An Enhancer Located between the Neutrophil Elastase and Proteinase 3 Promoters Is Activated by Sp1 and an Ets Factor*

The adjacent neutrophil elastase, proteinase 3, and azurocidin genes encode serine proteases expressed specifically in immature myeloid cells. Subclones of a 17-kilobase (kb) murine neutrophil elastase genomic clone were assessed for their ability to stimulate the neutrophil elastase promoter in 32D cl3 myeloid cells. Region −9.3 to −7.3 kb stimulated transcription 7-fold, whereas other genomic segments were inactive. This enhancer is located in the second intron of the proteinase-3 gene and so may regulate more than one gene in the myeloid protease cluster. Deletional analysis of the enhancer identified several segments which activated the neutrophil elastase and thymidine kinase promoters 3–6-fold. The most active segment was a 220-base pair region centered at −8.6 kb, which activated transcription 31-fold. This segment contains an Sp1 consensus site, which bound Sp1, flanked by two Ets family consensus sequences, which bound PU.1, GABP, and an Ets factor present in myeloid cell extracts. Mutation of the Sp1-binding site reduced enhancer activity 8-fold in 32D cl3 cells, and mutation of either or both Ets-binding sites reduced activity 3–4-fold. Sp1 activated the distal enhancer 5-fold, GABP 3-fold, and the combination 8-fold in Schneider cells.

To identify transcriptional events that determine myeloid cell determination, we have been investigating the regulation of the neutrophil elastase (NE) 1 gene. NE is a microbicidal serine protease present in the primary granules of granulocytes and monocytes (1). In bone marrow, NE mRNA is present mainly in promyelocytes (2), and transcriptional regulation plays a key role in its lineage-specific expression (3). We isolated a 17-kb genomic murine NE clone and found that, like the human NE gene, the murine gene contains five exons and initiates transcription 30 bp downstream from a TATAA ho-mology (4). We demonstrated that the 5Ј-flanking region of the NE gene contains a 60-bp proximal enhancer, located just upstream of the TATAA homology, which is strongly active in 32D cl3 myeloid cells differentiating in response to granulocyte-colony-stimulating factor (G-CSF), but only weakly active in uninduced 32D cl3 cells or in fibroblasts (4). This proximal enhancer is activated cooperatively by C/EBP, c-Myb, CBF, and Ets family members such as PU.1 or GABP (4 -6). The NE gene is located just downstream of the proteinase-3 (PR-3) gene, which in turn is located just downstream of the azurocidin gene (7). These three serine protease genes are expressed specifically and coordinately in immature myeloid cells (7), and the promoter of each contains conserved binding sites for C/EBP, c-Myb, and PU.1 (7,8).
To determine whether the murine NE gene contains additional regulatory regions, we assessed the activity of NE genomic subclones functionally after transient transfection into 32D cl3 cells. A 2-kb NE genomic region extending from Ϫ7.3 to Ϫ9.3 kb stimulated the NE promoter 7-fold, and stimulated the herpes TK promoter as well, whereas none of the other subclones, from Ϫ7.3 to ϩ7.5 kb, were active. Recent results indicate that this enhancer maps to the second intron of the adjacent PR-3 gene, 2 and so might stimulate both the NE and PR-3 promoters in vivo. We therefore designate it the myeloid protease enhancer. Functional analysis identified a 220-bp "core" enhancer, centered at Ϫ8.6 kb, which activated transcription 15-31-fold. Deletional analysis of this segment identified an Sp1-binding site and two flanking Ets familybinding sites, which contributed to its activity in 32D cl3 cells. Also, Sp1 and GABP cooperatively activated transcription via these sites in Schneider cells. Notably, Sp1 had been shown previously to cooperate with Ets factors to regulate several myeloid promoters.
