ABSTRACT
Haemophilus ducreyi causes chancroid, a genital ulcer disease. In human inoculation experiments, most volunteers fail to clear the bacteria despite the infiltration of innate and adaptive immune cells to the infected sites. The immunosuppressive protein indoleamine 2,3-dioxygenase (IDO) is a rate-limiting enzyme in the l-tryptophan-kynurenine metabolic pathway. Tryptophan depletion and tryptophan metabolites contribute to pathogen persistence by inhibiting T cell proliferation, inducing T cell apoptosis, and promoting the expansion of FOXP3+ regulatory T (Treg) cells. We previously found that FOXP3+ Treg cells are enriched in experimental lesions and that H. ducreyi induced IDO transcription in dendritic cells (DC) derived from blood of infected volunteers who developed pustules. Here, we showed that enzymatically active IDO was induced in DC by H. ducreyi. Neutralizing antibodies against interferon alpha/beta receptor 2 chain (IFNAR2) and tumor necrosis factor alpha (TNF-α) inhibited IDO induction. Inhibitors of the mitogen-activated protein kinase (MAPK) p38 and nuclear factor-κB (NF-κB) also inhibited IDO expression. Neither bacterial contact with nor uptake by DC was required for IDO activation. H. ducreyi culture supernatant and H. ducreyi lipooligosaccharides (LOS) induced IDO expression, which required type I interferons, TNF-α, and the three MAPK (p38, c-Jun N-terminal kinase, and extracellular signal regulated kinase) and NF-κB pathways. In addition, LOS-induced IFN-β activated the JAK-STAT pathway. Blocking the LOS/Toll-like receptor 4 (TLR4) signaling pathway greatly reduced H. ducreyi-induced IDO production. These findings indicate that H. ducreyi-induced IDO expression in DC is largely mediated by LOS via type I interferon- and TNF-α-dependent mechanisms and the MAPK, NF-κB, and JAK-STAT pathways.
INTRODUCTION
The Gram-negative bacterium Haemophilus ducreyi is a strict human pathogen that causes chancroid, a sexually transmitted genital ulcer disease that facilitates the acquisition and transmission of HIV-1 (48). H. ducreyi also causes a chronic limb ulceration syndrome that does not appear to be sexually transmitted (37, 41, 54). To study the immunopathogenesis of H. ducreyi infection, we developed a human challenge model in which healthy adult volunteers were inoculated on the skin of the upper arm with H. ducreyi strain 35000HP (where HP indicates human passaged) or its derivatives (25). Within 24 h of experimental infection, papules formed at infected sites and evolved into pustules within 2 to 5 days, mimicking the early stages of natural infection.
Despite the infiltration of infected sites by several types of innate and adaptive immune cells such as neutrophils, macrophages, myeloid dendritic cells (DC), NK cells, and memory/effector T cells (6, 32, 49), H. ducreyi replicates and persists extracellularly (8, 9). Recently, we reported that the CD4+ FOXP3+ regulatory T (Treg) cells were enriched in experimental pustules and that Treg cells suppress anti-H. ducreyi CD4 T cell responses (33). Treg cells at the infected sites could be composed of either naturally occurring Treg cells, which are generated in the thymus, or inducible Treg cells that are converted from CD4+ CD25− effector T cells at peripheral sites under immunosuppressive conditions.
Human DC expressing the immunosuppressive enzyme indoleamine 2,3-dioxygenase (IDO) induce the conversion of effector T cells to FOXP3+ Treg cells (12, 13, 22, 36). IDO is an intracellular heme-containing protein and is the rate-limiting enzyme in the pathway that degrades the essential amino acid l-tryptophan to generate several biologically active metabolites known as kynurenines. In addition to its role in expanding Treg cells, IDO inhibits T cell activation/proliferation and promotes T cell death through tryptophan depletion and the production of proapoptotic metabolites. This suppression of T cell responses by IDO promotes immune tolerance in pregnancy, autoimmune diseases, organ transplantation, neoplasia, and chronic infection (39, 43, 53, 56).
IDO expression is induced in DC and several other cell types under various physiological conditions, such as inflammation induced by viral and bacterial infections (56). Many soluble and membrane-bound factors mediate IDO induction, mostly through pathways involving type II interferon (IFN-γ) or type I interferons (IFN-α and IFN-β) (43, 56). In addition, microbial components such as lipopolysaccharide (LPS) and proinflammatory mediators such as tumor necrosis factor alpha (TNF-α) activate IDO through interferon-independent mechanisms or synergistically enhance IFN-γ-mediated signaling (19, 26, 45). Interferon-dependent activation of IDO is mediated by the JAK-STAT (Janus kinase-signal transducer and activator of transcription) signaling pathways, whereas interferon-independent induction is mediated by the p38 and JNK mitogen-activated protein kinase (MAPK), phosphatidylinositol 3-kinase (PI3K), and nuclear factor-κB (NF-κB) pathways (19, 26).
We previously reported that myeloid DC are enriched relative to plasmacytoid DC in lesions of experimentally infected volunteers (6). We also reported that monocyte-derived DC from volunteers who develop pustules after inoculation with H. ducreyi express high levels of IDO transcripts (24). In this study, we investigated the mechanisms by which H. ducreyi induces DC to express IDO. Our data demonstrate that H. ducreyi and its lipooligosaccharide (LOS) induced IDO activation in DC through type I interferons and TNF-α and through the MAPK, NK-κB, and JAK-STAT pathways but not through IFN-γ-mediated signals. We propose that H. ducreyi-induced IDO expression in DC may contribute to bacterial persistence through the suppression of anti-H. ducreyi immune responses.
