IFN-αβ functions in the transition from innate to adaptive immunity and may impinge on the interaction of Mycobacterium tuberculosis with its host. Infection by M. tuberculosis causes IFN-αβ secretion and down-regulation of IFN-αβ signaling in human APC and the human monocytic cell line THP-1, which provides a model for these studies. Neutralization of secreted IFN-αβ prevents inhibition of IFN-α signaling during infection, but several lines of evidence distinguish inhibition due to infection from a negative feedback response to only IFN-αβ. First, greater inhibition of IFN-α-stimulated STAT-1 tyrosine phosphorylation occurs 3 days postinfection than 1 or 3 days after IFN-αβ pretreatment. Second, LPS also induces IFN-αβ secretion and causes IFN-αβ-dependent down-regulation of IFN-α signaling, yet the inhibition differs from that caused by infection. Third, IFN-α signaling is inhibited when cells are grown in conditioned medium collected from infected cells 1 day postinfection, but not if it is collected 3 days postinfection. Because IFN-αβ is stable, the results with conditioned medium suggest the involvement of an additional, labile substance during infection. Further characterizing signaling for effects of infection, we found that cell surface IFN-αβ receptor is not reduced by infection, but that infection increases association of protein tyrosine phosphatase 1c with the receptor and with tyrosine kinase 2. Concomitantly, IFN-α stimulation of tyrosine kinase 2 tyrosine phosphorylation and kinase activity decreases in infected cells. Moreover, infection reduces the abundance of JAK-1 and tyrosine-phosphorylated JAK-1. Thus, the distinctive down-regulation of IFN-α signaling by M. tuberculosis occurs together with a previously undescribed combination of inhibitory intracellular events.
Interferon-αβ is important for both innate and adaptive immunity to both viral and bacterial pathogens (for recent reviews, see Refs.1 and 2, 3, 4). Recent studies are beginning to characterize the production of and response to IFN-αβ during infection with Mycobacterium tuberculosis. The observation that infected macrophage cultures contain IFN-αβ-stimulated gene factor 3 (ISGF-3),3 a transcription factor activated by IFN-αβ, first suggested production of and response to IFN-αβ during infection (5). Subsequently, production of IFN-αβ by human APC was confirmed (6, 7), and expression of IFN-αβ genes in mouse lungs during infection with M. tuberculosis was reported (8). Infection by M. tuberculosis not only induces IFN-αβ, but also inhibits the response to it (9). The inhibition of IFN-αβ signaling by M. tuberculosis is specific. Although both IFN-αβ and IFN-γ activate the transcription factor STAT-1, its activation by IFN-γ is unaffected by infection (10) (Y. Qiao, S. Prabhakar, and R. Pine, unpublished observations). Although the basis for induction of IFN-αβ during infection with M. tuberculosis has been determined at least in part (11), the mechanism by which IFN-αβ-stimulated signaling is inhibited during infection is largely unknown.
We hypothesized that an autocrine/paracrine loop contributes to the inhibition of the IFN-αβ signal transduction pathway, because some responses to IFN-α are inhibited at least 90% in cultures of cells infected with M. tuberculosis, although the proportion of infected cells does not exceed 50% (9). Moreover, studies relating greater strain or species virulence to induction of IFN-α gene expression (8) and inhibition of IFN-α signaling (9) led us to ask whether the IFN-αβ secreted during infection provides the autocrine/paracrine link. To test this hypothesis and relate it to intracellular events, we focused on the inhibition of IFN-α-stimulated STAT-1 tyrosine phosphorylation and DNA-binding activity for two reasons. First, of two transcription factors that mediate the response to IFN-α, STAT-1 homodimers and ISGF-3, STAT-1 homodimer formation is inhibited much more than ISGF-3 formation in cells infected with M. tuberculosis (9). Second, the almost complete inhibition of STAT-1 tyrosine phosphorylation and DNA-binding activity indicates that the mechanism responsible operates in most or all cells present.
Determining the mechanism(s) of inhibition is aided by knowing the signaling pathway that leads to IFN-α-stimulated STAT-1 tyrosine phosphorylation and DNA-binding activity (reviewed in Refs.12, 13, 14, 15). Binding of these ligands to a common cell surface receptor, which is composed of IFNAR1 and IFNAR2 subunits (also called the α- and β-chains), brings the constitutively associated tyrosine kinases tyrosine kinase 2 (TYK-2) and JAK-1 into proximity, and allows their activation by trans- or autophosphorylation. Activation then leads to phosphorylation of the receptor subunits and transcription factors STAT-1 and -2. STAT-1 can then form a homodimer. Alternatively, STAT-1 can be part of ISGF-3 through heterodimerization with STAT-2 and association with IFN regulatory factor 9. Transcriptional regulation is the primary means by which IFN-αβ exert their effects; STAT-1 homodimers and ISGF-3 regulate distinct sets of genes.
Infection by M. tuberculosis might mimic or foster the normal negative feedback regulation of the response to cytokines that ultimately terminates or greatly reduces stimulation of signal transduction and transcription. Receptor internalization stimulated by ligand binding is a well-known mechanism that pertains to the IFN-αβ receptor (16). Recent studies have demonstrated a role for TYK-2 in control of IFN-αβ receptor recycling (17, 18). Down-regulation of IFN-αβ signaling is also mediated by tyrosine phosphatases, including enzymes that dephosphorylate STAT-1 in the nucleus (19, 20), and perhaps by protein tyrosine phosphatase 1c (PTP1c), which limits the response to IFN-αβ, most likely by decreasing initiation of signaling (21).
