Mycobacterium tuberculosis Inhibits IFN-γ Transcriptional Responses Without Inhibiting Activation of STAT1¹

Mycobacterium tuberculosis is a highly successful human pathogen, as indicated by its ability to infect and cause disease in up to 10 million people annually worldwide. Active tuberculosis occurs in people with apparently normal immune systems and in HIV-infected people before profound depletion of circulating CD4⁺ lymphocytes (1), suggesting that M. tuberculosis itself has evolved specific mechanisms for evading destruction by the human immune system.

Macrophages are essential in host defense against many infections. When macrophages encounter most bacteria, they phagocytose and kill them. In contrast, macrophages phagocytose, but do not kill, M. tuberculosis. Indeed, macrophages support the intracellular growth of M. tuberculosis (reviewed in Ref. 2), and the observation that M. tuberculosis has evolved ligands for at least seven distinct macrophage surface receptors suggests that entry into macrophages provides an advantageous environment for induction of mycobacterial gene expression and subsequent growth and spread (3). Although several cytokines and hormones can modulate the activity of macrophages, the predominant activator of macrophage microbicidal activity is IFN-γ (4, 5). IFN-γ is secreted by lymphocytes in response to Ag stimulation and has been shown to be a determinant of resistance to tuberculosis in mice, as demonstrated by the severe, disseminated tuberculosis seen in IFN-γ knockout mice (6, 7). Studies of patients with tuberculosis have demonstrated the presence of IFN-γ in pleural fluid (8, 9), lung fluid (10), and lymph nodes (11), suggesting that a defect in response to IFN-γ rather than the absence of its production allows tuberculosis to progress. In vitro, IFN-γ activates human macrophages to control the growth of intracellular pathogens, including Toxoplasma gondii, Leishmania donovani, Chlamydia psittaci, and Legionella pneumophila (4, 12, 13, 14), but is unable to activate human macrophages to restrict or kill virulent M. tuberculosis (12, 15). This suggests that M. tuberculosis might interfere with cellular signal transduction pathways that are activated by IFN-γ and thereby avoids being killed within macrophages.

The signal transduction pathway initiated by IFN-γ is becoming increasingly well characterized. Binding of IFN-γ to cell surface receptors results in activation of the tyrosine kinases JAK1³ and JAK2, leading to phosphorylation of cytoplasmic STAT1. Tyrosine-phosphorylated STAT1 homodimerizes through interaction of the SH2 domain on one molecule with phosphotyrosine on another and translocates to the nucleus. In the nucleus, STAT1 homodimers activate transcription of specific genes that possess γ-activation sequences (GAS; consensus sequence is TTNCNNNAA). Human genes that contain GAS include Fcγ receptor type I (CD64), guanylate binding protein-2, class II trans-activator, and indoleamine-2,3-dioxygenase (16, 17, 18, 19). Although phosphorylation of STAT1 on Tyr⁷⁰¹ is sufficient for dimerization, nuclear translocation, and activation of transcription, maximal transcriptional activity of STAT1α also requires phosphorylation on a single serine (Ser⁷²⁷) (20). Interaction of STAT1α with the basal transcriptional apparatus is mediated by interaction with the CREB binding protein (CBP)/p300 family of transcriptional coactivators, an interaction that may be regulated in part by phosphorylation of STAT1α at Ser⁷²⁷ (21). Dephosphorylation of STAT1 is one means of terminating IFN signaling (22), although at least a portion is ubiquinated and catabolized by proteasomes (23).

In the present study we tested the hypothesis that M. tuberculosis blocks IFN-γ-initiated activation of macrophages, using human macrophages infected with a virulent strain of M. tuberculosis. We found that M. tuberculosis infection inhibits IFN-γ-activated microbicidal activity and IFN-γ-induced gene expression without inhibiting proximal steps in the JAK-STAT signaling pathway. Rather, M. tuberculosis infection disrupts the essential interaction of STAT1 with CBP and p300. We also found that the effect of live bacteria could be reproduced by gamma-irradiated M. tuberculosis and by crude cell wall fragments but not by purified lipoarabinomannan.

Human monocytes were isolated from buffy coats and cultured for 6 days in 24- or 6-well tissue culture plates (Corning, Corning, NY) as previously described (24) with a minor modification: cells were cultured in RPMI 1640 with 2 mM l-glutamine (Life Technologies, Gaithersburg, MD) and 2.5% autologous serum overnight after plating to improve adherence. Culture medium and nonadherent cells were removed by aspiration after 1 and 5 days of culture, and monolayers were subsequently incubated with fresh culture medium supplemented with 10% autologous serum.

