GA-Binding Protein and p300 Are Essential Components of a Retinoic Acid-Induced Enhanceosome in Myeloid Cells

Abstract

Expression of CD18, the β chain of the leukocyte integrins, is transcriptionally regulated by retinoic acid (RA) in myeloid cells. Full RA responsiveness of the CD18 gene requires its proximal promoter, which lacks a retinoic acid response element (RARE). Rather, RA responsiveness of the CD18 proximal promoter requires ets sites that are bound by GA-binding protein (GABP). The transcriptional coactivator, p300, further increases CD18 RA responsiveness. We demonstrate that GABPα, the ets DNA-binding subunit of GABP, physically interacts with p300 in myeloid cells. This interaction involves the GABPα pointed domain (PNT) and identifies p300 as the first known interaction partner of GABPα PNT. Expression of the PNT domain, alone, disrupts the GABPα-p300 interaction and decreases the RA responsiveness of the CD18 proximal promoter. Chromatin immunoprecipitation and chromosome conformation capture demonstrate that, in the presence of RA, GABPα and p300 at the proximal promoter recruit retinoic acid receptor/retinoid X receptor from a distal RARE to form an enhanceosome. A dominant negative p300 construct disrupts enhanceosome formation and reduces the RA responsiveness of CD18. Thus, proteins on the CD18 proximal promoter recruit the distal RARE in the presence of RA. This is the first description of an RA-induced enhanceosome and demonstrates that GABP and p300 are essential components of CD18 RA responsiveness in myeloid cells.

Gene expression is tightly regulated during the differentiation of myeloid progenitor cells into monocytes and granulocytes. Although most myeloid genes are controlled at the level of transcription, there is no single master myeloid transcription factor. Rather, the expression of myeloid genes requires the combinatorial actions of several key transcription factors, which together regulate the pattern of gene expression that mediates myeloid differentiation (28, 36, 37).

Retinoic acid (RA) plays an important role in the regulation of myeloid differentiation, and it induces normal granulocytic differentiation of the myeloid cell lines HL-60 and U937 (5). Typically, RA-induced gene expression occurs via its binding of the retinoic acid receptors (RARs) and retinoid X receptors (RXRs). RARs are nuclear receptors that bind to retinoic acid response elements (RAREs) and, when bound by ligand, recruit a complex of proteins that activate transcription (34). In the absence of ligand, RARs are associated with a corepressor complex that silences transcription.

While RA plays an important role in normal myeloid differentiation, alterations in retinoid receptors block myeloid differentiation and are associated with leukemia. The response to RA is blocked in HL-60 cells that express RARα with ligand-binding domain mutations (10). Similarly, overexpression of RARα and expression of C-terminal truncations of RARα block myeloid differentiation. Clinically, chromosomal translocations that create fusion proteins of the RARα N terminus with other proteins cause acute promyelocytic leukemia; this block can be overcome in vitro and clinically by pharmacological levels of RA (17).

CD18 is the β chain of the leukocyte adhesion molecules LFA-1, Mac-1, and p150/95 (CD11a, -b, and -c, respectively) (3). Expression of CD18 is important for leukocyte adhesion, immune function, and the inflammatory response; abnormalities in CD18 expression cause leukocyte adhesion deficiency (40). The CD18 gene is transcriptionally regulated in myeloid cells, and its expression increases dramatically during granulocytic differentiation (15, 25). Myeloid cell-specific transcription of CD18 is regulated in part by RA (6).

Two distinct promoter elements control the transcription of CD18. The proximal promoter, which includes the first 96 nucleotides upstream of the transcriptional start site, consists of two Sp1 sites and three ets-binding sites. Functional cooperation between Sp1 and the ets factors, GA-binding protein (GABP) and PU.1, increases expression of CD18 in myeloid cells (23, 24, 26). The distal enhancer, which lies nearly 1 kb upstream of the transcriptional start site, contains an RARE (6).

An increasing body of evidence indicates that RA responsiveness is not mediated by retinoid receptors alone (6, 14, 18, 39). Surprisingly, the CD18 proximal promoter, alone, accounts for one-half of the RA responsiveness. This promoter region is RA responsive despite the absence of an RARE or direct binding by the RAR/RXR heterodimer. Full RA responsiveness of the CD18 proximal promoter requires intact Sp1 and ets-binding sites and integration into chromatin (6). The ets sites, alone, can confer RA responsiveness to a heterologous promoter, indicating a possible role for GABP in mediating the RA responsiveness of CD18.

GABP is a widely expressed ets factor that regulates the transcription of several myeloid genes, including CD18, neutrophil elastase, lysozyme, and folate receptor β (21, 22, 24, 29). GABP is unique among the more than two dozen mammalian ets transcription factors in that it is an obligate heterotetramer. It is composed of two distinct and unrelated proteins: GABPα and GABPβ (27). GABPα contains the ets DNA-binding domain in its carboxy terminus and a pointed (PNT) domain (which serves as a protein-protein interaction domain in other ets factors) in its amino terminus. The transcriptional activation domain of GABP is located in the distinct protein, GABPβ, which also contains Notch-like ankyrin repeats that mediate the interaction with GABPα.

The transcriptional coactivator p300 interacts with several transcription factors that are involved in myeloid differentiation (4). p300 coactivates transcription via its histone acetyltransferase activity as well as by mediating interactions with the basal transcription machinery; p300 has also been identified as a key scaffolding protein in the formation of enhanceosomes (8). p300 directly interacts with RAR/RXR as part of the RA-induced coactivator complex and has been shown to interact with several ets factors (4). p300 has been shown to increase the RA responsiveness of the CD18 promoter (6). We have previously proposed that p300 physically interacts with the factors present at the CD18 proximal promoter in order to mediate the RA responsiveness of this element.

