Role of transcriptional repressor ICER in cyclic AMP-mediated attenuation of cytokine gene expression in human thymocytes.

Proliferating human medullary thymocytes can exhibit characteristic T helper cell type 1 cytokine responses exemplified by the immediate early expression of interleukin-2, interferon-gamma, tumor necrosis factor-alpha, and lymphotoxin-beta. Here we report that cAMP-mediated attenuation of the transcription of T helper-1-specific cytokine genes in human medullary thymocytes correlates with the induction of the cAMP-mediated transcriptional repressor ICER (inducible cAMP early repressor). We show that ICER binds specifically to several NFAT/AP-1 (nuclear factor of activated T cells/activating protein-1) composite DNA sites essential for the activation of the interleukin (IL)-2 promoter as well as to a homologous DNA motif present in the proximal segment of the interferon-gamma promoter. In the presence of the minimal NFAT DNA-binding domain, which is sufficient for both DNA binding and AP-1 complex formation, ICER and NFAT form NFAT/ICER ternary complexes on several NFAT/AP-1 DNA composite sites previously identified as essential for the expression of the immunoregulatory cytokines such as IL-2, IL-4, granulocyte-macrophage colony-stimulating factor, and tumor necrosis factor-alpha. In extracts prepared from human medullary thymocytes treated with forskolin and ionomycin, these composite sites bind endogenously expressed ICER either singly or in complexes. Moreover, in Jurkat cells, ectopically expressed ICER represses transcription from NFAT-mediated, phorbol ester/ionophore-activated IL-2, granulocyte-macrophage colony-stimulating factor, and tumor necrosis factor-alpha promoters. We present evidence that ICER interactions with NFAT/AP-1 composite DNA sites correlate with its ability to repress transcription. These findings provide further insight into the mechanisms involved in cAMP-mediated transcriptional attenuation of cytokine expression.

It is well established that cAMP signaling is inhibitory to T cell proliferation and effector functions. In particular, cAMP inhibits the expression of T helper-1 cytokine genes (1)(2)(3). Earlier reported studies of fibroblasts showed that elevated levels of intracellular cAMP inhibit upstream signal transduction pathways involved in cell growth and differentiation (4,5). In contrast to fibroblasts, in which elevated levels of intracel-lular cAMP inhibit extracellular signal-regulated kinases 1 and 2 and c-Jun NH 2 -terminal kinases involved in the signal transduction of mitogen-activated protein kinase pathways, T cell extracellular signal-regulated kinases 1 and 2 are insensitive to elevated levels of intracellular cAMP (6). Moreover, the cAMP-mediated inhibition of c-Jun NH 2 -terminal kinase in T cells shows delayed kinetics, an observation that correlates with the induction of the cAMP-inducible early repressor ICER 1 (7). In addition, overexpression of NFAT achieved by transfection of NFAT-encoding cDNAs to lymphoma cells abrogates the sensitivity of cAMP-mediated inhibition of IL-2 gene expression (8,9). Importantly, phosphorylation of aminoterminal serines of NFAT by protein kinase A does not prevent calcineurin-mediated translocation of NFAT to the nucleus, despite its ability to prevent IL-2 gene expression (10,11). The notion that a newly synthesized transcriptional repressor rather than inhibition of upstream signal transduction pathways could be involved in the cAMP-mediated transcriptional attenuation of T helper-1 cytokine expression was further strengthened by the reported alleviation of cAMP-mediated inhibition of IL-2 expression in the presence of inhibitors of both RNA and protein synthesis (6).
ICER is a transcriptional repressor that appears to serve as a generalized negative regulator of the CREB and CREM families of transcription factors as well as other related bZIP family members (12)(13)(14)(15). ICER isoforms represent a unique cAMPinducible CREM subfamily of transcription factors containing cAMP-response elements within an internal P2 promoter. Because of autoregulation of the cAMP-inducible P2 promoter, the expression of ICER can be intrinsically rhythmical. The rhythmical expression of ICER was first described in the pineal gland and in the hypothalamic-pituitary-gonadal axis (16,17). However, the P2 promoter of ICER is also inducible in organs other than the pineal and hypothalamic-pituitary gonadal axis such as in specific subsets of T lymphocytes including human medullary thymocytes (7). Importantly, in the Jurkat T cell line ectopically expressed ICER can substitute for the inhibitory effects of cAMP on the transcriptional attenuation of IL-2 promoter activity (7).