M. tuberculosis (Erdman) was used to infect macrophages as previously described (3). All experiments with M. tuberculosis-infected macrophages were performed with cells 3 days after infection unless stated.

T. gondii (RH strain) were obtained from Dr. Kami Kim (Albert Einstein College of Medicine, New York, NY) and maintained by passage in human foreskin fibroblasts grown in DMEM with 10% heat-inactivated FBS and 2 mM l-glutamine (Life Technologies). Freshly isolated parasites from lysed human foreskin fibroblast monolayers were resuspended in RPMI 1640 with 2 mM l-glutamine and 2.5% heat-inactivated FBS and added to macrophage monolayers at an MOI of 1. After 1 h of incubation at 37°C, extracellular parasites were removed by washing the monolayers three times with PBS. Some cells were fixed for enumeration of toxoplasma at this time corresponding to time zero, while others were incubated with RPMI 1640 with autologous serum alone or medium containing IFN-γ (100 ng/ml; human rIFN-γ; 3 × 10⁷ U/mg; Genentech, South San Francisco, CA) at 37°C for 8–24 h. Cells were then fixed, and intracellular toxoplasma were enumerated after staining with 4′,6-diamidino-2-phenylindole, dihydrochoride (DAPI; 0.5 μg/ml; Molecular Probes, Eugene, OR).

M. tuberculosis-infected and uninfected human macrophages cultured in six-well plates were either treated or untreated with IFN-γ (20 ng/ml) for 24 h at 37°C. Cell monolayers were washed twice with PBS, chilled in PBS containing 0.5 mM EDTA on ice, and scraped from the wells. Cells were washed and resuspended in PBS with 1% human serum and 0.1% NaN₃ at 5 × 10⁶ cells/ml. Aliquots of 5 × 10⁵ cells were incubated for 45 min on ice with FITC-conjugated anti-FcγRI (anti-CD64; Ancell, Bayport, MN) using the concentrations recommended by the manufacturer. Cells were then washed three times with PBS and fixed overnight in 1% paraformaldehyde. Ten thousand cells were analyzed for FcγRI expression on a FACSort flow cytometer with CellQuest software (Becton Dickinson, Mountain View, CA).

Total cellular RNA was isolated (RNeasy, Qiagen, Chatsworth, CA) from either M. tuberculosis-infected or uninfected macrophages cultured in six-well plates that were treated with IFN-γ (20 ng/ml) at 37°C for the indicated times. RNA was quantitated by UV absorbance, and equal amounts (10 μg) were fractionated by electrophoresis, transferred to nylon membranes, and hybridized with a ³²P-radiolabeled FcγRI cDNA probe (10⁶ cpm/ml) containing a fragment of human FcγRI cDNA (nucleotides 488–845 of the a1 splicing product). After washing and exposure to film, blots were stripped for 2 h at 65°C in stripping solution (5 mM Tris-HCl (pH 8.0), 2 mM EDTA, and 0.1× Denhardt’s solution) to remove bound FcγRI probe and then rehybridized with a ³²P-radiolabeled GAPDH probe containing nucleotides 536–899 of the human GAPDH cDNA to verify the amount of RNA in each lane. Quantitation of signals on Northern blots was performed on IS-1000 Digital Imaging System (Alpha Innotech, San Leandro, CA). Both radiolabeled FcγRI and GAPDH cDNA probes were generated by random priming with [α-³²P]dCTP (Amersham, Arlington Heights, IL) using a Prime-It RmT Random Primer Labeling Kit (Stratagene, La Jolla, CA).