In this report we demonstrate that p300 physically interacts with the ets factor GABPα in myeloid cells in the absence of RA and that this interaction is required for maximal RA responsiveness of the CD18 promoter. This interaction is direct and is mediated by the GABPα PNT domain and the cysteine- and histidine-rich domains of p300. This demonstrates p300 as the first known interaction partner for the GABPα PNT domain. Disruption of the GABPα-p300 interaction via expression of the PNT domain alone (as a dominant negative construct) decreases the RA responsiveness of the CD18 proximal promoter, and this effect can be reversed by the overexpression of p300. These results demonstrate the requisite role for the GABPα-p300 interaction in mediating CD18 RA responsiveness. We used chromatin immunoprecipitation (ChIP) to demonstrate that, in the presence of RA, GABPα and p300 at the CD18 proximal promoter recruit RAR/RXR from the distal RARE to form an RA-inducible enhanceosome. We independently confirmed the formation of this enhanceosome using the chromosome conformation capture (3C) assay. Disruption of the GABPα/p300/RXR complex with a dominant negative p300 molecule (which can bind to RXR, but not to GABPα) blocks formation of the enhanceosome, reduces the RA responsiveness of the CD18 gene, and disrupts the RA responsiveness of the CD18 proximal promoter alone. Thus, we demonstrate that GABPα, via its interaction with p300, is essential for formation of an RA-induced enhanceosome that regulates CD18 expression in myeloid cells.

MATERIALS AND METHODS

Cell culture.

U937 cells (ATCC CRL 1593; ATCC, Rockville, MD) were passaged twice weekly in RPMI 1640 (GIBCO BRL, Gaithersburg, MD) supplemented with 10% heat-inactivated fetal calf serum (ICN, Costa Mesa, CA), l-glutamine, and penicillin-streptomycin (complete RPMI medium) in an atmosphere of 5% CO₂. Human embryonic kidney cells (293 cells) were grown in Dulbecco modified Eagle medium (ATCC) supplemented with 10% fetal calf serum, l-glutamine, and penicillin-streptomycin.

DNA constructs.

The pGEX2T-GABPα and pGEX2T-GABPβ constructs were previously described (24). GABPα cDNA constructs are derived from the human clone E4TF1 (GenBank D13318), and derivative constructs are numbered so that the translational start ATG (nucleotide 181) is considered nucleotide 1. pGEX2T-GABPαN1-203 was created by deleting nucleotides 609 to 1362 from pGEX2T-GABPα by digestion with EcoRI. pGEX2TGABPαC263-454 was created by inserting a 0.6-kb MfeI/EcoRI fragment of GABPα (nucleotides 789 to 1362) into pGEX2T. pGEX2T-PNT was created by amplifying nucleotides 430 to 758 of GABPα, using a 5′ primer containing a BamHI linker and a 3′ primer containing an MfeI linker and inserting this fragment into the BamHI and EcoRI sites of pGEX2T. pcDNA3Flag-PNT was created by inserting the same PCR-amplified sequence into pcDNA3-Flag. pGEX2T-ΔPNT was created by first removing a 0.6-kb EcoRV/MfeI fragment, containing nucleotides 214 to 789 of GABPα, from pCAGGS-GABPα (24) followed by ligation with a 0.2-kb PCR-amplified insert containing nucleotides 214 to 428 of GABPα with EcoRV and MfeI linkers, to generate the pCAGGS-ΔPNT construct. The ΔPNT region was then transferred to pGEX2TGABPα by digesting both pCAGGS-ΔPNT and pGEX2T-GABPα with BseRI and XhoI and ligating the 1-kb fragment from pCAGGS-ΔPNT into the pGEX2T-GABPα vector to create pGEX2T-ΔPNT. pcDNAFlag-N143 was created by amplifying nucleotides 4 to 429 of GABPα using a 5′ primer containing a BamHI linker and a 3′ primer containing an EcoRI linker and inserting this fragment into the BamHI and EcoRI sites of pCDNA3Flag. pGEX4T-CH2 was created by amplifying nucleotides 1162 to 1450 of p300 using 5′ and 3′ primers containing EcoRI linkers and inserting this fragment into pGEX4T-1. MSCV-HAp300N-GFP was created by inserting a BamHI fragment from pCMVβ-HAp300 containing the hemagglutinin (HA) tag and the first 1,785 nucleotides of p300 into the MSCV BamHI site upstream of the internal ribosome entry site.

Immunoprecipitation.

Whole-cell extracts were prepared from U937 cells, and the protein concentration was determined by bicinchoninic acid protein assay (Pierce, Rockford, IL). A 1.0-mg aliquot of whole-cell extract was incubated with antiserum to GABPα (Santa Cruz Biotechnology, Santa Cruz, CA) and normal rabbit immunoglobulin G (IgG; Santa Cruz Biotechnology) or p300 (Santa Cruz Biotechnology) at 4°C overnight, followed by 1-hour incubation with protein G-Sepharose (Amersham Bioscience) at 4°C for 1 h. Precipitated material was boiled in 2× loading buffer, electrophoresed in a 7.5% polyacrylamide gel, and transferred to nitrocellulose (Bio-Rad, Hercules, CA). Nitrocellulose blots were immunodetected with polyclonal antiserum against GABPα (Santa Cruz Biotechnology), p300 (Santa Cruz Biotechnology), or the Flag tag (Sigma, St. Louis, MO) and horseradish peroxidase-conjugated goat anti-rabbit secondary antiserum (Sigma) and detected by enhanced chemiluminescence (Amersham Bioscience, Piscataway, NJ).

GST pull-down assays.

