Mycobacterium tuberculosis overcomes macrophage bactericidal activities and persists intracellularly. One mechanism by which M. tuberculosis avoids macrophage killing might be through inhibition of IFN-γ-mediated signaling. In this study we provide evidence that at least two distinct components of M. tuberculosis, the 19-kDa lipoprotein and cell wall peptidoglycan (contained in the mycolylarabinogalactan peptidoglycan (mAGP) complex), inhibit macrophage responses to IFN-γ at a transcriptional level. Moreover, these components engage distinct proximal signaling pathways to inhibit responses to IFN-γ: the 19-kDa lipoprotein inhibits IFN-γ signaling in a Toll-like receptor (TLR)2-dependent and myeloid differentiation factor 88-dependent fashion whereas mAGP inhibits independently of TLR2, TLR4, and myeloid differentiation factor 88. In addition to inhibiting the induction of specific IFN-γ responsive genes, the 19-kDa lipoprotein and mAGP inhibit the ability of IFN-γ to activate murine macrophages to kill virulent M. tuberculosis without inhibiting production of NO. These results imply that inhibition of macrophage responses to IFN-γ may contribute to the inability of an apparently effective immune response to eradicate M. tuberculosis.
One third of the world’s population is infected with Mycobacterium tuberculosis. In most people, M. tuberculosis persists as a latent infection in which the mycobacterium is contained but not eradicated by the host immune response. Many critical elements of the immune response to M. tuberculosis have been defined through the study of human disease and in murine models. IFN-γ, an intact CD4 T cell response, IL-12, and TNF-α are required for control of mycobacterial infection in humans and in mice (reviewed in Ref. 1). It remains unclear, however, why the immune response controls M. tuberculosis but does not eradicate infection.
IFN-γ is a critical macrophage activator, inducing the transcription of more than 200 genes of which the functional consequences include the up-regulation of MHC class II and costimulatory molecule expression and the production of antimicrobial effectors such as oxygen radicals and NO (2). IFN-γ is found at the site of M. tuberculosis infection and even within the granuloma (3, 4). Nonetheless, M. tuberculosis persists within macrophages and neither humans nor mice infected with M. tuberculosis are able to completely clear the infection (5).
Several mechanisms have been described that may allow M. tuberculosis to evade host immune responses (reviewed in Ref. 6), potentially contributing to its persistence in the infected host. M. tuberculosis attenuates the response of infected macrophages to IFN-γ (7, 8, 9, 10) in part by blocking the transcription of a subset of IFN-γ responsive genes including the type I receptor for the Fc domain of IgG (FcγRI or CD64) (11) and the MHC class II transactivator (CIITA),3 which regulates MHC class II expression (12). The bacterial components responsible for this phenomenon are not well defined. Live bacteria are not necessary for macrophage inhibition as gamma-irradiated M. tuberculosis also inhibit IFN-γ responsiveness in macrophages (9). Although unfractionated cell wall also inhibits IFN-γ signaling (9), mannosylated lipoarabinomannan, a major component of the M. tuberculosis cell wall with previously described immunomodulatory activity (6), is not responsible for the effect (9). In contrast, previous work has shown that the 19-kDa lipoprotein of M. tuberculosis inhibits IFN-γ induction of CIITA mRNA, surface MHC class II, and class II-dependent Ag presentation in a Toll-like receptor (TLR)2-dependent manner (10, 13, 14). However, it is not clear that the 19-kDa lipoprotein is the only component of M. tuberculosis with this capacity.
In this study, we show that multiple mycobacterial cell wall components including the lipoproteins and the peptidoglycan fraction of the mycolylarabinogalactan peptidoglycan (mAGP) cell wall core inhibit macrophage responses to IFN-γ. Furthermore, we provide evidence that these components act through different signaling pathways to inhibit macrophage responses to IFN-γ: specifically, the 19-kDa lipoprotein inhibits macrophage IFN-γ responses in a TLR2-dependent and myeloid differentiation factor 88 (MyD88)-dependent fashion, whereas mAGP inhibits macrophage IFN-γ responsiveness in a TLR2-, TLR4-, and MyD88-independent manner. Finally, we demonstrate that inhibition of the IFN-γ signaling pathway may contribute to persistence of mycobacterial infection by attenuating the ability of murine macrophages to kill M. tuberculosis after stimulation with IFN-γ.
