IL-23 plays crucial roles in both immunity against pathogens and autoimmunity against self. Although it is well recognized that IL-23 expression is restricted to the myeloid lineage and is tightly regulated at the transcriptional level, the nature of transcription factors required for IL-23 expression is poorly understood. We report, in this study, that murine dendritic cells deficient in c-Rel, a member of the NF-κB family, are severely compromised in their ability to transcribe the p19 gene, one of the two genes that encode the IL-23 protein. The p19 gene promoter contains three putative NF-κB binding sites, two of which can effectively bind c-Rel as determined by chromatin immunoprecipitation and EMSA. Unexpectedly, mutation of either of these two c-Rel binding sites completely abolished the p19 promoter activity induced by five TLRs (2, 3, 4, 6, and 9) and four members of the NF-κB family (c-Rel, p65, p100, and p105). Based on these observations, we conclude that c-Rel controls IL-23 p19 gene expression through two κB sites in the p19 promoter, and propose a c-Rel-dependent enhanceosome model for p19 gene activation.
Interleukin-23 is a newly described cytokine important for both immunity against pathogens and autoimmunity against self (1). IL-23 consists of an IL-12 p40 chain and a novel IL-23 p19 chain (2, 3, 4). p19 shares sequence homology with the IL-12 p35 chain (2). The formation of a biologically active p19p40 heterodimer requires the synthesis of both subunits within the same cell. The p19p40 interaction is stabilized by an interchain disulfide bond. Similar to IL-12, IL-23 stimulates IFN-γ production by, and proliferation of, T cells. Recent studies indicate that IL-23 also induces strong proliferation of memory T cells, whereas IL-12 preferentially induces proliferation of naive T cells (2, 3, 4). Like IL-12, IL-23 also activates STAT4 in T cells (2, 3, 4). Despite these similarities, IL-12 and IL-23 appear to play different roles in vivo. Targeted gene mutations of p35, p40, and p19 result in distinct phenotypes in mice. For example, p19 mutation, but not that of p35, significantly abolished autoimmune encephalomyelitis and arthritis, whereas p40 mutation led to an intermediate phenotype (2, 3, 4). IL-23 is associated with the differentiation of Th cells into a ThIL-17 phenotype, characterized by the production of IL-17, IL-6, and TNF-α (5). In contrast, IL-12, but not IL-23, plays a major role in Th1 cell differentiation (2, 3, 4). Deficiency in either IL-23 or IL-12 significantly compromises the host’s ability to eliminate pathogens (1).
In the past decade, much has been learned about the regulation of p40 and p35 genes. It is now known that their expression is tightly regulated at the transcriptional level by several transcription factors, including NF-κB, C/EBP, ets-2, PU.1, and AP-1. However, because of the short history of the p19 research, very little is known about the regulation of p19, and for that matter, of the new cytokine IL-23. Like IL-12, IL-23 is primarily produced by dendritic cells (DCs)3 and monocytes (1). Using several complementary approaches, we show in this study that NF-κB and its binding sites in the p19 promoter are essential for TLR-induced activation of the p19 gene. This is the first characterization of the p19 promoter and its regulation by NF-κB.
The mammalian NF-κB family consists of five members: c-Rel, RelA (p65), RelB, NF-κB1 (p50/p105), and NF-κB2 (p52/p100). These proteins share a highly conserved 300-aa Rel homology domain at their NH2 termini, which encompasses sequences required for DNA binding, protein dimerization, and nuclear localization. The C termini of these proteins are less conserved, with those of c-Rel, RelA, and RelB containing transcriptional trans activation domains. All of the members of the NF-κB family are constitutively expressed in a variety of cell types as inactive homo- or heterodimeric proteins in association with the inhibitory protein IκB. A wide variety of stimuli, including TLR ligands, cytokines, Ags, and stress factors, can activate NF-κB. Activation of NF-κB involves phosphorylation and subsequent proteolytic degradation of IκB through the specific IκB kinase complex. Once activated, the free NF-κB dimers enter the nucleus, and bind to the 9- to 10-bp κB sites of gene promoters. The exact number and identity of NF-κB target genes in various cell types are not clear. For lymphoid and myeloid cells, only ∼42 NF-κB target genes have been identified by either promoter trans activation or chromatin immunoprecipitation (ChIP) (6, 7). Of these, 11 were confirmed for c-Rel and 30 were confirmed for p50. To date, at least 122 NF-κB target genes have been identified in various mammalian cell types by either ChIP or promoter trans activation, and the total number of NF-κB target genes in the mouse or human genome is estimated to be in the range of 250–300 (6, 7). Identifying the NF-κB target genes in various physiological and pathological conditions is crucial for understanding the functions of NF-κB in health and disease. We report in this study that p19 is a novel target gene of NF-κB in DCs following TLR activation.
