Specific activation of human interleukin-5 depends on de novo synthesis of an AP-1 complex.

It is clear from the biology of eosinophilia that a specific regulatory mechanism must exist. Because interleukin-5 (IL5) is the key regulatory cytokine, it follows that a gene-specific control of IL5 expression must exist that differs even from closely related cytokines such as IL4. Two features of IL5 induction make it unique compared with other cytokines; first, induction by cyclic adenosine monophosphate (cAMP), which inhibits other T-cell-derived cytokines, and second, sensitivity to protein synthesis inhibitors, which have no effect on other cytokines. This study has utilized the activation of different transcription factors by different stimuli in a human T-cell line to study the role of conserved lymphokine element 0 (CLE0) in the specific induction of IL5. In unstimulated cells the ubiquitous Oct-1 binds to CLE0. Stimulation induces de novo synthesis of the AP-1 members JunD and Fra-2, which bind to CLE0. The amount of IL5 produced correlates with the production of the AP-1 complex, suggesting a key role in IL5 expression. The formation of the AP-1 complex is essential, but the rate-limiting step is the synthesis of AP-1, especially Fra-2. This provides an explanation for the sensitivity of IL5 to protein synthesis inhibitors and a mechanism for the specific induction of IL5 compared with other cytokines.

Eosinophilia is a biologically specific phenomenon that can occur in the absence of increases in other leukocytes and is mediated by IL5 1 (1,2). Such a close relationship between a gene and a biological effect is unusual in the cytokine field where pleiotropy and redundancy are more common. By mediating eosinophilia, IL5 is involved in the pathogenesis of a number of allergic diseases, most notably asthma (3), rhinitis, and dermatitis (4). Although IL5 is reported to be produced by mast cells and eosinophils, the regulation of eosinophilia is primarily mediated by T-lymphocytes, which produce IL5 after activation (5). Activation of T-cells requires the interaction of the T-cell receptor with antigen, which leads to an increase in intracellular calcium concentration and activation of protein kinase C (6). A second signal provided by antigen-presenting cells is also required for optimal activation (7). Antigen-presenting-stimulated T-cells produce a wide range of cytokines, and IL5 is often but not always co-expressed with cytokines such as IL4 and IL13 (8). It is not clear why certain antigens induce IL5 and others do not; however, this indicates that the gene is individually regulated, and control mechanisms for IL5 expression can be based on antigen-specific T-cell activation.
Efficient production of IL5 in vitro requires activation by both phorbol myristate acetate (PMA) and a second signaling pathway, which can be stimulated by anti-CD28 antibody (9,10), cyclic adenosine monophosphate (cAMP), or calcium ionophore A23187 (CaI) (11)(12)(13). PMA/anti-CD28 stimulation activates expression of a variety of other cytokine genes including IL2, IL3, IL4, IL10, and granulocyte macrophage-colony stimulating factor (GM-CSF) (14). However, cAMP has an inhibitory effect on IL2, IL3, IL10, and GM-CSF (12,15,16) and no effect on IL4 (17). This suggests that IL5 expression is controlled by at least two independent costimulatory pathways. One of these is the CD28 pathway, which may be regarded as a common pathway for activation of cytokine genes. The other is the cAMP-dependent pathway, which in the context of T-cell cytokine genes, appears to be unique for induction of IL5 expression.
IL5, like many other cytokines, is regulated at the transcriptional level (5). Although control of IL5 expression is not fully understood, there is increasing evidence that unique mechanisms are involved in activation of the gene. For example, T-cell hybrids expressing IL5 and no other lymphokine have been produced (18), treatment of Th2 cells with IL2 induces IL5 messenger RNA (mRNA) expression but does not induce detectable amount of IL4 or GM-CSF messengers (19), and T-cells purified from peripheral blood of non-atopic asthmatic patients secrete elevated amounts of IL5 but not of IL4 (20). In addition, protein synthesis inhibitors cycloheximide (CHX) and anisomycin completely inhibit IL5 mRNA synthesis in primary T-cells and the murine T-cell clone D10.G4.1, but they do not inhibit the expression of IL2, IL3, IL4, IL10, and GM-CSF mRNAs in these cells (21)(22)(23). Thus, at least one protein critical for the induction of IL5 gene expression, but not for expression of other cytokine genes, is newly synthesized in response to T-cell stimulation. In this study we sought to identify the mechanism behind this dependence on de novo protein synthesis because this represents an important factor in the specific regulation of the gene.
