Recent studies have implicated a family of mammalian Toll-like receptors (TLR) in the activation of macrophages by Gram-negative and Gram-positive bacterial products. We have previously shown that different TLR proteins mediate cellular activation by the distinct CD14 ligands Gram-negative bacterial LPS and mycobacterial glycolipid lipoarabinomannan (LAM). Here we show that viable Mycobacterium tuberculosis bacilli activated both Chinese hamster ovary cells and murine macrophages that overexpressed either TLR2 or TLR4. This contrasted with Gram-positive bacteria and Mycobacterium avium, which activated cells via TLR2 but not TLR4. Both virulent and attenuated strains of M. tuberculosis could activate the cells in a TLR-dependent manner. Neither membrane-bound nor soluble CD14 was required for bacilli to activate cells in a TLR-dependent manner. We also assessed whether LAM was the mycobacterial cell wall component responsible for TLR-dependent cellular activation by M. tuberculosis. We found that TLR2, but not TLR4, could confer responsiveness to LAM isolated from rapidly growing mycobacteria. In contrast, LAM isolated from M. tuberculosis or Mycobacterium bovis bacillus Calmette-Guérin failed to induce TLR-dependent activation. Lastly, both soluble and cell wall-associated mycobacterial factors were capable of mediating activation via distinct TLR proteins. A soluble heat-stable and protease-resistant factor was found to mediate TLR2-dependent activation, whereas a heat-sensitive cell-associated mycobacterial factor mediated TLR4-dependent activation. Together, our data demonstrate that Toll-like receptors can mediate cellular activation by M. tuberculosis via CD14-independent ligands that are distinct from the mycobacterial cell wall glycolipid LAM.

The discovery of mammalian homologues of the Drosophila Toll receptor protein has elicited interest in the role of these proteins in innate immunity (reviewed in Ref. 1). Several published reports have illustrated the potential importance of Toll-like receptors (TLR)3 in intracellular signaling. Janeway and colleagues (2) reported a human homologue of the Drosophila Toll protein, a protein later to be designated Toll-like receptor 4 (TLR4). There were three features of Toll that linked this protein with innate immunity and intracellular signaling. First, Drosophila Toll participates in an antifungal response in the adult fly (3), and it is likely that mammalian homologues would participate in similar innate immune responses. Second, the Drosophila Toll participates in a signal transduction pathway leading to the activation of the transcription factor Dorsal, the fly homologue of NF-κB. The central role played by NF-κB in signal transduction pathways activated by cytokines, and in the regulation of cytokine genes themselves, implicates mammalian Toll proteins in cellular responses similar to those evoked by cytokines. Third, the intracellular domains of Toll proteins share significant sequence similarity with the type I IL-1 receptor, the known mammalian Toll-like receptors, and the cytosolic adapter protein MyD88 (4, 5, 6).

Recent data have demonstrated that mammalian TLR proteins participate in intracellular signaling initiated by Gram-negative bacterial LPS. CD14 has been recognized for many years as the major receptor responsible for the effects of LPS on macrophages, monocytes, and neutrophils (reviewed in Ref. 7). Two groups independently reported that TLR2 could function as a signaling receptor for LPS in the presence of CD14 (8, 9). These investigators reported that human HEK293 cells stably transfected with TLR2 could respond to LPS in the presence of CD14 and LPS-binding protein, as judged by activation of a reporter gene under the control of the NF-κB-dependent ELAM-1 promoter. Deletion mutants of TLR2 that lack a region of the intracellular domain that shares sequence homology with the IL-1 receptor failed to mediate LPS responsiveness in this assay.

Subsequent to these findings, two other groups identified the gene responsible for the LPS-hyporesponsive phenotype of the C3H/HeJ mouse (10, 11). Macrophages from this mouse are hyporesponsive to LPS, even though they express normal amounts of CD14 on their surface. The gene locus responsible for this defect (Lpsd) mapped to the Tlr4 gene. In the C3H/HeJ mice, a single missense mutation within the Tlr4 coding sequence was identified (P712H). Supporting evidence for the hypothesis that this mutation is responsible for the LPS-hyporesponsive phenotype of the C3H/HeJ mouse comes from the finding that the C57BL/10ScCr LPS-nonresponsive mouse does not express TLR4. These results raised the possibility that, at least in mice, TLR2 is not sufficient to confer LPS responsiveness. Data from our own laboratories have shown that TLR4 is predominantly responsible for LPS signaling in murine and hamster cells, whereas TLR2 mediates cellular activation by a distinct CD14 ligand, the mycobacterial cell wall glycolipid lipoarabinomannan (LAM).4 This conclusion was further supported by our demonstration that LPS hyporesponsive macrophages from C3H/HeJ mice were not hyporesponsive to LAM.4 Most recently, Chow et al. (13) reported that human TLR4 could mediate LPS responsiveness in HEK293 cells. Together, these results suggest that TLR2 and TLR4 mediate CD14-dependent signals in both a ligand- and species-specific manner.

