Myeloid ELF1-like Factor Is a Potent Activator of Interleukin-8 Expression in Hematopoietic Cells*

Myeloid ELF1-like factor (MEF), also known as ELF4, is a member of the ETS family of transcription factors which is expressed in hematopoietic cells. MEF-defi-cient mice have defects in natural killer cell and natural killer T cell development, suggesting a role for MEF in regulating innate immunity. MEF also functions in myeloid cells, where it can transactivate target genes. To identify MEF target genes in a “myeloid” environment, we created an inducible expression system and used oligonucleotide microarrays to examine the transcript profile of HEL cells after induction of MEF expression. Sixteen genes were reproducibly turned on or off more than 2-fold, 8 h after induction of MEF expression, and we examined one of the genes, interleukin-8 (IL-8), in greater detail. IL-8 is a CXC chemokine involved in neutrophil chemoattraction, angiogenesis, and stem cell mobilization. It is expressed by several tumor types, and its expression is regulated primarily transcriptionally. The IL-8 promoter contains three ETS binding sites, and we identified the specific site that binds MEF and is required for MEF responsiveness. MEF, but not the closely related ETS factors PEA3, ETS1, ETS2, ELF1, or PU.1, strongly activates the IL-8 promoter. MEF a p (cid:1)

Myeloid ELF1-like factor (MEF), 1 also known as ELF4, is a member of the ETS family of transcriptional regulators and was originally isolated from a myeloid hematopoietic cell line (1). ETS proteins generally regulate transcription in coordination with other transcription factors, binding the core ETS binding sequence GGAA, via their ETS domain. Because this sequence occurs frequently in genomic DNA, the identification of true ETS target genes has been difficult. Furthermore, the significant homology in the ETS domains of these proteins makes it daunting to determine which ETS factor regulates a particular gene in vivo. Although cytokine and cytokine receptor genes and many other genes have been proposed as targets of various ETS factors (for review, see Ref. 2), their validation as in vivo targets has rarely been performed.
Microarrays have been used extensively to profile the transcripts expressed in tumor or normal tissue, both as a way to define their biology and also to determine which genes serve as determinants for their classification. We and others have used microarrays to identify genes regulated by particular transcription factors such as C/EBP␣ (3), c-Myc (4), WT-1 (5), and BRCA-1 (6), among others. These studies have identified target genes not previously described and have failed to find gene targets previously identified by reporter gene assays. Thus, subsequent validation of direct (and indirect) targets will generally require the use of multiple additional approaches to confirm the results.
MEF is a member of the Drosophila E74 subfamily of ETS proteins which includes the mammalian NERF and ELF1 proteins. We have shown that MEF regulates cytokine gene expression, including GM-CSF and IL-3 (1). Other ETS factors have been shown to regulate the expression of chemokines, such as RANTES (regulated on activation normal T cell expressed and secreted) (7) and PF4 (8), and cell surface receptors including the M-CSF receptor (9), G-and GM-CSF receptors (10,11) and CXCR1 (12). Although MEF and ELF1 are highly homologous, especially in their ETS domain, and they bind to the same DNA recognition sequence, MEF appears to be a more potent transactivator of promoter function (1). MEF also is required for perforin expression by NK cells but not activated cytotoxic T lymphocytes (34).
To identify MEF targets in hematopoietic cells, we established an inducible expression system in HEL cells, which have no detectable MEF mRNA or protein. We identified a number of MEF target genes using Affymetrix microarrays and focused primarily on IL-8, a member of the chemokine family with pleiotropic functions, including roles in stem cell mobilization (13), angiogenesis, and chemotaxis (14). The promoter of IL-8 has been identified previously, and both nuclear factor-B and c-Jun have been shown to regulate its activity (15). IL-8 is expressed by a variety of tumor types and is inducibly expressed in several myeloid leukemia cell lines after 12-O-tetradecanoylphorbol-13-acetate stimulation (16). PEA3 has been suggested to be the ETS factor that controls IL-8 expression, although direct effects of PEA3 on IL-8 promoter function were not assessed (17). A recent study showed that MEF and ETS2 antagonistically regulate IL-8 expression in a lung carcinoma cell line, with MEF acting as a repressor (18). We show that MEF strongly activates the IL-8 promoter in hematopoietic and nonhematopoietic cells and that overexpression of MEF is sufficient to increase significantly the level of IL-8 protein expression. Furthermore, knock-down of MEF expression in cells that express both MEF and IL-8 by RNA interference, decreases IL-8 expression, demonstrating the physiologic relevance of MEF in regulating IL-8 production.