Before treatment with IFN-γ, M. tuberculosis-infected and uninfected macrophages were preincubated with fresh culture medium for 2 h at 37°C. The cells were then treated with IFN-γ (20 ng/ml) for the indicated times, washed with ice-cold PBS, scraped into PBS, and pelleted by centrifugation (500 × g) at 4°C for 10 min. Cell pellets were resuspended in buffer A (10 mM HEPES (pH 7.9), 10 mM KCl, 0.1 mM EDTA, 1 mM EGTA, 1 mM sodium orthovanadate, 1 mM PMSF, 20 μg/ml aprotinin, 20 μg/ml antipain, 20 μg/ml leupeptin, and 10 μg/ml pepstatin A) and incubated on ice for 10 min. Nonidet P-40 was added to a final concentration of 0.2%, and the cell suspension was passed through a 26-gauge needle to break open cells. After centrifugation (15,000 × g, in a microcentrifuge) at 4°C for 1 min, supernatants were collected as cytoplasmic extracts, and the pellets (crude nuclei) were washed with buffer A and then resuspended in buffer C (20 mM HEPES (pH 7.9), 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 1 mM sodium orthovanadate, and the same protease inhibitors as in buffer A). After incubation at 4°C for 60 min, insoluble materials were pelleted by centrifugation (15,000 × g) at 4°C for 10 min, and supernatants were collected as nuclear extracts. Whole cell lysates were prepared by lysis of cells in RIPA lysis buffer containing phosphatase and protease inhibitors (50 mM Tris (pH 7.5), 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 0.15 M NaCl, 2 mM EDTA, 50 mM sodium fluoride, 1 mM sodium orthovanadate, 1 mM PMSF, 20 μg/ml aprotinin, 20 μg/ml antipain, 20 μg/ml leupeptin, and 10 μg/ml pepstatin) at 4°C for 10 min. The protein concentration was measured by the method of Bradford (25) using a protein assay kit and BSA standards (Bio-Rad Laboratories, Hercules, CA).

Proteins were subjected to SDS-PAGE on 8% polyacrylamide, transferred to nitrocellulose membranes, and blocked with 5% nonfat milk in Tris-buffered saline-Tween-20 (20 mM Tris (pH 7.6), 0.14 M NaCl, and 0.1% Tween-20) for 1 h at room temperature. The membranes were then incubated with a rabbit polyclonal Ab specific for phosphorylated (at Tyr⁷⁰¹) STAT1 (New England Biolabs, Beverly, MA), a rabbit polyclonal Ab specific for serine-phosphorylated (at Ser⁷²⁷) STAT1α (provided by Dr. David A. Frank, Dana-Farber Cancer Institute, Boston, MA) (26), a mAb that recognizes total STAT1 (Transduction Laboratories, Lexington, KY), or anti-annexin I polyclonal Ab 4800 (27) at a 1/50,000 dilution. After washing with Tris-buffered saline-Tween-20, blots were incubated with HRP-conjugated goat anti-rabbit IgG secondary Ab (for anti-phosphorylated STAT1 and anti-annexin I) or goat anti-mouse (for anti-total STAT1; Zymed, South San Francisco) for 1 h at room temperature, and bound Ab was visualized by enhanced chemiluminescence (Amersham, Arlington Heights, IL). Densitometry was performed by measuring the density of bands using an IS-1000 Digital Imaging System (Alpha Innotech).

Cytoplasmic extract (10 μg protein) or nuclear extract (1.5 μg protein) from equal numbers of cells was incubated with 10⁵ cpm of ³²P end-labeled double-stranded oligonucleotide probe as previously described (28). The phosphorylated probe containing the sequence GTATTTCCCAGAAAAAGGA found in the promoter region of the FcγRI gene was provided by Drs. Sarah Gaffen and Mark Goldsmith (Gladstone Institute of Virology and Immunology, San Francisco, CA). DNA-protein complexes were separated from free probe by electrophoresis through 4% nondenaturing polyacrylamide gels in 0.25× TBE buffer (22.5 mM Tris-borate and 0.5 mM EDTA). The gels were dried, and bands were visualized by autoradiography.

Whole cell extracts of macrophages were prepared as described by Nandan et al. (29). An extended oligonucleotide containing a biotinylated 5′ nucleotide, a six-base spacer, and the GAS from the human guanylate binding protein gene (29) was annealed to its complementary strand, and 100 pmol was used to bind dimerized STAT1 in whole cell extracts (200 μg of protein/condition). After allowing the GAS oligonucleotide to bind STAT1 in cell extracts (2 h, 4°C, on a rotator), streptavidin-agarose beads (Ultralink Streptavidin, Pierce, Rockford, IL; 20 μl of a 50% slurry) were added, and incubation was continued for an additional hour at 4°C. After washing three times with lysis buffer containing 100 mM KCl with protease and phosphatase inhibitors, GAS binding proteins were eluted by boiling in Laemmli sample buffer, and bound proteins were analyzed by SDS-PAGE and immunoblotting. Initial experiments established the optimal quantity of GAS oligo and streptavidin-agarose beads needed to bind all the competent STAT1 in macrophage cell extracts and established that STAT1 binding was displaceable with the addition of excess GAS oligonucleotide (not shown). Bound STAT1 was detected with anti-Tyr⁷⁰¹-phospho-STAT1 (New England Biolabs), and bound CBP and p300 were detected with an Ab from Zymed that recognizes both proteins. The density of immunoreactive STAT1 and CBP/p300 bands was quantitated using an image analysis system (Alpha Innotech).

