Lipopolysaccharide (LPS)-binding Protein Inhibits Responses to Cell-bound LPS*

Lipopolysaccharide (LPS)-binding protein (LBP) is an acute phase reactant that may play a dual role in vivo , both potentiating and decreasing cell responses to bacterial LPS. Whereas low concentrations of LBP potentiate cell stimulation by transferring LPS to CD14, high LBP concentrations inhibit cell responses to LPS. One inhibitory mechanism involves the ability of LBP to neutralize LPS by transferring it to plasma lipoproteins, whereas other inhibitory mechanisms, such as the one described here, do not require exogenous lipoproteins. Here we show that LBP can inhibit monocyte responses to LPS that has already bound to membrane-bound CD14 (mCD14) on the cell surface. LBP caused rapid dissociation of LPS from mCD14 as measured by the ability of LBP to inhibit cross-linking of a radioiodi-nated, photoactivatable LPS derivative to mCD14. Whereas LBP removed up to 75% of the mCD14-bound LPS in 10 min, this was not accompanied by extensive release of the LPS from the cells. The cross-linking data suggest that much of the LPS that remained bound to the cells was associated with LBP. The ability of LBP to inhibit cell responses could not

of the most potent of the known bacterial agonists in animal host cells. Whereas LPS recognition benefits the host by sensing the presence of bacteria and mobilizing defense mechanisms, an exaggerated response to LPS may contribute to the harmful sequelae of severe sepsis, which include coagulation disorders, organ failure, shock, and death. The host, therefore, has numerous mechanisms that down-regulate responses to LPS and remove it from the circulation and tissues (1)(2)(3)(4)(5)(6)(7).
The potency of LPS is due to the sensitivity of the host recognition system, which requires the Toll-like receptor 4 (Tlr4) (8) signaling receptor. At least three LPS-binding proteins (LBP (9), CD14 (10), and MD-2 (11)) are required for sensitive Tlr4-mediated LPS recognition. Plasma LBP potentiates LPS recognition by transferring it from bacterial membranes (6) to CD14 (12,13), which is expressed as a glycosylphosphatidylinositol-anchored protein on the surfaces of myeloid lineage cells (mCD14) and as an anchorless plasma protein, soluble CD14 (sCD14) (14,15). CD14 presents LPS to Tlr4, presumably by transferring the LPS to MD-2, an accessory protein that is constitutively associated with the extracellular domain of Tlr4 (16). MD-2 binds LPS with high affinity (17), and it is required for Tlr4 signaling (11). Although normal expression of the Tlr4 and MD-2 are too low for direct measurements of LPS binding to the receptor complex, indirect evidence of ligandreceptor binding has been provided by the ability of structurally different forms of Tlr4 (18,19) or MD-2 (20,21) to discriminate between differences in lipid A structure. Direct evidence of LPS binding to Tlr4/MD-2 was provided by LPS cross-linking experiments in cells that overexpressed these proteins (22).
LBP and sCD14 can both potentiate and down-regulate responses to LPS. sCD14 was originally described as an LPS inhibitor (23), and other studies have described inhibitory effects of high concentrations of sCD14 under various conditions (24,25). Recent work from our laboratory has shown that in plasma, sCD14 decreases monocyte responses to LPS by transferring cell-bound LPS to plasma lipoproteins (26).
LBP can prevent lethal shock caused by LPS or Gram-negative bacteria when administered intraperitoneally in mice (27). High acute phase concentrations of LBP in the plasma of septic humans decrease monocyte responses to LPS (28). The inhibitory mechanism of LBP can be explained in part by its ability to neutralize the bioactivity of LPS by transferring it to plasma lipoproteins (6,29). Other inhibitory mechanisms are suggested by the ability of LBP to inhibit cell responses to LPS in serum-free medium (27,28). Possible mechanisms might include the formation of large LPS⅐LBP complexes that are internalized by mCD14 but that contribute little to LPS signaling (30). Here we report that LBP can inhibit LPS signaling even after LPS monomers have bound to mCD14.

