Mechanism of IFN-β-Mediated Inhibition of IL-8 Gene Expression in Astroglioma Cells¹

Interleukin-8 is a small m.w. chemokine of the CXC family (1, 2, 3, 4). Like other CXC members, IL-8 contains four conserved cysteines that form intramolecular disulfide bonds to impart biological activity (1, 2, 3, 4). In vivo, IL-8 is strongly chemotactic for neutrophils, is moderately motogenic for other cell types (3), and also displays angiogenic properties (1). Although produced most abundantly by activated monocytes, many cell types are capable of producing IL-8 in response to stimuli, including LPS, PMA, IL-1, and TNF-α, or stress factors, including hypoxia, anoxia, acidosis, hyperglycemia, hyperosmotic pressure, and oxidative stress (3, 4, 5, 6, 7). In general, these stimuli or stress factors activate transcription factors that bind to response elements within the IL-8 promoter to initiate expression (6, 7).

The minimal IL-8 promoter is 133 bp and contains response elements for NF-κB, AP-1, and C/EBPβ. Among these sites, the NF-κB element (κB) is the sole cis-acting element regulating IL-8 expression in all cell types, while the AP-1 and C/EBPβ elements are required for maximal IL-8 expression in a cell type-specific manner (6, 7). Normally, NF-κB is inactive and cytoplasmically sequestered by the inhibitory protein IκB (8). However, in response to activating stimuli such as IL-1, TNF-α, or PMA, IκB is phosphorylated and subsequently degraded (8). This liberates and enables NF-κB to translocate to the nucleus, where it undergoes activating posttranslational modifications and binds to κB elements to regulate gene transcription (8). NF-κB signaling is terminated by IκB resynthesis, nuclear export, and inactivating posttranslational modifications (8). Unfortunately, these measures sometimes fail and constitutive NF-κB and dysregulated IL-8 expression have been associated with many respiratory diseases and various cancers, including human gliomas. Regarding gliomas, IL-8 is believed to increase tumorigenicity via its motogenic, mitogenic, and angiogenic properties (3). As such, mechanisms to restore IL-8 regulation or reduce IL-8 expression are considered attractive therapeutic avenues in future cancer therapies.

Previous studies have determined that small molecules or cytokines including glucocorticoids, IL-4, IL-10, TGF-β1, and IFN-β, are capable of inhibiting IL-8 expression in a variety of cell types (9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19). In particular, glucocorticoids inhibit IL-8 expression by inducing IκB synthesis and by binding to and inhibiting NF-κB activity at the IL-8 promoter (20, 21). Likewise, IL-10 also inhibits IL-8 expression by preventing NF-κB activation, interfering with NF-κB’s ability to bind DNA, and increasing the turnover of IL-8 mRNA (12, 13, 14). Studies regarding IL-4 are not as detailed, but indicate that IL-4 reduces transcription of the IL-8 gene by affecting NF-κB transcriptional activity (22). The mechanism of IFN-β-mediated inhibition of IL-8 remains undefined.

IFN-β, a type I IFN, is expressed by many cells in response to bacterial and viral infection and mediates antiproliferative and antiviral effects (23, 24). IFN-β exerts these effects by binding to its cell surface receptor, which is composed of two subunits, IFNAR1 and IFNAR2 (23, 24). Upon ligand binding, these subunits dimerize and bring into proximity two cytoplasmically bound kinases, Tyk2 and Jak1 (23, 24). These kinases auto- and transphosphorylate each other and the cytoplasmic domain of the IFN-β receptor (23, 24). The phosphorylated receptor then recruits STAT-1 and STAT-2, which are also tyrosine phosphorylated and then released (23, 24). Once free, a minor population of phosphorylated STAT-1 molecules may homodimerize to form the GAF transcription factor complex (24). However, in response to IFN-β, most STAT-1 molecules heterodimerize with STAT-2, translocate to the nucleus and recruit a third protein, p48/IFN regulatory factor 9, to form the heterotrimeric IFN-stimulated gene factor 3 (ISGF3)³ transcription factor complex (24). Both the γ-activated factor and ISGF3 complexes positively regulate gene expression by binding to IFN-γ activation site (GAS) or IFN-stimulated response element (ISRE) elements, respectively (24). Interestingly, although IFN-β inhibits the expression of IL-8, the IL-8 promoter does not contain GAS or ISRE elements, and at present, how IFN-β inhibits the expression of IL-8 is not known. However, early studies investigating this phenomenon identified two important details: 1) the κB element (−70 to −80) was required for IFN-β-mediated inhibition of IL-8 expression, and 2) de novo protein synthesis was not required for IFN-β to mediate inhibition of IL-8 expression (25, 26).

