Abstract
Activation of lamina propria (LP) T cells via the CD2 pathway enhances IFN-γ (IFN-γ) secretion with further enhancement after CD28 coligation. The molecular mechanisms regulating IFN-γ expression in LP T cells remain unknown. Previous studies in PBL and T cell lines identified cis- and trans-regulatory elements in TCR-mediated expression of IFN-γ. This study examines CD2 and PMA/ionophore-responsive IFN-γ promoter elements. Activation of LPMC via CD2-induced IFN-γ secretion and a parallel up-regulation of mRNA expression. CD28 coligation enhanced mRNA stability without up-regulating transcription as measured by nuclear run-on. Transfection of a −2.7-kb IFN-γ promoter-reporter construct into PBL and LP mononuclear cells (LPMC) revealed significant promoter activity after CD2 activation, with additional transactivation after CD2/CD28 costimulation in PBL, but not in LPMC. Functional analysis using truncated promoter fragments identified distinct cis-regulatory regions selectively transactivating IFN-γ expression in PBL compared with LPMC. In PBL, CD2 activation elements reside within the −108- to +64-bp region. However, in LPMC the upstream region between −204 and −108 bp was essential. Transfection of the proximal and distal AP-1-binding elements, as well as TRE/AP-1 constructs, revealed functional activation of AP-1 subsequent to CD2 signaling, with activation critical in PBL but diminished in LPMC. Electromobility shift analysis using oligonucleotides encompassing the proximal, distal, and BED/AP-1-binding regions failed to demonstrate selective transactivation after CD2 signaling of LPMC. This report provides evidence that activation of LPMC results in transactivation of multiple promoter elements regulating IFN-γ expression distinct from those in PBL.
Interferon is an important immunoregulatory protein with tightly controlled expression predominantly in activated T cells and NK cells (1, 2). As has been shown with other inflammatory mediators, expression of IFN-γ in the mucosa is different from that in the periphery. IFN-γ transcription and mRNA synthesis closely parallel those observed with IL-2, suggesting that differences in the capacity of T cells to produce IFN-γ are determined primarily at the transcriptional level (3, 4). Although the IL-2 promoter has been well characterized, the mechanisms involved in regulation of IFN-γ gene transcription are less well defined. The human IFN-γ gene structure consists of four exons and three introns and is highly conserved among mammalian species (5, 6). A number of 5′-flanking cis-regulatory promoter regions have been identified in PMA/ionophore-activated T cell lines, including proximal (−73 to −48) and distal (−96 to −80) regions that are conserved between human and rodent (7). Both regions are believed to contain binding sites for jun/fos AP-1 nucleoprotein complexes (8). The distal region contains a GATA-regulatory motif similar to that found in the promoter regions of GM-CSF and macrophage-inflammatory protein-1αβ and has been shown to contain GATA-3 as part of the nucleoprotein complex interacting with this region (9). The proximal region displays homology with the NFIL2A region of the human IL-2 gene and interacts with cAMP response element-binding protein/activating transcription factor (ATF)3 and AP-1 nuclear binding factors (9).
It has been proposed that in T cell lines the selective binding of c-Jun or c-Jun/ATF-2 to the proximal region results in a positive signal and activation of transcription. A diminution of transcription results from binding of other competing factors such as cAMP response element-binding protein/ATF-1, or after methylation of this site (10). Additional cis-binding elements that inhibit IFN-γ expression, including a silencer repressor element (−251 to −215 bp) that can bind Yin Yang 1 (YY-1) and an AP-2-like protein have been reported upstream of the minimal promoter region (−108 to +64) (11, 12). An additional YY-1 site (−211 to −186) has been identified that overlaps with an AP-1-binding site. It has been suggested that binding of YY-1 to this site in resting T cells blocks constitutive transcription of IFN-γ, whereas displacement of YY-1 by AP-1 results in transactivation of the IFN-γ gene. Limited analyses of expression of this site in PMA + PHA-activated PBMC suggests that recruitment of AP-1 to this site triggers selective transactivation of IFN-γ expression during the differentiation of naive T-cells to memory T cells (13).
Two signals are required to achieve maximal activation of PB T cells. The first signal is generated by engagement of the TCR, whereas a second signal is provided by a costimulatory molecule. One major costimulatory T cell surface molecule is CD28 (14, 15). Previous studies have reported that costimulation of TCR activated PBL with CD28 results in enhanced T cell proliferation, as well as cytokine production including IFN-γ (16). The molecular mechanisms that regulate increased IFN-γ production, however, remain undefined. Initial nuclear run-on studies of TCR-activated PBL indicated that CD28 costimulation resulted in enhanced IFN-γ mRNA stability without transcriptional up-regulation (17). Subsequent transfection studies using PMA/ionophore activation of T cell lines, however, suggested a modest up-regulation of IFN-γ mRNA expression after CD28 costimulation but failed to identify a CD28 response element (18). The presence of AU-rich sequences within the 3′-untranslated region of the mRNA of many cytokines, including IFN-γ, has been demonstrated to facilitate the rapid turnover and instability of the mRNA (19). Numerous proteins that bind to these AUUUA motifs regulate cytokine mRNA stability; however, the precise mechanism of posttranscriptional regulation of IFN-γ remains to be determined.
