TNFα and IFNγ Synergistically Enhance Transcriptional Activation of CXCL10 in Human Airway Smooth Muscle Cells via STAT-1, NF-κB, and the Transcriptional Coactivator CREB-binding Protein

Asthmatic airway smooth muscle (ASM) expresses interferon-γ-inducible protein-10 (CXCL10), a chemokine known to mediate mast cell migration into ASM bundles that has been reported in the airways of asthmatic patients. CXCL10 is elevated in patients suffering from viral exacerbations of asthma and in patients with chronic obstructive pulmonary disease (COPD), diseases in which corticosteroids are largely ineffective. IFNγ and TNFα synergistically induce CXCL10 release from human ASM cells in a steroid-insensitive manner, via an as yet undefined mechanism. We report that TNFα activates the classical NF-κB (nuclear factor κB) pathway, whereas IFNγ activates JAK2/STAT-1α and that inhibition of the JAK/STAT pathway is more effective in abrogating CXCL10 release than the steroid fluticasone. The synergy observed with TNFα and IFNγ together, however, did not lie at the level of NF-κB activation, STAT-1α phosphorylation, or in vivo binding of these transcription factors to the CXCL10 promoter. Stimulation of human ASM cells with TNFα and IFNγ induced histone H4 but not histone H3 acetylation at the CXCL10 promoter, although no synergism was observed when both cytokines were combined. We show, however, that TNFα and IFNγ exert a synergistic effect on the recruitment of CREB-binding protein (CBP) to the CXCL10, which is accompanied by increased RNA polymerase II. Our results provide evidence that synergism between TNFα and IFNγ lies at the level of coactivator recruitment in human ASM and suggest that inhibition of JAK/STAT signaling may be of therapeutic benefit in steroid-resistant airway disease.

Lung diseases such as asthma and COPD 2 cause significant morbidity and mortality in Western societies. 5-10% of the asthmatic population are unresponsive to corticosteroids, the mainstay for asthma therapy, and patients with COPD are treated ineffectively with steroids. Additionally, viral exacerba-tions of asthma are a major cause of morbidity and mortality associated with asthma and are relatively unresponsive to steroids (1,2).
CXCL10, a member of the CXC chemokine subfamily, is a potent chemoattractant for mast cells and T lymphocytes, cells implicated in the pathophysiology of asthma and COPD. CXCL10 is elevated in the airways of asthmatic patients and has been implicated in mast cell migration to the ASM bundles (3). CXCL10 also is elevated in the bronchial mucosa and broncho alveolar lavage (BAL) fluid of subjects with moderate/severe asthma, which is largely steroid-resistant (4). Enhanced CXCL10 secretion also has been demonstrated in COPD (5)(6)(7), and more recently, it has been reported that serum CXCL10 levels are increased to a greater extent in asthmatics with acute virus-induced asthma compared with nonvirus-induced acute asthma, and increased levels are predictive of virus-induced asthma exacerbations (8). CXCL10 is released in response to IFN␥ or TNF␣ from a number of inflammatory cell types (6, 9 -12), and several reports also have shown that TNF␣ and IFN␥ synergistically enhance CXCL10 release (13)(14)(15)(16). Given the heightened cytokine milieu observed in severe asthma and COPD, understanding the mechanism for this increase clearly is a central question to address and requires an understanding of the mechanisms involved in transcription and translation of inflammatory products by ASM cells.
Transcriptional regulation involves transcription factors binding to recognition sequences in gene promoters (17)(18)(19). In quiescent cells, DNA is packaged tightly in chromatin, which must unravel to allow access of basal transcription factors and RNA polymerase II. Chromatin consists of four core histones, an H3-H4 tetramer and two H2A-H2B dimers that undergo covalent modifications (acetylation, phosphorylation, methylation), which control the unraveling process and thereby gene transcription (20 -23). Our previous studies of CCL11 and CXCL8 have identified important transcription factors for these genes and, moreover, identified histone H4 acetylation as a key event in inflammatory gene transcription (18 -19, 24 -25). Little information is present regarding CXCL10 regulation by transcription factors and histone modifications in ASM. One study by Hardaker and colleagues (6) using pharmacological inhibitors implicates NF-B in TNF␣-induced CXCL10 release but did not study which transcription factors are utilized by IFN␥ because although IFN␥ produced a transient activation of NF-B, no inhibitory action was observed with the NF-B inhibitor used in the study. This study also highlighted a synergistic interaction between TNF␣ and IFN␥ on CXCL10 release but did not explore the mechanism involved. TNF␣ also has been reported to cooperate with IFN␥ to synergistically induce CCL5 (26), CX3CL1 (fractalkine, 27), and CD38 (28), although here, too, the underlying mechanisms are unclear.