32D cl3 cells proliferating in IL-3 were transiently transfected by a DEAE-dextran procedure as described, using 10 -15 g of reporter DNA (4). After transfection, the 32D cl3 cells were cultured in G-CSF or split between IL-3-and G-CSF-containing media. NIH 3T3 cells were transfected by calcium phosphate precipitation using 3 g of reporter DNA, followed 18 h later by a 3.5-min, 15% glycerol shock in phosphatebuffered saline (10). Schneider cells were transfected by calcium phosphate precipitation with 5 g of reporter DNAs and 2.5 g of effector DNAs. pMSV-CAT (0.25-0.5 g) or pCMV-␤Gal (1 g) were employed as internal control plasmids where indicated. Total cell extracts were prepared 42 h after transfection and analyzed for luciferase and chloramphenicol acetyltransferase or ␤-galactosidase activities as described (4 -6).
Nuclear Extracts and EMSA-32D cl3 and U937 nuclear extracts were prepared as described (5,6). GST-PU.1, GST-GABP␣, and GST-GABP␤ were expressed in Escherichia coli and isolated with glutathione-Sepharose as described (6). Recombinant human Sp1 was obtained from Promega. EMSA was carried out as described (5,6). In brief, 10 -12 g of nuclear extract, 1 footprinting unit of Sp1, or approximately 100 ng of the GST fusion proteins were incubated on ice for 30 min with 1 ng of an oligonucleotide probe, which had been radiolabeled with [␣-32 P]dCTP by Klenow fill-in of 5Ј-overhangs. For detection of Sp1 in 32D cl3 nuclear extracts, the binding solution was 10 mM Tris (pH 7.5), 50 mM KCl, 70 mM NaCl, 100 g/ml poly(dI-dC), 0.1 mg/ml bovine serum albumin, 0.5 mM EDTA, 1 mM dithiothreitol, 10% glycerol. For the U937 extracts, recombinant Sp1, or the GST fusion proteins, the binding solution was 10 mM Tris (pH 7.5), 50 mM NaCl, 1 mM EDTA, 50 or 100 g/ml poly(dI-dC), 1% Ficoll. 10 -100-Fold excess of unlabeled competitor oligonucleotides were added 5 min prior to probe addition, and 1 l of Sp1 antisera (Santa Cruz Biotechnology) or Sp1 antisera and specific peptide were added 30 min prior to probe addition. Reaction mixtures were then subject to electrophoresis on 5% acrylamide gels in 0.33-0.5 ϫ Tris borate-EDTA at 180 -200 V. The gels were then dried and subjected to autoradiography.
Plasmids and Oligonucleotides-A restriction map of the murine NE genomic locus is presented in Fig. 1A, top panel. The 1.8-kb BamHI-NcoI segment located just upstream of the first exon was studied previously (4). Six other segments, designated I-VI, were subcloned upstream of the proximal enhancer and the luciferase reporter, by insertion between the HindIII and XhoI sites of pNE(mϪ107/ Ϫ103)LUC (4). Segments of NE-I were subcloned upstream of the proximal enhancer similarly. These segments are diagrammed in Fig.  1A, bottom panel. Division of NE-I.11 into NE-X, NE-Y, and NE-Z was facilitated by insertion of BglII sites into NE-I.11 by site-directed mutagenesis after annealing mutagenic oligonucleotides and singlestranded DNA as described (4). pNE-XY(Ϫ103)LUC contains both NE-X and NE-Y upstream of the NE promoter and the luciferase reporter, and plasmids containing other DNA segment combinations are designated similarly. The sequence of pNE-Z, the core protease enhancer, is given in Fig. 1B. NE-Z was further divided into NE-A, NE-B, and NE-C by insertion of BamHI sites at bp 76 -81 or 157-162 using polymerase chain reaction mutagenesis. The final polymerase chain reaction products were sequenced to confirm that no additional mutations had been introduced. A 39-bp segment of NE-C, NE-C1, was ligated upstream of the NE promoter as a synthetic double-stranded oligonucleotide. Several mutant versions of NE-C1 were ligated similarly. The sequences of the top