The activation pathways of LP T cells are distinct from those of PB T cells (20, 21). LP T cells do not respond well to activation via the TCR/CD3 receptor, yet they do exhibit increased proliferation and cytokine production when activated via the CD2 pathway (21, 22). CD28 coligation further enhances the activation, and LP T cells are generally thought to manifest a heightened activation state compared with PB T cells. This activated state can be further amplified in conditions of dysregulated inflammation, such as Crohn’s disease and ulcerative colitis. Crohn’s disease is characterized by widespread mucosal inflammation involving an enhanced T cell activation state characterized by increased production of inflammatory mediators, including IFN-γ (23).
Previous studies have demonstrated that there are mucosa-specific mechanisms for T cell cytokine gene regulation. Recent studies suggest that regulation of IL-2 production in LPMC after CD2 activation may involve a different mechanism from that observed in PBL and T cell lines (24). The experiments described herein were designed to determine 1) the regulatory mechanisms involved in enhanced IFN-γ secretion in CD2/CD28-activated LP T cells, and 2) whether regulatory elements distinct from those previously reported for PBL and T cell lines play a role in IFN-γ secretion by LP T cells. In this study, evidence demonstrates that the IFN-γ promoter possesses multiple CD2-responsive enhancer elements located between the −204- and +64-bp region and that transactivation of IFN-γ expression in PBL and LPMC occurs through the use of different cis-regulatory elements and requires the recruitment of different transactivating factors.
Materials and Methods
Monoclonal Abs
Anti-CD2 mAbs (clones CB6 and GD10) were a gift from Chris Benjamin (Biogen, Cambridge, MA). Anti-CD28 ascites (clone 9.3) was obtained from Bristol-Meyers Squibb Pharmaceutical Research Institute (Princeton, NJ). The ascites was purified over a protein G column and quantified by ELISA.
Purification of LPMC
Intestinal specimens were obtained from patients undergoing surgical resection of the colon at Cedars-Sinai Medical Center, Los Angeles. Approval for the use of human subjects was granted by the Institutional Review Board at Cedars-Sinai Medical Center. In this study, all tissue specimens were taken from an uninvolved area of resected colon from patients with colonic carcinoma (normal), involved areas from patients with ulcerative colitis, and uninvolved and involved areas from patients with Crohn’s disease.
LPMC were isolated from the resection samples by a technique modified from that described previously (25). Briefly, the intestinal specimen was washed with HBSS, and the mucosae were dissected away from the underlying layers. The mucosal layer was incubated in a shaking water bath (100 rpm) in calcium- and magnesium-deficient HBSS, containing 1 mM EDTA, 50 μg/ml gentamicin, 100 U/ml penicillin, 100 μg/ml streptomycin, and 50 μg/ml Fungizone, with the solution changed every 30 min until the supernatant was free of epithelial cells. The remaining LP was minced into 1- to 2-mm pieces and digested for 10 min in RPMI 1640 containing 10% FCS, 0.5 mg/ml collagenase B (Boehringer Mannheim, Indianapolis, IN), 1 mg/ml hyaluronidase (Sigma, St. Louis, MO), 0.1 mg/ml DNase I (Sigma), 50 μg/ml gentamicin, 100 U/ml penicillin, 100 μg/ml streptomycin, and 50 μg/ml Fungizone in shaker water bath (100 rpm). The supernatant was collected, filtered through 110-μm nylon mesh (Spectrum Laboratory Products, Houston, TX), and centrifuged at 500 × g for 5 min. The cell pellet was resuspended in 15 ml and centrifuged at 30 × g for 5 min to remove epithelial and other large cells. The supernatant was removed, and lymphocytes were isolated by separation on Ficoll-Hypaque gradients. The cells were then washed three times with HBSS and resuspended in RPMI 1640 containing 10% FCS.
Stimulation of mononuclear cells
For stimulation through the CD2 receptor, LPMC were stimulated with 0.1 μg anti-CD2 Abs (both CB6 and GD10 clones)/106 cells at 37°C for the times indicated for each experiment. CD28 costimulation was conducted with 0.1 μg anti-CD28 Ab. Stimulation of T cells with anti-CD2 Abs did not require further cross-linking because the combination of two anti-CD2 Abs directed against different epitopes was sufficient to induce activation.
