Multiple transcription factors are required for activation of human interleukin 9 gene in T cells.

The genetic elements and regulatory mechanisms responsible for human interleukin 9 (IL-9) gene expression in a human T cell leukemia virus type I-transformed human T cell line, C5MJ2, were investigated. We demonstrated that IL-9 gene expression is controlled, at least in part, by transcriptional activation. Transient expression of the luciferase reporter gene linked to serially deleted sequences of the 5'-flanking region of the IL-9 gene has revealed several positive and negative regulatory elements involved in the basal and inducible expression of the IL-9 gene in C5MJ2 cells. An AP-1 site at -146 to -140 was shown to be involved in the expression of the IL-9 gene. A proximal region between -46 and -80 was identified as the minimum sequence for the basal and inducible expression of the IL-9 gene in C5MJ2 cells. Within this region, an NF-kappaB site at -59 to -50 and its adjacent 20-base pair upstream sequence were demonstrated to play a critical role for the IL-9 promoter activity. DNA-protein binding studies indicated that NF-kappaB, c-Jun, and potentially novel proteins (around 35 kDa) can bind to this important sequence. Mutations at different sites within this proximal promoter region abolished the promoter activity as well as the DNA binding. Taken together, these results suggest that the cooperation of different transcription factors is essential for IL-9 gene expression in T cells.

Interleukin 9 (IL-9) 1 is a T cell-derived cytokine with pleiotropic activities on various cell types (1). The expression of IL-9 is detectable mainly in activated CD4ϩ helper T cells (2,3). The targets of IL-9 include T cells (1,4), B cells (5,6), mast cells (7,8), erythroid and myeloid precursors (9,10), and fetal thymocytes (11). It has been shown that IL-9 can promote the proliferation of activated and transformed T cells, the production of immunoglobulins by B cells, the proliferation and differentiation of mast cells, and erythroid progenitors. The involvement of IL-9 in tumorigenesis has also been suggested, since: 1) the preferential expression of IL-9 was found in cell lines derived from patients with Hodgkin's disease and ana-plastic large cell lymphoma (12,13), 2) IL-9 was shown to be a proliferation inducer and a major anti-apoptotic factor for mouse thymic lymphomas (14,15), and 3) the ratio of occurrence of lymphoma by mutagen or x-ray irradiation increased dramatically in transgenic mice with dysregulated expression of IL-9 (16). Therefore, it is important to unveil the mechanisms of IL-9 gene regulation in the hematopoietic system.
Although it has been shown that the production of IL-9 in T cells can be induced by mitogen phorbol ester or anti-CD3 antibody (3), little is known about the control mechanisms of IL-9 expression at the transcriptional level. Human IL-9 gene has been mapped to the long arm of chromosome 5 at 5q31-32 (17,18). Interestingly, human IL-3, IL-4, IL-5, GM-CSF, and IL-13 gene clusters have also been localized to the same locus (19,20). The expression of IL-3, IL-4, GM-CSF, and IL-5 has been well studied in a variety of systems. Certain portions within the promoter regions of these genes were found to be highly conserved, and a common regulatory mechanism was suggested to be involved in the coordinated expression of these cytokines in certain cell types (21,22). It was noted that the 5Ј-flanking region of the IL-9 gene contains several putative transcriptional elements, some of which are also identified in the promoter regions of other cytokine genes mapped to the same chromosomal location (3,18). However, unlike other cytokines, up to now, the expression of the IL-9 gene has only been detected in T cells. To understand the basis of the T cell-restricted expression of IL-9, we undertook the characterization of IL-9 upstream regulatory sequences.
Our previous studies have demonstrated that a 0.9-kilobase SmaI-SacI fragment of the 5Ј-flanking region of the IL-9 gene contains the DNA sequences required for the basal and inducible expression of the IL-9 gene in a human T cell leukemia virus type I-transformed T cell line, C5MJ2 (18). In the present study, we further defined the cis-acting DNA elements and characterized the trans-acting transcription factors involved in IL-9 expression following 12-O-tetradecanoylphorbol-13-acetate (TPA)/phytohemagglutinin (PHA) stimulation. By using deletion and site-directed mutagenesis, we identified several DNA sequences in the 5Ј-flanking region of the IL-9 gene as important regulatory elements. Several transcription factors, including AP-1, NF-B, and potentially novel DNA-binding proteins (around 35 kDa), were found to play critical roles in the regulation of IL-9 gene expression in C5MJ2 cells.

