitutions on the N-linked 2- and 2′-acyloxy myristates (Fig. 1). We have compared LOSwt to LOSwt from NMB treated with AOAH (LOSAOAH; Fig. 1) and LOS from the msbB NMB mutant (LOSmsbB; Fig. 1) for their ability to interact with LBP, soluble CD14 (sCD14), and soluble rMD-2 to form monomeric endotoxin:MD-2 complexes and provoke TLR4-dependent cell activation. Metabolically labeled endotoxin of high specific radioactivity was used to allow studies with endotoxin at concentrations <1 ng/ml. We now show that hexa-, penta-, and tetra-acylated endotoxins have similar ability to form monomeric complexes with sMD-2 in an LBP- and CD14-dependent manner. However, in contrast to hexa-acylated endotoxin, these monomeric underacylated endotoxin:MD-2 complexes have a reduced ability to evoke TLR4-dependent cell activation. In fact, an underacylated endotoxin:MD-2 complex antagonizes TLR4-dependent activation, presumably by competing with hexa-acylated endotoxin:MD-2 for interaction with TLR4.

FIGURE 1.

Lipid A structure of LOS of NMB wt, msbB mutant, and wt after AOAH treatment (15 ).

FIGURE 1.

Lipid A structure of LOS of NMB wt, msbB mutant, and wt after AOAH treatment (15 ).

Close modal

LBP and sCD14 were provided by Xoma (US) LLC. Both parental HEK293 and cells stably transfected with TLR4 (HEK293/TLR4) were provided by Dr. J. Chow (Eisai Research Institute, Andover, MA). HEK293 transfected with CD14, MD-2, and TLR4 were purchased from InvivoGen. Soluble rMD-2 containing a hexapolyhistidine tag on the C-terminal end was prepared as previously described (12). AOAH was a gift from Dr. R. Munford (University of Texas Southwestern, Dallas, TX).

Chromatography matrices and electrophoresis supplies were purchased from GEHealthcare. Human serum albumin was obtained as an endotoxin-free, 25% stock solution (Baxter Healthcare). HisLink resin (Promega) was used for interactions with polyhistidine-tagged MD-2 as previously described (12).

[14C]LOS (sp. act., 600 cpm/ng) or [3H]LOS (spec. act., 6,000–8, 000 cpm/ng) was isolated from an acetate auxotroph of NMB after metabolic labeling as previously described (7). [14C]- or [3H]LOSagg (apparent Mr, >20 million) and [14C]- or [3H]LOS:CD14 (Mr, ∼60,000) was prepared and purified by size exclusion chromatography as previously described (7, 24). An msbB derivative of NMB, NMBA11K3cap, was obtained from Dr. M. Apicella (University of Iowa, Iowa City, IA). The strain was grown in medium supplemented with radioactive [3H]- or [14C]acetate, and LOS (500 or 80 cpm/ng, respectively) was isolated as described by Post et al. (25). Tetra-acylated LOS (spec. act. of 3H, 5000 cpm/ng; 14C, 500 cpm/ng) from NMBACE1 was prepared by deacylation with AOAH according to the method of Munford and Erwin (26). The specific radioactivities of various LOS preparations were estimated as previously described (7). The [14C]fatty acid compositions of 14C-labeled endotoxins (purified agg and protein:endotoxin complexes) were determined by sequential treatment with 4 N HCl and 4 N NaOH at 90°C to release fatty acids. After chloroform/methanol extraction, labeled 3-hydroxy and nonhydroxylated fatty acid recovered in the chloroform phase were analyzed by reverse phase TLC (7). Quantitation of 14C-labeled fatty acids resolved by TLC was performed using the Typhoon (GEHealthcare) image analysis program. Purified LOSagg, used as the starting material to generate endotoxin:protein complexes, was prepared by either gel sieving chromatography or ultracentrifugation (7).

HEK293 cells with or without TLR4 or HEK293/CD14/MD-2/TLR4 have been extensively characterized and were cultured as described previously (27) or according to the recommendations of InvivoGen. For cell activation assays, cells were grown to confluence, then washed twice with warm PBS, pH 7.4 (PBS), and incubated for 4 h at 37°C in 5% CO2 and 95% humidity in HBSS (HBSS+)/0.1% human serum albumin (HSA) with the indicated supplements in 48-well plates. Activation of HEK293 cells was assessed by measuring accumulation of extracellular IL-8 by ELISA (BD Clontech).

Columns of Sephacryl HR S200 (1.6 × 30 or 70 cm) or Sephacryl HR S500 (1.5 × 18 cm) were equilibrated in HBSS+, 10 mM HEPES, or PBS, pH 7.4, with or without 0.1% HSA. Aggregates and complexes were prepared as described using recombinant proteins and either [14C]- or [3H]LOSagg by incubation at 37°C for 30 min before gel filtration chromatography (7, 12). Fractions (1 ml) were collected (flow rate, 0.5 ml/min) at room temperature using GEHealthcare AKTA or AKTA Purifier fast performance liquid chromatography. Aliquots of the collected fractions were analyzed by liquid scintillation spectroscopy using a Beckman LS liquid scintillation counter to detect radioactive LOS. Recoveries of LOS were >70%. All solutions used were pyrogen free and sterile filtered. After chromatography, selected fractions to be used in the bioassays were pooled and passed through sterile syringe filters (0.22-μm pore size) with >90% recovery of radiolabeled material in the sterile filtrate. Fractions were stored under sterile conditions at 4°C until needed. LOSwt and LOSAOAH were stable for at least 1 mo at 4°C, with no detectable changes in chromatographic or functional properties. LOSmsbB was somewhat less stable; generated fractions were used within 1 wk. Sephacryl S200 columns were calibrated with Bio-Rad gel filtration standards that included thyroglobulin, γ-globulin, OVA, myoglobin, vitamin B12, and human serum albumin.