IFN-γ assay
IFN-γ was measured by an amplified ELISA assay (26). Dynatech (Burlington, MA) Immulon 3 microtiter plates were coated overnight with 100 μl 5 μg/ml monoclonal anti-IFN-γ (Endogen, Woburn, MA). Samples and standards were added for 24 h followed by addition of 100 μl of 2.5 μg/ml polyclonal rabbit anti-IFN-γ (Endogen) for 2 h. This was followed by addition of 100 μl 1:1000 diluted mouse anti-rabbit alkaline phosphatase-conjugated Ab (Jackson ImmunoResearch, West Grove, PA) for 2 h. Substrate, 0.2 mM NADP (Sigma), was added for 30 min followed by addition of amplifier (3% 2-propanol, 1 mM iodonitrotetrazolium violet, 75 μg/ml alcohol dehydrogenase, and 50 μg/ml diaphorase, Sigma) for 30 min. Plates were read at 490 nm using an E max plate reader (Molecular Devices, Sunnyvale, CA). All data acquisition and reduction were performed with the ELISA Master program for Macintosh computers, developed by R. L. Deem.
Northern blot analysis
Total cellular RNA was extracted using the RNeasy kit (Qiagen, Chatsworth, CA). RNA was separated electrophoretically on a denaturing 1% agarose gel containing 7% formaldehyde. Gels were transferred to nylon membrane (Amersham, Arlington Heights, IL) and hybridized to 32P-labeled DNA probe as previously described (24).
Nuclear run-on
PBL or LPMC (5 × 107) were stimulated and nuclei were isolated as previously described (27). In vitro transcription was conducted at 26°C for 20 min in transcription buffer (50 mM HEPES (pH 7.9), 100 mM KCl, 2 mM DTT, 30 μM EDTA, 1 mM ATP, 0.5 mM GTP, 0.5 mM CTP, 2 mM MnCl2, 35 mM (NH4)2SO4, 8.8 mM creatine phosphate, 40 μg/ml creatine phosphokinase) and 100 μCi [α-32P]UTP. Labeled mRNA transcripts were purified with the RNeasy kit for liquid samples and hybridized to 2 μg cDNA insert immobilized on a nylon membrane.
Preparation of nuclear protein extracts
Nuclear protein extractions were conducted with 5–10 × 106 LPMC. After activation, cells were centrifuged, washed in cold PBS, and kept on ice for subsequent extraction steps. The cell pellet was resuspended in 0.9 ml of RSB (10 mM Tris (pH 7.4), 10 mM NaCl, 3 mM MgCl2, 0.5 mM DTT, 2 μM leupeptin, 1 μg/ml aprotinin, 1 mM PMSF, 0.1 mM EGTA), and 0.1 ml of 5% Nonidet P-40 was added. Samples were mixed by gentle inversion and kept on ice for 10 min followed by centrifugation. The pellet was resuspended in 25–60 μl (volume is dependent on the starting number of cells) cold buffer C (20 mM HEPES (pH 7.4), 0.42 mM NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 25% v/v glycerol, 0.5 mM DTT, 20 μM leupeptin, 10 μg/ml aprotinin, 1 mM PMSF). Samples were incubated on ice for 30–40 min during which time they were pipetted twice. Cellular debris was removed by centrifugation, and nuclear proteins were diluted with an equal volume of buffer D (20 mM HEPES (pH 7.4), 50 mM KCl, 0.2 mM EDTA, 20% v/v glycerol, 0.5 mM DTT, 20 mM leupeptin, 10 μg/ml aprotinin, 1 mM PMSF). Protein concentrations were determined by Coomassie Plus assay (Pierce Chemical, Rockford, IL).
Gel (EMSA)
Double-stranded oligonucleotide was end-labeled with [γ-32P]ATP and T4 polynucleotide kinase. Nuclear extract protein (3–6 μg) was incubated at 25°C with 0.25 mg/ml poly(dI-dC), in 20% glycerol, 5 mM MgCl2, 2.5 mM EDTA, 2.5 mM DTT, 250 mM NaCl, 50 mM Tris (pH 7.5) for 10 min. The oligonucleotide was then added (20,000 cpm), and the binding reactions were incubated for an additional 30 min. Specificity was determined by the addition of 100-fold excess unlabeled oligonucleotide as competitor. The DNA-protein complexes were separated from unbound probe on a prerun native 5% polyacrylamide gel in low ionic strength buffer (22.3 mM Tris (pH 7.4), 22.3 mM borate, 0.5 mM EDTA (pH 8.0)). After 2 h, the gel was dried under vacuum and exposed to x-ray film. The oligonucleotides used were as follows: proximal conserved element, TTGTGAAAATACGTAATCC; distal conserved element, GCCTATCTGTCAAACTCTCAT; BED region AP-1-binding site, ATGGGTCTGTCTCATCGTCAAAGGA.