Materials and Methods
M. tuberculosis and strain fractions
M. tuberculosis strain H37Rv was cultured to late log-phase in glycerol-alanine-salts medium, and the cells were harvested and inactivated by gamma irradiation (15). Working stocks of H37Rv were prepared as previously described (11). The 19-kDa lipoprotein was purified from the TX-114 extract (lipoprotein pool) of M. tuberculosis H37Rv (16) by preparative SDS-PAGE and whole gel electroelution (15). M. tuberculosis soluble cell wall proteins (SCWP) were obtained from inactivated H37Rv by SDS extraction (17). mAGP and its constituents were purified as previously described (18). The crude cell wall, SCWP, and lipoprotein pool were quantitated by protein concentration estimated by protein assay (Bio-Rad, Richmond, CA) against BSA standards, and mAGP and peptidoglycan were quantitated by dry weight. Stocks of mAGP and peptidoglycan were resuspended at concentration of 5 mg/ml in DMSO, sonicated, diluted to 1 mg/ml in PBS and stored at −80°C until use. By Limulus amoebocyte assay (Cambrex, Walkersville, MD), a 5 μg/ml solution of mAGP contained <1.25 pg/ml of contaminating endotoxin. The 19-kDa lipopeptide was previously described (19) and was a kind gift of R. Modlin (Division of Dermatology, Department of Microbiology and Immunology and Molecular Biology Institute, University of California, Los Angeles, School of Medicine, Los Angeles, CA).
Flow cytometry analysis of human monocytes and murine macrophages
Human PBMC were isolated and treated with M. tuberculosis or M. tuberculosis components as previously detailed (9) with minor modifications: on day 6 after isolation, cells were incubated with killed bacteria or components for 4 h followed by washing, subsequent IFN-γ stimulation, and analysis as described. RAW 264.7 murine macrophage cells were plated at a density of 2 × 105 cells/well of a six-well tissue culture-treated plate and allowed to adhere overnight. Monolayers were then washed, treated with killed bacteria or bacterial components at the indicated concentrations for 24 h, and then stimulated with IFN-γ at 100 U/ml (BD Biosciences, Cambridge, MA and Roche, Indianapolis, IN) for 24 h.
For flow cytometric analysis for FcγRI (CD64) expression in human macrophages or MHC class II expression in RAW 264.7 macrophages, the treated monolayers were processed as previously described (9) except that after Ab staining, cells were analyzed immediately in the presence of 1 μg/ml propidium iodide. Anti-mouse MHC class II (I-A/I-E)-PE mAb was obtained from BD Biosciences. Five thousand live cells were analyzed for expression on a FACSort flow cytometer, and data were collected using CellQuest software (BD Biosciences) and expressed as mean channel fluorescence for each sample.
To generate the CIITA reporter cell line, 5 × 106 RAW 264.7 cells were cotransfected with 10 μg of CIITA-luciferase construct (a kind gift of Drs. G. O’Keefe and E. Benveniste, Department of Cell Biology, University of Alabama, Birmingham, AL) and pCDNA 3.1 (Invitrogen, Walkersville, MD), which contains the neomycin resistance gene, using the Superfectamine kit (Qiagen, Valencia, CA) according to the manufacturer’s instructions. One day after transfection, cells were placed in selective medium containing G418 (Invitrogen) at 600 μg/ml. After 20 days, the surviving transfectants were cloned by limiting dilutions. Each clone was individually screened for induction of luciferase by IFN-γ. Clone E2 was selected based on its relatively low background and a 2- to 3-fold induction of luciferase activity with IFN-γ stimulation and hereafter used as RAW-CIITA-lux.
For use in luciferase assays, transformants were plated in a 96-well tissue culture-treated white plate/clear bottom assay plate (Corning, Corning, NY) at a density of 5 × 104 cells/well. Macrophages were treated with M. tuberculosis, mycobacterial components, or LPS (Sigma-Aldrich, St. Louis, MO) and then treated with IFN-γ for 24 h as previously described. Luciferase assays were performed with the Luciferase Assay System from Promega (Madison, WI) according to the manufacturer’s instructions. Luciferase activity was read in a HTS 7000 Plus-Bioassay Reader (PerkinElmer, Norwalk, CT).
Mice and bone marrow-derived macrophages
C57/B6, C3H/HeJ, and C3H/HeOuJ female mice were obtained from The Jackson Laboratory (Bar Harbor, ME) and sacrificed between 8 and 12 wk of age. TLR2−/− and MyD88−/− mice were a kind gift from D. Golenbock (Division of Infectious Diseases and Immunology, University of Massachusetts Medical School, Worcester, MA). Mice were housed under specific pathogen-free conditions. Macrophages were derived from bone marrow precursors that were harvested from femur and tibia marrow, which were differentiated for 6–7 days in DMEM/F12 tissue culture medium (Invitrogen) supplemented with 10% FBS, 10% L929 cell-conditioned medium, and 1% penicillin/streptomycin (DMEM/F12/FBS/LCS/PS). For the first 2 days of differentiation, the medium was further supplemented with 100 ng/ml murine IL-3 (PeproTech, Rocky Hills, NJ). Macrophages to be infected with live M. tuberculosis were kept in medium containing penicillin/streptomycin through day 5 of differentiation and subsequently maintained in medium free of antibiotics. After differentiation, cells were plated at a density of 2–4 × 105 cells/well of six-well tissue culture treated plates and allowed to adhere overnight. Monolayers were then washed and treated with bacterial products and IFN-γ as previously described. After 24 h from the addition of IFN-γ, monolayers were washed and processed for harvest of RNA or infection with live M. tuberculosis.
