The Cxcl12, Periostin, and Ccl9 Genes Are Direct Targets for Early B-cell Factor in OP-9 Stroma Cells*

The development of blood cells from hematopoietic stem cells in the bone marrow is dependent on communication with bone marrow stroma cells, making these cells central for the appropriate regulation of hematopoiesis. To identify transcription factors that may play a role in gene regulation in stroma cells, we performed comparative gene expression analysis of fibroblastic NIH3T3 cells, unable to support hematopoiesis in vitro, and OP-9 stroma cells, highly efficient in this regard. These experiments revealed that transcription factors of the early B cell factor (EBF) family were highly expressed in OP-9 cells as compared with the NIH3T3 cells. To identify potential targets genes for EBF proteins in stroma cells, we overexpressed EBF in fibroblasts and analyzed the pattern of induced genes by microarray analysis. This revealed that EBF was able to up-regulate expression of among others the Cxcl12, Ccl9, and Periostin genes. The identification of relevant promoters revealed that they all contained functional EBF binding sites able to interact with EBF in OP-9 cells. Furthermore, ectopic expression of a dominant negative EBF protein or antisense EBF-1 RNA in OP-9 stroma cells resulted in reduced expression of these target genes. These data suggest that EBF proteins might have dual roles in hematopoiesis acting both as intrinsic regulators of B-lymphopoiesis and as regulators of genes in bone marrow stroma cells.


The development of blood cells from hematopoietic stem cells in the bone marrow is dependent on communication with bone marrow stroma cells, making these cells central for the appropriate regulation of hematopoiesis. To identify transcription factors that may play a role in gene regulation in stroma cells, we performed comparative gene expression analysis of fibroblastic NIH3T3 cells, unable to support hematopoiesis in vitro, and OP-9 stroma cells, highly efficient in this regard. These experiments revealed that transcription factors of the early B cell factor (EBF) family were highly expressed in OP-9 cells as compared with the NIH3T3 cells.
To identify potential targets genes for EBF proteins in stroma cells, we overexpressed EBF in fibroblasts and analyzed the pattern of induced genes by microarray analysis. This revealed that EBF was able to up-regulate expression of among others the Cxcl12, Ccl9, and Periostin genes. The identification of relevant promoters revealed that they all contained functional EBF binding sites able to interact with EBF in OP-9 cells. Furthermore, ectopic expression of a dominant negative EBF protein or antisense EBF-1 RNA in OP-9 stroma cells resulted in reduced expression of these target genes. These data suggest that EBF proteins might have dual roles in hematopoiesis acting both as intrinsic regulators of B-lymphopoiesis and as regulators of genes in bone marrow stroma cells.
The development of hematopoietic cells from multipotent progenitors is a process regulated by the coordinated action of transcription factors and extrinsic signals from the bone marrow environment. This can be exemplified by one developmental pathway that has been under intense investigation, the maturation of B-lymphocytes from hematopoietic stem cells in the bone marrow. This pathway is dependent of a set of transcrip-tion factors including Pu.1 (1,2), Pax-5 (3), E12/47 (4,5), Sox-4 (6), and early B-cell factor 1 (EBF-1) 2 (7), all acting as crucial and non-redundant intrinsic factors in early B lymphopoiesis. However, the early stages of this developmental pathway is also critically dependent on the active support of bone marrow stroma cells that provide signals crucial for homing, proliferation, and differentiation of progenitor cells. One such factor is Cxcl12 (8) that appears to be produced by stroma cells in an anatomically restricted region of the bone marrow (9). These bone marrow cells are involved in the support of the earliest stages of B cell development and possibly also participate in the regulation of the hematopoietic stem cell niche (9). Subsequent differentiation of the progenitor cell is suggested to be associated with relocation into another anatomical niche composed of stroma cells producing interleukin-7, a cytokine shown to be crucial for normal hematopoiesis (10 -12). interleukin-7 appears to act in concert with the FL-ligand, because mice deficient in both interleukin-7 and FL signaling display a more severe phenotype than either of the single deficient mice (13). Thus, the importance of the stroma cells for normal B cell development is undisputable. However, information regarding transcription factor function and genetic networks in the stroma cells are limited. Stroma cells from fetal livers in c-mybdeficient mice has been shown to have a reduced ability to support blood cell development (14) and recently a role for the homeo-box transcription factor pitx-2 was suggested from in vitro differentiation experiments using stroma from pitx-2-deficient embryos (15).
To understand more about the biology of hematopoiesis supporting stroma cells, we decided to compare gene expression patterns in fibroblasts (NIH3T3) unable to support B-lymphocyte development without the addition of cytokines in vitro, and stroma cells (OP-9) highly efficient in this regard (16). One group of transcription factors that turned out as highly expressed in the stroma cells was EBF proteins. Identification of target genes for EBF in the stroma indicates that these factors among other things regulate a set of cytokine/chemokine genes suggesting dual roles for EBF proteins in hematopoiesis.