The effects of M. tuberculosis on feedback regulation might be direct or indirect. Lipoarabinomannan (LAM) from M. tuberculosis activates PTP-1c and increases its association with cellular membranes (22), but whether that affects IFN-αβ signaling is unknown. Host gene expression induced by M. tuberculosis might also contribute to negative feedback. For example, M. tuberculosis induces IL-10 (23, 24), and IL-10 can inhibit IFN-αβ signaling, perhaps through induction of suppressor of cytokine signaling (SOCS) 3 (25). Induction of SOCS proteins is a common means for down-regulation of cytokine signaling (for recent reviews, see Refs.26 and 27). Direct evidence for SOCS-mediated down-regulation of IFN-αβ signaling has not been reported, nor has induction of SOCS by M. tuberculosis or IFN-αβ. However, the normal termination of IFN-αβ signaling does require protein synthesis, and cells are unresponsive to restimulation until 24 h after the start of an initial cycle of activation and repression (28). The desensitization involves T cell PTP, a nuclear enzyme that acts specifically on STAT-1 (29).
In testing the contribution of IFN-αβ secreted during infection with M. tuberculosis to inhibition of signaling, we found that it is necessary, but not sufficient. Moreover, inhibition by M. tuberculosis is distinct from inhibition caused by the mere presence of IFN-αβ or by LPS, another inducer of IFN-αβ. Among the intracellular mechanisms potentially involved, we observed a novel combination associated with inhibition of IFN-αβ signaling by M. tuberculosis.
Materials and Methods
Abs and other protein reagents
Anti-phosphotyrosine mAb 4G10 was obtained from Upstate Biotechnology; neutralizing anti-IL-10Rα mAb (MAB274) was obtained from R&D Systems; mAbs against STAT-1, tyrosine-phosphorylated STAT-1, and rabbit anti-mouse IgG Ab were obtained from Zymed Laboratories; and mAbs against TYK-2, JAK-1, phosphotyrosine (PY20), and PTP-1c were purchased from BD Transduction Laboratories. Neutralizing Abs against human IFN-αβ were provided by J. Vilcek (New York University School of Medicine, New York, NY); Ab against protein inhibitor of activated STAT 1 (PIAS1) was obtained from Santa Cruz Biotechnology; Ab against tyrosine-phosphorylated TYK-2 was obtained from Cell Signaling Technology; Ab against GST was obtained from Amersham Biosciences. Isotype-matched mAbs used to control for specificity in immunoprecipitation and flow cytometry were obtained from BD Pharmingen. HRP-conjugated goat anti-mouse IgG and anti-rabbit IgG were obtained from Calbiochem. The mAb IFNaR3, specific for the extracellular domain of the IFN-αβ receptor α-chain (also called IFNAR1) and fusion proteins with the GST moiety linked to intracellular domains of the IFN-αβ receptor α-chain were previously described (30, 31). Recombinant human STAT-1 was also previously described (32).
M. tuberculosis and eukaryotic cell culture
Human monocytic THP-1 cells (33) obtained from the American Type Culture Collection were grown and infected with M. tuberculosis strain TN913 (obtained from B. Kreiswirth), as previously described (6). TN913 is one of the predominant C strain isolates from the 1990s New York City tuberculosis outbreak (34) maintained in the PHRI Tuberculosis Center collection. We and others have demonstrated that THP-1 cells provide an accurate model for infection of human monocyte-derived and alveolar macrophages by M. tuberculosis (6, 9, 10, 35, 36). The multiplicity of infection was ∼1. Cells were infected for 1 or 3 days. Conditioned media (CM) were collected from infected cells and parallel cultures of uninfected cells and used to culture uninfected cells for 1 or 3 days, as previously described (37). Cells were stimulated with various doses of IFN-α-2a (Roferon; Hoffmann-La Roche) for various times, as indicated for individual experiments. As indicated, neutralizing Abs against human IFN-αβ or against IL-10Rα were added at a dilution of 1/500 or at 250 ng/ml, respectively, to cell cultures that subsequently were infected or remained uninfected. Negative controls were performed using normal sera from the appropriate species instead of neutralizing Abs. Cells cultured with anti-IFN-αβ Abs were washed three times before final treatment with IFN-α or IFN-β. Control experiments demonstrated effective neutralization of IFN-α and IFN-β (data not shown) or the ability to prevent response to IL-10 (as described by the vendor). LAM of M. tuberculosis was obtained through National Institutes of Health, National Institute of Arthritis and Infectious Diseases contract no. 1 (AI-75320), from Colorado State University. When included, it was added to cell cultures at concentrations from 1–3 μg/ml. Salmonella LPS (Sigma-Aldrich) was added to cultures at 50 ng/ml as indicated.
Extract preparation and EMSA
Whole cell, cytoplasmic, and nuclear extracts were prepared and analyzed by EMSA using radiolabeled IFN-γ activation site (GAS) oligonucleotide to measure STAT-1 DNA-binding activity, as previously described (9). The identity of the complex was confirmed by specific reaction with anti-STAT-1 Ab (data not shown).