The NFAT (nuclear factor of activated T cells) and AP-1 (activating protein 1) represent two major transcription factor families implicated in the transcription of the IL-2 promoter in proliferating T lymphocytes (18 -20). To address the possible mechanism by which ICER down-regulates IL-2 gene expres-sion, we examined the binding of bacterially expressed ICER to all five NFAT motifs of the IL-2 promoter reported to be essential for the full induction of the IL-2 gene (21) either alone or in the presence of the minimal DNA-binding domain of NFAT (NFAT DBD). The highest affinity of ICER binding was found on a CD28-responsive element (CD28RE; Ϫ160 NFAT/AP-1 composite site) and Ϫ90 site, which is the motif in the IL-2 promoter that has striking sequence homology with the conserved proximal region (base pairs Ϫ73 to Ϫ48) of both the human and mouse promoters of the IFN␥ genes (21,22).
Moreover, certain NFAT/AP-1 composite sites that reside within the IL-4, GM-CSF, and TNF-␣ promoters resemble those located within the IL-2 promoter (23)(24)(25)(26)(27). It is believed that the mechanism underlying the actions of NFAT requires the binding of NFAT and/or NFAT/AP-1 to the NFAT/AP-1 composite binding motifs as ternary complexes (18). These complexes are believed to be essential for the transcriptional expression of immunoregulatory cytokines during T cell proliferation, such as IL-2, IL-4, GM-CSF, and TNF-␣ (19). Here we demonstrate that ICER binds to these NFAT/AP-1 composite DNA sites in vitro, either directly or indirectly via complex formation with the rel homology region of NFAT (NFAT DBD). Furthermore, we detect the induction of ICER-immunoreactive complexes in extracts prepared from human medullary thymocytes treated with forskolin and ionomycin. Ectopically expressed ICER represses transcription from the IL-2, GM-CSF, and TNF-␣ promoters activated by ionomycin and phorbol ester, suggesting that the induction of ICER in response to cAMP may be responsible for the observed cAMP-mediated transcriptional attenuation of T helper-1 cytokine responses.

MATERIALS AND METHODS
Preparation of Human Medullary Thymocytes-Human thymus glands were obtained from children (ages 3 months to 4 years) undergoing corrective cardiac surgery. Thymocytes were fractionated over discontinuous Percoll gradients (Amersham Pharmacia Biotech) (28). Cells with densities of 1.060 Ͻ o Ͻ 1.070 and p Ͼ 1.070 were collected and classified into large (significantly enriched for medullary thymocytes) and small (cortical) thymocytes according to an established protocol (7). Separated human thymocytes were maintained in short term cultures in RPMI 1640 medium, supplemented with 10% fetal calf serum, 50 units/ml penicillin, and 25 g/ml streptomycin, and treated as indicated.
RNase Protection Analysis-RNA extraction was performed as described (Qiagen). RNA probes hCK1 and hCK3 were purchased from Pharmingen and labeled with [␣-32 P]CTP using reagents from an RNA probe kit (Ambion). These probes were used for RNase protection studies according to the protocol provided by Ambion (RPAII ribonuclease protection assay kit).
Western Blot Analysis-Separation of whole cell proteins (50 g) was carried out by SDS-polyacrylamide gel electrophoresis (10%) for 2 h at 40 mA in Tris/glycine buffer (25 mM Tris, 250 mM glycine, and 0.1% SDS) at room temperature. The proteins were electrotransferred (0.4 A) overnight at 4°C in 10 mM Tris/glycine buffer with 12% methanol on Immobilon P membrane (Millipore Corp.). The membrane was blocked in a Tris-buffered saline solution containing 0.05% Tween 20 (Sigma) and 5% nonfat dry milk (Bio-Rad) for 1 h with gentle agitation at room temperature. For immunological detection, the same solution without dry milk but containing the ICER or CREM-specific antiserum (CS4) diluted 1:10,000 was agitated for 1 h, followed by three washes, with subsequent incubation with horseradish peroxidase conjugated to a secondary antibody (Amersham Pharmacia Biotech) diluted 1:5000 for 1 h, followed by nine washes, and finally developed using an ECL kit (Amersham Pharmacia Biotech).
