Early Growth Response-1 Regulates Lipopolysaccharide-induced Suppressor of Cytokine Signaling-1 Transcription*

Suppressor of cytokine signaling (SOCS)-1 is a critical regulator of lipopolysaccharide (LPS) tolerance and LPS-induced cytokine production. The mechanisms regulating the transcription of SOCS-1 in response to LPS are not entirely understood. Functional analysis of the SOCS-1 promoter demonstrates that early growth re-sponse-1 (Egr-1) is an important transcriptional regulator of SOCS-1. Two Egr-1 binding sites are present within the SOCS-1 promoter as shown by EMSA and supershift analysis. Further, mutation of the Egr-1 binding sites significantly reduces both the basal and LPS-induced transcriptional activity of the promoter. Chromatin immunoprecipitation experiments confirm LPS-induced binding of Egr-1 to the SOCS-1 promoter in vivo . Additionally, Egr-1 (cid:1) / (cid:1) macrophages show reduced levels of LPS-induced SOCS-1 expression in comparison with macrophages derived from Egr-1 (cid:2) / (cid:2) littermate controls. These results demonstrate an important role for Egr-1 in regulating both the basal and LPS-induced activity of the SOCS-1 promoter.

The Toll-like receptor (TLR) 1 family is a diverse group of transmembrane receptors that recognize microbial products called pathogen-associated molecular patterns. The proper regulation of TLR signaling is important in limiting septic shock, inflammation, and various aspects of both innate and adaptive immunity. The ligation of the TLR4-MD2 complex by LPS induces the recruitment of the MyD88 and subsequent activation of both interleukin-1 receptor-associated kinase and TNF receptor-associated factor-6. Activation of these molecules leads to signaling via the NF-B and the c-Jun N-terminal kinase mitogen-activated protein kinase pathways. Whereas a significant response to a bacterial infection is often required for protective immunity, elevated levels of cytokine protection can have deleterious effects; for example, toxic shock and systemic inflammatory syndrome. Repeated exposure to LPS induces a protective condition, termed LPS tolerance, that prevents shock following challenge with otherwise lethal levels of endo-toxin. Various mechanisms have been proposed for the development of this refractory state: TLR4 down-regulation (1); upregulation of interleukin-1 receptor-associated kinase-M, a negative regulator of TLR signaling (2); reduced NF-B activation (3); and the induction of SOCS-1 (4,5).
SOCS-1, also called JAB or SSI-1, is a critical regulator of cytokine signaling and has been shown to be capable of regulating JAK-STAT signaling (6 -8). SOCS-1 is one of eight members of the SOCS gene family, which is defined by a C-terminal SOCS box region and a central phosphotyrosine-binding Src homology 2 domain (9). SOCS-1 is up-regulated by many cytokines, most notably IFN-␥ (10), and functions to inhibit signaling activity with its Src homology 2 domain, which binds to the activation loop of the JAK family members. Furthermore, the SOCS box region has been shown to target SOCS-1-interacting proteins for proteosomal-mediated degradation via ubiquitination; such proteins include JAK2 (11), TEL-JAK2 (12,13), p65 NF-B (14), and Vav (15). SOCS-1 is a critical regulator of IFN-␥ signaling as demonstrated by the lethality of the SOCS-1 Ϫ/Ϫ neonates. While normal at birth, SOCS-1-deficient neonates display stunted growth, multiorgan disease characterized by severe lymphopenia, fatty degeneration of the liver, and macrophage infiltration of various tissues, and they die by 3 weeks of age (16).
SOCS-1 regulates various aspects of innate immunity, since it has been shown to limit LPS-induced cytokine production and be critical for the regulation of LPS tolerance (4,5,17). LPS and CpG DNA, which signal through the TLR-4 and TLR-9 receptors, respectively, have been shown to up-regulate SOCS-1 in macrophages (18 -20). Further, both SOCS-1 ϩ/Ϫ and SOCS-1 Ϫ/Ϫ mice are hyperresponsive to LPS, demonstrate reduced viability to LPS challenge, and produce elevated levels of LPS-induced proinflammatory cytokines such as interleukin-12 and TNF-␣ (4,5). Consistent with its role as a negative regulator of TLR signaling, forced expression of SOCS-1 inhibits TLR4 signaling in macrophages (4) and targets p65 NF-B for ubiquitin-mediated proteolysis (14).
At present, the mediators for LPS induction of SOCS-1 are not entirely clear. Regulated by both transcriptional and posttranscriptional mechanisms, SOCS-1 levels are most potently up-regulated by IFN-␥ in an STAT-1-dependent manner (21). The IFN-␥-inducible activity of the murine promoter is dependent upon an IRF binding element (IRF-E) (21). Interferons are capable of regulating transcription through a variety of transcriptional complexes; principle examples include the STAT-1 homodimer binding to an IFN-␥ activation site element and ISGF3, which binds the IFN-stimulated regulatory element. However, in the case of SOCS-1, neither of these complexes binds directly to the promoter. Instead, IRF-1 (which contains its own IFN-␥ activation site element (22)) is up-regulated in a STAT1-dependent manner to bind the AANNGAAA repeat sequence within the SOCS-1 promoter (21). In addition, evidence * 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.