Uninfected cells and cells that were infected for 3 days were unstimulated or stimulated by IFN-α for 30 min. Cells were placed on ice, gently scraped into their growth media, then washed (centrifuged at 400 × g, 5 min, 4°C, and resuspended) and incubated as follows. Cells were washed into cold (0–4°C) PBS containing 10 mM NaN3 (PBS-N3), then into cold PBS-N3 containing 10% human serum and incubated for 30 min on ice to block FcRs. One-half to 1 million cells were recovered by centrifugation, then mAb IFNaR3 or isotype-matched control mAb was added to the loosened cell pellet. After an additional 30 min on ice, the cells were washed and recovered by centrifugation. Biotin-conjugated goat anti-mouse IgG was added to the loosened cell pellet. After 15 min on ice, cells were washed and recovered by centrifugation. PE-conjugated streptavidin (Molecular Probes) was added to the loosened cell pellet. After 15 min on ice, the cells were washed into chilled PBS then fixed by adding formaldehyde to 1% and incubating overnight at 4°C. Staining was analyzed on a FACSCalibur (BD Biosciences). PE emission excited at 488 nm was detected by logarithmic amplification through a 585/42-nm barrier filter. Autofluorescence above the second log in the PE channel was subtracted by electronic compensation using the autofluorescence detected through a 670LP-nm barrier filter in another channel. All data were acquired in list mode and processed subsequently with CellQuest software (BD Biosciences) on a Macintosh computer system (Apple Computer). Forward and 90° angle scattered laser light intensities were used to exclude cellular debris.
Precipitation with GST fusion proteins or Abs
GST fusion proteins or the GST moiety alone were expressed in Escherichia coli DH5 α cells and purified by affinity chromatography on glutathione-Sepharose (Amersham Biosciences). Whole-cell extracts prepared as described for analysis by EMSA were incubated overnight at 4°C with the indicated GST fusion protein or GST alone bound to the glutathione-Sepharose beads or with mouse mAbs. For immunoprecipitation, rabbit Ab against mouse IgG was added, incubation was continued for 6 h at 4°C, then protein A-agarose beads (Zymed Laboratories) were added, and incubation was continued overnight at 4°C. The beads (glutathione-Sepharose or protein A-agarose) and bound proteins were washed five times with cold extraction buffer.
Proteins in cell extracts and those recovered by precipitation with Abs or GST fusion proteins were prepared for and separated by SDS-PAGE. The proteins were transferred onto polyvinylidene difluoride (PVDF) membrane (Amersham Biosciences), and the residual binding sites on the membrane were blocked by incubation in 5% nonfat dry milk in PBS overnight at 4°C. The membranes were subsequently analyzed by standard methods (38) using the indicated primary Abs, the appropriate secondary Abs, and ECL reagents (Kirkegaard & Perry Laboratories). Analysis of more than one target in a sample was performed by stripping and reprobing the same membrane for each different protein or protein isoform. Equal loading of cell extracts was confirmed by Coomassie Blue staining of residual protein in polyacrylamide gels after transfer to the PVDF membrane (data not shown). Equal loading of precipitates was confirmed by detection of the target protein on the PVDF membrane.
In vitro kinase assay
Whole-cell extracts were prepared as usual, except that phosphatase inhibitors (NaF, Na3VO4, and Na4P2O7) were omitted. TYK-2 was then immunoprecipitated as described above, and kinase activity was determined with minor modifications of a previously described protocol (39). The immunocomplex-protein A-agarose beads were washed three times in extraction buffer containing phosphatase inhibitors and twice in the in vitro kinase assay buffer (12 mM MgCl2, 50 mM Tris-HCl (pH 7.4), 5 mM Na3VO4, and 150 mM NaCl). The immunocomplex-protein A-Sepharose beads were resuspended in 30 μl of in vitro kinase buffer, in which MnCl2 at a final concentration of 2.5 mM and 30 μCi of γ-[32P]ATP (ICN Biomedicals) were included. GST-IFNAR1 was also added as an exogenous kinase substrate. The beads were incubated for 30 min at room temperature, and the reaction was terminated by addition of SDS-PAGE loading buffer. Proteins were separated by SDS-PAGE and transferred to PVDF membranes. Radiolabeled proteins were detected with a Storm 860 PhosphorImager (Molecular Dynamics) and were visualized with ImageQuant software. The identities of the indicated proteins were determined by immunoblot.
IFN-αβ secreted during infection is necessary for inhibition of IFN-α signaling
IFN-αβ signaling is markedly down-regulated between a few hours and 1 day after stimulation of cells by IFN-αβ, but the inhibition diminishes with increasing time after removal of IFN-αβ and does not occur after 3 days (28). We previously reported that cells infected with M. tuberculosis for 1 day secrete IFN-αβ (6), and that IFN-αβ signaling is inhibited at 3 days postinfection (9). We examined whether the secreted IFN-αβ would cause down-regulation of signaling 1 and/or 3 days postinfection. Cells infected with M. tuberculosis for 1 or 3 days and uninfected cells were cultured in parallel in the presence of control Abs or neutralizing anti-IFN-αβ Abs, then washed and stimulated, or not, by exogenous IFN-α for 30 min to test signaling. We first measured STAT-1 DNA-binding activity as a functional readout (Fig. 1 A). Without stimulation by exogenous IFN-α, uninfected and infected cells contained little or no STAT-1 DNA-binding activity, whether cultured with control or neutralizing Abs (lanes 1–4 and 9–12). Uninfected cells cultured with control and neutralizing Abs for 1 day contained similar levels of IFN-α-stimulated STAT-1 DNA-binding activity (compare lanes 5 and 7), whereas culture for 3 days with the neutralizing Abs limited the IFN-α-stimulated activity (compare lanes 13 and 15). As expected, IFN-α stimulation of STAT-1 DNA-binding activity was inhibited in cells infected with M. tuberculosis for 1 or 3 days in the absence of neutralizing Abs (lanes 6 and 14). In contrast, the inhibition failed to occur when cells infected for 1 or 3 days in the presence of neutralizing Abs were stimulated by IFN-α; STAT-1 DNA-binding activity was similar to or greater than that in the respective uninfected cells that were cultured and stimulated in parallel (compare lanes 7 and 8 and lanes 15 and 16). Thus, IFN-αβ secreted during infection with M. tuberculosis is necessary to attain inhibition of IFN-α-stimulated signaling evidenced by lack of STAT-1 DNA-binding activity both 1 and 3 days postinfection.
