Early growth response proteins EGR-4 and EGR-3 interact with immune inflammatory mediators NF-κB p50 and p65

Antigenic stimulation of T cells is induced by T-cell-receptor ligation and results in the release of immune mediators such as cytokines and chemokines. This activation initiates a nuclear program that results in activation of early growth response factors (EGRs), NF-κB, NF-AT, AP-1 and STAT proteins, and de novo transcription of the corresponding genes. In general terms, these immediate-early genes encode DNA-binding transcription factors. Following transcription, protein expression and nuclear translocation, these effector proteins bind to their cognate promoters and initiate transcription of cytokine, chemokine and inflammatory genes. The precise mechanism of how, in lymphocytes, individual transcription factors interact with coactivators and basal transcription factors, and how these DNA-binding proteins control transcription of inflammatory genes is currently unclear.

The four EGR zinc-finger transcription factors (EGR-1 to EGR-4) are transiently and coordinately induced in T cells upon antigenic stimulation and the individual proteins regulate expression of genes encoding immune effectors like interleukin 2 (IL-2) (Skerka et al., 1995), tumor necrosis factor α (TNF-α) (Krämer et al., 1994), the β chain of the IL-2 receptor (Lin and Leonard, 1997), Fas (Dinkel et al., 1997) and the Fas ligand (Li-Weber et al., 1999). In addition, EGR proteins are expressed in distinct cell types and regulate transcription of a wide range of genes, including genes involved in the control of cell growth and apoptosis (Gashler and Sukhatme, 1995).

There is a large panel of EGR-regulated genes and these target genes show tissue-specific expression (Gashler and Sukhatme, 1995). The restricted expression pattern of the target genes contrasts with the ubiquitous expression of EGR proteins and it is hypothesized that the tissue-specific transcription is mediated by additional nuclear proteins that cooperate with EGR proteins. We have previously shown that, in T cells, EGR-1 and EGR-4 form stable complexes with the nuclear factor of activated T cells (NFAT) (Decker et al., 2003).

NF-κB is a transcription factor that plays a key role in inflammation and the immune response. The NF-κB family has five members, including RelA (p65), RelB, c-Rel, NF-κB1 (p105-p50) and NF-κB2 (p100-p52) (Hayden and Gosh, 2004). In the cytoplasm, NF-κB forms an inactive complex through association with the inhibitor IκB. In response to extracellular signals, IκB is targeted to the proteasome and is degraded (Hayden and Gosh, 2004). The individual members interact with each other and form homo- and heterodimeric complexes, which regulate transcription in positive as well as negative fashion. The p50-p65 complex is the most abundant heterodimer and the individual combinations of the various family members determine the specificity of transcription. Upon activation, NF-κB translocates into the nucleus, binds to promoter sites and induces the transcription of the genes encoding inflammatory mediators such as cytokines (e.g. IL-1, IL-2, IL-6, IL-8), cell adhesion molecules (e.g. ICAM-1, VCAM-1, ECAM-1), chemokines (e.g. MCP-1, MIP1-α, MIP-1β) and immune receptors (e.g. IL-2 receptor, Ig-κ light chain) (Grossmann et al., 1999).

In order to characterize how EGR-4 regulates transcription of the genes involved in the inflammatory immune response, we used a yeast two-hybrid assay and identified NF-κB p50 as a partner for EGR-4. Complex formation of EGR-4 and NF-κB p50 is confirmed using recombinant proteins for glutathione-S-transferase (GST) pull-down assays and surface-plasmon-resonance analysis. Physical interaction of EGR-4 with p65 was observed and, similarly, EGR-3 formed complexes with p65 and p50. Fluorescence-resonance-energy transfer (FRET) assays show that the various EGR/NF-κB complexes are located in the nuclei of Jurkat cells in vivo. The EGR/NF-κB complexes strongly activated transcription of inflammatory-gene promoters, such as those encoding IL-2, TNF-α and ICAM-1. The use of deletion mutants in transient-transfection assays and peptide-spot analysis identified the third zinc-finger region of EGR-4 as the central domain for functional interaction with NF-κB p65. The interaction domain is located within an α-helical region and amino acids relevant for the interaction are surface exposed.

Yeast strains Y187 and CG-1945, which have the reporter genes HIS3 and lacZ under control of GAL-4-responsive promoters were used. The `bait' plasmid had a fragment of EGR-4, representing amino acids 158-325, ligated to the C-terminus of the GAL4 DNA-binding domain in the pAS2-1 vector (carrying a TRP-1 marker; Clontech). Saccharomyces cerevisiae strain CG-1945 was transformed according to the manufacturer's instructions (Clontech). A cDNA library prepared from human leukocytes stimulated with phytohemagglutinin (PHA) for 6 and 24 hours was ligated into vector pADGAL4 (carrying a Leu2 marker) and transfected into the S. cerevisiae strain Y185. Following mating, transformed yeast cells were selected by cultivation on agar plates without Trp, Leu or His in the medium. Yeast colonies were transferred to a membrane, lysed in liquid nitrogen and tested for the expression of β-galactosidase. One positive clone was isolated and further analysed.

