Cooperation between core binding factor and adjacent promoter elements contributes to the tissue-specific expression of interleukin-3.

Tissue-specific expression of interleukin-3 (IL-3) is mediated via cis-acting elements located within 315 base pairs of the transcription start. This is achieved in part through the positive activities of the AP-1 and Elf-1 sites in the IL-3 promoter. The contribution to T cell-specific expression by other promoter sites was assessed in a transient expression assay with IL-3 promoter constructs linked to a luciferase gene, focusing initially on the core binding factor (CBF) site, which is footprinted in vivo upon T cell activation. Activity of the CBF site is shown to be critically dependent on the adjacent activator site Act-1. Together the Act-1 and CBF sites form a functional unit (AC unit) with dual activity. The AC unit is demonstrated to enhance basal activity of promoters both in fibroblasts and T cells. This activity is further inducible in activated T cells, but not in fibroblasts. In addition to the already identified NIP repressor site, evidence is presented for a second repressor region that restricts promoter activity in fibroblasts. Finally, a novel positive regulatory element has been mapped in the IL-3 promoter between nucleotide -180 and -210 that leads to increased expression in T cells. Together these results demonstrate that T cell expression of IL-3 is not specified by the activity of a single tissue-specific element, but instead involves multiple interacting elements that provide both specific positive regulation in T cells and specific negative regulation in fibroblasts.


Tissue-specific expression of interleukin-3 (IL-3) is mediated via cis-acting elements located within 315 base pairs of the transcription start. This is achieved in part through the positive activities of the AP-1 and Elf-1 sites in the IL-3 promoter. The contribution to T cell-specific expression by other promoter sites was assessed in a transient expression assay with IL-3 promoter constructs linked to a luciferase gene, focusing initially on the core binding factor (CBF) site, which is footprinted in vivo upon T cell activation. Activity of the CBF site is shown to be critically dependent on the adjacent activator site Act-1. Together the Act-1 and CBF sites form a functional unit (AC unit) with dual activity. The AC unit is demonstrated to enhance basal activity of promoters both in fibroblasts and T cells. This activity is further inducible in activated T cells, but not in fibroblasts.
In addition to the already identified NIP repressor site, evidence is presented for a second repressor region that restricts promoter activity in fibroblasts. Finally, a novel positive regulatory element has been mapped in the IL-3 promoter between nucleotide ؊180 and ؊210 that leads to increased expression in T cells. Together these results demonstrate that T cell expression of IL-3 is not specified by the activity of a single tissue-specific element, but instead involves multiple interacting elements that provide both specific positive regulation in T cells and specific negative regulation in fibroblasts.
Interleukin-3 (IL-3) 1 is involved in the proliferation and differentiation of hematopoietic progenitor cells (1). Consistent with its potent biologic activity, IL-3 gene expression is highly regulated and is restricted to human T and NK cells (2). Therefore, the IL-3 promoter has been extensively studied, and multiple regulatory elements have been identified within 315 bp of the transcription start site (3)(4)(5)(6). These elements include sev-eral activator sites: AP-1, Elf-1, and Act-1 (NFIL3); a repressor site, NIP; and a "permissive" site that binds core binding factor (CBF).
The activity of individual regulatory elements has been extensively studied in the context of multiple promoters/enhancers (7)(8)(9)(10)(11). However, the emphasis on identifying novel regulatory elements has prevented a complete understanding of the functional interactions between elements already identified. These interactions are important as they represent another possible level of regulatory activity and may in part specify the differential expression of genes despite impressive similarities of their promoters. For example, the promoter for both granulocyte-macrophage-colony-stimulating factor (GM-CSF) and IL-3 bind many similar proteins but exhibit different patterns of expression, with GM-CSF being produced by fibroblasts, macrophages, and endothelial cells (4,5,8,10). Therefore, the activity of a particular site may vary according to the context within which it is found, as adjoining regulatory sites may modulate its activity (4,12).
The IL-3 promoter is well suited to address possible functional interactions between different regulatory elements, since a relatively small region of DNA contains several well characterized sites that are important outside the context of IL-3 regulation. We initially focused on the normal function of CBF, since mutations of this protein complex are associated with specific subtypes of acute myelogenous leukemia (AML) and myelodyspastic syndrome (MDS) (13)(14)(15)(16)(17). In the present study we demonstrate that the CBF site and adjacent Act-1 element form a regulatory unit that provides inducible promoter activity in Jurkat T cells, but not in NIH3T3 fibroblasts. We also demonstrate that two additional upstream portions of the IL-3 promoter inhibit promoter activity in NIH3T3 cells. Finally, an initial characterization of a previously unknown positive regulatory region of the IL-3 promoter between nt Ϫ180 and nt Ϫ207 is provided.

