Anionic Pulmonary Surfactant Phospholipids Inhibit Inflammatory Responses from Alveolar Macrophages and U937 Cells by Binding the Lipopolysaccharide-interacting Proteins CD14 and MD-2^*,♦

Cells and Reagents

LPS (0111:B4) purified from Escherichia coli was purchased from Sigma-Aldrich. The lipids PC, PG, sphingomyelin (SM), PE, phosphatidylserine (PS), and PI were purchased from Avanti Polar Lipids (Alabaster, AL). TNF-α was from Genzyme (Cambridge, MA). Rabbit polyclonal anti-p46 JNK, rabbit polyclonal anti-p38, mouse monoclonal phospho-specific p46-p54 JNK antibodies, and rabbit polyclonal anti-mitogen-activated protein kinase phosphatase (MKP)-1 antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit polyclonal phospho-specific p42 ERK, rabbit polyclonal anti-p42 ERK, phospho-specific p38 MAPK, rabbit polyclonal anti-IκBα, and phospho-specific IκBα antibodies were obtained from Cell Signaling Technology (Beverly, MA). [³H]Leucine was from PerkinElmer Life Sciences. Recombinant human CD14 and mouse anti-CD14 monoclonal antibodies, biG2 and biG14, were purchased from Cell Sciences Inc. (Canton, MA). Mouse anti-CD14 monoclonal antibody MEM-18 was purchased from Exbio (Vestec, Czech Republic). Mouse anti-His antibody and horseradish peroxidase-conjugated mouse anti-V5 antibody were obtained from Invitrogen. Mouse IgG₁ isotype control, mouse monoclonal anti-human CD14 antibody, and sheep anti-human CD14 polyclonal antibody were purchased from R&D systems (Minneapolis, MN). The macrophage-like cell line U937 (CRL-1593.2) was obtained from the American Type Culture Collection (Manassas, VA). The cells were maintained in endotoxin-free RPMI 1640 medium from Cambrex (East Rutherford, NJ) with 10% heat-inactivated bovine growth serum (Hyclone, Logan, UT).

Isolation of Rat and Human Alveolar Macrophages

Rat alveolar macrophages were isolated from bronchoalveolar lavage fluid (BALF) of Sprague-Dawley rats. The macrophages were plated at 5 × 10⁵ cells/well in 24-well plates (Falcon) in RPMI 1640 medium containing 10% bovine growth serum and allowed to adhere for 2 h prior to treatment with agonists and antagonists. Human alveolar macrophages were isolated from BALF obtained from healthy volunteers using protocols reviewed and approved by the National Jewish Medical Research Center Institutional Review Board (IRB) and the University of Colorado General Clinical Research Center. The macrophages were plated at 5 × 10⁴ cells/well in 96-well plates (Falcon) in RPMI 1640 medium containing 10% bovine growth serum and allowed to adhere for 48 h prior to the addition of agonists and antagonists.

Induction of TNF-α Secretion

U937 cells (1.3 × 10⁵/well in 96-well plates) were induced to differentiate by treatment with 10 nm phorbol myristate acetate for 48 h followed by recovery for 24 h in RPMI 1640 medium containing 10% bovine growth serum. Rat alveolar macrophages (5 × 10⁵/well in 24-well plates) were incubated for 2 h after isolation. Phospholipids were added to the cultures 30 min before adding LPS. After LPS addition, cultures were incubated for 6 h at 37 °C. At the end of the incubation period, the medium was collected and assayed for TNF-α concentrations using an ELISA kit. For all batches of LPS used in these studies, we verified that the induction of inflammatory mediator production was neutralized by treatment with 10 μg/ml polymyxin B. In additional studies, we also verified that the antagonistic actions of anionic lipids were equally effective against both LPS and synthetic KDO₂-lipid A.