The human helper T-cell line Jurkat and human 293T kidney cells were maintained in RPMI 1640 (Bio Whittaker) supplemented with 10% heat-inactivated fetal calf serum and 0.2 units ml^–1 streptomycin at between 0.5 ×10⁵ cells ml^–1 and 8 ×10⁵ cells ml^–1. When indicated, Jurkat cells were stimulated for 2-3 hours with 1 μg ml^–1 PHA (Murex Diagnostics) and 20 ng ml^–1 phorbol 12-myristate 13-acetate (PMA; Sigma). Transfection of 293 cells was performed with FUGENE (Roche) according to the manufacturer's instructions. In all experiments, the total amount of transfected DNA was kept constant by addition of plasmid pSG5. After transfection, cells were incubated for 48 hours in culture medium. Cell lysis and measurement of luciferase activities were performed as described (Decker et al., 2003).

Reporter construct p-191T-Luc contains 191bp (–191 to +5) of the promoter of the human gene encoding TNF-α, linked to the luciferase-encoding gene luc (Decker et al., 2003). Reporter construct pTNF-Luc represents three copies of the overlapping EGR-1, Sp1 and the NFAT sites of the TNF-α-encoding-gene promoter region (–174 to –143) linked to the luc gene. This promoter fragment was generated by PCR, using primers with the sequences 5′-TTTGGTACCCCGCCCC-3′ and 5′-CGCGGATCCAGATCTC-3′, and the template 5′-TTTGGTACCCCGCCCCCGCGATGGAGAAGAAACCGAGACAAAATCCCCGCCCCCGCGATGGAGAAGAAACCGAGACAAAATCCCCGCCCCCGCGATGGAGAAGAAACCGAGATCTGGATCCGCG-3′. The PCR product was ligated into vector pMILuc4, containing a minimal promoter region (–53 to +51) from the IL-2-encoding gene and the luc gene (pGL2 basic) (Skerka et al., 1995). The pCILuc1 reporter has the promoter region (–366 to +51) from the IL-2-encoding gene ligated to the luc gene (Skerka et al., 1995). The ICAM-1 promoter construct was kindly provided by J. Johnson (University of Munich, Munich, Germany), which has 1014 nucleotides of the promoter region ligated to the luc gene.

Expression plasmids pSG5-EGR-3 and pSG5-EGR-4 have the full-length human cDNA sequences under control of a SV40 promoter. The EGR-3 cDNA was amplified by PCR from RNA isolated from PHA- and PMA-stimulated PBMCs using the primers 5′-TTGAATTCGCTATGACCGGCAAACTCGCC-3′ (EGR-3for) and 5′-ATGGATCCTGGCTGGCTTTCCCGCTGCTTTCA-3′ (EGR-3rev). The PCR product was cloned into the pSG5 vector. The EGR-4 cDNA was amplified by PCR from the EGR-4 expression plasmid pBSV-8HISEGR-4 (Decker et al., 1998) with the primers 5′-AAGAATTCATATCTAACATGCTCCACCTTAGCGAGTTT-3′ (EGR-4for) and 5′-AAGGATCCTCAGAGAGAAGCGAAGGAGAG-3′ (EGR-4rev). Deletions mutants of EGR-4 were created by PCR and cloned into the vector pSG5. The following primers were used for PCR reactions:

• 5′-GAATTCATATCTAACATGGCTGCCCCTTTCCCAGAGGCGTTC-3′ (EGR-4_150-487for) and EGR-4rev;

• 5′-GAATTCATATCTAACATGAAACCTCTGGTGGCGGACATCCCT-3′ (EGR-4_325-487for) and EGR-4rev;

• 5′-GAATTCATATCTAACATGGCCTTCGCTTGCCCGGTGGAGAGT-3′ (EGR4_377-487for) and EGR-4rev;

• 5′-GAATTCATATCTAACATGGCCTTCGCTTGCCCGGTGGAGAGT-3′ (EGR-4_377-464for) and 5′-AAGGATCCTCACTGCTTGAGGTGCACCTTGCTGTG-3′ (EGR-4_377-464rev);

• 5′-GAATTCATATCTAACATGTGCCTGTATGAGCCTCAGCTCTCCCCG-3′ (EGR-4_171-325for) and 5′-GGATCCGCCGCCGCTGCGAAGGCCCAGCGGGGA-3′ (EGR-4_171-325rev).