We also determined whether IFN-αβ secreted during infection with M. tuberculosis was necessary for the observed changes in STAT-1 protein abundance and tyrosine phosphorylation. The increase in STAT-1 abundance (Fig. 1 B, upper panel) that occurs upon infection for 1 or 3 days was similar with and without stimulation by exogenous IFN-α, as expected, and the increase was reduced or eliminated by inclusion of the neutralizing Abs (compare lanes 2 to 4, 6 to 8, 10 to 12, and 14 to 16). STAT-1 tyrosine phosphorylation (lower panel) in cells infected with M. tuberculosis was also reduced or prevented by growth in the presence of neutralizing Abs (compare lanes 2 and 4 and lanes 10 and 12). Importantly, the inhibition of IFN-α-stimulated STAT-1 tyrosine phosphorylation was also prevented (compare lanes 6 and 8 and lanes 14 and 16). Thus, autocrine and/or paracrine stimulation by the IFN-αβ secreted during infection is necessary for the increases in abundance and tyrosine phosphorylation of STAT-1 that take place; yet the secreted IFN-αβ is also required for the down-regulation of signaling that is detected when infected cells are tested by stimulation with exogenous IFN-α.
Surprisingly, the inhibition of IFN-α-stimulated STAT-1 tyrosine phosphorylation and STAT-1 DNA-binding activity did not correlate at 1 day postinfection. The decrease in DNA-binding activity was substantial (Fig. 1,A, lanes 5 and 6), whereas the decrease in tyrosine phosphorylation was modest (Fig. 1,B, lanes 5 and 6). In contrast, both were strongly inhibited 3 days postinfection (Fig. 1, A and B, lanes 13 and 14). We considered whether infection with M. tuberculosis for 1 day induced an inhibitor that limited the DNA-binding activity of tyrosine-phosphorylated STAT-1, such as PIAS1 (40). However, immunoblot assays failed to detect it in uninfected and infected cells with and without stimulation by IFN-α (data not shown). We directly assayed for the presence of an inhibitory activity in extracts prepared 1 day postinfection by mixing varying amounts of extract with recombinant STAT-1 (Fig. 1 C). The effect on DNA-binding activity, a concentration-dependent decrease, was the same whether extracts were from uninfected or infected cells cultured in parallel (compare lanes 2–5 and lanes 6–9). This result suggests that an inherent property of STAT-1 in cells infected with M. tuberculosis for 1 day is responsible for the greater inhibition of its IFN-α-stimulated DNA-binding activity than its tyrosine phosphorylation. Moreover, the different relation between tyrosine phosphorylation and DNA-binding activity indicates that the mechanism leading to inhibition of IFN-α signaling in cells infected with M. tuberculosis varies with time postinfection.
Stimulation by IFN-αβ does not reproduce the inhibition of IFN-α signaling by M. tuberculosis
We next determined whether IFN-αβ alone might cause the inhibition of IFN-αβ signaling that occurs either 1 or 3 days postinfection with M. tuberculosis. We grew uninfected cells in the presence of various doses of IFN-α (0–100 U/ml) for 3 days, then measured STAT-1 DNA-binding activity without additional stimulation or after stimulation by 500 U of IFN-α/ml for 30 min (Fig. 2,A). The activity was not detected after pretreatment for 3 days at any tested dose (lanes 2–5), and pretreatment with 3 U/ml had little or no effect on the STAT-1 activity resulting from a final IFN-α stimulus (compare lanes 6 and 7). However, the STAT-1 activity present after the final stimulus was reduced by pretreatment with 10 U/ml (lane 8) and strongly or completely inhibited by pretreatment with 30 or 100 U/ml (lanes 9 and 10). We examined whether the inhibition that occurred with pretreatment was associated with decreased abundance and/or reduced tyrosine phosphorylation of STAT-1 (Fig. 2 B). Pretreatment with increasing doses of IFN-α from 3–100 U/ml led to increasing STAT-1 abundance and STAT-1 tyrosine phosphorylation (lanes 2–5). Cells stimulated by 500 U of IFN-α/ml after pretreatment with 30 or 100 U/ml contained less tyrosine-phosphorylated STAT-1 than cells that were not pretreated or that were pretreated with lower does (lanes 6–10). Pretreatment for 1 day with IFN-α or IFN-β reproduced the dose response for each of these effects (data not shown). The strong and partial inhibitions, respectively, of IFN-α-stimulated STAT-1 DNA-binding activity and tyrosine phosphorylation after IFN-αβ pretreatment were similar to the inhibition occurring 1 day postinfection, but were distinct from those occurring after 3 d; thus, the inhibition of signaling 3 days postinfection apparently involves another substance.