In this study, we addressed the question of how IFN-β inhibits IL-8-gene expression. Using chromatin immunoprecipitation (ChIP) assays, we defined the transcriptional program of the IL-8 promoter in response to PMA and describe the events required for regulated IL-8 expression in U87-MG cells. Additionally, we analyzed the effects of IFN-β on the transcriptional program and determined that IFN-β reduces histone acetylation at the IL-8 promoter and diminishes the levels of NF-κB p65, C/EBPβ, and RNA Pol II at the IL-8 promoter, while increasing the levels of NF-κB p50, silencing mediator of retinoic acid and thyroid hormone receptors (SMRT), histone deacetylase (HDAC)-1 and HDAC-3. We show that IFN-β does not affect IFN-γ-inducible protein-10 (IP-10) or MCP-1 expression, genes also regulated by NF-κB, nor does IFN-β disturb NF-κB activation, nuclear translocation, or binding to the IP-10 promoter. As such, we believe that the effects of IFN-β are specific to the IL-8 promoter, and characterize the IL-8 gene as an NF-κB-regulated gene that is negatively regulated by IFN-β.

The IL-8-Luc reporter plasmid containing 133 bp of the human IL-8 promoter has been previously described (27) and was used in this study. The pcDNA3 vector was purchased from Promega.

U87-MG cells were obtained from American Type Culture Collection and maintained in MEM medium supplemented with 2 mM l-glutamine, 100 U/ml penicillin, 10 μg/ml streptomycin, nonessential amino acids, sodium pyruvate, and 10% heat-inactivated FBS, as previously described (28).

PMA was purchased from Calbiochem. Recombinant human IFN-β and human TNF-α were purchased from R&D Systems. The secondary peroxidase-conjugated Abs and ECL reagents were purchased from Amersham. The anti-p65 and anti-SMRT Abs were purchased from Abcam. The anti-p50, anti-c/EBPβ, anti-c-Fos, anti-c-Jun, anti-p300, anti-HDAC-1, and anti-HDAC-3 Abs were purchased from Santa Cruz Biotechnology. The anti-RNA polymerase II (Pol II), anti-phosphoserine-2 RNA Pol II, and anti-phosphoserine-5 RNA Pol II Abs were purchased from Covance. The anti-acetylated histone 3 and histone 4 Abs were purchased from Upstate Cell Signaling Solutions. Normal rabbit and mouse serum were purchased from Santa Cruz Biotechnology. Protein A/G beads were purchased from Upstate Cell Signaling Solutions. The Alexa Fluor 594 goat anti-rabbit secondary Ab was purchased from Molecular Probes. Hoechst was purchased from Sigma-Aldrich.

Cells were incubated in the absence (unstimulated (UT)) or presence of PMA (50 ng/ml), TNF-α (50 ng/ml), IFN-β (500 U/ml), or combinations for 24 h in serum-free medium. Concentrations of IL-8 in the supernatants were assayed by using a dual-Ab solid-phase ELISA for IL-8 (BioSource International). Cells were washed with PBS and lysed, and then the amount of total protein was measured by using a Bio-Rad protein assay kit. IL-8 expression was normalized to the amount of total protein. Results of at least three experiments are shown as picograms of IL-8 protein per milliliter of supernatant (mean ± SEM).

Cells were incubated in the absence (UT) or presence of PMA (10 or 50 ng/ml), TNF-α (50 ng/ml), IFN-β (500 U/ml), or combinations for 3 h in serum-free medium. Total RNA was isolated using TRIzol reagent (Invitrogen Life Technologies) as previously described (29). Ten to 20 μg of total RNA was hybridized with the hCK5 riboprobe (10 × 10⁴ cpm) (BD Pharmingen) at 56°C overnight. The hybridized mixtures were then treated with RNase A/T1 (1/200) at 37°C for 30 min, precipitated, and analyzed by 5% denaturing (8 M urea) PAGE. The gels were dried and exposed to PhosphorImager cassettes. Quantification of protected RNA fragments was performed using the PhosphorImager (Molecular Dynamics). Values for IL-8, IP-10, and MCP-1 mRNA expression were normalized to GAPDH mRNA levels for each experimental condition. Fold induction is shown and is representative of three experiments.

The IL-8-Luc reporter plasmid containing 133 bp of the human IL-8 promoter has been previously described (27) and was used in this study. Transient transfection was performed using the Fugene 6 reagent (Roche Diagnostics) as previously described (29). Cells were incubated in the absence (UT) or presence of PMA (50 ng/ml), TNF-α (50 ng/ml), IFN-β (500 U/ml), or combinations for 18 h in serum-free medium. Cell extracts were assayed in triplicate for luciferase activity as previously described (29), and were normalized to total protein. Protein concentrations were measured using the Bio-Rad Protein Assay. The luciferase activity from the vector control was arbitrarily set at 1 for calculation of fold induction. Results of at least four experiments are shown as fold induction (mean ± SEM).

