Abstract
MD-2 is associated with Toll-like receptor 4 (TLR4) on the cell surface and enables TLR4 to respond to LPS. We tested whether MD-2 enhances or enables the responses of both TLR2 and TLR4 to Gram-negative and Gram-positive bacteria and their components. TLR2 without MD-2 did not efficiently respond to highly purified LPS and LPS partial structures. MD-2 enabled TLR2 to respond to nonactivating protein-free LPS, LPS mutants, or lipid A and enhanced TLR2-mediated responses to both Gram-negative and Gram-positive bacteria and their LPS, peptidoglycan, and lipoteichoic acid components. MD-2 enabled TLR4 to respond to a wide variety of LPS partial structures, Gram-negative bacteria, and Gram-positive lipoteichoic acid, but not to Gram-positive bacteria, peptidoglycan, and lipopeptide. MD-2 physically associated with TLR2, but this association was weaker than with TLR4. MD-2 enhanced expression of both TLR2 and TLR4, and TLR2 and TLR4 enhanced expression of MD-2. Thus, MD-2 enables both TLR4 and TLR2 to respond with high sensitivity to a broad range of LPS structures and to lipoteichoic acid, and, moreover, MD-2 enhances the responses of TLR2 to Gram-positive bacteria and peptidoglycan, to which the TLR4-MD-2 complex is unresponsive.
Mammalian innate immune system recognizes bacteria and their cell wall components through two pattern-recognition receptors, CD14 (1, 2) and Toll-like receptors (TLR)4 2 and 4. TLR2, first identified as the cell-activating receptor for Gram-negative LPS (3, 4), also serves as the receptor for both Gram-positive and Gram-negative bacteria, mycobacteria, Mycoplasma, and spirochetes, and their peptidoglycan (PGN), lipoteichoic acid (LTA), lipoarabinomannan, and lipoprotein cell wall or cell membrane components (5, 6, 7, 8, 9, 10, 11). TLR2 seems to serve as the primary receptor for Gram-positive bacteria and their cell wall components, whereas TLR4 (but not TLR2) serves as the primary LPS receptor, because TLR2 (but not TLR4) knockout mice are unresponsive to Gram-positive bacteria (12), whereas TLR4 (but not TLR2) knockout mice are unresponsive to Gram-negative bacteria and LPS (12, 13). This notion is consistent with the hyporesponsiveness to LPS, but not to Gram-positive bacteria, of C3H/HeJ and in C57BL/10ScCr mice, which have a mutation in the Tlr4 gene (14, 15), and with the responsiveness of TLR2-deficient cells to LPS, but not to Gram-positive bacteria (16). Moreover, in the initial studies the responses in TLR2-transfected nonmacrophage cells required much higher concentrations of LPS than the concentrations needed to activate macrophages (1, 2, 3, 4, 5, 6, 17), and most recently highly purified LPS was shown not to stimulate cells through TLR2 (18). Therefore, these results suggested that by itself TLR2 does not function as an LPS receptor.
However, TLR4 by itself does not function as an LPS receptor either. To function as an LPS receptor, TLR4 requires a helper molecule, MD-2 (19). MD-2 is a 160-amino acid protein that is associated with TLR4 on the cell surface and enables TLR4 to respond to LPS (19, 20). The discovery of MD-2 explained why TLR4-transfected cells (in which MD-2 was not expressed) were unresponsive to LPS (4, 5, 6, 8) and supported the proposed function of the TLR4-MD-2 complex as the LPS receptor. However, it is not known whether MD-2 enables TLR4 to respond to Gram-positive bacterial components and also what effect MD-2 has on TLR2-mediated responses. Therefore, the aim of this study was to determine whether MD-2 functions as a helper molecule for both TLR2 and TLR4 for the responses to both Gram-positive and Gram-negative bacterial components.
