Gene-specific Requirement of a Nuclear Protein, IκB-ζ, for Promoter Association of Inflammatory Transcription Regulators*

Expression of many inflammatory genes is induced through activation of the transcription factor NF-κB. In contrast to the advanced understanding of cytoplasmic control of NF-κB activation, its regulation in the nucleus has not been fully understood despite its importance in selective gene expression. We previously identified an inducible nuclear protein, IκB-ζ, and demonstrated that this molecule is indispensable for the expression of a group of NF-κB-regulated genes. In this study, we established a unique gene induction system, in which IκB-ζ is expressed independently of inflammatory stimuli, to specifically investigate the molecular basis underlying IκB-ζ-mediated gene activation. We show that in the presence of IκB-ζ other primary response genes are dispensable for the expression of the target secondary response genes. ChIP analyses revealed that IκB-ζ is required for stimulus-induced recruitment of NF-κB onto the target promoter in a gene-specific manner. Surprisingly, IκB-ζ is also necessary for the gene-selective promoter recruitment of another inflammatory transcription factor, C/EBPβ, and the chromatin remodeling factor Brg1. We propose a new gene regulatory mechanism underlying the selective expression of inflammatory genes.

the cytoplasm by its inhibitors IB-␣, -␤, and -⑀. Upon receptor stimulation, dozens of receptor-associated molecules induce activation of the IB kinase, resulting in degradation of the IBs by the ubiquitin-proteasome pathway. NF-B liberated from IBs is translocated into the nucleus, where it participates in the transcriptional activation of target genes. As the cytoplasmic activation of NF-B does not require new protein synthesis, this transcription factor is capable of inducing target inflammatory genes within a short period.
Nonetheless, the induction of NF-B-regulated genes is not solely defined by the nuclear entry of NF-B. Growing evidence suggests that each NF-B-regulated gene has its own expression profile (for example, kinetics, cell type/stimulusspecificity and requirements of regulators), indicating the importance of gene-specific regulation after the nuclear translocation of NF-B. Expression of individual inflammatory genes with their own profiles is considered to be crucial for the integrity of particular inflammatory responses. In contrast to the advanced comprehension of the regulatory machinery of NF-B in the cytoplasm, the regulation of NF-B in the nucleus is not fully understood (4,5).
Recently, it was proposed that inflammatory genes could be categorized into two groups based on their requirement of de novo protein synthesis for induction (6,7). While primary response genes are rapidly induced independently of new protein synthesis, induction of secondary response genes takes longer and is impeded in the presence of protein synthesis inhibitors. Thus, the expression of at least one primary response gene is assumed to be necessary for the induction of a set of secondary response genes.
We previously cloned an inducible nuclear protein, IB-, by screening the genes rapidly induced in response to a cell wall component of Gram-negative bacteria, lipopolysaccharide (LPS), in macrophages (8). Other two groups have reported the identical molecule through similar screening of inducible genes upon inflammatory stimulation (9,10). Expression of IBis barely detectable in unstimulated cells, but is robustly induced upon inflammatory stimulation. IBharbors six ankyrin repeats, which are most similar to those of other nuclear IB proteins, Bcl-3 and IBNS. Similar to them (11,12), IB-pref-erentially binds to the NF-B p50 subunit (8,13). Subsequent analyses revealed that IBis indispensable for the expression of a group of NF-B-regulated genes (13). IB-is encoded by a primary response gene, Nfkbiz, and its induction depends on NF-B activation (14 -16), suggesting that IB--regulated genes are induced via a two-step machinery. In fibroblast cells, IBwas induced in response to LPS and IL-1␤, but not TNF-␣ (8). We have shown that TNF-␣ induced the transcription of the Nfkbiz gene but did not stabilize IB-mRNA, indicating that the stimulus-specific expression of IBis determined post-transcriptionally (15).