MATERIALS AND METHODS
Cell Culture-The human Jurkat T cell line E6.1 and NIH3T3 cells were obtained from ATCC and maintained in 5% CO 2 at 37°C. Jurkat T cells were cultured in RPMI medium supplemented with penicillin (50 units/ml), streptomycin (50 g/ml), and 10% fetal calf serum (RPMI complete). NIH3T3 cells were cultured in DMEM supplemented with penicillin, streptomycin, 1 mM sodium pyruvate, and 10% fetal calf serum (DMEM complete). Jurkat cells were stimulated with a combination of TPA (10 ng/ml) and ionomycin (0.5 M). NIH3T3 cells were stimulated with TPA alone.
IL-3 Promoter Constructs-Isolation of the Ϫ315-bp IL-3 promoter and insertion into an IL-3 reporter gene has already been described (18). Similarly, G-C mutation of the CBF site of the IL-3 promoter has been described (18). Briefly, guanine Ϫ139, Ϫ137, and Ϫ136 within the CBF binding site were selectively mutated to cytosine using standard polymerase chain reaction methodology. This mutation abrogates CBF protein binding. Flanking HindIII sites were added to transfer the full-length Ϫ315-bp promoter, or G-C mutant (Ϫ315/GC), to the lucif-erase reporter gene previously described by DeWet et al. (Ϫ315pL, Ϫ315/GCpL) (19). An independent, higher yield luciferase reporter gene (pFlash*) was derived from the pFlash vector (SynapsSys Corp., Burlington, MA) by excision of the downstream polylinker between ApaI and SmaI, followed by replacement of the SstI-XbaI (luciferase nt 57) pFlash upstream polylinker with the equivalent fragment from the luciferase reporter gene, pXp1 (pFlash*). The IL-3 promoter was then inserted at the new upstream HindIII site (Ϫ315pF*/Ϫ315/GCpF*). All reporter genes used in a particular experiment were derived from the same root plasmid. Specific constructs are depicted schematically within individual figures and were derived as follows.
A second CBF site (CBF*) was created at nt Ϫ192 using standard polymerase chain reaction methodology to introduce a C-G and T-G mutation at nt Ϫ191 and nt Ϫ189, respectively. Replacement of the DNA fragment between nt Ϫ180 and nt Ϫ207 with an AC unit was done via two intermediate cloning steps. The resultant promoter had an inserted oligonucleotide, 5Ј-atggATGAATAATTACGTCTGTGGTTTtcta-, that contained an AC unit (upper case letters) in place of the 27-bp MseI (nt Ϫ207)ϪSau3a (nt Ϫ180) fragment in the Ϫ315pL expression vector (Ϫ315NACpL). The equivalent replacement was made in an IL-3 promoter containing a G-C mutation at the endogenous CBF site (Ϫ315NAC/GCpL).
AC units were also inserted at nt Ϫ180 of the Ϫ315 promoter in Ϫ315pL to create promoters without deletion of endogenous nucleotides. AC cassettes contained mutations in none, one, or both members of the AC unit. The individual cassettes were synthetic oligonucleotides obtained from the Dana Farber Cancer Institute core facility (Table IA).
Addition of AC units containing either wild type or mutant Act-1 and CBF sites to a minimal (Ϫ61 bp) IL-3 promoter was accomplished via replacement of the IL-3 promoter upstream of nt Ϫ61 with one of the synthetic oligonucleotides summarized in Table IB. This was accomplished using the Ϫ315pF* vector digested at the 5Ј SstI site of the Ϫ315pF* polylinker and the 3Ј SmaI site at nt Ϫ61 of the IL-3 promoter (Ϫ61pF*). This construct contained 61 bp of the IL-3 promoter upstream of the transcription start site, including the TATA box at nt Ϫ30. The various Act-1-CBF oligonucleotides were inserted in place of the 254 deleted nucleotides between nt Ϫ61 and nt Ϫ315.
Intermediate length IL-3 promoters were derived from Ϫ315pF* via deletion of upstream regions of the promoter from nt Ϫ173 (ScaI), nt Ϫ180 (Sau3a) or nt Ϫ210 (MseI). All constructs were confirmed by restriction analysis. Any construct made using polymerase chain reaction techniques or utilizing synthetic oligonucleotides was also confirmed by DNA sequencing. Plasmid DNA was twice purified over cesium chloride prior to transfection.