The 5Ј-flanking regions of the human and murine IL5 RNA initiation sites contain a number of cis-regulatory elements that are involved in the control of IL5 production (8). Although these promoter elements contribute to the overall transcriptional activity of IL5, the conserved lymphokine element 0 (CLE0) in particular plays a very important role in IL5 gene transcription. Studies performed on both murine and human IL5 regulation demonstrate through deletion and mutation analysis that CLE0 is critical to IL5 expression (24 -28). Considering the critical role of CLE0 in IL5 expression, it seemed possible that the factor(s) involved in its activation might be synthesized de novo and provide the target for CHX inhibition.
The human T-cell line PER-117, which inducibly expresses IL2, IL4, and IL5 (29,30), provides a useful model for the study of IL5 expression. As in primary T-cells there is no detectable constitutive expression of IL5, but expression can be induced with PMA and further enhanced by cAMP and CaI. The transcription factors Oct-1, Oct-2, and AP-1 are involved in activation of IL5 transcription and exert their effects through CLE0 (28).
This study shows that, as in primary T-cells, IL5 gene expression in PER-117 cells is de novo protein synthesis-dependent, and CHX completely inhibits IL5 but not IL4 mRNA synthesis. Stimulation of the cells initiates de novo synthesis of the AP-1 complex binding to CLE0. Differences in the activation requirements for Oct-2 and AP-1 suggest that the AP-1 complex and not Oct-2 is of primary importance in the induction of IL5 expression.
Analysis of mRNA-RNA was extracted from 10 7 cells using the RNAeasy ® mini kit (Qiagen) according to manufacturer's instructions. The RNA was treated with RQ1 RNase-free DNase (Promega) and reverse-transcribed into cDNA using avian myeloblastosis virus reverse transcriptase (RT) (Promega). Polymerase chain reaction (PCR) of the cDNA was performed with the following specific primers: IL4, AGTGC-GATATCACCTTACAGGAGA and TTAAAATATTCAGCTCGAACAC; IL5, ACCTTGGCACTGCTTTCTACTCAT and AGAAACTCTTGCAGG-TAGTCTAGG; Fra-2, GCTTCTACGGTGAGGAGCCCCTGCAC and GGGTTACAGAGCCAGCAGAGTGGGGG; and Jun D, CGCAGCCT-CAAACCCTGCCTTTCC and CAAACAGGAATGTGGACTCGTAGC. Primers specific for ␤-actin were obtained from Clontech. The PCR mixture contained 1 l of cDNA, 1.0 M each primer, 0.5 mM MgCl 2 (2.5 mM for IL4), 0.2 mM deoxynucleotide triphosphates, and 1 unit Thermus thermophilus DNA polymerase (Promega) in Mg 2ϩ -free reaction buffer. The tubes were transferred into the preheated (94°C) thermocycler (PTC-100™, MJ Research), and DNA was denatured for 3 min. The samples were then subjected to 40 cycles of amplification at 94°C for 1 min, 60°C for 1 min (5 cycles at 58°C for 1 min and 35 cycles at 55°C for 1 min for Fra-2), and 72°C for 1 min. A two-step PCR (94°C for 1 min, 70°C for 2 min) was performed in the case of JunD. A final incubation of 72°C for 10 min was carried out for all samples. PCR products were fractionated on 1.5% agarose gel and photographed.
Electrophoretic Mobility Shift Assays (EMSA) and Western Blot Analysis (WB)-Nuclear proteins used in EMSA and WB analysis were isolated as described by Schreiber et al. (32) with the following modification; protease inhibitor mixture (Complete TM , Roche Molecular Biochemicals), 1 mM Na 3 V0 4 (Sigma), and 0.5 mM dithiothreitol (Sigma) were added to the reaction buffers just before lysis. After lysis nuclei were separated, and cytoplasmic proteins were further purified by centrifugation for 5 min at 10,000 ϫ g at 4°C. Protein concentration was determined using the DC Bio-Rad protein assay in three independ-ent assays. The mean of these assays was used in subsequent experiments. Standard EMSA binding reactions contained 3 g of nuclear extract, 60 mM KCl, 8 mM MgCl 2 , 12 mM Hepes, pH 7.9, 0.1 mM EDTA, 1 mM dithiothreitol, 0.5 g poly(dI⅐dC), 12% glycerol, and 25 fmol of 32 P end-labeled oligonucleotide probe (31). The oligonucleotide used as a probe was human IL5 CLE0, spanning nucleotides Ϫ59 to Ϫ38 of the IL5 promoter and containing the sequence GAAATTATTCATTTCCT-CAAAG (one strand shown). Probe preparation, protein-DNA binding reactions, and polyacrylamide gel electrophoresis were performed as described (33). DNA supershift experiments were carried out using antisera specifically directed against Oct-1 (sc-232), Oct-2 (sc-233), c-Jun/AP-1 (sc-44, reactive with all Jun members), c-Jun (sc-45), Jun B (sc-46), Jun D (sc-74), c-Fos (sc-253, reactive with all Fos members), c-Fos (sc-52), Fos B (sc-48), Fra-1 (sc-605), and Fra-2 (sc-604) supplied by Santa Cruz Biotechnology, Inc. This involved the inclusion of antibodies (2 g) to the reaction mixture and incubation for 2 h on ice before the addition of the labeled IL5 CLE0 probe.