MATERIALS AND METHODS
Northern Blot Analysis and Nuclear Run-on Assays-Total RNA from C5MJ2 cells were prepared at various time points after stimulation with TPA (5 ng/ml) and PHA (10 g/ml) (23). 10 g of total RNA was used for Northern blot and hybridized with a 32 P-labeled IL-9 cDNA probe. A human ␤-actin cDNA probe was hybridized to the same filter to ensure equal loading in each lane. The expression level of the IL-9 message was quantitated by densitometric scanning of IL-9 signal relative to that of the ␤-actin internal control.
For nuclear run-on assays, the labeled run-on transcripts from 3 ϫ 10 7 C5MJ2 cells were prepared as described previously (24). 5 g of plasmids containing the inserts for human GM-CSF (as a positive control), human IL-9, human ␤-actin, and pBR322 (no insert) were slot-blotted onto nylon membranes and hybridized with the same amount of labeled run-on products.
Primer Extension, RNase Protection, and DNA Sequencing-The transcriptional initiation site of human IL-9 in C5MJ2 cells was determined by both primer extension and RNase protection methods as described (25). For the primer extension assay, 50 g of C5MJ2 total RNA was hybridized to 1 ϫ 10 5 cpm of a 32 P-labeled oligonucleotide primer. The sequence of the synthetic 35-base oligonucleotide primer is 5Ј-TGGCCTGCCACGGAGCACAGGAGCAGGGCAGAGGT-3Ј, which is complementary to a region within the first exon of the human IL-9 gene (18). The antisense RNA probe for the RNase protection assay was generated by in vitro transcription of ScaI-linearized p5928 (18) using T7 RNA polymerase as described previously (26). 1 ϫ 10 6 cpm of antisense RNA probe was hybridized to 50 g of C5MJ2 total RNA. The RNA-RNA hybrids were treated with RNase A and RNase T1, followed by proteinase K digestion, and finally fractionated on an 8% urea sequencing gel.
Plasmid Construction-DNA fragments containing various portions of the 5Ј-flanking region of the human IL-9 gene were generated either by restriction endonuclease digestion or by polymerase chain reaction (PCR) and were subcloned into the SmaI/SacI sites of pXP2, a firefly luciferase reporter gene expression vector (27). The nested deletion constructs were designated as plasmids pIL-9-D1 to pIL-9-D13. The exact position (relative to the transcription start site) of deletion in each construct is shown in Figs. 3 and 6A. Other mutation constructs, such as pIL-9-M1 to pIL-9-M4, pIL-9-AP1w, and pIL-9-AP1m, were generated by PCR and confirmed by DNA sequencing.
Transient Transfection and Luciferase Assay-C5MJ2 cells were grown in RPMI medium supplemented with 5% fetal calf serum to the density of 5 ϫ 10 5 /ml. Transient transfection of C5MJ2 cells was performed by electroporation with a Bio-Rad Gene Pulser. 5 g of each promoter-reporter construct plus 1 g of the pCMV-␤-gal reference plasmid were mixed with 1 ϫ 10 7 C5MJ2 cells in 0.3 ml of medium and transferred into a 0.4-cm cuvette. Following a single 200-V/960-micro-farad pulse, each sample was divided into two parts and maintained in 6 ml of medium with or without TPA (5 ng/ml)/PHA (10 g/ml). 24 h after electroporation, cells were lysed for luciferase and ␤-galactosidase (␤-gal) assays (25). The luciferase activity was normalized according to the ␤-gal control.
DNA/Protein Binding Studies-Nuclear protein extracts were prepared as described (28). The DNase I footprinting assay was performed according to the standard procedure with 20 g of nuclear extract and 2 ϫ 10 4 cpm of a 32 P-5Ј-end-labeled fragment covering Ϫ125 to ϩ8 of the 5Ј-flanking region of the IL-9 gene. After DNase I partial digestion and phenol/chloroform extraction, samples were electrophoresed through an 8% urea sequencing gel. A sequencing reaction was run in parallel to serve as a reference for the protected region.
