Nuclear receptor ERR alpha and coactivator PGC-1 beta are effectors of IFN-gamma-induced host defense

Abstract

Macrophage activation by the proinflammatory cytokine interferon-gamma (IFN-gamma) is a critical component of the host innate response to bacterial pathogenesis. However, the precise nature of the IFN-gamma-induced activation pathway is not known. Here we show using genome-wide expression and chromatin-binding profiling that IFN-gamma induces the expression of many nuclear genes encoding mitochondrial respiratory chain machinery via activation of the nuclear receptor ERR alpha (estrogen-related receptor alpha, NR3B1). Studies with macrophages lacking ERR alpha demonstrate that it is required for induction of mitochondrial reactive oxygen species (ROS) production and efficient clearance of Listeria monocytogenes (LM) in response to IFN-gamma. As a result, mice lacking ERR alpha are susceptible to LM infection, a phenotype that is localized to bone marrow-derived cells. Furthermore, we found that IFN-gamma-induced activation of ERR alpha depends on coactivator PGC-1 beta (peroxisome proliferator-activated receptor gamma coactivator-1 beta), which appears to be a direct target for the IFN-gamma/STAT-1 signaling cascade. Thus, ERR alpha and PGC-1 beta act together as a key effector of IFN-gamma-induced mitochondrial ROS production and host defense.

Figure 1.

ERRα-dependent induction of nuclear-encoded mitochondrial genes by IFN-γ or IL-4. (A) GO pathways whose expression were significantly different between wild type and ERRα KO and the number of genes identified for each pathway. (B) Functional grouping of 86 unique genes whose expression are suppressed both in IFN-γ- or IL-4-treated ERRα KO macrophages compared with wild-type macrophages. (C) The relative expression levels of 86 genes identified in B are indicated by representative colors as shown at the bottom. The assigned function for each feature is indicated at right by color as shown in B. (D) Q-PCR mRNA expression analysis in macrophages (n = 4) treated with the indicated cytokines for 12 h. Aco2, Idh3a, Atp5b, and Ndg2 encode mitochondrial proteins. (E) ERRα targets identified by genome-wide expression and chromatin-binding profiling. Functional locations (nucleus, plasma membrane, cytoplasm, or mitochondria) of the identified gene products are schematically shown. (Blue) Common between the expression profiling of IFN-γ- and IL-4-treated cells (B); (green) common between expression profiling and chromatin-binding profiling of ERRα in IFN-γ-treated cells (Supplementary Fig. S8); (red) common for both of these two criteria. (MRP) Mitochondrial ribosomal proteins; (PDH) pyruvate dehydrogenase complex; (TIMM) translocase of inner mitochondria membrane.

Figure 2.

IFN-γ induces mitochondrial ROS production and LM clearance in macrophages through ERRα. (A, left) Cellular O₂ consumption in digitonin-permeabilized macrophages (n = 6) cultured with or without IFN-γ for 36 h. (*) P < 0.002 (wild type vs. ERRα KO); (#) P < 0.004 (IFN-γ vs. mock-treated group). (Right) A representative oxygraph pattern. (Arrows) Addition of substrates (succinate and glutamate) or the inhibitor (KCN). (B) Macrophages were cultured with or without IFN-γ for 36 h, then cultured in medium containing Rhodamine-123 (R123) for 30 min, harvested, and subjected to FACS analysis, to determine the mitochondrial electron potential. The results represent relative median values of the fluorescence (n = 6). (*) P < 0.005 (wild type vs. ERRα KO); (#) P < 5E-6; (##) P < 1E-8 (IFN-γ vs. mock-treated group). (C) Macrophages were cultured with or without IFN-γ for 36 h, then cultured in medium containing DCF for 30 min, harvested, and subjected to FACS analysis to determine the intracellular ROS level. The results represent relative median values of the fluorescence (n = 3. (*) P < 0.01 (wild type vs. ERRα KO); (#) P < 0.001; (##) P < 0.05 (IFN-γ vs, mock-treated group). (D) The rate of DCF oxidation was determined by adding DCF (t = 0) to suspensions of IFN-γ-activated macrophages followed by FACS analysis at the indicated time points. Median fluorescence was plotted for each measurement. (E) A similar experiment as C, except IgG-conjugated DCF was used to measure the intracellular ROS level generated by NADPH oxidase. (F) The effects of rotenone (50 nM) on the intracellular ROS level in IFN-γ-activated macrophages. (*) P < 0.01. (G–I) The ROS level (F), viability (G), and total cellular ATP content (H), of IFN-γ-activated macrophages after 6 h of treatment with no drug (M), with 50 nM rotenone (R), or with 1 mM KCN (K). (*) P < 3E-05; (**) P < 3E-04; (***) P < 0.03. (J) Total cellular ATP content in IFN-γ-activated macrophages. (*) P < 0.03 (wild type vs. KO); (#) P < 0.03 (mock vs. IFN-γ-treated). (K) In vitro LM infection of resting and activated macrophages. (*) P < 0.02; (**) P < 0.0003 (wild type vs. KO). (L) A similar in vitro LM infection of activated macrophages. After infection, IFN-γ-pretreated cells were incubated for 6 h without mitochondrial inhibitor (M), with 50 nM rotenone (R), or with 1 mM KCN (K). (*) P < 0.05; (**) P < 0.01.

