Transforming growth factor-β-activated kinase 1 resistance limits glucocorticoid responsiveness to Toll-like receptor 4-mediated inflammation

Abstract

Glucocorticoids (GC) are among the most effective anti-inflammatory drugs, but are often associated with serious adverse effects or inadequate therapeutic responses. Here, we use activation of different Toll-like receptors (TLRs) by their respective ligands to evaluate context-specific GC sensitivity in the macrophage. Recruitment and activation of transforming growth factor-β-activated kinase 1 (TAK1), downstream of TLR engagement, is crucial in activating multiple inflammatory pathways, and contributes to inflammatory disorders. We hypothesize that GC exert anti-inflammatory effects through regulation of TAK1. Both in vivo and in vitro, in comparison to other TLRs, there was limited GC potency in restricting TLR4 ligand-mediated secretion of interleukin-6, tumour necrosis factor-α and interleukin-12. Also, we found that inactivation of TAK1 both in vivo and in vitro strongly inhibits TLR4-induced inflammation-associated genes beyond the suppressive effects from GC treatment. However, there was no effect of TAK1 inactivation on GC inhibition of TLR3- or TLR9-initiated inflammatory actions. Together, our findings demonstrate that GC resistance for TAK1 activation associated with TLR4 engagement may be an important contributor to GC resistance in inflammatory disorders.

Introduction

Inflammation-associated diseases are frequent threats to human health.¹ Glucocorticoids (GC) are among the most effective clinical interventions for treating these disorders.² Critical GC functions that are essential for organism survival and the therapeutic benefits have been demonstrated using mice with conditional inactivation of GC receptors, a major mediator of GC actions.^3–5 Many patients with asthma, rheumatoid arthritis, chronic obstructive pulmonary disease and inflammatory bowel disease, however, show poor clinical responses to GC treatment for reasons that have remained elusive.^6–8 Although the use of biological agents alleviates some autoimmune inflammatory disorders,^9,10 the currently available drugs to treat GC-resistant diseases have achieved limited clinical success and are associated with toxicity and unacceptable adverse effects.¹¹ Hence, GC therapies remain the mainstay for treating patients. A recent population-based study reflects a 34% increase in GC treatment over the past 20 years.¹² Although extensive research over decades identified several functions of GC,^2,13 mechanisms that determine GC effectiveness for different inflammatory disorders remain to be determined.

Toll-like receptors (TLR) are critically involved in the exacerbation of several inflammatory diseases.^14–16 We have used activation of different TLR by their respective ligands as a highly informative model system to evaluate context-specific GC sensitivity in the macrophage, a pivotal player in inflammation. Downstream of TLR engagement, recruitment and activation of transforming growth factor β-activated kinase 1 (TAK1) is crucial in activating multiple inflammatory pathways.¹⁷ Emerging evidence indicates that activation of TAK1 contributes to renal inflammation, fibrosis, cerebral ischaemia and autoimmune disorders.^18–21 TAK1 is a serine/threonine kinase in the mitogen-activated protein kinase (MAPK) kinase kinase (MAP3K) family and functions as a central regulator of immune signalling pathways induced by a spectrum of stimuli.²¹ Being an upstream kinase, TAK1 plays critical roles in the activation of both nuclear factor-κB (NF-κB) and MAPK. TAK1 phosphorylates MAP kinase kinase 3/6, MAP kinase kinase 4/7, MAPK/ERK kinase 1/2 and IκB kinase to recruit and activate p38, c-Jun N-terminal kinase (JNK), extracellular signal-regulated kinase and NF-κB respectively.²¹

We demonstrated that TLR3 or TLR9-induced TAK1 phosphorylation is strongly GC-sensitive, whereas TLR4-mediated TAK1 activation was GC-resistant.²² Among the members of the TLR superfamily, TLR4 is known to induce the most robust inflammatory response by recruiting both myeloid differentiation primary response 88 - MyD88 adapter like and Toll/IL-1 receptor domain containing adaptor inducing IFN beta - TRIF related adaptor molecule adapter sets that are unique for TLR4.¹⁷ Also, TLR4 is prominently involved in a spectrum of inflammation-associated disorders, such as asthma, chronic obstructive pulmonary disease, atherosclerosis, colitis, pancreatitis, alcoholic liver disease, non-alcoholic liver disease, sepsis and cardiovascular diseases that are often refractory to GC therapy.^23–30 Whether failure to suppress TAK1 activity contributes to compromised GC-responsiveness is highly clinically relevant but remains to be addressed. Our previous work, although demonstrating that GC is ineffective in attenuating TAK1 phosphorylation with TLR4 activation, did not determine whether TAK1 inhibition in the presence of GC would enhance anti-inflammatory actions. Also, as TAK1-independent pathways, such as MLK (Mixed-Lineage Kinase), ASK1/2 (apoptosis signal-regulating kinase 1/2), Tpl-1 (tumour progression locus-1), MEKK4, TAO (thousand and one amino acid) are also known to activate JNK and NF-κB and induce inflammatory reactions,^31–33 the importance of TAK1 inhibition in mediating the anti-inflammatory effects of GC after TLR3 or TLR9 engagement is not known.

Here, we test the hypotheses that (i) TAK1 inhibition along with GC treatment would enhance the efficacy to restrict TLR4-induced inflammatory reactions and (ii) inhibition of TAK1 activation is required in mediating GC suppression of TLR3- or TLR9-initiated inflammatory actions. We provide the first in vivo evidence for relative GC inefficacy in suppressing TLR4-initiated pro-inflammatory responses. Further, in GC-treated macrophages, inactivation of TAK1 both in vivo and in vitro results in strong inhibition of LPS-induced inflammatory genes in addition to the extent of GC suppression. However, for TLR3 or TLR9 engagement, we found no significant effect of TAK1 inactivation in addition to GC inhibition. We found that GC resistance for TAK1 may determine its limited efficacy in restraining TLR4-mediated inflammatory reactions.