The gel shift assay was performed as described previously with 4 g of the nuclear extract and 2 ϫ 10 4 cpm of the probe (29). In gel supershift experiments, the empirically determined amounts of antibodies (Abs) were preincubated, in parallel with the same amount of preimmune serum, for 40 min at 4°C with the nuclear proteins before the addition of probes. The following Abs were used for gel supershift assays: anti-c-Jun/AP-1 and anti-c-Fos/AP-1 (Santa Cruz Biotechnology, Santa Cruz, CA), anti-p50/NF-B, anti-c-Rel/NF-B (30), and anti-CREB (a gift of Dr. Chou-Zen Giam). In peptide and oligonucleotide competition experiments, 2 g of competitor peptides (30) or a 50-fold molar excess of various competitor oligonucleotides were preincubated with Abs or nuclear extracts for 10 min before the binding reactions. Cells were stimulated with TPA (5 ng/ml)/PHA (10 g/ml) for the indicated time period. Total RNAs were analyzed for IL-9 expression by Northern analysis, and the nuclei were isolated for the nuclear run-on assay. A, Northern blot analysis. 10 g of RNA from C5MJ2 cells was used to hybridize to a 32 P-labeled IL-9 cDNA probe, and then to a human ␤-actin probe. B, nuclear run-on assay. 3 ϫ 10 7 C5MJ2 nuclei were prepared as described under "Materials and Methods." 5 ϫ 10 6 cpm transcripts were hybridized to 5 g of denatured plasmid DNAs containing specific and nonspecific inserts immobilized on nylon membranes. C, kinetics of Northern blot and run-on analysis. Induction folds were calculated according to the densitometric scanning of IL-9 signals versus ␤-actin signals.
FIG. 2. 5 mapping of the IL-9 mRNA. The 5Ј-end of the human IL-9 mRNA was determined by primer extension and RNase protection assays as described under "Materials and Methods." DNA ladders (lanes 1-4) were generated by sequencing IL-9 genomic DNA. The same primer used for sequencing was hybridized to unstimulated (lane 5) or TPA/PHA-stimulated (lane 6) C5MJ2 total RNA and extended with reverse transcriptase. The antisense RNA probe transcribed from IL-9 genomic DNA was hybridized to the total RNA from unstimulated (lane 7) or stimulated (lane 8) C5MJ2 cells. Following digestion with RNase, the samples were fractionated on an 8% sequencing gel. The mapped transcription start point, the overextended primer extension products, the RNA fragments protected from RNase digestion, and the DNA sequences around the start site are indicated. The restriction enzyme SacI site for constructing IL-9-luciferase plasmids is underlined. In UV-cross-linking and immunoprecipitation assays, the bromodeoxyuridine and [ 32 P]dCTP were introduced into the double-stranded DNA covering Ϫ76 to Ϫ58 of the IL-9 5Ј-flanking sequence as described (25). After a standard DNA and protein binding reaction, the DNAprotein complexes were UV-irradiated for 1 h on ice. For immunoprecipitation experiments, 4 g of Abs or normal rabbit IgG were added to each reaction after UV irradiation. The precipitated and UV-crosslinked proteins were separated on a 7.5% SDS-polyacrylamide gel, transferred to a nitrocellulose membrane, and exposed to x-ray film.