pEGR-4_171-464 was created by cloning the EGR-4_171-325 PCR product into the vector pEGR-4_377-464. Expression vector pMT2T65, coding for the human p65 protein, and pMT2T50, coding for the p50 protein are described (Bours et al., 1992). The EGR-3 cDNA, amplified with the primers 5′-AGATCTTATAAATATGACCGGCAAACTCGCCGAGAAG-3′ (EGR-3-Bacfor) and 5′-CCCGTGGTCACCACCTGCGCCAGAATTCCC-3′ (EGR-3-Bacrev), was cloned into the baculovirus expression vector pBSV8His and recombinant virus was generated by homologous recombination (Kühn and Zipfel, 1995). Generation of the EGR-4-expressing baculovirus vector has been described previously (Zipfel et al., 1997). Expression vectors pMT2T65 (coding for the human p65 protein), pMT2T50 (coding for the p50 protein) and pGEXp50 and pGEXp65 (coding for human p65 and p50 proteins) have been described previously (Bours et al., 1992; Franzoso et al., 1996).

For FRET analysis, full-length EGR-3 (encoding amino acids 1-384) and EGR-4 (encoding amino acids 1-486) cDNAs were cloned into the vectors pECFP-C1 and pEYFP-C1 [N-terminal cyan fluorescent protein (CFP) and yellow fluorescent protein (YFP) vectors, Becton Dickinson] using the primers 5′-AAGCTTTAATGACCGGCAAACTCGCCGAGA-3′ (EGR-3for) and 5′-GGATCCTCAGGCGCAGGTGGTGACCACGGG-3′ (EGR-3rev) or 5′-AAGCTTTAATGCTCCACCTTAGCGAGTTTT-3′ (EGR-4for) and 5′-GGATCCTCAGAGAGAAGCGAAGGAGAGGCC-3′ (EGR-4rev), yielding EGR-3-CFP, EGR-3-YFP, EGR-4-CFP and EGR-4-YFP. In addition, a cDNA encoding NF-κB p65 (encoding amino acids 1-552) was cloned into the vectors pCFP-C1 and pYFP-C1 (Becton Dickinson) using the primers 5′-TGCTCGAGGAATGGACGAACTGTTCCCCCTC-3′ (p65for) and 5′-TCGGATCCTTAGGAGCTGACTGACTCAGCAG-3′ (p65rev), resulting in the recombinant constructs p65-CFP and p65-YFP.

Recombinant EGR-3 and EGR-4 were produced and purified in the baculovirus system as previously described (Zipfel et al., 1997). Recombinant p50 (amino acids 55-398) and p65 (amino acids 12-551) proteins were expressed as GST fusion proteins in Escherichia coli as described (Decker et al., 2003). Cell extract was prepared from 1 ×10⁹ Jurkat cells stimulated for 2.5 hours with PHA (1 μg ml^–1) and PMA (20 ng ml^–1). NF-κB p50 or p65 were bound to glutathione-agarose beads and protein binding of EGR-3 and EGR-4 was performed as described (Decker et al., 2003). SDS-PAGE and western blotting were performed as described (Decker et al., 1998) using rabbit anti-human-EGR-3, EGR-4 (Franzoso et al., 1996), p50- or p65 antisera for detection (Santa Cruz Biotechnology).

3 ×10⁸ Jurkat T cells were stimulated for 3 hours with PHA and PMA. Cells were harvested and protein extract was incubated overnight at 4°C in 700 μl binding buffer (50 mM Tris-HCl, pH 7) containing a 1:100 dilution of anti-p65 antiserum (Sc-109-G, Santa Cruz Biotechnology) and 40 μl protein-A/Sepharose beads (Amersham Pharmacia Biotech). Immunoprecipitates were collected by centrifugation and washed five times with binding buffer. The final pellet was resuspended in 50 μl loading buffer (Roth GmbH), boiled for 5 minutes and centrifuged at 20,000 g for 5 minutes at 4°C. Supernatant was separated by SDS-PAGE, followed by western-blot analysis using rabbit antiserum against EGR-4 or p65 (Sc 372, Santa Cruz Biotechnology).

Protein-protein interactions were analysed by the surface-plasmon-resonance technique using a Biacore 3000 instrument (Biacore). Recombinant EGR-4 protein (1 μg ml^–1, dialysed against 10 mM acetate buffer, pH 5.5) was coupled by to the sensor chip (carboxylated dextran chip CM5, Biacore) and recombinant NF-κB p50 and p65 were dialysed against running buffer (0.5X PBS, pH 7.4). Each ligand (15 μg ml^–1, 10 μg ml^–1 and 5 μg ml^–1) as well as cell extract prepared from mock-infected cells was injected separately at a flow rate of 5 μl minute^–1 at 25°C. Resonance units were corrected by using controls.