DNA constructs
The human IFN-γ cDNA clone was obtained from American Type Culture Collection (Manassas, VA). The human IFN-γ luciferase reporter plasmids containing −2.7-kb and −204-bp fragments of the IFN-γ promoter has been described previously (9) and were subcloned upstream to the luciferase gene. The −2.7 kb IFN-γ luciferase reporter plasmid was subcloned by Dr. Masahiro Shiroo. The IFN-γ promoter-reporter constructs containing truncated promoter fragments, −538, −108, −39 bp, and the internal deletion mutant Δ−214/−178, as well as the dimer of proxIFN-γ and tetramer of distIFN-γ (gifts from Laurie Penix, Yale University, New Haven, CT) were subcloned upstream of the luciferase gene as described previously (9, 28). The plasmid TRE2 luciferase (a gift from M. Karin, University of California, San Diego) used to determine AP-1-dependent transactivation was generated by subcloning two copies of the collagenase AP-1-binding sites into a luciferase reporter plasmid (29).
Transfection
Freshly isolated LPMC were primed for transfection competence by culturing for 16 or 20 h, respectively, in RPMI 1640 containing 10% FCS, 50 mM 2-ME, and 1 μg/ml PHA-L (Sigma) as previously described (24, 30). Cells were then washed and resuspended in 250 μl fresh medium at 2 × 107 cells/ml and electroporated in the presence of 50 μg reporter construct (250 V, 2250 μF, 48 ohms) using 4-mm (gap width) cuvettes in a BTX Electro Cell Manipulator (Genetronics, San Diego, CA). After electroporation, the cells were diluted in fresh medium, allowed to rest for 1 h before plating, and then stimulated with anti-CD2 or anti-CD2 + anti-CD28 mAbs for 4 h. Luminescence was measured with a Promega (Madison, WI) luciferase assay kit and counted on a 6-detector Wallac 1450 Microbeta liquid scintillation counter (Wallac, Gaithersburg, MD) with coincidence counting deactivated.
Results
CD28 cosignaling augments IFN-γ expression by LPMC activated via the CD2 pathway
LP T cells represent a distinct class of lymphocytes that are significantly more responsive than PB T cells to activation via the CD2 pathway (21). Stimulation of LP T cells through the CD2 pathway elicits proliferation as well as secretion of IFN-γ and other cytokines. In contrast to PB T cells, LP T cells do not respond well to activation via the TCR/CD3 pathway. Costimulation of the accessory CD28 molecule synergizes with CD2, resulting in a marked increase of IFN-γ secretion in LPMC (21).
To determine the molecular events involved in regulation of IFN-γ secretion by LP T cells, LPMC were isolated and activated with anti-CD2 Abs in the presence or absence of CD28 costimulation. As seen in Fig. 1,A, a measurable amount of IFN-γ was detected in supernatants of LPMC as early as 3 h after CD2 activation and continued to rise over 24 h. CD28 costimulation further enhanced IFN-γ production levels, particularly at early time points up to 6 h postactivation. As shown in Fig. 1 B, Northern blot analysis of mRNA isolated from LPMC after activation with anti-CD2 or anti-CD2 + anti-CD28 revealed that the increase in the level of IFN-γ was paralleled by an increase in IFN-γ mRNA expression. Expression of IFN-γ mRNA was detectable as early as 1 h after activation by CD2 and continued to rise over 24 h. CD28 costimulation of the CD2 activation pathway resulted in a significant increase in the levels of IFN-γ mRNA expression by LPMC at 2 h that was sustained over 24 h.
CD28 costimulation enhances the stability of IFN-γ mRNA
The enhanced accumulation of IFN-γ mRNA observed after CD28 costimulation of LPMC could be a result of an increase in the transcriptional rate or in posttranscriptional modification of IFN-γ mRNA. To determine whether CD28 augmentation of mRNA levels in LPMC was a result of enhanced mRNA stability, LPMC were activated with anti-CD2 or anti-CD2 + anti-CD28 for 2 h (maximum for mRNA expression). Actinomycin D was then added to prevent further transcriptional initiation, and mRNA decay was monitored for the indicated periods of time. Fig. 2 shows that after CD2 activation, newly synthesized IFN-γ mRNA exhibited rapid decay, with a half-life of 43 min. CD28 coligation resulted in stabilization of IFN-γ mRNA and extended the half-life of IFN-γ mRNA from 43 min to 228 min.
CD2, but not CD28, mediates transcriptional up-regulation of the IFN-γ promoter
In tumor T cell lines, CD28 costimulation enhanced transcriptional activation of transfected IFN-γ promoter elements (18), yet nuclear run-on studies conducted in PBL failed to detect transcriptional activation (17). To determine whether enhanced transcriptional activation of the IFN-γ promoter was involved after CD2 + CD28 costimulation of PBL or LPMC, nuclear run-on assays were performed on cells utilizing the same activation protocol as that used for transfection studies. As seen in Fig. 3, a basal level of IFN-γ gene transcription was detectable in unstimulated PBL and LPMC and was up-regulated after CD2 activation. However, as reported previously (17), CD28 costimulation did not result in enhanced transcription, as measured by run-on assays, over that observed by CD2 alone in PBL or LPMC. An equivalent level of β-actin was detected in both unstimulated and stimulated conditions. No signal was detected hybridizing to the plasmid vector alone (data not shown).