Materials and methods

Mice

Mice were housed on a 12-hr light and 12-hr dark cycle. Mice used for the experimentation were male, 6–10 weeks old and of C57BL/6 × 129/Sv background. The experimental protocols were approved by the Animal Care and Use Committee of Cincinnati Children's Research Foundation. Blood was collected by retro-orbital phlebotomy into heparinized capillary tubes, with the time from first handling the animal to completion of the bleeding not exceeding 30 seconds.

Materials

Lipopolysaccharide (LPS-EB Ultrapure, Escherichia coli 0111:B4), Poly(I:C) (Cat. Tlrl-pic) and CpG oligonucleotide ODN 1826 (5′-tcc atg acg ttc ctg acg tt-3′) were purchased from InvivoGen (San Diego, CA) and reconstituted according to the manufacturer's instructions. TAK1 inhibitors (5Z)-7-oxozeaenol (Oxo), inactive control (5Z)-zeaenol (EMD Millipore, Billerica, MA), AZ-TAK1 (3-[(aminocarbonyl) amino]-5-[4-(4-morpholinylmethyl)phenyl]-2-thiophenecarboxamide) (TRC, Toronto, Ontario, Canada) and dexamethasone (Dex; Sigma, St Louis, MO) were purchased and reconstituted according to the manufacturer's instructions. Antibodies used in this study were as follows: anti-COX-2 (160126) (Cayman, Ann Arbor, MI); anti-actin (A 5060) (Sigma); anti-TAK1 (5206, 4505), anti-phospho-stress-activated protein kinase (SAPK)/JNK (4668, 9251), anti-SAPK/JNK (9253, 9252), anti-phospho-p38 MAPK (4511, 9211), anti-p38 MAPK (9211, 9212), anti-phospho-IκBα (2859) and anti-IκBα (4814, 9242) (Cell Signaling Technology, Beverly, MA); anti-TAK1 (7162) (Santa Cruz Biotechnology, Santa Cruz, CA). anti-phospho-TAK1 (06-1425) (Millipore, Temecula, CA).

Isolation and culture of mouse peritoneal macrophages

Thioglycollate-elicited macrophages were isolated by peritoneal lavage 3 days following intraperitoneal (i.p.) injection of 1·0 ml of sterile 4% thioglycollate (Becton & Dickinson, Sparks, MD). Cells were plated in six-well pates (BD, Franklin Lakes, NJ) with complete medium containing Dulbecco's modified Eagle's medium, 2 mm l-glutamine, 50 U/ml penicillin, 50 U/ml streptomycin, 10% fetal bovine serum (Life Technologies, Grand Island, NY). After 3 hr of attachment, media were aspirated and the monolayer was rinsed twice with serum-free warm Dulbecco's modified Eagle's medium. Fresh complete medium was then added to each well, and the cells were cultured under a humidified atmosphere of 5% CO₂ in air at 37°.

Isolation and culture of human monocyte-derived macrophages

Human monocytes were isolated from male healthy volunteer blood donors. Peripheral blood mononuclear cells were obtained by density centrifugation, using Ficoll Paque (GE Healthcare, Pittsburgh, PA). Briefly, the blood was diluted 1 : 1 with sterile PBS (Life Technologies, Grand Island, NY), placed under a layer of Ficoll–Paque and centrifuged at 600 g for 15 min at room temperature. To remove platelet contamination, cells collected from the interphase were washed twice with PBS containing 5 mm of EDTA, and resuspended in RPMI-1640 (Life Technologies) supplemented with 2 mm l-glutamine, 50 U/ml penicillin, 50 U/ml streptomycin. The cells were then seeded in six-well plates, 10 × 10⁶ cells per well. After 1 hr, cells were washed three times with warm PBS before adding 1·5 ml of complete differentiation medium containing RPMI-1640, 2 mm l-glutamine, 50 U/ml penicillin, 50 U/ml streptomycin, 20 ng/ml macrophage-colony stimulating factor (PeproTech, Rocky Hill, NJ) and 10% fetal bovine serum. Media were replaced every 3 days and cultured for 7 days under a humidified atmosphere of 5% CO₂ in air at 37° before performing the experiments.

Cytokine measurement

For in vivo cytokine studies, male mice were injected intraperitoneally. with LPS (1, 5, 10 mg/kg), CpG (1, 5, 10 mg/kg), Poly(I:C) (1, 5, 10 mg/kg) or endotoxin-free physiological water (NaCl 0·9%) (InvivoGen). Blood was collected by retro-orbital phlebotomy after the indicated periods of treatment. Plasma samples were stored at −80° until assay. Pro-inflammatory cytokine concentrations in plasma samples were measured by ELISA for tumour necrosis factor-α (TNF-α), interleukin-6 (IL-6) and interleukin-12 (IL-12) according to the manufacturer's instructions (BD Biosciences PharMingen, San Diego, CA).

For in vitro cytokine studies cells were plated at 10⁶ cells/ml in a 24-well plate (BD, Franklin Lakes, NJ) with 1·0 ml of complete medium. Cells were treated with Dex for 3 hr at a final concentration of 100 nm followed by a variety of TLR ligands for 18 hr. LPS, Poly(I:C) and CpG were used²² to engage TLR4, TLR3 and TLR9, respectively. For studies with TAK1 inhibitors, cells were pre-treated with various does of Oxo or AZ-TAK1 for the indicated periods of time, followed by the treatment with or without Dex for 3 hr and finally stimulated with respective TLR ligand for another 18 hr. Culture supernatants were collected and assessed for TNF-α, IL-6 and IL-12 as described previously.

In vivo treatment with TAK1 inhibitor and inactive control

TAK1 inhibitor Oxo and the inactive control (5Z)-zeaenol were resuspended as a 10 mg/ml stock in DMSO. This was further diluted 10-fold in Peanut Oil (Sigma) to yield a 1·0 mg/ml stock in 10% DMSO. Male mice, 6–8 weeks of age, were injected with various doses of Oxo or the inactive control intraperitoneally. Alternatively, 10% DMSO in Peanut Oil was delivered as a vehicle control. After 3 hr of treatment with the TAK1 inhibitor or the inactive control, mice were further treated with LPS (5 mg/kg), CpG (5 mg/kg) or Poly(I:C) (10 mg/kg).