Transient Transfection-Jurkat T cells were transfected with 1 g of DNA/10 6 cells using either the DEAE-dextran method as described (Stratagene, La Jolla, CA) or by electroporation in serum free conditions using a 300-V, 980-F, and 14-ms discharge (PG 200 Progenitor 2, Hoeffer Scientific Instruments, San Francisco, CA). Cells were cultured in RPMI medium supplemented as above for 16 h after transfection and were then stimulated with a combination of TPA and ionomycin. Cell lysates were harvested 6 -9 h after stimulation and assayed for luciferase activity (Analytical Luminescence Laboratory, San Diego, CA).
NIH3T3 cells were grown in DMEM supplemented as above and transfected at 20% confluence with 15 g of DNA/10 cm 2 tissue culture plate (Becton Dickinson, Lincoln Park, NJ) using a standard calcium phosphate method. Sixteen to 24 h after adding CaP, plates were washed three times with calcium/magnesium-free phosphate-buffered saline. Cells were grown overnight in fresh DMEM complete prior to stimulation with TPA. Stimulation of NIH3T3 cells with a combination of TPA and ionomycin decreased expression of all reporter constructs as compared with unstimulated or TPA-stimulated cells. Cell extracts were obtained 6 -9 h after stimulation and assayed for luciferase activity (Analytical Luminescence Laboratory).
Normalizing data of a test reporter gene according to the relative activity of a second, ␤-galactosidase, gene is a standard methodology. However, we 2 and others have found that the measured ␤-galactosidase activity (or that of another co-transfected construct) is dependent on several factors other than efficiency of transfection (20,21). Therefore, it is an unreliable method for normalization. It has also been our experience that spectrophotometry has not quantitated plasmid DNA as precisely as expected and that the efficiency of transfection is critically dependent on the purity of the transfected plasmid. Therefore, we perform replicate experiments using independently isolated and purified plasmids. All luciferase constructs in a particular experiment are: (a) derived from the same root plasmid, (b) differ only by a small number of nucleotides, (c) are twice purified over cesium chloride, and (d) quantitated by both spectrophotometry and on agarose gel. This is consistent with the methods proposed by others (21). Using this procedure, replicate transfections using independent isolates of the same luciferase construct have yielded differences in luciferase activity of Ͻ10%. This is more consistent, and accurate, than normalizing data to the independently variable activity of a co-transfected second construct. Luciferase activity was measured as relative light units (RLU) (Monolight Luminometer model 2010). All data are presented as the mean relative luciferase activity as calculated by the formula (RLU test construct / RLU reference construct ). Error Bars indicate 1 standard deviation of the mean relative luciferase activity.
Eectrophoretic Mobility Shift Assay-Nuclear extracts were prepared from either Jurkat T cells or NIH3T3 cells using the Andrews procedure and were used in gel shift assays as described (22). Labeling of the oligonucleotide probe was accomplished either by filling the 5Ј ends using DNA polymerase large fragment (Klenow) in the presence of dATP, dGTP, dTTP, and [␣-32 P]dCTP (Ͼ3,000 Ci/mmol; DuPont NEN) or by phosphorylation of the oligonucleotide probe at the free 5Ј-hydroxyl group using polynucleotide kinase and [␥-32 P]ATP (Ͼ3,000 Ci/ mmol; DuPont NEN). Labeled probes were purified by gel electrophoresis. A 50 -100 molar excess of competitor oligonucleotides were added to the binding reaction 15-20 min prior to the probe. The AP-1-specific oligonucleotide was kindly provided by N. C. Andrews. Antiserum specific for NFATp and Oct-1 was kindly provided by A. Rao and G. R. Crabtree, respectively.

RESULTS
The CBF Site within the Human IL-3 Promoter Is Positionsensitive-We previously demonstrated that CBF is necessary for IL-3 promoter activity (3). To further characterize the functional role of CBF, a new site (CBF*) was introduced 51 bp upstream of the endogenous location and tested for its capacity to rescue the activity of an IL-3 promoter mutated at the resident CBF site. These experiments were carried out in the human Jurkat T cell line JKE6.1. As indicated in Fig. 1A, there is little promoter activity in unstimulated Jurkat cells 2 D. S. Taylor, unpublished data.

