Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2018 Oct 1.
Published in final edited form as: Nat Chem Biol. 2017 Aug 14;13(10):1102–1108. doi: 10.1038/nchembio.2458

Small-molecule studies identify CDK8 as a regulator of IL-10 in myeloid cells

Liv Johannessen 1,2,13, Thomas B Sundberg 3,13, Daniel J O’Connell 4, Raivo Kolde 5, James Berstler 3, Katelyn J Billings 6,7, Bernard Khor 5, Brinton Seashore-Ludlow 7, Anne Fassl 2,8, Caitlin N Russell 9, Isabel J Latorre 4, Baishan Jiang 1,2, Daniel B Graham 4,10, Jose R Perez 3, Piotr Sicinski 2,8, Andrew J Phillips 3, Stuart L Schreiber 7,11,12, Nathanael S Gray 1,2,*, Alykhan F Shamji 7,*, Ramnik J Xavier 4,5,9,*
PMCID: PMC5693369  NIHMSID: NIHMS918591  PMID: 28805801

Abstract

Enhancing production of the anti-inflammatory cytokine interleukin-10 (IL-10) is a promising strategy to suppress pathogenic inflammation. To identify new mechanisms regulating IL-10 production, we conducted a phenotypic screen for small molecules that enhance IL-10 secretion from activated dendritic cells. Mechanism-of-action studies using a prioritized hit from the screen, BRD6989, identified the Mediator-associated kinase CDK8, and its paralog CDK19, as negative regulators of IL-10 production during innate immune activation. The ability of BRD6989 to upregulate IL-10 is recapitulated by multiple, structurally differentiated CDK8 and CDK19 inhibitors and requires an intact cyclin C–CDK8 complex. Using a highly parallel pathway reporter assay, we identified a role for enhanced AP-1 activity in IL-10 potentiation following CDK8 and CDK19 inhibition, an effect associated with reduced phosphorylation of a negative regulatory site on c-Jun. These findings identify a function for CDK8 and CDK19 in regulating innate immune activation and suggest that these kinases may warrant consideration as therapeutic targets for inflammatory disorders.

The anti-inflammatory cytokine IL-10 promotes immune homeostasis via multiple mechanisms including suppressing inflammatory cytokine production by innate immune cells and by promoting regulatory T cell (Treg) function1. Genetic studies of inflammatory bowel disease (IBD) have linked single-nucleotide polymorphisms near IL10 to adult-onset IBD and identified rare, loss-of-function mutations in IL10 or its receptor that result in severe, pediatric-onset enterocolitis2. Conversely, IL-10-based therapy reduces disease activity in murine models of colitis, bacterial infection, psoriasis and arthritis3. In addition, oral administration of bacteria engineered to express IL-10 and injection of a fusion protein in which recombinant IL-10 (rIL-10) is coupled to an inflammation-targeting antibody have shown promise in clinical trials for Crohn’s disease and rheumatoid arthritis, respectively4,5. Hence, while impaired function of IL-10 or its receptor can lead to inflammation, augmenting IL-10 abundance or signaling may be viable therapeutic strategies to restore and maintain immune homeostasis.

Small-molecule probes have provided key insights into regulation of IL-10 in innate immune cells. Activators of cAMP-response element binding protein (CREB)-dependent transcription such as G-protein-coupled receptor agonists or inhibitors of phosphodiesterases (PDEs), salt-inducible kinases (SIKs), or glycogen synthase kinase-3β (GSK-3β) potentiate IL-10 production in macrophages and dendritic cells6–9. Alternatively, amplification of mitogen-activated-protein (MAP) kinase signaling by perturbation of microtubule dynamics or inhibition of cyclin-dependent kinase 5 (CDK5) can also upregulate IL-10 (refs. 10,11). Notably, these probes have identified therapeutically useful drug targets, exemplified by approval of the PDE4 inhibitor apremilast for treatment of psoriatic arthritis as well as preclinical evaluation for treatment of IBD12,13.

Based on these successes, we conducted an unbiased phenotypic screen for small-molecule enhancers of IL-10 production with the goal of identifying new mechanisms of IL-10 regulation and, potentially, new targets for development of anti-inflammatory therapies. Subsequent mechanism-of-action (MoA) studies identified the transcriptional regulatory kinase CDK8 and its paralog CDK19 as molecular targets for the prioritized screening hit BRD6989 (1). Along with BRD6989, multiple structurally differentiated CDK8 and CDK19 inhibitors also upregulated IL-10 in activated human and mouse dendritic cells, and this effect required an intact cyclin C–CDK8 complex. Finally, using a multiplexed pathway reporter assay, we identified a role for enhanced activator protein 1 (AP-1) transcriptional activity in IL-10 potentiation following inhibition of CDK8 and CDK19, and demonstrated that this effect is associated with reduced phosphorylation of a negative regulatory site on c-Jun. Together, these findings suggest that inhibitory phosphorylation of c-Jun by CDK8 is a previously uncharacterized mechanism regulating IL-10 production during innate immune activation.

RESULTS

Identification of small molecules that upregulate IL-10

To identify new mechanisms of IL-10 regulation, we screened 59,346 small molecules for the ability to enhance IL-10 production by activated bone-marrow-derived dendritic cells (BMDCs) using a high-throughput assay that we developed previously9. Compounds were derived from both commercial collections and structurally and skeletally diverse libraries prepared by diversity-oriented synthesis (DOS). We identified >60 hit compounds that reproducibly enhance IL-10 production in a concentration-dependent manner (Supplementary Results, Supplementary Table 1). We prioritized three distinct chemotypes for follow-up studies on the basis of potency and an initial analysis of structure–activity relationship (SAR) with respect to core scaffold, stereochemistry and side chains (Fig. 1a). Among these, the DOS-derived hits BRD10330 (2) and BRD70326 (3) were deprioritized on the basis of data showing that they stimulated production of the inflammatory cytokine IL-1β and suppressed microtubule polymerization, respectively (Supplementary Figs. 1 and 2).

Figure 1. BRD6989 prioritized from phenotypic screen for small-molecule enhancers of IL-10 production.

Figure 1

Open in a new tab

(a) Chemical structures of BRD6989 (1) and the deprioritized screening hits BRD10330 (2) and BRD70326 (3) with the rationale for their deprioritization. (b,c) Effect of BRD6989 on cytokine production in BMDCs stimulated with zymosan A (b) or R848 (c). Each point corresponds to mean ± s.d.; n = 3 biological replicates from one independent experiment; Data represent three independent experiments. Vehicle corresponds to DMSO (≤0.5%). (d) Effect of BRD6989 on IL-10 production in human DCs stimulated with R848. Each point corresponds to mean ± s.d.; n = 3 biological replicates from one independent experiment using cells derived from one donor; Data represent two independent experiments (see Supplementary Fig. 3).

Prioritized hit BRD6989 selectively upregulates IL-10

The pyridinyl tetrahydroquinoline BRD6989 emerged as a prioritized hit in our screen (Fig. 1a and Supplementary Table 2). Pretreatment of BMDCs with BRD6989 for 48 h followed by stimulation with the yeast cell wall extract zymosan A for 18 h increased IL-10 production with an effector concentration for half-maximum response (EC50) ~1 μM (Fig. 1b), while only modestly reducing cell viability in these assay conditions (Supplementary Fig. 3). In addition, BRD6989 suppressed zymosan A–induced release of the inflammatory cytokine IL-6, but left production of TNFα, IL-12p40 and IL-1β largely unchanged (Fig. 1c and Supplementary Fig. 3). BRD6989 induced similar cytokine responses in BMDCs stimulated with the viral RNA mimetic R848 but with a greater fold-increase in IL-10 production (Fig. 1d and Supplementary Fig. 3). In addition to BMDCs, BRD6989 increased IL-10 production in mouse bone-marrow-derived macrophages (BMDMs) activated with zymosan A or R848 (Supplementary Fig. 3). Lastly, BRD6989 upregulated IL-10 following R848 stimulation in human, monocyte-derived DCs from two independent donors at concentrations consistent with its activity in BMDCs (Fig. 1d and Supplementary Fig. 3). Thus, BRD6989 enhances IL-10 production in activated human and murine macrophages and dendritic cells by a mechanism that appears to be particularly pronounced following stimulation of toll-like receptor-7 (TLR7) and TLR8 by R848.