Cocapture by metal chelation chromatography of radiolabeled LOS associated with sMD-2-His6 was performed as previously described (12). Briefly, samples in PBS/0.1% HSA were incubated with HisLink resin (0.2–1 ng of radiolabeled LOS/100 μl; Promega) at 4°C, batchwise, for 1 h. The resin was spun down, supernatant was removed, and the resin was washed three or four times (500 μl) before elution into 2% SDS. The presence of radiolabeled LOS absorbed to the resin was evaluated by liquid scintillation spectroscopy.

The potency of endotoxins, including meningococcal endotoxin, has been shown to vary correlating with the extent of lipid A acylation after the rank order: hexa ≫ penta- ≥ tetra-acylated (15, 16, 17, 18). To better define the molecular basis of the reduced potency of underacylated endotoxins, we chose to compare the properties of hexa-, penta-, and tetra-acylated forms of meningococcal LOS endotoxins. To facilitate these studies and permit the use of low concentrations of endotoxins, the LOS species were metabolically labeled to high specific activity as previously described (7).

Endotoxin with a penta-acylated lipid A (LOSmsbB) was isolated from an isogenic strain of NMB (NMBA11K3cap) that carries an insertional inactivation of the acyltransferase gene, msbB (25). Fatty acid analysis of the isolated LOSmsbB confirmed that the lipid A from LOSmsbB contained a decreased amount of lauric acid (C12:0) compared with LOSwt lipid A, consistent with the predicted 50% reduction of C12:0 (Fig. 2,A) (25). To obtain a tetra-acylated form of NMB LOS, wt LOS NMBACE1 was treated with the deacylating enzyme AOAH according to the procedure described by Munford and Erwin (26). Fatty acid analysis of the recovered AOAH-treated LOS confirmed the selective and nearly complete loss of the lauric acid residues that are attached to the 3-hydroxymyristoyl chains in the wt lipid A (Fig. 2,A). [14C]fatty acid content is illustrated in Fig. 2 A, as both the TLC pattern of the [14C]fatty acids recovered after chemical hydrolysis of the various [14C]LOS species and a bar graph of the [14C]fatty acid compositional analyses derived from image analysis.

FIGURE 2.

Structural and functional properties of wt, msbB, and AOAH-treated LOS aggregates. A, TLC of [14C]fatty acids of wt, msbB, and AOAH-treated [14C]LOS released by chemical hydrolysis as described in Materials and Methods. [14C]fatty acids were quantified by image analysis and expressed as a percentage of the total recovered [14C]fatty acids. B, Radiolabeled LOS (1–10 ng/ml) was analyzed by gel filtration chromatography on Sephacryl S500 HR (1.5 × 18 cm) in HBSS+, 10 mM HEPES, and 0.1% HSA (representative graphs). Data are expressed as a percentage of the total radiolabeled LOS recovered; total recoveries were ≥70%. C, HEK293/CD14/MD-2/TLR4 cells were incubated for 4 h with increasing amounts of LOS aggregates and LBP (100 ng/ml) in DMEM/0.1% HSA. Extracellular IL-8 accumulated in cell supernatants was determined by ELISA. Results are from one experiment with triplicate samples and are representative of two similar experiments.

FIGURE 2.

Structural and functional properties of wt, msbB, and AOAH-treated LOS aggregates. A, TLC of [14C]fatty acids of wt, msbB, and AOAH-treated [14C]LOS released by chemical hydrolysis as described in Materials and Methods. [14C]fatty acids were quantified by image analysis and expressed as a percentage of the total recovered [14C]fatty acids. B, Radiolabeled LOS (1–10 ng/ml) was analyzed by gel filtration chromatography on Sephacryl S500 HR (1.5 × 18 cm) in HBSS+, 10 mM HEPES, and 0.1% HSA (representative graphs). Data are expressed as a percentage of the total radiolabeled LOS recovered; total recoveries were ≥70%. C, HEK293/CD14/MD-2/TLR4 cells were incubated for 4 h with increasing amounts of LOS aggregates and LBP (100 ng/ml) in DMEM/0.1% HSA. Extracellular IL-8 accumulated in cell supernatants was determined by ELISA. Results are from one experiment with triplicate samples and are representative of two similar experiments.

Close modal

To compare the aggregation state of the isolated wt and underacylated endotoxins, gel filtration chromatography (Sephacryl S500) was used. As shown in Fig. 2,B, differences in the acylation state of NMB LOS were associated with differences in aggregation as judged by gel filtration. In contrast to LOSwt (Mr, >20 × 106), the majority of the agg of LOSmsbB eluted with an Mr of ∼1–2 × 106, whereas LOSAOAH agg were even smaller (Fig. 1 B).

The reduced activities of penta- and tetra-acylated lipid A species from NMB have been described previously (25, 28). To confirm that the underacylated material used in these experiments also displayed reduced bioactivity, isolated agg of LOSwt, LOSmsbB, and LOSAOAH were compared for their abilities to activate HEK293/CD14/MD-2/TLR4 cells by measuring extracellular accumulation of IL-8. As shown in Fig. 1,C, each LOS species produced dose-dependent cell activation. However, compared with LOSwt, the potencies of LOSAOAH and LOSmsbB were reduced at least 10-fold. The residual activities of LOSmsbB and LOSAOAH may be due to residual hexa-acylated LOS in these preparations, more so in LOSAOAH (∼10%; Fig. 2,A) (26) than in LOSmsbB (∼5%; Fig. 2 A) (25), as indicated by the presence of C12:0 remaining in AOAH-treated LOS and the slightly greater than predicted amount of 12:0 remaining in LOSmsbB. Cell activation by each LOS species was LBP dependent. Little or no activation of HEK293 cells by LOS was observed in the absence of CD14, MD-2, or TLR4.