As a source of p300, whole-cell extracts were prepared from 293 cells 24 h after CaPO₄ transfection of pCMVβ-HAp300, and protein concentration was determined by bicinchoninic acid protein assay. Alternatively, GABPα (pCMX-GABPα) (31) Flag-PNT or Flag-N143 were in vitro translated in the presence of [³⁵S]Met using a T7 TNT-coupled reticulocyte lysate kit (Promega, Madison, WI). Input material was then precleared at 4°C for 1 h with glutathione beads (Amersham Bioscience), followed by incubation with glutathione S-transferase (GST) or the indicated GST fusion proteins bound to glutathione beads at 4°C for 1.5 h. Precipitated material was boiled in 2× loading buffer, electrophoresed in an 8% or 6% polyacrylamide gel, and either dried (for radiolabeled proteins) or transferred to nitrocellulose (Bio-Rad) (for whole-cell extracts). Nitrocellulose blots were immunodetected with polyclonal antiserum to p300 (Santa Cruz Biotechnology) and horseradish peroxidase-conjugated goat anti-rabbit secondary antiserum (Sigma) and detected by enhanced chemiluminescence (Amersham Bioscience). Dried gels of radiolabeled GABPα were exposed to film overnight.

Inhibition of RA responsiveness.

For Flag-PNT, 5 × 10⁶ log-phase U937 cells that were stably transfected with the CD18 −96/luc reporter plasmid (6) were electroporated (960 μF, 300 V) with no DNA, 1 μg of pCDNA3-Flag vector, or 0.1 or 0.3 μg of pCDNA-FlagPNT with pCDNA3-Flag to a total of 1 μg. For the rescue experiment, 5 μg HA-p300, 0.3 μg Flag-PNT, or 5 μg HA-p300 and 0.3 μg Flag-PNT were transfected via Lipofectamine (Invitrogen). For HA-p300N, 1 × 10⁶ log-phase CD18 −918/luc or CD18 −96/luc cells were transfected via Lipofectamine with no DNA, 10 μg of mouse stem cell virus (MSCV) vector, or 1, 5, or 10 μg of MSCV-HAp300N with MSCV to a total of 10 μg. Transfected cells were then split into two samples and treated with either all-trans RA (Sigma) at 10⁻⁵ M in dimethyl sulfoxide (DMSO), or with DMSO vehicle alone as a negative control. Twenty-four hours after transfection, cells were harvested and reporter gene expression was evaluated with a luciferase assay kit (Promega). Induction by RA was defined as luciferase activity in the presence of RA divided by the activity of the paired DMSO control. Each transfection was replicated at least three times.

Chromatin immunoprecipitation.

RA-treated (24 h) or untreated U937 cells were cross-linked for 20 min in a final concentration of 1% formaldehyde. A total of 1 × 10⁷ cells were used per immunoprecipitation mixture. In the p300 ChIP interference assay, cells were infected with MSCV-HAp300N-GFP 48 h prior to RA treatment. Cells were washed once with 1× phosphate-buffered saline and resuspended in lysis buffer (Upstate Biotechnology). Cell lysates were sonicated (four times for 10 seconds on, 40 seconds off) at 30% power to shear DNA. Lysates were then diluted and precleared with protein G-Sepharose (Pierce Chemicals) previously blocked with salmon sperm DNA and rabbit IgG for 1 h at 4°C. Precleared lysates were incubated with antibodies to HA (Sigma), GABPα, GABPβ, p300, RXR, or normal rabbit IgG (Santa Cruz Biotechnology) overnight at 4°C, followed by the addition of salmon sperm-blocked protein G-Sepharose for 1 h at 4°C. Precipitated beads were washed once in a low-salt buffer, high-salt buffer, LiCl buffer, and Tris-EDTA. Precipitated complexes were eluted from the beads via a sodium dodecyl sulfate (SDS)-NaHCO₃ buffer containing proteinase K, and cross-links were reversed via 4-hour incubation at 68°C. DNA was then eluted using the QIAGEN PCR clean-up kit. One to 5 μl was used for PCR with primers toward the CD18 proximal promoter (5′-AGAGCTGCTGTAGAGCGGAGA-3′ and 5′-GAGTGCTTCCCTCCAAAAATC-3′), CD18 distal enhancer (5′-GAAGGCAGAGGTGGATGGA-3′ and 5′-GGACTCAGACACATTACTACCCTAAA-3′), CD18 promoter mid-region (5′-TCCCTTTCGTGTTCTGTGGT-3′ and 5′-GGAACAGGCAAGAAAACCTG-3′), or irrelevant DNA in CD18 exon 13 (5′-TCTGCTTCTGCGGGAAGTG-3′ and 5′-ATGTACTTGCCACAGGGTGA-3′).

Chromosome conformation capture.

A total of 4 × 10⁷ RA-treated (10⁻⁵ M for 24 h) U937 cells were diluted to 50 ml of RPMI and cross-linked at room temperature for 5 min with formaldehyde (final concentration, 2%). The cells were then quenched with 2 M glycine (final concentration, 0.125 M) and spun down for 15 min at 3,500 rpm. Following centrifugation, the cell pellet was resuspended in 50 ml lysis buffer (10 mM Tris HCl, pH 8.0, 10 mM NaCl, 0.2% NP-40 with protease inhibitors) and incubated on ice for 90 min with mixing. The nuclei were then spun down for 15 min at 2,500 rpm and resuspended in 2 ml of NEB buffer 2 (New England Biolabs) with 0.3% SDS and incubated for 1 h at 37°C with shaking. Next, 1 × 10⁶ nuclei (∼15 μg) were then digested overnight at 37°C with 600 U of Nsp1 (New England Biolabs). SDS (final concentration, 1.6%) was added, and the sample was incubated at 65°C for 20 min. Approximately 2 μg of the sample was diluted to 0.8 ml in ligation buffer (30 mM Tris HCl, pH 8.0, 10 mM dithiothreitol, and 1 mM ATP) plus Triton X-100 (final concentration, 1%) and incubated 1 h at 37°C. Ligation was then performed at 16°C using 30 Weiss units of T4 ligase (Invitrogen) for 4 h. Samples were then treated with proteinase K (final concentration, 100 μg/ml) and incubated overnight at 65°C to reverse cross-links. DNA was then ethanol precipitated, and serial dilutions were used for PCR using primers corresponding to the CD18 distal enhancer (5′-GAAGGCAGAGGTGGATGGA-3′) and proximal promoter (5′-TAGAAACCTCAGCTGGAGGC-3′).