LPS stimulation activates many transcription factors downstream of the c-Jun N-terminal kinase/mitogen-activated protein kinase pathway, one of which is the zinc finger transcription factor early growth response-1 (Egr-1), also known as zif268, Krox-24, tis-8, and NGFI-A. Egr-1 is an immediate early gene that is up-regulated by a multitude of growth factors, cytokines, and environmental stresses such as hypoxia (24), vascular injury (25), and septic shock (26). Specifically, the induction of Egr-1 in response to LPS has been shown to be mediated by the activation and subsequent binding of Elk-1 to a number of serum response elements present within the Egr-1 promoter (27)(28)(29). Egr-1 is capable of regulating a number of genes involved in inflammation, including TNF-␣ (30, 31), tissue factor (29), and intercellular adhesion molecule-1 (32). Egr-1 preferentially binds to the GC-rich sequence 5Ј-GCGGGGGCG-3Ј (33)(34)(35) and is expressed in a wide variety of tissues including the thymus, muscle, bone, and parts of the nervous system during development (36). Egr-1-null mice are viable but have reduced body size and are sterile due to apparent defects in hormone regulation (37). Egr-1 is expressed in monocytes, and expression of antisense oligonucleotides to Egr-1 inhibits macrophage differentiation (38). In contrast to these results, macrophage differentiation appears normal in Egr-1 Ϫ/Ϫ mice (39). Further, many of the target genes of Egr-1 have additional regulatory mechanisms, because the Egr-1 Ϫ/Ϫ mice demonstrate only slightly reduced levels of MCP-1, tissue factor, intercellular adhesion molecule-1, and interleukin-6 at certain time points following LPS challenge (26).
We present evidence that the LPS-induced activity of the SOCS-1 promoter is regulated by the transcription factor Egr-1. SOCS-1 expression is reduced in LPS-stimulated macrophages from Egr-1 Ϫ/Ϫ versus Egr-1 ϩ/ϩ mice, and chromatin immunoprecipitation (ChIP) analysis indicates that the SOCS-1 promoter is bound by Egr-1 in vivo. Further, two Egr-1 binding sites are present within the SOCS-1 promoter, and mutational analysis shows that the LPS response requires these sites and an IRF-E for LPS-induced activity of the SOCS-1 promoter. Taken together, these data suggest that the SOCS-1 promoter is regulated through both Egr-1-dependent and Egr-1-independent regulatory pathways.

EXPERIMENTAL PROCEDURES
Human DNA Samples-NA12547 was purchased from the Coriell Cell Repositories.
Cell Lines-THP-1, a human monocytic leukemia cell line, was obtained from the American Type Culture Collection. The RAW/FPR.10 clone derived from the RAW 264.7 murine macrophage cell line was obtained from Dr. Steve Greenberg (40).
Luciferase Experiments-The RAW/FPR.10 cells were transfected using Lipofectamine reagent (Invitrogen). Cells were plated at a density of 3 ϫ 10 5 cells/well in a 6-well plate 24 h prior to transfection. Transfection was performed following the manufacturer's protocol with 1 g of reporter plasmid and 100 ng of Renilla luciferase control plasmid (Promega) for 4 h. Following this period, the transfection mixture was removed and replaced with media with or without 100 ng/ml LPS for 20 h. Cell extracts were subsequently prepared and assayed using the Dual Luciferase kit (Promega) as per the manufacturer's instructions. Luciferase activities were normalized to the Renilla control plasmid, and values shown are the mean of three independent experiments.
Isolation and Differentiation of Macrophages-Bone marrow-derived macrophages were isolated and generated with an adapted method (41). Instead of granulocyte-macrophage colony-stimulating factor, the culture medium was supplemented with 10% macrophage colony-stimulating factor-containing L929 culture supernatant at day 0. At day 3, an additional 10 ml of medium containing 10% L929 supernatant was added. Day 7 adherent cells were analyzed by fluorescence-activated cell sorting and used in experiments.
Plasmids-The IRF-2/pRSET plasmid used for the isolation of recombinant IRF-2 was provided by Dr. Kathryn Calame. Regions of the SOCS-1 promoter were cloned into the BglII and MluI sites of the pGL3 Basic Vector (Promega) using primers with the appropriate restriction sites, with the exception of the Ϫ1521/Ϫ659pGL3basic, Ϫ1160/ Ϫ659pGL3basic, and Ϫ945/Ϫ659pGL3basic plasmids. The latter were generated by digestion of the Ϫ2523/Ϫ659pGL3basic plasmid with SacI, SacI ϩ AvrII, or SacI ϩ XhoI, respectively, followed by digestion of the overhangs with mung bean nuclease and ligation. Sequence mutations were made with the appropriate primers in a PCR-based mutagenesis protocol (Stratagene). All plasmids were confirmed by sequencing. The pRL-TK Renilla luciferase plasmid was purchased from Promega.
Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extracts were prepared as described previously (42). Recombinant IRF-2 was purified as described in Ref. 43. Binding reactions were performed for 15 min with either the indicated amounts of recombinant protein or 10 g of nuclear extracts with a final concentration of 5.0 mM HEPES (pH 7.9), 2.5 mM MgCl 2 , 20% glycerol, 2.5 mM dithiothreitol, 1.0 g of dIdC and 40 g of bovine serum albumin. Reactions were run in gels containing 6.0% acrylamide, 2.5% glycerol, and 0.25ϫ TBE.
