Endotoxin-binding Proteins Modulate the Susceptibility of Bacterial Endotoxin to Deacylation by Acyloxyacyl Hydrolase*

Acyloxyacyl hydrolase (AOAH) is an eukaryotic lipase that partially deacylates and detoxifies Gram-negative bacterial lipopolysaccharides and lipooligosaccharides (LPSs or LOSs, endotoxin) within intact cells and inflammatory fluids. In cell lysates or as purified enzyme, in contrast, detergent is required for AOAH to act on LPS or LOS (Erwin, A. L., and Munford, R. S. (1990) J. Biol. Chem. 265, 16444–16449 and Katz, S. S., Weinrauch, Y., Munford, R. S., Elsbach, P., and Weiss, J. (1999) J. Biol. Chem. 274, 36579–36584). We speculated that the sequential interactions of endotoxin (E) with endotoxin-binding proteins (lipopolysaccharide-binding protein (LBP), CD14, and MD-2) might produce changes in endotoxin presentation that would allow AOAH greater access to its substrate, lipid A. To test this hypothesis, we measured the activity of purified AOAH against isolated, metabolically labeled meningococcal LOS and Escherichia coli LPS that were presented either as aggregates (LOSagg or LPSagg) ± LBP or as monomeric protein (sCD14 or MD-2)-endotoxin complexes. Up to 100-fold differences in the efficiency of endotoxin deacylation by AOAH were observed, with the following rank order of susceptibility to AOAH: E:sCD14 ≥endotoxin aggregates (Eagg):LBP (molar ratio of E/LBP 100:1) ≫Eagg, Eagg:LBP (E/LBP ∼1, mol/mol), or E:MD-2. AOAH treatment of LOS-sCD14 produced partially deacylated LOS still complexed with sCD14. The underacylated LOS complexed to sCD14 transferred to MD-2 and thus formed a complex capable of preventing TLR4 activation. These findings strongly suggest that LBP- and CD14-dependent extraction and transfer of endotoxin monomers are accompanied by increased exposure of fatty acyl chains within lipid A and that the acyl chains are then sequestered when LOS binds MD-2. The susceptibility of the monomeric endotoxin-CD14 complex to AOAH may help constrain endotoxin-induced TLR4 activation when endotoxin and membrane CD14 are present in excess of MD-2/TLR-4.

Tissue invasion by even minute quantities of many Gramnegative bacteria (GNB) 2 initiates rapid mobilization of the innate immune responses of the host. In these circumstances, both GNB recognition and many responses depend upon activation of the exquisitely sensitive Toll-like receptor 4 (TLR4) by endotoxins, structurally unique and abundant glycolipids that occupy much of the outer leaflet of the GNB outer membrane (3). Maximal TLR4-dependent host responses to endotoxin are orchestrated through a sequential set of interactions of endotoxin with lipopolysaccharide-binding protein (LBP), membrane or soluble CD14, and soluble or TLR4-associated MD-2 (3)(4)(5). Although timely mobilization of host responses is essential, equally important is the regulation of the duration and intensity of host responses to endotoxin to prevent over-exuberant and sustained responses that can result in severe pathological consequences (6).
One mechanism of dampening host responses to endotoxin is for the host to modify endotoxin itself, converting it from a potent TLR4 agonist to a much weaker agonist with antagonistic properties. To date, the best described host endotoxin-detoxifying enzyme is acyloxyacyl hydrolase (AOAH) (6). AOAH reduces endotoxin activity by catalyzing the release of secondary fatty acyl chains that are attached to primary 3-hydroxy fatty acyl chains within the bioactive lipid A region (7,8). AOAH thus converts hexaacylated endotoxin species that are potent TLR4 agonists to pentaacylated or tetraacylated forms that have reduced or no agonist properties (9 -13) and, at least in vitro, possess the ability to inhibit TLR4 activation by intact, hexaacylated species (10,11,13,14). Recent studies indicate that hexaacylated and partially deacylated or underacylated endotoxin react similarly with LBP, CD14, and MD-2 to form monomeric complexes with MD-2 (13). However, only hexaacylated endotoxin-human MD-2 is a potent TLR4 agonist (13,14).