Expression and Purification of Recombinant Proteins-Human ICER II cDNA was subcloned into the pGEXKG vector (Amersham Pharmacia Biotech) and expressed in bacteria as a glutathione S-transferase (GST) fusion protein. The pGSTagCREB construct was described previously (29). Purifications of both ICER and CREB were carried out with minor modifications according to the protocol previously established for CREB (29). NFATpXS-(1-187), encompassing the minimal DNA-biding domain of NFATp (gift from Dr. A. Rao), was expressed in bacteria as a hexahistidine-tagged protein and purified as reported previously (30). Recombinant c-Fos (Fos-(139 -243)) and c-Jun (Jun-(187-334)) (gift from Dr. T. Kerppola) were purified from Escherichia coli overexpression strains by nickel chelate affinity chromatography (31,32).
GST Pull-down Assays-GST-ICER (GST-CREB) Sepharose beads prepared as described above were diluted 1:10 in 50 mM Tris, pH 7.5, 150 mM NaCl, 0.5 mM EDTA, 10 mM Na 3 (PO 4 ) 2 , 10 mM NaF, 0.1% Triton X-100, 2.5 mM leupeptin, 20 mM phenylmethylsulfonyl fluoride, 100 g/ml aprotinin, and recombinant protein(s) were added to a final volume of 250 l and incubated at 4°C on a nutator for 1.5 h. The beads were then washed three times with the same buffer, resuspended directly in Laemmli buffer, and loaded on 10% SDS-polyacrylamide gel. Retained NFAT DBD protein was visualized by Western blotting described below using R59 anti-NFAT DBD antiserum.
Transient Overexpression in Jurkat T Cells-Transfection assays were performed by the DEAE-dextran technique. Typically, 10 7 Jurkat cells were transfected with 2 g of the reporter and the same amount of ICER expression vector and treated 18 h posttransfection with phorbol ester (10 Ϫ5 mg/ml) and ionomycin (1 g/ml) for 48 h. Luciferase and chloramphenicol acetyltransferase assays and quantification methods are described elsewhere (Promega; Ref. 33). The percentage of conversion of [ 14 C]chloramphenicol to its acetylated forms was quantified using ImageQuant (Molecular Dynamics).

cAMP-mediated Transcriptional Attenuation of T Helper-1responsive Cytokine Genes in Human Medullary Thymocytes-
Under the experimental conditions used, human medullary thymocytes exhibited the characteristics of naive T helper-0 and T helper-1 cells with predominant IL-2 and IFN␥ expression documented by RNase protection used for the evaluation of mRNA levels of multiple cytokines (IL-2, IL-4, IL-5, IL-9, IL-10, IL-13, IL-14, IL-15, interferon-␤, IFN␥, human migration inhibitory factor, TNF-␣, TNF-␤, Lt␤, and transforming growth factors ␤1, ␤2, and ␤3) (Fig. 1). Stimulation of medullary thymocytes with a combination of phorbol ester and ionomycin significantly induced the synthesis of mRNAs encoding IL-2, IFN␥ and to a lesser extent also TNF-␣ and Lt␤ (Fig. 1A, lane 4; Fig. 1B, lane 4). At the same time, cotreatment with forskolin or 8-bromo-cAMP reduced the cellular mRNA levels of IL-2, IFN␥, TNF-␣, and Lt␤ (Fig. 1A, lanes 5 and 6; Fig. 1B, lanes 5 and 6). We propose that at least part of this transcriptional attenuation is based on cAMP-mediated expression of the transcriptional repressor ICER and a subsequent blockade of NFAT/AP-1 composite DNA sites essential for T helper-1 cytokine expression. The inhibition by ICER may occur either directly through binding to the DNA element or indirectly via protein-protein interactions such as to the rel homology domain of NFAT (NFAT DBD).