Radioimmunoassays

Plasma concentrations of corticosterone were determined by radioimmunoassay from blood collected by retro-orbital phlebotomy from 6- to 8-week-old male mice as described previously.³⁴ The corticosterone assays were performed according to the manufacturer's instructions (MP Biomedicals, Orangeburg, NY).

Cell cytotoxicity assay

Effects of TAK1 inhibitors on cell cytotoxicity were evaluated by quantifying the release of lactate dehydrogenase (LDH), a stable cytoplasmic enzyme in the culture supernatants. LDH activity in the culture supernatants was determined by a colorimetric assay using an LDH Cytotoxicity Detection Kit (Clontech, Mountain View, CA).

Western blot analysis

For Western blot analysis, whole cell extracts or the cytoplasmic extracts were resolved through SDS–PAGE using 4–12% separating gel (Invitrogen, Carlsbad, CA). Proteins were transferred to Hybond enhanced chemiluminescence nitrocellulose membrane (Amersham Pharmacia Biotech, Piscataway, NJ) using a semi-dry transfer system (Bio-Rad, Hercules, CA) and blocked with 5% dried milk in PBS and 0·1% Tween-20 (Sigma). Blots were probed with primary antibody overnight at 4°. Binding of horseradish peroxidase-labelled goat anti-rabbit antibody (sc-2004) or goat anti-mouse antibody (sc-2005) (Santa Cruz Biotechnology, Santa Cruz, CA) was determined using SuperSignalWest Chemiluminescent Substrate (Thermo Scientific, Rockford, IL). Blots were stripped with Restore™ Western Blot Stripping Buffer (Thermo Scientific, Rockford, IL) and re-probed with different antibodies.

Statistical analysis

Data are expressed as the mean ± SEM. Statistical significance was tested by unpaired two-tailed Student's t-test. Differences were considered to be significant for P < 0·05.

Results

Hyper-elevation of plasma inflammatory cytokines after TLR4 engagement

Depending on the nature of the TLR, intracellular adapter molecules are recruited to activate distinct signalling intermediates, such as NF-κB and interferon regulatory factors 3 and 5 that induce different gene signatures. To evaluate the relative effect of in vivo engagement of individual TLRs on the induction of inflammatory cytokines, we injected mice with various doses of Poly(I:C) (TLR3 ligand), LPS (TLR4 ligand) or CpG (TLR9 ligand). Plasma samples were analysed to evaluate the circulating levels of pro-inflammatory cytokines. There was no detectable plasma level of IL-6, TNF-α or IL-12 in mice treated with vehicle only. We observed a dose-dependent increase in plasma IL-6 and TNF-α levels after 3 hr of treatment with TLR ligands (see Supporting information, Fig. S1). For the subsequent in vivo studies, we treated mice with 5 mg/kg of LPS or CpG and 10 mg/kg of Poly(I:C). We found a time-dependent increase in plasma cytokine levels whereas mice were treated with ligands to individual TLR (Fig.1). However, we observed significantly higher levels of plasma cytokines in LPS-injected mice compared with CpG or Poly(I:C) -injected mice. After 3 hr of treatment, fourfold to 30-fold increases of plasma IL-6, TNF-α and IL-12 concentration were observed in LPS-injected mice compared with CpG-injected or Poly(I:C)-injected mice (Fig.1). After 6 hr of treatment, four- to eightfold increases of cytokine concentrations were detected in LPS-treated mice in comparison with CpG- or Poly(I:C)-treated mice. Although, the circulatory levels of cytokines were reduced at later time-points, our late kinetic study indicated persistently higher levels of cytokines in LPS-treated mice compared with the other two TLR ligands (data not shown).

Figure 1.

Toll-like receptor 4 (TLR4) ligand induces hyper-elevation of inflammatory cytokines in vivo. Male mice (n = 6–10) were treated with lipopolysaccharide (LPS; 5 mg/kg), CpG (5 mg/kg) or Poly(I:C) (10 mg/kg) for the indicated periods of time. Concentrations of (a) interleukin-6 (IL-6) and (b) tumour necrosis factor-α (TNF-α) and (c) IL-12 in plasma samples were analysed by ELISA. Data presented as mean ± SEM. *P < 0·05 and **P < 0·01 for mice treated with CpG or Poly(I:C) compared with treatment group with LPS.

Ligands to different TLRs induce comparable plasma corticosterone levels

Hypothalamus–pituitary–adrenal axis-mediated elevation of serum corticosterone (in rodents) or cortisol (in humans) is a homeostatic mechanism to limit the magnitude of immune and non-immune challenges. To determine whether LPS-induced hyper-elevation of inflammatory cytokines is the result of impaired corticosterone secretion, we examined plasma levels of corticosterone after treatment with different TLR ligands. Low levels of plasma corticosterone in control mice were observed without any treatment (Fig.2). In LPS-treated control mice, we observed 14-fold and 12·5-fold increases in plasma corticosterone levels (compared with the baseline corticosterone level) after 3 and 6 hr of challenge with the ligand, respectively (Fig.2a). Treatment of control mice with CpG for 3 and 6 hr resulted in 14-fold and 9·5-fold increase in circulatory corticosterone, respectively (Fig.2b). There was a moderate but significant (P < 0·05) decrease in plasma corticosterone after 6 hr of CpG treatment compared with the 3 hr treatment group. Similarly, 16-fold and 14·5-fold increases of plasma corticosterone levels were found in Poly(I:C)-injected control mice after 3 and 6 hr of treatment, respectively (Fig.2c). After 3 hr treatment with LPS, CpG and Poly(I:C) the corticosterone levels were 630 ± 73, 470 ± 66 and 575 ± 37 ng/ml, respectively. Six hours of treatments with the same ligands resulted in corticosterone levels of 567 ± 52, 323 ± 48 and 519 ± 36 ng/ml, respectively. Also, we did not find any significant difference in plasma corticosterone release between 3 and 6 hr treatment groups in LPS- or Poly(I:C)-injected mice. Only a moderate reduction in corticosterone release was observed after 6 hr of CpG treatment compared with CpG treatment for 3 hr. Together, we found similar elevation of plasma corticosterone in response to ligands to each of the different TLRs.

Figure 2.

