A Novel Retinoic Acid-Responsive Element Regulates Retinoic Acid ...

MOLECULAR AND CELLULAR BIOLOGY, Mar. 2004, p. 2423–2443 0270-7306/04/$08.00⫹0 DOI: 10.1128/MCB.24.6.2423–2443.2004 Copyright © 2004, American Society for Microbiology. All Rights Reserved.

Vol. 24, No. 6

A Novel Retinoic Acid-Responsive Element Regulates Retinoic Acid-Induced BLR1 Expression Jianrong Wang and Andrew Yen* Department of Biomedical Sciences, Cornell University, Ithaca, New York 14853 Received 14 July 2003/Returned for modification 27 August 2003/Accepted 10 December 2003

The mechanism of action of retinoic acid (RA) is of broad relevance to cell and developmental biology, nutrition, and cancer chemotherapy. RA is known to induce expression of the Burkitt’s lymphoma receptor 1 (BLR1) gene which propels RA-induced cell cycle arrest and differentiation of HL-60 human myeloblastic leukemia cells, motivating the present analysis of transcriptional regulation of blr1 expression by RA. The RA-treated HL-60 cells used here expressed all RA receptor (RAR) and retinoid X receptor (RXR) subtypes (as detected by Northern analysis) except RXR␥. Treatment with RAR- and RXR-selective ligands showed that RAR␣ synergized with RXR␣ to transcriptionally activate blr1 expression. A 5ⴕ-flanking region capable of supporting RA-induced blr1 activation in HL-60 cells was found to contain a 205-bp sequence in the distal portion that was necessary for transcriptional activation by RA. Within this sequence DNase I footprinting revealed that RA induced binding of a nuclear protein complex to an element containing two GT boxes. Electromobility shift assays (EMSAs) and supershift assays showed that this element bound recombinant RAR␣ and RXR␣. Without RA there was neither complex binding nor transcriptional activation. Both GT boxes were needed for binding the complex, and mutation of either GT box caused the loss of transcriptional activation by RA. The ability of this cis-acting RAR-RXR binding element to activate transcription in response to RA also depended on downstream sequences where an octamer transcription factor 1 (Oct1) site and a nuclear factor of activated T cells (NFATc) site between this element and the transcriptional start, as well as a cyclic AMP response element binding factor (CREB) site between the transcriptional start and first exon of the blr1 gene, were necessary. Each of these sites bound its corresponding transcription factor. A transcription factor-transcription factor binding array analysis of nuclear lysate from RA-treated cells indicated several prominent RAR␣ binding partners; among these, Oct1, NFATc3, and CREB2 were identified by competition EMSA and supershift and chromatin immunoprecipitation assays as components of the complex. RA upregulated expression of these three factors. In sum the results of the present study indicate that RA-induced expression of blr1 expression depends on a novel RA response element. This cis-acting element approximately 1 kb upstream of the transcriptional start consists of two GT boxes that bind RAR and RXR in a nuclear protein complex that also contains Oct1, NFATc3, and CREB2 bound to their cognate downstream consensus binding sites.

cyte-specific chemokine receptor family (15, 16). Murine expression of the receptor, however, has also been found in defined areas of the cerebellum (17), monocytes, peripheral blood leukocytes (4), and neuronal tissues (31). Knockout of the murine blr1 function by gene targeting disrupted the formation of inguinal lymph nodes and severely affected the development of Peyer’s patches. The formation of follicles in the spleen of blr1-negative mice was aberrant, and B cells failed to migrate into the B-cell follicles (15). Two CXC chemokines, B-lymphocyte chemoattractant and B-cell-attracting chemokine 1 (BCA-1), were found to function as physiological BLR1 ligands. B-lymphocyte chemoattractant directed the migration of murine B lymphocytes to follicles in secondary lymphoid organs (24), and BCA-1 selectively attracted murine and human B lymphocytes via BLR1/CXCR5 (38). However, BCA-1 was chemotactic only for human B lymphocytes and not for T lymphocytes, monocytes, or neutrophils. Regulation of blr1 expression has been studied in a limited number of contexts. Alignment of the nucleotide sequences showed that neither the human blr1 promoter nor its murine counterpart has TATA or CCAAT boxes, a feature that is not uncommon for genes encoding serpentine receptors. The human blr1 promoter directs initiation of transcription from a

Certain features of the biology of Burkitt’s lymphoma receptor 1 (BLR1) have been established which suggest a regulatory role in cell proliferation and differentiation and development. The blr1 gene was originally identified in a screening for differentially expressed genes conferring metastatic capability to human B-cell lymphomas. It encodes a putative serpentine heterotrimeric G protein-coupled chemokine receptor (also known as CXCR5) (13, 14, 36). The blr1 cDNA sequence contains an open reading frame of 1,116 bp encoding a 372amino-acid protein. Analysis of the genomic structure of the blr1 gene indicated that the predicted protein is derived from two exons, with the first encoding 17 amino acids about 10 kb upstream of the second exon (13). The sequence is highly related to receptors for interleukin-8 (IL-8) and other neutrophil chemoattractants (13). Other than in the human Burkitt’s lymphoma cells from which it was originally isolated, BLR1 expression in human hematopoietic system cells is restricted to mature resting B cells (13) and a subset of T-helper memory cells (16), thus identifying BLR1 as a member of the lympho* Corresponding author. Mailing address: Department of Biomedical Sciences, T4-008 VRT, Cornell University, Ithaca, NY 14853. Phone: (607) 253-3354. Fax: (607) 253-3317. E-mail: [email protected]. 2423

2424

WANG AND YEN

single start site (54). In murine B lymphocytes, regulatory elements lying between positions ⫺78 and ⫹215 are sufficient to confer basal activity of the murine blr1 promoter. Three essential regulatory elements in the blr1 promoter have been found to be necessary for cell type-specific and differentiation lineage-specific blr1 expression (54). These include binding sites for the transcription factors (TFs) NF-␬B, Oct2, and Bob1, which cooperatively regulate the blr1 gene in B cells. As a B-lymphocyte-restricted octamer factor, Oct2 is the key determinant for blr1 promoter activity. Bob1 acts as a B-cell coactivator of octamer binding factors (54). Both are also expressed in myeloid cells and regulate myeloid lineage development (2, 48). BLR1 mRNA expression can be induced by retinoic acid (RA) (7). RA is a developmental morphogen, a necessary dietary factor for proper development in juveniles, and a cancer chemotherapeutic agent used in differentiation induction therapy. Reflecting these physiological functions, RA regulates cell proliferation, differentiation, and apoptosis (1, 12, 19, 30, 40, 47). RA and its retinoid metabolites act as ligands for RA receptors (RARs) RAR␣, RAR␤, and RAR␥ and retinoid X receptors (RXRs) RXR␣, RXR␤, and RXR␥, which are ligand-activated TFs that are members of the steroid thyroid hormone superfamily of nuclear receptors (32, 37, 42, 44, 62). RAR and RXR regulate transcription by binding to particular cis-responsive elements, RA response elements (RAREs), and regulate their transactivation function (8, 32, 42, 43, 62). In HL-60 human myelomonocytic precursor cells, RA induced BLR1 expression through RAR and RXR activation, which caused activation of the extracellular signal-regulated kinase 2 (ERK2) mitogen-activated protein kinase and propelled cell cycle arrest and myeloid cell differentiation. Ectopic expression of BLR1 in these myeloblastic leukemia cells enhanced ERK2 activation and accelerated RA-induced myeloid differentiation (6, 7). BLR1 also enhanced ERK2 activation and RA-induced differentiation along the monocytic lineage in U937 monoblastic leukemia cells (5). There is thus a role for BLR1 in regulating cell proliferation and differentiation. The transcriptional mechanism by which RA induced BLR1 expression in HL-60 cells to propel their differentiation and G0 arrest is thus of interest, but the cis-acting elements and TFs conferring RA responsiveness are not known. The present study identifies an approximately 1.3-kb 5⬘flanking region capable of supporting RA-induced blr1 expression. A cis-acting element that contains a novel RA-responsive element with two GT boxes is located in the distal portion of this region. The element binds a TF complex in response to RA. The complex contains RAR␣, RXR␣, octamer TF 1 (Oct1), and nuclear factor of activated T cells 3 (NFATc3) throughout blr1 activation (as well as cyclic AMP [cAMP] response element binding factor 2 [CREB2] transiently at an early stage of blr1 activation). The cis-transactivating capability of this element in response to RA depends on downstream Oct1, NFATc, and CREB sites. Activation of the blr1 gene by RA thus seems to require formation of a multimolecular TF complex to a novel RA-responsive element in the blr1 promoter.

MOL. CELL. BIOL. MATERIALS AND METHODS Reagents. Plasmids pGL3-basic, pRL-TK, and pGEM-T Easy and a luciferase reporter assay system as well as a footprinting kit were purchased from Promega. Rabbit polyclonal antibodies for each of RAR␣, RXR␣, Oct1, Oct2, NTATc1, NFATc2, NFATc3, NFATc4, NFATc5, CREB1, and CREB2 and normal antirabbit immunoglobulin G (IgG) were purchased from Santa Cruz Biotechnology Inc. [␣-32P]dCTP, [␣-32P]dATP, [␣-32P]dTTP, and [␥-32P]ATP were obtained from Perkin Elmer Life Sciences. A TITANIUM One-Step reverse transcriptase PCR (RT-PCR) kit was purchased from BD Biosciences. Probes for Northern analysis and oligonucleotides based on blr1 promoter sequences for electromobility shift assays (EMSAs) were ordered from Operon Qiagen. A random primer labeling system and mutagenesis reagents were obtained from Stratagene. An end extension labeling kit and a biotin labeling kit were obtained from Perkin Elmer Life Sciences. All-trans-RA and actinomycin D were purchased from Sigma-Aldrich. An EMSA kit, oligonucleotides for Sp1, early growth response factor 1 (EGR1), DR5, DR2, DR1, NFATc, CREB, NF1, Pbx1, and DNAprotein array and TF-TF array kits were purchased from Panomics (Redwood, Calif.). A rabbit reticulocyte lysate system was purchased from Novagen (Madison, Wis.). A chromatin immunoprecipitation (ChIP) assay kit was purchased from Upstate Cell Signaling Solutions (Lake Placid, N.Y.). Cell culture. HL-60 cells were grown in RPMI 1640 supplemented with 5% fetal calf serum in a 5% CO2 humidified atmosphere at 37°C. The cultures were initiated every 2 or 3 days at a density of 0.2 ⫻ 106 or 0.1 ⫻ 106 cells/ml, respectively. Cell viability (as determined by trypan blue exclusion) routinely exceeded 95%. For experimental cultures, cells were initiated at a density of 0.2 ⫻ 106 cells/ml. All-trans-RA, actinomycin D, and RAR-selective ligands were added at indicated concentrations from 1 or 5 mM stock dissolved in 100% ethanol and stored protected from light at ⫺20°C (RA and actinomycin D) and ⫺80°C (ligands). Plasmid constructions. The human blr1 promoter sequence spanning 1,360 bp upstream of the translational start codon (or 1,096 bp upstream of the initiation site for transcription) was obtained by PCR from a genomic DNA template that was prepared from HL-60 cells. The primers were designed on the basis of the reported sequences (54) with GenBank/EBI database accession numbers X83755 and X83756. The 1,360-bp PCR product was inserted at the KpnI-HindIII sites in pGL3-basic, a promoterless luciferase reporter vector, resulting in a blr1 promoter-controlled luciferase reporter construct, BLR1-Luc. 5⬘ end unidirectional deletion from ⫺1,096 to ⫹78, 3⬘ end deletion from ⫹1 to ⫹266, and deletion between ⫺891 and ⫹1 of the BLR1 promoter were performed by PCR using the BLR1-Luc plasmid as a template to generate 10 BLR1-Luc constructs with different length blr1 promoter fragments. Two pairs of primers for amplification of RAR␣ and RXR␣ cDNAs, each 1,386 bp in length, were synthesized (incorporated restriction sites NotI-SalI and NotI-PstI are denoted by underlining): RAR␣ forward, 5⬘-AATGCGGCCGCCATGGCCAGCAACAGCAGCTCCTG CCCG-3⬘; RAR␣ reverse, 5⬘-AATGTCGACTCACGGGGAGTGGGTGGCC GGGCTGC-3⬘; RXR␣ forward, 5⬘-AATGCGGCCGCCATGGACACCAAAC ATTTCCTGCCGC-3⬘; RXR␣ reverse, 5⬘-AATCTGCAGCTAAGTCATTTGG TGCGGCGCCTCC-3⬘. Total RNA from RA-treated HL-60 cells was prepared and used as a starting template to amplify the cDNAs by RT-PCR using a TITANIUM One-Step RT-PCR kit according to the manufacturer’s protocol. The expected PCR products were digested with NotI-SalI or NotI-PstI and ligated with a NotI-SalI- or NotI-PstI-digested plasmid (pGEM-T Easy) to produce the expression vectors (pGEM-RAR␣ and pGEM-RXR␣). The Kozak sequence was incorporated immediately upstream of the open reading frames of the receptors to achieve high-level in vitro expression. All of the PCR products for construct building were confirmed by sequencing. Transient transfection and luciferase assay. HL-60 cells were harvested by centrifugation and washed with RPMI 1640 medium supplemented with 5% fetal calf serum, and 10 million cells were suspended in 0.4 ml of this medium. Transient cotransfection into cells of the BLR1-Luc luciferase (firefly) reporter plasmid and luciferase (Renilla [sea pansy]) expression vector, pRL-TK (for normalization of transfection efficiency), was performed by DEAE-dextran electroporation. A total of 50 ␮g of BLR1-Luc plasmid (1 ␮g/␮l) and 15 ␮g of pRL-TK (1 ␮g/␮l) with 2 ␮l of DEAE-dextran (1 ␮g/␮l) were added to the 107 cells, and the mixture was incubated for 10 min at room temperature followed by 5 min on ice and electroporated with a Gene Pulser apparatus (Bio-Rad) at 320 V and 960 ␮F. Right after electroporation, the cells were placed on ice for 15 min prior to dilution with 5 ml of prewarmed RPMI 1640 medium containing 5% fetal calf serum without antibiotics. The cells were harvested by centrifugation, rinsed once with room temperature phosphate-buffered saline (PBS), and suspended in passive lysis buffer. The protein concentration of the samples was

VOL. 24, 2004

RA REGULATES BLR1 BY A NOVEL RARE

2425

TABLE 1. Primer pairs used for PCR amplifications Specificity

Primer sequence

RAR␣

5⬘-GGGCATGTCCAAGGAGTCTGTGAGAA 5⬘-GGGATCTCCATCTTCAGCGTGATCA 5⬘-CGTCTGCCTGGTTTCACTGGCTTGA 5⬘-CTGAGCTGGGTGAGATGCTAGGACTGT 5⬘-GGGTCTACAAGCCATGCTTCGTGTGCAA 5⬘-GGCAGGCAGCTTTGAGCAGAGTGATCT 5⬘-CCCCAGTGCTAGCTGCCGAAGCACCC 5⬘-GACTCCGGTACCTACTTGCAGGCTGGC 5⬘-GCTCCTCAGGCAAGCACTATGGAGTGTA 5⬘-GGAGAAGGAGGCGATGAGCAGCTCATT 5⬘-TGGGGCTGGCAAACGGCTATGTGCAAT 5⬘-GCCCACTCAACAAGCGTGAATAGCTGT 5⬘-GCAGCGCCTCCAGGAATCAACTTGGTT 5⬘-GGATGCCATCCTGCACGGAAACTG A 5⬘-GGAGGACCTGTTCTGGGAACTGGACAGA 5⬘-GCCCACCAGCCAGATGGTCCCACAG

RAR␤ RAR␥ RAR␥2 RXR␣ RXR␤ RXR␥ BLR1 a

Fragment (bp)

756 594 549 428 521 468 724 610

Locationa

555–580 1286–1310 1180–1204 1748–1774 668–695 1190–1216 1569–1594 1971–1997 490–517 984–1010 773–799 1214–1240 441–467 1140–1164 489–516 953–978

NCBI accession no.