MATERIALS AND METHODS
Tissue Culture Conditions and Cell Lines-BaF/3 cells were grown at 37°C and 5% CO 2 in RPMI supplemented with 10% fetal calf serum, 10 mM HEPES, 2 mM pyruvate, 50 M 2-mercaptoethanol, and 50 g/ml gentamicin (all purchased from Invitrogen AB, Täby, Sweden). The medium for the BaF/3 cells was supplemented with 10% of conditioned medium from confluent WEHI3 cells as a source of interleukin-3. NIH3T3 and S17N stroma cells were propagated in Dulbecco's modified Eagle's medium containing 10% (v/v) fetal calf serum and 50 g/ml gentamicin (all purchased from Invitrogen) at 37°C in a humidified atmosphere of 10% CO 2 . OP-9 cells were grown in Opti-MEM supplemented with 10% (v/v) fetal calf serum and 50 g/ml gentamicin (Invitrogen) at 37°C and 5% CO 2 .
Protein Extracts and Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extracts were prepared according to Schreiber et al. (17). DNA probes were labeled with [␥-32 P]ATP by incubation with T4 polynucleotide kinase (Roche), annealed with the complementary strand, and purified on a micro-spin column (Roche). 5-10 g of nuclear extract or 0.5-2 l of in vitro-transcribed/translated protein was incubated with the labeled probe (20,000 cpm, 3 fmol) for 30 min at room temperature in binding buffer (10 mM HEPES pH 7.9, 70 mM KCl, 1 mM dithiothreitol, 1 mM EDTA, 2.5 mM MgCl 2 , 5% glycerol, 1 mM ZnCl 2 ) with 0.75 g of poly(dI-dC) (Amersham Biosciences). DNA competitors were added 10 min before addition of the labeled probe. The samples were separated on a 6% acrylamide TBE gel, which was dried and subjected to autoradiography. Competitors based on synthetic oligonucleotides were added at the molar excesses indicated in the respective figures. Supershifts were performed under the same conditions but with the additional presence of 2 l of polyclonal anti-EBF (SC-15333) or anti-actin (SC-1616) (Santa Cruz) as a negative control.
In Vitro Transcription and Translation-Recombinant protein was generated by coupled in vitro transcription/translation using a reticulocyte lysate kit (Promega) and a cDNA3-based EBF-1 vector (18).
Retrovirus Production and Infection-Phoenix retroviral packaging cells were transfected with pBabepuro retrovirus vectors (p-Babe-puro, p-Babe-puro-EBF, or p-babe-puro-EBF engrailed (19), anti-EBF-1 or control short hairpin RNA virus (20)) in 100-mm dishes at 70% confluence and virus supernatants were harvested after 48 h. Target cells (NIH3T3 or OP-9) were infected in 100-or 60-mm dishes at 50% confluence, using 1 volume of virus supernatant diluted in 1 volume of Opti-MEM (Invitrogen) with 10% (v/v) calf serum, after which Polybrene (Sigma) was added at a final concentration of 6 g/ml. The medium was replenished 8 -12 h following infection. 24 h after infection the cells were split 1:3 to 100-mm dishes and selected with 2 g/ml of puromycin (Sigma) for 24 -48 h. Dishes with selected cells were trypsinized, pooled, and replated on 100-or 60-mm dishes for differentiation. BaF/3 cells were transduced as described in Ref. 21. The cells were selected with 2 g/ml of puromycin (Sigma) for 48 h where after dead, cells were removed by centrifugation on a Ficoll gradient.
Gene Expression Analysis-RNA was prepared using TRIzol (Invitrogen) and 7.5 g of total RNA was annealed to a T7-oligo(T) primer by denaturation at 70°C for 10 min followed by a 10-min incubation of the samples on ice. First strand synthesis was performed for 2 h at 42°C using 20 units of Superscript Reverse Transcriptase (Invitrogen) in buffers and nucleotide mixtures according to the manufacturers instructions. This was followed by a second strand synthesis for 2 h at 16°C, using RNase H, Escherichia coli DNA polymerase I, and E. coli DNA ligase (all from Invitrogen), according to the manufacturers instructions. The obtained double-stranded cDNA was then blunted by the addition of 20 units of T4 DNA polymerase and incubation for 5 min at 16°C. The material was then purified by phenol/chloroform/isoamyl alcohol extraction followed by precipitation with NH 4 Ac and ethanol. The cDNA was then used in an in vitro transcription reaction for 6 h at 37°C using a T7 IV Translation kit and biotin-labeled ribonucleotides. The obtained cRNA was purified from unincorporated nucleotides on an RNeasy column (Qiagen). The eluted cRNA was then fragmented by incubation of the products for 2 h in fragmentation buffer (40 mM Tris acetate, pH 8.1, 100 mM KOAc, 150 mM MgOAc). 20 g of the final fragmented cRNA was then hybridized to Affymetrix chips U74Av2 or MOE4302 (Affymetrix) in 200 l of hybridization buffer (100 mM MES buffer, pH 6.6, 1 M NaCl, 20 mM EDTA, 0.01% Tween 20) supplemented with herring sperm DNA (100 g/ml) and acetylated bovine serum albumin (500 g/ml) in an Affymetrix Gene ChIP Hybridization oven 320. The chip was then developed by the addition of fluorescein isothiocyanate-streptavidin followed by washing using an Affymetrix Gene Chip Fluidics Station 400. Scanning was performed using a Hewlett Packard Gene Array Scanner. Hierarchical tree clusters were generated using the dCHIP program (24). Expression values were calculated according to the PM-only model.