Transactivation of the IFN-γ gene appears to be complex with numerous, partially defined regulatory elements. However, transgenic mice generated with a full length human genomic fragment including the −2.7-kb 5′-flanking promoter region express human IFN-γ in a tissue-specific manner (31). These studies suggest that the key elements necessary for IFN-γ gene regulation reside within the 2.7-kb region of the IFN-γ promoter. To investigate the role of these elements in CD2 pathway activation of IFN-γ production, a −2.7-kb IFN-γ promoter-luciferase construct was transfected into PBL and LPMC. CD2 activation resulted in a marked up-regulation of reporter gene activity in both PBL and LPMC (Fig. 4). CD28 costimulation of PBL resulted in further enhancement of promoter activity (Fig. 4,A). However, in contrast to PBL, coligation of CD28 on LPMC did not increase IFN-γ transactivation over that observed by CD2 alone (Fig. 4 B). The results suggest the existence of a CD2 response element within the −2.7-kb promoter region of both PBL and LPMC with additional CD28 transcriptional response elements activated in PBL but not LPMC.
CD2 signaling results in transactivation of multiple regulatory elements within the IFN-γ promoter region
To identify the CD2 response elements within the −2.7-kb region of the IFN-γ promoter, we transfected both PBL and LPMC with a promoter-construct encompassing the entire −2.7-kb region. Expression was then compared with a series of constructs truncated to −538 bp, −204 bp and −108 bp immediately upstream from the IFN-γ transcriptional start site, as well as the basal −39-bp construct. Fig. 5 shows that the cis-regulatory regions involved in the transactivation of IFN-γ gene expression in PBL are different from those in LPMC. In CD2-activated PBL, deletion of the region between −538 and −204 bp resulted in reduction of promoter activity to levels of the basal −39-bp construct, although significant CD2 responsiveness was retained within the −108-bp region (Fig. 5,A). Moreover, the −108-bp construct consistently exhibited enhanced expression as compared with the −204-bp construct, suggesting the presence of a CD2 response repressor element within the −204 to −108 region. A similar pattern of cis-regulatory regions was involved after PMA/ionophore activation (Fig. 5,C). In LPMC, deletion of the region from −2.7 kb down to −204 bp did not diminish CD2-directed activation of the IFN-γ promoter-reporter construct (Fig. 5,B). Furthermore, the −204-bp reporter construct consistently exhibited enhanced response to CD2 activation after transfection compared with the −2.7-kb and −538-bp promoter constructs, suggesting the presence of a CD2 response repressor element upstream of −204 region. Truncation of the region between −204 and −108 bp resulted in a significant reduction of promoter activity, although a basal level of CD2 responsiveness was retained within the region from −108 to +64. A similar pattern of cis-regulatory regions to those identified after CD2 activation was involved in promoter transactivation after PMA/ionophore activation of LPMC (Fig. 5 D).
These results suggest that LPMC have multiple CD2 and PMA activation response elements within the IFN-γ promoter, with distinct cis-regulatory regions of transactivation. In PBL, the region of −108 to −39 bp is essential for promoter function, whereas in LPMC at least two regions are involved in controlling CD2-mediated transactivation of the IFN-γ promoter: 1) a major CD2-regulatory element contained within the region −204 to −108 bp upstream of the transcriptional start site; and 2) a minor regulatory element residing within −108 to −40 bp of the transcriptional start site.
Up-regulation of AP-1 nuclear factors is involved in CD2 signaling
Increased binding of transcriptional AP-1 nucleoprotein has been implicated in regulation of IFN-γ promoter activity in T cell lines (10). Furthermore, inhibition of the binding of AP-1 is implicated in the mechanism for glucocorticoid inhibition of IFN-γ expression in T cell lines (8). To determine whether induction of functionally active jun/fos, AP-1 nucleo-factors were involved in the regulation of IFN-γ expression in PBL and LPMC, transfection experiments were conducted using a multimeric AP-1-binding TRE2 reporter construct. CD2 ligation resulted in transactivation of a multimeric AP-1-binding TRE2 reporter construct in both PBL and LPMC (Fig. 6). The increased promoter activity induced in PBL was 100-fold (Fig. 6,A), whereas only a 20-fold increase of AP-1 transactivation occurred in LPMC (Fig. 6 B). Although the background level in unstimulated LPMC was high, PMA/ionophore-activated PBL and LPMC exhibited a similar level of AP-1 transactivation. These results indicate that there is a functional increase in AP-1 binding after CD2 pathway activation of LPMC and PBL, with AP-1 activation appearing essential for transactivation in PBL but probably less so for LPMC.