NM 000964 NM 000965 M 24857 X 57280 NM 002957 NM 021976 NM 006917 NM 032966

Sequence location 1 refers to the first base pair of the published sequence in National Center for Biotechnology Information (NCBI) database.

measured, and each sample was diluted to 3 ␮g/␮l with lysis buffer. Luciferase (30 ␮g) activity was measured with a luminometer (Lumat LB950; Berthold Japan K.K., Tokyo, Japan) using a Promega dual-luciferase reporter assay system. RT-PCR and Northern analysis. Total RNA was extracted from HL-60 cells by using an RNA-Bee kit (Tel-test, Friendswood, Tex.) according to the manufacturer’s protocol. Specific primers for detecting mRNA transcripts of RAR and RXR as well as BLR1 with a TITANIUM One-Step RT-PCR kit are given in Table 1. In each case, reactions with 500 ng of total RNA were run on a PTC-100 thermal cycler (MJ Research, Inc.) with the following program cycle: 50°C for 1 h (1 cycle); 94°C for 5 min (1 cycle); 94°C for 30 s, 65°C for 30 s, and 68°C for 1 min (35 cycles); and 68°C for 2 min (1 cycle). Negative controls with Taq but not RT were included in each set of samples to confirm a lack of genomic DNA contamination in the RNA samples. The product was resolved by agarose electrophoresis, and bands representing the appropriate size product were recovered and used as probes in subsequent Northern analysis of RAR and RXR expression. A probe for the RXR␥ transcript, which yielded no RT-PCR product from HL-60 cells, was amplified from a plasmid, pSG5/RXR␥ (36, 44), by PCR with the following program cycle: 95°C for 2 min (1 cycle); 95°C for 30 s, 74°C for 30 s, and 72°C for 1 min (35 cycles); and 72°C for 10 min (1 cycle). Due to the unavailability of the complete RAR␥2 mRNA sequence, the probe for RAR␥2 extended from its 5⬘ untranslated region into the first exon and was amplified from 500 ng of genomic DNA by PCR using the thermal cycler program described above. For Northern analysis, each 10-␮g RNA sample was size fractionated by electrophoresis on a 1% agarose gel containing the following: 20 mM MOPS (pH 7.4), 1 mM EDTA, 5 mM sodium acetate, 0.2 M formaldehyde, and 0.5 ␮g of ethidium bromide/ml. RNA samples contained 50% formamide. RNA was transferred to a GeneScreen (Stratagene) hybridization transfer membrane and UV cross-linked (with a Stratagene Stratalinker 1800 apparatus) to the membrane. The PCR-amplified fragments were gel purified and labeled with [␣-32P]dCTP by using a Prime-It RmT random primer labeling kit (Stratagene), and the labeled probes were purified with a NucTrap probe purification column (Stratagene) for detection of mRNA transcripts of blr1, RARs, and RXRs by hybridization at 50°C overnight after a 2-h prehybridization at 42°C for blr1 and 45°C for RARs and RXRs. After two washes at room temperature for 30 min in a 2⫻ SSC (1⫻ SSC is 0.15 M NaCl plus 0.015 M sodium citrate)–0.1% sodium dodecyl sulfate (SDS) solution and then two washes at 60°C for 15 min with 0.1⫻ SSC, blots were exposed to phosphor luminescence plates (Fuji) for 48 h; the hybridized bands for specific transcripts were then read and quantitated on a PhosphorImager. Preparation of nuclear extracts and in vitro translation receptors. HL-60 cells were grown in cultures for 12, 24, and 48 h in the presence or absence of 1 ␮M all-trans-RA. The nuclear extracts were prepared using a Sigma-Aldrich nuclear extract kit. Briefly, cells were rinsed with PBS, resuspended in hypotonic lysis buffer, and lysed in a tight-fitting Dounce homogenizer with 45 strokes. The cells were incubated on ice for 10 min and observed by phase-contrast microscopy to verify that 95% lysis had occurred. A nuclear pellet was obtained by centrifugation in a refrigerated microcentrifuge (30 min at 13,000 ⫻ g). Protein concen-

tration of the samples was determined by a Bradford assay. Nuclear extracts were aliquoted and stored at ⫺80°C until use. RAR␣ and RXR␣ proteins were prepared using a rabbit reticulocyte lysate translation system kit (Novagen). pGEM-RAR␣ and pGEM/RXR␣ plasmids were used as templates for RNA synthesis with T7 RNA polymerase, and the resultant mRNAs were used as templates for in vitro translation according to the manufacturer’s instructions. Translation of the appropriate protein was verified by Western blotting after SDS-polyacrylamide gel electrophoresis (SDS-PAGE). DNase I footprinting. DNase I footprinting was performed with reagents from Promega’s Core footprinting system and a modification of the manufacturer’s protocol. Briefly, the DNA fragment for footprinting experiments was amplified by PCR (using BLR1-Luc plasmid as a template) with a primer incorporating EcoRI at the 5⬘ end and a primer incorporating PstI at the 3⬘ end. To get better resolution of the 5⬘ end of the footprinting probe in the sequencing gel, the PCR-amplified probe covered 217 bp from the blr1 gene 5⬘-flanking region (⫺1096 to ⫺891) and 25 bp at the 5⬘end from the vector backbone plus 8 nucleotides (nt) from the incorporated restriction sites (EcoRI at the 5⬘ end of the fragment and PstI at the 3⬘ end). The footprinting forward primer was 5⬘-CGTGCGAATTCCAGGTGCCAGAACATTTCTC-3⬘ and the reverse primer was 5⬘-CGTGCCTGCAGGTCCTGTCTGAGGGCCGCTTCC-3⬘ (the incorporated EcoRI and PstI restriction sites, respectively, are denoted by underlining). The PCR product was then digested with EcoRI and PstI to produce a 250-bp DNA fragment with a 3⬘ recessed end and 3⬘ protruding end 5⬘ and 3⬘ of the double-stranded DNA (dsDNA) fragments, respectively, to ensure that there was only one end with a 3⬘ recessed terminus to be radioactively labeled. The digested fragment was then exclusively end labeled with [␣-32P]dATP and [␣-32P]dTTP by filling in the 3⬘ recessed end at the EcoRI restriction site with the Klenow fragment of DNA polymerase I and purified by a spin column. The DNA-protein binding reaction mixture (50 ␮l) contained 200,000 cpm of labeled DNA probe, 50 ␮g of nuclear extract (prepared as described above), 25 mM Tris-HCl (pH 8.0), 50 mM KCl, 6.125 mM MgCl2, 0.5 mM EDTA, 10% glycerol, and 1 mM dithiothreitol. After incubation for 20 min on ice, 50 ␮l of 10 mM MgCl2 and 5 mM CaCl2 were added followed by the addition of 0.1 U of DNase I and incubation was continued at room temperature for 3 min. DNase I digestion was quenched by the addition of 90 ␮l of stop solution (200 mM NaCl, 30 mM EDTA, 1% SDS), and the DNA template was purified by phenol-chloroform (1:1) extraction and ethanol precipitation. Samples were run on denaturing 6% polyacrylamide-urea sequencing gels. Gels were dried on Whatman 3MM paper for 3 h with a gel dryer and visualized by autoradiography. A 10-bp DNA ladder labeled (using polynucleotide kinase) with [␥-32P]dCTP was heat denatured and run along with the DNase-treated samples as a size marker, and alignment with the ladder identified the protected nucleotide within the known sequence of the 250-bp probe. EMSA and supershift assay. EMSAs were performed using a Panomics kit. Biotin-labeled probes and unlabeled competitors for DR5, DR2, DR1, EGR1, Sp1, NFATc, NF1, CREB, and Pbx1 were also purchased from Panomics. Oligonucleotides containing the GT box element or the putative RAR-RXR binding sites in the blr1 promoter (identified by a TransFac database analysis) were

2426

WANG AND YEN

MOL. CELL. BIOL. TABLE 2. Oligonucleotide sequences for EMSAs

Oligonucleotide sequencea

DNA element

5⬘-AAGGGTGGGGGTGGGGC-3⬘ .....................................................................................................................................Wild-type GT box motif 5⬘-AAGggtgGGGGTGGGGC-3⬘...........................................................................................................................................The first GT box deleted 5⬘-AAGGGTGGGggtgGGGC-3⬘...........................................................................................................................................The second GT box deleted 5⬘-AAGggtgGGggtgGGGC-3⬘................................................................................................................................................Both of the GT boxes deleted 5⬘-AAAGACCTGAGTTCAAGTCCCAGCTC-3⬘ ..............................................................................................................Putative 1st RARE 5⬘-GAGGAGAAGTGAGGGCAGCCTTTAAACTAG-3⬘................................................................................................Putative 2nd RARE 5⬘-CGAGGGTAGGGTTCACCGAAAGTTCACTCG-3⬘ .................................................................................................DR5 5⬘-AGCTTCAGGTCAGAGGTCAGAGAGCT-3⬘.............................................................................................................DR2 5⬘-GGTAAAGGTCAAAGGTCAATCGGC-3⬘ ..................................................................................................................DR1 5⬘-ATTCGATCGGGGCGGGGCGAG-3⬘ ..........................................................................................................................Sp1 5⬘-GATTCAATGACATCACGGCTGTG-3⬘.......................................................................................................................CREB 5⬘-GGATCCAGCGGGGGCGAGCGGGGGCGA-3⬘.......................................................................................................EGR 5⬘-GGCACCTGTTTCAATTTGGCACGGAGCCAACAG-3⬘.........................................................................................NF-1 5⬘-CGCCCAAAGAGGAAAATTTGTTTCATA-3⬘ ...........................................................................................................NFATc 5⬘-TGTCGAATGCAAATCACTAGAA-3⬘ .........................................................................................................................Oct1 5⬘-CTCCAATTAGTGCATCAATCAATTCG-3⬘ ...............................................................................................................Pbx1 a

Upper strands of oligonucleotides with consensus sequences are underlined; deleted sequences are in lowercase characters.

synthesized by Qiagen Operon Inc. and end labeled with biotin before annealing to make double-stranded probes according to the manufacturer’s instructions. The probes and competitors for EMSA are given in Table 2. HL-60 cell nuclear extracts with or without all-trans-RA induction or in vitro translation products (0.5 ␮g) from pGEM-RAR␣ and pGEM-RXR␣ expressed in a rabbit reticulocyte lysate translation system were first mixed with poly(dI-dC) in EMSA binding buffer and incubated on ice for 10 min, after which labeled probe was then added. In EMSA competition studies, a 100-fold excess of unlabeled competitor oligonucleotide was incubated with nuclear extract or receptors for 30 min before the addition of the labeled probe. The mixture with the labeled probe was incubated for another 30 min at room temperature. For the supershift assay, 0.5 ␮g of the indicated antibody was further added and the mixture was incubated for an additional 15 min at room temperature and then subjected to gel electrophoresis using 6% native polyacrylamide gels in 1⫻ Tris-borate-EDTA buffer at 4°C for 5 h. After electrophoresis, the DNA-protein complex was transferred to a Pall Biodyne B membrane and then cross-linked for 3 min with an UV cross-linker. The membrane-bound DNA-protein complex was treated with streptavidin-horseradish peroxidase conjugate according to the manufacturer’s protocol. Bands for DNA-protein complexes were visualized and quantified on a PhosphorImager. PCR-based mutagenesis. Block or single-base deletions of core sequences of putative cis-acting elements in the blr1 promoter were done by PCR using a Stratagene QuikChange XL mutagenesis kit or QuikChange mutagenesis kit, respectively. A pair of primers (for block deletion) or a single primer (for single-base deletion) was designed for each deletion. The sequences of the

primers lacking the underlined deleted portions are indicated in Table 3. The mutant sequence in the BLR1-Luc plasmid was verified by nucleotide sequence analysis. All primers were PAGE purified and phosphorylated at the 5⬘ end with T4 polynucleotide kinase. Cycling parameters were adjusted on the basis of the thermal denaturation values of the primers. Protein-DNA array and TF-TF array. RA-induced expression of TFs in HL-60 cells was assayed using a Panomics TranSignal protein-DNA array according to the manufacturer’s instructions. Briefly, the nuclear extracts of 1 ␮M all-transRA-treated or untreated HL-60 cells were incubated with TranSignal biotinlabeled probe mix, a cocktail of DNA oligonucleotides corresponding to the DNA consensus sequences for 58 TFs, to allow the formation of DNA-protein complexes, which were then separated from the free probes by agarose gel electrophoresis. The DNA probes in the DNA-protein complexes were then extracted from the gel, dissociated from the DNA-protein complexes, and hybridized to the TranSignal Array membrane that had been spotted with the consensus binding sequences of the 58 TFs. Hybridization signals were visualized using horseradish peroxidase-mediated chemiluminescence detected by exposing the blot to X-ray film. The protein-DNA array hybridization was repeated using separately prepared nuclear extracts, and the same results were obtained. TF-TF array analysis was performed using a Panomics TranSignal TF-TF interaction array kit according to the manufacturer’s instructions. Briefly, nuclear extracts from HL-60 cells treated with RA for 12 h were incubated with a cocktail of biotin-labeled double-stranded oligonucleotide probes (which represent a library of known TF binding consensus sequences), allowing these probes to bind TFs in the nuclear extracts. A RAR␣ antibody was used to pull out RAR␣ and its

