Monocyte polarization by IFN-γ or IL-4 drives a complex series of cellular responses leading to increased intracellular killing (IFN-γ) or enhanced healing (IL-4) among other functional responses. We studied the effect of IL-4 and IFN-γ polarization on histone modifications at the TNF-α locus in human primary monocytes. IFN-γ polarization markedly increased the expression of TNF-α, whereas IL-4 treatment decreased the expression. We found that IFN-γ alone increased histone H4 acetylation at the TNF-α promoter. The effect of IFN-γ on TNF-α expression was durable upon cytokine washout and even repolarization with IL-4. Concordantly, IFN-γ-mediated H4 acetylation was also durable. IFN-γ recruited activating transcription factor-2 via p38 to the TNF-α promoter, but inhibition of p38 had minimal effect on H4 acetylation. In a novel finding, we found that IFN-γ recruited RNA Pol II to the human TNF-α promoter via ERK signaling, but did so without initiating transcription, leading to a poised condition. These studies provide an important perspective on monocyte polarization. Polarization by IFN-γ has a durable effect on TNF-α expression, and histone acetylation may provide a mechanism for persistence of the effect.
Cytokines influence every aspect of the immune response from the innate to the adaptive. For monocytes and monocyte-derived dendritic cells and macrophages, cytokines influence the expression of chemokines, the expression of costimulatory molecules, and the execution of effector programs. Polarization with IL-4 and IFN-γ are two well-studied systems in both the mouse and the human (1, 2). IL-4 polarization, referred to as either alternative or M2a activation, promotes a response characteristic of wound healing and parasite immunity, whereas IFN-γ polarization, known as classical or M1 activation, programs monocytes for intracellular killing, tumor resistance, and IL-12 production (3).
Monocytes are the major source of TNF-α, and cytokine polarization has been shown in multiple models to affect the level of TNF-α production (4, 5, 6). Polarization of monocytes with IFN-γ increases production of TNF-α, whereas polarization with IL-4 decreases the amount of TNF-α produced (3). The mechanisms underlying polarization effects on TNF-α expression include altered expression of surface receptors and altered expression of signaling molecules; however, epigenetic regulation has not been previously examined (2, 7).
The transcriptional regulation of TNF-α is complex. TNF-α production has been shown to be differentially regulated depending on cell type and stimulus (8, 9, 10, 11, 12, 13, 14, 15, 16, 17). Transient transfection studies with reporter plasmids have identified the −200-1 region of the TNF-α promoter, containing an enhanceosome composed of Ets, activating transcription factor (ATF)3-2/c-Jun, NF-κB, CREB-binding protein (CBP)/p300, and Sp1, as being critical for TNF-α transcription (12, 13, 18). Our laboratory has previously demonstrated the importance of DNA methylation, nuclear localization, and histone modifications in the transcriptional control of TNF-α expression, and this current study was designed to evaluate the role of histone modifications in the regulation of TNF-α expression after polarization (19, 20, 21).
Most promoters demonstrate a chromatin signature that determines transcriptional competence, with active promoters exhibiting high levels of H3 and H4 acetylation as well as H3 lysine 4 trimethylation (22, 23). On many promoters, RNA polymerase II (RNA pol II) is preloaded onto the transcription start site, and short transcripts of uncertain function may be produced (23, 24). The TNF-α gene exhibits an unusual pattern with H3 lysine 9 methylation being erased coincident with RNA pol II recruitment after stimulation (25). The TNF-α gene also undergoes transient H4 acetylation in response to some stimuli (20). Endotoxin tolerance is mediated by a diminished ability to remodel chromatin, suggesting that chromatin is an important regulatory mechanism for TNF-α (26, 27).
In this study of polarization effects, IFN-γ was found to drive the TNF-α promoter into a poised condition. IFN-γ treatment led to increased H4 acetylation, and both ATF-2 and RNA Pol II were recruited to the promoter. A run-on assay confirmed that IFN-γ led to an increase in the rate of transcription. These novel findings contribute to the understanding of polarization and provide a framework for understanding the persistence of the effect of IFN-γ.