Real Time Quantitative-PCR and Reverse Transcriptase-PCR Analysis-RNA was isolated from cells using TRIzol (Invitrogen) and cDNA was generated by annealing 1 g of total RNA to 0.5 g of random hexamers in 10 l of diethyl pyrocarbonate-treated water. Reverse transcriptase reactions were performed with 200 units of SuperScript Reverse Transcriptase (Invitrogen) in the manufacturers buffer supplemented with 0.5 mM dNTP, 10 mM dithiothreitol, and 20 units of RNase inhibitor (Roche) in a total volume of 20 l. The real time PCR was based on the TaqMan TM technology (Applied Biosystems). The threshold cycles (Ct) for the endogenous control hypoxanthine-guanine phosphoribosyltransferase mRNA and the target signals were determined and the relative RNA quantification was calculated using the comparative Ct method as 2 Ϫ⌬Ct , where ⌬Ct is Ct(target) Ϫ Ct(hypoxanthine-guanine phosphoribosyltransferase). Oligonucleotides for quantitative TaqMan real time PCR were ordered as Assay on demand (Applied Biosystems): Ccl9 Mm00441260_m1, Cxcl12 Mm00445552_m1; Postn Mm00450111_m1; and EBF-1 Mm01288947_g1.
5Ј-Rapid Amplification of cDNA Ends (RACE) Analysis-Total RNA was prepared by TRIzol extraction of OP-9 stroma cells according to the manufactures instructions. 5 g of total RNA was then used for cDNA synthesis and RACE reaction using the RACE analysis kit (Invitrogen), according to the manufacturers protocol. Briefly, oligonucleotide primers located about 200 bp into the published cDNAs of the Perisotin, Ccl9, and Cxcl12 genes were annealed to the RNA and first strand synthesis was performed with Expand Moloney murine leukemia virus reverse transcriptase (Roche). The products were purified and the cDNAs were modified by terminal dioxynucleotide transferase-mediated addition of a 5Ј oligo(G) tail. The RACE products were then amplified by PCR using a second nested gene-specific primer and an oligo(C) primer aimed toward the generated 5Ј poly(G) tail. 1/200 dilutions of the primary PCR products were subjected to a second round of amplification using a third nested primer. The obtained products were purified and sequenced. The RACE primers were: Cxcl12:1, 5Ј-AAGCTTTCTCCAGGTACTCTTGG; Cxcl12:2, 5Ј-CAGCCGTGCAACAATCTGAAGG; Cxcl12:3, 5Ј-GGCA-CAGTTTGAGTGTTGAGG; Ccl9:1, 5Ј-GAATCCGTGAGT-TATAGGACAGG; Ccl9:2, 5Ј-GCTGTGCCTTCAGACTGC-TCTGG; Ccl9:3, 5Ј-TGGACTTCTTTTGTCTCTGTTGC; Periostin:1, 5Ј-CGTTCGAGACATCGGAGTAGTGC; Periostin:2, 5Ј-AACATGGTCAATAGGCATCACTGC; Periostin:3, 5Ј-CATCCCTTCCATTCTCATATAGC.
Transient Transfections and Luciferase Assays-HeLa cells were seeded into a 24-well tissue culture plate so that they reached 60 -80% confluence for transfection. A total of 400 ng of DNA was used to transfect the cells each well. 50 ng of each reporter construct was co-transfected with 300 ng of expression vector and all transfections included 50 ng of pRL-0 Renilla luciferase, received from Dr. Björn Olde, as internal control, and used for normalization of the luciferase activities. The DNA was first added to 50 l of serum-free Opti-MEM medium (Invitrogen), mixed with 7 l of PLUS Reagent (Invitrogen), and incubated for 15 min at room temperature. 0.75 l of Lipofectamine reagent (Invitrogen) was diluted in 25 l of serum-free medium, incubated for 30 min in room temperature, mixed with the diluted DNA, and then incubated for another 15 min at room temperature. Meanwhile, the complete RPMI medium was replaced with 300 l of serum-free medium, and the DNA-Plus/Lipofectamine mixture was added to the cells that were then incubated at 37°C in 5% CO 2 for 3 h, after which each well was supplemented with 1 ml of complete RPMI. Cells were harvested 48 h after transfection and protein extracts were prepared directly in the 24-well plates by adding 100 l of cell lysis buffer.
