|Home | About | Journals | Submit | Contact Us | Français|
IL-1β is a key proinflammatory cytokine with roles in multiple diseases. Monocytes package the IL-1β promoter into a “poised architecture” characterized by a histone-free transcription start site and constitutive transcription factor associations. Upon LPS stimulation, multiple proteins inducibly associate with the IL-1β gene. To understand how the complex combination of constitutive and inducible transcription factors activate the IL-1β gene from a poised structure, we measured temporal changes in NF-κB and IFN regulatory factor (IRF) association with IL-1β regulatory elements. Association of the p65 subunit of NF-κB peaks 30–60 min post-monocyte stimulation, and it shortly precedes IRF-4 recruitment to the IL-1β enhancer and maximal mRNA production. In contrast, IRF-8/enhancer association decreases poststimulation. To test the importance of delayed IRF-4/enhancer association, we introduced a mutated PU.1 protein shown to prevent PU.1-mediated IRF-4 recruitment to the enhancer sequence. Mutated PU.1 initially increased IL-1β mRNA followed by decreased mRNA levels 2–3 h poststimulation. Taken together, these data support a dynamic model of IL-1β transcriptional activation in which a combination of IRF-8 and p65 drives the initial phase of IL-1β transcription, while PU.1-mediated IRF-4 recruitment to the enhancer is important for the second phase. We further demonstrate that activation of both NF-κB and IRF-4 depends on CK2 kinase activity. Because IRF-4/enhancer association requires CK2 but not p65 activation, we conclude that CK2 triggers the IRF-4 and p65 pathways independently to serve as a master regulator of IL-1β transcription.
Interleukin-1β is a potent proinflammatory cytokine positioned at the apex of multiple pathological inflammatory cascades (reviewed in Ref. 1). Because IL-1β is a transcriptionally regulated gene, and transcript levels correlate with IL-1β protein levels in IL-1β-mediated disease (2, 3), understanding IL-1β transcription to artificially regulate protein levels is of high clinical interest. Human IL-1β transcription is regulated by two regions, a proximal promoter and an enhancer centered ~3 kb upstream from transcription start. Transient transfection studies on reporter constructs suggested that the promoter is as an on/off switch for basal transcription, but that inducible transcription is mediated through both the promoter and the enhancer (4–6). These early studies were important to define candidate cis elements and trans factors that regulate IL-1β mRNA production from the endogenous locus in monocytes/macrophages.
The following transcription factors identified by these studies activate the IL-1β promoter and enhancer: PU.1, the CCAAT-enhancer binding protein β (C/EBPβ),3 NF-κB, AP-1, STAT proteins, and IFN regulatory factors (IRFs) (4–12). More recent work analyzing IL-1β transcription in the context of chromatin has largely verified the importance of each of these factors in a more physiological context (12, 13). These studies showed the monocyte IL-1β promoter is packaged into a highly accessible chromatin structure that, in contrast to the other well-characterized cytokine promoters such as IL-12p40, IL-4, and IFN-β, does not change upon cellular stimulation (13–17). This “poised” chromatin structure probably characterizes many rapidly activated genes (18), although most cytokine genes must undergo remodeling of a blocking nucleosome for transcriptional initiation (19). The accessible chromatin structure of the IL-1β promoter is further characterized by constitutive association of PU.1 and C/EBPβ, but inducible association of RNA polymerase II (13).
Preliminary findings suggest the IL-1β enhancer also lacks regulation by changes in chromatin structure (13). PU.1 association with the enhancer, like that at the promoter, is constitutive, although whether the PU.1 partner C/EBPβ is constitutively or inducibly associated is debatable (12, 13). Recent evidence also shows IRF-8 and STAT-1 constitutively associate with the enhancer (12). In contrast, associations of IRF-4 and the kinase CK2 with the enhancer are inducible, and likely reflect CK2-mediated phosphorylation of enhancer-bound PU.1 at Ser148, a modification shown to be critical for IRF-4 recruitment to the enhancer sequence (13). Similarly, phosphorylation of enhancer-associated IRF-8 may contribute to IL-1β transcriptional activation, despite the demonstration that phosphorylation can decrease IRF-8/DNA association in some contexts (20). Whether additional activators of the promoter and enhancer identified in earlier studies constitutively or inducibly associate with the endogenous IL-1β gene remains unknown. Similarly, the roles of more general transcription factors such as TATA-binding protein (TBP) and structure-specific recognition protein 1 (SSRP1), a member of the transcript elongation complex FACT (21), are also unknown, although both of these factors may theoretically be recruited to the IL-1β gene through demonstrated protein-protein interactions with constitutively associated PU.1 (22, 23).