Previous studies have indicated similar LBP-dependent delivery of wt (hexa-acylated) and AOAH-treated (tetra-acylated) LPS to CD14 (23). Therefore, the differences in activity of hexa- and tetra-acylated endotoxin presumably derive from an event after interaction with CD14 (23), e.g., transfer of endotoxin from CD14 to MD-2 and/or activation of TLR4 by endotoxin:MD-2. To test the former possibility, we compared LBP/sCD14-dependent formation of monomeric LOS:MD-2-His6 complexes from LOSwtagg, LOSmsbBagg, and LOSAOAHagg. To promote formation of LOS:sCD14 before addition of sMD-2, the purified LOSagg were first treated with LBP and sCD14 for 15 min at 37°C in the presence of albumin (24). After 15 min, recombinant sMD-2-His6 was added to the incubation mixture and incubated an additional 15 min at 37°C to permit formation of LOS:MD-2. Reaction conditions were chosen that produced ∼50% conversion of LOSwtagg to the monomeric LOSwt:MD-2 complex (latest eluting peak in Fig. 3,A) so that either more or less efficient generation of LOS:MD-2 from LOSmsbBagg and LOSAOAHagg could be detected. Remarkably, both the overall elution pattern as well as the apparent degree of formation of LOS:MD-2 from LOSmsbBagg, and LOSAOAHagg were virtually identical with that observed with LOSwtagg. The latest eluting LOS species (peak) in Fig. 3 A is the monomeric LOS:MD-2 complex: 1) the elution volume predicts an Mr of ∼25,000, closely corresponding to the predicted Mr of a monomeric complex of LOS and MD-2 (12); 2) this peak was only observed after incubation of LOSagg with LBP, sCD14, and sMD-12 (12); and 3) the yield of this peak from LOSwtagg, LOSmsbBagg, and LOSAOAHagg was increased by addition of a higher concentration of sMD-2 (data not shown). Therefore, these findings demonstrate a close similarity in the efficiency of formation of LOS:MD-2 complexes from LOSwtagg, LOSmsbBagg, and LOSAOAHagg. This was observed under both analytical (≤1 ng/ml LOS) and preparative (∼500 ng/ml) conditions.

FIGURE 3.

LBP/sCD14-dependent formation of LOS:MD-2 complexes from hexa-, penta-, and tetra-acylated LOS aggregates. A, LOSagg (500 ng, 100–120 pmol) from wt, msbB, and AOAH-treated LOS were incubated with LBP (0.03 μg, 0.5 pmol) and sCD14 (9 μg, 150 pmol) for 15 min at 37°C before the addition of either insect culture medium (dialyzed against PBS) containing sMD-2-His6 or partially purified sMD-2-His6. Samples were incubated for an additional 15 min at 37°C and then applied to a Sephacryl S200 HR (1.6 × 70 cm) column in PBS. Column fractions were analyzed for [3H]- or 14C]LOS, and results are expressed as a percentage of the total cpm recovered; recoveries were ≥70%. Results shown are from one experiment that is representative of three similar experiments. B, Fractions from the latest eluting major peak (denoted by arrow for LOSmsbB and LOSAOAH), presumably LOS:MD-2 (Mr, ∼25, 000), as well as a control (wt LOS:sCD14) were incubated with HisLink (Promega) resin for 1 h at 25°C and washed, and the absorbed material was eluted with 2% SDS and analyzed for the amount of radioactivity. Results are expressed as the percentage of LOS captured; the results shown are from three separate experiments. D, The [14C]fatty acid content of fractions from the ∼25,000 Mr peak (LOS:MD-2) and LOSwtagg and LOSAOAHagg was determined as described in Materials and Methods. TLC of the fatty acids isolated from LOSwtagg and LOSAOAHagg is shown, and data are expressed as described in Fig. 2 A.

FIGURE 3.

LBP/sCD14-dependent formation of LOS:MD-2 complexes from hexa-, penta-, and tetra-acylated LOS aggregates. A, LOSagg (500 ng, 100–120 pmol) from wt, msbB, and AOAH-treated LOS were incubated with LBP (0.03 μg, 0.5 pmol) and sCD14 (9 μg, 150 pmol) for 15 min at 37°C before the addition of either insect culture medium (dialyzed against PBS) containing sMD-2-His6 or partially purified sMD-2-His6. Samples were incubated for an additional 15 min at 37°C and then applied to a Sephacryl S200 HR (1.6 × 70 cm) column in PBS. Column fractions were analyzed for [3H]- or 14C]LOS, and results are expressed as a percentage of the total cpm recovered; recoveries were ≥70%. Results shown are from one experiment that is representative of three similar experiments. B, Fractions from the latest eluting major peak (denoted by arrow for LOSmsbB and LOSAOAH), presumably LOS:MD-2 (Mr, ∼25, 000), as well as a control (wt LOS:sCD14) were incubated with HisLink (Promega) resin for 1 h at 25°C and washed, and the absorbed material was eluted with 2% SDS and analyzed for the amount of radioactivity. Results are expressed as the percentage of LOS captured; the results shown are from three separate experiments. D, The [14C]fatty acid content of fractions from the ∼25,000 Mr peak (LOS:MD-2) and LOSwtagg and LOSAOAHagg was determined as described in Materials and Methods. TLC of the fatty acids isolated from LOSwtagg and LOSAOAHagg is shown, and data are expressed as described in Fig. 2 A.