AOAH-dependent deacylation of endotoxin has been demonstrated in vivo (1,15,16) and in complex in vitro settings that roughly simulate the extracellular and intracellular conditions of inflammatory fluids (1,2,15,(17)(18)(19). In contrast, either as purified enzyme or as part of a cell lysate, AOAH is virtually * This work was supported by U.S. Public Health Service Grants PO144642 and inactive against purified or membrane-bound endotoxin in buffered salt solutions in the absence of detergent (20,21). This suggests that AOAH alone is unable to act efficiently on endotoxin when the fatty acyl chains of the amphipathic endotoxin are buried within pure supramolecular aggregates or the bacterial outer membrane. In contrast, the ability of AOAH to function physiologically suggests that other endotoxin binding molecules may act as "physiological detergents" to facilitate access of AOAH to the acyloxyacyl bonds that link the non-hydroxylated fatty acyl chains to lipid A.
The delivery of endotoxin to MD-2, as is needed for TLR4 activation, is markedly enhanced by LBP-and CD14-dependent extraction and transfer of endotoxin monomers from purified endotoxin aggregates (E agg ) (4) or the bacterial outer membrane (22). These steps are also albumin-dependent (23), suggesting that E agg (or membrane-bound endotoxin)-LBP and monomeric endotoxin-CD14 intermediates present endotoxin in a manner that partially exposes the fatty acyl chains and so require albumin to maintain endotoxin solubility in an aqueous environment. By making use of metabolically radiolabeled endotoxin from Neisseria meningitidis serotype B (24) and from Escherichia coli, we now show that AOAH can act on endotoxin-LBP aggregates and, to an even greater extent, on monomeric endotoxin-sCD14 complexes, but much less on monomeric endotoxin-MD-2. These findings strongly support a model in which LBP-and CD14-dependent extraction and transfer of endotoxin expose fatty acyl chains that then become less accessible when endotoxin binds MD-2. The ability of AOAH to act on these protein-endotoxin intermediates provides a mechanism by which partial deacylation of endotoxin can take place before transfer to MD-2 and thus down-regulate TLR4 activation by endotoxin.

EXPERIMENTAL PROCEDURES
LBP, BPI, and sCD14 were provided by Xoma (US) LLC (Berkeley, CA). Recombinant soluble human MD-2 containing a hexapolyhistidine tag on the carboxyl-terminal end was prepared using infection of High Five insect cells with baculovirus containing the cDNA for human MD-2 as described (25). Recombinant human AOAH was a generous gift from Zymo-Genetics (Seattle, WA). Chromatography matrices were purchased from GE Healthcare. Human serum albumin (HSA) was obtained as an endotoxin-free, 25% stock solution (Baxter Healthcare Corp., Glendale, CA). A soluble truncated CD14 (amino acids 1-156) containing a hexapolyhistidine tag at the carboxyl terminus was generated by infection of High Five insect cells with baculovirus containing the cDNA for this protein derived from recombination with the vector pBAC11 (Novagen) using the procedure previously described for soluble MD-2 (25). Insect supernatant containing sCD14-(1-156) was dialyzed against 20 mM phosphate, pH 7.4, 0.5 M NaCl before adsorption to Ni Fast Flow-Sepharose resin (GE Healthcare). The protein was purified on the AKTA Explorer 100 FPLC using an imidazole gradient for elution of adsorbed protein.
HisLink resin (Promega) was used analytically to capture complexes by the method previously described (25). Note no imidazole was used in the binding buffer or rinses but buffer did contain 0.1% bovine serum albumin.