Materials and Methods
Reagents and cells
LPS (Escherichia coli 011:B4) was obtained from Sigma-Aldrich and used at a concentration of 1 μg/ml in all experiments. Carrier-free human rIFN-γ and rIL-4 were obtained from eBioscience and were both used at a concentration of 50 ng/ml (330 IU IFN-γ and 500 IU IL-4). PD98059 and SB203580 were obtained from Sigma-Aldrich and used at a concentration of 10 μM, a concentration that led to maximum effect and minimal cell death. The following Abs were used for chromatin immunoprecipitation: acetylated histone 3, acetylated histone 4, pan-histone 3, RNA Pol II clone 8WG16 (Upstate Biotechnology/Millipore), ATF-2 N-96 (Santa Cruz Biotechnology), and anti-rabbit GST (Zymed Laboratories).
Human peripheral blood monocytes, isolated by countercurrent elutriation, were provided by the Human Immunology Core at the University of Pennsylvania and were confirmed to be >85% CD14+ by flow cytometry (28). Monocytes were cultured in RPMI 1640 supplemented with 10% cosmic calf serum, 2 mM glutamine, 200 U/ml penicillin, and 200 μg/ml streptomycin. Monocytes were rested overnight at 4°C on a nutator before being transferred to culture flasks. Cells were warmed in the incubator for at least 2 h before experiments. The only assay performed with a cell line was the run-on assay, in which THP-1 cells were used and were grown in RPMI 1640 with 10% cosmic calf serum.
TNF-α was detected with BD Biosciences’ BD OptEIA Human TNF ELISA set, according to manufacturer’s protocol. The IL-12p70 ELISA was performed with purified rat anti-human IL-12p70 capture mAb (catalogue 555065), biotinylated mouse anti-human IL-12 p40/p70 detection mAbs (catalogue 554660), avidin-HRP, and human rIL-12p70 standards from BD Biosciences. The ELISA was performed according to the manufacturer’s instructions with optimization of Ab concentrations for IL-12, as follows: 1 μg/ml capture Ab, 2 μg/ml detection Ab, and 1/1000 dilution of avidin-HRP. The plates were read at 450 nm.
RNA was isolated from 1 million primary human monocytes using Qiagen RNeasy kit and on-column DNase digestion using Qiagen’s RNase free DNase set. A quantity amounting to 0.2 μg of total RNA was used for reverse-transcription reactions using BD Bioscience’s Advantage RT-for-PCR kit. The cDNA was used for quantitative real-time Taqman PCR using Applied Biosystems’ TNF-α human mRNA primers and the 18S internal control. Calculations performed normalized to the 18S internal control signal and are expressed relative to nonpolarized unstimulated monocytes or as a simple normalized signal.
Five million monocytes were polarized for 18 h, as described, and stimulated with LPS for the indicated times. Cells were lysed immediately with cold radioimmunoprecipitation assay buffer. Protein lysates were quantitated, and equal amounts were loaded on a NuPAGE gel (Invitrogen Life Technologies). Blots were probed with Abs for tubulin, histone deacetylase (HDAC) 3, and phospho-ATF-2 (Santa Cruz Biotechnology), according to manufacturer’s specifications.
Chromatin immunoprecipitation (ChIP)
ChIP assays were performed, as described previously, according to a modified protocol based on Upstate Biotechnology (20, 21). Immunoprecipitated DNA was quantitated using a Taqman SDS 7900HT using previously published primer/probe combinations to specific regions of the TNF promoter, TNF1–4, and the intronic enhancer (20, 21). Cycle threshold (Ct) values were normalized to 10% input of each sample according to the formula 2^(10% input Ct − sample Ct). Standard curves using genomic thymic DNA were used as a positive control. Rabbit GST was used as nonspecific Ab control in all experiments, but is only shown in Fig. 2D for simplicity. Normalized Ct values for GST in subsequent experiments were similar and were always <0.01. Reported results represent two to five independent experiments with duplicates.