The dynamic nature of transcriptional regulation is appreciated for genes such as the estrogen responsive pS2 gene and Wnt targets such as c-myc and CycD1 (24, 25). Our understanding of the IL-1β promoter in the context of chromatin is thus far a snapshot, aimed at detailing multiple events occurring at a given point following stimulation. This approach has led to conflicting models of inducible IL-1β transcription (12, 13). We have taken advantage of the basic understanding of IL-1β transcriptional regulation described above and examined activation from a kinetic perspective toward detailing how the complex array of activators synergize and adjust over time to yield robust IL-1β transcription in monocytes. Our analyses revealed that mechanisms of inducible IL-1β transcription are divided into two separate phases. The first phase occurs independently of CK2-mediated PU.1 phosphorylation, but likely depends on the ability of CK2 to activate NF-κB. The second phase is characterized by a requirement for PU.1 phosphorylation by CK2 and a gain of IRF-4/enhancer binding. IRF-4 recruitment begins at about the time NF-κB binding is maximal, and it appears to usher in a second phase of more moderate transcriptional activation 2–3 h poststimulation. Overall, our data demonstrate the dynamic nature of protein associations triggered by a single kinase, CK2, in the context of a constitutively accessible promoter. These findings unite the various models to explain inducible activation of this tightly regulated gene, and they establish paradigms toward explaining the rapid activation of an array of early immediate genes in response to monocyte stimulation.
Mono-Mac-6 human monocytes (MM6) were maintained in RPMI 1640 medium containing 15% heat-inactivated FCS, 2 mM L-glutamine, and supplemented to a final concentration of 1 mM oxaloacetate, 0.45 mM pyruvate, and 0.2 U/ml insulin (OPI). Primary murine macrophages were differentiated from bone marrow as published (26). Human peripheral blood monocytes or B cells were isolated from fresh blood or leukopacks (New York Biologics) using negatively selecting magnetic beads (Miltenyi Biotec). Pro-B cells were selected from mouse bone marrow as the CD43+CD23−B220+ population by fluorescence-activated cell sorting. A murine pro-B cell line, 1–2, was analyzed in parallel with primary pro-B cells. Pre-B cells were similarly isolated as the CD43−CD23−B220+ population. This purification scheme isolated a sufficient number of cells from one mouse for reproducible biochemical analyses (~7000 pro-B cells per experimental condition). Cell populations were >99% pure as assayed by flow cytometry using additional lineage-specific markers. All B lineage cells or primary monocytes were cultured in RPMI 1640 medium or DMEM, respectively, containing 10% heat-inactivated FCS and 50 μM 2-ME. All culture media also contained 100 U of penicillin and 100 μg of streptomycin per milliliter.
Total RNA was isolated from 5 × 105 cells using RNEasy (Qiagen). RNA was further treated with DNase I for 15 min at 37°C, followed by heat inactivation at 65°C for ≥10 min. cDNA was prepared by standard methods. PCR reactions amplified equal volumes of cDNA in the presence of SYBR Green (Bio-Rad) in a Stratagene Mx3000p real-time PCR machine. Each amplification was performed in triplicate using the following conditions: hot start, 3 min at 95°C; melting, 15 s at 95°C; and then annealing and extension at 60°C for 1 min. The primers used for IL-1β transcript were either sense: 5′-ACGAATCTCCGACCACCACT-3′ and antisense: 5′-CCATGGCCACAACAACTGAC-3′ or sense: 5′-CCAAACCTCTTCGAGGC-3′ and antisense: 5′-AAGGAAGATTAAAGTCATACTTGAG-3′. Nascent IL-1β transcript primers were intron sequence: 5′-TGAACGTAGCCGTCATGGG-3′ and exon sequence: 5′-ACTGGCGAGCTCAGGTACTTCT-3′. PU.1 primers were 5′-AGATGCACGTCCTCGATACCC-3′ and 5′-TGGTGGCCAAGACTGGGAT-3′. PCR reactions conducted in parallel using β2-microglobulin primers (5′-CTCCGTGGCCTTAGCTGTG-3′ and 5′-TTTGGAGTACGCTGGATAGCCT-3′) were used to normalize for differences in cDNA synthesis and RNA input. mRNA copy number was calculated according to the equation: copies = 10(Ct−40)/−3.32. Fold induction was calculated as normalized copy number IL-1β in stimulated/unstimulated cells.
Accessibility of DNA to digestion with micrococcal nuclease (MNase) was analyzed using CHART-PCR as published (27) using 7000 to 106 cells. DNA (1–10 ng) as measured by PicoGreen was analyzed by SYBR Green incorporation during quantitative PCR. PCR products from each primer set were run on acrylamide gels to verify product size. All products displayed a single optimum on melt curve analysis of each amplification. Apparent negative accessibility is a mathematical artifact likely due to error in DNA quantitation and is neither reproducible nor biologically significant, and hence was set to 0% accessible. Different sources of MNase were used (see Fig. 1 vs Fig. 2), explaining the difference in range of units used for assays. Each enzyme lot was titrated to achieve partial to near maximal DNA cleavage in each assay.