It is still unknown how IBis involved in the transcriptional activation of its target genes. First, the requirement of NF-B for the expression of IB-itself makes it difficult to directly define whether or not NF-B is indeed involved in IB--mediated gene activation. In the present study, we devised a gene expression system in which IBis induced independently of NF-B activation. This system is composed of zincinduced IBexpression and cellular activation by TNF-␣ stimulation, the latter of which activates NF-B, but does not induce IBexpression. Using this system, we extended the characterization of IB--mediated gene regulation and identified components that are essential for this process. Importantly, our ChIP experiments revealed the gene-specific requirement of IBfor the promoter recruitment of NF-B and other transcription regulators involved in inflammatory responses. These findings suggest that IBplays a critical role in the selective gene expression by controlling the association of key transcription regulators with target promoters in the nucleus.
Construction of Plasmids and Stable Cell Lines-The DNA fragment containing the sheep metallothionein Ia (sMT-Ia) promoter (Ϫ600/ϩ72) was obtained by PCR using sheep genomic DNA as template. The CMV promoter region in pcDNA3 (Invitrogen) was replaced with the sMT-Ia promoter using the restriction sites BglII and HindIII. The IBfragment of coding region was inserted into the resultant vector. NIH3T3 cells were transfected using FuGENE6 according to the manufacturer's instructions (Roche Applied Science), and cells were selected in the presence of 1 mg/ml G418 (Nakalai) to obtain stably transfected cells.
Quantitation of mRNA Levels by Real Time PCR-Total RNA isolated using TRIsure (BIOLINE, London, UK) was reverse-transcribed by ReverTra Ace (TOYOBO, Osaka, Japan). The cDNA was analyzed by quantitative real-time PCR (RT-qPCR) using SYBR Premix Ex Taq (TaKaRa Bio, Shiga, Japan) on the Roter-Gene 6200 system (Corbett Life Science, Sydney, Australia). The primer sequences used for RT-qPCR analyses are available upon request.
Immunoblotting of Nuclear and Total Cellular Proteins-Nuclear protein was extracted as described (6,16). After cells were lysed with the buffer containing 10 mM HEPES (pH 8.0), 1.5 mM MgCl 2 , 10 mM KCl, 0.5% Nonidet P-40, and 1 mM dithiothreitol, on ice for 10 min, nuclei were pelleted by centrifugation at 2,500 ϫ g for 5 min at 4°C. Nuclear proteins were extracted with the buffer containing 10 mM HEPES (pH 8.0), 25% glycerol, 0.6 M KCl, 1.5 mM MgCl 2 , and 0.2 mM EDTA, on ice for 30 min and separated from debris by centrifugation at 15,000 ϫ g for 5 min. The immunoblot was performed as described previously (8).
ChIP (Chromatin Immunoprecipitation) Assay-ChIP was performed as previously described (16,19). Cells were fixed with formaldehyde (1%) at room temperature for 10 min and washed with ice-cold phosphate-buffered saline, twice. After a sonicated chromatin solution was pre-cleared with protein G-Sepharose 4 Fast Flow (GE Healthcare, Buckinghamshire, UK) at 4°C for 2 h, the antibody of interest (1.5 g) was added and incubated at 4°C overnight with continuous rotation. Antigen-antibody complexes captured by protein G-Sepharose were extensively washed, and DNA-protein complexes were eluted with elution buffer containing 1% SDS and 100 mM sodium bicarbonate. After cross-links were reversed by incubation at 65°C overnight, proteins were digested with proteinase K (0.1 mg/ml) at 45°C 3 h. 1/25 th of the DNA preparation was used for PCR analysis.
Luciferase Reporter Assay-Cells were transfected with a luciferase reporter plasmid and the internal control plasmid pRL-TK (Promega). Two days after transfection, cells were stimulated, and luciferase activities were determined by the Dual-Luciferase Reporter assay system (Promega). A, NIH3T3 cells were transiently transfected with the luciferase reporter plasmid containing the indicated promoters. Two days after transfection, cells were stimulated with ZnSO 4 (50 M) or LPS (100 ng/ml) for the indicated periods, and luciferase activities were measured. Luciferase activities were normalized to the control Renilla luciferase activities. B, NIH3T3 cells stably transfected with the indicated plasmids were stimulated with zinc for the indicated periods. Total cellular protein was subjected to immunoblotting using the indicated antibodies. Tg, transgene. C, NIH3T3 cells harboring sMT-Ia-IB-were stimulated with ZnSO 4 (50 M), TNF-␣ (10 ng/ml), or IL-1␤ (10 ng/ml) for the indicated periods, and total cellular protein was analyzed by immunoblotting. The results are representative of at least two independent experiments.