Other potentially important transcription factors involved in CXCL10 regulation by IFN␥ are STATs (signal transducers and activators of transcription). STATs are a family of latent cytoplasmic proteins that are activated when cells encounter various extracellular polypeptides, such as interferons and interleukins (29), and appear to be essential for responsiveness to IFN␣ and IFN␥ (30 -31). Ohmori and colleagues (32) reported that the frequently observed synergy between TNF␣ and IFN␥ depends in part upon cooperation between STAT-1␣ and NF-B in NIH3T3 fibroblasts and is likely mediated by their independent interaction with one or more components of the basal transcription complex. However, the authors but did not explore this possibility experimentally. IFN␥ alone, or in combination with TNF␣, also has been reported to involve p48, a transcription factor that complexes with STAT-1␣ and binds to interferon stimulated response element (ISRE) sites on the promoter. Induction of CXCL10 by TNF␣/IFN␥ also required NF-B binding sites that bound p65 homodimeric NF-B (15)(16).
The aim of the present study was to characterize the transcription factors, histone modifications, and RNA polymerase II recruitment to the CXCL10 promoter to determine at which point of regulation synergism can be explained between TNF␣ and IFN␥. We show that TNF␣ activates the classical NF-B pathway, IFN␥ signals via STAT-1, and their synergism on CXCL10 gene induction is due not to their effects on transcription factor activation or histone modification, but rather a direct effect on the recruitment of the transcriptional coactivator CBP and RNA polymerase II. We also demonstrate that inhibition of the janus kinase (JAK)/STAT signaling pathway may be a novel approach to targeting steroid-resistant inflammatory lung disease. This is the first observation in primary airway cells and could highlight a novel anti-inflammatory therapy for the treatment of airway inflammation.

EXPERIMENTAL PROCEDURES
Reagents and Plasmids-Polyclonal anti-human pJAK1, pJAK2, pSTAT-1␣, STAT-1␣, and RNA polymerase II antibodies were purchased from Cell Signaling Technology; anti-human p65 was from Santa Cruz Biotechnology; acetylated histone H3 and H4 was from Upstate Biotech, Inc.; JAK inhibitor 1 was from Calbiochem; recombinant human IFN␥ and TNF␣ were from R&D Systems (Abingdon, Oxon, UK); MB120L and fluticasone propionate were kind gifts from GlaxoSmithKline; FuGENE 6 was from Roche Applied Science; the Dual-Luciferase reporter assay system, Renilla luciferase pRL-SV40 was from Promega; SYBR Green and Excite Master Mix for realtime PCR was from Biogene; primers were from MWG Biotech; ChIP-IT kits were from Active Motif; human and murine CXCL10 ELISA duosets and human and murine recombinant IFN␥ and TNF␣ were from R&D Systems. All other reagents were purchased from Sigma.
The NF-B-dependent luciferase reporter, 6NF-Btkluc (33) was obtained from Dr. Newton (University of Calgary). The CXCL10 promoter-driven luciferase, mutation and deletion constructs has been described previously (34). In all experiments, empty vector experiments were performed, which had little or no significant effect on luciferase activity (data not shown).
Cell Culture-Human tracheas were obtained from three to six post-mortem individuals. Primary cultures of human ASM cells were prepared from explants of ASM according to methods reported previously (25,(35)(36)(37). This protocol was approved by the Nottingham City Hospital Research Ethics Committee. Cells at passages 6 -7 were used for all experiments. We have shown previously that cells grown in this manner have the immunohistochemical and light microscopic characteristics of typical ASM cells.
Experimental Protocol-HASM cells were cultured to confluence in either six-well plates for Western blotting experiments or 24-well culture plates for other all experiments in a humidified, 5% CO 2 , 37°C incubator using Dulbecco's modified Eagle's medium (Sigma) supplemented with 10% fetal calf serum (Seralab, Crawly Down, Sussex, UK), 100 units/ml penicillin, 100 g/ml streptomycin, 4 mM L-glutamine, and 2.5 g/ml amphotericin B (Sigma). The cells were growth-arrested in serum-free medium for 24 h prior to experiments. Immediately before each experiment, fresh serum-free medium containing cytokine or relevant vehicle was added. In concentration response experiments, cells were incubated for 24 h with 0.001-100 ng/ml TNF␣ or IFN␥. When cytokines were used in combination cells were stimulated with submaximal concentrations of each cytokine (0.1 ng/ml TNF␣ and 10 ng/ml IFN␥). Chemical inhibitors and steroids were incubated for 30 -60 min before cytokine exposure as indicated in the figure legend. At the indicated times, the culture media were harvested and stored at Ϫ20°C for subsequent ELISAs.
CXCL10 Assay-The enzyme-linked immunosorbent assay (ELISA) was used to measure CXCL10 (R&D Systems) according to the manufacturer's instructions.
RNA Isolation and Reverse Transcription-Cells in six-well plates were treated with cytokines and collected at times indicated in the text. Total RNA was isolated by using the RNeasy mini kit (Qiagen, West Sussex, UK) following the manufacturer's protocol with on-column DNase digestion. 1 g of total RNA was reverse transcribed in a total volume of 20 l including 200 units of Moloney murine leukemia virus reverse transcriptase (Promega, Madison, WI), 25 units of RNase inhibitor (Promega), 0.5 g of oligo(dT) 15 primer, 0.5 mM each dNTPs, and 1ϫ first-strand buffer provided by Promega. The reaction was incubated at 42°C for 90 min.
