IL-10 inhibits transcription elongation of the human TNF gene in primary macrophages

We initially examined the effects of IL-10 on LPS-induced mRNA expression in primary human macrophages. IL-10 inhibited TNF and IκB-ζ mRNA production at early and late time points and IL-6, a secondary response gene, at 120 min (Fig. 1 A). IL-10 did not inhibit the production of IκB-α mRNA, which is consistent with the gene-specific effects previously described (Lang et al., 2002; Murray, 2005). A longstanding question has been whether IL-10 inhibits gene expression at the transcriptional (Murray, 2005) or posttranscriptional level (Schaljo et al., 2009), or perhaps both (Denys et al., 2002), in activated macrophages. Chromatin immunoprecipitation (ChIP) was used to measure accumulation of RNA polymerase II (Pol II) at distal regions of genes (Fig. 1 B), which is considered a hallmark of active transcription (Sandoval et al., 2004). LPS induced rapid accumulation of Pol II at a distal region of the TNF gene and slower accumulation at a distal region of the IL6 gene (Fig. 1 C), which is consistent with the identification of these as primary and secondary response genes (Hargreaves et al., 2009; Natoli, 2009; Ramirez-Carrozzi et al., 2009). IL-10 inhibited the accumulation of Pol II in the distal regions of TNF, IL6, and NFKBIZ, but not NFKBIA (Fig. 1 D), mirroring its effects on the corresponding mRNAs (Fig. 1 A). Nascent unspliced TNF transcripts were also quantified by PCR (Fig. S1 A) and were seen to accumulate with identical kinetics to those of mature TNF mRNA in these cells. As estimated by both of these methods, IL-10 decreased transcription of TNF (Fig. 1 D and Fig. S1 A). These findings in human macrophages are consistent with previous reports describing IL-10’s gene-specific effects on transcription (Lang et al., 2002; Murray, 2005).

Activation of transcription elongation has recently emerged as an important mechanism controlling the rapid induction of proinflammatory genes (Adelman et al., 2009; Hargreaves et al., 2009; Ramirez-Carrozzi et al., 2009). We therefore investigated whether IL-10 inhibits transcription elongation by comparing Pol II densities at the transcription start site (TSS) to the downstream region of TNF. As described by others (Adelman et al., 2009; Hargreaves et al., 2009; Ramirez-Carrozzi et al., 2009), Pol II was present at the TNF TSS under resting conditions and increased relatively weakly in response to LPS (Fig. 2 A). In contrast, it was scarcely detectable at the downstream region in resting cells and very strongly up-regulated by LPS, which is consistent with control of gene expression largely through the release of a block to transcription elongation. IL-10 weakly inhibited the LPS-induced increase of Pol II at the TSS but strongly inhibited the increase of Pol II at the distal region (Fig. 2 A and Fig. S1 B). Phosphorylation of Ser2 of the Pol II C-terminal domain (CTD) by CDK9 occurs during the transition from the initiation of transcription to elongation (Price, 2000). We therefore performed ChIP using a phospho-Ser2–specific antibody. Where IL-10 did not inhibit the recruitment of total Pol II to the start of the TNF gene (Fig. 2 A; top left) it completely inhibited the phosphorylation of the Ser2 residue of the Pol II CTD (Fig. 2 A; bottom left). Therefore, at TNF the elongating form of Pol II is reduced by IL-10. When we conducted similar analysis at the NFKBIA gene, it was noted that neither the elongation of Pol II into the downstream region of the gene nor the phosho-Ser2 form of Pol II was decreased by IL-10 (Fig. 2 B).

We used the CDK9 inhibitor flavopiridol (Chao and Price, 2001) to confirm that the phosphorylation of Ser2 of the Pol II CTD controls elongation at TNF. Flavopiridol had negligible effects on total Pol II recruitment to the start of TNF in contrast to its inhibition of Pol II Ser2 phosphorylation (Fig. 2 C). Additionally, flavopiridol inhibited the accumulation of total Pol II at the downstream region of the TNF gene. These effects mirror IL-10’s inhibition of Pol II phospho-Ser2 and elongation into the downstream regions of the gene (Fig. 2 A).

