Reconstitution of T Cell-Specific Transcription Directed by Composite NFAT/Oct Elements

RNA analyses

RNA was prepared from unstimulated cells and cells stimulated for 9 h with 20 ng/ml PMA and 1 μM A23187. IL-3, GM-CSF, and OCA-B transcripts were assayed by Northern blot hybridization analysis. ³²P-labeled probes contained the 480-bp BamHI/XmnI 3′ region of the human IL-3 gene, the 700-bp EcoRI/BamHI fragment of the human GM-CSF gene, and the human OCA-B gene isolated as a 2-kb EcoRI/NotI fragment of the expression plasmid pCDNA/OBF-1 (see below). Membranes were reprobed with the 400-bp EcoRI/HindIII fragment from the human GAPDH gene to control for RNA integrity and uniformity of sample loading.

DH site analyses

GM-CSF and IL-3 enhancer DH sites in Jurkat cells, CEM cells, KG1a cells, 5637 cells, K562 cells, and Raji cells were assayed as described previously (9, 16, 17). For each cell line, a DNase I titration was performed and samples that had optimal extents of DNase I digestion were selected for Southern blot hybridization analysis of DH sites. The membrane containing BamHI-digested samples used in the analysis of the IL-3 enhancer DH site was reprobed with a probe designed to detect a ubiquitous DH site immediately downstream of the IL-3 gene (12, 15) to confirm that DH sites could be detected in each sample.

Plasmid construction

All of the luciferase reporter gene constructs described in this work contained oligonucleotides or enhancers inserted upstream of a minimal promoter in pXPG-GM55 (18), which was made by cloning a minimal core fragment of the human GM-CSF promoter (−55 to +28) into the HindIII site of the luciferase reporter plasmid pXPG (18). pXPG is a modified form of pXP1 (19) that has the firefly luciferase gene replaced by the Luc⁺ gene from pGL3 (Promega, Madison, WI), and the pBR322 ori converted to a high copy ori. Like pXP1, pXPG has four upstream polyadenylation signals to block read-through transcription effects that in the past have affected equivalent pGL3-based plasmids employed by us (16, 18).

pIL3E contained the 330-bp NruI/AccI fragment of the IL-3 enhancer (16), and pGME contained the 425-bp BamHI/BalI fragment of the GM-CSF enhancer (10). Mutations in the IL-3 enhancer were generated by site-directed mutagenesis of the 330-bp IL-3 enhancer using oligonucleotides containing the following sequences: IL140 ΔN, GGCAAGAACCCTTGCTTTggCCACTGGGCCTTTCTT; IL190 ΔN, GGGTGCTCCATGGccAATGCAAATCTACTTAACTGA; IL190 ΔOct, TCCATGGAAAATGCccATCTACTTAACTGA; IL280 ΔN, GCTGTGAGCTACAGTTggCCAGCCTCTAGAGCC; IL3-AP-1, CTCCATGGAAAATGagAgTCaACTTAACTGACTTT. (Altered bases shown in lower case.) pGM420 contained a single copy of GM420 (10). Because IL190 is a weaker enhancer than GM420, we created pIL190 by inserting three copies of IL190 into pXPG-GM55 to increase the ability to detect residual activity in cells other than T cells. pIL140/IL190 contains the sequence CGGCAAGAACCCTTGCTTTTTCCACTGGGCCTTTCTTCCTCCCACCCTGAGGGTGCTCCATGGAAAATGCAAATCTACT. All DNA fragments were inserted in the same orientation relative to the promoters as they exist in the GM-CSF/IL-3 locus. The NFATp expression plasmid pLGP-NFAT1c was a gift from A. Rao (Harvard Medical School, Boston, MA; Ref. 20), the human OCA-B expression plasmid pCDNA/OBF-1 was a gift from P. Matthias (Friedrich Miescher Institute, Basel, Switzerland), the Oct-1 expression plasmid pWS1 and the Oct-2 expression plasmid pWS2 originated from W. Schaffner (21) and were gifts from T. Wirth (Ulm University, Ulm, Germany), the mouse NIP45 expression plasmid pCI-NIP45 was a gift of L. Glimcher (Harvard Medical School, Ref. 22), and the Brn-2 expression plasmid pCDNA3.1/Brn-2 was a gift of R. Sturm (University of Queensland, Brisbane, Australia).

Oligonucleotides

Previously described oligonucleotide duplexes (10, 16) used as probes and competitors incorporated the following sequences: IL190, GGGTGCTCCATGGAAAATGCAAATCTACT; GM420, CCATCTTTCTCATGGAAAGATGACATCAGGG; GM430, TCACACATCTTTCTCATGGAAAGATGA; octamer, CCTAATTTGCATG; AP-1, TGGATCACCCGCAGCTTGACTCATCCTTGCA.

