Mouse Toll-like Receptor 4·MD-2 Complex Mediates Lipopolysaccharide-mimetic Signal Transduction by Taxol*

Taxol, an antitumor agent derived from a plant, mimics the action of lipopolysaccharide (LPS) in mice but not in humans. Although Taxol is structurally unrelated to LPS, Taxol and LPS are presumed to share a receptor or signaling molecule. The LPS-mimetic activity of Taxol is not observed in LPS-hyporesponsive C3H/HeJ mice, which possess a point mutation in Toll-like receptor 4 (TLR4); therefore, TLR4 appears to be involved in both Taxol and LPS signaling. In addition, TLR4 was recently shown to physically associate with MD-2, a molecule that confers LPS responsiveness on TLR4. To determine whether TLR4·MD-2 complex mediates a Taxol-induced signal, we constructed transformants of the mouse pro-B cell line, Ba/F3, expressing mouse TLR4 alone, both mouse TLR4 and mouse MD-2, and both mouse MD-2 and mouse TLR4 lacking the cytoplasmic portion, and then examined whether Taxol induced NFκB activation in these transfectants. Noticeable NFκB activation by Taxol was detected in Ba/F3 expressing mouse TLR4 and mouse MD-2 but not in the other transfectants. Coexpression of human TLR4 and human MD-2 did not confer Taxol responsiveness on Ba/F3 cells, suggesting that the TLR4·MD-2 complex is responsible for the species specificity with respect to Taxol responsiveness. Furthermore, Taxol-induced NFκB activation via TLR4·MD-2 was blocked by an LPS antagonist that blocks LPS-induced NFκB activation via TLR4·MD-2. These results demonstrated that coexpression of mouse TLR4 and mouse MD-2 is required for Taxol responsiveness and that the TLR4·MD-2 complex is the shared molecule in Taxol and LPS signal transduction in mice.

Taxol, a diterpene purified from the bark of the Western yew (Taxus brevifolia) (1), is an antitumor agent that blocks mitosis by binding and stabilizing microtubules (2,3). Ding et al. (4) found that Taxol induces the secretion of tumor necrosis factor and down-regulation of tumor necrosis factor receptors in murine macrophages. Although the structure of Taxol is quite different from that of lipopolysaccharide (LPS), 1 Taxol has been shown to possess many LPS-like activities, such as tyrosine phosphorylation of microtubule-associated protein kinases (5), induction of LPS-inducible gene expression (6), and activation of NFB (7). Interestingly, Taxol mimics the actions of LPS on murine macrophages but not on human LPS-responsive cells including macrophages (8,9).
LPS contains polysaccharide and lipid A portion, the latter of which mediates many LPS responses (10). Several synthetic and natural lipid A analogs, which lack LPS-like activities, have been shown to retain the ability to block LPS-induced cellular responses (11)(12)(13). Taxol-induced signaling events in murine macrophages are blocked by some of these LPS antagonists, suggesting that LPS and Taxol share a receptor or signaling molecule (14). Although the target of the antagonists was not defined well, membrane-bound CD14 (mCD14), which has been demonstrated to be involved in LPS-induced signaling events on macrophages (15), might not be the target, because the LPS antagonists suppress LPS-induced signaling under conditions in which it does not block LPS binding to mCD14 (16). These findings are consistent with the observation that Taxol signaling was blocked by an LPS antagonist in a mutant macrophage cell line defective in mCD14 expression (17). Identification of molecules shared in LPS-and Taxol-induced signaling will provide a new insight into the development of antiinflammatory drugs.