We have extended these earlier studies to test the hypothesis that Mycobacterium tuberculosis, bacteria that do not synthesize LPS, might also be recognized by TLR proteins. It was recently found that cellular activation by Gram-positive bacteria and Mycobacterium avium was mediated by TLR2, but not TLR4 (Ref. 14 and data not shown).5 Here we report that M. tuberculosis activates cells in a TLR-dependent manner, but unlike Gram-positive bacteria and M. avium, these organisms utilize both TLR2 and TLR4 proteins. Unlike Gram-positive bacteria, TLR-dependent cellular activation by M. tuberculosis does not appear to depend on the presence of CD14. Furthermore, the mycobacterial ligands responsible for this activation appear to be distinct from LAM.

M. tuberculosis strains H37Rv (ATCC 25618), H37Ra (ATCC 25177), and Mycobacterium bovis BCG (ATCC 35734) were purchased from the American Type Culture Collection (ATCC, Manassas, VA). Bacterial cultures were grown in Middlebrook 7H9 medium supplemented with Tween 80 and ADC (Difco, Detroit, MI) at 37°C under biosafety level 3 conditions. Bacterial culture medium was prepared in LPS-free flasks using LPS-free water (Baxter-Travenol, Deerfield, IL). Bacterial cultures were grown to midlogarithmic phase (OD620 nm = 0.4), and CFU per ml were determined by growth on Middlebrook agar plates. The CHO-K1 fibroblast (CCL-61) and RAW264.7 murine macrophage cell lines (TIB-71) were purchased from the ATCC. RAW264.7 cells were maintained in DMEM (BioWhittaker, Walkersville, MD) supplemented with 10% heat-inactivated fetal bovine serum (HyClone Laboratories, Logan, UT), 10 mM HEPES, 2 mM l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin (BioWhittaker). Chinese hamster ovary (CHO) cells were maintained in Ham’s F-12 culture medium (BioWhittaker) supplemented with 10% heat-inactivated FBS (HyClone, Logan, UT), 10 mM HEPES, 2 mM l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin (BioWhittaker). LPS levels in all medium components were <10 pg/ml final concentration as indicated by BioWhittaker or measured by the Limulus amebocyte lysate kit (BioWhittaker). Cells were cultured at 37°C in the presence of 5% CO2 in a humidified incubator. CHO/CD14, 3E10 (CHO/CD14/ELAM-CD25), 3E10/TLR2, 3E10/TLR4, and CHO/TLR2 were previously described4 (14, 15) and cultured as described above. All of these stable cell lines used expressed a similar level of surface CD14, TLR2, and TLR4. As described below, cells that were stably transfected with the TLR2 and TLR4 expression plasmids express FLAG-tagged TLR proteins. After transfection and selection of stable lines, clones were selected that expressed similar levels of TLR proteins, based on mean fluorescence intensity determined with the same anti-FLAG Ab. Furthermore, the TLR2- and TLR4-expressing cell lines were generated from the same CHO/CD14 parental clonal cell line, and thus each line also expresses the same amount of CD14.

LPS (purified from Escherichia coli 055:B5) was purchased from Sigma (St. Louis, MO). Mycobacterial LAM, purified from rapidly growing avirulent mycobacteria (AraLAM), M. bovis BCG, and M. tuberculosis strains H37Rv and H37Ra were all provided by Dr. John Belisle (Colorado State University, Fort Collins, CO) under the provisions of National Institutes of Health Contract NO1 AI25147. Levels of contaminating LPS in the LAM preparations were determined using a quantitative Limulus lysate assay (BioWhittaker) and were <1 pg/ml final concentration in all experiments. A neutralizing anti-LAM IgG3 mAb that recognizes LAM from these different mycobacteria (CS-35) was also provided by Dr. Belisle and was previously described (16). Recombinant human IL-1β was purchased from Genzyme (Cambridge, MA). FITC- and PE-conjugated anti-human CD25 mAbs were purchased from Becton Dickinson (Bedford, MA).

Nuclear extracts were prepared essentially as described by Schreiber et al. (17). Approximately 1.0 × 107 CHO or RAW264.7 cells were washed and harvested by scraping in Ca2+- and Mg2+-free PBS (BioWhittaker). Cells were pelleted by centrifugation at 800 × g for 10 min at 4°C. Cell pellets were resuspended in 400 μl of a buffer containing 10 mM KCl, 10 mM Tris-HCl (pH 7.8), 5 mM MgCl2, 0.5 mM DTT, 0.3 M sucrose, 10 mM β-glycerol phosphate, 0.1 mM EGTA, 1 mM PMSF, and 5 μg/ml each of aprotinin, leupeptin, chymostatin, and antipain and incubated on ice for 10 min. Subsequently, 25 μl of 10% Nonidet P-40 (Sigma) were added to each sample before vortexing. The nuclei were centrifuged for 1 min at 5000 × g to pellet the nuclei. Nuclear pellets were resuspended in a nuclear extraction buffer containing 320 mM KCl, 10 mM Tris-HCl (pH 7.8), 5 mM MgCl2, 0.5 mM DTT, 10 mM β-glycerol phosphate, 0.1 mM EGTA, 25% glycerol, 1 mM PMSF, and 5 μg/ml each of aprotinin, leupeptin, antipain, and chymostatin. Samples were extracted on ice for 15 min followed by centrifugation at 16,000 × g for 10 min at 4°C. Protein concentration was determined using the Bio-Rad assay kit (Bio-Rad, Hercules, CA). All nuclear extracts were stored at −70°C, and multiple freeze-thawing cycles were avoided.