TABLE 3. Primers for mutagenesis Primer sequence

Mutated sitea

5⬘-GGGAGCGTGATAACAAGGGTGGGGGTGGGGCCAAGAAGC-3⬘ ............................................................................GTm1 at ⫺1068/⫺1065 5⬘-GCGTGATAACAAGGGTGGGGGTGGGGCCAAGAAGCAGCC-3⬘.............................................................................GTm2 at ⫺1062/⫺1059 5⬘-GGGAGCGTGATAACAAGGGTGGGGGTGGGGCCAAGAAGGCCACC-3⬘ .............................................................GTm12 5⬘-GTGGATCAAGAGAGGAAATGCCCACTTCTGG-3⬘ .......................................................................................................NFATc at ⫺980 5⬘-GCCCACTTCTGGAAGAAAAAGCCACAAAATGAGACTTGG-3 .................................................................................NFATc at ⫺961 5⬘-CACATCATT-GTGGATCAAGAAATTGATCAACATC-3.................................................................................................NFATc at ⫺932 5⬘-CCCTAGTGGTGAGGAAAATGAAGGTTTGGAGG-3⬘....................................................................................................NFATc at ⫺207 5⬘-GGAGTCAAAAGACCTGAGTTCAAGTCCCAGCTCTGC-3⬘..........................................................................................putative RARE at ⫺783 5⬘-GGGAGGAGAAGTGAGGGCAGCCTTTAAACTAGTCATAGGC-3⬘ ...........................................................................putative RARE at ⫺45 5⬘-GGCCTCACGGACCTCCTGAATAAAAAATTGGAGGG-3⬘............................................................................................Oct1 at ⫺705 5⬘-CCTTGGCAGACTGGAATGGTTGATCACCCTAGTGG-3⬘.............................................................................................Oct1 at ⫺230 5⬘-GCGTGGTGGTTTCATTACAAGTTGTGAGCC-3⬘ ............................................................................................................Oct1 at ⫺93 5⬘-CGCAGCTCATTTGCTTAAATTTCGAGC-3⬘ ......................................................................................................................Oct2 at ⫹169/⫹168 5⬘-GCCAGGACTTAGGGGTTTCCCAAGTCAAGGG ............................................................................................................NF-␬B at ⫺274/⫺271 5⬘-CTGTTGATCGGAGTTTCCCTCATCAACCTGCTGAC-3⬘ ...............................................................................................NF-␬B at ⫺132/⫺128 5⬘-GCACTGATGCTGTGGGGGATTTTCCCTCTTTCTTC-3⬘................................................................................................NF-␬B at ⫹45/⫹41 5⬘-GCTTAAATTTCGAGCTGACGGCTGCCACCTCTCTAGAGGC-3⬘ ...............................................................................CREB at ⫹178/⫹181 a

Initiation site for blr1 transcription, ⫹1. Deleted sequences are underlined.

VOL. 24, 2004

RA REGULATES BLR1 BY A NOVEL RARE

2427

FIG. 1. Effect of all-trans-RA on BLR1 mRNA expression. Total RNA isolated from HL-60 cells treated with RA was analyzed by Northern blotting (middle panels) using a blr1 cDNA-specific probe. Ethidium bromide-stained 28S and 18S rRNA bands are shown (lower panels) to indicate the uniformity of RNA loading per lane. The upper charts show the relative Northern blot band intensities (as measured by a PhosphorImager). The values were normalized using 18S rRNA band intensity. (A) Effect of all-trans-RA concentration on blr1 mRNA expression. HL-60 cells were treated for 24 or 48 h with the indicated concentrations of RA. The highest blr1 mRNA expression levels observed were in cells treated with 1 to 10 ␮M RA for 24 h and 0.5 to 10 ␮M RA for 48 h. (B) Time course of RA-induced blr1 mRNA expression. HL-60 cells were treated with 1 ␮M RA for the indicated times. The maximum increase in blr1 mRNA expression occurred at 24 to 60 h.

associated factors on a column, bringing the corresponding TF consensus sequence oligonucleotides. After washing away free oligonucleotide probes and nonspecific binding proteins, the oligonucleotide probes were isolated as described above for the protein-DNA arrays and hybridized to the TranSignal array membrane. Hybridization signals were visualized as described for the proteinDNA array. The TF-TF array experiment was repeated using the nuclear extracts prepared separately, and the same results were obtained. ChIP assay. ChIP assays were performed with a kit and a procedure modified from the manufacturer’s instructions. Briefly, HL-60 cells (treated with or without RA for 12, 24, or 48 h) were cross-linked by formaldehyde at a final concentration of 1% and incubated for 15 min at 37°C. The cells were centrifuged to form a pellet and washed twice with ice-cold PBS containing protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 1 ␮g of aprotinin/ml, and 1 ␮g of pepstatin A/ml). After centrifugation at 300 ⫻ g for 5 min, the cell pellet was resuspended in prewarmed SDS lysis buffer containing the protease inhibitors and incubated for 10 min on ice. The HL-60 cell lysate was sonicated, using an 80-watt model ultrasonic processor (SmithKline, Shelton, Conn.) (with 40 sets of 30-s pulses), on an ice bath to shear DNA to a length of approximately 1 to 2 kb. Samples were centrifuged for 10 min at 13,000 ⫻ g at 4°C. The sonicated cell supernatant was diluted 10-fold in ChIP dilution buffer containing the protease inhibitors. After preclearing the diluted cell supernatant (2 ml) with salmon sperm DNA– protein A agarose–50% slurry (100 ␮l) for 30 min at 4°C with agitation, the agarose was precipitated by brief centrifugation and the supernatant fraction was collected. Antibodies specific for each of RAR␣, RXR␣, Oct1, NFATc3, and CREB2 were added and the suspension was incubated overnight at 4°C with rotation. Salmon sperm DNA–protein A agarose–50% slurry was added followed by incubation for 3 h at 4°C with rotation to collect the antibody-bound complex. After gentle centrifugation, the supernatant containing unbound, nonspecific DNA was carefully removed. The antibody-bound complex was subjected to a series of 5-min washes on a rotating platform with the buffers supplied by the manufacturer. The complex was eluted from the antibody with elution buffer (1% SDS, 0.1 M NaHCO3). The cross-linked portion was dissociated by adding 5 M NaCl and heating at 65°C for 5 h. The fragmented DNA was treated with proteinase K at 45°C for 2 h and purified with a Qiagen quick-spin column. The DNA was analyzed by PCR (100 pmol of each primer set, 1 U of Pfu Hotstart polymerase [Stratagene], 0.01 ␮Ci of [32P]dCTP, 40 cycles). Primer sets for amplifying the blr1 promoter region from fragmented genomic DNA consisted of the sequences from ⫺1096 to ⫺1078 and ⫹185 to ⫹205. Primer sets for amplifying the blr1 promoter region from transfected plasmid pBLR1-Luc are located in the vector backbone sequence directly linking to the two ends of the blr1 promoter fragment. PCR products were resolved on PAGE and exposed to phosphor luminescence plates for 24 h to 48 h. Each experiment was repeated, and the same results were obtained.

RESULTS RA-induced blr1 expression is dose and time dependent and actinomycin D sensitive. RA causes a dose- and time-dependent induction of blr1 expression in HL-60 cells, in which treatment with 1 ␮M all-trans-RA for 24 to 48 h elicits a largely maximized response. HL-60 cells (treated with various concentrations of all-trans-RA from 0.01 to 10 ␮M for 24 and 48 h) were harvested using total RNA for Northern analysis of blr1 expression. Figure 1A shows the results. Expression in untreated cells was undetectable. After 24 h, the response was largely maximized by 1 ␮M all-trans-RA. Even a 10-fold-higher concentration had no significant further effect. After 48 h, a concentration of 1 ␮M still elicited a largely maximized response. This identifies 1 ␮M as an effective dose. Using this dose, the kinetics of RA-induced blr1 expression over time was determined. HL-60 cells were treated with 1 ␮M all-trans-RA and harvested for Northern analysis at sequential times up to 96 h. Figure 1B shows the results. Induced expression maximized after 24 h, although it was first detectable at 9 h and weakly evident at 12 h. The induced expression was maximal from 24 to 60 h and decreased by approximately twofold every 12 h thereafter. This identifies treatment at 1 ␮M for 24 or 48 h as effective in eliciting largely maximized blr1 expression. These will be adopted as standardized treatment conditions in subsequent experiments. It is noteworthy that this typical HL-60 cell subline has a doubling time of approximately 24 h and that 1 ␮M alltrans-RA induces onset of G0 cell cycle arrest and functional differentiation (detected by inducible oxidative metabolism) by 48 h. The majority of the cells are not differentiated until 72 h. By 96 h, essentially all cells are G0 arrested and differentiated as functionally mature myeloid cells (7, 58, 59). Hence, alltrans-RA induced blr1 expression very early in this process, leading ultimately to G0 arrest and differentiation; in particu-

2428

WANG AND YEN

FIG. 2. Inhibition of RA-induced BLR1 mRNA expression by actinomycin D. HL-60 cells were treated with the indicated concentrations of actinomycin D for 24 h following pretreatment with 1 ␮M all-trans-RA for 24 h. (Middle panel) Total RNA isolated from the treated cells was analyzed by Northern blotting using a blr1 cDNAspecific probe. (Lower panel) Ethidium bromide-stained 28S and 18S rRNA bands are shown to demonstrate uniform loading of RNA per lane. (Upper panel) The chart shows the relative intensities of Northern blot bands (as measured with a PhosphorImager). The values were normalized using 18S rRNA band intensity.

lar, it occurred well before any overt changes in cell cycle or differentiation. To confirm transcriptional regulation of blr1 expression in all-trans-RA-treated cells, cells were treated with actinomycin D, which caused a loss of blr1 mRNA in RA-treated cells. HL-60 cells were treated with 1 ␮M all-trans-RA for 24 h to elicit blr1 expression and then with various concentrations of actinomycin D for a further 24 h before Northern analysis. Figure 2 shows the results. Actinomycin D caused a dosedependent loss of blr1 mRNA expression, with a concentration of 1 ␮M causing almost total loss. This concentration of drug did not inhibit HL-60 cell proliferation; hence, the mRNA is labile and was lost within the duration of one cell cycle when synthesis was inhibited. Interestingly, the mRNA of important regulatory molecules is often labile. Taken with previously reported results showing that RA-induced blr1 expression was cycloheximide insensitive (9), the data are consistent with the hypothesis of RA-induced primary transcriptional regulation of blr1.

MOL. CELL. BIOL.

Activation of RAR␣ and RXR␣ expressed in HL-60 cells induces blr1 expression. HL-60 cells express both RARs and RXRs. Using cells that were untreated or treated with alltrans-RA for 48 h, RT-PCR was performed with primers specific for each of the RAR␣, RAR␤, and RAR␥ and RXR␣, RXR␤, and RXR␥ subtypes. Detection of appropriately sized PCR products demonstrated expression of RAR␣, RAR␥, RXR␣, and RXR␤ mRNA transcripts in untreated and RAtreated cells. Expression of RAR␤ was not detected in untreated cells but was detected in RA-treated cells. RXR␥ expression was not detected in either untreated or RA-treated cells (data not shown). Northern analysis was performed to confirm the RT-PCR results. Probes consisted of 3⬘ sequences of the respective RAR␣, RAR␤, and RAR␥ subtype cDNAs and RXR␣, RXR␤, and RXR␥ subtype cDNAs so that transcript variants of each of the different RAR␣, RAR␤, and RAR␥ subtypes and RXR␣, RXR␤, and RXR␥ subtypes would be detected. The probes for RAR␣, RAR␤, and RAR␥ subtypes and RXR␣ and RXR␤ subtypes were generated by RT-PCR using total RNA from untreated or RA (48 h)treated HL-60 cells. Since RXR␥ was undetected by RT-PCR, the RXR␥ probe was generated by PCR from the pSG5/RXR␥ plasmid (39, 46). Figure 3 shows the resulting Northern blot. In consistency with the RT-PCR results, expression of RAR␣ and RAR␥ mRNA transcripts and RXR␣ and RXR␤ mRNA transcripts was detected in untreated and RA-treated cells. RAR␤ mRNA transcripts were detectable in RA-treated cells but not in untreated cells. RXR␥ transcripts were undetectable in either untreated or treated cells. Two transcript variants of RAR␣ and of RXR␤ were detected. Four variants of RAR␤ were observed. The presence of RA significantly increased the expression of RAR mRNA transcripts, with increases ranging from 125% for RAR␣ to 150% for RAR␥ and RAR␥2. RA also induced more modest increases in RXR expression, e.g., 30% for RXR␤ and 80% for RXR␣. Interestingly, Western analysis of protein expression has shown that all-trans-RA caused downregulation of RXR␣ (60) as well as RAR␣ (56), with which the hypo-phosphorylated RAR␣ protein was preferentially lost (59). Thus, for these receptors there is nonconcordance between changes in mRNA and protein expression. Earlier reports originally showed that the RAR␤ protein was detected (using an antiserum) in HL-60 cells with-

FIG. 3. Expression of RARs and RXRs in HL-60 cells. Total RNA (isolated from HL-60 cells that were left untreated [RA⫺] or treated with 1 ␮M RA for 48 h [RA⫹]) was analyzed by Northern blotting. RAR and RXR subtype-specific primers located in the 3⬘ ends of their corresponding cDNAs were used so that for each receptor subtype, all possible isoforms derived from alternative in-frame translational start codons would be detected. Except for the primer for RXR␥ (which was amplified from a plasmid), all other specific primers were prepared by RT-PCR from HL-60 cells. The probes were [␣-32P]dCTP labeled. Ethidium bromide-stained 28S and 18S rRNA bands are shown to serve as size markers. Constitutively expressed transcripts of RAR␣, RAR␥, RAR␥2, RXR␣, and RXR␤ were observed. RAR␣ and RXR␤ showed two isoforms. RAR␤ (detected only after RA treatment) showed four isoforms. RXR␥ was not detectable in untreated or RA-treated cells.