ChIPs were performed as previously published (28, 29), using 500,000 human monocytes or 1.5 × 106 murine bone marrow-derived macrophages per Ab. In some experiments, precipitated DNA is shown as a percentage of input DNA. Alternatively, fold enrichment was calculated by 2(Ct input – Ct ChIP) after quantitative amplification of equivalent picogram amounts of DNA (30). Relative enrichment was calculated as fold enrichment using Ab of interest/fold enrichment using a control Ab. The appropriate amount of each Ab was determined empirically (1–10 μg). ChIP-competent Abs were: rabbit anti-histidine tag or anti-GST Ab (Santa Cruz Biotechnology sc-803 or sc-459, respectively) as a nonspecific isotype/species matched control, anti-TBP (sc-273, ×10 μg), anti-c-Jun (sc-45, ×10 μg), anti-SSRP (sc-25382, 10 μg), anti-p65 subunit of NF-κB (sc-372, 2 μg), anti-IFN consensus sequence binding protein/IRF-8 (sc-13043x), and anti-IRF-4 (whole rabbit serum provided by Dr. Michael Atchison, University of Pennsylvania.). Goat anti-c-fos was obtained from Santa Cruz Biotechnology (sc-52x); this Ab was precipitated with protein G-coupled beads. Oligonucleotides used for amplifying the precipitated IL-1β promoter region were sense: 5′-TGCACTGGATGCTGAGAGAAA-3′ and antisense: 5′-GGCTGCTTCAGACACCTGTG-3′. IL-1β enhancer-specific primers were promoter proximal (EIV): 5′-AGCGGTCTCCTTGGGAAGA-3′ and 5′-ACGATCCCATCCATCTCAGG-3′; or promoter distal (EI): 5′-TGACCCTGACAGGGTAAAGAGG-3′ and 5′-CCAGGATGCTCCAGCTTTTG-3′.
MM6 cells were pretreated with pharmacological inhibitors of CK2 for 3 h as previously detailed (13). Efficacy of CK2 inhibition by apigenin was assayed as published (31) except γ-32P-ATP was used as the phosphate donor. The CK2 source for these assays was cytoplasmic extracts from MM6 cells stimulated with LPS for 2 h. Background counts in the absence of CK2-specific target peptide were subtracted from counts in the presence of peptide. Alternatively, SB203580 or a TNFR-associated factor 6 (TRAF6)-blocking peptide (Imgenex IMG-2002; DRQIKIWFQNRRMKWKKRKIPTEDEY) was added 10 min before further treatment. Escherichia coli LPS (Sigma-Aldrich; 50–100 ng/ml) was then added to stimulate IL-1β transcription. Transcripts were measured by quantitative RT-PCR and SYBR Green incorporation as detailed above. Primers approximated 97% efficiency. Values graphed are the ratio of normalized IL-1β copy number in stimulated cells to normalized IL-1β copy number in resting cells for a fold induction. All PCR reactions were run in triplicate to define average Ct values for each sample.
MM6 cells were split to 0.5 × 105/ml and after 3–4 days reached 3–4 × 105/ml, a confluency that was optimal for nucleofection. DNA was introduced using the Amaxa system for human monocytes (Nucleofector kit V). All constructs of interest were conucleofected with a GFP-expressing construct for visualization and/or sorting of transfected cell populations. GFP+ cells were isolated 48 h postnucleofection using a MoFlo cell sorter, then rested for 2 h before stimulation with LPS and mRNA quantitation as outlined above.
Nuclear or whole-cell extracts were separated on 10% polyacrylamide gels. Proteins were transferred to polyvinylidene difluoride membranes and probed using Abs outlined above, and standard protocols. Calnexin was the loading control for all Western blots.
We previously demonstrated that monocyte cell lines package the IL-1β promoter into an MNase-accessible, transcription factor-associated “poised” structure before transcriptional activation (13). Toward understanding the biological significance of this structure, we questioned whether primary human monocytes have a nucleo-some-depleted promoter. We isolated monocytes from pooled human peripheral blood and measured IL-1β promoter accessibility to MNase by CHART-PCR (13, 27). The IL-1β promoter is highly accessible (>50% of alleles cleaved by MNase) at ~100 bp upstream of the transcription start site in primary unstimulated monocytes (Fig. 1, A and B, left bars). In contrast, DNA upstream of the core promoter is inaccessible to MNase, as indicated by amplifying the same DNA samples with primers centered at ~400 bp upstream of the transcription start site (Fig. 1A, right bars). These findings support the interpretation that the IL-1β promoter has a nucleosome-free region at the transcription start site flanked by nucleosome-packaged (i.e., MNase-inaccessible) DNA. We conclude that the previously analyzed MM6 cell line accurately recapitulates the IL-1β promoter structure characteristic of primary human monocytes (13).