A double-stranded oligonucleotide containing a single copy of the IL-2 receptor α-chain NF-κB (GGGGAATTCC) was used as an EMSA probe. DNA probes were labeled with [α-32P]dNTPs (DuPont-NEN, Boston, MA) using E. coli DNA polymerase Klenow fragment (U.S. Biochemicals, Cleveland, OH) as recommended by the manufacturer. Unincorporated nucleotides were removed with Sephadex G-25 columns (5 Prime→3 Prime, Boulder, CO). Nuclear extracts (typically 3 μg) were incubated with radiolabeled probe DNA (0.1 ng, typically 10,000 cpm) in the presence of 2 μg poly(dI-dC) (Pharmacia, Piscataway, NJ), 1.0 mM EDTA, 10 mM Tris-HCl (pH 7.9), 25 mM glycerol, and 0.5 mM DTT in a final volume of 20 μl, as previously described (18). Binding reactions were then incubated at room temperature for 30 min. After incubation, a portion of each binding reaction (typically 6 μl) was loaded onto 7% nondenaturing low ionic strength polyacrylamide gel. The gels were then dried and visualized by autoradiography.

The ELAM-luc reporter plasmid was generated by subcloning the promoter of pUMS(ELAM)-Tac (15) into the promoterless pGL3 luciferase reporter plasmid (Promega, Madison, WI). The human TLR2 and TLR4 cDNAs cloned into the pFlag-CMV-1 mammalian expression plasmid were gifts of Drs. Carsten Kirschning and Mike Rothe (Tularik, South San Francisco, CA) and were previously described (9). These gene products are expressed as fusion proteins containing an N-terminal FLAG epitope tag. A second TLR4 expression plasmid that expresses native TLR4 (termed hToll) was a gift of Dr. Charles Janeway (Yale University, New Haven, CT) and was previously described (2). Plasmids were prepared using Qiagen (Valencia, CA) plasmid DNA purification columns, DNA was eluted from the columns using LPS-free buffers, and contaminating LPS levels were found to be <10 pg/ml. Furthermore, all plasmid preparations were unable to activate the LPS-sensitive CHO/CD14 cells, demonstrating that the plasmids were not contaminated with LPS.

Transient transfections were performed using SuperFect reagent (Qiagen, Valencia, CA) as per the manufacturer’s instructions. Briefly, cells were plated on six-well dishes 1–2 days before transfection, and transfections were performed when cells plated reached 80% confluence. Plasmid DNA was added to 100 μl of Opti-Mem reduced serum media (Life Technologies, Gaithersburg, MD). All transfections utilized a total of 4 μg of plasmid DNA consisting of 2 μg of reporter plasmid, 1 μg of each expression vector, and the balance made up with empty vector described above. SuperFect, 10 μl, was added to the DNA-medium mixture and incubated for 10 min at ambient temperature. Subsequently, 600 μl of serum-containing medium were added to the reaction mixture and added to the individual wells. Each reaction was prepared individually, and each condition was performed in triplicate. Reactions were incubated with the cells for 2–3 h, whereupon the reaction was removed from the cells and fresh medium containing serum was added. On the following day, individual wells were left untreated or were stimulated with either LPS or LAM as indicated in the figures. Cells were then incubated for an additional 5 h before harvesting. Luciferase assays were performed as described below. All transfection experiments were repeated at least three times with different plasmid preparations, and a single representative experiment is shown. Each single experiment represents triplicate independent transfections and data are expressed as average values ± S.D.

Luciferase activity was measured using the Luciferase Assay System (Promega), according to manufacturer’s instructions. Briefly, cells were washed and scraped on ice in cold PBS, pelleted by centrifugation, and resuspended in 100 μl of reporter lysis buffer. Samples were freeze-thawed once and centrifuged at 14,000 × g for 10 min at 4°C to remove cellular debris. Supernatants were recovered and assayed for total protein using the Bio-Rad protein assay according to manufacturer’s instructions. An equal amount of total protein from each lysate was assayed for luciferase activity as measured by light emissions in a scintillation counter.

In experiments with 3E10 cells, which contain a stably transfected CD25 reporter gene under the control of the ELAM-1 promoter, reporter gene expression was measured by flow cytometry as previously described4 (14, 15). Data were collected using FACScan software (Becton Dickinson, Mountain View, CA) and expressed as the ratio (fold activation) of the percent of CD25+ cells in unstimulated and stimulated cell populations. The 95% confidence limit for nonspecific fluorescence was established with the use of isotype control Abs. Each experiment was repeated at least three times in all cases.