The LPS-mimetic activity of Taxol was not observed in macrophages from a spontaneous LPS-hyporesponsive mutant, C3H/HeJ mice (4 -7). Analysis of recombinant inbred mice showed that the genes controlling the responses to LPS and Taxol were closely linked (4). The Lps gene, which has been shown to be responsible for LPS hyporesponsiveness in C3H/ HeJ mice, was recently mapped to the Toll-like receptor (TLR) 4. TLR4 from C3H/HeJ mice has a point mutation that causes a failure to activate NFB (18 -20). TLRs constitute a mammalian transmembrane protein family and are similar to Drosophila Toll in that they have extracellular domains containing leucine-rich repeats and a cytoplasmic portion homologous to the intracellular signaling domain of the type 1 interleukin-1 receptor (21,22). Studies over the past few years have demonstrated that Drosophila Toll and mammalian TLRs play crucial roles in innate immune recognition (23). Although C3H/HeJ and generated TLR4-deficient mice show LPS hyporesponsiveness (20), expression of TLR4 is not sufficient to confer LPS responsiveness on the human embryonic 293 cell line and the mouse pro-B cell line, Ba/F3, implying a lack of a factor in the transformants (24,25). Expression of MD-2, a molecule that physically associates with TLR4 on the cell surface, with TLR4 has been demonstrated to confer LPS responsiveness on Ba/F3, which expresses neither TLR4 nor MD-2 (25). These recent findings prompted us to determine whether the TLR4⅐MD-2 complex is the shared molecule in Taxol and LPS signaling. In this paper we demonstrate that the mouse TLR4⅐MD-2 complex mediates LPS-mimetic signal transduction by Taxol.

EXPERIMENTAL PROCEDURES
Reagents-RPMI 1640 medium and Taxol from Taxus brevifolia were purchased from Sigma. Fetal bovine serum was purchased from Atlanta Biologica. LPS prepared from Escherichia coli 0111:B4 was purchased from List Biological Laboratories. B464 was provided by Eisai Co., Ltd. (Tokyo, Japan). Tetra-His TM antibody was purchased from Qiagen. BCA protein assay reagents were purchased from Pierce. All other chemicals used were of reagent grade or better.
cDNAs and Expression Constructs-The cDNA encoding human MD-2 was described previously (25). The cDNAs encoding mouse TLR4 or mouse MD-2 were cloned in our own laboratory and will be described elsewhere. 2 Restriction sites (XhoI and BamHI) were introduced by polymerase chain reaction, and the cDNAs were cloned into an expression vector, pEFBOS (26). The DNA fragment encoding the FLAG epitope followed by the 6xHis epitope was introduced into the pEFBOS vector such that all expressed protein bore the FLAG and 6xHis epitope at the C termini. The mouse TLR4⌬ cDNA encodes the truncated protein in which the cytoplasmic region encompassing C-terminal 134 amino acids (662-835) was deleted.
Stable Transformants and Cell Culture-The Ba/F3 cell line was obtained from the RIKEN Cell Bank (Tsukuba, Japan). A Ba/F3 cell line stably expressing human TLR4, human MD-2, and p55IgLuc, an NFBdependent luciferase reporter construct, was established as described previously (25). The expression constructs were transfected into Ba/F3 cells, which express neither endogenous TLR4 nor MD-2 (25), by electroporation in the order of the construct encoding mouse TLR4 or TLR4⌬ and followed by that encoding mouse or human MD-2. The mouse MD-2 construct was also transfected into Ba/F3 cells stably expressing human TLR4 (named Ba/hTLR4), which was established previously (25). p55IgLuc was finally introduced into the transfectant expressing mouse TLR4 and mouse or human MD-2. Expression of mouse TLR4 or TLR4⌬ was confirmed by staining the C-terminal FLAG epitope, which was located inside the cells. Expression of mouse or human MD-2 was confirmed by immunoblotting with Tetra-His TM antibody. Ba/F3 and Ba/F3 stable transformants were maintained as described previously (27) except for the use of heat-inactivated (56°C for 30 min) fetal bovine serum.
Luciferase Assay-Unless otherwise indicated, cells were inoculated into a 48-well dish (Corning) or 1.5-ml tube (Treff) at 2 ϫ 10 5 cells/500 l of cell culture medium. Taxol, LPS, and LPS antagonists were added as described in the figure legends. After stimulation at 37°C for 4 h, cells were harvested and lysed in 50 l of cell culture lysis reagent (Promega Corp.), and then luciferase activity was measured using 5 l of lysate and 25 l of luciferase assay substrate (Promega Corp.). The luminescence was quantified with a luminometer (Berthold Japan, Tokyo, Japan).

