Regulation of c-myc Transcription by Interleukin-2 (IL-2)

Regulation of c-myc expression is known to occur at the level of transcription initiation. However, the participating promoter elements and their cognate binding proteins have not been fully characterized. c-myc transcription can be stimulated by a number of cytokines including interleukin-2 (IL-2). We have identified a novel IL-2-responsive element, located in the 5′-flanking region of the c-myc gene, between nucleotides −1406 and −1387 (relative to the P2 promoter). This element belongs to the family of interferon-γ activation site-like responsive elements and has the core sequence TTCCAATAA. We confirmed that IL-2-mediated signaling involves activation by phosphorylation of Jak2 tyrosine kinase and subsequently STAT4. The transcription factor STAT4 binds the TTCCAATAA motif within this responsive element and, therefore, is probably involved in enhancing c-myctranscription upon IL-2 stimulation. Our results propose participation of Jak2 and STAT4 in IL-2-induced up-regulation of c-myc.

The c-myc proto-oncogene is involved in controlling cell proliferation and differentiation (1). The human c-myc gene is composed of three exons (2). The first exon is untranslated and contains two principal promoters: P2, which is the predominant c-myc promoter, generating up to 90% of all transcripts, and P1, which is separated from P2 by 165 bp 1 and generates 10 -25% of c-myc RNA (3). All minor promoters, including P0 (located 750 bp 5Ј of P2), P3 (in the first intron), and an antisense promoter (found in the second intron), generate less than 5% of c-myc mRNA. The human c-myc gene encodes two polypeptides with apparent molecular masses of 64 and 67 kDa (4). The c-Myc proteins contain a basic region, which mediates sequence-specific DNA-binding, and also helix-loop-helix and leucine zipper motifs, which promote protein-protein interactions (1). Heterodimers formed by MYC and its partner MAX are able to bind DNA and function as transcription activators (5), while MAX homodimers act as transcriptional repressors (5).
The role of c-myc in the development of neoplasia is now well established (3-4, 6 -10). The c-myc locus is interrupted by reciprocal chromosomal translocation in human Burkitt's lymphoma and murine plasmacytomas (11). c-myc is amplified in some myeloid leukemia cell lines and in some cases of human breast cancer (7). The c-myc gene is capable of inducing multiple neoplasms in transgenic mice when fused to immunoglobulin enhancers (6) or a mouse mammary tumor virus long terminal repeat (9). Collectively, each of these perturbations results in the constitutive activation of c-myc transcription. Unlike the ras gene family, mutations within the coding sequences appear not to be an important feature in converting myc from a proto-oncogene to an oncogene. Rather, abnormally high transcription of c-myc, at an inappropriate stage of the cell cycle or during differentiation, leads to oncogenic transformation (2). Therefore, to define the precise role of the c-myc gene in tumorigenesis, a better understanding of its regulation is warranted. Regulation of c-myc expression is extremely complicated and may occur at the levels of transcription initiation (12), transcript elongation (3,13), and messenger RNA stability (10). At the level of initiation of transcription, regulation appears to occur via cis-acting regulatory elements (2,11). Although two enhancer elements have been described, 3Ј of the c-myc exons (14,15), the majority of cis-acting regulatory sequences have been identified within the 5Ј-flanking domain of the human c-myc gene (11,16). Two sequences, ME1a1 and ME1a2, are located between the principal c-myc promoters P1 and P2. A regulatory region close to P2 (Ϫ58 to Ϫ68) was found to mediate activation of the P2 promoter by E2F (17,18) and by the product of the RB1 gene (19). A palindromic purine/pyrimidine-rich positive regulatory element, also described as DNase I hypersensitivity site III 1 , has been identified in positions Ϫ142 to Ϫ115 relative to the P2 promoter (20,21). A negative regulatory element located between bp Ϫ293 and 253, relative to the P1 promoter, has been shown to interact with a transcription factor complex formed by Fos, Jun, and octamerbinding factors (22,23). An additional regulatory element about 2.2 kilobase pairs upstream of P1 was found to bind nuclear proteins (24,25). However, all promoter elements and their cognate binding proteins that are necessary for optimal transcription initiation have yet to be fully characterized. Furthermore, the significance of many of these binding sites and their corresponding factors, during physiological regulation of c-myc expression, remains largely unknown.