EMSA and run-on assay
Primary monocytes were treated as indicated in Fig. 6, with nuclear extracts made as previously described (29, 30). Nuclear extracts were bound to the biotinylated double-stranded ATF-2 consensus sequence (TGAGAGATTGCCTGACGTCAGAGAGCTAGCA) and run on a precast gel (Invitrogen Life Technologies), according to the manufacturer’s instructions. Blotting and detection were performed using the methods of the Lightshift Chemiluminescent EMSA kit (Pierce). The modified run-on assay (31) required the use of a cell line, THP-1, due to the number of cells required. Ten to 12 million cells per condition were treated as follows: 100 nM trichostatin A for 1 h before LPS, LPS at 1 μg/ml for 2 h, and IFN-γ at 50 ng/ml for 18 h either alone or followed by LPS. Nuclei were isolated and transcription was allowed to proceed in the presence of biotin-16-UTP for 2 h (Roche). The nascent transcripts were collected on avidin magnetic beads (Dynal Biotech; Invitrogen Life Technologies), and AMV reverse transcriptase was used to generate cDNA. The cDNA was quantitated by RealTime PCR using 18S primers (Applied Biosystems) for normalization and TNF-α primers in the first exon: 5′-GCGGGAAATATGACAGCTAAGG-3′ and 5′-TCTTTCCCTGAGTGTCTTCTGTGT-3′ and the probe: 5′-AGGAGAGAAGAAGATAGGGTG-3′.
Significance calculations were performed using Student’s t tests with a Bonferroni correction for the four amplimers. All reported p values include the correction. Error bars indicate SD.
IFN-γ polarization of human monocytes significantly increases TNF-α production and kinetics
To establish a system of classical and alternative polarization using IFN-γ and IL-4, we performed a dose-response study for a range of IFN-γ and IL-4 cytokine concentrations (data not shown). Monocytes were pretreated for 18 h with IFN-γ or IL-4 before being stimulated with 1 μg/ml LPS for 6 h. A cytokine dose of 50 ng/ml was associated with consistently strong polarization with no cell death.
Treatment with either cytokine alone resulted in minimal TNF-α production as measured by ELISA (Fig. 1,A), but polarization with IFN-γ before LPS stimulation resulted in significantly increased TNF-α production compared with nonpolarized cells (p = 0.002). IL-4 polarization before LPS stimulation significantly decreased TNF-α production (p = 0.006). As an independent confirmation of polarization, polarized cultures were stimulated with LPS for 24 h before supernatants were collected and an ELISA was performed for IL-12p70 (Fig. 1 B). As expected, IL-12p70 production only occurred in LPS-stimulated monocytes that had been polarized with IFN-γ.
Next, we examined the kinetics of TNF-α mRNA accumulation by quantitative RT-PCR in polarized and nonpolarized monocytes (Fig. 1, C and D). Treatment with IFN-γ or IL-4 alone did not induce appreciable amounts of TNF-α message. Induction of transcription was observed upon LPS stimulation. IL-4-polarized, LPS-stimulated cells had decreased TNF-α message, but with similar kinetics of accumulation as nonpolarized cells. In contrast, IFN-γ-polarized, LPS-stimulated cells had greatly increased kinetics of TNF-α mRNA accumulation from 1–2 h of LPS treatment (triangles). The increased kinetics suggested that IFN-γ polarization was affecting TNF-α at the transcriptional level. The pretreatment with IFN-γ was required for this increase in kinetics because treatment with IFN-γ concurrently with LPS stimulation (square symbols) resulted in kinetics and accumulation identical with LPS alone (heavy line, circles).
To ensure that this represented an increased rate of transcription after IFN-γ polarization, a modified run-on assay was performed in THP-1 cells. IFN-γ polarization or treatment with an agent to increase histone acetylation (trichostatin A) before LPS markedly increased the rate of transcription compared with the rate of transcription in LPS-treated cells (Fig. 1 E).
Cytokine polarization alters histone acetylation of the TNF-α promoter and third intron enhancer
Previously, our laboratory showed that LPS stimulation results in chromatin remodeling of the TNF-α promoter in human monocytes, specifically by transiently modifying histones H3 and H4 with a peak effect at 30–60 min (20). We hypothesized that cytokine polarization was affecting histone modifications at the TNF-α promoter and thereby exerting effects on TNF-α production. Initially, H3 acetylation, H4 acetylation, H3 lysine 4 dimethylation, and H3 lysine 4 trimethylation were examined by CHiP (data not shown). Only histone acetylation was affected by polarization, and this study focused on that finding.
