Molecular determinants of crosstalk between nuclear receptors and toll-like receptors

Abstract

Nuclear receptors (NRs) repress transcriptional responses to diverse signaling pathways as an essential aspect of their biological activities, but mechanisms determining the specificity and functional consequences of transrepression remain poorly understood. Here, we report signal- and gene-specific repression of transcriptional responses initiated by engagement of toll-like receptors (TLR) 3, 4, and 9 in macrophages. The glucocorticoid receptor (GR) represses a large set of functionally related inflammatory response genes by disrupting p65/interferon regulatory factor (IRF) complexes required for TLR4- or TLR9-dependent, but not TLR3-dependent, transcriptional activation. This mechanism requires signaling through MyD88 and enables the GR to differentially regulate pathogen-specific programs of gene expression. PPARgamma and LXRs repress overlapping transcriptional targets by p65/IRF3-independent mechanisms and cooperate with the GR to synergistically transrepress distinct subsets of TLR-responsive genes. These findings reveal combinatorial control of homeostasis and immune responses by nuclear receptors and suggest new approaches for treatment of inflammatory diseases.

Figure 1

Inhibition of LPS-Dependent Gene Expression by Nuclear Receptor Agonists (A) Relative expression of LPS-inducible genes in peritoneal macrophages under control conditions and after 6 h of LPS treatment in the absence or presence of 1 μM Dex. Left panel; genes induced by LPS >2-fold. Middle panel; genes induced by LPS >2-fold and inhibited by Dex more than 50% (Dex-sensitive). Right panel; genes induced by LPS >2-fold and resistant to Dex. Expression data was collected using Affymetrix U74A microarrays and represent results obtained from four independent experiments. (B) Dex-mediated transrepression of LPS-inducible genes in macrophages derived from fetal liver-derived macrophages of wild-type (WT) and NCoR^−/− mice. The illustrated gene expression values are for the 12 most highly repressed genes in wild-type macrophages. Expression data was collected using Affymetrix U74A microarrays and represents results obtained from two independent experiments. (C) Effect of nuclear receptor agonists on responsiveness of 208 LPS-target genes in peritoneal macrophages. Genes are ordered based on magnitude of average LPS induction over 4 experiments from >40-fold at the top to 2-fold at the bottom. Genes in which the LPS response was not altered by agonist treatment are illustrated in gray. Red indicates ligand-dependent upregulation and blue ligand-dependent downregulation of the LPS response. The magnitude of the effect is indicated by the key at lower left. Expression data was collected using Affymetrix U74A microarrays and represents results obtained from a minimum of two independent experiments for each drug treatment. Data for Ro is from experiments performed under identical conditions (Welch et al., 2003). (D) Confirmation of negative regulation of LPS-target genes by Northern blotting. Macrophages were treated with LPS for 6 h in the presence of 1 μM concentrations of the indicated agonists.

Figure 2

Differential Repression of TLR Responses by GR, PPARγ, and LXRs (A) Identification of IRF3-binding sites as highly enriched sequence motifs in the promoters of LPS-inducible genes. The top sequence logo is representative of the most significant motif present in the promoters of genes that are LPS-inducible and not present in promoters of non-LPS-inducible genes. This motif was found de novo without previous knowledge of known transcription factor binding sites. The sequence logos representing the known consensus IRF3-binding and ISRE motifs are shown for comparison. (B) A scatter plot illustrating fold responses to poly I:C compared to genes activated at least 3-fold by LPS. Expression data was collected using Codelink Uniset Mouse 1 microarrays and is representative of results obtained from four independent experiments. (C) Effect of IRF3-deficiency and GR (Dex), LXR (GW3965) or PPARγ (GW7845) agonists on transcriptional responses to LPS and poly I:C. The panel illustrates the 208 most highly induced LPS-responsive genes. Sensitivity of each gene to loss of IRF3 (column 1) or treatment with Dex, GW3965 and GW7845 is color coded as indicated in the key at the bottom. Expression data was collected using Amersham Codelink Mouse Uniset 1 microarrays and represents results from two independent experiments for each ligand. (D) Confirmation of signal-specific repression of LPS- and poly I:C-inducible genes by GR, LXR and PPARγ-specific agonists. Macrophages were treated with LPS or poly I:C for 6 h in the presence of 1 μM concentrations of the indicated agonists.