MATERIALS AND METHODS
Bacterial growth conditions.H. ducreyi strain 35000HP was grown on chocolate agar plates and GC medium broth as described previously (5, 25). Bacteria were grown to mid-log phase and washed three times with phosphate-buffered saline (PBS) before use for infection of DC. To obtain heat-killed H. ducreyi, washed bacteria were suspended in DC growth medium (see below) and incubated at 65°C for 60 min.
Preparation of H. ducreyi culture supernatant and LOS.To prepare a cell-free culture supernatant, an overnight broth culture of H. ducreyi was filtered through a 0.22-μm-pore-size filter and stored at −20°C. LOS was prepared from H. ducreyi as described previously (11) with some minor modifications. Briefly, the bacteria were grown to mid-log phase, washed with PBS, sonicated in a solution containing 50 mM NaH2PO4 and 5 mM EDTA, treated with lysozyme, DNase I (30 μg/ml), and RNase A (30 μg/ml) in a buffer with 50 mM NaH2PO4 and 15 mM MgCl2, and subsequently treated with proteinase K. The treated cell lysates were subjected to microphenol extraction. LOS was precipitated with NaCl and ethanol and suspended in deionized water.
Generation and stimulation of DC.Peripheral blood mononuclear cells (PBMC) were isolated by Ficoll-Paque Plus gradient centrifugation of leukopacks derived from 29 anonymous donors (the Central Indiana Regional Blood Center) or from blood drawn from three healthy uninfected adult volunteers. Informed consent was obtained from the volunteers for participation, in accordance with the human experimentation guidelines of the U.S. Department of Health and Human Services and the Institutional Review Board of Indiana University-Purdue University at Indianapolis.
CD14+ monocytes were isolated from PBMC by positive selection using magnetic CD14 microbeads (Miltenyi Biotech) according to the manufacturer's instructions. DC were differentiated from either freshly isolated or frozen CD14+ cells in RPMI 1640 medium supplemented with 2 mM l-glutamine, 50 μM beta-mercaptoethanol, 10% heat-inactivated fetal bovine serum, and 5 ng/ml of recombinant human granulocyte-macrophage colony-stimulating factor (GM-CSF) and interleukin-4 (IL-4; R&D Systems) as described previously (6). Cultured cells were washed and used as immature DC (iDC), which consisted of approximately 95% CD1a+ cells, <5% CD14+ cells, and <1% CD3+ T cells, as assessed by flow cytometry using antibodies against CD1a (HI149; BD Biosciences), CD14 (61D3; eBioscience), and CD3 (SK7; eBioscienc). Heat-killed or live bacteria at multiplicities of infection (MOI) ranging from 0.003:1 to 10:1 were centrifuged onto wells containing 1.5 × 105 iDC in 96-well flat bottom plates. After 90 min of incubation at 35°C, cells were further incubated at 37°C for various periods of time.
For antibody blocking experiments, DC were preincubated with 10 μg/ml of neutralizing antibodies against IFN-α/β receptor chain 2 (IFNAR2) (murine IgG2a [mIgG2a], clone MMHAR2; Millipore), IFN-γ (mIgG1, clone NIB42; eBioscience), TNF-α (mIgG1, clone MAb1; eBioscience), Toll-like receptor 4 (TLR4) (mIgG2a, clone HTA125; eBioscience), or isotype-matched control antibodies mIgG2a and mIgG1 (eBioscience) for 60 min at 37°C before addition of live (MOI of 0.003:1) or heat-killed (MOI of 0.1:1 to 1:1) H. ducreyi or LOS at a final concentration of 1 to 2 ng/ml. For some experiments, DC were preincubated for 60 min at 37°C with 5 to 10 μM MEK (extracellular signal-regulated kinase [ERK] kinase) inhibitor U0126 (Cell Signaling), 10 μM p38 inhibitor SB203580 (Calbiochem), 10 μM JNK inhibitor SP600125 (Calbiochem), 0.3 μM PI3K inhibitor wortmannin, 20 to 40 μM NF-κB inhibitor NF-κB SN50 (Calbiochem), or 5 to 10 μM actin polymerization inhibitor cytochalasin D (Calbiochem). Dimethyl sulfoxide (DSMO), which was used to dissolve the MAPK inhibitors and cytochalasin D, served as a vehicle control; the inactive control peptide NF-κB SN50M (Calbiochem) was used as a negative control for NF-κB SN50. DC were subsequently infected with live or heat-killed H. ducreyi at an MOI of 3:1 or 1:1, respectively. For transwell experiments, DC were seeded in 24-well tissue culture plates. An insert with a 0.2-μm-pore-size membrane (Nalgene Nunc) was placed in the well, and live H. ducreyi (MOI of 3:1) was added to the top chamber.
To block H. ducreyi LOS-induced activation of DC, 10 and 30 μg/ml polymyxin B (InvivoGen) was preincubated with 3 μl of cell-free overnight bacterial culture supernatant, 0.3 to 1 ng/ml purified LOS, and heat-killed organisms (MOI of 0.1:1 to 1:1) at 37°C for 60 min before coculturing with 1 × 105 DC.