RNA harvest and quantitative real-time PCR
RNA was harvested using the RNeasy kit from (Qiagen) according to the manufacturer’s instructions for preparation of RNA from adherent mammalian cells except that the time of the DNase digest was increased to 20 min, which was required for complete digestion of the genomic DNA. RNA yield was determined by taking the OD260 using a Beckman DU520 spectrophotometer. A total of 0.5 μg of RNA was used in a 40 μl reverse transcription reaction using the Omniscript RT kit (Qiagen) according to the manufacturer’s instructions using oligo-dT (Ambion, Austin, TX) at a final concentration of 5 μM and RNase inhibitor (Promega) at a final concentration of 0.5 U/μl. Reverse transcription was conducted at 37°C for 2 h followed by heat inactivation at 95°C for 5 min.
The cDNA was then diluted 1/2, and 5 μl were used as template in a quantitative real-time PCR. Duplex PCR contained primers and probe for both β-actin and CIITA or TNF-α. The CIITA and TNF-α probes were labeled with TET and BlackHole Quencher 1, and the β-actin probe was labeled with HEX and BlackHole Quencher 1 (IDT, Coralville, IA). Primers and probes used are as follows: CIITA: forward 5′-GAA GTT CAC CAT TGA GCC ATT TAA-3′, reverse 5′-CTG GGT CTG CAG GAG ACG AT-3, probe 5′-CCA AAT CCC CAA AGG ATG TGG AAG ACC-3′; TNF-α: forward 5′-AAA ATT CGA GTG ACA AGC CTG TAG-3′, reverse 5′-CCC TTG AAG AGA ACC TGG GAG TAG-3′, probe 5′-CAC GTC GTA GCA AAC CAC CAA GTG GA-3′; β-actin: forward 5′-ACA GTG TGG GTG ACC CCG TC-3′, reverse 5′-CCC TGT GCT GCT CAC CGA-3′, probe 5′-CCC CTG AAC CCT AAG GCC AAC CG-3′.
RT-PCR was performed on a DNA Engine Opticon (MJ Research, Waltham, MA). Data were analyzed using the Opticon Monitor 2.01 operating software. Cycle thresholds were set manually and quantities were calculated by the operating software by applying the sample’s threshold cycle to a standard curve of log quantity vs cycle threshold cycle. Standards were made with gel-purified PCR products and standard curves (all R > 0.995) were run with every set of reactions. Expected product size was confirmed by visualization on an agarose gel. Specificity of the product was further ensured with no-reverse transcriptase and no-template controls.
For every PCR, the quantity of CIITA or TNF-α was normalized to the quantity of β-actin in the same well. Real-time PCR was performed in triplicate and their normalized values were averaged to give the value for each biologic replicate. To calculate the percentage of maximal increase, we divided the value of each biologic replicate by the average of all of the IFN-γ stimulated replicates. These ratios were averaged and SD determined.
To prepare M. tuberculosis stock culture, strain H37Rv was grown as described (20). In mid-log phase, aliquots were frozen and the stock’s titer was determined by plating a representative aliquot for CFU. Before use, aliquots were thawed and cultured overnight in 7H9-Tween-OADC with shaking. Mycobacteria were then resuspended in tissue culture medium, sonicated, then diluted for addition to the macrophage monolayers at a multiplicity of infection (MOI) of 5. Four hours after infection, monolayers were pulsed for 45 min with amikacin (Sigma-Aldrich) at 200 μg/ml to kill extracellular mycobacteria. Monolayers were then washed three times with PBS-1% FCS and cultured in fresh medium until the time of harvest. At the time of harvest, monolayers were again washed three times and then lysed in PBS-1% Triton (Sigma-Aldrich). Lysates were serially diluted in PBS-0.5% Tween 80 and plated onto 7H10 agar (Difco, Detroit, MI) supplemented with Tween and OADC. Colonies were counted 21–28 days after plating.