Quantitative Real-time RT-PCR-CXCL10 expression was determined using the following primer sequences: sense, 5Ј-GAAATTATTCCTGCAAGCCAATTT-3Ј and antisense, 5Ј-TCACCCTTCTTTTTCATGTAGCA-3Ј (34). ␤ 2 -Microglobulin was used validated for use in this assay and used as the housekeeping gene. 1 ng of reverse-transcribed cDNA was subjected to real-time PCR using Excite Real-time Mastermix with SYBR Green (Biogene, Cambridge, UK) and the ABI Prism 7700 detection system (Applied Biosystems, Warrington, Cheshire, UK) as described previously (25). CXCL10 expression was normalized to the housekeeping gene by dividing the mean of the CXCL10 triplicate value by the mean of the ␤ 2 -microglobulin triplicate value. This was then expressed as fold increase over nonstimulated cells at each time point.
Western Blot Analysis-HASM cells were treated for times indicated in the text, cells were washed, and proteins were extracted as described previously (25). Anti-human pSTAT-1, total STAT-1, or pJAK1 and pJAK2 polyclonal antibody were used at 1:1,000 dilution.
CXCL10 mRNA Stability-Serum-starved HASM cells were treated with or without TNF␣ (1 ng/ml), IFN␥ (100 ng/ml), or TNF␣ (0.1 ng/ml) and IFN␥ (10 ng/ml) in combination for 3 h before the addition of actinomycin D (5 g/ml) for the times indicated to block new transcript generation. RNA was extracted and quantitative real-time RT-PCR of CXCL10 expression was measured as detailed above.
Transient Transfection of CXCL10 Promoter-driven Luciferase Constructs-All transient transfections were conducted by using FuGENE 6 according to the recommended protocol of the manufacturer and as described previously (25). A 1:3 ratio was used; 0.4 g/well plasmid per 1.2 l FuGENE 6, together with 0.8 ng/well of Renilla luciferase reporter/well as an internal control. Relative luciferase activity was obtained by normalizing the firefly luciferase activity against the internal control Renilla luciferase activity. The fold increase was obtained by comparing relative luciferase activity from experiment groups against that from their respective controls. Data are expressed as fold difference, as little and no significant difference was seen in baseline levels observed between each construct when nonstimulated (data not shown). The transfection rate was between 25 and 30% as measured by transfection with a green fluorescent protein expression vector.
Chromatin Immunoprecipiation Assay (ChIP)-HASM cells were cultured to confluence in T75 cm 2 flasks, growth arrested, and incubated with media or cytokines for 60 min. The ChIP assay was performed using the ChIP-IT kit (Active Motif, Rixensart, Belgium) following the manufacturer's protocol and as described previously (25). 4 g of antibody or IgG control was used for each immunoprecipitation and stimulation parameter. IgG binding control levels did not chance significantly from nonstimulated levels for each antibody tested indicating specificity of antibodies used (data not shown). The CXCL10 primers used yielded a 134-bp product corresponding to Ϫ224 to Ϫ90 of the CXCL10 gene promoter and were as follows: forward, 5Ј-TTTGGAAAGTGAAACCTAATTCA-3Ј and reverse, 5Ј-CAGGAACAGCCAGCAGGTTTT-3Ј.
Statistical Analysis-To determine whether there was synergy between ⌻⌵F␣ and IFN␥, the sum of ⌻⌵F␣ (0.1 ng/ml) and IFN␥ (10 ng/ml) alone was compared with ⌻⌵F␣ and IFN␥, Wilcoxon signed rank test was used to determine differences, and a p value Յ 0.05 was assumed to be significant. For all other experiments, statistical analysis was performed on the absolute data using a t test with a Mann-Whitney posttest for single comparisons and a Kruskal-Wallis test with a Dunn's post-test for multiple comparisons. A plus sign (ϩ) represents a significant (p Ͻ 0.05) differences from the nonstimulated control group; and the number sign represents a significant (p Ͻ 0.05) difference from stimulated control group.
To explore synergy between TNF␣ and IFN␥, submaximal concentrations of each cytokine were chosen (10 ng/ml IFN␥ and 0.1 ng/ml TNF␣) to ensure that cytokine release was not supramaximal and could be modulated by pharmacological inhibitors in future experiments. Using these concentrations, IFN␥ induced 61.9 Ϯ 29.2 pg/ml, and TNF␣ induced 567.3 Ϯ 197.5 pg/ml CXCL10. However, addition of these cytokines together synergistically induced CXCL10, inducing 19,920.9 Ϯ 3454.8 pg/ml (Fig. 1A).