Western blot analysis.DC were washed with PBS, lysed in Laemmli buffer, sonicated, and boiled. Cell lysates were electrophoresed in 12.5% acrylamide gels and transferred to nitrocellulose membranes. To detect IDO protein expression, DC were activated with H. ducreyi, bacterial culture supernatants, or LOS for 24 h, unless otherwise indicated on the figures or in the legends, and the membranes were probed with mouse monoclonal antibodies to IDO (clone 10.1; Millipore). To ensure that equal amounts of cellular proteins were analyzed, the mouse monoclonal antibody 6C5 (Abcam) was used to detect glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The level of expression of IDO normalized to GAPDH was determined by densitometry using Adobe Photoshop CS4 (Adobe Systems Inc.). To examine the activation of MAPK, PI3K, NF-κB, and JAK-STAT pathways, cell lysates were prepared from DC infected with H. ducreyi at an MOI of 10:1 and probed with specific antibodies against individual components of these pathways (all antibodies were obtained from Cell Signaling). The active forms of the three MAPKs (p38, JNK, and ERK), Akt (a downstream target of PI3K), the p65 and IkBα subunits of NF-κB complex, and STAT1 were detected with rabbit monoclonal antibodies to phosphorylated p38 (Thr180/Tyr1824, clone 12F8), JNK (Thr183/Tyr185, clone 98F2), ERK (Thr202/Tyr204, clone 20G11), Akt (Ser473, clone D9E), IκBα (Ser32, clone 14D4), NF-κB p65 (Ser536, clone 93H1), and STAT1 (Tyr701, clone 58D6). Total amounts of MAPK, Akt, and NF-κB proteins were analyzed with rabbit monoclonal antibodies to JNK (clone 56G8), Akt (clone C67E7), NF-κB p65 (clone C22B4), rabbit polyclonal antibodies to p38 and ERK, and mouse monoclonal antibody to IκBα (clone L35A5).
Measurement of IDO enzymatic activity.IDO enzymatic activity was assessed by measuring the stable tryptophan metabolite kynurenine as described previously with minor modifications (13). Briefly, iDC were incubated with live (MOI of 3:1) or heat-killed (MOI of 1:1) H. ducreyi for 24 h, washed three times with Hanks' balanced salt solution (HBSS), and incubated in HBSS with 100 μM tryptophan at a cell concentration of 3 × 106/ml at 37°C for 4 h. The culture supernatants were collected, mixed with 30% trichloroacetic acid (2:1) by vortexing, and spun at 4°C in a microcentrifuge at maximal speed for 10 min. The amount of kynurenine in the treated culture supernatants was measured by a spectrophotometric assay with Ehrlich reagent (Sigma-Aldrich) with a Bio-Kinetics EL 312e MicroPlate Reader (Bio Tek) at 490 nm. Purified kynurenine (Sigma-Aldrich) served as a standard.
Cytokine assays.Culture supernatants were collected and frozen at −20°C. IFN-β and TNF-α levels were measured by cytokine enzyme-linked immunosorbent assay (ELISA) kits from R&D Systems and BD Biosciences, respectively, according to the manufacturer's instructions.
Statistical analysis.Paired t tests were used to analyze data unless otherwise indicated. We report the nominal P value for each comparison without adjustment for multiple testing. Thus, these results should be interpreted cautiously due to possible inflation of type 1 error. A P value that is less than or equal to 0.05 was considered statistically significant.
RESULTS
H. ducreyi induces IDO protein expression and enzymatic activity in DC.We previously reported that H. ducreyi induced the expression of IDO transcripts in DC derived from blood of experimentally infected volunteers who developed pustules (24). To confirm that H. ducreyi induces the expression of IDO protein in DC, monocyte-derived DC were incubated with live and heat-killed bacteria for various periods of time and at different MOI. iDC did not express detectable levels of IDO protein; treatment with both live and heat-killed H. ducreyi led to a time-dependent increase in IDO production, which peaked at 18 to 24 h (Fig. 1A and B). Heat-killed H. ducreyi induced IDO protein expression in a dose-dependent manner, which saturated at MOI of 0.3:1 to 10:1 (Fig. 1B). Live H. ducreyi induced IDO over a wide range of MOI (0.003:1 to 10:1) that did not appear to be dose dependent (Fig. 1A and data not shown). However, live H. ducreyi replicated in the DC cocultures, resulting in an MOI of 3:1 to 10:1 at the end of the culture period for all doses tested (data not shown), which likely explains the lack of a dose response to live bacteria.
H. ducreyi-induced IDO expression in DC. IDO protein production by DC exposed to live (A) and heat-killed (B) H. ducreyi. Shown are representative Western blots of samples obtained from three donors. DC were incubated with medium (med) or with live or heat-killed H. ducreyi at MOI of 3:1 and 1:1, respectively, in the time course studies (left panels) and at various MOI for 24 h in the dose-response studies (right panels). (C) Induction of IDO enzymatic activity. DC were incubated with medium or with live or heat-killed H. ducreyi for 24 h, washed, and cultured in HBSS containing 100 μM tryptophan for 4 h. Supernatants were assessed for kynurenine production with the Ehrlich reagent. Bars represent the mean ± SD of results from six donors. *, P ≤ 0.05 compared with the medium-treated DC group (the Wilcoxon signed rank tests).