ChIP-THP-1 cells were treated with a final concentration of 1% formaldehyde to cross-link for 10 min and subsequently stopped with the addition of glycine at a final concentration of 0.125 M for 5 min. Cells were washed twice with TBS (20 mM Tris-Cl, pH 7.4, 150 mM NaCl), counted so that 1 ϫ 10 7 cells were used in each immunoprecipitation and lysed in cold SDS buffer (0.5% SDS, 100 mM NaCl, 50 mM Tris-Cl, pH 8.1, 5 mM, EDTA, pH 8.0, and 0.02% NaN 3 ). Cells were then pelleted and resuspended in cold IP buffer (1 volume of SDS buffer plus 0.5 volume of Triton dilution buffer (100 mM Tris-Cl, pH 8.6, 100 mM NaCl, 5 mM EDTA, pH 8.0, 0.02% NaN 3 , and 5% Triton X-100)). The protease inhibitors aprotinin, pepstatin, and phenylmethylsulfonyl fluoride were previously added to the IP buffer. Cells were then sonicated at 4°C and spun to remove cell debris. Extract volume was adjusted to 1.0 ml per immunoprecipitation in 1.5-ml siliconized Eppendorf tubes and precleared with 30 l of Protein A beads (Amersham 50% slurry containing 0.2 mg/ml sonicated salmon sperm DNA and 0.5 mg/ml lipid-free bovine serum albumin) for 30 min at 4°C. Extracts were spun to remove beads, and 25 l of lysate was removed as a total control. Primary antibodies were then added and rotated overnight at 4°C. 30 l of single-stranded DNA/Protein A beads were then added the following day and rotated for an additional 2 h at 4°C before precipitating beads and washed as follows: 2 ϫ 1.0 ml of mixed micelle buffer (150 mM NaCl, 20 mM Tris-Cl, pH 8.1, 5 mM EDTA, pH 8.0, 0.052% (w/v) sucrose, 0.02% NaN 3 , 1% Triton X-100, and 0.2% SDS), 2 ϫ 1.0 ml of buffer 500 (0.1% (w/v) deoxycholic acid, 1 mM EDTA, 50 mM HEPES, pH 7.5, 500 mM NaCl, 1% Triton X-100, and 0.02% NaN 3 ), 2 ϫ 1.0 ml of LiCl/detergent wash (0.5% (w/v) deoxycholic acid, 1 mM EDTA, 250 mM LiCl, 0.5% Nonidet P-40, 10 mM Tris-Cl, pH 8.0, and 0.02% NaN 3 ), and 2 ϫ 1.0 ml of TE (10 mM Tris-Cl, pH 8.0, and 1 mM EDTA, pH 8.0). 0.3 ml of a 1% SDS, 100 mM NaHCO 3 solution was then added to both the immunoprecipitates and the total controls and incubated at 65°C to elute complexes from beads and reverse the cross-links. Eluate was transferred to a fresh microcentrifuge tube with 0.25 ml of a proteinase K solution (30 g of glycogen, 100 g of proteinase K in TE, pH 7.6) and incubated at 37°C for 2 h. 50 l of 4 M LiCl was then added, followed by 0.5 ml of 1:1 phenol/chloroform mixture. Tubes were then shaken vigorously for 1 min and phases separated by centrifugation for 10 min at room temperature. The upper phase was transferred to a separate tube with 1.0 ml of 100% ethanol and incubated on dry ice for 15 min to precipitate DNA. DNA was centrifuged at 4°C for 30 min, washed with 70% ethanol, recentrifuged for 5 min at 4°C, allowed to air dry for 10 min, and resuspended in 30 l of H 2 O. Input samples represent 0.05, 0.01, and 0.002% of the total DNA, whereas the PCRs of the immunoprecipitations include 10, 5, and 2.5% of the resuspended DNA. The general conditions for PCR are as follows: an initial step of 94°C for 5 min, followed by 94°C for 30 s, 55°C for 30 s, and 72°C for 30 s repeated for 30 cycles and 72°C for 7 min. Each 25-l PCR contained 1ϫ High Fidelity PCR buffer, 2.0 mM MgSO 4 , 0.8 mM deoxynucleoside triphos-phate, 0.8 M each primer, and 1.25 units of HiFi Platinum TaqDNA polymerase (Invitrogen). Primers for ChIP were as follows: promoter-5Ј, 5Ј-GTCGCCAAGTCCGAAGGA-3Ј; promoter-3Ј, 5Ј-CCCAGCTCCACT-TTTGGT-3Ј; 3Ј-control-5Ј, 5Ј-CACTAGGCAACCGGAGGA-3Ј; 3Ј-control-3Ј, 5Ј-CTTTTGAGGAGGACCTGT-3Ј.

Regulation of the SOCS-1 Promoter in Response to LPS-
The mechanism by which LPS induces SOCS-1 transcription has not been fully elucidated. Previous evidence has suggested that the regulation is dependent upon autocrine and paracrine cytokine signaling mechanisms, whereby the treatment of macrophages with LPS induces cytokines such as IFN-␣/␤ to upregulate SOCS-1 expression (19,20). The JAK-STATindependent nature of TLR signaling suggested the possibility of an additional pathway for the regulation of SOCS-1 transcription. To determine the transcriptional elements required for regulation of SOCS-1 in response to LPS, promoter constructs were cloned using DNA isolated from a human cell line. The cloned sequence corresponds with the most common allele of human SOCS-1 (data not shown). The sequence ( Fig. 1) illustrates the SOCS-1 genomic locus from the Ϫ882-position relative to the translation ATG start codon. The intron/exon structure (shown by the boxed area) is defined by the cDNA sequence of SOCS-1 (GenBank TM accession number NM_003745). Both the human and murine sequence of SOCS-1 contain two exons, with the entire coding region present within the second exon.