Materials and Methods
Reagents and cell culture
LPS from Escherichia coli (O55:B5) was purchased from Sigma-Aldrich. Pam3CSK4, peptidoglycan (PGN), zymosan, and poly(I:C) were obtained from InvivoGen. The CpG oligonucleotide 1826 was obtained from IDT. The murine macrophage cell line RAW 264.7 was cultured in DMEM supplemented with 10% FCS. Bone marrow-derived DCs were generated from mouse bone marrow cells after culturing them for 7 days in RPMI 1640 containing 10% FCS, 20 ng/ml GM-CSF, 10 ng/ml IL-4 (PeproTech), 100 μg/ml streptomycin, and 10 U/ml penicillin.
Normal C57BL/6 mice were obtained from The Jackson Laboratory. c-Rel-deficient C57BL/6 mice were generated by inserting the neo resistance cassette into the fifth exon of the c-Rel gene, as described previously (8). The mice used in this study have been backcrossed to the C57BL/6 background for >10 generations. All animal procedures were preapproved by the University of Pennsylvania Animal Care and Use Committee.
The p19 promoter containing the genomic fragment −1180 to +110 of the p19 gene was amplified by PCR from C57BL/6 genomic DNA and cloned into the pGL3-basic vector (Promega). Putative NF-κB binding sites were identified using the Transcriptional Element Search System program (www.cbil.upenn.edu/tess/). Site-directed mutagenensis of identified NF-κB binding sites was performed using the QuickChange kit (Stratagene), according to manufacturer’s instructions. RAW cells were transfected with reporter constructs using FuGene-6 reagent (Roche). The luciferase activities of whole cell lysates were analyzed using the dual-luciferase reporter assay system (Promega). Cotransfection of the Renilla-luciferase expression vector pRL-TK (Promega) was used as an internal control for all reporter assays. For all samples, the data were normalized for transfection efficiency by dividing firefly luciferase activity by that of the Renilla luciferase.
A total of 5 μg of nuclear protein was used for each sample. NF-κB consensus (5′-AGTTGAGGGGACTTTCCCAGG-3′) double-stranded oligonucleotides were purchased from Santa Cruz Biotechnology. Oligonucleotides were end labeled with [γ-32P]ATP (Amersham Biosciences) using the T4 polynucleotide kinase (Promega). Binding reactions were prepared using 5 μg of nuclear extract with 50,000 cpm of oligonucleotide in a 25- μl reaction volume containing 10 mM HEPES-KOH (pH 7.9), 50 mM KCl, 2.5 mM MgCl2, 1 mM DTT, 10% glycerol, 1 μg DNase-free BSA, and 2.5 μg of poly(d(I-C)) at room temperature for 30 min. For supershift assay, Abs were added to the reaction mixture on ice for 20 min before the addition of radiolabeled probes. Binding reactions were resolved on a 4% nondenaturing polyacrylamide gel at 22 mA for 3 h at 4°C in 1× TBE (0.089 M Tris-base, 0.089 M boric acid, and 0.002 M EDTA). Gels were subsequently dried, exposed to a phosphor screen, and visualized on a phosphor imager (Amersham Biosciences).