The ChIP assays reported in this study used the amplimers demonstrated in Fig. 2,A. Polarization for 18 h with IFN-γ or IL-4 alone had little effect on the acetylation of H3 at the TNF-α promoter (Fig. 2,B). This is in contrast to its effects on acetylation of H4 (Fig. 2 C) in the distal TNF4 region.
The effect of IFN-γ polarization was more obvious following LPS stimulation for 30 min. IFN-γ polarization increased H3 acetylation of the distal region (TNF4, p = 0.002; TNF3, p < 0.00001) 2-fold over LPS alone and 4-fold over nontreated cells. Following LPS stimulation for 30 min, the acetylation of H3 in the TNF-α promoter of IL-4-polarized cells was virtually identical with that of LPS alone. Similarly, IFN-γ polarization followed by LPS stimulation resulted in significantly increased acetylation of H4 in TNF4 and TNF3 compared with LPS-stimulated monocytes (TNF4, p = 0.0002; TNF3, p = 0.0019). This effect appeared to be an additive effect of IFN-γ and LPS effects. These effects were specific to TNF-α because we detected no significant alterations in H3 or H4 acetylation in the GAPDH or β-globin promoters (data not shown). IL-4 treatment did not lead to significant changes in histone modifications at the locations tested, suggesting that IL-4 polarization occurs via a separate mechanism. A negative control Ab, anti-GST, was run with each experiment, but always gave negligible signal and is shown in this study as a representative example (Fig. 2,D). We also explored whether histone acetylation was occurring in the third enhancer region. IFN-γ also appeared to increase H3 and H4 acetylation at the enhancer (Fig. 2 E).
The discordance of the signal from the distal TNF3, 4 probes compared with the signal from the proximal TNF2, 1 probes after LPS treatment suggested that nucleosome depletion might be occurring. We used a pan-H3 Ab to define nucleosome content at the TNF-α promoter. The signal at TNF4 and 3 was higher than the signal at TNF2 and 1 and was constant following LPS stimulation (Fig. 2 F). These data suggest that the proximal promoter has limited nucleosome content and there is little change after LPS stimulation.
The effects of IL-4 on TNF-α production are reversible, but the effect of IFN-γ is durable
Alterations of chromatin modifications may be a mechanism by which terminally differentiated cells maintain gene expression patterns even upon withdrawal of activating stimuli or alteration of the microenvironment. To determine whether IL-4 and IFN-γ polarization confer a durable phenotype with respect to TNF-α production, monocytes were cultured for 18 h with IL-4, IFN-γ, or medium before being washed with warm PBS and recultured for another 18 h in medium alone or with IFN-γ or IL-4 (Fig. 3). Cells were then stimulated with LPS and TNF-α production measured after 6 h by ELISA.
Monocytes not stimulated with LPS had no detectable TNF-α production in any of the conditions (light gray bars). Compared with nonpolarized cells, IFN-γ-polarized cells that were then rested in medium (IFNG NC) still had significantly increased TNF-α production (p = 0.0055) compared with nonpolarized cells (NC NC). The withdrawal of IFN-γ did result in significantly decreased TNF-α production compared with monocytes polarized immediately before LPS stimulation (NC IFNG) (p = 0.00014), indicating that the continued presence of IFN-γ is necessary to maintain some aspects of polarization with respect to intensity of TNF-α production. This is confirmed by the maintenance of high TNF-α production in cells receiving only IFN-γ during both polarization periods (IFNG IFNG).
To determine whether the durability of IFN-γ polarization on protein production was paralleled by durability of chromatin changes, histone modifications were examined. Polarization with IFN-γ alone (NC IFNG) increased H4 acetylation at TNF3 and TNF4, but this effect was diminished when IFN-γ was washed out (IFNG NC) (Fig. 4,A). This decrease paralleled the effects on protein production seen in Fig. 3, but neither protein production nor histone modifications returned to baseline. IL-4-treated cells repolarized with IFN-γ (IL-4 IFNG) showed significantly increased H4 acetylation at TNF4 and TNF3 compared with nonpolarized cells (NC NC) (TNF4, p = 0.0018; TNF3, p = 0.01) and IL-4 only (IL-4 NC)-polarized cells (TNF4, p = 0.0001; TNF3, p = 0.001). Although IL-4 repolarization of IFN-γ-treated cells (IFNG IL-4) showed significantly decreased H4 acetylation compared with NC IFNG (TNF4 p = 0.03), this decrease was not significantly different from that observed when IFN-γ was washed out and the cells rested (IFNG NC) (p = 0.5).