To perform WB, 50 g of cytoplasmic fraction or 28 g of nuclear proteins were fractionated on 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to nylon membranes. Antiseras listed above (dilutions recommended by supplier) were used to detect protein expression. The amount of Oct-1 in each lane was used as loading controls. After incubation with the first antibody, the proteins were detected with an appropriate secondary antibody conjugated with horseradish peroxidase and chromogen substrate (ECL; Amersham Biosciences).
The PCR products, EMSA, or WB gel image was generated using a Kodak DC200 digital camera and Micrografx Picture Publisher 5.0 and printed on an HP laser jet 6p/6mp printer. Levels of formation of EMSA complexes were quantified by densitometric analysis using Scion Image, Release Beta 4.0.2.
Antisense Co-transfection Experiments-The pCR2.1hIL5p plasmid was used to isolate a 553-bp hIL5 promoter fragment. The fragment BamHI (Ϫ509) to EcoRI (ϩ44) from pCR2.1hIL5p was then subcloned in the BamHI/EcoRI site of pSP72 (Promega) creating the pSP-hIL5 construct. The pSP-hIL5 construct was subsequently digested with XbaI and BglII, isolated, and ligated into NheI/BglII digested pGL3-Basic (Promega). The resulting construct was called hIL5p and was used as the hIL5 reporter construct for transfection experiments. The Fra2 and JunD antisense constructs were made by ligating the PCR fragments obtained using the Fra2 and JunD primers described earlier into pGEM-T (Promega) then digesting the resulting plasmids with SalI/XhoI and inserting the fragments into similarly cut pSI (Promega). The final constructs pSIasFra2 and pSIasJunD contain 224 and 292-bp fragments, respectively, in reverse orientation.
All transfections were carried out in triplicate. 10 g of both antisense construct and hIL5 reporter construct DNA were electroporated at 960 microfarads and 280 V into 10 7 PER-117 cells in 400 l of growth media using the Bio-Rad gene pulser. After electroporation, the cells were incubated at 37°C for a period of 4 h before activation. For T-cell activation, PMA (Sigma), cAMP (Sigma), and CaI (Sigma) were used at concentrations of 10 ng/ml, 1 mM, and 0.25 M respectively. After an additional incubation period of 16 h, the cells were harvested and resuspended in 100 l of reaction buffer containing 50 mM Tris-HCl pH 7.8, 15 mM MgSO 4 , 33.3 mM dithiothreitol, 0.1 mM EDTA, 250 M lithium-CoA (Sigma), 500 M sodium luciferin (Molecular Probes), and 0.5% Triton X-100. Luciferase activity was measured in a Victor 1420 multilabel reader (Wallac, Finland).

RESULTS
CHX Inhibits IL5 but Not IL4 mRNA Synthesis-To investigate whether the activation of IL5 transcription in PER-117 cells was sensitive to inhibitors of protein synthesis, cells were treated with CHX. RT-PCR analysis showed no IL5 or IL4 mRNA in resting cells (Fig. 1, lane 1). Treating the cells for 4 h with 10 ng/ml PMA and 0.25 M CaI induced synthesis of these mRNAs (lane 2). CHX completely inhibited IL5 mRNA synthesis but had no effect on IL4 (lane 3). This indicates that, as in primary T-cells, at least one protein critical for the induction of IL5, but not IL4, transcription is newly synthesized in stimulated PER-117 cells.