Semiquantitative and real-time RT-PCR
Total RNA was isolated using RNeasy kits (Qiagen), according to manufacturer’s instructions. RNA was primed with oligo(dT) oligonucleotides and reversely transcribed with Moloney murine leukemia virus reverse transcriptase (BD Clontech). Semiquantitative PCR of p19 and β-actin cDNAs were performed using the following oligonucleotides: p19p19
ChIP was performed using the ChIP assay kit, per the manufacturer’s instructions (Upstate Biotechnology). Briefly, cells were fixed with 1% formaldehyde at room temperature for 10 min and lysed in lysis buffer. DNA was then fragmented by sonication. After preclearance for 1 h at 4°C with salmon sperm DNA-saturated protein A-agarose, chromatin solutions were immunoprecipitated overnight at 4°C using 1 μg of rabbit anti-mouse c-Rel IgG (Santa Cruz Biotechnology), or control rabbit IgG. Input and immunoprecipitated chromatin were incubated for 4 h at 65°C to reverse cross-links. After proteinase K digestion, DNA was extracted with phenol/chloroform and precipitated with ethanol.
To determine the identity of the c-Rel target genes, ChIP DNA was further analyzed by PCR. Genomic DNA or input DNA was used as a positive control. The following primer sets were used: IL-2 promoter (IL-2-promoter-forward (F), CATACAGAAGGCGTTCATTG; IL-2-promoter-reverse (R), TACCTGTGTGGCAGAAAGC); p19 κ1 site (p19-promoter-F1, CCCGACCTAGGCCTCTAGC; p19-promoter-R1, TTCCAGGGACGTGACATTCC); p19 κ2 site (p19-promoter-F2, ACTGTAGGCTAAGCAGGCTG; p19-promoter-R2, AAAACTTCCACTATCTTCGG). The PCR products were then analyzed on 2% agarose gels.
c-Rel is essential for TLR-induced IL-23 p19 gene expression
Previous studies have demonstrated a critical role for the transcription factor c-Rel in the inducible expression of the IL-12 p35 and p40, the latter of which can bind to IL-23 p19 to form IL-23 (1). To determine what role c-Rel may play in the regulation of IL-23 p19 expression, we examined the effect of c-Rel deficiency on p19 gene expression. Thus, bone marrow-derived DCs from wild-type (WT) and c-Rel−/− C57BL/6 mice were stimulated for 6 h with LPS. Using standard RT-PCR techniques, we observed strong p19 signals in WT DCs, but a near complete absence of p19 mRNA in c-Rel−/− DCs (Fig. 1⇓A). Quantitative p19 expression analysis was then performed using real-time PCR following stimulation of DCs with five different TLR ligands, including LPS (for TLR4), PGN (for TLR2/6), CpG oligonucleotide 1826 (for TLR9), poly(I:C) (for TLR3), and Pam3CSK4 (for TLR2). All TLR ligands tested in this study robustly induced p19 expression in WT cells, although considerable variation was evident in the levels of induction for different ligands (Fig. 1⇓B). PGN was the most potent inducer of p19 (∼615-fold over the unstimulated control), whereas poly(I:C) was the least effective (∼15-fold over the unstimulated control), a difference that may have important ramifications on the roles of IL-23 in host defenses against bacterial vs viral infections. Remarkably, c-Rel−/− DCs failed to produce high levels of p19 mRNA following stimulation with any of these TLR ligands (Fig. 1⇓B). The absence of c-Rel led to a 6-fold reduction of p19 mRNA in response to PGN stimulation, and ∼3-fold reductions in responses to LPS, CpG, Pam3CSK4, and and poly(I:C). Western blot analysis revealed that high levels of p19 protein were produced in WT, but not c-Rel−/− DCs following stimulation with either LPS or PGN, further confirming the c-Rel requirement for p19 expression (Fig. 1⇓C). These results demonstrate that c-Rel is essential for p19 gene expression in DCs following TLR stimulation, and that other transcription factors, including other members of the NF-κB family, cannot compensate for the roles of c-Rel.
TLRs effectively activate the IL-23 p19 promoter
We next examined the promoter sequence and activity of the IL-23 p19 gene. Sequence analysis of the proximal 4-kb nucleotides of the murine p19 promoter region using the Transcriptional ElementSystem Search revealed three putative NF-κB binding sites at 95, 600, and 935 bp upstream of the translational start site (Fig. 2⇓A). These sequences share close homology to those binding sites found in the promoters of genes regulated by the classical or canonical pathway of the NF-κB activation, which appear to be different from those used by the alternative or noncanonical pathway. The p19 κ2 site sequence is identical with the κB site sequence of the CD95 (Fas) promoter. Similarly, human genomic sequences upstream of the p19 gene contain two conserved putative κB sites, suggesting a shared regulatory mechanism for murine and human p19 gene (data not shown). Additionally, this region of the genomic DNA contains putative binding sites for C/EBP, AP-1, and IFN regulatory factor (IRF)-1 transcription factors (Fig. 2⇓).