Materials and Methods
Materials
All materials were purchased from Sigma (St. Louis, MO), unless otherwise indicated. Smooth LPS from Escherichia coli K12 LCD25 (List, Campbell, CA), E. coli O127:B8, E. coli O113 (refined endotoxin standard, Ribi, Hamilton, MT), Salmonella minnesota, Salmonella typhimurium, and Shigella flexneri 1A were obtained by phenol-water extraction. Rough LPS mutants from E. coli EH100 (Ra), S. typhimurium TV119 (Ra), E. coli J5 (Rc), S. minnesota R5 (Rc), S. typhimurium SL684 (Rc), E. coli F583 (Rd), S. minnesota R7 (Rd), E. coli K12 D31 m4 (Re) (List), S. minnesota Re595 (Re), S. typhimurium SL1181 (Re), and S. flexneri (Re) were obtained by phenol-chloroform-petroleum ether extraction. Ra mutants have the entire lipid A and core polysaccharide, but are devoid of the O-polysaccharide, Rc mutants are devoid of the O-polysaccharide and four terminal sugars in the core polysaccharide, Rd mutants are devoid of the O-polysaccharide and five or six terminal sugars in the core polysaccharide, and Re mutants are devoid of the O-polysaccharide and seven terminal sugars of the core polysaccharide, with only three KDO units of the core polysaccharide that remain bound to lipid A. Diphosphoryl lipid A were from E. coli F-583 and S. minnesota Re595. Detoxified (delipidized) LPS from E. coli O127:B8 was chromatographically purified. The purity of LPS was analyzed by periodate-modified silver staining of 12% SDS-polyacrylamide gels (17, 21) and by silver-enhanced colloidal gold staining (22) using a Bio-Rad kit (Hercules, CA) as recommended by the manufacturer.
Soluble PGN (sPGN) from Staphylococcus aureus was purified by affinity chromatography and contained <24 pg endotoxin/mg (23). LTA from S. aureus, Streptococcus pyogenes, and Streptococcus mutans, prepared by phenolic extraction and purified by hydrophobic interaction chromatography on octyl-Sepharose (17), contained <500 pg endotoxin/mg. Synthetic lipopeptide (Pam3Cys-Ser-Lys4OH) was obtained from Boehringer Mannheim (Indianapolis, IN). Micrococcus luteus ATCC 4698, Bacillus subtilis ATCC 6633, S. aureus 845 (all contained <500 pg endotoxin/mg), E. coli K12, Enterobacter cloace ATCC 13047, Serratia marcescens ATCC 13880, S. typhimurium ATCC 10289, and Proteus vulgaris ATCC 13315 were heat killed (70°C, 30 min).
Cell cultures
Human HEK 293 cells were cultured in DMEM with 10% endotoxin-free (<6 pg/ml) FCS as before (5). Stable transfectants of B lymphocytic cell line Ba/F3, expressing human TLR4 and human MD-2, were generated as before (19). RAW264.7 cells were stimulated, and supernatants were assayed for TNF-α and IL-6 and cell lysates for NF-κB (2, 17).
Transfection and luciferase assay
For cell activation, 293 cells were cultured at 0.3–0.35 × 106/ml in 48-well plates (0.25 ml/well) for 16–20 h and transfected with Lipofectamine in a serum-free medium with optimal concentrations of the following plasmids: 0.22 μg/ml NF-κB reporter endothelial leukocyte adhesion molecule 1 luciferase plasmid (5); 0.027 μg/ml human CD14 (5); 0.22 μg/ml of Flag-tagged human TLR1, TLR2, or TLR4 (5); and 0.22 μg/ml of 6His and Flag-tagged human MD-2 (19). When TLRs, MD-2, or CD14 were not used, equivalent amounts of appropriate control vectors were included. In addition, the total DNA concentration was brought up to 1.25 μg/ml with salmon sperm DNA. After 5 h, 10% FCS was added, and, after an additional 16–18 h, fresh medium with 10% FCS was added and cells were stimulated for 6 h as indicated in Results. The lysates were assayed for luciferase activity (5), and the data were expressed as normalized relative luciferase units.