VOL. 24, 2004

RA REGULATES BLR1 BY A NOVEL RARE

2429

FIG. 4. Effect of RAR- and RXR-selective agonists on BLR1 mRNA. HL-60 cells were treated with the indicated retinoid ligands for 48 h, after which total cellular RNA was isolated and analyzed by Northern blotting using a blr1 cDNA-specific probe (middle panel). Ethidium bromidestained 28S and 18S rRNA bands are shown to indicate the uniformity of RNA loading per lane (lower panel). The concentrations of the compounds used are as follows: RA, 1 ␮M; Ro 40-6055 (RAR␣-selective agonist), 1 ␮M; AM80 (RAR␣- and RAR␤-selective agonist), 0.5 ␮M; AGN190168 (RAR␤- and RAR␤␥-selective agonist), 0.5 ␮M; Ro25-7386 (RXR␣-selective agonist), 1 ␮M; AGN194204 (RXR panagonist), 0.3 ␮M. The upper chart shows the relative intensities of bands on the Northern blot (as measured by a PhosphorImager). The values were normalized using 18S rRNA band intensity.

out RA treatment (18, 25). However, according to a later report neither RAR␤ nor RAR␥ mRNA was detected and RAR␣ showed four isoforms in HL-60 cells (47). According to a more recent report, RAR␤ was found only in RA-treated HL-60 cells and RAR␥ was present in HL-60 cells regardless of RA treatment (3), which is in accordance with our results. The various profiles of RARs found by independent laboratories may reflect potential HL-60 subline differences. The currently used HL-60 cells thus express RAR␣, RAR␤, RAR␥, RXR␣, and RXR␤ mRNA, whereas only RAR␤ was detected in RA-treated cells.

Activation of RAR␣ plus RXR␣ induced blr1 expression, but additional activation of other RAR or RXR subtypes did not substantially augment the response. Activation of an RAR or RXR alone failed to induce blr1 expression. RAR- or RXRselective ligands were used to identify which RARs and RXRs could regulate blr1 expression. HL-60 cells were treated for 48 h with 1 uM all-trans-RA, 1 ␮M Ro 40-6055 (an RAR␣selective agonist), 0.5 ␮M AM80 (an RAR␣- and RAR␤-selective agonist), 0.5 ␮M AGN190168 (an RAR␤- and RAR␥selective agonist), 1 ␮M Ro25-7386 (an RXR␣-selective agonist), or 0.3 ␮M AGN194204 (an RXR panagonist) used

FIG. 5. Mapping of the RA-regulatory elements in the BLR1 promoter. Reporter constructs resulting from 5⬘ deletions from the 1,360 bp of the BLR1 5⬘-flanking region were prepared by PCR and transiently transfected into HL-60 cells as described in Materials and Methods. The lines and numbers on the left side of the figure indicate the blr1 promoter sequence contained in each reporter construct (⫹1 denotes the start of transcription). The chart on the right side indicates relative luciferase activity levels of the reporter constructs. All transient transfection studies were conducted in triplicate on three separate occasions, with similar results. The data reported in the chart represent the means of three replicate experiments.

2430

WANG AND YEN

FIG. 6. RA-induced nuclear extracts protect sequences in the distal region of the BLR1 promoter. A dsDNA fragment of 250 bp (spanning 217 bp [⫺1096 to ⫺879] in the BLR1 promoter plus 25 bp at the 5⬘ end from the plasmid backbone sequence in the pBLR1-Luc promoterreporter construct and 8 nt from the incorporated EcoRI and PstI site) was prepared by PCR. After digestion with EcoRI and PstI, the amplified fragment was [␣-32P]dATP and [␣-32P]dTTP end labeled at the 3⬘ recessed end with the Klenow fragment of Escherichia coli DNA polymerase I and used in the DNase I footprinting assay with nuclear extracts from HL-60 cells that were either left untreated (RA⫺) or treated (RA⫹) with all-trans-RA for 48 h. A DNA sequencing ladder (10 bp) was end labeled (using T4 polynucleotide kinase) with [␥32 P]ATP, heat denatured, and corun with the DNase I-treated samples as a size marker. The nucleotide sequence of the DNase I-protected site was determined by alignment of the protected region with the sequencing ladder. An approximately 17-bp region (⫺1071 to ⫺1055) with the indicated sequence was specifically protected from DNase I digestion in the nuclear extracts from RA-treated cells. No footprint was visible with nuclear extracts from untreated cells. An autoradiograph of the DNA footprint is shown. The sizes of the denatured DNA sequence markers that were corun with the samples are indicated with arrows on the left side of the right panel. The 5⬘ and 3⬘ ends of the DNA probe used in the footprinting assay are indicated by arrows pointing up and down. The nucleotide sequence of the DNA footprint is shown on the right. Numbers indicate the positions of start and end points of the protection region relative to ⫹1, the transcriptional initiation site.

either singly or in combinations. The final concentrations used were chosen to optimize response and minimize toxicity. Total RNA from the treated cells was harvested for Northern analysis of blr1 expression. Figure 4 shows the resulting Northern blot. Untreated cells showed no expression, whereas all-transRA-treated cells showed prominent expression. Cells treated with agonists selective for just RARs or just RXRs also showed

MOL. CELL. BIOL.

no expression, indicating that activation of just RARs or just RXRs was insufficient to activate blr1. Treatment with an RAR␣- plus an RXR␣-selective agonist elicited strong expression (comparable to that seen with all-trans-RA), indicating that activation of RAR␣ plus RXR␣ was needed to activate blr1. Additional activation of RAR␤ and RAR␥ or RXR␤ and RXR␥ or all RARs and RXRs by using additional agonists or agonists with broader specificity caused modest increases in expression in some instances but in general failed to substantively increase expression. For example, treatment with the combination of an RAR␣-selective agonist, an RAR␤- and RAR␥-selective agonist, and a pan-RXR agonist resulted in only a slight increase compared to treatment with an RAR␣selective agonist plus an RXR␣-selective agonist. Although there is the caveat that the relative subtype-specific transactivation activities of each of these agonists in the context of the blr1 promoter are not known, making their direct quantitative comparison unfeasible, the overall general indication is that activation of RAR␣ plus RXR␣ is sufficient to induce blr1 expression characteristic of RA, with no further substantive augmentation attributable to additional activation of other RAR or RXR subtypes. A 1,360-bp 5ⴕ-flanking region of the blr1 gene confers RA response and contains a distal 205-bp sequence necessary for response. A 5⬘-flanking region extending from ⫺1096 to the 5⬘ end of the first exon at ⫹266 of the blr1 gene confers response to RA. A Transfac database (http://www.gene-regulation.com /pub/databases.html#transfac) (53) search for putative TF binding sites in the 5⬘ region of the blr1 gene showed that the 1,360 bp 5⬘-flanking region before the first exon was rich with putative sites potentially including two single putative RARRXR half-sites at ⫺785 to ⫺780 and ⫺47 to ⫺42. This region extended from ⫺1096 through the transcriptional start (⫹1) to ⫹266, which is the beginning of the first coding exon. This fragment was isolated by PCR and cloned into pGL3-basic, the promoterless reporter plasmid, to create a blr1 promoter-luciferase reporter construct, BLR1-Luc. HL-60 cells were transiently transfected and then regrown in medium with 1 ␮M all-trans-RA for 24, 48, 72, or 96 h. Luciferase activity was then fluorometrically measured. Figure 5 shows the reporter activity after 48 h. The all-trans-RA-induced expression level of BLR1Luc was 75 to 80 times higher than the expression level of pGL3-basic vector and 50 to 55 times higher than the non-RAinduced expression level of the BLR1-Luc construct. Similar results were observed after 24 h of all-trans-RA treatment (data not shown). Luciferase expression mirrored the kinetics of RA-induced blr1 expression, with comparable maximal activity seen at 24 and 48 h which progressively decreased at 72 h and was hardly detectable at 96 h. The blr1 promoter activation pattern in the transient experiment thus mimicked that observed by Northern analysis for the in vivo pattern (data not shown), indicating that the RA-induced reporter activity bore kinetic fidelity to RA-induced blr1 expression and suggesting that this 1,360-bp 5⬘-flanking region harbors complete cis elements required for the full activity of the blr1 promoter for RA responsiveness. 5⬘ deletion mapping revealed that a 205-bp 5⬘ sequence at the distal portion of this region was critical for RA responsiveness. 5⬘ deletion mutants were created and transiently trans-

VOL. 24, 2004

RA REGULATES BLR1 BY A NOVEL RARE

2431

FIG. 7. Retinoid receptors specifically bind to the GT box element in the blr1 promoter. Electrophoretic mobility shift assays were conducted using the in vitro translation products of RAR␣ and RXR␣ or nuclear extracts of HL-60 cells with (⫹) or without (⫺) 1 ␮M RA treatment and the GT box element (identified in the blr1 promoter as a biotin-labeled double-stranded oligonucleotide probe as described under Materials and Methods). DNA on the membrane was visualized and quantified on a PhosphorImager. The upper charts indicate the relative binding affinities of the DNA-protein complex and supershifted complex. (A) Lane 1 (first lane from the left), free biotin-labeled GT box element probe; lanes 2, 3, and 8, RAR␣-incubated probe; lanes 4, 5, and 9, RXR␣-incubated probe; lanes 6, 10, 11, and 12, RAR␣- and RXR␣-coincubated probe; lanes 3, 5, and 7, probe in the presence of a 100-fold excess of unlabeled GT box competitor; lanes 8 and 10, probe in the presence of RAR␣ antibody; lane 9 and 11, probe in the presence of RXR␣ antibody; lane 12, probe in the presence of anti-RAR␣ IgG and anti-RXR␣ IgG. (B) Lanes 1, 3, and 5, GT box probe and in vitro-translated RAR without RA; lanes 2, 4, and 6, GT box probe and in vitro-translated RAR incubated with 1 ␮M RA for 24 h at 37°C; lane 7, GT box probe and nuclear extracts from non-RA-treated HL-60 cells incubated without RA; lane 8, GT box probe and nuclear extracts from non-RA-treated HL-60 cells incubated with 1 ␮M RA for 24 h at 37°C; lanes 9 and 10, GT box probe and nuclear extracts from RA-treated HL-60 cells incubated without or with 1 ␮M RA for 24 h at 37°C.

fected as described above. Figure 5 shows the luciferase expression level seen after the transfected cells were grown in cultures in all-trans-RA containing medium for 48 h. Deletion of the sequence from ⫺1096 to ⫺891 caused a 95% loss of luciferase expression. Sequential removal of more 5⬘ distal sequences caused no further reduction in reporter gene expression. The distal 205-bp sequence thus contains elements necessary for RA responsiveness. The distal 205-bp sequence contains two GT boxes constituting a novel RARE. RA induces the binding of a nuclear protein complex to an element containing two GT boxes within the critical 205-bp distal sequence from ⫺1096 to ⫺891 that is necessary for RA response. To determine whether RA induces TFs to bind within the 205-bp sequence identified above, DNase I footprinting was performed using nuclear extracts from RA-treated and untreated cells. The probe was a dsDNA fragment of 250 bp that contained the sequence spanning ⫺1096 to ⫺879 from the blr1 promoter plus 25 bp at the 5⬘end from the BLR1-Luc plasmid backbone and 8 nt from the incorporated EcoRI and PstI sites at the 5⬘ and 3⬘ ends of the fragment, respectively. The probe was generated by PCR, digested with the two restriction enzymes, [␣-32P]dATP and [␣32 P]dTTP labeled with Klenow on the 3⬘ recessed end created by EcoRI digestion, and used for DNase I footprinting. Figure 6 shows the resulting sequencing gel. Nuclear lysate from alltrans-RA-treated cells protected a region of the probe from nt 74 to 58, according to the sequencing ladder, which corre-

sponds to an element between ⫺1071 and ⫺1055 in the blr1 promoter. The sequence of the 17-nt protected element (deduced from the known sequence of the blr1 5⬘-flanking region) is shown in Fig. 6 and includes two truncated GT boxes located at ⫺1068 to ⫺1065 and ⫺1062 to ⫺1059. Interestingly, this element has no sequence homology with known RAR-RXR consensus sequences. The GT box-containing element binds recombinant RAR␣ and RXR␣ in vitro independently of RA. EMSAs were performed using the 17-nt GT box-containing element as a probe. The binding of in vitro-translated recombinant RAR␣ (462 amino acids) and RXR␣ (462 amino acids) at high concentrations was assayed. Figure 7A shows the gel. RAR␣ bound the biotin-labeled element and retarded its mobility. The binding could be competed off with excess unlabeled probe. A supershift induced by the addition of anti-RAR␣ antibody confirmed the existence of RAR␣ bound to the probe. Likewise, RXR␣ bound the probe and was competed off with excess probe. A supershift also verified RXR␣ binding to the probe. When both RAR␣ and RXR␣ were added to the probe, a more pronounced band (representing approximately 3.8 times the DNA binding affinity of RAR␣ or RXR␣ of the same mobility) was detected. A supershift caused by the addition of anti-RAR␣ or of anti-RXR␣ verified the presence of RAR␣ and RXR␣ bound to the probe. The addition of both antiRAR␣ and RXR␣ caused a further supershift (representing approximately 3.5 times the DNA binding intensity of RAR␣

2432

WANG AND YEN

or RXR␣ apparent as a single shifted band), indicating that RAR␣ and RXR␣ occurred together on the probe. This is consistent with the occurrence of a RAR␣ and RXR␣ heterodimer on the probe. Since the mobility of the probe with RAR␣ and the probe with RXR␣ was the same as that of the probe with RAR␣ plus RXR␣, the GT box-containing element also binds both RAR␣ and RXR␣ homodimers. The greater signal intensity of the RAR␣-RXR␣-bound probe compared to that of the RAR␣- or the RXR␣-bound probe suggests that the heterodimer binds more efficiently to the GT box-containing element than the homodimers. The EMSA and supershift assay results thus indicate that homo- and heterodimers of RAR␣ and RXR␣ bind to the GT box-containing element. While high concentrations of the in vitro-translated RAR␣ and RXR␣ or RAR␣-RXR␣ heterodimer can bind the GT box sequence without all-trans-RA treatment, binding of the nuclear protein complex seen in footprinting (Fig. 6) occurred with the nuclear extracts from RA-treated (but not untreated) HL-60 cells. To compare the RA dependence of binding to the GT box sequence by RAR-RXR in vitro versus that seen with the nuclear protein complex in vivo, the binding of in vitrotranslated RAR␣ (0.5 ␮g) and RXR␣ (0.5 ␮g) and in vivoexpressed nuclear extracts (80 ␮g) of HL-60 cells was determined in RA-treated and untreated binding reactions. Figure 7B shows that when in vitro-prepared RAR␣ and RXR␣ were used, high concentrations of RAR␣, RXR␣, or RAR␣-RXR␣ (0.5 ␮g each) could efficiently bind to the GT boxes and that RA did not alter the binding of RAR␣ and RXR␣ or RXR␣-

MOL. CELL. BIOL.

