The macrosialin promoter directs high levels of transcriptional activity in macrophages dependent on combinatorial interactions between PU.1 and c-Jun.

Macrosialin is a transmembrane glycoprotein that is highly expressed in macrophages. In the present studies, macrosialin mRNA levels are shown to be markedly up-regulated during macrophage differentiation of bone marrow progenitor cells in response to macrophage colony-stimulating factor and granulocyte-macrophage colony-stimulating factor. To investigate the mechanisms responsible for regulation of macrosialin expression, we have isolated the macrosialin gene and performed an initial analysis of its transcriptional regulatory elements. The macrosialin promoter and 7.0 kilobase pairs of 5'-flanking information direct high levels of reporter gene activity in monocyte/macrophage-like cells, but little or no expression in nonmyeloid cells. This pattern of expression is dependent on regulatory elements located between -7.0 and -2.5 kilobase pairs from the transcriptional start site that exhibit strong enhancer activity in macrophages and repressor activity in nonmyeloid cells. Analysis of the proximal macrosialin promoter indicates that combinatorial interactions between at least four classes of transcriptional activators, including PU.1/Spi-1 and members of the AP-1 family are required for basal promoter function. PU.1/Spi-1 and c-Jun act synergistically to activate the macrosialin promoter in a nonmyeloid cell line, suggesting that combinatorial interactions between these proteins are involved in regulating macrosialin expression during macrophage differentiation.

Macrosialin is a transmembrane glycoprotein that is highly expressed in macrophages. In the present studies, macrosialin mRNA levels are shown to be markedly up-regulated during macrophage differentiation of bone marrow progenitor cells in response to macrophage colony-stimulating factor and granulocyte-macrophage colony-stimulating factor. To investigate the mechanisms responsible for regulation of macrosialin expression, we have isolated the macrosialin gene and performed an initial analysis of its transcriptional regulatory elements. The macrosialin promoter and 7.0 kilobase pairs of 5-flanking information direct high levels of reporter gene activity in monocyte/macrophage-like cells, but little or no expression in nonmyeloid cells. This pattern of expression is dependent on regulatory elements located between ؊7.0 and ؊2.5 kilobase pairs from the transcriptional start site that exhibit strong enhancer activity in macrophages and repressor activity in nonmyeloid cells. Analysis of the proximal macrosialin promoter indicates that combinatorial interactions between at least four classes of transcriptional activators, including PU.1/Spi-1 and members of the AP-1 family are required for basal promoter function. PU.1/Spi-1 and c-Jun act synergistically to activate the macrosialin promoter in a nonmyeloid cell line, suggesting that combinatorial interactions between these proteins are involved in regulating macrosialin expression during macrophage differentiation.
Macrosialin is a heavily glycosylated murine transmembrane protein that belongs to the lysosomal/endosomal-associated membrane glycoprotein (lamp) 1 family (1)(2)(3)(4). Unlike other members of this family, which are constitutively and ubiquitously expressed in all cell types, macrosialin is specifically expressed in macrophages, and to a lesser extent in dendritic cells (reviewed by Gordon et al. (5)) (6,7). CD68, the human homologue of macrosialin (8), is expressed very early during granulomonopoietic differentiation, with intracellular staining found in bone marrow progenitor cells that express CD34 and myeloperoxidase (9). During further myeloid differentiation, CD68 remains strongly expressed in myeloperoxidase-positive, lactoferrin-negative, and CD14-negative cells. These cells represent myeloblasts, promyelocytes, and promonocytes. Interestingly, terminal differentiation toward the neutrophil/granulocyte lineage results in a marked decrease in CD68 expression, while continued differentiation toward the monocytic lineage is accompanied by further up-regulation of CD68 (9). In contrast to macrosialin expression, CD68 expression also has been localized, although at much lower levels, in certain lymphocyte subsets, megakaryocytes, and malignant hematopoietic cells (9 -11). This somewhat more extended pattern of expression may reflect differences in antibody specificities or slightly different functions between macrosialin and CD68 (7,12).
The biological functions of macrosialin and CD68 are not known. Although some studies have found CD68 to be exclusively localized in intracellular membrane compartments (13) in resting macrophages, more sensitive methods have detected a small fraction on the cell surface (9). Because macrosialin and CD68 are highly expressed, this small percentage of surface expression may nevertheless be of functional significance. Macrosialin and CD68 possess unique mucin-like extracellular domains located at the N-terminal region. In response to inflammatory stimuli, these regions undergo complex alterations in their patterns of N-and O-linked glycosylation (7,14) (reviewed by da Silva et al. (15)), and an increased fraction of macrosialin is found on the cell surface (16). The glycosylated regions of macrosialin and CD68 may play a role in protecting these proteins from the harsh hydrolytic environment found in lysosomes and/or may act as ligands for cell adhesion molecules, such as selectins. Saitoh et al. (17) have demonstrated that lamp-1 on leukemic cells can bind to E-selectin (17). Because macrosialin and CD68 are strongly expressed in mono-cytes and undergo changes in cell surface expression during an inflammatory response, it has been postulated that they might play roles in phagocytosis, and cell-cell and cell-pathogen interactions (1). Recently, macrosialin has been demonstrated to bind oxidized low density lipoproteins and account for 30 -50% of its uptake by activated THP-1 cells in vitro (16,18), suggesting that it may contribute to the development of macrophage foam cells in atherosclerotic lesions.
Because expression of macrosialin is up-regulated during macrophage differentiation, it also provides a model for understanding mechanisms that control macrophage-specific gene expression. The development of macrophages from bone marrow progenitor cells is regulated by a myriad of cytokines and colony-stimulating factors that include M-CSF and GM-CSF (19). Although significant progress has been made in identifying components of the signal transduction pathways that are activated by M-CSF and GM-CSF, the mechanisms by which these factors act to regulate the transcription of specific target genes so as to coordinate the proliferation and development of the monocytic lineage remain poorly understood.