Stimulation Conditions Modulate Protein Binding to the CLE0 -It has previously been shown that PMA/cAMP and PMA/CaI stimulation induce binding of an AP-1 complex consisting of Jun D and Fra-2 to CLE0 (28). Here we also show that three protein complexes were formed on a CLE0 probe with PMA-stimulated cell nuclear extracts (Fig. 2, lane 1). An antibody specific to the Oct-1 protein inhibited formation of the low mobility complex (lane 2). Similarly, an antibody to Oct-2 shifted the high mobility complex (lane 3). The third complex was partially inhibited by anti-Jun D and anti-Fra-2 (lanes 4 and 5) but not antibodies against c-Jun (lane 6) or other members of the Jun/Fos family (data not shown). Combined shift with both anti-Jun D and anti-Fra-2 completely removed the AP-1 complex (data not shown). These results indicate that Oct-1, Oct-2, and AP-1 (Jun D and Fra-2) bind to the CLE0 in PMA induced PER-117 cells.
EMSA were carried out with CLE0 probe, and nuclear extracts were prepared from PER117 cells stimulated under different conditions for 0, 2, 6, 12, and 18 h. Although different stimulation induces the same AP-1 complex, there were differences in the intensity and time course of interaction of AP-1 proteins to CLE0 (Fig. 3A). PMA stimulation gives a weak induction of AP-1, which reaches a maximum after 12 h. PMA/ cAMP induced AP-1 more rapidly, peaking between 6 and 12 h and declining by 18 h. In contrast, PMA/CaI induced a more intense band, which was maintained for at least 18 h. To determine whether the time course of binding of AP-1 factors to CLE0 correlated with IL5 protein production, densitometry analysis of AP-1 binding to CLE0 was graphed with IL5 protein production over time (Fig. 3B). The production of IL5 did correlate with the pattern of AP-1 binding, suggesting that AP-1 is critical for IL5 expression. A combination of all three stimuli (Fig. 3A) provided the best conditions for AP-1 induction and has been shown to induce the highest levels of IL5 production (data not shown). In contrast, Oct-2 binding did not correlate with the production of IL5 protein. For example, Oct-2 always appeared later than IL5 protein and is not binding under PMA/cAMP stimulation, where IL5 is produced. This suggests that Oct-2 is not involved in all signaling pathways and that different signaling pathways probably use alternate combinations of factors. CHX completely inhibited the formation of AP-1 and Oct-2 complexes (Fig. 3A), suggesting that de novo synthesis is involved in inducible protein binding to the CLE0.
Synthesis of IL5 CLE0-binding Proteins-To confirm that the CLE0-binding proteins were synthesized after cell activation, nuclear and cytoplasmic extracts were prepared from resting and stimulated cells and subjected to Western blot analysis. Similar amounts of Oct-1 were observed in cytoplasmic extracts of resting and activated cells (Fig. 4A, lanes 1-4). Fra-2 was only detectable in the cytoplasm of PMA/cAMPstimulated cells (lane 3). Oct-2 was present after activation except when cAMP was included (lanes 2 and 4). Jun D was not detected in cytoplasmic extracts (data not shown).
Analysis of the nuclear extracts showed that Oct-1 was constitutively present (Fig. 4B, lanes 1-5). In contrast, Jun D and Fra-2 protein was not detected in the nuclei of resting cells. Fra-2 was induced by all stimulations but was only detected after 6 h (lanes 3, 7, and 11), whereas Jun D was detected after 2 h (lanes 2, 6, and 10). Both cAMP and CaI enhanced PMAinduced Fra-2 synthesis (compare lanes 3-5, 7-9, and 11-13). However these coactivators did not affect PMA-induced Jun D. Oct-2 was present in resting cells, and it was further induced by stimulation with PMA or PMA/CaI (lanes 1-5 and 10 -13). Treatment of the cells with CHX abolished the induction of Fra-2, Jun D, and Oct-2 (Fig. 4C).
Fra-2 Is the Rate-limiting Factor for IL5 Production-RT-PCR for Fra-2 and Jun D mRNA in activated PER-117 cells indicates the clear induction of Fra-2 transcription compared with Jun D (Fig 5). The latter shows some constitutive expression (lane 1), and although no protein could be detected in unstimulated cells by Western blot (Fig. 4B), this suggests that it is the production of Fra-2, which is rate-limiting for IL5 expression.