Generation of Cells with Inducible MEF Expression-HEL cells and
NB4 -306 (a subclone of NB4, which is ATRA-resistant) cells were maintained in RPMI 1640 medium and COS-7 cells in Dulbecco's modified Eagle's medium, all supplemented with 10% fetal calf serum, penicillin, streptomycin, and glutamine. HEL cells were electroporated (using a Gene Pulser (Bio-Rad) set at 250 mV, 960 microfarads) with the pVgRXR plasmid (Invitrogen), which expresses a fusion of the ecdysone receptor and the VP-16 transactivating domain and the RXR cDNA. Stably transfected clones were then selected in 500 g/ml Zeocin (Invitrogen). These cells were electroporated with the pIND-MEF vector (which can express MEF in response to the active ecdysone receptor fusion protein), and clones were selected in 100 g/ml hygromycin to isolate double transfected cells. To induce MEF expression, HEL/IND-MEF cells were incubated with the synthetic ecdysone analog, ponasterone A (Invitrogen), at a final concentration of 5 M for 4, 8, 12, and 24 h. MEF levels were assessed by Western blot analysis using a 1:1,000 dilution of the MEF polyclonal antiserum (which is described elsewhere (19)).
Oligonucleotide Array-based Expression Profiling-Total RNA was isolated from HEL/IND-MEF cells after an 8-h incubation with 5 M ponasterone A, using TrIzol (Invitrogen) followed by RNeasy purification (Qiagen). The RNA was amplified, labeled, and hybridized to oligonucleotide arrays, as described elsewhere (20,21), using the Hu-GeneFL Affymetrix GeneChip (Affymetrix, Santa Clara, CA) containing probe sets for 6,800 genes and expressed sequence tags. The results were analyzed with Affymetrix microarray suite (MAS 5.0). Only genes showing greater than 2-fold induction or repression, called "present" in the induced sample and called "induced" or "repressed" at a p Ͻ 0.001 by MAS 5.0, compared with base-line expression in HEL/VgRXR cells incubated with ponasterone A, were chosen for further consideration.
Plasmid Construction-The MEF cDNA containing the entire open reading frame was cloned into the HindIII/XhoI site of the pIND vector to generate pIND-MEF. The 5Ј-region (Ϫ165 to ϩ58) of the human IL-8 promoter was generated by PCR using genomic DNA as a template. The 5Ј-primer contained an XhoI site to facilitate cloning (5Ј-CCCTCGAG-CATACTCCGTATTTGATAAGGAAC), and the 3Ј-PCR primer contained an HindIII site (5Ј-GGCTCTTGTCCTAGAAGCTT). The PCR product was then digested with XhoI and HindIII and subcloned into pGL3Basic (to create pGL3-Basic-IL8). To mutate ETS binding sites in this construct, the QuikChange mutagenesis kit (Promega, Madison, WI) was used with the following primers (the mutated bp are in lowercase and underlined): for EBS1, 5Ј-CATCAGTTGCAAATCGTGacATT-TCCTCTGACATAATG and 5Ј-CATTATGTCAGAGGAAATgtCACGTA-TTTGCAACTGATG; for EBS2, 5Ј-GGAACAAATAGacAGTGTGATGA-CTCAGG and 5Ј-CCTGAGTCATCACACTgtCTATTTGTTCC; and for EBS3, 5Ј-CTCCGTATTTGATAAGacACAAATAGGAAGTGTG and 5Ј-CACACTTCCTATTTGTgtCTTATCAAATACGGAG. The DNA sequence of all wild type and mutated IL-8 promoter constructs was confirmed. PEA3 was cloned by reverse transcription PCR from K562 cells using the Superscript one-step reverse transcription PCR system (Invitrogen) and the following primers: 5Ј-ATGGAGCGGAGGATGAAAGC and 5Ј-ATG-GAGCGGAGGATGAAAGC. The PEA3 cDNA insert was sequenced (anti-PEA3, Santa Cruz, SC-113, mouse monoclonal IgG1). The ETS1 and ETS2 plasmids were generous gift from Dr. Dennis Watson. The ELF1 cDNA was kindly provided by Craig Thompson and the PU.1 cDNA by Robert Maki. Protein expression from each of the expression vectors was confirmed by Western blot analysis of transiently transfected COS-7 cell extracts (data not shown).