ChIP assays were performed as previously described (29, 30, 31). Nuclei from cross-linked cells were resuspended in Tris/EDTA buffer and sonicated. The soluble chromatin was adjusted into radioimmunoprecipitation buffer (0.1% SDS, 1% Triton X-100, 0.1% sodium deoxycholate, and 140 mM NaCl) and precleared. Immunoprecipitation was performed with 2–5 μg of appropriate Abs, and the immune complexes were absorbed with protein A beads or protein A/G beads (Upstate Cell Signaling Solutions) blocked with BSA and salmon sperm DNA. Immunoprecipitated DNA was subjected to semiquantitative PCR. The PCR products were resolved in 2.0% agarose gels in 1× Tris/acetate/EDTA buffer, and the gels were stained with ethidium bromide. Densitometry was used to quantify the PCR results, and all results were normalized by respective input values. Fold induction is shown and is representative of at least three experiments. The human IL-8 promoter was analyzed using primers previously described: IL-8 forward 5′-TTC CTT CCG GTG GTT TCT TC-3′ and IL-8 reverse 5′-GGG CCA TCA GTT GCA AAT C-3′ (21). The human IP-10 promoter was analyzed using the following primers: IP-10 forward, 5′-GAG CTG AAC CCC ATC GTA AA-3′ and IP-10 reverse, 5′-ATA GGA CG CCT GCT TTG A-3′.

Cells were seeded on 8-well chamber slides and grown overnight. At 24 h, cells were serum starved and then grown in the absence (UT) or presence of IFN-β (500 U/ml), PMA (50 ng/ml), or both for 30 min. Slides were then washed with PBS and fixed for 15 min in 10% formalin. The slides were washed twice with PBS and blocked in PBS supplemented with 10% goat serum for 15 min. Human NF-κB p65 expression was detected with rabbit anti-p65 Ab (Abcam) followed by Alexa Fluor 594 goat anti-rabbit Ab (Molecular Probes) for 1 h at room temperature. Nuclei were visualized by staining with Hoechst (Sigma-Aldrich).

Levels of significance for comparisons between samples were determined using Student’s t test distribution.

Many cell types are capable of expressing IL-8 protein in response to stimuli that activate NF-κB, such as TNF-α and PMA (3, 4, 5, 6, 7). As shown in Fig. 1,A, although human glioblastoma U87-MG cells constitutively expressed modest levels of IL-8 protein, TNF-α and PMA increased IL-8 protein levels ∼4-fold. To determine whether the changes induced by TNF-α and PMA were due to changes in the levels of IL-8 mRNA, U87-MG cells were grown in the absence or presence of TNF-α or PMA for 3 h, and levels of IL-8 mRNA were analyzed by RPA. In the absence of any stimuli, IL-8 mRNA levels are low but detectable, however, both TNF-α and PMA substantially increased the levels of IL-8 mRNA (Fig. 1,B). Finally, to confirm that the changes in IL-8 protein and mRNA induced by TNF-α and PMA were due to de novo transcription, U87-MG cells were transiently transfected with a vector encoding the luciferase gene under the regulation of the 133-bp region of the IL-8 promoter (IL-8-Luc) in the absence or presence of TNF-α or PMA. Both TNF-α and PMA treatment stimulated IL-8 promoter activity by ∼5.7- and 7.0-fold, respectively (Fig. 1,C). Although the fold induction in IL-8 mRNA is higher than that for either IL-8 protein or IL-8 promoter activity, IL-8 mRNA levels were measured using densitometry, which reveals relative differences and thus these numbers may overestimate the true fold difference. As such, we have greater confidence in the more conservative measurements made via ELISA (Fig. 1,A) or reporter assays (Fig. 1 C). Therefore, these data confirm that TNF-α and PMA are able to induce IL-8 gene expression in U87-MG cells.

