Nrf2 suppresses macrophage inflammatory response by blocking proinflammatory cytokine transcription

Abstract

Nrf2 (NF-E2-related factor-2) transcription factor regulates oxidative/xenobiotic stress response and also represses inflammation. However, the mechanisms how Nrf2 alleviates inflammation are still unclear. Here, we demonstrate that Nrf2 interferes with lipopolysaccharide-induced transcriptional upregulation of proinflammatory cytokines, including IL-6 and IL-1β. Chromatin immunoprecipitation (ChIP)-seq and ChIP-qPCR analyses revealed that Nrf2 binds to the proximity of these genes in macrophages and inhibits RNA Pol II recruitment. Further, we found that Nrf2-mediated inhibition is independent of the Nrf2-binding motif and reactive oxygen species level. Murine inflammatory models further demonstrated that Nrf2 interferes with IL6 induction and inflammatory phenotypes in vivo. Thus, contrary to the widely accepted view that Nrf2 suppresses inflammation through redox control, we demonstrate here that Nrf2 opposes transcriptional upregulation of proinflammatory cytokine genes. This study identifies Nrf2 as the upstream regulator of cytokine production and establishes a molecular basis for an Nrf2-mediated anti-inflammation approach.

Figure 1. Genetic activation of Nrf2 inhibits M1 induction of proinflammatory cytokine genes.

(a) Nrf2 protein expression in BMDMs from two independent WT and KpCKO mice, respectively. (b) Relative Nqo1 gene expression. BMDMs from WT and KpCKO mice were either untreated (Con) or treated with M1 stimulation (M1, 5-ng ml⁻¹ LPS and 10-ng ml⁻¹ IFNγ) or M2 stimulation (M2, 10-ng ml⁻¹ IL-4) for 6 h. (c) Similarity in gene expression of M1- and M2-induced genes between WT and KpCKO mice. With >2-fold change, 4,162 genes are upregulated by M1 stimulation (M1-induced genes), while 856 genes are upregulated by M2 stimulation (M2-induced genes), both in two independent WT mice. The expression pattern of M1- and M2-induced genes in WT and KpCKO BMDMs are compared to each other. The Pearson correlation coefficient for each comparison is depicted as colours from white (low) to magenta (high). (d) The number of M1-induced genes regulated by Nrf2. In 4,162 M1-induced genes, the number of genes upregulated or downregulated by Keap1-deficiency (with >2-fold change in KpCKO compared with WT in two independent BMDM cultures) are represented as orange and blue numbers, respectively. The numbers given in parentheses are the numbers of genes within 50 kb from Nrf2 peaks detected in the ChIP-seq analysis of M1-activated KpCKO BMDMs described below. (e–m) Relative expression of M1- and M2-induced genes in WT and KpCKO BMDMs. Expression of IL6 (e), IL1b (f), IL1a (g), Nos2 (h), IL12b (i), Tnf (j), Irf1 (k), Arg1 (l) and Ccl17 (m) are presented. Data in b,e–m are mean±s.d. from three to four mice (*P<0.05, unpaired t-test).

Figure 2. Nrf2 binding to the proximity of the proinflammatory cytokine genes.

(a–d) Nrf2 ChIP-sequencing tracks of KpCKO BMDMs and DEM-treated Hepa1 cells. KpCKO BMDMs were either untreated (KpCKO-Con) or treated with M1 stimulation (KpCKO-M1, 5-ng ml⁻¹ LPS and 10-ng ml⁻¹ IFNγ) or M2 stimulation (KpCKO-M2, 10-ng ml⁻¹ IL-4) for 4 h. Fragment pileup per million reads at the genomic loci proximal to the IL6 (a), IL1b (b), IL1a (c) and Nqo1 (d) genes are presented. DEM-Hapa1 shows the Nrf2 ChIP-seq track of DEM-treated Hepa1 cells. Orange bars represent the peak regions. Primers for ChIP-qPCR described below were designed to amplify peak and non-peak (negative control) regions indicated by blue and grey triangles, respectively. (e) A consensus motif enriched in a ±25 bp region from Nrf2 peak summits in KpCKO-Con samples. (f) Phylogenetic conservation of ARE motifs in peak regions proximal to the IL6. Nucleotides identical to the core sequence of ARE, TGAG/CnnnGC, are in blue shades, and nucleotides in ARE identical to the murine sequences are depicted with the blue characters. (g–j) Nrf2 ChIP-qPCR analyses of KpCKO BMDMs. The Nrf2 signal was detected at the peak regions proximal to IL6 (g), IL1b (h) and IL1a (i) genes (blue characters), similarly to the Nqo1 promoter (j) used as a positive control. The intron of the Txs gene (Txs, g–j) and the non-peak regions proximal to the IL1b (h) and IL1a (i) genes were used as negative controls. Data in g–j are mean±s.d. from three mice (*P<0.05, unpaired t-test).

Figure 3. Microarray and ChIP-seq analyses of Keap1-deficient BMDMs.