These experiments were mirrored at the level of mRNA. After a 6-h stimulation, TNF␣ (1 ng/ml) and IFN␥ (100 ng/ml) increased CXCL10 mRNA from 0 to 0.42 Ϯ 0.14 and 0.096 Ϯ 0.02, respectively. Synergy also was observed at the mRNA level as the mRNA level increased to 1.44 Ϯ 0.07 in the presence of both cytokines (10 ng/ml IFN␥ and 0.1 ng/ml TNF␣; Fig. 1B, n ϭ 3). Actinomycin D (5 g/ml) addition indicated that the increase in CXCL10 mRNA observed was mediated transcriptionally (Fig. 1B).
TNF␣ and IFN␥ Do Not Regulate CXCL10 via Post-transcriptional Mechanisms-Regulation of CXCL10 has been reported previously to be mediated via both transcriptional and posttranscriptional mechanisms (38); therefore, we investigated whether costimulation with TNF␣ and IFN␥ affect CXCL10 mRNA stability. As can be seen in Fig. 1C, CXCL10 decay over time was not affected by TNF␣ or IFN␥ alone, or when cells were costimulated with these cytokines, indicating post-transcriptional regulation does not play a role in this pathway. These data support that shown in Fig. 1B suggesting CXCL10 is regulated transcriptionally under these experimental conditions.

JOURNAL OF BIOLOGICAL CHEMISTRY 29103
TNF␣-and IFN␥-induced CXCL10 Release Is Steroidinsensitive-To determine whether cytokine-induced CXCL10 release is steroid-sensitive, we tested the corticosteroid fluticasone. TNF␣-induced CXCL10 release was concentrationdependently inhibited by fluticasone (from 1811.6 Ϯ 252.2 pg/ml using 1 ng/ml TNF␣ to 62.3 Ϯ 31.9 pg/ml using 10 Ϫ6 M fluticasone; 96.5 Ϯ 1.9% inhibition; Fig. 2A). IFN␥-induced CXCL10 was less sensitive to a steroid, with the highest concentration of fluticasone inhibiting CXCL10 release by 30.6 Ϯ 6.8% (from 535.5 Ϯ 19.8 pg/ml to 369.6 Ϯ 29.8 pg/ml; Fig. 2B). In keeping with data published previously (28,39,40), TNF␣ and IFN␥, when used in combination, induced additional steroid insensitivity, with the highest concentration of fluticasone (10 Ϫ6 M) inhibiting CXCL10 release by 19.2 Ϯ 3.4% (from 17,070.7 Ϯ 4642.9 pg/ml to 14,095 Ϯ 4428.2 pg/ml; Fig. 2C). Preliminary data in our groups also showed that other CC and CXC T cell chemokines such as CXCL11 and CXCL9, which are induced synergistically by IFN␥ and TNF␣, exhibited the same pattern of steroid sensitivity as CXCL10. Indeed, TNF␣-induced ITAC and MIG mRNAs were steroid-sensitive, whereas IFN␥ or IFN␥ and TNF␣ combined induced a steroid-insensitive response. RANTES was induced by TNF␣ only in a steroidsensitive manner (data not shown). Given the functional implications of the lack of effectiveness of steroids in severe disease where the cytokine milieu is heightened, we wanted to determine which signaling pathways were activated by TNF␣ and IFN␥ to induce CXCL10 release with the aim to highlight novel therapeutics for the treatment of steroid-resistant disease.
Synergy Does Not Lie at the Level of NF-B Activation-It is believed the anti-inflammatory effects of glucocorticoids are exerted in part through their ability to inhibit the function of NF-B (41)(42). To probe the role of NF-B in cytokineinduced CXCL10 release, we first used the specific and selective IKK2 inhibitor ML120B (43). TNF␣-induced CXCL10 release was concentration-dependently inhibited by ML120B yet had little effect on IFN␥-induced CXCL10 release (Fig.  3, A and B). This data suggests that TNF␣ uses NF-B to signal, whereas IFN␥ does not. However, as these results do not determine whether synergy lies at the level of NF-B activation, we assessed the ability of these cytokines alone and in combination to activate an NF-B reporter (33). TNF␣ induced a 5.8 Ϯ 0.8fold increase in NF-B reporter activity (raw values, from 1.4 Ϯ 0.4 to 1.7 Ϯ 0.2), yet IFN␥ minimally activated the reporter (1.7 Ϯ 0.2-fold; raw values from 1.4 Ϯ 0.4 to 2.2 Ϯ 0.6). However, when TNF␣ was added in the presence of IFN␥ (at 0.1 ng/ml and 10 ng/ml, respectively), no synergy was observed (4.2 Ϯ 0.5-fold, raw values from 1.4 Ϯ 0.4 to 6.5 Ϯ 2.5; Fig. 3C).