c-myc is known to be a cytokine-responsive gene (26). Among several cytokines, interleukin-2 (IL-2) is one of the critical regulators of proliferation and differentiation of hematopoietic cells. The functional interleukin-2 receptor (IL-2R) consists of three subunits: the IL-2R␣, IL-2R␤, and IL-2R␥ chains. Both IL-2R␤ and IL-2R␥ subunits are required to transmit the IL-2 signal to the cell interior (27,28). The membrane-proximal cytoplasmic region of IL-2R␤, termed the serine-rich region, has been shown to play a critical role in IL-2-mediated c-myc induction followed by cell proliferation (27,28). Recently, the * Portions of this work were supported by NCI, National Institutes of Health, Public Health Service Grants CA-41165 and CA-55819. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
role of IL-2 in stimulation of c-myc transcription has been confirmed (26, 29 -31). IL-2 has been shown to selectively stimulate transcription from the P2 promoter (8,32). However, the participating IL-2 response elements and their cognate binding proteins have not yet been identified.
Previously, we determined that c-myc transcription is rapidly induced in the Natural Killer cell line NK3.3 in response to exogenous IL-2 (33). Accordingly, NK3.3 cells have been chosen as the model IL-2-responsive cell line in our experiments. Using a functional reporter gene assay, we have found an IL-2responsive element within 537 bp of the 5Ј-flanking region of the c-myc locus (from bp Ϫ1429 to Ϫ892 relative to the P2 promoter). Analysis of protein-DNA interactions within this 537-bp region has localized the IL-2-inducible response element and identified its binding protein.

EXPERIMENTAL PROCEDURES
Reagents, Enzymes, and Cytokines-Cell culture medium (RPMI) and fetal calf serum were purchased from Life Technologies, Inc. Lymphocult was obtained from Biotest (Dreieich, Germany), and recombinant IL-2 was from Genzyme (Cambridge, MA). Poly(dI-dC) and protein A-Sepharose Fast Flow were purchased from Amersham Pharmacia Biotech. High pressure liquid chromatography-purified oligonucleotides were obtained from Bio-Synthesis (Lewisville, TX). Antibodies to STAT proteins (supershift quality, concentration of 1 mg/ml) were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Monoclonal antibodies to anti-phosphotyrosine for Western blotting were obtained from Upstate Biotechnology, Inc. (Lake Placid, NY). All restriction endonucleases, T4 polynucleotide kinase, calf intestine alkaline phosphatase, and DNase I were purchased from New England Biolabs (Beverly, MA). All reagents for PCR were from Promega (Madison, WI).
Cell Lines and Cultures-Natural Killer cell line NK3.3 was used as a model in all experiments. Cells were grown in RPMI 1640 supplemented with 15% fetal calf serum, 15% Lymphocult, penicillin, streptomycin, and L-glutamine. Cells were kept in a humidified incubator at 5% CO 2 . Prior to stimulation with IL-2, NK3.3 cells were starved for 18 h in the same medium with a decreased concentration (5%) of serum and without Lymphocult. For the purposes of the experiment, cells were stimulated for 1 h with recombinant IL-2 (200 units/ml).
Transient Transfection and Reporter Assays-A variety of DNA fragments were generated from exon 1 and upstream 5Ј-flanking sequences of the c-myc locus (Fig. 1A) and cloned upstream of the promoter-lucϩ transcriptional unit into luciferase reporter construct pGL3-promoter vector (Promega). NK3.3 cells (7-10 ϫ 10 6 cells/sample) were transiently transfected with 10 g of the appropriate recombinant vector by electroporation at 800 microfarads at 300 V using "CellPorator" (Life Technologies). All transfections were normalized to ␤-galactosidase activity by cotransfection of 0.5 g of a ␤-galactosidase (pRSV-␤-Gal) expression vector. Transfected cells were starved for 18 h, as described earlier, and then were split in two sets. One set of cells was left untreated; another was stimulated with recombinant IL-2 (200 units/ ml). In 48 h, all cells were harvested for luciferase and ␤-galactosidase assays, which were performed according to the manufacturer's protocols (Promega and ICN Pharmaceuticals, Inc. (Costa Mesa, CA), respectively). The light intensity was measured with a luminometer. To exclude variation due to differences in transfection efficiency, signals obtained with the reporter genes were normalized to the levels of the internal ␤-galactosidase control at each point. The statistical analysis of the data was performed using "Origin" software (Microcal, Northampton, MA).