Antisense Fra-2 but Not Antisense JunD Reduces IL5 Expression-Antisense experiments were carried out by co-transfecting constructs expressing antisense sequences for JunD and Fra-2 with a 500-bp hIL5 reporter construct (hIL5p), and the effect on luciferase activity was measured (Fig 6). The effect of the antisense or control constructs on the hIL5 500-bp promoter is shown as fold. The Fra-2 antisense construct reduced activity of the hIL5p construct by up to 75% the level of the hIL5p construct co-transfected with the control vector. This reduction was observed in PMA, PMA/cAMP, and PMA/CaIstimulated cells. The JunD antisense construct had no effect under any of the stimulation conditions. DISCUSSION The significance of IL5 in the production of eosinophils suggests a unique and tight control of gene expression. However, the gene-specific mechanisms of IL5 regulation are poorly understood. We report that de novo synthesis of AP-1 proteins, Fra-2, and Jun D controls activation of human IL5 CLE0 in T-cells. This identifies a gene-specific mechanism of induction of IL5 transcription and a regulatory mechanism involved in formation of the AP-1 complex on the IL5 promoter.
Considering the tight control of IL5 expression as well as the similarity in the development of eosinophilia in different species, it might be expected that the specific regulatory element would be highly conserved. However, it is surprising that the flanking region of the human and mouse gene has very limited homology. The region Ϫ20 to Ϫ80 is highly conserved, but upstream and 3Ј regions give little indication of common ele- ments that might be involved in the specific regulation of the gene. This conserved proximal region of the IL5 promoter harbors a cluster of regulatory sequences composed of CLE0 and GATA sites. Both of these sites act as positive regulatory elements and are critical for IL5 gene expression. The GATA element located immediately upstream of the CLE0 was shown to complex with transcription factor GATA-3. This factor was reported to drive development of Th2 cells and directly activate transcription of IL4 and IL5 genes (24, 27, 34 -38), suggesting that the GATA site is an essential, but "nonspecific" part of the proximal IL5 regulatory complex.
IL5 CLE0 and the interactions with Octamer and AP-1 factors are a good candidate for a gene-specific switch (27,28,39).
Although related elements exist in other cytokine genes, in each case these elements bind a unique protein complex. NF-AT, Fos, and Octamer transcription factors were shown to activate the analogous human IL4 P0 site (40). The GM-CSF CLE0, which differs by only a single base, has been shown to bind AP-1, NF-AT, and Ets (41)(42)(43). In addition, the important difference between the regulation of IL5 and these other cytokine genes is the sensitivity to inhibition by CHX. It is intriguing that AP-1 binds to GM-CSF CLE0, but expression is not CHX-sensitive. It is possible that AP-1 may not be as crucial to GM-CSF expression as we show here for IL5 or that different family members that are not CHX-sensitive are involved.
This study utilized a human T-cell line, PER-117, which inducibly expresses IL5 and IL4. As in primary T-cells there is no detectable constitutive expression of these cytokines, but expression can be induced with PMA in combination with costimulators. Cytokine production is inhibited by dexamethasone and cyclosporin A, so the genes appears to be functioning in a physiological fashion (30). We have used the differential activation of transcription factors by different stimuli to dissect the role of octamer and AP-1 factors on the CLE0 element. The effect of CHX was confirmed for IL5 in PER-117 cells compared with IL4. Thus, as in primary T-cells, IL5 gene transcription in PER-117 cells is de novo protein synthesis-dependent. Taken together these data indicate that the PER-117 cell line is an appropriate model to study signal transduction and transcriptional activation of the human IL5 gene.
EMSA results indicated that the constitutively expressed Oct-1 was not significantly affected by CHX, whereas binding of all the inducible proteins (Oct-2, Jun D, and Fra-2) was inhibited by CHX. Thus, any one of these proteins could provide the basis for the mechanism of action of CHX on IL5 expression.
Binding of the AP-1 complex to the DNA element was detected as early as 2 h after stimulation, whereas Fra-2 protein was only detected at 6 h, suggesting the EMSA is more sensitive than Western blotting. However, there is a remarkable correlation between AP-1 binding in EMSA and IL5 expression. Differences in induction of the binding proteins suggested a key role for AP-1 in IL5 production because there appeared to be a correlation between the timing and intensity of the bands and the levels of IL5 produced. For example, compare the effect of PMA and PMA/CaI in Fig. 3B. Also, the more sustained induction of AP-1 by PMA/CaI, resulting in higher IL5 production, compared with the transient effect of PMA/cAMP. This confirms the crucial role of AP-1 in the induction of IL5. The large increase of JunD at 6 h could be explained by the recently reported translational regulation of JunD in combination with the increased expression of JunD mRNA after cell stimulation (44,45).