strand of these oligonucleotides are given below. Each has a 4-bp 5Ј-overhang compatible with HindIII, and the bottom strand had a similar overhang compatible with XhoI. The wildtype or mutant Ets and Sp1 core motifs are underlined, 5Ј-AGCTTGGCCTCAAGCATTAAGGGGCGG-GGGTTGGATTGGCGATC-3Ј NE-C1-mGG1AA2Sp5: 5Ј-AGCTTGGCCTCAAGCATTAAGGTTTTT-GGGTTGGATTGGCGATC-3Ј NE-C1, NE-C1-mSp5, and NE-C1-mGG1AA2 were also ligated just upstream of the TATAA homology in the NE promoter by insertion into HindIII/XhoI digested pNE(mϪ47/Ϫ42)LUC (4). The 109-bp herpes simplex virus TK promoter was ligated upstream of the luciferase reporter in p19LUC (11) to create ptk-LUC. NE genomic segments NE-I.11 and NE-Z were ligated upstream of the TK promoter in this plasmid to create pNE(I.11)tkLUC and pNE(Z)tkLUC. I.11 was also subcloned in reverse orientation in front of the 103-bp NE promoter region, to generate pNE-I.11r(Ϫ103)LUC. The sequences of oligonucleotides containing known Sp1-and Etsbinding sites, and of an irrelevant oligonucleotide, used as competitors in EMSA are given below.

Sp1:
5Ј-ATTCGATCGGGGCGGGGCGAGC-3Ј Ets: For trans-activation experiments, Sp1 and the GABP subunits were expressed using pPac-Sp1 (12) (gift from R. Tjian, University of California, Berkeley, CA) and pPac-GABP␣ ϩ pPac-GABP␤ (gift from N. Speck, Dartmouth University, Hanover, NH). Fig. 1A diagrams our murine NE genomic clone. The five NE exons span 1.7 kb (13). The 1.8-kb region located just upstream of the first exon was studied previously and shown to contain a 60-bp proximal NE enhancer located between Ϫ90 and Ϫ30, just upstream of the TATAA homology (4). Six additional genomic regions were positioned upstream of Ϫ103 in the NE promoter linked to the luciferase reporter. The activities of these constructs were assessed relative to the Ϫ103 promoter segment alone after transient transfection into 32D cl3 myeloid cells and culture in G-CSF for 2 days ( Fig. 2A). Region I stimulated transcription 7.4-fold, on average, whereas the other regions were inactive. Several segments of Region I were then also subcloned upstream of the promoter and luciferase and analyzed similarly. These segments are diagrammed in the bottom panel of Fig. 1A, and the results of transient transfection analysis are shown in Fig. 2B. I.1 stimulated transcription 6.5-fold, whereas I.2 only increased activity 2.2-fold. Also, when I.2 was divided into I. 21 and I.22, neither of these DNA segments stimulated the NE promoter. On the other hand, subdomains of I.1, I.11 and I.12, both retained significant activity. I.11 activated transcription 14-fold, I.12 activated transcription 6-fold, and both together stimulated activity 17-fold. Thus, the 1.4-kb genomic region contained in I.11 and I.12 is the smallest portion of the 16.8-kb murine NE clone, which has enhancer activity. As this enhancer maps in the second intron of the adjacent PR-3 gene, 2 we designated it as the myeloid protease enhancer.

Identification of a Distal NE Enhancer-
We chose to focus on I.11, the most active segment of this region, for further analysis. Using BglII sites inserted by sitedirected mutagenesis, region I.11 was divided into domains X, Y, and Z. Domain X stimulated the NE promoter 4-fold, domains X and Y together stimulated activity 2.5-fold, and domains Y and Z stimulated activity 3-fold. Strikingly, domain Z alone stimulated activity 31-fold. Apparently domain X is minimally active, domain Y is actually repressive, and domain Z is very stimulatory. We designate domain Z as the core protease enhancer, realizing that nearby regions X and I.12 may also contribute to the activity of the NE enhancer. The sequence of domain Z is given in Fig. 1B.