Cell culture

The Jurkat T leukemic cell line, CEM T leukemic cell line, Raji B leukemic cell line, KG1a myeloid leukemic cell line, and K562 proerythroid/megakaryotic CML cell line were cultured in RPMI containing 10% FCS. The 5637 bladder carcinoma epithelial cell line was cultured in RPMI containing 7% FCS.

Transient transfections and luciferase assays

All cells were transfected with CsCl-purified plasmid DNA by electroporation, cultured for 20–24 h, stimulated with 20 ng/ml PMA and 1 μM calcium ionophore A23187 for 9 h, and assayed for luciferase reporter gene activities, as previously described (16). In all transfections, 5 μg of luciferase reporter plasmid was used. In cotransfection experiments, 5 μg of each expression plasmid or parent vector was also transfected.

Gel EMSAs

Nuclear extracts were prepared and EMSAs performed as previously described (9), except that assays employed 4 μg of nuclear extract and 2 μg of poly(dI-dC) in a 15 μl volume. EMSAs employed nuclear extracts prepared from unstimulated cells and cells stimulated for 2–3 h with 20 ng/ml PMA and 2 μM A23187. Some assays included specific antisera. The Brn-2 antisera raised against the C terminus of human Brn-2 (23) and the Oct-1 Ab (24) were gifts from R. Sturm. The Oct-2 antisera raised against the N terminus of mouse Oct-2 was a gift from L. Corcoran (Walter and Eliza Hill Institute, Melbourne, Australia).

Differential regulation of the IL-3 and GM-CSF genes

Before studying the tissue-specific regulation of the IL-3/GM-CSF locus, it was necessary to establish a well-defined model system that included a range of cell types that reflected the different properties of the two genes, and mirrored the activities of normal cells. To this end, we assayed IL-3 and GM-CSF gene expression in a variety of hemopoietic and nonhemopoietic cell lines following stimulation with a combination of the phorbol ester PMA and the calcium ionophore A23187 (Fig. 1⇑B). These signals directly activate the protein kinase C and calcium pathways in T cells and are known to synergize in the activation of IL-3 and GM-CSF gene expression. For consistency, the same activation protocol was adopted for each of the cell types employed in this study. As anticipated, the Jurkat and CEM human T cell lines both expressed IL-3 and GM-CSF mRNA in a highly inducible fashion. The myeloblastic cell line KG1a, the epithelial cell line 5637, and the CML cell line K562 all expressed GM-CSF, but not IL-3 in response to stimulation with PMA and A23187. These signals did not elicit expression of either gene in the Raji human B cell line. The ubiquitously expressed GAPDH gene was used as a control in this study to enable comparison of mRNA levels in the different cell lines.

To assess the activation status of the endogenous IL-3 and GM-CSF enhancers in the same panel of cells, we assayed the enhancer regions for the presence of inducible DH sites (top two panels, Fig. 1⇑C). This approach identifies chromatin-remodeling events and typically provides the first indication that a locus is undergoing activation. To confirm that we had selected appropriate DNase I-digested samples for this analysis, we also assayed the same samples for a ubiquitous constitutive DH site immediately downstream of the IL-3 gene (IL-3 +2.8 kb, Fig. 1⇑C (15). A strong correlation between the induction of DH sites in the enhancers and the activation of gene expression was observed for both genes. The IL-3 enhancer DH site was restricted to activated T cells (Jurkat and CEM), while the GM-CSF enhancer DH site was also induced in KG1a cells and K562 cells, but not in Raji cells. Although 5637 cells expressed moderate levels of GM-CSF, they did not contain a DH site within the GM-CSF enhancer. However, 5637 cells do exhibit a strong constitutive DH site in the GM-CSF promoter (15), and the GM-CSF promoter is highly inducible in transfected 5637 cells (15). These results infer that the GM-CSF gene is activated via enhancer-dependent mechanisms in cells such as Jurkat, CEM, and KG1a, which express NFAT, but enhancer-independent mechanisms in 5637 cells, allowing us to employ 5637 cells as a useful control in subsequent studies.