Analyzing the effects of BRD6989 pretreatment on Il10 mRNA levels after R848 stimulation revealed a sustained increase beginning at 4 h and extending to 24 h (Supplementary Fig. 3). These Il10 transcript dynamics contrast with activators of CREB-dependent transcription such as the pan-SIK (salt-inducible kinase) inhibitor HG-9-91-01 (4)8, which elevated Il10 mRNA during the first 2 h following microbial stimulation before returning to levels similar to DMSO-treated cells (Supplementary Fig. 3). Based on its ability to selectively potentiate IL-10 production by a mechanism that appears temporally distinct from CREB activation, we prioritized BRD6989 for MoA studies.

Mediator-associated kinase CDK8 is a target of BRD6989

Given that the pyridine core of BRD6989 is a common kinase-binding motif14, we profiled its effects on 414 kinases using a combination of binding and activity assays. We found that BRD6989 binds a complex of cyclin C–CDK8, a Mediator-associated kinase not previously linked to IL-10 regulation15, with a half-maximal inhibitory concentration (IC50) ~200 nM and remarkable selectivity (Fig. 2a and Supplementary Fig. 4 and Supplementary Data Set 1). Binding of BRD6989 to CDK8 and its paralog CDK19 was confirmed using an orthogonal kinase-profiling format (Supplementary Data Set 2). In addition to CDK8, both kinase profiling experiments identified phosphatidylinositol-4,5-biphosphate 3-kinase C2A (PI3KC2A) and PI3KCG as secondary targets of BRD6989, suggesting an overall consistency between these approaches. Of note, PI3K inhibitors with varying isoform specificities screened in our assay system fail to enhance IL-10 production9, suggesting that inhibition of PI3Ks does not contribute to IL-10 potentiation by BRD6989.

Figure 2. CDK8 is a molecular target of BRD6989.

Figure 2

Open in a new tab

(a) Kinase profiling identifies cyclin C–CDK8 as a molecular target of BRD6989. Each point corresponds to the mean of three biological replicates from one independent experiment normalized to a DMSO (≤0.5%) control (see Supplementary Data Set 1). (b) Effect of the indicated concentrations of BRD6989 on the activity of cyclin C–CDK8 or cyclin C–CDK19 complexes. Each point corresponds to mean ± s.d.; n = 3 biological replicates from one independent experiment for cyclin C–CDK8 or n = 2 biological replicates from one independent experiment for cyclin C–CDK19 normalized to a DMSO (1%) control. (c) BRD6989 suppresses phosphorylation of the STAT1 transactivation domain at Ser727 in IFNγ-stimulated BMDCs. Data represent three independent experiments; full gel images are presented in Supplementary Figure 11. (d) CDK8 binding affinity correlates with the potency of IL-10 upregulation for BRD6989 and its derivatives. R2 = 0.91 for linear regression; IC50 and EC50 data are from one independent experiment and have been transformed to a negative log10 (p) scale (see Supplementary Fig. 5).

In agreement with the kinase profiling results, BRD6989 inhibits the kinase activity of recombinant cyclin C–CDK8 or cyclin C–CDK19 complexes with IC50s ~0.5 μM and >30 μM, respectively (Fig. 2b), but not the activity of several CDKs involved in cell cycle regulation (Supplementary Fig. 4). Furthermore, pre-incubation of BMDCs with BRD6989 inhibited IFNγ-induced phosphorylation of signal transducer and activator of transcription 1 (STAT1) at the known CDK8-regulated position Ser727 (ref. 16), but did not affect Janus kinase (JAK)-mediated phosphorylation of Tyr701 (Fig. 2c). Preliminary SAR analysis of BRD6989’s effects on CDK8 and IL-10 suggest that the pyridine and amino substituents and the methyl cyclohexyl core of BRD6989 make critical contacts with CDK8 (Supplementary Fig. 5). Notably, the potency of CDK8 binding correlates with that of IL-10 induction for these analogs (Fig. 2d), whereas BRD6989 only inhibits CDK19 at concentrations much greater than the EC50 for IL-10 induction. Together, these kinase profiling, binding assay, enzyme inhibition and cell-based data identify the Mediator-associated kinases CDK8 and, to a lesser degree, CDK19 as molecular targets of BRD6989.

Structurally distinct CDK8 inhibitors upregulate IL-10

Burgeoning interest in Mediator-associated kinases as chemotherapeutic targets has prompted development of several structurally distinct CDK8 and CDK19 inhibitors including Δ16-cortistatin A (dCA; 5), an analog of the potent CDK8- and CDK19-selective natural product cortistatin A17,18, and the recently published highly selective CDK8 and CDK19 inhibitor CCT251921 (6)19 (Fig. 3a). Pretreatment with these CDK8 and CDK19 inhibitors recapitulated the IL-10 potentiating activity of BRD6989 in both BMDCs (Fig. 3b) and human DCs derived from two independent donors with R848 stimulation (Fig. 3c and Supplementary Fig. 6). Notably, the IL-10 potentiating activity of these three inhibitors occurs at concentrations consistent with their potencies for CDK8 and CDK19 inhibition. By contrast, IL-10 production is not enhanced by the CDK7-targeting inhibitor THZ-1 (ref. 20) (7) (Supplementary Fig. 6), suggesting that this effect is not a nonspecific consequence of inhibiting transcriptional regulatory kinases. Lastly, the effect of CDK8 inhibition on IL-10 production may be cell-type-specific, because neither dCA nor BRD6989 dramatically increased IL-10 production during ex vivo differentiation of Tregs (Supplementary Fig. 6). Together, these data support CDK8 inhibition as the mechanism driving IL-10 potentiation in response to BRD6989 in activated human and murine DCs.

Figure 3. Pharmacological and genetic data identify CDK8 as a negative regulator of IL-10 production.

Figure 3

Open in a new tab

(a) Chemical structures of the potent, specific CDK8 and CDK19 inhibitors Δ16-cortistatin A (dCA; 5) and CCT251921 (6). (b) CCT251921 and dCA recapitulate the IL-10-enhancing activity of BRD6989 in R848-stimulated BMDCs. Each point corresponds to mean ± s.d.; n = 3 biological replicates from one independent experiment; data represent three independent experiments. (c) CCT251921 and dCA recapitulate IL-10 potentiation by BRD6989 in human DCs stimulated with R848. Each point corresponds to mean ± s.d.; n = 3 biological replicates from one independent experiment using cells derived from one donor; data represent two independent experiments (see Supplementary Fig. 6). (d,e) Deletion of cyclin C in BMDCs derived from CcncΔ/Δ mice impairs induction of IL-10 and suppression of IL-6 by dCA (d) or BRD6989 (e) following activation with R848. Each point corresponds to mean ± s.d.; n = 4 biological replicates from one independent pair of CcncΔ/Δ or Ccncf/f mice; data represent three independent experiments (see Supplementary Fig. 7).

IL-10 elevation requires intact cyclin C–CDK8 complexes

Next, to determine whether IL-10 potentiation by CDK8 inhibitors requires functional cyclin C–CDK8 complexes, we differentiated between BMDCs from CcncΔ/Δ mice21, in which the floxed Ccncfl/fl allele had been specifically disrupted in hematopoietic cells by induction of Mx1-Cre following stimulation with polyinosinic–polycytidylic acid (poly(I:C)) (Supplementary Fig. 7). As a control, BMDCs were differentiated from poly(I:C)-treated littermates expressing the Ccncfl/fl allele, but lacking Mx1-Cre. Notably, there was not a statistically significant difference in the levels of IL-10 secreted between CcncΔ/Δ and Ccncf/f following R848 stimulation (Supplementary Fig. 7), suggesting that other regulatory mechanisms can compensate for a sustained loss of CDK8 activity and maintain low basal levels of IL-10 production. However, CcncΔ/Δ BMDCs failed to upregulate IL-10 in response to treatment with dCA or BRD6989, whereas the response remained intact in Ccncf/f BMDCs (Fig. 3d,e and Supplementary Fig. 7). In addition, CcncΔ/Δ BMDCs displayed a ~50% reduction in IL-6 production relative to Ccncf/f BMDCs, and saturating concentrations of dCA or BRD6989 caused no further reduction (Fig. 3d,e). Cyclin C deletion does not appear to inhibit cytokine production nonspecifically, as it does not affect IL-10 and IL-6 responses induced by the pan-SIK inhibitor HG-9-91-01 (Supplementary Fig. 7). The impaired responses of BMDCs lacking cyclin C to treatment with BRD6989 and dCA are consistent with a model in which CDK8 restrains IL-10 production during activation of wild-type myeloid cells.