Cyclic AMP-mediated Attenuation of T Helper-1 Cytokine Transcription in Human Medullary Thymocytes Correlates with cAMP-mediated Induction of the Transcriptional Repressor ICER-To test the proposed mechanism involved, we first sought evidence for the presence of ICER protein in human medullary thymocytes after forskolin treatment to explore whether ICER may interact with important NFAT/AP-1 enhancer motifs of the IL-2 and IFN␥ promoters. Western immunoblotting analysis using an ICER-specific antiserum (7) confirmed that ICER protein is indeed detectable in human medullary thymocytes but is absent in cortical thymocytes after 3 h of forskolin treatment (Fig. 1C). This finding is in agreement with the previously observed delayed appearance of ICER mRNA in medullary thymocytes after their exposure to forskolin and its subsequent deinduction (7). These observations suggest that the cAMP-mediated inhibition of cytokine expression may occur in a stage-specific manner in cells of T cell lineage.
ICER Binds to NFAT/AP-1 Composite Sites of IL-2 Promoter-To further address the possible mechanism by which ICER down-regulates IL-2 gene expression, we examined the binding of bacterially expressed ICER to all five NFAT motifs of the IL-2 promoter reported to be essential for the full induction of the IL-2 gene (21) (Fig. 2A). Four out of five NFAT sites in positions Ϫ90, Ϫ135, Ϫ160, and Ϫ280 were previously characterized as NFAT/AP-1 composite sites due to their inherent ability to bind to the NFAT/AP-1 complex in a cooperative fashion (34). The fifth NFAT site in the most proximal position, NFAT Ϫ45, does not bind to the NFAT/AP-1 complex and was determined to be exclusively an NFAT binding site (34). Bacterially expressed ICER, as well as ICER expressed in COS cells (data not shown), binds to all five NFAT sites, although to different extents (Fig. 2B). The strongest binding was observed to the NFAT/AP-1 composite sites in positions Ϫ90 and Ϫ160, an intermediate binding to the Ϫ135 NFAT/AP-1 composite site, and weak binding to the most proximal and distal NFAT and NFAT/AP-1 sites in positions Ϫ45 and Ϫ280, respectively. The binding specificity of recombinant ICER was evaluated using a CREM-specific antiserum (CS4) that "supershifts" ICER bound to specific oligonucleotides containing individually the five DNA motifs, leaving nonspecific binding unaffected (Fig. 2B, lanes 1-10) as well as by using control oligonucleotide encompassing the first 21-base pair repeat (H21) of the human T-cell lymphotrophic virus type I long terminal repeat promoter containing the CRE-like motif (35) (Fig. 2B, lanes 11 and 12).
ICER Binding to NFAT/AP-1 Composite Sites in the Presence of NFAT DBD-To better understand the role of the interactions of ICER with NFAT/AP-1 composite DNA motifs, we examined the effectiveness of ICER binding to the DNA motifs in the presence of NFAT DBD. This domain of NFAT, which has the highest degree of conservation among NFAT family members (36), is both necessary and sufficient for DNA binding as well as to associate AP-1 with NFAT (30). All NFAT motifs on the IL-2 promoter examined, with the exception of the NFAT/AP-1 composite site in position Ϫ90, bound NFAT effi-  2 and 3) or presence of phorbol ester and ionomycin (PMA ϩ Iono; lanes 5 and 6) were scored for cytokine expression using RIBOQUANT probes hCK1 (A) and hCK3 (B) in the RNase protection assay. Also shown are the corresponding RNase-protected probes following hybridization with yeast tRNA in the presence (ϩ) (lane 7) or absence (Ϫ) (lane 8) of RNase. Templates for the analysis of hL32 and hGAPDH housekeeping genes were included to allow assessments of total RNA levels (Pharmingen). Note that each probe (lane 8) migrates more slowly than its protected band; this is due to flanking sequences in the probe that are not protected by mRNA. C, Western immunoblotting using ICER-specific antiserum generated against a peptide encompassing the ICER-specific exon (7) shows induction of ICER in medullary thymocytes after 3 h but not after 12 h of forskolin treatment. There is no ICER protein detectable before treatment or in forskolin-treated human cortical thymocytes, which corresponds to a lack of detectable ICER mRNA reported previously in cortical thymocytes (7). ciently (Fig. 2C). The most pronounced ability of NFAT to associate with ICER was observed on the sites with the highest DNA binding efficiencies to both proteins, particularly at site position Ϫ160 (CD28RE), and to a lesser extent to the NFAT site in position Ϫ45. Interestingly, this NFAT Ϫ45 site tends to show a stronger binding of ICER to NFAT in the NFAT/ICER complex compared with ICER by itself (Fig. 2C, lanes 1-3). Because the NFAT Ϫ45 of the IL-2 promoter is the only one of the five sites examined without an adjacent AP-1 site, the finding that this site can form an NFAT/ICER complex suggests that ICER itself may tether to NFAT by a protein-protein interaction in addition to a protein-DNA action.