Preparation of Surfactant Lipids

Surfactant lipids were isolated from the bronchoalveolar lavage of Sprague-Dawley rats, 28 days after intratracheal instillation of 25 mg of silica (∼125 mg/kg). Initially, the surfactant was purified by the method of Hawgood et al. (30) using NaBr density gradient centrifugation. The purified surfactant was extracted with butanol (31) and segregated into butanol-soluble and -insoluble material. The butanol-soluble surfactant lipids were recovered by drying under vacuum and resuspending in chloroform. The phospholipid content was determined by the method of Rouser et al. (32), and the mixture was stored at −20 °C. Prior to use, an aliquot of surfactant lipids was initially dried under nitrogen and subsequently hydrated in 20 mm Tris (pH 7.4), 150 mm NaCl buffer at 37 °C for 1 h. Finally, the surfactant lipids were probe-sonicated in 5–30-s bursts with 1 min of cooling between bursts to make a vesicle preparation for use in experiments.

Analysis of Nitric Oxide Accumulation

Nitric oxide (NO) accumulation in cell culture supernatants was determined as reported previously (33). Briefly, rat alveolar macrophages were stimulated with LPS (10 ng/ml) or LPS plus phospholipids for 24 h. Culture supernatants (usually 100 μl) were combined with an equal volume of Griess reagent (33), and the samples were incubated at room temperature for 10 min before the absorbance was quantified at 550 nm. With the use of a standard curve, the nmol of NO produced were determined and normalized to the total cell number in each sample.

Analysis of Cytokine Production

Human and mouse TNF-α ELISA kits were purchased from Invitrogen. Mouse KC and MIP-2 Quantikine kits were purchased from R&D Systems. Measurements of these cytokines were according to the manufacturers' protocols.

Measurement of MAPK, IκBα, and MKP-1

Monolayers of unstimulated or stimulated macrophages were lysed on ice with 250 μl of ice-cold lysis buffer (50 mm Tris·HCl buffer, pH 8.0, containing 137 mm NaCl, 10% (v/v) glycerol, 1% (v/v) Nonidet P-40, 1 mm NaF, 10 μg/ml leupeptin, 10 μg/ml aprotinin, 2 mm Na₃VO₄, and 1 mm phenylmethylsulfonyl fluoride) (34). Insoluble nuclear material was removed by centrifugation at 14,000 × g for 10 min at 4 °C, and the resultant supernatants were collected. Aliquots of 15 μg of lysate protein were subjected to SDS-PAGE and transferred onto nitrocellulose membranes (35). The blots were then washed in Tris-Tween-buffered saline (20 mm Tris-HCl buffer, pH 7.6, containing 137 mm NaCl and 0.05% (v/v) Tween 20), blocked with 5% (w/v) nonfat dry milk for 1 h, and probed according to the method described by Towbin et al. (35) with phospho-specific antibodies to p46-p54 JNK, p42/p44 ERK, and p38 MAPK, or IκBα or with a polyclonal MKP-1 or IκBα antibodies in 5% (w/v) bovine serum albumin dissolved in Tris-Tween-buffered saline. Bound primary antibodies were detected by enhanced chemiluminescence (ECL Plus, Amersham Biosciences) with the use of horseradish peroxidase-conjugated secondary anti-rabbit or anti-mouse antibody. To determine loading of proteins between samples, the membranes were probed with rabbit polyclonal p46 JNK, p42/p44 ERK, and p38 MAPK antibodies.

Administration of LPS and Phospholipids in Vivo

Female BALB/c mice from 6 to 8 weeks of age were obtained from Jackson Laboratories (Bar Harbor, ME). Experiments were conducted under a protocol approved by the Institutional Animal Care and Use Committee of the National Jewish Medical and Research Center. Liposomes were formed using a Liposofast^TM (Avestin; Ottawa, Canada), which produces unilamellar liposomes of 100 nm of diameter, and then mixed with an aqueous solution containing LPS. The mixture of LPS and phospholipids was sprayed into murine trachea using a MicroSprayer^TM aerosolizer (PennCentury, Philadelphia, PA) under isoflurane anesthesia. Delivery by MicroSprayer^TM has been shown to result in lung deposition fractions of more than 93% in primates (36). For collection of BALF, lethally anesthetized mice were lavaged via the trachea with 1 ml of Hanks' balanced salt solution (Invitrogen). The volume of collected BALF was measured in each sample, and the number of leukocytes was counted (Coulter Counter; Beckman Coulter). Differential cell counts were determined from at least 300 cells on cytocentrifuged preparations (Cytospin; Shandon Ltd., Runcorn, Cheshire, UK). Slides were stained with modified Wright-Giemsa (Hema; Protocol, Swedesboro, NJ), and the cell populations were differentiated by standard hematologic procedures. Cytokine levels in the BALF or in the supernatants of cultured airway macrophages were measured using ELISA kits.