LPS causes inhibition of IFN-α signaling that depends on secreted IFN-αβ, but differs from inhibition caused by M. tuberculosis
We questioned whether inhibition of IFN-α signaling after infection with M. tuberculosis would also be distinct from inhibition that might occur in cells exposed to another inducer of IFN-αβ secretion. Recent studies have demonstrated that cells stimulated by LPS secrete IFN-αβ, which, in turn, stimulates STAT-1 activation (41, 42); therefore, we examined the effect of LPS on IFN-α signaling to address this question. THP-1 cells stimulated by a dose of LPS sufficient to strongly inhibit IFN-α-stimulation of STAT-1 DNA-binding activity 1 day later were fully responsive to IFN-α 3 days later (Fig. 3,A, compare lanes 5 and 6 to lanes 13 and 14). Although this effect differed from that of infection, the inhibition that did occur was nonetheless dependent on secreted IFN-αβ, as shown by the lack of inhibition from stimulation by LPS for 1 day in the presence of anti-IFN-αβ Abs (compare lanes 7 and 8). Another difference from the effect of infection was that little or no inhibition of IFN-α-stimulated STAT-1 tyrosine phosphorylation occurred (Fig. 3 B, compare lanes 5 and 6 and lanes 13 and 14). Thus, inhibition of IFN-α signaling by M. tuberculosis is distinct from inhibition caused by LPS despite the requirement for secreted IFN-αβ in both cases.
CM from infected cells does not reproduce the inhibition of IFN-α signaling by M. tuberculosis
To corroborate the distinctive effect of M. tuberculosis on inhibition of IFN-α signaling and begin to characterize whatever acts along with secreted IFN-αβ to cause the down-regulation, we cultured uninfected cells for 1 or 3 days in CM that had been collected in parallel from uninfected and infected cells 1 or 3 days postinfection. The cells in CM were not additionally treated or were stimulated with IFN-α for the final 30 min before harvest, then STAT-1 DNA-binding activity, abundance, and tyrosine phosphorylation were measured (Fig. 4 and data not shown). Regardless of whether cells were grown in CM from uninfected or infected cells, no DNA-binding activity was detected without IFN-α stimulation (Fig. 4 A, lanes 1, 2, 5, and 6). IFN-α stimulation of STAT-1 DNA-binding activity was inhibited by growth in CM from cells infected for 1 day, but, surprisingly, not by growth in CM from cells infected for 3 days, compared with growth in CM from parallel cultures of uninfected cells (compare lanes 3 and 4 and lanes 7 and 8). As expected, addition of neutralizing anti-IFN-αβ Ab to the CM from cells infected for 1 day prevented it from inhibiting IFN-α stimulation of STAT-1 DNA-binding activity (data not shown). However, growing cells for 3 days in CM collected 1 day postinfection also failed to inhibit IFN-α stimulation of STAT-1 DNA-binding activity (data not shown).
We confirmed that even small amounts of IFN-αβ secreted during infection would maintain activity for at least 4 days. CM collected 3 days postinfection was kept at 37°C in a 5% CO2 atmosphere without or with addition of various doses of IFN-α or IFN-β (1–100 U/ml) 3 days or immediately before the CM was added to unstimulated THP-1 cells. The next day, cells that had been pretreated with the CM lacking or containing exogenous IFN-α or IFN-β and cells that were not incubated in CM received no additional stimulation or were stimulated by 500 U/ml IFN-α for 30 min. With this approach, we found that preincubation failed to alter the dose response for the effect of pretreatment on the final IFN-α-stimulated STAT-1 DNA-binding activity (data not shown). Thus, when CM fails to inhibit IFN-α signaling, the failure is not due to lability of IFN-αβ.
We next characterized whether the abundance of STAT-1 or its tyrosine phosphorylation was affected by growth in CM from infected cells. Cells grown in CM obtained 1 or 3 days postinfection contained more STAT-1 than cells grown in CM from uninfected cells (Fig. 4 B, compare lanes 1 and 2, lanes 3 and 4, lanes 5 and 6, and lanes 7 and 8). As expected, the abundance was unchanged by stimulation for 30 min with exogenous IFN-α (compare lanes 2 and 4 and lanes 7 and 8). Cells contained little or no tyrosine-phosphorylated STAT-1 when grown in CM from uninfected or infected cells (lanes 1, 2, 5, and 6). IFN-α-stimulated STAT-1 tyrosine phosphorylation was moderately inhibited by growth in CM collected 1 day postinfection, whereas growth in CM collected 3 days postinfection failed to inhibit, compared with the effect of growth in CM collected in parallel from uninfected cells (compare lanes 3 and 4 and lanes 7 and 8). Thus, growth in CM from cells infected for 1 day recapitulates the induction of STAT-1 protein and the partial inhibition of IFN-α-stimulated STAT-1 tyrosine phosphorylation that occurs at the respective time postinfection, as does induction of STAT-1 protein by CM collected 3 days postinfection. However, the failure of CM collected 3 days postinfection to inhibit IFN-α-stimulated STAT-1 tyrosine phosphorylation contrasts markedly with the strong inhibition 3 days postinfection. Together, these differences between cells grown in CM collected 1 and 3 days postinfection and also between cells grown for 1 and 3 days in CM collected 1 day postinfection suggest that a labile substance in the CM is required to inhibit IFN-α signaling.