Close modal

The latest eluting LOS species (Mr ∼25, 000), i.e., monomeric LOS:MD-2-His6 complex formed from LOSwtagg, LOSmsbBagg, and LOSAOAHagg were also assayed by cocapture of the radiolabeled LOS from these fractions by a metal-chelating resin. Fig. 3,B shows that, like LOSwt, complexes that contain LOSmsbB and LOSAOAH were also specially adsorbed by the resin, consistent with association of each LOS species with sMD-2-His6. The reduced capture of LOSmsbB by the resin paralleled the reduced stability of the LOSmsbB:MD-2 complex compared with LOSwt:MD-2 (see below). The reduced stability may be responsible for the reduced percentage of the LOS remaining bound to MD-2 during the incubations and washing of the metal-chelating resin as well as reduced access of the polyhistidine tag when MD-2 is bound to LOSmsbB, possibly due to aggregation of LOSmsbB:MD-2 complexes (see Fig. 4).

FIGURE 4.

Stability of various monomeric LOS:MD-2 complexes. Purified radiolabeled LOS:MD-2 for each type of LOS recovered after Sephacryl S200 chromatography (Fig. 3 A) were sterile filtered and stored at 4°C until rechromatography. Samples of 4–6 ng of LOS:MD-2 were either applied directly or preincubated at 37°C for the indicated time before chromatography (Sephacryl S200; 1.6 × 30 cm). Column fractions were analyzed for radioactivity, and results are expressed as a percentage of the total cpm recovered; recoveries were ≥70%. The profiles shown are representative of three separate experiments.

FIGURE 4.

Stability of various monomeric LOS:MD-2 complexes. Purified radiolabeled LOS:MD-2 for each type of LOS recovered after Sephacryl S200 chromatography (Fig. 3 A) were sterile filtered and stored at 4°C until rechromatography. Samples of 4–6 ng of LOS:MD-2 were either applied directly or preincubated at 37°C for the indicated time before chromatography (Sephacryl S200; 1.6 × 30 cm). Column fractions were analyzed for radioactivity, and results are expressed as a percentage of the total cpm recovered; recoveries were ≥70%. The profiles shown are representative of three separate experiments.

Close modal

Because the AOAH-treated LOS almost certainly contains some residual hexa-acylated endotoxin (Fig. 2,A) (25, 28), we compared the fatty acid contents of the starting LOSagg and the isolated LOS:MD-2 complex formed from AOAH-treated LOS. As shown in Fig. 3 C, the fatty acid compositions of the LOSAOAHagg and LOS:MD-2 complex formed from LOSAOAHagg were virtually identical, indicating little or no fatty acid selectivity in the sequence of reactions from aggregated LOS to formation of a monomeric LOS:MD-2 complex.

The findings described above indicate that the differences in potency among LOSwt, LOSmsbB, and LOSAOAH are not explained by differences in LBP/CD14-dependent delivery of LOS to MD-2. These findings suggest that differences in bioactivity correspond to differences in TLR4-dependent cell activation by monomeric LOS:MD-2, depending on the acylation state of the LOS complexed to MD-2. However, before testing this prediction directly by comparing activation of HEK293/TLR4 cells by LOSwt:MD-2, LOSmsbB:MD-2, and LOSAOAH:MD-2, it was necessary to confirm that each of these complexes had sufficient stability to remain intact during collection at room temperature, storage at 4°C, and incubation at 37°C as needed in cell activation assays. As shown in Fig. 4, LOSwt:MD-2 complexes were virtually unchanged after weeks at 4°C and after 6 h at 37°C. In fact, even overnight incubation at 37°C had little effect on the state of LOSwt:MD-2 (data not shown). LOSAOAH:MD-2 was also stable for at least 1 wk at 4°C, but gradually converted to a higher Mr species with incubation at 37°C. LOSmsbB:MD-2 was the least stable at both 4 and 37°C. These findings demonstrate time- and temperature-dependent differences in the stability of monomeric LOS:MD-2 complexes as a function of the acylation state of LOS. However, the stability of all three species was sufficient to permit direct comparison of their abilities to trigger TLR4-dependent cell activation.

To test the abilities of the various LOS:MD-2 complexes to activate cells through TLR4, HEK293/TLR4 cells were incubated with increasing amounts of the three isolated LOS:MD-2 complexes, and the degree of activation was measured by determining the extracellular IL-8 accumulated. We measured extracellular accumulation of IL-8 at 4 h, which reflects the effect of cell activation during the first 2–3 h of incubation (data not shown) at a point in time when each of the LOS:MD-2 complexes showed little or no aggregation at 37°C (Fig. 4; (data not shown). Parental HEK293 cells (no TLR4) were not activated by these complexes (data not shown). All three complexes produced dose-dependent TLR4-dependent cell activation (Fig. 5,A). However, compared with LOSwt:MD-2, the potencies of LOSAOAH:MD-2 and LOSmsbB:MD-2 were reduced at least 10-fold, comparable to the differences in LBP-, CD14-, MD-2-, and TLR4-dependent activity of the LOSwt, LOSmsbB, and LOSAOAH aggregates shown in Fig. 2 C. These findings strongly suggest that the differences in TLR4-dependent activity of hexa-, penta-, and tetra-acylated forms of meningococcal LOS do not derive from differences in interactions of LOS aggregates with LBP and CD14, but, rather, from differences in the interaction of the monomeric LOS:MD-2 complexes with TLR4.

FIGURE 5.

Bioactivities of hexa-, penta-, and tetra-acylated LOS:MD-2 complexes. A, Purified LOS:MD-2 complexes from wt (•), msbB (♦), and AOAH-treated (○) LOS were incubated with HEK293/TLR4 cells in HBSS+, HEPES, and 0.1% HSA for 4 h. Extracellular IL-8 was measured by ELISA. The x-axis indicates the amount of LOS added. Note that for LOSmsbB:MD-2 ≤50% of LOS remains as a monomeric complex (see Fig. 4). B, HEK293/TLR4 cells were incubated with increasing amounts of LOSmsbB:MD-2 with (•) or without (♦) LOS:MD-2 (0.2 ng/ml) for 4 h. Extracellular IL-8 was determined by ELISA. Data shown represent one experiment that is representative of three experiments, each containing duplicate samples.