RXR␣, suggesting that RA neither induces nor enhances in vitro binding of RAR␣, RXR␣, or RAR␣-RXR␣ to the GT box. In contrast, nuclear extracts from non-RA-treated HL-60 cells failed to bind the GT box but incubation of the nuclear extracts of untreated HL-60 cells with the GT box sequence probe and 1 ␮M RA at 37°C for 24 h resulted in specific binding of a nuclear protein complex. Nuclear extracts of 24 h-RA-treated HL-60 cells also exhibited similarly bound nuclear protein complexes. The slower migration of nuclear protein complex-bound probes compared to that seen with probes bound to just RAR-RXR suggests that RA might induce the formation of a nuclear protein complex binding to the GT box sequence. RA induces binding of a TF complex to the GT box-containing element which could be competed off with DR2 oligonucleotides. If, as the data thus far suggest, the presence of this GT box-containing element is critical to all-trans-RA-induced activation of the blr1 promoter, then one might anticipate that RA induces the binding of a nuclear protein complex to it. Furthermore, if it contains RAR-RXR, then DR1, DR2, or DR5 oligonucleotides, the idealized consensus sequences for RAR-RXR binding, should compete for binding of the complex to the element. EMSAs were performed using the 17-nt GT box-containing element as a probe for in vitro-translated RAR-RXR and nuclear extracts, respectively. The nuclear extracts were from untreated cells and cells treated with alltrans-RA for 48 h, respectively. Figure 8A shows results demonstrating that both DR2 and DR5 compete with the GT box

FIG. 8. DR2 oligonucleotides compete with the GT box element for binding nuclear proteins of RA-treated (⫹) HL-60 cells. EMSAs with competing test oligonucleotides were performed using in vitro-translated RARs and nuclear extracts from untreated (⫺) and all-trans-RA (1 ␮M)-treated HL-60 cells. The probe was the wild-type GT box element. The numbers above the blots (lower panels) indicate the severalfold excesses of the competing oligonucleotides added. The arrows indicate the specific GT box element-nuclear protein binding complex (upper part of the gel) and the free GT box probe (bottom of the gel). The upper chart indicates the relative binding levels of the DNA-protein complex. (A) Lanes 1, 2, and 3 (lane 1 is the first lane from the left), GT box probe and in vitro-translated RAR␣ and RXR␣ receptors; lanes 4, 7, and 10, GT box probe and in vitro-translated RARs incubated with DR1; lanes 5, 8, and 11, GT box probe and in vitro-translated RARs incubated with DR2; lanes 6, 9, and 12, GT box probe and in vitro-translated RARs incubated with DR5. The EMSAs with in vitro-translated RARs were repeated, and the same results were obtained. (B) GT box probe was incubated with nuclear extracts from 48-h RA (1 ␮M)-treated HL-60 cells (lanes 1 and 5 to 10, 80 ␮g each; lanes 2, 3, and 4, respectively, 5, 20, and 80 ␮g). Unlabeled synthetic oligonucleotides for typical RAR binding DR5, DR2, and DR1 sequences, as well as the Sp1 DNA binding consensus sequence, were used as competitors for GT box element binding. The assays repeated using nuclear extracts from cells treated with 1 ␮M RA for 24 h yielded similar results.

VOL. 24, 2004

RA REGULATES BLR1 BY A NOVEL RARE

2433

FIG. 9. Mutation of the GT box element diminishes binding of the RA-induced complex to the sequence motif and abolishes activation of the blr1 promoter following RA induction. (A) EMSAs were done with nuclear extracts of HL-60 cells with (⫹) or without (⫺) RA treatment and the biotin-labeled 17-nt wild-type GT box element (GTbox) from the blr1 promoter and its mutants with either or both GT boxes deleted (13-bp fragment, GTm1 and Gtm2; 9-bp fragment, GTm12). Binding was detected with wild-type GT boxes; this binding was not detected with either of the GT boxes mutated. The EMSA was repeated, and the same result was achieved. (B) Mutation of GT boxes within the wild-type (W.T.) GT box element used in the BLR1-Luc promoter reporter construct was performed using PCR with three pairs of primers as described in Materials and Methods. The BLR1-Luc reporter constructs containing the indicated mutations in either one or both of the GT boxes were cotransfected with pRL-TK into HL-60 cells. Cells were collected after 48 h for assaying luciferase levels. The chart at the right represents the activity of the BLR1-Luc reporters relative to that of the promoterless reporter control. The transient transfection with the mutated constructs was repeated twice with similar results.

sequence for binding in vitro-translated RARs (RAR, RXR, and RAR-RXR) but that DR1 failed to compete. Interestingly, Fig. 8B shows that compared to the binding of in vitro-translated receptors, the binding of the nuclear protein complex from nuclear extracts was not as susceptible to competition. Without RA, there was no protein complex binding to the probe. RA induced the binding of a nuclear protein complex which increased as the amount of lysate was increased from 5 to 20 to 80 ␮g. The binding was progressively competed off by increasing the amounts of cold probe. DR2 oligonucleotide also diminished binding but DR1 and DR5 oligonucleotides did not (Fig. 8B), suggesting that the GT box has a DR2-like binding motif character. This again suggested that the nuclear protein complex has more factors than just RAR and RXR which may be involved in the binding to the GT box motif and that the presumed multiprotein complex has much higher binding affinity for the GT box sequence than for DR2 and DR5. An Sp1 consensus sequence oligonucleotide failed to compete with the GT box-containing element for all-trans-RAinduced binding of the nuclear protein complex. An Sp1 binding motif within the GT box-containing element was predicted by sequence analysis using the Transfac database. Sp1 was found to interact with the heterodimer RAR-RXR in the regulation of thrombomodulin gene transcription by retinoid acid (28). This motivated a test of whether an Sp1 consensus sequence oligonucleotide would compete with the GT box-containing element. An EMSA performed with the addition of a competing Sp1 consensus sequence oligonucleotide failed to compete for binding (Fig. 8B), indicating that assembly of the TF complex on the GT box-containing element was not critically dependent on Sp1. Assembly of the TF complex on the GT box-containing element depends on both GT boxes. Whereas a cold GT box-

containing element could compete with the GT box-containing element probe, deletion mutants of either GT box could not compete with the GT box-containing element probe. EMSAs using the 17-nt GT box-containing element as a probe of nuclear extracts from all-trans-RA-treated and untreated cells were performed as described above. The ability of the unlabeled GT box-containing element or the element mutated in either the first or the second or both GT boxes (the mutants are shown in Fig. 9B) to compete for binding of the nuclear protein complex was tested. Figure 9A shows the results. Mutating either or both of the GT boxes resulted in the loss of the ability to compete for binding. Binding of the TF complex to the GT box-containing element thus depended on both GT boxes being intact. Consistent with the loss of TF binding, mutation of either GT box within this distal element of the 1,360-bp region conferring RA responsiveness caused loss of transcriptional activation in response to the presence of RA (Fig. 9B). HL-60 cells were transiently transfected (as described above) with the BLR1-Luc promoter reporter construct and deletion mutants of either the GT box at ⫺1068 to ⫺1065 or the GT box at ⫺1062 to ⫺1059 or both (respectively denoted GTm1, GTm2, and GTm12). The transfected cells were treated with alltrans-RA and assayed for luciferase expression as before. Figure 9B shows the results. Loss of either box caused loss of approximately 90% of the reporter activity. Loss of both GT boxes reduced the level of activity to that of the promoterless reporter vector control (i.e., that of the pGL3 basic vector). RA-induced transcriptional activation of the blr1 promoter segment is thus strongly dependent on both GT boxes. In sum, the distal portion of the 5⬘-flanking region supporting RA responsiveness in the BLR1 promoter contains an element with two GT boxes. This element binds an RA-induced TF complex. The element binds in vitro both RAR␣ and

2434

WANG AND YEN

MOL. CELL. BIOL.

FIG. 10. RA responsiveness by the GT box element RARE depends on the downstream context in the blr1 promoter. Promoter-reporter constructs from which 5⬘ sequences of ⫺891 to ⫹1 or ⫹1 to ⫹266 in the 1,360 bp of the BLR1 5⬘-flanking region were deleted were prepared by PCR and transiently transfected into HL-60 cells which were then treated with all-trans-RA for 48 h and assayed for luciferase activity as described in Materials and Methods. The lines and numbers indicate the blr1 promoter sequence contained in each promoter-reporter construct. The chart on the right side indicates the relative luciferase activity level for each promoter-reporter construct. With the loss of either segment, the distal sequence from ⫺1096 to ⫺891 containing the GT box element was no longer able to confer responsiveness to all-trans-RA. All transient transfection studies were conducted in triplicate on three separate occasions with similar results. The chart reports the means of three experiments.

RXR␣ of high concentration. Assembly of the TF complex on this element depends on both GT boxes. Consistent with this, transcriptional activation in response to RA also depends on both GT boxes. The data suggest that this GT box-containing element binds a TF complex containing RAR␣ and RXR␣ (in response to the presence of RA) to transcriptionally activate blr1. RA responsiveness conferred by the distal element depends on downstream sequences containing critical Oct1, NFATc, and CREB sites. The ability of the TF complex-bound distal promoter element to confer RA responsiveness depends on downstream sequences. In particular, mutation of downstream Oct1, NFATc, and CREB sites (but, interestingly, not mutation of either of the two RAR-RXR half-sites) caused loss of RA responsiveness. Deletion mutants of the BLR1-Luc promoter reporter described above were created by deletion of ⫺891 to ⫹1 or ⫹1 to ⫹266 and transiently transfected into cells, and their luciferase activity levels were assayed after 48 h of all-trans-RA treatment as described earlier. Figure 10 shows the results. With the loss of either segment, the distal sequence from ⫺1096 to ⫺891 containing the GT box sequence was no longer able to confer responsiveness to RA. There were thus sequences within the deleted segments that were essential for the distal domain to activate transcription in response to RA. A TransFac database analysis of potential TF binding motifs in these segments indicated a pair of putative RAR-RXR halfsites at ⫺785 to ⫺780 and ⫺47 to ⫺42. To test their significance, BLR1-Luc promoter-reporter transfection were again performed after mutating either or both of the putative halfsites. Figure 11 shows the results. Mutation of either or both sites within the BLR1-Luc caused no loss of RA responsiveness, indicating that these putative RAR-RXR half-sites downstream of the GT box-containing sequence are not needed for RA responsiveness and do not constitute a RARE. Consistent with this, EMSAs using annealed double-stranded oligonucleotides containing either of the single putative RAR-RXR halfsites, a 26-nt probe containing the first site, or a 30-nt probe containing the second site and nuclear extracts from 24- and 48-h RA-treated HL-60 cells showed no protein-DNA complex shift (data not shown). The TransFac database analysis indicated three potential

Oct1 binding motifs at ⫺706 to ⫺703, ⫺231 to ⫺228, and ⫺94 to ⫺91. To test their potential functional significance, mutants of BLR1-Luc were created in which each site was mutated either singly or in all combinations (including mutation of all three sites). The luciferase activity was again measured after all-trans-RA treatment of transiently transfected cells. Figure 12 shows the results. Mutation of the Oct1 site at ⫺705 caused loss of response, whereas mutation of the other two caused no loss of response. This was consistent in all seven combinations of mutated Oct1 sites (representing all combinations). Thus, one specific site of the three Oct1 sites is needed for RA response but the other two are not. The TransFac database search also indicated four potential NFATc sites at ⫺980 to ⫺987, ⫺961 to ⫺958, ⫺932 to ⫺929, and ⫺207 to ⫺204. Mutation of each of these sites was again

FIG. 11. The putative RAR-RXR half-sites identified by TransFac database analysis do not control blr1 response to RA. Mutations within either or both of two putative RAR-RXR half-sites at ⫺783 to ⫺782 and ⫺45 to ⫺44 in the blr1 promoter identified by TransFac database analysis were created in the BLR1-Luc construct with two pairs of primers specific to these putative sites as described in Materials and Methods. The wild-type (W.T.) BLR1-Luc reporter and its mutants were cotransfected with pRL-TK into HL-60 cells. The cells were treated with 1 ␮M all-trans-RA for 48 h and harvested to assay luciferase activity. The values represent the activity of the BLR1-Luc reporters relative to that of the promoterless reporter control. Transient transfections with these mutated constructs were repeated twice, and the same results were obtained.

VOL. 24, 2004

FIG. 12. Mutation of Oct1, NFATc, or CREB motifs downstream of the GT box element abolishes RA inducibility of the BLR1 promoter. Deletion mutagenesis targeting the three Oct1, four NFATc, and one CREB consensus sequences in the blr1 promoter was performed on the BLR1-Luc construct as described in Materials and Methods. Bases in lowercase are deleted. The consensus sequence (ATTC) in italics is located in the antisense strand of DNA. The BLR1-Luc reporter constructs containing the indicated mutations were cotransfected with pRL-TK into HL-60 cells, which were treated for 48 h with all-trans-RA and then collected to assay luciferase activity levels. The values represent the activity of the BLR1-Luc reporters relative to that of the promoterless reporter control. Transient transfections with the mutated constructs were repeated twice, and the same results were obtained.

RA REGULATES BLR1 BY A NOVEL RARE

2435

performed to test (using BLR1-Luc transfection as described above) their functional significance with respect to RA responsiveness. Figure 12 shows the results. Mutation of the site at ⫺207 caused loss of RA responsiveness, but mutations at the other sites had little effect. Thus, only the ⫺207 site was necessary for RA responsiveness. Finally, the TransFac database search indicated a potential CREB site at ⫹178 to ⫹181. Testing (using BLR1-Luc transfection) the functional significance of this site indicated that mutation of the site caused loss of RA responsiveness. (Figure 12 shows the results of the luciferase activity assay.) The CREB site thus appears to be necessary for RA responsiveness. To confirm that the TransFac database-identified, functional NFATc, Oct1 and CREB sites in the blr1 promoter bind their cognate factors, supershift EMSAs were performed. The oligonucleotide sequences corresponding to the functional binding sites (for Oct1, TGAATAAAAAATTG ⫺707 to ⫺694; for NFATc, GTGAGGAAAATG ⫺212 to ⫺201; for CREB, GAGCTGACGGCT ⫹174 to ⫹185) in the blr1 promoter were synthesized and used as probes. Nuclear extracts from HL-60 cells were incubated with the biotin-labeled probe, and antibodies specific to each of the individual factors were added to induce supershifts. Since Oct1 and Oct2 recognize very similar binding sites, EMSAs with Oct1- or Oct2-specific antibodies were used to detect which actually bound the blr1 promoter sequences. The results are shown in Fig. 13. Anti-Oct1, but not anti-Oct2, caused a supershifted band (Fig. 13, left panel), suggesting that Oct1 binds the putative binding site in the blr1 promoter. NFATc1-, NFATc2-, NFATc4-, and NFATc5-specific antibodies did not produce super shifts, but NFATc3 did (Fig. 13, center panel), suggesting that NFATc3 is the factor binding to the putative site in the blr1 promoter. While CREB1-specific antibody did not cause a supershift, the CREB2-specific antibody did (Fig. 13, right panel), suggesting that CREB2 but not CREB1 binds the putative site in the blr1 promoter. The data thus demonstrate that the sequences identified in silica and shown genetically to be functional do biochemically bind their corresponding TFs.