Various regions of the human SOCS-1 promoter were cloned upstream of a luciferase reporter construct, and promoter activity was assayed in the RAW 264.7 macrophage cell line (Fig.  2). Previous experimental evidence has demonstrated that LPS is capable of inducing endogenous SOCS-1 expression in the RAW 264.7 cell line (5). Further, overexpression of SOCS-1 is also capable of inhibiting the LPS signaling pathway in the RAW 264.7 cell line (5). LPS treatment elicited a moderate increase in the transcriptional activity starting with the Ϫ2523/Ϫ659 region of the SOCS-1 promoter (Fig. 2). The Ϫ945/ Ϫ659 and Ϫ882/Ϫ659 regions of the SOCS-1 promoter demonstrated the most significant basal and LPS-induced activity. LPS stimulation yielded an approximate 7-fold increase in activity, which was diminished in promoter constructs lacking promoter sequence 3Ј of the Ϫ864-position. The Ϫ768/Ϫ659 construct was transcriptionally unresponsive to LPS; nor did it demonstrate any significant basal promoter activity. The Ϫ768/ Ϫ659 construct results were similar to the vector control lack-ing any SOCS-1 promoter sequence. Interestingly, we observed a significant decrease in promoter activity with the loss of promoter sequence between Ϫ850 and Ϫ840, indicating that this region of the promoter may contain an LPS-inducible cisacting element. These data indicate that the Ϫ864 to Ϫ659 region of SOCS-1 includes response elements required for the maximal LPS-induced transcriptional response of the promoter.
Identification of Egr-1 Binding Sites within the SOCS-1 Promoter-Initial efforts to characterize the DNA binding factors that regulate this promoter used probe sequences within the Ϫ2523/Ϫ659 region of SOCS-1, which resembled the NF-B consensus motif, by EMSA. No NF-B-or LPS-inducible DNA binding complexes were observed with any of the four putative sites tested, which suggests that NF-B does not mediate the LPS response of the SOCS-1 promoter (data not shown). To determine the identity of the transcription factors responsible for the LPS-induced activity, EMSAs were performed with labeled probe sequences from Ϫ882 to Ϫ727, which included sequences found to be functionally important in the luciferase reporter experiments. A series of seven partially overlapping sequences spanning this region were utilized as probes in EMSA with nuclear extracts from unstimulated and LPS-stimulated RAW cells (data not shown). Four probes formed complexes with Sp1 and Sp3, which were found to be unaltered in EMSA with unstimulated and LPS-stimulated cell extracts (data not shown). The sequences within these particular probes that match the GC-box consensus 5Ј-GGGCGG-3Ј are underlined in the SOCS-1 promoter region (Fig. 1). Importantly, two of these seven probes demonstrated LPS-induced complexes (Fig. 3A, lanes 2 and 3) (Fig. 3B, lanes 2 and 3); these probes are subsequently termed site 1 (S1) and site 2 (S2).
The S1 probe contained Ϫ864 to Ϫ838 of the SOCS-1 sequence, the region of the promoter that demonstrated the most significant loss in LPS-induced activity (Fig. 2), consistent with the notion that a cis-acting element may be located within this region. Analysis of the S1 probe sequence suggested a putative binding site 5Ј-CGCCACCGC-3Ј (Ϫ849 to Ϫ841), which conforms to the Egr-1 consensus binding motif 5Ј-GCGGGGGCG-3Ј (33). Binding of Egr-1 to S1 was confirmed with the addition of anti-Egr-1 antibody, which blocked the formation of the inducible complex. However, the antibody did not prevent binding of the slower migrating unknown complex (Fig. 3A, lane 5). Both the LPS-inducible complex and the unknown complex were competed away with excess unlabeled S1 probe (Fig. 3A, lanes 6 -8). The The translation ATG initiation site is depicted in boldface letters. The GAAA repeat IRF-E is illustrated in italic type. Four GC-boxes are underlined (GGGCGG consensus). Egr-1 site 1 and site 2 (GCGGGGGCG consensus) are shown in outline and labeled as S1 and S2, respectively. The proximal GC-box overlaps with S2. The transcription initiation site is denoted with a bent arrow, and the numbers indicate the relative position from the A(ϩ1) of the ATG initiation site. mutation in the Egr-1 binding element abolished the ability of excess unlabeled S1m to compete for binding of the LPS-induced complex; however, the unknown complex was competed away as with unlabeled S1 (Fig. 3A, lanes 9 -11). The LPS-inducible complex was also effectively competed away with unlabeled Egr-1 consensus sequence from the TNF-␣ promoter (31). These results suggest that Egr-1 binds to the Ϫ849 to Ϫ841 region of the SOCS-1 promoter.