Studies have shown that IL-8 expression is regulated by a sequence spanning nucleotides −1 to −133 of the upstream DNA flanking the IL-8 gene, and that this region is essential and sufficient for IL-8 expression (6, 7). Within this defined region exists a TATA box and response elements for AP-1 (−120 to −126), C/EBPβ (−81 to −92), and NF-κB (−70 to −80) (6, 7). To understand the timing and the order of events required for PMA-induced expression of IL-8, we used ChIP assays to study the in vivo interactions between proteins involved in transcription and/or chromatin remodeling at the IL-8 promoter in U87-MG cells. In the absence of stimulation, the IL-8 promoter contained high basal levels of acetylated histones 3 (Ac-H3) and 4 (Ac-H4) (Fig. 2, group 1). Interestingly, in the absence of any stimuli, we were unable to detect CREB-binding protein (data not shown) or p300 (Fig. 2, group 1) on the IL-8 promoter, suggesting that another histone acetyltransferase may be responsible for the basal levels of histone acetylation present at the IL-8 promoter. Because histone acetylation is believed to render chromosomal domains more accessible to DNA-binding proteins (32), our data suggest that transcription factors such as NF-κB and C/EBPβ may be able to readily access the IL-8 promoter. In the absence of PMA, the levels of NF-κB (p65 or p50) and C/EBPβ associated with the IL-8 promoter were low compared with the levels of c-Fos and c-Jun (AP-1 proteins) (Fig. 2, group 2). Unlike NF-κB and C/EBPβ, which require cytoplasmic activation, AP-1 proteins are usually bound to their response elements and are regulated by their abundance and posttranslational modifications (33, 34). Together, these data suggest that the IL-8 promoter may be accessible to DNA-binding proteins, yet transcriptionally inactive due to the presence of other factors, including corepressors.

SMRT is a corepressor that has been shown to interact with HDACs, leading to histone hypoacetylation and/or transrepression of target transcription factors, including NF-κB (35, 36). As shown in Fig. 2, group 3, SMRT is present at the IL-8 promoter in the absence of PMA. Although SMRT has been shown to interact with both HDAC-1 and HDAC-3 (35, 36, 37), only HDAC-1 but not HDAC-3 was present in appreciable levels at the IL-8 promoter in the absence of stimuli. These data suggest that while the IL-8 promoter may be acetylated and accessible, other factors, including the SMRT-HDAC-1 complex, may keep the IL-8 gene transcriptionally silent.

RNA Pol II is the eukaryotic RNA polymerase responsible for transcription. The activity of RNA Pol II is regulated by posttranslational modifications to its C-terminal domain (CTD). Within the CTD, RNA Pol II is phosphorylated on Ser² (pS2) and Ser⁵ (pS5). Previous reports have determined that pS5-RNA Pol II correlates with RNA Pol II that is positioned at a promoter and in a “ready” phase, while pS2-RNA Pol II correlates with an RNA Pol II that is in the elongation phase (38, 39). In the absence of PMA, RNA Pol II was undetectable using an Ab against the CTD, although we were able to detect RNA Pol II at the IL-8 promoter using Abs that recognize RNA Pol II (pS2) or RNA Pol II (pS5) (Fig. 2, group 4). This apparent discrepancy may be the result of the RNA Pol II CTD Ab, which may not or may only poorly recognize the phosphorylated forms of RNA Pol II. Thus, our data suggest that while RNA Pol II is present at the IL-8 promoter in the UT state, it may not be transcriptionally active.

Above, our data indicate that the IL-8 promoter is acetylated, yet is transcriptionally silent for several reasons: 1) the presence of SMRT and HDAC-1, 2) low levels of NF-κB and C/EBPβ, and 3) low levels of RNA Pol II (pS2). Therefore, we hypothesized that stimuli that induce IL-8 expression may alter conditions at the IL-8 promoter to promote transcription. Previously, we had determined that IL-8 mRNA is detectable at 1 h, peaks at 3 h, and begins to taper off at 4 h (data not shown). As such, we used ChIP assays to analyze the IL-8 promoter in response to PMA over a time period of 15 min to 4 h. PMA was chosen as a representative stimulus of IL-8 expression in the U87-MG cells. The levels of Ac-H3 increase upon PMA stimulation (Fig. 2, group 1), peaking at 1 h, suggesting that the promoter may undergo additional acetylation. Indeed, following stimulation with PMA, there was a noticeable increase in the level of p300 associated with the IL-8 promoter at 30–45 min (Fig. 2, group 1), which may account for the increased levels of Ac-H3 histone acetylation.

Because IL-8 expression is rapidly induced by PMA, we speculated that NF-κB and perhaps C/EBPβ would be recruited to the IL-8 promoter following PMA stimulation. Interestingly, recruitment of C/EBPβ and NF-κB p65, but not NF-κB p50, is enhanced at the IL-8 promoter as early as 30-min poststimulation (Fig. 2, group 2). Although p65/p50 is the prototypic NF-κB form, our results are consistent with previous reports showing that the κB element within the IL-8 promoter is unique and readily bound by p65 homodimers, but not the p65/p50 form of NF-κB (40). Regarding the AP-1 proteins, while the levels of c-Jun are modestly affected by PMA stimulation, the levels of c-Fos increase appreciably between 45-min and 3-h poststimulation (Fig. 2, group 2).