Interestingly, we found that the effects of IFN-γ polarization were dominant to those of IL-4, because monocytes that received IFN-γ at any point relative to IL-4 produced significantly more TNF-α compared with monocytes polarized only with IL-4 (IL-4 IL-4) (IL-4 IFNG, p = 0.000015; IFNG IL-4, p = 0.002).
Next, we repeated these experiments, but defined the acetylation of H4 by CHiP both before and after 30 min of stimulation with LPS. Once again, we found that IL-4 polarization alone or in combination with LPS caused little change in H4 acetylation compared with nonpolarized cells (Fig. 4,B). Following LPS treatment, we again saw that cells that received IFN-γ at any time point had significantly increased H4 acetylation of the TNF-α promoter compared with nonpolarized cells or those that only received IL-4 (Fig. 4 B). In the setting of LPS stimulation, the durability of the IFN-γ effect appeared enhanced with H4 acetylation levels comparable in all IFN-γ-treated cells.
Levels of ATF-2 and HDAC3 in IFN-γ-polarized monocytes
The effect of IFN-γ on H4 acetylation at the TNF-α promoter was hypothesized to be indirect because there are no STAT1 binding sites in the TNF-α promoter. Two histone acetyl transferases are known to bind the TNF-α promoter (CBP and ATF-2) (11, 12, 32). ATF-2 demonstrates a preference for H4 acetylation over H3 acetylation, and we hypothesized that ATF-2 could be responsible for IFN-γ-mediated H4 acetylation. ATF-2 is primarily activated via phosphorylation by the MAPK, p38, although other mechanisms of activation exist (10, 33). It is negatively regulated by HDAC3, which dephosphorylates ATF-2 (34). To define the role of ATF-2 in polarized monocytes, we examined the cellular levels of phosphorylated ATF-2 and HDAC3 (33). The cells were polarized for 18 h with either IFN-γ or IL-4 as usual and then treated for various times with LPS. After 18 h of culture, both IL-4-polarized cells and IFN-γ-polarized cells had phosphorylated ATF-2 present (Fig. 5). Following LPS stimulation, phospho-ATF-2 increased in both polarized and nonpolarized cells. Unexpectedly, IL-4-polarized cells stimulated with LPS had more phospho-ATF-2 than nonpolarized cells, comparable to those in IFN-γ-polarized cells. For these experiments, all membranes were incubated in the same Ab, and the film exposures were simultaneous to allow cross-comparisons.
Before stimulation, nonpolarized monocytes had higher levels of HDAC3 than the polarized cells. Following LPS stimulation, all cells experienced an acute decrease in HDAC3 levels. IFN-γ-polarized cells had the lowest level of HDAC3 at the outset; however, the pattern of HDAC3 expression after LPS was comparable in IFN-γ- and IL-4-polarized cells. These data suggest that the absolute levels of phospho-ATF-2 and HDAC3 may contribute to H4 acetylation, but are unlikely to represent the main mechanism regulating H4 acetylation. Recruitment to a specific regulatory region is likely to be more important than cellular levels.
The role of MAPKs in IFN-γ-regulated H4 acetylation
Both p38 and ERK are well recognized as important regulatory pathways for TNF-α (35, 36, 37, 38). To define the roles of ERK and p38 in the acetylation of H4, we treated monocytes with 10 μM PD58059, a MEK inhibitor that blocks ERK activation, or 10 μM SB203580, a specific p38 inhibitor, for 30 min before adding IFN-γ to cultures (39). In nonpolarized cells and IFN-γ-polarized cells, there was no effect of ERK or p38 inhibition on resting levels of H4 acetylation (Fig. 6 A). Inhibition of p38 completely blocked LPS-mediated H4 acetylation in nonpolarized monocytes (TNF4, p = 0.0001; TNF3, p = 0.0011; TNF2, p = 0.038), whereas ERK inhibition had no effect. In IFN-γ-polarized, LPS-stimulated cells, there was only a slight decrease in H4 acetylation with either inhibitor compared with the IFN-γ-polarized, LPS-stimulated vehicle control (p38 TNF4, p < 0.00001; TNF3, p = 0.05; ERK TNF4, p = 0.0008; TNF3, p = 0.06). These data demonstrate that LPS and IFN-γ use different pathways to induce H4 acetylation at the TNF-α promoter.