To evaluate the activity of the putative p19 promoter region, we next generated a luciferase reporter plasmid designated pLucp19. The reporter plasmid carries the murine genomic DNA fragment 1180 bp upstream and 110 bp downstream of the p19 translation start site containing all three putative κB sites, which is fused to the firefly luciferase gene (Fig. 2⇑B). Transient transfection of RAW 264.7 cells (a murine macrophage cell line) with this reporter plasmid demonstrated efficient induction of the reporter activity in response to stimulation with LPS, PGN, CpG oligonucleotide 1826, and poly(I:C). Interestingly, the relative levels of the p19 promoter activity induced by different TLRs in RAW cells differed from those seen in DCs (Fig. 2⇑C). This difference may be due to variations in TLR signal transduction between these cell types or may reflect a divergence in the regulation of episomal vs genomic DNA promoter regions. Nonetheless, the inducibility of the pLucp19 reporter by TLRs demonstrates its suitability for the study of p19 promoter activity.
Two proximal NF-κB sites in the IL-23 p19 promoter are both required for TLR-induced IL-23 p19 expression
Our observation that c-Rel−/− DCs produce little or no p19 indicates that c-Rel is required for the activation of the p19 promoter. To directly test this theory, we transiently cotransfected RAW cells with the pLucp19 reporter plasmid and increasing amounts of a c-Rel expression vector. Subsequent analysis of the luciferase activity in transfected cells demonstrated a clear dose-dependent induction of reporter activity by c-Rel (Fig. 3⇓A). The near complete abrogation of p19 gene expression in c-Rel−/− DCs as shown in Fig. 1⇑ is striking because c-Rel is only one of the five members of the NF-κB family. To determine whether p19 promoter can also be activated by other members of the NF-κB family, we transfected RAW cells with expression vectors for p65, NF-κB1 (p105), NF-κB2 (p100), and RelB. We found that in addition to c-Rel, p65, p105, and p100, but not RelB, significantly induced the p19 reporter activity (Fig. 3⇓B). Cotransfection with c-Rel and another member of the NF-κB family revealed that only p105 could significantly synergize with c-Rel to increase the reporter activity (Fig. 3⇓C), indicating that the c-Rel:p50 heterodimer might be the most effective c-Rel containing transcriptional activator of p19 gene. c-Rel:p65 dimer may also be effective in activating p19 promoter, although it may be less so than the p65 homodimer.
To directly test the role of the three putative NF-κB sites in the p19 promoter region, we next generated pLucp19 reporter constructs in which one of the three sites was mutated (Fig. 3⇑D). RAW cells were transfected with the control pLucp19 plasmid or the pLucp19 plasmid containing mutations in one of the three NF-κB binding sites (κ1, −95 bp; κ2, −600 bp; κ3, −935 bp) and then stimulated with 100 ng/ml LPS for 8 h. Unexpectedly, we found that both NF-κB binding sites located at −95 and −600 bp, designated κ1 and κ2, respectively, were essential for LPS-induced p19 reporter activity, whereas the NF-κB site most distal from the transcriptional start site, κ3 (−935 bp), was dispensable (Fig. 3⇑E). Similar results were observed in the promoter trans activation assay in which p19 promoter was activated by enforced expression of c-Rel, p65, p105, or p100 (Fig. 3⇑F). Mutations in either the κ1 or κ2 site of the p19 promoter completely abrogated the promoter activity. The inability of LPS or NF-κB members to activate the pLucp19 reporter when either the κ1 or κ2 site is mutated reveals a critical, nonredundant role for these sites in TLR-induced, NF-κB-mediated p19 gene expression.