CDK8 represses a small subset of genes in macrophages

To define systematically the role of CDK8 during innate immune activation, we profiled the transcriptional responses elicited by dCA or BRD6989 following stimulation of BMDMs with a panel of ten microbial ligands or Sendai virus for time points from 0.25 to 4 h. BMDMs were chosen for these experiments because, like human and mouse DCs, they upregulate IL-10 in response to CDK8 inhibition (Supplementary Fig. 3), and they have been previously studied using the multiplexed pathway reporter assay described below22. In this experiment, the transcriptional responses elicited by BRD6989 and dCA were highly correlated (Supplementary Fig. 8), again supporting CDK8 as a principle cellular target of BRD6989. For instance, both of the CDK8 inhibitors reduced expression of the interferon-inducible genes Ifit2, Cxcl10 and Rsad2 (Supplementary Fig. 8). Given that CDK8-dependent phosphorylation of STAT1 at S727 is required for expression of many IFNγ-inducible genes16, these suppressive effects suggest that CDK8 inhibition may limit autocrine interferon signaling following microbial stimulation.

In addition to suppressing interferon-inducible genes, BRD6989 and dCA increased abundance of a similar, small subset of transcripts in both quiescent and activated BMDMs (Supplementary Fig. 8). Consistent with the effects of CDK8 inhibition on cytokine responses in activated myeloid cells, the genes induced include chemokines, metallopeptidases and regulators of cellular second messengers linked to inflammatory responses and migration of innate immune cells (P = 0.0255 and P = 1.52 × 10−5, respectively) (Supplementary Data Set 3). In the context of acute myeloid leukemia (AML), CDK8 inhibition preferentially induces genes near ‘super enhancers’ (SEs), defined by dense binding of the Mediator complex and transcription factors along with accumulation of histone modifications associated with active transcription18. Acknowledging the limitations in comparing data sets from different myeloid cell types, we queried genes upregulated in BRD6989- or dCA-treated BMDMs for the presence of histone modifications linked to active transcription that were previously identified in lipopolysaccharide (LPS)-stimulated BMDCs23. In this analysis, monomethylated Lys-4 on histone 3 (H3K4me1), a mark associated with poised enhancers24,25, is enriched (P = 0.0093) near genes induced by CDK8 inhibition compared to all highly expressed genes (Supplementary Fig. 8). Although we detect a similar trend in regard to enrichment of histone 3 Lys27 acetylation (H3K27ac), a modification associated with active enhancers24,25, near genes that are induced by CDK8 inhibition, this association did not achieve statistical significance (Supplementary Fig. 8). These data indicate that CDK8 primarily represses a small set of genes in quiescent or activated macrophages, although the link between CDK8-regulated genes and histone modifications associated with active transcription appears to be less well correlated in this context.

Enhanced AP-1 activity links CDK8 inhibition to IL-10

To identify CDK8-responsive signaling pathways, we applied an integrative genomics approach termed transcription factor sequencing (TF-seq) in which consensus transcription factor binding sites drive expression of 58 reporter constructs bearing unique sequence tags22. Consistent with the established role of CDK8 in STAT1 activation16, BRD6989 and dCA both suppressed induction of STAT1–STAT2 activity ~2 h after stimulation of BMDMs (Fig. 4a; Supplementary Fig. 9 and Supplementary Data Set 4). Furthermore, similar to results from previous studies of CDK8 function in other contexts including RPMI8226 myeloma cells activated by TLR9 stimulation26–28, pretreatment with dCA or BRD6989 suppressed NF-κB activation to a varying degree for all stimuli tested (Fig. 4a and Supplementary Fig. 9). Because dCA pretreatment does not perturb upstream signaling events that link sensing of LPS or R848 to NF-κB activation (Supplementary Fig. 9), it appears that inhibiting CDK8 may suppress the ability of nuclear NF-κB to activate transcription.

Figure 4. Modulation of c-Jun–AP-1 links CDK8 inhibition to enhanced IL-10 production.

Figure 4

Open in a new tab

(a) CDK8-responsive signaling pathways in activated BMDMs as identified by the highly parallel reporter gene assay TF-seq. Reporter data are integrated for time points between 0.25 and 4 h from one independent experiment (See Online Methods and Supplementary Data Set 4). Dashed lines correspond to false discovery rate (FDR) < 0.05 as corrected by Benjamini–Hochberg testing. no stim, no stimulation; FLA, flagellin A; LPS, lipopolysaccharide; MDP, muramyldipeptide; CpG, synthetic CpG-containing oligonucleotide; Pam3cys, synthetic triacylated lipoprotein; poly(I:C), polyinosinic–polycytidylic acid; SeV, Sendai virus; TDB, trehalose-6,6-dibehenate. (b) In comparison to all highly expressed genes, JunB binding sites are enriched among the subset of genes induced by CDK8 inhibition (see Online Methods). (c) Pretreatment with dCA (100 nM; 18 h) suppresses phosphorylation (p) of c-Jun on Ser243 in BMDCs activated with R848 for the indicated times. Data represent two independent experiments; full images are presented in Supplementary Figure 12. (d) The indicated concentrations of dCA suppress phosphorylation of recombinant c-Jun at Ser243 by recombinant cyclin C–CDK8 in an in vitro kinase reaction. Data represent two independent experiments; full images are presented in Supplementary Figure 12. (e,f) Co-treatment of BMDCs with the c-Fos–AP-1 inhibitor T-5224 (100 μM) suppresses the IL-10 enhancing activity of dCA in BMDCs stimulated with R848 (e) or zymosan A (f) relative to DMSO (≤0.5%) control. Each point correspond to mean ± s.d.; n = 3 biological replicates from one independent experiment; data represent three independent experiments.

TF-seq results also indicate that CDK8 inhibition increases AP-1 activity in BMDMs (Fig. 4a and Supplementary Fig. 10). In support of this observation, binding sites for the AP-1 subunit JunB are enriched among genes upregulated following CDK8 inhibition compared to to all highly expressed genes (Fig. 4b). Given that c-Jun’s ability to activate transcription is tightly regulated by phosphorylation, we tested whether CDK8 inhibition affects c-Jun phosphorylation in activated BMDCs. We found that pretreatment with dCA only moderately delayed the activating phosphorylation of c-Jun on Ser63, an effect that is rapidly induced by R848 stimulation and peaks after 30 min (Fig. 4c). In contrast, inhibiting CDK8 with dCA dramatically reduced phosphorylation of Ser243, a suppressive mark shown to destabilize c-Jun and interfere with its ability to bind DNA29–31, which is first observed 60 min after R848 stimulation (Fig. 4c). Notably, dCA induces concentration-dependent suppression of c-Jun Ser243 phosphorylation both in R848-stimulated BMDCs (Supplementary Fig. 10) and an in vitro kinase reaction with recombinant cyclin C–CDK8 and c-Jun (Fig. 4d). Together, these data suggest that c-Jun Ser243 is a direct CDK8 substrate that regulates AP-1 activity in activated myeloid cells.