Interestingly, transcription elongation of NFKBIA was also sensitive to flavopiridol (Fig. S1 D), indicating that although IL-10 had no effect on elongation of NFKBIA, inhibiting the kinase activity of CDK9 with flavopiridol elicited a broader effect. Therefore it is unlikely that IL-10 has inhibitory effects on the kinase activity of CDK9 (as was the case for flavopiridol). We therefore looked at the effect IL-10 had on the recruitment of CDK9 itself to the TNF and NFKBIA genes (Fig. 2 D). At the start of TNF, CDK9 was recruited in a stimulus-dependent manner and this was inhibited by IL-10 (mean 47.2%; SEM 6.5; n = 4). In contrast, at the start of NFKBIA, CDK9 was seen to be more constitutively associated, inducing only slightly in response to LPS and exhibiting no inhibition upon IL-10 costimulation. Recently, NFKBIA was described as a housekeeping primary response gene that is continuously transcribed in macrophages (Hargreaves et al., 2009), perhaps accounting for these differences. These differential effects on CDK9 recruitment between TNF and NFKBIA may underlie IL-10’s gene-specific, as opposed to general, inhibition of transcription elongation.

The current findings contradicted a previous observation from our laboratory, where the TNF 3′ untranslated region (UTR) was required for IL-10 to suppress the TNF reporter gene activity (Denys et al., 2002). This implicated a mechanism of posttranscriptional control; however, on closer inspection, the construct in question, 5′luc3′(1037) (Fig. 3 A), was found to include both the 785-bp TNF 3′UTR and an additional 250 bp of downstream genomic sequence (originally included in the construct for ease of cloning). We hypothesized that this 250-bp region, rather than the 3′UTR, could have mediated the IL-10–dependent inhibition of 5′luc3′(1037). Indeed, when this construct was compared with a 5′luc3′(785) that lacked the additional genomic sequence, IL-10–dependent inhibition was lost (Fig. 3 A and Fig. S2 A). This also resulted in a partial loss of LPS responsiveness, indicating that the 3′ downstream region may contribute to TNF transcriptional control.

The downstream 3′ region of the TNF gene is known to contain the NF-κB binding site κB4 (Tsytsykova et al., 2007; Krausgruber et al., 2010). Although it is well documented that NF-κB plays a role in TNF transcription in macrophages (Foxwell et al., 1998; Udalova et al., 2000), previously we and others were unable to demonstrate any IL-10–dependent suppression of NF-κB activity (Denys et al., 2002; Murray, 2005). Therefore, to establish to what extent NF-κB contributes toward TNF transcription, we depleted the levels of RelA protein using a small interfering (si) RNA approach, which was recently optimized for primary human macrophages in our laboratory (Behmoaras et al., 2008; Fig. 3 B). The effect of RelA knockdown was then determined on the LPS induction of TNF (Fig. 3 C), as well as IL-6, IκB-ζ, and IκB-α mRNA (Fig. S2 B), and on the secretion of TNF and IL-6 protein (Fig. S2 C), and it could be seen that their expression was dependent on RelA. This, along with the fact that the IL-10–sensitive downstream 3′ region of the TNF gene contains the κB4 binding site, prompted us to explore previously unexamined aspects of NF-κB signaling that IL-10 may regulate. To examine effects on the nuclear translocation of RelA, we compared nuclear versus cytoplasmic protein fractions from macrophages stimulated with LPS or IL-10 and found no major role for IL-10 in preventing RelA nuclear translocation (Fig. S3). The posttranslational modification, by means of LPS-induced phosphorylation of RelA Ser536 (a known mark of RelA function), was also unaffected (Fig. S3). We also rechecked whether IL-10 modulated the LPS-induced degradation and resynthesis of IκB-α in the cytoplasm of human macrophages yet found no effect, which is consistent with previous findings in mouse macrophages (Murray, 2005).