We next questioned the specificity of the constitutively accessible monocyte IL-1β promoter by measuring promoter structure in a closely related hematopoietic cell type, B cells. In contrast to monocytes, B cells isolated from human peripheral blood or tonsil package the transcription start site into a relatively inaccessible structure (Fig. 1B, ~15–20% accessible to MNase). This finding in part explains the demonstration that B cells do not produce significant amounts of IL-1β mRNA or protein (Fig. 1C and Ref. 1). Finally, murine B lineage precursors, pro- and pre-B cells, package the IL-1β transcription start site into a relatively inaccessible chromatin structure (Fig. 1D). Because the common hematopoietic stem cell that gives rise to both B cells and monocytes produces IL-1β (32, 33), the inaccessible promoter structure in early B lineage cells indicates that either monocytes engage active processes to maintain locus accessibility during development, or B cells actively close the locus before the pro-B cell stage. The moderate accessibility (20–40%) upstream of the transcription start site in pro- and pre-B cells, as quantified by the – V amplicon (Fig. 1E) contrasted with the totally inaccessible (i.e., nucleosome-protected) structure of this region in monocytes (Fig. 1A). These findings support the interpretation that nucleosomes flanking the transcription start site are precisely positioned in monocytes but not in B cells. Overall, these data show that the constitutively accessible IL-1β promoter is limited to cells capable of inducible IL-1β production, and that it is associated with nucleosome-protected DNA upstream of the nucleosome-free transcription start site.
The IL-1β promoter is not transcribed in unstimulated monocytes, despite the demonstrated lack of a repressive chromatin structure (Fig. 1). We previously showed constitutive association of two proteins, PU.1 and C/EBPβ, to the promoter and enhancer regions in unstimulated monocytes. This finding was complemented by the demonstration that IRF-4 inducibly associated with the enhancer, likely due to recruitment by CK2-phosphorylated PU.1 (11, 13, 34). Other studies strongly indicated that inducible association of NF-κB proteins with IL-1β regulatory regions was critical for inducible transcription (35–37). We therefore questioned how the PU.1/IRF and NF-κB pathways converge to achieve IL-1β transcription. The first step toward answering this question was to more completely identify constitutively vs inducibly binding transcription factors by ChIP. We focused analyses on c-Jun and c-fos, which have been shown to be important for inducible IL-1β transcription in overexpression and in vitro binding assays (38–40). We also measured association of two ubiquitous factors, TBP and the transcription elongation factor SSRP1, based on previous demonstrations that these proteins form physical interactions with PU.1 (22).
ChIP assays showed that TBP associated with the IL-1β promoter in both unstimulated and stimulated monocytes (Fig. 2A). Note that we conservatively designate a 2-fold average increase in fold increase over the control anti-histidine tag Ab as “association”. Constitutive association of TBP with the IL-1β promoter contrasts with the inducible association of RNA polymerase II we detected upon cellular stimulation (13) and it demonstrates that TBP associates with the promoter before mRNA production. TBP is therefore a component of the “poised architecture” of the IL-1β promoter characterized by an accessible promoter structure and constitutive transcription factor association in the absence of significant transcription. Control assays showed TBP did not associate with the IL-1β enhancer (Fig. 2B). In contrast, ChIP assays using an Ab recognizing c-Jun showed inducible association to the promoter (Fig. 2A).
The demonstrations that the transcription factor PU.1 constitutively associates with the IL-1β promoter (13) and that PU.1 can form protein-protein interactions with SSRP1 (22), a member of the transcriptional elongation complex FACT (21), raise the possibility that IL-1β transcription requires this mRNA elongation factor. To test this possibility, we measured association between SSRP1 and the IL-1β gene by ChIP. SSRP1 associated with the IL-1β promoter weakly in unstimulated monocytes, but association increased ~2.5-fold in stimulated monocytes (Fig. 2A; 1.9- vs 4.8-fold control Ab on average in unstimulated and stimulated monocytes, respectively). This finding indicates that SSRP1 plays a role in inducible IL-1β transcription by promoting mRNA elongation only poststimulation. Therefore, IL-1β elongation probably occurs through relatively standard processes. As expected, SSRP1 did not associate with the IL-1β enhancer in unstimulated or stimulated monocytes (Fig. 2B). Overall, ChIP assays demonstrate constitutively associated PU.1, C/EBPβ, and TBP form the core of the poised promoter structure that synergizes with inducibly associated c-Jun, p65, SSRP1, and pol II to achieve rapid IL-1β transcription (see Figs. 2 and and77 and Refs. 12, 13, 35, 41).
To further understand the complexities of IL-1β transcriptional activation, we measured association of transcription factors to the IL-1β enhancer by ChIP. c-Jun inducibly associated with the promoter-distal region (EI) of the IL-1β enhancer (Fig. 2B, right bars; 1.64- vs 4.03-fold control antibody). The resolution of the ChIP assay did not allow us to distinguish between inducible association with AP-1 sites at −3114 vs −3491, although the lack of functional significance of the −3114 element (38) suggests inducible binding may be favored at the functional −3491 enhancer element (39). Interestingly, analysis of the same ChIP products with promoter-proximal enhancer-specific primers (EIV) demonstrated constitutive association of c-Jun to this region (Fig. 2B, left bars; 2.44- vs 3.29-fold control antibody), most likely at the −2782 AP-1 element (42). Pairing constitutive and inducible association of c-Jun to nearby elements of the IL-1β enhancer illustrates the degree of specificity driving the tight regulation of this potent proinflammatory cytokine. However, analyses using a ChIP-competent anti-c-fos Ab failed to detect c-fos association with any region of the IL-1β gene (Fig. 2B and data not shown), consistent with the demonstration that c-Jun homodimers activate the IL-1β promoter (41).