Generation of Nuclear Extracts and Electrophoretic Mobility Shift Assay (EMSA)-Generation of the nuclear extract from control and IL-2-stimulated NK3.3 cells was performed according to the procedure described by Marzluff (34) and Peterson (35) with a modification that included extraction of nuclear proteins with 0.4 M KCl. The protein concentration was determined by using the Bio-Rad protein assay and bovine serum albumin (Sigma) as a standard. Three types of the doublestranded 32 P-end-labeled DNA probes were used in the experiments. A 537-bp probe was generated by digestion of the plasmid pMC41-HE (36), containing a 12-kilobase pair c-myc genomic clone, with the restriction endonucleases ClaI and TthIII. A 100-bp probe was the product of the 537-bp fragment digestion with HpaI. A 20-bp probe was the double-stranded oligonucleotide 5Ј-GCATTTCCAATAATAAAAGG-3Ј, corresponding to the nucleotide sequence within the ClaI-HpaI frag-ment. A mutated 20-bp probe was designed by substitution TTGTT for TTTCC in the original 20-bp probe. The binding reaction, in EMSA studies, used 10 g of total protein from nuclear extracts and 1 ng (approximately 20,000 -30,000 cpm) of the 32 P-end-labeled DNA probe. Incubation was generally performed as described (37) with some corrections. In brief, the binding reaction was carried out for 30 min at room temperature in 25 l of binding buffer (10 mM HEPES, pH 7.8, 100 mM KCl, 5 mM MgCl 2 , 1 mM dithiothreitol, 0.1 mM phenylmethylsulfonyl fluoride, and 10% glycerol) in the presence of 2 g of nonspecific competitor poly(dI-dC). DNA-protein complexes were separated by electrophoresis on 3.5% polyacrylamide gel at 250 V. Supershift experiments were performed to determine the nature of DNA-binding transcription factors, present in the nuclear proteins, using specific antibodies for STAT1 to -5 or normal rabbit IgG (1 mg/ml). 1 g of corresponding antibodies was added to nuclear protein samples prior to mixing with the probe and kept for 10 min at 4°C. EMSA was then performed, as described.
The "stairway assay" (modification of standard EMSA for localization of protein/DNA-binding sites in large DNA segments with known sequence) was performed as described by Van Wijnen (38). Briefly, two samples of 537-bp DNA fragment ClaI-TthIII containing a single 5Ј-32 P-labeled terminus were prepared. Aliquots (25 ng) of each probe were separately digested to completion with each of a series of restriction enzymes chosen to shorten the probe subsequently by 100 bp per cut. After organic extraction and ethanol precipitation, the equimolar quantities of these various shortened DNA fragments were dissolved in TE buffer and used as probes for the standard EMSA protocol.
Site-directed Mutagenesis-The oligonucleotide sequence TTTCC was replaced by TTGTT in the 100-bp probe using the site-directed mutagenesis procedure described by Ling and Robinson (39) and elsewhere. Mutagenesis involved two rounds of PCR using recombinant plasmid pGL3 with a 100-bp insert as a wild type template. First PCR was performed using RVprimer3 from pGL3 plasmid (Promega) as the forward flanking primer and the mutagenic internal primer 5Ј-TTAT-TAACAATGCGGTCATGC-3Ј (annealing temperature ϭ 58°C). The product of that reaction, the "megaprimer," was purified by a PCR purification kit (Qiagen Inc., Valencia, CA) and used, along with the reverse flanking primer Glprimer2 (Promega), as a primer for the second PCR (annealing temperature ϭ 53°C). The final PCR product contained the desired mutation (described above) in a particular DNA sequence. The mutated 100-bp fragment was excised from the PCR product by SacI and XhoI restriction endonucleases (Promega) and cloned into pGL3-promoter vector. The presence of the desired mutation was verified by sequencing with the ABI Prism Big Dye Terminator Cycle Sequencing Ready Reaction Kit (Perkin-Elmer).