We then went on to examine the actual recruitment of NF-κB to κB binding sites at proinflammatory genes using ChIP. Initially the kinetics of RelA recruitment was assessed in these cells (Fig. S4 A). We observed a time-dependent transient recruitment of RelA to the κB sites of the TNF gene peaking at 60 min after LPS stimulation. IL-10 inhibited the recruitment of RelA to all of the regions in the TNF gene containing κB sites (Fig. 3 E), including the downstream region containing κB4 (Fig. 3 D). As a control, PCR was performed using primers within the TNF introns (Fig. 3 D, region D), where no κB sites were found and no recruitment was seen at this position, confirming the specificity of the assay.

To ascertain whether the IL-10 inhibition of RelA recruitment occurred at additional genes, we performed similar analysis at the κB sites in the NFKBIA, TNFAIP3, and CCL20 promoters, to which RelA was similarly inhibited (Fig. S4 B) This also occurred at the IL6 κB site (albeit with later kinetics of recruitment; Fig. S4 C). It therefore appears that the mechanism elicited by IL-10 is a widespread block in recruitment of RelA to κB sites in the genome.

Numerous studies have shown a role for RelA in the activation of CDK9-mediated control of transcription elongation in alternate systems (Barboric et al., 2001; Huang et al., 2009). Considering that IL-10 inhibits RelA recruitment to the TNF gene, we wondered whether depletion of RelA in our experimental system could replicate the effects of IL-10 on transcription elongation. As expected, using an RNAi approach to deplete RelA, after LPS stimulation no RelA was detected to the TNF downstream region containing the κB4 site (Fig. 4 A). Concomitantly, Rel A depletion had no significant effect on the recruitment of total Pol II to the start of TNF, yet it did result in an inability of LPS to induce its progression down the length of TNF (Fig. 4 B). This also coincided with a loss of inducible phosphorylation of Pol II at Ser2, and an inability to recruit CDK9 to the start of TNF in RelA-depleted macrophages (Fig. 4 C). These data show that artificial inhibition of RelA recruitment to its target κB sites replicates IL-10’s effects on TNF transcription elongation.

It is notable that IL-10 also inhibited the recruitment of RelA to the NFKBIA κB site, despite the fact that transcription of this gene is spared from inhibition. The fact that neither CDK9 recruitment nor Pol II Ser2 phosphorylation was inhibited by IL-10 at NFKBIA suggests that CDK9 recruitment here is independent of RelA. Certain aspects of these data bear resemblance to a study of the gene-specific actions of glucocorticoids on CDK9 recruitment, where inhibition was observed at the gene coding for IL-8 but not at NFKBIA (Luecke and Yamamoto, 2005). The up-regulation of IκB-α is part of a negative-feedback loop that limits inflammatory responses; therefore, it is unsurprising that this escapes negative regulation by IL-10 and dexamethasone.

To conclude, we provide novel insights into the mechanisms of the IL-10 antiinflammatory response. We have shown for TNF that IL-10 predominantly inhibits transcription through a unique mechanism targeting the rapid and immediate induction of transcription elongation by Pol II. IL-10 achieves this in a gene-specific manner through the inhibition of RelA-mediated recruitment of CDK9 to the TNF but not the NFKBIA promoter, thus preventing the phosphorylation of RNA Pol II at Ser2. This study, performed entirely in the physiologically relevant primary human macrophage, has addressed long-standing contradictions within the field. The discovery of a unique negative regulatory checkpoint within the human innate immune system (CDK9 modulation of transcription elongation) has implications for the development of therapeutics that harness the central role that IL-10 plays in the suppression of inflammatory processes in humans.

IL-10 was a gift (Schering Plough), macrophage CSF (M-CSF) was a gift (Pfizer), and LPS was purchased (Enzo Life Sciences, Inc). Flavopiridol (Enzo Life Sciences, Inc.) was used at a 500-nM concentration for 30 min before stimulation with LPS.