2 ×10⁷ human Jurkat T cells were transiently transfected with plasmids encoding EGR-3-CFP and p65-YFP or EGR-4-CFP and p65-YFP using FuGene (Roche). Following incubation for 24 hours, cells were washed with PBS, fixed in 3.8% paraformaldehyde in PBS for 15 minutes and mounted on microscope slides using the antifade mounting medium Vectashield (Vector Laboratories). DAPI was used for DNA staining. Fluorescence was visualized with a 63 ×/1.4 NA Zeiss Plan-Apochromat oil-immersion objective (Carl Zeiss) and twofold zoom factor using a confocal laser-scanning microscope (Zeiss-LSM-510 META; Carl Zeiss). CFP fluorescence was excited at 458 nm and signals were detected with the META detector set for 490-520 nm. YFP fluorescence was achieved at an excitation of 514 nm and detected at 559-615 nm. For the acquisition of FRET signals, CFP was excited with a laser set to 458 nm and FRET signals were detected in the YFP channel of the META detector set to 559-615 nm. Interaction of the proteins was further confirmed by acceptor photobleaching (Kenworthy, 2001). Cells were treated with a laser set to YFP setting by scanning a region of interest (ROI) 100 times using a laser set to 514 nm and used at 100% intensity. Before and after treatment, CFP images were collected to monitor changes in donor fluorescence. In order to minimize the effect of photobleaching caused by imaging, images were collected at 5% laser intensity, which is 20 times lower than the bleaching intensity used. Each image was collected first for FRET, followed by CFP and at last for YFP fluorescence. In all measurements, the background (pixel values outside the cells) was very low (<5% of the signal). The FRET efficiency was calculated as a percentage (E_F) using the formula E_F=[(I_{after bleach}–I_{before bleach}) ×100]÷I_{after bleach} (Karpova et al., 2003). Identical calculations were performed in selected regions of the images that had not been used for bleaching. The distance between CFP and YFP signals was determined using the formula for Foerster distances (Karpova et al., 2003). According to this procedure, photobleaching of the acceptor (YFP) results in a significant increase in donor fluorescence (CFP).

Peptides that represent the zinc-finger regions of EGR-3 (amino acids 272-364), EGR-4 (amino acids 377-468) and mutated EGR-4 (EGR-4mut; amino acids 377-468, with mutations K455G, K456G, K457G, K460G, K464G, K466G and R468G) were synthesized with length of 13 amino acids and an overlap of ten amino acids, and were coupled to a cellulose membrane (Jerini Peptide Technologies). The membranes were blocked with 5% albumin (Sigma) in buffer I (50 mM Tris-HCl, 137 mM NaCl, 2.7 mM KCl, pH 8) overnight and the filters were washed with buffer II (50 mM Tris-HCl, 137 mM NaCl, 2.7 mM KCl, pH 8, 0.05% Tween 20). Following incubation with NF-κB p65 for 3 hours at 4°C, the filters were incubated with rabbit antiserum against p65 (Biomol) for 3 hours. The membranes were incubated with an horseradish-peroxidase-labeled antirabbit antibody and developed using enhanced chemiluminescence (ECL™; Amersham Pharmacia Biotech) and exposed to X-ray film (Kodak).

Synthetic peptides representing the interaction domain of zinc finger III of EGR-4 (peptide 1, amino acids EKKRHSKVHLKQKAR, residues 453-466) as well as an unspecific scrambled peptide (peptide 2, amino acids TQPQLSPLGLRSAAA) were synthesized (Jerini Peptide Technologies). The inhibitory activity of the two peptides were tested in GST pull-down assays and by surface-plasmon-resonance analysis. Recombinant EGR-4 was immobilized to Ni²⁺-NTA agarose. The NF-κB p65 protein was preincubated with 50 μg ml^–1 peptide 1 or peptide 2 for 60 minutes on ice and then applied to the EGR-4 matrix for 2 hours at 4°C. The resin was precipitated and protein complexes were eluted. The eluates were separated by SDS-PAGE and identified by western blotting using an antibody against EGR-4, horseradish-peroxidase-labeled antirabbit antibody and ECL reagent (Amersham Pharmacia Biotech). For surface-plasmon-resonance assays, 1 μg ml^–1 recombinant EGR-4 was coupled to the surface of a sensor chip. Purified NF-κB p65 preincubated with either peptide 1 or peptide 2 for 60 minutes in running buffer was injected into the flow cell, using a flow rate of 5 μl minute^–1 at 25°C. Resonance units were recorded and corrected by controls.

A molecular model for EGR-4 was produced using the published structure of Zif268 [the mouse homolog of EGR-1; protein databank (PDB), www.rscb.org, protein databank code 1zaa] (Pavletich and Pabo, 1991) bound to the consensus binding site (5′-GCGGGGGCG-3′). The amino acids of the identified p65 interaction domain in zinc finger III is highlighted to visualize position of this domain in the zinc-finger region.