Although transcription factor recruitment would indicate that the IL-8 gene is active, the presence of corepressors may keep the IL-8 silent gene. However, the levels of SMRT and HDAC-1 are substantially reduced at 15-min through 3-h poststimulation, and little HDAC-3 is detected during through 2-h poststimulation (Fig. 2, group 3).

Eukaryotic transcription is typically characterized by notable changes in the phosphorylation status of RNA Pol II. Above, we determined that in the absence of stimulation, RNA Pol II is present and in the ready phase (pS5) at the IL-8 promoter. However, upon stimulation, the levels of RNA Pol II (pS5) are substantially reduced and the levels of total and pS2 RNA Pol II rapidly increase (Fig. 2, group 4). As such, these data suggest that the IL-8 gene uses a transcriptional program to rapidly induce IL-8 expression.

Because we had determined that IL-8 mRNA expression begins to diminish at 4-h poststimulation (data not shown), this suggested that events occur at the IL-8 promoter to attenuate IL-8 expression. Indeed, at 1-h poststimulation, the levels of p300 are diminished and by 2 h, the levels of Ac-H3 and Ac-H4 are also substantially reduced (Fig. 2, group 1), suggesting that the IL-8 promoter may become less accessible and hence less transcriptionally active.

Interestingly, as the levels of histone acetylation diminish, less p65 and C/EBPβ are associated with the IL-8 promoter, while at 4 h, the levels of p50, are substantially increased (Fig. 2, group 2). Because p50 lacks a transactivation domain, its presence at the IL-8 promoter may help explain the reduction in IL-8 expression noticed at this time. Additionally, while the levels of c-Jun appear unaffected over time, the levels of c-Fos are reduced at 4-h poststimulation (Fig. 2, group 2).

Because our data above indicate that corepressors regulate IL-8 expression in the absence of PMA, we postulated that these proteins may also participate in attenuating IL-8 expression. As shown in Fig. 2, group 3, at 3-h poststimulation, the levels of HDAC-3, but not HDAC-1, are substantially increased, and by 4 h the levels of SMRT are also increased. This suggests that the IL-8 promoter is undergoing events to attenuate IL-8 gene transcription. Indeed, by 1-h poststimulation, there are substantial reductions in the total and phosphorylated levels of RNA Pol II present at the IL-8 promoter (Fig. 2, group 4).

To ensure that equal amounts of DNA were used in each experiment, we analyzed DNA before immunoprecipitation (Input) (Fig. 2, group 5). Furthermore, the specificity of each Ab was confirmed using IgG controls (IgG), which failed to immunoprecipitate the IL-8 promoter. As such, the above results define the transcriptional program used at the IL-8 promoter.

Previous studies demonstrated that IFN-β could inhibit the expression of IL-8 in some cell types (15, 18, 25). However, one report indicated that this phenomenon was sometimes lost in tumor-derived or transformed cell lines (25). Therefore, to determine whether IFN-β inhibits IL-8 protein expression in the U87-MG astroglioma line, cells were grown in the absence or presence of PMA and/or IFN-β for 24 h, and IL-8 protein levels determined. As shown in Fig. 3 A, in the absence of stimulation, U87-MG cells express basal amounts of IL-8 protein, which are reduced by IFN-β treatment alone. PMA increased the levels of IL-8 protein ∼6-fold, and when cells were simultaneously treated with both PMA and IFN-β, the levels of IL-8 protein were inhibited by 60% compared with PMA treatment alone. These data indicate that expression of IL-8 protein is inhibited by IFN-β treatment in U87-MG cells.

We sought to determine whether the IFN-β inhibitory effect was mediated at the level of mRNA expression. U87-MG cells were grown in the absence or presence of PMA (10 ng/ml or 50 ng/ml) and/or IFN-β for 3 h and total RNA analyzed by RPA. As shown on the left side of Fig. 3,B, the levels of IL-8 mRNA are minimal in the absence of stimulation or upon treatment with IFN-β alone. However, the levels of IL-8 mRNA increase 10.9-fold and 67.9-fold when stimulated with 10 or 50 ng/ml PMA, respectively, indicating a dose responsiveness to PMA stimulation. Interestingly, when U87-MG cells were simultaneously treated with both IFN-β and either concentration of PMA (Both), the levels of IL-8 mRNA were substantially reduced. To determine whether the inhibitory effect of IFN-β was dependent on the type of stimuli used to induce IL-8, we repeated these experiments using TNF-α. As shown on the right side of Fig. 3 B, IL-8 mRNA levels were robustly increased by TNF-α (48.7-fold), and inhibited in the presence of IFN-β (Both).