RESULTS
Up-regulation of IL-9 Gene Expression in TPA/PHA-stimulated C5MJ2 Cells-The kinetics of TPA/PHA-induced expression of the human IL-9 gene in C5MJ2 cells was examined by Northern blot analysis. As shown in Fig. 1A, the steady-state level of the IL-9 mRNA was very low in unstimulated cells and can be transiently increased following TPA/PHA stimulation. The induction of the IL-9 mRNA by TPA/PHA reached the maximum level 11 h after stimulation and slowly decreased thereafter. 24 h after stimulation, the steady state level of IL-9 mRNA was still about 3 times higher than that in unstimulated cells. To determine the control mechanisms of IL-9 gene expression after TPA/PHA stimulation, nuclear run-on assays were performed to measure the transcription rate of the IL-9 gene in C5MJ2 cells following TPA/PHA treatment (Fig. 1B). In agreement with the Northern blot analysis, the rate of IL-9 gene transcription was low in unstimulated cells. The transcription rate of the IL-9 gene was drastically increased (7-fold) 2 h after stimulation, reached the peak (10-fold) at 5 h, and then gradually decreased. These results, together with the observation that the steady-state level of the IL-9 mRNA peaked at 11 h of induction (15.5-fold) in Northern blot analysis, suggest that different mechanisms are involved in the accumulation of the IL-9 message. The mRNA induction observed at the first 5 h resulted mainly from the transcriptional activation since both Northern blot and nuclear run-on analyses showed identical induction folds of the IL-9 mRNA (Fig.  1C). In contrast, a predominantly posttranscriptional event was evidenced at 11 h of activation, suggesting a shift in the mechanisms regulating IL-9 mRNA accumulation from the transcriptional to the posttranscription level, most likely through increasing the stability of the message.
5Ј-Mapping of the IL-9 mRNA-Transcription initiation site for the IL-9 gene in C5MJ2 cells was determined by primer extension and RNase protection methods. In primer extension experiments, two bands with different densities were observed by using total RNA from activated C5MJ2 cells (Fig. 2, lane 6). The major band (82%) was 80 bases in length, indicating IL-9 gene transcription mainly initiated at 24 bases upstream from the translation start codon ATG and 25 bases downstream from the TATA box. The minor band (18%), 3 bases longer than the main band, was probably an overextended product. In RNase protection assays, one protected band was obtained with total RNA from both unstimulated and stimulated cells (Fig. 2, lanes  7 and 8) following RNase digestion. The size of the protected fragment corresponds to the size of the major band derived from the primer extension experiment, suggesting that transcription initiates at base 880 (according to the numbering system in the published human IL-9 genomic sequence; Ref. 18). This base is therefore considered as the transcription initiation site of the IL-9 gene and all the IL-9 sequences in this paper will be numbered accordingly. Our 5Ј-mapping results were slightly different from the ones mapped in stimulated PBMC cells (3), which probably reflect the differences in the control mechanisms of IL-9 gene expression in different cell lines.
Positive and Negative Regulatory Regions in the IL-9 Promoter-We have demonstrated previously that a 0.9-kilobase 5Јflanking region of the human IL-9 gene was able to direct the reporter gene expression in response to TPA/PHA stimulation in C5MJ2 cells (18). To further identify regions of possible functional significance within this 0.9-kilobase sequence, a series of deletion mutants were constructed for transient expression in C5MJ2 cells. As shown in Fig. 3, the expression level of the luciferase reporter gene was steadily increased when the DNA sequence between Ϫ878 and Ϫ379 was deleted, implying the existence of negative control elements in this region. The luciferase activity fluctuated between Ϫ379 and Ϫ143: the expression level reached the maximum with construct pIL-9-D4 (Ϫ379) and decreased about 2-fold with construct pIL-9-D5 (Ϫ299); After deletion of another 86-base pair sequence (Ϫ213, pIL-9-D6), the reporter gene expression increased again, especially after TPA/PHA induction; When