Like c-Jun, IRF-4 inducibly associated with the IL-1β enhancer, as assayed by primers specific for the promoter-proximal enhancer region (Fig. 2C; 1.66- vs 2.88-fold control Ab in unstimulated vs stimulated cells; p = 0.046 by Student's t test). We confirmed association between the IL-1β enhancer and IRF-8 in unstimulated cells, which modestly decreases 1 h poststimulation (Ref. 12 and Fig. 2C; 5.02-vs 3.17-fold control antibody). Overall, ChIP analyses of the IL-1β enhancer suggest that constitutive PU.1, C/EBPβ, and c-Jun associations are augmented by inducible c-Jun and IRF-4 associations to culminate in IL-1β transcription from an accessible promoter structure. Identifying these associations was the first step in understanding how inputs from multiple proteins converge to choreograph rapid IL-1β activation in monocytes.
CK2 activity plays a critical role in inducible IL-1β transcription, as evidenced by the ability of CK2 inhibitors to block mRNA production in response to LPS (Fig. 3A, black bar, and Ref. 13). Importantly, CK2 inhibitors do not affect the constitutively accessible chromatin structure of the IL-1β promoter in monocytes, suggesting that CK2 functions downstream of this first step in gene activation (Fig. 3, B and C). In contrast, the MAPK inhibitor PD98059 has no effect on inducible IL-1β transcription (Fig. 3A and Ref. 13). The quantity of mRNA produced by stimulated monocytes in the presence of the CK2 inhibitors apigenin or emodin (data not shown) approximates mRNA levels in stimulated cells treated with either of two NF-κB inhibitors: a TRAF6 blocking peptide or the proteosome inhibitor lactacystin (Fig. 3A). Because CK2 is a constitutively active kinase that inducibly associates with the IL-1β enhancer (13), one method that monocytes may use to increase CK2 activity would be to increase CK2 production to promote CK2/enhancer association. We therefore quantified CK2 mRNA in monocytes responding to LPS (Fig. 3D). Although IL-1β mRNA levels peak 1–2 h following LPS stimulation (Fig. 3D, right bars), CK2 mRNA levels remain unchanged in response to LPS stimulation (Fig. 3D, left bars), consistent with the interpretation that CK2 activity increases in LPS-stimulated monocytes independent of CK2 production.
To define specific steps in IL-1β activation that are triggered by CK2 activity, we measured the effect of CK2 inhibition on the two pathways that activate IL-1β transcription: PU.1/IRF-4 and NF-κB/p65. We first questioned the effect of CK2 inhibition on IRF-4 recruitment to the IL-1β enhancer. IRF-4 recruitment to the IL-1β enhancer sequence requires LPS-inducible PU.1 phosphorylation by CK2 (13, 34), suggesting that inhibiting CK2 may block IRF-4/enhancer association. To test this possibility, we treated monocytes with the CK2 inhibitor apigenin, then stimulated the cells and measured IRF-4/enhancer association by ChIP. Apigenin potently inhibited IRF-4 recruitment to the IL-1β enhancer (Fig. 3E, right bars; compare white stippled and dark gray bars). This finding identifies one mechanistic explanation for the dramatic decrease in IL-1β transcript levels in the presence of apigenin (Fig. 3A): the IRF-4 nucleoprotein complex at the enhancer fails to assemble in the absence of CK2 activity. Interestingly, we measured only a partial block in IRF-4/IL-1β enhancer association in the presence of the p38 inhibitor SB203580 (Fig. 3E, 51% decrease, right light gray bar), consistent with the partial (58%) decrease in IL-1β mRNA in the presence of SB203580 (13).
Because of the possible off-target effects of pharmacological inhibitors, it was important to verify that apigenin blocked CK2 activity in monocytes as expected. We therefore added apigenin to extracts from LPS-stimulated monocytes and tested the ability of the extracts to transfer phosphate to a CK2 consensus peptide. Fig. 3F demonstrates that apigenin completely blocked CK2-dependent phosphate transfer in monocyte extracts. p38 inhibition in response to SB203580 and TRAF6 inhibition by the blocking peptide have been published (43, 44), suggesting that all three inhibitors block activity of their intended targets. Taken together, our results strongly support the interpretation that CK2 activity is required to form an active enhancer complex by promoting inducible IRF-4/ enhancer association.