Soluble Extracellular Domain of TLR4, CD14, and MD-2

A soluble form of the extracellular domain of TLR4 (sTLR4) consists of the putative extracellular sequence (Met¹–Lys⁶³¹) and a His₆ epitope tag at its C-terminal end. sTLR4 and MD-2 cDNAs were described previously (37). sTLR4-His₆ was subcloned into pcDNA3.1(+) (Invitrogen). MD-2-V5-His₆ that contains the C-terminal fusion V5 epitope tag and His₆ epitope tag was generated by using PCR and subcloned into pcDNA3.1D/V5-His-TOPO (Invitrogen). A control protein, yeast PstB2-V5-His₆ that contains the C-terminal fusion V5 tag epitope and His₆ tag epitope, was generated by using PCR and subcloned into the baculovirus vector pVL1392 (38). The epitope-tagged cDNA constructs for sTLR4 and MD-2 were subcloned into PVL1392, and in addition to PstB2, they were independently expressed using a baculovirus-insect cell expression system according to the methods described by O'Reilly et al. (39). The sTLR4 protein, MD-2 protein, sCD14 protein, and PstB2 protein were purified using an epitope affinity column of nickel-nitrilotriacetic acid beads (Qiagen, Valencia, CA) and elution with 50–200 mm imidazole, by the method described previously (40).

Binding of CD14 and MD-2 to Phospholipids

Phospholipids (1.25 nmol) in 20-μl aliquots of ethanol were pipetted onto 96-well half-area plates (Corning Inc., Corning, NY), and the solvent was evaporated using a warm air blower. After nonspecific binding was blocked with 20 mm Tris buffer (pH 7.4) containing 0.15 m NaCl, 5 mm CaCl₂ or 2 mm EGTA, and 5% (w/v) bovine serum albumin (buffer A), various concentrations of human CD14 or MD-2 in 25 μl of buffer A were added and incubated at 37 °C for 1 h. The wells were then washed with 20 mm Tris buffer (pH 7.4) containing 0.15 m NaCl and 5 mm CaCl₂ or 2 mm EGTA (buffer B), and 1 μg/ml anti-human CD14 IgG or anti-His antibody (50 μl/well) in buffer A was added and incubated overnight at 4 °C followed by incubation with horseradish peroxidase-labeled anti-mouse IgG (1:5000) for 1 h. After washing the wells with buffer B, the peroxidase reaction was finally performed using o-phenylenediamine as a substrate. The binding of CD14 or MD-2 to phospholipids was detected by measuring absorbance at 490 nm.

Binding of LPS to CD14

LPS (2 μg) in 20-μl aliquots of ethanol was pipetted onto a 96-well plate, and the solvent was evaporated using a warm air blower. After the nonspecific binding was blocked with buffer A, mixtures of CD14 (1 μg/ml) and phospholipid liposomes (20 μg/ml) in buffer A, which were preincubated at 37 °C for 1 h, were added and further incubated at 37 °C for 1 h. The amount of bound CD14 was detected using the method described above.

Phospholipid Inhibition of MD-2 and sTLR4 Interaction

Preparations of sTLR4 (100 ng) in aliquots of 20 μl of buffer B were pipetted onto a 96-well plate, and the solvent was evaporated using a warm air dryer. After the nonspecific binding was blocked with buffer A, mixtures of MD-2 (1 μg/ml) and varying concentrations of phospholipid liposomes in buffer A were added and incubated at 37 °C for 1 h. The amount of bound MD-2 was detected using specific antibodies directed against the V5-epitope of the recombinant protein.

Statistical Analysis

All results were expressed as mean ± S.E. Analysis of variance was used to determine the levels of significant difference between all groups. Groups were compared by unpaired two-tailed t test. The p value for significance was set at 0.05.