Co-stimulation with cAMP induces IL5 but not other cytokines and, thus, provides a model for specific induction of IL5. It is clear from Fig. 3A that PMA/cAMP strongly induces AP-1; thus, at least part of the specificity of cAMP-induced IL5 probably results from the activation of Fra-2 and JunD. Why cAMP inhibits other genes is not clear.
The question then is whether one or both of Fra-2 and JunD are synthesized de novo to provide the major difference between the control of IL5 compared with other cytokine genes. JunD mRNA is present in unstimulated cells (Fig. 5), although it does increase after stimulation, suggesting some induction process. However, this constitutively expressed mRNA is not translated, because no protein is detectable in the cytoplasm and is only present in the nucleus after the cells are stimulated (Fig. 4B). Furthermore, the production of JunD is inhibited by CHX, indicating that new translation is occurring. The antisense experiments with JunD had no effect, which is consistent with pre-existing protein, but in the absence of clear data on the effectiveness of the antisense, this result needs to be interpreted with caution. Taken together, these experiments point to a requirement for newly synthesized JunD. In addition, the experiments with Fra-2 show conclusively that de novo synthesis is required. After stimulation the protein appears in the cytoplasm and the nucleus, protein production is inhibited by CHX, and the antisense Fra-2 caused a 2-4-fold reduction in IL5 expression. Because Fra-2 appears later than JunD, we propose that it may be the rate-limiting step in IL5 expression.
A distinguishing feature of these AP-1 proteins is their ability to be activated by either protein kinase C or protein kinase A pathways (46,47). Jun D was activated by the protein kinase C pathway (Fig. 4B, PMA stimulation, lanes 1-5) and apparently not increased by activation of additional pathways. However, for the full induction of Fra-2 (Fig. 4B, PMA/cAMP stimulation, lanes 6 -9) and the AP-1 complex (Fig. 3A) multiple pathways were required. Because the rate-limiting step for IL5 expression appears to be the synthesis of Fra-2, it is clear that multiple pathways are required for efficient IL5 expression. The previous data provide an explanation for the strong requirement for a costimulator in IL5 expression.
The EMSA analysis show constitutive binding of Oct-1 and stimulation specific binding of Oct-2 (Fig. 3A). Because there is only one Oct binding site in the CLE0, it is reasonable to assume that either Oct-1 or Oct-2 bind at any one time, and the appearance of both in EMSA is due to the excess of labeled probe in the EMSA reaction. Although present in the nucleus of unstimulated cells, the failure of Oct-2 to bind to hIL5 CLE0 indicates post-translational modification is required for bind- ing or that a different isoform is induced upon specific stimulation.
These results question the role of Oct-2 in IL5 expression as IL5 can be produced in the absence of Oct-2 (Fig. 3, A and B, PMA/cAMP stimulation). However, its importance should not be underestimated. Mutation of the Oct site in CLE0 is just as effective at blocking expression as mutation of the AP-1 site (28). Furthermore, overexpression of either Oct-1 or Oct-2 induces IL5 expression in unstimulated PER117 cells (28) (hence, in the absence of AP-1); thus, either Oct-1 or Oct-2 appear to be able to support IL5 expression. It should also be noted that mutation of the Ϫ400 GATA3 repressor site in the IL5 promoter gives high level constitutive expression, which overrides the need for any of the induced CLE0 transcription factors (48). Under physiological conditions it can be expected that different pathways of activation may vary the induction of Oct and AP-1.
Finally, it should be noted that other known IL5 transcription activators, GATA-3 and NF-AT (24,27,36,49), were observed in resting PER-117 cells (data not shown). This indicates that the modification but not de novo synthesis of these proteins is involved in induction of IL5 transcription. This is also in agreement with the properties of Th2 cytokine-producing cells, where GATA-3 and NF-AT transcription factors were shown to be present in resting cells in inactive form and to be activated by cell stimulation (34, 50 -52). This concept is also relevant to our findings with Oct-2, where it is constitutively present in the nucleus but binding to hCLE0 only occurs after stimulation. All these data support our hypothesis that the highly conserved CLE0 element provides an ON/OFF switch for IL5, and other promoter and 3Ј elements modulate the amount of IL5 production in individual T-cells (8). The present results point to a crucial role for CLE0 in the specific regulation of IL5, where the formation of an AP-1 complex is essential, and the rate-limiting step appears to be the synthesis of Fra-2.