Chromatin Immunoprecipitations (ChIP)-ChIP experiments were conducted using the ChIP kit (Upstate Biotech) according to the protocol supplied by the manufacturer. Briefly, formaldehyde was added to a concentration of 1% to cultures at 80% confluence of the OP-9 cells. The cells were incubated at 37°C for 10 min before the formaldehyde was removed and the cells were washed 2 times with ice-cold PBS supplemented with protease inhibitors (Complete TM , Roche). After lysis of the cells and sonication of the chromatin, the lysate was pre-cleared and 2 l of either anti-EBF or preimmune rabbit antisera was added (23). After incubation overnight at 4°C, protein A-Sepharose was added and the precipitated material was washed according to the Upstate protocol. Protein A-Sepharose-bound material was eluted by incubation in 500 l of 0.1 M Na 2 CO 3 buffer supplemented with 0.1% SDS. The protein/DNA cross-links were removed by addition of 20 l of 5 M NaCl and incubation at 65°C for 3 h. The DNA was then purified by incubation with 10 g of proteinase K followed by phenol/chloroform extraction and precipitation with 0.1 volumes of 3 M NaAc, pH 5.6, and 2.5 volumes of 95% ethanol. After centrifugation and wash with 70% ethanol the material was dried and redissolved in 25 l of water. 1 l of precipitated material was added to a 25-l PCR. The PCRs were performed using the conditions described above with a PCR cycle 94°C for 45 s, 60°C for 45 s, and 72°C for 45 s, using the following specific primers: Cxcl12 180 sense, CAGCGGAGCCGCGGACACTGAGG; Cxcl12 60 antisense, GcGCTTTAGAGGCGAAAACCAGG; Ccl9 80 sense, GGGT-GTTATGTAGTCAAAGGAGG; Ccl9 10 antisense, CTTGCT-ATTTATACATGTAAGAGG; Periostin 250 sense, AtAGGA-GACAGAGTTCAGATTGC; Periostin 200 antisense, CTTG-CGATGGCCTTACAGAAGC; CD53 sense, TACCATGCTG-GCTTGGTCAGTATGG; CD53 antisense, AGTGCTTC-TCAATTGAAGCAGC.
Isolation and Purification of Bone Marrow Cells-Bone marrow was flushed and passed through a cell strainer. 15-20 ϫ 10 6 cells were incubated in a 75-ml cell culture flask using ␣-minimal essential medium (Invitrogen) supplemented with 15% fetal calf serum, 50 M 2-mercaptoethanol (Invitrogen), and 50 g/ml gentamicin (Invitrogen) for 24 h at 37°C in a humidified atmosphere of 10% CO 2 . Non-adherent cells were removed and washed twice with PBS (Invitrogen), adherent cells were washed with PBS, trypsinized, and washed twice with PBS, RNA was isolated using TRIzol (Invitrogen) as described above.
ELISA-ELISA was made using the DuoSet ELISA kits against CCL9 and CXCL12 (R&D Systems UK). ELISA plates were treated with 100 l of capture antibody diluted in PBS and incubated in room temperature overnight. Unspecific binding were avoided by blocking the plate with 200 l of 0.5% bovine serum albumin in PBS in 37°C for 1 h. Plates were then incubated with 100 l of supernatant from virally infected OP-9 or NIH 3T3 cells for 2 h in 37°C and standards from the kit were used as positive controls. The plates were then incubated with 100 l of detection antibody diluted in 0.5% bovine serum albumin in PBS for 2 h at 37°C. Streptavidin/horseradish peroxidase was added and incubated for 20 min at room temperature. Between all incubation steps washing were performed with 400 l of washing buffer (0.9% NaCl, 0.05% Tween 20) and repeated four times. For antibody detection 100 l of Dako TMB Blue Substrate-Chromogen was used. The TMB reaction was stopped by the addition of 100 l of 2 M H 2 SO 4 . Substrate conversion was quantified in an automatic ELISA reader (VERSA max) at 450 nm.

RESULTS
Bone Marrow Stroma Cells Express EBF Proteins-To identify transcription regulatory networks in stroma cells it is crucial to identify cells capable of supporting blood cell development. Because the exact nature of such cells in the bone marrow remains elusive, we decided to compare gene expression patterns in two cell types with different abilities to support lymphopoiesis in vitro. Two candidates for such a comparison is the stroma cell line is OP-9, able to support the development of B-lymphocytes, and the fibroblast line NIH3T3, being much less efficient in this regard (supplementary data) (16). To get information about the overall gene expression pattern in these cells, we performed global gene expression profiling using MOE4302 Affymetrix microarray chips. The data were analyzed using the dCHIP software (24) revealing clear differences in gene expression patterns ( Fig. 1A  and supplementary data Fig. S1). Among the transcription factors were Fox G1 as well as Fox F2 expressed at a higher level in the NIH3T3 cells, whereas the expression levels of Gata-6 and Id genes (Id1-4), were higher in the Op-9 cells. Another clear difference between two cell types was the expression of EBF genes, where three different probe sets indicated 11-15-fold higher levels of EBF-1 message in the OP-9 cells as compared with the NIH3T3 cells (Fig. 1A, supplementary data Fig. S1). To investigate the relative levels of EBF-1 message in the stroma cell line OP-9 and the pre-B cell lines, 230 -238 and 18 -81, we A, display of dChIP analysis of cDNA microarray data generated by hybridization of two independent cRNAs generated from either NIH3T3 or OP-9 cells to MOE4302 Affymetrix chips. Red indicates high and blue indicates low relative expression. The filter criteria used were: minimal expression value 500 in 50% of the arrays, P call % in the array used Ն50% and 10-fold difference in expression with a 90% confidence interval. Probe sets for transcription factors defined by these criteria are indicated. The complete register of differentially expressed genes is presented in supplementary Fig. S1. Quantitative-PCR data of EBF-1 expression in the hematopoietic progenitor cells, BaF/3, the pre-B cell lines 230 -231 and 18 -81, the fibroblast cell line NIH3T3, and the stroma cell line OP-9 are shown in B. The data were normalized to the expression of hypoxanthine-guanine phosphoribosyltransferase and the results represent the mean values from two representative quantitative-PCR experiments using duplicate PCR. C, displays ethidium bromide-stained agarose gels with the resulting PCR products after amplification of either the control gene ␤-actin (25 cycles) or EBF-1, -2, and -3 (35 cycles) using cDNA from the stroma cell line S17N, the stroma cell line OP-9, and the pre-B cell line 70Z3 as indicated. The autoradiograms in D are generated from an electrophoretic mobility shift assay where the mb-1 promoter EBF binding site (36) was incubated with nuclear extracts from OP-9 cells in the absence or presence of anti-actin (SC-1616) or anti-EBF (SC-15333) specific antibody as indicated. Primary bone marrow cells were screened for expression of EBF proteins as shown in E where ethidium bromide-stained agarose gels with the resulting PCR products from one representative analysis of EBF-1 and -2 expression is displayed. ␤-Actin was amplified by 25 cycles, whereas 35 cycles were used to amplify EBF-1 and EBF-2 message.