Immunoprecipitation
Human MD-2 (19) was tagged with protein C epitope followed by the 6His epitope and subcloned into the pEFBOS expression vector. For coimmunoprecipitation experiments, 293 cells were cultured in 10-cm plates (six per group) until 60–70% confluent and transfected with Lipofectamine with 0.22–0.44 μg/ml of Flag-tagged human TLR2 or TLR4 (5), 0.22 μg/ml human Flag-tagged c-Jun N-terminal kinase (JNK)-1 (24), and 0.22–0.6 μg/ml 6His and protein C (PC)-tagged human MD-2. When TLRs or MD-2 were not used, equivalent amounts of appropriate control vectors were included. In addition, the total DNA concentration was brought up to 1.25 μg/ml with salmon sperm DNA. Thirty-three hours after transfection, the cells were lysed in 1 ml of lysis buffer per plate (0.05 M Tris-HCl, pH 8.0, with 0.3 M NaCl, 1% Triton X-100, 10 mM imidazole, and EDTA-free protease inhibitor mixture) at 4°C. Insoluble cell debris were removed by centrifugation (10,000 × g) and 5 ml of each supernatant was rotated for 5–10 h with 5 μl of anti-Flag (M-2)-agarose (Sigma) for anti-PC Western blot to detect coprecipitated MD-2, 0.5 ml of each supernatant was rotated with 2.5 μl of Ni-NTA-agarose (Qiagen, Chatsworth, CA) for anti-PC Western blot to detect the expression of MD-2, and another 0.5 ml of each supernatant was rotated with 1.25 μl of anti-Flag (M-2)-agarose for anti-Flag Western blot to detect the expression of TLRs. For other immunoprecipitation experiments, done to detect the expression of each construct, lysates from cells from one 10-cm plate/group were used. The agarose immunoprecipitates were washed in the lysis buffer and subjected to SDS-PAGE (12%) and Western blot analysis with anti-Flag M2 (Sigma) or anti-PC (Boehringer Mannheim) primary Abs, peroxidase-labeled anti-mouse γ-chain secondary Abs, and ECL-Plus ECL reagent (Amersham Pharmacia Biotech, Piscataway, NJ).
Results
TLR2 does not efficiently recognize highly purified endotoxic LPS and LPS partial structures
We tested the responsiveness of 293 cells transfected with TLR2 to 20 different LPS and LPS partial structures, including a number of smooth LPS from various bacterial species, as well as a series of Ra, Rc, Rd, and Re LPS mutants (which have progressively truncated polysaccharide portion), lipid A (the minimum endotoxic structure of LPS), and detoxified LPS. In all experiments, the cells were also cotransfected with CD14 to mimic the natural expression of CD14 in monocytic cells. Several LPS preparations induced very high responses, whereas other LPS preparations induced intermediate or very low responses, or did not stimulate the cells at all (Fig. 1⇓). However, there was no correlation between the LPS structure and the ability to activate cells, e.g., some smooth LPS (E. coli LCD25) and some ReLPS (E. coli) were highly stimulatory, whereas other smooth LPS (E. coli O113 or S. flexneri), some ReLPS (S. minnesota or S. typhimurium), or lipid A were nonstimulatory or very weakly stimulatory (Fig. 1⇓). All of these preparations induced very high secretion of cytokines in mouse macrophages or human monocytes (Refs. 1, 2, 17 and data not shown).
TLR2 efficiently recognizes some, but not all endotoxic LPS and LPS partial structures. 293 cells, transfected with TLR2, CD14, and NF-κB-luciferase reporter plasmids, were stimulated with the indicated LPS preparations, and cell lysates were assayed for luciferase activity. The results are means ± SE of three to five experiments. Ec, E. coli; Sm, S. minnesota; St, S. typhimurium; Sf, S. flexneri.
This lack of correlation between the LPS structure and activating capacity suggested that possibly some contaminants could have been responsible for the stimulatory activity of these LPS preparations. Indeed, it was previously reported that three of six commercial LPS preparations were contaminated with endotoxin protein (22), and very recently it was shown that highly purified LPS did not stimulate cells through TLR2 (18). We tested the purity of our LPS preparations with a periodate-silver stain that stains both LPS and proteins (with the sensitivity of 10 ng protein/lane) and with a silver-enhanced colloidal gold that stains proteins only, but not LPS (with the sensitivity of 0.4 ng protein/lane). The most stimulatory LPS preparations (ReLPS from E. coli, LPS from E. coli LCD25, RaLPS from E. coli, LPS from E. coli 0127) had the highest protein contamination (Fig. 2⇓), and the extent of contamination (Fig. 2⇓) correlated with the stimulating capacity of these LPS preparations (Fig. 1⇑). On the other hand, LPS preparations that were not contaminated with proteins (lipid A from S. minnesota, ReLPS from S. minnesota and S. typhimurium, and LPS from E. coli 0113, which were >99.99% protein free) (Fig. 2⇓) did not stimulate cells through TLR2 (Fig. 1⇑).