FIG. 13. Oct1, NFATc3, and CREB2 bind to their corresponding consensus sites in the blr1 promoter. EMSAs were performed to confirm that the Oct1, NFATc, and CREB sites identified in the blr1 promoter by TransFac database analysis and functional mutagenesis experiments are actual consensus binding sites for these factors. The oligonucleotide sequences harboring the database-deduced binding sites (Oct1, TGAATAAAAA ATTG [⫺707 to ⫺694]; NFATc, GTGAGGAAAATG [⫺212 to ⫺201]; CREB, GAGCTGACGGCT [⫹174 to ⫹185]) were synthesized and used as probes. Nuclear extracts (NE) from non-RA-treated HL-60 cells were incubated with the biotin-labeled probe. Antibodies specific to the individual factors were added to induce supershifts. Supershifts were observed with anti-Oct1 (left panel), anti-NFATc3 (center panel), and anti-CREB2 (right panel), suggesting that Oct1, NFATc3, and CREB2 bind their putative consensus binding sites in the blr1 promoter.

2436

WANG AND YEN

MOL. CELL. BIOL.

FIG. 14. TF-TF array for detecting TFs binding to RAR␣. Array analysis of TFs binding to RAR␣ in RA-induced HL-60 cells was carried out using a Panomics TransSignal TF-TF array system as described in Materials and Methods. The organization of the TF-TF array was identical to the that of the TransSignal DNA-protein array described for Fig. 15. (A) The right panel shows hybridization signals of TFs binding to RAR␣ using anti-RAR␣, and the left panel shows the results obtained using normal IgG with a control. (B) The table shows the identities of TFs in the array that bind to RAR␣ after treatment of HL-60 cells with 1 ␮M all-trans-RA for 12 h. The most prominent RAR␣ binding TFs detected in the array are circled.

The RA-induced TF complex assembled at the RAR␣-RXR␣ binding GT box-containing element includes at least Oct1 and NFATc3. Aside from RAR␣ and RXR␣, other potential components of the nuclear protein complex binding the GT box sequence are not known, although the mutagenesis and EMSA data suggest that Oct1, NFATc3, and CREB2 might be involved. The differences in RA dependence and DR5 competition levels between the RAR-RXR binding and nuclear extract-derived complex binding to the GT sequence suggest more complexity in the in vivo complex compared to that seen with recombinant RAR and RXR. This may reflect the presence of factors other than RAR and RXR in the complex. To identify candidates that might bind RAR␣ in the TF complex, a TF-TF array analysis was performed. A commercial kit was used in which nuclear lysate is incubated with a cocktail of labeled dsDNA representing the cognate consensus sequences for a panel of 58 TFs. TFs that are present bind their cognate DNA sequences. Those protein-DNA complexes that contain RAR␣ are separated out using an anti-RAR␣ antibody. Nonspecific Ig is used as a control. Free probe is separated from the protein-DNA complexes. The DNA is dissociated from the DNA-protein complexes and hybridized against a membrane spotted with oligonucleotides representing the consensus binding sequences of 58 TFs. The labeled probe then betrays which TFs were detected according to their binding to their cognate consensus sequence. Figure 14 shows the resulting blot prepared using nuclear lysate from cells treated with all-trans-RA for 48 h. The presence of Oct1, Pbx1, CREB, NF-1, EGR1, and NFATc was detected as candidate RAR␣ binding partners. The DNA-protein array assay, which is similar to the TF-TF assay, was performed without the anti-RAR␣ antibody (thereby not restricting the assay to TFs complexed with

RAR␣) and used to compare nuclear lysate from all-trans-RAtreated versus untreated cells. The results showed that alltrans-RA increased binding of Oct1, Pbx1, CREB, NF-1, EGR1, and NFATc to their respective consensus sequences. Figure 15 shows the resulting array blots. Among the RAR␣-complexing candidates, Oct1 and NFATc consensus sequences competed with the GT box-containing element for binding of the all-trans-RA-induced transcription complex. EMSA assays using the 17-nt GT box-containing element as a probe and nuclear lysate from cells treated with all-trans-RA for 48 h were performed as described above. The ability of oligonucleotides representing the Oct1, Pbx1, CREB, NF-1, EGR1, and NFATc cognate consensus sequences to compete for binding was tested. Figure 16A shows the resulting blot. Oct1 and NFATc oligonucleotides competed with the probe. The signal intensity of the original TF complexbound probe diminished, and a single new signal corresponding to a smaller complex appeared. CREB, EGR1, and Pbx1 oligonucleotides did not compete with the probe for binding of the 48-h RA-induced transcription complex under the conditions used. The data imply the participation of Oct1 and NFATc in the TF complex bound to the GT box sequence along with RAR␣ and RXR␣. To further test whether RAR␣, RXR␣, NFATc3, Oct1, and CREB2 TFs are the components of the GT-bound complex, supershift EMSAs were performed using nuclear extracts from cells treated with RA for 24 and 48 h, the GT box sequence as a probe, and antibodies to these factors. The results presented in Fig. 16B show that the addition of anti-RAR␣, anti-RXR␣, anti-NFATc3, and anti-Oct1 resulted in supershifted bands, suggesting that RAR␣, RXR␣, NFATc3, and Oct1 are components of the complex in RA-treated HL-60 cells. The addi-

VOL. 24, 2004

RA REGULATES BLR1 BY A NOVEL RARE

2437

FIG. 15. Protein-DNA array analysis for detecting RA-inducible TFs in HL-60 cells. DNA-protein array analysis was carried out using a Panomics TransSignal DNA-protein array system as described in Materials and Methods. The results of array hybridizations in which nuclear extracts of HL-60 cells that were left untreated (48-h control) or treated with 1 ␮M all-trans-RA for 48 h were analyzed are shown. Experiments repeated using the nuclear extracts prepared after 12, 24, and 48 h of all-trans-RA treatment all gave the same results. Each TF was represented in four spots on the blot in a two-by-two grouping of two rows and two columns. The first row consisted of DNA spotted normally, and the second row consisted of DNA diluted 1:10. (A) The panels show the resulting hybridization signals of TFs in cells with (right panel) or without (left panel) all-trans-RA induction. (B) The table shows the TFs represented in the array and identifies those with strong hybridization signals induced by the presence of all-trans-RA in HL-60 cells. Roman lightface letters indicate no induced expression; roman boldface letters without underlining indicate low-level induced expression; roman boldface letters with dotted underlining indicate medium-level induced expression; roman boldface letters with regular underlining indicate high-level induced expression; boldface letters with thick underlining indicate very-high-level induced expression; boldface italic letters with underlining indicate expression at lower levels before induction and at higher levels after all-trans-RA induction.

tion of anti-CREB2 did not cause a supershift, indicating that CREB2 was not in the complex after RA treatment for 24 and 48 h (Fig. 16B). To determine the level of dependence on the presence of RA for incorporation of these factors in the TF complex binding the GT box sequence, EMSA supershifts were likewise performed to compare the effects of adding or deleting RA. When RA was added, binding of the TF complex to the GT box sequence occurred and all the factors (including RAR␣, RXR␣, NFATc3, and Oct1) were observed in the multiprotein complex. However, without RA none of the components could be detected by specific antibodies and no protein complex was observed to bind the GT box sequence of the blr1 promoter in HL-60 cells (Fig. 16C). These data suggest that RA induces the association of these components into a multicomponent complex binding the GT box sequence. RAR␣, RXR␣, Oct1, NFATc3, and CREB2 bind in vivo to their consensus sites in the blr1 promoter. The promoterreporter, mutagenesis, EMSA, and supershift assay results described above provide evidence for the assembly of a TF complex including RAR␣, RXR␣, Oct1, and NFATc3 for transcriptional activation of the blr1 gene by RA. To test whether these TFs bind to the blr1 promoter in vivo, ChIP assays were performed. HL-60 cells and HL-60 cells transfected with the wild-type GT box construct or GT box mutant construct (GTm12) treated with or without all-trans-RA treatment were harvested after 12, 24, or 48 h and cross-linked with formaldehyde. Nuclear extracts were prepared, DNA was fragmented by sonication, and DNA bound by proteins was immu-

noprecipitated with antibodies specific for RAR␣, RXR␣, Oct1, NFATc3, or CREB2. DNA was released from the immunoprecipitates and assayed by PCR for the presence of the blr1 promoter sequence. As shown in Fig. 17, the endogenous blr1 promoter was detected (using PCR primers located within the blr1 promoter region) in the RAR␣ and RXR␣ immunoprecipitates from 12-h RA-treated HL-60 cells. The promoter sequence was also detected in HL-60 cells transfected with the wild-type GT box construct by using primers located in the construct backbone regions flanking the blr1 promoter insert in the plasmid (Fig. 17A). In both cases, the promoter sequence was not detected in the absence of RA (Fig. 17B). In vivo binding of RARs and RXRs to the blr1 promoter thus depends on the presence of RA. This is consistent with the results previously shown for EMSAs using nuclear extracts containing physiological concentrations of RARs and RXRs in which RAR␣, RXR␣, or RAR-RXR binding to the GT box sequence was RA dependent. In contrast, the blr1 promoter was detected in the Oct1, NFATc3, and CREB2 immunoprecipitates from both nontransfected and transfected HL-60 cells regardless of RA treatment (Fig. 17B), suggesting that binding of Oct1, NFATc3, and CREB2 to the blr1 promoter might be RA independent. Regardless of RA induction, the blr1 promoter was not detected (by ChIP assays using PCR with primers located in the construct backbone sequences flanking the blr1 promoter region) in the RAR and RXR immunoprecipitates from HL-60 cells transfected with the GT box mutant construct. Mutating the

2438

WANG AND YEN

MOL. CELL. BIOL.

FIG. 16. RA induces formation of a multimolecular complex that includes RAR␣, RXR␣, Oct1, and NFATc3 and binds the GT box element in the blr1 promoter. (A) (Lower panel) EMSAs with competing test oligonucleotides were performed using the 17-bp GT box element labeled with biotin as the probe of nuclear extracts from HL-60 cells that were left untreated (⫺) or treated with 1 ␮M all-trans-RA for 48 h (⫹). A total of 100-fold of each unlabeled GT box element (lane 3 [lane 1 is the first lane from the left]), CREB (lane 4), EGR1 (lane 5), NF1 (lane 6), NFATc (lane 7), Oct1 (lane 8), and Pbx1 (lane 9) oligonucleotide was added as a competitor for their cognate factors’ possible binding to the GT box element DNA-protein complex. The arrows indicate the specific DNA-protein complex (upper part of the gel) and the free GT box probe (bottom part of the gel). The assay was repeated three times using different preparations of nuclear extracts of HL-60 cells with 24- and 48-h all-trans-RA treatment. All replicates yielded similar results. (Upper panel) The chart indicates relative binding levels of the DNA-protein complex with the GT box element or targeted competitor oligonucleotides. (B) EMSAs were performed using antibodies specific to RAR␣, RXR␣, Oct1, NFATc3, and CREB2 added to the incubation reactions of GT box probe and nuclear extracts (NE) of 48-h RA (1 ␮M)-treated HL-60 cells. Supershifts were observed with antibodies for RAR␣, RXR␣, Oct1, and NFATc3 but not CREB2, suggesting that RAR␣, RXR␣, Oct1, and NFATc3 are included in the complex induced by 48-h RA treatment. (C) EMSAs were performed with the GT box sequence probe and nuclear extracts from untreated cells in the absence or presence of RA in the binding reaction to test the dependence of binding on RA. Nuclear extracts from non-RA-treated HL-60 cells showed no GT box-bound complexes regardless of the presence or absence of added antibodies. Nuclear extracts from cells with RA added showed the GT box-bound complex which was supershifted by the addition of antibodies specific to the components of the complex. GT box probe and nuclear extracts from non-RA-treated HL-60 cells incubated with 1 ␮M all-trans-RA for 24 h (shown) or 48 h yielded the same pattern of protein-DNA interaction. RA thus induces formation of a multiprotein complex binding to the GT box sequence.

GT box thus abolished binding of RAR␣ and RXR␣ to the promoter sequence. In contrast to HL-60 cells transfected with the GT box mutant promoter, cells transfected with the wildtype GT box construct gave the same ChIP results for proteinpromoter interactions as nontransfected HL-60 cells. The data suggest that GT boxes are direct target sequences for RARs and RXRs. In contrast, GT box mutation did not prevent the binding of Oct1, NFATc3, and CREB2 to the blr1 promoter, thereby excluding the possibility that Oct1, NFATc3, and CREB2 bind to the GT boxes. The data are consistent with a model in which these three TFs bind to their consensus sites and the GT boxes bind RARs and RXRs; while the former are

RA independent, the latter are RA dependent and may precipitate for formation of the transcriptionally active complex. CREB in association with RAR␣ was detected in TF-TF array assays after 12 h of all-trans-RA treatment; CREB bound in vivo to the blr1 promoter was detected in ChIP assays at 12 h with and without RA treatment. Furthermore, mutagenesis of its consensus binding site in the blr1 promoter caused loss of blr1 transcriptional activity, suggesting its involvement in blr1 activation by RA. However, it was not found bound to GT boxes in EMSA assays of all-trans-RA-induced nuclear protein complexes at 24 or 48 h, which is when blr1 reaches peak expression. ChIP assays (using both cells treated with all-

VOL. 24, 2004

RA REGULATES BLR1 BY A NOVEL RARE

2439

FIG. 17. Physical interaction of RAR␣, RXR␣, Oct1, NFATc3, and CREB2 with the blr1 promoter. ChIP assays were conducted to assess the interaction of RAR␣, RXR␣, Oct1, NFATc3, and CREB2 with the blr1 promoter in intact HL-60 cells. HL-60 cells and HL-60 cells transfected with GT box wild-type (W.T.) or mutant constructs were treated with (⫹) or without (⫺) 1 ␮M all-trans-RA for 12, 24, and 48 h. The cells were cross-linked with formaldehyde. Chromatin- and TF-bound transfected plasmids were immunoprecipitated with antibodies to RAR␣, RXR␣, Oct1, NFATc3, or CREB2. After reversing the cross-link, DNA released from the immunoprecipitates was used for radioactive PCR analysis by two different pairs of primers (detecting the endogenous or transfected promoter). Specific bands representing the promoter sequence were detected by resolving the PCR products on PAGE. (A) The positions of the two sets of primers used for detecting the blr1 promoter sequence are shown. (B) ChIP assays using 12-h all-trans-RA-treated or untreated cells were conducted with antibody specific to RAR␣, RXR␣, Oct1, NFATc3, or CREB2. (C) Anti-CREB2 ChIP assays using 24- and 48-h all-trans-RA-treated or untreated cells.

trans-RA for 24 and 48 h or left untreated and cells transfected with constructs carrying wild-type or mutant GT box motifs) were thus performed with CREB2 antibody to determine in vivo binding at these times. The results showed in all cases that without RA, CREB2 bound to the blr1 promoter in HL-60 cells but that when cells were treated with RA this in vivo binding was no longer detected at 24 or 48 h (Fig. 17C), indicating that in vivo binding of CREB2 to the blr1 promoter is transient after RA treatment. Since RA induces blr1 expression after 9 to 12 h, CREB2 may be involved in early stages of blr1 activation without the need for a sustained presence. DISCUSSION RA is known to regulate cell proliferation and differentiation in a variety of cells (9, 10, 12, 23, 26, 29, 41, 52). One archetypal cell used to study its mechanism of action is the HL-60 human myeloblastic leukemia cell, which is an uncommitted myelomono-

cytic precursor cell capable of undergoing either myeloid or monocytic differentiation (55, 57). RA induces the myeloid differentiation of these cells accompanied by G0 cell cycle arrest (58, 59). Although blr1 is not expressed in untreated cells, RA induces an early upregulation of blr1 expression antecedent to any overt evidence of cell differentiation or cell cycle arrest. And its expression propels RA-induced differentiation and G0 arrest (7). Blr1 is thus an instance of a gene whose expression changes (from undetectable to prominently detectable levels) early in the RAinduced cascade leading to cell differentiation and cell cycle arrest and whose expression propels the RA-induced differentiation and G0 arrest of the cells. Furthermore, blr1 is a receptor whose expression obviously engenders consequential signaling to elicit cellular consequences. These considerations motivate interest in how RA regulates early expression of a receptor consequential to cell differentiation cell cycle arrest. The present studies used blr1 as a prototype of this class of genes in these cells and pursued the mechanism of its regulation by RA.