We next probed this pathway by looking at the recruitment of p65 to the CXCL10 promoter by ChIP. TNF␣ induced a 3.28 Ϯ 0.78-fold increase in p65 recruitment, whereas IFN␥ had no effect (1.2 Ϯ 0.07-fold increase). Synergy was not observed when TNF␣ and IFN␥ were added together (2.47 Ϯ 0.58-fold FIGURE 1. TNF␣ and IFN␥ induce CXCL10 transcriptionally. A, HASM cells were incubated with increasing cytokine concentrations (0.01-100 ng/ml) or both cytokines in combination (0.1 ng/ml TNF␣ and 10 ng/ml IFN␥) for 24 h, and CXCL10 was released into the supernatant measured by ELISA (mean Ϯ S.E., n ϭ 7 from independent donors). An asterisk indicates statistically significant synergy observed when cytokines used in combination, p value Յ 0.05). B, for mRNA, cells were pretreated (30 min) with actinomycin D (5 g/ml) prior to cytokine stimulation for 6 h. CXCL10 and ␤ 2 -microglobulin mRNA was measured by real-time PCR, and the relative expression was calculated by dividing CXCL10 values by ␤ 2 -microglobulin at the same data point (mean Ϯ S.E., n ϭ 3 from independent donors). C, serum-starved HASM cells were treated with or without TNF␣ (1 ng/ml), IFN␥ (100 ng/ml), or TNF␣ (0.1 ng/ml) and IFN␥ (10 ng/ml) in combination for 3 h before the addition of actinomycin D (5 g/ml) for the times indicated to block new transcript generation. RNA was extracted, and quantitative real-time RT-PCR of CXCL10 gene expression was measured by real-time PCR. D, 50 to 60% confluent human ASM cells in 24-well plates were cotransfected with Renilla luciferase internal control reporter (0.8 ng/well), and the full-length CXCL10 promoter firefly luciferase reporter construct (0.4 g/well) using FuGENE 6 for 18 h and then stimulated with or without TNF␣ (1 ng/ml), IFN␥ (100 ng/ml), and both cytokines in combination (0.1 ng/ml TNF␣ and 10 ng/ml IFN␥) for 6 h. Relative luciferase activity was obtained by normalizing the firefly luciferase activity against the internal control Renilla luciferase activity. The results are expressed as mean Ϯ the S.E. of three separate experiments performed in quadruplicate. Statistical analysis was performed on the absolute data using a t test with a Mann-Whitney posttest for single comparisons and a Kruskal-Wallis test with a Dunn's post-test for multiple comparisons. ϩ, significant (p Ͻ 0.05) differences from nonstimulated control group; *, significant (p Ͻ 0.05) difference from stimulated control group (mean Ϯ S.E.).
increase; raw values from 0.58 Ϯ 0.1 to 1.79 Ϯ 0.14 for TNF␣, 0.71 Ϯ 0.1 for IFN␥, and 1.35 Ϯ 0.09 together; Fig. 3D). These data suggest that IFN␥-induced CXCL10 release is independent of NF-B activation and that synergy observed with TNF␣ and IFN␥ on CXCL10 release from HASM cells does not lie at the level of NF-B activation and p65 DNA binding.
Synergy Does Not Lie at the Level of STAT-1␣ Activation-IFN␥ is known to activate the JAK/STAT signaling kinases to regulate gene production (44 -45), hence, we probed this pathway in HASM cells. We first used a nonselective JAK inhibitor, JAK inhibitor 1. JAK inhibitor 1 concentration-dependently inhibited IFN␥-induced CXCL10 release, inhibiting release by 100% (from 336.1 Ϯ 60.9 pg/ml to 0 pg/ml; n ϭ 3), indicating that IFN␥ is signaling via this pathway (Fig. 4A).
We also tested the effects of the JAK inhibitor on the release of CXCL10 induced by both cytokines in combination. TNF␣and IFN␥-induced (1 and 10 ng/ml, respectively) CXCL10 release was abrogated by 65.9 Ϯ 6.1% (from 6948.4 Ϯ 1234.4 pg/ml to 2232.9 Ϯ 260.6 pg/ml using 10 Ϫ5 M Jak 1 inhibitor). Fluticasone was ineffective at reducing CXCL10 release under these conditions (Fig. 4D). These data suggest that inhibitors of the JAK/STAT pathway may prove more effective than a steroid in steroid-resistant disease.

Essential Elements in the Human CXCL10 Promoter Region in Response to TNF␣, IFN␥. and Both Cytokines in Combination-
We wanted to confirm the transcription factors involved in cytokine-induced CXCL10 gene transcription at a more molecular level, hence, we transfected HASM cells with luciferase reporter plasmids containing mutations and deletions of the wild-type CXCL10 promoter sequence (Fig. 5A) (34). Data are expressed as fold difference, as little and no significant difference was observed in baseline levels observed between each construct when nonstimulated (data not shown). TNF␣ (1 ng/ml) and IFN␥ (100 ng/ml) increased luciferase activity 2.5 Ϯ 0.2-fold and 4.5 Ϯ 0.8-fold, respectively (Fig. 5, B and C). To determine whether synergism also exists at the promoter level, transfected cells were stimulated with TNF␣ and IFN␥ together (0.1 ng/ml TNF␣ plus 10 ng/ml IFN␥). This resulted in a 13.4 Ϯ 3.4-fold increase in promoter activity, which was significantly higher than either cytokine alone (Fig. 5D).