To investigate molecular mechanisms that regulate early events in the program of macrophage differentiation, we have isolated the macrosialin gene, defined the exon structure, and performed an initial characterization of its promoter. The macrosialin promoter directs high levels of reporter gene expression in several monocyte/macrophage cell lines. Genomic sequences residing between Ϫ2.5 and Ϫ7 kb upstream of the translational start site contain cis-active elements that exhibit enhancer activities in monocyte/macrophage cells and silencer activities in nonmyeloid cells. Analysis of the proximal promoter suggests that combinatorial interactions between several classes of transcription factors are required for high levels of activity. Mutations or deletions of binding sites for PU.1/ Spi-1 and AP-1 severely impair macrosialin promoter activity. Conversely, high levels of macrosialin promoter activity can be established in a nonmyeloid cell by co-expression of c-Jun and PU.1/Spi-1. These observations suggest that PU.1 and c-Jun functionally cooperate to regulate macrosialin expression during macrophage differentiation.

EXPERIMENTAL PROCEDURES
Materials-The murine embryonic stem cell P1 library was obtained from Genomic Systems. Human recombinant M-CSF, IFN-␥, murine recombinant GM-CSF, TNF-␣, and IL-3 were from R&D Systems. Murine recombinant IFN-␥ was purchased from Genzyme. TPA and LPS were from Sigma. Plasmids Bluescript SK II and KS II were from Stratagene; pCDNA-3 was from Invitrogen; and pGEM T and the TNT-coupled reticulocyte lysate system were from Promega. DNA sequencing was performed using the Sequenase 7-diaza-dGTP DNA sequencing kit (U. S. Biochemical Corp.) and the 35  Isolation of Macrophages and Bone Marrow Progenitor Cells-Thioglycollate-induced mouse peritoneal macrophages were obtained from B6/D2 mice, and bone marrow cells were isolated and purified on Ficoll-Hypaque gradients as described previously (20). After initial purification, the bone marrow progenitor cells (5.0 ϫ 10 7 cells) were plated onto 150-mm tissue culture plates in 50 ml of bone marrow medium (21). The cells were treated with either recombinant human M-CSF (20 ng/ml), recombinant murine GM-CSF (4 ng/ml), or recombinant murine IFN-␥ (1000 units/ml) and assayed 24 -72 h later.
RNA Analysis-Total RNA was isolated by the guanidium thiocyanate method (22). RNase protection assays were performed as described previously (23). The antisense RNA probe for macrosialin corresponded to nucleotides 787-1072. The antisense probe for rat ␤-actin corre-sponded to nucleotides 2452-2594. The probes were hybridized with 20 g of total RNA or tRNA as a control, digested with RNase A (30 g/ml), and analyzed on a 10% denaturing polyacrylamide gel.
Isolation and Characterization of Genomic Clones-A mouse embryonic stem cell P1 library was screened using PCR oligonucleotides of sequence 5Ј-CTGATCTTGCTAGGACCGCT-3Ј and 5Ј-GCTGGTCG-TAGGGCTG-3Ј corresponding to nucleotides 110 -129 and 222-241 of the macrosialin cDNA, respectively. Three overlapping P1 clones were obtained. The PCR product generated with the two oligonucleotides was used to probe restriction fragments of one of these clones. An 8.5-kb BamHI fragment hybridizing to the PCR product was subcloned into Bluescript SK II, generating the plasmid JC59. Sequence analysis confirmed that this fragment contained the 5Ј end of the macrosialin cDNA, but lacked the 3Ј end of the macrosialin-coding sequences. Therefore oligonucleotides corresponding to the 3Ј end of the macrosialin cDNA (5Ј-CAGAATTCATCTCTTCGAGAGCTC-3Ј and 5Ј-GATGCT-CAGAGGGGCTGGT-3Ј corresponding to nucleotides 785-807 and 1058 -1077, respectively) were used to amplify genomic macrosialin sequences. A 500-bp PCR product was generated and subcloned into pGEM T, JC67. Sequence confirmed that it contained the 3Ј end of the macrosialin gene. A 180-bp fragment from JC67 was used to probe the P1 clone that had been digested with BamHI in a Southern blot. A 3.2-kb fragment was subcloned into Bluescript SK II and confirmed by sequencing to contain the 3Ј end of genomic macrosialin and the flanking region generating plasmid, JC74. Sequence analysis was performed using an ABI automated DNA sequencer and the MacDNAsis program (Hitachi).
Identification of the Transcriptional Start Site by Primer Extension and RNase Protection Analysis-The transcriptional start site was mapped by primer extension and RNase protection assays. In primer extension assays, antisense oligonucleotides (5Ј-CCAGCTAGGCTA-CACCAGTTCCTTC-3Ј and 5Ј-AGGGAGAAGCTTGGCAGAGATGC-3Ј) corresponding to nucleotides Ϫ41 to Ϫ17 and Ϫ11 to ϩ11, relative to the translation start site in the macrosialin gene were labeled with [␥-32 P]ATP using T 4 polynucleotide kinase and annealed at 30°C with 10 g of total murine peritoneal macrophage RNA. Annealing conditions and subsequent procedures were carried out as described previously (22). For RNase protection studies plasmid JC59, containing the genomic fragment of the macrosialin 5Ј-flanking region, was linearized with NsiI, which is located 235 bp upstream of the translational start site. A 32 P-labeled antisense cRNA probe was generated from this plasmid using T 7 RNA polymerase, and RNase protection assays were performed using 20 g of total RNA as described above.