Bone marrow-derived macrophages were plated and treated with whole mycobacteria or mycobacterial components as previously described. At 24 h later, culture supernatants were collected and stored at −20°C until analysis. Concentrations of murine TNF-α and IL-6 were determined by ELISA following the manufacturer‘s instructions (R&D Systems, Minneapolis, MN). The data were recorded and analyzed using the SOFTmax version 2.0 software (Molecular Devices, Menlo Park, CA). The detection limits of the assays were 5.1 pg/ml for TNF-α and 16 pg/ml for IL-6.
Multiple components of M. tuberculosis inhibit macrophage responses to IFN-γ
We have previously found that infection of human macrophages with live M. tuberculosis inhibits the IFN-γ-inducible expression of CD64 (9) To better characterize the M. tuberculosis components responsible for this inhibition, we fractionated bacteria and tested each fraction for its ability to inhibit IFN-γ induction of surface CD64 expression on human monocyte-derived macrophages. Subcellular fractions of M. tuberculosis H37Rv were generated including a cell wall core composed of the mAGP, SDS-extracted SCWP, and a Triton X-114-extracted lipoprotein pool (21). Flow cytometric analysis demonstrated that human macrophages expressed CD64 constitutively and that this expression increased ∼3-fold in response to IFN-γ (Fig. 1,A). As previously observed for live M. tuberculosis (at an infecting MOI of 5), pretreatment of the macrophages with gamma-irradiated M. tuberculosis and crude M. tuberculosis cell wall inhibited IFN-γ induction of cell surface CD64 (Fig. 1,A). M. tuberculosis mAGP, SCWP, and the lipoprotein pool also inhibited IFN-γ induction of CD64 in a dose-dependent manner (Fig. 1 A). In contrast, M. tuberculosis membranes, total lipids, and culture filtrate exhibited little, if any, inhibitory activity (data not shown).
Whole M. tuberculosis inhibits responses to IFN-γ in murine bone marrow-derived macrophages and the murine macrophage cell line, RAW 264.7 (22), as well as in human primary macrophages. We therefore assessed the ability of the mycobacterial components to inhibit the response to IFN-γ in RAW 264.7 cells as assessed by MHC class II surface expression. Treatment of RAW 264.7 cells with IFN-γ markedly induced cell surface expression of MHC class II molecules. As we observed in human macrophages, M. tuberculosis mAGP, SCWP, and lipoprotein pool each resulted in a dose-dependent inhibition of the MHC class II induction (Fig. 1 B). Moreover, as we have previously observed for live M. tuberculosis in human macrophages (9), inhibition of the IFN-γ response induced by the M. tuberculosis subcellular fractions was exerted at a transcriptional level and did not involve inhibition of tyrosine phosphorylation of the transcription factor STAT1 in either human or murine macrophages (data not shown).
Mycobacterial peptidoglycan inhibits IFN-γ-induced expression of CD64
To determine the identity of the active component(s) in the mAGP complex, this tripartite structure was separated into its three primary constituents, mycolic acids, arabinogalactan, and peptidoglycan. We assessed the ability of each of these fractions to inhibit the response to IFN-γ in human macrophages as assessed by surface expression of CD64. Again IFN-γ induced a roughly 3-fold increase in CD64 expression that was inhibited by M. tuberculosis. Neither mycolic acids nor arabinogalactan (up to 50 μg/ml) inhibited the macrophage response to IFN-γ (Fig. 2,A). In contrast, purified M. tuberculosis peptidoglycan strongly inhibited the IFN-γ-induced up-regulation of CD64 expression in human macrophages (Fig. 2 B) and MHC class II expression in murine macrophages (data not shown). Because peptidoglycan was the only component of the mAGP complex that exhibited activity, and preparation of the complex in large quantities is significantly simpler than the preparation of peptidoglycan, subsequent experiments were performed with mAGP as the source of peptidoglycan.
The 19-kDa lipoprotein and triacylated 19-kDa lipopeptide inhibit IFN-γ-induced transcription from the CIITA class IV promoter
The detergent soluble M. tuberculosis fractions, the lipoprotein pool and soluble cell wall proteins, contain a complex mixture of compounds. However, both preparations of detergent-extracted proteins contained the 19-kDa lipoprotein as determined by immunoblotting (data not shown), which has been found to inhibit macrophage surface expression of MHC class II (13, 14), at least partly by inhibiting IFN-γ induction of CIITA expression in a TLR2-dependent fashion (10). To confirm these findings, we tested the ability of the 19-kDa lipoprotein and a synthetic lipopeptide derived from 19-kDa lipoprotein, which acts as a TLR2 agonist only when triacylated (23), to inhibit IFN-γ-induced gene transcription. IFN-γ-induced gene transcription was assessed in RAW 264.7 cells stably transfected with a reporter construct containing −231 to +83 of the murine CIITA type IV promoter fused to the firefly luciferase gene (CIITA-lux). IFN-γ treatment increased transcription from the CIITA promoter roughly 2-fold over baseline. Pretreatment with the 19-kDa lipoprotein inhibited IFN-γ induced transcription from the CIITA promoter in a dose-dependent manner (Fig. 3,A). As expected, pretreatment of the CIITA-lux reporter cell line with the triacylated 19-kDa lipopeptide inhibited IFN-γ-induced transcription from the CIITA promoter equivalently to the 19-kDa lipoprotein whereas the nonacylated form, which does not activate TLR2, did not (Fig. 3 B).