We next evaluated ATF-2 recruitment. Inhibition of either p38 or ERK resulted in significantly increased ATF-2 recruitment in nonpolarized, nonstimulated cells compared with the vehicle control (SB203580 TNF4, p = 0.107; TNF3, p = 0.024; TNF2, p = 0.001; TNF1, p = 0.001; PD58059 TNF4, p = 0.006; TNF3, p = 0.0004; TNF2, p = 0.495; TNF1, p = 0.092). There is no clear explanation for this finding at this time. Monocytes polarized with IFN-γ or stimulated with LPS showed recruitment of ATF-2 to the TNF-α promoter (Fig. 6,B) (IFN-γ TNF4, p = 0.06; TNF3, p = 0.008; TNF2, p = 0.014; TNF1, p = 0.001; LPS TNF4, p = 0.07; TNF3, p = 0.004; TNF2, p = 0.021; TNF1, p = 0.00001). Inhibition of p38 led to decreased ATF-2 recruitment in IFN-γ-treated cells, but there was no decrease in LPS-treated cells. ERK inhibition had no effect in either single treatment setting, but did decrease ATF-2 recruitment after IFN-γ polarization with LPS stimulation. Due to the dual binding sites within the TNF4 and TNF2 amplimers, as well as the overlapping nature of the probes, the binding appears across the entire promoter. From these data, we conclude that IFN-γ-mediated ATF-2 recruitment to the TNF-α promoter is at least partly p38 dependent, but ATF-2 is unlikely to be responsible for IFN-γ-mediated H4 acetylation. To confirm the activation of ATF-2 by IFN-γ, an EMSA was performed (Fig. 6,D). IFN-γ activated ATF-2 binding, as did LPS. Binding was inhibited in SB203580-treated cells. To determine whether protein synthesis was required for IFN-γ-mediated H4 acetylation, cells were treated with IFN-γ either with or without 2.5 μM cycloheximide (Fig. 6 E). Viability was >95% at 18 h of polarization. Cycloheximide clearly inhibited H4 acetylation after IFN-γ.
IFN-γ recruits RNA Pol II to the TNF-α promoter independently of transcription initiation
MAPK pathways have been implicated in the activation of RNA Pol II, and ATF-2 has been reported to directly recruit RNA pol II (40, 41). We therefore examined the effect of ERK and p38 inhibition on RNA Pol II recruitment. Interestingly, IFN-γ alone was able to recruit RNA Pol II to the TNF-α promoter (TNF2, p < 0.00001; TNF1, p = 0.00034), and yet it did not activate transcription in human primary monocytes (Figs. 1 and 6 C). It has previously been shown that stable RNA Pol II recruitment can occur in the absence of transcription for certain genes, but this is the first description of the phenomenon for TNF-α (23, 42, 43). This finding provides a mechanism by which IFN-γ could increase the kinetics and level of LPS-mediated TNF-α transcription by preassembling the transcriptional machinery at the promoter. Blockade of p38 for IFN-γ-polarized cells reduced the level of recruited RNA Pol II (TNF2, p = 0.0002; TNF1, p = 0.259), but blockade of ERK completely abolished IFN-γ-mediated RNA Pol II recruitment to the TNF-α promoter (TNF2, p < 0.00001; TNF1, p = 0.00087). This represents a novel role for ERK.
LPS alone also recruited RNA Pol II and did so in a p38-dependent manner (TNF2, p = 0.007; TNF1, p = 0.04). ERK had no effect on LPS-mediated RNA Pol II recruitment. Interestingly, neither blockade of ERK or p38 was able to abolish RNA Pol II recruitment in LPS-stimulated, IFN-γ-polarized cells, indicating that the pathways may be compensatory.
p38 and ERK signaling regulate TNF-α mRNA levels
We examined the effects of inhibiting p38 or ERK on TNF-α mRNA production. The cells were polarized as before, with RNA being collected at 0, 1, 2, and 3 h after LPS stimulation. We found that p38 inhibition nearly abolished LPS-mediated TNF-α transcription (Fig. 7 A). In contrast, we saw that ERK inhibition was associated with increased TNF-α mRNA accumulation after 1 h.