To determine whether H. ducreyi could induce the expression of enzymatically active IDO, DC were incubated with both live and heat-killed H. ducreyi for 24 h to induce IDO protein expression and cultured further with medium containing tryptophan for 4 h. The culture supernatants were then assessed for the level of l-kynurenine, a stable product of IDO-mediated tryptophan metabolism. Little l-kynurenine was detected in the supernatant from iDC. However, DC activated by live or heat-killed bacteria generated a significant amount of l-kynurenine (Fig. 1C). Therefore, H. ducreyi infection of DC leads to the production of active IDO enzyme, and heat-stable components of H. ducreyi were sufficient to induce this activation.
Type I interferons and TNF-α induced by H. ducreyi play a critical role in IDO expression.IFN-γ and type I interferons are potent inducers of IDO in many cell types, and DC infected by H. ducreyi release TNF-α, another IDO-inducing cytokine (6, 24). Therefore, we examined the possible involvement of those cytokines in IDO expression induced by both live and heat-killed H. ducreyi using neutralizing antibodies against IFN-γ, type I interferon receptor IFNAR2, and TNF-α. A representative result is shown in Fig. 2A. Compared to their respective isotype-matched control antibodies, anti-IFNAR2 or anti-TNF-α antibodies significantly inhibited IDO production induced by heat-killed, but not by live, H. ducreyi. However, the combination of anti-IFNAR2 and anti-TNF-α antibodies significantly reduced IDO expression by both live and heat-killed H. ducreyi and nearly completely blocked IDO production induced by heat-killed organisms. In contrast, anti-IFN-γ antibodies had no effect on IDO expression, either alone (Fig. 2A) or in combination with anti-TNF-α antibodies (data not shown).
Critical role of type I interferons and TNF-α in H. ducreyi-induced IDO production. (A) DC were incubated with medium (med) and with live or heat-killed H. ducreyi in the presence of neutralizing antibodies to IFN-γ, IFNAR2, TNF-α, IFNAR2 and TNF-α, or isotype-matched control antibodies mIgG1 and mIgG2a. Top panels are representative Western blots, and bottom panels are densitometry data obtained from at least four individual donors. The ratio of IDO/GAPDH in each sample was normalized to that of DC incubated with H. ducreyi alone, which was set to a value of 100, and was expressed as a mean ± SD. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001. α, anti. (B) Time course of IFN-β and TNF-α production by DC stimulated with live or heat-killed H. ducreyi. Data represent results from multiple measurements made in one of two independent experiments.
To confirm the role of type I interferons and TNF-α in H. ducreyi-induced IDO activation, culture supernatants from H. ducreyi-treated DC were assessed for the production of IFN-α, IFN-β, and TNF-α. While the IFN-α level was below the limits of detection (data not shown), both live and heat-killed H. ducreyi induced substantial amounts of IFN-β and TNF-α after 4 h, i.e., before IDO protein production was detected (Fig. 2B). The IFN-β concentration was maximal at 6 to 8 h after infection by live bacteria while IFN-β production peaked later at 18 h following treatment with heat-killed H. ducreyi. TNF-α production appeared to have a two-phase expression pattern, with an early peak at 6 h and a higher peak at 24 h (Fig. 2B). IFN-γ production was undetectable (data not shown), consistent with the results of the antibody blocking experiments. Together, these results indicated that H. ducreyi-activated DC release IFN-β and TNF-α, which act synergistically to promote IDO production in DC by an autocrine mechanism.
H. ducreyi-activated p38 and NF-κB pathways mediate IDO induction.Activation of the MAPK, PI3K, and NF-κB signal transduction pathways is critical for DC cytokine production as well as IFN-independent induction of IDO. We investigated whether H. ducreyi infection of DC activates the MAPK (p38, JNK, and ERK), PI3K, and NF-κB pathways by Western blot analysis using antibodies against phosphorylated (activated) forms of MAPKs, Akt (a downstream target of PI3K), and NF-κB subunits. The requirement of these pathways in H. ducreyi-induced expression of IDO and the IDO-inducing cytokines IFN-β and TNF-α was assessed by using specific pharmacological inhibitors.
As shown in Fig. 3A, live H. ducreyi rapidly induced the phosphorylation of p38, JNK, ERK, and Akt, which generally peaked around 45 to 60 min postinfection. Compared to the vehicle control DMSO, the p38 inhibitor SB202190 nearly completely abolished IDO induction. The JNK inhibitor SP600125 reduced IDO production to 56% ± 17% (mean ± standard deviation [SD]; n = 5), but this difference was not significant (P = 0.23). In contrast, the respective ERK and PI3K inhibitors U0126 and wortmannin had little effect on IDO induction (Fig. 3B). Next, we examined the effect of these inhibitors on the production of IFN-β and TNF-α. The p38 and JNK inhibitors significantly decreased the induction of both cytokines, and the ERK inhibitor had little effect on IFN-β production but significantly decreased TNF-α production (Fig. 3C). The PI3K inhibitor failed to block TNF-α or IFN-β production (data not shown). Thus, H. ducreyi activates the p38, JNK, ERK, and PI3K pathways, and the p38 pathway plays an essential role in the production of both IDO and the IDO-inducing cytokines IFN-β and TNF-α.