TABLE I Synthetic oligonucleotides used in the synthesis of IL-3 promoter constructs
1A: synthetic oligonucleotides used to insert various AC units at nt Ϫ180 of the full-length (Ϫ315 bp) IL-3 promoter. 1B: synthetic oligonucleotides used to place various AC units immediately upstream of the minimal (Ϫ61 bp) IL-3 promoter. The portion of the oligonucleotide sequence corresponding to the wild type or mutated Act-1 and CBF sites is depicted with uppercase letters.
5Ј-SstI-cTCGAGAAGGTTGGATCCTGTGGTTTtccc-SmaI (Ϫ315pL). Upon stimulation with TPA and ionomycin, there is up-regulation of promoter activity, which is significantly reduced by a G-C mutation of the endogenous CBF binding site (Ϫ315/GCpL). Introduction of a second CBF site (Ϫ315*CBFpL) neither increased the base-line promoter activity nor completely rescued activity of the CBF/GC mutant promoter (Ϫ315*CBF/GCpL). To ensure that the created *CBF site bound the CBF complex, gel shift assays were performed. Both the CBF and the *CBF sites bind a protein complex with a similar mobility and cross-compete with one another for binding to that complex (Fig. 1B). Equivalent results were obtained using nuclear extract from both stimulated and unstimulated Jurkat cells (data not shown). Since the *CBF site is capable of binding the CBF protein complex, but does not alter promoter activity, the location of the CBF site in the IL-3 promoter appears critical for its biologic activity. The Act-1 and CBF Sites Form a Functional Unit-The endogenous CBF site within the IL-3 promoter is interesting as it is juxtaposed to another important regulatory element, Act-1. Such juxtaposition of CBF and another regulatory site is a recurring motif in other promoter/enhancer elements (Table  II), and suggests a functional interdependence between the CBF and Act-1 sites in the context of the IL-3 promoter.
To evaluate the functional interdependence of the Act-1 and CBF sites, a second Act-1/CBF (AC) unit, or a mutant thereof, was introduced upstream of the wild type site at nt Ϫ180. Introduction of either the wild type, or one of several mutants of the AC unit, did not involve the deletion of endogenous nucleotides and resulted in extension of the IL-3 promoter by 34 -42 bp. In contrast to our previous experiments with promoters containing the *CBF site (Ϫ315*CBFpL), the AC unit significantly up-regulates promoter activity (Fig. 2) in Jurkat T cells upon stimulation with TPA and ionomycin. Insertion of this unit also resulted in a tendency to increase the basal (constitutive/unstimulated) activity of the IL-3 promoter. In addition, the inserted AC unit overcomes the inhibition of promoter activity caused by a G-C mutation of the endogenous CBF site. Mutation of either the Act-1 or CBF component of the AC unit resulted in a sharp decrease in promoter activity. Thus, in Jurkat T cells the Act-1 and CBF sites function only as a unit (not individually) to provide inducible promoter activity.
The AC Unit Is Functional in a Minimal Promoter-Having established the functional interdependence of the Act-1 and CBF sites, we next evaluated its function outside the context of an intact, full-length proximal IL-3 promoter. To this end, a normal AC unit or an AC unit containing a mutant CBF site was inserted 31 bp upstream of the TATA box in the Ϫ61-bp minimal IL-3 promoter. As seen in Fig. 3A, the minimal promoter has significant activity in unstimulated Jurkat cells compared with the full-length Ϫ315-bp proximal IL-3 promoter. This activity is further augmented by stimulation with TPA and ionomycin, and addition of a complete AC unit increases both basal and stimulated activity of the minimal promoter. The enhancement in activity was abrogated upon introduction of a G-C mutation to the CBF site. Thus, the AC unit is functional both in the absence of upstream elements within the IL-3 promoter and within 31 bp of the TATA box.
An AC Unit Is Functional in Heterologous Cells-The activity of the full-length, Ϫ315-bp proximal IL-3 promoter is highly tissue restricted. It is functional only in human T and NK cells (2). Such observed tissue-specific activity may result from specific positive regulation in a permissive cell type, specific inhibition in nonpermissive cells, or a combination thereof. Previous work in this laboratory identified a T cell-specific complex that, upon activation, binds the 5Ј portion of the Act-1 region, suggesting that the AC unit may contribute to the observed Sixteen hours after transfection cells were mock-stimulated or stimulated with a combination of TPA and ionomycin. Six hours after stimulation luciferase activity was measured in cell lysates. Luciferase activity was measured as RLU. The data from three independent experiments are presented as the mean relative luciferase activity as calculated by the formula (RLU test construct /RLU reference construct ). Error bars indicate 1 standard deviation of the mean relative luciferase activity. The stimulated wild type IL-3 promoter (Ϫ315pL) is the reference construct in this experiment and has a relative luciferase activity of 1.0. B, nuclear extracts were prepared from Jurkat T cells. The nuclear extracts were incubated with the indicated competitor oligonucleotide and an oligonucleotide probe corresponding to either the endogenous CBF site or the newly introduced CBF site (*CBF). The CBF binding site is indicated by a bracket.