Initial analysis of S2 found two potential Egr-1 binding sites overlapping a central GC-box: 5Ј-GGGCGG-3Ј (Ϫ737 to Ϫ732). The locations of these putative Egr-1 sites are 5Ј-GCAGGGGCG-3Ј (Ϫ741 to Ϫ733) and 5Ј-GCGGGCGCC-3Ј (Ϫ735 to Ϫ728) (Fig. 3C). The identity of Egr-1 in the LPSinducible complex was confirmed with the addition of an anti-Egr-1 antibody (Fig. 3B, lane 5). Furthermore, the complexes that bind to S2 with both unstimulated and LPS-stimulated extracts (Fig. 3, lane 2) were identified as Sp1 and Sp3 by supershift analysis (Fig. 3B, lanes 6 and 7). To determine the exact binding sites of these transcription factors, a number of mutant oligonucleotides were tested in cold competition assays. Unlabeled excess S2, S2m1, and S2m5 (Fig. 3B, lanes 8, 9, and  13) were capable of competing away the Sp1/Sp3 binding activity, yet the addition of S2m2, S2m3, and S2m4 (Fig. 3B,  lanes 10 -12) failed to inhibit binding. Hence, the binding site for these factors was determined to be from Ϫ737 to Ϫ732. Excess unlabeled S2, S2m4, and S2m5 (Fig. 3B, lanes 8, 12, and  13) were able to compete for Egr-1 binding activity, but the addition of S2m1, S2m2, and S2m3 (Fig. 3B, lanes 9 -11) did not impede the formation of the Egr-1⅐S2 complex. These results suggest that the Ϫ741 to Ϫ733 region of the SOCS-1 promoter binds Egr-1 as denoted by the solid line above S2, whereas the Ϫ735 to Ϫ728 region does not interact with Egr-1 as shown by the dashed line above S2 (Fig. 3C). Probe S2m3 (Fig. 3B, lanes  15 and 16) did not demonstrate significant complex binding in comparison with the S2 probe (Fig. 3B, lanes 2 and 3), confirming the inability of unlabeled S2m3 to impede the formation of both the Sp1-, Sp3-, and Egr-1-containing complexes (Fig. 3B,  lane 11). Collectively, these data from the EMSAs suggest the presence of two LPS-induced Egr-1 binding sites within the SOCS-1 promoter, one of which overlaps with a Sp1-and Sp3binding GC-box.

Results from the luciferase reporter experiments demonstrated a reduction in LPS-induced activity with the loss of the Ϫ840 to
Ϫ820 promoter sequence, as demonstrated by the comparison between the Ϫ840/Ϫ659 and Ϫ819/Ϫ659 constructs (Fig. 2). Sequence alignment of the murine and human SOCS-1 promoters demonstrated a high degree of homology within this GAAA repeat region (Fig. 4B). The ability of IFN-␥ to regulate the murine SOCS-1 promoter is mediated via this series of three tandem GAAA units (21). Consistent with the expectation that this region of the human promoter, termed the human IRF-E, is regulated in a manner similar to the murine IRF-E, the overexpression of IRF-1 increased activity to mimic IFN-␥ treatment, and the overexpression of IRF-2 negatively regulated the activity of the human SOCS-1 promoter (data not shown). Additionally, the loss of the IRF-E between the Ϫ840/ Ϫ659 and the Ϫ819/Ϫ659 luciferase constructs led to a signif-

Egr-1 Regulation of SOCS-1
icant reduction in the ability of IFN-␥ to increase transcription of the SOCS-1 promoter (data not shown). Further, the mutation of the central GAAA repeat (Ϫ833 to Ϫ830) to CTTT, as shown by the mIRF-E promoter construct, significantly reduces the activity of the promoter in response to IFN-␥ (as shown below in Fig. 6B). EMSA analysis was performed with the Ϫ840 to Ϫ813 region of the SOCS-1 promoter to determine whether IRF-2 is capable of binding to this region in the human SOCS-1 promoter (Fig. 4A). Murine thymocyte nuclear extracts form two separate complexes (Fig. 4A, lane 2), which are supershifted upon incubation with an anti-IRF-2 antibody (Fig. 4A,  lane 4). These complexes involve either one molecule of IRF-2 bound to probe (1IRF-2:DNA) or 3 molecules bound to probe (3IRF-2:DNA) as determined by binding of increasing amounts of purified recombinant IRF-2 (rIRF-2) (Fig. 4A lanes 5-9). In total, these results indicate that the IRF-1 and IRF-2 can bind the human SOCS-1 promoter via the IRF-E, and this region is required for the maximal promoter response to LPS.