Purity of LPS and LPS partial structures. The indicated LPS preparations (25 μg/lane in A and 15 μg/lane in B) were subjected to 12% SDS-PAGE and staining with periodate-sliver stain, which stains LPS and proteins (A), or blotting and staining with enhanced colloidal gold, which stains proteins only (B). For abbreviations, see the legend to Fig. 1⇑.
These results demonstrate that TLR2 does not efficiently recognize highly purified LPS endotoxic structures and are consistent with the recent report (18) that contaminating endotoxin proteins might have been responsible for the previously observed TLR2-activating capacity of LPS.
MD-2 enables TLR2-mediated responses to LPS and LPS partial structures and enhances TLR2-mediated responses to sPGN, LTA, and both Gram-positive and Gram-negative bacteria
Because activation of cells by LPS through TLR4 requires MD-2 (19), we next tested the hypothesis that MD-2 may also enable TLR2 and CD14 to respond to LPS. When 293 cells were cotransfected with TLR2, CD14, and MD-2, they became highly responsive to all LPS preparations and LPS partial structures (except detoxified LPS), including highly purified protein-free LPS, ReLPS, and lipid A preparations that did not activate or very poorly activated TLR2/CD14-transfected cells (Fig. 3⇓ and data not shown). In the presence of MD-2, LPS preparations efficiently activated TLR2/CD14-transfected cells even at 0.1–1 ng/ml (Fig. 4⇓). For example, 0.1 ng/ml S. minnesota ReLPS and 1 ng/ml E. coli 0113 LPS yielded about 20% of maximum responses, which is higher than the response to 10 μg/ml of these ReLPS or LPS without MD-2 (Fig. 4⇓). Thus, MD-2 enhanced 104–105 times the responses of TLR2/CD14-transfected cells to LPS and ReLPS. These results demonstrate that MD-2 enables very sensitive TLR2/CD14-mediated recognition of protein-free LPS and LPS partial structures and enhances responses to all endotoxic LPS.
MD-2 enables or enhances TLR2-mediated responses to LPS, LPS partial structures, sPGN, and LTA. 293 cells, transfected with TLR2, CD14, and NF-κB-luciferase reporter plasmids, with (+) or without (−) MD-2 plasmid, were stimulated with the indicated preparations, and cell lysates were assayed for luciferase activity. The results are means ± SE of three to five experiments. Sa, S. aureus; Sp, S. pyogenes; others as in the legend to Fig. 1⇑.
MD-2 enables TLR2-mediated responses to low concentrations of LPS. 293 cells, transfected with TLR2, with or without MD-2, and with CD14 and NF-κB-luciferase reporter plasmids, were stimulated with the indicated concentrations of protein-free ReLPS or LPS, and cell lysates were assayed for luciferase activity. The results are from one of two similar experiments. For abbreviations, see the legend to Fig. 1⇑.
We next tested the hypothesis that MD-2 may also enhance TLR2/CD14-mediated responses to sPGN and LTA from Gram-positive bacteria, a synthetic lipopeptide (an analogue of bacterial lipoproteins), as well as intact bacteria. MD-2 highly enhanced the responses to sPGN, LTA (Fig. 3⇑), and both Gram-positive and Gram-negative bacteria (Fig. 5⇓), but not to lipopeptide. Cell activation by Gram-positive bacteria or their cell wall components was not due to endotoxin contamination, because these preparations were endotoxin free and because cells transfected with TLR4, CD14, and MD-2 were unresponsive to these stimuli, whereas they were highly responsive to endotoxin and Gram-negative bacteria (see next section below).
MD-2 enhances TLR2-mediated responses to both Gram-positive and Gram-negative bacteria. 293 cells were transfected as in Fig. 3⇑ and stimulated with the indicated bacteria. The results are means ± SE of three to five experiments.
Cells transfected with TLR2 or TLR2 and MD-2 without CD14 showed lower responses to all of the stimulants than cells transfected with all three plasmids (data not shown), which confirms that CD14 is not indispensable, but that it enhances TLR2-mediated responses (5, 6). Cells transfected with MD-2 alone, CD14 alone, or with MD-2 and CD14 were unresponsive to all of the stimulants (data not shown), thus confirming the requirement for TLR2 for the above responses and for the enhancing effect of MD-2. Cells transfected with TLR1, CD14, and MD-2 were also unresponsive to all of the stimuli tested (data not shown).