FIGURE 5.

Bioactivities of hexa-, penta-, and tetra-acylated LOS:MD-2 complexes. A, Purified LOS:MD-2 complexes from wt (•), msbB (♦), and AOAH-treated (○) LOS were incubated with HEK293/TLR4 cells in HBSS+, HEPES, and 0.1% HSA for 4 h. Extracellular IL-8 was measured by ELISA. The x-axis indicates the amount of LOS added. Note that for LOSmsbB:MD-2 ≤50% of LOS remains as a monomeric complex (see Fig. 4). B, HEK293/TLR4 cells were incubated with increasing amounts of LOSmsbB:MD-2 with (•) or without (♦) LOS:MD-2 (0.2 ng/ml) for 4 h. Extracellular IL-8 was determined by ELISA. Data shown represent one experiment that is representative of three experiments, each containing duplicate samples.

Close modal

MD-2-reduced agonist properties of underacylated LOS:MD-2 could reflect reduced TLR4 association and/or reduced TLR4 activation after engagement of TLR4 by the underacylated LOS:MD-2 complex. Even with [3H]LOS preparations of relatively high specific radioactivity (5000 cpm/ng), measurable TLR4-dependent cell association of LOS:MD-2 required ≥8-h incubation at 37°C; such incubation conditions precluded testing of the underacylated LOS:MD-2 complexes because of their greater instability during prolonged incubation at 37°C. Therefore, we instead tested whether the least active LOS:MD-2 complex (LOSmsbB:MD-2) could inhibit activation of HEK293/TLR4 cells by LOSwt:MD-2, reasoning that only if LOSmsbB:MD-2 could bind TLR4 would an inhibitory effect be seen. As shown in Fig. 5 B, LOSmsbB:MD-2 produced a dose-dependent inhibition of TLR4-dependent cell activation by LOSwt:MD-2. Note that at >3 ng/ml LOSmsbB:MD-2, all the observed cell activation could be accounted for by the activity inherent in LOSmsbB:MD-2, which may be due to the presence of residual hexa-acylated LOS. These findings strongly suggest that an underacylated LOS:MD-2 complex such as LOSmsbB:MD-2 can compete with LOSwt:MD-2 for engagement of TLR4 and has reduced agonist properties, because the interaction with TLR4 does not as efficiently trigger TLR4 activation.

Variations in endotoxin potency have long been recognized (for review, see Refs.14, 15, 16, 17 and 29). This has stimulated great interest in the structural and molecular bases of these functional differences. This effort was initially driven by pharmacologic interests in generating compounds with either immunostimulatory or immunosuppressive properties. However, it is now apparent that variations in endotoxin structure within the lipid A region among different Gram-negative bacteria are more common than previously appreciated and may play an important role in defining specific features of host interactions with specific Gram-negative bacteria (14).

A prevailing view of the molecular basis of differing agonist/antagonist properties of endotoxin variants has been that variation of endotoxin (lipid A) structure confers differences in the supramolecular organization of aggregates of endotoxins that modify interactions with the LBP/CD14/MD-2/TLR4-dependent pathway of cell activation (for review see Refs.15 , 16 , 29 , and 30). However, it has also been demonstrated that functional differences between an endotoxin species that is a potent agonist and a partially deacylated derivative that is a potent antagonist involve interactions that occur after CD14 (22, 31, 32). Because, at least in solution, the active form of endotoxin complexed to CD14 is a monomeric protein:endotoxin complex (7, 24), these findings suggest variables that are independent of differences in the aggregate properties of the different endotoxin species. We have previously speculated that, at least for some endotoxin species, decreased acylation would decrease endotoxin agonist properties not by effects on endotoxin aggregate properties, but by altering properties of monomeric protein:endotoxin complexes that regulate delivery of endotoxin to MD-2 and/or activation of TLR4 (8, 12).

Using LOS isolated from the msbB mutant of NMB as a source of penta-acylated endotoxin and LOSwt deacylated by AOAH as a source of tetra-acylated endotoxin, we have shown that aggregates of both LOSmsbB and LOSAOAH, when treated with LBP, sCD14, and sMD-2, generate monomeric LOS:MD-2 complexes in a manner virtually indistinguishable from that seen with LOSwt (Fig. 3). Partial deacylation by AOAH of LOSwt is incomplete (Fig. 2,A), making it likely that some residual hexa-acylated (wt) LOS remains in LOSAOAH preparations (26). Nevertheless, LOSAOAH:MD-2 complexes have the identical fatty acid composition as LOSAOAH aggregates, also indicating that LBP, sCD14, and sMD-2 treatment of endotoxin aggregates to form monomeric endotoxin:MD-2 is equally efficient with tetra-acylated as with hexa-acylated LOS. It should be noted, as shown in Fig. 2, that the aggregate size of LOSmsbB and LOSAOAH differs markedly from that of LOSwt. The latter migrates with a Mr of ≥20 million, suggesting ≥4000 molecules of LOS/aggregate. In contrast, the chromatographic behavior of both LOSmsbB and LOSAOAH suggest that these species form aggregates of ≥200 LOS/aggregate. In the case of LOSAOAH, exposure to Triton X-100 during AOAH treatment to facilitate enzyme action accounts at least in part for the marked reduction in aggregate size (data not shown). Despite these differences in aggregate size, disaggregation by LBP/sCD14 to form LOS:sCD14 as a prelude to transfer of endotoxin to sMD-2 (12) (A. Teghanemt, unpublished observations) is closely similar (Fig. 3) and, hence, apparently independent of the initial size of the endotoxin aggregate. Thus, at least for these LOS variants, the aggregate properties of these species do not appear to be important determinants of their agonist potencies. Studies are in progress to determine whether, by contrast, the reduced agonist potency of hyperacylated (e.g., hepta-acylated) endotoxin variants or species containing long-chain fatty acids (15, 29) reflects reduced efficiency of LBP/CD14-dependent extraction and transfer of single molecules of endotoxin to CD14, as is generally needed for potent MD-2/TLR4-dependent cell activation.