Preparation of Metabolically Labeled Endotoxin and Endotoxin-Protein Complexes-[ 3 H]LOS
(specific activity 25,000 cpm/pmol) was isolated from an acetate auxotroph of N. meningitidis serogroup B after metabolic labeling as previously described (24). LOS agg (apparent M r Ͼ 20 million) used as starting material to generate complexes was prepared using Sephacryl HR S500 or by ultracentrifugation (24). Aggregates containing various concentrations of LBP (LOS agg -LBP) or BPI (LOS-BPI agg ) to be used in AOAH assays were prepared in situ (0.   For determination of the ability of AOAH-treated [ 14 C]LOS-sCD14 to transfer partially deacylated [ 14 C]LOS to MD-2, [ 14 C]LOS-sCD14 (40 nM) was treated with AOAH (concentration adjusted based on method as described below) for 4 h at 37°C. An aliquot (10 l) was removed to evaluate the extent of deacylation by measuring ethanol-soluble radioactivity (i.e. 14 C-free fatty acids) described below. The remainder of the sample was incubated with insect culture medium containing MD-2-His 6 for 30 min at 37°C. The reaction mixture was resolved on Sephacryl HR S200 in PBS, pH 7.4, and the fractions corresponding to [ 14 C]LOS-MD-2 were subjected to 14 C-fatty acid analysis. [ 14 C]lipopolysaccharide (LPS) (specific activity ϳ6,000cpm/pmol) was isolated from an acetate auxotroph generated in E. coli BW25113 (BW25113 aceE). This strain was created by deletion-insertion mutagenesis according to the method of Datsenko and Wanner (26) resulting in insertion of a chloramphenicol acetyltransferase (cat) gene cassette within the aceE gene. Disruption of the aceE gene and insertion of the chloramphenicol acetyltransferase cassette was confirmed by PCR analysis and growth was shown to be acetate-dependent. Analysis of the 14 C-fatty acids released by acid and alkaline hydrolysis of isolated LPS was performed as described below for LOS to determine the specific activity and composition of 14  In gel filtration chromatography, fractions (1 ml) were collected (flow rate 0.5 ml/min) at room temperature using AKTA Purifier TM FPLC. 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 and LPS were Ͼ70%. All solutions used were pyrogen-free and sterilefiltered. After chromatography, selected peak fractions to be used in bioassays were pooled and passed through sterile syringe filters (0.22 or 0.45 m) with greater than 90% recovery of radiolabeled material in the sterile filtrate. Fractions were stored under sterile conditions at 4°C until needed. Sephacryl S200 columns were calibrated with Bio-Rad gel filtration standards that included thyroglobulin (V o ), ␥-globulin, human serum albumin, ovalbumin, myoglobin, and vitamin B12 (V i ).

JOURNAL OF BIOLOGICAL CHEMISTRY 7879
NaCl, 25% glycerol, 0.05% Triton X-100) (27) immediately prior to usage and added to samples as indicated. The volume of AOAH in storage buffer added to samples was Ͻ1% of the total incubation volume. In addition to samples prepared in HBSS ϩ , pH 7.4, 10 mM HEPES, 0.1% HSA, LOS agg was incubated with AOAH in 20 mM sodium acetate, pH 6, 150 mM NaCl, 5 mM CaCl 2 , 0.1% bovine serum albumin, 0.1% Triton X-100 (27) (27). The quantity of fatty acid released was evaluated using the ethanol precipitation method (27) wherein undigested and partially deacylated endotoxin are precipitated and released free fatty acids remain in the ethanol supernatant. Briefly, after incubation Ϯ AOAH, samples (100 l) were treated with 0.2 ml of ice-cold ethanol and kept at Ϫ20°C overnight. Samples were spun for 10 min at 14,000 ϫ g. Supernatants were removed; the pellets were washed with an additional 0.2 ml of ice-cold ethanol, incubated for 1 h at 0°C, and spun for 10 min at 14,000 ϫ g. The supernatant and rinses were combined for analysis of released radiolabeled free fatty acids by liquid scintillation spectroscopy. Pellets were treated with 0.2 ml of 2% SDS for 30 min at 37°C and then also evaluated by liquid scintillation spectroscopy. Overall recovery of radioactivity was Ն90%. The results of reaction with AOAH are reported as % deacylation. Because only two of six moles of fatty acid of LOS/LPS attached to lipid A can be released by AOAH, 33% fatty acid release was defined as 100% deacylation.
Fatty Acid Analysis-The 14 C fatty acid content of various 14 C-endotoxin and 14 C-endotoxin-protein preparations was analyzed as previously described (22,24). In brief, these complexes were subjected to acid hydrolysis (4 N HCl, 90°C), neutralization, and base hydrolysis (4 N NaOH), followed by Bligh-Dyer extraction and recovery of the released 14 C-labeled fatty acids in the chloroform phase. The 14 C-fatty acids were resolved by reversephase TLC (0.2 mm HPTLC, RP-18; Merck) using acetonitrile/acetic acid (1/1) as the solvent system. 14 Cfatty acids were identified by comparison to migration of 14 C-fatty acid standards. Analysis was done using the Typhoon imaging system (GE Healthcare).