TNF-α mRNA was produced in IFN-γ-polarized monocytes following LPS stimulation even in the presence of p38 inhibitors (Fig. 7 B); however, the amounts were significantly reduced compared with the vehicle control. In the IFN-γ-polarized, LPS-stimulated cells, we again found that ERK negatively regulated TNF-α mRNA accumulation because blocking ERK increased TNF-α production over that of IFN-γ-treated, LPS-stimulated cells. This finding is incompletely explained by the chromatin data and may indicate that ERK inhibition is acting at a posttranscriptional level. From these data, we conclude that both ERK and p38 MAPKs participate in the regulation of TNF-α production in IFN-γ-polarized cells, but they appear to act at many levels with an end result of opposite effects on transcript accumulation.
The role of chromatin in regulating gene expression is well studied in cells of the adaptive immune system (44, 45, 46). Acetylation of H3 and H4, in general, correlates with competence for gene transcription and has been shown to enhance binding of transcription factors to nucleosomal DNA (47, 48). A specific pattern of histone modifications, including methylation as well as acetylation, has a high correlation with active gene expression in genome-wide studies (49, 50, 51, 52). These patterns are set in part by the differentiation program of the cells, with dynamic changes in histone modifications occurring in response to acute stimuli (53).
Histones are modified at the TNF-α promoter during monocyte differentiation and in response to LPS (20, 21). Given the influence of cytokines on the maturation and activation state of monocytes, macrophages, and dendritic cells, we investigated whether the classical models of monocyte polarization using IFN-γ and IL-4 regulated TNF-α expression through modification of histones. We were surprised that IL-4 had little effect on histone modifications despite its ability to significantly decrease TNF-α protein production, whereas IFN-γ increased histone acetylation significantly and this mechanism appears to be important for the regulation of expression. Both LPS stimulation and IFN-γ polarization led to significantly increased acetylation of H3 and H4 in the distal −100 to −345 region. The data from this study demonstrate that durable polarization effects on H4 acetylation and transient effects on H4 acetylation due to LPS stimulation are clearly regulated distinctly.
Monocyte polarization is a complex process, and there are limited data on the durability of the effect. One study evaluated sequential treatment with IL-10 or IL-4 and IFN-γ and found that the effects of polarization were largely, but not completely reversible for the expression of six cytokines examined after LPS stimulation (54). Neither a washout nor repolarization diminished the effect of IFN-γ on IL-1β, and the effect of IFN-γ appeared to be dominant over IL-10. Similarly, in IL-4-treated cells, the effect on MCP-1 expression was not altered by a washout nor by repolarization with IFN-γ. For most of the other genes examined, the polarization effect was transient. These data suggest that the polarization effect is not durable for all genes, but for a subset of genes. Durability of polarization effects has important implications for concepts such as the hygiene hypothesis and immunologic memory (55). We found that IFN-γ polarization mediated increased LPS-induced TNF-α protein production, increased transcription kinetics, and increased H4 acetylation. This increase in TNF-α protein production and H4 acetylation was diminished, but still detectable after a washout and 18-h rest period. Additionally, IFN-γ was capable of increasing TNF-α production and acetylation of H4 in cells previously polarized by IL-4; however, IL-4 treatment of cells previously polarized with IFN-γ did not decrease either TNF-α production or acetylation of H4. This indicates two nonmutually exclusive possibilities, as follows: 1) the effect of IFN-γ is dominant over the effect of IL-4, or 2) IFN-γ and IL-4 polarize cells via different mechanisms.