Previously, data had shown that the effects of IFN-β on IL-8 expression were dependent upon the presence of the κB element situated at position −70 to −80 within the IL-8 promoter (25, 26, 41). For this reason, it is possible that IFN-β may inhibit the expression of other genes whose promoters also contain κB elements. To determine this, we analyzed the expression patterns of two other chemokines, MCP-1 and IP-10, by RPA using the same samples previously described. The human MCP-1 promoter contains two κB elements (42), while the IP-10 promoter contains two κB elements and an ISRE element (43). Interestingly, the ISRE element is recognized by the ISGF3 transcription factor complex, which is activated by IFN-β (23, 24).

As shown in Fig. 3 B, the levels of IP-10 (second row) and MCP-1 (third row) mRNA are low in untreated U87-MG cells and are relatively unaffected by treatment with IFN-β alone or low doses of PMA alone. In response to TNF-α (right side) or 50 ng/ml PMA (left side), the levels of MCP-1 were increased 3.1- and 1.8-fold, respectively, while the levels of IP-10 remained relatively unchanged. However, in the presence of IFN-β and PMA or TNF-α, the levels of IP-10 were increased ∼2.6- and 5.2-fold, respectively. This is consistent with other reports indicating that the IP-10 gene is synergistically activated by NF-κB and ISGF3 (43). Interestingly, the presence of both IFN-β and PMA or TNF-α slightly enhanced MCP-1 expression. Together, these data suggest that IFN-β does not inhibit the expression of all NF-κB-regulated genes.

We next analyzed IL-8 promoter activity in the absence or presence of PMA and/or IFN-β. As shown in Fig. 3 C, simultaneous treatment of cells with PMA and IFN-β reduced IL-8 promoter activity by 59%, compared with PMA alone. Moreover, we confirmed the inhibitory effect of IFN-β in several additional human astroglioma (U251-MG, CRT-MG, CH235-MG) or fibrosarcoma (HT1080) cell lines (data not shown). Therefore, we propose that IFN-β inhibits IL-8 expression by acting at the promoter level, perhaps by disrupting the transcriptional program required by IL-8 for gene expression.

Although the data in Fig. 3,B suggest that IFN-β does not prevent NF-κB activation or nuclear translocation, we assessed this formally using immunofluorescence to monitor NF-κB movement in response to IFN-β, PMA or both. Cells were grown overnight on coverslips and then left UT or stimulated with IFN-β and/or PMA for 30 min. The activation and nuclear localization of NF-κB was assessed using Abs to detect p65/RelA, while nuclei were visualized with Hoechst dye. As shown in Fig. 4, in the absence of stimuli or in the presence of IFN-β alone, NF-κB is localized to the cytoplasm. However, when cells were treated with PMA, NF-κB was largely detected in or near the nucleus, indicating its nuclear migration. Moreover, in the presence of IFN-β and PMA, the pattern of NF-κB expression appeared similar to cells treated with PMA alone, suggesting that IFN-β does not disrupt or affect NF-κB activation or nuclear translocation.

To analyze changes affected by IFN-β at the IL-8 promoter, we analyzed U87-MG cells grown in the absence or presence of IFN-β, PMA or both at 30-min poststimulation using ChIP assays. Above, we had determined that 30 min represented the earliest time point at which noticeable changes in the transcriptional program could be detected, i.e., decreased levels of corepressors and increased levels of transcription factors and active RNA Pol II, thus this time point was chosen for the experiments described below.

As shown in Fig. 5,A, group 1, the histones of the IL-8 promoter are constitutively acetylated and the levels of Ac-H3 and Ac-H4 increase slightly when stimulated with PMA. Interestingly, although treatment with IFN-β alone had no apparent effect on histone acetylation, the levels of Ac-H3 and Ac-H4 were substantially reduced by cotreatment with both PMA and IFN-β (Fig. 5,A, group 1). As before, we were unable to detect p300 at the promoter in the absence of any stimuli, yet we did detect increased levels of p300 at 30 min following PMA stimulation (Fig. 5,A, group 1). Moreover, while IFN-β alone had no effect on p300 recruitment, the levels of p300 were reduced following stimulation with PMA and IFN-β (Fig. 5 A, group 1). Together, these data indicate that IFN-β reduces the level of histone acetylation at the IL-8 promoter, perhaps by reducing recruitment of p300.

Because our data above showed that cotreatment with PMA and IFN-β substantially reduced the levels of Ac-H3 and Ac-H4, this suggested that the IL-8 promoter may be less accessible to NF-κB and C/EBPβ. As before, in the absence of stimuli, we detect moderate levels of c-Jun and low levels of p65, p50, C/EBPβ, and c-Fos at the IL-8 promoter (Fig. 5,A, group 2). Moreover, the levels of p65, C/EBPβ, and c-Fos at the IL-8 promoter are substantially increased by PMA. Interestingly, while IFN-β alone had no apparent effect, the use of PMA and IFN-β reduced the levels of p65, C/EBPβ and c-Fos, and increased the levels of p50, compared with PMA alone (Fig. 5 A, group 2). These data indicate that IFN-β may inhibit IL-8 expression by reducing histone acetylation and the presence of “positive” transcription factors, while increasing the presence of “negative” transcription factors such as NF-κB p50.