Next, we compared the responses to the full-length CXCL10 promoter to constructs within which each of the two proximal NF-B recognition sites and the proximal ISRE individually mutated for TNF␣ and IFN␥, respectively. Compared with the native promoter, mutation of either of the NF-B sites led to a significant reduction in promoter activity after stimulation with TNF␣ (Fig. 5B). A reduction also was observed when mutating the proximal ISRE binding site in the full promoter after stimulation with IFN␥ ( Fig. 5C) but not when either NF-kB binding sites or the AP-1 binding site was mutated (data not shown). In keeping with either cytokine alone, mutation of the NF-B and ISRE binding sites also abrogated promoter activity after stimulation with TNF␣ and IFN␥ combined (Fig. 5D). These data confirm the pharmacological data suggesting TNF␣-induced CXCL10 released is controlled by NF-B, IFN␥-induced release by STAT transcription factor and that both play a role when both cytokines are added together. These data support that published by Ohmori and colleagues (32), whereby TNF␣ and IFN␥ independently activated nuclear factors capable of specific interaction with the ISRE and NF-kB sites, respectively.
TNF␣ and IFN␥ Do Not Acetylate Histone H3 or Synergistically Induce Histone H4 Acetylation at the CXCL10 Promoter-Given the lack of synergy observed at the cellular transcription factor activation level, we then postulated that both cytokines in combination could be having an indirect effect by modifying the chromatin environment. Histone H4 acetylation at the promoter was low basally, yet increased by both TNF␣ and IFN␥ (13.6 Ϯ 3.0-fold and 7.4 Ϯ 1.4-fold, respectively; Fig. 6A). How-  6A). We also report a lack of synergy on H4 acetylation at the CXCL10 promoter when TNF␣ and IFN␥ were used at 0.1 ng/ml and 10 ng/ml, respectively, confirming these data. No acetylation of histone H3 at the CXCL10 promoter was observed with either cytokine alone or in combination (Fig. 6B). These data in would suggest that the level of synergy is not due to modification of the histone H3 or H4 acetylation status at the CXCL10 promoter.
TNF␣ and IFN␥ Synergistically Recruit CBP and RNA Polymerase II to the CXCL10 Promoter-We have reported previously that TNF␣ recruits p300/CBP-associated factor to the CCL11 promoter in a PKC␤-dependent manner in HASM cells; therefore, we investigated the effects of TNF␣ and IFN␥ on p300, CBP, and p/CAF recruitment to the CXCL10 promoter (25). Both cytokines, either on their own or in combination, did not recruit p300 and p/CAF to the CXCL10 promoter. However, TNF␣ and IFN␥ in combination induced a 5.5 Ϯ 1.2-fold increase in CBP binding to the CXCL10 promoter (Fig. 6C).
We also report that TNF␣ and IFN␥ in combination recruit RNA polymerase II to the CXCL10 promoter, allowing heightened transcription of the CXCL10 gene. TNF␣ and IFN␥ alone had a small effect on recruitment, inducing a 2.26 Ϯ 0.18-fold and 2.39 Ϯ 0.29-fold increase, respectively. However, when both cytokines were added in combination, a 5.85 Ϯ 0.29-fold increase in RNA polymerase II at the CXCL10 promoter was achieved (raw values of immunoprecipitated RNA polymerase II/input were as follows: control, 0.014 Ϯ 0.003; TNF␣, 0.029 Ϯ 0.005; IFN␥, 0.29 Ϯ 0.004; TNF␣ and IFN␥, 0.072 Ϯ 0.013; Fig. 6D).
These data suggest that TNF␣ and IFN␥ induce CXCL10 release synergistically via RNA polymerase II activation, which results in a heightening in gene transcription. This is the first report in primary airway cells.

DISCUSSION
The first aim of the current study was to find the mechanism of the synergistic enhancement of CXCL10 release in HASM cells after TNF␣ and IFN␥ treatment. We found that the synergy between TNF␣ and IFN␥ on CXCL10 release was not due to the combination of these agents causing increased activity of the two transcription factors NF-B and STAT-1␣ that these two cytokines act through, respectively. We also report that it is not due to increased binding of these transcription factors to the CXCL10 promoter or enhanced acetylation of histone H4, an epigenetic change that we have shown is responsible for the activation of other inflammatory gene promoters in these cells, such as CCL11 and CXCL8 (24 -25). We found that TNF␣ and IFN␥ synergistically recruited the coactivator CBP to the CXCL10 promoter along with heightened recruitment of RNA polymerase II to the promoter, suggesting that this was the main cause of there synergistic effect on CXCL10 release.