Ligands to different Toll-like receptors (TLR) induce comparable plasma corticosterone levels. Male control mice (n = 6–8) were treated with (a) lipopolysaccharide (LPS; 5 mg/kg), (b) CpG (5 mg/kg) or (c) Poly(I:C) (10 mg/kg) for the indicated periods of time. Corticosterone concentrations in plasma samples were analysed by radioimmunoassay. Baseline indicates the corticosterone concentration in plasma after treatment with PBS for 3 hr. Data presented as mean ± SEM. *P < 0·05 and **P < 0·01 for mice treated with LPS, CpG or Poly(I:C) compared with treatment group with PBS.

Limited GC efficacy to suppress TLR4 ligand-mediated inflammation in vitro

To determine the direct actions of glucocorticoid on innate immune cells, we treated peritoneal macrophages from control mice with LPS, CpG or Poly(I:C), in the presence or absence of Dex, a synthetic GC. In the absence of TLR ligands or in the presence of Dex (only) IL-6, TNF-α and IL-12 were undetectable in the culture supernatants. As our in vivo study indicated that LPS was significantly more effective at inducing inflammatory responses than CpG or Poly(I:C), we performed an in vitro dose–response study for individual TLR ligands. Peritoneal mouse macrophages were treated with various doses of LPS (0·001–1 μg/ml), CpG (6·25–100 μg/ml) or Poly(I:C) (6·25–100 μg/ml). Saturations of IL-6 levels in the culture supernatants were observed at ligand concentrations above 0·1 μg/ml of LPS, 12·5 μg/ml of CpG and 50 μg/ml of Poly(I:C) (see Supporting information, Fig. S2). For the subsequent in vitro studies, we used these concentrations of ligands to study individual ligand-mediated release of inflammatory cytokines. Consistent with in vivo study, treatment with LPS resulted in a markedly elevated induction of all pro-inflammatory cytokines evaluated in this study compared with CpG or Poly(I:C) (Table1). Indeed, we found 60–75% suppression of TLR ligand-induced IL-6, TNF-α and IL-12 secretion by Dex pre-treatment. However, suppressed cytokine levels in Dex-treated macrophages remain significantly higher after LPS treatment than after Poly(I:C) or CpG treatments (Table1). Notably, with the exception of TNF-α, for LPS treatment, the Dex-suppressed cytokine levels were higher than the treatment groups with CpG or Poly(I:C) only (without Dex). Previously, we demonstrated that Dex fails to inhibit TLR4 ligand-induced activation of TAK1 and downstream signalling mediators, such as NF-κB and JNK in mouse macrophages.²² Consistently, a similar inefficacy of Dex in inhibiting TAK1 or JNK phosphorylation was observed in human monocyte-derived macrophages that were treated with LPS for 30 min (Fig.3). In contrast, 79% and 67% inhibition of TAK1; 87% and 73% inhibition of JNK; 64% and 92% inhibition of p38 phosphorylation were detected in response to Dex in CpG- and Poly(I:C)-treated cells, respectively (Fig.3). There was no detectable change in total protein levels of TAK1, JNK and p38 following treatment with Dex and/or TLR ligands.

Table 1.

Dexamethasone sensitivity for Toll-like receptor ligand-induced inflammatory cytokine secretion

	IL-6 (ng/ml)	TNF-α (ng/ml)	IL-12 (ng/ml)
LPS	31·536 ± 1·046	16·373 ± 0·678	19·403 ± 0·940
Dex + LPS	8·390 ± 0·372***	5·695 ± 0·336**	6·402 ± 0·308**
CpG	2·818 ± 0·112	6·439 ± 0·348	3·622 ± 0·142
Dex + CpG	1·076 ± 0·068***	2·599 ± 0·093**	1·257 ± 0·039**
Poly(I:C)	0·953 ± 0·046	5·681 ± 0·313	0·201 ± 0·004
Dex + Poly(I:C)	0·344 ± 0·013**	1·842 ± 0·0 77*	0·043 ± 0·002**

Mouse peritoneal macrophages were treated with dexamethasone (Dex) for 3 hr, followed by the treatment with lipopolysaccharide (LPS; 0·1 μg/ml), CpG (12·5 μg/ml) or Poly(I:C) (50 μg/ml) for another 18 hr. Concentrations of interleukin-6 (IL-6), tumour necrosis factor-α (TNF-α) and interleukin-12 (IL-12) in the culture media were measured by ELISA. Data presented as mean ± SEM.

^*P < 0·05

^**P < 0·01

^***P < 0·001 for Dex + Toll-like receptor ligand treatment versus ligand only treatment.

Figure 3.

Dexamethasone (Dex) sensitivity for Toll-like receptor (TLR) ligand-induced activation of transforming growth factor-β-activated kinase 1 (TAK1) and downstream signalling mediators. Effect of Dex on TAK1 and mitogen-activated protein kinase (MAPK) activation in monocyte-derived macrophages. Macrophages were treated with 100 nm of Dex for 3 hr, followed by treatment with lipopolysaccharide (LPS; 0·1 μg/ml), CpG (12·5 μg/ml) or Poly(I:C) (50 μg/ml) for 30 min. Whole cell lysates were analysed by Western blot using anti-phospho TAK1 (p-TAK1), total TAK1 (TAK1), anti-phospho c-Jun N-terminal kinase (p-JNK), total JNK (JNK), anti-phospho p38 MAPK (p-p38) and total p38 MAPK (p38) antibodies. Data presented are representative of three or four independent experiments.