Given the presence of an AP-1 consensus motif in Il10 and the evidence that potentiation of MAP kinase signaling can increase IL-10 production1,10,11, we hypothesized that an increase in AP-1 activity resulting from reduced c-Jun Ser243 phosphorylation may mediate activation of IL-10 following CDK8 inhibition. Supporting this theory, we found that co-treatment of BMDCs with dCA and T-5224 (8), a small-molecule inhibitor of AP-1 reported to bind c-Fos and inhibit its dimerization with c-Jun32 (Supplementary Fig. 10), suppressed IL-10 production following stimulation with R848 or zymosan A (Fig. 4e,f). Although the concentration of T-5224 (100 μM) that suppressed IL-10 potentiation by dCA is consistent with concentrations needed to prevent c-Jun binding to AP-1 elements32, it is possible that the decrease in IL-10 induction could be due to nonspecific effects of T-5224 on myeloid cell function. However, the same concentration of T-5224 did not: 1) Affect the ability of BMDCs to increase in IL-10 production in response to potentiation of CREB signaling by the pan-SIK inhibitor HG-9-91-01; or, 2) Suppress production of IL-6 or TNFα by R848-stimulated BMDCs (Supplementary Fig. 10). These data suggest that T-5224 does not nonspecifically impair the ability of BMDCs to respond to microbial stimulation. Together with the increase in AP-1 transcriptional activity and reduced c-Jun Ser243 phosphorylation, the inhibitory effect of T-5224 suggests that CDK8 inhibition upregulates IL-10 production by a c-Jun–AP-1-dependent mechanism that is distinct from cAMP and CREB signaling.

DISCUSSION

Using unbiased phenotypic screening and MoA studies, we have identified the Mediator-associated kinase CDK8 and, likely, its paralog CDK19 as regulators of IL-10 production, a mechanism that may be harnessed to enhance anti-inflammatory functions of innate immune cells. Our initial screen uncovered three new structural classes of small molecules that enhance IL-10 production including BRD6989, a probe found to bind to CDK8 through kinase profiling. The growing interest in Mediator-associated kinases as therapeutic targets in cancer has spurred development of several chemically distinct CDK8 and CDK19 inhibitors such as the potent, dual inhibitor CCT251921 (ref. 19), which shares a pyridinyl tetrahydroquinoline core with BRD6989, and a close analog of the natural product cortistatin A (dCA)17,27. Though not as potent as these inhibitors, BRD6989 appears to be unique in its ability to differentially inhibit CDK8 relative to its paralog CDK19, suggesting that it may inform further development of CDK8-specific inhibitors. Importantly, we found that both CCT251921 and dCA, similarly to BRD6989, promote IL-10 production in R848-stimulated human and mouse DCs in a manner that depends on an intact cyclin C–CDK8 complex. Together, these data identify CDK8 as a druggable regulator of IL-10 production in activated myeloid cells.

In comparison to CDK7 and CDK9, which broadly affect transcriptional initiation by activating phosphorylation of the C-terminal domain of RNA polymerase II20,33, CDK8’s function appears to be more context dependent, having been linked to specific effects on interferon, TGF-β, Wnt and Notch signaling15. For example, phosphorylation of the transactivation domain of STAT1 by CDK8 is required for expression of many interferon-inducible genes16, and cortistatin A specifically elevates expression of SE-associated genes in a subset of AML lines18. Similarly, our transcriptional profiling data in BRD6989- or dCA-treated BMDMs suggest that CDK8’s kinase function primarily restrains expression of a limited number of genes in myeloid cells. In contrast to the AML context, our preliminary analysis suggests that genes regulated by CDK8 in myeloid cells are not correlated with regions of active transcription (i.e., SE-associated regions). However, it will be interesting to define experimentally how inhibiting CDK8 activity alters chromatin organization (for example, H3K27ac localization) in myeloid cells.

Measuring the response to CDK8 inhibition using a multiplex pathway reporter assay (TF-seq) provided insights into CDK8-sensitive signaling pathways in activated myeloid cells. Consistent with previous studies implicating CDK8 in IFNγ signaling16, we observed that inhibition of CDK8 reduced activation of STAT1–STAT2 and triggered a corresponding decrease in expression of several interferon-inducible genes. We also detected decreased NF-κB activity to varying degrees using different microbial stimuli that appears to be insufficient in magnitude to suppress expression of NF-κB targets on a genome-wide scale. This intermediate suppressive effect may explain why CDK8 inhibition does not block expression of Il10, itself an NF-κB target gene28,34. Our data with dCA and BRD6989 contrasts with the recent identification of an essential role for CDK8 and CDK19 in NF-κB-dependent Il10 expression following stimulation of TLR9 in the B cell–derived myeloma cell line RPMI8226 (ref. 28). Whereas differences in cell type and microbial stimuli may alter the degree to which NF-κB activation depends on CDK8 and CDK19, contrasting effects of depleting CDK8 and CDK19 by RNA interference compared to specifically inhibiting their kinase activity with small molecules may also play a role.

CDK8 inhibition also increased AP-1 reporter activity in the TF-seq assay for nearly all microbial stimuli tested, and binding sites for the AP-1 subunit JunB are enriched near genes upregulated in response to CDK8 inhibition. Enhanced AP-1 activity appears to be important for potentiation of IL-10 following CDK8 inhibition, as the c-Fos-targeting inhibitor T-5224 suppresses this response. In contrast, disrupting AP-1 activity does not impair the ability of BMDCs to increase IL-10 production through a parallel, CREB-dependent pathway triggered by the pan-SIK inhibitor HG-9-91-01. Reduced phosphorylation of the negative regulatory site Ser243 on c-Jun may be central to the mechanism linking CDK8 inhibition to enhanced AP-1 transcriptional activity and IL-10 production in myeloid cells. Importantly, CDK8 appears to directly phosphorylate Ser243 in a cell-free biochemical assay and is inhibited by concentrations of dCA that stimulate IL-10 production in cells. Our data identifying CDK8 as a negative regulator of c-Jun–AP-1 in activated myeloid cells contrasts with the role of CDK8 in promoting AP-1 activity during serum stimulation of prostate cancer cells35. As such, although CDK8 appears to regulate c-Jun Ser243 phosphorylation following microbial stimulation of postmitotic BMDCs, other kinases targeting this site (for example, casein kinase-2 and Dyrk2a29,31) may play this role in cycling cancer cells.

Our data identify the CDK8–c-Jun module as a regulator of IL-10 production that is distinct from previously described pathways such as cAMP–CREB signaling. It remains to be determined whether using CDK8 inhibitors to upregulate IL-10 will be an efficacious and tolerated treatment for inflammatory disorders. The clinical observations that targeted delivery of IL-10 to inflamed tissues can suppress inflammation suggest that recapitulating this strategy with small molecules may hold promise4,5. Notably, CDK8 selectively upregulates IL-10 production by activated myeloid cells, suggesting its immunomodulatory activity could be localized to sites of inflammation, circumventing dose-limiting toxicities associated with systemic delivery of IL-10 (refs. 36,37). In addition, given that increased IL-10 expression and other mechanisms that suppress antitumor immune responses have well-established links to tumor progression and resistance to chemotherapy38–40, the anti-inflammatory effects of CDK8 and CDK19 inhibition described here warrant consideration in the context of targeting these kinases in cancer.

Given that CDK8 and CDK19 are fundamental regulators of transcription in numerous cell types, it is perhaps not surprising that small-molecule inhibitors of these Mediator-associated kinases are poorly tolerated in vivo at doses that achieve the sustained target occupancy required for antitumor activity41. Although these toxicities may preclude therapeutic targeting of CDK8 and CDK19, it is possible that the anti-inflammatory consequences of CDK8 and CDK19 inhibition can be achieved in vivo using less aggressive dosing regimens. Our data suggest that inhibition of CDK8 and CDK19 affects a limited number of pathways in myeloid cells including several targeted by clinically approved inhibitors of Janus kinases and the proteasome, respectively42,43. Thorough investigation of potential toxicities associated with immunomodulatory doses of CDK8 and CDK19 inhibitors in vivo will be a critical next step for evaluating the potential of developing therapeutics targeting these Mediator-associated kinases for treatment of inflammatory disorders.

METHODS

Methods, including statements of data availability and any associated accession codes and references, are available in the online version of the paper.