Second, we wanted to determine the effect of transcription factor inhibitors in this steroid resistant model of airway inflammation, given the pathogenic role CXCL10 exerts in steroid-resistant airway disease, e.g. viral-induced asthma exacerbations and COPD (3,8). We found that TNF␣-induced CXCL10 release was inhibited by both the steroid fluticasone . C, 50 to 60% confluent human ASM cells in 24-well plates were cotransfected with Renilla luciferase internal control reporter (0.8 ng/well) and an NF-B reporter (0.2 g/well) using FuGENE 6 for 18 h and then stimulated with TNF␣ (1 ng/ml), IFN␥ (100 ng/ml), and both cytokines in combination (0.1 ng/ml TNF␣ and 10 ng/ml IFN␥) for 6 h. Relative luciferase activity was obtained by normalizing the firefly luciferase activity against the internal control Renilla luciferase activity. The results are expressed as mean Ϯ the S.E. of four separate experiments performed in quadruplicate. D, confluent and serum-starved HASM cells in 75-cm 2 dishes were incubated with TNF␣ (1 ng/ml), IFN␥ (100 ng/ml), or both cytokines in combination (0.1 ng/ml TNF␣ and 10 ng/ml IFN␥) for 1 h. The in vivo protein-DNA complexes were cross-linked by formaldehyde treatment, and chromatin pellets were extracted and sonicated. p65 was immunoprecipitated with a specific antibody, and the associated CXCL10 promoter DNA was amplified by real-time PCR. The input represents PCR products from chromatin pellets prior to immunoprecipitation and is used to normalize the data. An IgG control sample (IgG) also was used to determine nonspecific background levels. Statistical analysis was performed on the absolute data using a t test with a Mann-Whitney posttest for single comparisons and a Kruskal-Wallis test with a Dunn's post-test for multiple comparisons. ϩ, significant (p Ͻ 0.05) differences from nonstimulated control group; *, significant (p Ͻ 0.05) difference from stimulated control group (mean Ϯ S.E., n ϭ 3). DMSO, dimethyl sulfoxide. and the selective IKK2 inhibitor ML120B (43). However, in contrast to previous reports (6), IFN␥ neither induced NF-B activation, nor was the release of CXCL10 induced by IFN␥ inhibited by IKK2 kinase inhibition, indicating that this release was due to the activation of another transcription factor in these cells. Reports have emerged in the literature suggesting that IFN␥ is able to modulate CXCL10 release in human bronchial epithelial cells in an IKK2 manner, yet independently of NF-B (47). The data presented here suggest that in HASM cells this is not the case. Further investigation highlighted the JAK2/STAT-1␣ pathway to be instrumental in the induction of CXCL10 by IFN␥. Pharmacological inhibition using a JAK inhibitor abrogated CXCL10 release by IFN␥ in addition to when both cytokines were used in combination and was more effective than the steroid fluticasone, highlighting a potential novel therapy for steroid resistant airway inflammation through the inhibition of JAK/STAT signaling pathway. Preliminary data from our group also suggested that this phenomenon is not restricted to CXCL10. We report that other IFN␥-inducible chemokines such as ITAC and MIG are synergistically enhanced by TNF␣ and IFN␥ and that the steroid flucticasone is ineffective at inhibiting this response. We have not as yet tested the effects of inhibition of JAK on these genes; however, these data support the finding that inhibition of this pathway may be of clinical relevance in severe steroid resistance disease.
We used HASM cells as they are a good model system for studying inflammatory gene transcription in primary human  ). B, HASM cells grown on six-well plates were growth-arrested and treated with IFN␥ (100 ng/ml) for 0 -240 min. Protein was extracted and separated on a 10% SDS-PAGE gel and probed with an anti-human pJAK1, pJAK2, pSTAT-1␣, or STAT-1␣ antibody. HT-29 cells stimulated with IL-4 were used as a positive control. Blots are representative of three independent experiments with similar results. C, confluent and serum-starved HASM cells in 75-cm 2 dishes were incubated with TNF␣ (1 ng/ml), IFN␥ (100 ng/ml), or both cytokines in combination (0.1 ng/ml TNF␣ and 10 ng/ml IFN␥) for 1 h. The in vivo protein-DNA complexes were cross-linked by formaldehyde treatment, and chromatin pellets were extracted and sonicated. STAT-1␣ was immunoprecipitated with a specific antibody, and the associated CXCL10 promoter DNA was amplified by real-time PCR. The input represents PCR products from chromatin pellets prior to immunoprecipitation and is used to normalize the data. An IgG control (IgG) sample also was used to determine nonspecific background levels. Statistical analysis was performed on the absolute data using a t test with a Mann-Whitney posttest for single comparisons and a Kruskal-Wallis test with a Dunn's post-test for multiple comparisons. ϩ, significant (p Ͻ 0.05) differences from nonstimulated control group; *, significant (p Ͻ 0.05) difference from stimulated control group (mean Ϯ S.E., n ϭ 3). DMSO, dimethyl sulfoxide. cells. We have previously used these cells to characterize the transcription factors and epigenetic events regulating the production of other chemokines and growth factors in airway and pulmonary vascular smooth muscle cells (24 -25, 48 -50).