Egr-1, IRF-1, and IRF-2 Bind the SOCS-1 Promoter in Vivo-To confirm the role of Egr-1 in LPS transcriptional regulation, ChIP assays were performed to determine whether this protein interacts with the SOCS-1 promoter in vivo. Stimulation of THP-1 cells with LPS increased binding of Egr-1 to the promoter, consistent with Egr-1 having a positive regulatory role in LPS-induced SOCS-1 transcription (Fig. 5A). Further, LPS increased binding of IRF-1 to the SOCS-1 promoter, supporting the idea that this stimulus utilizes both transcrip- FIG. 3. Egr-1 interacts with the SOCS promoter sequences in response to LPS. A and B, nuclear extracts from RAW cells, which were either unstimulated (Ϫ) or treated with 100 ng/ml LPS (ϩ) for 1 h, were used in EMSA with sequences from the SOCS-1 promoter. Antibodies (1.0 g) for the supershift assay, and excess unlabeled probe for cold competition were added for a 10-min room temperature incubation period prior to the addition of labeled probe. A, the labeled S1 probe (Ϫ864 to Ϫ838) is incubated with a 10-, 50-, and 100-fold molar excess of the indicated unlabeled probe. B, labeled S2 (Ϫ750 to Ϫ727) is incubated with a 200-fold molar excess of the unlabeled probes. The labeled S2m3 probe (Ϫ750 to Ϫ727) used as a control is shown on the right. C, alignment of the SOCS-1 sequence from the probes used in the cold competition. The Egr-1 and Sp1/3 sites defined by the supershift and cold competition experiments are depicted with a line over the respective sequences. The S2 region that is eliminated as a putative Egr-1 binding site is depicted with a dashed line.  5-9, respectively). B, alignment of the IRF-E sequences from the human and murine SOCS-1 promoters. The boxed regions indicate defined IRF binding sites (21).

Egr-1 Regulation of SOCS-1
tion factors to regulate SOCS-1 expression (Fig. 5A). Additionally, IFN-␥ increased IRF-1 binding activity to the SOCS-1 promoter, which is consistent with the previous results that implicated IRF-1 as a key regulator of IFN-␥-induced SOCS-1 expression (Fig. 5A). Primers were also used to amplify sequence 3Ј of the SOCS-1 gene (ϩ3387 to ϩ3583 relative to the ATG start codon and starting ϳ1.6 kb 3Ј of the poly(A) tract) to control for sufficient sonication of the DNA and nonspecific immunoprecipitation (Fig. 5A).
EMSA analysis implicated IRF-2 as a transcriptional regulator of the SOCS-1 promoter. Consistent with this observation, significant IRF-2 binding activity was demonstrated by ChIP analysis in unstimulated THP-1 cells, a human monocytic cell line (Fig. 5A). It has been shown that IRF-2 contains a Cterminal repression domain (44) and is capable of inhibiting transcription by preventing the recruitment of the CBP-polymerase II holoenzyme complex (45). Therefore, significant levels of IRF-2 binding to the SOCS-1 promoter in unstimulated cells would be consistent with low basal levels of SOCS-1 expression which have been observed in THP-1 cells (data not shown). Interestingly, treatment with IFN-␥ and LPS reduced this binding of IRF-2 by 32 and 70%, respectively (Fig. 5B). These data would suggest that direct competition by IRF-1 for one or more of the GAAA repeats within the IRF-E, which is bound by IRF-2 in unstimulated conditions, increases SOCS-1 transcription after stimulation with either LPS or IFN-␥.
Mutation of Egr-1 Binding Sites within the SOCS-1 Promoter-The S1 mutation and some of the S2 mutations tested by EMSA were generated within the context of the SOCS-1 pro-moter luciferase construct to determine the importance of the Egr-1 sites. Constructs bearing S1 and S2 mutations were tested alone or in combination with mutations at both S1 and S2 (Fig. 6A). Mutation of the distal site (S1) reduced the total LPS-induced transcriptional activity of the promoter 50%, whereas mutation of the proximal site (S2m1), which abolishes binding of Egr-1 but not Sp1/3, demonstrates a 33% reduction in the total activity. These results indicate that each site is responsible for a portion of the LPS-induced activity. Interestingly, the double mutant (S1S2m1), which contains both mutations, exhibits a 96% reduction in the total LPS-induced activity. In addition to this reduction in the total luciferase activity, the 9-fold activation of the wild type promoter is reduced to 7-, 8-, and 7-fold with the S1, S2m1, and S1S2m1 mutant promoters, respectively. The S1S2m1 Egr-1 mutant promoter construct demonstrates an 84% reduction in the IFN-␥-induced total promoter activity to suggest that Egr-1 is also a mediator of IFN-␥ activation of SOCS-1; however, there was a significantly higher -fold activation with the mutant (23) in comparison with the wild type promoter (15). These data suggest that although the mutation of both Egr-1 binding sites significantly reduced the basal and LPS-and IFN-␥-induced activity of the promoter, a degree of regulation is retained through the SOCS-1 IRF-E or additional cis-acting elements.
The S1S2m1 double Egr-1 site mutant was compared with the human promoter luciferase construct with a mutation within the central GAAA repeat motif of the IRF-E (Ϫ833 to Ϫ830), termed mIRF-E, to determine the relative importance of the IRF-E and the Egr-1 binding sites in response to LPS and FIG . 5. Egr-1, IRF-1, and IRF-2 bind the SOCS-1 promoter in vivo. THP-1 cells are untreated or activated with either 100 ng/ml IFN-␥ or 100 ng/ml LPS for 4 h prior to formaldehyde cross-linking. Cross-linked chromatin was immunoprecipitated with the antibodies indicated and extensively washed prior to elution and precipitation of DNA. A, quantitative PCR was then performed on the input samples, which represent 0.05, 0.01, and 0.002% of the total starting DNA, and the immunoprecipitations, which include 10, 5, and 2.5% of the resuspended DNA. The PCR was performed with primers for the SOCS-1 promoter region to amplify the sequence from Ϫ1014 to Ϫ813 or with primers to amplify the ϩ3387 to ϩ3583 region 3Ј of the SOCS-1 gene. B, graph depicting the quantitation of the band intensities in the ChIP assay as determined by NIH Image version 1.62. The immunoprecipitation faction amplifications were normalized to the individual input intensities and subsequently plotted relative to unstimulated IRF-2. Only the dilutions of fractions that fell in the linear range were used in the calculations.