(a) KEGG pathway analysis on M1-induced and Nrf2-repressed genes within 50 kb of Nrf2 peaks detected in KpCKO-M1 samples. The top five pathways are presented. (b) Expression profiles of representative M1-induced genes repressed by Nrf2. BMDMs from two independent WT and KpCKO mice were either untreated (Con) or treated with M1 stimulation (M1, 5-ng ml⁻¹ LPS and 10-ng ml⁻¹ IFNγ) for 6 h and rendered for microarray analysis. Each row represents the mRNA expression level in the corresponding mice depicted as colours from turquoise (low) to coral (high). Each column represents one of 26 genes located within 50 kb of the Nrf2-binding peaks in KpCKO-M1 BMDMs, whose expression was induced by M1 stimulation in WT BMDMs (fold change >2) and repressed by Keap1-deficiency in M1-stimulated BMDMs (fold change <−4). The expression of Nqo1 is displayed to show the expression pattern of typical Nrf2 target genes. Blue dots under the figure represent genes within 50 kb of the Nrf2-binding peaks obtained in DEM-treated Hepa1 cells. Note that the inflammation-related genes are accumulated in the left group of genes, which is inhibited by Nrf2 in a macrophage-specific manner.

Figure 4. Inhibition of M1-induced proinflammatory cytokine genes by chemical Nrf2 inducers.

(a) Nrf2 protein accumulation in the WT BMDMs treated with 100-μM DEM for 4 h (representative data, n=4). (b–e) Inhibition of proinflammatory gene expression by Nrf2 inducers. BMDMs from WT and Nrf2-KO mice were untreated (Con) or M1-stimulated (M1, 5-ng ml⁻¹ LPS and 10-ng ml⁻¹ IFNγ) in the presence of 100-μM DEM (DEM), 10-μM 15d-PGJ₂ (15d-PGJ₂) or vehicle for 6 h. Relative expressions of Nqo1 (b), IL6 (c), IL1b (d) and IL1a (e) genes were examined. Note that the inhibition of IL6, IL1b and IL1a genes and Nqo1-induction are cancelled by Nrf2-knockout. (f–h) Dose-dependent inhibition of proinflammatory genes by the Nrf2 inducer. Relative expressions of IL6 (f), IL1b (g) and IL1a (h) genes after 6 h of M1-stimulation in the presence of increasing doses of DEM. *P<0.05 (unpaired t-test) against vehicle-treated M1-BMDMs. (i–l) Nrf2 ChIP-qPCR of DEM-treated WT BMDMs. WT BMDMs were treated with 100-μM DEM for 4 h. Nrf2 signal was detected at the peak regions proximal to the IL6 (i), IL1b (j), IL1a (k) and Nqo1 (l) genes (blue triangles in Fig. 2a–d). The Txs intron (Txs, i–l) and the non-peak regions near the IL1b (j) and IL1a (k) genes were used as negative controls (grey triangles in Fig. 2b,c). (m,n) The concentration of IL-6 (m) and IL-1β (n) secreted in the culture media. BMDMs were stimulated with 5-ng ml⁻¹ LPS and 100-μg ml⁻¹ alum in the presence of 100-μM DEM or vehicle for 24 h. (o–r) Inhibition of proinflammatory genes by the Nrf2 inducer in human cells. THP-1 cells were untreated (Con) or LPS-treated (LPS, 5-ng ml⁻¹ LPS) in the presence of 100-μM DEM or vehicle for 6 h. Relative expressions of IL6 (o), IL1B (p), IL1A (q) and NQO1 (r) are presented. Data are mean±s.d. from three mice (b–n) or three individual samplings (o–r). *P<0.05, unpaired t-test; NS, not significant.

Figure 5. Nrf2-mediated inhibitions of proinflammatory cytokine genes are independent of ROS elimination.

(a) Evaluation of the ROS level in DEM- or NAC-treated BMDMs. WT BMDMs were stimulated with 5-ng ml⁻¹ LPS or vehicle, with or without 100-μM DEM or 1-mM NAC, for 4 h. The ROS level was evaluated by fluorescence after treatment with CellROX reagent during the last 30 min of 4 h stimulation. Data are mean±s.d. from four mice (*P<0.05, paired t-test). (b–d) Evaluation of the ROS-mediated effect. The relative expressions of IL6 (b), IL1b (c) and IL1a (d) were examined by RT-qPCR after the same stimulations as (a). Note that NAC-treatment does not affect these cytokine mRNA expressions. (e–h) Time-course change of inflammatory cytokine gene expression after the chemical Nrf2 inducer treatment. Relative mRNA expression of IL6 (e), IL1b (f), IL1a (g) and Nqo1 (h) in WT BMDMs were examined at indicated times from the start of M1 stimulation in the presence of 100-μM DEM or vehicle. DEM and vehicle were added in the culture media at the same time as the M1 stimulus. The value of M1+Vehicle at 6 h is set to 1. *P<0.05 (unpaired t-test) against M1+Vehicle; ^#P<0.05 (unpaired t-test) against Con+Vehicle. Note that DEM inhibits the expression of the inflammatory cytokine mRNAs, but induces the expression of Nqo1 mRNA.