Effect of TAK1 inhibition on individual TLR ligand-induced inflammatory reactions

We examined the relative contribution of TAK1 activation to the inflammatory responses initiated by the ligands for TLR3, TLR4 and TLR9. TAK1 catalytic activity is selectively inhibited by Oxo,^35,36 a pharmacological inhibitor widely used both in vitro and in vivo.^19,37 As TAK1 contributes critical roles for both immune and non-immune cellular functions, we first determined the optimal Oxo concentration that efficiently inhibits TAK1 activation without toxicity. To evaluate dose- and time-dependent Oxo effects, cells were pre-treated with various doses (0–1000 nm) of Oxo or the inactive control for 1, 2, 3 or 24 hr. Continuous treatment with Oxo for 24 hr was found to be highly cytotoxic, as shown by changes in cell morphology and the release of LDH in the culture supernatant (data not shown). In the next set of experiments, we treated the cells with Oxo at various doses for 1, 2 or 3 hr, followed by wash out of Oxo and subsequent culture for another 24 hr. In the washout experiments, in the presence of TLR ligands, Oxo at lower concentrations (up to 125 nm), was associated with only 0–4% cytotoxicity, comparable with the treatment group without Oxo (Fig.4a–c). There was no detectable Oxo cytotoxicity in the experiments with a shorter duration of Oxo exposure to study the effects of TAK1 inhibition on the activation of the signalling intermediates (data not shown). The data presented here are from the experiments with wash out of Oxo for 1 hr of treatment.

Figure 4.

Effect of transforming growth factor-β-activated kinase 1 (TAK1) inhibitor on cell cytotoxicity and individual Toll-like receptor (TLR) ligand-induced inflammatory reactions. Dose-dependent effect of (5Z)-7 oxozeanol (Oxo) treatment on cell cytotoxicity. Mouse peritoneal macrophages were pre-treated with Oxo (31·25–1000 nm) for 1 hr, washed, and stimulated with (a) 0·1 μg/ml of lipopolysaccharide (LPS), (b) 12·5 μg of CpG or (c) 50 μg/ml of Poly(I:C) for another 18 hr. The percent cell cytotoxicity was evaluated by measuring lactate dehydrogenase (LDH) release in the culture media (as described in the Materials and methods). Dose-dependent effect of Oxo on LPS (d, g), CpG (e, h) or Poly(I:C) (f, i) -induced cytokine secretion. Concentrations of interleukin-6 (IL-6) and tumour necrosis factor-α (TNF-α) in the culture media were analysed by ELISA. Data are shown as mean ± SEM; n = 3. (j) Dose- and time-dependent effect of Oxo treatment on LPS-induced c-Jun N-terminal kinase (JNK) phosphorylation. Peritoneal macrophages were pre-treated with various concentrations of Oxo (62·5–1000 nm) for 1, 2 or 3 hr, washed, and stimulated with LPS for 30 min. Whole cell lysates were analysed by Western blot using anti-phospho JNK (p-JNK) and total JNK (JNK) antibodies. Data presented are representative of three independent experiments.

We evaluated the effect of Oxo treatment on TLR ligand-induced pro-inflammatory cytokine secretion. Macrophages were treated with various concentrations of Oxo for 1 hr, washed and treated with TLR-specific ligand for 24 hr. At a concentration of 125 nm of Oxo, we observed 57%, 73% and 51% of inhibition of IL-6 secretion induced by LPS, CpG and Poly(I:C), respectively (Fig.4d–f). The same concentration of Oxo resulted in 54%, 74% and 47% of inhibition of TNF-α secretion induced by LPS, CpG and Poly(I:C), respectively (Fig.4g–i). We observed the strongest inhibitory effect of Oxo on cytokine secretion for CpG > LPS > Poly(I:C) treatment. Oxo concentrations of 500 nm and above almost abolished TLR ligand-induced cytokine release. We also assessed dose- and time-dependent effects of Oxo on LPS-induced JNK phosphorylation (Fig.4j). Oxo treatment at a concentration of 500 nm and above completely inhibited JNK phosphorylation, which is consistent with the results from the cytokine and cytotoxicity assay. After 1, 2 and 3 hr of treatment with 125 nm of Oxo, LPS-induced JNK phosphorylation was suppressed by 78%, 86% and 65%, respectively.

Effect of TAK1 inhibitors on Dex suppression of TLR ligand-induced inflammatory responses

Our current study and previous reports indicated GC suppression of pro-inflammatory genes and TAK1. How important is TAK1 inhibition in mediating the anti-inflammatory actions of GC? To answer this question, the efficacy of Dex in inhibiting TLR ligand-induced inflammatory reactions was determined in Oxo (125 nm) -treated cells. The specificity of the Oxo effect was determined by using (5Z)-zeaenol, an inactive control (InC) for Oxo. We could not detect any cytotoxicity from InC treatment at a concentration of 125 nm (data not shown). For the LPS-treated macrophages, treatment with Dex caused more inhibitory effect on IL-6, TNF-α and IL-12 secretion than the treatment with of Oxo. Oxo pre-treatment (without Dex) resulted in 61%, 47% and 52% of inhibition of LPS-mediated IL-6, TNF-α and IL-12 secretion, respectively (Fig.5a–c). In the presence of Dex, we observed significant additional suppression (P < 0·005) of cytokine secretion in the same treatment groups (with Oxo). Inhibition of 97%, 95% and 97% of LPS-induced IL-6, TNF-α and IL-12 secretion was detected when treated with both Oxo and Dex (Fig.5a–c). We anticipated an additive suppressive effect of Dex and Oxo only if Dex inhibits inflammatory reactions in a TAK1-independent manner. Similar experiments with InC had no effect on LPS-induced cytokine secretion (see Supporting information, Fig. S3). Also, we could not detect additional suppression of LPS-induced IL-6 secretion by InC pre-treatment in Dex-treated cells (Fig. S3). For CpG-treated cells, compared with Dex treatment, Oxo itself has similar or stronger inhibitory effects on cytokine release in the culture supernatant. Notably, there was no additional effect of Dex on Oxo-mediated suppression of CpG-induced pro-inflammatory cytokine secretion. We observed 88%, 76% and 67% of inhibition in Oxo (only) treated macrophages compared with 80%, 78% and 68% of inhibition of IL-6, TNF-α and IL-12 secretion in Dex + Oxo-treated cells (Fig.5d–f). For the treatment with Poly(I:C), we found Oxo to be less effective than Dex in suppressing cytokine secretion. Inhibition of 42–52% of IL-6, TNF-α and IL-12 secretion was detected in Oxo-treated macrophages compared with 62–79% inhibition in Dex-treated cells (Fig.5g–i). However, there was no significant difference in Poly(I:C)-induced cytokine levels between Dex and Dex + Oxo treatment groups.

Figure 5.