LBP and sCD14
Increase LOS Susceptibility to AOAH-To determine whether LBP-and/or CD14-dependent alterations of endotoxin presentation render endotoxin more sensitive to AOAH, we examined the ability of AOAH to deacylate purified meningococcal LOS when LOS was presented either alone, as purified LOS agg , after treatment with LBP to produce LOS agg -LBP aggregates, or after sequential treatment with LBP and sCD14 to yield monomeric LOS-sCD14 complexes. All assays were performed in HBSS ϩ , 10 mM HEPES, 0.1% HSA, pH 7.4. The susceptibility of LOS to AOAH was markedly increased by pretreatment of LOS agg with LBP ϩ sCD14 to convert LOS agg to monomeric LOS-sCD14 complex (Fig. 1A) (23). Incubation of LOS agg with LBP alone also markedly increased the susceptibility of LOS to AOAH at specific LBP:LOS ratios (Fig. 1B). The dose-dependence of LBP alone on AOAH susceptibility of LOS agg showed a complex bell-shaped curve (Fig. 1B) with maximum AOAH activity at a molar ratio of ϳ1 mol LBP/100 mol LOS. At the concentration of LBP most favorable for AOAH activity, there was no detectable change in the gel sieving profile of the LOS agg (i.e. no detectable disaggregation of LOS agg ) (28). The bell-shaped LBP dose-dependence of AOAH activity on LOS agg closely paralleled the bell-shaped LBP dosedependence for extraction and delivery of LOS monomers to sCD14 (Fig. 1C). This suggests that LBP-catalyzed extraction and delivery of endotoxin monomers to sCD14 produce alter- ations in the orientation of endotoxin within LOS agg that favor access of AOAH to endotoxin, independent of CD14. In contrast to the effects of LBP, the closely related endotoxin-binding protein bactericidal permeability increasing protein (BPI) did not promote formation of LOS-sCD14 (28) or significantly increase LOS agg susceptibility to AOAH (Fig. 1B).
Under optimal conditions, the susceptibility of LOS agg -LBP (100:1 mol/mol) and LOS-sCD14 to AOAH appeared to be similar (Fig. 1, A and B). We focused further biochemical characterization on the substrate properties of LOS-sCD14 because of the greater molecular homogeneity of LOS-sCD14 in comparison to LOS agg -LBP. Fig. 2 compares the rate of fatty acid release versus substrate concentration for two substrates, LOS-sCD14 and LOS agg . The data demonstrate a Ͼ10-fold increase in AOAH sensitivity conferred by extraction and delivery of endotoxin monomers from LOS agg to LOS-sCD14. Lineweaver-Burk analysis of the data for LOS-sCD14 indicated that the K m is ϳ5 nM and the V max is 9.1 Ϯ 0.05 fmol/min (n ϭ 3).

Reaction of LOS-sCD14 with AOAH Yields Partially Deacylated LOS AOAH -sCD14 Complex That Can React with MD-2 to
Generate LOS AOAH -MD-2-The ability of LBP and CD14 to promote AOAH-dependent deacylation of LOS suggested a mechanism by which non-covalent host protein-mediated modifications in lipid A presentation might facilitate not only TLR4 activation but also AOAH-mediated lipid A deacylation that can reduce TLR4 activation (9 -14). Underacylated and partially deacylated endotoxin can be transferred from CD14 to MD-2 (13, 14, 29) and the underacylated AOAH-treated endotoxin-MD-2 complex interacts with TLR4 without efficiently inducing TLR4 activation (13). Thus, provided the partially deacylated LOS remains bound to CD14 during and after AOAH treatment, deacylation of monomeric endotoxin-sCD14 complex (e.g. LOS-sCD14) could provide substrate for generation of AOAH-treated endotoxin-MD-2 complex (e.g. LOS AOAH -MD-2) that can then act as a TLR4 antagonist.