ONLINE METHODS

Reagents

Unless otherwise indicated, compounds were obtained from the Broad Institute Compound Management Group including (6R,7R,8S)-7-(4-bromophenyl)-8-(hydroxymethyl)-1,4-diazabicyclo[4.2.0]octan-2-one (21), a core scaffold used to prepare analogs of BRD10330 (Supplementary Note). Prostaglandin E2 (PGE2) and zymosan A from Saccharomyces cerevisiae were purchased from Sigma. BRD6989 (1; cat# STL241555) was purchased from Vitas M-Laboratory. BRD8408 (16; cat# 75329634) was purchased from ChemBridge. BRD4064 (17; cat# F9994-0073) and BRD9548 (20; cat# F9994-0072) were purchased from Life Chemicals. BRD5817 (18; cat# 6233-1031) and BRD2299 (19; cat# 6233-0271) were purchased from ChemDiv. Ultrapure lipopolysaccharide (LPS) from Escherichia coli O111:B4. R848, flagellin A (FLA), muramyl dipeptide (MDP), synthetic triacylated lipoprotein (Pam3cys), polyinosinic–polycytidylic acid (Poly(I:C)), and trehalose-6,6-dibehenate (TDB) were purchased from InvivoGen. Sendai virus (Cantell) was purchased from ATCC. Recombinant murine IFNγ, GM-CSF and IL-4 were obtained from Peprotech. T-5224 was obtained from ApexBio. HG-9-91-01 and CCT251921 were prepared as described previously8,19. Δ16-cortistatin A (dCA) was a generous gift from P. Baran (The Scripps Research Institute). Small-molecule reagents were confirmed to have ≥95% purity by HPLC–MS.

Isolation and culture of murine macrophages and dendritic cells

Bone marrow was flushed from femurs and tibias of C57BL/6 mice, and preparation of bone-marrow-derived macrophages (BMDMs) and dendritic cells (BMDCs) was conducted as described previously44. Unless otherwise indicated, cells were treated for 24 h with BRD6989 (5 μM), dCA (0.1 μM), HG-9-91-01 (0.5 μM) or an equivalent concentration of DMSO (≤0.5%) and then stimulated for 18 h with R848 (2 μg/mL), LPS (100 ng/mL) or zymosan A (4 μg/mL).

Isolation and culture of human dendritic cells (DCs)

For culture of human DCs, CD14+ peripheral blood monocytes (PBMCs) were first enriched from buffy coats obtained from healthy volunteers (Research Blood Components) using RosetteSep Human Monocyte Enrichment Cocktail (cat# 15068) and SepMate 50 centrifugation tube (cat# 85450) according to the manufacturer’s protocol (StemCell Technologies). Isolated PBMCs were differentiated into DCs by culturing in X-VIVO10 (Lonza; cat #04-380Q) + FBS (10% v/v) and penicillin–streptomycin (1% v/v) along with GM-CSF (50 ng/mL; Peprotech; cat #300-03) and IL-4 (50 ng/mL; Peprotech; cat #200-04) for 7 d with additional media and growth factors added on day 4. On day 7, DCs were collected by scraping and plated in 96-well tissue culture plates at 30,000 cells/well before treatment with indicated concentrations of BRD6989, dCA or CCT251921 and an equivalent concentration of DMSO (≤0.5%), and then stimulated for 18 h with R848 (2 μg/mL). After stimulation, tissue culture media was collected and stored at –80 °C before cytokine detection.

High-throughput detection of IL-10 release and cell viability

High-throughput detection of viability and IL-10 release by zymosan A–stimulated BMDCs was conducted as described previously9.

Connectivity mapping of transcriptional profiling data

Changes in gene expression elicited by BRD0326 (7) in U-2 OS osteosarcoma cells have been previously characterized by the Luminex 1000 (L1000) assay45. The correlation between the transcriptional responses elicited by BRD0326 and microtubule modulators was identified by comparison with a library of >1.5 × 106 gene expression signatures using Broad Institute connectivity mapping database (https://www.broadinstitute.org/cmap/) as described previously46.

Tubulin polymerization assays

The effect of BRD0326 and related analogs on tubulin dynamics was determined using a fluorescence-based tubulin polymerization assay (Cytoskeleton) according to the manufacturer’s protocol.

Kinase profiling

BRD6989 (1 μM) was tested for the ability to bind or inhibit the activity of 414 kinases using either LanthaScreen Eu binding, Adapta or Z-LYTE assays (Invitrogen). Results are presented as percent displacement of dye-labeled probe from the protein kinase domain or remaining kinase activity relative to DMSO (≤0.5%) control (Supplementary Data Set 1). Alternatively, interaction of BRD6989 (10 μM) with a panel of 395 unique kinases was evaluated using the KINOMEscan active site–directed competitive binding assay (DiscoverX). Data are presented as percent displacement of a catalytic-site-directed probe relative to DMSO (≤0.5%) control (Supplementary Data Set 2).

Cyclic C–CDK8 binding assays

Binding of BRD6989 and related analogs to the human cyclin C–CDK8 complex was determined using a LanthaScreen binding assays (Life Technologies) and expressed as percent displacement of a dye-labeled probe from the CDK8 protein kinase domain.

CDK8 and CDK19 kinase activity assays

Effects of BRD6989 on CDK8 or CDK19 activity were quantified using a radiometric filter-binding assay (ProQinase). Briefly, purified recombinant cyclin C–CDK8 or cyclin C–CDK19 were incubated with the indicated dilution series of BRD6989 and changes in activity quantified in terms of ATP[γ-32P] transfer to the recombinant RBER-IRStide substrate using a radiometric filter binding assay relative to a DMSO (1%) control.

Generation of CnccΔ/Δ mice

Male Ccncf/f and CnccΔ/Δ (Ccncf/f × Mx1-Cre) mice were generated as described previously using a protocol approved by the Dana–Farber Cancer Institute Animal Care and Use Committee21. Briefly, 6–8 week old Ccncf/f and CnccΔ/Δ mice were injected with poly(I:C) (2.5 μg g−1 intraperitoneally) every other day for 3 d. Femurs and tibias were then harvested from mice sacrificed by CO2 inhalation overdose followed by cervical dislocation 10 d after the final injection. Recombination of the Cncc locus in genomic DNA from BMDCs derived from CnccΔ/Δ mice was confirmed using primers P1: 5°-GGGGTGATGGTGAAAACACTGA-3°; P2: 5°-CAAGCAAGTAATCCAGGACCA-3° and P3: 5°-CTGCACATACTGGAGCAAAGGTCT-3° and the following PCR protocol: 95 °C for 3 min, followed by 30 cycles of 95 °C for 1 min, 60 °C for 1 min, 72 °C for 1 min and a final elongation step at 72 C for 5 min.

CDK8–c-Jun kinase reaction

In vitro kinase reactions were carried out for 30 min at 30 °C in assay buffer (60 mM HEPES pH 7.5, 3 mM MgCl2, 3 mM MnCl2, 1.2 mM DTT) using 60 ng recombinant cyclin C–CDK8 (ProQinase; cat# 0376-0390-1), 60 ng recombinant human c-Jun (Abcam; cat# 54318), and 20 μM ATP. CDK8 and c-Jun were pre-incubated with the indicated dilution series of dCA for 15 min before addition of ATP. Reactions were quenched with 10 μL of 80 mM EDTA. To detect c-Jun, 15 μL of the quenched kinase reactions were separated by SDS–PAGE and visualized using the immunoblotting procedure described below.