Previous studies demonstrated that both bacterial cell wall components (LPS, LAM, and peptidoglycan) and heat-killed Gram-positive bacteria (Staphylococcus aureus and Streptococcus pneumoniae) could activate cells in a TLR-dependent manner4 (14, 19). Here we tested whether live M. tuberculosis bacilli (Mtb) could also activate cells in a TLR-dependent manner. To measure TLR-dependent activation, we used CHO fibroblast cell lines that were stably transfected with expression plasmids encoding human CD14 (CHO/CD14) and human TLR2 proteins (CHO/CD14/TLR2). These lines were also stably transfected with a reporter plasmid consisting of a human CD25 cDNA under the control of an NF-κB-dependent ELAM-1 promoter. As we previously published, activation of these cells in a CD14-dependent manner leads to the rapid activation of NF-κB and the subsequent expression of CD25 on the cell surface4 (14, 15). Furthermore, CD14- and TLR-independent stimuli that activate NF-κB (e.g., IL-1β protein) also activate CD25 expression. Here we incubated viable Mtb with CHO/CD14 and CHO/CD14/TLR2 cells (10 Mtb/CHO cell) for 16 h and subsequently measured CD25 expression by flow cytometry. As shown in Fig. 1, both the virulent H37Rv and attenuated H37Ra strains of Mtb did not activate the CHO/CD14 cells, and overexpression of TLR2 conferred Mtb responsiveness on these cells. This TLR-dependent activation did not appear to vary with the virulence of the Mtb strain used. Similar data were obtained with avirulent M. bovis BCG bacilli (data not shown). CD25 expression was also measured at 1, 4, 8, 16, and 24 h after stimulation by H37Ra and H37Rv. At each time point analyzed, we found that both Mtb strains activated the cells to a similar extent (data not shown). Lastly, surface expression of either CD14 or TLR proteins did not alter Mtb binding or uptake by the CHO cells compared with untransfected CHO cells (data not shown).

Our earlier data showed that TLR-dependent activation of cells by the mycobacterial cell wall glycolipid LAM was substantially enhanced in the presence of LPS-binding protein4 (LBP (20)). To determine whether LBP present in serum was required for TLR-dependent activation of cells by Mtb, we repeated the preceding experiment in the presence and absence of serum. As shown in Fig. 2, the presence of serum did not significantly affect responsiveness of cells to Mtb, suggesting that LBP is not required for TLR-dependent activation by Mtb. Because the FBS used in these studies was heat inactivated, our data also demonstrate that neither opsonization nor activated complement was required for this TLR-dependent activation. This is not surprising because CHO cells do not express either complement receptors or other receptors known to mediate the binding of Mtb to macrophages (e.g., macrophage mannose receptor). Because LBP, at least in part, transfers bacterial glycolipids to CD14, our data also suggested that CD14 was not required for this TLR-mediated activation of cells by Mtb.

We recently showed that TLR-dependent activation of cells by either LAM or Gram-positive bacteria required the presence of CD14 as either a membrane-bound or soluble protein4 (14). To test the hypothesis that Mtb activate cells in a TLR-dependent, but CD14-independent manner, we compared the capacity of Mtb to activate CHO/TLR2 cells (i.e., CHO fibroblasts that express TLR2, but not CD14) in the absence and presence of serum, which serves as a source of soluble CD14. Nuclear extracts were prepared from the CHO/TLR2 cells 1 h after Mtb stimulation, and NF-κB levels were subsequently measured by EMSA. As shown in Fig. 3,A, similar levels of NF-κB were induced in a TLR-dependent manner in the presence and absence of soluble CD14. NF-κB levels in CHO/CD14 cells were not altered by exposure to Mtb. Furthermore, we compared the capacity of Mtb to activate NF-κB in CHO/CD14/TLR2 vs CHO/TLR2 cells and found that the presence of membrane-bound CD14 did not augment Mtb responsiveness (Fig. 3 B). These data demonstrate that TLR-dependent activation by Mtb is not mediated by CD14. This contrasts with Gram-positive bacteria where activation is dependent on both TLR2 and CD14 (14, 19).

Recent studies revealed that both Gram-positive bacteria and M. avium can activate cells via TLR2, but not TLR4 (Ref. 14 and data not shown).5 We sought to determine whether Mtb also activated cells via the same selective use of TLR proteins. Mtb were added to CHO/CD14/TLR cells. As shown in Fig. 4,A, we found that Mtb could activate the cells via both TLR2 and TLR4. Thus, Mtb differed from Gram-positive bacteria and M. avium in their utilization of TLR proteins during cellular activation. Virtually identical results were obtained with either H37Ra or H37Rv Mtb, or when the experiment was performed in the presence or absence of serum (data not shown). Furthermore, activation via either TLR2 or TLR4 was not dependent on the presence of membrane-bound CD14 (data not shown). We subsequently examined whether Mtb could also activate macrophages via both TLR proteins. To test this possibility, we transiently cotransfected RAW264.7 murine macrophages with a luciferase reporter plasmid under the control of the NF-κB-dependent ELAM-1 promoter and the various TLR expression plasmids. After transfection, the cells were stimulated with Mtb (10 Mtb/RAW264.7 cell) and harvested 5 h later. Lysates were prepared from the harvested cells, and luciferase activity was measured as described in Materials and Methods. As shown in Fig. 4 B, Mtb activated the macrophages in a TLR-dependent manner. Like the CHO cells, this activation could be mediated by both TLR2 and TLR4.

Our previous studies revealed that LAM purified from rapidly growing Mycobacteria could activate cells in a TLR-dependent manner.4 This form of LAM contains highly branched arabinofuranosyl side chains (termed AraLAM), whereas Mtb and M. bovis BCG contain LAM that is terminally capped with mannose residues (ManLAM). These chemical differences are believed to be responsible for the different biological activities of these glycolipids (reviewed in Ref. 21). We sought to determine whether LAM on the surface of the mycobacteria was the ligand responsible for TLR-dependent activation of the cells. First, we compared the capacity of LAM isolated from rapidly growing Mycobacteria (AraLAM), Mtb (ManLAM), and M. bovis BCG (BCG LAM) to activate cells in a TLR-dependent manner. As shown in Fig. 5, AraLAM was capable of activating CHO/CD14 cells that expressed TLR2, but not TLR4. In contrast to AraLAM, neither Mtb LAM nor BCG LAM were capable of activating any of the cell lines tested. This finding is consistent with the known relative inability of these ManLAMs to activate macrophages (22, 23, 24).