Constructions of Reporter Constructs and Site-directed Mutagenesis-Macrosialin-luciferase reporter genes were constructed by changing the translational start site ATG to a EcoRI site by PCR mutagenesis using an antisense oligonucleotide of sequence 5Ј-AGGGAGAATTCTG-GCAGAGATGC-3Ј and a sense primer of sequence 5Ј-CAAGCCTTTA-ATTCCCAGCAT-3Ј corresponding to the sequence starting 664 bp upstream from the macrosialin translational start site in plasmid JC59. The 675-bp PCR product was sequenced and verified to be correct. The PCR product was digested with NheI and EcoRI and used to replaced the corresponding sequence within the wild type macrosialin gene and subcloned into Bluescript KS II (AL9). The Ϫ7.0 kb 5Ј-flanking region was excised with BamHI and EcoRI and subcloned into ⌬5ЈPSV2 luciferase to generate Mac 7.0-luciferase. Deletions of the 5Ј-flanking region were created by restriction enzyme digestion using the following enzymes: NotI (Ϫ5.5 kb), HindIII (Ϫ2.5 kb), Spe I (Ϫ803 bp), NheI (Ϫ614 bp), and PvuII (Ϫ250 bp). Additional 5Ј deletions were generated by PCR using an antisense primer 5Ј-AGGGAGAAGCTTGGCAGATGC-3Ј, replacing the ATG site with a HindIII site, and the following sense primers containing BamHI restriction sites: Ϫ1060, 5Ј-ATTTGCTG-GATCCAATCTACAG-3Ј; Ϫ221, 5Ј-GAGGTAACGGATCCTTTGTAC-3Ј; Ϫ203, 5Ј-CGCCCGGATCCGAACGTCAC3-Ј; Ϫ133, 5Ј-GCTGAGGATC-CTGAGTCAGGT-3Ј; Ϫ108, 5Ј-GTGGGATCCTTTTAGTTAAGG-3Ј; Ϫ77, 5Ј-GGCTTTGGATCCCCTCTTCCA-3Ј. The Ϫ31 to Ϫ1 construct was generated by annealing two complementary oligonucleotides of sequence 5Ј-GATCCTGTGTAGCCTAGCTGGTCTGAGCATCTCTGCC-A-3Ј and 5Ј-AGCTTGGCAGAGATGCTCAGACCAGCTAGGCTACA-CAG-3Ј and subcloning these into the ⌬5Ј PSV2-luciferase reporter gene at the BamHI and HindIII sites. In addition, a 3Ј deletion was generated, deleting a putative Ets binding site at Ϫ46, using an antisense oligonucleotide of sequence 5Ј-CCTCAAGCTTATCCCCTTTGCCT-TCTC-3Ј and the Ϫ221 sense oligonucleotide described above for PCR amplification. Similarly, mutations in the AP-1, PU.1/Spi-1, GC region and the CCAAT binding sites were generated by PCR using the Ϫ221 construct as a template. A general strategy for developing the mutations of these binding sites using overlapping PCR mutagenesis was employed. For example, the mutation of the AP-1 site at Ϫ132 was constructed by creating external primers containing BamHI and Hin-dIII sites at Ϫ221 and Ϫ1 primers respectively as described above. The internal primers 5Ј-TGAGGTGTCCTCGAGAGGTTT-3Ј (forward) and 5Ј-AAACCTCTCGAGGACACCTCA-3Ј (reverse) introduced an XhoI restriction site into the AP-1 site. The two PCR products were digested with either BamHI and XhoI, or XhoI and HindIII, ligated together and subcloned into the luciferase expression vector at BamHI and HindIII sites. Mutations of the PU.1/Spi-1, GC region and CCAAT boxes were similarly done introducing a NotI restriction site into the PU.1/SPi-1 site (5Ј-TATTTTAGTGCGGCCGCGTGAGGCTTT-3Ј (forward) and 5Ј-AAAGCCTCACGCGGCCGCACTAAAATA-3Ј (reverse)); an EcoRI restriction site into the GC region (5Ј-TAACGGATCCTTTGTAGAAT-TCACTGA-3Ј (forward) and 5ЈTCAGTGAATTCTACAAAGGATCCGT-TA-3Ј (reverse)); and a BglII restriction site into the second CCAAT box (5Ј-TGTGAAAAGATCTGGCTTGAGTGG-3Ј (forward) and 5Ј-CCACT-CAAGCCAGATCTTTTCACA-3Ј (reverse)). For the 1st CCAAT box mutation, a PCR product was made using a primer containing a XbaI restriction site at the CCAAT box region (5Ј-TAACGGATCCTTTGTAC-CGCCCACTGAGAACGTCACTGTCTAGAACAGCCTAAT-3Ј) and the Ϫ1 antisense HindIII oligonucleotide primer. All constructs generated by PCR were confirmed by dideoxy sequencing and restriction enzyme digestion.
Cell Culture and Transient Expression Analysis-Transient transfections using THP-1 (monocytic leukemia), U937 (histocytic leukemia), HL-60 (acute promyelocytic leukemia), Jurkat (T cells), and BaF/3 (murine pro-B cell) and BaF/3 cells expressing the murine GM-CSF ␣-receptor (gifts from A. D. D'Andrea) were performed by electroporation as described previously (24) using 5-10 g of total plasmid DNA. HeLa (cervical endothelial), P-19 (mouse embryonic carcinoma), and GC-3 (anterior pituitary) were transfected by the calcium phosphate method (25) using 2 g of total plasmid DNA. MCF-7 (breast adenocarcinoma) and RAW 264.7 (murine monocyte) were transfected with 2 g of total plasmid DNA using DOTAP and Lipofectin, respectively, following the manufacturer's instructions. The myeloid cell lines were harvested for luciferase and ␤-galactosidase activity 24 h after the time of transfection, whereas the nonmyeloid and the RAW 264.7 cells were harvested 48 h after transfection. Luciferase activity was measured in a Monolight 2010 luminometer (Analytical Luminesence) as described previously (26). Luciferase activity was normalized to ␤-galatosidase activity directed by a co-transfected plasmid containing the ␤-actin promoter linked to the ␤-galatosidase gene. ␤-Galactosidase activity was assayed using Galacto-Light following the manufacturer's instructions and also measured in the Monolight 2010 luminometer. An equimolar amount of the ␤-actin luciferase construct was used as an external standard. U937 and BaF/3 cell lines were also treated with various cytokines and chemokines for 14 -48 h prior to harvesting. After electroporation, the U937 cells were resuspended in 0.5% fetal bovine serum in RPMI prior to treatment. The BaF/3 cells were maintained in either RPMI with 10% fetal bovine serum and 5% WEHI-3 conditioned medium or RPMI with 1% fetal bovine serum and 0.5 ng/ml murine IL-3. Prior to electroporation, the BaF/3 cells were cytokinestarved for 6 h and resuspended in 0.5-1.0% fetal bovine serum without cytokines in RPMI after electroporating the cells. The cells were treated with either TPA, 1 ϫ 10 Ϫ7 M; LPS, 10 ng/ml; retinoic acid, 1 ϫ 10 Ϫ6 M; recombinant human and murine GM-CSF, 4 ng/ml; recombinant human IFN-␥, 1000 units/ml; TNF-␣, 100 ng/ml; and recombinant murine IL-3, 10 ng/ml. A ␤-actin/␤-galatosidase construct was used as an internal control in the U937 cells and a SV40-␤-galatosidase construct was used in the BaF/3 cells. Co-transfections of expression vectors into P-19 cells used 1.0 g of reporter plasmid and either or both 100 ng of a CMV-PU.1 expression vector and 100 ng of a CMV-c-Jun expression vector. 50 ng of ␤-actin/␤-galatosidase construct was also co-transfected and used as an internal standard. 100 -200 ng of the empty vector, pcDNA-3, were used as a control. Salmon sperm DNA was used as a carrier to equalize the total amount of transfected DNA. All transfections were performed with triplicate points, at least three times.