Coexpression of Mouse TLR4 and Mouse MD-2 Confers Taxol
Responsiveness on Ba/F3-Genetic characterization of C3H/ HeJ mice suggested that a point mutation in TLR4 affects Taxol responsiveness in mice (4,18,19). It was left undetermined whether TLR4 directly mediates Taxol-induced signaling. To address this question, we examined Taxol responsiveness by measuring NFB activation in a Ba/F3 stable transfectant expressing epitope-tagged mouse TLR4 (mTLR4), named Ba/mTLR4, and in a Ba/F3 stable transfectant expressing both mTLR4 and epitope-tagged mouse MD-2 (mMD-2), named Ba/mTLR4/mMD2. After Taxol stimulation, nuclear extracts were prepared to measure NFB activation by means of an electrophoretic mobility shift assay (EMSA). As shown in Fig. 1A, Taxol evidently induced NFB activation in Ba/ mTLR4/mMD2 cells under the condition in which NFB activation was not found and slightly induced in Ba/F3 and parental Ba/mTLR4 cells, respectively. Similar results were obtained with LPS stimulation (Fig. 1A) as described previously (25).
Requirement of coexpression of mTLR4 and mMD-2 for noticeable NFB activation by Taxol was more evident when Taxol dose response of Ba/mTLR4/mMD2 was compared with that of other strains (Fig. 1B). Activation of NFB by Taxol was also determined by measuring luciferase activity in Ba/mTLR4/ mMD2 cells transfected with an NFB-dependent luciferase reporter plasmid, p55IgLUC. Taxol increased the luciferase activity in a dose-dependent manner (Fig. 2B). These results showed that the Ba/mTLR4/mMD2 transfectant acquired Taxol responsiveness.
To determine whether a Taxol-induced signal was mediated by the TLR4 cytoplasmic portion, which is thought to be involved in LPS signaling events (21, 23, 28), we measured NFB activation by Taxol in a Ba/F3 stable transfectant expressing both mMD-2 and epitope-tagged mouse TLR4 lacking the cytoplasmic portion (mTLR4⌬), named Ba/mTLR4⌬/mMD2, by EMSA. As shown in Fig. 1, NFB was not activated through Taxol stimulation on Ba/mTLR4⌬/mMD2 cells, implying that the TLR4 cytoplasmic portion is required for Taxol signaling, and that expression of mMD-2 without TLR4 on Ba/F3 is not sufficient for Taxol responsiveness. To eliminate the possibility that the difference in Taxol responsiveness between the transfectants was due to the expression levels of TLR4 and MD-2 proteins, we examined the expression level of the tagged proteins by immunoblot analysis. The expression levels of mTLR4 or mTLR4⌬ protein in Ba/mTLR4, Ba/mTLR4/mMD2, and Ba/ mTLR4⌬/mMD2 were similar, and that of mMD-2 protein was higher in Ba/mTLR4⌬/MD2 cells than in Ba/mTLR4/mMD2 cells (data not shown).
These results, taken together, demonstrate that coexpression of mTLR4 and mMD-2 enables Ba/F3 cells to activate NFB noticeably by Taxol stimulation. Such a requirement for coexpression of TLR4 and MD-2 has also been reported in the case of LPS responsiveness on Ba/F3 cells (Ref. 25 and Fig. 1A), suggesting that the mTLR4⅐mMD-2 complex is the shared molecule in Taxol and LPS signaling.