DNase I Footprinting Analysis-The binding reaction was performed as described for electrophoretic mobility shift assay, except the reaction contained 40 g of the nuclear extract and 4 ng of labeled DNA. After incubation, samples were digested in the same binding buffer containing 1 mM CaCl 2 with 0.01-0.9 units of DNase I for 2 min. Digested samples were precipitated with ice-cold ethanol in the presence of saturated ammonium acetate and carrier tRNA, washed twice with 70% ethanol, and resuspended in the electrophoresis loading buffer to 10,000 cpm/l. Equal counts were loaded onto 6% acrylamide/8 M urea sequencing gel. The relative intensity of radioactive bands was determined by PhosphorImager analysis, utilizing a PhosphorImager SI (Molecular Dynamics, Inc., Sunnyvale, CA).
Scanning Densitometry-Autoradiographic bands were quantified within the linear range of film on a model 300A laser densitometer and ImageQuant software (Molecular Dynamics).
Immunoprecipitation and Western Blotting-NK3.3 cells (10 7 cells/ sample) were starved for 18 h prior to IL-2 stimulation and then treated with 200 units/ml of recombinant IL-2 for 1 h. Untreated (control) and IL-2-treated cells were lysed in 20 mM Tris buffer, pH 7.5, containing 150 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40, 10 mM NaF, 1 mM Na 3 VO 4 , 5 g/ml leupeptin, 1 g/ml aprotinin, and 1 mM phenylmethylsulfonyl fluoride. For Immunoprecipitation, lysis supernatants were incubated with 1 g of specific polyclonal antibodies (anti-JAK2 or anti-STAT4) for 1 h at 4°C. 40 l of protein A-Sepharose was then added to each sample for an additional incubation, with agitation, at 4°C for 1 h. Precipitated proteins were separated on 7.5% Laemmli polyacrylamide gels, blotted onto polyvinylidene difluoride membranes, probed with anti-phosphotyrosine antibodies, and followed by horseradish peroxidase-conjugated sheep anti-mouse antibodies. Immunoreactive bands were visualized with chemiluminescent SuperSignal Substrate for Western blotting (Pierce).

RESULTS
Identification of a Functional IL-2 Response Element within the 5Ј-Flanking Sequence of c-myc-To identify the region of the human c-myc gene essential for IL-2-regulated transcriptional activity, various luciferase reporter constructs were assembled. These constructs incorporated the known and putative c-myc regulatory regions, including exon 1 and upstream 5Ј-flanking sequences (2500 bp linked to the P2 promoter, described in Fig. 1A). These regions were cloned upstream of the SV40 promoter in the luciferase reporter plasmid pGL3-promoter. We measured luciferase activity in lysates of NK3.3 cells transiently transfected with the aforementioned constructs. None of the constructs was able to increase the level of luciferase activity in untreated cells. Upon stimulation of the transfected cells with recombinant IL-2, only the constructs containing the fragment ClaI-TthIII (bp Ϫ1429 to Ϫ892 relative to P2 promoter) cloned in both direct and reversed orientations exhibited a 4 -6-fold increase of luciferase activity (Fig.  1B). The observed differences were statistically significant (p Ͻ 0.05). Other constructs had no effect on the level of luciferase in IL-2-stimulated cells (data not shown). Our results suggested the presence of a functional IL-2-inducible element within 537 bp ClaI-TthIII DNA fragment (bp Ϫ1429 to Ϫ892 relative to the P2 promoter).
Precise Localization of the IL-2-responsive Element within the 5Ј-Flanking Region of c-myc Gene-The EMSA was used to analyze protein-DNA interactions within this putative 537-bp, cis-acting, IL-2-inducible regulatory domain. The end-labeled 537-bp ClaI-TthIII fragment was incubated with nuclear extracts from untreated and IL-2-stimulated cells and analyzed  5). The double-stranded 537-bp probe (ClaI-TthIII fragment from the plasmid pcm41) was 32 P-end-labeled. Lane 1 contains the probe only and no NE. For each binding reaction, 10 g of total NE protein and 20,000 cpm (1 ng) of the probe were taken. For the cold competition (lanes 4 and 5), the binding reaction was performed in the presence of 100 ng of the unlabeled probe. The position of the DNA-protein complexes is marked with the arrow. on nondenaturing gel electrophoresis. As seen in Fig. 1C, only nuclear extracts from IL-2-stimulated cells form a slowly migrating DNA-protein complex. This protein-DNA complex was eliminated when a 100-fold molar excess of unlabeled fragment was included in the binding reaction. Cold competitors of unrelated sequence had no effect (not shown).