Whether or not deacylated LOS remained associated with sCD14 could not be tested readily by comparing the gel sieving properties of the LOS-sCD14 complex before and after treatment with AOAH because of the closely similar M r of sCD14 and AOAH. Therefore, we made use of a truncated form of recombinant CD14 (amino acid residues 1-156) containing a carboxyl-terminal hexapolyhistidine tag (sCD14-(1-156)-His 6 ). This recombinant sCD14-(1-156)-His 6 fully retains the reactivity of full-length sCD14 (356 residues) toward LOS agg -LBP and MD-2 (data not shown). LOS-sCD14-(1-156)-His 6 was also a good substrate for AOAH (Fig. 3A). After AOAH treatment, the majority of the To verify that treatment of LOS-sCD14 with AOAH yielded partially deacylated LOS-sCD14 complex capable of transferring the partially deacylated LOS to MD-2, [ 14 C]LOS-sCD14 (full-length sCD14) was deacylated with AOAH, subsequently treated with MD-2-His 6 , and the product(s) separated using Sephacryl S200 (Fig. 4A). Fatty acid analysis of the peak fractions of [ 14 C]LOS corresponding to [ 14 C]LOS-MD-2 by migration on Sephacryl S200 indicated that the recovered LOS-MD-2 contained partially deacylated LOS (Fig. 4B). Experimental conditions were used that yielded ϳ60% release of the nonhydroxylated fatty acid (12:0) from LOS-sCD14 (data not shown). Because cleavage by AOAH of the two non-hydroxylated fatty acids present in each molecule of endotoxin normally follows in rapid succession (2,19,27), the major product under our experimental conditions is likely tetraacylated LOS-sCD14 with ϳ40% of the hexaacylated LOS-sCD14 exposed to 4 T. L. Gioannini, data not shown. AOAH remaining undigested. The close match in 14 C-fatty acid composition of the AOAH-treated [ 14 C]LOS AOAH -sCD14 and the recovered [ 14 C]LOS AOAH -MD-2 suggests, therefore, essentially equal efficiency of transfer of undigested and partially deacylated LOS from sCD14 to MD-2, as was previously reported (13).
Monomeric Endotoxin-MD-2 Complex Is Relatively Resistant to AOAH-In contrast to LOS-sCD14, monomeric LOS-MD-2 complexes do not depend upon albumin to maintain aqueous solubility and induce TLR4-dependent cell activation. Therefore, we have speculated that the fatty acids of lipid A are more fully buried within a deeper hydrophobic cavity within MD-2 and so might be a poorer substrate for AOAH. To test this hypothesis, we compared the effect of increasing concentrations of AOAH on LOS-sCD14 and LOS-MD-2 and, for comparison, unmodified LOS agg . As predicted, LOS-MD-2, in comparison to LOS-sCD14, showed much reduced sensitivity to added AOAH (Fig. 5A). As illustrated in Fig. 5B, a closely similar pattern of endotoxin sensitivity to AOAH comparing purified endotoxin aggregates (LPS agg ) Ϯ LBP and monomeric LPS-sCD14 or MD-2 was seen with a rough form of E. coli [ 14 C]LPS isolated from E. coli strain BW25113 aceE. Monomeric LPS-sCD14 complex and LPS agg -LBP at substoichiometric ratios of LBP-LPS, which most readily resulted in the formation of LPS-sCD14, were deacylated most effectively by AOAH in HBSS ϩ , 10 mM HEPES, 0.1% HSA, pH 7.4. As seen with LOS, LPS in the form of LPS agg and LPS-MD-2 was much more resistant to deacylation by AOAH (Fig. 5B).

DISCUSSION
LBP and CD14 have important and complex roles in host interactions with endotoxin. The most potent activation of mammalian cells by many endotoxins requires sequential protein-endotoxin and protein-protein interactions involving LBP, C14, MD-2, and TLR4 (3)(4)(5). The combined action of LBP and CD14 results in extraction of endotoxin monomers from purified endotoxin aggregates (30 -32) and the GNB outer membrane to yield a monomeric endotoxin-CD14 complex that is the preferred substrate of MD-2/TLR4 (4,25,33,34). Maximal generation of endotoxin-CD14 requires a very low molar ratio of LBP to endotoxin and CD14 (Fig.  1C) (30 -32). Higher molar ratios of LBP to endotoxin and CD14 reduce MD-2/TLR4-dependent cell activation by endotoxin (31,32,35,36) and instead promote cellular and extracellular (e.g. plasma lipoprotein-dependent) clearance of endotoxin (32,35,(37)(38)(39). The studies described here demonstrate a new role for LBP and CD14 in negative regulation of the pro-inflammatory action of endotoxin: they can participate as cofactors in the catalytic detoxification of endotoxin by AOAH. Our findings clearly show that both LBP alone and LBP and sCD14 in concert (i.e. by yielding endotoxin-sCD14) substantially increase the sensitivity of both meningococcal LOS and E. coli LPS to AOAH (Figs. 1 and 5). The stimulatory effects of LBP Ϯ sCD14 on AOAH activity require substoichiometric concentrations of LBP (Figs. 1B and 5B) that correspond closely to the concentrations required for LBP-dependent extraction and delivery of endotoxin monomers from endotoxin aggregates to sCD14 (Fig. 1C). The increased sensitivity to AOAH of endotoxin aggregates treated with very low concentrations of LBP suggests that, under these conditions, LBP induces a rearrangement of endotoxin within the aggregates that increases exposure of the lipid A region of endotoxin, facilitating AOAH action as well as extraction and transfer of endotoxin monomers to CD14. The inhibitory effect of higher concentrations of LBP on AOAH action and formation of endotoxin-sCD14 monomers (Fig. 1, B and C) may reflect increased competition by added LBP on the interaction of endotoxin with CD14 or AOAH. The inability of BPI to significantly increase endotoxin sensitivity to AOAH (Fig. 1B) mirrors its inability to promote extraction and transfer of endotoxin to CD14 (28,40). This may reflect the higher affinity of BPI for endotoxin and further illus- trates differences in the interactions of LBP and BPI with endotoxin-rich interfaces (28,41).