Role of MAPK and PI3K activation in the expression of H. ducreyi-induced IDO, IFN-β, and TNF-α. (A) Time course of MAPK and PI3K activation. Western blots of whole-cell lysates of DC infected with live H. ducreyi for various times (in minutes) were probed with antibodies to phosphorylated or total forms of MAPKs and Akt. (B) Effects of MAPK and PI3K inhibitors on IDO expression. DC were incubated in medium (med) or infected with live H. ducreyi in the presence of SB203580 (SB; p38 inhibitor), SP600125 (SP; JNK inhibitor), U0126 (ERK inhibitor), the PI3K inhibitor wortmannin (WM), and the vehicle control DMSO. Top panels are representative Western blots, and the bottom panel shows densitometry data obtained from five donors. Bars represent mean ± SD of the ratio of IDO/GAPDH. (C) Effects of MAPK inhibitors on the production of IFN-β and TNF-α. Culture supernatants from DC infected for 6 and 24 h were assessed for the accumulation of IFN-β (n = 6) and TNF-α (n = 5), respectively. Due to donor-to-donor variation in cytokine production, the cytokine level in each sample was normalized to that of DC incubated with H. ducreyi alone, which was set at 100%. Bars represent mean ± SD. **, P ≤ 0.01; ***, P ≤ 0.001.
In resting cells, the p50 and p65 heterodimer of the transcription factor NF-κB are sequestered in the cytoplasm by the IκB protein. NF-κB activation involves the phosphorylation and subsequent degradation of IκB, phosphorylation of p65, and nuclear translocation of the NF-κB active complex into the nucleus. As shown in Fig. 4A, infection of DC with live H. ducreyi led to an increase in the phosphorylation of IκBα and p65 within 10 min and persisted for at least 24 h. Total IκBα was degraded between 30 to 90 min and reappeared after 2 h. The involvement of the NF-κB pathway in H. ducreyi-induced IDO expression was assessed by using NF-κB SN50, a peptide that blocks the translocation of the NF-κB active complex into the nucleus. Compared to its inactive control peptide NF-κB SN50M, NF-κB SN50 significantly suppressed IDO expression induced by heat-killed H. ducreyi but not by live H. ducreyi (Fig. 4B). The effect of NF-κB SN50 on cytokine production is shown in Fig. 4C. IFN-β production induced by both live and heat-killed bacteria was almost completed abolished by NF-κB SN50. Although NF-κΒ SN50 significantly inhibited TNF-α production induced by heat-killed H. ducreyi, it failed to reduce TNF-α production induced by live organisms. Replication of live H. ducreyi in the DC cocultures might have overwhelmed the inhibitory capacity of NF-κB SN50 on the production of TNF-α and IDO. Taken together, the results indicate that the p38 and NF-κB signaling pathways are critical for H. ducreyi-induced IDO activation in DC, likely through their role in TNF-α and IFN-β production.
Role of NF-κB activation in the production of H. ducreyi-induced IDO, IFN-β, and TNF-α. (A) Time course of NF-κΒ activation. Western blots of whole-cell lysates of DC infected with live H. ducreyi for various periods of time were probed with antibodies against phosphorylated or total forms of IκBα and p65 and GAPDH. Shown are representative Western blots of samples obtained from three donors. (B) Effect of NF-κB inhibitor NF-κB SN50 on IDO production. DC were incubated in medium (med) or with live or heat-killed (HK) H. ducreyi in the presence or absence of NF-κB SN50 or its inactive control NF-κB SN50M. Top panels are representative Western blots, and bottom panels are densitometry data obtained from four donors. The ratio of IDO/GAPDH in each sample was normalized to that of DC incubated with H. ducreyi alone, which was set to a value of 100, and is expressed as a mean ± SD. (C) Effect of NF-κB SN50 on the production of IFN-β and TNF-α. In culture supernatants from DC infected with live H. ducreyi for 6 h and 24 h, the levels of IFN-β (n = 4) and TNF-α (n = 4), respectively, were measured. Supernatants from DC treated with heat-killed bacteria for 18 h were assessed for the production of IFN-β (n = 4) and TNF-α (n = 5). Due to donor-to-donor variation in cytokine production, cytokine levels were normalized to that of DC incubated with H. ducreyi alone, which was set at 100%. Bars are means ± SD. *, P ≤ 0.05; **, P ≤ 0.01.
H. ducreyi uptake is not required for IDO induction.We previously showed that iDC ingest H. ducreyi (6). Here, we investigated whether bacterial uptake by DC was necessary for IDO induction. The actin polymerization inhibitor cytochalasin D, which prevents bacterial phagocytosis, slightly but significantly inhibited H. ducreyi-induced IDO expression (Fig. 5A, left panels). To further assess the role of bacterial ingestion and to determine the requirement for a direct contact between the bacteria and DC, we used a 0.2-μm-pore-size transwell membrane to physically separate DC from H. ducreyi. IDO induction was not contact dependent (Fig. 5A, right panels), suggesting that acellular components derived from H. ducreyi were sufficient to induce IDO expression. Bacterial cell-free culture supernatants induced IDO expression; polymyxin B, which binds to the lipid A portion of LOS and neutralizes its function, completely blocked supernatant-induced IDO production (Fig. 5B, left panels), suggesting that H. ducreyi LOS is a major bacterial component for IDO induction.