an additional 70-base pair sequence was deleted (Ϫ143, pIL-9-D7), the luciferase activity decreased 5-6-fold. These results suggested that there are several positive and negative regulatory elements present between Ϫ379 and Ϫ143. Through computer search, several DNA sequences within Ϫ379 to Ϫ143 were found to be similar to the binding sites of certain known transcription factors. These sites included a B-like binding motif at Ϫ331 to Ϫ317, an Sp1-like site at Ϫ317 to Ϫ307, an AP-3-like site at Ϫ306 to Ϫ298, and an AP-1-like site at Ϫ146 to Ϫ140. Except the AP-1-like site, none of the other sites were demonstrated to be important in controlling IL-9 gene transcription by mutagenesis studies (data not shown). The AP-1-like site at Ϫ146 to Ϫ140 was demonstrated to be functional in IL-9 gene expression, since both basal and TPA/PHA-inducible luciferase activities decreased about 2-3-fold when this site was mutated (Fig.  4A). This AP-1-like site differs from the AP-1 consensus sequence by only one nucleotide. The nuclear proteins binding to this site were found to be inducible by TPA/PHA stimulation in the gel shift assay (Fig. 4B, lane 2). Mutations at this AP-1 site abolished formation of the DNA-protein complexes (Fig. 4B,  lane 3). These DNA-binding proteins were suggested to be c-Jun and c-Fos by the gel supershift assay, since anti-c-Jun antibody supershifted the retarded band and anti-c-Fos antibody inhibited the formation of the DNA-protein complexes (Fig. 4B, lanes 5 and 6). The sequence between Ϫ125 and Ϫ46 appeared to be essential for the constitutive and TPA/PHA-inducible expression of the IL-9 gene in C5MJ2 cells (Fig. 3). The construct pIL-9-D9 (Ϫ46) showed very low basal level reporter gene expression and failed to respond to TPA induction. An 8-fold (uninduced) and 24-fold (TPA/PHA-induced) increase in promoter activity was noted when Ϫ125 to Ϫ47 sequence was added. To localize the DNA sequences interacting with nuclear proteins, DNase I footprinting was performed using a 5Ј-end-labeled probe covering Ϫ125 to ϩ7. As indicated in Fig. 5, the region from position Ϫ45 to Ϫ80 was protected from DNase I digestion, suggesting that DNA-protein interactions exist within this region.
Minimum Sequences for the Basal and Induced Expression of the IL-9 Gene in C5MJ2 Cells-To identify the minimum sequence that mediates the basal and TPA-induced expression of the IL-9 gene, the deletion mutants within the proximal region from Ϫ90 to Ϫ63 of the IL-9 promoter were constructed. They were designated as phIL-9-D10 to phIL-9-D13, beginning at Ϫ90, Ϫ80, Ϫ74, and Ϫ63, respectively. Transient expression of these constructs in C5MJ2 cells showed that phIL-9-D10 and phIL-9-D11 generated luciferase activities comparable to phIL-9-D8 (Ϫ125), but a 5-10-fold decrease in luciferase activity was observed in transfections with constructs phIL-9-D12 and IL-9-D13, respectively (Fig. 6A). Since phIL-9-D9 (Ϫ47) had very low level of luciferase expression (Fig. 3), DNA cis-elements were proposed to be localized in the region between Ϫ47 and Ϫ80. This result is consistent with that of the DNase I footprinting assay. A closer inspection of the sequences between Ϫ47 and Ϫ80 revealed two sites highly homologous to the consensus sequences for NF-B and cAMP response element (CRE). The NF-B-like binding motif (GGGTTTTTCC) is located at Ϫ59 to Ϫ50, and the CRE-like motif (TGATGTCA) is immediately upstream from this NF-B site. To investigate whether these two sites and their neighboring sequences contribute to the promoter activity, mutations were introduced into these two potential binding sites as well as their adjacent upstream sequences. As shown in Fig. 6B, any mutation within the NF-B-like binding motif (phIL-9-M3), the CRE-like site (phIL-9-M2) or its adjacent 6-base pair upstream sequence (phIL-9-M1) caused a 90% decrease in the basal level and TPA/PHA-induced expression of the reporter gene. Simultaneous mutations at these three sites (phIL-9-M4) almost completely abolished the promoter activity. These results indicated that the three different sites are very important for IL-9 gene expression and they are likely to regulate transcription of the IL-9 gene in a cooperative manner in C5MJ2 cells.