Our observations contrast with a recent report by Tsuneyoshi et al. (33) that concluded that MD-2 could not bind penta-acylated endotoxin. In these studies, MD-2:LPS complexes could be recovered from wt Escherichia coli expressing recombinant MD-2 and hexa-acylated LPS, but not from lpxM (equivalent to msbB) E. coli that produced MD-2 and penta-acylated LPS. Many differences distinguish the experimental design used in that study from our assays of MD-2:endotoxin interactions. Perhaps most important are the differences in the manner of endotoxin presentation to MD-2. We have previously shown that the reactivity of MD-2 with LOS (or LPS) is markedly enhanced by presentation of endotoxin as a monomeric complex with CD14 (12). This is what physiologically occurs after LBP- and CD14-dependent interactions with endotoxin (8). By contrast, interactions of MD-2 and LPS within E. coli or after extraction bear little resemblance to the natural and ordered sequence of interactions that take place in the host. Moreover, the reduced stability we observed for LOSmsbB:MD-2 complexes could mean that LPSlpxM:MD-2 complexes from E. coli, even if formed, might not withstand the treatment used for coprecipitation analysis.

Because we were able to recover and purify monomeric complexes of LOSwt:MD-2, LOSmsbB:MD-2 and LOSAOAH:MD-2, we were able to demonstrate directly that for these meningococcal LOS species, the agonist potency of LOS:MD-2 complexes depends on the acylation state of the endotoxin bound to MD-2 (Fig. 5,A). The differences in dose-dependent activation of HEK293/TLR4 cells by these LOS:MD-2 complexes mirrored the differences in LBP-dependent bioactivity of LOSwt, LOSmsbB, and LOSAOAH aggregates toward HEK293/mCD14/MD-2/TLR4 cells, further supporting the conclusion that the differences in activity of different LOS aggregates are due to differences conferred after extraction and delivery of endotoxin monomers to MD-2. Although the reduced stability of the LOS:MD-2 complexes containing underacylated endotoxins could contribute to their reduced activity, we assayed cell activation within a time frame (4 h) in which an appreciable percentage of the complexes remained in monomeric form (Fig. 4). Aggregates that formed during storage had no effect on bioactivity. We cannot fully exclude the possibility that complexes of MD-2 with underacylated endotoxins bind less avidly to TLR4, leading to less potent TLR4-dependent cell activation. However, the ability of the LOSmsbB:MD-2 complex to inhibit TLR4-dependent cell activation by LOSwt:MD-2 seems most compatible with a model (Fig. 6) in which the occupation of MD-2 by an underacylated endotoxin species leads to occupation of TLR4 by MD-2 in a form that cannot efficiently trigger receptor activation (e.g., oligomerization) (34).

FIGURE 6.

Model of agonist and antagonist properties of wt and underacylated endotoxin. Reduced acylation of endotoxin, such as by AOAH treatment, does not decrease the efficiency of LBP-, CD14-, and MD-2-dependent transfer of endotoxin from endotoxin aggregates to MD-2. These transfer reactions in aqueous solution also depend on albumin (1224 ). We propose that hexa-acylated, but not underacylated, endotoxins promote conformational/configurational changes in MD-2 and/or TLR4 that lead to oligomerization of TLR4, an event necessary for cellular activation (363738 ).

FIGURE 6.

Model of agonist and antagonist properties of wt and underacylated endotoxin. Reduced acylation of endotoxin, such as by AOAH treatment, does not decrease the efficiency of LBP-, CD14-, and MD-2-dependent transfer of endotoxin from endotoxin aggregates to MD-2. These transfer reactions in aqueous solution also depend on albumin (1224 ). We propose that hexa-acylated, but not underacylated, endotoxins promote conformational/configurational changes in MD-2 and/or TLR4 that lead to oligomerization of TLR4, an event necessary for cellular activation (363738 ).

Close modal

Finally, it should be noted that the agonist potencies of the various LOS:MD-2 complexes we studied were directly related to the stability of these complexes (i.e., LOSwt:MD-2 > LOSAOAH:MD-2 ≥ LOSmsbB:MD-2; Figs. 4 and 5,A). Thus, unless binding of TLR4 by MD-2 markedly alters the stability of endotoxin-MD-2 interactions, our findings suggest that TLR4 activation is not dependent upon a transfer of endotoxin from MD-2 to TLR4. We propose, instead, that endotoxin is an allosteric regulator of MD-2, inducing conformational changes in MD-2 that are needed to induce changes in TLR4 resulting in receptor activation (Fig. 6). The computational model of MD-2 suggested by Ichikawa et al. (35) supports this view. Accordingly, we predict that underacylated endotoxins function as antagonists by occupying MD-2 without inducing the conformational changes needed to trigger TLR4 activation and oligomerization (36, 37, 38) (Fig. 6). This is consistent with the findings of Akashi et al. (39), who showed that CD14-dependent binding of the lipid A antagonist E5531 does not induce the changes in mAb reactivity of MD-2/TLR4 that accompany cell binding and activation by an agonist. Studies are in progress using the wt and underacylated endotoxin:MD-2 complexes described in this study to test this hypothesis.