Because our data above showed that corepressors regulate IL-8 expression in the absence of any stimuli, or after 4 h of PMA, we hypothesized that these proteins may also be involved in changes exerted by costimulation with IFN-β and PMA. As shown in Fig. 5,A, group 3, in the absence of any stimuli, the IL-8 promoter is bound by SMRT, HDAC-1, and HDAC-3 to varying degrees, yet each is reduced upon stimulation with PMA. Although IFN-β alone does not affect the levels of HDAC-1, the presence of both PMA and IFN-β increased these levels (Fig. 5,A, group 3). In contrast, IFN-β alone increased the levels of SMRT and HDAC-3, while the presence of both PMA and IFN-β restored the levels of SMRT and HDAC-3 to the levels seen in the absence of any stimulation (Fig. 5 A, group 3).

Next, we assessed whether IFN-β affects the phosphorylation, and presumably the activity, of RNA Pol II. PMA alone increased the levels of total and pS2 RNA Pol II while reducing the levels of pS5 RNA Pol II compared with untreated samples. Interestingly, either IFN-β alone or costimulation with PMA and IFN-β reduced the levels of total and phosphorylated RNA Pol II (Fig. 5,A, group 4). These data suggest that another effect of costimulation with PMA and IFN-β may be to reduce the total levels of RNA Pol II associated with the IL-8 promoter. As before, to ensure that equal amounts of DNA were used in each experiment, we analyzed DNA before immunoprecipitation (Input). Furthermore, we assessed the specificity of each Ab using IgG controls (IgG), which failed to immunoprecipitate the IL-8 promoter (Fig. 5 A, group 5).

Because our data above indicate that IFN-β inhibits the ability of NF-κB p65 to associate with the IL-8 promoter, it is possible that IFN-β signaling affects NF-κB such that its nuclear localization is unperturbed, but the DNA binding and/or transactivation abilities of NF-κB are diminished. To assess this possibility, we analyzed the IP-10 promoter using the ChIP assay. The human IP-10 promoter contains two κB elements and one ISRE element (43). As described previously, neither IFN-β nor PMA alone induces IP-10 expression, however, IP-10 expression can be induced by costimulation with both PMA and IFN-β (Fig. 3,B). In the absence of stimulation, relatively little NF-κB p65 is associated with the IP-10 promoter (Fig. 5,B). Moreover, IFN-β or PMA alone had only moderate effects on p65 recruitment to the IP-10 promoter. Most striking, in the presence of both PMA and IFN-β, p65 increased, suggesting that IFN-β does not negatively influence the ability of NF-κB p65 to associate with DNA (Fig. 5 B).

IL-8 is the prototypic and best-characterized member of the CXC chemokine family (1, 2, 3, 4, 44). IL-8 can be expressed by many cell types in response to inflammatory conditions or other stimuli that activate NF-κB, AP-1, and/or C/EBPβ proteins (1, 2, 3). Previous investigations determined that IFNs, most notably IFN-β, were capable of inhibiting stimuli-induced IL-8 expression (15, 25). However, while IFN-β can regulate gene expression by activating the ISGF3 transcription factor, which binds to ISRE elements present within target genes (24), the IL-8 promoter lacks any recognizable ISRE or GAS elements, which suggests that the mechanism by which IFN-β inhibits IL-8 expression may be more complex. Herein, we describe our approach and the mechanism by which IFN-β inhibits IL-8 expression.

In this study, we analyzed the U87-MG human astroglioma cell line, which has been shown to express low levels of IL-8 in the absence of stimuli. However, in response to either PMA and TNF-α, which activate NF-κB, IL-8 expression is elevated in these cells (Fig. 1). From our own data (Fig. 3) and consistent with previous observations, we determined that IFN-β inhibited IL-8 protein and mRNA expression and reduced IL-8 promoter activity. Using ChIP assays, we characterized the transcriptional program used at the IL-8 promoter (Fig. 2), and determined that the IL-8 promoter is heavily acetylated but bound by the corepressors SMRT and HDAC-1, and to a lesser extent HDAC-3, in the absence of any stimuli (Fig. 6,A). However, because we detected elevated levels of acetylated histones, our data suggest that the IL-8 promoter may be more accessible than other promoters that do not exhibit basal histone acetylation (45), but that the presence of SMRT and the HDACs may limit or prevent unintended transcription. In response to stimulation with PMA, the HDACs and SMRT exit the IL-8 promoter, and both transcription factors and RNA Pol II bind, indicating that the IL-8 promoter is rapidly responsive and readily activated by such stimuli (Fig. 6 B). These data correlate well with IL-8 transcription, which is detectable within 1 h and peaks at 3–4 h poststimulation (data not shown). Interestingly, at 3–4 h poststimulation, the IL-8 promoter undergoes changes that include histone deacetylation, diminished levels of RNA Pol II and transcription factors, and increased levels of NF-κB p50, SMRT, HDAC-1, and HDAC-3.