Immunoblotting

Cell lysis and immunoblotting were conducted as described previously9. Briefly, single cell suspensions were rinsed in ice-cold PBS and extracted in lysis buffer (50 mM Tris–HCl at pH 7.4, 1 mM EDTA, 50 mM NaF, 10 mM sodium β-glycerol 1-phosphate, 1 mM DTT, 1 mM sodium orthovanadate, 1% (v/v) Triton X-100, 0.2% (w/v) SDS and 1× Complete EDTA-free Protease Inhibitor Mixture (Roche)). Cell extracts were clarified by centrifugation at 14,000 × g for 10 min at 4 °C, and protein concentrations were determined using the Bradford assay (Bio-Rad). To detect proteins in cell lysates, 10–20 μg of protein extract was separated by SDS–PAGE. After transfer to PVDF membranes, proteins were detected by immunoblotting and visualized by treating the blots with SuperSignal West Femto Chemiluminescent Substrate (Thermo Fisher Scientific) followed by autoradiography. The following antibodies were used for immunoblotting: β-actin (cat# A2228; 1/10,000 (v/v) dilution) was from Sigma-Aldrich; cyclin C (cat# 558903; 1/1000 (v/v) dilution) was from BD Biosciences; CDK8 antibody was obtained from Bethyl (cat# A302-501; 1/1,000 (v/v) dilution); total STAT1 (cat# 14994; 1/1,000 (v/v) dilution), p-STAT1 Tyr701 (cat# 9167; 1/1,000 (v/v) dilution), p-STAT1 Ser727 (cat# 9177; 1/1,000 (v/v) dilution), total IκBα (cat #4814; 1/1,000 (v/v) dilution), pIKKα/β S176/180 (cat# 2697; 1/1,000 (v/v) dilution), total IKKα (cat# 11930; 1/1,000 (v/v) dilution), total c-Jun (cat #9165; 1/1,000 (v/v) dilution), phospho-c-Jun Ser63 (cat# 9261; 1/1,000 (v/v) dilution) and phospho-c-Jun Ser243 (cat# 2994; 1/1000 (v/v) dilution) were from Cell Signaling Technologies.

mRNA isolation

mRNA was isolated from cultured cells using a Dynabeads mRNA DIRECT Kit (Thermo Fisher Scientific) according to the manufacturer’s protocol. Briefly, culture media was removed from BMDCs cultured in a 96-well tissue culture treated plate and cells washed with PBS before suspension in lysis buffer. Lysates were incubated with Oligo (dT)25 Dynabeads at room temperature for 5 min. Beads were magnetically separated from the lysate and washed 3× with wash buffer A and 1× with wash buffer B before being transferred to a fresh 96-well plate. mRNA was released by suspension of beads in dH2O and beads removed magnetically. cDNA was prepared from purified mRNA using an iScript cDNA Synthesis kit (Bio-Rad) and diluted 1/10 in dH2O.

qPCR

Diluted cDNA was quantified by qPCR using the iQ SYBR Green Supermix (Bio-Rad) on a CFX96 real-time system (Bio-Rad). The relative abundance of each mRNA was calculated from Ct values and normalized relative to the abundance of B2m mRNA using the 2−ΔΔCt method47. The primers used are as follows:

  • Il10-F, 5′-GCTCTTACTGACTGGCATGAG-3′;

  • Il10-R, 5′-CGCAGCTCTAGGAGCATGTG-3′;

  • B2m-F, 5′-TTCTGGTGCTTGTCTCACTGA-3′;

  • B2m-R, 5′-CAGTATGTTCGGCTTCCCATTC-3′;

Cytokine detection

Release of IL-10 from activated BMDMs and BMDCs or human DCs was quantified using BD ELISA kits (cat #555252 and cat #555157, respectively) according to the manufacturer’s protocol. Alternatively, culture medium from stimulated BMDCs was removed and clarified by centrifugation for 5 min at 14,000 × g. Levels of TNFα, IL-6, IL-10, or IL-1β in culture medium were quantified using FlexSet Cytokine Bead Array (BD Biosciences) according to the manufacturer’s protocol.

Simultaneous signaling pathway activity inference and global gene expression analysis

Preparation of the TF-Seq multiplex reporter library from BMDMs was conducted essentially as described previously22. Briefly, BMDMs were differentiated in pen–strep containing DMEM plus 30% FBS and 10% L929 M-CSF conditioned medium (CMG). BMDMs were transduced with an equimolar pool of lentiviral sequencing-based gene reporters (TF-seq) in 90% concentrated lentiviral supernatant + 10% CMG on day 4. On day 7 BMDMs were frozen in 90% FBS and 10% DMSO and stored in liquid nitrogen. BMDMs were thawed in DMEM plus 30% FBS and 10% L929 M-CSF for 2 d before they were harvested from petri dishes (BD Falcon), counted on a hemocytometer, and plated overnight in DMEM + 10% FBS at 100,000 cells/well in 96-well tissue-culture-treated plates (Corning). BMDMs were pretreated with dCA (0.1 μM) or BRD6989 (5 μM) or an equivalent concentration of DMSO (≤0.5%) for 24 h at 37 °C before stimulation with a panel of innate immune ligands (final concentrations: R848 (10 μg/mL); LPS (100 ng/mL) or zymosan A (5 μg/mL). Pam3Cys (300 ng/mL); poly(I:C) (20 μg/mL); CpG (5 μM); FLA (5 μg/mL); MDP (10 μg/mL); TDB (20 μg/mL); or SeV (10 MOI)) in a short 3-h time series. After defined periods of stimulation, cells were lysed in RLT lysis buffer (Qiagen) and stored at –80 °C. Total RNA was precipitated in 10% PEG 8000 and 1.5 M NaCl, and then washed once in 10% PEG 8000 and 1.5 M NaCl. Purified total RNA was then then resuspended in 50 μL with 2 units TURBO DNase (Life Technologies) and incubated at 37 °C for 30 min. The DNase was denatured by adding 75 μL of RLT lysis buffer and the remaining RNA re-precipitated and washed twice in 10% PEG 8000 and 1.25 M NaCl. Maxima reverse transcriptase (Thermo Scientific) was run according to the manufacturer’s instructions using a multiplexed primed reverse transcriptase reaction containing 96-well sequence tagged gene reporter primers, specific to luciferase, and 96-well tagged polydT primers were used at 750 nM and 250 nM final concentrations respectively with 50 units of Maxima. After sequencing tagging all first-strand cDNA during reverse transcriptase, each 96-well plate was pooled and the unincorporated primers were washed away from the cDNA by precipitating with 10% PEG 8000 and 1.25 M NaCl. After RNase A treatment at 37 °C for 30 min, 10% of the first strand cDNA was used in a PCR reaction with Illumina-compatible sequencing primers targeted to the luciferase gene reporter transcripts. The gene reporter amplicon was then sequenced using a 50 cycle PE Illumina Miseq kit. The remaining first-strand cDNA was then converted to double-strand cDNA using the Second Strand Synthesis Module (NEB). 1 ng of the double-strand cDNA (ds-cDNA) libraries was then tagmented with Nextera XT (Illumina) according to the manufacturer’s protocol. The pooled cDNA libraries for global gene expression are sequence-tagged only on the 3′ end of the sense transcript, therefore, after full-length ds-cDNA tagmentation we performed enrichment PCR of only the sequence-tagged 3′ end of the mRNA transcript by combining the Nextera N700 series of primers with the Tru-Seq P5 adaptor by means of sequence complementarity introduced during reverse transcriptase priming. This process creates 3′ digital gene expression libraries and was sequenced using 1 flow cell on the Illumina NextSeq machine.

Analysis of gene expression and TF-seq data

For pathway reporter data, we counted the number of unique RNA molecules for every well and reporter, requiring a perfect match for the respective tags. In 3′ gene expression data we first mapped the genomic read to the RefSeq (mm10) transcript sequences using Burrows–Wheeler alignment with default settings48. Then we counted the number of unique RNA molecules mapping to each gene in each well, using the RNA UMI and well barcodes embedded into the other read. The resulting count matrices for reporters and genes were analyzed identically using negative binomial generalized linear models from the EdgeR package that are optimized for analyzing count data49. We analyzed all of the stimulations individually with a linear model in which we compared the drug effect to DMSO while subtracting the effect of time, using Benjamini–Hochberg FDR < 0.05 as a significance threshold. The set of consistently regulated genes was found by aggregating significant genes from different stimulations using Robust Rank Aggregation algorithm50. Significance threshold of 0.05 was applied to Bonferroni corrected aggregation P values. To identify enriched Chip-seq binding within this group of genes, we first downloaded the binding scores for multiple transcription factors and chromatin modifications from LPS stimulated BMDMs23. First, we associated the binding peaks to the closest gene and then added up all the scores for all peaks associated to a gene. Based on these scores we ranked all genes and used Wilcoxon test to compare the rankings of our gene set to 1,000 most highly expressed genes in our data set.