To confirm that TLR-dependent cellular activation by Mtb was not mediated by LAM, we determined whether an anti-LAM Ab could block activation of cells by Mtb bacilli. Mtb were pretreated with the neutralizing CS-35 anti-LAM mAb (107 H37Ra bacilli treated with 500 μg/ml CS-35 for 15 min), and then added to the CHO/CD14/TLR cells (10 Mtb/CHO cell, 12.5 μg/ml CS-35 final concentration). As shown in Fig. 6,A, the anti-LAM mAb was incapable of blocking Mtb activation of cells via TLR2 or TLR4. The ability of CS-35 to block LAM-induced cellular activation was confirmed by the finding that this mAb could block TLR2-dependent activation by AraLAM (Fig. 6,B). Normal rabbit serum had no effect on either Mtb- or AraLAM-induced cellular activation (data not shown). Thus, using two distinct experimental approaches (Figs. 5 and 6), we have demonstrated that TLR-dependent activation of cells by Mtb was not mediated by the cell wall glycolipid LAM. Although we have shown that CD14 does not appear to play a role in TLR-dependent cellular activation by Mtb, it remains to be determined whether the mycobacterial ligands activate TLR proteins directly or in association with other cell surface receptors.

We sought to determine whether soluble factors released by cultured Mtb could activate the cells in a TLR-dependent manner. H37Ra Mtb were grown to mid-log phase as described in Materials and Methods, bacilli were removed by centrifugation and filtration, and the Mtb-free culture medium was used to stimulate the CHO/CD14/TLR cells. As shown in Fig. 7,A, we found that Mtb-conditioned culture medium activated the cells via TLR2, but not TLR4. Fresh culture medium did not activate the cells under any conditions tested (data not shown). We subsequently examined whether the soluble factors responsible for TLR-dependent activation were proteins. The Mtb-conditioned culture medium was digested with proteinase K (50 μg/ml, 55°C, 1 h) and boiled for 5 min to inactivate the protease. This protein-free Mtb-conditioned culture medium was then evaluated for its capacity to activate cells in a TLR2-dependent manner. As shown in Fig. 7 B, the protein-free and heat-stable preparation was still capable of inducing TLR2-dependent cellular activation. Thus, it appears that a nonproteinaceous heat-stable factor is the TLR2 ligand. Based on the chemical nature of other known TLR ligands (e.g LPS, LAM, peptidoglycan), it is likely that this factor is a polysaccharide or a glycolipid. It is also possible that this preparation contains several distinct ligands for TLR2.

To test whether LAM present in the Mtb-conditioned culture medium was responsible for the TLR2-dependent cellular activation, we assessed the capacity of the CS-35 anti-LAM neutralizing Ab to block TLR2-dependent cellular activation by the Mtb-conditioned culture medium. CHO/CD14/TLR2 cells were stimulated with Mtb-conditioned culture medium in the presence or absence of either CS-35, or a matching isotype control Ab. The CS-35 Ab had no effect on the capacity of Mtb-conditioned culture medium to activate the CHO/CD14/TLR2 cells (data not shown). This finding demonstrates that LAM is not responsible for the observed activation, and is consistent with our earlier finding that purified Mtb LAM could not activate cells in a TLR-dependent manner (Fig. 5).

Because viable Mtb bacilli could induce both TLR2- and TLR4-dependent signaling, but the soluble mycobacterial factor only activated cells via TLR2, we sought to determine whether it was a cell-associated factor that induced TLR4-dependent signaling. As shown in Fig. 7 C, we found that heat-killed Mtb that were washed free of culture medium activated cells via TLR2. In contrast to the viable Mtb, the heat-killed bacilli could not activate cells via TLR4. These differences imply that the ligand for TLR4-dependent activation is a heat-labile cell-associated mycobacterial factor. Cellular activation via TLR4 was not caused by LPS contamination because all materials used to perform this experiment were essentially LPS free and cellular activation by the bacilli was not CD14 dependent. Thus, soluble mycobacterial factors selectively activate cells in a TLR2-dependent manner, whereas heat-labile cell-associated mycobacterial factors selectively activate cells in a TLR4-dependent manner. These findings demonstrate that whole Mtb and soluble mycobacterial factors differ in their utilization of TLR proteins leading to cellular activation.

Mammalian TLR proteins are novel mediators of cellular activation by bacterial products. Our previous studies have shown that distinct bacterial products (LPS and LAM) bind to a common receptor (CD14) but utilize different TLR proteins (TLR4 and TLR2, respectively). We have now shown that Mtb bacilli can activate both CHO cells and murine macrophages via both TLR2 and TLR4. This contrasts with Gram-positive bacteria and M. avium which activate cells via TLR2, but not TLR4. Both virulent and attenuated Mtb strains can activate the cells in a TLR-dependent manner. Neither membrane-bound nor soluble CD14 appears to be required for Mtb to activate cells in a TLR-dependent manner. We also found that TLR2, but not TLR4, can confer responsiveness to LAM isolated from rapidly growing mycobacteria. In contrast, neither LAM isolated from Mtb nor M. bovis BCG can activate cells in a TLR-dependent manner. Lastly, both soluble and cell wall-associated mycobacterial factors can mediate activation via distinct TLR proteins. Together, our data demonstrate that mammalian Toll-like receptors can mediate cellular activation by Mtb via CD14-independent ligands that are distinct from LAM.