Because the IL-8 promoter had undergone noticeable changes in response to PMA stimulation as soon as 30-min poststimulation (Fig. 2), we analyzed the response to IFN-β alone or IFN-β plus PMA at this time point. We determined that IFN-β has a multifaceted effect on the IL-8 promoter, that is, we saw reduced levels of histone acetylation, transcription factor and RNA Pol II binding, and increased levels of SMRT and HDAC-1. Together, these data suggest that IFN-β disrupts the IL-8 transcriptional program by preventing or reducing the positive influences on IL-8 transcription (i.e., transcription factors and RNA Pol II) and increasing negative influences (SMRT and HDAC-1) (Fig. 6 C). However, at present it is difficult to discern whether one of these events is an initiating catalyst that proceeds and/or causes others events to ensue.

We are still left to ponder how IFN-β specifically inhibits the expression of IL-8 but not other NF-κB-regulated genes. Other inhibitors, including IL-10 and glucocorticoids, suppress IL-8 expression by preventing or disrupting NF-κB activation (13, 14, 20, 21). However, we have shown that IFN-β does not disturb NF-κB activation or inhibit the expression of two other chemokines, IP-10 and MCP-1, which are also regulated by NF-κB (Fig. 3,B). Moreover, using immunofluorescence, we show that IFN-β does not disrupt NF-κB nuclear translocation (Fig. 4), and the ChIP analysis demonstrates that despite reduced NF-κB p65 levels at the IL-8 promoter in the presence of PMA and IFN-β, NF-κB is still active and competent to bind the IP-10 promoter (Fig. 5 B).

Our data suggest that the effect of IFN-β on IL-8 expression is not simple, that is, IFN-β does not merely prevent the activation of necessary transcription factors, i.e., NF-κB. Instead, IFN-β may alter specific factors or events at the IL-8 promoter that are critical for IL-8 expression. One possibility is that IFN-β may promote or diminish the presence of certain posttranslational modifications to NF-κB p65 that in turn alter the ability of NF-κB p65 to associate with proteins at certain promoters (46, 47, 48). Conversely, the specificity of IFN-β’s influence may be dictated by the nucleotide sequence within the IL-8 κB element (TGGAATTTCC), which is quite different with respect to sequence and orientation from either of the κB elements within the MCP-1 promoter (κB-2: GGGAAATTCC; κB-1: GGGACTTCCC) or those within the IP-10 promoter (κB-2: GGGAATTTCC; κB-1: GGGAACTTCC). Moreover, unlike the MCP-1 and IP-10 promoters, the IL-8 promoter does not bind the prototypic NF-κB p65/p50 heterodimer but appears to prefer either p65 or p50 homodimers (26). As such, the exact sequence of the κB element may determine both NF-κB family member specificity and which coactivators will form productive interactions with the bound NF-κB dimer (49, 50, 51). Indeed, a recent interesting study by Luecke and Yamamoto (52) supports this idea. They showed that glucocorticoids repress IL-8 transcription by interfering with the ability of p-Tefb, a kinase that phosphorylates Ser² of RNA Pol II, to associate in a promoter-specific manner with the IL-8 promoter, while IκB, another NF-κB-regulated gene, was unaffected by glucocorticoids because IκB expression does not rely on p-Tefb recruitment to the IκB promoter (52).

Still, a final question remains unanswered. In particular, what role if any does the ISGF3 transcription factor complex have in mediating IFN-β mediated inhibition of the IL-8 gene? At present, the absence of any ISRE or GAS elements and the absence of detectable ISGF3 components at the IL-8 promoter suggest that IFN-β does not inhibit IL-8 expression by causing ISGF3 to bind to and act as a transrepressor at the IL-8 promoter. However, these data do not exclude ISGF3 or various components of the ISGF3 complex from participating in the effects of IFN-β on the IL-8 promoter. It remains possible that one or more of the ISGF3 components participate in this process in a manner not yet described. Our current efforts are working to address these questions.

We thank Shaun Sparacio, Dr. Zhendong Ma and Dr. Younhee Choi for technical assistance.

The authors have no financial conflict of interest.