performed a quantitative-PCR of EBF-1 expression. This confirmed the microarray results where the OP-9 cells expressed high levels of EBF-1 message as compared with the NIH3T3 as well as the pre-B cell lines (Fig. 1B). EBF-1 belongs to a group of transcription factors designated EBF 1-4 (25,26) and the microarray analysis of 3T3 and OP-9 cells suggested differential expression of EBF-1 and -3 (supplementary data Fig. S1). To further investigate the spectra of EBF genes expressed in stroma cell lines, we performed reverse transcriptase-PCR with mRNA from the stroma cell lines OP-9 and S17N (previously shown to express EBF mRNA (27)), or the pre-B cell line 70Z3 (Fig. 1C). This revealed that whereas the pre-B cells only expressed EBF-1, EBF-1 as well as the EBF-3 message were detected in OP-9 cells. The S17N cells also expressed high levels of EBF-2, whereas none of the tested cell types were found to express EBF-4 (data not shown). The expression of EBF proteins in OP-9 cells could also be detected in EMSAs using nuclear extracts from OP-9 cells and the mb-1 promoter EBF site (Fig. 1D). This resulted in one prominent DNA-protein complex that was supershifted by the anti-EBF, but not the control anti-actin antibody. To investigate if primary bone marrow stroma cells express EBF genes, bone marrow cells were incubated on a plastic surface and EBF expression in adherent and non-adherent cells were investigated by PCR (Fig. 1E). The non-adherent cells, including B-cell progenitors, expressed EBF-1 exclusively, whereas the adherent population expressed EBF-1 as well as EBF-2 message (Fig.  1E). Thus, we conclude that bone marrow stroma cells lines as well as primary bone marrow stroma cells express EBF mRNA.

Ectopic Expression of EBF-1 Induces Cytokine/Chemokine Gene Expression in Fibroblastic Cells-
Knowing that EBF proteins were expressed in stroma cells, we wanted to learn more about the role of these factors in stroma cells by the identification of target genes. Developing a useful loss of function model is complicated by the fact that EBF-1 as well as EBF-2-deficient mice display dramatic phenotypes outside of the hematopoietic system (28 -31) resulting in prenatal or early postnatal death 3 or infertility (31) and that they do not contain regions of homology to which a common small interfering RNA can be designed. 4 Therefore, we developed a gain of function model by transduction of NIH3T3 cells with an EBF-1 encoding retrovirus and selection of the cells in puromycin for 48 h. Changes in gene expression patterns were then analyzed by Affymetrixbased microarrays. The obtained data were analyzed using dCHIP (24) revealing that the increased expression of EBF, as verified by EMSA analysis using the mb-1 promoter EBF site ( Fig. 2A), resulted in higher expression levels of a set of cytokine, chemokine, and growth factor genes ( Fig. 2B and supplementary data Fig. S2). We noted up-regulation of mRNA for among others insulin-like growth factor II, stroma cell-derived 3 M. Sigvardsson, unpublished observation. 4 P. Åkerblad, personal communication.

FIGURE 2. EBF activates the expression of cytokine/chemokine genes in a tissue-restricted manner.