A neutralizing mAb to TGF-β1, -β2, and -β3 was purchased from Genzyme (Cambridge, MA), and latency-associated peptide was purchased from R & D Systems (Minneapolis, MN). Antagonists or carrier were added (at the indicated concentrations) to macrophage culture medium at the same time as addition of M. tuberculosis.

Gamma-irradiated M. tuberculosis, whole cell lysates, cytosol, membranes, total lipid extract, cell wall, and lipoarabinomannan were all derived from M. tuberculosis H37Rv and were provided by Dr. John Belisle (Colorado State University, Fort Collins, CO). The monoclonal Ab CS-35 was also provided by Dr. Belisle. Killed bacteria or isolated components were incubated with macrophages for 1 h, followed by washing with fresh medium, then macrophages were incubated for 2 days before stimulation with IFN-γ.

The observation that human macrophages cannot kill M. tuberculosis despite IFN-γ treatment could be due to one of two possible general mechanisms. One is that M. tuberculosis is intrinsically resistant to the microbicidal mechanisms of human macrophages. The other is that M. tuberculosis inhibits IFN-γ-initiated activation of macrophages by blocking one or more steps in the IFN-γ signal transduction pathway. Because IFN-γ can activate macrophages to restrict the growth of T. gondii, we reasoned that if M. tuberculosis blocks IFN-γ signaling, M. tuberculosis-infected macrophages should not be able to respond to IFN-γ by restricting T. gondii. Shown in Fig. 1 are results that demonstrate that coinfection of macrophages with M. tuberculosis inhibits the ability of IFN-γ to cause macrophages to restrict T. gondii. T. gondii invaded as efficiently and replicated within M. tuberculosis-infected macrophages at the same rate as in uninfected macrophages. Although IFN-γ activated control macrophages to restrict T. gondii, coinfection of macrophages with M. tuberculosis completely abrogated the ability of IFN-γ to activate macrophages to kill or restrict growth of T. gondii. These results support the hypothesis that M. tuberculosis blocks IFN-γ signaling in human macrophages.

To further test the hypothesis that M. tuberculosis inhibits IFN-γ signaling, we examined another macrophage response: induction of expression of the type I receptor for the Fc domain of IgG (FcγRI or CD64). Analysis by flow cytometry demonstrated that cell surface FcγRI was expressed constitutively on human macrophages and that IFN-γ increased expression of cell surface FcγRI on uninfected and M. tuberculosis-infected macrophages ∼3-fold (Fig. 2). In this assay, M. tuberculosis infection inhibited IFN-γ induction of FcγRI expression by ∼50% (range in 16 independent experiments using cells from different donors was 34–60%) compared with that in uninfected cells (Fig. 2). The effect of M. tuberculosis was not simply a consequence of phagocytosis, because macrophages that phagocytosed serum-opsonized zymosan or latex beads did not exhibit any reduction in IFN-γ up-regulation of FcγRI (not shown).

The increase in expression of FcγRI in response to IFN-γ is due to direct transcriptional activation of the FcγRI gene (30). IFN-γ induced expression of mRNA in both uninfected and M. tuberculosis-infected macrophages; however, M. tuberculosis markedly inhibited the increase in FcγRI mRNA in response to IFN-γ (Fig. 3). Densitometric analysis of the blot revealed that M. tuberculosis infection resulted in reduction of IFN-γ-induction of FcγRI mRNA transcription, with up to 60% reduction in cells treated with IFN-γ for 4–8 h. These results indicate that M. tuberculosis inhibits IFN-γ induction of FcγRI cell surface expression by inhibiting activation of transcription, the end point in the IFN-γ signaling pathway. M. tuberculosis inhibition of IFN-γ transcriptional activation was exerted on all IFN-γ-responsive genes that we examined, including all three alternatively spliced transcripts of the FcγRI gene, indoleamine 2,3-dioxygenase (31), and guanylate binding protein-2 (32) (not shown).