Luciferase Reporter Assays-5 g of either the pCMV5 cytomegalovirus promoter-based expression vector alone or pCMV5 containing the MEF, ETS1, ETS2, PU.1, ELF1, or PEA3 cDNA (1) was used to cotransfect COS-7 cells using the calcium phosphate precipitation method together with 1 g of pGL3-Basic-IL8 and 10 ng of pRL-CMV control plasmid, used to control for transfection efficiency. HEL cells were electroporated in RPMI with 10% fetal calf serum without antibiotics at 250 mV, 960 microfarads with 10 g of reporter plasmid, 10 g of expression vector, and 100 ng of pRL-CMV. Cell lysates were prepared 24 h after transfection, and 10 l of lysate was assayed using the dual luciferase assay system according to the manufacturer's protocol (Promega).
Western Blot Analysis-Total cell lysates were prepared from COS-7 cells or HEL cells using radioimmune precipitation assay buffer, and protein concentrations were determined using Bio-Rad protein assay reagent. Samples were boiled, separated by SDS-PAGE, and trans-ferred to Hybond-ECL membranes (Amersham Biosciences) for blotting with rabbit polyclonal anti-MEF antiserum (previously generated (19)) using goat anti-rabbit horseradish peroxidase (Amersham Biosciences) as the secondary antibody. Blots were developed using the SuperSignal West Dura Extended Duration Substrate (Pierce).
Measurement of IL-8 Protein-Conditioned medium was collected from HEL/VgRXR and HEL/Ind-MEF cells at various time points after stimulation with ponasterone A and assayed using an IL-8 Quantikine kit (R&D Systems, Minneapolis) per the manufacturer's directions.
Real Time PCR Assay-One g of total RNA was processed to cDNA by reverse transcription with Superscript II (Invitrogen) and an oligo(dT) primer, according to the manufacturer's protocol. All PCR primers and TaqMan probes were purchased from Applied Biosystems (MEF, Hs00154964_m1; IL-8, Hs00174103_m1) and were labeled with the reporter dye FAM or VIC. Amplification reactions contained 10 ng of cDNA and were performed using the Universal TaqMan 2X PCR master mix (Applied Biosystems) in a total volume of 50 l. All reactions were performed in triplicate using the ABI Prism 7700 sequence detection system (Applied Biosystems); the thermal cycling conditions were 2 min at 50°C, 10 min at 95°C, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. Hypoxanthine-guanine phosphoribosyltransferase amplification was used to normalize the expression data for IL-8. Primers for amplifying the IL-8 and hypoxanthine-guanine phosphoribosyltransferase cDNA were obtained from Applied Biosystems.
Chromatin Immunoprecipitation-2 ϫ 10 7 NB4 -306 or HEL cells were processed using the chromatin immunoprecipitation assay kit (Upstate Biotechnology, Lake Placid, NY) according to the manufacturer's instructions. Briefly, asynchronously growing NB4 -306 or HEL cells were cross-linked with 1% formaldehyde for 10 min, and the reaction was stopped by the addition of glycine. The cells were washed in Tris-buffered saline and harvested in SDS buffer. After centrifugation, the cells were resuspended in immunoprecipitation buffer and sonicated (Fisher Scientific, model 500) to yield DNA-protein complexes with an average length of DNA of 300 -500 bp. The lysate was precleared and then immunoprecipitated at 4°C overnight with 1-2 g of polyclonal antibodies specific for MEF, acetyl-histone H3 (Upstate Biotechnology, 06 -866), or c-Jun (SC-122), PEA3 (SC-113), ETS2 (SC-112), ELF1 (SC-631), and PU.1 (SC-352) (all from Santa Cruz). To recover immune complexes, blocked protein A beads were added and incubated for 2 h at 4°C. The beads were washed thoroughly, the complexes eluted, the cross-links reversed, and the material recovered by phenolchloroform extraction and ethanol precipitation. The amount of IL-8 genomic DNA in the immunoprecipitate was determined by PCR using the following oligonucleotide primers (which amplify a region spanning all three EBSs): 5Ј-CCCTCGAGCATACTCCGTATTTGATAAGGAAC and 5Ј-GGCTCTTGTCCTAGAAGCTT.