Mycobacterial components inhibit the macrophage IFN-γ response through TLR2-dependent and -independent pathways
Similar to the 19-kDa lipoprotein, mAGP is a TLR2 agonist (24) and our preparation of mAGP activated NF-κB in CHO cells transfected with TLR2 and CD14 (data not shown). We therefore hypothesized that multiple mycobacterial components including the 19-kDa lipoprotein and mAGP signal through a common, TLR-mediated pathway, to inhibit the subsequent macrophage response to IFN-γ. To test this hypothesis, we compared the ability of the 19-kDa lipoprotein and mAGP to inhibit IFN-γ signaling in bone marrow-derived macrophages from C57/B6 and TLR2−/− mice. The macrophage IFN-γ response was assessed by measuring the level of endogenous CIITA mRNA transcripts by quantitative RT-PCR. In both wild-type and TLR2−/− macrophages, treatment with IFN-γ up-regulated the level of CIITA mRNA transcripts by 10- to 15-fold. Pretreatment with the 19-kDa lipoprotein inhibited up-regulation of CIITA transcripts by 60% in wild-type macrophages but had no effect in TLR2−/− macrophages (Fig. 4,A). Pretreatment of wild-type macrophages with mAGP also inhibited IFN-γ induction of CIITA mRNA. Surprisingly, however, mAGP inhibited IFN-γ-induced CIITA mRNA equally well in TLR2−/− macrophages and wild-type macrophages (Fig. 4 A). Thus, although mAGP activates TLR2, it can inhibit IFN-γ-induced CIITA up-regulation in a TLR2-independent manner.
Several TLR agonists have been shown to inhibit macrophage MHC class II surface expression including LPS (Refs. 13 , 25 and data not shown), which activates TLR4. Although the inhibitory activity of the crude mycobacterial fractions described in Fig. 1 was not affected by the presence of polymyxin B (20 μg/ml; data not shown) and although we did not find biologically significant LPS contamination in our mAGP preparation by Limulus assay (as described in Materials and Methods), an unidentified heat-labile TLR4 agonist has been detected in whole M. tuberculosis (26) raising the possibility that the cell wall core might stimulate TLR4. To test this possibility we used bone marrow-derived macrophages from C3H/HeJ mice that have a mutation in TLR4 rendering it unresponsive to LPS (27). In C3H/HeJ macrophages, mAGP markedly inhibited induction of CIITA transcripts whereas LPS did not (Fig. 4,B). Interpretation of these results is complicated by the consistently poor induction of CIITA transcript by IFN-γ in the control C3H/OuJ (Fig. 4) and C3H/FeJ (data not shown) macrophages. Although interesting in light of data on differences between C3H substrains (28, 29), this discrepancy precludes direct comparison of inhibition of IFN-γ signaling in HeJ and OuJ macrophages. We cannot therefore formally rule out the possibility that mAGP inhibits IFN-γ signaling in part through TLR4 but our data indicate that mAGP does not inhibit IFN-γ signaling solely or even largely in a TLR4-dependent manner.
Mycobacterial components inhibit macrophage response to IFN-γ through MyD88-dependent and -independent pathways
MyD88 is a critical adapter protein in the TLR signaling pathway, necessary for TLR-mediated cytokine production (30). Signaling by all TLRs occurs, at least in part, through MyD88 although MyD88-independent signaling by TLR3 and TLR4 has also been described (30). To assess the role of MyD88 in inhibition of IFN-γ signaling by mycobacterial components, we used bone marrow macrophages derived from wild-type and MyD88−/− mice. In both, IFN-γ up-regulated CIITA mRNA roughly 10-fold. As expected, pretreatment of wild-type macrophages with the 19-kDa lipoprotein inhibited the induction of CIITA transcript by IFN-γ but the 19-kDa lipoprotein had no effect on IFN-γ signaling in MyD88−/− macrophages (Fig. 4,C). Thus, 19-kDa inhibits the macrophage response to IFN-γ in a TLR2- and MyD88-dependent fashion. In contrast, pretreatment of macrophages with mAGP inhibited the subsequent response to IFN-γ equally in wild-type and MyD88−/− macrophages (Fig. 4 C). This result is consistent with the evidence previously presented that neither TLR2 nor TLR4 are essential for mAGP inhibition of macrophage responses to IFN-γ. In addition, these data clearly establish that the effects of the 19-kDa lipoprotein and mAGP/peptidoglycan are not mediated by a common component.