Statistical analysis

Results of pilot studies and/or previously completed experiments in similar assay systems were used to estimate sample sizes9. Unless otherwise indicated, error bars represent mean ± s.d. for three biological replicates from one independent experiment that is representative of at least two independent experiments. Variance could not be determined for comparisons of small sample sizes. EC50s for IL-10 upregulation and IC50s for CDK8 inhibition were estimated using 4-parameter nonlinear regression in Prism6 (GraphPad). Statistical significance of differences between experimental groups was assessed using unpaired, two-tailed Student t test or one-way ANOVA with Dunnett post-test as indicated with Prism6. Unless otherwise indicated, *P < 0.05; **P < 0.01; ***P < 0.001.

Data availability

All data generated or analyzed during this study are included in this published article (and its Supplementary Information files) or are available from the corresponding author on reasonable request. Primary RNA-seq data used to generate Supplementary Figure 8b are available from the NIH Gene Expression Omnibus (accession number GSE99759).

Supplementary Material

Dataset 1
Dataset 2
Dataset 3
Dataset 4
S1
S2

Editorial Summary.

Inhibitors of CDK8 enhance IL-10 production during innate immune activation in human and mouse primary macrophages and dendritic cells via diminished phosphorylation of the c-Jun subunit of the AP-1 transcription regulation complex.

Acknowledgments

This work was supported by funding from the National Institutes of Health Grants K08DK104021 (B.K.), R01CA190509 (P.S.), U01DK062432 and P30DK043351 (R.J.X.); Crohn’s and Colitis Foundation of America Grant 500229 (R.J.X.); and The Leona M. and Harry B. Helmsley Charitable Trust Grant 500203 (S.L.S. and R.J.X.). S.L.S. is an Investigator of the Howard Hughes Medical Institute. We thank P. Baran (The Scripps Research Institute) for the generous gift of Δ16-cortistatin A.

Footnotes

Competing financial interests

L.J., T.B.S., B.K., J.R.P., N.S.G., A.F.S., and R.J.X. are co-inventors on a provisional patent application based on these results.

The authors declare competing financial interests: details accompany the online version of the paper.

Author contributions

L.J. designed studies, conducted experiments, analyzed results and wrote the paper. T.B.S. devised the project, developed the screen, designed studies, conducted experiments, analyzed results and wrote the paper. D.J.O. designed studies, conducted experiments and analyzed results. R.K. analyzed gene expression and multiplex reporter data. J.B. designed studies, conducted experiments and analyzed results. K.J.B. synthesized BRD0330 analogs. B.K. designed studies, conducted experiments and analyzed results. B.S.-L. designed studies, conducted experiments and analyzed results. A.F. generated CcncΔ/Δ mice. C.N.R. conducted the screen. I.J.L. designed studies and analyzed results. B.J. synthesized CCT251921. D.B.G. designed studies, conducted experiments and analyzed results. J.R.P. designed studies, conducted experiments and analyzed results. P.S. generated CcncΔ/Δ mice. A.J.P. designed studies, conducted experiments and analyzed results. S.L.S. designed studies, analyzed results and wrote the paper. N.S.G. designed studies, analyzed results and wrote the paper. A.F.S. devised the project, developed the screen, synthesized CCT251921, designed studies, analyzed results and wrote the paper. R.J.X. devised the project, developed the screen, designed studies, analyzed results and wrote the paper.

Additional information

Any supplementary information, chemical compound information and source data are available in the online version of the paper. Reprints and permissions information is available online at http://www.nature.com/reprints/index.html. Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Correspondence and requests for materials should be addressed to N.S.G., A.F.S. or R.J.X.