selective activity of the IL-3 promoter (18). Therefore, the activity of the minimal IL-3 promoter with, and without, a functional AC unit was compared with the full-length Ϫ315-bp IL-3 promoter in heterologous, murine NIH3T3 cells (Fig. 3B). In contrast to the full-length IL-3 promoter, the minimal IL-3 promoter has significant activity in unstimulated NIH3T3 cells. The activity of the minimal promoter in these fibroblasts is enhanced approximately 2-fold by the addition of a single AC unit. The AC unit increases basal (unstimulated) promoter activity when upstream of the minimal (Ϫ61 bp) IL-3 promoter in both Jurkat T-cells and NIH3T3 fibroblasts (Fig. 3, A and B). However, the AC unit increases the stimulation index (stimulated/unstimulated) in the context of Jurkat T cells but not NIH3T3 fibroblasts. By virtue of its inducible positive activity in T cells, the AC unit contributes to the tissue-specific regulation of IL-3 gene expression.
Both the Act-1 and CBF segments of the AC unit bind protein complexes common to nuclear extracts from both Jurkat and NIH3T3 cells (Fig. 4). The CBF segment of the AC unit binds a major complex in NIH3T3 cells that migrates and competes identically to the previously identified CBF protein complex present in nuclear extracts from Jurkat T cells (Fig. 4A). The more diffuse appearance of this complex in NIH3T3 cells may result from more than one closely migrating complex and suggests the presence of more than a single isoform of CBF in these cells (23,24). The Act-1 region of the AC unit binds to a slowly migrating complex present in both cell types (Fig. 4B). This band was previously identified in this laboratory as Oct-1 or an Oct-1-like protein. It is possible that the faster migrating complex that is specifically competed by Act-1 in the Jurkat nuclear extract represents the recently identified E4BP4 protein, NFIL3A (25).
A Positive Regulatory Element (CK3) Is Present between nt Ϫ180 and nt Ϫ207 of the IL-3 Promoter-Insertion of the AC unit into the full-length proximal IL-3 promoter resulted both in elongation of the promoter and a tendency to increase basal (unstimulated) promoter as noted in Fig. 2. As the NIP repressor element maps to the upstream region of the IL-3 promoter, it was possible that the inserted nucleotides diminished its effect by increasing the distance between NIP and other important downstream elements. To investigate this possibility, the promoter fragment between nt Ϫ180 and nt Ϫ207 was deleted and replaced with an AC unit. This region was chosen as it contained no previously identified promoter elements. As shown in Fig. 5, constancy of the distance between NIP and other downstream elements of the IL-3 promoter restores a low basal activity of the IL-3 promoter in the presence of an additional AC unit. Furthermore, mutation of the endogenous CBF is fully rescued in the context of the upstream AC unit (data not shown). However, deletion of the region between nt Ϫ180 and nt Ϫ207 causes a decrease in overall promoter activity, suggesting that a previously unidentified positive regulatory element is perturbed in this promoter construct.
To further analyze this region, the activity of a Ϫ180-bp IL-3 promoter was compared with the Ϫ207 and Ϫ315 IL-3 promoter constructs in both Jurkat and NIH3T3 cells (Fig. 6). In all experiments performed, regardless of cell type, inclusion of the 27 bp between nt Ϫ180 and nt Ϫ207 induced a significant increase in promoter activity.
Two Specific Complexes Are Bound by the CK3 Element-The coding strand of DNA between nt Ϫ180 and nt Ϫ207 bears no homology to known regulatory elements. However, the noncoding strand contains sequences that are similar, but not identical, to known binding sites for nuclear factor of activated T cells (NFAT), AP-1, and the 3Ј portion of Act-1 (Table III). To investigate protein binding to the IL-3 promoter in this region (CK3), gel shift assays were performed using as a probe both a restriction fragment of the IL-3 promoter containing these nucleotides (MseI-Sau3a) and a synthetic oligonucleotide containing the nucleotides of interest (Fig. 7A). Both probes bind a faster migrating complex (CK3a) and a slower migrating doublet (CK3b).