2440

WANG AND YEN

The nuclear factors and cis-elements regulating the blr1 promoter in HL-60 cells treated with RA are thus of interest. Alignment of the blr1 promoter nucleotide sequences with known TF binding motifs by a TransFac database analysis revealed two single putative RAR-RXR half-sites in the blr1 promoter (5⬘-AGTTCA-3⬘ at ⫺785 to ⫺780 and 5⬘-AGGGC A-3⬘ at ⫺47 to ⫺42). However, mutating these putative consensus sites caused no loss of RA-induced blr1 promoter activity. Neither did they cause a mobility shift in EMSAs using nuclear extracts of RA-treated HL-60 cells (data not shown). These two nonrepeated single RAR-RXR half-sites thus do not confer RA response. This motivates efforts to identify other cis elements in the blr1 promoter responsible for conferring RA responsiveness. A novel RARE (5⬘-AAGGGTGGGGGTGGGCC-3⬘ [arbitrarily named the GT box element]) in the distal region of the human BLR1 promoter (⫺1071 to ⫺1055) was identified and characterized by DNase I footprinting, EMSAs, and mutagenesis in the present study. This element in the blr1 promoter confers RA-responsive activation of blr1 expression by a transcription complex containing RAR␣ and RXR␣. Two types of RARE have hitherto been identified in various natural promoters. One type is a direct repeat (5⬘-PuGGTCA-3⬘ [Pu ⫽ A or G]) that follows the 1- to 5-spacer rule for receptor specificity whereby separation by 1, 3, 4, or 5 nt determines specificity for RXRs, vitamin D receptors, thyroid hormone receptors, or RARs (which also bind with a direct repeat separation of 2) (42, 44). The second type of RARE has various spacers that are from 7 to 15 (or even more) bp in size (21, 33) but are still derivatives of the same hexameric DNA core motif. The novel RARE presented here has several unique characteristics. (i) The binding sequence motif differs from the ones commonly known in RA-responsive genes, and the unique sequence of the RARE features two GT boxes which are G rich and resemble Sp1 sites but do not interact with Sp1. (ii) This GT box sequence hosts a unique binding complex. Nuclear extracts from non-RA-treated HL-60 cells do not bind this element, although high concentrations of both RAR-RXR heterodimers and RAR or RXR homodimers produced by in vitro translation bind this element regardless of the presence or absence of RA. The suggestion that RAR-RXR heterodimers bind more efficiently than RAR or RXR homodimers is consistent with previous reports comparing heterodimer and homodimer binding to various DR1 to DR5 elements (35, 39, 55, 61). (iii) The ability of the element to confer RA responsiveness requires the formation of a complex containing multiple TFs associating with RAR and RXR. (iv) The ability of the element to confer RA responsiveness also depends on downstream cis elements of the blr1 promoter. These features may reflect potentially complex synergism between RAR and RXR and additional transactivators in cell-type-specific and differentiation-specific control of RA-induced BLR1 expression. The novel GT box motif is a DR2-like RARE. Although both DR2 and DR5 can compete with the GT boxes in binding RAR␣ and RXR␣ in vitro, the GT boxes more closely resemble a DR2 element for RAR-RXR. DR2 can compete with the GT boxes to some extent in binding to the complex, while DR5 cannot. The GT box binding motif covers 17 bp with a 2-base spacer between the two GT boxes, in similarity to the consensus sequence of DR2 covering 18 bp with a 2-base spacer

MOL. CELL. BIOL.

between its half-sites. The similarity in structural organization to a DR2 suggests the potential for RAR-RXR binding to this RARE in the blr1 promoter. Although RA induces binding of a multiprotein complex to GT boxes such that Oct1 and NFATc3 (and probably CREB2 at an early stage) are implicated components of the TF complex assembled at this novel RARE, it is unlikely that they all directly bind DNA there. There is no structural basis within this GT box sequence for binding these factors except for that of RAR-RXR, and their cognate DNA consensus sequences have no evident homology to any part of the novel RARE. In the in vivo studies with ChIP assays, furthermore, GT box mutation prevented binding by RAR␣ and RXR␣, but not by Oct1, NTATc3, and CREB2, to the blr1 promoter. It thus appears that Oct1, NFATc3, and CREB2 bind their own corresponding sites in the blr1 promoter and that only RAR␣ and RXR␣ bind the GT boxes. The interaction of the multiprotein complex with the GT boxes is thus apparently through RAR-RXR bound to the GT boxes. RA induces association of a multifactor complex that binds to this novel RARE through RAR␣ and RXR␣. A survey of RAR␣-associated TFs in the nuclear lysate of all-trans-RAtreated cells identified NFATc, Oct1, EGR1, CREB, nuclear factor 1 (NF1), and pre-B-cell leukemia TF 1 (Pbx1) as candidates that might also be in the complex binding to this RARE. Among those that are associated with the complex, competitive EMSAs identified two, Oct1 and NFATc. ChIP assays demonstrated that five TFs, RAR␣, RXR␣, Oct1, NFATc3, and CREB2, bound the blr1 promoter, suggesting that CREB2 might also be required for activation of blr1. The GT box-containing element identified here does not confer RA responsiveness by itself but depends on downstream cis elements. In particular, RA response depended on specific intact downstream Oct1, NFATc, and CREB sites. Evidence in this paper demonstrates that Oct1, NFATc3, and probably CREB2 were involved in the formation of a multiprotein complex with RAR␣ and RXR␣. It is thus possible that these downstream Oct1, NFATc, and CREB sites play a role in the assembly of Oct1, NFATc3, and CREB2 into the RA-induced TF complex on the novel RARE. Since in vivo RAR and RXR are not found bound to the GT box sequence except as part of the RA-induced nuclear protein complex, it is likely that RAactivated RAR and RXR precipitate formation of the nuclear protein complex at the GT box sequence through the use of Oct1, NFATc3, and CREB2 tethered to their downstream sequences. One can speculate that the DNA thus bends back on itself, possibly serving as a tether to facilitate assembly of the complex. Figure 18 shows an illustration representing blr1 transcription activation. Oct1 is a ubiquitously expressed TF. It binds the promoters of Igs, IL-2, and IL-3 as well as housekeeping genes (51). It can cooperate with numerous other factors, including GR, SP1, AP-1, TATA binding protein, TFIIB, and SNAPc (51). Oct1 has been implicated in RA-induced HL-60 cell differentiation through its regulation of H2B gene activity (34). Oct1 and Oct2 are equally capable of binding to the octamer sequence. Oct2 is the dominant form in B lymphocytes, where it activates cell-specific genes such as Ig genes, Crisp3, and CD36 (54). Bob1, a B cell-specific coactivator of octamer binding factors, binds the POU domains of either Oct1 or Oct2, forming a ternary complex on DNA. After binding Bob1, Oct1 is con-

VOL. 24, 2004

FIG. 18. An illustration of the model for blr1 transcription activation by RA through a novel RARE. Without RA, RARs at in vivo levels are unable to bind the GT box motif whereas Oct1, NFATc3, and CREB2 bind their corresponding consensus binding sites in the blr1 promoter in a RA-independent manner. Treatment with RA causes a nuclear protein complex containing TFs RAR␣, RXR␣, Oct1, and NFATc3 to form at a novel GT box RARE. CREB2 may be involved transiently in the complex at an early stage of activation. RA thus causes the DNA-tethered Oct1, NFATc3, and CREB2 to associate with the RAR␣-RXR␣ which binds to the novel GT box RARE to finally activate transcription of blr1 gene in HL-60 cells.

verted from a ubiquitous to a lymphocyte-specific factor with increased DNA binding specificity (22). In B lymphocytes, Oct2 regulates blr1 gene expression, binding the blr1 promoter such that a mutation at ⫹133 inhibited blr1 gene activation (54). To determine whether it might have a similar role in HL-60 cells, the Oct2 site at ⫹132 to ⫹133 in the blr1 promoter of HL-60 cells was deleted from the BLR1-Luc construct, which was then transiently transfected into RA-treated and untreated cells. The mutation had no effect (data not shown), suggesting that Oct1 but not Oct2 functions in octamer-dependent RA-responsive blr1 gene expression in HL-60 cells. This was also confirmed by EMSA, which showed that only antiOct1 caused a band retardation. NFATc interacts with various receptors, and its activation can be regulated by T- and B-cell membrane receptors as well as G protein-associated receptors (49). Its transcript is found in leukocytes from peripheral blood and spleen. NFATc can bind certain ␬B-like sites (44). It can also cooperatively interact with AP-1 (11). NFATc modulates gene expression of IL-2, IL-3, IL-4, IL-5, IL-8, IL-13, granulocyte-macrophage colony-

RA REGULATES BLR1 BY A NOVEL RARE

2441

stimulating factor, alpha interferon, gamma interferon, and CD40L and has been implicated in immune response and immune system development (49). The suggested potential regulatory role for NFATc in leukocyte differentiation is consistent with the present results. CREB is known to be a transcription activator that binds to the cAMP response element within a variety of cAMP-responsive promoters to regulate gene expression (20, 27, 45). CREB was not detected in the DNA-protein complex from HL-60 cells by competitive EMSAs and supershift EMSAs after 24 or 48 h of RA treatment. In vivo binding of CREB2 to the blr1 promoter was also not observed at 24 and 48 h of RA treatment in ChIP assays. However, CREB binding to RAR␣ was detected in the TF-TF array assay, and in vivo binding of CREB2 to the blr1 promoter was observed in ChIP assays at 12 h of RA treatment. Since activation of blr1 by RA begins at 9 to 12 h, this suggests that CREB2 might be needed in an early stage of blr1 activation. In consistency with this putative transient need, mutagenesis of the CREB DNA consensus site downstream of the novel RARE abolished RA-induced blr1 promoter activation, indicating a critical CREB dependence in RA-induced blr1 expression. CREB2 is thus likely involved in activation of the blr1 promoter in a transient manner at an early stage of blr1 activation. Taken together, the data of this study suggest that RAR␣, RXR␣, Oct1, NFATc3, and CREB2 act in a synergistic fashion to regulate blr1 expression at a novel RARE that contains two GT boxes. RA induces binding of a multimolecular complex containing RAR␣, RXR␣, Oct1, NFATc3, and CREB2 to this RARE. RA responsiveness depends on downstream sequences that include intact Oct1 and NFATc binding motifs as well as a CREB site. Formation of the complex at this RARE may involve a loop in the blr1 promoter region that brings multiple DNA consensus sites into close proximity to each other, enabling all the involved factors to associate with RAR and RXR and bind to the RARE through RAR-RXR to initiate transcription activation of the blr1 gene. An illustration representing the model of RA-induced blr1 activation through the novel RARE is shown in Fig. 18. ACKNOWLEDGMENTS We are very grateful to Willie Mark (Memorial Sloan-Kettering Cancer Center) for critically reading the manuscript and helpful suggestions, to P. Walker and R. Chandraratna (Allergen Co.), M. Klaus (Roche Co.), and K. Shudo (U. Tokyo) for their generous gifts of retinoids reagents, and also to S. Varvayanis for her technical support. This work was supported in part by grants from the National Institutes of Health (U.S. Public Health Service) and the U.S. Department of Agriculture Human Nutrition Program. REFERENCES 1. Altucci, L., A. Rossin, W. Raffelsberger, A. Reitmair, C. Chomienne, and H. Gronemeyer. 2001. Retinoic acid-induced apoptosis in leukemia cells is mediated by paracrine action of tumor-selective death ligand TRAIL. Nat. Med. 7:680–686. 2. Anderson, K. L., K. A. Smith, K. Conners, S. R. McKercher, R. Maki, and B. E. Torbett. 1998. Myeloid development is selectively disrupted in PU. 1 null mice. Blood 91:3702–3710. 3. Babina, M., K. Mammeri, and B. M. Henz. 2001. Retinoic acid up-regulates myeloid ICAM-3 expression and function in a cell-specific fashion—evidence for retinoid signaling pathways in the mast cell lineage. J. Leukoc. Biol. 69:361–371. 4. Barella, L., M. Loetscher, A. Tobler, M. Baggiolini, and B. Moser. 1995. Sequence variation of a novel heptahelical leucocyte receptor through alternative transcript formation. Biochem. J. 309:773–779.