To further characterize the effect of M. tuberculosis infection on IFN-γ signaling, we examined an intermediate step in the IFN-γ signal transduction pathway: tyrosine phosphorylation of STAT1. As shown in Fig. 4,A, IFN-γ induced tyrosine phosphorylation of STAT1α and -β in both uninfected and M. tuberculosis-infected macrophages in a similar time-dependent manner. Instead of inhibition of STAT1 tyrosine phosphorylation, we found that STAT1 phosphorylation was greater in M. tuberculosis-infected macrophages than in uninfected cells. The ultimate extent of tyrosine phosphorylation of STAT1α and -β in M. tuberculosis-infected macrophages was 2.1-fold higher than that in uninfected cells at 30 min of IFN-γ treatment, as quantitated by densitometry. When the same samples were analyzed using an Ab that detects total STAT1α and -β, densitometric analysis revealed that the abundance of STAT1 protein was increased 2-fold in M. tuberculosis-infected macrophages compared with that in uninfected cells, suggesting that M. tuberculosis infection results in up-regulation of both STAT1α and STAT1β expression in macrophages either by increasing synthesis or by decreasing degradation of the protein (Fig. 4,B). To address the possibility that M. tuberculosis nonspecifically increases the abundance of cytoplasmic proteins in macrophages, we examined the level of another cytosolic protein, annexin I. The annexin I content was the same in all the samples (Fig. 4 C), indicating that increase in STAT1 abundance does not extend to other cytoplasmic proteins.

Because tyrosine-phosphorylated STAT1 must dimerize and translocate from the cytoplasm to the nucleus to activate transcription, we investigated the effect of M. tuberculosis infection on translocation of STAT1 from the cytoplasm to the nucleus. As shown in Fig. 5, tyrosine-phosphorylated STAT1 is present in nuclei isolated from either uninfected or M. tuberculosis-infected macrophages treated with IFN-γ, indicating that IFN-γ-induced nuclear translocation of STAT1 is not inhibited by M. tuberculosis infection. Immunofluorescence microscopy using Abs to total STAT1 or to tyrosine-phosphorylated STAT1 confirmed that IFN-γ induces nuclear translocation of STAT1 in M. tuberculosis-infected macrophages (not shown).

Because M. tuberculosis infection did not inhibit IFN-γ-induced tyrosine phosphorylation or nuclear translocation of STAT1, we examined the effect of M. tuberculosis infection on IFN-γ-activated DNA binding activity of STAT1 by EMSA. Treatment of macrophages with IFN-γ induced formation of STAT1 complexes in the cytoplasmic and nuclear extracts that were competent to bind a synthetic GAS derived from the human FcγRI gene (Fig. 6). M. tuberculosis infection resulted in enhanced formation of the IFN-γ-activated DNA-protein complex, reflecting the increased amount of cytoplasmic and nuclear STAT1 in M. tuberculosis-infected macrophages (Fig. 5 A). These results demonstrate that M. tuberculosis infection does not affect the ability of STAT1 to dimerize, translocate to the nucleus, or bind specific DNA target sequences.

Although phosphorylation of STAT1 on tyrosine 701 is sufficient to cause dimerization, nuclear translocation, and DNA binding of STAT1, full transcriptional activity of STAT1α also requires phosphorylation at serine 727 (20, 33). Using an Ab that specifically recognizes STAT1α phosphorylated at Ser⁷²⁷ (26), we found that M. tuberculosis-infected macrophages contain a small amount of serine-phosphorylated STAT1α, and that addition of IFN-γ stimulated serine phosphorylation of STAT1α to a similar extent in M. tuberculosis-infected and uninfected macrophages (Fig. 7). Therefore, the mechanism of M. tuberculosis inhibition of IFN-γ responses is not exerted through reduced phosphorylation or enhanced dephosphorylation of STAT1α Ser⁷²⁷.