The faster migrating protein complex (CK3a) is specifically competed by unlabeled "self" oligonucleotide and nucleotides containing the complete Act-1 site, but is not competed by an oligonucleotide containing only the 5Ј portion of the Act-1 site  16 h after transfection cells were mock-stimulated or stimulated with a combination of TPA and ionomycin. 6 h after stimulation luciferase activity was measured in cell lysates. Data from three independent experiments are presented as outlined in Fig. 1. The stimulated wild type IL-3 promoter (top construct) is the reference construct in this experiment and has a relative luciferase activity of 1.0. (Fig. 7A). Oct-1 is known to bind to the 3Ј portion of the Act-1 site, but antiserum specific for Oct-1 does not supershift CK3a (data not shown) (18). Also, despite similarity to an AP-1 site, the CK3a complex migrates much more rapidly than the AP-1 complex (Fig. 7A) and the same AP-1-specific oligonucleotide does not compete for CK3a binding (Fig. 7B). Protein binding associated with the slower migrating doublet (CK3b) is specifically inhibited by unlabeled oligonucleotide containing either the 5Ј portion of the NFAT region of the IL-2 promoter (IL2NFAT) or the NFAT-like sequence of the IL-3 promoter (IL3NFAT) (Fig. 7B). However, antiserum specific for NFATp does not supershift the CK3b complex (data not shown). Thus, the CK3 region of the IL-3 promoter identifies proteins that bind DNA with specificities that are similar to, but distinct from, Oct-1 and NFAT. DISCUSSION IL-3 is a lymphokine important for normal hematopoiesis (1). Its expression is highly tissue-specific, being expressed primarily in T and NK cells (2). Therefore, its promoter has been extensively studied (3)(4)(5)(6). Most recently, in vivo footprinting of the IL-3 promoter identified a CBF binding site as one of the few regions specifically protected upon stimulation of T cells (3). The initial characterization of this site indicated that CBF binding to the IL-3 promoter is necessary, but not sufficient, for IL-3 promoter activity (3).
The recent identification of mutant CBF proteins created by the t(8;21), t(3;21), and inv(16) chromosomal rearrangements and the association of these rearrangements with specific subtypes of AML or MDS has encouraged intense study of the mutant CBF fusion proteins (14 -16, 26). However, the understanding of normal CBF function remains incomplete. The presence of a functional CBF site within the IL-3 proximal promoter provided an ideal context in which to more closely evaluate the activity of this important regulatory complex.
The present data demonstrate that the CBF complex cannot function alone. This is supported by the observation that a second CBF binding site (*CBF) that binds the CBF complex in a gel shift assay provides neither positive promoter activity nor overcomes a mutation of the endogenous CBF site (Fig. 1). In contrast, a CBF site in conjunction with the adjacent Act-1 region of the IL-3 promoter both augments promoter activity and overcomes a null mutation of the endogenous CBF site. Both effects are abrogated by mutation of either the Act-1 or CBF component of the unit (Fig. 2). The AC unit functions similarly in the context of a minimal (Ϫ61 bp) IL-3 promoter (Fig. 3). Thus, the AC unit is an autologous functional unit that retains its activity outside the context of a full-length IL-3 promoter and requires both the Act-1 and CBF sites to be intact.
The AC unit functions in both human T cells (Jurkat) and murine fibroblasts (NIH3T3), and nuclear extracts from both cell types contain proteins that bind similarly to both the Act-1 and CBF sites of the AC unit (Fig. 4). In the context of T cells, the AC unit augments both basal (constitutive/unstimulated) and inducible (stimulated) promoter activity. However, the augmentation of inducible promoter activity in T cells is greater than the augmentation of basal activity as indicated by an increase in the stimulation index of an IL-3 promoter containing a second AC unit (Figs. 2 and 3). This effect is not

FIG. 3. The Act-1/CBF unit enhances transcription activity of a minimal promoter in both Jurkat T cells and NIH3T3 fibroblasts.