Preparation of Nuclear Extracts and Electrophoretic Mobility Shift Assays-Nuclear extracts were prepared from TPA-treated THP-1 cells as described previously (27). To determine the binding of nuclear factors to regions within the macrosialin promoter, double-stranded oligonucleotides with 5Ј overhangs were synthesized. The oligonucleotides were labeled either with [␣-32 P]dCTP using the Klenow fragment of DNA polymerase or with [␥-32 P]dATP using T 4 polynucleotide kinase. The sequences of the oligonucleotides corresponding to the macrosialin AP-1 element were 5Ј-gatccAGGTGTCTGAGTCAGGTTTGG-3Ј (sense) and 5Ј-gatctCCAAACCTGACTCAGACACCT-3Ј (antisense), where lowercase letters denote non-native sequences added to facilitate cloning.
The sequences of the sense and antisense oligonucleotide corresponding to the macrosialin PU.1/Spi-1 element were 5Ј-gatccGTTAAGG-GAAGTGA-3Ј and 5Ј-gatctTCACTTCCCTTAAC-3Ј, respectively. Five micrograms of nuclear extract were incubated with 0.5 g of poly(dI⅐dC) and 5-50-fold molar excess of specific or mutant unlabeled competitor (5Ј-TGAGGTGTCCTCGAGAGGTTT-3Ј for the AP-1 mutation and 5Ј-TATTTTAGTGCGGCCGCGTGAGGCTTT-3Ј for the PU.1/Spi-1 mutation) in 150 mM KCl, 10 mM Tris, pH 8.0, 0.1 mM EDTA, 50 mM dithiothreitol, 5 g of bovine serum albumin, and 5% glycerol for 30 min on ice. One microliter of probe (100,000 cpm) was added to the reaction mixture and incubated for another 30 min on ice. For supershift assay, 1-2 l of antibody were added to the mixture and preincubated on ice for 1 h. The reaction mixtures were spun down at 4°C and run on a 6% nondenaturing polyacrylamide gel at 300 V. Anti c-Jun monoclonal IgG raised against a peptide corresponding to amino acids 56 -69 of the human c-Jun and anti-Jun-B polyclonal IgG raised against amino acids 45-61 of mouse Jun-B were used. Guinea pig anti-PU.1 antiserum was raised against a recombinant peptide, corresponding to amino acids 157-272, that is specific to the DNA binding domain. In vitro translated products of c-Jun and PU.1 were synthesized using the TNT-coupled reticulocyte lysate system following the manufacturer's instructions.

Macrosialin mRNA Is Up-regulated during Macrophage
Differentiation-To assess macrosialin expression during macrophage differentiation, RNase protection assays were performed. Bone marrow progenitor cells were enriched by purification through Ficoll gradients and cultured in the presence of M-CSF or GM-CSF to induce proliferation and differentiation. Macrosialin mRNA was detected in the bone marrow progenitor cells, consistent with the observation that CD68 is expressed in CD34 ϩ cells (9). In addition, macrosialin mRNA levels were markedly up-regulated in response to both M-CSF and GM-CSF within 24 h, reaching maximal levels by 72 h in M-CSF-treated cells. Intriguingly IFN-␥, a potent regulator of numerous inflammatory responses in macrophages (28), inhibited the induction of macrosialin mRNA in response to M-CSF by more than 50%. Thioglycollate-elicited peritoneal macrophages demonstrated the highest levels of macrosialin mRNA, which were approximately half as abundant as ␤-actin mRNA when corrected for the relative specific activities of the macrosialin and ␤-actin probes (Fig. 1).
Cloning and Characterization of the Macrosialin Gene-To investigate molecular mechanisms controlling macrosialin gene expression at the transcriptional level, genomic clones containing macrosialin regulatory elements and coding sequences were isolated from a mouse P1 clone ( Fig. 2A). Two adjacent BamHI fragments were subcloned that together contain 7.0 kb of 5Ј-flanking information, the macrosialin structural gene, and 1.35 kb of 3Ј-flanking information. Comparison of the genomic and cDNA sequences indicate that the primary macrosialin transcript contains six exons. With the exception of the junction between exon 4 and exon 5, all of the assigned splice donor and acceptor sites exhibit close matches to consensus splice donor and acceptor sequences. Exon 1, which is of variable length due to multiple transcriptional start sites, as described below, contains the translational start site and encodes the first 14 amino acids of the signal peptide. Exon 2, which is the largest exon (439 bp), encodes the remaining 7 amino acids of the signal peptide and the majority of serine and threonine residues thought to be substrates for O-linked glycosylation. In addition, exon 2 encodes five potential N-linked glycosylation sites (Asn-X-Ser/Thr), the so-called proline hinge, and the first of four cysteine residues that are conserved in members of the lamp family and are involved in intramolecular disulfide bonds. Exon 3 contains the second conserved cysteine residue and one potential N-linked glycosylation site. Exon 4 contains a single potential N-linked glycosylation site, while exon 5 contains two potential N-linked glycosylation sites and the third conserved cysteine residue. Exon 6 contains the final conserved cysteine residue and encodes the transmembrane domain and cytoplasmic tail. The predicted cytoplasmic tail sequence is RRRQSTYQPL, which is similar to other members of the lamp family in containing three basic amino acids and a conserved tyrosine. Previous studies have suggested that an alternative form of macrosialin might be expressed with a truncated C-terminal tail of sequence RR*, due to either a second, highly related gene or an alternative exon (1). No evidence was found for an alternative exon that would encode a truncated cytoplasmic domain in the 1.35 kb of the further 3Ј-flanking sequence. As only one genomic P1 clone was characterized, the possibility of a second highly related gene cannot be excluded. However, Southern blotting experiments using probes to the 3Ј end of the cloned macrosialin gene have thus far been consistent with the presence of a single gene. These observations suggest that versions of the macrosialin cDNA that encode truncated proteins may have arisen as artifacts during PCR amplification or cDNA library construction.