MD-2 enables TLR4-mediated responses to LPS, LPS partial structures, Gram-negative bacteria, and LTA, but not to sPGN, lipopeptide, and Gram-positive bacteria
We next tested whether MD-2 could enable TLR4 to recognize the same large variety of LPS, LPS partial structures, and other cell wall components, as well as intact Gram-positive and Gram-negative bacteria. None of these stimulants activated cells transfected with TLR4 and CD14 without MD-2. However, cells transfected with TLR4, CD14, and MD-2 were highly responsive to all LPS and LPS partial structures (except detoxified LPS) (Fig. 6⇓). Effective cell activation could be even achieved with 0.1 ng/ml endotoxin (data not shown). These cells were also responsive to LTA from Gram-positive bacteria, but not to sPGN and lipopeptide (Fig. 6⇓), and were highly responsive to Gram-negative, but not to Gram-positive bacteria (Fig. 7⇓). Similarly, a B lymphocytic cell line Ba/F3, stably expressing TLR4 and MD-2 (19), was also unresponsive to sPGN and S. aureus, whereas, as expected (19), it was highly responsive to E. coli LPS and S. minnesota ReLPS (Fig. 8⇓).
MD-2 enables TLR4-mediated responses to LPS, LPS partial structures, and LTA, but not to sPGN and lipopeptide. 293 cells, transfected with TLR4, CD14, and NF-κB-luciferase reporter plasmids, with (+) or without (−) MD-2 plasmid, were stimulated with the indicated preparations, and cell lysates were assayed for luciferase activity. The results are means ± SE of three to five experiments. Abbreviations: see Figs. 1⇑ and 3⇑.
MD-2 enables TLR4-mediated responses to Gram-negative, but not to Gram-positive bacteria. 293 cells were transfected as in Fig. 6⇑ and stimulated with the indicated bacteria. The results are means ± SE of three to five experiments.
Ba/F3 cells expressing TLR4 and MD-2 are responsive to LPS, but not to sPGN and S. aureus. Stable Ba/F3 transfectants expressing NF-κB reporter luciferase plasmid and TLR4 alone or TLR4 and MD-2 were stimulated with low, medium, or high concentrations of sPGN (0.4, 2, or 10 μg/ml), S. aureus (0.8 × 108, 4 × 108, or 20 × 108 cells/ml), or LPS (0.1, 1, or 10 μg/ml), and cell lysates were assayed for luciferase activity. The results are from one of two similar experiments.
MD-2 associates with both TLR2 and TLR4 and enhances expression of TLR2 and TLR4
To begin to understand the mechanism of MD-2 enhancement of TLR2-mediated responses, we first tested whether MD-2 physically associates with TLR2, because it was earlier shown that MD-2 physically associates with TLR4 (19, 20). Indeed, in cells transiently cotransfected with PC-tagged MD-2 and Flag-tagged TLR2 or TLR4, MD-2 could be detected with anti-PC Abs on Western blots of Flag immunoprecipitates (Fig. 9⇓), thus indicating physical association of MD-2 with TLR2 and TLR4. MD-2 was not detectable in Flag immunoprecipitates from cells transfected with MD-2 alone, control vectors alone, TLR2 or TLR4 alone, or MD-2 and Flag-tagged unrelated molecule, JNK-1 (Fig. 9⇓), thus demonstrating the specificity of MD-2-TLR association, as well as the specificity of immunoprecipitation and anti-Flag and anti-PC Abs. However, the association of MD-2 with TLR2 was weaker than with TLR4, as judged by a much lower amount of MD-2 in the TLR2 than in the TLR4 coprecipitations, despite equivalent amounts of MD-2, TLR2, and TLR4 expressed in both groups and loaded on the gel (Fig. 9⇓).
MD-2 coprecipitates with both TLR2 and TLR4. 293 cells were transfected with the indicated Flag-tagged TLRs or JNK-1 and/or PC-tagged MD-2, and cell lysates were subjected to immunoprecipitation. MD-2 was detected in anti-Flag TLR2 and TLR4 immunoprecipitates by Western blotting with anti-PC Abs. The expression of MD-2 was demonstrated in Ni-NTA-agarose precipitates by Western blotting with anti-PC Abs and of TLRs and JNK-1 in anti-Flag immunoprecipitates by Western blotting with anti-Flag Abs. The results are from one of three similar experiments.