The AP-1-binding proximal and distal conserved elements are not critical regulators of CD2 response in LPMC
A number of AP-1-binding elements have been defined within the region −108 bp to +64 bp of the transcriptional start site necessary for regulation of IFN-γ gene expression in T cell lines. The conserved proximal (−73 to −48) and distal (−96 to −80) regions are believed to be essential for regulation of IFN-γ gene expression in activated T cell lines. Activation of both the proximal and distal conserved elements is mediated through the binding of AP-1, jun/fos-transacting factors (8). To ascertain the role of these elements in regulation of IFN-γ expression, both LPMC and PBL were transfected with promoter-reporter constructs containing multiple copies of the proximal or distal element upstream of a minimal IFN-γ promoter. IFN-γ expression was then monitored after activation with CD2. CD2 activation of PBL increased transactivation of the proximal and the distal conserved regions by 200- and 100-fold, respectively, showing that both represent CD2 response elements (Fig. 7,A). Likewise, the proximal and distal regions responded to PMA/ionophore activation with 100- and 700-fold increases, respectively (Fig. 7,C). In contrast, transfection of the proximal AP-1-binding elements in LPMC did not restore loss of CD2 responsiveness above the expression of the basal −39-bp construct (Fig. 7,C; note difference in scale range). There was only a modest 2-fold increase over basal expression by concatamers of four repeats of the distal AP-1 binding region transfected into LPMC (Fig. 7,B). Similarly, after PMA/ionophore activation, no additional transactivation was detected with the distal region and only a modest (<2-fold) increase was detected over the minimal promoter following transfection of the proximal AP-1-binding elements (Fig. 7 D).
Because there was only marginal transactivation of the proximal and distal conserved regions in LPMC, we wished to determine whether any nuclear proteins were bound to these elements after CD2 activation of LPMC. Nuclear proteins were extracted from LPMC before or after activation with CD2 and analyzed by EMSA analysis for binding to proximal and distal conserved regions. Fig. 8 shows the kinetics of induction of nuclear proteins binding to the proximal conserved region of the IFN-γ promoter. After CD2 activation, up-regulation of protein complexes binding to the proximal AP-1 region of the IFN-γ promoter is marginal. A similarly marginal up-regulation of proteins binding to the distal conserved element of the IFN-γ promoter following CD2 activation was detected (data not shown). Thus, the proximal and distal elements may be critical for activation of IFN-γ expression in PBL and T cell lines, although it is unlikely that they play a similarly important role after CD2 activation of LPMC.
The IFN-γ BED region does not appear essential for regulation of IFN-γ expression following CD2 activation of LPMC
The region −204 to −108 bp upstream of the transcriptional start site has been defined above by functional analysis as being a major CD2-regulatory element in LPMC. This region almost completely overlaps with the previously defined BED region (−211 to −186 bp) in T cell lines (12). Previous studies have shown that there is a complex interaction between competitive binding of the YY-1 and AP-1 transcriptional factors in T cell lines. It has been hypothesized that transactivation of the IFN-γ gene occurs through the displacement of YY-1 by AP-1 (12). Subsequent studies confirmed the importance of this region (−196 to −183 bp) in activation of primary memory T cells, through the recruitment of AP-1 binding to this site (13). To assess the effect of CD2 activation on functional regulation of the CD2 in our system, LPMC were transfected with a promoter-reporter construct containing an internal deletion spanning the BED region AP-1-binding site. In contrast to what has been reported for PBL, deletion of the entire region between −214 and −178 bp did not result in a loss of CD2 responsiveness, but rather an increase in CD2 responsiveness was detected, probably due to the loss of the repressor YY-1 binding site (Fig. 9,A). This finding suggests that binding of AP-1 to a CD2 enhancer response element occurs outside of the YY-1 site. Additionally, LPMC were activated with CD2 and nuclear proteins were analyzed for binding to the −198- to −180-bp AP-1-binding site by EMSA. Fig. 9 B shows that there was constitutive binding of nuclear proteins to this site in unstimulated LPMC, which remained unchanged after CD2 activation. Thus, functional activation of AP-1 appears to be up-regulated after CD2 activation; however, only modest changes, if any, were observed in binding of trans-factors to the previously characterized AP-1-binding sites. These results suggest that additional distinct sites, other than those previously described in PBL and T cell line systems, are involved in regulation of IFN-γ gene expression in LPMC.