Transcription of the Macrosialin Gene Is Initiated at Multiple Start Sites-The transcription start sites of the macrosialin gene were determined by primer extension and RNase protection analysis using total RNA from murine peritoneal macrophages that were thioglycollate-elicited to maximize the macrosialin mRNA levels. Two oligonucleotides that are complementary to the macrosialin mRNA from Ϫ17 to Ϫ41 and ϩ11 to Ϫ11 bp from the translational start site were used.
Primer extension experiments utilizing the antisense oligonucleotide corresponding to nucleotides Ϫ17 to Ϫ41 of the macrosialin cDNA relative to the translational start site resulted in several extension products, with the majority ending 55-79 nucleotides upstream of the initiator methionine (Fig. 3A). Extension products were not detected for the ϩ11 to Ϫ11 oligonucleotide, possibly due to secondary structure of the macrosialin mRNA. To confirm the presence of multiple start sites, RNase protection assays were performed using the 5Ј-flanking region of the genomic sequence as a template for generation of an antisense RNA transcript. The RNase protection assay also demonstrated multiple sized fragments (Fig. 3B) corresponding to the start sites identified in the primer extension analysis using the Ϫ17 to Ϫ41 primer. These observations are consistent with the presence of sequences that match the consensus for Inr elements (YYANT/AYY) (29) at Ϫ83 and Ϫ111, with the sequence at Ϫ83 closely associated with several transcriptional start sites. Intriguingly, the Inr element at Ϫ111 is only 1 bp removed from a consensus TATA box and is favorably positioned with respect to the Inr element at Ϫ83. Because there was no predominant start site, we have utilized a convention in which upstream regulatory elements are numbered relative to the translational start site, with the majority of transcription initiated at clusters of start sites between Ϫ55 and Ϫ79 bp upstream (Figs. 2 and 3). The 5Ј end of the previously published macrosialin cDNA ends at Ϫ93, indicating that it was generated from a relatively rare mRNA initiated upstream of the major start sites (1).
The Macrosialin Promoter Is Preferentially Expressed in Myeloid Cells-To begin to characterize transcriptional regulatory elements that control macrosialin expression, a 7.0-kb fragment of the 5Ј-flanking region of the macrosialin gene containing the transcriptional start sites was linked to a luciferase reporter gene and transiently transfected into myeloid and nonmyeloid cells (Fig. 4A). The activity of the macrosialin promoter was normalized to the activity of the ␤-actin promoter to correct for different transfection efficiencies observed in the various cell types. As shown in Fig. 4A, the macrosialin promoter displayed a preferential activity in myeloid cell lines versus nonmyeloid cells lines. RAW cells, which represent a macrophage-like cell, demonstrated the highest levels of basal promoter activity. High levels of expression were also observed in the monocytic THP-1 cells and U937 which, represent monoblast-like cells. The macrosialin promoter was also expressed in HL-60 cells, which are capable of differentiating into either monocyte, macrophage, or granulocyte-like cells in response to vitamin D, TPA, or retinoic acid, respectively (30,31). Jurkat cells exhibited the lowest level of promoter activity among the hematopoietic cell lines. Treatment of HL60, U937, and THP-1 cells with TPA to induce macrophage differentiation increased the level of promoter activity in each of these cell lines by 3-5-fold. The results are consistent with the effect of TPA on CD68 protein levels observed in THP-1 cells (16). TPA did not have any significant effect on the more fully differentiated RAW cell line. Interestingly, HeLa but not other nonmyeloid cells examined, showed significant levels of promoter activity that was also inducible by TPA. The 7.0-kb macrosialin regulatory elements exhibited a level of promoter activity in TPAtreated THP-1 cells that was approximately one-fourth the promoter activity directed by the ␤-actin promoter. This observation is consistent with the relative levels of macrosialin and ␤-actin mRNA determined by RNase protection assays in primary macrophages (Fig. 1) and suggests that the macrosialin promoter is one of the most highly active macrophage-specific promoters yet identified. A direct comparison with the scavenger receptor A gene regulatory elements, which also direct macrophage-specific expression (20,25,32), indicated that the macrosialin promoter is approximately 100-fold more active with respect to both basal and TPA-stimulated activity (Fig. 4B).
Regulation of Macrosialin Promoter Activity-To determine whether or not the macrosialin promoter responds to cytokines and other signaling molecules that affect the proliferation and maturation of macrophages, U937 cells were transiently transfected with the Mac-7.0 luciferase reporter gene and treated with combinations of TPA, GM-CSF, IFN-␥, retinoic acid, TNF-␣, and LPS. GM-CSF treatment alone increased promoter activity slightly above basal activity, and a more than additive effect was observed when cells were treated with both TPA and GM-CSF (Fig. 4C). IFN-␥ and retinoic acid had no effect on basal promoter activity, but inhibited TPA-dependent transcription by approximately 50%. TNF-␣ and LPS inhibited both basal and TPA-dependent promoter activity to a similar extent (Fig. 4C).
We next evaluated regulation of the macrosialin promoter in BaF/3 murine proB cells, which are dependent on IL-3 for growth (33). When the Mac 7.0-luciferase reporter gene was transfected into BaF/3 cells in the absence of IL-3, very little promoter activity was observed, and TPA treatment had only a slight stimulatory effect. Treatment with GM-CSF, either alone or in combination with TPA, had no effect on Mac 7.0 promoter activity, consistent with the lack of expression of GM-CSF receptors in BaF/3 cells. In contrast, treatment with IL-3 led to a 23-fold induction in Mac 7.0 promoter activity (Fig. 4D).
To further evaluate regulation of the macrosialin promoter by GM-CSF, experiments were performed in BaF/3 cells that had been stably transfected with the ␣-subunit of the GM-CSF receptor (34). Transient transfections of the Mac 7.0-luciferase reporter in this cell line exhibited a level of basal activity that was as much as 100-fold higher than the parental BaF/3 cell line (Fig. 4E). In contrast to the parental BaF/3 cells, treatment with GM-CSF lead to a 3-4-fold induction in promoter activity in BaF/3-GM cells. Treatment of these cells with IL-3 also resulted in a 3-4-fold increase in promoter activity, resulting in an absolute level of promoter activity that was similar to that observed in GM-CSF-treated cells, but representing a much lower fold of induction than that observed in the parental BaF/3 cells (Fig. 4E). TPA treatment did not contribute significantly to promoter activity when added to GM-CSF or IL-3 treatments.