EXPERIMENTAL PROCEDURES
Reagents and Cells-Recombinant human LBP (C-terminal His tag) was expressed in a baculovirus system in SF-9 cells in SF-900 II SFM (Invitrogen). The protein was purified to homogeneity on a His-Trap™ nickel column (Amersham Biosciences). Recombinant human sCD14 and soluble MD-2 (sMD-2) were produced by baculovirus expression as previously described (17,26). The recombinant proteins did not stimulate cells when added alone at the indicated concentrations. THP-1 cells (provided by LiWu Li, Wake Forest University, Winston-Salem, NC) were cultured in 0.05 M 1,25-dihydroxyvitamin D 3 (VD 3 ) for 3-4 days to induce mCD14 expression (THP-1 (VD 3 )). In experiments in which IL-1␤ was used as an agonist, the cells were cultured in 10 Ϫ6 M prostaglandin E 2 (PGE 2 ) during the last 24 h of culture in VD 3 (THP-1 (VD 3 -PGE 2 )). PGE 2 and VD 3 induced responsiveness of the cells to IL-1␤ without inhibiting responsiveness to LPS. In some experiments (Figs. 3 and 4B), mCD14 expression was enhanced by stable transfection of THP-1 cells with human CD14 cDNA as previously described (31), and the cells were also differentiated by culturing them in VD 3 as described above. Peripheral blood mononuclear cells were isolated from normal human blood as previously described (26) with the approval of the Institutional Review Board of UT Southwestern Medical Center. HEK 293 cells (ATCC) were transiently transfected with human CD14 cDNA (pRcRSV-CD14) (31) using Fugene 6 (Roche Applied Science). Anti-human CD14 monoclonal antibodies, 60bca and 63D3, were purified from the culture supernatants of hybridomas obtained from the ATCC. 60bca was covalently bound to agarose beads using Carbolink™ from Pierce. Affinity-purified rabbit polyclonal anti-human LBP antibody, K1970602, was a gift of XOMA Corp. (Berkeley, CA). IgG fractions of goat polyclonal antibodies to recombinant human soluble CD14 (number 4-2051-5) and recombinant human LBP (number 4-1855-9) were produced in the Tobias laboratory. Peroxidase-conjugated donkey anti-goat IgG was obtained from Jackson Immunoresearch (West Grove, PA). All other reagents were from Sigma unless otherwise stated. LPS-[ 3 H]LPS was biosynthetically labeled in the fatty acyl chains in Escherichia coli LCD25 (Ra chemotype) (32). 125 I-ASD-LPS was made by derivatization of RcLPS (Salmonella minnesota R5) with sulfosuccinimidyl-2-[p-azidosalicylamido]ethyl-1-3Ј-dithiopropionate (SASD) (Pierce), a cleavable, photoactivatable cross-linker, as previously described (33). The stimulatory activity of the 125 I-ASD-LPS was tested in THP-1 (VD 3 ) cells and found to be similar to that of the underivatized RcLPS. Each LPS preparation stimulated cytokine production at threshold concentrations of 0.03-0.1 ng of LPS/ml in the presence of LBP (0.1 g/ml) or as preformed LPS-sCD14 complexes (data not shown). Fluorescent LPS (Alexa Fluor488™-LPS derived from E. coli 055:B5 LPS) was obtained from Molecular Probes, Inc. (Eugene, OR) and is referred to here as Alexa-LPS. Protein contamination was not found in any of the LPS preparations on silver-stained SDS-PAGE gels.
LPS-Cell Binding and Stimulation-LPS-cell binding was performed in the absence of LBP using preformed LPS⅐sCD14 complexes (34). For cell stimulation assays, the cells (7 ϫ 10 5 cells/0.1 ml) were incubated with LPS-sCD14 at 1 ng of LPS/ml for 2 min at 37°C. The cells were then washed in cold RPMI 1640 medium to remove unbound LPS and were incubated in serum-free medium (SFM) (RPMI 1640, 20 mM HEPES, pH 7.4, 0.1 mg/ml bovine serum albumin) containing the indicated proteins for 2 h at 37°C with brief mixing at 2-min intervals. Cytokines were measured in culture supernatants by enzyme-linked immunosorbent assay (BD Pharmingen, San Diego, CA). The cell-bound LPS stimulated the production of ϳ44 ng of IL-8/ml, 5.5 ng of tumor necrosis factor-␣/ml, and 1.7 ng of IL-1␤/ml during 2-h incubations, whereas unstimulated cells produced less than 2% of those cytokine levels.