To correlate enhancer function with the above structural and expression data, the two enhancers were placed upstream of a minimal promoter in a luciferase reporter gene plasmid and tested in a representative group of four of the six cell lines (Fig. 2⇓A). Enhancer activity was assessed in transfected cells stimulated with PMA and A23187. The IL-3 and GM-CSF enhancers were both very active in the two T cell lines, increasing promoter activity by up to 80-fold. However, the two enhancers behaved very differently in KG1a cells in which the IL-3 enhancer was completely inactive, while the GM-CSF enhancer increased promoter activity by 110-fold. The two enhancers were both inactive in 5637 cells, in contrast to the full-length GM-CSF promoter and the SV40 enhancer that are both highly active when analyzed in 5637 cells by the same procedure in parallel assays (15). Hence, the activities of the two transfected enhancers directly paralleled the appearance of DH sites within the endogenous enhancers, reinforcing the direct link between chromatin remodeling and cytokine gene activation. These data are also consistent with other published and unpublished studies, demonstrating that the GM-CSF enhancer is active and the IL-3 enhancer is inactive in endothelial cells (15, 16) and mast cells (data not shown). All of the enhancer activities reported in this work represented inducible activity, as no significant luciferase activity was detected in any instance in the absence of stimulation (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Tissue-specific regulation of enhancers and NFAT sites in the IL-3/GM-CSF locus. Enhancers and composite NFAT elements were linked to a luciferase reporter gene and tested for enhancer function in transiently transfected cells stimulated for 9 h with 20 ng/ml PMA and 1 μM A23187. A, Luciferase activities of the IL-3 and GM-CSF enhancer plasmids pIL3E and pGME. B, Luciferase activities of pIL190 and pGM420. In each panel, error bars represent the SEM of at least six transfections, and enhancer activity is expressed as relative luciferase activity in which the promoter alone has an activity of 1.

Composite NFAT sites have activities that mirror their enhancers of origin

Having established that the IL-3 and GM-CSF enhancers had inducible activities that paralleled the activities of the endogenous genes, we sought the mechanisms that enabled activation of both enhancers in T cells, but just the GM-CSF enhancer in other cell types. As both enhancers consist primarily of arrays of NFAT sites, we focused on the properties of the two distinct classes of composite NFAT sites found within these two enhancers (Figs. 1⇑A and 2B). To select an example of a composite NFAT/AP-1 site, we chose the GM420 element from the GM-CSF enhancer. The GM420 element is the most active element in the enhancer and the only functional NFAT site with the capacity to bind NFAT independently of AP-1 (10). The GM420 element is located within the minimum enhancer core region associated with the DH site and is likely to contribute to cyclosporin A-sensitive chromatin-remodeling events within the enhancer. From the IL-3 enhancer, we selected the IL190 NFAT/Oct element that has overlapping NFAT and Oct elements that are both essential to its function as an enhancer element (16). The IL190 element is also the feature that best distinguishes the IL-3 enhancer from the GM-CSF enhancer. Although the IL-3 enhancer also contains a composite NFAT/AP-1 element (IL280, Fig. 1⇑A), it was not regarded as significant in this analysis, as it does not fit the preferred consensus needed for cooperativity (16), and it did not contribute to overall enhancer activity (see below).

The two different composite NFAT elements were assayed for enhancer function in the same system used above to test the full-length enhancers (Fig. 2⇑B). Oligonucleotides containing either the IL190 NFAT/Oct site or the GM420 NFAT/AP-1 site were introduced upstream of the promoter in the luciferase reporter gene plasmid. As observed with the full-length IL-3 enhancer, the IL190 element was very active in Jurkat and CEM T cells, but had little activity in KG1a or 5637 cells. Likewise, the GM420 element had the same profile of activity as the GM-CSF enhancer, as it was active in Jurkat, CEM, and KG1a cells, but generated little activity in 5637 cells. Additional analyses demonstrated that the IL190 element was also inactive in endothelial cells and mast cells under conditions in which the GM420 element was highly inducible (data not shown). To extend the observations made with GM420 to other composite NFAT/AP-1 sites, we performed similar assays with the GM550 element (Fig. 1⇑A) (10), which also functioned as an inducible enhancer in both Jurkat cells and KG1a cells (data not shown).

To ensure that the pXPG-GM55 promoter was not partly responsible for the tissue specificity of pIL190 and pGM420, similar studies were performed linking just one copy of either an NFAT/AP-1 site or an NFAT/Oct site to a promoter construct truncated at −33 to include just the TATA box. The NFAT/AP-1 element was sufficient to convert the TATA box to an inducible promoter in both Jurkat cells and endothelial cells. In contrast, the NFAT/Oct-TATA construct functioned in Jurkat cells, but not endothelial cells (data not shown). These observations not only confirmed the results in Fig. 2⇑B, but defined a single composite NFAT site linked to a TATA box as an entity sufficient to constitute an inducible tissue-specific promoter.