Discussion
In this study, we examined molecular events and mechanisms involved in the regulation of IFN-γ production after CD2 and CD2 + CD28 as well as PMA/ionophore activation of LPMC. Our results identify novel regulatory mechanisms in the mucosa that are distinct from those in PBL and T cell lines. IFN-γ gene expression in LPMC is highly sensitive to CD2 activation. Coligation of the CD28 molecule further enhances IFN-γ secretion by mucosal T cells. Our data show that in CD2-activated PBL and LPMC, similar to what has been reported for TCR-activated PBL, CD28 costimulation enhances IFN-γ mRNA stability without increasing the rate of transcriptional activation. Transfection of a −2.7-kb IFN-γ promoter-reporter construct into both PBL and LPMC reveals significant promoter activity after CD2 activation. After CD2 + CD28 costimulation, there was a significantly greater increase of transactivation in PBL, but not in LPMC. There are multiple CD2 and PMA activational response elements within the IFN-γ promoter with transactivation of IFN-γ involving distinct cis-regulatory regions in PBL as compared with LPMC.
Studies performed in PBMC and tumor T cell lines activated via TCR and CD28 costimulation demonstrated that CD28 ligation of activated T cells results in enhanced secretion of multiple cytokines, including IFN-γ (16, 17). The molecular mechanisms of this effect remain uncertain. Initially, nuclear run-on studies performed with PBL revealed enhanced IFN-γ mRNA stability after CD28 costimulation. Although TCR stimulation results in IFN-γ mRNA expression, CD28 costimulation did not enhance transcriptional activity (16). Subsequent transfection studies in T cell lines suggest that a modest increase in transcriptional activation of the IFN-γ promoter occurs after CD28 costimulation. Analysis of several cytokine promoter elements, including IFN-γ, reveal a conserved region (−161 to −153 bp) in the 5′-flanking sequence that closely resembles a promoter motif defined as the CD28 response element in the IL-2 promoter (16). Expression of the IL-2 gene after CD28 costimulation has been well studied and occurs through the binding of κB-like transcription factors to the CD28 response element (32). Multiple c-Rel and NFκB binding sites have been identified throughout the IFN-γ promoter, although none of these regions, including the conserved region −161 to −153 bp, have been demonstrated to be selectively transactivated after CD28 costimulation (33).
At first glance, it would seem that there is a discrepancy between the nuclear run-on data and the results from transfection analysis. However, several explanations could account for this discordance. Although traditionally the transcriptional run-on assay has been favored as the method for analyzing alterations in the rate of transcriptional initiation, it is relatively insensitive in part, due to measurement of transcription at a single time point outside the context of the cell environment. Additionally, run-on assays do not directly measure initiation, but rather, they are indicative of the rate of transcription through the measurement of elongation and run-on of previously initiated transcript. Likewise, run-on assays are insensitive for measuring changes in the rate of transcriptional elongation, splicing, and mRNA transport and are susceptible to attenuation through DNA sequences within a gene (34). Indeed, regulation of IL-2 expression after cycloheximide treatment results in inconsistencies between the nuclear run-on data and enhanced mRNA levels, due to posttranscriptional up-regulation (35). Moreover, the steady state levels of unspliced IL-2 mRNA precursors have been shown to differ from those of IL-2-luciferase (36) due to a posttranscriptional mechanism. In contrast, transfection studies directly measure the accumulative kinetics of transcriptional initiation within the intact cell. Our nuclear run-on data are in agreement with the literature (17, 18). Although one could argue that transfection experiments measure promoter activity in the absence of chromatin structure, marked differences in promoter response were displayed between PBL and LPL despite the fact that both would be subject to identical absence of chromatin structure. More importantly, the transfection studies emphasize that although CD2 leads to activation of the IFN-γ promoter in both PBL and LPL, different and distinct promoter elements are involved in this augmentation.
In addition to the CD28 pathway, several additional signaling systems affecting IFN-γ mRNA stability have been described. Expression of IFN-γ mRNA in PHA-stimulated blasts or tumor T cells is stabilized after treatment with both IL-2 and IL-12 (37). IL-7, in a dose-dependent manner, up-regulates IFN-γ secretion of CD28-coactivated T cells by increasing the transcriptional rate as well as enhancing IFN-γ mRNA stability (38). Similarly, in a murine system, activation of either cAMP or protein kinase C resulted in enhanced IFN-γ mRNA expression without having any effect on transactivation of the IFN-γ gene (39). Interestingly, studies of patients with atopic dermatitis suggest that disease pathology correlates with reduced IFN-γ secretion (40). A posttranscriptional defect has been proposed as a mediator of the disease process because high levels of IFN-γ mRNA are observed in the absence of IFN-γ protein secretion. Despite these studies, the precise molecular mechanism regulating IFN-γ mRNA stability remains unclear.