Identification of Cell-specific Distal Enhancer Elements-To
identify regulatory elements within the macrosialin gene necessary for high levels of promoter activity, a 5Ј-deletion analysis was performed (Fig. 5). Deletion of the region from Ϫ7.0 to Ϫ5.5 kb resulted in a 50% reduction in basal promoter activity and an 85% reduction in TPA-dependent activity in U937 cells (Fig. 5A). Deletion to Ϫ2.5 kb led to a further reduction in basal promoter activity in U937 cells to 20% of that observed for the Mac 7.0 reporter, and a reduction of TPA-dependent activity to approximately 10% of that observed for Mac 7.0. In contrast, deletions of the macrosialin promoter to Ϫ5.5 and 2.5 kb resulted in progressive increases in basal and TPA-dependent transcriptional activity in HeLa cells (Fig. 5B). In P19 cells, which express the Mac 7.0 reporter gene at very low levels, deletions to Ϫ5.5 kb also led to a 3-fold increase in reporter gene activity (Fig. 5C). In concert, these observations suggest the existence of complex regulatory elements residing between Ϫ2.5 and Ϫ7.5 kb upstream of the major transcriptional response site that confer enhancer activities in myeloid cells and silencer activities in nonmyeloid cells.
Multiple cis-Active Elements Are Required for Activity of the Proximal Macrosialin Promoter-Computer-assisted analysis of the macrosialin promoter revealed several potential binding sites for sequence-specific transcription factors. Four putative binding sites for AP-1 transcription factors were identified at Ϫ925, Ϫ899, Ϫ257, and Ϫ131 bp from the translational start site. Three putative Ets binding sites, were identified at Ϫ919, Ϫ332, and Ϫ43. A sequence motif located between Ϫ104 to Ϫ89 is similar to the PU.1/Spi-1 binding site identified in the macrophage scavenger receptor A gene (32). Three CCAAT boxes were identified at Ϫ338, Ϫ191, and Ϫ167. A GC-box was noted at Ϫ213. To determine whether these or other elements were required for function of the macrosialin promoter, an extensive series of 5Ј-deletion mutants was evaluated in U937 cells (Fig.  6). Progressive deletions from 1.06 kb to Ϫ250 bp resulted in progressive decreases in basal promoter activity (Fig. 6A). Deletion to Ϫ221 resulted in a marked increase in activity, suggesting that the region between Ϫ250 and Ϫ221 contains a silencer element (Fig. 6A). Further deletion to Ϫ77 abolished promoter activity. Based on these results, a more detailed analysis was performed of the Ϫ221 promoter (Fig. 6C). Mutations were made in the putative GC box, the CCAAT boxes, the AP-1, and PU.1/Spi-1 binding sites to determine their potential roles. Mutating each of these regions substantially decreased the basal promoter activity and all but the first CCAAT box had a decrease in TPA response. Deletion of the Ets binding site did  6. cis-Active elements required for activity of the proximal macrosialin promoter in U937 cells. A, a series of 5Ј-deletion mutants of the macrosialin promoter extending from Ϫ1.06 kb to Ϫ31 bp from the translational start site were electroporated into the U937 cell line as described under "Experimental Procedures." Ten micrograms of each constructs were electroporated into the U937 cell line. One microgram of the ␤-actin/␤-galactosidase was co-transfected and used as an internal standard. Cells were treated with or without TPA (10 Ϫ7 M) as indicated for 14 -16 h prior to harvesting for analysis of luciferase and ␤-galactosidase activity. Error bars represent standard deviations. The results are representative of three independent experiments. B, location of putative binding sites for PU.1/Spi-1, AP-1, CCAAT-box binding proteins, and GC-box binding proteins. C, effects of mutations in the putative cis-acting binding sites and their effects on promoter activity. Ten micrograms of each construct were electroporated into the U937 cell line. One microgram of the ␤-actin/␤-galactosidase was co-transfected and used as an internal standard. Cells were treated with and without TPA (10 Ϫ7 M). The cells were harvested 14 -16 h after TPA treatment and assayed for luciferase and ␤-galactosidase activity. Error bars represent standard deviations. The results are representative of at least three independent experiments. not have any effect in promoter activity. In concert, these studies establish the region from Ϫ221 to Ϫ1 of the macrosialin gene as a proximal promoter that is capable of conferring a transcriptional response to TPA in monocyte-like cells. At least four classes of cis-active elements appear to be required for proximal promoter activity in these cells; a GC-rich sequence at Ϫ213, two CCAAT boxes at Ϫ191 and Ϫ167, a putative binding site for PU.1/Spi-1 at Ϫ104 and a binding site for AP-1 at Ϫ131.
The Proximal Macrosialin Promoter Mediates Transcriptional Activation by GM-CSF-Based on the regulation of the endogenous macrosialin mRNA levels by GM-CSF (Fig. 1), we wished to determine whether macrosialin regulatory elements could confer a transcriptional response to GM-CSF in a model cell line. BaF/3 cells containing the murine GM-CSF ␣-receptor were transfected with the macrosialin deletion constructs (Ϫ7.0 to Ϫ221) (Fig. 7A). Each construct exhibited a transcriptional response to GM-CSF with the fold of induction being similar for the Ϫ7.0 kb and Ϫ221 bp. The results demonstrated that the proximal promoter is alone sufficient to confer a transcriptional response to GM-CSF. To determine if the same cis-acting elements that directed the TPA response in the U937 also are responsible for the GM-CSF response, the mutated versions of the proximal promoter constructs were transfected into the BaF/3-GM cells (Fig. 7B). As was observed in the U937 cells, these regions also affected the basal activity of the proximal promoter in the BaF/3-GM cells. However, only mutations in the GC-rich region exhibited a decrease in fold induction when corrected to the new basal level. The results suggest that additional sequences are primarily responsible for the GM-CSF response within proximal promoter.