A, displays autoradiograms from electrophoretic mobility shift assays where an mb-1 promoter EBF binding site was incubated with 2.5 g of nuclear extracts from vector or EBF transduced NIH3T3 cells as indicated. B, displays a dChIP analysis of cDNA microarray data generated by hybridization of cRNA generated from NIH3T3 cells transduced with either empty pBabe (Vector) or EBF (EBF) encoding retroviruses to a U74AV2 Affymetrix chip. The data are generated from two transduction experiments and two hybridizations 48 h after initiation of selection with 2 g/ml puromycin. Red indicates high and blue indicated the low relative expression. The filter criteria used were: variation across samples: 0.50 Ͻ S.D./mean Ͻ 10.00, P call % in the array used Ն10%. A selection of induced genes is indicated, whereas the complete data set I are presented in supplementary Fig. S2. C, displays diagrams with real time PCR data from vector (V) or EBF-1 (E-1) transduced BaF/3 and NIH3T3 cells 24 and 48 h after transduction, as well as the Abelson transformed pre-B cell line 230 -238 and the stroma cell line OP-9 as indicated. The data were generated using TaqMan technology and assay on demand primer and probe sets. The transduced BaF/3 cells were analyzed 48 h after completed infection and initiation of selection, whereas NIH3T3 cells were analyzed 24 and 48 h after initiation of puromycin selection. All data were normalized to the expression levels of hypoxanthine-guanine phosphoribosyltransferase. The expression of Periostin, Cxcl12, and Ccl9 was set to one in NIH3T3 at 24 h (control). The data presented represents 4 data points collected from two independent transduction experiments. The diagram to the far right displays ELISA data monitoring Ccl9 expression in supernatants from vector-or EBF-transduced cells as indicated. The data are collected from three measurements from one representative of two experiments. The error bars indicate S.D.
factor I (Cxcl12), OSF-2 (Periostin), and Ccl9. To verify part of these data, we performed quantitative-PCR analysis to investigate the expression of a set of genes using mRNA from vector or EBF-1-transduced NIH3T3 cells (Fig. 2C). This revealed that the level of Periostin message was up-regulated 10 times 48 h after transduction of the NIH3T3 cells, resulting in expression levels comparable with that in the stroma cell line OP-9. Ccl9 and Cxcl12 expression was up-regulated 26-and 28-fold, respectively, in the EBF-transduced NIH3T3 cells but the expression levels were low as compared with those found in OP-9 stroma cells. None of these genes were induced by ectopic expression of EBF in the hematopoietic cell line BaF/3 (Fig. 2C) even though other EBF targets was induced (21,32), 5 indicating that EBF induce target genes in a tissue-restricted manner. To investigate if the increase in mRNA was reflected in increased secretion of cytokines we analyzed supernatants from vector or EBF-transduced cells for the presence of Cxcl12 and Ccl9 by ELISA. Whereas the level of Cxcl12 was to low to give reproducible data from either the vector or EBF-transduced cells, a marked increase in Ccl9 levels could be detected in supernatant from the EBF-transduced cells (Fig. 2C). These data suggest that EBF is able to activate a set of genes expressed to high levels in stroma cells.
The Cxcl12, Periostin, and Ccl9 Promoters Are Targets for EBF Proteins-To investigate if the induced cytokine genes were direct targets for EBF, we wanted to identify relevant promoter elements to search for EBF binding sites. To this end we performed 5Ј-RACE analysis using total RNA from OP-9 cells. This resulted in the generation of PCR products (Fig. 3A) that were purified on an agarose gel and sequenced. The obtained cDNA sequence was then aligned to genomic DNA allowing us to get relevant information about the transcription start sites used in OP-9 cells. To investigate if the presumed promoters were functional in stroma cells we cloned ϳ500 base pairs (see "materials and Methods") of genomic DNA covering the presumed transcription start sites of each of the genes in a luciferase reporter vector and transfected OP-9 cells with the resulting plasmids (Fig. 3B). The Cxcl-12 promoter element gave a functional activity 45-fold over the pGl-3 vector carrying an immunoglobulin light chain promoter TATA-box only, whereas the Ccl-9 and Periostin promoters gave 9-and 13-fold this level of transcription. These data support the idea that the identified DNA elements indeed contains sequences that allow for initiation of transcription and for a degree of induced activity, confirming that we have identified functional promoter elements by RACE. Visual inspection of the promoters suggested the existence of TATA boxes 40 or 24 base pairs 5Ј of the presumed transcription start sites in the Periostin and Ccl9 genes, respectively. The Cxcl12 promoter has a cryptic TATA box (TAAA) 28 base pairs 5Ј of the defined cDNA end but the initiation of transcription at an A nucleotide in a putative initiator element indicates that the promoter lacks a functional TATA box and instead relies on an initiator element for the initiation of transcription. Sequence analysis using Match (www.gene-regulation.com) suggested that all three promoters contained potential binding sites for GATA proteins and LMO-2 but no other obvious similarities could be detected. Visual inspection of the promoter sequences allowed us to identify potential EBF binding sites (variants of ATTCCC-NNGGGAAT (33)) in all three promoters (Fig. 3C). To investigate if these sites were able to interact with EBF in vitro, we performed EMSA experiments where the binding of recombinant EBF to the mouse mb-1 promoter EBF binding site was FIGURE 3. Identification of functionally active Cxcl12, Ccl9, and Periostin promoters suggests shared ability to interact with EBF. A, displays an ethidium bromide-stained agarose gel of the PCR products obtained by 5Ј RACE analysis of the Cxcl12, Ccl9, and Periostin cDNAs using total RNA from OP-9 cells. B, displays a diagram over the resulting luciferase activity obtained when approximately 500 base pairs of genomic DNA from the putative Cxcl12 (nucleotides Ϫ450 to ϩ50), Ccl9 (nucleotides Ϫ400 to ϩ10), and Periostin promoters (Ϫ450 to ϩ15) were cloned in the luciferase reporter vector pGl-3, when the plasmids were transfected into OP-9 cells. pGl-3 vector carrying only a immunoglobulin-light chain promoter TATA box were used as control for basal promoter activity and a SV40 virus promoter controlled plasmid were used as an example of a strong promoter element. The data shown are collected from three transfections in one representative of two independent transfection experiments. The error bars indicate S.D. C, shows schematic drawings of the promoter elements and the nucleotide composition, as well as the position of the EBF binding sites relative to the transcription start sites defined by RACE. The EBF binding core is underlined.