An LPS Antagonist, B464, Blocks Taxol-induced NFB Activation via mTLR4⅐mMD-2-To further confirm that the mTLR4⅐mMD-2 complex is the shared molecule in Taxol and LPS signaling, the effect of a synthetic lipid A analog, B464, which blocks LPS-induced TNF production in macrophages (13,29), on Taxol-induced NFB activation was examined. B464 inhibited LPS-induced NFB activation in Ba/mTLR4/mMD2 and Ba/hTLR4/hMD2 cells in a dose-dependent manner, LPSinduced NFB activation being completely blocked with a 100fold higher dose of B464 in both the transfectants (Fig. 3, A and  B). Next, we examined whether B464 inhibits Taxol-induced NFB activation in Ba/mTLR4/mMD2 cells. As shown in Fig.  3C, Taxol-induced NFB activation was also inhibited in these cells by B464 in a dose-dependent manner. Reporter activity induced by 1 M Taxol was reduced by 73% when the cells were preincubated with a 6.3-fold concentration of B464 (10 g/ml, 6.3 M), which did not affect the basal reporter gene activity. This result is consistent with the previous finding that LPS antagonists inhibit the LPS-mimetic activity of Taxol in murine macrophages (14).
Serum Is Not Required for Taxol Signaling via mTLR4⅐mMD-2-It has been demonstrated that LPS-induced signaling is dependent on serum as a source of LPS-binding protein and/or soluble CD14 (15). Induction of signals by LPS in Ba/hTLR4/hMD2 cells, which do not express mCD14 (25), has been demonstrated to be dependent on soluble CD14 in serum. 2 To determine whether serum factor(s) is required for Taxol-induced signaling via mTLR4⅐mMD-2, the dependence of NFB activation by Taxol on serum was examined in the Ba/mTLR4/mMD2 transfectant. As   FIG. 4. Serum is not required for Taxol-induced NFB activation via mTLR4⅐mMD-2. Ba/mTLR4/mMD2 cells were harvested and then resuspended in CHO serum-free medium (Sigma) supplemented with or without 10% fetal bovine serum. Cells were incubated at 37°C for 4 h with the indicated amounts of Taxol or LPS, which was followed by the measurement of luciferase activity. The bars indicate the averages of more than three independent assays. Error bars indicate standard deviations.

FIG. 2. Taxol responsiveness is conferred by mouse TLR4⅐MD-2 but not by human TLR4⅐MD-2.
A, Ba/mTLR4/mMD2 or Ba/hTLR4/ hMD2 cells were cultivated in cell culture medium or the medium containing 10 ng/ml LPS, 10 M Taxol or 1% Me 2 SO. After a 4-h cultivation, cells were harvested, and luciferase activity was measured.
The bars indicate averages of duplicate wells; deviations from the mean were Ͻ10%. B, Ba/mTLR4/mMD2 (q) and Ba/hTLR4/hMD2 (E) cells were exposed to the indicated concentrations of Taxol at 37°C for 4 h, and then luciferase activity was measured. Data represent the averages of triplicate wells. Error bars indicate standard deviations. C, Indicated cells were incubated without (Ϫ) or with (ϩ) 30 M Taxol or 100 ng/ml LPS as described in the legend of Fig. 1A. After incubation, nuclear extracts were prepared and subjected to EMSA to measure NFB activation as described previously (33). The results shown are representative of two independent experiments.

FIG. 3. An LPS antagonist inhibits Taxol-induced NFB activation via mTLR4⅐mMD-2.
A and B, the indicated amounts of B464 were added to Ba/hTLR4/hMD2 (A) or Ba/mTLR4/mMD2 (B) cells, together with (ϩ) or without (Ϫ) 10 ng/ml LPS. After 4 h of cultivation at 37°C, cells were harvested, and luciferase activity was measured. C, Ba/mTLR4/mMD2 cells were precultured with the indicated amounts of B464 for 30 min and then further cultivated at 37°C for 4 h without (Ϫ) or with (ϩ) 1 M Taxol, which was followed by the measurement of luciferase activity. The bars indicate the averages of triplicate wells.
Error bars indicate standard deviations.