To determine whether the myeloid protease enhancer could function in either orientation, region I.11 was positioned in reverse orientation upstream of the NE promoter and luciferase. I.11 activated transcription 8-fold in the forward orientation and 18-fold in the reverse orientation in induced 32D cl3 cells (Fig. 2C).
To determine whether the enhancer could activate a heterologous promoter, regions I.11 and Z were positioned upstream of the herpes simplex TK promoter. Region I.11 stimulated the TK promoter 5-fold, and region Z stimulated the promoter 10-fold in induced 32D cl3 cells (Fig. 2C).
The NE promoter region alone was 5-fold more active when transfected 32D cl3 cells were cultured for 48 h in G-CSF, compared with IL-3 (4). To determine whether the myeloid protease enhancer might also be more active in G-CSF, the activity of several of these constructs was also assessed in IL-3 (Fig. 2C). When the observed increased activities of the NE and TK promoters in cells cultured in G-CSF compared with IL-3 are taken into account, we conclude that the protease enhancer is equally active in 32D cl3 cells cultured in G-CSF or in IL-3.
The activity of the myeloid protease enhancer was also assessed in NIH 3T3 cells. Region I.11 activated the NE promoter 8-fold in induced 32D cl3 cells and 3-fold in NIH 3T3 cells. Region Z activated the promoter 23-fold in 32D cl3 cells (the average of the data in Fig. 2B and Fig. 3A) and 7-fold in NIH 3T3 cells. Thus, the protease enhancer was more active in 32D cl3 myeloid cells than in NIH 3T3 fibroblasts. The activity observed in NIH 3T3 cells could indicate binding by factors not normally expressed in myeloid cells or that do not have access to the enhancer in the chromatin of non-myeloid cells.
Functional Analysis of the Core Protease Gene Enhancer-Region Z was divided into segments A, B, and C by inserting BamHI sites in place of bp 76 -81 or 157-162. Segments ABC (NE-Z), A, AB, AC, BC, and C were then assessed for their ability to enhance the activity of the NE promoter region in induced 32D cl3 cells (Fig. 3A). Segment A only activated transcription 2-fold, when assayed alone or when added to segment C. Segment B increased the activity of segment A 3-fold and of segment C 2-fold. On the other hand, segment C activated the promoter 10-fold alone and 6-fold with segment A. Thus the 75-bp segment C was more stimulatory than seg-ments A or B. To map the active region of segment C, segment C1 was synthesized and linked to the proximal enhancer. Segment C1 activated transcription 12-fold (Fig. 3A), indicating that this 39-bp region was critical to the activity of the core enhancer. Of note, introduction of a BamHI site at bases 157-162, between segments A and B, disrupted a consensus site for CBF. This mutation did not decrease the activity of the core enhancer. 3 Inspection of the sequence of C1 revealed an Sp1 consensus site, GGGCGG, flanked by two GGAA motifs, which are potential binding sites for Ets family members. Oligonucleotides containing 2-bp mutations in each of these sites were synthesized and linked to the NE promoter. The 5Ј-GGAA was mutated to TTAA (GG1), and the 3Ј-GGAA was mutated to GGTT (AA2), so that the introduced mutations would be separate from the Sp1 consensus. A double GGAA mutant version of C1 was also prepared (GG1/AA2). Similarly, the Sp1 consensus was mutated to GTTCGG (Sp2) in an effort to avoid interfering with the GGAA sites. This oligonucleotide still weakly bound Sp1 (data not shown), and so a 5-bp Sp1 site mutation, GTTTTT (Sp5), was also prepared. Finally, an oligonucleotide containing mutations in both GGAA sites and a 5-bp mutation in the Sp1 site were also ligated into the basal reporter. The activities of these constructs relative to that of C1 linked to the NE promoter were then assessed in induced 32D cl3 cells (  3B). Mutation of the 5Ј-GGAA reduced activity 3.6-fold, mutation of the 3Ј-GGAA reduced activity 2.6-fold, and mutation of both Ets sites reduced activity 4.4-fold. The 2-or 5-bp mutation in the Sp1 site had a similar effect, reducing activity 7.6-fold, and mutation of all three sites reduced activity 6.3-fold. Thus, integrity of the Sp1 site and both Ets sites is important for the activity of segment C1 in the core myeloid protease enhancer in myeloid cells.