Retroviral Infection and Small Interfering RNA (siRNA)
Experiments-Gene transduction via retroviral infection was carried out as previously described (17). Supernatant of 293T cells transfected with Moloney Murine Leukemia virus-⌿A helper plasmid and pBABEpuro construct was used for the infection. Transfection of cells with siRNA was performed using Lipofectamine2000 as previously described (18) according to the manufacturer's instructions (Invitrogen). Duplexed modified RNA oligonucleotides (Stealth RNAi) were obtained from Invitrogen. The sequences of the sense strands of the siRNAs are as follows: 5Ј-GAGAAUGGACAGAACAGCAG-GAUGU-3Ј for the mouse NF-B p50 subunit 1, 5Ј-GCAGAU-GGCCCAUACCUUCAAAUAU-3Ј for the mouse NF-B p50 subunit 2, 5Ј-GCAGUAUCCAUAGCUUCCAGAACCU-3Ј for the mouse NF-B p65 subunit 1 and 5Ј-UCCCAAUGGUCUC-UCCGGAGAUGAA-3Ј for the mouse NF-B p65 subunit 2.
Stealth RNAi negative control duplexes (Invitrogen) were used as controls. Two days after transfection, cells were stimulated and subjected to RT-qPCR or immunoblotting.

RESULTS
A Gene Expression System for IB--regulated Genes-We attempted to construct a gene induction system, in which IBis expressed independently of NF-B activation. Because the basal level of IBis very low, we chose an inducible expression system to avoid undesirable artificial effects caused by its constitutive overexpression. The sheep metallothionein Ia (sMT-Ia) promoter is reportedly activated in response to zinc treatment exclusively (20,21). The sMT-Ia promoter was activated when NIH3T3 fibroblast cells were treated with zinc, but was unresponsive to LPS stimulation (Fig. 1A). By contrast, the control ELAM1 reporter selectively responded to LPS. We generated a transgene, in which the coding region of IBwas placed downstream of the sMT-Ia promoter, and stably introduced it into NIH3T3 cells. Treatment of the transfected cells with zinc induced IBprotein with similar kinetics to the induction of endogenous IBprotein in response to IL-1␤ and LPS (Fig. 1, B and C, Ref. 15). As we have shown previously, TNF-␣ stimulation minimally induced IBprotein (Fig. 1C, Refs. 8 and 15).
The expression of Lcn2 (Lipocalin-2, 24p3), Csf3 (G-CSF), and Cebpd (C/EBP␦) is highly dependent on IB-in MEFs ( Fig.  2A). In the cells harboring the sMT-Ia-IB-transgene, these IB--dependent genes were not significantly induced in response to either TNF-␣ or zinc alone, but co-treatment of the cells with TNF-␣ plus zinc elicited their robust expression (Fig.  2B, left panels). These genes were not expressed in control cells (Fig. 2B, right panels). By contrast, the primary response genes Tnfaip3 (A20), Cxcl1 (KC) (Fig. 2B), and Cxcl2 (MIP-2) (data not shown) were induced in response to TNF-␣ alone, and their expression was largely unaffected by the addition of zinc. The zinc-induced expression of IB-protein and TNF-␣-induced nuclear translocation of p65 were confirmed by immunoblotting (Fig. 2C).
IB-Is the Only Primary Response Gene Necessary for Its Target Gene Induction-IB--regulated genes are secondary response genes, because their expression was impaired in the FIGURE 2. A gene expression system for focused analyses of IB--mediated gene regulation. A, primary MEFs prepared from the littermate embryos were stimulated with LPS (100 ng/ml) for the indicated periods. Total RNA was analyzed by RT-qPCR, and mRNA expression levels were normalized to those of Rpl32 (L32). Similar results were obtained in immortalized MEFs. B, NIH3T3 cells stably transfected with sMT-Ia-IB-or sMT-Ia-GFP were stimulated with TNF-␣ (5 ng/ml), ZnSO 4 (50 M), or TNF-␣ plus ZnSO 4 for the indicated periods. Total RNA was extracted and the expression levels of the indicated genes were determined by RT-qPCR. mRNA expression levels were normalized to those of Rpl32. C, NIH3T3 cells harboring sMT-Ia-IBwere stimulated as indicated, and nuclear protein was analyzed by immunoblotting using the indicated antibodies. USF2 is a control nuclear protein that was used as a loading control. The results are representative of at least three independent experiments.