Transcriptional activation of IFN-γ-responsive genes depends on association of STAT1 dimers with the transcriptional coactivators CBP and p300, which link STAT1 to the basal transcriptional apparatus and RNA polymerase II (21, 34). Initial efforts to examine the association of CBP and p300 with STAT1 by immunoprecipitation were unsuccessful, because the anti-STAT1 Abs used recognize the N- and C-terminal domains of STAT1, which are the same domains that interact with CBP and p300. Consequently, we used oligoprecipitation with a GAS-containing oligonucleotide bound to agarose beads to isolate STAT1 dimers and STAT1-associated proteins from M. tuberculosis-infected and uninfected macrophages. These studies revealed an IFN-γ- and time-dependent association of STAT1 with the GAS oligonucleotide in extracts of infected and uninfected macrophages (Fig. 8). Although the amount of STAT1 that bound the GAS oligonucleotide in lysates of M. tuberculosis-infected macrophages was the same as (or slightly greater than) that in lysates of uninfected macrophages, lower amounts of CBP/p300 were present in eluates from M. tuberculosis-infected compared with uninfected macrophages at all time points examined. Densitometric analysis of the samples harvested after 30 min of IFN-γ treatment (when the quantity of oligo-bound STAT1 was maximal), revealed an apparent 77% reduction in the amount of CBP/p300 associated with STAT1 (when normalized to the amount of bound STAT1) in M. tuberculosis-infected macrophages compared with that in uninfected macrophages. This reduction in CBP/p300 association with STAT1 closely resembles the extent of reduction in transcriptional responses to IFN-γ in M. tuberculosis-infected macrophages.

The finding that M. tuberculosis inhibits IFN-γ transcriptional responses without affecting known steps in the JAK-STAT signaling pathway suggested that its mechanism of inhibition resembles that of TGF-β (29, 35, 36). Moreover, TGF-β is secreted by human monocytes containing an avirulent strain of M. tuberculosis, H37Ra (37). However, neither an mAb to TGF-β nor latency-associated peptide, a protein that binds TGF-β with high affinity and prevents receptor binding, restored the ability of M. tuberculosis-infected macrophages to respond to IFN-γ (Fig. 9). To further evaluate the possibility that TGF-β mediates the inhibition of IFN-γ responses, we assayed TGF-β in medium of macrophages infected with the Erdman strain of M. tuberculosis. We found that there was no biologically active TGF-β detectable in untreated medium, but we found that acid activation of the medium revealed up to 120 pg/ml of activatable TGF-β. When we used this concentration of active TGF-β to treat macrophages, we did not observe any inhibition in IFN-γ responses (data not shown). Taken together, these results strongly indicate that the inhibitory effect of M. tuberculosis on macrophage responses to IFN-γ is not mediated by the secretion of TGF-β.

To determine whether inhibition of IFN-γ responses by M. tuberculosis is due to a bacterial component induced after infection of macrophages or is due to a constitutively expressed bacterial component, we substituted killed (gamma-irradiated) M. tuberculosis (H37Rv) for live bacteria before stimulation with IFN-γ. As shown in Fig. 10, there was a dose-related inhibition of IFN-γ-induced expression of FcγRI that closely corresponded to the inhibition observed with live M. tuberculosis.

The finding that killed M. tuberculosis are capable of causing inhibition of IFN-γ responses indicates that one or more preformed components of the bacteria are sufficient for initiating a pathway that results in inhibition of IFN-γ responses. In a subsequent experiment, we found that a whole cell lysate of M. tuberculosis H37Rv also inhibited IFN-γ-induced up-regulation of cell surface FcγRI, indicating that the essential bacterial component was able to exert its effects even when it was not presented in the form of whole particulate bacteria (not shown). We therefore surveyed several subcellular fractions of gamma-irradiated M. tuberculosis H37Rv. We found that cytosol (up to 100 μg/ml), crude membranes (up to 50 μg/ml), or total lipids (up to 50 μg/ml) had no effect on the ability of macrophages to respond to IFN-γ (data not shown). In contrast, we found that the cell wall fraction possessed a potent ability to initiate inhibition of IFN-γ up-regulation of FcγRI (Fig. 11). As little as 0.5 μg/ml (quantitated by protein assay) of M. tuberculosis cell wall was capable of maximum inhibition of IFN-γ up-regulation of FcγRI. Because lipoarabinomannan (LAM) is a major component of the M. tuberculosis cell wall, we examined the ability of purified LAM to cause inhibition of IFN-γ up-regulation of FcγRI. Although LAM exhibited some activity, it did not inhibit the response to IFN-γ to the extent observed with unfractionated cell walls, even at very high concentrations of LAM (Fig. 11B). To determine the relationship between the concentration of purified LAM that exerted a partial inhibitory effect and the concentration present in highly active unfractionated cell walls, we quantitated LAM in cell walls by immunoblotting, compared with a standard curve constructed using purified LAM. This revealed that the content of LAM in unfractionated cell walls did not exceed 350 ng per 0.5 μg of cell wall protein. Because 25 μg/ml of LAM caused less inhibition of IFN-γ than 0.5 μg/ml of unfractionated cell wall, we conclude that LAM cannot be the sole component of the M. tuberculosis cell wall that initiates the inhibition of IFN-γ responses.