The yeast two-hybrid system (Matchmaker, Clontech) was used according the manufacture's protocol to search for proteins that interact with EGR-4. To prevent activation by the bait itself, a domain of EGR-4 that lacks the zinc-finger DNA-binding region of EGR-4 (amino acids 158-325) was used. The EGR-4/GAL4 construct was used as a bait to screen a cDNA library (HybriZAP, Stratagene) prepared form activated T cells. One positive clone was isolated, sequenced and identified as encoding NF-κB p50 (position 18-1580, coding sequence NM_003998).

Recombinant proteins were used to confirm the observed interaction between EGR-4 and NF-κB p50. NF-κB p50 with an N-terminal GST tag was bound to glutathione beads and recombinant EGR-4 was applied to this matrix. After extensive washing, bound proteins were eluted, separated by SDS-PAGE and analysed by western blotting. EGR-4 was identified in the cell extract as a 70 kDa protein (Fig. 1A, lane 1). The protein was adsorbed to the p50 matrix and was recovered in the eluate (Fig. 1A, lane 2). In addition, binding of native Jurkat-derived EGR-4 to immobilized p50 was assayed. Again, EGR-4 was identified in cell extract prepared from stimulated Jurkat cells and also in the eluate (Fig. 1A, lanes 3, 4). Binding was specific, because no protein was detected in the wash fraction (data not shown). These results demonstrate specific binding of recombinant and native EGR-4 to immobilized NF-κB p50 protein and confirm the results of the yeast two-hybrid screen.

Given the homology between NF-κB proteins, the same approach was used to study the interaction of EGR-4 with the related p65. Recombinant and native EGR-4 were applied to the p65 matrix and EGR-4 was detected in the eluate (Fig. 1A, lanes 6, 8), demonstrating binding of EGR-4 to p65.

We further asked whether EGR-3, another member of the EGR protein family, can interact with NF-κB proteins. EGR-3 was identified as a 60 kDa protein in cell extracts prepared from both baculovirus-infected insect cells and Jurkat T cells (Fig. 1A, lanes 9, 13). The recombinant and also the native EGR-3 proteins bound to p50 as well as p65 (Fig. 1A, lanes 10, 12, 14, 16). This binding was specific, because no protein was detectable in the wash fractions (data not shown). Binding of EGR-3 and EGR-4 to the GST matrix in the absence of the ligand was excluded (Fig. 1A, lanes 18, 20, 22, 24).

Interaction of EGR-4 with p50 and p65 proteins was further analysed by surface plasmon resonance. Recombinant EGR-4 was immobilized to the sensor-chip surface and recombinant p50 or p65 was injected to the flow cell. Dose-dependent association and dissociation of p50 to EGR-4 (Fig. 1B) is indicated by the strong association and dissociation profile. Similar results were obtained for p65 (Fig. 1C).

To verify EGR/NF-κB complex formation, immunoprecipitation was performed. Cell extract from stimulated Jurkat T cells was incubated with p65 antiserum and protein-A/Sepharose. After extensive washing, bound proteins were eluted, separated by SDS-PAGE and analysed by western blotting. NF-κB p65 and EGR-4 were identified in the flow through, first wash and elute fractions (Fig. 1D, lanes 2, 3, 5, top and bottom, respectively) and are absent from the second wash fractions (Fig. 1D, lane 4). Identification of p65 and EGR-4 in the eluates (Fig. 1D, lane 5) demonstrates interaction of EGR-4 with p65. The identified proteins (Fig. 1D, lanes 2-5) have comparable mobilities to the recombinant proteins shown in Fig. 1D, lane 1. Unspecific binding of EGR-4 to the column was excluded (data not shown).

The in vivo interaction of EGR and NF-κB proteins was further studied using fluorescently tagged proteins following transient production in Jurkat T cells. DAPI was used for nuclear staining in order to follow the cellular distribution of the proteins. EGR-3 and EGR-4 were localized predominantly to the nuclei. Most NF-κB p65 was identified in the cytoplasm but a small proportion (about 15%) was detected in the nucleus (Fig. 2). Upon co-transfection with plasmids expressing EGR-3-CFP and p65-YFP, interaction of the proteins was assayed by FRET, which allows detection of acceptor fluorescence after donor excitation when proteins are in very close proximity (∼6-10 nm) to each other. A FRET signal was detected in the nuclei of the transfected cells, thus demonstrating an interaction between EGR-3 and p65 (Fig. 2, top, FRET). Transfection of an unrelated YFP-labeled protein did not result in a FRET signal (data not shown). The relevance of the EGR-p65 interaction was further confirmed by acceptor photobleaching (Fig. 2, top, Bleach). Following acceptor (p65-YFP) bleaching, the fluorescence of the donor (EGR-3-CFP) was enhanced by about 25% because energy transfer between the proteins was eliminated and the FRET signal was absent (Fig. 2, top, Bleach). The presence of FRET signal in the normal setting together with the increase in fluorescence after photobleaching show that EGR-3 interacts with p65 in vivo and reveals that the two proteins are close together (calculated to be ∼6 nm). Almost identical results were obtained for EGR-4 (Fig. 2, bottom), thus demonstrating an interaction between both EGR-3 and EGR-4, and the NF-κB protein p65 in vivo.