RESULTS

GABPα and p300 physically interact in myeloid cells.

CD18 is transcriptionally activated by RA in myeloid cells (6). This response is mediated, in part, by the binding of RAR/RXR to a distal RARE that lies nearly 1 kb upstream of the transcriptional start site (6). However, we have previously shown that this distal RARE only accounts for one-half of the RA responsiveness of CD18. The remainder of RA responsiveness is mediated by the CD18 proximal promoter, which lacks an RARE and is not directly bound by RAR or RXR. CD18 RA responsiveness requires both ets- and Sp1-binding sites, which are bound by GABPα, PU.1, and Sp1 (Fig. 1) (6). Furthermore, the ets sites, alone, confer RA responsiveness on a heterologous promoter (6). The RA responsiveness of CD18 is further increased by the transcriptional coactivator p300. p300 interacts with numerous transcription factors, including both retinoid receptors and ets factors. Thus, we hypothesized that a physical interaction between GABPα and p300 mediates RA responsiveness of the CD18 proximal promoter. We have previously demonstrated that GABPα and p300 interact in RA-treated U937 myeloid cells (6). Therefore, in order to characterize the role of the GABPα-p300 interaction in RA responsiveness, we sought to determine if this interaction also occurs in untreated U937 cells or if it requires the presence of RA.

FIG. 1.

Schematic representation of transcription factors that bind and activate the CD18 promoter in myeloid cells. Numbers indicate base pairs upstream of the transcriptional start site.

Whole-cell extracts from U937 cells were immunoprecipitated with an antibody to GABPα or control IgG. Antiserum to GABPα immunoprecipitated both GABPα (Fig. 2A, lower panel, lane 2) and p300 (upper panel, lane 2); the negative control IgG did not precipitate either protein (lane 3). Therefore, GABPα and p300 physically interact in untreated U937 myeloid cells.

FIG. 2.

GABPα and p300 physically interact in myeloid cells; p300 interacts with the GABPα Pointed domain. (A) Whole-cell extracts of U937 cells were incubated with antiserum to GABPα (lane 2) or negative control IgG (lane 3), followed by incubation with protein G-Sepharose. Immunoprecipitated material and the equivalent of 10% input protein (lane 1) were analyzed by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) followed by immunoblotting for GABPα (bottom panel) or p300 (top panel). (B) Schematic representation of the GST-GABPα fusion proteins used for GST pull-down assays; numbers indicate amino acid residues. The black box represents the GST tag, the white box indicates the PNT protein interaction domain, and the gray box indicates the ets DNA-binding domain. (C) Whole-cell extracts of 293 cells transfected with pCMVβ-p300 were precleared with glutathione beads, followed by incubation with GST alone (lane 2), GST-GABPα (lane 3), GST-N1-208 (lane 4), or GST-C236-454 (lane 5). Precipitated material and 10% input protein (lane 1) were analyzed by SDS-PAGE and immunoblotted for p300 (upper panel). Input GST proteins were stained with Coomassie blue (lower panel). (D) Same experiment as in panel C, except GST-PNT (lane 4) and GST-ΔPNT(lane 5) were used for the pull-down assay. Asterisks indicate fusion proteins; smaller bands are degradation products of these proteins. Lane 6 is a shorter exposure of lane 1.

p300 interacts with the GABPα pointed domain.

We next sought to determine the domain(s) that mediates the GABPα-p300 interaction so that we could disrupt this interaction and determine its role in the RA responsiveness of the CD18 gene. We expressed GST fusion proteins of the GABPα N terminus (N 1-203) and C terminus (C 263-454) (shown schematically in Fig. 2B) in order to localize its interaction with p300. HEK 293 cells transfected with pCMVβ-p300 were used as the source of p300 protein. Incubation of the GST-GABPα constructs with 293 whole-cell extracts showed that the N terminus, but not the C terminus, of GABPα interacted with p300 (Fig. 2C).

The amino-terminal fragment of GABPα that interacts with p300 contains a pointed domain (PNT). The PNT domain consists of five α-helices that form a distinct protein-protein association domain (38). In other ets factors, the PNT domain mediates self-association (16) or other protein-protein interactions (33); to date, no binding partner for the GABPα PNT domain has been identified. Therefore, we expressed the GABPα PNT domain as a GST fusion protein to determine if it interacts with p300. As shown in Fig. 2D, the GABPα PNT domain, alone, interacts with p300 (lane 4). The physical interaction between p300 and the GABPα PNT occurs at levels comparable to the full-length protein (lane 3).

In order to determine if the PNT domain is the sole domain in the GABPα N terminus that is capable of binding p300, we created a GABPα construct that lacks the PNT domain (ΔPNT). This construct was also capable of interacting with p300 (Fig. 2D, lane 5), although the interaction was considerably weaker than the isolated PNT domain. This result indicates that residues within the first 143 amino acids of GABPα also mediate the interaction with p300.

GABPα physically interacts with p300 C/H domains.

The p300 protein contains three cysteine- and histidine-rich (C/H) regions that are involved in its physical interactions with other proteins, including multiple ets factors (4). We sought to determine which, if any, of the p300 C/H domains (illustrated schematically in Fig. 3A) mediate the physical interaction with GABPα. [³⁵S]methionine-labeled, in vitro-translated GABPα was incubated with GST fusion proteins that correspond to the p300 C/H domains (2). GABPα failed to interact with the C/H1 domain (Fig. 3B, lane 3), but it interacted with both the p300 C/H2 (lane 4) and C/H3 (lane 5) domains. We thus sought to better define the interactions between the p300 C/H2 and C/H3 domains and the GABPα N terminus and PNT domain. In vitro-translated GABPαN143 weakly interacted with both the C/H2 and C/H3 domains (Fig. 3C, top panel, lanes 3 and 4), while the PNT domain only interacted with the C/H3 domain (middle panel, lane 4). Thus, the GABPα PNT domain interacts with the p300 C/H3 domain, while the N terminus, which can weakly bind both the C/H2 and C/H3 domains, most likely interacts with the C/H2 domain in the context of the full-length proteins. The interaction with more than one domain of p300 is not unique to GABPα; ets-1 has been shown to interact with both the C/H1 and C/H3 domains (43). Because these interactions occurred between in vitro-translated GABPα and bacterially expressed, purified GST-p300 constructs, these results suggest that the physical interaction between GABPα and p300 constructs is direct and does not require other proteins. Thus, we reasoned that we could use these domains as dominant negative constructs to disrupt the GABPα/p300 complex and perturb its function.