A, Jurkat T cells were transfected with the indicated promoter constructs and stimulated as described in the legend to Fig. 1. An AC unit containing either a wild type or mutant CBF binding site was appended to the 5Ј portion of the minimal (Ϫ61 bp) IL-3 promoter as outlined under "Materials and Methods" and compared with the minimal IL-3 promoter. Data from five independent experiments are presented as outlined in Fig. 1. The stimulated minimal (Ϫ61 bp) promoter (second construct) is the reference construct in this experiment and has a relative luciferase activity of 1.0. In one of five experiments the full-length (Ϫ315 bp) proximal IL-3 promoter (bottom construct) was included for the purpose of comparison. B, murine NIH3T3 fibroblast cells were transfected with the indicated promoter constructs using the calcium phosphate method. Cells were grown for 24 h after transfection, then stimulated for 7 h with TPA only as stimulation of NIH3T3 cells with a combination of TPA and ionomycin substantially decrease activity of all promoter constructs. Cell lysates were isolated and assessed for luciferase activity. Data from four independent experiments are presented as outlined in Fig. 1. The stimulated minimal (Ϫ61 bp) promoter (second construct) is the reference construct in this experiment and has a relative luciferase activity of 1.0. The Ϫ173 bp IL-3 promoter construct was present in one of four experiments. observed in NIH3T3 cells (Fig. 3). Thus, the AC unit contributes to the tissue-specific expression of IL-3 by its selective positive contribution to inducible promoter activity in the context of T cells.
The mechanism underlying the interdependence of the Act-1 and CBF sites is unknown. As shown for other DNA binding factors, it may involve alteration in DNA configuration via bending (27). Alternatively, CBF may function as a chaperon protein and augment/inhibit binding of other proteins to the promoter. For example, several isoforms of CBF have been identified and found to have variable activity in vitro (24,28). Thus, the capacity of CBF to modulate the association between other proteins, such as the Act-1-binding proteins, and DNA may be dependent on the relative amount of the various CBF isoforms under various conditions. Previous in vivo footprinting of the IL-3 promoter is consistent with this hypothesis. These data demonstrate protection of the CBF site only upon stimulation, despite similar amounts of CBF binding activity in nuclear extracts obtained from either stimulated or unstimulated Jurkat cells (3). This hypothesis is further supported by recent data that demonstrate a similar functional interdependence between PEBP2␣, the murine homologue of CBF, and Ets-1, which binds an adjacent regulatory element in the T cell receptor ␣ chain enhancer (29).
Although the AC unit contributes to the specific up-regulation of the IL-3 promoter activity in T cells, as evidenced by its effect on a minimal promoter, the full-length (Ϫ315 bp) IL-3 promoter is not active in fibroblasts. Therefore, additional elements between nt Ϫ61 and nt Ϫ315 of the IL-3 promoter contribute to tissue-specific promoter activity by limiting activity in non-T cells. As suggested by our results in Fig. 3B, there are at least two elements that inhibit IL-3 promoter activity in NIH3T3 cells. One element resides between nt Ϫ173 and nt Ϫ315 (Fig. 3B). The presence of this upstream region of the IL-3 promoter is also associated with a low basal activity of the promoter in Jurkat T cells (Fig. 3A). The NIP site of the IL-3 promoter is located at nt Ϫ266, is a negative regulator of transcription, and has been extensively studied in this laboratory (5,30). The relative proximity of the upstream NIP region of the IL-3 promoter to the AP-1 (nucleotide Ϫ303) and Elf-1  5. Replacement of IL-3 promoter between nt ؊180 and nt ؊207 decreases promoter activity. Jurkat T cells were transfected with the indicated promoter constructs and stimulated as described in Fig. 1. The NAC constructs contain a second AC unit and flanking nucleotides that replaces the endogenous IL-3 promoter between nt Ϫ207 and nt Ϫ180. Data from two independent experiments are presented as outlined in Fig. 1. The stimulated wild type Ϫ315-bp IL-3 promoter (second construct) is the reference construct in this experiment and has a relative luciferase activity of 1.0.
(nucleotide Ϫ288) elements is intriguing. The latter two regions are known positive regulators of promoter activity, and Elf-1 has been implicated in the specific positive regulation of the IL-3 promoter in T cells (4). Thus, the AP-1⅐Elf-1⅐NIP complex may represent a second example within the IL-3 promoter of a functional interdependence between adjacent regulatory regions. A second element is present between nt Ϫ61 and nt Ϫ173. This element functionally negates the positive contribution of the endogenous AC unit, which is evidenced by a decrease in activity of the Ϫ173-bp promoter as compared with the Ϫ61-bp promoter containing an AC unit. A highly con-served CK1/CK2 DNA sequence is present in this region and may be important in this activity, since a concatamer of this sequence has been shown to inhibit the basal transcription of a heterologous promoter and proteins from the transcriptionally active Rel/NFB family bind to the CK1 element (31). The combined effect of these two negative elements is abrogation of IL-3 promoter activity in the context of NIH3T3 cells.