Several papers have recently implicated HDAC proteins as being important for transcriptional regulation of inflammatory genes as well as regulating the activity of transcription factors (56, 57, 58). HDAC3 has been implicated in direct gene repression through interactions with the silencing mediator of retinoid and thyroid receptors corepressor and is also involved in suppressing the activity of the histone acetyltransferase, ATF-2 (34, 59). Surprisingly, we found that both IL-4 polarization and IFN-γ polarization led to increased phosphorylation of ATF-2 and decreased levels of HDAC3 in monocytes. IFN-γ did decrease HDAC3 levels more than IL-4, both before and after LPS treatment; however, the absolute levels of these two proteins do not appear to be as important as their recruitment to specific genes.
The TNF-α promoter is known to bind at least two different histone acetyltransferases, as follows: ATF-2 and CBP/p300. ATF-2 appeared to be the more promising candidate regulating H4 acetylation at the TNF-α promoter. ATF-2 has a preference for H4 acetylation and had been shown in preliminary experiments to regulate TNF-α (data not shown). However, we ultimately found that it had a complex relationship with transcription, and we could not directly implicate it in H4 acetylation. In our study, ATF-2 did not appear to be responsible for H4 acetylation after IFN-γ, but was recruited to the promoter by IFN-γ in a p-38-dependent fashion. ATF-2 recruitment after LPS occurred in a p-38-independent fashion, underscoring the differences between polarization and acute stimulation. The roles of p38 have been characterized for both transcriptional control and posttranscriptional control of TNF-α production (60, 61, 62, 63, 64). There is also evidence that the MAPK pathways control TNF-α transcription via RNA Pol II activation (41). Our study contributes to a growing body of literature in the importance of p38 in chromatin (65).
We have shown for the first time that IFN-γ polarization alone is capable of recruiting RNA Pol II to the human TNF-α promoter independent of any observable transcription of mRNA. We hypothesize that IFN-γ polarization may lead to increased TNF-α transcription by preassembling the transcriptional machinery at the TNF-α promoter (66). Both artificial histone modifications by trichostatin A and treatment with IFN-γ led to enhanced transcription kinetics. We also found that IFN-γ-mediated RNA Pol II recruitment was dependent on ERK signaling, whereas LPS stimulation led to recruitment of RNA Pol II via p38. ERK has not been previously shown to have a role in RNA Pol II recruitment.
Thus, we have described a situation in which IFN-γ and TLR4 signaling could collaboratively acetylate H4 and recruit RNA Pol II via different pathways. A potential model includes the TLR4-LPS pathway driving a rapid increase in H4 acetylation and recruitment of RNA pol II, both dependent on p38 activation. This effect is transient and may serve to produce a pulse of TNF-α message (20). This rapid remodeling is similar to what has been observed for other acutely activated genes (67, 68). The effect of IFN-γ is to enhance the transcriptional competence of the TNF-α locus. No effect on message levels was seen when IFN-γ was added simultaneously with LPS. Its effect on H4 acetylation was slower in onset (2–4 h vs 30 min), and the effect, although not permanent, was substantially more durable than the effect of LPS-driven H4 acetylation. The effect was still detectable 18 h after the IFN-γ was washed out. The polarization effect was mediated in part by ATF-2 and RNA pol II recruitment to the promoter. These phenomena were p38 dependent, but the effect of IFN-γ on H4 acetylation did not appear to be via p38. Thus, IFN-γ and LPS led to increased H4 acetylation by different pathways, and this may relate to the kinetics of acetylation. The existence of a pathway leading to durable changes in histone modifications in terminally differentiated cells is of particular interest. Understanding the pathways regulating TNF-α transcription are important for the development of therapeutic interventions directed at inflammation. The pathways identified in this study could potentially be targeted for autoimmune granulomatous diseases in which IFN-γ effects are believed to perpetuate the granulomas (69, 70, 71).
We acknowledge the Human Immunology Core at the University of Pennsylvania, which provided the primary monocytes for this study.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by National Institutes of Health Grants R01 AI051323 and R01 AI44127. S.G. was supported in part by American Academy of Allergy, Asthma and Immunology’s Strategic Training in Allergy Research (ST*AR) Award, University of Pennsylvania School of Medicine Cancer Research Institute Predoctoral Emphasis on Pathways in Tumor Immunology Training Grant, and the Goldie Simon Award.
Abbreviations used in this paper: ATF, activating transcription factor; CBP, CREB-binding protein; ChIP, chromatin immunoprecipitation; Ct, cycle threshold; HDAC, histone deacetylase.