M. tuberculosis, the etiological agent of tuberculosis, is a major worldwide public health threat. Mtb bacilli are uniquely adapted to survive and grow within macrophages (reviewed in Ref. 25). Paradoxically, macrophages are the primary effector cells of the innate immune response. On binding microbial pathogens, these cells are activated to release a variety of cytokines, nitric oxide, and reactive oxygen intermediates. Cytokines released by activated macrophages can augment both innate and cell-mediated immune responses. The precise mechanisms by which Mtb activate macrophages remain unclear. Our data revealed two unexpected features of these Mtb-induced responses. The first feature is that CD14 does not appear to be required for cellular activation by Mtb. This contrasts with both Gram-positive and Gram-negative bacteria that predominantly activate cells in a CD14-dependent manner (14, 24, 26). Although CD14 has been reported to function as an adhesion receptor for Mtb bacilli in some cell types (27), blocking Abs against the complement receptors CR3, CR4, the macrophage mannose receptor, and the class A scavenger receptor can together prevent almost all binding of Mtb to macrophages (28, 29, 30). It is possible that complement receptors, like CD14, can serve as coreceptors for TLR proteins. Support for this possibility comes from the finding that the CR3 ligand taxol, a plant-derived antitumor agent, cannot activate macrophages from the TLR4-deficient C3H/HeJ mouse whereas taxol can activate macrophages from the normal C3H/FeJ mouse (31). The second novel feature of our findings is that Mtb differs in the utilization of TLR proteins to initiate intracellular signaling compared with Gram-positive bacteria (14).5 Whether these differences in TLR signaling induced by Mtb and other bacteria result in different qualitative or quantitative antimicrobial responses remains to be determined.

Like Gram-positive bacteria, the cell wall of Mtb does not contain LPS. Nevertheless, both types of bacteria can activate cells in a TLR-dependent manner via chemically distinct bacterial products that serve as ligands to mediate this activation. Previous studies have shown that LAM, LPS, and peptidoglycan all activate TLR-dependent signaling in association with CD14 (14, 19).4 In contrast, we have shown that TLR proteins can be activated by CD14-independent ligands. We found both soluble and cell-associated mycobacterial factors can activate TLR-dependent signaling in a CD14-independent manner. A heat-labile cell-associated mycobacterial factor activated cells in a TLR4-dependent manner, whereas a soluble heat-stable mycobacterial factor activated cells in a TLR2-dependent manner. This soluble factor was also resistant to protease treatment, suggesting that it is a polysaccharide or a glycolipid. In addition, we found that a heat-stable cell-associated factor could activate cells in a TLR2-dependent manner (data not shown), although it remains to be determined whether this factor is identical with the soluble heat-stable factor. Studies are currently under way to identify these factors. Interestingly, mycobacterial envelopes contain peptidoglycan, a known CD14 ligand that can activate cells in a TLR2-dependent manner (14, 19). Our finding that Mtb activates cells in a CD14-independent manner suggests that peptidoglycan is not the heat-stable factor we found that activated cells via TLR2. The extensive waxy cell wall of Mtb may mask the underlying peptidoglycan, although this possibility remains to be tested.

Pathogenic mycobacteria, such as Mtb, are part of a family of slow-growing mycobacteria that contain mannose-capped LAM (i.e., ManLAM) in their cell walls. In contrast, rapidly growing mycobacteria are nonpathogenic in immunocompetent hosts and possess arabinofuranosyl-capped LAM (i.e., AraLAM). Part of the survival strategy of Mtb may depend on bacilli gaining entry into the host macrophage without evoking a strong antimicrobial response. Our data demonstrate that Mtb can induce TLR-dependent cellular activation, but the extent of this activation may differ qualitatively from TLR signaling induced by other types of bacteria. CD14 ligands present in the cell walls of Gram-positive and Gram-negative bacilli that activate TLR signaling may induce potent antimicrobial responses that are not induced by Mtb. Our studies have only assessed signaling via TLR2 and TLR4, but how Mtb and other types of bacteria might differentially utilize other TLR proteins remain to be determined.

The Drosophila Toll and 18-wheeler proteins are receptors that mediate antifungal and antibacterial immune responses, respectively (3, 32). These findings support the possibility that mammalian TLR proteins also participate in innate immunity. Mammalian cells possess at least 10 distinct TLR genes, and it is possible that these proteins can bind a wide variety of bacterial products (Patent Cooperation Treaty Publication number WO9850547A3). Given the dimeric nature of the Drosophila Toll receptor, it is also possible that mammalian TLR proteins might heterodimerize, thereby further extending the variety of ligands that might be capable of inducing TLR signaling. Furthermore, mammalian ligands for TLR proteins may also exist. In Drosophila, the natural ligand for Toll is the Spatzle protein (33). Spatzle, a primitive member of the NGF family of cysteine-knot proteins (34), is secreted as an inactive precursor protein that is cleaved into a biologically active form by the serine protease Easter (12). Similar paradigms are well known in mammalian innate immune responses, and natural mammalian ligands for TLR proteins may yet be discovered. Our data do not exclude the possibility that Mtb, and mycobacterial factors do not bind directly to TLR proteins, but instead stimulate the production of a mammalian Spatzle-like factor that binds directly to the TLR proteins.