H. ducreyi LOS is a major inducer of IDO. (A) Nonessential role of bacterial ingestion in IDO production. DC were cultured in medium (med) or infected with live H. ducreyi in the presence of DMSO or cytochalasin D (CD) or with live H. ducreyi separated by a 0.2-μm-pore-size transwell (TW). Western blots represent results from five cytochalasin D and six transwell experiments, respectively. Bars represent mean ± SD of the ratio of IDO/GAPDH, which was determined by densitometry. (B) Inhibitory effect of polymyxin B (PMB) on the expression of IDO induced by H. ducreyi culture supernatants (sup) and purified H. ducreyi LOS. DC were stimulated with supernatants or LOS in the presence and 0, 10, or 30 μg/ml of polymyxin B. Shown are representative Western blots of samples obtained from five donors. (C) Effect of anti-TLR4 antibody and polymyxin B on H. ducreyi-induced IDO production. DC were stimulated with heat-killed bacteria in the presence of anti-TLR4 antibody or isotype-matched control antibody (iso) (left panels) or polymyxin B (right panels). Top panels are representative Western blots, and bottom panels are densitometry data obtained from four antibody inhibition and five polymyxin B inhibition experiments. The ratio of IDO/GAPDH in each sample was normalized to that of DC incubated with H. ducreyi alone, which was set to a value of 100, and is expressed as a mean ± SD. *, P ≤ 0.05. α, anti.
H. ducreyi LOS is a potent inducer of IDO.Purified H. ducreyi LOS activated IDO expression, which was completely inhibited by polymyxin B (Fig. 5B, right panels). DC sense LOS through TLR4 complexes. Neutralizing antibodies against TLR4 significantly blocked IDO production induced by heat-killed H. ducreyi (Fig. 5C, left panels). IDO expression induced by heat-killed H. ducreyi was reduced by 59% ± 28% and 76% ± 17% with 10 and 30 μg/ml of polymyxin B, respectively (Fig. 5C, right panels). Together, the data indicated that H. ducreyi LOS is a major stimulator of IDO in DC and explains why both live and heat-killed H. ducreyi induced IDO.
LOS activates IDO via type I interferons and TNF-α and the MAPK and NF-κB pathways.We examined whether H. ducreyi LOS required the same cytokines and signaling pathways as the whole organisms to induce IDO expression. Neutralizing antibodies against the type I interferon receptors or TNF-α significantly reduced IDO expression induced by LOS, and the combination of both antibodies almost completely blocked IDO induction (Fig. 6A). The p38, JNK, and NF-κB inhibitors SB202190, SP600125, and NF-κB SN50, respectively, nearly abolished the production of IDO (Fig. 6B, top panels) and TNF-α and IFN-β stimulated by LOS (Fig. 6B, bottom panels). The ERK inhibitor U0126 also significantly inhibited the production of IDO (P = 0.03) and the cytokines induced by LOS (Fig. 6B). Thus, similar to whole bacteria, H. ducreyi LOS activates IDO expression through IFN-β and TNF-α and requires p38 and NF-κB signaling. In addition, the JNK and ERK pathways also contribute significantly to LOS-induced IDO production.
H. ducreyi LOS-induced IDO production is mediated by type I interferons, TNF-α, and the MAPK and NF-κB signaling pathways. (A) Effect of neutralizing antibodies against type I interferon receptor and TNF-α on IDO induction. DC were incubated in medium (med) or stimulated with LOS in the presence of antibodies against IFNAR2, TNF-α, IFNAR2 and TNF-α, or isotype-matched control antibodies mIgG1 and mIgG2a. Top panels are representative Western blots, and bottom panels are densitometry data obtained from four donors. The ratio of IDO/GAPDH in each sample was normalized to that of DC incubated with LOS alone, which was set to a value of 100, and is expressed as a mean ± SD. α, anti. (B) Inhibitory effect of the MAPK and NF-κΒ inhibitors on IDO induction. Cell lysates of DC treated with medium or LOS in the presence of DMSO, the MAPK inhibitors SB203580 (SB), SP600125 (SP), and U0126, or the NF-κB inhibitor NF-κB SN50 (SN50) and its inactive control NF-κB SN50M (SN50M) were analyzed for IDO and GAPDH protein expression. Representative Western blots of samples obtained from four donors are shown at the top. Culture supernatants from unstimulated DC or DC stimulated with LOS for 6 h were assayed for the production of IFN-β (n = 3) and TNF-α (n = 5 for the MAPK inhibitors and n = 3 for the NF-κB inhibitor), respectively. Due to donor-to-donor variation in cytokine production, cytokine levels were normalized to those of DC incubated with LOS alone, which was set at 100%. Bars indicate mean ± SD. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.