Having demonstrated physical interaction between EGR and NF-κB proteins, we were interested to test whether these protein complexes are relevant for gene transcription. Human 293 cells were transfected with expression vectors coding for the individual EGR or NF-κB proteins and a reporter construct that has the complete 366 bp of the human IL-2-encoding-gene promoter linked to luc. Used as single proteins, EGR-3, EGR-4 and p50 have little effect on transcription (increases of 1.0, 0.9 and 1.8 times, respectively), whereas p65 as a homodimer or as a heterodimer with p50 increased transcription 17.6 and 15.5 times, respectively (Fig. 3A). EGR-3 and also EGR-4 together with p50 showed no effect but, when combined with p65, each zinc-finger protein induced reporter-gene activity strongly, resulting in 108- and 87-fold induction, respectively. Thus, with p50, both zinc-finger proteins lack transcriptional activity and, together with p65, they strongly activate transcription. Both EGR/p65 complexes are actually more efficient than the p65-p50 heterodimer.

The role of the EGR/NF-κB complexes on transcription was further tested with a TNF-α reporter construct. Again, EGR-3 and EGR-4 alone displayed little effect on transcription (1.2 times and 1.6 times, respectively), and a combination of p50 and p65 increased reporter gene activity 21-fold (Fig. 3B). Coproduction of EGR-3 or EGR-4 with p50 showed no transcriptional response but the EGR-3/p65 and EGR-4/p65 complexes resulted in a strong activation of transcription (84- and 90-fold, respectively).

ICAM-1 is a central inflammatory marker protein and its human gene promoter includes binding sites for EGR-1, NF-κB and NFAT (van der Stolpe and Saag, 1996). Used as single factors EGR-3, EGR-4 and p50 had no or minor effects on transcription (inductions of 0.3, 0.6 and 1.3 times), whereas p65 as homodimer induced transcription sevenfold and, as a heterodimer with p50, 13-fold (Fig. 3C). Both zinc-finger proteins combined with p50 showed no activation but the EGR-p65 complex induced transcription 24- and 36-fold. These results demonstrate a strong transcriptional effect for the EGR-p65 heterodimers on the transcription of human inflammatory-cytokine-encoding genes. We conclude that EGR-p65 synergistic activation is specific, because EGR-4/p65 complexes do not activate a reporter plasmid like the promoter of the p21-encoding gene, which includes an EGR binding site but lacks a NF-κB site in the promoter region (data not shown).

Having demonstrated complex formation between EGR-4 and p65, it was of interest to localize the protein domain(s) relevant for this interaction. To this end, deletion mutants of EGR-4 were generated and produced together with p65 in the presence of a TNF-α reporter construct. Wild-type EGR-4 together with p65 increased transcription of the reporter 105-fold, but p65 as homodimer or as a heterodimer with p50 caused a 25- or 68-fold induction (Fig. 4). Deletion of the N-terminal 149 (EGR-4_150-487) or 324 (EGR-4_325-487) amino acids reduced reporter activity by 62% and 79%, respectively. Further deletion of the N-terminal domain to amino acid 377 (EGR-4_377-487) surprisingly enhanced reporter activity, resulting in a 162-fold induction.

The construct that encodes just the DNA-binding zinc-finger domain (EGR-4_377-464) showed strong transcriptional activity (131-fold induction, which is 25% higher than that of the wild-type protein). Thus, the zinc-finger domain by itself is necessary and sufficient for interaction with the NF-κB subunit p65. In the absence of the zinc finger, the N-terminal region between amino acids 171 and 325 (EGR-4_171-325) lacked transcriptional activity. However, when linked to the zinc-finger domain (EGR-4_377-464), this domain showed a strong activity, resulting in a 273-fold activation of the reporter construct, which is 2.6 times that of wild-type EGR-4. Differences in reporter gene activation caused by different production levels of p65 were excluded, because production of p65 was tested by western-blot analysis and was similar in all co-transfection assays (data not shown).