To narrow the position of this protein-binding site within the ClaI-TthIII fragment, we used the stairway assay (a modification of standard EMSA for localization of protein/DNA-binding sites in large DNA segments with known sequence). As described under "Experimental Procedures," we prepared two 537-bp probes, labeled only on a single 5Ј terminus (either ClaI or TthIII end). Fig. 2A presents the way each of the probes was subsequently shortened by restriction endonucleases HpaI, KpnI, HaeII, BsrI, and ApaI. After cleavage, each probe contained two fragments with only one 32 P-labeled. Only complexes between nuclear proteins and the 32 P-labeled portion could be visualized in EMSA assays. As shown in Fig. 2B, ClaI end-labeled full-length probe and all fragments, generated by restriction from this probe, were able to generate similar DNAprotein complexes. The ability of the shortest ClaI end-labeled fragment (lane 16, ClaI-HpaI fragment) to form a complex with nuclear proteins from IL-2-stimulated cells suggests the presence of the protein-binding site within this 100-bp fragment. In the case of the TthIII end-labeled fragment, we observed DNA-protein complexes only when the full-length probe was used (lane 5a). Shortening of the probe by 100 bp (lane 6a, cleavage with HpaI) eliminated the band of slowly migrating complex. These data map the position of the protein-binding site within the 100-bp DNA fragment (Ϫ1429 to Ϫ1329 bp relative to the P2 promoter).
To examine whether this shortened DNA fragment (ClaI-HpaI) contains the same enhancer element(s) responsive to IL-2 activation and is acting in the same manner as the fulllength 537-bp probe, a 100-bp fragment was cloned into the pGL3-promoter luciferase reporter construct, and transient transfections were carried out, as described previously. As shown in Fig. 3A, insertion of the ClaI-HpaI fragment in any orientation (forward or reverse) did not activate this reporter in the absence of IL-2. IL-2 induction led to a 12-15-fold increase in luciferase activity using both constructs pGL3/100(ϩ) and pGL3/100(Ϫ). These data are statistically significant (p ϭ 0.05). Comparison of luciferase transcriptional assays performed with reporter constructs containing the 537-bp insert (Fig. 1B) and the 100-bp insert (Fig. 3A) revealed that the smaller insert induced higher levels of transcriptional activation than the larger one. This suggests that replacement of DNA sequence ClaI-TthIII in the reporter gene construct with the smaller ClaI-HpaI fragment eliminated additional negative effects of other regulatory elements in this portion of the c-myc 5Ј-flanking region.
To confirm the ability of the 100-bp fragment to form the same type of DNA-protein complex as the 537-bp fragment, we performed another EMSA experiment using both fragments as probes. Incubation of each of these probes with nuclear extracts from IL-2-stimulated cells leads to formation of specific DNAprotein complexes (Fig. 3B, lanes 3 and 6). The remainder of the 537-bp probe (HpaI-TthIII, 400-bp fragment) was unable to bind nuclear proteins from IL-2-stimulated cells (data not shown).