c-Rel binds to the κ1 and κ2 sites of the IL-23 p19 promoter following TLR activation
To establish whether c-Rel physically interacts with κ1 and κ2 sites of the p19 promoter, we used two independent approaches. The first approach is the EMSA, which enables the visualization of protein:DNA interaction in vitro. Thus, nuclear extracts from LPS-treated RAW cells were first incubated with radiolabeled oligonucleotides containing the κ1 and κ2 sites and then resolved on a nondenaturing polyacrylamide gel. As Fig. 4⇓A shows, both κ1 and κ2 sites displayed significant NF-κB-binding activities, with the κ1 site demonstrating a higher binding activity than κ2. To examine whether the binding activity of these two sites is mediated by NF-κB, we performed a supershift assay, using Abs specific for c-Rel, p65, and p50. This supershift assay revealed a c-Rel-binding component in the complexes associated with both κB sites. Comparison of the supershifted complexes between samples incubated with anti-c-Rel Ab and Abs directed against p65 or p50 indicates that the κ1- and κ2-containing oligonucleotides also bind p50- and p65-containing complexes (Fig. 4⇓A).
Although these results indicate that NF-κB is capable of binding oligonucleotides containing the κ1 or κ2 sites of the p19 promoter, they do not prove that this interaction occurs in vivo, i.e., inside the cell. To address this issue, a second approach, which is based on ChIP, was used. ChIP is the most direct way for identifying target genes of transcription factors because it measures the in vivo interaction of endogenous transcription factors with their target promoters. Thus, WT and c-Rel−/− DCs were cultured with 100 ng/ml LPS for 6 h and ChIP was performed using a c-Rel-specific Ab. PCR amplification of the resulting immunoprecipitates using primers flanking the κ1 and κ2 sites of the p19 promoter region revealed c-Rel-binding activity at both of these sites in WT DCs (Fig. 4⇑B). As expected, no binding was detected in c-Rel−/− DCs, serving as a control for the specificity of the ChIP.
In parallel experiments, we also examined c-Rel binding to p19 promoter by ChIP in splenocytes of mice treated with LPS i.v. As shown in Fig. 5⇓A, little c-Rel was detected in the nuclear extracts of splenocytes before the treatment. The c-Rel nuclear level was dramatically increased following a single injection of LPS. The c-Rel nuclear signal was reduced to the background level 15 h after the treatment, reflecting the short duration of the LPS effect. To identify c-Rel target genes in this system, PCR was performed following ChIP with a c-Rel-specific Ab using primers flanking c-Rel binding sites in candidate target genes. IL-2, a known target gene of c-Rel, was used as a positive control. As shown in Fig. 5⇓B, both p19 and IL-2 promoter DNAs were specifically precipitated by anti-c-Rel Ab, indicating that they are specific targets of c-Rel. Taken together, these results establish that c-Rel binds to the p19 promoter both in vivo and in vitro.
Results reported in this study establish that p19 gene expression is absolutely dependent on c-Rel and its two κB binding sites, and that other transcription factors, including other members of the NF-κB family, are unable to compensate for the c-Rel function. These results raise the following basic questions: 1) How is c-Rel specificity determined at the NF-κB binding sites? 2) Why are two c-Rel binding sites required for the expression of p19 gene in response to TLR signaling?
The absolute requirement for c-Rel binding at two distinct sites of the p19 promoter prompted us to propose a c-Rel-dependent enhanceosome model of p19 gene transcription. The enhanceosome model states that gene transcription is the result of multiple transcription factors and cofactors occupying and functionally interacting on the promoter/enhancer region of the gene (9). It argues that a specific gene would be selected for activation only if all enhanceosome components are present in the same nucleus (9). This model may explain cell- and signal-specific gene regulation, as the presence of factors common to multiple cell types and/or signaling pathways is insufficient to drive gene expression. With regard to p19, although multiple transcription factors may bind to its promoter (the p19 promoter contains putative binding sites for several transcription factors, including AP-1, C/EBP, and IRF proteins), it is not activated unless c-Rel binds to both of its κB sites. Therefore, two distinct and nonredundant interactions between c-Rel and p19 κB sites are absolutely required for p19 activation. This model also argues that the functional interactions between other transcription factors and NF-κB required for transcription of the p19 gene may only be provided by c-Rel-containing NF-κB complexes. Indeed, our EMSA analysis of the κ1 and κ2 binding sites suggests a preference toward particular c-Rel-containing complexes at both of these sites, indicating that subunit specificity may be encoded in the NF-κB binding sites and the flanking sequences.