2442

WANG AND YEN

5. Battle, T. E., and A. Yen. 2002. Ectopic expression of CXCR5/BLR1 accelerates retinoic acid- and vitamin D3-induced monocytic differentiation of U937 cells. Exp. Biol. Med. 227:753–762. 6. Battle, T. E., M. S. Roberson, T. Zhang, S. Varvayanis, and A. Yen. 2001. Retinoic acid-induced blr1 expression requires RAR␣, RXR, and MAPK activation and uses ERK2 but not JNK/SAPK to accelerate cell differentiation. Eur. J. Cell Biol. 80:59–67. 7. Battle, T. E., R. A. Levine, and A. Yen. 2000. Retinoic acid-induced blr1 expression promotes ERK2 activation and cell differentiation in HL-60 cells. Exp. Cell Res. 254:287–298. 8. Benoit, G., L. Altucci, M. Flexor, S. Ruchaud, J. Lillehaug, W. Raffelsberger, H. Gronemeyer, and M. Lanotte. 1999. RAR-independent RXR signaling induces t(15; 17) leukemia cell maturation. EMBO J. 18:7011–7018. 9. Breitman, T. R., S. E. Selonick, and S. J. Collins. 1990. Induction of differentiation of the human promyelocytic leukemia cell line (HL-60) by retinoic acid. Proc. Natl. Acad. Sci. USA 77:2936–2940. 10. Castaigne, S., C. Chomienne, M. T. Daniel, P. Fenaux, R, Berger, and L. Degos. 1990. All-trans retinoic acid as a differentiation therapy in acute promyelocytic leukemia. I. Clinical results. Blood 76:567–572. 11. Chen, L., M. G. Oakley, J. N. M. Glover, J. Jain, P. B. Dervan, P. G. Hogan, A. Rao, and G. L. Verdine. 1995. Only one of the two DNA-bound orientations of AP-1 found in solution cooperates with NFATp. Curr. Biol. 5:882– 889. 12. De Luca, L. 1991. Retinoids and their receptors in differentiation, embryogenesis, and neoplasia. FASEB J. 5:2924–2933. 13. Dobner, T., I. Wolf, T. Emrich, and M. Lipp. 1992. Differentiation-specific expression of a novel G protein-coupled receptor from Burkitt’s lymphoma. Eur. J. Immunol. 22:2795–2799. 14. Emrich, T., R. Forster, and M Lipp. 1994. Transmembrane topology of the lymphocyte-specific G protein-coupled receptor BLR1: analysis by flow cytometry and immunocytochemistry. Cell. Mol. Biol. 40:413–419. 15. Forster, R., A. E. Mattis, E. Kremmer, E. Wolf, G. Brem, and M. Lipp. 1996. A putative chemokine receptor, BLR1, directs B cell migration to defined lymphoid organs and specific anatomic compartments of the spleen. Cell 87:1037–1047. 16. Forster, R., I. Wolf, E. Kaiser, and M. Lipp. 1994a. Selective expression of the murine homologue of the G protein-coupled receptor BLR1 in B cell differentiation, B cell neoplasia and defined areas of the cerebellum. Cell. Mol. Biol. 40:381–387. 17. Forster, R., T. Emrich, E. Kremmer, and M. Lipp. 1994b. Expression of the G protein-coupled receptor BLR1 defines mature, recirculating B cells and a subset of T-helper memory cells. Blood 84:830–840. 18. Gaub, M. P., Y. Lutz, E. Ruberte, M. Petkovich, N. Brand, and P. Chambon. 1989. Antibodies specific to the retinoic acid human nuclear receptors ␣ and ␤. Proc. Natl. Acad. Sci. USA 86:3089–3093. 19. Gianni, M., I. Ponzanelli, L. Mologni, U. Reichert, A. Rambaldi, M. Terao, and E. Garratini. 2000. Retinoid-dependent growth inhibition, differentiation and apoptosis in acute promyelocytic leukemia cells. Expression and activation of caspases. Cell Death Differ. 7:447–460. 20. Gonzalez, G. A., K. K. Yamamoto, W. H. Fischer, D. Karr, P. Menzel, W. Biggs, W. W. Vale, and M. R. Montminy. 1989. A cluster of phosphorylation sites on the cyclic AMP-regulated nuclear factor CREB predicted by its sequence. Nature 337:749–752. 21. Gronemeyer, H., and R. Miturski. 2001. Molecular mechanism of retinoid action. Cell. Mol. Biol. Lett. 6:3–52. 22. Gstaiger, M., O. Georgiev, H. Van Leeuwen, P. Van der Vliet, and W. Schaffner. 1996. The B cell coactivator Bob1 shows DNA sequence-dependent complex formation with Oct1/Oct-2 factors, leading to differential promoter activation. EMBO J. 15:2781–2790. 23. Gudas, L. J. 1994. Retinoids and vertebrate development. J. Biol. Chem. 269:15399–15402. 24. Gunn, M. D., V. Ngo, K. M. Ansel, E. H. Ekland, J. G. Cyster, and L. T. Williams. 1998. A B-cell-homing chemokine made in lymphoid follicles activates Burkitt’s lymphoma receptor-1. Nature 391:799–803. 25. Hashimoto, Y., M. Petkovich, M. P. Gaub, H. Kagechika, K. Shudo, and P. Chambon. 1989. The retinoic acid receptors ␣ and ␤ are expressed in the human promyelocytic leukemia cell line HL-60. Mol. Endocrinol. 3:1046– 1052. 26. Higuchi, T., G. E. Hannigan, D. Malkan, H. Yeger, and B. R. Williams. 1991. Enhancement by retinoic acid and dibutyryl cyclic adenosine 3⬘:5⬘-monophosphate of the differentiation and gene expression of human neuroblastoma cells induced by interferon. Cancer Res. 51:3958–3964. 27. Hoeffler, J. P., T. E. Meyer, Y. Yun, J. L. Jameson, and J. F. Habener. 1988. Cyclic AMP-responsive DNA-binding proteins: structure based on cloned placental cDNA. Science 242:1430–1433. 28. Horie, S., H. Ishii, F. Matsumoto, M. Kusano, K. Kizaki, J. Matsuda, and M. Kazama. 2001. Acceleration of thrombomodulin gene transcription by retinoic acid. J. Biol. Chem. 276:2440–2450. 29. Huang, M. E., Y. C. Ye, S. R. Chen, J. R. Chai, J. X. Lu, L. Zhoa, L. J. Gu, Z. Y. Wang. 1988. Use of all-trans retinoic acid in the treatment of acute promyelocytic leukemia. Blood 72:567–572. 30. Jozan, S., S. Paute, M. Courtade-Saidi, S. Julie, S. Vidal, R. Bugat, and A.

MOL. CELL. BIOL.

31.

32. 33.

34.

35. 36. 37. 38.

39.

40.

41. 42. 43. 44. 45. 46.

47. 48. 49. 50. 51. 52.

53. 54.

Valette. 2002. All-trans retinoic acid enhances CDDP-induced apoptosis: modulation of the CDDP effect on cell cycle progression. Int. J. Oncol. 20:1289–1295. Kaiser, E., R. Forster, I. Wolf, C. Ebensperger, W. M. Keuhl, and M. Lipp. 1993. The G protein-coupled receptor BLR1 is involved in murine B cell differentiation and is also expressed in neuronal tissues. Eur. J. Immunol. 23:2532–2539. Kastner, P., M. Mark, and P. Chambon. 1995. Nonsteroid nuclear receptors: what are genetic studies telling about their role in real life? Cell 83:859–869. Kato, S., H. Sasaki, M. Suzawa, S. Masushige, L. Tora, P. Chambon, and H. Gronemeyer. 1995. Widely spaced, directly repeated PuGGTCA elements act as promiscuous enhancers for different classes of nuclear receptors. Mol. Cell. Biol. 15:5858–5867. Kim, K. Y., K. R. Kweon, M. S. Lee, S. T. Kwak, K. E. Kim, B. D. Hwang, and K. Lim. 1995. Reduced level of octamer binding transcription factor (Oct1) is correlated with H2B histone gene expression during differentiation of HL-60 cells by all-trans retinoic acid. Biochem. Biophy. Res. Commun. 213:616–624. Kliewer, S. A., K. Umesono, D. J. Mangelsdorf, and R. M. Evans. 1992. Retinoic X receptor interacts with nuclear receptors in retinoic acid, thyroid hormone and vitamin D3 signaling. Nature 355:446–449. Kouba, M., M. Vanetti, X. Wang, M. Schafer, and V. Hollt. 1993. Cloning of a novel putative G-protein-coupled receptor (NLR) which is expressed in neuronal and lymphatic tissue. FEBS Lett. 321:173–178. Krust, A., P. Kastner, M. Petkovich, A. Zelent, and P. Chambon. 1989. A third retinoic acid receptor, hRAR␥. Proc. Natl. Acad. Sci. USA 86:5310– 5314. Legler, D. F., M. Loetscher, R. S. Roos, I. Clark-Lewis, M. Baggiolini, and B. Moser. 1998. B-cell-attracting chemokine 1, a human CXC chemokine expressed in lymphoid tissue, selectively attracts B lymphocytes via BLR1/ CXCR5. J. Exp. Med. 187:655–660. Leid, M., P. Kastner, R. Lyons, H. Nakshatri, M. Saunders, T. Zacharewski, J. Y. Chen, A. Staub, J. M. Garnier, S. Mader, and P. Chambon. 1992. Purification, cloning, and RXR identity of the Hela cell factor with which RAR or TR heterodimerizes to bind target sequences efficiently. Cell 68: 377–395. Lin, B., G. Q. Chen, D. Xiao, S. Kumar Kolluri, X. Cao, H. Su, and X. K. Zhang. 2000. Orphan receptor COUP-TF is required for induction of retinoic acid receptor ␤, growth inhibition, and apoptosis by retinoic acid in cancer cells. Mol. Cell. Biol. 20:957–970. Lotan, R. 1997. Roles of retinoids and their nuclear receptors in the development and prevention of upper aerodigestive tract cancers. Environ. Health Perspect. 105(Suppl. 4):985–988. Manglesdorf, D. J., and R. M. Evans. 1995. The RXR heterodimers and orphan receptors. Cell 83:841–850. Manglesdorf, D. J., U. Kazukiko, and R. M. Evans. 1994. The retinoid receptors, p. 319–349. In M. B. Sporn, A. Roberts, and D. S. Goodman (ed.), The retinoids, 2nd ed. Raven Press, New York, N.Y. McCaffrey, P. G., A. E. Goldfeld, and A. Rao. 1994. The role of NFATp in cyclosporin A-sensitive tumor necrosis factor-␣ gene transcription. J. Biol. Chem. 269:30445–30450. Montminy, M. R., and L. M. Bilezikjian. 1987. Binding of a nuclear protein to the cyclic-AMP response element of the somatostatin gene. Nature 328: 175–178. Nagpal, S., M. Saunders, P. Kastner, B. Durand, H. Nakshatri, and P. Chambon. 1992. Promoter context- and response element-dependent specificity of the transcriptional activation and modulating functions of retinoic acid receptors. Cell 70:1007–1019. Nagy, L., V. A. Thomazy, G. L. Shipley, L. Fesus, W. Lamph, R. A. Heyman, R. A. Chandraratna, and P. J. Davies. 1995. Activation of retinoid X receptors induces apoptosis in HL-60 cell lines. Mol. Cell. Biol. 15:3540–3551. Nerlov, C., and T. Graf. 1998. PU. 1 induces myeloid lineage commitment in multipotent hematopoietic progenitors. Genes Dev. 12:2403–2412. Rao, A., C. Luo, and P. G. Hogan. 1997. Transcription factors of the NFAT family: regulation and function. Annu. Rev. Immunol. 15:707–747. Siebenlist, U., G. Franzoso, and K. Brown. 1994. Structure, regulation, and function of NF-␬B. Annu. Rev. Cell Biol. 10:405–455. Veenstra, G. T. C., P. C. Van der Vliet, and O. H. J. Destree. 1997. POU domain transcription factors in embryonic development. Mol. Biol. Rep. 24:139–155. Warrell, R. P., Jr., S. R. Frankel, W. H. Miller, D. A. Scheinberg, L. M. Itri, W. N. Hittelman, R. Vyas, A. Andreff, A. Tafudi, A. Jakubowski, J. Gabrilove, M. Gordon, and E. Dmitrovsky. 1991. Differentiation therapy of acute promyelocytic leukemia with tretinoin (all-trans retinoic acid). N. Engl. J. Med. 324:1385–1393. Wingender, E., X. Chen, R. Hehl, H. Karas, I. Liebich, V. Matys, T. Meinhardt, M. Prub, I. Reuter, and F. Schacherer. 2000. TRANSFAC: an integrated system for gene expression regulation. Nucleic Acids Res. 28:316–319. Wolf, I., V. Pevzner, W. Kaiser, G. Bernhardt, E. Claudio, U. Siebenlist, R. Forster, and M. Lipp. 1998. Downstream activation of a TATA-less promoter by Oct-2, Bob1, and NF-␬B directs expression of the homing receptor BLR1 to mature B cells. J. Biol. Chem. 273:28831–28836.

VOL. 24, 2004 55. Yen, A. 1990. HL-60 cells as a model of growth control and differentiation: the significance of variant cells. Hematol. Rev. 4:5–46. 56. Yen, A., L. Placanica, L, S. Bloom, S. Varvayanis. 2001. Polyomavirus small t antigen prevents retinoic acid-induced retinoblastoma protein hypophosphorylation and redirects retinoic acid-induced G0 arrest and differentiation to apoptosis. J. Virol. 75:5302–5314. 57. Yen, A., M. Forbes, G. deGala, and J. Fishbaugh. 1987. Control of HL-60 cells differentiation lineage specificity, a late event occurring after precommitment. Cancer Res. 47:129–134. 58. Yen, A., M. S. Roberson, S. Varvayanis, and A. T. Lee. 1998. Retinoic acid induces mitogen-activated protein (MAP)/extracellular signal-regulated kinase (ERK) kinase-dependent MAP kinase activation needed to elicit HL-60 cell differentiation and growth arrest. Cancer Res. 58:3163–3172. 59. Yen, A., R. Sturgill, and S. Varvayanis. 2000. Retinoic acid increases amount

RA REGULATES BLR1 BY A NOVEL RARE

2443

of phosphorylated RAF; ectopic expression of cFMS reveals that retinoic acid-induced differentiation is more strongly dependent on ERK2 signaling than induced G0 arrest is. In Vitro Cell. Dev. Biol. 36:249–255. 60. Yen, A., V. Cherington, B. Schaffhausen, K. Marks, and S. Varvayanis. 1999. Transformation defective polyoma middle T antigen mutants defective in PLC␥, PI-3 or SRC kinase activation enhance ERK2 activation and promote retinoic acid-induced, cell differentiation like wild-type middle T. Exp. Cell Res. 248:538–551. 61. Zhang, X. K., B. Hoffman, P. B. V. Tran, G. Graupner, and M. Pfahl. 1992. Retinoid X receptor is an auxiliary protein for thyroid hormone and retinoic acid receptors. Nature 355:441–446. 62. Zhang, X. K., and M. Pfahl. 1993. Regulation of retinoid and thyroid hormone action through homodimeric receptors. Trends Endocrinol. Metab. 4:156–162.