To locate the position of the protein-binding site within the 100-bp fragment, DNase I footprinting analysis was performed, as described under "Experimental Procedures." As shown in Fig. 4, only nuclear extracts from IL-2-stimulated cells were able to generate the DNase I-protected region in the 100-bp fragment (lane 7). Using densitometry and PhosphorImager analysis, we found at least 70% reduction of the bands' inten-sity in the protected region (marked with the arrows) while compared with the bands' intensity outside this area. The arrows indicate the position of the 20-bp protected sequence 5Ј-GCATTTCCAATAATAAAAGG-3Ј (Ϫ1406 to Ϫ1387 relative to P2 promoter). Based on this sequence, a 20-bp probe was generated for EMSA experiments. Fig. 5 presents the results of EMSA and competition experiments with 20-and 100-bp fragments. First, the 20-bp probe was able to generate complexes with nuclear extracts from IL-2-stimulated cells (lane 11, marked with the arrows). Minor complexes, other than those indicated by arrows, were not reproducibly induced by IL-2 in multiple experiments and thus should be disregarded. Second, the presence of a 100-fold excess of unlabeled 20-bp fragment removed the complexes formed between nuclear proteins of stimulated cells and the 100-bp fragment (lane 8). Third, the presence of a 100-fold excess of cold 100-bp probe eliminated 82% (determined by densitometry) of the complexes, formed by the 20-bp fragment and nuclear proteins (lane 13). Collectively, these results provide evidence of identical protein-binding sites in both fragments.

Identification of the Putative Transcription Factor Associated with (or within) the Characterized 20-bp Protein-binding Site-
Using the data base "Transfac Matrix," computer analysis of putative binding sites for known transcription factors was performed for the aforementioned 20-bp sequence. A STAT binding-related sequence 5Ј-TTCCAATAA-3Ј was found within this area. This region, located between bp Ϫ1402 and Ϫ1394 relative to the P2 promoter, is highly homologous (approximately 85.6%) to STAT1␣ and -␤, STAT2, STAT3, STAT4, and STAT5 binding sites. Therefore, the ability of anti-STAT antibodies to alter the mobility of slowly migrating DNA-protein complexes in "supershift" experiments was examined. As seen in Fig. 6A 2, 6, and 10) and IL-2- stimulated (lanes 3, 4, 7, 8, and 11-13) NK3.3 cells was performed for the 100-bp ClaI-HpaI 32 P-labeled fragment (lanes 1-8) and 20-bp double-stranded synthetic 32 P-labeled oligonucleotide (lanes 9 -13). For each binding reaction 10 g of total NE protein and 1 ng of the corresponding probe were taken. Lanes 1, 5, and 9 have probes only, no NE. For the cold competition, binding was performed in the presence of a 100-fold excess (100 ng) of unlabeled 100-bp fragment (lanes 4 and 13) or 20-bp fragment (lanes 8 and 12). Positions of the DNA-protein complexes are marked with the arrows .   FIG. 6. A, IL-2-mediated complexes between 537-bp probe and nuclear proteins reacted specifically only with anti-STAT4 antibodies. Reactivity to various anti-STAT antibodies was determined by using the supershift technique. EMSA with NE from untreated (lane 1) and IL-2-treated (lanes 2-7) NK3.3 cells was performed for 537-bp ClaI-TthIII end-labeled fragment in the absence (lanes 1 and 2) or presence (lanes 3-7) of various anti-STAT antibodies. 2 g of appropriate anti-STAT antibodies were added to the binding mixture 30 min prior to the addition of the probe. Positions of the DNA-protein complexes are marked with arrows. B, anti-STAT4 antibodies impair the complex formation between nuclear proteins and shortened (100-and 20-bp) probes. EMSA with NE from nonstimulated (lanes 2 and 7) and IL-2-stimulated (lanes 3-5, 8, and 9) NK3.3 cells was made for the 100-bp ClaI-HpaI probe (lanes 1-5) and 20-bp double-stranded synthetic oligonucleotide (lanes 6 -9). For each binding reaction, 10 g of total NE protein and 1 ng of the corresponding probe were taken. Lanes 1 and 6 have probes only, no NE. 2 g of anti-STAT4 (lanes 4 and 9) or 2 g of nonspecific human IgG (lane 5) were added to the binding mixture 30 min prior to the probe. Positions of the DNA-protein complexes are marked with arrows.
used. Fig. 6B shows that the addition of anti-STAT4 antiserum to 100-bp (lane 4) or 20-bp (lane 9) probes reduced the abundance of slow migrating DNA-protein complexes in a similar manner. Collectively, these results indicate the presence of STAT4 in nuclear extracts from IL-2-stimulated cells and its ability to bind the DNA sequence within the c-myc locus and, therefore, enhance c-myc transcription after IL-2 stimulation.