mAGP stimulates TNF-α production in a MyD88-independent manner
Because activation of TLR-mediated signaling by the 19-kDa lipoprotein leads to both macrophage activation and inhibition of the IFN-γ response, we hypothesized that the TLR-independent pathway triggered by mAGP also results in macrophage activation. Indeed, we found that both the 19-kDa lipoprotein and mAGP strongly induced TNF-α mRNA in wild-type macrophages (Fig. 5,A). Induction of TNF-α by the 19-kDa lipoprotein was completely abrogated in macrophages deficient in MyD88 (Fig. 5,A), which is required for TLR2, 3, 4, 5, 7 and 9-mediated cytokine production (30). In contrast, mAGP robustly stimulated TNF-α mRNA in both wild-type and MyD88−/− macrophages (Fig. 5,A). To confirm these results at a protein level, we measured mAGP-induced TNF-α production by ELISA in wild-type and MyD88−/− macrophages in the absence of IFN-γ. mAGP strongly stimulated TNF-α production in both wild-type and MyD88−/− macrophages (Fig. 5 B), although the lower baseline TNF-α production by the MyD88−/− macrophages, consistent with the altered resting cytokine expression profile in MyD88−/− macrophages reported by Shi et al. (31), makes it difficult to determine whether mAGP induction of TNF-α production is partially attenuated in MyD88−/− macrophages.
Macrophage inhibition by 19-kDa and mAGP is not mediated by IL-6
We have found that M. tuberculosis infected macrophages produce IL-6, which in some contexts, contributes to the inhibition of their subsequent response to IFN-γ (22). Moreover, TLR2 activation may stimulate macrophages to produce IL-6 (32) although TLR4 activation is a more potent stimulus of IL-6 production (33). Therefore, we tested the ability of the 19-kDa lipopeptide and mAGP to induce IL-6 expression in our system. Treatment of bone marrow-derived macrophages with whole, irradiated M. tuberculosis induced significant IL-6 production (Table I) at a concentration (1000 pg/ml) sufficient to partially inhibit macrophage responsiveness when exogenously added to macrophage cultures (22). However, neither the 19-kDa lipopeptide nor mAGP induced biologically significant IL-6 production (<35 pg/ml) (Table I). Thus, the inhibition of IFN-γ-mediated signaling by these components cannot be accounted for by the production of IL-6. This is consistent with our recent observation that whole M. tuberculosis also inhibits responses to IFN-γ in macrophages derived from IL-6−/− mice (22).
|Stimulusa .||IL-6 Production (pg/ml)b .|
|Unstimulated||5.0 ± 0.7|
|M. tuberculosis||1049.0 ± 352.0|
|19-kDa lipopeptide||9.3 ± 3.1|
|mAGP||32.0 ± 2.6|
|Stimulusa .||IL-6 Production (pg/ml)b .|
|Unstimulated||5.0 ± 0.7|
|M. tuberculosis||1049.0 ± 352.0|
|19-kDa lipopeptide||9.3 ± 3.1|
|mAGP||32.0 ± 2.6|
C57/B6 bone marrow-derived macrophages were treated with the 19 kDa-derived lipopeptide (5 μg/ml), mAGP (5 μg/ml), irradiated M. tuberculosis, or left untreated for 24 h.
Culture supernatants were collected and analyzed for IL-6 by ELISA. Each condition was performed at least in triplicate, and the value is expressed ± SD of the replicates. The data are representative of two independent experiments.
Modulation of macrophage function
Murine macrophages activated with IFN-γ before infection are able to kill infecting M. tuberculosis. We reasoned that an important functional effect of the inhibition of IFN-γ signaling could, therefore, be to abrogate the ability of the infected macrophage to respond to IFN-γ by killing the infecting mycobacterium. To assess this possibility, murine bone marrow-derived macrophages were pretreated with the 19-kDa-derived lipopeptide or mAGP and then with IFN-γ. Macrophages were then infected with virulent M. tuberculosis, and the number of intracellular mycobacteria at the time of infection and at various time points thereafter was determined.