References

  • 1.Saraiva M, O’Garra A. The regulation of IL-10 production by immune cells. Nat Rev Immunol. 2010;10:170–181. doi: 10.1038/nri2711. [DOI] [PubMed] [Google Scholar]
  • 2.Shouval DS, et al. Interleukin 10 receptor signaling: master regulator of intestinal mucosal homeostasis in mice and humans. Adv Immunol. 2014;122:177–210. doi: 10.1016/B978-0-12-800267-4.00005-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Asadullah K, Sterry W, Volk HD. Interleukin-10 therapy–review of a new approach. Pharmacol Rev. 2003;55:241–269. doi: 10.1124/pr.55.2.4. [DOI] [PubMed] [Google Scholar]
  • 4.Braat H, et al. A phase I trial with transgenic bacteria expressing interleukin-10 in Crohn’s disease. Clin Gastroenterol Hepatol. 2006;4:754–759. doi: 10.1016/j.cgh.2006.03.028. [DOI] [PubMed] [Google Scholar]
  • 5.Galeazzi M, et al. A phase IB clinical trial with Dekavil (F8-IL10), an immunoregulatory ‘armed antibody’ for the treatment of rheumatoid arthritis, used in combination wiIh methotrexate. Isr Med Assoc J. 2014;16:666. [PubMed] [Google Scholar]
  • 6.Rodríguez M, et al. Polarization of the innate immune response by prostaglandin E2: a puzzle of receptors and signals. Mol Pharmacol. 2014;85:187–197. doi: 10.1124/mol.113.089573. [DOI] [PubMed] [Google Scholar]
  • 7.Martin M, Rehani K, Jope RS, Michalek SM. Toll-like receptor-mediated cytokine production is differentially regulated by glycogen synthase kinase 3. Nat Immunol. 2005;6:777–784. doi: 10.1038/ni1221. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Clark K, et al. Phosphorylation of CRTC3 by the salt-inducible kinases controls the interconversion of classically activated and regulatory macrophages. Proc Natl Acad Sci USA. 2012;109:16986–16991. doi: 10.1073/pnas.1215450109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Sundberg TB, et al. Small-molecule screening identifies inhibition of salt-inducible kinases as a therapeutic strategy to enhance immunoregulatory functions of dendritic cells. Proc Natl Acad Sci USA. 2014;111:12468–12473. doi: 10.1073/pnas.1412308111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Wang B, et al. Microtubule acetylation amplifies p38 kinase signalling and anti-inflammatory IL-10 production. Nat Commun. 2014;5:3479. doi: 10.1038/ncomms4479. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Na YR, et al. The early synthesis of p35 and activation of CDK5 in LPS-stimulated macrophages suppresses interleukin-10 production. Sci Signal. 2015;8:ra121. doi: 10.1126/scisignal.aab3156. [DOI] [PubMed] [Google Scholar]
  • 12.Souto A, Gómez-Reino JJ. Apremilast for the treatment of psoriatic arthritis. Expert Rev Clin Immunol. 2015;11:1281–1290. doi: 10.1586/1744666X.2015.1102057. [DOI] [PubMed] [Google Scholar]
  • 13.Gordon JN, et al. CC-10004 but not thalidomide or lenalidomide inhibits lamina propria mononuclear cell TNF-α and MMP-3 production in patients with inflammatory bowel disease. J Crohns Colitis. 2009;3:175–182. doi: 10.1016/j.crohns.2009.03.001. [DOI] [PubMed] [Google Scholar]
  • 14.Xing L, Rai B, Lunney EA. Scaffold mining of kinase hinge binders in crystal structure database. J Comput Aided Mol Des. 2014;28:13–23. doi: 10.1007/s10822-013-9700-4. [DOI] [PubMed] [Google Scholar]
  • 15.Allen BL, Taatjes DJ. The Mediator complex: a central integrator of transcription. Nat Rev Mol Cell Biol. 2015;16:155–166. doi: 10.1038/nrm3951. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Bancerek J, et al. CDK8 kinase phosphorylates transcription factor STAT1 to selectively regulate the interferon response. Immunity. 2013;38:250–262. doi: 10.1016/j.immuni.2012.10.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Shi J, et al. Scalable synthesis of cortistatin A and related structures. J Am Chem Soc. 2011;133:8014–8027. doi: 10.1021/ja202103e. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Pelish HE, et al. Mediator kinase inhibition further activates super-enhancer-associated genes in AML. Nature. 2015;526:273–276. doi: 10.1038/nature14904. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Mallinger A, et al. Discovery of potent, selective, and orally bioavailable small-molecule modulators of the mediator complex-associated kinases CDK8 and CDK19. J Med Chem. 2016;59:1078–1101. doi: 10.1021/acs.jmedchem.5b01685. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Kwiatkowski N, et al. Targeting transcription regulation in cancer with a covalent CDK7 inhibitor. Nature. 2014;511:616–620. doi: 10.1038/nature13393. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Li N, et al. Cyclin C is a haploinsufficient tumour suppressor. Nat Cell Biol. 2014;16:1080–1091. doi: 10.1038/ncb3046. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.O’Connell DJ, et al. Simultaneous pathway activity inference and gene expression analysis using RNA sequencing. Cell Syst. 2016;2:323–334. doi: 10.1016/j.cels.2016.04.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Garber M, et al. A high-throughput chromatin immunoprecipitation approach reveals principles of dynamic gene regulation in mammals. Mol Cell. 2012;47:810–822. doi: 10.1016/j.molcel.2012.07.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Witte S, O’Shea JJ, Vahedi G. Super-enhancers: Asset management in immune cell genomes. Trends Immunol. 2015;36:519–526. doi: 10.1016/j.it.2015.07.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Shlyueva D, Stampfel G, Stark A. Transcriptional enhancers: from properties to genome-wide predictions. Nat Rev Genet. 2014;15:272–286. doi: 10.1038/nrg3682. [DOI] [PubMed] [Google Scholar]
  • 26.Porter DC, et al. Cyclin-dependent kinase 8 mediates chemotherapy-induced tumor-promoting paracrine activities. Proc Natl Acad Sci USA. 2012;109:13799–13804. doi: 10.1073/pnas.1206906109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Poss ZC, et al. Identification of mediator kinase substrates in human cells using cortistatin A and quantitative phosphoproteomics. Cell Reports. 2016;15:436–450. doi: 10.1016/j.celrep.2016.03.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Yamamoto S, et al. Mediator cyclin-dependent kinases upregulate transcription of inflammatory genes in cooperation with NF-κB and C/EBPβ on stimulation of Toll-like receptor 9. Genes Cells. 2017;22:265–276. doi: 10.1111/gtc.12475. [DOI] [PubMed] [Google Scholar]
  • 29.Lin A, et al. Casein kinase II is a negative regulator of c-Jun DNA binding and AP-1 activity. Cell. 1992;70:777–789. doi: 10.1016/0092-8674(92)90311-y. [DOI] [PubMed] [Google Scholar]
  • 30.Huang CC, et al. Calcineurin-mediated dephosphorylation of c-Jun Ser-243 is required for c-Jun protein stability and cell transformation. Oncogene. 2008;27:2422–2429. doi: 10.1038/sj.onc.1210888. [DOI] [PubMed] [Google Scholar]
  • 31.Taira N, et al. DYRK2 priming phosphorylation of c-Jun and c-Myc modulates cell cycle progression in human cancer cells. J Clin Invest. 2012;122:859–872. doi: 10.1172/JCI60818. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Aikawa Y, et al. Treatment of arthritis with a selective inhibitor of c-Fos/activator protein-1. Nat Biotechnol. 2008;26:817–823. doi: 10.1038/nbt1412. [DOI] [PubMed] [Google Scholar]
  • 33.Lim S, Kaldis P. Cdks, cyclins and CKIs: roles beyond cell cycle regulation. Development. 2013;140:3079–3093. doi: 10.1242/dev.091744. [DOI] [PubMed] [Google Scholar]
  • 34.Saraiva M, et al. Identification of a macrophage-specific chromatin signature in the IL-10 locus. J Immunol. 2005;175:1041–1046. doi: 10.4049/jimmunol.175.2.1041. [DOI] [PubMed] [Google Scholar]
  • 35.Donner AJ, Ebmeier CC, Taatjes DJ, Espinosa JM. CDK8 is a positive regulator of transcriptional elongation within the serum response network. Nat Struct Mol Biol. 2010;17:194–201. doi: 10.1038/nsmb.1752. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Schreiber S, et al. Safety and efficacy of recombinant human interleukin 10 in chronic active Crohn’s disease. Crohn’s disease IL-10 cooperative study group. Gastroenterology. 2000;119:1461–1472. doi: 10.1053/gast.2000.20196. [DOI] [PubMed] [Google Scholar]
  • 37.Marlow GJ, van Gent D, Ferguson LR. Why interleukin-10 supplementation does not work in Crohn’s disease patients. World J Gastroenterol. 2013;19:3931–3941. doi: 10.3748/wjg.v19.i25.3931. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Itakura E, et al. IL-10 expression by primary tumor cells correlates with melanoma progression from radial to vertical growth phase and development of metastatic competence. Mod Pathol. 2011;24:801–809. doi: 10.1038/modpathol.2011.5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Pyonteck SM, et al. CSF-1R inhibition alters macrophage polarization and blocks glioma progression. Nat Med. 2013;19:1264–1272. doi: 10.1038/nm.3337. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Wang S, et al. Expression of CD163, interleukin-10, and interferon-gamma in oral squamous cell carcinoma: mutual relationships and prognostic implications. Eur J Oral Sci. 2014;122:202–209. doi: 10.1111/eos.12131. [DOI] [PubMed] [Google Scholar]
  • 41.Clarke PA, et al. Assessing the mechanism and therapeutic potential of modulators of the human Mediator complex-associated protein kinases. eLife. 2016;5:e20722. doi: 10.7554/eLife.20722. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Yamaoka K. Janus kinase inhibitors for rheumatoid arthritis. Curr Opin Chem Biol. 2016;32:29–33. doi: 10.1016/j.cbpa.2016.03.006. [DOI] [PubMed] [Google Scholar]
  • 43.Miller SC, et al. Identification of known drugs that act as inhibitors of NF-kappaB signaling and their mechanism of action. Biochem Pharmacol. 2010;79:1272–1280. doi: 10.1016/j.bcp.2009.12.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Sundberg TB, et al. Development of chemical probes for investigation of salt-inducible kinase function in vivo. ACS Chem Biol. 2016;11:2105–2111. doi: 10.1021/acschembio.6b00217. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Wawer MJ, et al. Toward performance-diverse small-molecule libraries for cell-based phenotypic screening using multiplexed high-dimensional profiling. Proc Natl Acad Sci USA. 2014;111:10911–10916. doi: 10.1073/pnas.1410933111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Duan Q, et al. LINCS Canvas Browser: interactive web app to query, browse and interrogate LINCS L1000 gene expression signatures. Nucleic Acids Res. 2014;42:W449–W460. doi: 10.1093/nar/gku476. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Pfaffl MW. A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Res. 2001;29:e45. doi: 10.1093/nar/29.9.e45. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics. 2009;25:1754–1760. doi: 10.1093/bioinformatics/btp324. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Robinson MD, McCarthy DJ, Smyth GK. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics. 2010;26:139–140. doi: 10.1093/bioinformatics/btp616. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Kolde R, Laur S, Adler P, Vilo J. Robust rank aggregation for gene list integration and meta-analysis. Bioinformatics. 2012;28:573–580. doi: 10.1093/bioinformatics/btr709. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Dataset 1
NIHMS918591-supplement-Dataset_1.xlsx (22.1KB, xlsx)
Dataset 2
NIHMS918591-supplement-Dataset_2.xlsx (59.7KB, xlsx)
Dataset 3
NIHMS918591-supplement-Dataset_3.xlsx (14.5KB, xlsx)
Dataset 4
NIHMS918591-supplement-Dataset_4.xlsx (269.1KB, xlsx)
S1
NIHMS918591-supplement-S1.pdf (4.6MB, pdf)
S2
NIHMS918591-supplement-S2.pdf (375.1KB, pdf)

Data Availability Statement

All data generated or analyzed during this study are included in this published article (and its Supplementary Information files) or are available from the corresponding author on reasonable request. Primary RNA-seq data used to generate Supplementary Figure 8b are available from the NIH Gene Expression Omnibus (accession number GSE99759).

RESOURCES