Although our data support a role for CK2 in formation of an IRF-4-dominated IL-1β enhancer complex, published work suggested that CK2 may also function to regulate a second IL-1β activator, NF-κB (35). To test whether CK2 plays a role in NF-κB-mediated IL-1β activation, we treated monocytes with the CK2 inhibitor apigenin, then stimulated cells for 0 – 60 min before quantifying IκB-α, one of a family of proteins that sequesters p65 in the cytoplasm in unstimulated monocytes. IκB-α levels decreased 15–30 min poststimulation, rebounding to unstimulated levels 60 min post-LPS (Fig. 4A, lanes 1–4). This pattern of expression was due to proteosome-mediated IκB-α degradation 15–30 min post-stimulation, which was blocked by the proteosome inhibitor MG132 (Fig. 4A; compare lanes 1–4 with lanes 9–12). Increased IκB-α levels 60 min poststimulation are due to NF-κB activation, which drives IκB-α transcription (45). In contrast, IκB-α levels in apigenin-treated cells decreased steadily over time in response to LPS (Fig. 4A, lanes 5–8>), indicating IκB-α degradation proceeded as expected, but new IκB-α production was not initiated in response to NF-κB activity. This interpretation is consistent with the demonstration that apigenin blocks NF-κB activation in breast cancer cells (46). Pretreatment of cells with MG132 delayed IκB-α degradation in response to LPS as expected (Fig. 4A, compare lanes 5–8 with lanes 13–16). Control experiments showed apigenin treatment did not affect expression levels of IRF-4, IRF-8, or PU.1 in unstimulated or stimulated monocytes (see below; data not shown). Overall, these data suggest that CK2 regulates IL-1β transcription through effects on NF-κB activation.
To further investigate the role CK2 plays in NF-κB activation, we measured the effect of CK2 inhibition on a hallmark of p65 activation, phosphorylation at Ser276 (47). p65 is phosphorylated in response to monocyte activation (Fig. 4B, lanes 1–6). However, monocytes pretreated with the CK2 inhibitor apigenin fail to phosphorylate p65 at Ser276 (Fig. 4B, lanes 7–9). This finding extends results showing apigenin decreases LPS-stimulated p65 phosphor-ylation at Ser536 in primary human monocytes (48). Because p65 Ser276 is a target of protein kinase A in LPS-stimulated cells (47), these data support an indirect requirement for CK2 in p65 phosphorylation at Ser276. Overall, these data demonstrate CK2 plays a role in at least two aspects of NF-κB activation: IκB-α degradation and p65 phosphorylation. We conclude that CK2 function is critical for posttranslationally activating two transcription factors that synergize to induce IL-1β transcription: IRF-4 and NF-κB.
Models explaining inducible IL-1β transcription are largely based on a snapshot of protein associations at a single time point following transcriptional activation. However, transcriptional activation complexes are often highly dynamic, as measured by both changes in chromatin structure and protein associations (24, 25, 49–51). Toward understanding the dynamic nature of the IL-1β activating complex, and to further test the functional significance of inducible IRF-4 recruitment to the enhancer, we took advantage of published findings that mutation of PU.1 at Ser148, a major CK2 target, prevents IRF-4 recruitment to the enhancer in vitro and decreases IL-1β promoter activity (34, 52). We expressed PU.1-mutated serine to alanine at position 148 (mPU.1) in monocytes, then sort-purified monocytes expressing mPU.1 based on expression of cotransfected GFP. The cells were then stimulated with LPS and collected at intervals for quantitative analysis of IL-1β mRNA transcript. Surprisingly, cells that expressed mPU.1 produced higher levels of IL-1β mRNA 30 min following stimulation as compared with control cells transfected with GFP alone (Fig. 5A, black line). In contrast, cells expressing mPU.1 contained markedly reduced levels of IL-1β mRNA 2–3 h poststimulation as compared with controls. Analysis of nascent IL-1β transcripts (data not shown) confirmed that the pattern of IL-1β mRNA was due to changes in transcription vs the steady-state transcript levels shown. Furthermore, total PU.1 mRNA levels were similar in mPU.1 vs control-transfected cells (Fig. 5B). This biphasic pattern of IL-1β mRNA production revealed by mPU.1 overexpression strongly suggested that the IL-1β-activating complex changed over time. The proposed change in transcriptional-activating complex was not determined by overall transcription factor levels, as demonstrated by quantification of PU.1, IRF-4, and IRF-8 on Western blots in unstimulated vs stimulated cells (Fig. 6A and data not shown) (52). In contrast to untransfected cells that have maximal IL-1β mRNA levels 1–2 h poststimulation (Figs. 1C and and3D),3D), transfected cells increase IL-1β transcript from 2 to 3 h poststimulation (Fig. 5A). The act of transfection itself can activate IFN pathways, and a combination of IFN + LPS stimulation prolongs p65/IL-1β association without significantly altering other mechanisms of transcriptional activation (data not shown); therefore, the delay in maximal IL-1β mRNA in transfected cells was predictable. Overall, these data show that in the first phase of IL-1β transcription, 30 min poststimulation, PU.1 phosphorylation was dispensable. However, because mPU.1 that cannot be phosphorylated by CK2 (13) inhibits transcription 2–3 h poststimulation, we conclude that CK2-mediated PU.1 phosphorylation plays a role in sustaining transcription.