In order to identify and localize the amino acids relevant for the interaction with NF-κB p65 peptides representing the entire zinc-finger domain of EGR-3 and EGR-4 were spotted and probed with p65. A linear interaction domain was localized to amino acids 348-361 of EGR-3 (Fig. 5A) and to amino acids 453-466 of EGR-4 (Fig. 5B). In both EGR proteins, the interaction domains are located in the conserved C-terminal region of zinc finger III. Replacing the positively charged Lys and Arg residues in the EGR-4 interaction domain (K453, K454, R455, K458, K462, K464 and R466) with glycine abrogated the binding of p65 (Fig. 5C).

Inhibition experiments with synthetic peptides were performed in order to confirm the relevance of the linear EGR-4 interaction domain (amino acids 453-466). Preincubation of NF-κB p65 with synthetic peptides representing the major interaction domain of EGR-4 (peptide 1) blocked binding of p65 to EGR-4 in the GST pull-down assay (Fig. 6A). An unrelated scrambled sequence (peptide 2) had no effect (Fig. 6B). Similar inhibitory activities were observed by surface plasmon resonance. Peptide 1 strongly reduced binding of p65 to immobilized EGR-4, and the unspecific peptide 2 affected the interaction weakly (Fig. 6C). These experiments confirm the relevance of this interaction domain in EGR-4.

The zinc-finger domains of the individual EGR proteins are highly homologous to each other and are conserved between man and mouse (Fig. 7A). Molecular modeling based on the structure of the mouse EGR-1 homolog (ZIF 268) (Pavletich and Pabo, 1991) was therefore used to localize the interaction domain and the position of the charged residues within the zinc finger of EGR-4. The corresponding structure identifies the p65 binding region within the α-helical region of zinc finger III and shows that four of the seven positively charged amino acids, which are required for p65 interaction are surface exposed and thus accessible for interaction (Fig. 7B).

A yeast two-hybrid screen identified the NF-κB protein p50 as an interaction partner of EGR-4 and this interaction was confirmed with recombinant and native proteins. GST-pull-down and surface-plasmon-resonance assays show physical-complex formation between either EGR-3 or EGR-4 and either p50 or p65. Complex formation was further demonstrated in vivo by immunoprecipitation and FRET analysis in Jurkat T cells. The individual complexes affected transcription of inflammatory genes differently. EGR-p50 complexes lack transcriptional activity, but EGR-p65 complexes are potent activators of inflammatory gene promoters such as those for IL-2, TNF-α and ICAM-1. Transcriptional activation of the EGR-p65 complex exceeds those of the p65 homodimer and the p65-p50 heterodimer. The use of deletion constructs in functional assays and mutagenesis studies identified the zinc-finger region of EGR-4 as central element for interaction with p65. Thus, we show direct interaction and complex formation between EGR/NF-κB proteins, which are central in human T cells for transcription of inflammatory genes.

Upon T-cell activation, EGR, NFAT and NF-κB transcription factors are coordinately induced. These proteins function as nuclear signal transmitters and induce the expression of immune effector genes. In transfection assays, EGR proteins alone show no transcriptional activity. However, they form complexes with additional nuclear transcription factors such as NF-κB p65 and NFAT (Fig. 1) (Decker et al., 1998; Decker et al., 2003), and these complexes act as strong transcriptional regulators of immune effector genes (Fig. 3). Particular EGR-p65 complexes are potent inducers of inflammatory genes, because (i) the transcriptional activity of the EGR-p65 complex increases reporter gene activity about 100-fold (Fig. 3), (ii) this transcriptional activity exceeds that of the p50-p65 heterodimer and the p65 homodimer (Fig. 3), and (iii) corresponding EGR and NF-κB binding sites are present in a range of inflammatory and immune effector genes.

This is the first characterization of a physical interaction of the zinc-finger proteins EGR-3 and EGR-4 with NF-κB proteins. However, a transcriptional interaction between EGR-1 and RelA in regulation of NF-κB1 gene transcription (Cogswell et al., 1997) and a DNA-dependent interaction between EGR-1 and RelA (Chapman and Perkins, 2000) have been described previously. EGR proteins form physical complexes with NF-κB proteins p65 and p50 (Fig. 1) and also with other lineage-specific nuclear factors, such as (i) NFAT proteins (NFATc and NFATp) (Decker et al, 1998; Decker et al., 2000), (ii) the homeobox protein Ptx1 and steroidogenic factor 1 (Tremblay and Droiun, 1999), (iii) the cAMP-responsive-element-binding-protein-binding protein CBP/p300 (Silverman et al., 1998), and (iv) the tumor suppressor p53 (Liu et al., 2001). In addition interaction of EGR-1 with viral proteins like HBx (Yoo et al., 1996a), IE2 (Yoo et al., 1996b), tax (Trejo et al., 1996) and tat (Yang et al., 2002), or with NAB repressor proteins (Russo et al., 1995; Svaren et al., 1996) has been demonstrated. It is likely that the EGR/NF-κB heterodimer is part of a multiprotein complex, because both EGR-1 and p65 interact with basal factors such as CBP/p300 (Silverman et al., 1998; Gerritsen et al., 1997), TATA-box-associated factors (TAFs), as well as histone acetylase (Yami-Hezi et al., 2000; Zhong et al., 2002). Thus, EGR and p65 proteins participate in the formation of a specific regulatory transcriptional complex that links inducible and basal promoter components, and that initiates gene transcription.