The release of cell-bound [ 3 H]LPS was measured by incubating the cells with 100 ng/ml [ 3 H]LPS in LPS⅐sCD14 complexes for 2-5 min at 37°C, washing with cold SFM to remove unbound LPS, and incubating the cells in SFM at 37°C for the indicated times. Released and cellassociated [ 3 H]LPS were measured as previously described (35). LPS internalization was measured by protease protection of [ 3 H]LPS and by surface quenching of fluorescent LPS with trypan blue as previously described (31,34). Briefly, cell surface-bound [ 3 H]LPS was removed by incubating the cells in 0.02% proteinase K for 30 min at 4°C, or the fluorescence of cell surface-bound Alexa-LPS was quenched by suspending the cells in trypan blue and measuring the mean fluorescence intensity by flow cytometry using a FACScan™ (BD Biosciences).
In LPS cross-linking experiments, 125 I-ASD-LPS was prebound to rsCD14, the mixture was then incubated in SFM with the indicated soluble proteins or with cells at 37°C, and the mixtures were then exposed to UV light to covalently cross-link the 125 I moiety to the associated proteins. After cross-linking the LPS to cellular proteins, the cells (ϳ4 ϫ 10 6 cells/0.5 ml) were washed and lysed in 50 l of lysis buffer (50 mM HEPES, pH 7.4, 100 mM NaCl, 2 mM EDTA, 1% Triton X-100, 20 mM octyl glucoside, 5 mM iodoacetamide, 1 mM benzamidine, 0.25 mM phenylmethylsulfonyl fluoride, 100 mM sodium fluoride, and 1 g/ml each of pepstatin A, leupeptin, and aprotinin) for 30 min at 4°C and centrifuged to remove insoluble material. CD14 and LBP were immunoprecipitated sequentially by incubating the lysate supernatants with 12 l of 60bca-agarose beads for 2 h at 4°C with brief agitation (3 s) at 2-min intervals in an Eppendorf Thermomixer™ (Brinkmann Instruments). After removing the beads, LBP was immunoprecipitated from the lysates by incubating them with 2 g of rabbit anti-human LBP for 1 h followed by 12 l of protein A-Sepharose for 1 h as described above. The beads were then washed and eluted in SDS sample buffer as previously described (33). After reductive cleavage of LPS from the ASD group with dithiothreitol, the 125 I-proteins were separated on 11% SDS-PAGE gels. 125 I radioactivity was quantitated by exposing the dried gel to a phosphor screen and measuring the radioactivity in a Cyclone PhosphorImager System (Packard Instrument Co.). Background activity, defined by equally sized control areas of the gel, was subtracted using the software provided by the manufacturer. In experiments in which LPS was cross-linked to soluble proteins, autoradiography was performed by exposing the gels to x-ray film.
To test the specificity of the immunoprecipitations, CD14 and LBP were immunoprecipitated as described above using specific antibodies or control antibodies (an irrelevant isotype-matched (IgG1) mAb bound to beads or nonimmune rabbit IgG and protein A-Sepharose, respectively). After separating the proteins by SDS-PAGE, they were transferred to Immobilon-P membranes (Millipore Corp., Bedford, MA) according to the manufacturer's instructions. The membranes were blocked with 1% fish gelatin in PBS plus 0.05% Tween 20, and the proteins were detected with goat anti-human CD14 (7 g/ml) or goat anti-human LBP (3 g/ml) followed by peroxidase-conjugated donkey anti-goat IgG. The blots were developed with ECL reagents (Amersham Biosciences) and exposed to x-ray film. The CD14 and LBP bands were found at their expected molecular sizes (ϳ55 and 60 kDa, respectively), whereas no bands were detected in control antibody immunoprecipitates (data not shown).