presence of protein synthesis inhibitors (Ref. 16 and Fig. 3A). We attempted to determine whether new synthesis of any primary response genes other than IBis also necessary for the expression of IB--regulated genes (Fig. 3B). Induction of Lcn2 and Csf3 in response to TNF-␣ plus zinc was abrogated in the presence of the protein synthesis inhibitor cycloheximide (CHX), while expression of the primary response genes Tnfaip3 and Cxcl1 was strongly augmented by CHX (Fig. 3B, columns 4 and 5) due to an mRNA-stabilizing effect (16). Similar to the results seen after simultaneous treatment of cells with TNF-␣ plus zinc, Lcn2 and Csf3 were induced when cells were pretreated with zinc and then stimulated with TNF-␣ (Fig. 3B,  columns 4 and 6). Importantly, when the cells were pretreated with zinc, inhibition of new protein synthesis by CHX treatment did not affect induction of IB--regulated genes by subsequent TNF-␣ stimulation (Fig. 3B, columns 5 and 7). Thus, we conclude that IBis the only primary response gene neces-sary for expression of the target secondary response genes tested.
Required Components in IB--mediated Gene Induction-Next, we verified the requirement of NF-B for induction of IB--regulated genes. Blockade of NF-B activation by expression of a super-repressor mutant of IB-␣ (IB-␣(SR)) resulted in impaired induction of Lcn2 and Csf3, as well as that of primary response genes, following stimulation of cells with TNF-␣ plus zinc (Fig. 4A). The expression of IB-␣(SR) did not affect zinc-induced IBexpression (Fig. 4B).
To better understand the molecular machinery involved, we examined the requirement of individual NF-B subunits using siRNAs (Fig. 5). While both p65 and p50 subunits were necessary for maximal induction of Lcn2, another IB--regulated gene, Csf3, required only p50 for its induction when cells were treated with TNF-␣ plus zinc (Fig. 5, A and C). Induction of Tnfiap3 was impaired by p65 siRNA, but not by p50 siRNA. In contrast to the results of stimulation with TNF-␣ plus zinc, the induction of Csf3 in response to LPS required the p65 subunit, which could be explained by a requirement of p65 for IBinduction in response to LPS (Fig. 5E). The efficacy of siRNAmediated gene knockdown was confirmed by immunoblotting (Fig. 5, B and D). It has been reported that Helenalin, a sesquiterpene lactone, is an alkylating agent specific for NF-B p65 subunit while it does not modify the DNA binding of p50 (22). To validate the gene-specific requirement of p65 subunit, we examined sensitivity of the Lcn2 and Csf3 expression to Helenalin. Induction of Lcn2 in response to TNF-␣ plus zinc, but not that of Csf3, was specifically suppressed in the presence of Helenalin (Fig. 5F).
We have recently shown an involvement of C/EBP family proteins in addition to NF-B in IB--mediated activation of target promoters (18). To explore the involvement of C/EBP proteins in IB--mediated gene induction, we tested the effect of expression of a dominant-negative isoform of C/EBP␤, LIP (Fig. 6A). As LIP was supposed to form non-functional heterodimers with other C/EBP family members (23,24), expression of LIP seems to be effective for the functional suppression of members of this redundant protein family. The expression of LIP inhibited induction of Lcn2 and Csf3, but did not largely affect that of Tnfaip3 and Cxcl1. The expression of LIP did not affect zinc-inducible expression of IBprotein (Fig. 6B).

IB-Is Required for Promoter Recruitment of Inflammatory Transcription Regulators in a Gene-specific
Manner-To elucidate the role of IBin the nucleus, we examined the requirement of IBfor the recruitment of NF-B onto target promoters by ChIP experiments (Fig. 7). When the cells harboring the sMT-Ia-IBtransgene were stimulated with either TNF-␣ or zinc alone, the NF-B p65 subunit was not detectable at the Lcn2 promoter. By contrast, p65 was recruited to the Cxcl1 and Cxcl2 promoters upon TNF-␣ stimulation. Remarkably, co-stimulation of cells with TNF-␣ plus zinc induced p65 recruitment to the Lcn2 promoter with slow kinetics, suggesting a requirement of IBexpression for this recruitment process.