The second observation resulting from the study of the AC unit, and immediately relevant to IL-3 promoter function, is the identification of a previously unknown positive regulatory region between nt Ϫ180 and nt Ϫ207 (Fig. 6). This region of the IL-3 promoter is 67-100% conserved between human, new world monkey, sheep, rat, and mice (32). Furthermore, the majority of differences within these 27 bp represent conservative changes. Preliminary characterization of Ϫ180/Ϫ207 region by gel shift analysis demonstrates that at least two protein complexes bind independently to this region. Although there are sequence similarities between the noncoding strand of this region and the Act-1/Oct-1, AP-1, and NFATp sites, preliminary experiments suggest that these factors are not involved in binding to these 27 nucleotides. In addition to Oct-1, the E4BP4 protein, NFIL3, binds to the 3Ј portion of the Act-1 site (6,33). Although the 3Ј portion of the Act-1 site is important for competition of CK3a binding, the sequence of the CK3a site contains only 5 (5Ј-ATTAC-) of 10 nucleotides that define the E4BP4 consensus site (25). Antiserum specific for the NFIL3 protein is not available, so the relationship between E4BP4/ NFIL3 and the observed binding at the CK3a region is unknown at this time. Regardless, a more complete mutational analysis of this region will be necessary to confirm that the protein complexes identified by this gel shift analysis are as- FIG. 6. Positive regulatory activity maps between nt ؊180 and nt ؊207.

Jurkat T cells (A) and NIH3T3 cells (B)
were transfected with the indicated promoter constructs and stimulated as described in Fig. 3. Data from four independent experiments are presented as outlined in Fig. 1. The stimulated Ϫ207-bp IL-3 promoter is the reference construct in this experiment and has a relative luciferase activity of 1.0.

TABLE III The IL-3 promoter contains conserved sequences between nt
Ϫ184 and nt Ϫ206 The noncoding strand of the human IL-3 promoter is compared with both AP-1/Act-1 sequences (IL3CK3a) and the NFAT binding sequence found in either the intergenic region of the IL-3/GM-CSF locus or the IL-2 promoter (IL3CK3b).
Ϫ207 Խ 5Ј-TTAAGTAATCTTTTTTCTTGTTTCA--AATTCATTAGAAAAAAGAACAAAGT-5Ј In conclusion, we demonstrate that the Act-1 and CBF sites of the IL-3 promoter form a functional unit. We further demonstrate that the tissue-specific activity of the proximal IL-3 promoter results from the combined activity of this unit and other regulatory elements within the promoter rather than from a single, tissue-specific protein. The contribution of the Act-1/CBF unit to this activity is important as it mediates inducible promoter activity in the context of T cells (but not fibroblasts). It remains important to determine how, or whether, mutant forms of the CBF protein, relevant to AML and MDS, alter this tissue-specific activity. However, the recognized association of the M4Eo subtype of AML with both a mutation in the ␤-chain of CBF and eosinophilia indirectly suggests that IL-3, a positive regulator of eosinophil differentiation, is abnormally up-regulated in these cells.
FIG. 7. The IL-3 promoter between nt ؊180 and nt ؊207 contains two protein binding regions. Nuclear extracts were prepared from Jurkat T cells that were stimulated for 6 h with a combination of TPA and ionomycin. A, the extracts were incubated with the competitor oligonucleotide indicated above each lane and one of three probes indicated below the lane. AP1, tggggaacctgtgcTGAGTCActggag; MseI-Sau3a, TTAAGTAATCTTTTTTCTTGTTTCACT; or CK3, agctTGAAA-CAAGAAAAAAGATTACTTAg. B, nuclear extracts were incubated with the competitor oligonucleotide indicated above each lane and with the CK3 probe. Sequences of competitor oligonucleotide are as follows: IL3NFAT, agctTGAAACAAGAAAAAAG; IL2NFAT, agcttGAAAGGAG-GAAAAAg; ACT-CBF, gatcctaatggATGAATAATTACGTCTGTGGTT-Tctagagct; Act, catggATGAATAATTACGTCTGca; Act-5Ј, agcttgcAT-GAATTAgagccc; CBF, gatcACGTCTGTGGTTTTCTATG. Uppercase letters are used to designate nucleotides corresponding to portions of the IL-3 promoter. Relevant binding sites within oligonucleotides derived from exogenous promoters are also noted in uppercase letters.