M. tuberculosis is a facultative intracellular pathogen that infects and replicates within human macrophages despite the development of a cell-mediated immune response. Because IFN-γ activates macrophages to kill other pathogens but cannot activate human macrophages to kill M. tuberculosis, we tested the hypothesis that M. tuberculosis survives in macrophages by inhibiting IFN-γ signaling. We found that M. tuberculosis infection of macrophages indeed inhibits macrophage responses to IFN-γ, and that inhibition is exerted at the level of transcription of IFN-γ-responsive genes. At least two other pathogens have been found to inhibit IFN-γ signaling. Leishmania donovani inhibits IFN-γ responses by inhibiting tyrosine phosphorylation of STAT1 by JAK1 and JAK2 (38), and human CMV inhibits responses to IFN-γ in infected fibroblasts and endothelial cells, by depleting cells of JAK kinases through degradation by proteosomes (39). In contrast, infection with M. tuberculosis does not inhibit STAT1 tyrosine or serine phosphorylation, dimerization, nuclear translocation, or recognition of specific DNA sequences. Rather, infection with M. tuberculosis inhibits IFN-γ responses by directly or indirectly disrupting the essential interaction of STAT1α with the transcriptional coactivators CBP and p300. The underlying mechanism by which M. tuberculosis infection disrupts the STAT1α-CBP/p300 interaction remains to be elucidated, and one or more mechanisms could be responsible. First, another transcription factor could compete with STAT1 for the same binding site(s) on CBP/p300. CBP and p300 are present in limiting quantities in most, if not all, cells. A recent study found that activation of Jurkat cells with IFN-α inhibits TNF-α activation of transcription of the HIV-1 long terminal repeat because STAT2 (activated by IFN-α) binds the same domain on p300 as does NF-κB (activated by TNF-α). Because Jurkat cells possess ∼28,000 molecules of p300 and ∼150,000 molecules of STAT2 per cell, activation of STAT2 sequestered p300 so that it was unavailable to NF-κB (40). Similarly, STAT1α and AP-1/ets have been found to inhibit one another’s actions by competition for a limiting quantity of CBP (34). In view of these findings, M. tuberculosis might indirectly inhibit IFN-γ responses by activating a macrophage signaling pathway that requires CBP and/or p300 and thereby restricts the availability of these coactivators for use by STAT1α. Alternatively, M. tuberculosis could directly target the domains of either STAT1 or CBP that are involved in their protein-protein interaction. The interaction of STAT1α with CBP is mediated by binding of the N-terminal domain of STAT1α to Cys/His-rich domain 1 of CBP and binding of the C-terminal domain of STAT1α to Cys/His-rich domain 3 (21), and M. tuberculosis could target any of these domains on either protein to disrupt their association. A precedent for such a mechanism was recently reported; in addition to the well-established interaction of adenovirus E1A with p300 (41, 42), E1A also directly interacts with the C-terminal domain of STAT1α and blocks IFN-γ activated transcription (43). Another potential target of such a mechanism is the N-Myc interactor protein (Nmi). Nmi markedly stabilizes the interaction of STAT1α and CBP and thereby enhances IFN-γ responses (44). Therefore, interference with Nmi could yield the decrease in STAT1-CBP/p300 association that we observed in M. tuberculosis-infected macrophages. Additional experiments will be necessary to determine which of these mechanisms may account for the decreased association of STAT1 and CBP/p300 observed in M. tuberculosis-infected macrophages.

An alternative, distinct, mechanism for M. tuberculosis inhibition of IFN-γ responses is disruption of a pathway that is distinct from the JAK-STAT pathway but that converges with it to activate transcription of IFN-γ-responsive genes. Considering the findings that we report here, a plausible candidate target of such a pathway is CBP. Although there is not yet any information regarding post-translational modification of CBP in response to IFN-γ, CREB/CBP-dependent transcriptional activation in pituitary cells requires a CBP activating signal that includes nuclear calcium signaling and calcium/calmodulin-dependent kinase IV (45). If an analogous CBP-activating signal is required for IFN-γ signaling, it represents an additional potential target of M. tuberculosis.