As described by Yamamoto and co-workers (40) and recommended for use by Santa Cruz Biotechnology, replacement of one or more C for T nucleotides in the STAT4 consensus binding site eliminates binding between this mutated probe and STAT4 protein in the gel shift assay. Thus, we performed another EMSA experiment using two probes: the original 20-bp probe and a mutated one, where part of the consensus binding site TTTCC was changed to TTGTT. Fig. 7A shows that the mutated probe lost the ability to bind STAT4 in IL-2-stimulated cells (lane 6).
To investigate whether the mutations in the STAT element could disengage the IL-2-dependent activation of the promoter in vivo, the mutated 100-bp (ClaI-HpaI) fragment was cloned into the luciferase reporter plasmid pGL3-promoter. Using site-directed mutagenesis (40), we mutated the sequence TT-TCC to TTGTT in the STAT4 binding site of the 100-bp fragment (for details, see "Experimental Procedures"), introduced this fragment to the pGL3-promoter plasmid, and tested this recombinant construct (pGL3/mut.100) for its ability to respond to IL-2 in the transient transfection assays (as described above). As shown in Fig. 7B, no significant increase over control in luciferase activity was found in either nonstimulated or IL-2-stimulated cells, transfected with the construct pGL3/ mut.100. Therefore, we conclude that the construct containing the mutated STAT4 binding site was unable to mediate response to IL-2 in contrast to the pGL3/100 construct containing the wild type consensus element within the 100-bp fragment (see Fig. 3A).
Previous studies have demonstrated that activation of STAT4 is a result of its phosphorylation by activated Jak2 protein tyrosine kinase (41). Accordingly, we examined the ability of IL-2 to induce activation of the Jak2/STAT4 signal transduction pathway. Since tyrosine phosphorylation is critical to the activation of Jaks and STATs, Jak2 and STAT4 were immunoprecipitated with corresponding antibodies from the lysates of IL-2-stimulated NK3.3 cells, and Western blotting with anti-phosphotyrosine antibodies was performed. Phosphorylated Jak2 and STAT4 were found only in IL-2-activated cells (Fig. 8A) and not in untreated cells. These results are consistent with our hypothesis that IL-2 activates the Jak2/STAT4 pathway, and STAT4 binds to an IL-2 response element in the c-myc gene, thereby positively regulating c-myc transcription. DISCUSSION Three distinct signaling pathways, linked to IL-2R, have been recently identified (28,42): the p56 lck pathway leading to c-fos/c-jun induction, the bcl-2 induction pathway, and the cmyc induction pathway. Importantly, none of these pathways affect activation of the other. However, the combination of any two of the three pathways is sufficient to promote cell growth in the absence of cytokines (28). Therefore, it is reasonable to suggest that cross-talk between intermediate members of the different IL-2-mediated pathways could provide sufficient induction of target genes. The induction pathway of the c-myc gene is primarily linked to the serine-rich region of IL-2R␤ chain (27,28), which is associated with Jak1 (43,44) and Syk protein-tyrosine kinases (45). However, involvement of IL-2R␥ and association of this subunit with Jak3 in the same transduction pathway has been confirmed (46). Another member of the family Syk/ZAP70 protein-tyrosine kinase, ZAP70, is also suspected to be IL-2R-associated (42,47). As yet, no association has been found for Jak2 and IL-2R. The distal components, linking the IL-2 receptor to c-myc transcription, are largely unknown.
Although several positive and negative regulatory elements within the c-myc locus have been described, none have been identified as IL-2-responsive. In addition, no known transcription factor(s) have been shown to be directly involved in IL-2mediated up-regulation of c-myc. The lack of specific information regarding this important cytokine-mediated effect prompted our search for IL-2-responsive elements within the c-myc locus and the corresponding transcription factor(s).