Such sequence specificity has also been noted in recent studies using cells deficient in one or more members of the NF-κB family. Besides dictating which NF-κB dimer binds with the highest affinity, the sequence of the κB sites may affect with which coactivator NF-κB complexes can functionally interact. Grumont et al. (10) and Sanjabi et al. (11) reported that unlike other members of the NF-κB family, c-Rel is essential for the activation of the IL-12 p35 and p40 promoters in DCs and macrophages, respectively. Thus, c-Rel may be a specific transcriptional regulator of both IL-12 and IL-23. Other studies using NF-κB-deficient mice also indicate that the roles of NF-κB in vivo may be member specific. Thus, mice deficient in c-Rel develop normally and acquire a structurally normal immune system (8, 12). However, T cells derived from these mice fail to respond to activation signals mediated by the TCR, and produce much reduced levels of IL-2, IL-3, IFN-γ, and GM-CSF. In addition, the induction of IL-2Rα is also decreased, whereas apoptosis is increased in c-Rel-deficient cells. By contrast, mice deficient in RelA die in utero, presumably due to enhanced hepatocyte apoptosis. RelB-deficient mice develop normally, but suffer from severe disorders ranging from splenomegaly to chronic microbial infections. Similarly, NF-κB1 (p50)-deficient mice do not have any developmental defects, but are more susceptible to intracellular and extracellular Gram-positive bacterial infections, and are partially compromised in their B cell responses to LPS (13). Surprisingly, they are resistant to viral and Gram-negative bacterial infections (13). In contrast, NF-κB2-deficient mice suffer from severe developmental defects. Both their spleens and lymph nodes are bereft of B lymphocytes, undermining their capacity to form germinal centers. These results have led to the recognition that members of the NF-κB family perform nonoverlapping functions, and that a loss-of-function mutation of a NF-κB gene cannot be fully compensated for by other members of this family.
In addition to its well-documented roles in immunity against pathogens, recent studies from several laboratories, including ours, indicate that NF-κB plays indispensable roles in the development of autoimmune diseases (14, 15). In both mice and humans, development of type 1 diabetes is associated with heightened NF-κB activation in DCs, monocytes, and tissues infiltrated by these cells (16, 17). Conversely, inhibiting NF-κB activity is effective in suppressing type 1 diabetes in NOD mice (18), CD1 mice (19), and C57BL/6 mice (20). Additionally, modulating NF-κB activity in islet β cells significantly affects their viability in vitro (21). To directly test the roles of NF-κB in the development of autoimmune diseases, we and others have recently studied autoimmune diabetes, arthritis, and encephalomyelitis in c-Rel- and NF-κB1-deficient mice (14, 22, 23, 24). These studies have led to the discovery that both c-Rel and NF-κB1 play indispensable roles in the development of autoimmune diseases. Because IL-23 is crucial for the development of autoimmune diseases, our demonstration that its production in c-Rel-deficient mice is blocked indicates that it may be a major mediator of the c-Rel effect in these diseases. Therefore, strategies targeting the c-Rel-IL-23 axis should be effective for the treatment of these disorders.
We thank Dr. Brendan Hilliard, Dr. Christopher Hunter, Dr. Li Li, and Stacey Garrett for valuable discussions and technical support.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work was supported by National Institutes of Health Grants AI50059, AI055934, and AI55934.
↵2 Address correspondence and reprint requests to Dr. Youhai H. Chen, Department of Pathology and Laboratory Medicine, University of Pennsylvania School of Medicine, 421 Curie Boulevard, 614 Biomedical Research Building II/III, Philadelphia, PA 19104. E-mail address:
↵3 Abbreviations used in this paper: DC, dendritic cell; ChIP, chromatin immunoprecipitation; IRF, IFN regulatory factor; PGN, peptidoglycan; WT, wild type; F, forward; R, reverse.
- Received June 14, 2006.
- Accepted October 24, 2006.
- Copyright © 2007 by The American Association of Immunologists