TLR4⅐MD-2 Mediates Taxol Signaling
shown in Fig. 4, luciferase activity was clearly increased by Taxol even in the absence of serum, and the reporter activity was further increased when cells were stimulated by Taxol in medium supplemented with 10% serum. On the other hand, LPS-induced NFB activation was absolutely dependent on serum. These results show that serum factor(s) is not required for Taxol-induced signaling mediated by mTLR4⅐mMD-2.

DISCUSSION
In this study, we show that Taxol responsiveness is acquired through the coexpression of mTLR4 and mMD-2 on Ba/F3 cells. Furthermore, the LPS antagonist, B464, blocks both LPS-and Taxol-induced signals mediated by mTLR4⅐mMD-2 complex. Taken together, these results demonstrate that mTLR4⅐mMD-2 complex mediates the LPS-mimetic activity of Taxol and that the complex is the shared molecule in Taxol and LPS signaling.
The Taxol-induced signal via mTLR4⅐mMD-2 complex activates NFB. It has been demonstrated that expression of a constitutively, active human TLR4 construct, the extracellular portion of which was replaced by CD4, induced the activation of NFB and that the induced signal was mediated by adaptor protein MyD88, which associates with the cytoplasmic portion of TLR4, and subsequently by IRAK and TRAF6 (28). Taxolinduced NFB activation might be mediated by a similar pathway, as we found in this study that the cytoplasmic portion of mouse TLR4 was required for Taxol signaling. The fact that Taxol-induced NFB activation shares this pathway with LPS signaling might be responsible for its LPS-mimetic activity.
An important question to be addressed is whether TLR4⅐MD-2 complex can directly recognize extracellular stimulants such as LPS and Taxol. To determine whether mTLR4⅐mMD-2 complex is a receptor for Taxol, [ 3 H]Taxol binding to Ba/mTLR4/mMD2 cells was compared with that of parental Ba/F3 cells, and we did not detect any significant Taxol binding to Ba/mTLR4/mMD2 cells. 3 Previously, we had also examined whether LPS binds to hTLR4⅐hMD-2, but we did not detect significant LPS binding to hTLR4⅐hMD-2 complex. 2 Thus, the mechanism of recognition of the stimulant by TLR4⅐MD-2 remains to be elucidated. If mTLR4⅐mMD-2 were a receptor for Taxol, the binding of only a minute amount of Taxol to mTLR4⅐mMD-2, undetectable under our binding assay conditions, might be sufficient to induce a signal. Alternatively, another molecule(s) might serve as an initial receptor for Taxol, and then the Taxol⅐receptor complex might be recognized by mTLR4⅐mMD-2. Recently, several proteins were identified as Taxol-binding proteins, such as CD18 (30) and HSP-90 (31). One or both of these Taxol-binding proteins might serve as an initial receptor and mediate signals to mTLR4⅐mMD-2 complex.
Interestingly, Taxol-induced signaling was mediated by mTLR4⅐mMD-2 complex but not by hTLR4⅐hMD-2 complex, implying that a Taxol-responsive domain exists in mTLR4⅐mMD-2 complex. Human TLR4 exhibited 69% amino acid sequence identity with mouse TLR4, and human MD-2 exhibited 66% amino acid sequence identity with mouse MD-2. 2 The relatively low homology of TLR4 and/or MD-2 between man and mouse may be responsible for the species specificity of Taxol responsiveness on TLR4⅐MD-2. Our results shown in Fig. 2C suggest that mouse MD-2, but not human MD-2, is important for mediating Taxol signaling. How-ever, the precise role of TLR4 and/or MD-2 in Taxol signaling remains to be elucidated. Because the structure of Taxol is quite different from that of LPS, the molecular mechanism underlying Taxol recognition might be different from that of LPS recognition. Therefore, identification of the Taxol-responsive domain in mTLR4 (or hTLR4)⅐mMD-2 complex will provide a new clue for understanding the mechanism underlying ligand recognition by TLR4⅐MD-2 complex. Such fundamental information concerning signal transduction through TLR4⅐MD-2 complex will also provide a novel insight for the development of anti-inflammatory drugs.