Mechanisms of IL-3 enhancer activation

To formally identify the DNA elements responsible for IL-3 enhancer activation in T cells, we performed site-directed mutagenesis on each of the three NFAT sites known to be able to function as enhancer elements and assayed these altered enhancers for function in transfection assays (Fig. 3⇓, A and C). Mutations in either the Oct or the NFAT motif of the IL190 element reduced the activity of the enhancer in Jurkat cells by 60–70%. Mutation of the adjacent high affinity IL140 NFAT similarly reduced enhancer activity by 60%. In contrast, a mutation in the IL280 NFAT/AP-1 site had no effect, but this was not unexpected, as this is a weak enhancer element with a poor AP-1 site. When an array of three copies of the IL280 element was tested in pXPG-GM55, it was only 5–10% as active as equivalent arrays of IL140 or IL190 elements (data not shown). This low activity may be due to the fact that the IL280 AP-1 site is 1 bp further apart from the NFAT site than is observed for sites that bind NFAT and AP-1 cooperatively (10, 16), and previous studies of the GM-CSF enhancer have found that lack of cooperativity in binding equates to lack of function (10).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Mutagenic analysis of the IL-3 enhancer in Jurkat T cells. Relative luciferase activity of plasmids containing IL-3 enhancer fragments was determined after stimulation of the cells for 9 h with 20 ng/ml PMA and 1 μM A23187 and calculated on the basis that pIL3E had an activity of 1. A, Site-directed mutagenesis of NFAT and Oct elements in the IL-3 enhancer plasmid pIL3E. Error bars represent the SEM of at least seven transfections. B, Deletion analysis of the IL-3 enhancer. Error bars represent the SEM of at least four transfections. C, Schematic diagram showing the presence or absence (by mutation) of the NFAT, Oct, and AP-1 elements in each of the constructs used in the above experiments and in Fig. 6⇓C.

Having demonstrated that the closely linked IL140 and IL190 elements were critical for enhancer activity, we next assessed whether the IL140/IL190 region was sufficient for enhancer function. A 79-bp DNA fragment encompassing just the IL140 and IL190 elements (Fig. 3⇑C) was found to be even more active than the 330-bp IL-3 enhancer fragment, indicating that this region does indeed represent the core of the enhancer. The higher apparent activity of the IL140/IL190 region could indicate the existence of a repressor in the proximal region of the enhancer, but we suspect that this increase in activity is more likely the result of positioning the enhancer core 150 bp closer to the minimal promoter. Similar increases in activity have been observed in parallel studies of the GM-CSF enhancer in this system when proximal deletions of nonessential sequences bring the enhancer core closer to the minimal promoter (data not shown).

The above findings implied that the NFAT and Oct elements were not only necessary, but perhaps sufficient for the T cell-specific activity of the IL-3 enhancer. However, as the Oct family protein Oct-1 is ubiquitous and NFAT controls both the IL-3 and the GM-CSF enhancer, we assumed that additional factors would be required to drive the T cell-specific activation of the IL-3 enhancer. To identify potential regulators of the IL-3 and GM-CSF enhancers, we conducted gel mobility shift analyses of NFAT, AP-1, and Oct family proteins present in nuclear extracts prepared from the six cell lines employed in the above studies (Fig. 4A⇓). In each case, AP-1-like proteins were induced by PMA and A23187. Strong induction of NFAT-like proteins was detected in three cell lines in which the GM-CSF enhancer was active (Jurkat, CEM, and KG1a), whereas K562 expressed very low levels and no NFAT was detected in Raji⁶ or 5637 cells. As expected, all of the cells expressed Oct-1 ubiquitously, and Raji B cells also expressed substantial amounts of Oct-2. However, we were surprised to also detect another Oct-like factor that comigrated with the N-Oct-3 complex derived from the human Brn-2 Oct family protein that is expressed in the brain and in melanoma (26, 27). Brn-2 was expressed at a high level in Jurkat cells and at low levels in CEM and K562 cells. The identities of the Oct-1, Oct-2, and Brn-2 factors seen in Fig. 4⇓A were confirmed in this study using a combination of Oct family Abs that eliminated the specific complexes described above and generated supershifted complexes above the Oct-1 band (Fig. 4⇓B). In addition, this analysis revealed that CEM cells also expressed a small amount of Oct-2. A minor species resembling Oct-2 in Jurkat, KG1a, and 5637 cells was not considered further, as it did not react with any of the Abs and may represent a simple artifact. The properties of the NFAT and AP-1 complexes have been described previously (10) and were not analyzed further. Note, however, that the GM420 AP-1 site can also associate with the closely related cAMP response element binding protein/activating transcription factor family of proteins (10).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Regulation of the IL-3 enhancer by inducible and tissue-specific transcription factors. A, Gel EMSA of transcription factors present in nuclear extracts prepared from either unstimulated cells (−) or cells stimulated for 3 h with 20 ng/ml PMA and 2 μM A23187 (+). The GM430 probe used to identify the presence of NFATp represents just the NFAT-binding component of the GM420 element. An AP-1 site from the stromelysin gene was used as an AP-1 probe, and an Oct consensus sequence was used as a probe for Oct family proteins. As these complexes have been described in previous studies, only the specific bands of interest are shown. B, Identification of specific Oct family proteins. Nuclear extracts from unstimulated cells were assayed as in A, but in the presence or absence of Abs for Brn-2 or Oct-2. An Oct-1 Ab was used with the Raji extract to verify that the ubiquitous species was Oct-1. C, Northern blot analysis of OCA-B mRNA. RNA was made from either unstimulated cells (−) or cells stimulated for 9 h with 20 ng/ml PMA and 1 μM A23187 (+). Filters were reprobed with GAPDH to verify RNA integrity (not shown). D, Activation of the IL190 element in Jurkat T cells by cotransfection of pIL190 with transcription factor expression plasmids. Cells were cotransfected with the Oct-1 plasmid WS1, the Oct-2 plasmid WS2, the OCA-B plasmid pCDNA/OBF-1, the Brn-2 plasmid pCDNA3.1/Brn-2, the NFATp plasmid pLGP-NFAT1c, or the NIP45 plasmid pCI-NIP45. Luciferase activity was determined after stimulation of the cells for 9 h with 20 ng/ml PMA and 1 μM A23187. In each case, luciferase activity is expressed relative to the pIL190 construct cotransfected with a parent vector. Error bars represent the SEM of at least four transfections.