Interactions of Nuclear Proteins with the Proximal Region between Ϫ47 and Ϫ80 of the IL-9 Promoter-To determine whether NF-B, CREB, and other transcription factors are involved in the binding to the proximal region between Ϫ47 and Ϫ80, the gel shift assays were performed with DNA fragments spanning from Ϫ47 to Ϫ90 (Fig. 7A). As shown in Fig.  7B, nuclear extracts from both unstimulated and stimulated cells formed DNA-protein complexes with this sequence. It was noticed that TPA/PHA stimulation increased the intensity of the major DNA-protein complexes. To identify individual binding sites, competition experiments were carried out with DNA fragments bearing different binding motifs of several transcrip- Transcriptional Regulation of IL-9 Gene Expression 15820 tion factors. Our data showed that proteins binding to this sequence are specific and the binding could be competed by unlabeled DNA fragment containing human immunodeficiency virus NF-B motif, but not by unlabeled DNA fragments containing the binding motifs for NF-AT p/c, AP-1, and CREB (Fig.  7B). To further characterize proteins binding to this region, gel supershift assays were performed using specific antibodies against NF-B, CREB, c-Jun, and c-Fos. As shown in Fig. 7C, NF-B was demonstrated to be one of the components in the DNA-protein complexes since antibodies raised against both Rel and p50 of NF-B could supershift the complexes (lanes 5 and 7) and the supershift pattern could be reversed by the addition of excess amount of NF-B peptides (lanes 6 and 8).
The anti-c-Jun antibody was also shown to be able to shift the DNA-protein complexes in the supershift assay (lane 2), but anti-c-Fos and anti-CREB antibodies failed to shift the retarded band. The CRE-like motif 5ЈTGATGTCA 3Ј at Ϫ67 to Ϫ60 differs from CRE consensus sequence 5ЈTGACGTCA 3Ј by a single base substitution. This variant CRE sequence has been shown to have a markedly reduced binding affinity for CREB (32). Our gel supershift assays and competition experiments also suggested that CRE and CREB may not be involved in the regulation of IL-9 expression.
To characterize proteins binding to the upstream region of the NF-B site, the UV-cross-linking assay was performed to identify proteins binding to the sequence between Ϫ77 and Ϫ58. As shown in Fig To further investigate whether proteins binding to the proximal region correlates with regulation of the IL-9 gene expression, we used oligonucleotides with mutations within the NF-B and its adjacent upstream region as the probes in the gel shift assay (Fig. 7, A and E). As compared with wild type sequence, mutations within the NF-B motif or its adjacent 20-base pair upstream sequence hindered the DNA-protein complexes formation (Fig. 7E, lanes 2-4). Simultaneous mutations of NF-B and its adjacent 20-base pair upstream sequence completely abolished the formation of DNA-protein complexes (lane 5). These results correlated well with the functional study (Fig. 6B), in which IL-9 promoter constructs containing the same mutations generated much lower luciferase activities as compared to the wild type construct, demonstrating that synergistic binding of these different transcription factors to the same promoter region is crucial for the expression of the IL-9 gene in C5MJ2 cells. DISCUSSION In this study, the up-regulation of IL-9 gene expression in C5MJ2 cells was shown to be regulated, at least in part, by transcriptional activation. Functional analysis of the 5Ј-flanking sequence of the IL-9 gene revealed several positive and negative regions controlling IL-9 gene expression. One positive region includes an AP-1 site at Ϫ146 to Ϫ140, which was demonstrated to be involved in the expression of the IL-9 gene in C5MJ2 cells. Members of the AP-1 protein family have been reported to be involved in the expression of IL-2, IL-3, IL-4, IL-5, and GM-CSF in T cells by working independently or cooperatively with other transcription factors, such as NF-AT and Oct1 (33)(34)(35)(36)(37). In this study, AP-1 protein was found not only to bind to the AP-1 site at Ϫ146 to Ϫ140, but also to participate in the formation of DNA-protein complexes binding to another important proximal region (Ϫ80 to Ϫ47) of the IL-9 promoter. Therefore, AP-1 appears to be an important general transcription factor involved in the up-regulation of a large number of cytokines in activated T cells.