POPG and PI Inhibit LPS-induced Production of Proinflammatory Cytokines

In our initial studies, we investigated the ability of purified lipids normally present as minor components of pulmonary surfactant to modulate LPS-induced cytokine secretion. Macrophages were stimulated with LPS in the presence or absence of purified phospholipids (Fig. 1). Culture supernatants were collected, and TNF-α production by U937 cells and NO production by rat alveolar macrophages were determined. Palmitoyl-oleoyl-phosphatidylglycerol (POPG) and PI significantly attenuated TNF-α and NO production in a concentration-dependent manner with the maximal inhibitory effect ≤ 2.5 μg of phospholipids/ml. Another anionic phospholipid, PS, was less effective than PI and POPG. In contrast, the aminophospholipids and sphingolipids DPPC, PE, and SM had no significant effect on TNF-α or NO production. The major molecular species of PG in humans is POPG, whereas rodent surfactant contains a mixture of disaturated and unsaturated PG. We next examined the effect of saturation and acyl chain length of PGs on the inhibition of LPS-induced inflammation. As shown in Fig. 2, disaturated PGs containing two palmitic (16:0), stearic (18:0), or octanoic (8:0) fatty acids failed to antagonize LPS-induced TNF-α or NO production. However, PGs with two myristic (14:0) fatty acids were as potent as POPG as antagonists of LPS. PGs with two lauric (12:0) fatty acids were also modest antagonists of LPS-induced cytokine production. Although we did not have the reagents to compare different molecular species of PI for LPS antagonism, the PIs that we used were unsaturated with the major form containing 16:0 and 18:2 fatty acids. From the above findings, we conclude that unsaturated PGs and PIs and selected saturated PGs act as potent antagonists of LPS action upon macrophages.

POPG Antagonism of LPS-induced Inflammation Occurs in the Context of Surfactant Phospholipids

The POPG present in the alveolar compartment is in a lipid-rich environment with concentrations of total phospholipids of 15–30 mg/ml (21), and we examined whether these other lipids can interfere with LPS antagonism. In one set of experiments, POPG was added to organic solvent extracts of surfactant using a method that ensured ideal mixing of all components, and liposomes were prepared by sonication. In a second set of experiments, vesicles composed of surfactant lipids and vesicles composed of POPG were prepared separately and then combined. In this latter situation, there will be two populations of vesicles, one containing randomly mixed surfactant lipids and a second containing pure POPG. The results in Fig. 3A reveal that surface dilution and randomization of POPG within a single vesicle significantly diminishes the potency of the lipid as an antagonist of LPS action. To approximate the activity of POPG alone, the randomized POPG must now constitute nearly 50% of the total lipid present in a surfactant lipid-containing vesicle. In contrast to the results in Fig. 3A, the data presented in Fig. 3B demonstrate that admixture of pure POPG vesicles and randomized surfactant lipid vesicles has essentially no effect upon the activity of POPG as an LPS antagonist, measured by TNF-α production. This result also indicates that the combination of pure POPG vesicles with surfactant lipid vesicles does not result in significant fusion and intermixing of lipids between vesicle bilayers and suggests that introduction of POPG into the lung may potently antagonize LPS action.

POPG Inhibits the LPS-induced Phosphorylation of MAPK and IκBα and Expression of MKP-1

We next investigated the influence of phospholipids upon the intracellular signaling pathways of LPS-induced TNF-α secretion. In these studies, we used concentrations of antagonistic phospholipids of 200 μg/ml, which is ∼10% of the levels present in pulmonary surfactant. Host cells recognize many specific microbial components through Toll-like receptors that mediate immune responses. On alveolar macrophages, LPS binds to membrane CD14 and a TLR4-MD2 complex. The signals from TLR4 are transmitted through MyD88 and TRAF6 (41) to IκBα or MAPKs such as ERK, JNK, and p38. These signals regulate transcription factors and induce proinflammatory cytokine production. In experiments summarized in Fig. 4, differentiated U937 cells were stimulated with LPS in the absence or presence of POPG, and cell lysates were electrophoresed and immunoblotted (Fig. 4A). Significant increases in phosphorylated forms of p38, p42 ERK, and JNK, as well as IκBα, were detected between 15 and 60 min after LPS treatment. LPS treatment also reduced the steady state levels of IκBα due to protein degradation. Treatment of cells with POPG in addition to LPS eliminated the phosphorylation of p38, p42 ERK, JNK, and IκBα and also abrogated the reduction in the steady state levels of IκBα. In addition to inducing phosphorylation of MAPKs and IκBα, LPS induced synthesis of MKP-1 that functions to turn off MAPKs signaling (42, 43). The POPG treatment blocked the synthesis of new MKP-1, indicating that the lipid is likely to act upstream of MAPK activation rather than downstream of the process by induction of MKP-1.