RESULTS
To determine whether LBP can inhibit cell responses after LPS-mCD14 binding has occurred, we incubated LPS-sCD14 complexes with mCD14-expressing THP-1 cells for 2 min at 37°C. We previously found that nearly maximal binding of LPS monomers to mCD14 occurs under these incubation conditions (34). After washing to remove unbound LPS, we found that incubating the cells with LBP inhibited cytokine production. As shown in Fig. 1, 50% inhibition of cytokine production occurred at LBP concentrations of 1-2 g/ml. In experiments not shown, LBP also inhibited responses to cell-bound LPS in normal human monocytes in peripheral blood mononuclear cell mixtures with similar potency. Because the LPS receptor (Tlr4) and the IL-1 receptor use many of the same intracellular signaling proteins to induce the expression of inflammatory cytokine genes, we stimulated the cells with IL-1␤ to test whether the inhibitory effect of LBP is LPS-specific. As shown in Table  I, the cells produced cytokines in response to both LPS and IL-1␤, and LBP inhibited responses only to LPS.
As shown in Fig. 2, LBP had its strongest inhibitory effect when added immediately after LPS-cell binding, and delaying the addition of LBP caused a progressive loss of its ability to inhibit the responses. Under the conditions of our 2-h incubation, LBP measurably inhibited IL-8 and tumor necrosis factor-␣ production when its addition was delayed for up to 30 min, and it inhibited IL-1␤ production when added up to 1 h after LPS-cell binding had occurred.
We next asked whether LBP could enhance the release or internalization of cell-bound LPS. LPS binding and internalization were measured by two independent methods (31,34). In the first (Fig. 3A), cell surface-bound [ 3 H]LPS was measured by the ability of proteinase K to release the [ 3 H]LPS from the cells. Protection of cell-associated [ 3 H]LPS from proteolytic release may either reflect its internalization or its transfer to a protease-resistant site on the cell surface. Therefore, we use a second method in which the binding and internalization of a fluorescent derivative of LPS were measured by flow cytometry; trypan blue was used to quench the fluorescence of the surface-exposed LPS (Fig. 3B). The results showed that whereas LBP slightly increased the release of LPS from the cells, most of the LPS remained associated with the cells after they were exposed to 3 g/ml of LBP, an LBP concentration that produced a 60 -75% inhibition of the cell response ( Fig. 1 and Table I). Whereas the LBP-induced reduction of cell-associated Alexa-LPS (Fig. 3B) was greater than that of [ 3 H]LPS (Fig. 3A), it is unclear whether more Alexa-LPS was actually released from the cells or whether the binding of LBP to the cell-associated Alexa-LPS caused a slight reduction in its fluorescence intensity. In any case, LBP did not significantly increase the rate of release of cell-associated Alexa-LPS except at the earliest time point. Also shown in Fig. 3, LBP did not significantly increase the internalization of cell-bound LPS.
To determine whether the cell-associated LPS remained bound to mCD14 after the cells were incubated with LBP, we derivatized LPS with a radioiodinatable, photoactivatable cross-linker (SASD) and measured the effect of LBP on the ability of cell-bound 125 I-ASD-LPS to cross-link to mCD14. The results shown in Fig. 4 suggest that LBP dramatically increased the dissociation of cell-bound 125 I-ASD-LPS from mCD14 in both CD14-transfected HEK 293 cells and THP-1 cells. The ability of LBP to remove LPS from mCD14 occurred rapidly; 60% inhibition of cross-linking occurred after the cells were incubated with LBP for 5 min (data not shown), and 64 -75% inhibition occurred after 10 min (Fig. 4). Immunopre-cipitation of LBP from lysates of the LBP-treated cells revealed that a significant amount of cell-bound LPS was associated with LBP (Fig. 4). No labeled protein band was immunoprecipitated from cells that were not incubated with LBP (data not shown). The specificities of the immunoprecipitations were demonstrated by Western blotting (see "Experimental Procedures"). In other experiments not shown, we confirmed that the 125 I-labeled CD14 immunoprecipitates were not derived from sCD14 that may have bound to the cells. When we cross-linked 125 I-ASD-LPS to sCD14 and ran this product on the gel beside the 125 I-labeled immunoprecipitate from the cell lysate, the 125 I-sCD14 was smaller (ϳ46 kDa) than the 125 I-mCD14 band (ϳ55 kDa), as expected.