The lymphoid-specific Oct cofactor OCA-B (28), otherwise known as OBF-1 (29) or Bob-1 (30), was also considered as a potential tissue-specific activator of the IL-3 enhancer. OCA-B is known to be expressed constitutively in B cells and is induced in activated T cells (31, 32). In T cells, the inducible activity of OCA-B is also dependent on serine phosphorylation via an inducible kinase (31). We assessed OCA-B mRNA expression by Northern blot analysis (Fig. 4⇑C). As anticipated, OCA-B was expressed ubiquitously in Raji B cells and was induced to similar levels in activated Jurkat and CEM T cells. OCA-B mRNA was also expressed at very low levels in activated KG1a cells, but not in 5637 or K562 cells.

The preceding analyses of the factors and cofactors present in T cells highlighted Oct-1, Oct-2, Brn-2, OCA-B, and NFAT family proteins as potential activators of the IL-3 enhancer. An additional cofactor that had to be considered as a candidate was the NFAT-binding cofactor NIP45 that is expressed predominantly in lymphoid cells (22). However, the NIP45 expression pattern in the cell lines employed in this study was not determined, as a human NIP45 probe was unavailable, and we were unsuccessful in our attempts to detect NIP45 using a mouse NIP45 probe.

The roles of these potential regulators of the IL-3 enhancer were investigated by introducing them on expression plasmids in cotransfections with the pIL190 luciferase plasmid containing three copies of the IL190 element (Fig. 4⇑D). In activated Jurkat cells, Oct-1, OCA-B, and NFAT were the most potent activators of pIL190, increasing its inducible activity by 5–8-fold, whereas Oct-2 increased expression by about 2-fold. The NFAT cofactor NIP45 also increased pIL190 activity 2-fold and had the capacity to cooperate with NFAT in this process to increase the degree of enhancement to 8-fold compared with the 5-fold activation seen with NFAT alone. In contrast, Brn-2 was a strong repressor of pIL190. Parallel assays in CEM T cells confirmed that Brn-2 was a repressor of the IL190 element in cells, in which it was normally active, as overexpression of Brn-2 reduced activity by 60% and transfection of a Brn-2 antisense RNA expression plasmid increased IL190 activity by 40% (data not shown). Along these lines, it was also interesting to note that Jurkat cells produced the highest levels of Brn-2 and that the IL-3 enhancer was less active in Jurkat cells than CEM cells. Hence, the activation of the IL190 NFAT/Oct element in T cells is likely to involve a complex assembly of at least four distinct factors, with Brn-2 working to dampen Oct-dependent activation.

Reconstitution of T cell-specific Oct/NFAT-dependent transcription

The above studies indicated that at least two distinct transcription factors plus their specific cofactors were necessary for efficient activation of the IL190 NFAT/Oct element. In a bid to recapitulate the environment of a T cell, we introduced OCA-B, NFAT, and NIP45 expression plasmids in cotransfections of K562 cells that express high levels of Oct-1, but low levels of NFAT and no detectable OCA-B. Following activation with PMA and A23187, the combination of NFATp, NIP45, and OCA-B gave rise to a striking 50-fold increase in the activity of pIL190, which had previously supported an activity indistinguishable from that of a plasmid containing just the core promoter (Fig. 5⇓A). The effects of these three factors were specific for the IL190 element, as they had no influence on the plasmid containing just the promoter. The restoration of IL190 activity was highly dependent upon OCA-B, as the combination of NFATp and NIP45 had negligible effect, while OCA-B alone increased pIL190 activity by 10-fold.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