The second positive regulatory region resides within positions Ϫ80 to Ϫ47 of the IL-9 promoter. Our study revealed that this region is absolutely required for IL-9 gene expression. The DNA sequence within this proximal region is highly conserved in evolution (identical between human and mouse IL-9 promoters). Within this region, an NF-B motif at Ϫ59 to Ϫ50 as well as its 20-base pair adjacent upstream sequence were demonstrated to be critical since mutations at one or both sites drastically decreased the promoter activity. The direct interaction of NF-B with this region was also confirmed by competition experiments and the gel supershift assay. NF-B or related proteins have been shown to play important roles in the expression of several cytokines (37)(38)(39)(40)(41). NF-B is required for the activation of GM-CSF gene by various mitogens in T cells (37,41). Unlike GM-CSF gene, the position of this NF-B motif in IL-9 promoter is located much closer to the TATA box. Compared with this NF-B site in the IL-9 promoter, a stretch of DNA sequence referred to as the conserved lymphokin element 0 (CLE0) was found in the promoter regions of the GM-CSF, IL-4, and IL-5 genes (22). CLE0, regarded as an essential cis-element, is required for the expression of GM-CSF, IL-4, and IL-5 genes in activated T cells and contains composite binding site for AP-1/Ets or NF-AT factors (22,31,37,(42)(43). The binding sequences for NF-B, Ets and NF-AT proteins are quite similar, all of which have the AAAGG motif. The similarity suggests that promoters of those cytokines on chromosome 5 may be evolutionarily related and also raises the possibility that the T cell-restricted expression of IL-9 may be attributed to the unique feature of its promoter sequence.
It should be pointed out that in addition to NF-B binding to the proximal region between Ϫ80 and Ϫ47, c-Jun and the proteins of 35 kDa have also been shown to specifically bind to this functional region. c-Jun may bind to the CRE-like site at Ϫ67 to Ϫ60, since probes with mutations within this CRE-like motif failed to give rise to the supershifted band using anti-c-Jun in our gel supershift assay (data not shown). The nature of 35-kDa proteins remains unclear. It was noticed that a GATA binding motif 5Ј-TGATAC-3Ј exists at position Ϫ76 to Ϫ71 on the minus strand of the IL-9 promoter, and mutations containing this GATA binding site (pIL-9-M1) reduced reporter gene expression. At present, four members of GATA family tran-B, gel shift assay with oligonucleotide competitors. 50-fold molar excess of competitors were preincubated with nuclear proteins for 5 min before addition of labeled wild type probe (PW1). The position of the specific binding complex is indicated by an arrow. C, gel supershift assay with antibodies and peptide competitors. For the gel supershift assay, empirically determined amount of Abs and same amount of preimmune serum were preincubated with the nuclear proteins before adding probes (PW1). In the corresponding peptide competition assay, 2 g of the specific peptide was preincubated with Abs before binding reactions. D, UV-cross-linking studies. The sequences of upper strand of wild type and mutation probes used in UV-cross-linking study are shown with mutated nucleotides highlighted. After UV irradiation of the DNA-protein complexes, equal volumes of the same sample were treated either with anti-c-Jun or preimmune IgG, followed by the addition of anti-rabbit-IgG agarose. The UV-cross-linked proteins and immunoprecipitates were fractionated on a 7.5% SDS-polyacrylamide gel. E, gel shift assay with DNA fragments containing substitution mutations. The position of the specific binding complexes is marked by an arrow. scription factors have been characterized (44 -45). Among them, only GATA-3 is predominantly present in T cells and is involved in the regulation of several T cell-specific genes (46 -48). Recently, IL-5 was shown to be activated by GATA-3 in T cells (43). Since the molecular masses of GATA-3 proteins are around 45 kDa, it is unlikely that the 35-kDa UV-cross-linked proteins are GATA-3.
In summary, our results suggest that synergistic interactions of known and unknown transcription factors are likely to play a critical role in the regulation of IL-9 expression. As indicated by our study, several proteins can bind to the proximal region between Ϫ80 and Ϫ47. Mutations at different sites within this region not only decreased reporter gene expression but also abolished the formation of specific DNA-protein complexes. In addition to this important proximal region, we have also demonstrated that the AP-1 site at Ϫ146 to Ϫ140 can be regulated by AP-1 proteins to enhance IL-9 promoter activity, implying "cross-talk" and cooperation may exist between these two regions and among the different transcription factors. The cooperation of multiple transcription factors has been shown to be involved in the expression of IL-3, IL-5, and GM-CSF in T cells (37,49). For IL-9 gene expression, its T cell-restricted expression pattern may be mediated by certain unknown T cell-specific transcription factors and/or by a unique profile of combinatorial interactions of common transcription factors.