The use of deletion constructs characterized EGR-4 as a multidomain protein containing several activation- and repressor domains (Fig. 4). The high transcriptional activity of the various EGR-4 deletion constructs, particularly of EGR-4_171-464, (i) identifies an activation domain between residues 171 and 325, (ii) identifies the presence of a repressor domain (residues 325-377) immediately upstream of the zinc-finger domain, and (iii) highlights the role of the zinc-finger domain for cooperation with p65. Thus, EGR-4 is a multidomain protein that, in addition to DNA-binding and interaction domains, has activating and repressing functions. The repressor domain of the related EGR-1 protein, which is also located upstream of the zinc-finger domain represents a binding site for NAB repressor proteins (Russo et al., 1995; Svaren et al., 1996). Further experiments will show whether the repressor domain of EGR-4 also interacts with NAB factors.

Transfection assays localized one region within the DNA-binding zinc-finger domain of EGR-4 that is central for interaction with NF-κB p65 (amino acids 377-464) (Fig. 4). Similarly, the zinc-finger domain of the highly related zinc-finger protein Sp1 interacts with the v-Rel oncoprotein (Sif and Gilmore, 1994) and the zinc-finger domain of EGR-1 is sufficient for interaction with RelA (p65) (Chapman and Perkins, 2000). Here, we localize the interaction domain for EGR-3 and EGR-4 to a linear 14 amino acid sequence (i.e. EGR-3, amino acids 347-361; EGR-4, amino acids 453-466) located within the third zinc-finger (Fig. 5). The zinc-finger domain is also necessary for nuclear translocation (Matheny et al., 1994) of EGR-1; in particular, an Arg at position 412 in EGR-1 is essential for nuclear localization. In EGR-4, this position is occupied by a Lys (K453) (Fig. 7A) and participates in p65 binding. Mutations of the relevant amino acids (K453N, K454K, R455Q, K458Q) in EGR-4 resulted in complete cytoplasmic localization of the protein and no FRET between EGR-4 and p65 was detectable (data not shown). Thus, the zinc finger III EGR-4 has multiple functions: nuclear translocation, DNA binding and interaction with p65. The molecular model of the EGR-4 zinc-finger domain in its DNA-bound conformation confirms the dual role for DNA binding and protein interaction (Fig. 7B). Similar to the conserved DNA-binding domain in the EGR zinc-finger proteins, the conserved Rel homology domain (RHD) of the NF-κB proteins is probably involved in interaction with EGR-1 (Chapman and Perkins, 2000) and the related Sp1 proteins (Sif and Gilmore, 1994; Perkins et al., 1994). Further experiments will show whether EGR-3 and EGR-4 also interact with the RHD domain, and will identify amino acids involved in binding.

Based on the high sequence identity of the four EGR proteins, particularly within the zinc-finger domains, it is expected that all EGR proteins (i.e. EGR-1 and EGR-2 as well as EGR-3 and EGR-4) interact with NF-κB proteins. The initial yeast two-hybrid screen identified a N-terminal domain (amino acids 158-325) of EGR-4 as a ligand of p50. Further characterization mapped the major binding of p65 to the zinc-finger domain of EGR-4. Together with the sequence similarity of p50 and p65, these results suggest that EGR-4 has two interaction domains for p50 and for p65. One low-affinity interaction domain is located within the N-terminus and the second is positioned in the zinc-finger domain of EGR-4. Co-transfection assays also show a higher synergistic activity when an N-terminal region of EGR-4 is linked to the zinc-finger domain (Fig. 4). The concept of two interaction domains is confirmed by the strong but incomplete inhibition of the specific synthetic peptide representing the interaction site located in the zinc finger of EGR-4 (Fig. 6).

In T cells, the immediate-early EGR zinc-finger proteins function as transcriptional regulators of inflammatory genes, particularly during the initial phase of T-cell activation. Experiments with challenged EGR-1-knockout mice show a significant reduced TNF-α response (Silverman et al., 2001). Thus, EGR zinc-finger proteins regulate the immune response and transcription of central inflammatory genes by specific interaction and complex formation with inducible nuclear inflammatory regulators such as NF-κB and NFAT. These potent immune regulators could represent ideal targets for the development of inhibitors of the immune reaction.

This work was funded by the Thüringer Ministerium für Wissenschaft, Forschung und Kultur (TMWFK).