FIG. 3.

GABPα interacts with p300 C/H domains. (A) Schematic representation of p300, from which GST-C/H domain fusion proteins were prepared for use in GST pull-down assays. Numbers indicate amino acid residues. (B) In vitro-translated [³⁵S]Met-labeled GABPα was precleared with glutathione beads followed by incubation with GST alone (lane 2), GST-300-528 (C/H1) (lane 3), GST-1162-1450 (C/H2) (lane 4), or GST-1570-1820 (C/H3) (lane 5). Precipitated material and 5% input protein (lane 1) were analyzed by SDS-polyacrylamide gel electrophoresis and autoradiography. Input GST proteins were stained with Coomassie blue (lower panel). (C) Same experiment as in panel B, except input material was in vitro-translated [³⁵S]Met-labeled N143 (upper panel) or PNT (middle panel), and only GST alone (lane 2) GST-1162-1450 (C/H2) (lane 3), and GST-1570-1820 (C/H3) (lane 4) were used.

The GABPα PNT domain disrupts the GABPα-p300 interaction and inhibits RA responsiveness of the CD18 proximal promoter.

We previously showed that p300 increases the RA responsiveness of the CD18 promoter in myeloid cells. We sought to determine if the interaction between p300 and GABPα (Fig. 2A) is required for the RA responsiveness of CD18. We cloned the GABPα PNT domain into the mammalian expression vector pCDNA-Flag to act as a dominant negative inhibitor of the interaction between GABPα and p300. U937 cells were transiently transfected with the dominant negative Flag-tagged PNT (Flag-PNT) construct and subjected to an immunoprecipitation for the p300 protein. In the absence of the Flag-PNT construct, p300 physically interacted with GABPα (Fig. 4A, lane 1). As increasing concentrations of the Flag-PNT protein were introduced, the amount of full-length GABPα associated with p300 was greatly diminished (Fig. 4A, lanes 2 and 3). Thus, the Flag-PNT protein acts as a dominant negative molecule that disrupts the GABPα-p300 interaction by competing with GABPα for p300.

FIG. 4.

The GABPα PNT domain inhibits RA responsiveness of the CD18 proximal promoter. (A) U937 cells that were stably transfected with CD18(−96)/luc were transiently transfected with the indicated quantities of the Flag-tagged PNT construct. Whole-cell extracts of these cells were incubated with antiserum to p300, followed by incubation with protein G-Sepharose. Immunoprecipitated material was analyzed by SDS-polyacrylamide gel electrophoresis followed by immunoblotting for Flag-PNT (bottom panel), GABPα (middle panel), or p300 (top panel). (B) U937 cells that were stably transfected with CD18(−96)/luc were transiently transfected with the indicated quantities of the Flag-tagged PNT construct. Transfected cells were divided into two equal aliquots, which were treated with 10⁻⁵ M RA or DMSO, and RA activation was measured 24 h later. All data represent at least three independent transient transfections. (C) Same experiment as in panel B, except cells were transfected with p300, Flag-PNT, or p300 and Flag PNT.

In order to determine the effect of disrupting the GABPα-p300 interaction on the RA responsiveness of the CD18 proximal promoter, we used U937 cells that were stably transfected with the CD18 proximal promoter upstream of a luciferase reporter [CD18 (−96)/luc)]. These cells were transiently transfected with the Flag-PNT construct followed by treatment with 10⁻⁵ M RA or DMSO vehicle for 24 h. CD18 proximal promoter activity increased fourfold following RA treatment (Fig. 4B). Mock transfection or transfection of vector DNA alone had no effect on the RA responsiveness of the CD18 proximal promoter. Transfection of Flag-PNT decreased the RA responsiveness of the CD18 proximal promoter by one-half (Fig. 4B), a degree of inhibition that is comparable to the deletion of the CD18 proximal promoter ets sites (6).

In order to confirm that the Flag-PNT protein reduces CD18 RA responsiveness by competing for p300, we overexpressed p300 in the presence of Flag-PNT. While RA responsiveness of the CD18 proximal promoter was nearly doubled by overexpression of p300 alone, cotransfection of p300 and Flag-PNT restored RA responsiveness to a level comparable to untransfected cells (Fig. 4C). Thus, the presence of excess p300 rescued the inhibition of CD18 RA responsiveness caused by Flag-PNT. Together, these results indicate that p300 is required for full RA responsiveness of the CD18 proximal promoter, and we conclude the physical interaction of GABPα and p300 in U937 cells is required for full RA responsiveness of CD18 in myeloid cells.

GABP and p300 recruit RAR/RXR from the distal enhancer to create a retinoic acid-induced enhanceosome.

In order to understand how the physical interaction between p300 and GABP affects RA responsiveness, we sought to determine how these proteins interact with the CD18 promoter in the absence and presence of RA. We envisioned two possible models for the mechanism of CD18 RA responsiveness: (i) independent recruitment of p300 to both the CD18 proximal promoter and distal enhancer (Fig. 5A), or (ii) formation of an RA-induced enhanceosome in which p300 physically links the CD18 proximal promoter and the distal enhancer (Fig. 5B). For both models we expect that in the absence of RA, GABP and p300 will be present at the CD18 proximal promoter. In the presence of RA, ligand binding by RAR/RXR creates a conformational change that allows for interaction with p300; thus, in both models p300 should be bound to the distal element following RA treatment. According to the first model, the RA-dependent interaction between RAR/RXR and p300 would act independently at both the proximal promoter and distal enhancer to effect RA responsiveness (Fig. 5A). In the second model, p300 at the proximal promoter would recruit RAR/RXR already present at the distal enhancer, thereby bridging these two elements to create an RA-induced enhanceosome on the CD18 promoter (Fig. 5B).