Short Interfering (si)RNA Preparation-21-nucleotide-long RNA molecules were chemically synthesized by Dharmacon (Lafayette, CO). The complementary oligonucleotides were 2Ј-deprotected, annealed, and purified by the manufacturer. The siRNA sequence selected for these experiments was compared with the human genome sequence using a BLAST search to ensure specificity for the desired sequence. The sequence targeting enhanced GFP was 5Ј-CCUGUCUCUCAGUA-CAAUCUU-3Ј (antisense); 5Ј-GGCUACGUCCAGGAGCGCACC-3Ј (sense), and MEF was 5Ј-GGAGGUACUGAAGAUCUGCUU-3Ј (antisense); 5Ј-AAGCAGAUCUUCAGUACCUCC-3Ј (sense). 200 pmol of each siRNAs was introduced by electroporation of NB4 -306 cells in RPMI ϩ 10% fetal calf serum without antibiotics at 250 mV, 960 microfarads daily for 2 consecutive days. RNA was extracted (as above) 72 h after the first electroporation. MEF and IL-8 mRNA levels were determined by real time PCR.

RESULTS
MEF Target Gene Identification Using Microarrays-We used high density oligonucleotide microarrays to search for genes whose expression is altered after the inducible expression of MEF in HEL human erythroleukemia cells (Fig. 1A). In these cells with tightly regulated, ponasterone A-inducible MEF expression, MEF protein is not present in the absence of ponasterone A but is detectable by Western blot as early as 2 h postinduction (Fig. 1B). To identify direct transcriptional targets, cells were grown in the presence of ponasterone A for 8 h, at which time total RNA was isolated and used to interrogate oligonucleotide arrays representing 6800 known genes and expressed sequence tags. Expression profiles in cells expressing MEF were compared with ponasterone A-stimulated HEL cells that contain the RXR/EcR heterodimer alone and do not inducibly express MEF (data not shown).
Using strict criteria, 13 genes were up-regulated greater than 2-fold, and three genes were down-regulated (Table I). If the -fold change threshold is lowered to 1.5 times, the list was expanded to 39 genes that were up-regulated and 14 genes that were downregulated (data not shown). The most highly induced gene was IL-8 (which increased an average of 11.7-fold). Other genes induced by MEF included another chemokine, CXCL2 (gro-beta) and several other genes important to hematopoietic function including c-kit, MDR1, and phosphatidylinositol 3 kinase.
MEF Regulates the IL-8 Promoter-Given the magnitude of IL-8 mRNA induction by MEF, we sought to determine whether the IL-8 promoter might constitute a direct transcriptional target of MEF. We generated a reporter gene construct containing the Ϫ165 to ϩ44 region of the IL-8 promoter, which contains three potential EBSs upstream of the luciferase gene. Cotransfection assays in both COS-7 cells and HEL cells showed that MEF activates the IL-8 promoter (72-fold in COS-7 cells and 14-fold in HEL cells). In contrast, none of the related ETS transcription factors ELF1, PEA3, ETS1, ETS2, or PU.1 transactivates this construct more than 2-fold (Fig. 2, A  and B). Mutation of two of the three potential ETS binding sites in the IL-8 promoter (EBS2 and EBS3), did not alter its MEF responsiveness or its basal activity (Fig. 2C). However, MEF responsiveness was lost when the most 3Ј-ETS binding site (EBS1) was mutated. Mutations in the individual EBSs did not significantly affect the basal IL-8 promoter activity (mEBS1, 1.6-fold; mEBS2, 1.2-fold, mEBS3, 1.3-fold relative to the construct with three wild type EBSs).