IL-10 and LAM were candidates for additional secreted substances involved in inhibition of IFN-α signaling. However, neutralizing IL-10R during infection, including IL-10 during pretreatment with IFN-α, and adding LAM during culture of THP-1 cells in CM from cells infected for 3 days failed to alter the response to final stimulation by IFN-α that was otherwise obtained (data not shown). Thus, neither IL-10 nor LAM appears to contribute to the negative feedback regulation that we observed.
Inhibition of IFN-α signaling in cells infected with M. tuberculosis occurs in the cytoplasm
Having determined that IFN-α secreted during infection is necessary, but not sufficient, for inhibition of IFN-α signaling in cells infected with M. tuberculosis, we next considered the intracellular mechanisms involved in the inhibition. We further studied cells infected for 3 days for this purpose, because at that time the ability to stimulate both STAT-1 tyrosine phosphorylation and DNA-binding activity is inhibited. To determine whether inhibition reflects a failure to generate STAT-1 DNA-binding activity or an enhanced loss of activity after STAT-1 translocates to the nucleus, we compared cytoplasmic and nuclear extracts prepared from the same cells (Fig. 5,A and data not shown). Without IFN-α stimulation, neither uninfected nor infected cells contained activity in the cytoplasm (lanes 1 and 2). Upon stimulation with IFN-α, uninfected cells, but not infected cells, contained cytoplasmic STAT-1 DNA-binding activity (compare lanes 3 and 4). Total STAT-1 and tyrosine phosphorylation increased in cytoplasmic extracts from infected cells (Fig. 5,B, lane 2). Cytoplasmic extracts from uninfected cells had abundant IFN-α-stimulated tyrosine phosphorylation of STAT-1, which was inhibited by infection (compare lanes 3 and 4). The nuclear extracts from these cells produced equivalent results for all these measurements (data not shown), in accord with previous analyses (Fig. 1) (9). Thus, inhibition of signaling occurs in the cytoplasm.
Cell surface IFN-α receptor is not reduced in cells infected with M. tuberculosis
Before investigating the molecular interactions involved in signaling, we examined down-regulation of cell surface receptor abundance, the farthest upstream event that might alter IFN-α signaling in cells infected with M. tuberculosis. We tested for this possibility by staining uninfected and infected cells that were unstimulated or stimulated by IFN-α with anti-receptor Ab (Fig. 6,A). Cells infected with M. tuberculosis had equal or greater levels of receptor as uninfected cells whether the cells were unstimulated or stimulated by IFN-α (compare upper and lower panels). Moreover, IFN-α stimulation failed to decrease the cell surface receptor on infected cells. As expected, cell surface receptor did decrease upon IFN-α stimulation of uninfected cells. Additional quantification of these changes in cell surface IFNAR1 are shown in Fig. 6 B. Together, the data indicate that inhibition of IFN-α signaling in cells infected with M. tuberculosis is not due to a reduction in the level of cell surface receptor.
Infection by M. tuberculosis increases association of PTP-1c with IFN-α receptor and TYK-2
We next investigated whether the inhibition of IFN-α-stimulated signal transduction in cells infected with M. tuberculosis involved PTP-1c because of reports on the role of PTP-1c in IFN-α-stimulated signal transduction (21, 39) and on the ability of M. tuberculosis LAM to activate PTP-1c and foster its association with cellular membranes (22). PTP-1c was recovered from extracts of unstimulated, infected cells upon incubation with the recombinant cytoplasmic domain of IFNAR1 (Fig. 7,A, upper panel, lane 5); less was recovered from IFN-α-stimulated, infected cells (lane 6), and none was recovered from uninfected cells whether unstimulated or IFN-α-stimulated (lanes 3 and 4). As judged by coimmunoprecipitation, PTP-1c also associated with TYK-2 in extracts from infected cells that had not been stimulated by IFN-α (Fig. 7,B, upper panel, lane 5), but not in extracts from IFN-α-stimulated, infected cells (lane 6) or in extracts from uninfected cells (lanes 3 and 4). The abundance of PTP-1c was unchanged by infection and/or stimulation (Fig. 7,B, lower panel). PTP-1c was not recovered with recombinant GST moiety lacking the IFNAR1 domain (Fig. 7,A, lanes 1 and 2), and neither PTP-1c nor TYK-2 was recovered in immunoprecipitates prepared with isotype control Ab (Fig. 7 B, lanes 1 and 2). The increased association of PTP-1c with IFNAR1 and TYK2 in cells infected with M. tuberculosis before stimulation with IFN-α is consistent with inhibition of IFN-α signaling.