LOS-induced IFN-β activates the JAK-STAT pathway.The triggering of the IFN-α/β receptor IFNAR by all type I interferons including IFN-β activates the receptor-associated JAKs (TYK2 and JAK1), which then leads to the phosphorylation of the tyrosine 701 of STAT1. To examine whether H. ducreyi LOS-induced IFN-β activates the JAK-STAT pathway, we used an antibody against the phospho-Y701 STAT1 to detect the activated form of STAT1. Stimulation of DC with LOS led to phosphorylation of STAT1, which appeared to have a two-phase expression pattern, with an early peak at 2 h and a later peak at 18 h (Fig. 7A), similar to the profile of LOS-induced IFN-β production (data not shown). Since STAT1 can be activated downstream of receptors for several cytokines such as IL-6 and IFN-λ (29, 40), the contribution of IFN-β signaling in STAT1 activation was examined by using a neutralizing antibody against IFNAR. As shown in Fig. 7B, the antibody significantly reduced STAT1 phosphorylation induced by LOS. Thus, H. ducreyi LOS-induced IFN-β activates the JAK-STAT1 pathway, which likely cooperates with the MAPK and NF-κB pathways to promote IDO production in DC.
The JAK-STAT pathway is activated by H. ducreyi LOS-induced IFN-β. (A) Time course of STAT1 activation by LOS. Western blots of whole-cell lysates of DC stimulated with H. ducreyi LOS for various periods of time (hours) were probed with antibodies against phosphorylated STAT1 and GAPDH. Shown are representative Western blots of samples obtained from two donors. (B) Effect of neutralizing antibodies against type I interferon receptor on STAT1 activation. DC were incubated in medium (med) or stimulated with LOS in the presence of antibodies against IFNAR2 or isotype-matched control antibody mIgG2a. Top panels are representative Western blots, and bottom panels are densitometry data obtained from five donors. The ratio of phosphorylated STAT1/GAPDH in each sample was normalized to that of DC incubated with LOS alone, which was set to a value of 100, and is expressed as a mean ± SD. ***, P ≤ 0.001.
DISCUSSION
We previously reported that FOXP3+ Treg cells are expanded at sites of H. ducreyi infection and that myeloid DC infected with H. ducreyi upregulate transcripts that encode IDO. Here, we found that H. ducreyi infection of monocyte-derived DC induced the production of enzymatically active IDO. Intact bacteria or bacterial uptake was not required for H. ducreyi-induced IDO production while cell-free culture supernatants and LOS were sufficient to induce IDO. Gram-negative bacteria generally shed outer membrane vesicles, which contain LOS/LPS, lipids, outer membrane and periplasmic proteins, and DNA. These vesicles are recognized by innate immune cells and play a role in immune modulation (16, 31). Our data indicate that H. ducreyi LOS is a major inducer of IDO in DC.
IFN-γ is a potent activator of IDO in various cell types, and other cytokines such as TNF-α can synergistically enhance IFN-γ-mediated IDO induction (1, 4, 14, 39, 43, 45). DC infected with several species of bacteria are capable of producing IFN-γ and IDO. For example, DC infected with Listeria monocytogenes express both IFN-γ and IFN-β as well as TNF-α, and IFN-γ synergizes with TNF-α to induce IDO in Listeria-infected cells (42). Chlamydia pneumoniae also induces human DC to produce IFN-γ and IDO (28). However, our data indicate that H. ducreyi infection of DC does not induce detectable levels of IFN-γ, and IFN-γ is not involved in IDO induction. Instead, type I interferons and TNF-α act synergistically to promote IDO expression induced by H. ducreyi. To our knowledge, this is the first demonstration that a bacterial pathogen induces IDO in DC via an autocrine mechanism involving type I interferons and TNF-α.
LPS/LOS recognition by TLR4 initiates multiple intracellular signaling cascades, leading to the production of proinflammatory mediators, such as TNF-α and IFN-β. Several studies indicate that LPS induces the expression of IDO in an IFN-independent manner. For example, Salmonella enterica serovar Abortus equi LPS promotes IDO expression in a human monocyte cell line via the activation of the p38 and NF-κB pathways but not IFN-initiated JAK-STAT signaling (19, 20). Escherichia coli LPS-induced IDO expression in murine bone marrow-derived DC does not require JAK-STAT signaling either but depends on the JNK and PI3K pathways (26). Our data indicate that IDO expression induced by H. ducreyi LOS is mediated by TNF-α and IFN-β, which activates the JAK-STAT pathway, and requires the p38, JNK, ERK, and NF-κB pathways. Thus, unlike LPS, H. ducreyi LOS induces IDO via IFN-dependent as well as IFN-independent mechanisms.
The possible mechanisms by which H. ducreyi LOS induces IDO production in DC are summarized in a model shown in Fig. 8. Stimulation of TLR4 by LPS/LOS initiates myeloid differentiation factor 88 (MyD88)-dependent and TRIF (Toll/IL-1 receptor domain-containing adaptor inducing IFN-β)-dependent signaling cascades, which activate the MAPK and NF-κB pathways. The TRIF-dependent pathway also activates interferon response factor 3 (IRF3), which is essential for LPS/LOS-induced IFN-β production in DC (27, 46). Activation of the p38, JNK, ERK, and NF-κB pathways leads to production of the IDO-inducing cytokines IFN-β and TNF-α, which act in an autocrine manner to stimulate the JAK-STAT pathway and the MAPK and NF-κB pathways, respectively. The JAK-STAT signaling activates the IFN-stimulated response element (ISRE) in the IDO gene promoter (3, 38). The MAPK and NF-κB pathways contribute to IDO induction indirectly by promoting the production of TNF-α and IFN-β and directly by regulating IDO gene promoter activity (19) in concert with the JAK-STAT pathway.