We thank Dr. Robert Munford (University of Texas Southwestern Medical Center, Dallas, TX) for the gift of purified AOAH; Dr. Jesse Chow (Eisai Research Institute, Andover, MA) for the gift of HEK293 and HEK293/TLR4 cells; Dr. Michael Apicella (University of Iowa, Iowa City, IA) for the msbB derivative of NMB, NMBA11K3cap; and Xoma (Berkeley, CA) for rLBP and sCD14.

The authors have no financial conflict of interest.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

This work was supported by U.S. Public Health Service Grants PO144642 and AI59372 (to J.P.W.) and a Veterans Administration Merit Review grant (to T.L.G.).

3

Abbreviations used in this paper: LBP, LPS-binding protein; agg, aggregate; AOAH, acyloxyacylhydrolase; HEK, human embryonic kidney; HSA, human serum albumin; LOS, lipooligosaccharide; NMB, Neisseria meningitidis serotype B; sCD14, soluble CD14; wt, wild type; MD-2, myeloid differentiation protein 2.

1
Beutler, B..
2002
. TLR4 as the mammalian endotoxin sensor.
Curr. Top. Microbiol. Immunol.
270
:
109
.-120.
2
Ulevitch, R. J., P. S. Tobias.
1999
. Recognition of Gram-negative bacteria and endotoxin by the innate immune system.
Curr. Opin. Immunol.
11
:
19
.-22.
3
Heumann, D., T. Roger.
2002
. Initial responses to endotoxins and Gram-negative bacteria.
Clin. Chim. Acta
323
:
59
.-72.
4
Nagai, Y., S. Akashi, M. Nagafuku, M. Ogata, Y. Iwakura, S. Akira, T. Kitamura, A. Kosugi, M. Kimoto, K. Miyake.
2002
. Essential role of MD-2 in LPS responsiveness and TLR4 distribution.
Nat. Immunol.
3
:
667
.-672.
5
Hailman, E., H. S. Lichenstein, M. M. Wurfel, D. S. Miller, D. A. Johnson, M. Kelley, L. A. Busse, M. M. Zukowski, S. D. Wright.
1994
. Lipopolysaccharide (LPS)-binding protein accelerates the binding of LPS to CD14.
J. Exp. Med.
179
:
269
.-277.
6
Viriyakosol, S., P. S. Tobias, R. L. Kitchens, T. N. Kirkland.
2001
. MD-2 binds to bacterial lipopolysaccharide.
J. Biol. Chem.
276
:
38044
.-38051.
7
Giardina, P. C., T. Gioannini, B. A. Buscher, A. Zaleski, D. S. Zheng, L. Stoll, A. Teghanemt, M. A. Apicella, J. Weiss.
2001
. Construction of acetate auxotrophs of Neisseria meningitidis to study host-meningococcal endotoxin interactions.
J. Biol. Chem.
276
:
5883
.-5891.
8
Gioannini, T. L., A. Teghanemt, K. A. Zarember, J. P. Weiss.
2003
. Regulation of interactions of endotoxin with host cells.
J. Endotoxin Res.
9
:
401
.-408.
9
Yu, B., E. Hailman, S. D. Wright.
1997
. Lipopolysaccharide binding protein and soluble CD14 catalyze exchange of phospholipids.
J. Clin. Invest.
99
:
315
.-324.
10
Tapping, R. I., P. S. Tobias.
1997
. Cellular binding of soluble CD14 requires lipopolysaccharide (LPS) and LPS-binding protein.
J. Biol. Chem.
272
:
23157
.-23164.
11
Muroi, M., T. Ohnishi, K. Tanamoto.
2002
. MD-2, a novel accessory molecule, is involved in species-specific actions of Salmonella lipid A.
Infect. Immun.
70
:
3546
.-3550.
12
Gioannini, T. L., A. Teghanemt, D. Zhang, N. P. Coussens, W. Dockstader, S. Ramaswamy, J. P. Weiss.
2004
. Isolation of an endotoxin-MD-2 complex that produces Toll-like receptor 4-dependent cell activation at picomolar concentrations.
Proc. Natl. Acad. Sci. USA
101
:
4186
.-4191.
13
Kennedy, M. N., G. E. Mullen, C. A. Leifer, C. Lee, A. Mazzoni, K. N. Dileepan, D. M. Segal.
2004
. A complex of soluble MD-2 and lipopolysaccharide serves as an activating ligand for Toll-like receptor 4.
J. Biol. Chem.
279
:
34698
.-34704.
14
Miller, S. I., R. K. Ernst, M. W. Bader.
2005
. LPS, TLR4 and infectious disease diversity.
Nat. Rev. Microbiol.
3
:
36
.-46.
15
Alexander, C., E. T. Rietschel.
2001
. Bacterial lipopolysaccharides and innate immunity.
J. Endotoxin Res.
7
:
167
.-202.
16
Brandenburg, K., A. Wiese.
2004
. Endotoxins: relationships between structure, function, and activity.
Curr. Top. Med. Chem.
4
:
1127
.-1146.
17
Darveau, R. P..
1998
. Lipid A diversity and the innate host response to bacterial infection.
Curr. Opin. Microbiol.
1
:
36
.-42.
18
Caroff, M., D. Karibian, J. M. Cavaillon, N. Haeffner-Cavaillon.
2002
. Structural and functional analyses of bacterial lipopolysaccharides.
Microbes Infect.
4
:
915
.-926.
19
Hajjar, A. M., R. K. Ernst, J. H. Tsai, C. B. Wilson, S. I. Miller.
2002
. Human Toll-like receptor 4 recognizes host-specific LPS modifications.
Nat. Immunol.
3
:
354
.-359.
20
Lien, E., T. K. Means, H. Heine, A. Yoshimura, S. Kusumoto, K. Fukase, M. J. Fenton, M. Oikawa, N. Qureshi, B. Monks, et al
2000