Infection with M. tuberculosis reduces IFN-α-stimulated tyrosine phosphorylation of TYK-2 and abundance of JAK-1
To assess possible functional consequences from the interaction of PTP-1c with the IFN-α receptor and TYK-2 in cells infected with M. tuberculosis, we first examined the tyrosine phosphorylation and abundance of TYK-2. TYK-2 was immunoprecipitated from cells that were uninfected or infected with M. tuberculosis and unstimulated or stimulated by IFN-α (Fig. 8,A). IFN-α stimulated tyrosine phosphorylation of TYK-2 only in uninfected cells (upper panel, lane 3), although the amount of TYK-2 recovered was comparable in each case (lower panel, lanes 2–5). The interaction of TYK-2 with the IFNAR1 chain of the receptor was also unchanged by infection and/or stimulation (data not shown). For comparison, JAK-1 was immunoprecipitated (Fig. 8 B). Tyrosine phosphorylation of JAK-1 (upper panel) occurred constitutively in uninfected cells and increased in cells stimulated by IFN-α (lanes 2 and 3), but was almost undetectable in cells infected with M. tuberculosis whether unstimulated or stimulated by IFN-α (lanes 4 and 5). The decrease in JAK-1 tyrosine phosphorylation in cells infected with M. tuberculosis correlated with a decrease in JAK-1 abundance (lower panel, compare lanes 2 and 3 and lanes 4 and 5). Thus, tyrosine-phosphorylated TYK-2 and JAK-1 were reduced in cells infected with M. tuberculosis and then stimulated by IFN-α, although by apparently different mechanisms. Infection limits IFN-α-stimulated tyrosine phosphorylation of TYK-2 without affecting its abundance, whereas the amount of tyrosine-phosphorylated JAK-1 is limited by the decrease in its abundance.
Infection with M. tuberculosis reduces IFN-α-stimulated TYK-2 kinase activity
We tested whether inhibition of IFN-α-stimulated TYK-2 kinase activity would occur in cells infected with M. tuberculosis, given the observed inhibition of TYK-2 tyrosine phosphorylation. TYK-2 was immunoprecipitated from extracts of uninfected and infected cells that had been unstimulated or stimulated by IFN-α. The recovered material was supplemented with the recombinant cytoplasmic domain of IFNAR1 and incubated with [γ-32P]ATP, then incorporation of 32P into TYK-2, STAT-1 that was coimmunoprecipitated, and the added IFNAR1 domain was determined (Fig. 8 C). The identity of the indicated proteins was confirmed by immunoblot, which also demonstrated the recovery of TYK-2 by immunoprecipitation (lower panel). Comparing total incorporation of 32P to recovery of TYK-2 indicates that a background incorporation of 32P occurred using isotype control Ab for immunoprecipitation (lanes 1 and 2) and that activity recovered from uninfected cells was low relative to the amount of TYK-2 (lane 3). Although less TYK-2 was recovered from infected cells, it had similar kinase activity, indicating activation (lane 4). TYK-2 kinase activity recovered from cells stimulated by IFN-α was greater, although the recovery of TYK-2 was still less than that from unstimulated cells (lane 5). Infection by M. tuberculosis inhibited the IFN-α-stimulated increase in activity, as judged by the incorporation of 32P relative to the recovery of TYK-2 (compare lanes 6 and 5). The inhibition of IFN-α-stimulated TYK-2 kinase activity is consistent with inhibition of its tyrosine phosphorylation and enhanced association with PTP-1c.
We have examined the autocrine/paracrine and intracellular mechanisms involved in the inhibition of IFN-αβ signaling during infection with M. tuberculosis. This study reveals an unexpected dynamic host response that governs the ability of cells to respond to secreted IFN-αβ during the course of infection. The inhibition includes a negative feedback response to IFN-αβ secreted within the first day postinfection, but by 3 days postinfection, the inhibition is distinct from a general effect that is solely due to stimulation by IFN-αβ. Analyzing inhibition of IFN-α signaling in cells that are 1) infected with M. tuberculosis, 2) stimulated by LPS, 3) subject to prior stimulation by exogenous IFN-αβ, and 4) grown in CM from infected cells demonstrates that that the inhibition due to infection is distinct from other modes of negative feedback regulation involving IFN-αβ. Because IFN-αβ is necessary, but not sufficient, to cause the inhibition of IFN-α signaling that occurs during infection, we conclude that another stimulus is also required. Direct confirmation of its function will require its identification, which will be hindered by its apparent lability. We identified enhanced association of PTP-1c with IFNAR1 and TYK-2 and decreased abundance of JAK-1 as the likely intracellular mechanism(s) for the inhibition of IFN-α signaling at 3 days postinfection and discounted other possibilities. These results can account for the inhibition of IFN-αβ-stimulated STAT-1 activation that occurs without affecting STAT-1 activation stimulated by IFN-γ during infection with M. tuberculosis. The data also demonstrate a previously unrecognized link between independent studies of PTP-1c in host response to M. tuberculosis (22) and in IFN-αβ signaling (21), and they characterize a novel combination of intracellular events associated with the distinct inhibition of IFN-α signaling by M. tuberculosis.
Negative feedback regulation could be mediated by a reduced level of cell surface receptor, but it was unaffected by infection, and surprisingly, typical IFN-α-stimulated down-regulation was inhibited by infection. The C-terminal 16-aa domain of IFNAR1, which has the most highly conserved tyrosine-containing motifs, is critical for IFN-α-stimulated down-regulation of the IFN-αβ receptor (43). Independent studies have described a role for TYK-2 in normal cell surface expression of the IFN-αβ receptor (17). Additional study is required to determine whether the failure to down-regulate the cell surface receptor is the consequence of a failure to initiate signaling that affects the function of the C-terminal motif or the contribution of TYK-2 to cell surface receptor expression.