Single-donor platelet phoresis residue packs were purchased from the North London Blood Transfusion Service. Mononuclear cells were isolated by Ficoll-Hypaque centrifugation (specific density, 1.077 g/ml) preceeding T cell/monocyte separation in a JE6 elutriator (Beckman Coulter). T cell purity was assessed by flow cytometry using directly conjugated anti-CD3 (BD), and monocyte purity, assessed using anti-CD45 and anti-CD14 antibodies (Leucogate; BD), was routinely >90%. All media and sera were routinely tested for endotoxin using the Limulus amebocyte lysate test (BioWhittaker; Lonza) and were rejected if the endotoxin concentration exceeded 0.1 U/ml. Macrophages were derived from elutriated monocytes by culturing the cells with M-CSF at 100 ng/ml (Pfizer) in 10% heat-inactivated FCS RPMI 1640 for 3 d.

The anti-IκBα (C-15) antibody, anti–RNA Pol II, and anti-CDK9 (H-169) were purchased from Santa Cruz Biotechnology, Inc. Anti-tubulin was purchased from Sigma-Aldrich. Anti-lamin A/C was purchased from BD. Anti Cox-2 was purchased from Cayman Chemical. The anti-Pol II CTD phosphor-Ser2, rabbit isotype control antibody, and anti-p65 were all purchased from Abcam.

The TNF-luciferase reporters AdV.Luc.5′only and AdV.Luc.3′(1037) were generated as previously described (Denys et al., 2002). The AdV.Luc.3′(785) vector was generated by PCR using the primer pairs 5′-CAGTCTAGAG­AGGACGAACAT­CCAACCTTC-3′ and 5′-CGAGTCGACG­CAAACTTTATT­TCTCGCCAC-3′. The PCR product was cloned from the pCRII-TOPO TA vector using XbaI and SpeI into the 3′ region of the pAdTrack vector used to make the AdV.Luc.5′only virus. This was used to generate the adenoviral vector by homologous recombination with the pADeasy vector in Escherichia coli BJ cells and purified as per previously devised methods (He et al., 1998). Macrophages were infected with virus at a multiplicity of infection of 100 as previously described (Foxwell et al., 1998).

After stimulation, cells were scraped into ice-cold PBS and then resuspended in hypotonic lysis buffer (10 mM Hepes, pH 7.9, 10 mM KCl, 0.1 mM EGTA, 0.1 M EDTA, 1 mM NaV0₃, 0.5 mM NaF, 1 mM PMSF, protease inhibitor tablet, and 1 mM DTT). Cells were kept on ice for 15 min before the addition of NP-40 to a final concentration of 0.0064%, vortexed for 10 s, and nuclei were harvested by centrifugation (13,000 rpm for 1 min). The supernatant containing the cytosolic extract was kept at −80°C and the nuclear pellet was washed once in hypotonic lysis buffer. Nuclear protein extracts were prepared by incubating the nuclei in hypertonic extraction buffer (20 mM Hepes, pH 7.9, 400 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM NaV0₃, 0.5 mM NaF, 1 mM PMSF, protease inhibitor tablet [Roche], and 1 mM DTT) for 2 h with constant agitation at 4°C. Postnuclear lysates were then isolated after a 15-min spin at 13,000 rpm and frozen at −80°C. Protein assays was performed using BCA protein assay (Thermo Fischer Scientific) according to manufacturer’s instructions. Proteins were resolved by SDS-PAGE and transferred to polyvinyl difluoride membranes (Millipore), which were blocked for 1 h with blocking buffer (5% wt/vol fat-free milk and 0.1% vol/vol Tween-20 in PBS), followed by a 1-h incubation with the antibodies, which were diluted 1:1,000 in blocking buffer. Horseradish peroxidase–conjugated anti–mouse IgG or anti–rabbit IgG (GE Healthcare) were used as secondary antibodies at a dilution of 1:2,000. Bound antibody was detected using the enhanced chemiluminescence kit (GE Healthcare) and visualized using Hyperfilm MP (GE Healthcare).