M. tuberculosis LAM has been reported to inhibit mitogen-activated protein kinase (MAPK) activation in human monocytes by activating the protein phosphatase SHP-1 (Src homology domain 2-containing tyrosine phosphatase-1) (46). Because MAPK can catalyze the phosphorylation of STAT1α Ser⁷²⁷ (20), it was essential for us to consider inhibition of MAPK activity as a possible mechanism by which M. tuberculosis inhibits IFN-γ responses. However, our finding that STAT1α is phosphorylated on Tyr⁷⁰¹ and Ser⁷²⁷ to the same extent and with the same kinetics in uninfected and M. tuberculosis-infected macrophages makes this effect unlikely to account for inhibition of IFN-γ responses. Moreover, we have not observed any inhibition of phosphorylation of ERK2 at Thr¹⁸³ or Tyr¹⁸⁵ in response to IFN-γ in M. tuberculosis-infected macrophages (l.-M. Ting and J. D. Ernst, unpublished observation). An additional possible target of M. tuberculosis is one or more additional kinases activated by IFN-γ, such as the renaturable tyrosine kinases whose activation by IFN-γ is sensitive to inhibition by TGFβ (29). Although the effects of M. tuberculosis in our experiments are not attributable to TGFβ, these or similar kinases may be sensitive to inhibition by other mediators activated by M. tuberculosis.

Previous studies have demonstrated that Mycobacterium leprae-infected murine macrophages are refractory to IFN-γ induction of microbicidal activity, cytotoxicity for tumor cells, superoxide anion production, and surface Ia Ag expression (47, 48). In addition, exposure of macrophages to high concentrations of purified M. tuberculosis LAM results in defective responses to IFN-γ, including transcriptional activation, intracellular microbicidal activity, expression of MHC class II molecules, and cytotoxicity for tumor cells (49, 50). We found that these effects are also found in human macrophages infected in vitro with a virulent strain of M. tuberculosis. Although M. tuberculosis infection did not completely inhibit transcriptional activation of FcγRI in response to IFN-γ, IFN-γ-induced killing of T. gondii was completely abolished by M. tuberculosis infection in macrophages, suggesting that the block of IFN-γ-activated signaling is functionally significant and may at least partially account for the failure of IFN-γ to activate macrophages to kill M. tuberculosis. Because these experiments were performed using high concentrations of IFN-γ (20–100 ng/ml; 600-3000 U/ml) it is likely that a more complete block may occur in vivo, where the local concentration of IFN-γ is lower.

In contrast to the findings of the aforementioned studies, we found that LAM is unlikely to be the sole component of M. tuberculosis that initiates the inhibition of IFN-γ responses. Although the most potent inhibitory activity was found in a cell wall fraction, the concentration of LAM required for inhibition of IFN-γ responses was >100-fold greater than that found in the cell wall fraction that exerted even greater inhibitory activity. Although this observation does not completely exclude a role for LAM in initiating a pathway that results in inhibition of IFN-γ responses, it suggests that if LAM is involved at all, it depends on another cell wall component for its effect. The role for this other hypothetical cell wall component could be a structural one, in which it orients LAM in a manner that enables it to interact more productively with macrophages than it can when it is in a purified (probably micellar) form. The alternative role for the other hypothetical cell wall component is to provide an additional signal to macrophages that, when combined with a signal initiated by LAM, causes potent inhibition of IFN-γ responses. Such a mechanism is reminiscent of the need for both lipoteichoic acid and peptidoglycan of Staphylococcus aureus to initiate septic shock (51). Further analysis of the precise components and conformations of the cell wall will be necessary to definitively identify the molecule(s) responsible for initiating the inhibition of IFN-γ responses.

The survival of M. tuberculosis in macrophages and resistance to the human immune system are likely to involve more than one mechanism, and identification of these mechanisms will be crucial to understanding the pathogenesis of this common and serious disease. The ability of M. tuberculosis to block macrophage responses to IFN-γ is likely to be an important trait developed by the bacteria in response to the development of cell-mediated immunity. Overcoming this blockade may allow the immune system to better contain and eradicate M. tuberculosis and may be a valuable adjunct in developing improved therapies for latent and active tuberculosis.

We thank Dr. David Frank (Dana-Farber Cancer Institute) for the Ab to Ser⁷²⁷-phosphorylated STAT1, and Drs. Mark Goldsmith and Sarah Gaffen for the FcγRI GAS oligonucleotide probe and advice concerning the EMSA experiments.