The quantification of Western blotting results from multiple experiments performed as shown in Fig. 4A is presented in Fig. 5. The POPG treatment significantly reduces p38, ERK, JNK, and IκBα phosphorylation in LPS-stimulated cells to values nearly equivalent to untreated cells. The expression of MKP1 was also reduced to control levels by POPG treatment. In addition, the total amount of IκBα present in the cells remained constant when LPS-treated cells were also given POPG. This latter result demonstrates that POPG prevents degradation of IκBα that occurs subsequent to LPS treatment alone. Experiments identical to those described in the legends for were performed to directly compare the antagonism of LPS activation of U937 cells by POPG, POPC, and dipalmitoylphosphatidylglycerol (DPPG) (Fig. 4B). The cells treated with POPG and LPS reproduced the results shown in Figs. 4A and 5. The cells treated with LPS and POPC or DPPG failed to show any significant antagonism of LPS activation of p38, p42/44, JNK, IκBα, or MKP-1 expression. Thus, the inhibition of LPS-induced signaling pathways is dependent upon both the specific acyl chains present in the PG molecule and the specific polar substituent, which defines the class of phospholipid.

We also conducted control experiments to test whether POPG treatment had a general toxic effect upon U937 cells. Protein synthesis was measured by determining [³H]leucine incorporation into trichloroacetic acid precipitable material in the presence and absence of 200 μg/ml POPG. No changes in protein synthesis occurred over a 6-h period in POPG-treated U937 cells when compared with untreated cells in three independent experiments. Additional studies were performed to examine whether POPG pleiotropically inhibited signaling by macrophages. In these studies, macrophages were treated with TNF-α (10 ng/ml), and the degradation of IκBα was measured. POPG treatment of TNF-α-stimulated macrophages failed to alter IκBα degradation when compared with stimulated cells without POPG treatment. Collectively, these studies indicate that the actions of the anionic lipids upon LPS signaling are specific and not attributable to generalized suppression of cell growth or signaling processes. Additional experiments were performed examining the activity of POPG as an antagonist of flagellin-mediated TLR5 activation on BEAS2B cells and CG-rich oligonucleotide (CpG)-mediated TLR9 activation of RAW 264.7 cells. Results from three independent experiments revealed that POPG had no significant effect upon interleukin 8 production from BEAS2B cells (flagellin treatment = 5.5 ± 0.9 ng/ml; flagellin plus POPG = 6.3 ± 0.2 ng/ml) or TNF-α production from RAW 264.7 cells (CpG treatment = 9.2 ± 1.7 ng/ml; CpG plus POPG = 7.4 ± 1.9 ng/ml). These latter findings clearly demonstrate that the POPG effects are relatively specific for CD14·TLR4·MD-2 related activation of responsive cells.

Anionic Phospholipids Antagonize LPS Activation of Human Alveolar Macrophages

We next examined the lipid antagonism of primary human alveolar macrophages, isolated by bronchoalveolar lavage, and challenged with 10 ng/ml LPS for 6 h. The results presented in Fig. 6 demonstrate that POPG, dimyristoylphosphatidylglycerol (DMPG), and PI markedly attenuate the inflammatory response to LPS of freshly isolated human alveolar macrophages. In contrast, DPPG and DPPC had no significant effect upon the human alveolar macrophage response to LPS. These results demonstrate that human alveolar macrophages are susceptible to regulation of their inflammatory response by resident anionic surfactant phospholipids and the synthetic lipid DMPG.