We detected a gain of IRF-4, but a modest decrease in IRF-8 association, with the IL-1β enhancer 1 h post-monocyte stimulation (Fig. 2C and Ref. 13), while recent work showed IRF-8, not IRF-4, associated with the enhancer 30 min poststimulation (12). The dynamic nucleoprotein complex uncovered by mPU.1 overexpression may explain this apparent disagreement over the identity of the IRF family member that activates IL-1β, if in fact phospho-PU.1-dependent IRF-4/enhancer association occurs only at later time points in IL-1β activation. To test this possibility, we used ChIP to quantify IRF-4 and IRF-8 recruitment to the IL-1β enhancer at multiple time points poststimulation. Fig. 6B shows that IRF-4 does not associate with the IL-1β enhancer in monocytes 0–30 min following LPS treatment (Fig. 6B, middle black or gray bars, respectively). However, IRF-4 associates robustly both 1 and 2 h poststimulation (Fig. 6B, middle white and white stippled bars, respectively). In contrast, IRF-8 constitutively associates with the IL-1β enhancer in unstimulated cells (Fig. 6B, rightmost black bar) and in monocytes stimulated with LPS for 30–60 min (Fig. 6B, rightmost gray and white bars, respectively). The transient decrease in IRF-8 association 30 min poststimulation was highly variable and was not recapitulated in published analyses (12). In contrast to IRF-4, IRF-8 probably dissociates from the IL-1β enhancer in monocytes stimulated for ≥2 h (Fig. 6B, rightmost stippled bars).
Pharmacologically inhibiting NF-κB activation completely blocked inducible IL-1β transcription (Fig. 3A), consistent with the interpretation that CK2-activated PU.1/IRF and NF-κB pathways converge to activate IL-1β transcription in monocytes. To determine whether NF-κB activation reinforces the biphasic mechanism of inducible IL-1β transcription revealed by mPU.1 and dynamic IRF-4 recruitment, we measured p65 association with the IL-1β promoter and enhancer by ChIP 0–4 h post-LPS stimulation. p65 maximally associated with the IL-1β promoter at approximately the same time as IRF-4 recruitment began at the enhancer: 1 h poststimulation (Fig. 7A). In contrast, p65 association with the enhancer occurs more rapidly and is maximal at ~30 min poststimulation (Fig. 7B), at about the same time as robust IRF-8/enhancer association and immeasurable IRF-4/enhancer association (Fig. 6B). This finding is reminiscent of maximal p65 association with the IL-10 promoter 30 min postinduction (53). p65/enhancer association therefore precedes IRF-4 recruitment to the enhancer, indicating that p65 functions in concert with constitutively associated PU.1 and IRF-8 during the first (phospho-PU.1-independent) phase of IL-1β activation. Because CK2 regulates both IRF-4/enhancer recruitment (Fig. 3E) and NF-κB activation (Fig. 4, A and B), our data highlight CK2 as a key modulator of these two major pathways that function in combination to activate IL-1β transcription.
To test the possibility that NF-κB activation plays a direct role in IRF-4 recruitment, we blocked LPS-mediated NF-κB activation using the TRAF6-blocking peptide, then measured the effect of TRAF6 neutralization on IRF protein/IL-1β enhancer association. The TRAF6-blocking peptide had little effect on associations between the enhancer and either IRF-4 (Fig. 7C, left panel) or IRF-8 (Fig. 7C, right panel), indicating that NF-κB activity was not required for changes in IRF-4/enhancer association. These data strongly suggest that CK2 activates two pathways, NF-κB and PU.1/IRF-4, independently, and that these CK2 targets converge only at the level of protein/enhancer binding to activate and maintain inducible IL-1β transcription.
IL-1β is activated from an accessible chromatin structure by a complex array of DNA-binding proteins. A subset of these proteins constitutively associates with IL-1β regulatory elements before transcription initiation, while a second subset associates only upon monocyte stimulation. Our data indicate that one kinase, CK2, activates two processes that function in parallel to drive IL-1β transcription from the poised promoter only in stimulated cells. First, CK2 regulates p65 translocation and activity by controlling IκB-α levels and p65 phosphorylation, respectively. Second, CK2 regulates IRF-4 recruitment by modulating phosphorylation of PU.1 at Ser148. Kinetic analyses further indicate that the timing of inducible p65/IL-1β enhancer association likely primes initial IL-1β transcription in combination with constitutive IRF-8 association, while PU.1 phosphorylation and the resulting IRF-4 recruitment likely contribute to prolonging IL-1β transcription beyond early time points. This association temporally bridges the two proposed phases of IL-1β transcriptional activation. Taken together, these findings demonstrate how two pathways resulting in NF-κB and IRF association are both triggered by CK2 activity to culminate in transcriptional activation of a potent proinflammatory cytokine, IL-1β.