Interaction of c-Jun and PU.1 with Macrosialin Regulatory Elements-To identify nuclear proteins that bind to the putative AP-1 and PU.1/Spi-1 elements present in the macrosialin promoter, electrophoretic mobility shift assays were performed. Incubation of nuclear proteins obtained from THP-1 cells with an oligonucleotide probe containing the putative PU.1/Spi-1 element resulted in several complexes, labeled Ia, Ib, and Ic in Fig. 8A. All complexes were effectively competed by the unlabeled macrosialin probe, while Ia and Ic were effectively competed by an oligonucleotide corresponding to a consensus PU.1/ Spi-1 binding site. None of these binding activities was competed for by an oligonucleotide in which the PU.1/Spi-1 site was mutated. PU.1 and Spi-1 are the murine and human orthologues of a B cell and macrophage-specific transcription factor belonging to the ets domain gene family (35,36). PU.1 and Spi-1 are equivalent in size and can be recognized by a specific antiserum raised against PU.1 binding DNA domain (32). To determine whether these complexes contained Spi-1, THP-1 nuclear extracts were incubated with this antiserum prior to addition of the radiolabeled PU.1/Spi-1 probe. As illustrated in Fig. 8B, the PU.1/Spi-1 antibody almost completely abolished complexes Ia and Ic, and a faint supershifted band was observed (lane 3). Complex Ib did not change, consistent with its failure to be competed effectively by the consensus PU.1/Spi-1 binding site. As a control, in vitro translated PU.1 was incubated with the probe and antibody was also added. The in vitro translated product of PU.1 also bound with high affinity to the macrosialin PU.1/Spi-1 site (Fig. 8B, lane 4). The major complex migrated at the same position as complex Ic, which results from partial proteolysis of full-length PU.1. These results indicate that complexes Ia and Ic contain Spi-1 and suggest that complex Ia may represent either multimers of Spi-1 or a ternary complex containing Spi-1 and other nuclear proteins.
To characterize proteins binding to the putative macrosialin AP-1 site, electrophoretic mobility shift assay experiments were performed with nuclear extracts prepared from TPAtreated THP-1 cells. Incubation of THP-1 nuclear extracts with the macrosialin AP-1 probe (lane 2) resulted in the formation of a cluster of bands that were effectively competed by the unlabeled, wild-type oligonucleotide and a consensus AP-1 binding site, but not by an oligonucleotide containing a mutation in the AP-1 binding site (Fig. 8C). Antibody raised against amino acids 56 -69 of the human c-Jun strongly inhibited formation of the majority of these complexes and resulted in a supershifted complex (Fig. 8D, lane 3, arrow). A Jun-B-specific antibody also slightly decreased the lower bands in this cluster. Addition of both c-Jun and Jun-B antibodies to the nuclear extract nearly abolished all of the specific complexes (Fig. 8D, lane 5). These observations indicate that c-Jun and Jun-B are components of the AP-1 complexes that bind to the macrosialin AP-1 element in TPA-treated THP-1 cells.
Functional Cooperation between c-Jun and PU.1 on the Macrosialin Promoter-To directly assess whether PU.1 and c-Jun can functionally cooperate to stimulate macrosialin promoter activity, experiments were performed in P19 cells, which lack c-Jun and PU.1. The Mac 221-luciferase construct was cotransfected into P19 cells with expression vectors containing cDNAs for either c-Jun or PU.1 (Fig. 9). Cells were treated with and without TPA for 14 -16 h and harvested 48 h after transfection. Cells that were transfected with the Ϫ221 construct and an empty expression vector (pcDNA-3) demonstrated very FIG. 7. Macrosialin proximal promoter directs tissue-specific activity in response to GM-CSF. A, the indicated macrosialin regulatory elements directing the expression of a luciferase reporter gene were transfected into BaF/3-GM cells as described under "Experimental Procedures." An equimolar amount of each construct were electroporated. One microgram of the SV40/␤-galactosidase construct was cotransfected and used as an internal standard. Cells were treated with and without murine GM-CSF (4 ng/ml) as indicated and harvested for luciferase and ␤-galactosidase activities 48 h later. B, effects of mutations in putative cis-acting binding sites and their effects on promoter activity. Five micrograms of each construct were electroporated into the BaF/3-GM cell line. 500 nanograms of the SV40/␤-galactosidase construct was co-transfected and used as an internal standard. Cells were treated with and without GM-CSF (4 ng/ml). The cells were harvested 48 h after GM-CSF treatment and assayed for luciferase and ␤-galactosidase activities. Error bars represent standard deviations. The results are representative of at least three independent experiments. low levels of promoter activity in the presence or absence of TPA (Fig. 9). Co-transfection of the PU.1 expression vector did not significantly alter promoter activity (Fig. 9). Similar results were obtained over a wide range of PU.1 expression vector concentrations (data not shown). When the c-Jun expression vector was co-transfected with the Ϫ221 construct, the basal level of promoter activity also did not significantly change; however, a 10-fold induction was seen when TPA was added. Significantly, co-expression of PU.1 and c-Jun resulted in synergistic increases in both basal and TPA-dependent transcription.

DISCUSSION
Macrosialin is a transmembrane glycoprotein of uncertain function that has been shown to be highly expressed in macrophages and dendritic cells (9). Consistent with these observations, macrosialin mRNA was observed in murine bone marrow progenitor cells and was markedly up-regulated by M-CSF and GM-CSF, factors that promote the proliferation and differentiation of the monocyte-macrophage lineage. These observations suggest that the macrosialin gene will be a useful model for the investigation of molecular mechanisms that control early events in macrophage differentiation.
In the present studies, we have cloned the macrosialin gene and have performed an initial characterization of its transcriptional regulatory elements. A fragment of the macrosialin gene containing the promoter and 7.0 kb of 5Ј-flanking information was demonstrated to direct high levels of reporter gene activity in several monocyte-like cell lines. Maximal levels of promoter activity in U937 cells required the presence of enhancer elements located between Ϫ7.0 and Ϫ2.5 kb of the translational start site. This region also contained regulatory elements that inhibited promoter activity in nonmyeloid cells. It will thus be of interest to identify the proteins that bind to these regulatory elements and determine their roles in controlling macrosialin expression in vivo.
In comparison to several other myeloid-specific promoters that we have evaluated, including the scavenger receptor A promoter (32), the macrosialin promoter is very active, directing levels of reporter gene activity that are approximately 100 times higher than the scavenger receptor A promoter. In addition to facilitating the analysis of transcriptional regulatory elements, the relative strength and specificity of the macrosialin promoter is likely to have practical applications in transgenic animal experiments in which the macrosialin promoter is used to overexpress genes of interest in macrophages and dendritic cells.