To determine whether the EBS1 ETS binding sequence was responding directly to the expression of MEF, we performed EMSAs to examine the binding of in vitro translated MEF (or PEA3) to EBS1, EBS2, and EBS3. Sequence-specific binding of MEF to the radiolabeled EBS1 probe was observed (Fig. 3A) which was competed by the wild type EBS1 oligonucleotide but not by a mutant EBS1 (mEBS1) or the EBS2/3 oligonucleotide that contains the other two ETS binding sites. MEF does not bind to EBS2/3 (Fig. 3B, lane 9). However we confirmed the work of Iguchi et al. (17) who showed that PEA3 binds to the 5Ј-AGGAAG-3Ј sequence within EBS2 (Fig. 3B, lane 11). PEA3 also binds to EBS1 (Fig. 3B, lane 5); however, the IL-8 pro-  MEF Is Bound to the IL-8 Promoter in Vivo-To establish whether MEF is bound to the IL-8 promoter in vivo, chromatin immunoprecipitation assays were performed using anti-MEF antiserum. Chemical cross-linking of chromatin from NB4 -306 cells, which are known to express constitutively both MEF and IL-8 (16), was followed by immunoprecipitation with the indicated antibodies (Fig. 3C). Using PCR to amplify the Ϫ165 to   5, 6, 11, and 12), or rabbit reticulocyte lysate alone (lanes 1, 2, 7, and 9) were incubated with an EBS1 or EBS2/3 oligonucleotide probe labeled with ␥-32 P. The reactions were done in the absence or presence of a 100-fold excess of unlabeled competitor oligonucleotide. The specific protein-probe complexes are indicated by arrows.

MEF Expression Increases IL-8 Protein
Production-To confirm our microarray data and define the time course of IL-8 expression in response to MEF induction, we performed real time PCR to monitor changes in the level of IL-8 mRNA (Fig.  4A). IL-8 mRNA is expressed at a minimal level in HEL/IND-MEF cells, increases when MEF is induced at 4 h (27.5-fold), and is ϳ10-fold higher than its baseline at 8 h. Given the profound transactivating effects of MEF on the IL-8 promoter, we sought to determine whether MEF overexpression was sufficient to drive the expression of IL-8 protein in cells. We performed enzyme-linked immunosorbent assays on conditioned media obtained from HEL/IND-MEF cells at various time points after MEF induction (Fig. 4B). IL-8 significantly accumulates in the medium (ϳ52-fold compared with control at 8 h), demonstrating that overexpression of MEF alone is sufficient to induce IL-8 protein expression. The concentration of IL-8 in the medium, which ranged from 1.9 to 6.8 nM during the 24-h period, is biologically relevant as it exceeds the concentration required for high affinity binding to the CXCR1 and CXCR2 IL-8 receptors (K d ϭ 2 nM).
Effect of siMEF on Endogenous IL-8 Expression in NB4 -306 Cells-To determine whether MEF is an important regulator of IL-8 expression in vivo, we used RNA interference to knockdown the level of MEF mRNA (and protein) in NB4 -306 cells.
For these studies, we used 21-nucleotide siRNA complementary to MEF sequences, whereas siRNA duplexes specific for GFP were used as a control. Introduction of MEF siRNA into NB4 -306 cells decreased MEF mRNA levels by ϳ90% (Fig. 5). This reduction of MEF expression resulted in a 40% decrease in IL-8 mRNA expression, demonstrating the physiologic role of MEF in regulating IL-8 expression. DISCUSSION The identification of true ETS target genes is complicated by the simplicity of their consensus recognition elements (GGAA), which are present at high frequency throughout genomic DNA and by the great number of related ETS proteins, which are often coexpressed in several different cell types. We have used an inducible expression system in the human HEL leukemia cell line to identify MEF target genes in vivo and have extensively characterized the promoter of one of these genes, namely IL-8. Several of these genes play important roles in hematopoietic cell growth and function including c-kit, phosphatidylinositol-3 kinase, and MDR1, but it will be necessary to further evaluate each of these genes, to examine the importance of MEF in their constitutive or induced expression.