FIG. 5.

Potential mechanisms for the role of p300 in the RA responsiveness of the CD18 promoter. (A) p300 is independently recruited to the CD18 proximal promoter and distal enhancer, mediating the RA responsiveness of the proximal promoter by recruiting the RAR/RXR heterodimer. (B) p300 at the proximal promoter recruits the RAR/RXR heterodimer at the distal enhancer, thus acting as a platform for the creation of a myeloid enhanceosome that mediates RA responsiveness of the CD18 gene.

We performed ChIP to identify the proteins that bind to the CD18 promoter and to thereby test these two models. The primer pairs used for ChIP are indicated in Fig. 6A. In the absence of RA treatment, GABPα and GABPβ are bound to the CD18 proximal promoter (Fig. 6B, lanes 3 and 4, row 1) but not to the distal enhancer (row 2). Conversely, RXR is present at the distal element (lane 6, row 2) but not at the proximal element (lane 6, row 1). p300 is also bound to the CD18 proximal promoter in the absence of RA (lane 5, row 1) due to its interaction with GABPα. As expected, p300 was not found at the distal element in the absence of RA (lane 5, row 2).

FIG. 6.

p300 and GABP are components of an RA-induced enhanceosome on the CD18 promoter. (A) Schematic of primers used for ChIP of the CD18 promoter. (B) ChIP was performed on untreated U937 cells using antibodies to GABPα (lane 3), GABPβ (lane 4), p300 (lane 5), RXR (lane 6), or IgG (lane 7). PCR of immunoprecipitated material was performed using primers that correspond to the CD18 proximal promoter (open arrows; row 1), distal promoter (hatched arrows; row 2), mid-region (black arrows; row 3), and control primers located with an exon of CD18 (dotted arrows; row 4). Lane 1 is a no-DNA PCR control, and lane 2 is input DNA. (C) Same experiment as in panel B, except ChIP was performed on U937 cells 24 h following RA treatment.

The interactions at the CD18 promoter undergo a significant change following treatment with RA. Strikingly, RXR is now associated with the proximal promoter (Fig. 6C, lane 6, row 1). PCR of the region between the CD18 proximal and distal elements indicated that the presence of RXR at the proximal promoter is not due to incomplete shearing of the DNA (lane 6, row 3). The presence of RXR at the proximal promoter is consistent with either of the two models (Fig. 5). Thus, it is the changes seen at the distal element that are most informative for distinguishing between these models. In the presence of RA, GABPα, GABPβ, and p300 are now bound to the distal element (lanes 3 to 5, row 2). These results suggest that p300 (via its interactions with both RXR and GABPα) bridges the CD18 proximal promoter and distal enhancer to form an enhanceosome complex induced by RA.

3C confirms the CD18 enhanceosome.

In order to confirm that RA treatment of U937 cells leads to the formation of an enhanceosome at the CD18 promoter, we performed the 3C assay (11). Untreated and RA-treated U937 cells were treated with formaldehyde to link DNA and proteins. These nuclei were digested with Nsp1, which cuts twice between the CD18 proximal promoter and distal enhancer, thereby removing a 300-bp fragment. DNA ligation was performed at a low concentration of chromatin in order to favor intramolecular ligation. The cross-links were then reversed, and PCR was performed with a 5′ primer that corresponds to the CD18 distal enhancer and a 3′ primer that corresponds to the CD18 proximal promoter. If an RA-induced enhanceosome forms at the CD18 promoter that bridges these two regions, this complex will keep the Nsp1 ends in close proximity and allow intramolecular ligation. However, if two distinct complexes are formed, the Nsp1 ends will no longer be in proximity and therefore will not be ligated. Thus, following complete Nsp1 digestion and ligation, if an RA-induced CD18 enhanceosome has formed, PCR should produce a unique 600-bp product that is distinct from the 900-bp product that results from PCR of undigested chromatin.

As predicted, the 3C assay demonstrated that, in the absence of RA treatment, PCR performed on undigested, digested, or digested/ligated DNA produced a 900-bp product that corresponds to the full-length promoter (Fig. 7A). The presence of this band in digested and digested/ligated material indicates the existence of residual, undigested chromatin. Upon RA treatment, the 3C assay demonstrated a significant change in the conformation of the CD18 promoter. While PCR of undigested and digested material only produced the 900-bp band, the digested/ligated material generated a novel 600-bp band. This product represents loss of the 300-bp fragment between the Nsp1 sites and religation of the proximal promoter and distal enhancer that remain in close proximity due to the bridging enhanceosome proteins (Fig. 7). Thus, the 3C assay confirms the presence of an RA-induced enhanceosome at the CD18 promoter.

FIG. 7.

Chromosome conformation capture confirms the RA-induced CD18 enhanceosome. (A) Untreated or RA-treated U937 cells were cross-linked with 2% formaldehyde, and isolated nuclei were digested with Nsp1 overnight, followed by ligation with T4 ligase. The cross-links were then reversed, and PCR was performed on serial dilutions of undigested (top panel), digested (middle panel), or digested/ligated (bottom panel) DNA using primers that correspond to the full-length CD18 promoter. The first lane of each panel (φ) is a no-DNA PCR control. (B) Schematic of the chromosome conformation capture assay in the presence of the CD18 enhanceosome.

Dominant negative p300 disrupts the CD18 enhanceosome and reduces RA responsiveness.

In order to confirm the role of the GABPα-p300 interaction in the RA-induced enhanceosome and to determine the effect of disrupting this complex on RA responsiveness of CD18, we created a dominant negative p300 construct (Fig. 8A). HA-p300N consists of only the first 595 amino acids of p300; thus, this construct contains the residues required for binding RXR but not the C/H2 and C/H3 domains that bind to GABPα (Fig. 3). Expression of the HA-p300N protein in transfected U937 cells is shown in Fig. 8B.