competed for by the addition of unlabeled duplex oligonucleotides covering the presumed sites (Fig. 4A). This suggested that the sites selected in the 5Ј regions of the Periostin, Cxcl12, and Ccl9 genes all competed for EBF binding thus represents functional EBF binding sites. To investigate if these sites interact with EBF proteins from stroma cells, we performed EMSA using nuclear extracts from the OP-9 cells and labeled binding sites from the Periostin, Cxcl12, and Ccl9 genes (Fig. 4B). All the putative binding sites interacted with EBF protein as supported by supershift analysis with EBF reactive antibodies. To investigate if EBF proteins are interacting with the promoter elements in vivo, we performed chromatin immunoprecipitation experiments using a polyclonal anti-EBF antisera (23) and fragmented chromatin from OP-9 stroma cells. PCR analysis of the precipitated fractions suggested an enrichment of the Ccl9, Cxcl12, and Periostin promoter elements when the EBF antisera was used as compared with when we used a preimmunization rabbit antisera (Fig. 4C). To investigate the specificity of the precipitation, we also analyzed the enrichment of the promoter controlling the CD53 gene (Fig. 4C), not expressed in the OP-9 cells (data not shown). This element could not be detected in any of the precipitated fractions, whereas the input DNA gave rise to an amplification product, arguing for a specificity of the ChIP experiment. These data support the idea that EBF is involved in the regulation of genes in stroma cells and that the presumed target genes are direct targets for EBF. To further investigate the ability of EBF to activate the Cxcl12, Ccl9, and Periostin promoters we transfected the luciferase reporter constructs into epithelial human HeLa cells, lacking expression of endogenous EBF (18), together with either control (6x9E10-myc tag, MD) of EBF-1 (MDEBF) containing cDNA3 expression plasmids (18) (Fig. 5A). The control vector was constructed by cloning the three mb-1 promoter EBF binding sites in front of a TATA box (3xmb-1) (22). This reporter was induced 23-fold by the inclusion of 100 ng of EBF expression plasmid and 61 times by the inclusion of 300 ng of EBF expression plasmid. The Cxcl12 and Periostin promoters generated a rather high basal functional activity but the inclusion of 300 ng of EBF encoding expression plasmid induced these promoters 7-and 16-fold, respectively, whereas the Ccl9 promoter responded by 20-and 77-fold inductions of promoter activity. We could not detect any induction of the Prl0-Renilla control plasmid or a cytomegalovirus promoter-driven reporter plasmid was induced less then 2-fold by the inclusion of 300 ng of EBF (data not shown) arguing against a nonspecific activation induced by EBF. To test if EBF-2 was able to activate these promoters, we transfected the reporter constructs together with 300 ng of EBF-2 encoding expression plasmid (Fig. 5B). This resulted in a 15-fold activation of the 3xmb-1 promoter, a 74-fold activation of the Ccl9 promoter, as well as 4-and 8-fold activations of the Cxcl12 and Periostin promoters, respectively. This supports the idea that the Cxcl12, Ccl9, and Periostin genes are direct genetic targets for EBF proteins and that EBF-1 and EBF-2 are able to activate the promoters.

Reduction of EBF Function Modulates the Transcription of
Target Genes in OP-9 Cells-To further investigate a potential role of EBF proteins in the regulation of genes in stroma cells we transduced OP-9 cells with a dominant negative form of EBF where the transactivation domain of EBF-1 is replaced with the repressor domain from engrailed (19). This protein has previously been shown able to impair fat cell differentiation without any obvious negative effects on cell proliferation or survival (19). Analysis of the levels of the Periostin, Cxcl12, and Ccl9 message in transduced and selected OP-9 cells by quantitative-PCR suggested that the level of mRNA from all the presumed EBF target genes were reduced even though the effect on Ccl9 transcription was modest (Fig. 6). To investigate if this reduction of mRNA was reflected in protein secretion we analyzed the supernatants for the levels of Cxcl12 and Ccl9 proteins by ELISA. This suggested a marked reduction in the secretion of Cxcl12, whereas the effect on the levels of Ccl9 was modest (Fig.  6A, white bars). To verify these findings using another loss of function model we transduced OP-9 cells with an EBF-1 targeting short hairpin RNA expressing retrovirus (20). Cells containing this virus displayed a reduced level of EBF-1 message as compared with control virus-transduced cells. We could also detect reductions in Cxcl12, Ccl9, and Periostin mRNA levels (Fig. 6B). These data support the idea that EBF has a role in the regulation of genes in stroma cells even though there may be partial redundancy in the regulation of some of the target genes.