RNA was extracted from macrophages using the Blood RNA extraction kit (QIAGEN), and PCR reactions were performed and measured on a Corbett Rotor-Gene 6000 (QIAGEN) using the Superscript III platinum one-step RT-PCR kit (Invitrogen) and Assay-On-Demand premixed Taqman probe master mixes(Applied Biosystems). The relative gene expression was calculated using the ΔΔCt method with the GAPDH gene for normalization of RNA levels. Primary transcript PCR was performed as follows: contaminating genomic DNA was removed from RNA samples using TURBO DNA-free kit (Applied Biosystems). The procedure routinely removed DNA contamination to levels below the detection limits of a standard real-time PCR. Reverse transcription of total RNA was performed at 42°C for 2 h with reagents purchased from Promega, followed by heat inactivation at 65°C for 10 min to inactivate the enzyme. cDNA was then subjected to real-time PCR analysis using SYBR Premix Ex Taq (Lonza). The primers used for measuring primary transcript and mature transcripts were 5′-GCAGTCAGATCATCTTCTCG-3′ and 5′-AGGTACAGGCCCTCTGATGGCAC-3′ and 5′-CCTGCTGCACTTTGGAGTGATCGG-3′ and 5′-GTACAGGCCCTCTGATGGCACCACC-3′, respectively.

Macrophages were plated in 10-cm dishes for culture at 8.5 × 10⁵ cells per dish. After stimulation, the cells were fixed for 10 min with formaldehyde (37% stock solution; Sigma-Aldrich) at a concentration of 1% and quenched using 125 mM Tris, pH 7.5. The nuclei were sonicated using conditions optimised for primary human macrophages. Extracts were precleared for 2 h once with 60 µl of a 50% suspension of salmon sperm-saturated protein G (GE Healthcare) and again with another 60 µl of protein G and 2 µg ChIP-grade rabbit isotype control antibody (Abcam). Immunoprecipitations were performed at 4°C overnight. Immunocomplexes were then collected with protein G Sepharose beads for 30 min, rigorously washed, and eluted. Protein DNA cross-links were reversed by heat at 65°C for 4 h, and DNA was purified using the QIAquick PCR Purification kit (QIAGEN) and subjected to real-time PCR analysis using SYBR Premix Ex Taq. Primer pairs used for ChIP in this study were optimised and shown to amplify with similar efficiencies.

After stimulation, the amount of GFP fluorescence was measured using a FLUOstar Omega plate reader (BMG Labtech) for normalizing the levels of viral infection. Luciferase activity was measured with a Labsystem luminometer after the addition of 30 µl luciferin (Bright-Glo luciferase assay system; Promega) per well. Levels of luciferase were then normalised to the amounts of GFP.

Macrophages were transfected with siRNA (Thermo Fisher Scientific) targeted to either RelA (RelA) or a nonspecific scrambled control (siControl), using methods previously devised and published (Behmoaras et al., 2008).

The concentration of TNF and IL-6 in cell culture supernatants was determined by ELISA (BD), according to the manufacturer’s instructions. Absorbance was read and analyzed at 450 nm on a spectrophotometric ELISA plate reader (Labsystems Multiskan Biochromatic) using Ascent software (version 2.4.2).

Fig. S1 shows that IL-10 inhibits transcription elongation at TNF. Fig. S2 shows that IL-10 requires elements 3′ and downstream of the TNF 3′UTR for inhibition. Fig. S3 shows that IL-10 does not inhibit NF-κB activation. Fig. S4 shows that IL-10 inhibits the recruitment of RelA to proinflammatory genes.

This paper is dedicated to the late Professor Brian Foxwell, who was the driving force behind this work.

We are grateful to Matthew Pierce who critically read this manuscript and we thank Rachel Simmonds for editorial help.

This work was supported by the Arthritis Research Campaign, the Kennedy Institute Trustees, and a Medical Research Council New Investigators Award.

The authors have no conflicting financial interests.