Reconstitution of IL190 NFAT/Oct enhancer function in cells other than T cells. pIL190 was cotransfected with plasmids expressing NFATp, NIP45, or OCA-B, as in Fig. 4⇑D. Relative luciferase activity was determined following 9-h stimulation with 20 ng/ml PMA and 1 μM A23187, and expressed on a scale in which pIL90 cotransfected with a parent vector had an activity of 1. A model indicates in each case, the transcription factors present in the cell line before and after cotransfections leading to maximal activity from the IL190 element. A, Mean of at least five transfections in K562 cells. B, Mean of at least three transfections in Raji B cells. C, Mean of at least four transfections in KG1a cells. In each case, error bars represent the SEM.

The Raji B cells employed in this study presented a distinctly different model for reconstituting IL190 activity, as they expressed Oct-1, Oct-2, and the Oct cofactor OCA-B, but not NFAT. In cotransfections of Raji cells, NFATp and NIP45 synergized to increase the activity of pIL190 by 10-fold (Fig. 5⇑B). NIP45 clearly needed its DNA-binding partner NFATp to play any role in this study, as no change in activity was detected after cotransfection of just NIP45. NFATp and NIP45 also synergized to activate the IL140 and GM420 NFAT elements in Raji cells (data not shown), but had no effect on the promoter alone (Fig. 5⇑B).

The pIL190 plasmid was next assayed in cotransfections of KG1a myeloid cells, which express high levels of Oct-1 and NFAT, but relatively low levels of OCA-B compared with Jurkat, CEM, and Raji. In this instance, OCA-B and NIP45 cooperated to increase inducible pIL190 activity 6-fold (Fig. 5⇑C), whereas OCA-B alone increased activity 3-fold and NIP45 alone increased activity 2-fold. The actions of these factors were again seen to be specific for the IL190 element, as they had no effect on the promoter alone. As in Jurkat cells (Fig. 4⇑D), Brn-2 was also implicated as an antagonist of Oct/OCA-B-dependent transcription in KG1a cells (Fig. 5⇑C). Cotransfection of a Brn-2 expression plasmid suppressed the weak inducible enhancer activity supported by the IL190 element in KG1a cells, and suppressed its activation by cotransfection of either OCA-B or NIP45 (Fig. 5⇑C).

The otherwise inactive IL190 element was similarly activated by the combination of OCA-B and NIP45 in mast cells and endothelial cells that express Oct-1 and NFAT, but not OCA-B (data not shown).

The above transfections of pIL190 (containing an array of three Oct/NFAT elements) (Fig. 5⇑) produced some Oct-dependent activities that were not observed below with the native IL-3 enhancer, which contains just one Oct element (Fig. 6⇓). Assays of pIL190 in K562 and Raji cells indicated that three linked Oct/NFAT elements were sufficient to support a modest degree of activity in the presence of just Oct-1 and OCA-B, but required NFAT for maximum effect (Fig. 5⇑, A and C). Even in the absence of any cotransfected factors, the IL190 plasmid had some activity in Raji B cells that express Oct-1, Oct-2, and OCA-B constitutively (Fig. 5⇑C). In contrast to the normally inducible native IL-3 enhancer, the weak pIL190 activity seen in Raji cells in the absence of cotransfection was entirely constitutive, while the increased activity seen after NFATp cotransfection was inducible (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

Reconstitution of IL-3 enhancer activity in cells other than T cells by cotransfection and mutagenesis. Cells were transfected with luciferase plasmids and stimulated for 9-h stimulation with 20 ng/ml PMA and 1 μM A23187. Relative luciferase activity was calculated on a scale in which pIL3E had an activity of 1. A, K562 cells were transfected with pXPG-GM55 or pIL3E and cotransfected with NFATp-, NIP45-, and OCA-B-expressing plasmids or with parent vector. Error bars represent the SEM of at least three transfections. B, Raji B cells were transfected with pXPG-GM55 or pIL3E and cotransfected with NFATp- and NIP45-expressing plasmids or with parent vector. Error bars represent the SEM of at least four transfections. C, KG1a cells were transfected with either pXPG-GM55, pIL3E, or pIL3AP-1, in which the Oct element in the enhancer was changed to an AP-1 site (Fig. 3⇑C). Error bars represent the SEM of at least four transfections. D, Model of differential regulation of the IL-3 and GM-CSF enhancers in different cell types.