To look for other possible mechanisms of MD-2 enhancement of TLR2-mediated responses, we also studied the effect of MD-2 on the expression of TLR2 and TLR4. Cotransfection of cells with MD-2 and TLR2 or TLR4 greatly enhanced the expression of TLR2, TLR4, and MD-2 (Fig. 10⇓). This effect was specific for MD-2, TLR2, and TLR4, because cotransfection of cells with another unrelated Flag-tagged molecule (JNK-1) or control vectors did not enhance the expression of MD-2 (Figs. 9⇑ and 10⇓) or TLR2 or TLR4 (data not shown). The enhancement of MD-2, TLR2, and TLR4 expression was observed with both Flag-tagged MD-2 (Fig. 10⇓) and PC-tagged MD-2 (Fig. 9⇑ and data not shown).
MD-2 enhances expression of TLR2 and TLR4, and TLR2 and TLR4 enhance expression of MD-2. 293 cells were transfected with Flag-tagged MD-2, TLR2, TLR4, or JNK-1 alone or in combination, and the expression of each protein was detected in anti-Flag immunoprecipitates by Western blotting with anti-Flag Abs. The results are from one of three similar experiments.
These results suggest that MD-2 enables or enhances TLR2- and TLR4-mediated responses by physically associating with TLR2 and TLR4 and by increasing the amount of expressed TLR2 and TLR4 protein.
Discussion
Our results demonstrate that MD-2 can work not only with TLR4, as discovered earlier (19, 20), but also with TLR2 to enable TLR2 to respond to nonactivating or low-activating stimuli, such as protein-free LPS, LPS mutants, or lipid A. MD-2 can also enhance the responses of TLR2 to both Gram-positive and Gram-negative bacteria, as well as their sPGN, LTA, and LPS-associated cell wall components that can activate cells through TLR2 without MD-2. Thus, TLR2 is able to respond to several (but not all) stimuli without MD-2, and MD-2 enhances its responses to these stimuli, whereas TLR4 cannot be activated by any of these stimuli without MD-2 and, therefore, TLR4 has an absolute requirement for MD-2 for cell activation. However, for some ligands (e.g., protein-free LPS), TLR2 behaves like TLR4, i.e., it also requires MD-2 for cell activation by these ligands. Therefore, MD-2 both broadens the spectrum of stimulants that can activate cells through TLR2 and enhances the responsiveness of TLR2 to ligands that do not absolutely require MD-2 for TLR2-mediated cell activation.
The mechanism of the enhancement of TLR2-mediated responses could be based on both the physical association of MD-2 with TLR2 (which, however is much weaker than with TLR4) and on the enhancement by MD-2 of TLR2 protein expression (or it could be based on either of these mechanisms alone). However, the ability of MD-2 to enable the responses of TLR2 and TLR4 to the stimulants that are not active in the absence of MD-2 is unlikely to be solely due to the enhancement of TLR expression, because cells overexpressing TLR2 or TLR4 without MD-2 are unresponsive to these stimulants (4, 5, 6, 18, 19). Therefore, the enabling function of MD-2 for TLR2- and TLR4-mediated responses is likely to be based on both the physical association of MD-2 with TLRs and the enhancement of their expression. The exact mechanism of this enhanced expression is unknown, but, because of the physical association of TLRs with MD-2, it may rely on a greater stability of MD-2-TLR complexes, compared with the stability of TLR and MD-2 alone.
Our data confirm previous results that TLR4 associates with MD-2 and requires MD-2 for the responses to LPS (19, 20) and demonstrate that MD-2 makes TLR4 highly responsive to a wide variety of endotoxic LPS and LPS partial structures, and also to intact Gram-negative bacteria and isolated LTA from Gram-positive bacteria. They also demonstrate that the TLR4-MD-2 complex does not respond to Gram-positive bacteria, their sPGN component, and lipopeptide. Thus, the TLR4-MD-2 complex has a narrower specificity than the TLR2-MD-2 complex. Our data on the specificity of TLR4 responses are consistent with the experiments on TLR4 knockout mice (12, 13), which showed that TLR4 responds to Gram-negative LPS and Gram-positive LTA. Our data on the specificity of TLR2 responses are consistent with the previous reports showing TLR2 responsiveness to both Gram-negative LPS and Gram-positive bacteria and their PGN and LTA components, as well as lipopeptide (5, 6, 8, 25), and with the recent report showing unresponsiveness of TLR2 (without MD-2) to highly purified protein-free LPS (18).