The sensitivity of monomeric endotoxin-sCD14 complex to AOAH is consistent with structural data derived from x-ray crystallography of mouse CD14 (42). Structural and functional data suggest that the binding site for endotoxin in CD14 includes a wide, flexible hydrophobic pocket near the amino terminus (42). The characteristics of this pocket predict that the fatty acyl chains of bound endotoxin will be partially exposed, explaining the need for albumin to maintain the stability of endotoxin-sCD14 in aqueous solution and rendering the acyloxyacyl linkages in lipid A accessible to AOAH. The relative resistance of LOS-MD-2 and LPS-MD-2 to AOAH (Fig.  5, A and B), by contrast, suggests a more complete sequestration of the fatty acyl chains within MD-2. This is consistent with models of MD-2 structure that suggest a deep hydrophobic cavity (43,44) and with the solubility and stability of endotoxin-MD-2 in aqueous solution even in the absence of albumin (13).
AOAH action on endotoxin-CD14 yields a partially deacylated endotoxin-CD14 complex (Fig. 3, B-D). AOAH-treated endotoxin can be readily transferred from CD14 to MD-2 (see Fig. 4) (13), producing a complex of underacylated endotoxin with MD-2 that can occupy TLR4 without efficiently inducing receptor activation (13,34). Thus, the underacylated endotoxin-MD-2 complex inhibits responsiveness of TLR4 by competing with the fully hexaacylated endotoxin-MD-2 complex (13). Soon after bacterial invasion, however, levels of extracellular AOAH are likely to be limiting. Moreover, comparison of the reactivity of LOS-sCD14 with AOAH (K m ϳ 5 nM; Fig. 2) versus MD-2 (apparent "K m " for transfer ϳ100 pM) (34) strongly suggests that at low (pM) concentrations of endotoxin-CD14, AOAH is unlikely to blunt MD-2/TLR4-dependent cell activation by endotoxin. This difference in reactivity between AOAH and MD-2 with endotoxin-CD14 may be necessary to permit induction of robust host innate and adaptive immune response to small numbers of invading GNB and pM endotoxin (15). However, under conditions when the levels of endotoxin-sCD14 produced substantially exceed those of MD-2/TLR4 and the concentrations of other potential host reactants with endotoxin-sCD14 (e.g. plasma lipoproteins) are limiting, AOAH activity against endotoxin-CD14 may help constrain potentially protracted endotoxin-driven inflammatory and immune responses.
It should be noted that the experimental conditions we have used roughly model extracellular AOAH action (pH 7.4) against extracellular endotoxin or, perhaps, cell surface-bound monomeric endotoxin-membrane CD14. However, within an inflammatory exudate, intracellular AOAH action within phagocytic cells (macrophages, dendritic cells, neutrophils) may predominate (2,15,45,46). In these cells, most endotoxin that is internalized is taken up in the form of large aggregates or membrane remnants, distinct from the endotoxin-sCD14 complex that is the preferred substrate for MD-2/TLR4 and extracellular AOAH. The identities of the intracellular factors and conditions that promote intracellular AOAH action remain to be determined, as does the biological significance of the partially deacylated endotoxin that is released from phagocytes into their extracellular environment.