Sp1 and Ets Family Members Bind C1-A nuclear extract from induced 32D cl3 cells was prepared and incubated with radiolabeled C1. The reaction was then resolved on a 5% native acrylamide gel (Fig. 4A). Two gel shift species were observed. Both were competed by 10-or 50-fold excess of unlabeled C1 or of a consensus Sp1 site. A C1 oligonucleotide carrying a 2-bp mutation in the Sp1 site competed far less efficiently, whereas the GG1 or AA2 mutations did not prevent efficient competition. When Sp1 antisera was added to the reaction, the more slowly migrating gel shift species was supershifted efficiently, whereas the lower species was only mildly affected. The specific peptide prevented the supershift. Thus, C1 binds Sp1 via the GGGCGG consensus and may bind a less abundant, Sp1-related protein or degradation product present in 32D cl3 nuclear extracts as well. No additional bands, which might correspond to Ets family members, were observed, even on long exposures of the autoradiographs. In particular, we easily detected binding of PU.1 to the NE proximal enhancer (5), and C1 only weakly competed for this binding (data not shown).
To further characterize proteins capable of binding C1, we

FIG. 2. Functional identification of a myeloid protease enhancer.
A, the activities of genomic regions I-VI linked to the NE promoter at bp Ϫ103 and to the luciferase reporter were assessed relative to the promoter alone, the activity of which was set as 1 in each experiment. All transfections included pMSVCAT as an internal control. B, the activities of segments of genomic region I were assessed similarly. C, plasmids containing segment I.11 linked in the forward or reverse (I.11r) orientation to the NE promoter (NE) in pNE(⌬Ϫ103)LUC, or segments I.11 or Z linked to the tk promoter (TK) in ptkLUC, were transiently transfected into 32D cl3 cells proliferating in IL-3. Transfected cells were then split into IL-3 or G-CSF and cultured for an additional 2 days. Luciferase and chloramphenicol acetyltransferase activities were then determined. The activity of each construct is shown relative to that of the NE or TK promoter alone in G-CSF. The activity of region I.11 was assessed also in NIH 3T3 cells, and the activity of region Z linked to the NE promoter was assessed in 32D cl3 cells cultured in G-CSF and in NIH 3T3 cells. The mean and S.E. from three experiments are shown. also employed several purified proteins and U937 nuclear extracts (Fig. 4B). Purified Sp1 bound efficiently (lane 2), and binding was also detected using bacterially expressed GST-PU.1, GST-GABP␣, or GST-GABP␣ with GST-GABP␤ (lanes [3][4][5]. GABP is an Ets family member, which also binds and activates the NE promoter (6). Several bands were detected in EMSA using a U937 extract, including a prominent doublet (open arrows) reminiscent of the doublet detected using 32D cl3 extracts (lane 6). Interestingly, competition with a consensus Sp1 oligonucleotide inhibited binding by all of the bands except for a minor band present between the doublet (lane 8, filled arrow). And competition with an oligonucleotide derived from the CD18 promoter, which is known to bind PU.1 and GABP (14), prevented binding by this minor species but not by the major doublet (lane 9). An irrelevant oligonucleotide competitor had no effect. These results indicate that the Ets factors PU.1 and GABP can bind C1. They also identify an endogenous Ets factor that can bind C1. This endogenous factor was not PU.1 or GABP, as it did not supershift with the corresponding antisera (data not shown).