POPG Inhibits LPS-induced Proinflammatory Cytokine Production in Vivo

We next examined whether POPG could act to attenuate LPS effects in mice. Phospholipid liposomes were formed using a Liposofast^TM pressure filtration apparatus. Mixtures of LPS and phospholipids were sprayed into the trachea of mice using a MicroSprayer^TM positioned at the vocal cords. At 18 h after stimulation, the lungs of mice were lavaged (Fig. 7), and TNF-α, neutrophil infiltration, the mouse interleukin-8 equivalent, KC, and MIP-2 were measured in the recovered lavage fluid (44). These inflammatory mediators are important prognostic determinants of ALI/ARDS in humans (45). LPS-induced TNF-α was ∼300 pg/ml and was unaffected by DPPC instillation. In contrast, POPG, DMPG, and PI significantly attenuated the TNF-α secretion in the lung. These results clearly indicate that the intratracheally administered POPG, DMPG, and PI can reduce the inflammation in the lung in vivo. These results correlate well with in vitro results. LPS stimulation also induced the infiltration of neutrophils. POPG, DMPG, and PI, but not DPPC, modestly attenuated the LPS-induced neutrophil infiltration. DMPG, PI, and especially POPG attenuated the KC and MIP-2 secretion in BALF. These findings predict that the antagonistic phospholipids will inhibit the secretion of interleukin-8 from human alveolar macrophages in vivo.

Next, we examined the effect of phospholipids in a non-pulmonary sepsis model in mice. LPS (50 μg/200 μl) was injected into mice via the tail vein, and phospholipids were administered intratracheally at the same time. 3 h after stimulation, BALF was collected (Fig. 8). Intratracheally administered POPG, but not DPPC, significantly inhibited the TNF-α secretion in BALF, indicating that this anionic lipid has an anti-inflammatory effect for sepsis originating outside the lung. POPG administered via the trachea effectively inhibited the infiltration of neutrophils in BALF when compared with DPPC. POPG also significantly attenuated the KC and MIP-2 levels in BALF, although the magnitude of this effect was not very large. These results suggest that POPG may be useful for attenuating lung inflammation caused by sepsis, originating at a non-pulmonary site.

CD14 Binds POPG and PI in a Concentration-dependent Manner

LPS mainly binds to CD14 on cell surfaces and is subsequently transferred to an MD-2·TLR4 complex on the same membrane to initiate signaling. We prepared recombinant forms of the extracellular domains of CD14 and TLR4 and soluble MD-2 as shown in Fig. 9A. We first investigated whether CD14 is a target for phospholipid interaction that antagonizes LPS action. The anionic surfactant lipids POPG and PI, when adsorbed as a solid phase to microtiter wells, strongly bound to CD14 in a concentration-dependent manner (Fig. 9B). The binding of CD14 to the zwitterionic surfactant lipid, DPPC, was ∼30% of the levels attained for POPG binding. The CD14-lipid interactions were not attenuated by EGTA, and CaCl₂ was not required for the binding. These results demonstrate that POPG and PI can directly bind CD14, and these interactions are of higher affinity than those with DPPC. These binding interactions for the anionic phospholipids are consistent with the effect of these same lipids upon inflammatory mediator production. The molecular species of PG were also evaluated for their direct binding interactions with CD14. POPG and DMPG exhibited the strongest direct binding interactions with CD14. However, CD14 also bound to DPPG to nearly the same extent as POPG and DMPG (Fig. 10A). These results clearly indicate that anti-inflammatory anionic phospholipids can bind strongly to CD14. However, some lipids without demonstrable anti-inflammatory effect will also directly bind CD14.

POPG Blocks the Interaction of CD14 with LPS

Another method to evaluate CD14-lipid interaction is to perform competition experiments in which CD14 binding to solid phase LPS is subjected to competition using fluid phase phospholipid liposomes. This series of experiments is described in Fig. 10B. Two lipids that function as potent LPS antagonists in vitro and vivo, DMPG and POPG, are the most effective inhibitors of CD14 binding to solid phase LPS. DPPG, which is inactive as an LPS antagonist, weakly competes for CD14 binding to LPS. These latter findings are consistent with earlier findings about PG antagonism of LPS activation of macrophages. Paradoxically, PI, which is a potent LPS antagonist, fails to compete for CD14 binding to solid phase LPS. These latter results strongly suggest that PI and POPG do not act by identical mechanisms in producing LPS antagonism.