Effect of transforming growth factor-β-activated kinase 1 (TAK1) inhibitors on dexamethasone (Dex) suppression of Toll-like receptor (TLR) ligand-induced inflammatory reactions. (a–i) Effect of Oxo on Dex-mediated inhibition of pro-inflammatory cytokines induced by TLR ligands. Mouse peritoneal macrophages were pre-treated with 125 nm of Oxo for 1 hr, washed and given 100 nm of Dex for 3 hr, followed by the stimulation with lipopolysaccharide (LPS) (a–c), CpG (d–f) or Poly(I:C) (g–i) for another 18 hr. Concentrations of interleukin-6 (IL-6), tumour necrosis factor-α (TNF-α) and interleukin-12 (IL-12) in the culture media were analysed by ELISA. Data are shown as mean ± SEM; n = 3; *P < 0·05; **P < 0·01 and ***P < 0·001 for Oxo + Dex treated macrophages compared with treatment group treated with Dex. (j–l) Effect of Oxo on Dex-mediated inhibition of TLR ligands-induced COX-2 expression. Macrophages were treated with Oxo and Dex (as described) followed by the treatment with (j) LPS, (k) CpG or (l) Poly(I:C) for 3 hr. Whole cell lysates were analysed by Western blot using anti-COX-2 and β-actin antibodies. Data presented are representative of three independent experiments. (m–r) Effect of AZ-TAK1 on Dex-mediated inhibition of pro-inflammatory cytokines induced by TLR ligands. Macrophages were pre-treated with 1 µm of Oxo for 1 hr, administered with 100 nm of Dex for 3 hr followed by the treatment with LPS (m, n), CpG (o, p) or Poly(I:C) (q, r) for another 18 hr. Concentrations of IL-6 and TNF-α in the culture media were analysed by ELISA. Data are shown as mean ± SEM; n = 3.

Previously, we reported Dex-mediated inhibition of TLR ligand-induced expression of cyclo-oygenase 2 (COX-2), a critical mediator of inflammatory reactions. In this study we examined whether Oxo contributes any additional inhibition to Dex suppression of COX-2. Densitometric analysis indicated that after 3 hr of LPS treatment, in Dex-treated cells there was 77% inhibition of COX-2 expression (Fig.5j). In the presence of Oxo, Dex inhibition of COX-2 expression was further increased to 92%. There was no effect of Oxo treatment on Dex-mediated inhibition of either CpG- or Poly(I:C)-induced COX-2 expression (Fig.5k,l).

To further validate the results from the study with Oxo we performed similar experiments with AZ-TAK1, a novel ATP-competitive small molecule inhibitor of TAK1.³⁸ Dose-dependent study indicated that AZ-TAK1 is not cytotoxic at a concentration of 2·5 µm and below, as shown by LDH release in the culture supernatant (see Supporting information, Fig. S4); 1 µm of AZ-TAK1 was non-cytotoxic and efficiently inhibited TLR ligand-mediated IL-6 secretion (Fig. S4). Further, at this concentration AZ-TAK1 markedly inhibited (but did not abolish) the phosphorylation of both TAK1 and JNK (Fig. S4). Consistent with the results from the study with Oxo, we found that AZ-TAK1 caused significant additional suppression (P < 0·001) of pro-inflammatory cytokines in LPS + Dex-treated macrophages. We observed 62% and 55% inhibition in AZ-TAK1 (only)-treated macrophages compared with 97% and 99% inhibition of IL-6 and TNF-α secretion in Dex + AZ-TAK1-treated cells (Fig.5m,n). Also, there was little or no effect of AZ-TAK1 on Dex-mediated suppression of CpG- or Poly(I:C)-induced IL-6 and TNF-α secretion (Fig.5o–r).

Effect of TAK1 inhibition in vivo on TLR ligand-induced inflammatory cytokine secretion

Next, we investigated the effect of TAK1 inhibition in vivo on circulatory pro-inflammatory cytokine levels. Control mice were injected with various doses of Oxo (1, 5, 10 or 15 mg/kg) to examine cytotoxicity. We did not detect overt toxicity in the animals for Oxo treatment at a concentration of 5 mg/kg body weight. Male mice were treated with Oxo or InC followed by the treatment with ligands to TLR3, TLR4 or TLR9. Since treatment with LPS, CpG or Poly(I:C) elevated plasma corticosterone concentrations, the basal cytokine levels in response to individual ligand treatment would be considered as cytokine concentrations suppressed by endogenous corticosterone. Pre-treatment with InC had no effect on TLR ligand-induced pro-inflammatory cytokine secretion (Fig.6). Also, after 3 or 6 hr of treatment with ligands, there was no significant difference between Oxo and InC treatment groups that were injected with CpG or Poly(I:C) (Fig.6d–i and m–r). In accord with the in vitro data, after 3 hr of LPS treatment, we observed 47%, 37% and 32% of reduction in plasma IL-6, TNF-α and IL-12 concentrations in Oxo-treated mice compared with InC-treated mice (Fig.6a–c). More pronounced reduction in cytokine levels was found after 6 hr of LPS treatment. Decreases of 73%, 38% and 69% in plasma IL-6, TNF-α and IL-12 concentrations were detected in the treatment group injected with Oxo (Fig.6j–l). Of note, this inhibition of IL-6 is in addition to the suppression by elevated endogenous corticosterone. Hence, Oxo in vivo ensures additional suppression to Dex-mediated inhibition of inflammatory reactions following LPS but not CpG or Poly(I:C) treatment.

Figure 6.

Effect of transforming growth factor-β-activated kinase 1 (TAK1) inhibition in vivo on Toll-like receptor (TLR) ligand-induced inflammatory cytokine secretion. Male control mice (n = 8–10) were pre-treated with vehicle, vehicle + InC (5 mg/kg) or vehicle + Oxo (5 mg/kg) for 3 hr. This is followed by the treatment with (a–c, j–l) LPS (5 mg/kg), (d–f, m–o) CpG (5 mg/kg) or (g–i, p–r) Poly(I:C) (10 mg/kg) for the indicated periods of time. Concentrations of interleukin-6 (IL-6) and tumour necrosis factor-α (TNF-α) and interleukin-12 (IL-12) in plasma samples were analysed by ELISA. Data presented as mean ± SEM. *P < 0·05; **P < 0·01 and ***P < 0·001 for Oxo + vehicle-treated mice compared with treatment group treated with InC + vehicle.