FIG. 8.

Dominant negative p300 disrupts the CD18 enhanceosome and reduces RA responsiveness. (A) Schematic representation of the HA-p300N construct (circled region), which consists of amino acids 1 to 595 of p300 and therefore can interact with RXR but not GABPα. (B) Western blot of untransfected U937 cells (lane 1) or U937 cells transfected with an increasing concentration of the HA-p300N terminus construct (lanes 2 and 3), probed with anti-HA, indicating expression of the construct. (B) Following 24-h RA treatment, ChIP was performed on U937 cells transfected with HA-p300N, using antibodies to GABPα (lane 3), GABPβ (lane 4), p300 (lane 5), HA (lane 6), RXR (lane 7), and IgG (lane 8). PCR was performed using the same primers as in Fig. 7. (C) U937 cells that were stably transfected with CD18(−918)/luc were transiently transfected with the indicated quantities of the dominant negative HA-p300N construct. Transfected cells were divided into two equal aliquots, which were treated with 10⁻⁵ M RA or DMSO, and RA activation was measured 24 h later. (D) Same experiment as in panel C, except U937 cells stably transfected with the CD18(−96)/luc were used.

We performed ChIP on these cells 24 h after RA treatment and, as predicted, ChIP using an HA antibody indicated the presence of HA-p300N at the CD18 distal enhancer (Fig. 8C, lane 6, row 2) but not at the proximal promoter (row 1). The presence of this protein blocks native, full-length p300 from binding the distal enhancer (lane 5, row 2). GABPα and -β and p300 were still present at the proximal promoter; however, the presence of HA-p300N blocked their ability to recruit the distal element (lanes 3 to 5, rows 1 and 2). Similarly, RXR at the distal element, bound by HA-p300N, was incapable of recruiting the proximal element (lane 7, rows 1 and 2). Thus, dominant negative p300 effectively disrupts enhanceosome formation at the CD18 promoter.

We also examined the effects of HA-p300N on RA-induced expression of CD18. U937 cells stably transfected with CD18(−918)/luc (the CD18 full-length promoter upstream of a luciferase reporter) were transiently transfected with HA-p300N, followed by treatment with 10⁻⁵ M RA or DMSO vehicle for 24 h. Mock-transfected cells and cells transfected with the empty vector showed a nearly fivefold increase in gene expression in response to RA (Fig. 8D). Transfection of HA-p300N decreased the RA responsiveness of the CD18 promoter by more than one-half (Fig. 8D). Thus, disruption of the CD18 enhanceosome prevents full RA responsiveness of the CD18 gene.

We also sought to determine if the HA-p300N construct disrupts RA-induced expression of the CD18 proximal promoter alone. Similar to the full-length promoter, transfection of HA-p300N reduced by half the RA-induced expression of the CD18 proximal promoter (Fig. 8E). This response is remarkably similar to (i) the effects of ets site deletions (6) and (ii) the effect of transfection with Flag-PNT. Because there is no recognizable RARE in this construct, we propose that this effect is due to the HA-p300N protein binding to non-DNA-bound RAR/RXR and preventing its recruitment to the CD18 proximal promoter. Our results demonstrate that RA induces an enhanceosome on the full-length CD18 promoter in the presence of RA, but the data in Fig. 8E suggest that the CD18 proximal promoter is capable of recruiting non-DNA-bound RXR to effect RA-induced gene expression in the absence of a local RARE.

DISCUSSION

RA signaling via the activation of RARs is an important mediator of normal myeloid differentiation. Furthermore, chromosomal translocations that disrupt RARα are present in nearly all cases of acute promyelocytic leukemia (13). Thus, RA is a powerful regulator of normal myelopoiesis, and abnormalities of RARs contribute importantly to aberrant myeloid differentiation. However, it is increasingly evident that RA responsiveness is more complex than the conventional model of direct signaling exclusively through RARs. The RA responsiveness of several genes requires transcription factors other than the RARs (6, 14, 18, 39).

We previously showed that intact ets and Sp1-binding sites are required for the transcriptional regulation of CD18 by RA in myeloid cells and that the CD18 ets sites alone are sufficient to confer RA responsiveness on a heterologous promoter. We also demonstrated that the transcriptional coactivator p300 contributes to this RA responsiveness (6). We now demonstrate that a direct physical interaction between GABPα and p300, mediated by the GABPα PNT domain and C/H domains of p300, is required for RA responsiveness of CD18. Furthermore, we show that a dominant negative form of GABPα that interferes with this interaction reduces the RA responsiveness of the CD18 proximal promoter. Finally, we have shown that, in the presence of RA, p300 interacts with GABPα to form an enhanceosome at the CD18 promoter. Furthermore, a dominant negative p300 molecule, which can bind RAR/RXR but not GABPα, disrupts this complex and reduces CD18 RA responsiveness.

This study has focused on the role of GABP in mediating the RA responsiveness of the CD18 ets sites in myeloid cells. To this end we have demonstrated that GABPα physically interacts with the coactivator p300 in myeloid cells (Fig. 2). Furthermore, we have shown that the GABPα-p300 interaction directly involves the GABPα PNT domain (Fig. 2 and 3) and that this domain is capable of disrupting RA responsiveness (Fig. 4). The PNT motif, which can mediate both homodimerization and heterodimerization, was originally characterized in the Drosophila melanogaster ets protein Pointed (30, 38). Based on sequence homology to this ∼80-amino-acid region, PNT domains were identified in approximately one-third of ets factors (30). The GABPα PNT domain was defined by sequence homology with other ets proteins, but no interaction partners or functional role had previously been defined for this domain. Importantly, the PNT domains of various ets factors cannot be considered interchangeable or functionally equivalent. For example, the Ets-1 PNT domain interacts with ERK2, which leads to phosphorylation of the N terminus of Ets-1 (32), and recent work by Foulds et al. demonstrated that phosphorylation of Ets-1 and Ets-2 enhances their interaction with p300 via the PNT domain (12). However, the PNT domain of GABPα was unable to mediate an interaction with ERK2 when substituted for the corresponding Ets-1 PNT domain (32). Also, GABPα is unable to form homodimers in solution, suggesting that the GABPα PNT domain cannot self-associate, as is seen with the TEL PNT domain (9). Furthermore, a derivative of the leukemogenic TEL-PDGFR fusion protein, in which the TEL PNT domain was replaced by that of GABPα, failed to coimmunoprecipitate GABPα (16). Thus, the physical interaction between GABPα and p300 demonstrates the first known binding partner of the GABPα PNT domain.

We have also demonstrated an important role for the interaction between p300 and the GABPα PNT domain in the RA responsiveness of CD18. Inhibition of the GABPα-p300 interaction by expression of its PNT domain, alone, decreased the RA responsiveness of the CD18 proximal promoter by one-half (Fig. 4), which is comparable to the decrease in RA responsiveness when the CD18 ets sites are disrupted (6). This inhibition was rescued by overexpression of p300 (Fig. 4). This result indicates that the role of the CD18 ets sites in RA responsiveness is to recruit the coactivator p300 via a physical interaction with GABPα.

Discovering the importance of the p300-GABPα interaction leads to provocative new models for the mechanism of CD18 RA responsiveness in myeloid cells. How might the recruitment of p300 to the CD18 proximal promoter by GABPα mediate RA responsiveness? The simplest model would involve an RA-dependent interaction between GABPα and p300. However, p300 and GABPα interact in U937 cells even in the absence of RA (Fig. 2). Also, we demonstrated via ChIP that p300 is already present at the CD18 promoter in the absence of RA (Fig. 6).

We have also demonstrated that p300 interacts with GABPα via the C/H2 and C/H3 domains (Fig. 3B), while the retinoid receptors are known to interact with the N terminus of p300 (7). Therefore, the C/H2 and C/H3 domains could tether p300 to GABPα at the CD18 promoter, while leaving the N terminus free to recruit RAR/RXR. Based on these interactions, we proposed two potential models for the role of the GABPα-p300 interaction in the RA-induced activation of CD18 (Fig. 5). p300 could independently recruit a coactivator complex, including RAR/RXR, to the proximal promoter, thereby creating RA-induced activation distinct from the distal element (Fig. 5A). Alternatively, p300 could recruit the RAR/RXR heterodimer bound to the distal RARE to the proximal promoter, thereby creating a larger RA-dependent coactivation complex, an enhanceosome (Fig. 5B).

An enhanceosome has been broadly defined as multiple cis elements brought into close association by an assortment of trans factors (19). These factors can include site-specific, DNA-binding transcription factors as well as transcriptional coactivators (1). The association of these factors is proposed to create synergy by establishing a distinct activation surface that more efficiently recruits the basal transcription machinery (19). A central role for p300 as a scaffold has been described for the two best-understood enhanceosome complexes: the virus-induced beta interferon enhanceosome and enhanceosomes associated with the tumor necrosis factor alpha promoter (20, 41). Thus, it is enticing to hypothesize that p300 functions as a scaffold in an RA-induced enhanceosome at the CD18 promoter.

We used ChIP analysis in the presence of RA to determine if an RA-induced enhanceosome is present at the CD18 promoter. The presence of GABP at the CD18 distal enhancer following RA treatment, but not in its absence, clearly demonstrated the formation of an RA-induced enhanceosome (Fig. 6). The 3C assay independently confirmed the formation of the RA-induced CD18 enhanceosome (Fig. 7). This assay was recently used to demonstrate the association of distant elements, such as regions of the β-globin locus (42). In the present setting, the 3C assay was useful to demonstrate the protein-dependent association of relatively nearby DNA regions. Together, the ChIP data and 3C assay clearly demonstrate the first report of an RA-induced enhanceosome.

The ChIP and 3C data suggest that it is p300 that acts as a scaffold to form the RA-induced enhanceosome at the CD18 promoter. This was confirmed by the use of a dominant negative p300 construct that can bind to RXR but not to GABPα, which blocked the RA-dependent protein interactions between the CD18 proximal promoter and distal element (Fig. 8). Furthermore, the dominant negative HA-p300N reduced RA responsiveness of the CD18 gene. These results confirm that p300 does, in fact, bridge the two CD18 regulatory elements to mediate RA responsiveness. We also found that when the distal RARE was absent, the HA-p300N construct reduced RA responsiveness of the CD18 proximal promoter alone (Fig. 8E). This suggests that GABPα/p300 can also recruit non-DNA-bound RAR/RXR.

RA responsiveness is clearly not solely dependent on the retinoid receptors. Along with CD18, the RA responsiveness of another myeloid gene, folate receptor β, was shown to be dependent on GABP and Sp1 (14, 29). For that gene, it appears that RARs can be recruited to the promoter by Sp1-binding sites (14, 29), and Sp1 has previously been shown to physically interact with RAR and RXR. Therefore, it is thought that Sp1 recruits retinoid receptors during the RA-induced transcription of folate receptor β and other nonmyeloid promoters (6, 18, 35, 39). However, the role of the GABP-bound ets-binding sites in the folate receptor β promoter was not fully examined (14, 29). In this work, we have demonstrated the important role of GABP and its interaction with p300 in mediating the RA responsiveness of the CD18 ets sites.

We conclude that the RA responsiveness of CD18 in myeloid cells is dependent on the physical interaction between GABP and p300 and, together, these factors are critical for the formation of an RA-induced enhanceosome. The CD18 enhanceosome represents a novel mechanism for the RA-induced differentiation of myeloid cells; therefore, it will be important to determine if similar complexes form at the promoters of RA-induced genes as well. This further implies that this enhanceosome contributes to normal development of myeloid cells and may contribute to the aberrant differentiation associated with acute myelogenous leukemia.