Computer modeling of the c-myc flanking DNA suggests that two slow-moving DNA fragments, spanning nucleotides (relative to the c-myc promoter P2) Ϫ983 to Ϫ617 and Ϫ1855 to Ϫ1219, form large left-handed superhelices or curved structures (48). It is possible that curved DNA segments may play a regulatory role in DNA transcription (48). In the current studies, we first identified a functional IL-2-responsive element within the area Ϫ1429 to Ϫ892 bp relative to the P2 promoter (partially overlapping with the first regulatory DNA fragment determined by Kumar and Leffak (48)). Then we localized the precise position of this element and determined (using EMSA and DNase I footprinting) its core nucleotide sequence. Finally, STAT4 was identified as at least one transcription factor that binds the core sequence (TTCCAATAA) within this element.
The DNA sequence elements in the promoters of genes that bind STAT proteins can be classified into two groups (49). The prototype of the first class is the interferon-stimulated response element (50). Interferon-stimulated response elements have the consensus sequence AGTTTCNNTTTCN(C/T) (where N is any nucleotide). The second class comprises the IFN-␥ activation site (GAS)-like response elements. A significant number of GAS-like sequences have been identified in promoters of genes activated by different extracellular signaling proteins. Various STATs have been shown to bind at least one of the GAS-like elements (40,(51)(52)(53). All of these elements have the palindromic core sequence TTNNNNNAA, but they differ in five inner nucleotides. Selective and specific activation of genes, by different STAT dimers, with different binding affinities, involves slightly different response elements. We have identified a new IL-2 response element, which also contains the palindromic core sequence TTCCAATAA, belonging to the GAS-like response elements. Under the conditions described, STAT4, upon activation by Jak2, binds to this response element and probably promotes c-myc expression. Until recently, only IL-12 and IFN-␣ were shown to mediate signaling through the phosphorylation of Jak2/Tyk2 and subsequently STAT4 (54 -57). Therefore, overlapping biological responses to IL-2 and IL-12 could possibly be explained by the synergistic effect on IFN-␥ production (58,59) or induction of differential expression of specific sets of genes (54). However, recently K. Wang and co-workers (60) described the direct involvement of IL-2 in the Jak2/STAT4 signaling pathway. They demonstrated the ability of IL-2 to activate target genes through phosphorylation of Jak2 and STAT4 in primary NK cells and also in the NK3.3 cell line. Moreover, they reported the absence of IL-2-mediated Jak2 and STAT4 activation in primary resting T cells or mitogen-activated T cells (60). Thus, K. Wang suggested that this unique activation of the STAT4-signaling pathway only in NK cells might underlie the distinct functional effect of IL-2 on this cell population. Our data generated on the same model of NK cells propose c-myc as one of the target genes in this signaling pathway.
Mechanisms of Jak2/STAT4 involvement in IL-2-mediated c-myc up-regulation are still largely unknown. The activation of STAT4, in response to IL-2, is not due to the autocrine production of IL-12 or IFN-␣, because the presence of IFN-␣-or IL-12-neutralizing antibodies did not affect the activation of STAT4 in response to IL-2 (60). IL-2R is known to be associated with Jak1 and Jak3 kinases but not with Jak2. Although unproved, it is provocative to suggest that some novel docking proteins are required to transmit IL-2-mediated signals to Jak2 kinase. A signal-transducing adapter molecule (STAM), which contains a Src homology 3 domain and immunoreceptor tyrosine-based activation motif (ITAM) could play the role of such a docking protein (61). STAM is associated with Jak2 and Jak3 tyrosine kinases via its ITAM region and phosphorylated by Jak2 and Jak3 upon stimulation with IL-2 and other cytokines. The wild-type STAM, but not STAM mutants deleted of Src homology 3 domain or immunoreceptor tyrosine-based activation motif, significantly enhances c-myc induction mediated by IL-2 (61). Therefore, STAM is considered involved in the IL-2induced c-myc pathway. These signals are positioned immediately downstream of Jak kinases and potentially could transmit the signals between different Jaks. Further experiments are needed to identify all intermediates in this signal pathway. However, the data reported in this work represent, to our knowledge, the first demonstration of IL-2 up-regulation of c-myc expression upon activation of the Jak2/ STAT4 signaling pathway. We have identified a novel IL-2responsive element in the 5Ј-flanking region of c-myc and confirmed the role of STAT4 as the cognate binding protein for this element.
We believe that results of these studies will help in a better understanding of the complex control of c-myc expression and also of various molecular mechanisms by which cytokines control and regulate transcription of proto-oncogenes.