ICER Can Interact Directly with NFAT DBD-To examine whether or not ICER and NFAT may directly interact in the absence of DNA, GST pull-down assays were performed using a GST-ICER fusion protein linked to a Sepharose matrix in the presence of a truncated NFAT consisting of the NFAT DBD (Fig. 3, lane 1). GST-ICER formed a complex with NFAT DBD as demonstrated by retention of NFAT DBD. Interestingly, GST-CREB failed to associate with NFAT DBD under the same conditions (Fig. 3, lane 10). To examine the specificity of both components in the NFAT/ICER ternary complex found on the CD28RE (Ϫ160 motif) of the IL-2 promoter, we used both CREM-specific antiserum (CS4) and antiserum raised against the minimal DNA-binding domain of NFAT (R59). Both antisera either prevented or reduced the amount of the NFAT/ ICER complex (Fig. 2D, lanes 4 and 5, respectively). Competi-

FIG. 3. ICER protein can interact directly with NFAT DBD.
Decreasing amounts of NFAT DBD protein retained on Sepharose matrix with equal amounts of GST-linked ICER (GST-ICER, lanes 1, 3, and 6) beside a negative control represented by equivalent amounts of GST matrix alone (GST, lanes 2, 4, and 7) interact specifically in a GST pull-down assay. Lane 5 represents the input of NFAT DBD, which is equivalent to protein added to GST-ICER beads retained in GST pulldown in lane 3. In contrast, GST-CREB (lane 10) does not interact with comparable amounts of NFAT DBD protein (input, lane 8). Lane 9 represents GST-CREB beads alone. NFAT DBD retained on GST-ICER Sepharose beads was separated by SDS-polyacrylamide gel electrophoresis and visualized by Western blotting using the NFAT-specific antibody R59.

FIG. 2. ICER binding and formation of an NFAT/ICER complex with NFAT/AP-1 composite DNA-binding sites of the IL-2 promoter.
A, list of NFAT/AP-1 composite sites of IL-2 promoter delineated previously to be essential for IL-2 expression (34) used in electromobility shift assay analysis. NFAT and AP-1 (top) denote domains of homology between NFAT/AP-1 composite sites and consensus sequences for NFAT and AP-1 in human IL-2 promoter, respectively. Numbers on the left correspond to the relative distance of the depicted DNA-binding motifs from the TATA box of the IL-2 promoter. B, bacterially expressed ICER binds in boiled total bacterial lysate specifically to Ϫ90 (lanes 3 and 4) and Ϫ160 CD28RE motifs (lanes 7 and 8) and to a limited extent also to the rest of the motifs. The binding of ICER to these motifs is specific, since it is recognized by CS4 CREM-specific antiserum (CS4), causing a specific supershift (sICER), while the nonspecific complex (NS) remains unperturbed both in the presence of CS4 (ϩ) and normal rabbit antisera (Ϫ). The control CRE consists of an oligonucleotide encompassing the 21-base pair repeat of human T-cell lymphotrophic virus type I long terminal repeat (H21 CRE) (35).  lanes 6 and 13) and AP-1 (ap) (lanes 7 and 14) efficiently compete for complex. tion with unlabeled oligonucleotides containing NFAT (nf oligonucleotide spanning mouse IL-4 NFAT, positions Ϫ69 to Ϫ79) or the AP-1 motif (ap oligonucleotide spanning the human metallothionein IIA AP-1 site, positions Ϫ114 to Ϫ88) eliminated the NFAT/ICER complex (Fig. 2D, lanes 6 and 7, respectively), suggesting that both ICER and NFAT are essential components of the observed complex.