. Toll-like receptor 4 imparts ligand-specific recognition of bacterial lipopolysaccharide.
J. Clin. Invest.
105
:
497
.-504.
21
Poltorak, A., P. Ricciardi-Castagnoli, S. Citterio, B. Beutler.
2000
. Physical contact between lipopolysaccharide and Toll-like receptor 4 revealed by genetic complementation.
Proc. Natl. Acad. Sci. USA
97
:
2163
.-2167.
22
Backhed, F., S. Normark, E. K. Schweda, S. Oscarson, A. Richter-Dahlfors.
2003
. Structural requirements for TLR4-mediated LPS signalling: a biological role for LPS modifications.
Microbes Infect.
5
:
1057
.-1063.
23
Kitchens, R. L., R. S. Munford.
1995
. Enzymatically deacylated lipopolysaccharide (LPS) can antagonize LPS at multiple sites in the LPS recognition pathway.
J. Biol. Chem.
270
:
9904
.-9910.
24
Gioannini, T. L., D. Zhang, A. Teghanemt, J. P. Weiss.
2002
. An essential role for albumin in the interaction of endotoxin with lipopolysaccharide-binding protein and sCD14 and resultant cell activation.
J. Biol. Chem.
277
:
47818
.-47825.
25
Post, D. M., M. R. Ketterer, N. J. Phillips, B. W. Gibson, M. A. Apicella.
2003
. The msbB mutant of Neisseria meningitidis strain NMB has a defect in lipooligosaccharide assembly and transport to the outer membrane.
Infect. Immun.
71
:
647
.-655.
26
Munford, R. S., A. L. Erwin.
1992
. Eukaryotic lipopolysaccharide deacylating enzyme.
Methods Enzymol.
209
:
485
.-492.
27
Yang, H., D. W. Young, F. Gusovsky, J. C. Chow.
2000
. Cellular events mediated by lipopolysaccharide-stimulated toll-like receptor 4. MD-2 is required for activation of mitogen-activated protein kinases and Elk-1.
J. Biol. Chem.
275
:
20861
.-20866.
28
van der Ley, P., L. Steeghs, H. J. Hamstra, J. ten Hove, B. Zomer, L. van Alphen.
2001
. Modification of lipid A biosynthesis in Neisseria meningitidis lpxL mutants: influence on lipopolysaccharide structure, toxicity, and adjuvant activity.
Infect. Immun.
69
:
5981
.-5990.
29
Erridge, C., E. Bennett-Guerrero, I. R. Poxton.
2002
. Structure and function of lipopolysaccharides.
Microbes Infect.
4
:
837
.-851.
30
Seydel, U., L. Hawkins, A. B. Schromm, H. Heine, O. Scheel, M. H. Koch, K. Brandenburg.
2003
. The generalized endotoxic principle.
Eur. J. Immunol.
33
:
1586
.-1592.
31
Kitchens, R. L., R. J. Ulevitch, R. S. Munford.
1992
. Lipopolysaccharide (LPS) partial structures inhibit responses to LPS in a human macrophage cell line without inhibiting LPS uptake by a CD14-mediated pathway.
J. Exp. Med.
176
:
485
.-494.
32
Kitchens, R. L., R. S. Munford.
1995
. Modulation of protein tyrosine phosphorylation in human cells by LPS and enzymatically deacylated LPS.
Prog. Clin. Biol. Res.
392
:
353
.-363.
33
Tsuneyoshi, N., K. Fukudome, J. Kohara, R. Tomimasu, J. F. Gauchat, H. Nakatake, M. Kimoto.
2005
. The functional and structural properties of MD-2 required for lipopolysaccharide binding are absent in MD-1.
J. Immunol.
174
:
340
.-344.
34
Saitoh, S., S. Akashi, T. Yamada, N. Tanimura, F. Matsumoto, K. Fukase, S. Kusumoto, A. Kosugi, K. Miyake.
2004
. Ligand-dependent Toll-like receptor 4 (TLR4)-oligomerization is directly linked with TLR4-signaling.
J. Endotoxin Res.
10
:
257
.-260.
35
Ichikawa, S., T. Takai, T. Inoue, T. Yuuki, Y. Okumura, K. Ogura, F. Inagaki, H. Hatanaka.
2005
. NMR study on the major mite allergen Der f 2: its refined tertiary structure, epitopes for monoclonal antibodies and characteristics shared by ML protein group members.
J. Biochem.
137
:
255
.-263.
36
Saitoh, S., S. Akashi, T. Yamada, N. Tanimura, M. Kobayashi, K. Konno, F. Matsumoto, K. Fukase, S. Kusumoto, Y. Nagai, et al
2004
. Lipid A antagonist, lipid IVa, is distinct from lipid A in interaction with Toll-like receptor 4 (TLR4)-MD-2 and ligand-induced TLR4 oligomerization.
Int. Immunol.
16
:
961
.-969.
37
Visintin, A., E. Latz, B. G. Monks, T. Espevik, D. T. Golenbock.
2003
. Lysines 128 and 132 enable LPS binding to MD-2, leading to Toll-like receptor 4 aggregation and signal transduction.
J. Biol. Chem.
278
:
48313
.-48320.
38
Zhang, H., P. N. Tay, W. Cao, W. Li, J. Lu.
2002
. Integrin-nucleated Toll-like receptor (TLR) dimerization reveals subcellular targeting of TLRs and distinct mechanisms of TLR4 activation and signaling.
FEBS Lett.
532
:
171
.-176.
39
Akashi, S., S. Saitoh, Y. Wakabayashi, T. Kikuchi, N. Takamura, Y. Nagai, Y. Kusumoto, K. Fukase, S. Kusumoto, Y. Adachi, et al
2003
. Lipopolysaccharide interaction with cell surface Toll-like receptor 4-MD-2: higher affinity than that with MD-2 or CD14.
J. Exp. Med.
198
:
1035
.-1042.