We could not measure LPS binding to MD-2 and Tlr4, which are present in very low abundance on the cell surface. We therefore used soluble recombinant proteins in a cell-free assay to measure the impact of LBP on the transfer of 125 I-ASD-LPS from CD14 to MD-2. As shown in Fig. 5, sCD14 (lane 1) transferred LPS to sMD-2 (lane 2). LBP inhibited LPS transfer from sCD14 to sMD-2 (lanes 3 and 4) in keeping with the previous finding that LBP can inhibit the binding of free LPS to sMD-2 (17). When LBP was added after LPS had already been transferred to sMD-2 ( lanes 5 and 6), LBP removed a significant amount of the LPS from sMD-2 (lanes 7 and 8). Taken together, our data support the hypothesis that LBP may inhibit the transfer of LPS from mCD14 to the Tlr4-MD-2 signaling receptor. Fig. 1. The cells were washed and incubated in SFM to which rLBP was added either immediately (time 0) or at the indicated times after the incubation had begun. The total incubation time measured from time 0 in all samples was 120 min. Cytokines were measured in the culture supernatants by enzyme-linked immunosorbent assay. The means Ϯ S.D. for three experiments are shown as percentages of control.

TABLE I LBP does not inhibit cell responses to IL-1␤
THP-1 (VD 3 -PGE 2 ) cells were preincubated with LPS as described in the legend to Fig. 1 or with SFM. The cells were washed and incubated for 2 h in SFM or in SFM containing rIL-1␤ (2 ng/ml) in the presence or absence of 3 g/ml rLBP. The cytokine content of the culture supernatant is expressed as ng/ml and as a percentage of control (% of Ctrl) production obtained in the absence of LBP. The data are shown as mean Ϯ range of two experiments that were each performed in duplicate. Cytokines produced by unstimulated cells were less than 1% of the indicated control levels. Our data reveal a new mechanism by which LBP can inhibit cell responses to LPS. LBP rapidly removed up to 75% of cell-bound LPS from mCD14 (Fig. 4) and a significant amount of LPS from MD-2 (Fig. 5), suggesting that LBP attenuates signal responses by interfering with LPS interactions with the extracellular domains of mCD14 and the Tlr4⅐MD-2 receptor complex. Whereas our data show that LBP promoted some release of the cell-associated LPS into the medium (Fig. 3), it seems unlikely that this low percentage of LPS release could account for the strong inhibitory effect. The ability of LBP to rapidly and almost completely remove cell-bound LPS from mCD14 provides a more likely explanation for the impact of LBP on cellular responses. Our data also suggest that a significant amount of the cell-bound LPS forms a complex with LBP that remains associated with the cells. It is unclear if the LPS⅐LBP complexes remain bound to mCD14 or if they move to another membrane location. If the complexes remain associated with mCD14, their inability to stimulate the cells is consistent with the previous finding that most ternary LPS⅐LBP⅐mCD14 complexes do not trigger signal responses and are eventually internalized (30). If the LPS-LBP complexes become bound to another membrane structure, they may not be able to induce signaling, because LPS⅐LBP complexes can induce little or no signal response without the help of CD14 (10). Seydel and co-workers (36) showed that LBP may insert into monocyte membranes and enhance cell responses to LPS. In their experiments, LBP did not inhibit monocyte activation when LBP and LPS were added together, possibly because the LBP concentrations (Յ0.2 g/ml) in their experiments were below the threshold required for inhibition.