Egr-1 Regulation of SOCS-1
IFN-␥. Mutation of the central GAAA to CTTT in the context of the murine SOCS-1 promoter had previously been shown to abolish the ability of IRF-1 to both bind oligonucleotide sequences by EMSA and induce transcription when overexpressed in luciferase experiments (21). The mIRF-E promoter construct demonstrated a 63 and 85% reduction in LPS-and IFN-␥-induced activity, respectively. This construct responded to LPS to suggest that a portion of LPS-induced activity of the SOCS-1 promoter is mediated through S1 and S2 or additional cis-acting elements. A promoter mutated in all three binding sites, S1S2m3mIRF-E, was generated to determine whether the remainder of the activity was dependent upon the Egr-1 sites. The S1S2m3mIRF-E triple mutant did not demonstrate significant levels of basal transcription or a response to either LPS or IFN-␥. These data demonstrate that all three sites are required for SOCS-1 promoter activity and suggest that Egr-1 and IRF-1 are important factors for LPS-and IFN-␥-induced SOCS-1 transcription.
Mutation of the Sp1/Sp3-binding site lowered both the basal promoter activity and the total activity of the promoter in response to LPS, as demonstrated by the S2m4 and S1S2m4 constructs. However, the LPS-induced response appears to be mediated through the Egr-1 binding site rather than the Sp1/ Sp3 site, because S1S2m4 retains a moderate response to LPS in contrast to the reduction in the S1S2m1 construct. The S1S2m3 promoter with mutations in both Egr-1 sites and the Sp1/Sp3 site curiously demonstrated a slightly higher response to LPS than S1S2m1, which may be explained by the ability of Sp3 to negatively regulate certain promoters (46). Reduced Levels of LPS-induced SOCS-1 in Egr-1-deficient Macrophages-In order to determine whether Egr-1 is important in mediating LPS-induced SOCS-1 expression, cells were isolated from Egr-1 Ϫ/Ϫ mice and assayed for SOCS-1 levels in comparison with Egr-1 ϩ/ϩ littermate controls. Freshly isolated macrophages and splenocytes do not demonstrate a measurable induction of SOCS-1 in response to LPS; therefore, to assess LPS-induced responses, macrophages were derived from the bone marrow of these mice. Consistent with published reports Egr-1 Regulation of SOCS-1 that indicate that macrophage differentiation in Egr-1-null bone marrow appears normal (39), fluorescence-activated cell sorting analysis demonstrated that these bone marrow-derived macrophages had equivalent levels of CD11b expression after 7 days in macrophage colony-stimulating factor (data not shown). Northern blot analysis demonstrated that SOCS-1 mRNA expression levels were reduced in the Egr-1 Ϫ/Ϫ macrophages after 2, 3, and 4 h of LPS stimulation (Fig. 7A, lanes [11][12][13] in comparison with cells derived from Egr-1 ϩ/ϩ controls (Fig. 7A, lanes 4 -6). This difference (3-fold) was most pronounced at 3 h post-LPS stimulation (Fig. 7B), which is consistent with the rapid induction of Egr-1 in response to this stimulus. Isolated splenocytes from Egr-1 Ϫ/Ϫ mice demonstrated slightly lower basal levels of SOCS-1 protein (Fig. 7,  lane 2); however, IFN-␥-induced levels were not reduced (Fig.  7C, lane 4). Additionally, the LPS-induced levels of SOCS-1 protein were 34% lower in Egr-1 Ϫ/Ϫ macrophages (Fig. 7C, lane 8) than in Egr-1 ϩ/ϩ macrophages (Fig. 7C, lane 7). This observation confirms that the reduction in LPS-induced SOCS-1 mRNA levels is also observed at the level of protein expression. Taken together, these results indicate that basal and LPSinduced levels of SOCS-1 expression are dependent upon the Egr-1 transcription factor and that there also exists an Egr-1independent pathway for SOCS-1 induction by LPS. DISCUSSION In addition to the extensive amount of literature that describes the role of SOCS-1 inhibiting JAK-STAT signaling (reviewed in Ref. 47), recent published evidence has shown a role for SOCS-1 as an important negative regulator of LPS signaling (4,5). This report demonstrates a clear role for the immediate early gene Egr-1 in LPS induction of SOCS-1. Two Egr-1 binding sites, S1 (Ϫ849 to Ϫ841) and S2 (Ϫ741 to Ϫ733), are required for the maximal transcriptional response to LPS stimulation. Mutation of either of these sites lowers the LPS-induced activity of the promoter, whereas the double mutant demonstrates a severe loss of both basal and LPS-induced activity. The Sp1/Sp3-binding GC-box (Ϫ737 to Ϫ732) is also important for maintaining the activity of the promoter but is unlikely to be responsible for mediating the LPS response. This is an important distinction, because Sp1 can mediate the induction of other promoters; one such example is the interleukin-10 promoter in LPS-stimulated macrophages (48). Additionally, most previously illustrated examples of overlapping Egr-1 and Sp1 sites implicated Egr-1 as having a role as a transcriptional repressor through the displacement of Sp1 (49 -53). Only a smaller number of overlapping sites exhibit Egr-1mediated transactivation (54). In most instances, a role in transcriptional repression for Egr-1 has been defined by Egr- 1   FIG. 7. Egr-1 is required for normal SOCS-1 levels in response to LPS. A, SOCS-1 mRNA levels were examined by Northern blot in bone marrow-derived macrophages isolated from an Egr-1 Ϫ/Ϫ mouse or its Egr-1 ϩ/ϩ littermate control. Cells were treated with 100 ng/ml LPS or 100 ng/ml IFN-␥. B, levels of SOCS-1 mRNA were quantitated by densitometry and normalized to the 28 S ribosomal expression level using NIH Image version 1.62. C, SOCS-1 protein levels were monitored by Western blot in splenocytes (Spl) that were isolated from an Egr-1 Ϫ/Ϫ mouse and its Egr-1 ϩ/ϩ littermate control. Cells were cultured for 4 h with or without 100 ng/ml IFN-␥. Bone marrow-derived macrophages (M⌽) from these mice were cultured with or without 100 ng/ml LPS for 4 h. Band intensities were quantitated using densitometry and normalized to the STAT1 loading control. The percentage of wild type values indicates the normalized SOCS-1 intensities of the Egr-1 Ϫ/Ϫ cells versus the Egr-1 ϩ/ϩ controls.