We next explored the possibility that the endogenous Ets factor, which bound C1, might do so more efficiently if the Sp1-binding site is mutated (Fig. 4C). Using radiolabeled C1 and a U937 extract in an EMSA (lane 2) we again identified a minor gel shift species (dark arrow) between the two Sp1related species (open arrows). When the C1-Sp5 oligonucleotide was used as a probe with the same extract, the Sp1 species were not detected, and the minor band was more prominent (lanes 4 and 7). Binding by this species was competed more efficiently by the homologous probe (lanes 5 and 6) than by C1-mGG1/AA2, carrying mutations in both Ets consensus sites (lanes 8 and 9). Thus, Sp1 and an Ets factor present in U937 extracts can simultaneously bind C1. Of note, when purified Sp1 and GST-PU.1 or GST-GABP␣/␤ were used in an EMSA with C1 as probe, no additional, more slowly migrating species was detected, indicating that Sp1 and PU.1 or GABP did not bind C1 simultaneously under these conditions. 3 C1 Is Activated Cooperatively by Sp1 and GABP-Sp1 is abundant in mammalian cells. Therefore, we employed Drosophila Schneider cells for trans-activation experiments. Initial experiments suggested that Sp1 could activate the proximal NE enhancer in pNE(⌬Ϫ103)LUC. Therefore, for these experiments we positioned C1 just upstream of the TATAA homology in the NE promoter, followed by the luciferase reporter. The empty pPac Drosophila expression vector did not affect the activity of this reporter. 4 Sp1 did not activate the NE promoter in pNE(⌬Ϫ42)LUC alone, but activated the C1 segment of pC1-NE(Ϫ42)LUC 3-fold, on average (Fig. 5A). Neither PU.1 nor GABP alone activated segment C1 in Schneider cells, and PU.1 did not increase stimulation by Sp1. On the other hand, the combination of GABP and Sp1 stimulated pC1-NE(Ϫ42)LUC 8-fold, but did not stimulate pNE(⌬Ϫ42)LUC. Thus, Sp1 and GABP cooperated to stimulate transcription via the enhancer segment C1 in Schneider cells.
To verify that cooperation between Sp1 and GABP was dependent upon binding to the Sp1-and Ets-binding sites in C1, additional transfection experiments were carried out using re-4 C. P. Simkevich, M. Luo, and A. G. Rosmarin, unpublished data.  4 and 7), or with 10-or 100-fold excess of C1-mSp5 (lanes 5 and 6) or C1-mSp5ϩmGG1/AA2 containing mutations in the Sp1 and in the two Ets sites (lanes 8 and 9). porters carrying either C1, C1-mGG1/AA2, or C1-mSp5, linked to NE(Ϫ42)LUC (Fig. 5B). In this set of experiments, Sp1 activated C1 5-fold, and Sp1 ϩ GABP activated C1 12-fold. Mutation of the two Ets-binding sites did not prevent activation by Sp1, but prevented additional activation by GABP. Mutation of the Sp1-binding site prevented activation by Sp1, alone or with GABP. DISCUSSION We previously studied 1.8 kb of the murine NE 5Ј-flanking region and identified a proximal NE enhancer within the first 100 bp, which activated transcription 300-fold (4). We have now assessed the ability of an additional 15 kb of the murine NE genomic locus to enhance transcription from the proximal NE enhancer/promoter in 32D cl3 myeloid cells as they differentiate in response to G-CSF. Only a 2-kb region centered at Ϫ8.3 kb was active, and this region includes a highly active 220-bp core enhancer, which activated the proximal enhancer/promoter 30-fold in 32D cl3 cells.
Using a similar functional approach we identified an enhancer at Ϫ3 kb in the murine myeloperoxidase (MPO) gene, which was also localized using a DNase I hypersensitivity assay (15,16). A distal enhancer in the MyoD gene was also identified by functional assay of multiple genomic subclones (17). We made numerous attempts to identify DNase I hyper-sensitive sites within the murine NE locus. Although most were unsuccessful, two experiments indicated a hypersensitive site in the region that we have now identified as functionally important. 3 The NE, PR-3, and MPO genes are transcribed specifically in immature myeloid cells (2-4, 18 -20). The MPO distal enhancer is regulated by PU.1 and C/EBP (21) and perhaps by CBF as well (15). Besides NE and MPO, the only other gene expressed specifically in myeloid cells known to contain a distal enhancer is the chicken lysozyme gene. The lysozyme enhancer, which is most active in mature macrophages, is also activated by PU.1 (22).