Our results also explain the variable and often low responses or the lack of responses through TLR2 (in TLR2-transfected cells) to various endotoxic LPS preparations (that are much more active in monocytic cells; Refs. 1, 2, 17) by showing that TLR2 alone is unresponsive to highly purified protein-free LPS, LPS mutants, and lipid A, and that the responses to “active” LPS preparations correlate with the level of their protein contamination. Our results also extend these observations by showing that in the presence of MD-2 both TLR2 and TLR4 are highly and equally sensitive to all endotoxic LPS, including protein-free LPS.
However, our results in 293 cells are different from the initial results obtained in TLR2 and TLR4 knockout mice and in Tlr4 mutant mice, which showed that TLR4 knockout and Tlr4 mutant mice, which had an intact Tlr2 gene, were unresponsive to LPS and LTA, thus indicating that in these mice TLR2 could not function as an effective LPS and LTA receptor (12, 13, 14, 15). This difference may be due to the inducible expression of TLR2 (26) or due to the insufficient expression or function of MD-2 in TLR4 knockout and Tlr4 mutant mice. This possibility is consistent with much stronger association of MD-2 with TLR4 than with TLR2 observed by us. Therefore, the amount of MD-2 in TLR2 knockout mice may be sufficient to support the function of TLR4, but in TLR4 knockout or Tlr4 mutant mice may not be sufficient to support the function of TLR2. Moreover, in most primary cells, the expression of TLR2 is much lower than TLR4 (26, 27) and the expression of TLR2 is inducible by cell stimulation (26). Because the expression of MD-2 is enhanced by the expression of TLRs, it is possible that TLR4 knockout and mutant mice have insufficient expression of MD-2, because of the lack of the enhancing effect of TLR4 on the expression of MD-2 and because of the low expression of TLR2. Insufficient expression of MD-2 and low expression of TLR2 would then explain the unresponsiveness of TLR4 knockout and mutant mice to pure LPS, which requires MD-2, but not to Gram-positive bacteria, which do not require MD-2.
Moreover, our results are consistent with two recent studies. The first one (28) demonstrated that unresponsiveness of TLR4 knockout mice was only limited to some LPS preparations; however, because there is no data on the purity of the LPS preparations used, these results need to be interpreted with caution. The second one indicated that γδ T cells respond to LPS and lipid A through TLR2 (29).
Our results confirm the specificity of the TLR4-MD-2 complex for glycolipids, which is a structural feature shared by both LPS and LTA, and suggest the requirement for both the glycan and the lipid components of LPS and LTA, because neither detoxified (delipidated) LPS nor the lipid-containing lipopeptide (that lacks the glycan component) could stimulate cells through the TLR4-MD-2 complex. Our results also suggest that LTA does not contribute to the responses to intact Gram-positive bacteria, because cells transfected with TLR4, CD14, and MD-2 were responsive to isolated LTA, but were unresponsive to LTA-containing Gram-positive bacteria. This lack of cell activation by intact bacteria may be due to the anchoring of the lipid part of LTA in the bacterial cell membrane, which is located underneath the thick PGN layer.
In summary, our results demonstrate that MD-2 enhances TLR2-mediated responsiveness to both Gram-negative and Gram-positive bacteria and their LPS, PGN, and LTA components and that it enables TLR2 to respond with high sensitivity to a much broader variety of stimulants, including protein-free LPS.
Acknowledgments
We thank Dr. Roger J. Davis for providing the JNK-1 plasmid.
Footnotes
-
↵1 This work was supported by the U.S. Public Health Service Grant AI2879 from the National Institutes of Health.
-
↵2 Address correspondence and reprint requests to Dr. Roman Dziarski, Indiana University School of Medicine, 3400 Broadway, Gary, IN 46408. E-mail address: rdziar{at}iun.edu
-
↵3 Current address: Washington University School of Medicine, St. Louis, MO 63110.
-
↵4 Abbreviations used in this paper: TLR, Toll-like receptor; LTA, lipoteichoic acid; PGN, peptidoglycan; sPGN, soluble PGN; JNK-1, c-Jun N-terminal kinase; PC protein C.
- Received July 26, 2000.
- Accepted November 13, 2000.
- Copyright © 2001 by The American Association of Immunologists
References
In this issue