1

This work was supported by National Institutes of Health Grants GM54060 (to D.T.G. and A.Y.) and HL55681 (to M.J.F. and T.K.M.). E.L. was supported by the Norwegian Cancer Society and the Research Council of Norway.

3

Abbreviations used in this paper: TLR, Toll-like receptor; LAM, lipoarabinomannan; Mtb, Mycobacterium tuberculosis; CHO, Chinese hamster ovary; LBP, LPS-binding protein; BCG, bacillus Calmette-Guérin; EMSA, electrophoretic mobility shift assays.

4

T. K. Means, E. Lien, A. Yoshimura, S. Wang, D. T. Golenbock, and M. J. Fenton. 1999. The CD14 ligands lipoarabinomannan and lipopolysaccharide differ in their requirement for Toll-like receptors. Submitted for publication.

5

E. Lien, T. J. Sellati, A. Yoshimura, J. D. Carroll, R. R. Ingalls, J. D. Radolf, and D. T. Golenbock. 1999. Toll-like receptor 2 mediates pattern recognition by Borrelia burgdorferi, Treponema pallidum, and Mycobacterium avium. Submitted for publication.

1
Medzhitov, R., C. A. Janeway, Jr.
1997
. Innate immunity: the virtues of a nonclonal system of recognition.
Cell
91
:
295
2
Medzhitov, R., P. Preston-Hurlburt, C. A. Janeway, Jr.
1997
. A human homologue of the Drosophila Toll protein signals activation of adaptive immunity.
Nature
388
:
394
3
Lemaitre, B., E. Nicolas, L. Michaut, J. M. Reichhart, J. A. Hoffmann.
1996
. The dorsoventral regulatory gene cassette Spatzle/Toll/Cactus controls the potent antifungal response in Drosophila adults.
Cell
86
:
973
4
Rock, F. L., G. Hardiman, J. C. Timans, R. A. Kastelein, J. F. Bazan.
1998
. A family of human receptors structurally related to Drosophila Toll.
Proc. Natl. Acad. Sci. USA
95
:
588
5
O’Neill, L. A. J., C. Greene.
1998
. Signal transduction pathways activated by the IL-1 receptor family: ancient signaling machinery in mammals, insects, and plants.
J. Leukocyte Biol.
63
:
650
6
Takeuchi, O., T. Kawai, H. Sanjo, N. G. Copeland, D. J. Gilbert, N. A. Jenkins, K. Takeda, S. Akira.
1999
. TLR6: a novel member of an expanding Toll-like receptor family.
Gene
231
:
59
7
Fenton, M. J., D. T. Golenbock.
1998
. LPS-binding proteins and receptors.
J. Leukocyte Biol.
64
:
25
8
Yang, R. B., M. R. Mark, A. Gray, A. Huang, M. H. Xie, M. Zhang, A. Goddard, W. I. Wood, A. L. Gurney, P. J. Godowski.
1998
. Toll-like receptor-2 mediates lipopolysaccharide-induced cellular signaling.
Nature
395
:
284
9
Kirschning, C. J., H. Wesche, T. M. Ayres, M. Rothe.
1998
. Human Toll-like receptor 2 confers responsiveness to bacterial lipopolysaccharide.
J. Exp. Med.
188
:
2091
10
Poltorak, A., X. He, I. Smirnova, M.-Y. Liu, C. Van Huffel, X. Du, D. Birdwell, E. Alejos, M. Silva, C. Galanos, M. Freudenberg, P. Ricciardi-Castagnoli, B. Layton, B. Beutler.
1998
. Defective LPS signaling in C3H/HeJ and C57BL/10ScCr mice: mutations in the Tlr4 gene.
Science
282
:
2085
11
Qureshi, S. T., L. Lariviere, G. Leveque, S. Clermont, K. J. Moore, P. Gros, D. Malo.
1999
. Endotoxin-tolerant mice have mutations in Toll-like receptor 4 (Tlr4).
J. Exp. Med.
189
:
615
12
DeLotto, Y., R. DeLotto.
1998
. Proteolytic processing of the Drosophila Spatzle protein by easter generates a dimeric NGF-like molecule with ventralising activity.
Mech. Dev.
72
:
141
13
Chow, J. C., D. W. Young, D. T. Golenbock, W. J. Christ, Fabian Gusovsky.
1999
. Toll-like receptor-4 mediates lipopolysaccharide-induced signal transduction.
J. Biol. Chem.
274
:
10689
14
Yoshimura, A., E. Lien, E. Toumanen, R. Dziarski, D. T. Golenbock.
1999
. Recognition of Gram-positive bacterial cell wall components by the innate immune system occurs via Toll-like receptor 2.
J. Immunol.
163
:
1
15
Delude, R. L., A. Yoshimura, R. R. Ingalls, D. T. Golenbock.
1998
. Construction of a lipopolysaccharide reporter cell line and its use in identifying mutants defective in endotoxin, but not TNF-α, signal transduction.
J. Immunol.
161
:
3001
16
Prinzis, S., D. Chatterjee, P. J. Brennan.
1993
. Structure and antigenicity of lipoarabinomannan from Mycobacterium bovis BCG.