Although we used sCD14 to promote the binding of LPS to I-ASD⅐LPS⅐sCD14 complexes (120 -200 ng of LPS/ml) for 3 min at 37°C. The cells were washed and incubated for 10 min in SFM or in SFM containing 3 g/ml rLBP (SFM ϩ LBP) and were then exposed to UV light to induce cross-linking of the 125 I-ASD⅐LPS to protein. The cells were then washed, lysates were prepared, and mCD14 (CD14 i.p.) and LBP (LBP i.p.) were sequentially immunoprecipitated from the lysates. The immunoprecipitated 125 I-proteins were separated by SDS-PAGE, and the radioactivity in the CD14 and LBP bands was quantitated by phosphor imaging. To measure the inhibitory effect of LBP, the average CD14 radioactivity in lanes 3 and 4 is expressed as the percentage of the average CD14 radioactivity in lanes 1 and 2. The experiments were repeated twice with similar results. Similar results were obtained when mCD14 was immunoprecipitated with an antibody (63D3) that binds to a different CD14 epitope (data not shown).
FIG. 5. LBP inhibits LPS transfer from sCD14 to sMD-2. 125 I-ASD-LPS was preincubated with a 1.5-fold molar excess of sCD14 to produce LPS-sCD14 complexes. The complexes (4.2 pmol of LPS and 6.2 pmol of sCD14 in 25 l) were then incubated for 10 min in SFM containing the indicated proteins (shown as pmol/25 l). In lanes 5-8, the LPS⅐sCD14 complexes were preincubated with sMD-2 for 10 min followed by LBP for an additional 10 min. The 125 I-ASD⅐LPS was then covalently cross-linked to the binding proteins by exposing the samples to UV light. The 125 I-proteins were then separated on SDS-PAGE after reductive cleavage of the LPS, and the autoradiograph was produced by exposing the dried gel to film. The experiment was repeated with similar results. our cells, our data suggest that there was no interaction between LBP and sCD14 in our experiments. After the cells were washed to remove unbound LPS⅐sCD14 complexes, the cellbound LPS cross-linked to mCD14 and not to sCD14. We previously found that, under these conditions, sCD14 transferred LPS to mCD14 in an ϳ1:1 molar ratio of LPS to mCD14 (34). Under these conditions, all of the sCD14 is probably released from the cells, since the presence of LBP is required to promote measurable cellular retention of sCD14 (37).
Other inhibitory mechanisms of LBP include its ability to form large extracellular LPS-LBP complexes that have a reduced ability to stimulate cells (30,36,38). As noted above, these complexes appear to target most of the LPS for internalization. The ability of LBP to enhance internalization of LPS may be restricted to LPS⅐LBP complexes that are formed before LPS-cell binding occurs; we found that, once LPS monomers were bound to mCD14, LBP had little or no ability to enhance LPS internalization (Fig. 3). LBP also inhibits LPS activity by transferring it to plasma lipoproteins, which bind and neutralize LPS (6,29). In a physiological environment, such as plasma, it is likely that each of these mechanisms of inhibition contributes to the cumulative inhibitory effect of LBP, but the relative importance of each mechanism is difficult to determine.
Previous studies suggest that the balance between the stimulatory and inhibitory effects of LBP depends primarily upon the LBP concentration. When LBP and LPS were coincubated with macrophages in serum-free medium, concentrations of LBP that are much lower than those found in normal plasma (0.1 g of LBP/ml) promoted maximal cell stimulation, whereas higher concentrations (1-10 g/ml) were less stimulatory, suggesting that LBP was exerting a significant inhibitory effect at the higher concentrations (27,39). Whereas LBP can promote LPS signaling at all LBP concentrations, the inhibitory mechanisms require relatively high LBP concentrations to have a significant impact. In the same experiments, high concentrations of LBP were inhibitory only when the LPS concentration was low (Յ1 ng/ml), whereas at higher LPS concentrations, much higher concentrations of LBP were required to inhibit responses to the LPS (27,28). LBP concentrations often increase to very high levels in the blood of septic patients, and progressive immunodepletion of LBP from the serum of these patients revealed that the LBP can have a strong inhibitory effect on the activity of LPS (28).
The balance between the stimulatory and inhibitory effects of sCD14 are also concentration-dependent (26), although the inhibitory mechanism of sCD14 differs from that of LBP. Taken together, these findings suggest mechanisms by which the moderate to high concentrations of LBP and sCD14 that are found in human blood may help to prevent LPS-induced systemic inflammation, whereas lower concentrations of these proteins, which presumably occur in extravascular fluids, may promote beneficial inflammation at local sites of infection (40 -42).