NFAT and Oct factors and cofactors are sufficient to activate the intact IL-3 enhancer outside of T cells

Having shown that an array of NFAT/Oct elements could be activated by a specific combination of factors, it was now necessary to determine whether the same group of four factors was sufficient to activate the native IL-3 enhancer that contains an array of NFAT sites, but only one Oct element. This question was addressed by assaying the activity of pIL3E in K562 and Raji cells that had been cotransfected with the NFAT, NIP45, and OCA-B expression vectors and stimulated with PMA and A23187, as above. In the absence of cotransfected factors, the IL-3 enhancer was completely inactive in both K562 and Raji cells (Fig. 6⇑, A and B). However, the activity of the IL-3 enhancer was restored in K562 cells by the combination of NFATp, NIP45, and OCA-B (Fig. 6⇑A) and in Raji cells by NFATp and NIP45 (Fig. 6⇑B). As was the case with pIL190, NIP45 and NFATp synergized in the activation of pIL3E and, as expected, NIP45 was highly dependent on NFAT for its ability to activate. This restoration of IL-3 enhancer activity required just the core IL140/IL190 region because a similar 19-fold activation of pIL140/IL190 was observed in transfected Raji cells overexpressing NFATp and NIP45 (data not shown).

The above studies implied that the NFAT/Oct element was the significant site that distinguished the IL-3 enhancer from the GM-CSF enhancer and dictated its T cell specificity. To determine whether a single site in a complex enhancer could in fact govern cell specificity, we performed the simple operation of converting the Oct site to an AP-1 site to recreate an element resembling the NFAT/AP-1 sites present in the GM-CSF enhancer (pIL3-AP1, Fig. 3⇑C). Remarkably, this mutation was sufficient to generate an enhancer that could now function in KG1a cells, whereas the native IL-3 enhancer is completely inactive in the cells (Fig. 6⇑C). From these studies, we conclude that NFAT can activate transcription by at least two distinct mechanisms (Fig. 6⇑D): a T cell-specific pathway in which NFAT recruits or assists binding of Oct factors and the cofactors NIP45 and OCA-B, and a more general pathway in which NFAT activates transcription by recruiting the ubiquitous transcription factor AP-1 with no apparent need for additional tissue-specific cofactors.

Contributions of NFAT and Oct elements to tissue-specific gene expression

There are many examples in which NFAT and Oct elements contribute to activation of tissue-specific gene expression, but their role in directing a specific pattern of expression has not previously been determined. The T cell-specific IL-2 promoter contains an array of four composite NFAT/AP-1 elements (39) that, while essential for its activity, cannot account for its T cell specificity. When introduced into transgenic mice, the distal IL-2 NFAT/AP-1 element was sufficient to support activation of transcription, but its activity was not confined to T cells (40). The IL-4 promoter depends on interactions between NFAT and AP-1 (41), but this is insufficient for T cell-specific expression as it also requires factors such as c-maf (42).

Our studies have highlighted the importance of Oct elements in T cell-specific transcription, and the replacement of the sole Oct element by an AP-1 element was sufficient to relieve restrictions on the activity of the IL-3 enhancer. The T cell-specific IL-2, IL-3, and IL-4 promoters each contain Oct elements in addition to NFAT and AP-1 elements (43, 44, 45, 46, 47 ; unpublished data). Hence, the combination of NFAT, AP-1, Oct, NIP45, and OCA-B proteins may provide a general mechanism for activating the T cell-specific expression of a subset of cytokine genes.

Oct elements also contribute to B cell-specific transcription (reviewed in Ref. 48), and they are conserved features of the promoters of B cell-specific Ig genes. An isolated Oct element linked to a TATA box supports transcription in B cells, but not fibroblasts (49), and OCA-B is essential for Oct-dependent transcription in B cells (50). However, although the predominantly lymphoid proteins Oct-2 and OCA-B have both been proposed as mediators of B cell-specific transcription (28, 29, 30, 51), neither protein is necessary or sufficient for activation of B cell-specific promoters. Ig genes are still expressed in both OCA-B and Oct-2-deficient B cells (32, 52), and Oct-2 expression has been detected in a variety of hemopoietic cell types (53). In the case of the IgH enhancer, the octamer is dispensable for enhancer function as a tripartite unit comprising sites for Ets-1, TFE3, and PU.1 and is sufficient to reconstitute enhancer activity in B cells, but not fibroblasts (reviewed in Ref. 54). These observations highlight the degree of redundancy employed in the combinatorial regulation of Ig genes.

The activity of Oct elements may also be regulated by the ratio in the expression of different Oct family proteins, and in this study the IL-3 enhancer had a greater relative activity in the T cell line that expressed the least amount of Brn-2 relative to Oct-1. Unlike Oct-1 and Oct-2, Brn-2 lacks the ability to recruit OCA-B (30). Hence, we hypothesize that Brn-2 may in some situations function as a repressor of Oct elements by serving as a competitive inhibitor of the Oct/OCA-B complex.