Macrophages treated with either the 19-kDa-derived lipopeptide or mAGP before activation with IFN-γ were less able to control intracellular replication of bacteria than IFN-γ-activated macrophages. Unactivated macrophages were not able to control mycobacterial growth resulting, at 65 h postinfection, in a roughly 1.5- to 3.5-fold increase in CFU relative to the infecting inoculum (Fig. 6,A) and at 90 h, in a significant loss of the macrophage monolayer. In contrast, at 65 h IFN-γ-activated macrophages killed ∼50% of the infecting inoculum (Fig. 6,A). Treatment of the macrophages with either the 19-kDa-derived lipopeptide or mAGP before activation with IFN-γ abrogated the ability of the macrophage to control mycobacterial growth (Fig. 6). Exposure of the macrophages to higher doses of the 19-kDa-derived lipopeptide or mAGP rendered them progressively less able to control mycobacterial growth, although the maximal effect of mAGP exceeded that of the 19-kDa-derived lipopeptide (Fig. 6 B).
The 19-kDa lipopeptide and mAGP do not alter IFN-γ-induced NO production
IFN-γ induces murine macrophages to produce NO that has been found to be a critical mediator of their antimycobacterial effect (34, 35, 36, 37). We therefore assessed whether the 19-kDa lipopeptide or mAGP inhibited the ability of IFN-γ to induce NO production. Macrophage NO production was measured after treatment with the 19-kDa lipopeptide or mAGP, after treated or untreated macrophages were activated with IFN-γ, and during the course of infection with M. tuberculosis.
Before activation with IFN-γ, macrophages in all groups produced little NO (Fig. 7). There was a roughly 20-fold induction of NO production with IFN-γ alone, which was not significantly altered by either the 19-kDa lipopeptide or mAGP (Fig. 7). After infection with M. tuberculosis, there was no inhibition of NO production by the 19-kDa lipopeptide. There was slight, non-dose responsive inhibition of NO production by mAGP (Fig. 7). Thus, 19-kDa does not inhibit NO production under any of the circumstances we examined and mAGP does not inhibit NO production to an extent that can account for its inhibition of mycobacterial killing, suggesting that both 19-kDa and mAGP modulate the other antimicrobial effects of IFN-γ, which contribute to the ability of macrophages to kill M. tuberculosis.
We have previously shown that exposure of macrophages to live M. tuberculosis inhibits their subsequent responsiveness to IFN-γ (9) by down-modulating the transcription of a subset of IFN-γ responsive genes that includes CIITA and CD64 (11). In this study we find that multiple mycobacterial components are capable of inhibiting macrophage IFN-γ responsiveness. Both the peptidoglycan fraction of the mAGP complex and the 19-kDa lipoprotein, a significant component of the soluble cell wall protein and lipoprotein pool, inhibit transcription of IFN-γ-induced CIITA and CD64 expression, recapitulating the inhibitory phenotype of whole mycobacteria.
The 19-kDa lipoprotein is one of many well-described TLR agonists that signals through TLR2 and MyD88 to stimulate a canonical proinflammatory response (30). Our data show that the 19-kDa lipoprotein also inhibits IFN-γ-induced CIITA transcription through TLR2 and MyD88, consistent with recently published data showing that the 19-kDa lipoprotein inhibits IFN-γ-regulated gene expression in human and macrophages in a TLR2-dependent fashion (10, 14). In addition, we find that a minimal TLR2 agonist, a triacylated hexapeptide derived from the 19-kDa lipoprotein, inhibits IFN-γ signaling. Because other mycobacterial lipoproteins are triacylated and some have been shown to activate macrophages like the 19-kDa lipoprotein (17), it is likely that other mycobacterial lipoproteins similarly inhibit IFN-γ signaling.
Although our data and that of others indicate that TLR2 activation is sufficient for inhibition of IFN-γ signaling, it does not appear to be the only mechanism used by M. tuberculosis for inhibition of IFN-γ signaling. Mycobacterial mAGP is a TLR2 agonist in our hands as well as others (24), but mAGP inhibits the IFN-γ response in TLR2−/−, C3H/HeJ, and MyD88−/− macrophages, indicating that mAGP also acts through an alternate signaling pathway. In addition, we have shown that mAGP stimulates TNF-α production independently of MyD88, which is required for TLR-mediated cytokine production (30). We hypothesize that mAGP activates innate immune receptor(s) other than the currently identified TLRs, and that one consequence is inhibition of the IFN-γ signaling cascade. Recently, MyD88-independent regulation of macrophage gene expression by live M. tuberculosis has been described (31). Our results indicate that mAGP may mediate this MyD88-independent macrophage activation with significant functional consequences. The innate immune receptor engaged by mAGP remains to be determined but candidates include a growing number of host innate immune receptors beyond TLR2 that recognize bacterial peptidoglycan including the peptidoglycan recognition proteins (38, 39), triggering-receptor expressed by myeloid cells (TREM)-2 (40), and members of the nucleotide-binding oligomerization domain (NOD) family of proteins that recognize specific peptidoglycan fragments in the cytosol of eukaryotic cells (41, 42).