We also examined the promoter recruitment of C/EBP␤ (Fig.  7). Unlike p65, Cxcl1 and Cxcl2 promoters were not enriched with C/EBP␤ in a stimulus-dependent manner. Surprisingly, FIGURE 3. IB-is the only primary response gene indispensable for the induction of its target secondary response genes. A, induction of IB-regulated secondary response genes is impaired by a blockade of new protein synthesis. NIH3T3 cells were stimulated with LPS (100 ng/ml), CHX (10 g/ml), or LPS plus CHX for the indicated periods, and total RNA was analyzed by RT-qPCR. mRNA expression levels were normalized to those of Rpl32. B, NIH3T3 cells stably transfected with sMT-Ia-IB-were untreated or pretreated with ZnSO 4 (50 M) for 2 h, and then stimulated as indicated in the absence or presence of CHX for 4 or 6 h. Total RNA was analyzed by RT-qPCR and mRNA expression levels were normalized to those of Rpl32. The results are representative of three independent experiments. NOVEMBER 21, 2008 • VOLUME 283 • NUMBER 47 JOURNAL OF BIOLOGICAL CHEMISTRY 32407 C/EBP␤ was recruited onto the Lcn2 promoter only when cells were stimulated with TNF-␣ plus zinc, indicating that the promoter recruitment of C/EBP␤ is also dependent on IB-. It was recently reported that induction of secondary response genes is accompanied by remodeling of nucleosomes, and that this process is mediated by the two ATPase subunits of the SWI/ SNF complex, Brg1 and Brm1 (7). Intriguingly, as with p65 and C/EBP␤, stimulus-dependent Brg1 recruitment to Lcn2 promoter also requires IB- (Fig. 7).

Gene-specific Regulation by IB-in the Nucleus
The requirement of IBfor the association of these transcription regulators with specific target promoters was also investigated in LPS-stimulated macrophages. The expression of Lcn2 gene was highly dependent on the presence of IBin BMMs, and we observed a gene dose dependence (Fig. 8A). Consistent with the ChIP results obtained from the fibroblasts harboring the sMT-Ia-IBtransgene, p65, c-Rel, C/EBP␤, and Brg1 were all localized to the Lcn2 promoter only when IBis present (Fig. 8B). The recruitment of these transcription regulators to the control promoters (Cxcl1 and Cxcl2) was not dependent on IB-. Taken together, these findings suggest that IBdetermines the accessibility of target promoters to these transcription regulators in the nucleus.

DISCUSSION
In the present study, we established a unique expression system to specifically analyze the IB--mediated gene regulation. This system allowed us to directly define essential components involved in the process.
Overexpression of IB-␣(SR) in our system revealed that the activation of NF-B is not only required for IBinduction but also is substantially involved in the transcriptional up-regulation of the IB--regulated secondary response genes. As the NF-B p65 subunit is necessary for induction of IBexpression and IB-controls the  participation of p65 in the transcriptional up-regulation of selective target genes, these secondary response genes are not regulated by a simple two-step machinery in which distinct molecules are sequentially activated. Rather, it is likely that NF-B and IB-comprise a reciprocally limiting loop to activate the selective genes. The physiological significance of this feed-forward mechanism will be discussed below. It has been reported that the stimulus-induced recruitment of p65 to the secondary response promoters is slower than that to the primary response promoters (19). This delay could be accounted for by the time necessary for IBinduction.
Using our gene expression system, we also demonstrated the direct involvement of the C/EBP protein in activation of IB--regulated genes (Fig. 6). The ChIP experiments suggested that IBregulates accessibility of the selective promoters to C/EBP proteins as well as NF-B p65 (Figs. 7 and 8). It is noteworthy that comprehensive cDNA microarray analyses have identified common target genes between IBand C/EBP␤ in LPSstimulated macrophages (13,25). In addition, the binding sites for C/EBP proteins have been identified in the promoter regions of many IB--regulated genes, and we have previously shown the importance of these sequences in the promoter activation (18). As expression of IB-␣(SR) suppressed the association of C/EB␤ with the Lcn2 promoter, but LIP expression did not perturb p65 recruitment, 3 promoter recruitment of NF-B seems to be a prerequisite for C/EBP␤ recruitment.