Mechanisms of T cell-specific transcription

Although we have presented the first clear mechanism for inducible T cell-specific transcription, there are other examples of developmentally regulated T cell-specific enhancers, and individual transcription factors that contribute to programs of T cell-specific gene regulation (reviewed in Refs. 55, 56). The best example is the TCR α enhancer, which contains an array of individual binding sites for the transcription factors ATF (activating transcription factor), AML-1, and Ets-1, and the architectural proteins lymphocyte enhancer-binding factor-1 or T cell-specific factor-1 (TCF-1) that are required for the assembly of an enhanceosome (reviewed in Refs. 12, 55). In this instance, T cell specificity is also derived from combinatorial regulation as the predominantly lymphoid factors Ets-1, lymphocyte enhancer-binding factor-1, and TCF-1 cooperate with AML-1, which is expressed by T cells and myeloid cells, but not B cells. Interestingly, AML-1 also regulates the activities of the IL-3 promoter (57) and the GM-CSF promoter and enhancer (58), and thereby has the potential to contribute to the T cell-specific activation of IL-3 in T cells and the activation of GM-CSF expression in both T cells and myeloid cells.

The IL-3 locus is regulated via several distinct mechanisms

Although the combination of NFAT, Oct, NIP45, and OCA-B is able to activate the IL-3 enhancer, it is not sufficient to activate the whole IL-3 locus, as the endogenous IL-3 gene remained silent upon cotransfection of these factors into a variety of non-T cell types (data not shown). The enhancer may not even be essential for IL-3 expression, as OCA-B-deficient T cells still express normal levels of IL-2 and IL-3 (32). Similar to the situation in Ig genes (54), the IL-3 locus appears to be controlled by a complex and redundant set of mechanisms. Activation of the IL-3 locus outside of T cells also requires activation of the T cell-specific IL-3 promoter, which has binding sites for AML-1 and GATA proteins in addition to NFAT, Oct, and AP-1 sites.

In the IL-3 locus, the IL-3 enhancer responds to TCR signals when an immune response is elicited, but we believe this only represents the end stage of a complex series of events that begin much earlier in T cell development. An array of constitutive T cell-specific DH sites extends up to 5 kb upstream of the IL-3 gene (9). Because they exist even in primitive T cell lines such as Molt4 that do not express IL-3 and in which the IL-3 enhancer is inactive (unpublished data), we suggest that they function as a chromatin-opening domain that primes the locus for subsequent activation via the TCR. This developmental pathway resembles the chromatin remodeling that occurs in the IFN-γ and IL-4 genes during the course of T cell differentiation (59). When naive T cells differentiate down distinct pathways, DH sites appear in the IL-4 locus in Th2 T cells and in the IFN-γ locus in Th1 T cells, where these genes gain the potential to be induced via the TCR.

Given the complexity of the IL-3 locus, it is not surprising then that activation of the IL190 element alone was insufficient to activate IL-3 gene expression outside of T cells. This is in contrast to studies of the IL-4 locus, in which association of NFAT, NIP45, and c-maf with proximal elements of the promoter was sufficient to induce expression of the endogenous IL-4 gene in B lymphoma cells (22). Although our studies have not identified a mechanism that is sufficient to drive IL-3 expression, they have clearly identified an important mechanism driving T cell-specific transcription that may be widely used.

Combinatorial regulation maintains tight control over gene expression

In this study, we demonstrated that complex enhancers can be broken down to simple composite elements that retain the properties of the intact enhancers. In doing so, we have provided an example of the power of combinatorial regulation to fix patterns of gene expression. Composite NFAT/Oct and NFAT/AP-1 elements retained the inducible tissue-specific properties of their parent enhancers, and these patterns were essentially the same as those of the endogenous IL-3 and GM-CSF genes to which they were linked. The IL190 element represented the minimal unit that could function as a T cell-specific enhancer, but in the context of the native IL-3 enhancer it appeared to function in concert with the closely linked IL140 NFAT site. An array of IL190 elements was not as active as the IL140/IL190 enhancer core region, suggesting that correct enhancer function may require assembly of a specific enhanceosome-like complex. Such a complex could also include architectural proteins such as HMGI(Y) because NFAT binding is promoted by HMGI(Y) (60) and the NFAT and Oct sites in the IL-3 enhancer encompass A/T-rich segments likely to favor HMGI(Y) binding.

Composite NFAT sites also represent points at which major signaling pathways converge, thus ensuring very tight regulation of cytokine gene expression. Furthermore, NFAT/AP-1 complexes and OCA-B each require both a calcium and a kinase signal to efficiently activate transcription (5, 31). This presents a requirement for additional specific kinases to phosphorylate proteins such as Jun and OCA-B. The complexity and interdependence of the signaling pathways and factors involved means that cytokine genes require correctly delivered signals in the right cell type for gene activation to occur.