Our data identify changes in JAK-1 abundance and TYK-2 activation as mechanisms that can account for the inhibition of IFN-α-stimulated signaling in cells infected with M. tuberculosis. If SOCS proteins were induced, they could also inhibit STAT tyrosine phosphorylation (26, 27). Although we did not obtain consistent evidence for an increase in SOCS-1 or -3 protein, transcription of the genes was increased 2- to 3-fold by infection (S. Prabhakar and R. Pine, unpublished observations). Therefore, a role for induction of SOCS genes remains possible. We have not investigated whether the secreted protein tyrosine phosphatases of M. tuberculosis (44, 45) contribute to inhibition of IFN-α signaling at either 1 or 3 days postinfection. The decreased abundance of JAK-1 may be an important contributor to the down-regulation of IFN-α-stimulated signaling, because in other circumstances lack of JAK-1 has a greater negative impact on IFN-α signaling than lack of TYK-2 (46, 47, 48, 49, 50). However, the association of PTP-1c with the α-chain of the IFN-α receptor and with TYK-2 and the associated inhibition of TYK-2 activation and its resultant kinase activity is likely to contribute to inhibition of IFN-α-stimulated signaling by infection. Initiation of IFN-α signaling has been associated with a transient decrease in the interaction of PTP-1c with IFNAR1, JAK-1, and TYK-2, and lack of PTP-1c was reported to enhance tyrosine phosphorylation of JAK-1 and STAT-1, but not TYK-2, in cells stimulated by IFN-αβ (21). The data presented in this study demonstrate distinct roles for modulation of JAK-1 abundance and for PTP-1c interaction with TYK-2 in negative feedback regulation of IFN-αβ signaling that occurs in the cytoplasm due to infection with M. tuberculosis, apparently independent of down-regulation involving nuclear tyrosine phosphatases that inactivate STAT-1 (19, 20, 29).
Although we have more fully characterized the inhibition of IFN-α signaling at 3 days than at 1 day postinfection, the basis for the greater reduction in STAT-1 DNA-binding activity than in tyrosine phosphorylation observed in cells stimulated by IFN-α 1 day postinfection is also of interest. Reduction in STAT-1 tyrosine phosphorylation is expected to cause a greater reduction in its DNA-binding activity because the ability to form homodimers that bind DNA will be exponentially decreased. Notably, a mutation in STAT-1 that causes increased human susceptibility to mycobacterial infection acts as a dominant negative allele that causes a greater decrease in STAT-1 DNA-binding activity than in tyrosine phosphorylation in heterozygous cells stimulated by IFN-α or IFN-γ (51). Nonetheless, we considered whether additional mechanisms contributed to the proportionally greater decrease in DNA-binding activity than in tyrosine phosphorylation that occurs 1 day postinfection. Failure to detect induction of PIAS-1 or any other inhibitor of STAT-1 DNA binding suggests that an inherent difference in STAT-1 due to infection accounts for the different extents of inhibition. One such modification, a decrease in STAT-1 arginine methylation due to infection, which would increase interaction with PIAS-1 (52), could involve an amount of PIAS-1 that is below the detection limit and that is not increased by infection. Although also speculative with regard to STAT-1, acetylation is a possibility that should not be overlooked in light of its known interactions with acetyltransferases (53, 54) and the increasing appreciation for regulation of transcription factors by acetylation (reviewed in Refs.55 and 56).
The production of and response to IFN-αβ may be especially important during initial stages of bacterial infection, including infection with M. tuberculosis, both for the infected cells and as a determinant of adaptive immune response (2, 6, 8, 9). Following the course of infection through 3 days revealed negative feedback regulation, suggesting that the response to the secreted IFN-αβ is not as great as it would be otherwise. Variations in the amount of IFN-αβ secreted and the duration of its presence are consequences of the specific scenario of induction that will influence the nature of the feedback inhibition, together with other influences specific to the inducing stimulus. Thus, a general mode of feedback inhibition may not exist, although common molecular mechanisms may be involved in different situations. Moreover, IFN-αβ is necessary, but not sufficient, for the greatest inhibition of IFN-α signaling during infection with M. tuberculosis; thus, its effect on this pathway is distinct from a general negative feedback due solely to the presence of IFN-αβ. Finally, we note that the association of greater virulence among M. tuberculosis strains with greater induction of IFN-α gene expression (8) could also be related to the extent of feedback inhibition, whether strictly through a dose-dependent effect or together with specific responses to M. tuberculosis that might also vary among strains.
We thank Karl Drlica and Michael Weiden for critically reading the manuscript. We thank Oscar Colamonici for providing mAbs against the IFN-α receptor and plasmids encoding GST-IFN-α receptor fusion proteins, Christian Schindler for providing recombinant STAT-1, Michael Brunda for supplying rIFN-α, and Jan Vilcek for a generous gift of neutralizing anti-IFN-αβ Abs.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by National Institutes of Health Grants AI37877 (to R.P.) and AI27742 (to the New York University School of Medicine Center for AIDS Research) and by a fellowship from the Heiser Program for Research in Leprosy and Tuberculosis (to S.P.). LAM of M. tuberculosis was obtained through National Institutes of Health, National Institute of Arthritis and Infectious Diseases, Contract 1 (AI75320) from Colorado State University.
Abbreviations used in this paper: ISGF-3, IFN-α-stimulated gene factor 3; CM, conditioned medium; GAS, IFN-γ activation site; IFNAR1, IFNα receptor α-chain; IFNAR2, IFNα receptor β-chain; LAM, lipoarabinomannan; PIAS, protein inhibitor of activated STAT; PTP, protein tyrosine phosphatase; PVDF, polyvinylidene difluoride; SOCS, suppressor of cytokine signaling; TYK, tyrosine kinase.