Schematic overview of IDO induction by H. ducreyi LOS. Engagement of TLR4 on the DC surface by H. ducreyi LOS activates MyD88-dependent and MyD88-independent signaling pathways. The MyD88-dependent pathway leads to the activation of MAPK (p38, JNK, and ERK) and NF-κB signals. The MyD88-independent pathway is mediated by TRIF, which activates IRF3 as well as MAPKs and NF-κB. Signaling through MAPKs and NF-κB leads to the expression of TNF-α, whereas activation of IRF3 (27, 46), MAPKs, and NF-κB is required for the expression of IFN-β. Secreted TNF-α and IFN-β bind to their cognate receptors, TNF receptor (TNFR) and IFNAR, via autocrine mechanisms to activate the MAPK and NF-κB and JAK-STAT signal pathways, respectively. The JAK-STAT signaling promotes the formation of the transcription factor complex ISGF3 (a heterotrimer of STAT1, STAT2, and IRF9). The cooperative activation of these pathways leads to transcriptional activation of the INDO gene, which encodes IDO, and the production of IDO protein.
While polymyxin B completely blocked the induction of IDO by H. ducreyi culture supernatants or LOS, polymyxin B only partially blocked the induction by heat-killed bacteria, suggesting that other bacterial components might synergize with LOS to foster IDO induction. In addition to TLR4, DC express many other pattern recognition receptors that detect various bacterial components such as lipoproteins and pore-forming toxins. Lipoproteins and outer membrane porin proteins can activate TLR2 and induce inflammatory cytokine production (34, 55), and TLR2 signaling has been shown to induce IDO expression (35, 42). The intracellular pattern recognition receptor NOD2 recognizes the muramyl dipeptide component of peptidoglycans and activates NF-κB and MAPK signaling pathways to induce the expression of TNF-α and other inflammatory cytokines. H. ducreyi expresses multiple lipoproteins and porins OmpP2A and OmpP2B (7, 23, 44, 51). In addition, we previously observed that DC infected with H. ducreyi upregulate the transcription of the NOD2 gene (24). Whether TLR2- and NOD2-mediated pathways contribute to IDO induction by H. ducreyi requires further investigation.
Our results also do not exclude the possibility that other inflammatory factors could also contribute to the production of IDO induced by H. ducreyi or LOS. LPS-induced IDO in human mononuclear cells is mediated by TNF-α, IL-1β, and IL-6 (19). Prostaglandin E2 (PGE2) is also known to synergize with TNF-α or TNF-α/IL-1β/IL-6 to promote IDO induction in monocyte-derived DC (10, 30). H. ducreyi and LOS induce the production of many inflammatory mediators, including IL-6, IL-1β, and PGE2 (6, 24). The observation that a combination of anti-IFNAR2 and anti-TNF-α antibodies did not completely abolish IDO production induced by live H. ducreyi supports the possible involvement of additional molecules. Additionally, live bacteria can induce qualitatively and quantitatively different cytokine responses from heat-killed bacteria or bacterial components in DC or mononuclear cells (21, 47, 52). Therefore, further study is needed to identify other IDO-inducing factors and the mechanisms by which IFN-dependent and IFN-independent factors synergize to promote IDO activation by H. ducreyi.
In human inoculation experiments, different hosts are differentially susceptible to H. ducreyi infection (50). Some participants repeatedly clear infection (resolvers) while others fail to do so (pustule formers). In pustule formers, H. ducreyi evades phagocytosis and persists extracellularly at infected sites. DC derived from resolvers and pustule formers have distinct transcript profiles in response to H. ducreyi. DC from the pustule formers simultaneously express high levels of transcripts for many inflammatory cytokines as well as suppressor markers, such as IDO (24). These IDO-expressing DC might downregulate anti-H. ducreyi T cell responses through inhibition of T cell proliferation, induction of T cell death, and the expansion of FOXP3+ Treg cells. In addition, tryptophan metabolites can also inhibit NK cell proliferation and cytokine production (15, 18) and downregulate Th-17 responses (2, 17). H. ducreyi-infected DC express several additional immunosuppressive molecules, including IL-10, IL-2 receptor α chain, and cyclooxygenase-2 (6, 24). The immune suppression mediated by IDO and other inhibitory mechanisms may collectively inhibit H. ducreyi-specific T cell responses, reduce the production of T cell and NK cell effector cytokines, such as IFN-γ, and dampen the phagocytosis and bactericidal activity of neutrophils and macrophages, leading to H. ducreyi persistence. Further studies are required to explore this hypothesis and to dissect the mechanisms by which inhibitory signals interact with each other to promote immune evasion by H. ducreyi.
ACKNOWLEDGMENTS
We thank Margaret E. Bauer, Chadi A. Hage, and David S. Wilkes for their critical review of the manuscript.
This work was supported by grant AI059384 from the National Institute of Allergy and Infectious Diseases.
FOOTNOTES
- Received 4 March 2011.
- Returned for modification 4 April 2011.
- Accepted 10 May 2011.
- Accepted manuscript posted online 16 May 2011.
- Copyright © 2011, American Society for Microbiology. All Rights Reserved.