Figure 3.

ERRα KO mice are defective for LM clearance. (A,B) Bacteria counts of the whole liver and gallbladder (n = 8) (A) and the spleen (n = 6) (B) at day 2 post-2 × 10⁴ LM i.v. injection. (C) Spleen weights at day 10 post-1 × 10⁴ LM i.v. injection (n = 8). (D) H&E straining of spleens after 1 × 10⁴ LM infection. (Arrowheads) The PALS structures. (E) Serum IFN-γ concentrations at day 2 post-infection (2 × 10⁴ cfu). No significant difference between wild type and KO. (#) P < 0.05 (−LM vs. +LM). (F) Q-PCR analysis of peritoneal exudates cells (n = 6) at day 4 post-LM infection (1 × 10⁶ cfu). (*) P < 0.02; (**) P < 0.04; (***) P < 0.0004. Note that expression of ERRα target genes, but not of iNOS, is significantly reduced in the KO cells. (G) Kaplan-Meier survival plot after i.v. injection of the indicated dose of LM. (H) Bacteria counts of the whole liver and gallbladder (left) and the spleen (right) (n = 10) at day 2 post-2.4 × 10⁴ LM i.v. injection. (I) Kaplan-Meier survival plot for bone marrow-transplanted mice after 2 × 10⁵ LM i.v. injection. P values shown are of KO + KO BM to the indicated group.

Figure 4.

PGC-1β expression is regulated by IFN-γ and correlates with ERRα activation. (A) Q-PCR mRNA expression analysis in macrophages (n = 4) treated with the indicated cytokines for 12 h. (B) Transient transfection reporter assay of murine macrophage cell line RAW264. Schematic diagrams for the experiments are shown at the top. (Left) Cells were cotransfected with an expression vector for mouse PGC-1β and/or mouse ERRα, together with a luciferase reporter harboring three copies of ERRα-binding site from the murine Idh3a gene. (Right) Cells were cotransfected with an expression vector for mouse PGC-1β and/or human ERRα ligand-binding domain fused to GAL4 DNA-binding domain (GAL-ERRαLBD), together with a luciferase reporter harboring three copies of GAL4-binding site (UAS). The results represent the average of triplicate experiments for luciferase activity normalized by β-gal activity. (C) Schematic diagram of murine PGC-1β promoter. (GAS) IFN-γ activation sequence; (4AS) IL-4 activation sequence. (D) ChIP assay using anti-STAT-1 or anti-acetylated histone H3 antibody in IFN-γ-activated macrophages. (E) EMSA was used to test specific binding of IFN-γ- or IL-4-induced nuclear activities to the potential GAS and 4AS from the PGC-1β promoter. Supershift with anti-STAT-1 antibody shows that the IFN-γ-induced binding complex contains STAT-1. (F) Transient transfection reporter assay of RAW264 cells with luciferase reporter driven by three copies of GAS and 4AS sites. After the transfection, cells were treated with the indicated cytokines for 12 h. The results represent the average of triplicate experiments for normalized luciferase activity.

Figure 5.

PGC-1β KO mice are defective for LM clearance. (A) Northern blot (left) and Q-PCR (right) mRNA expression analyses in PGC-1β-deficient macrophages activated with the indicated cytokines. (Left) cDNA encoding a part of exon 5 (top) or entire PGC-1β (middle) was used as probe. (GAPDH) Loading control. (B) The relative expression levels of 86 genes identified in Figure 1B in wild-type and PGC-1β KO cells are indicated by colors as shown at the bottom. (C,D) IFN-γ-activated macrophages were cultured in medium containing Rhodamine-123 (C) or DCF (D) for 30 min, harvested, and subjected to FACS analysis to determine the mitochondrial membrane potential (C) and the intracellular ROS level (D), respectively. The results represent relative median values of the fluorescence (n = 4). (*) P < 0.05 (wild type vs. ERRα KO); (#) P < 0.05 (IFN-γ vs. mock-treated group). (E) In vitro LM infection of activated macrophages. (*) P < 0.05 (wild type vs. KO). (F) Bacteria counts of the whole liver and gallbladder (left) and the spleen (right) (n = 6–8) at day 3 post-2 × 10⁶ LM i.p. injection. (G) Kaplan-Meier survival plot after injection of the indicated dose of LM. (H) A model for the role of mitochondria ROS in pathogen resistance. Disruption of ERRα impairs inducible mitochondrial ROS production and pathogen resistance (this study), whereas disruption of mitochondrial uncoupling protein (UCP2), whose function is to reduce mitochondrial ROS (Echtay et al. 2002), results in constitutive pathogen resistance (Arsenijevic et al. 2000; Rousset et al. 2006).