Effect of TAK1 inhibition on Dex suppression of TAK1, MAPK and NF-κB activation

As TAK1 is an upstream regulator of MAPK and NF-κB activation, application of TAK1 inhibitor to Dex-treated cells is expected to render additional inhibitory effects on these signalling intermediates if not suppressed by Dex. To test this hypothesis, we determined the effects of Oxo on Dex-dependent suppression of MAPK and NF-κB activation following TLR engagement with respective ligand. Treatment with Oxo alone resulted in 28% and 45% inhibition of p38 phosphorylation after 15 and 30 min of LPS treatment, respectively (Fig.7a). Notably, while LPS-induced p38 phosphorylation is Oxo and Dex-sensitive, treatment with both Oxo and Dex resulted in additional inhibitory effects (Fig.7a). After 15 min of LPS treatment, in the presence of Dex alone, we observed 77% inhibition of p38 phosphorylation. Pre-treatment with Oxo + Dex resulted in 94% inhibition of p38 phosphorylation for the same duration of LPS treatment. Compared with Dex only treatment, a more prominent inhibitory effect of Oxo + Dex was observed after 30 min of LPS treatment. At this time-point, whereas treatment with Dex alone resulted in 49% inhibition of p38 phosphorylation, treatment with Oxo + Dex caused 95% inhibition. In contrast, although LPS-induced JNK phosphorylation was strongly suppressible by Oxo but not Dex pre-treatment, there was no additional inhibitory effect from the combinatorial treatment with Dex and Oxo (Fig.7a). Also, while phosphorylation of TAK1, IκBα and degradation of IκBα are markedly inhibited by Oxo, combination of Dex and Oxo had no additional inhibitory effect (Fig.7b).

Figure 7.

Effect of transforming growth factor-β-activated kinase 1 (TAK1) inhibition on dexamethasone (Dex) suppression of TAK1, mitogen-activated protein kinases (MAPK) and nuclear factor-κB activation. Mouse peritoneal macrophages were pre-treated with 125 nm of Oxo for 1 hr, washed and administered with or without 100 nm of Dex for 3 hr, followed by the stimulation with (a, b) LPS, (b) CpG and (c) Poly(I:C) for the indicated periods of time. Whole cell lysates were analysed by Western blot using anti-phospho TAK1 (p-TAK1), total TAK1 (TAK1), anti-phospho c-Jun N-terminal kinase (p-JNK), total JNK (JNK), anti-phospho p38 MAPK (p-p38) and total p38 MAPK (p38) antibodies. Cytoplasmic extracts were analysed by Western blotting using anti-phospho IκBα and anti-IκBα antibodies. Data presented are representative of two to four independent experiments.

For CpG treatment, although the phosphorylation of TAK1, IκB, p38 and JNK or degradation of IκBα are Dex-suppressible we could not detect an additional inhibitory effect in Dex + Oxo treatment groups (Fig.7c). Also, there was no distinguishable change for Poly(I:C)-mediated p38, JNK and TAK1 phosphorylation between Dex versus Dex + Oxo treatment groups (Fig.7d). These results are concordant with the minimal or no effect of Oxo or AZ-TAK1 on Dex-mediated suppression of CpG or Poly(I:C) -induced pro-inflammatory cytokines (Fig.5a–c and m,n).

Discussion

In the present study, we evaluate the requirement for TAK1 inhibition in mediating the anti-inflammatory actions of GC. We hypothesized that TAK1 sensitivity to GC suppression determines GC efficacy in limiting inflammation. We found inefficient GC suppression of both inflammatory cytokines and key signalling intermediates that are induced or activated following TLR4 engagement with LPS. Also, for the treatment with LPS, we found that inactivation of TAK1 results in robust inhibition of pro-inflammatory responses that was in addition to the inhibitory effect caused by endogenous or pharmacological GC. In contrast, TAK1 inactivation has no additional inhibitory effect on GC suppression of TLR3 or TLR9 ligand-induced inflammatory actions. These findings are consistent with our previous report implicating GC resistance for LPS-induced activation of TAK1 and downstream signalling mediators, such as NF-κB and JNK.²²

Glucocorticoids, produced endogenously or administered therapeutically, provide an essential restraint to limit the magnitude and duration of inflammatory responses.² Ligands for TLRs have been found to enhance corticosterone secretion.^39,40 In this study, we compare the relative potency of the ligands for TLR3, TLR4 and TLR9 to induce corticosterone secretion. Following treatment with these ligands, we find substantial elevation but no marked difference in plasma corticosterone levels among the mice that were treated with different TLR ligands (Fig.2). This is in sharp contrast to our results that indicate significantly higher levels of circulatory pro-inflammatory cytokines in LPS-treated mice compared with CpG- or Poly(I:C)-treated mice (Fig.1). Therefore, in vivo, elevated endogenous GC following LPS treatment fails to efficiently attenuate the inflammatory reactions. Also, although GC in vitro is capable of inhibiting approximately 65–75% of LPS-stimulated cytokine production by macrophages, the absolute suppressed levels still exceed those associated with stimulation of the other TLR that individually signal through either MyD88 or Trif (Table1). Despite suppression with Dex, the persistent high levels of inflammatory cytokines following TLR4 engagement demonstrate the limited effectiveness of GC. Dysfunction of TLR4 has been demonstrated in several inflammatory disorders that are less responsive/unresponsive to GC therapy. As an example, whereas adrenal insufficiency is associated with sepsis, exogenous administration of GC to patients with septicaemia and/or septic shock is often not beneficial.^41,42 In intestinal inflammation-associated diseases, such as colitis, patients undergo therapy with GC but frequently fail to respond acutely or chronically.^7,43 These clinical observations are in agreement with compromised GC efficacy in suppressing TLR4-augmented inflammation both in vivo and in vitro.