Several of the genes activated by MEF in this system have been reported previously to be regulated by ETS proteins including c-kit (22) and IL-8 (17). We reported previously that MEF can transactivate the IL-3 and GM-CSF promoters in transient transfection studies (1); however, neither gene was identified as an MEF target in this study, possibly because of the lack of specific cofactors required for their induction. In fact, IL-3 and GM-CSF expression was not found in HEL cells by microarray either in the absence or presence of MEF.
We have shown that MEF, and not several other ETS factors, strongly activates IL-8 protein expression. IL-8 is produced by myeloid and lymphoid leukemia cells, both constitutively and in response to 12-O-tetradecanoylphorbol-13-acetate (16,23). Previous studies have suggested that PEA3 may mediate the induction of IL-8 expression through one of these ETS response elements in hepatoma cells (17). However, no transactivation studies were performed. We have shown that PEA3 binds to IL-8 promoter sequences in vitro, but not in vivo, and it does not activate the promoter. In contrast, overexpression of MEF strongly positively regulates IL-8 expression in hematopoietic cells, and reduction in MEF levels using RNA interference, in hematopoietic cells that express both MEF and IL-8 (NB4 - 306), leads to a significant reduction in IL-8 levels, supporting a critical role for MEF in IL-8 regulation. IL-8 is known to act as a potent neutrophil chemoattractant, and it can function as a proangiogenic factor in some systems. Although perhaps counterintuitive, myeloid leukemias appear to have some dependence on angiogenesis, as they express proangiogenic molecules, and increased bone marrow high microvessel density may confer a poor prognosis in AML (24). Expression of MEF by myeloid (and lymphoid) leukemia cells could increase IL-8 expression, stimulating bone marrow angiogenesis, and enhancing leukemia cell growth and survival in the marrow microenvironment. Although the best prognosis AMLs (FAB subtypes M2 and M3) have the lowest level of MEF mRNA (25), IL-8 production has not been quantitatively examined or correlated with specific FAB subtypes of AML.
Vascular endothelial growth factor (26) is an angiogenic factor secreted by leukemic blasts whose expression correlates with microvessel density, and blockade of vascular endothelial growth factor signaling is currently being evaluated as anticancer therapy. Other factors, such as IL-8, may play a role in the angiogenic process as well and could serve as another possible target. Although many murine chemokines have been identified, there is no known murine homolog of IL-8. Even the closely related chemokine, KC, which shares some functional overlap with IL-8, does not appear to be its direct murine homolog. This limitation has precluded further use of mouse models, such as generating "IL-8-deficient" mice to define its role in angiogenic and other processes. Likewise, the lack of a murine homolog has precluded us from using MEF-deficient mice to characterize further its role in IL-8 expression. IL-8 induces the rapid mobilization of hematopoietic progenitor cells in mice and primates (13,27), and gro-beta (CXCL2), and its truncated form, gro-beta-T are also involved in hematopoietic stem cell mobilization (28,29). Stem cell mobilization is currently being investigated in MEF-deficient mice.
MEF was recently reported to function as a transcriptional repressor of IL-8 promoter activity (18); however, the same group showed that MEF induces lysozyme gene expression (30,31). The difference in our results may be the because of critical cell type-specific effects. Although many genes were induced by MEF in our assay, only three genes were repressed, suggesting that MEF most often serves to activate gene expression. Some transcription factors act as repressors or activators, depending on the specific promoter context (e.g. AML1 or NF-IL3/E4BP4) or specific cofactor associations. For example, AML1 associates with ETS proteins to activate the Moloney murine leukemia virus enhancer (32) but associates with corepressor molecules such as mSin3A to repress transcription of the p21 promoter (33).
We have shown that MEF is bound to the IL-8 promoter in vivo and is a strong activator of IL-8 expression in hematopoietic cells. Although other ETS factors such as PEA3 may bind to the IL-8 promoter and regulate its expression in other contexts, MEF appears to play a most important role in regulating IL-8 production. The combined use of an inducible expression system with expression profiling represents a convergence of powerful techniques that allowed us to identify direct MEF transcriptional targets despite the significant overlap in ETS gene expression and the 3-or 4-nucleotide ETS binding recog-nition sequence ( . . . GGA(A/T) . . . ) present in innumerable locations throughout the genome.