The roles of IRF-4 vs IRF-8 in IL-1β transcription have been highly controversial. Transient transfection experiments have shown 1) either IRF-4 or IRF-8 can activate IL-1β transcription (11), or 2) IRF-4 plays a more important role in IL-1β activation specifically in macrophages (4). Similarly, ChIP experiments differed on whether IRF-4 or IRF-8 association correlates with inducible IL-1β transcription (12, 13). Finally, experiments in mice genetically null for IRF-4 or IRF-8 demonstrated that proinflammatory cytokine production increased or decreased, respectively, indicating that IRF-8 activates IL-1β, but that IRF-4 actually inhibits cytokine production (54, 55). All of these findings are unified in the following model based on the demonstrated shift from a p65/PU.1/IRF-8 to a phospho-PU.1/IRF-4-dominated complex at the IL-1β enhancer. IRF-8 associates with the IL-1β enhancer in unstimulated monocytes, and this association is maintained for 30–60 min poststimulation (Fig. 6). We speculate that constitutive IRF-8 and PU.1 association along with rapid recruitment of NF-κB to the enhancer (Fig. 7B) contribute to the nearly instantaneous recruitment of pol II upon LPS stimulation and drive the extraordinarily rapid activation of the IL-1β gene. We propose that the demonstrated tyrosine phosphorylation of IRF-8 at 30 min poststimulation (12) opens a window for PU.1-mediated IRF-4 recruitment to the enhancer and decreased p65 and IRF-8/enhancer association (Figs. 6 and and7)7) that correlates with maximal IL-1β mRNA levels at 60–120 min poststimulation (Figs. 1C and and3D).3D). The resolution of ChIP does not allow us to distinguish between IRF-4 and IRF-8 binding adjacent to PU.1 in the enhancer vs, for example, IRF-4 being tethered to the enhancer complex. Because IRF-4 can negatively regulate cytokine production (55), we propose that the shift from a p65/PU.1/IRF-8 to a phospho-PU.1/IRF-4-dominated complex results in a controlled moderation of IL-1β transcription, as evidenced by decreased transcript levels at ~3h poststimulation (Fig. 3D). The proposed mechanism of IL-1β activation by an initial IRF-8 dominance followed by a shift to IRF-4 dominance contrasts with a recently published model of sustained IFN-β activation, which is promoted by IRF-8/enhancer association (56).
This detailed model of IL-1β transcriptional activation explains multiple apparent conflicts in the literature describing the roles of various sequence-specific transcription factors. However, this model does not take into account the demonstration that both IRF-4 and IRF-8 association with composite PU.1/IRF elements in other genes requires PU.1 phosphorylation at Ser148 (or Ser147 in the rat PU.1 protein) (57–59). However, in vitro transcribed/translated PU.1 and PU.1 from unstimulated macrophage nuclear extracts facilitates IRF-8/IL-1β enhancer binding (4, 12), indicating PU.1 phosphorylation may be dispensable for early IRF-8-mediated activation. This possibility is supported by the demonstration that expression of the nonmodifiable mPU.1, in combination with LPS stimulation (i.e., NF-κB activation), resulted in IL-1β hyper-production only at the earliest time point tested.
Synergy between NF-κB proteins and LPS-induced PU.1 and/or IRFs plays a role in activation of many immune system genes, including IL-1 receptor antagonist, IL-12p40, and RANTES (60–63). Similarly, feedback loops between NF-κB and IRF induction have been identified in several experimental systems. For example, NF-κB family members regulate IRF-4 levels in lymphocytes (64, 65), and IRF-8 is required for activity of the IκB kinase that initiates the cascade of events leading to NF-κB nuclear translocation in dendritic cells (66). However, the almost instantaneous activation of IL-1β transcript occurs before the times these feedback loops have been analyzed, so the contribution of direct NF-κB/PU.1/IRF cross-talk to inducible IL-1β transcription is uncertain. It is possible that serial activation of NF-κB and PU.1 phosphorylation/IRF-4 recruitment by CK2 does not require additional interaction in the earlier phases, but that feedback loops become operational at about the same time IL-1β transcription is moderated 2–3 h poststimulation. Regardless, our data support the model that CK2 is a unifying regulator of inducible IL-1β transcription, based on its ability to regulate NF-κB and IRF-4 activation.
We thank Tom Rothstein for the lactacystin, David Seldin and Esther Landesman-Bollag for MG132 and for assistance with the radioactive CK2 kinase assays, Michael Atchison for the IRF-4 specific Ab, and Ron Corley and Ann Marshak-Rothstein for Western blot Abs. Sylvia Sardi provided the murine pro- and pre-B cells. MoFlo cell sorts were done by the Boston University School of Medicine Flow Core Facility. Gerald Denis and Gerd Blobel provided valuable critiques of the manuscript.
This work was funded by R01 AI54611 and an American Diabetes Association Research Grant to B.S.N.
3Abbreviations used in this paper: C/EBPβ, CCAAT-enhancer binding protein β; CHART-PCR, chromatin accessibility by real-time PCR; ChIP, chromatin immunoprecipitation; IRF, IFN regulatory factor; MM6, Mono-Mac-6 human monocytes; MNase, micrococcal nuclease; mPU.1, PU.1 mutated serine to alanine at position 148; SSRP, structure-specific recognition protein; TBP, TATA-binding protein; TRAF6, TNFR-associated factor 6.
Disclosures The authors have no financial conflicts of interest.