In this study, we initially determined whether TNF␣ and IFN␥ were able to synergistically enhance CXCL10 release via the recruitment of p65 or STAT-1␣ to the CXCL10 promoter and then further determined whether these cytokines in combination altered histone H3 and/or H4 acetylation. IFN␥ has been reported to synergize with TNF␣ to augment expression of CD38 (51) and several chemokines, including CCL5, CXCL10, and CX3CL1 (6, 26 -27). Most studies that used a combination of IFN␥ and TNF␣ showed that the synergistic action involves several molecular mechanisms. In some instances, their cooperatively may be explained by the IFN␥induced up-regulation of TNF␣ receptors (52) or vice versa (53). Furthermore, both cytokines may collaborate at the gene level by increasing promoter activation through a synergistic interaction between transcription factors activated by IFN␥ (STATs, IFN regulatory factor-1) and TNF␣ (NF-B) (54 -55). In the current experiments, the submaximal concentrations used were able to enhance synergistically CXCL10 release but have no effect on the activation of transcription factor binding; thus, we believe this mechanism does not explain our findings. Another mechanism underlying such cooperation could be the induction of defined genes including RANTES as well as CD38 by TNF␣ via activation of the autocrine action of IFN␤ (39,51). Although we saw no synergistic recruitment of p65, STAT-1␣, or histone H4 when TNF␣ and IFN␥ were used in combination, we did observe a synergistic recruitment of the coactivator CBP and RNA polymerase II to the CXCL10 promoter, heightening CXCL10 inflammatory gene regulation. With respect to the steroid resistance observed under these conditions, it is not known whether or not regulation of CXCL10 by transcription factors or cofactor recruitment to the CXCL10 promoter confer steroid insensitivity. Further work is needed to pinpoint the precise mechanism of steroid resistance.
Although we and others (6) report a synergistic induction of CXCL10 after stimulation of HASM cells with TNF␣ and IFN␥, it has been reported that IFN␥ can inhibit TNF␣-induced NF-B reporter activity and mediator release (56). The molecular mechanisms are unclear but may involve the recruitment of other transcription factors in addition to effects on coactivators such as CBP in this study and/or additional cofactors as yet undefined. This is the first study in HASM cells reporting the synergistic regulation of a coactivator and RNA polymerase II by inflammatory cytokines. Previously, Lee and colleagues (57) reported that in A549 cells, IL-1␤ recruits RNA polymerase II, along with p65 and c-Jun to the TGF␤1 promoter, in addition to histone H4 and H3 acetylation. Wada and colleagues (58) also reported that IL-1␤ enhanced recruitment of RNA polymerase II to the secretory leukocyte protease inhibitor gene. Additionally, Hiroi and colleagues (59) report similar findings to our results, in that the STAT-1/NF-B-dependent transcriptional synergy observed after TNF␣ and IFN␥ stimulation on CXCL9 release in mouse NIH3T3 and human embryonic kidney 293 cells could result from the enhanced recruitment of RNA polymerase II complex to the CXCL9 promoter via simultaneous interaction of CBP with STAT-1 and NF-B.
Although steroid resistance is observed in only 5-10% of the asthmatic population, this leads to morbidity and mortality, and these subjects account for Ͼ50% of the United Kingdom healthcare costs for asthma (Ͼ2 billion pounds/annum; 60). FIGURE 5. Identification of promoter cis-elements in cytokine-induced CXCL10 gene transcription. A, fulllength promoter used in these studies (Ϫ972). 50 -60% confluent HASM cells in 24-well plates were cotransfected with the internal control Renilla luciferase (Luc) reporter plasmids (0.8 ng/well) and the CXCL10 promoter firefly reporter plasmids (0.4 g/well) depicted in the figure for 18 h using FuGENE 6 transfection reagent and then treated with or without TNF␣ (B; 1 ng/ml), IFN␥ (C; 100 ng/ml), and both cytokines in combination (D; 0.1 ng/ml TNF␣ and 10 ng/ml IFN␥) for 6 h. Relative luciferase activity was obtained by normalizing the firefly luciferase activity against the internal control Renilla luciferase activity. Statistical analysis was performed on the absolute data using a t test with a Mann-Whitney posttest for single comparisons and a Kruskal-Wallis test with a Dunn's post-test for multiple comparisons. ϩ, significant (p Ͻ 0.05) differences from nonstimulated control group; *, significant (p Ͻ 0.05) difference from stimulated control group (mean Ϯ S.E., n ϭ 3). ISRE, interferon-stimulated response element.
Additionally, steroids are relatively ineffective in acute exacerbations of asthma induced by virus (1,2), where CXCL10 is increased (8). Alternative therapeutic interventions are therefore needed to address this unmet clinical need.
Inhibition of JAK3 has been reported to be an effective target in a murine collagen-induced and rat adjuvant-induced model of rheumatoid arthritis (61) and can drastically improve allograft survival post kidney transplantation (62). More recently, inhibition of this kinase has proven to be an effective anti-inflammatory therapy in a murine model of allergic pulmonary inflammation (63); therefore, clearly, these isoforms are tractable and can be developed for inflammatory diseases. Our data suggest that abrogation of JAK2 and subsequent STAT-1 signaling is more effective than the steroid fluticasone in an in vitro model of steroid-resistant inflammation and may provide a point of intervention in diseases such as severe asthma and COPD that are refractory to such treatment.