NFAT/ICER Ternary Complex Shows Distinct Features in Comparison with NFAT/AP-1 Complex on Various NFAT Motifs of IL-2 Promoter-Analogous experiments performed with the NFAT DBD and truncated forms of Fos and Jun confirmed
in vitro that both NFAT/AP-1 and NFAT/ICER complexes could exist in the context of the CD28RE (Ϫ160 NFAT/AP-1 composite site; Fig. 2D, lanes 8 -14). Interestingly, in the presence of NFAT DBD, the majority of ICER protein was retained in the complex with NFAT DBD, although NFAT/ICER and NFAT/ AP-1 complexes appear to show unequal binding affinities to different NFAT/AP-1 motifs. For example, the Ϫ280 motif of the IL-2 promoter, known from NFAT studies as the principal site for binding of the NFAT/AP-1 complex, binds NFAT/AP-1 complexes effectively (32,37), showing little or no detectable formation of an NFAT/ICER complex (Fig. 2C, lanes 13-15). In contrast, the CD28RE (Ϫ160 NFAT/AP-1 composite site) of human IL-2 promoter creates a ternary NFAT/ICER complex with an equal or slightly higher efficiency than the NFAT/AP-1 complex (Fig. 2D).
ICER Binds Directly to the Conserved Proximal Motif of IFN␥ Promoter-Unlike the CD28RE (Ϫ160 motif), which shows equally high affinity for both ICER and NFAT, the NFAT/AP-1 motif of the IL-2 promoter in position Ϫ90, which has striking homology to the conserved proximal element of the INF␥ promoter, does not interact with NFAT DBD (Fig. 4).
Studies performed on the conserved proximal motifs of both human and mouse IFN␥ promoters demonstrated high affinity ICER binding and a lack of NFAT binding or NFAT/ICER complex formation in the presence of NFAT/DBD (Fig. 4), a situation similar to that observed on the homologous Ϫ90 motif of the IL-2 promoter (Fig. 2).

NFAT/AP-1 Composite Sites in the Context of the GM-CSF, IL-4, and TNF-␣ Promoters Bind ICER either Alone or in Com-
plexes-NFAT/AP-1 binding sites have been shown previously to be essential for the efficient activation of the GM-CSF, IL-4, and TNF-␣ promoters. Therefore, we examined the binding to these sites of ICER and NFAT, both as purified recombinant proteins and in extracts prepared from human medullary thymocytes treated with forskolin and ionomycin (Fig. 5). These studies demonstrated that ICER can bind either by itself or in complexes with NFAT DBD to these composite sites in the promoters of the GM-CSF, IL-4, and TNF-␣ promoters, similar to the experiments using the binding site motifs of the IL-2 and IFN␥ promoters.
The GM-420 DNA motif strongly bound the purified NFAT/ ICER complex (Fig. 5B, lane 6), whereas the GM-330 and GM-550 motifs bound the complex much more weakly (Fig. 5B,  lanes 3 and 9). It is noteworthy that the GM-420 motif has been shown to constitute the essential enhancer core of the GM-CSF promoter (23). Likewise, NFAT/ICER readily formed a complex with the Ϫ80 element of the IL-4 and Ϫ95 element of the TNF-␣ promoter (Fig. 5B, lanes 12 and 15, respectively). Interestingly, the 3 motif of the TNF-␣ promoter, which contains an "inverted CRE" motif adjacent to the NFAT composite site (27), created a complex with an electrophoretic mobility different from those observed on the motifs of the IL-2, IL-4, and GM-CSF promoters (Fig. 2C and Fig. 5B). Furthermore, NFAT DBD alone formed a much slower mobility complex (lane 14), suggesting that NFAT may bind to the TNF-␣ motif as oligomers.
The treatment of isolated human medullary thymocytes with forskolin and ionomycin readily induced the expression of ICER (Fig. 5, D and E) not seen in uninduced thymocytes (Fig.  5C). The binding of ICER to the oligonucleotides containing the NFAT/AP-1 composite sites is inhibited by competition of the binding with a CRE-containing oligonucleotide (Fig. 5D, lanes  4, 7, and 10) or interference of the binding with antiserum to ICER (C-Ab) (Fig. 5E, lanes 4, 7, and 10).