Figure 3

Signal-specific Repression by GR Correlates with a Requirement for IRF3 for Transcriptional Activation (A) Transcriptional responses of 54 genes highly induced by LPS, poly I:C and CpG1668, exhibiting sensitivity to Dex when activated by LPS (column 2) and resistance to Dex when activated by poly I:C (column 3). The dependence of the LPS response on IRF3 is indicated in the first column and the effect of CpG1668 (1 μM) is indicated column 4. Effects of IRF3-deficiency or nuclear receptor agonists on LPS, CpG1668 or poly I:C responses are color coded as in Figure 2C. (B) Promoter structure and expression profile of Ifit1 in response to LPS, poly I:C and Dex in wild-type (WT) and IRF3^−/− peritoneal macrophages. (C) An ISRE-specific promoter exhibits LPS-specific repression by Dex. U373 cells were transfected with a 5xISRE-Luc reporter plasmid. Cells were treated with the indicated combinations of LPS (100 ng/ml), poly I:C (50 μg/ml) and Dex, and analyzed for luciferase activity 18 h later. (D) Expression of Ifit1 in response to LPS (1 μg/ml) in control and NF-κB^−/− fetal liver-derived macrophages. (E) p65 recruitment to the proximal promoter region of Ifit1 is specifically induced by LPS and inhibited by activation of GR. Primary macrophages were treated with LPS (100 ng/ml), poly I:C (50 μg/ml), and the indicated agonists for GR, PPARγ and LXRs for 1 h. ChIP assays were performed with antibodies against IRF3, CBP, p65 and control IgG, respectively. Immunoprecipitated DNA was analyzed by PCR using primers specific for the promoter.

Figure 4

GR Specifically Inhibits the Interaction of p65 with IRF3 (A) GR-DBD preferentially interacts with the p65 RHD. GST pull-down assays were performed using the indicated GST-NR-DBD fusion proteins and in vitro translated full-length p65 or p65 RHD, respectively. (B) Wild-type GR preferentially interacts with p65 in vivo in a ligand-dependent manner. The mammalian two-hybrid assay was performed in RAW264.7 cells using Gal-p65 as bait and the indicated VP16-nuclear receptor fusion proteins as preys in the presence and absence of agonists. (C) GR-DBD inhibits interaction of IRF3 with p65. GST pull-down assays were performed using GST-p65 and increasing amounts of in vitro translated full-length IRF3 and/or GR-DBD as indicated. (D) IRF3 interacts with the p65 RHD in vitro. GST pull-down assays were performed using GSTp65 RHD and in vitro translated full-length IRF3. (E) GR-DBD mutant GR^K471A is unable to interact with the p65 RHD in vitro. GST pull-down assays were performed using GST-p65 RHD and in vitro translated full-length GR and GR^K477A, respectively. (F) Inhibition of Gal-p65 transactivation by liganded wild-type GR but not by GR^K477A. Endogenous GR expression in mouse RAW264.7 cells was knocked down by pretreatment with a GR-specific siRNA for 48 h. Cells were then transfected with expression vectors for Gal4 (Gal), Gal4-p65 (Gal-p65), wild-type human GR or GR^K477A as indicated in the presence of Dex. Luciferase activity was analyzed 24 h later. (G) Coactivation of Gal-IRF3 by p65 is inhibited by liganded wild-type GR but not by GR^K477A. Endogenous GR expression in RAW264.7 cells was knocked down by pretreatment with a GR-specific siRNA for 48 h prior to transfection with expression vectors for human wild-type GR or GR^K477A and treatment with Dex as indicated.

Figure 5

Utilization of IRF3 as a Coactivator of p65 Determines Gene-specific Sensitivity (A) Location of NF-κB sites in proximal promoter regions of Scyb9, Clic4, Nfkbia and Gro1, and their expression profiles in response to LPS in wild-type (WT) and IRF3^−/− peritoneal macrophages. (B) Expression profiles of Scyb9, Clic4, Nfkbia and Gro1 in response to LPS and Dex in wild-type peritoneal macrophages. (C) Recruitment of IRF3 to the proximal promoter regions of Scyb9 and Clic4, but not Nfkbia or Gro1, in response to LPS. Recruitment of IRF3 was largely inhibited by Dex. Proximal regions of Scyb9, Clic4, Nfkbia and Gro1 promoters that includes an NF-κB site were analyzed by ChIP assays using the indicated antibodies. Crosslinking was performed 1 h after treatment with LPS and Dex. (D) Expression profiles of IP10, Clic4 and Nfkbia in response to LPS and Dex in control (WT) and MyD88^−/− peritoneal macrophages. Gene expression was determined by real-time quantitative PCR.