Consistent with the finding that GM-CSF up-regulates macrosialin mRNA in bone marrow progenitor cells, the macrosialin promoter and 7.0 kb of 5Ј-flanking information conferred a strong positive transcriptional response to GM-CSF in BaF/3 cells transfected with the ␣ subunit of the GM-CSF receptor. Furthermore, we have determined that the proximal 221 bp of promoter information are sufficient to mediate the GM-CSF response, indicating that the proximal promoter is likely to play an important role in the GM-CSF-dependent induction of macrosialin transcription, but do not appear to correspond to elements required for phorbal ester responsiveness.
The macrosialin promoter and 5Ј-flanking sequences also conferred positive transcriptional responses to the phorbol ester, TPA, which induces macrophage differentiation of THP-1, U937, and HL60 cells. TPA regulates the activities of several classes of transcription factors, including AP-1 proteins, as a consequence of stimulating protein kinase C (37). Among other events, protein kinase C activates the Raf/mitogen-activated protein kinase pathway, which has been suggested to play an important role in directing the proliferation and differentiation of macrophage progenitor cells in response to M-CSF (37)(38)(39)(40)(41)(42). TPA-dependent induction of the macrosialin promoter was inhibited by several cytokines and regulatory molecules that influence macrophage development and function. Inhibition of TPA-dependent expression of macrosialin by retinoic acid is of interest because of the potent effects of retinoic acid as an inducer of granulocyte differentiation in several myeloid leukemic cell lines (31,43), which would be anticipated to lead to repression of the endogenous macrosialin gene. The inhibitory effects of IFN-␥, bacterial LPS, and TNF-␣ on macrosialin promoter activity are very similar to the inhibitory effects of these substances on the SR-A gene (44), suggesting that macrosialin and SR-A are coordinately regulated in terminally differentiated macrophages. Mice lacking the TNF receptor R1 (p55) develop more atherosclerosis on a high fat diet than did control animals (45). Thus, the effects of TNF-␣ on macrosialin and SR-A expression may be relevant to the development of atherosclerosis if these proteins play important roles in the uptake of oxidatively modified lipoproteins. Hsu et. al. (46) demonstrated that down-regulation of the macrophage scavenger receptor was not due to a transcriptional decrease, but was mainly due to the destabilization of the macrophage scavenger receptor mRNA (46).
Characterization of the proximal macrosialin promoter indicates that at least four classes of cis-active elements are required for full activity. Mutation of a GC-box at Ϫ213 bp reduces promoter activity by approximately 10-fold. This element is recognized by a DNA binding activity that is present in many cell types and confers enhancer activity to a heterologous minimal promoter. 2 Antibody-perturbated gel shift studies, competition experiments with consensus binding sites, and electrophoretic mobility shift assays with recombinant proteins have thus far excluded NF-B, SP-1, SP-3, AML-1, and Egr-1 as potential factors binding to these elements in U937 cells. Two putative CCAAT boxes have also been identified within the macrosialin promoter. Mutation of these regions also reduced the activity of the Ϫ221 promoter by 10-fold. We are currently investigating whether the two CCAAT bases between Ϫ191 and Ϫ162 are binding sites for CCAAT binding proteins such as C/EBP or CREB-binding protein (CBP).
Several lines of evidence indicate that AP-1 and PU.1/Spi-1 cooperate to activate macrosialin transcription via regulatory elements in the proximal macrosialin promoter. Mutations in either the PU.1/Spi-1 or AP-1 binding sites reduced activity of the Ϫ221 bp promoter by nearly a factor of 10. Antibodyperturbated gel shift experiments confirmed that PU.1/Spi-1 and c-Jun are indeed components of protein complexes that bind to the PU.1 and AP-1 elements, respectively. Furthermore, coexpression of PU.1 and c-Jun led to synergistic transcriptional activation of the Ϫ221 macrosialin promoter in P19 cells.
These observations are consistent with the proposed roles of PU.1/Spi-1 and c-Jun in regulating critical aspects of macrophage development and function. PU.1/Spi-1 is a B cell and macrophage-specific transcription factor (35) that has been demonstrated to activate several genes that are selectively expressed in these cell types (32,47,48). PU.1/Spi-1 has been found in CD34 ϩ cells, indicating that it is present when macrosialin first becomes expressed (49,50). Disruption of the PU.1 gene results in complete absence of B cells and macrophages, indicating its requirement for the development of these lineages (51). Intriguingly, PU.1 has been found to be located at or near the transcriptional start site of several TATA-less promoters and to interact with TATA binding protein (52,53). The unusual structure of the macrosialin promoter, in which the PU.1 site lies between a downstream consensus Inr element and an upstream Inr element that is only 1 bp removed from a consensus TATA box, may permit several options for transcriptional initiation and account in part for the relative strength of the macrosialin promoter.
In the present studies, high and low mobility protein DNA complexes were abolished by the anti-PU.1 antibody. These observations raise the possibility that the low mobility complex consists of a ternary complex between PU.1 and another factor. PU.1 has previously been demonstrated to form a ternary complex with the lymphocyte-specific factor NF-EM5/Pip or regulatory elements present in the immunoglobulin 3Ј enhancer (54,55). It will be of interest to determine whether analogous factors exist in macrophages that function to enhance PU.1 activity on macrophage-specific genes.
Recent studies have also implicated c-Jun as an important factor in mediating at least some of the actions of M-CSF in macrophages (reviewed in Roussel (56)). c-Jun binds as a homodimer or a heterodimer with other basic leucine zipper proteins to AP-1 elements in a large number of genes that are activated in macrophages in response to M-CSF, and has been found to activate a number of these genes in cotransfection assays. In the case of the SR-A gene, mutation of the AP-1 binding sites abolishes the transcriptional response of the promoter to M-CSF (57). In addition to mediating the positive transcriptional effects of M-CSF, AP-1 factors have also been proposed to be targets of negative regulation by IFN-␥ and retinoic acid. Recent studies suggest that transcriptional activation by c-Jun, STAT1␣, and retinoic acid receptor requires the recruitment of coactivator complexes that contain CBP or p300 (57,58). CBP and p300 appear to be present in ratelimiting amounts in cells, suggesting that competition for these complexes may account for antagonistic interaction between activators of AP-1, IFN-␥, and the retinoic acid receptor. These observation also raise the possibility that cooperative recruitment of CBP⅐p300 complexes by several locally bound transcription factors could potentially account for synergistic interaction between pathways. It will therefore be of interest to determine the mechanistic basis for synergy between PU.1 and AP-1.