Egr-1 Regulation of SOCS-1
overexpression or with the activation of signaling pathways that negatively regulate promoters rather than activation with stimuli that positively regulate transcription.
The phenotype of the Egr-1 Ϫ/Ϫ mice suggests that other regulatory factors are involved in the induction of SOCS-1 because animals from an acute model of endotoxemia demonstrate only slightly reduced survival rates (26). If the regulation of SOCS-1 were entirely dependent upon Egr-1, the Egr-1 Ϫ/Ϫ mice would demonstrate a dramatic susceptibility to endotoxin challenge, a phenotype displayed by both the SOCS-1 ϩ/Ϫ and SOCS-1 Ϫ/Ϫ animals. It may be that there exist other disrupted regulatory pathways that compensate for the lack of SOCS-1 induction. However, the simplest explanation is that there must be other mechanisms for the expression of SOCS-1 in the absence of Egr-1; for example, additional regulatory elements situated outside of the promoter region that we evaluated, regulatory elements within the promoter that were not induced by LPS in our system, or regulation via the IRF-E in the absence of Egr-1 is sufficient to promote SOCS-1 transcription. This would also be consistent with our observations that the levels of SOCS-1 in Egr-1 Ϫ/Ϫ macrophages, although reduced, can still be induced by LPS stimulation.
Our study not only identified Egr-1 as a regulator of SOCS-1 but also clarified the role of IRF-1 and IRF-2 in response to LPS. ChIP experiments demonstrated a clear increase in the binding of the SOCS-1 promoter by IRF-1 in response to both LPS and IFN-␥. Concomitant with this observed increase was a reduction in the promoter-associated IRF-2. These data confirmed the respective positive and negative regulation of the SOCS-1 promoter by IRF-1 and IRF-2 in vivo. Interestingly, IRF-2 Ϫ/Ϫ mice are extremely resistant to LPS-induced shock (55). Our data would suggest that this resistance may be due to elevated levels of SOCS-1 in the absence of the repressor IRF-2.
It has been proposed that LPS does not directly induce SOCS-1 but rather that its expression is dependent upon the LPS-induced autocrine/paracrine factors IFN-␣/␤ that would in turn up-regulate SOCS-1 expression (19,20). Consistent with type I interferon being a major regulator of LPS-induced SOCS-1 expression, reduced levels are observed with the addition of anti-IFN-␣/␤ antibodies (19). Interestingly, a strong correlation exists between the regulation of IRF-1 and that of SOCS-1 in response to LPS. Similar to the regulation of SOCS-1, LPS-induced IRF-1 levels are reduced with the addition of anti-IFN-␣/␤ antibodies or with the use of IFN-␣/␤ R Ϫ/Ϫ macrophages (56). The significant reduction in LPS-induced SOCS-1 promoter activity with the mutation of the IRF-E suggests that a significant portion of the activation is mediated by IRF-1. Further, levels of LPS-induced IRF-1 are significantly reduced in STAT1 Ϫ/Ϫ macrophages (56) to suggest that the LPS induction of IRF-1 expression is dependent upon type I interferon's activation of STAT1 rather than the direct activation of IRF-1 via NF-B.
SOCS-1 was originally identified as a regulator of adaptive cytokine signaling pathways but has recently been shown to be important in regulating innate immune responses. The identification of an IRF-1-independent signaling pathway is important in understanding how SOCS-1 is regulated in response to stimuli that do not activate the presumed JAK/ STAT1/IRF-1 cascade. Understanding the transcriptional regulation of SOCS-1 in response to various stimuli will no doubt be critical in understanding its role in limiting inflammatory responses. Collectively, these data presented here suggest multiple mechanisms for the regulation of SOCS-1: pathways dependent upon JAK/STAT and pathways dependent upon mitogen-activated protein kinases, which mediate signals via IRF-1 and Egr-1, respectively.