Our ChIP analyses using two types of cells revealed that IBis required for recruitment of the chromatin-remodeling factor Brg1 to the selective promoter (Figs. 7 and 8). As the experiment using CHX indicated that IBis the only primary response gene necessary for the expression of its target secondary response genes (Fig. 3), the requirement of de novo protein synthesis for the Brg1-mediated nucleosome remodeling (7) could be explained by expression of IBprotein. During the preparation of this report, Kayama et al. (26) reported the involvement of MyD88dependent signaling and IBin selective expression of pro-inflammatory genes. As they showed that the promoter association of Brg1 and nucleosome remodeling at the target promoters are defective in MyD88-deficient cells but are normal in IB-deficient cells, they hypothesized at least one other MyD88-dependent molecule that is required for Brg1-mediated nucleosome remodeling. Although the exact reason for the differences between this study and the published report regarding the role of IBis unknown, there could be differences in regulation 3 S. Yamazaki, unpublished observation.  among IB--dependent genes. Furthermore, we observed a cell type-specific requirement of IBfor induction of some genes. 3 In this study, we focused on the target genes whose expression is highly dependent on IBboth in fibroblasts and macrophages. Although IL-6 production was severely impaired in IB--deficient macrophages (13), the expression of IL-6 in response to TNF-␣ was not significantly augmented by zincinduced IBas that of Lcn2 and Csf3 (data not shown). As the TNF-␣-induced IL-6 production in MEFs was shown to be unaffected by IBdeficiency (13), it is likely that IB-is not a limiting factor in this response.
Gene knockdown experiment using specific siRNAs indicated the close functional relationship between IBand the NF-B p50 subunit. We have previously shown a preferential physical interaction of IBwith p50 rather than other NF-B subunits, and an overlap in the target genes of IBand p50 (8,13). The experiments using siRNA and Helenalin revealed that while p65 was necessary for induction of Lcn2, it was dispensable for Csf3 expression (Fig. 5). It is likely that p50 and IBcould form a core element for transcriptional activation of target genes (see below), and transcriptional activity of p65 might be required for full activation of some target genes. Because p50 does not contain a transactivation domain, transcriptional activity derived from other sources should be involved in the induction of Csf3, which is possibly mediated by IBitself and C/EBP proteins (18,27). Requirement of C/EBP␤ for expression of Csf3 has been reported thus far (28).
It is still an enigma how IBregulates the promoter accessibility of the key transcription regulators in a gene-selective manner. Because IBhas no obvious DNA binding motif, it is unlikely that IBassociates with the target promoter prior to the DNA-binding proteins. It is conceivable that IBinduced upon stimulation could form a complex with the p50 homodimer, which is known to be constitutively present in the specific promoters (13,29). Then, this complex might act as a landmark for recruitment of Brg1, p65, and C/EBP. IBcould stabilize or assist the promoter binding of these transcription regulators by an unknown mechanism. Despite a careful comparison among the promoter sequences of IB--regulated genes, no consensus structural features have been deduced thus far. Epigenetic factors, including covalent modifications of promoters and histones, could be a determinant of IBdependence. Intriguingly, Kayama et al. (26) demonstrated the involvement of IBin stimulus-dependent histone H3K4 trimethylation of the secondary response promoters. Along with the constitutively associated p50 subunit, low level of this modification at the specific target promoters could be directly or indirectly recognized by IB-.
The biological importance of the IB--mediated regulation of its target secondary response genes seems to be different from that of the interferon-␤-mediated induction of secondary response genes, which potentially amplifies antiviral gene programs in an autocrine/paracrine fashion (6). IBcould play a key role in forming a stimulus-specific gene expression pattern. In fact, it has been reported that IB--regulated genes are preferentially induced in response to IL-1␤ rather than TNF-␣ (30,31). In addition, given that expression of IB--regulated genes is inevitably delayed, IBmight act as a timing device to determine the durations of gene expression. IB--mediated gene regulation appears to be crucial for the accomplishment of a specific inflammatory response (32).