Mycobacteria shed cell wall components both in culture in which the 19-kDa lipoprotein (43) and peptidoglycan (44) have been identified, and into the phagolysosome of the infected macrophage (45). Although the mechanism by which live mycobacteria engage innate immune receptors and concomitantly inhibit IFN-γ signaling remains unclear, the finding that TLR2 localizes to the phagosome after pathogen uptake (46) suggests that TLR2 may be triggered there by shed mycobacterial components. Moreover, because the mycobacterial vacuole is permeable to at least some macromolecules (47), cytoplasmic innate immune receptors such as the NOD family members may also have access to mycobacterial components shed in the phagosome.
The way in which activation of the innate immune response is associated with inhibition of IFN-γ signaling is also unclear with other work having excluded inhibition of STAT1 activation or function (10, 11). TLR2 agonists either alone (32) or in conjunction with TLR4 activation (33) stimulate macrophages to produce IL-6 and, in some circumstances, IL-6 production may contribute to inhibition of IFN-γ signaling (22). However, we found that neither mAGP nor the 19-kDa hexapeptide induced significant IL-6 production, indicating the existence of an IL-6-independent mechanism of inhibition.
IFN-γ is required by murine macrophages to kill infecting M. tuberculosis (34, 35) and likely plays a critical role in allowing human macrophages do so as well (48, 49), although modeling this in vitro has proved remarkably difficult (1). In this work, we have demonstrated that M. tuberculosis inhibits IFN-γ signaling in human and murine macrophages using several assays to assess the integrity of the IFN-γ signaling pathway at a transcriptional level and at the level of protein expression. Functionally, inhibition of IFN-γ signaling correlates strongly with attenuation of mycobacterial killing in murine macrophages.
Interestingly, neither mAGP nor the 19-kDa lipoprotein significantly alters macrophage NO production. NO is an important mediator of mycobacterial killing both in vitro and in vivo, but several studies suggest that it is not the only means by which IFN-γ-activated macrophages kill mycobacteria. Although mice that lack inducible NO synthase (NOS) are highly susceptible to mycobacterial infection (36, 37), mice deficient in IFN-γ production are even more susceptible to infection with M. tuberculosis (50). In similar models of infection, wild-type mice survive >90 days postinfection (36), IFN-γ−/− mice die at an average of 18 days (50), and NOS2−/− mice die at an average of 38 days postinfection (36). The discrepancy in survival suggests that IFN-γ stimulates macrophages or other cells to control mycobacterial growth through means other than NO production. The difference in survival is not accounted for by the production of reactive oxygen species as mice deficient in the production of reactive oxygen species as well as NOS2 (Phox/NOS−/−) mice are no more susceptible to infection than NOS2−/− mice (37). Thus, our data indicate that mAGP and 19-kDa inhibit the mechanism(s) other than the production of NO by which IFN-γ activated macrophages control mycobacterial growth, but that mechanism remains to be identified.
Taken together, these data show that several components of M. tuberculosis down-modulate IFN-γ signaling in macrophages. The components we examined, the mycobacterial mAGP complex and the 19-kDa lipoprotein, have a similar functional effect but act through different proximal signaling pathways. During the course of infection, inhibition of IFN-γ signaling through these redundant signaling mechanisms may allow M. tuberculosis to persist in the infected macrophage by down-modulating the macrophage’s mycobactericidal activity. Inhibition of killing may act together with other consequences of inhibition of IFN-γ signaling such as down-modulation of MHC class II expression and inhibition of MHC class I and class II mycobacterial Ag presentation (10, 14, 51), to allow M. tuberculosis to escape eradication by the host immune system.
We thank Robert Modlin for providing acylated and nonacylated lipopeptide, Douglas Golenbock for TLR2−/− and MyD88−/− mice, and George O’Keefe and Etty Benveniste for the CIITA-luciferase construct.
This work was supported by National Institutes of Health Grants AI46097 and AI053074 (to J.D.E.), AI48704 and AI51929 (to E.J.R.), AI07118 and AI23545 (to B.R.B.), AI50734-02 (to S.M.F.). S.M.F. was also supported by a Roche Postdoctoral Fellowship Award from the Infectious Diseases Society of America and a Postdoctoral Research Fellowship for Physicians from the Howard Hughes Medical Institute.
Abbreviations used in this paper: CIITA, MHC class II transactivator; mAPG, mycolylarabinogalactan peptidoglycan; TLR, Toll-like receptor; MyD88, myeloid differentiation factor 88; SCWP, soluble cell wall protein; NOS, NO synthase.