Among the members of the TLR superfamily, TLR4 is unique. Excluding TLR3 that uses Trif, most of the TLRs recruit MyD88 as the adapter molecule that propagates the activation signal following ligand–receptor interaction.¹⁷ TLR4 is distinct because its engagement recruits both MyD88-MAL and/or Trif-TRAM adapter sets.¹⁷ Recent studies indicate that TLR4 engagement results in temporal recruitment of MyD88 and Trif, which leads to both overlapping and discrete biological functions.⁴⁴ Hence, in comparison with other TLR, TLR4 renders a more robust inflammatory response both in magnitude and duration. As these inflammatory actions are mediated by a variety of signalling molecules, it is plausible that GC may not efficiently inhibit the activation of key signalling mediators that function downstream of TLR4 engagement. In support of this hypothesis, our previous report indicates that multiple steps of LPS-induced NF-κB activation are Dex-resistant.²² Consistently, for TLR4 engagement, phosphorylation of JNK and TAK1 are also found to be Dex-resistant (Fig.3a).²² In contrast, for CpG treatment, Dex suppresses the phosphorylation, degradation of IκBα, and phospho p65 redistribution (Fig.7c).²² Further, CpG or Poly(I:C)-mediated p38, JNK and TAK1 phosphorylation are Dex-suppressible (Fig.3b,c). Together, while following TLR3-Trif or TLR9-MyD88 engagement the major signalling intermediates are GC-sensitive; those are largely GC-resistant after TLR4-MyD88-Trif engagement. Following treatment with LPS, only p38 phosphorylation is Dex-suppressible (Fig.3a); this is anticipated to be the result of an indirect GC effect via the induction of MAPK phosphatase 1 (MKP1).³ We did not find such an up-regulation of MKP1 expression by Dex treatment in CpG- or Poly(I:C)-treated cells (S. Bhattacharyya, unpublished data) suggesting a distinct mechanism for GC regulation of p38 MAPK activation that requires further investigation.

It is noteworthy that non-TAK1 pathways, such as MLK (Mixed-Lineage Kinase), ASK1/2 (Apoptosis signal-regulating kinase 1/2), Tpl-1 (Tumour progression locus-1), MEKK4, TAO (Thousand And One amino acid) are also known to activate JNK and NF-κB.^31–33 We find that treatment with TAK1 inhibitors result in 50–60% inhibition of TLR ligand-mediated activation of inflammatory genes (Figs5 and 7). If TAK1 is not GC-suppressible, additive inhibitory effects of Dex and TAK1 inhibitors on ligand-induced inflammatory responses are predicted. In contrast, if TAK1 is GC-suppressible further inhibition of inflammatory reactions is not expected in the experimental sets with Dex + TAK1 inhibitor(s). As there was no additional suppression by Oxo/AZ-TAK1 treatment to Dex inhibition of CpG or Poly(I:C)-induced cytokine secretion or COX-2 expression (Fig.5d–i and k,l), the major Dex effects are mediated by the impairment of TAK1 functions. Although, we found a very low level of Oxo cytotoxicity, the similar regulation of inflammatory cytokines by AZ-TAK1 treatment indicates TAK1-specific effects. Consistent with the results from the in vitro studies, in CpG or Poly(I:C)-injected mice there was no significant effect on plasma cytokine levels by Oxo pre-treatment (Fig.7d–i and m–r). Also, Oxo has no additional effect on Dex-mediated inhibition of CpG or Poly(I:C)-induced p38 and JNK phosphorylation (Fig.7b,c). In contrast, for LPS, while we observe 60–75% suppression of cytokine secretion by Dex treatment, addition of Oxo/AZ-TAK1 almost abolished secretion of IL-6, TNF-α and IL-12 (Fig.5a–c and m,n). Also, pronounced suppressive effects on Dex inhibition of LPS-induced COX-2 are observed by Oxo pre-treatment (Fig.5d). In a similar experiment, densitometric analysis indicated additional inhibition of p38 phosphorylation in Oxo-treated cells (Fig.7a). Further, our in vivo data demonstrate significant inhibitory effects of Oxo treatment on the circulatory levels of LPS-induced pro-inflammatory cytokines. It is noteworthy that this inhibition is in addition to the suppressive effects mediated by endogenous corticosterone, induced by LPS treatment (Fig.2a). Together, our data implicate that for TLR4 engagement with LPS, TAK1 is not a GC target. The physiological ramification of such differential GC regulation of TLR4 versus TLR3- or TLR9-initiated inflammatory actions is substantial. The non-responsiveness of several inflammation-associated diseases toward GC therapy would be due to GC-resistance for TAK1 and/or TAK1 inducible downstream signalling mediators. One attractive alternative strategy to overcome such limitations of GC therapy would be through pharmacological inhibition of TAK1.

Although the genomic functions of GC are well studied, many rapid, non-genomic GC functions are emerging.^45,46 Our studies suggest new transcription-independent GC suppression of TAK1 for TLR3 and TLR9 engagement (F.K. Fansheng Kong, S.B. Sandip Bhattacharyya, unpublished data). Biochemical and genetic studies indicated that TLR ligand-induced TAK1 activation is tightly regulated by the actions of the ubiquitin proteasome system (UPS).^47,48 Recent studies indicate that GC enhances proteasome-mediated degradation of glycogen synthase kinase 3, Cyclin D1, MyoD and c-maf, while inducing the transcription of ubiquitin C mRNA.^49–53 Nonetheless, the role of GC to modulate UPS in regulating inflammatory reactions is largely unknown. Whether GC modulation of UPS functions determines its ability to modulate TAK1 activation and thereby the anti-inflammatory efficacy in inflammatory diseases will be an important area of future investigation.

Glossary

AZ-TAK1

TAK1 inhibitor from AstraZeneca

Dex

dexamethasone

GC

glucocorticoid

Oxo

(5Z)-7-oxozeaenol

TAK1

TGF-β- activated kinase 1

TLR

Toll-like-receptor

Authors contribution

Sandip Bhattacharyya designed the study. All authors conducted experiments. Sandip Bhattacharyya wrote the manuscript.

Disclosures

The authors declared no financial or commercial conflicts of interest.

Supporting Information

Additional Supporting Information may be found in the online version of this article: