Identification of a HoxA10 Activation Domain Necessary for Transcription of the Gene Encoding β3 Integrin during Myeloid Differentiation*

Transcription of the ITGB3 gene, which encodes β3 integrin, increases during myeloid differentiation. αvβ3 integrin mediates adhesion to fibronectin or vitronectin and regulates various aspects of the inflammatory response in mature phagocytes. In these studies, we found that the homeodomain transcription factor HoxA10 interacted with a specific ITGB3 cis element and activated transcription of this gene during myeloid differentiation. We also found that increased fibronectin adhesion in differentiating myeloid cells was dependent upon this HoxA10-induced increase in β3 integrin expression. We determined that activation of ITGB3 transcription required a HoxA10 domain that was not identical to the “hexapeptide” that mediates interaction of Hox and Pbx proteins. This activation domain was also not identical to a previously identified HoxA10 repression domain that mediates interaction with transcriptional co-repressors. Instead, this HoxA10 activation domain had homology to “PQ” protein-protein interaction domains that have been described previously in other transcription factors. Consistent with this, we found that the HoxA10 PQ-like domain recruited the CREB-binding protein (CBP) to the ITGB3 promoter. This was associated with an increase in local histone acetylation in vivo. In immature myeloid cells, we previously determined that HoxA10 repressed transcription of the CYBB and NCF2 genes, which encode the phagocyte oxidase proteins gp91PHOX and p67PHOX, respectively. Therefore, our studies indicated that HoxA10 either activates or represses gene transcription at various points during myelopoiesis. Our studies also suggested that HoxA10 is a bifunctional protein that is involved in dynamic regulation of multiple aspects of phagocyte phenotype and function.

Transcription of the ITGB3 gene, which encodes ␤3 integrin, increases during myeloid differentiation. ␣v␤3 integrin mediates adhesion to fibronectin or vitronectin and regulates various aspects of the inflammatory response in mature phagocytes. In these studies, we found that the homeodomain transcription factor HoxA10 interacted with a specific ITGB3 cis element and activated transcription of this gene during myeloid differentiation. We also found that increased fibronectin adhesion in differentiating myeloid cells was dependent upon this HoxA10induced increase in ␤3 integrin expression. We determined that activation of ITGB3 transcription required a HoxA10 domain that was not identical to the "hexapeptide" that mediates interaction of Hox and Pbx proteins. This activation domain was also not identical to a previously identified HoxA10 repression domain that mediates interaction with transcriptional co-repressors. Instead, this HoxA10 activation domain had homology to "PQ" protein-protein interaction domains that have been described previously in other transcription factors. Consistent with this, we found that the HoxA10 PQ-like domain recruited the CREB-binding protein (CBP) to the ITGB3 promoter. This was associated with an increase in local histone acetylation in vivo. In immature myeloid cells, we previously determined that HoxA10 repressed transcription of the CYBB and NCF2 genes, which encode the phagocyte oxidase proteins gp91 PHOX and p67 PHOX , respectively. Therefore, our studies indicated that HoxA10 either activates or represses gene transcription at various points during myelopoiesis. Our studies also suggested that HoxA10 is a bifunctional protein that is involved in dynamic regulation of multiple aspects of phagocyte phenotype and function.
␤3 integrin, which is encoded by the ITGB3 gene, is expressed in a variety of cell types. ITGB3 transcription occurs in hematopoietic stem cells (HSC) 2 and in various blood cell lineages (1). In cells of the myeloid lineage, ␤3 integrin participates in ␣v␤3 heterodimers that mediate cell adhesion to fibronectin and vitronectin (2,3). Fibronectin is a component of the bone marrow extracellular matrix, and fibronectin binding promotes survival of HSC and differentiating myeloid progenitor cells (4,5). Expression of ␤3 integrin increases during differentiation of neutrophils and monocytes (1). In mature monocytes, ␣v␤3 mediates phagocytosis of apoptotic neutrophils, thereby contributing to down-regulation of the inflammatory response (6,7). In neutrophils, ␣v␤3 provides a "braking" mechanism during transmigration from the vascular space and is also involved in activation of the phagocyte-specific respiratory burst oxidase (8,9). In addition to participation in the innate immune response, ␣v␤3-ligand binding activates Syk protein-tyrosine kinase and may thereby be involved in generation of cell survival signals in phagocytic cells (8).
These results suggest that understanding the mechanisms that regulate ␣v␤3 expression would be relevant to understanding phagocytic cell function and survival. However, regulation of ITGB3 transcription during myelopoiesis or in mature myeloid cells has not been studied. Indeed, the only previous studies of the ITGB3 promoter were performed in epithelial carcinoma cells (10). These studies identified a region of the ITGB3 promoter that interacts in vitro with the homeodomain transcription factor HoxA10. In these studies, overexpressed HoxA10 activated an artificial promoter construct with this region of the ITGB3 promoter in transfection experiments with a breast cancer cell line (10). Because HoxA10 is expressed in differentiating myeloid cells, the first hypothesis of our current studies was that HoxA10 regulates ITGB3 transcription during myelopoiesis.
The HOX genes are arranged in four paralog groups (A-D) on four different chromosomes in mouse and man. During embryogenesis, HOX gene transcription occurs 3Ј to 5Ј through each paralog group with the more 3Ј genes expressed in cephalad and the more 5Ј genes in caudal organs (11). HOX genes are also sequentially expressed during definitive hematopoiesis with the 3Ј-most genes being expressed in HSCs and 5Ј-most genes in more differentiated hematopoietic cells (12). Consistent with this, ABD HOXA genes (HoxA7-11) are maximally expressed in committed myeloid progenitors. Abnormal expression of various Hox proteins is associated with acute myeloid leukemia, implying that tight regulation of the Abd HoxA proteins is important for normal myelopoiesis (13). Additionally, our previous studies indicated that Abd HoxA proteins are involved in regulating transcription of genes involved in phagocyte-specific functional activities (14,15). Therefore, the second hypothesis of these studies was that dynamic regulation of ITGB3 transcription by HoxA10 contributes to phagocyte function and survival during myelopoiesis. In our previous studies, we found that HoxA10 dynamically regulates transcription of the genes encoding the phagocyte oxidase proteins gp91 PHOX and p67 PHOX (the CYBB and NCF2 genes, respectively) (14,15). These genes are expressed in myeloid cells after the promyelocyte stage of differentiation and are actively transcribed until cell death (16,17). We determined that HoxA10 represses CYBB and NCF2 transcription in undifferentiated myeloid cells. In these cells, HoxA10 binds to homologous CYBB and NCF2 cis elements as a heterodimer with the homeodomain protein Pbx1 (14,15).
Hox and Pbx proteins are frequent DNA-binding partners, and transcriptional regulation by some such heterodimers is mediated by Pbx interaction with transcriptional co-activators or co-repressors (18). However, we found that HoxA10-repression of CYBB and NCF2 transcription was Pbx-independent and involved direct interaction of HoxA10 with the corepressor histone deacetylase 2 (HDAC2) (15,19). This interaction required a unique HoxA10 domain, not conserved in other Abd HoxA proteins (19). As differentiation proceeds, HoxA10 binding to the CYBB and NCF2 genes decreases, relieving repression (14,15). Because gp91 PHOX and p67 PHOX are the rate-limiting phagocyte oxidase components, this results in differentiation stage-specific acquisition of respiratory burst competence in mature phagocytes. If HoxA10 also regulates ITGB3 transcription in a differentiation stage-specific manner, this would further suggest that HoxA10 is an important regulator of phagocyte function.
Target genes that are activated by HoxA10 in myeloid cells have been not been previously identified. However, based on previous studies, HoxA10 would be anticipated to activate ITGB3 transcription during myelopoiesis. Therefore, the first goal of these studies was to determine whether HoxA10 increases ␤3 integrin expression and ITGB3-transcription in differentiating myeloid cells. The second goal of these studies was to determine the functional relevance of Hox10 regulation of ␤3 integrin expression for differentiation-induced phagocytic adhesion. The third goal of these studies was to determine whether HoxA10 activation of ITGB3 transcription requires a Pbx partner. If transcriptional activation is not Pbx-dependent, ITGB3 transcription will be used as a model to identify a HoxA10 activation domain. Therefore, these studies address mechanisms regulating phagocyte function and also fundamental mechanisms of HoxA10 transcription factor activity.

Plasmids and PCR Mutagenesis
Artificial Promoter Constructs-Artificial promoter/reporter constructs were generated as described previously (14,15), in the minimal promoter/reporter vector, p-TATACAT (20) (obtained from Dr. A. Kraft, Hollings Cancer Center at the Medical University of South Carolina, Charleston). Constructs were generated with four copies (in the forward direction) of the Ϫ1973 to Ϫ1933 bp sequence from the ITGB3 promoter (p-B3ITATACAT) (10,21).
cDNA Sequences and Mutagenesis-The cDNA for human HoxA10 was obtained from C. Largman (University of California, San Francisco) (22). This cDNA sequence represents the major transcript in mammalian hematopoietic cells, encoding a 393-amino acid, 55-kDa protein (22). Wild type HoxA10 cDNA sequence was subcloned into the pSR␣ vector for expression in mammalian cells (23). Wild type HoxA10 was also subcloned into the pM2 vector, for expression in mammalian cells as a fusion protein with the DNA-binding domain of the yeast transcription factor GAL4, as described previously (14).
HoxA10 5Ј truncation mutant cDNAs were generated by PCR, using primers that incorporate the genuine HoxA10 Kozak consensus sequence and an ATG. Truncation mutants were generated that included HoxA10 amino acids (aa) 1-219, 1-179, 1-112, 102-146, 124 -146, and 145-219. These sequences were subcloned into the PM2 vector for expression as a fusion protein with the DNA-binding domain of the yeast Gal4 transcription factor. HoxA10 truncations were also generated representing amino acids 219 -403, 146 -403, and 60 -403. These sequences were subcloned into the pSR␣ vector for expression in mammalian cells. HoxA10 cDNAs with mutation of the Pbx1 interaction domain were generated by PCRbased site-directed mutagenesis, by a previously described technique (19). The wild type HoxA10 was mutated to change amino acid 312 from asparagine to alanine and amino acid 313 from tryptophan to threonine (N312A/W323T HoxA10). This mutant cDNA was subcloned into a pSR␣ vector for expression in U937 transfection experiments. All mutant cDNA sequences generated by PCR were completely sequenced on both strands to verify that no unintended mutations had been introduced.
Plasmids for shRNA Expression-Plasmids were generated to express shRNA to human ␤3 integrin or HoxA10 using the pLKO.1 vector (kindly provided by Dr. K. Rundell, Northwestern University, Chicago). Oligonucleotide sequences for ␤3 integrin-specific shRNA, HoxA10-specific shRNA, or scrambled control shRNAs were designed with the assistance of the software on the Promega web site (Promega, Madison, WI).

Oligonucleotides
Oligonucleotides were synthesized by the Core Facility of the Robert H. Lurie Comprehensive Cancer Center at Northwestern University. Double-stranded, synthetic oligonucleotides were generated representing the Ϫ1973 to Ϫ1933 bp sequence of the ITGB3 promoter (5Ј-GGGGGGCTTATAATGT-TATTTTTAGTTTACAGGTTCTTAC-3Ј) or the Ϫ94 to Ϫ134 bp sequence of the CYBB promoter, 5Ј-TTCAGTTGAC-CAATGATTATTAGCCAATTTTCTGATAAAA-3Ј. In these oligonucleotides, the HoxA10 core is in boldface type; the Pbx core is in italic type, and the ccaat boxes are underlined. In some experiments, an irrelevant control sequence from the NF1 promoter was also used.

Myeloid Cell Line Culture
The human myelomonocytic cell line U937 (24) was obtained from Andrew Kraft (Hollings Cancer Center at the Medical University of South Carolina, Charleston). Cells were maintained and differentiated as described (14,15,19). U937 cells were treated with 500 units per ml of human recombinant IFN␥ for 24 or 48 h, as indicated (Hoffmann-La Roche).

Murine Bone Marrow Cell Culture
Bone marrow was harvested from HoxA10 ϩ/Ϫ mice or wild type litter mates, and myeloid progenitor cells were cultured in GM-CSF (20 ng/ml), IL3 (10 ng/ml), and stem cell factor (10 ng/ml) for 48 h. These cells were differentiated ex vivo in M-CSF (20 ng/ml) for 72 h. Animal experiments were performed according to a protocol approved by the Northwestern University Animal Care and Use Committee.

Electrophoretic Mobility Shift Assays (EMSA)
Nuclear extract proteins were prepared by the method of Dignam et al. (25), with protease inhibitors as described previously (14). Oligonucleotides probes were prepared; EMSA and antibody supershift assays were performed, as described (14,15). Antiserum to HoxA10 (not cross-reactive with other Hox proteins) was obtained from Covance Research Products (Richmond, CA) and from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodies to Pbx1 and Pbx2 (noncross-reactive with each other or other Pbx proteins) were obtained from Santa Cruz Biotechnology. All EMSA were performed several times with at least two different batches of nuclear proteins, and representative data are shown.
In other experiments, cells were transfected with 30 g of p-(GAL4)5TKCAT and 60 g of a vector to expression HoxA10 or various mutants as fusion proteins with the DNA-binding domain of the yeast GAL4 transcription factor. In some experiments, cells were also co-transfected with a vector to express the E1a viral oncoprotein. The amount of GAL4 fusion protein expression vector used in these experiments was determined in preliminary studies in which the various HoxA10 expression plasmids were titrated to determine, for each construct, the amount of plasmid for maximal repression activity (10, 30, 60, and 120 g). For all constructs, maximal repression activity was obtained with 60 g of expression vector, consistent with our previous investigations with this assay (19).
Stable Transfectants-Stable U937 transfectants were generated with a vector to overexpress HoxA10 or empty vector control. Stable transfectants were selected in G418 or puromycin as described previously (14). At least three transfectant pools were tested for each construct. In some experiments, U937 cells were co-transfected with vectors to express shRNA for ␤3 integrin or scrambled shRNA control.

Immunoprecipitation and Western Blotting
Western Blots-Total cell lysate proteins from U937 cells or various U937 stable transfectants (50 g) were separated by SDS-PAGE and transferred to nitrocellulose. Western blots were serially probed with antibodies to various proteins. Antibody to ␤3 integrin was obtained from Chemicon-Millipore (Billericon, MA). Antibodies to tubulin, HoxA10, Syk, and phospho-Syk were obtained from Santa Cruz Biotechnology. Each experiment was performed at least three times, and representative blots are shown.
Co-immunoprecipitation Experiments-In other experiments, lysate proteins (1.5 mg) from U937 transfectants with 124 -146 aa of HoxA10-Gal4 or Gal4-DBD vector control were immunoprecipitated with anti-GAL4 DNA-binding domain antibody (rabbit polyclonal from Santa Cruz Biotechnology) or irrelevant control antibody (rabbit anti-mouse IgG from Santa Cruz Biotechnology), as described above. Immunoprecipitated proteins were analyzed by Western blots and sequentially probed with antibodies to CBP or the GAL4 DNA-binding domain (all from Santa Cruz Biotechnology). This experiment was repeated twice, and representative blots are shown.

Quantitative Real Time PCR
RNA was isolated from U937 cells using the TRIzol reagent, according to manufacture's instructions (Invitrogen). RNA was tested by denaturing gel electrophoresis to determine the integrity of the 18 S and 28 S ribosomal bands. cDNA was generated using standard techniques. Primers were designed with the software from Integrated DNA Technologies, and real time PCR was performed using SYBR green according to the "standard curve" method. Result were normalized to 18 S and actin to control for RNA abundance in various samples. In other experiments, real time PCR was performed on chromatin that coimmunoprecipitated from U937 cells with an antibody to HoxA10, the CREB-binding protein (CBP), pCAF, p300, acetylhistone 2A, or irrelevant control. Co-precipitating chromatin was amplified with PCR primers flanking the HoxA10-binding cis element in the ITGB3 promoter. Un-precipitated chromatin was a control to normalize for DNA abundance in the various samples.

Cell Adhesion Assays
Tissue culture dishes were coated overnight at 4°C with 20 g/ml fibronectin. U937 cells were incubated with or without IFN␥ for 48 h, counted, and transferred to fibronectin precoated tissue culture dishes. Cells were incubated for 16 h at 37°C, 5% CO 2 . In some experiments, cells were preincubated with 20 g/ml of antibody to ␤1 integrin, ␤3 integrin, or irrelevant control antibody (Chemicon, Temecula, CA) prior to plating on fibronectin. The dishes were washed with phos-phate-buffered saline, and adherent cells were fixed for 40 min in 3.7% formaldehyde at room temperature. Fixed cells were stained for 40 min at room temperature with 0.1% crystal violet and solubilized in 10% acetic acid, and absorbance at 540 nm was determined as a measure of adhesion (27).
In other studies, murine bone marrow-derived, ex vivo differentiated monocytes were assayed for fibronectin adhesion. For these studies, nonadherent cells were transferred to fibronectin-coated dishes and assayed for adhesion as above.

Statistical Analysis
Statistical significance was determined by Student's t test and analysis of variance methods using the Sigmaplot and Sigmastat software.

RESULTS
␤3 Integrin Expression and ␤3 Integrin-dependent Fibronectin Adhesion Increased during IFN␥-induced Differentiation of U937 Myeloid Cells-The goal of these studies was to determine whether HoxA10 regulates ITGB3 promoter activity, ␤3 integrin expression, and phagocyte adhesion in differentiating myeloid cells. Therefore, we performed initial experiments to determine whether the U937 myelomonocytic cell line was a good model for these studies. U937 cells are a myeloid leukemia line that undergoes monocytic differentiation upon treatment with various cytokines, including IFN␥ (24). IFN␥ differentiation of U937 cells results in acquisition of functional characteristics of mature phagocytes over 48 -72 h. These functional characteristics include respiratory burst activity, phagocytosis, and cell adhesion. In previous studies, we determined that HoxA10 protein is abundantly expressed in U937 cells and that expression is not altered by IFN␥ differentiation (14,19). Therefore, this cell line represents a reasonable model to study these events of late myeloid differentiation.
We initially determined the impact of IFN␥ differentiation on ␤3 integrin expression. In Western blots of U937 lysate proteins, we found that abundance of ␤3 integrin protein increased during the 72 h of IFN␥-induced differentiation (Fig. 1A). To determine whether this increase in ␤3 integrin protein correlated with an increase in mRNA abundance, we determined the impact of IFN␥ treatment on ␤3 integrin mRNA expression by real time PCR. For these studies, mRNA abundance of the respiratory burst oxidase components gp91 PHOX and p67 PHOX were positive controls for differentiation, and message abundance was normalized to actin and 18 S. Consistent with the protein expression data, we found that IFN␥ treatment increased ␤3 integrin mRNA in U937 cells (Fig. 1B). These studies indicated a statistically significant increase in ␤3 integrin, gp91 PHOX , and p67 PHOX mRNA during differentiation (p Ͻ 0.01, n ϭ 9 for all three messages).
However, these studies did not indicate a functional significance to increased ␤3 integrin expression during myeloid differentiation. Therefore, we investigated the role of ␤3 integrin in fibronectin adhesion of differentiating U937 cells. We chose to study this interaction because fibronectin is an extracellular matrix component involved in enhancing survival of hematopoietic cells (5). Interaction with fibronectin is also implicated in phagocyte rolling (8). In these studies, we found that 48 h of IFN␥ differentiation significantly increased U937 cell adhesion to fibronectin (p Ͻ 0.03, n ϭ 4) (Fig. 1C). Because both ␣v␤3 and ␣5␤1 heterodimers mediate myeloid cell adhesion to fibronectin, we determined the impact of blocking antibodies to ␤1 or ␤3 integrin (or irrelevant control). We found that preincubation with anti-␤1 antibody significantly decreased adhesion of untreated U937 cells to fibronectin (p Ͻ 0.001, n ϭ 4) (Fig. 1C). However, anti-␤1 antibody did not significantly impact fibronectin adhesion of IFN␥-treated cells (p ϭ 0.2, n ϭ 4). In contrast, preincubation with an antibody to ␤3 integrin did not alter adhesion of undifferentiated U937 cells to fibronectin (p ϭ 0.5, n ϭ 4) (Fig. 1C). However, IFN␥ differentiation did not significantly alter fibronectin adhesion of U937 cells pretreated with ␤3 integrin antibody.
To verify the specific impact of ␤3 integrin on IFN␥-induced fibronectin adhesion, stable U937 transfectants were generated with a vector to express a ␤3 integrin-specific shRNA or scrambled shRNA control (Fig. 1D). We found that IFN␥ differentiation significantly increased fibronectin adhesion of scrambled shRNA expressing control U937 transfectants, consistent with control U937 cells (p Ͻ 0.05, n ϭ 15) ( Fig. 1C). However, expression of a ␤3 integrin-specific shRNA blocked IFN␥-induced fibronectin adhesion of U937 transfectants. ␤3 integrinspecific shRNA to did not alter adhesion to fibronectin of untreated U937 cells, consistent with the blocking antibody studies above.
HoxA10 Expression Influenced ␤3 Integrin Expression and Fibronectin Adhesion of U937 Myeloid Cells-Therefore, we next determined the impact of HoxA10 overexpression on ␤3 integrin expression and fibronectin adhesion of U937 cells. For these studies, U937 cells were stably transfected with a vector to overexpress HoxA10 or empty control vector. Cell lysates from untreated and IFN␥-differentiated transfectants were tested by Western blot for expression of ␤3 integrin and HoxA10 (as a positive control). We found increased ␤3 integrin expression in HoxA10-overexpressing U937 transfectants in comparison with empty vector control transfectants with and without IFN␥ differentiation ( Fig. 2A). Neither endogenous nor overexpressed HoxA10 were altered in abundance by IFN␥ differentiation of the transfectants, consistent with our previous results (14,19).
We also investigated activation of Syk protein-tyrosine kinase, a downstream target of ␤3 integrin that mediates cell survival signals (8). For these studies, the blots were serially probed with antibodies to phosphorylated (activated) and total Syk. We found increased phospho-Syk in IFN␥-treated U937 transfectants overexpressing HoxA10, consistent with increased ␤3 integrin in these cells ( Fig. 2A). In contrast, little phospho-Syk is seen in untreated HoxA10-overexpressing transfectants or control transfectants with or without IFN␥.
We also performed the reciprocal investigation to determine the impact of decreased HoxA10 protein abundance on ␤3 integrin expression and fibronectin adhesion. For these studies, stable U937 transfectants were generated with a vector to express a HoxA10-specific shRNA or scrambled control shRNA. Cell lysates from these transfectants were analyzed for expression of ␤3 integrin and HoxA10, with and without IFN␥ differentiation (Fig. 2B). We found that inhibition of HoxA10 expression decreased ␤3 integrin expression in differentiating U937 cells.
We also investigated fibronectin adhesion of U937 transfectants with either HoxA10 overexpression or HoxA10 knockdown to determine the physiologic relevance of altered ␤3 integrin expression in these cells. We found that HoxA10 overexpression significantly increased adhesion to fibronectin in both undifferentiated and IFN␥-treated U937 transfectants (p Ͻ 0.02, n ϭ 6) (Fig. 2C). We also found that expression of a HoxA10-specific shRNA significantly decreased fibronectin adhesion in IFN␥-treated transfectants in comparison with transfectants with scrambled shRNA control vector (p Ͻ 0.02, n ϭ 6) (Fig. 2C). These studies suggested that altered HoxA10 expression altered both ␤3 integrin expression and fibronectin adhesion in differentiating myeloid leukemia cells.
HoxA10 Deficiency Decreased ␤3 Integrin Expression and ␤3 Integrin-dependent Adhesion in Differentiating Primary Myeloid Progenitor Cells-We also investigated the impact of HoxA10 deficiency on ␤3 integrin expression and fibronectin adhesion of murine bone marrow-derived monocytes. For these studies, myeloid progenitor cells were isolated from the bone marrow of HoxA10 ϩ/Ϫ mice or wild type control littermates. Progenitor cells were cultured in GM-CSF, IL3, and stem cell factor followed by ex vivo differentiation to monocytes with M-CSF over 4 days (see Ref. 15).
We initially determined the effect of HoxA10 haploinsufficiency on ␤3 integrin expression by Western blot of monocyte lysate proteins. We found increased ␤3 integrin expression in FIGURE 1. ␤3 integrin expression and ␤3 integrin-dependent adhesion increased during IFN␥ differentiation of U937 cells. A, IFN␥ treatment of U937 cells increased ␤3 integrin protein expression. U937 cells were differentiated with IFN␥ over a 72-h period, as indicated. Cells were harvested at various times and lysate proteins separated by SDS-PAGE. Western blots (WB) were serially probed with antibodies to ␤3 integrin and tubulin (as a loading control). IFN␥ treatment increased ␤3 integrin protein abundance over 72 h. B, IFN␥ treatment of U937 cells increased ␤3 integrin RNA expression. U937 cells were harvested with and without 48 h of IFN␥ treatment. Total cellular RNA was analyzed for abundance of ␤3 integrin, gp91 PHOX , and p67 PHOX mRNA by quantitative real time PCR. The latter two represent messages known to be induced in differentiating U937 cells. IFN␥ treatment of U937 cells significantly increased mRNA abundance of ␤3 integrin (indicated by *), gp91 PHOX (indicated by **), and p67 PHOX (indicated by #). Results were normalized to abundance of 18 S and actin mRNA to control for differences between samples in the total amount of RNA. C, ␤3 integrin-specific adhesion to fibronectin increased during IFN␥ differentiation of U937 cells. U937 cells were tested for adhesion to fibronectin with and without IFN␥-induced differentiation. Cells were preincubated with an antibody to ␤1 integrin, ␤3 integrin, or control bovine serum albumin. Incubation with ␤1 integrin antibody significantly decreased fibronectin adhesion of untreated U937 cells (indicated by *) but did not impact adhesion of IFN␥-differentiated U937 cells. In contrast, incubation with ␤3 integrin antibody significantly decreased fibronectin adhesion of IFN␥-treated U937 cells (indicated by **) but not of undifferentiated cells. In other studies, U937 cells were stably transfected with a vector to express a ␤3 integrin-specific shRNA or scrambled control shRNA. Adhesion of the stable transfectants to fibronectin was determined with and without IFN␥ treatment. No IFN␥-induced increase in adhesion to fibronectin was observed in cells expressing ␤3 integrin-specific shRNA. In contrast, expression of a ␤3 integrin-specific shRNA inhibited IFN␥-induced fibronectin adhesion of the transfectant cells (statistically significant difference in adhesion of differentiated transfectants is indicated by #). D, B3 integrin protein expression was decreased in U937 cells stably transfected with a vector to express a ␤3 integrin-specific shRNA. U937 cells were stably transfected with a vector to express a ␤3 integrin-specific shRNA to or scrambled control shRNA, as described above. Cells were treated with IFN␥, and total cell lysates were separated by SDS-PAGE. Western blots were serially probed with antibody to ␤3 integrin and tubulin (as a loading control). Decreased ␤3 integrin expression was seen in transfectants with the ␤3 integrin-specific shRNA vector in comparison with scrambled control vector. ex vivo differentiated wild type monocytes in comparison with HoxA10 haploinsufficient monocytes (Fig. 3A). These results were consistent with the studies with HoxA10 knockdown in U937 cells.
We also assayed these ex vivo differentiated cells for fibronectin adhesion. In these studies, we found that ex vivo differentiated wild type monocytes demonstrated significantly more fibronectin adhesion in comparison with HoxA10 ϩ/Ϫ monocytes (p Ͻ 0.02, n ϭ 4) (Fig. 3B). These studies further suggested a role for HoxA10 in ␤3 integrin expression and cell adhesion in differentiating myeloid cells.
Increased Fibronection Adhesion of HoxA10-overexpressing U937 Cells Was ␤3 Integrin-dependent-To determine whether increased fibronectin adhesion of HoxA10-overexpressing U937 cells was dependent on ␤3 integrin, stable transfectants were generated with a vector to overexpress HoxA10 or control vector and a vector to express ␤3 integrin-specific shRNA or scrambled control shRNA. Cell lysate proteins from IFN␥-treated transfectants were tested by Western blot to verify inhibition of ␤3 integrin expression by ␤3 integrin-specific shRNA in HoxA10-overexpressing transfectants (Fig. 4A).
These stable transfectants were assayed for fibronectin adhesion assays with and without IFN␥ differentiation. We found that HoxA10 overexpression significantly increased fibronectin adhesion of U937 cells co-transfected with a vector to express scrambled control shRNA, with and without IFN␥ differentiation, similar to the experiments in the section above (p Ͻ 0.001, n ϭ 4) (Fig. 4B). In contrast, HoxA10 overexpression did not significantly increase fibronectin adhesion in untreated cells integrin protein abundance in U937 cells with and without IFN␥ differentiation. U937 cells were stably transfected with a vector to express HoxA10 or empty vector control. Lysate proteins from undifferentiated and IFN␥-treated transfectants were separated by SDS-PAGE, and Western blots (WB) were serially probed with antibodies to ␤3 integrin, HoxA10 (to verify overexpression), and total and phospho-Syk (as a downstream activation target). Increased ␤3 integrin protein abundance was seen in HoxA10-overexpressing transfectants, with and without differentiation. This increase in ␤3 integrin expression was accompanied by increased phosphorylation of Syk in IFN␥-treated transfectants stably overexpressing HoxA10. Total Syk protein was the loading control in these studies. Expression of endogenous HoxA10 in control U937 transfectants was not altered by IFN␥ treatment, consistent with previous results in this cell line (14). B, decreased HoxA10 expression was associated with decreased ␤3 integrin expression during IFN␥-induced differentiation of U937 cells. U937 stable transfectants were generated with a vector to express HoxA10-specific shRNA or scrambled control shRNA. Cell lysates from IFN␥-treated or undifferentiated transfectants were analyzed by Western blot with antibodies to HoxA10, ␤3 integrin, or glyceraldehyde-3-phosphate dehydrogenase (to control for loading). Inhibition of HoxA10 expression did not alter ␤3 integrin expression in undifferentiated transfectants. In contrast, inhibition of HoxA10 expression was associated with decreased IFN␥-induced ␤3 integrin expression in differentiating U937 cells. ␤3 integrin expression was increased, and HoxA10 expression was not altered by IFN␥ differentiation of transfectants with control, scrambled shRNA, consistent with results of the experiments above. C, HoxA10 overexpression increased and HoxA10 knockdown decreased adhesion of U937 cells to fibronectin. U937 cells were stably transfected with a vector to express HoxA10 or empty vector control, or a HoxA10-specific shRNA or scrambled shRNA control. Adhesion to fibronectin was determined with and without IFN␥ differentiation of the transfectants. HoxA10 overexpression significantly increased fibronectin adhesion of U937 transfectants without (indicated by *) and with (indicated by **) IFN␥ treatment, consistent with the effect of HoxA10 expression on ␤3 integrin expression. In contrast, adhesion of IFN␥-treated transfectants expressing HoxA10-specific shRNA was significantly less than control, scrambled shRNA transfectants (indicated by #).
HoxA10 Activated a Cis Element in the ITGB3 Promoter-We next investigated the impact of HoxA10 on the activity of the ITGB3 promoter. Previous studies demonstrated in vitro interaction of HoxA10 with a Hox/Pbx-binding consensus sequence in the ITGB3 5Ј flank by EMSA using nuclear proteins from endometrial carcinoma cells (10). To determine whether this sequence functions as a HoxA10-activated cis element in myeloid cells, we generated an artificial promoter construct with multiple copies of the putative cis element linked to a minimal pro- Murine myeloid progenitor cells were isolated from the bone marrow of HoxA10 ϩ/Ϫ mice or wild type control littermates and cultured in GM-CSF and IL3. Myeloid progenitors were differentiated ex vivo with M-CSF, and cell lysates were analyzed by Western blots (WB) with antibodies to HoxA10, ␤3 integrin, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (to control for protein loading). HoxA10 ϩ/Ϫ monocytes expressed less ␤3 integrin in comparison with wild type monocytes. As expected, HoxA10 ϩ/Ϫ monocytes also expressed less HoxA10 than wild type monocytes. B, HoxA10 deficiency decreased fibronectin adhesion of ex vivo differentiated murine bone marrow monocytes. Murine myeloid progenitor cells were isolated from the bone marrow of HoxA10 ϩ/Ϫ and wild type mice and ex vivo differentiated with M-CSF, and fibronectin adhesion of the cells was determined. Adhesion of HoxA10 ϩ/Ϫ monocytes was significantly less than wild type monocytes (indicated by *). ␤3 integrin-dependent. A, expression of a ␤3 integrin-specific shRNA decreased ␤3 integrin protein expression in HoxA10-overexpressing U937 cells. U937 cells were stably transfected with a vector to overexpress HoxA10 or control vector and a vector to express ␤3 integrin-specific shRNA or scrambled control shRNA. Cell lysate proteins were separated by SDS-PAGE and Western blots (WB) serially probed with antibodies to ␤3 integrin, HoxA10, and tubulin (as a loading control). Expression of the ␤3 integrin-specific shRNA decreased ␤3 integrin expression in IFN␥-differentiated U937 transfectants, with and without HoxA10 overexpression. B, increased fibronectin adhesion of HoxA10-overexpressing U937 cells was mediated by ␤3 integrin. U937 cells were stably transfected with a vector to overexpress HoxA10 or control vector and a vector to express ␤3 integrin-specific shRNA or scrambled control shRNA. Fibronectin adhesion was determined with and without IFN␥ differentiation of the transfectants. Fibronectin adhesion was significantly increased by HoxA10 overexpression in untreated (indicated by *) and IFN␥-differentiated (indicated by **) transfectants expressing control, scrambled shRNA. In contrast, HoxA10 overexpression did not increase adhesion to fibronectin in transfectants expressing ␤3 integrin-specific shRNA, with or without IFN␥ differentiation. moter and chloramphenicol acetyltransferase reporter (B3ITATACAT). This construct or empty minimal promoter/ reporter control (pTATACAT) was co-transfected into U937 cells with a vector to express HoxA10 or with empty control vector. We also performed the reciprocal experiment and transfected U937 cells with these reporter constructs and a vector to express a HoxA10-specific shRNA or scrambled control shRNA. Transfectants were assayed with or without 48 h of IFN␥ differentiation.

FIGURE 4. Increased fibronectin adhesion in HoxA10-overexpressing U937 cells is
We found that reporter activity from the B3ITATACAT construct was almost 20-fold greater than from empty pTATA-CAT control vector (p Ͻ 0.001, n ϭ 7), indicating that this sequence functioned as a positive cis element in myeloid cells (Fig. 5A). We also found that this cis element was approximately twice as active in IFN␥-treated transfectants than in untreated transfectants (p ϭ 0.02, n ϭ 7). This was consistent with involvement of the cis element in ITGB3 transcription during myeloid differentiation.
Therefore, we next investigated the impact of HoxA10 overexpression on reporter activity in this system. We found that activity of the ITGB3 cis element was significantly greater in HoxA10-overexpressing U937 transfectants, with and without IFN␥ treatment (p Ͻ 0.004, n ϭ 7) (Fig. 5A). These results were consistent with the effect of HoxA10 overexpression on ␤3 integrin protein and message above. In contrast, overexpression of HoxA10 did not influence reporter activity from empty pTATACAT control plasmid, with or without IFN␥ treatment.
Also, we found that expression of a HoxA10-specific shRNA did not significantly influence activity of the ITGB3 cis element in undifferentiated U937 transfectants in comparison with transfectants with scrambled control shRNA (Fig. 5A). In contrast, expression of HoxA10-specific shRNA blocked IFN␥-induced increase in ITGB3-cis element activity, although expression of scrambled control shRNA did not have this effect. These studies are consistent with the results of Western blots above (Fig. 2B). These results are also consistent with the hypothesis that HoxA10 influences ITGB3 transcription during differentiation of myeloid cells.
A HoxA10/Pbx2 Heterodimer Binds the ITGB3 Promoter in a Multiprotein Complex-Because this ITGB3-cis element was homologous to derived consensus sequences for DNA binding of HoxA10/Pbx heterodimers, we investigated whether HoxA10 partners with a Pbx protein to bind this cis element. In our previous studies of the CYBB and NCF2 genes, we found that HoxA10 interacted with negative cis elements in these genes in vitro and in vivo in assays with undifferentiated U937 cells. During IFN␥ differentiation, HoxA10 binding to these negative CYBB and NCF2 cis elements decreased, resulting in decreased repression. However, the functional assays above suggested that HoxA10 bound to the ITGB3 cis element in IFN␥-treated cells and activated transcription during differentiation. This suggested the possibility that differentiation increased HoxA10 binding to the ITGB3 cis element, the opposite of the effect on HoxA10 binding to the CYBB and NCF2 genes.
Therefore, we performed EMSA to determine the impact of IFN␥ differentiation on HoxA10 interaction with the ITGB3 promoter cis element. For these studies, nuclear proteins were isolated from U937 cells, with or without 48 h of IFN␥ treat-ment, and used in binding assays with a radiolabeled probe representing the ITGB3 cis element (bp Ϫ1973 to Ϫ1933). As a control for IFN␥-induced U937 differentiation, EMSA were also performed with a probe representing the negative CYBB cis element (bp Ϫ94 to Ϫ134). In these studies, we found that IFN␥ treatment of U937 cells increased binding of a low mobility protein complex to the ITGB3 cis element (Fig. 5B). In additional studies, we determined that this complex represented specific protein binding to the Hox/Pbx consensus sequence in the probe (not shown; consistent with Ref. 10). Consistent with our previous results, IFN␥ treatment decreased binding of a low mobility protein complex of similar mobility to the CYBB probe. We have shown previously that this complex included HoxA10 and Pbx1.
Therefore, we performed additional experiments to determine whether the low mobility protein complex, which bound to the ITGB3 cis element, included HoxA10 and/or a Pbx protein. For these studies, nuclear proteins from IFN␥-treated U937 cells were preincubated with antibodies to HoxA10, Pbx1, Pbx2, or irrelevant control antibody and used in EMSA with the ITGB3 probe. We found that anti-HoxA10 antibody disrupted binding of the protein complex to this probe (Fig. 5C). Additionally, we found that antibody to Pbx2, but not Pbx1, also disrupted this complex. These results suggested that the ITGB3 cis element was activated by a HoxA10/Pbx2 heterodimer.
In these studies, we found that HoxA10 bound to the ITGB3 cis element in vivo in undifferentiated U937 cells. These results further suggested the hypothesis that HoxA10 activated the ITGB3 promoter in differentiating myeloid cells.
The Pbx Interaction Domain Is Not Identical to the HoxA10 Activation Domain-For some previously described Hox/Pbx target genes, the Pbx protein recruited transcriptional co-activator or co-repressor proteins to Hox/Pbx binding cis elements (18). The Pbx protein was therefore essential for modulation of transcription of such target genes. Therefore, we investigated whether HoxA10 activation of the ITGB3 promoter cis element required interaction with a Pbx protein. To do this, we used a previously described form of HoxA10 with mutation in the Pbx binding "hexapeptide" (referred to as (N312A/W313T) HoxA10) (28). Mutation of these residues was demonstrated previously to prevent interaction of HoxA10 with Pbx proteins but not HoxA10-DNA binding (15,19,28).
We co-transfected U937 cells with the reporter vector containing multiple copies of the ITGB3 cis element or empty vector control and a vector to overexpress (N312A/W313T) HoxA10. Reporter gene activity was determined for IFN␥treated and -untreated U937 transfectants, and we compared it with transfectants with wild type HoxA10 or empty control vector (Fig. 5A). We found that activation of the ITGB3 cis element was not significantly different in transfectants with wild type versus (N312A/W313T) HoxA10, with or without IFN differentiation (p Ͼ 0.2, n ϭ 7). In control experiments, neither expression of wild type nor (N312A/W313T) HoxA10 influenced expression of empty pTATACAT control reporter vector. Consistent with previous results, wild type and (N312A/ W313T) HoxA10 were equivalently expressed in U937 transfectants (not shown) (19).
These results suggested that the Pbx interaction domain was not identical to the HoxA10 activation domain. Therefore, we pursued identification of the domain involved in HoxA10 activation of ITGB3 transcription during myeloid differentiation.
HoxA10 Activates Transcription via a "PQ-like" Domain-We employed two different reporter assays to identify the HoxA10 activation domain. For the first assay system, we expressed various HoxA10 truncation mutants as fusion proteins with the DNA-binding domain of the yeast Gal4 transcription factor (referred to as Gal4-DBD). These HoxA10-Gal4-fusion proteins were assayed for activity by co-transfecting U937 cells with a vector to express the fusion protein and a reporter construct containing multiple copies of a Gal4-DNA-binding site linked to a minimal promoter. The advantage to this approach was that HoxA10 peptide sequences could be tested in isolation from the DNA-binding homeodomain. In the second set of assays, we tested the ability of HoxA10 truncation mutants to activate transcription from the cis element in the ITGB3 gene using the minimal promoter/ reporter constructs discussed above. The advantage to this approach was that it permitted functional assessment of the putative activation domain in the context of a genuine target gene. A, HoxA10 activated an IFN␥-inducible cis element in the ITGB3 promoter and decreased HoxA10 expression impaired IFN␥-induced activation of this cis element. U937 cells were cotransfected with a minimal promoter/reporter construct with four copies of the Ϫ1973 to Ϫ1933 bp sequence of the ITGB3 promoter (B3ITATACAT) or empty vector control (pTATACAT) and a vector to overexpress HoxA10 or N312A/W313T HoxA10 (a Pbx interaction domain mutant) or empty vector control. Other cells were transfected with these reporter vectors and a vector to express HoxA10-specific shRNA or control scrambled shRNA. Reporter gene activity was assayed with and without IFN␥ differentiation of the transfectants. IFN␥ differentiation significantly increased reporter expression from the ITGB3 sequence-containing construct. Also, reporter expression from the B3ITATACAT construct was significantly less in control transfectants in comparison with transfectants overexpressing either wild type HoxA10 or N312A/W313T HoxA10 without (indicated by *) and with (indicated by **) IFN␥ differentiation. Consistent with this, activity of the ITGB3 cis element was significantly less in IFN␥-differentiated transfectants expressing HoxA10-specific shRNA in comparison with control, scrambled shRNA transfectants (indicated by #). In contrast, none of these expression vectors significantly altered reporter expression from empty pTATCAT vector control. B, IFN␥ differentiation increased binding of a low mobility complex to the ITGB3 cis element in EMSA with nuclear proteins from U937 cells. Nuclear proteins were isolated from U937 cells with and without IFN␥ treatment. These proteins were used in EMSA with a radiolabeled probe representing either the positive cis element from the ITGB3 gene or the negative cis element from CYBB promoter. IFN␥ differentiation increased binding of a low mobility complex to the ITGB3 probe. In contrast, IFN␥ decreased binding of a similar HoxA10/ Pbx1-containing low mobility complex to the CYBB probe, as demonstrated previously (12,17). C, low mobility complex binding the ITGB3 cis element in EMSA with nuclear proteins from IFN␥-treated U937 cells is cross-immunoreactive with HoxA10 and Pbx2. EMSA were performed with nuclear proteins from IFN␥-treated U937 cells and the ITGB3 cis element probe. Binding reactions were preincubated with antibody (Ab) to HoxA10, Pbx1, Pbx2, or irrelevant control antibody. Antibodies to either HoxA10 or Pbx2 disrupted binding of the protein complex to the ITGB3 probe.
Based on these results, we generated a series of N-terminal HoxA10 truncation mutants to assay for activation of the ITGB3 cis element, as described above. These truncation mutants were engineered to include the genuine HoxA10 Kozak consensus sequence, an ATG start codon, and the DNAbinding homeodomain (amino acids 319 -403). U937 cells were co-transfected with the ITGB3 cis element-containing vector or empty vector control (B3ITATACAT or pTATA-CAT), and a vector to express HoxA10, 60 -403-aa HoxA10, 146 -403-aa HoxA10, 219 -403-aa HoxA10, or empty vector control. Transfectants were assayed for reporter gene activity with and without IFN␥ differentiation.
We found that activation of the ITGB3 cis element was not significantly different in transfectants overexpressing wild type HoxA10 versus 60 -403-aa HoxA10, with and without IFN␥ treatment (p Ն 0.4, n ϭ 7) (Fig. 6B). In contrast, both 146 -203-aa HoxA10 and 219 -403-aa HoxA10 significantly A, cryptic HoxA10 activation domain was located between amino acids 124 and 146. U937 cells were co-transfected with a vector to express various HoxA10 peptides as fusion proteins with the DNA-binding domain of the yeast transcription factor Gal4 or empty Gal4-DBD vector control and a minimal promoter/reporter vector with multiple copies of a Gal4-DNA-binding site. In comparison with control Gal4-DBD vector transfectants, reporter activity was significantly increased by transfection with a vector to express Gal4 fusion proteins HoxA10 amino acids 1-219, 1-179, 102-146, and 124 -146 (indicated by *). In contrast, reporter activity was significantly repressed in transfectants with a vector to express the full-length HoxA10-Gal4 fusion protein, consistent with our previous results (indicated by **). Transfectants with vectors to express fusion proteins with HoxA10 amino acids 1-112 or 145-219 exhibit reporter activity that is not significantly different from Gal4-DBD control transfectants. B, HoxA10 activation of the ITGB3 cis element required the cryptic HoxA10 activation domain. U937 cells were cotransfected with a minimal promoter/reporter construct with four copies of the Ϫ1973 to Ϫ1933 bp sequence of the ITGB3 promoter (B3ITATACAT) or empty vector control (pTATACAT) and a vector to express wild type HoxA10, 60 -403-aa HoxA10, 146 -403-aa HoxA10, 219 -403-aa HoxA10, or empty vector control. Reporter gene assays were performed with and without IFN␥ differentiation. Transcriptional activation of the ITGB3 cis element-containing reporter construct was significantly greater in untreated transfectants with wild type or 60 -403-aa HoxA10 in comparison with control vector transfectants (indicated by *). ITGB3 cis element activation was also significantly greater in IFN␥-treated transfectants overexpressing wild type or 60 -403-aa HoxA10 in comparison with control (indicated by **). In contrast, reporter activity from the ITGB3 cis element is significantly repressed by expression of 145-403-aa HoxA10 or 219 -403-aa HoxA10 in undifferentiated transfectants (indicated by #). Repression of B3ITATACAT by 145-403-aa HoxA10 or 219 -403-aa HoxA10 is relieved by IFN␥ treatment of the transfectants. None of these proteins influence reporter expression from control pTATACAT vector. C, HoxA10 activation domain is homologous to PQ domains found in transcriptional activator proteins. The HoxA10 124 -146-aa sequence was used to search a peptide data base for homology to known transcription factors. This domain has homology to previously described PQ domains in other proteins with transcriptional activation function such as the E1a-binding protein and bromodomain containing protein 4.
repressed reporter expression from the ITGB3 cis elementcontaining construct in comparison with control in untreated U937 transfectants, although this repression activity was abolished by differentiation. Indeed, reporter activity from the ITGB3 cis element-containing reporter construct in IFN␥treated transfectants with 146 -203-aa HoxA10 or 219 -403-aa HoxA10 was not significantly different from reporter activity in vector control transfectants (p ϭ 0.74, F ϭ 0.3, n ϭ 4). None of these proteins influence expression from empty pTATACAT control vector, with or without IFN␥ differentiation.
These studies suggested that the HoxA10 activation domain was not identical to the Pbx interaction domain but resided between amino acids 124 and 146. Therefore, we were interested in further examining this peptide sequence to identify homology with known functional domains in other proteins. To do this, we performed a data base search using the National Center for Biotechnology Information web site (National Institutes of Health, National Library of Medicine). We found that this HoxA10 peptide sequence had homology to "PQ" domains in other transcription factors, including the E1a-binding protein p300 (29) and bromodomain containing protein 4 (30) (Fig.  6C). These proteins activate gene transcription via interaction with transcriptional co-activators. Therefore, we next investigated whether HoxA10 aa 124 -146 recruited transcriptional co-activators as a potential mechanism for transcriptional activation.
E1a Expression Inhibits HoxA10 Activation of the ITGB3 Cis Element-Based on homology of the HoxA10 activation domain with the PQ domain in the E1a-binding protein, we hypothesized that this domain might recruit E1a-interacting transcriptional co-activators to the ITGB3 promoter. E1a is a viral oncoprotein that interacts with the transcriptional co-activators, including the CREB-binding protein (CBP), p300, and pCAF (31). E1a interacts with specific domains in these coactivators and prevents interaction with transcription factors, thereby blocking gene transcription. Therefore, we tested the impact of E1a expression on the HoxA10 activation domain. For these studies, U937 cells were co-transfected with a vector to express HoxA10-Gal4, 124 -146-aa HoxA10-Gal4, or control Gal4-DBD vector, a vector to express E1a or empty control vector, and a reporter vector with multiple copies of a Gal4binding cis element linked to a minimal promoter.
We found that E1a expression significantly decreased activation of the reporter construct by 124 -146-aa HoxA10-Gal4 (p Ͻ 0.0001, n ϭ 9) (Fig. 7A). Indeed, reporter expression in transfectants expressing 124 -146-aa HoxA10-Gal4 ϩ E1a was not significantly different from reporter expression in transfectants with empty Gal4-DBD vector control (p ϭ 0.98, n ϭ 9). E1a expression also significantly increased repression activity of HoxA10-Gal4 in this assay (p ϭ 0.01, n ϭ 9). These results suggested that expression of E1a in U937 cells interfered with interaction of a transcriptional co-activator and the HoxA10 PQ-like domain.
We were also interested in determining the impact of E1a expression on HoxA10 activation of the ITGB3 cis element. For these studies, U937 cells were co-transfected with the minimal promoter/reporter vector with multiple copies of the ITGB3 cis element (B3ITATACAT) or empty control vector (pTATA-CAT), a vector to express HoxA10 or empty vector control, and a vector to express E1a or empty vector control. Reporter expression was determined with and without IFN␥ differentiation of the transfectants.
In these studies, we found that E1a expression inhibited HoxA10 activation of the ITGB3 cis element in both untreated and IFN␥-differentiated transfectants (p Յ 0.01, n ϭ 4) (Fig.  7B). Specifically, reporter activity in these transfectants was not significantly different from transfectants with B3ITATACAT without HoxA10. In control experiments, HoxA10 did not activate the pTATACAT control vector, with or without IFN␥, with or without E1a.
The HoxA10 PQ-like Domain Recruited CBP to the ITGB3 Cis Element-The results of our functional studies suggested that HoxA10 activation of the ITGB3 cis element required interaction with a transcriptional co-activator. Detecting interaction of transcriptional co-activator proteins with DNA-bound transcription factor complexes can be difficult because these large complexes do not migrate well into acrylamide gels. Therefore, we performed chromatin co-immunoprecipitation using U937 cells with and without IFN␥ differentiation. For these studies, proteins were cross-linked to chromatin in vivo, and cell lysates were immunoprecipitated with an antibody to HoxA10 or the transcriptional co-activators CBP, pCAF, p300, or irrelevant control antibody. Immunoprecipitates were quantitated by real time PCR with primers flanking the HoxA10-activated cis element in the ITGB3 promoter (generating an ϳ90 bp product). Results were normalized to nonprecipitated chromatin to control for differences in DNA abundance between the samples. In these studies, we found that in vivo HoxA10 binding to this cis element increased significantly upon IFN␥ differentiation of U937 cells (p Ͻ 0.02, n ϭ 6) (Fig. 7C). We also found that IFN␥ differentiation significantly increased in vivo CBP binding to the ITGB3 cis element (p Ͻ 0.02, n ϭ 4). In contrast, there was no significant co-precipitation of ITGB3 cis element chromatin in experiments with antibody to pCAF or p300 in comparison with control, irrelevant antibody (not shown).
Because CBP has histone acetyltransferase activity, recruitment to the ITGB3 cis element by HoxA10 would be expected to increase acetylation of locally bound histones, decreasing chromatin compaction, and favoring transcription. Therefore, we used chromatin immunoprecipitation to analyze binding of acetylated histone 2A (H2A) to the ITGB3 cis element during IFN␥ differentiation. This antibody was chosen because histone H2 is one of the preferred substrates for histone acetyltransferase activity of CBP. In these studies, we found that IFN␥ differentiation significantly increased abundance of ITGB3bound acetyl H2A (p Ͻ 0.001, n ϭ 4).
These studies suggested that HoxA10 recruited CBP to the ITGB3 cis element, and this resulted in transcriptional activation. Therefore, we performed additional investigations to determine whether CBP interacted with the HoxA10 PQ-like activation domain. For these studies, U937 cells were transfected with a vector to express 124 -146-aa HoxA10-Gal4 or control Gal4-DBD. Lysate proteins from these transfectants were immunoprecipitated under nondenaturing conditions with an antibody to the Gal4-DBD or an irrelevant control antibody. Precipitated proteins were separated by SDS-PAGE, and Western blots were serially probed with antibodies to CBP and the Gal4-DNA-binding domain (as an expression control).
Consistent with our hypothesis, we found that CBP specifically co-precipitated with the HoxA10-PQ domain (Fig. 7D). We also found that this association was greater in assays with proteins from the differentiated transfectants. Control blots probed with the Gal4-DBD antibody demonstrated that 124 -146-aa HoxA10-Gal4 and control Gal4-DBD were equivalently expressed, and neither protein was altered in abundance by IFN␥-treatment of the transfectants.

DISCUSSION
In these studies, we found that HoxA10 was required for increased ITGB3 transcription and ␤3 integrin expression dur-FIGURE 7. HoxA10 activated transcription by recruiting the CBP to the ITGB3 cis element. A, function of the HoxA10 PQ-like activation domain was impaired by expression of the E1a oncoprotein. U937 cells were co-transfected with a vector to express either HoxA10 or 124 -146-aa HoxA10 as fusion proteins with the Gal4-DNA-binding domain, or empty Gal4-DBD control vector, a vector to express the E1a viral oncoprotein, or empty vector control, and a minimal promoter/reporter vector with multiple copies of a Gal4-DNA-binding site. Repression of reporter expression in transfectants with HoxA10-Gal4 was significantly increased by expression of E1a (indicated by *). Additionally, activation of reporter expression in transfectants expressing 124 -146-aa HoxA10-Gal4 was significantly decreased by co-transfection with the E1a expression vector in comparison with vector control (indicated by **). Indeed, reporter expression was not significantly different from transfectants with 124 -146-aa HoxA10-Gal4 ϩ E1a in comparison with empty Gal4-DBD control transfectants. B, expression of E1a impaired activation of the ITGB3 cis element by HoxA10. U937 cells were co-transfected with a minimal promoter/reporter construct with four copies of the cis element from the ITGB3 promoter (B3ITATACAT), or empty vector control (pTATACAT), a vector to express HoxA10, or empty vector control, and a vector to express E1a, or empty vector control. Reporter gene assays were performed with and without IFN␥ differentiation of the transfectants. Reporter gene activity is reported as % increase with HoxA10 overexpression in comparison with empty vector control. E1a expression significantly decreased HoxA10-induced activation of the ITGB3 cis element in untreated transfectants (indicated by *) and in IFN␥-treated transfectants (**). Neither HoxA10 nor E1a expression significantly influenced reporter expression from control vector pTATACAT. C, in vivo interaction of HoxA10, CBP, and acetylated histone 2A interaction with the ITGB3 cis element was increased by IFN␥ differentiation of U937 cells. U937 cells with and without IFN␥ differentiation were used in chromatin immunoprecipitation studies. Lysate proteins were immunoprecipitated with an antibody to HoxA10, CBP, acetylated histone 2A (acetyl-H2A), or irrelevant control antibody. Co-precipitating chromatin was amplified by real time PCR using primers flanking the ITGB3 cis element. Nonprecipitated chromatin was a control for DNA abundance in the samples. HoxA10 interacted with the ITGB3 cis element, and this interaction increased significantly in IFN␥-differentiated U937 cells (indicated by *). This was associated with a significant increase in interaction of CBP and acetyl-H2A with the cis element in differentiating U937 cells (indicated by ** and #, respectively). In contrast, the ITGB3 cis element did not co-precipitate significantly with irrelevant control antibody. D, CBP interacted with the HoxA10 PQ-like activation domain. U937 cells were transfected with a vector to express 124 -146-aa HoxA10 as a fusion protein with the Gal4-DNA-binding domain, or control Gal4-DBD vector. Cells were harvested with or without IFN␥-differentiation, and lysate proteins were immunoprecipitated under nondenaturing conditions with an antibody to the Gal4-DNA-binding domain or irrelevant control antibody. Immunoprecipitates were separated by SDS-PAGE and Western blots (WB) serially probed with antibody to CBP and Gal4-DNA-binding domain (as a control). CBP specifically co-precipitated from U937 transfectants with HoxA10-aa 124 -146 but not Gal4-DBD control. This interaction was greater in IFN␥-treated transfectants. Control Gal4-DBD Western blot indicated equivalence of protein expression and co-precipitation. HoxA10-aa 124 -146-Gal4-DBD fusion protein and control Gal4-DBD protein appear similar in size because of the relatively small size of the HoxA10 PQ domain in comparison with the Gal4-DBD (46 versus 43 kDa).
ing myeloid differentiation. We also found that this increase in ␤3 integrin expression was functionally relevant to the increase in fibronectin adhesion observed in differentiating myeloid cells. Although HoxA10 bound to an ITGB3 cis element as a heterodimer with Pbx2, we found that interaction with Pbx2 was dispensable for HoxA10 activation of this cis element. Consistent with this, we found that the HoxA10 activation domain was not identical to the Pbx-interacting hexapeptide but was a novel domain, not conserved in other Abd HoxA proteins. This domain was homologous to PQ domains that mediate proteinprotein interactions in other transcriptional activator proteins. In additional studies, we found that the HoxA10 PQ-like domain recruited CBP to the ITGB3 promoter, in vitro and in vivo. Therefore, these studies identified ITGB3 as a genuine target gene activated by HoxA10 in differentiating myeloid cells, determined that HoxA10 expression levels were functionally significant for fibronectin adhesion of differentiating myeloid cells, identified a unique HoxA10 activation domain, and determined that this domain recruited a transcriptional co-activator protein to a target gene promoter.
The Abd HoxA proteins, including HoxA10, are known to play key regulatory roles in embryogenesis and hematopoiesis (reviewed in Ref. 32). Despite this, relatively few genuine target genes have been identified for these homeodomain proteins. The issue of Hox target genes is of some complexity, because Hox proteins can either activate or repress transcription of various genes, depending upon context. Consistent with this, both activation and repression target genes have been described for HoxA10. For example, previous studies identified a sequence in the ITGB3 promoter that was activated by overexpressed HoxA10 in epithelial carcinoma cells (10). In contrast, we found that the CYBB and NCF2 genes were repressed by HoxA10 in undifferentiated myeloid cells (14,15). Because expression of ␤3 integrin increases during myelopoiesis, we investigated ITGB3 transcription as a potential model for transcriptional activation by HoxA10 during myeloid differentiation. During the course of these studies, we determined that HoxA10 activated ITGB3 transcription during myelopoiesis. Therefore, these studies identified a target gene that is activated by HoxA10 and is involved in the functional activities of mature phagocytes.
In these studies, we found that fibronectin adhesion of U937 cells undergoing IFN␥ differentiation is dependent on HoxA10-induced ␤3 integrin expression. We also found that HoxA10 overexpression increased Syk activation in differentiating myeloid cells. Because Syk kinase activity is associated with generation of cell survival signals (8), these results suggested a mechanism by which HoxA10 overexpression might antagonize apoptosis of differentiating myeloid cells. This is of potential importance because HoxA10 overexpression has been associated with poor prognosis in human acute myeloid leukemia and induces a myeloproliferative disorder in murine models.
Although the various functions of ␤3 integrin in mature phagocytes are well established, mechanisms that regulate ␤3 integrin activity are not. Myeloid cell adhesion via ␣5␤1 integrin is regulated by signal-dependent changes in integrin glycosylation, specifically in ␣2-6-sialylation of the ␤1 subunit (27,33). However, no similar modification of ␤3 integrin occurs (see Ref. 34). 3 Our studies identified an impact of IFN␥-induced myeloid differentiation on ␤3 integrin expression via increased ITGB3 transcription. These results suggested that cytokinestimulated expression may regulate ␤3 integrin function in differentiating myeloid cells or during the inflammatory response.
Although we found that IFN␥ differentiation of U937 cells increased ␤3 integrin-dependent fibronectin adhesion, this stimulation did not influence ␤1 integrin-dependent adhesion. In contrast, stimulation of U937 cells with phorbol myristate acetate increased fibronectin adhesion mediated by ␤1 integrin (27). The system of integrin regulation and function is obviously complex, involving post-translational modification and gene transcription. The importance of such fine-tuning of integrin expression and function in differentiating myeloid cells is not known but will be of considerable interest for understanding phagocyte function.
In current studies, we determined that HoxA10 bound the ITGB3 promoter as a heterodimer with the homeodomain protein Pbx2. However, in further studies, we found that interaction with Pbx2 was not essential for HoxA10 activation of the ITGB3 cis element. DNA-bound Hox/Pbx heterodimers have been shown to activate or repress some target genes via the ability of the Pbx partner to recruit transcriptional co-activator or co-repressor proteins (18). In such cases, the Hox partner is hypothesized to select the specific binding site for the heterodimer. In the current studies, we identified a novel HoxA10 activation domain that recruited a transcriptional co-activator protein to the ITGB3 promoter in a Pbx-independent manner. Although Pbx2 may increase the affinity of HoxA10 to the ITGB3 cis element, this effect was not apparent in our transfection studies. It is possible that increased HoxA10 binding affinity, because of interaction with Pbx2, may be relevant at endogenous HoxA10 levels but not in cells overexpressing HoxA10.
In previous studies, we determined that the CYBB and NCF2 are HoxA10 target genes (14,15). In contrast to the current studies, HoxA10 repressed these genes in undifferentiated myeloid cells. In those previous studies, we identified a novel HoxA10 repression domain that interacts directly with HDAC2 (19). This domain was not identical to the activation domain identified in the current studies, but it was similarly not conserved in other Abd HoxA family members. Although Abd HoxA proteins are highly conserved throughout the homeodomain and Pbx-interacting hexapeptide, the remainder of the proteins are divergent. This suggests that the roles that various family members play in gene regulation are likely to be at least partly nonredundant. We found this to be true for the CYBB gene, which is activated in differentiating myeloid cells by interaction of HoxA9 with the same cis element that is repressed by HoxA10 in undifferentiated cells (35).
In the current studies, interaction of HoxA10 with the positive cis element in the ITGB3 gene increased during myeloid differentiation. In contrast, interaction of HoxA10 with the repressor cis elements in the CYBB and NCF2 genes decreased during myelopoiesis (14,15). Decreased HoxA10 binding to the CYBB and NCF2 cis elements was related to inhibitory phosphorylation of tyrosine residues in the homeodomain in response to differentiating cytokines (36). The mechanism for increased HoxA10 interaction with the positive ITGB3 cis element in response to differentiation is not known but is a topic of current investigations in the laboratory.
Of additional interest are mechanisms involved in regulating activity of the HoxA10 activation domain versus the HoxA10 repression domain at various stages of differentiation or for various target genes. It is possible that function of these two domains is regulated by HoxA10 post-translational modification in response to cytokines. Alternatively, the presence of partner proteins might dictate whether the HoxA10 activation or repression domain is functionally dominant at a particular differentiation stage. Another possibility is that various binding sites alter HoxA10 conformation differently, favoring activity of either the activation or repression domain. Superficial examination of the CYBB and NCF2 versus ITGB3 cis elements did not provide an immediate explanation for variation in differentiation stage-specific interaction of HoxA10 with these genes or differential function. All three cis element have homology to previously described derived Hox/Pbx binding consensus sequences (37). Identification of additional HoxA10 target genes may clarify this issue.
Despite the importance of Abd HoxA proteins to normal and malignant myelopoiesis, the function and regulation of these proteins is poorly understood. In these studies, we identified a HoxA10 target gene that is activated during myeloid differentiation. We use this target gene as a model system to identify the HoxA10 activation domain and the mechanism by which this domain influences transcription. Such studies of Abd HoxA proteins may shed light on the oncogenic effect of overexpression of these proteins in myeloid malignancies.