DISCUSSION
We here report that EBF proteins are expressed in stroma cell lines as well as in stroma cells from mouse bone marrow and that the transcription factor directly targets and activates the transcription of the Ccl9, Cxcl12, and Periostin genes. EBF-1 is essential for B cell development (7) and is expressed mainly at the earliest stages of differentiation (35,36) where it acts in concert with the gene products from the E2A gene (E12 and E47) (18,32,37). Among the identified target genes are the surrogate light chain genes 5 and VpreB (18,21,32) as well as the signal transduction molecules Ig␣ (35,36,38), Ig␤ (39), and Blk (40) genes. Whereas these target genes display tissue-and stage-specific expression patterns, EBF itself is expressed outside of the B-lymphoid compartment (41,42). For instance, EBF-1 as well as its homologues EBF-2, -3, and -4, (O/E1-4) are expressed in, and play important roles in the central nervous system (26,43,45). In olfactory neurons EBF proteins have been suggested to control the expression of the OcNC, ACIII, and Golf genes and the phenotype of mice deficient of EBF family proteins suggest that these factors play important roles in the functional development of both the central and the peripheral nervous system (25,28,29,31,46). EBF has also been shown to support adipogenesis in vitro (19,48). These collected findings provide a complex picture of the ability of EBF proteins to act and induce transcription of target genes where it appears as if these transcription factors activates lineage-specific target genes rather than general inducers of differentiation or cell growth. One possibility can be selective activation of target genes by the different EBF proteins, however, the high homology between these proteins (25) and their shared ability to activate target promoters (25) (Fig. 5) would suggest a functional redundancy among EBF-1, -2, and -3. Thus, the differential phenotypes observed in the mice carrying mutations in the different EBF genes (7,28,29,31) could be a result of restricted expression patterns rather than distinct biochemical features. However, this does not exclude that there are differences in the ability to interact with lineage-restricted co-factors. One striking finding in the OP-9 cells is the high expression of Id-proteins ( Fig. 1, supplementary data Fig. S1), known to act as inhibitors of E-proteins that in turn are proposed to act in synergy with EBF in the activation of B-lineage target genes (49). The 3T3 and OP-9 cells differed largely also with regard to the expression of other transcription factors (Fig. 1A, supplementary data Fig. S1) and another co-factor could aid in the selective target gene activation. In addition, the finding that the EBF mainly up-regulate the expression genes already expressed to a low level (Fig. 2) could indicate an important role also for epigenetic settings in the regulation of EBF function as previously suggested from studies of EBF function in neuroblastoma cells (50).
The OP-9 cells we used as a model for a bone marrow stroma cell are able to support the development of several types of blood cells (16). However, without the addition of cytokines these cells mainly support B lymphopoiesis (16). Among the EBF targets investigated, we find the Cxcl12 gene to be of special interest due to its role in the homing of hematopoietic stem cells and progenitor B cells (9, 51). The protein has been shown FIGURE 6. A dominant negative form of EBF-1 reduces the expression of Cxcl12 and Periostin mRNA in OP9 stroma cells. The diagram in A displays real-time PCR data from OP-9 cells transduced with an empty vector (Vector) or a vector containing a dominant negative form of EBF (Engrailed). The expression of Cxcl12, Ccl9, and Periostin in RNA was set to one in Vector (control). The data were normalized to the expression levels of hypoxanthineguanine phosphoribosyltransferase and represents 4 data points collected from two transduction experiments. The white bars in A represents ELISA data monitoring the levels of Cxcl12 and Ccl9 in the supernatants taken from OP-9 cells transduced either with control vector or EBF-engrailed virus. The data are collected from 6 measurements and two independent experiments. B, displays quantitative-PCR data from OP-9 cells transduced with either a control or EBF-1 short hairpin RNA expressing virus (20). The data are collected from at least four PCR and two experiments. In all panels, error bars indicate S.D.
to be expressed by a specific niche in the bone marrow attracting the earliest progenitor B-cells (9). The Periostin gene, induced selectively in the NIH3T3 cells encodes a fashilin-like protein involved in cell adhesion and spreading in vitro (52). The protein is expressed in early osteoblastic cells but has also been associated with growth and metastasis of breast and colon cancer tumors (44,47,52). We have not been able to find published data supporting a direct role for periostin or Ccl-9 in normal B-lymphopoesis but the ability of periostin to stimulate cells via the vitronectin receptor (␣ v ␤ 3 integrin) expressed on human B-lymphoid cells suggests that this could be the case (34).
Our results support the idea that EBF is expressed and participate in gene regulation in a model for a bone marrow stroma cell. Thus, even though we cannot from our data conclude about the relative importance for EBF proteins in stroma cell function, it is clear that EBF have the ability to bind promoters of target genes in stroma cells. This suggests that EBF proteins have dual roles in hematopoiesis both as regulators of B-lineage-specific gene expression and production of growth/differentiation modulating factors in the stroma cell compartment.