Expression of the linked NE, PR-3, and azurocidin myeloid serine protease genes is coordinately regulated during myeloid differentiation (6). This coordinate regulation might be accounted for by conserved PU.1-, C/EBP-, and c-Myb-binding sites within their promoters (4,7,8,23), as we found that these three transcription factors cooperatively activate the NE proximal enhancer (4,5). Recent genetic mapping studies indicate that the most 3Ј PR-3 exon is located 3 kb upstream of the first NE exon, placing the myeloid protease enhancer within the large PR-3 second intron, 1 kb downstream of the PR-3 promoter. 2 Perhaps coordinate regulation of the myeloid protease gene cluster also results from activation via this enhancer. Determining whether the myeloid protease enhancer activates the NE and/or PR-3 promoters in vivo will be a key question for our future investigations. As only 10 kb separate these two promoters, it should prove feasible to develop transgenic murine lines in which this question can be addressed. On the other hand, transient transfection assays are not expected to identify insulator or other chromatin boundary elements, which might prevent this enhancer from activating the NE or PR-3 promoters.
This core protease enhancer activated the NE proximal region 5-fold in NIH 3T3 cells, and unlike the proximal NE enhancer, its activity was not increased when 32D cl3 cells were transferred from IL-3 to G-CSF. Perhaps in the context of chromatin, greater tissue and developmental specificity would be observed.
We found that integrity of an Sp1-binding site and of flanking Ets factor-binding sites were essential for the activity of the core protease enhancer. Although ubiquitously expressed, Sp1 was found to be particularly highly expressed in developing murine granulocytic cells (24). Sp1 was shown to be essential for the myeloid-specific activity of the CD11b and CD14 promoters (25,26) and for induction of CD14 expression during monocytic differentiation (27). Sp1 has also been implicated in the regulation of the c-fes, human MPO, and CD18 promoters in myeloid cells (28 -30). Sp1 DNA binding activity did not increase during U937 or THP-1 monocytic differentiation (25,27), and we similarly did not detect a change in Sp1 DNA binding activity during 32D cl3 granulocytic differentiation. 3 Interaction with other factors may allow Sp1 to participate in transcriptional induction during myeloid differentiation.
The active Sp1-and PU.1-binding sites in the CD11b promoter are separated by 40 bp, and in the c-fes promoter they are separated by about 50 bp (25,28,31). In the PTHrP promoter and in the human T-cell lymphotrophic virus, type I long terminal repeat, the central base pairs of the Sp1-and Ets1binding sites are separated by 15 bp (36,37). In the myeloid protease enhancer, only 6 bp separates the centers of the GGAA motifs and the Sp1 consensus. Sp1 binding partially excluded Ets factor binding from U937 nuclear extracts. Simultaneous binding by Sp1 and GABP to the protease enhancer was not observed in vitro, but this assay may not reflect in vivo conditions. Also, the growth hormone gene promoter contains overlapping Sp1-and Pit-1-binding sites, both of which are required for full promoter activity despite mutually exclusive binding in vitro (39). Perhaps, in vivo additional proteins mediate cooperative binding by Sp1 and Ets factors to the protease enhancer. Sp1 and GABP cooperatively activated the myeloid protease enhancer via their binding sites in Schneider cells, although the Ets factor which cooperates with Sp1 in vivo to activate the enhancer remains to be determined. GABP alone did not activate the enhancer. Similarly, AML1 and C/EBP␣ synergistically activate the M-CSF receptor gene, but C/EBP␣ alone is inactive (40).
Most functional Sp1 sites have been identified in promoter regions. However the immunoglobulin , collagen I, collagen II, ␤-globin, CD13, and gb110 genes contain more distal, functional Sp1 sites (41)(42)(43)(44)(45)(46). We anticipate that further characterization of the myeloid protease enhancer and its interaction with the NE and PR-3 promoters in vivo will contribute to our understanding of early myeloid differentiation and of transcriptional mechanisms which enable coordinate gene activation.