Figure 6. Nrf2 inhibits the recruitment of RNA Pol II on the proinflammatory cytokine gene loci without affecting the NF-κB p65 recruitment.

(a) ChIP-sequencing track specific for p65 at the IL6 and IL1b loci and Nrf2 ChIP-sequencing track of KpCKO-M1 BMDMs depicted in Fig. 2. Primers for the ChIP-qPCR were designed to amplify the regions indicated with triangles. Grey triangles represent the negative control regions. (b) Pol II ChIP-qPCR analyses of WT BMDMs. BMDMs were stimulated with 5-ng ml⁻¹ LPS, in the presence of 100-μM DEM or vehicle, for 4 h. Numbers under the bar graph represent the peak or non-peak regions depicted in (a). The binding of Pol II was upregulated in LPS-treated cells at the TSS of IL6 (2), the gene body of IL6 (3) and the TSS of the IL1b gene (5), similar to the TSS of the Nqo1 gene in LPS+DEM-treated cells used as a positive control (Nqo1 TSS). The intron of the Txs gene and the promoter regions of IL6 (1) and IL1b (4) were used as negative controls. (c) p65 ChIP-qPCR analyses of WT BMDMs. BMDMs were treated with LPS and/or DEM as in (b). The binding of p65 was upregulated in LPS-treated cells at the p65-binding sites proximal to IL6 (8) and IL1b (9, 10), similarly to the 60 kb upstream site of the IL6 gene (6) used as a positive control. The Txs intron, the Nrf2-binding site near Nqo1 gene (Nqo1 ARE) and the regions without p65 binding near the IL6 (7) and IL1b (11) genes were used as negative controls. Data are mean±s.d. from three mice (*P<0.05, unpaired t-test; NS, not significant).

Figure 7. Nrf2-mediated inhibition of proinflammatory cytokine genes is independent of ARE in upstream regulatory region.

(a) The schema of luciferase reporter vectors. Approximately 1 kb region from the translation start site of IL6 gene was conjugated to the translation start site of NanoLuc reporter vector (IL6-wt). The ARE motif in WT IL6 locus, GCTGAGTCA (Fig. 2f), is mutated to AGATCTGAC in the ARE-mutant reporter vector (IL6-mARE). (b) Relative luciferase activity of IL6-wt and IL6-mARE reporter vectors in the Raw264.7 cells transfected with constitutive active form of NRF2 (NRF2^T80R). (c) Relative luciferase activity of IL6-wt and IL6-mARE reporter vectors in the Raw264.7 cells transfected with NRF2^T80R and p65. (d,e) Relative IL6 and IL1b mRNA expressions in the parental Raw264.7 cells and derived cells in which Nrf2-binding region proximal to the gene is deleted. We deleted the ARE-containing regulatory regions from the IL6 (d) and IL1b (e) genes (Supplementary Fig. 8) by means of the CRISPR/Cas9 genome-editing technology. Cells were treated with LPS and/or DEM for 6 h. The value of LPS-stimulated parental Raw264.7 cells is set to 1 (d,e).

Figure 8. Nrf2 inhibits IL6 expression and alleviates inflammation in vivo.

(a) The schema of in vivo monitoring of IL6 expression using WIM-6 mice. WIM-6 mice express the luciferase gene under the control of the human IL6 promoter, and IL6-inducing stimuli lead to the emission of luciferase luminescence in WIM-6 mice. IL6-luc luminescence intensity in the following figure is depicted as colours from red (high) to blue (low). (b,c) In vivo monitoring of IL6 expression in the EAE model. WIM-6 mice with WT Keap1 (WIM-6) or with Keap1-knockdown (WIM-6::Keap1-KD) on the C57BL/6 background were subjected to EAE induction. (b) Representative images of IL6-luc luminescence of WIM-6 mouse (left) and WIM-6::Keap1-KD mouse (right) at day 12 are shown. (c) Time course of the IL6-luc luminescence intensity in each mouse from WIM-6 group (red line, n=6) and WIM-6::Keap1-KD group (blue line, n=9) are shown in left panel. The maximum IL6-luc luminescence in the observation period is shown in right panel. Red and blue bars represent the means of the maximum IL6-luc luminescence from WIM-6 and WIM-6::Keap1-KD mice, respectively. (d) Clinical EAE scores in the same experiment as in c. The mean±s.d. of EAE score (left) and the maximum EAE score in the observation period (right) are shown. Note that the EAE score was significantly repressed in WIM-6::Keap1-KD mice. Red and blue bars represent the mean of the maximum EAE score. *P<0.05, unpaired t-test. (e) The illustration shows the schema of the Nrf2-mediated anti-inflammatory effect. Nrf2 binds to the proximity of inflammatory cytokine genes, including IL6 and IL1b, and inhibits their transcription. At the same time Nrf2 upregulates expression of genes coding antioxidant proteins. These antioxidant proteins eliminate ROS and subsequently contribute to the anti-inflammation.