J. Gen. Microbiol.
139
:
2649
17
Schreiber, E., P. Matthias, M. M. Muller, W. Schaffner.
1989
. Rapid detection of octamer binding proteins with “mini-extracts,” prepared from a small number of cells.
Nucleic Acids Res.
17
:
6419
18
Hunninghake, G. W., B. G. Monks, L. J. Geist, M. M. Monick, M. A. Monroy, M. F. Stinski, A. C. Webb, J. M. Dayer, P. E. Auron, M. J. Fenton.
1992
. The functional importance of a cap site-proximal region of the human prointerleukin 1β gene is defined by viral trans-activation.
Mol. Cell. Biol.
12
:
3439
19
Schwandner, R., R. Dziarski, H. Wesche, M. Rothe, C. J. Kirschning.
1999
. Peptidoglycan- and lipoteichoic acid-induced cell activation is mediated by toll-like receptor 2.
J. Biol. Chem.
274
:
17406
20
Savedra, R., R. J. Delude, R. R. Ingalls, M. J. Fenton, D. T. Golenbock.
1996
. Mycobacterial lipoarabinomannan recognition requires a receptor that shares components of the endotoxin signaling system.
J. Immunol.
157
:
2549
21
Strohmeiher, G., M. J. Fenton.
1999
. Role of lipoarabinomannan in mycobacterial pathogenesis and immunity.
Microbes Inf.
2
:
1
22
Chatterjee, D., A. D. Roberts, K. Lowell, P. J. Brennan, I. M. Orme.
1992
. Structural basis of capacity of lipoarabinomannan to induce secretion of tumor necrosis factor.
Infect. Immun.
60
:
1249
23
Roach, T. I., C. H. Barton, D. Chatterjee, J. M. Blackwell.
1993
. Macrophage activation: lipoarabinomannan from avirulent and virulent strains of Mycobacterium tuberculosis differentially induces the early genes c-fos, KC, JE, and tumor necrosis factor-α.
J. Immunol.
150
:
1886
24
Zhang, Y., M. Broser, H. Cohen, M. Bodkin, K. Law, J. Reibman, W. N. Rom.
1995
. Enhanced interleukin-8 release and gene expression in macrophages after exposure to Mycobacterium tuberculosis and its components.
J. Clin. Invest.
95
:
586
25
Fenton, M. J..
1998
. Macrophages and tuberculosis.
Curr. Opin. Hematol.
5
:
72
26
Haziot, A., E. Ferrero, F. Kontgen, N. Hijiya, S. Yamamto, J. Silver, C. L. Stewart, S. M. Goyert.
1996
. Resistance to endotoxin shock and reduced dissemination of Gram-negative bacteria in CD14-deficient mice.
Immunity
4
:
407
27
Peterson, P. K., G. Gekker, S. Hu, W. S. Sheng, W. R. Anderson, R. J. Ulevitch, P. S. Tobias, K. V. Gustafson, T. W. Molitor, C. C. Chao.
1995
. CD14 receptor-mediated uptake of nonopsonized Mycobacterium tuberculosis by human microglia.
Infect. Immun.
63
:
1598
28
Stokes, R. W., I. D. Haidl, W. A. Jefferies, D. P. Speert.
1993
. Mycobacteria-macrophage interactions: macrophage phenotype determines the nonopsonic binding of Mycobacterium tuberculosis to murine macrophages.
J. Immunol.
151
:
7067
29
Schlesinger, L. S..
1993
. Macrophage phagocytosis of virulent but not attenuated strains of Mycobacterium tuberculosis is mediated by mannose receptors in addition to complement receptors.
J. Immunol.
150
:
2920
30
Zimmerli, S., S. Edwards, J. D. Ernst.
1996
. Selective receptor blockade during phagocytosis does not alter the survival and growth of Mycobacterium tuberculosis in human macrophages.
Am. J. Respir. Cell Mol. Biol.
15
:
760
31
Bhat, N., P. Perera, J. M. Carboni, J. Blanco, D. T. Golenbock, T. N. Mayadas, S. N. Vogel.
1999
. Use of a photoactivatable taxol analogue to identify unique cellular targets in murine macrophages: identification of murine CD18 as a taxol-binding protein and a role for Mac-1 in taxol-induced gene expression.
J. Immunol.
162
:
7335
32
Williams, M. J., A. Rodriguez, D. A. Kimbrell, E. D. Eldon.
1997
. The 18-wheeler mutation reveals complex antibacterial gene regulation in Drosophila host defense.
EMBO J.
16
:
6120
33
Morisato, D., K.V. Anderson.
1994
. The spatzle gene encodes a component of the extracellular signaling pathway establishing the dorsal-ventral pattern of the Drosophila embryo.
Cell
76
:
677
34
Mizuguchi, K., J. S. Parker, T. L. Blundell, N. J. Gay.
1998
. Getting knotted: a model for the structure and activation of Spatzle.
Trends Biochem. Sci.
23
:
239