POPG and PI Bind to CD14 at the LPS Binding Site

CD14 has four LPS binding sites located at the N terminus of the protein (46). Monoclonal antibodies biG14 and MEM-18 recognize amino acids 39–44 and amino acids 57–64, respectively, that constitute part of the LPS binding site. The biG14 and MEM-18 antibodies are also proven inhibitors of LPS binding to CD14. Another antibody, biG2, recognizes amino acids 147–152, which are not part of the LPS binding site, and biG2 ligation does not inhibit LPS binding to CD14. The epitope for another antibody, RDIg, has not been determined but appears to recognize a site distinct from that used for LPS binding. Solution of the crystal structure of mouse CD14 at a resolution of 2.5 Å provides evidence that LPS binds a defined pocket in the protein (47). The biG14 and MEM-18 binding sites are close to the pocket and predicted to sterically occlude LPS binding. In contrast, the biG2 site when ligated by antibody should not interfere with LPS binding. We examined the action of the above described monoclonal antibodies to determine the relationship between the LPS binding site and the anionic phospholipid binding site on CD14.

In experiments, shown in Fig. 11, the CD14 was preincubated with specific monoclonal antibodies, and the effect of this interaction upon the recognition of solid phase phospholipid by CD14 was measured. The CD14 bound to the solid phase was detected using anti-CD14 polyclonal antibody. As shown in Fig. 11A, the monoclonal antibodies biG14 and MEM-18 significantly reduced the CD14 binding to solid phase POPG by 40–60%, whereas other antibodies that did not recognize the LPS binding site (mouse IG, RDIg, and biG2) failed to significantly alter CD14 recognition of POPG. The addition of biG14 and MEM-18 antibodies together gave slightly higher inhibition of CD14 binding to lipid than either antibody alone. Nearly identical results were obtained when PI was used as the solid phase ligand as shown in Fig. 11B, with the monoclonal antibodies inhibiting the binding reaction by 40–70%. Interestingly, none of the antibodies tested inhibited the binding of CD14 to DPPG (data not shown). Thus, the site of interaction between CD14 and POPG and PI is different from the site of interaction with DPPG. In Fig. 11C, we conducted control experiments with solid phase CD14 to show that ligation of the protein by RDIg, biG14, MEM-18, and biG2 does not attenuate the binding of anti-CD14 polyclonal antibody. Thus, the loss of polyclonal antibody detection of CD14 reflects a reduction in interaction of the protein with phospholipids and is not due to monoclonal antibody interference with polyclonal antibody recognition.

POPG Binds to MD-2 and Blocks the Interaction of MD-2 with sTLR4

TLR4 requires MD-2 for CD14-dependent cellular response to LPS. It is known that LPS binds to CD14 and MD-2 but not TLR4. Next, we examined whether phospholipids directly interact with MD-2 or TLR4. Recombinant MD-2, sTLR4, and the yeast protein PstB2, all with a His₆ epitope tag, were expressed using the baculovirus-insect cell expression system (Fig. 9A). Solid phase POPG strongly bound to MD-2 but not to sTLR4 or the control epitope-tagged protein PstB2 (Fig. 12A). The lipid recognition specificity of MD-2 was evaluated using PI, two molecular species of PC, and three molecular species of PG (Fig. 12B). Relative to POPG, only DMPG showed significant binding (∼50% of the POPG value) to MD-2. Neither saturated nor unsaturated PC, unsaturated PI, nor DPPG showed any significant binding to MD-2.

We next probed the influence of lipids upon the interactions between MD-2 and TLR4. The extracellular domain of TLR4 was adsorbed onto microtiter wells, and the direct binding of MD-2 was measured by ELISA using a monoclonal antibody directed against a V5 epitope on the protein. At low levels of lipid competitor added as liposomes, only POPG interfered with the MD-2·TLR4 interaction (Fig. 13A), producing 40% inhibition. In Fig. 13B, the concentration of lipid competitors was varied up to 200 μg/ml, and only POPG showed any significant inhibition (∼75%) of the MD-2-TLR4 interaction. The action of POPG as an inhibitor increased with increasing concentration of the lipid between 20 and 200 μg/ml. These results clearly demonstrate that another site of action of POPG occurs between MD-2 and TLR4. These results further indicate that PI and POPG have non-identical mechanisms of interaction with the innate immune system that result in suppression of inflammation.