Ectopically Expressed ICER Represses NFAT-mediated Activation of IL-2, GM-CSF, and TNF-␣ Promoters-To determine whether ICER expression could supplant the effect of forskolin in transcriptional attenuation of various cytokine promoters observed in medullary thymocytes, ICER (isoform II) was expressed in Jurkat T cells in transient transfection assays. Expression of ICER down-regulated the human IL-2, GM-CSF, and TNF-␣ promoters activated by the combined treatment of the cells with PMA and ionomycin, whereas ectopic expression of neither isoform of ICER did not prove to have any significant effect on VP16-mediated transactivation of (3ϫ GAL4)-CR-CAT under the same conditions (Fig. 6). Thus, ICER can be induced by, and substituted for, forskolin in the transcriptional down-regulation of the calcineurin-dependent, NFAT/AP-1-mediated transactivation of IL-2, GM-CSF, and TNF-␣ promoters when induced by PMA and ionomycin. DISCUSSION The mechanism of cAMP-mediated inhibition of cytokine expression in proliferating T lymphocytes has been attributed to cAMP-mediated inactivation of upstream signal transduction pathways directing the proliferation of T lymphocytes. Although this hypothesis was supported by studies of fibroblasts (5), it was not confirmed by studies of T cells (6). Surprisingly, several protein kinases required for T cell prolifera- FIG. 4. ICER binds to a conserved proximal element of the IFN␥ promoter homologous to the ؊90 motif of IL-2 promoter. A, ICER binds to both human and mouse motifs of the IFN␥ promoter (lanes 2 and 9), despite the fact that these motifs fail to bind NFAT DBD (NFAT) (lanes 3 and 10) in a fashion analogous to the Ϫ90 motif of the IL-2 promoter. No complex formation between ICER and NFAT was detected in electromobility shift assay (lanes 4 and 11). ICER bound to the proximal element of the IFN␥ promoter is specifically "supershifted" (sICER) by a CS4 CREM-specific antibody (C) (lanes 5 and 12). The mobility of the ICER complex is altered using competition with unlabeled CRE oligonucleotide (cre) (lanes 6 and 13) but unchanged in the presence of unlabeled NFAT oligonucleotide (nf) (lanes 7 and 14). B, sequences of human and mouse proximal elements of the IFN␥ promoter show significant homology to the Ϫ90 motif of the human IL-2 promoter.
tion were found to be insensitive or to exhibit a delayed response to high levels of intracellular cAMP (6). Here we provide evidence that in human medullary thymocytes expression of the transcriptional repressor ICER correlates with a delayed cAMP-mediated transcriptional attenuation of T helper-1 cytokine responses.
Footprinting and electrophoretic mobility shift analysis of the IL-2 promoter revealed (34) that the originally defined AP-1 site at position Ϫ150 of the IL-2 promoter (37, 38), a major CD28RE that contains an upstream NFAT binding site (39,40), represents a novel NFAT/AP-1 composite site at position Ϫ160 (34). A reexamination of the original observations in which NF-B was identified as a major component of the complex (41,42) determined that NFAT is a prevalent component of the complex that binds the CD28RE in vivo (34,43). Our findings indicate that the CD28RE (Ϫ160 composite site of IL-2 promoter) effectively binds ICER either alone or in an NFAT/ ICER complex. These findings may be important for obtaining a better understanding of a direct cAMP-mediated transcriptional attenuation of IL-2 expression. In addition, a potential indirect role of ICER has been demonstrated in transgenic mice overexpressing the dominant negative CREB mutant (a functional homologue of ICER), which impairs the expression of IL-2 in thymocytes (44). Since the induction of IL-2 and IFN␥ expression is dependent on the activity conferred by each of the individual DNA motifs (22,34), a demonstration of a direct binding of ICER and/or the formation of an inhibitory NFAT/ ICER complex on any of these NFAT/AP-1 composite sites could provide an explanation for the mechanism involved in the transcriptional attenuation of IL-2 and IFN␥ expression mediated by cAMP. These findings correlate with observations in which the conserved proximal motif of the IFN␥ promoter was reported to be inhibited by forskolin in proliferating thymocytes of mice made transgenic with an IFN␥ promoter-luciferase reporter gene (45). These findings further suggest that both NFAT/AP-1 motifs, either those that directly bind ICER or those that form NFAT/ICER complexes, could convey ICERmediated transcriptional attenuation.