Figure 6

GR and PPARγ Function in a Combinatorial Manner to Inhibit LPS Responses (A) Combinatorial interactions between GR and PPARγ agonists at a genome-wide level. Peritoneal macrophages were stimulated with LPS in the absence or presence of Dex alone, GW7845 alone, or the combination of Dex plus GW7845. Each agonist was used at 1 μM. The panel illustrates LPS-target genes exhibiting a > 40% reduction of the LPS response in the presence of at least one agonist. Effects of agonists on the LPS response are color coded according to the legend in Figure 1C. Red arrows indicate genes in which one agonist abolished strong inhibitory effects of the other agonist. Blue arrows indicate genes in which the combination of Dex and GW7845 resulted in stronger inhibition of the LPS response than either agonist alone. Expression data was collected using Affymetrix U74A microarrays and represents results obtained from two independent experiments. (B) Confirmation of combinatorial effect of GR and PPARγ agonists on regulation of LPS-target genes by Northern blotting. Macrophages were treated with LPS for 6 h in the presence of the indicated concentrations (10 nM, 1 μM) of agonists. (C) Confirmation of combinatorial effect of GR and LXR agonists on regulation of LPS-target genes by Northern blotting. (D) Confirmation of combinatorial effect of GR and PPARγ agonists (1 μM) on iNOS expression by real-time quantitative PCR. (E) Combinatorial interactions between GR and PPARγ at the promoter levels. RAW 264.7 cells were transfected with a luciferase reporter plasmid under transcriptional control of the iNOS promoter, PPARγ and RXRα expression plasmids. Cells were treated with the indicated combinations of LPS (100 ng/ml), Dex (10 nM, 1 μM) and GW7845 (10 nM, 1 μM), analyzed for luciferase activity 24 h later. (F) In vivo effects of combinations of GR and PPARγ agonists on the response to intraperitoneal injection of LPS. Six C57BL6 mice were pretreated with the indicated combinations of GW7845 (1 mg/kg) and Dex (1 mg/kg) for 7 days, injected intraperitoneally with LPS (1 mg) and circulating levels of IL-12 p40 were measured by ELISA 8 h later. (G) In vivo effects of combinations of GR and LXR agonists on the response to intraperitoneal injection of LPS. Six C57BL6 mice were pretreated with the indicated combinations of Dex (1 mg/kg) and T1317 (10 mg/kg) for 7 days, injected intraperitoneally with LPS (1 mg) and circulating levels of TNFα were measured by ELISA 6 h later.

Figure 7

Receptor-, Signal- and Gene-specific Counter-regulation of Inflammatory Responses by GR, PPARγ and LXRs (A) Venn diagram indicating sensitivity of LPS-responsive genes to GR, PPARγ and LXR-specific agonists. Nuclear receptor sensitivity was defined as >40% repression of the LPS response in a minimum of two independent microarray experiments (left). Venn diagram indicating sensitivity of poly I:C-responsive genes to GR, PPARγ and LXR-specific agonists. Nuclear receptor sensitivity was defined as >40% repression of the poly I:C response in a minimum of two independent microarray experiments (right). Venn diagrams are derived from the complete data set used to generate Figures 2B and 2C. (B) Representative functional annotations corresponding to Biological Process terms derived from the Gene Ontology database that were significantly enriched in the sets of LPS- and poly I:C-responsive genes. p represents the probability of obtaining the indicated number n genes within the category by chance determined as previously described (Ogawa et al., 2004). Color coding corresponds to the following p values; [unk]; p less than 0.01, [unk]; p less than 0.0001, [unk]; p less than 10^-6. (C) Model for signal-specific GR-mediated transrepression, determined by utilization of p65 as an obligate TLR4-specific co-activator of IRF3. IRF-mediated activation of ISRE-containing genes by TLR4 and TLR9 through MyD88-pathway requires that p65 function as a signal-specific co-activator. The p65/IRF interaction is disrupted by liganded GR, resulting in transrepression. TLR3-specific activation of IRF3 through the TRIF pathway is p65-independent, and hence GR-resistant. (D) Model for gene-specific GR-mediated transrepression, determined by utilization of IRF3 as an obligate promoter-specific co-activator of NF-κB. The IRF3/p65 interaction is disrupted by liganded GR, providing an explanation for promoter-specific inhibition of the LPS response.