It appears that numerous NFAT/AP-1 composite sites previously identified in the context of GM-CSF, IL-4, and TNF-␣ with affinities comparable with the CD28RE motif (Ϫ160) of the IL-2 promoter (Fig. 2). C-E, NFAT/AP-1 composite sites in GM-CSF, IL-4, and TNF-␣ promoters bind to ICER and form ICER containing complexes in whole cell extracts prepared from freshly prepared human medullary thymocytes. Bacterially expressed, purified ICER II and ectopically expressed full-length NFAT1 protein in COS cells served as controls (lanes 1 and 2, respectively) for evaluation of extracts (lanes 3-11) prepared from untreated human medullary thymocytes (C) or human medullary thymocytes treated with forskolin (3 h) and ionomycin (15 min) (D and E). ICER and ICER-containing complexes were competed by unlabeled oligonucleotides containing CRE motifs (cre) (lanes 4, 7, and 10) or NFAT motifs (nf) (lanes 5, 8, and 11) (panels C and D) and were immunoreactive with CS4 CREM-specific antisera (C) (panel E, lanes 4, 7, and 10), whereas the R59 NFAT-specific antisera (R) (lanes 5, 8, and 11) affected the mobility of numerous complexes but showed little effect on the binding of ICER (panel E).
promoters as essential determinants of their expression (23,25,27) can associate with ICER. This property does not seem to be a universal feature shared by all NFAT/AP-1 composite sites tested because we find that relatively minor differences in DNA sequences have a profound effect on both the binding of ICER and the formation of NFAT/ICER complexes. An example of this circumstance is represented by the GM-CSF promoter in which the GM Ϫ420 element shows binding for ICER and NFAT/ICER complexes, whereas both neighboring motifs GM Ϫ330 and GM Ϫ550 show only a modest binding of ICER and/or formation of NFAT/ICER complexes. The strong binding of ICER to the GM Ϫ420 element, previously defined by deletion analysis as the essential core of the GM-CSF enhancer (25), suggests that ICER may play an important role in transcriptional attenuation of GM-CSF expression. Similar binding studies performed with several NFAT/AP-1 composite sites important in the context of IL-4 and TNF-␣ promoters show that these sites, previously shown to be essential for efficient expression (23,26,27), bind ICER either alone or in complexes similarly to the motifs of the IL-2 and IFN␥ promoters. It remains to be determined whether the induction of ICER can selectively modulate T helper-1 versus T helper-2 cytokine expression in peripheral blood T lymphocytes.
We have reported previously that human medullary but not cortical thymocytes synthesize ICER mRNA after 3 h of forskolin treatment (7). Western immunoblot analysis using an ICER-specific antiserum confirmed that in these conditions the ICER mRNA is translated efficiently into ICER protein. Moreover, endogenously expressed ICER protein was detected in extracts prepared from human medullary thymocytes treated with forskolin and ionomycin using oligonucleotide probes containing NFAT/AP-1 DNA motifs that are able to form NFAT/ ICER complexes in vitro. In contrast to bacterially expressed ICER, endogenously expressed ICER in medullary thymocytes shows an altered mobility in gel shift assays, suggesting that posttranslational modification(s) may be involved in the regulation of the binding properties of ICER and/or degradative pathways involved in its proteolysis in vivo. The ICER-containing complexes that are immunoreactive to ICER-supershifting antisera are efficiently competed by oligonucleotides containing CRE or NFAT motifs. NFAT antisera that are unable to recognize directly bound ICER still affect the mobility of ICERcontaining complexes, suggesting the possibility of the formation of NFAT/ICER complexes in vivo. The ambiguity of the DNA-protein complexes in extracts of thymocytes observed on gel shift assays may be due to posttranslational modifications of the proteins involved (data not shown) and/or their potential consequences for DNA binding. At this point, the possibility cannot be excluded that proteins other than ICER and NFAT that contain homologous bZIP or rel homology regions may also participate in the formation of ICER-containing complexes. Finally, ectopic expression of ICER in Jurkat cells demonstrates that ICER, in agreement with its binding capabilities, can also effectively inhibit NFAT-mediated, phorbol ester/ ionophore-induced expression of IL-2, GM-CSF, and TNF-␣ promoters.
In conclusion, inducible ICER expression in developing human medullary thymocytes as well as in certain subset(s) of human peripheral blood lymphocytes (7) and monocytes (work in progress) could significantly influence their respective effector functions(s). The proposed inhibitory effects on effector function of the immune system mediated by ICER may be related to its ability to bind (mask) a wide range of CRE and AP-1 motifs and/or its ability to inactivate certain transcription complexes via protein-protein interactions.