The mRNA of IFN-γ possesses multiple AUUUA motifs in the 3′-untranslated region of the mRNA (19). These sequences are believed to function as rapid turnover elements mediating mRNA degradation. Indeed, the addition of AUUUA motifs onto normally stable transcripts has been shown in chimeric constructs to be sufficient to generate rapid turnover (41). Nevertheless, CD28 costimulation enhances IFN-γ and IL-2 mRNA stability in PBL; however, no preferential enhancement in the stability of c-myc or c-fos was detected, notwithstanding the presence of AU-rich motifs in these mRNAs (17). Recent studies of the IL-2 gene have suggested that both 5′ and 3′ sequences within the untranslated region of the IL-2 mRNA are critical for mRNA stabilization (42). Indeed, the 5′-untranslated region appears to be an important region, targeting the activation of c-jun amino-terminal kinase, which leads to phosphorylation and activation of c-jun. Thus, c-jun amino-terminal kinase activation not only results in an increase in binding of AP-1 and up-regulated gene transcription but now also has been shown to directly promote mRNA stability. It is conceivable that similar unidentified sequences are present within the IFN-γ gene that would directly link up-regulation of AP-1 activity with enhanced IFN-γ mRNA stability.
Functional studies indicate an increase in AP-1 activity after CD2 activation of PBL and LPMC; however, the effect varies between cell types. In PBL, a 120-fold increase in AP-1 activity was detected after transfection of an AP-1-binding construct, whereas in LPMC a more modest 20-fold enhancement of AP-1 activity was observed. Transfection of progressive truncations of the IFN-γ promoter and constructs encompassing the AP-1-binding sites suggest that the region between −108 and +64 bp, which is critical for CD2 and PMA/ionophore transactivation in PBL, is of only modest importance in transactivation in LPMC. In LPMC, a significant CD2 response element resides between −204 and −108 bp, a region previously reported in PBL and T cell lines to possess an essential AP-1-binding site. CD2-mediated transactivation of the IFN-γ promoter in LPMC is not altered by deletion of this AP-1 site. Moreover, in direct contrast to what has been previously reported for PBMC, EMSA analysis of nucleoprotein binding to the known AP-1-binding elements described within both the −108- and +64-bp and −204- and −108-bp regions supports the conclusion that CD2 activation does not appreciably alter nuclear proteins binding to these sites in LPMC. However, a change in the composition of these nuclear protein complexes cannot be ruled out. These results illustrate the complexity of molecular events involved in transcriptional regulation of IFN-γ expression in LPMC and highlight unique regulatory mechanisms distinct from those we observed in PBL and previously noted T cell lines.
In T cell lines, the IFN-γ promoter possesses multiple AP-1-binding sites. The two promoter elements designated the proximal (−73 to −48) and distal (−96 to −80) binding sites have been shown to be critical for transactivation of IFN-γ expression in T cell lines. The proximal element has been shown to be a target for selective hypomethylation in cells that express IFN-γ (43) and is believed to generate activation-specific expression in T-cell lines through binding of jun/ATF-2 heterodimers to this element (10). These sites are targets for selective inhibition of IFN-γ gene transactivation after glucocorticoid treatment (8), but not following retinoid treatment (44, 45). Additionally, whereas these sites are critical in regulation in PBL and T cell lines, they are of modest importance in modulating CD2 response in LPMC. Thus, it appears that both the cell type and the mode of activation play a role in the selective transactivation of promoter elements.
Recent studies in our laboratory evaluating expression of IL-2 in LPMC support the notion that regulation of cytokine gene expression in PBL differs from that observed in LPMC (24, 46). Additional studies by other groups have indicated that regulation of IFN-γ gene expression in primary T cells differs from that observed in tumor T cell lines and likewise differs in naive compared with memory T cell subsets. On a structural level, the IFN-γ gene is virtually completely methylated in thymocytes or CD45RAhigh CD45ROlow neonatal or adult T cells that do not express IFN-γ (47). In contrast, the IFN-γ gene is in a hypomethylated state in adult CD45RAlow CD45ROhigh T cells that express IFN-γ. It has been suggested that the interplay between inducible and constitutive nucleoprotein interactions directs IFN-γ gene transcription in vivo. For example, in a transgenic murine primary T cell system, expression of IFN-γ in memory cells was under the control of both the proximal and distal element, yet naive T cells required priming to activate transcription from these elements (28). Furthermore, cAMP inhibited transactivation directed by the proximal element in primed mouse CD8+ T cells; however, transactivation of the distal element was increased in these same cells after induction of cAMP (28). In fact, distinct differences were seen in the transactivation of these two elements when comparing murine CD4+ and CD8+ T cells subsets (28).
The mononuclear cells in this study are comprised of predominantly CD45RAlow CD45RO high memory T cells of both CD4+ and CD8+ T cell subsets (21). Regulation of transcription in these cells appears distinct from that observed in PB memory T cells. The −183- and −196-bp AP-1 binding site, located within the BED element first identified in T cell lines, encompasses a number of overlapping cis elements capable of binding YY-1, AP-1 and SP-1 (12). It was hypothesized that the selective binding of AP-1 to this region, displacing YY-1, was critical for transcriptional activation. Studies of PBL supported this hypothesis and suggested that the composition of DNA-protein interactions binding to the −183- and −196-bp AP-1 site in human memory T cells differs from that seen in naive T-cells (13). Indeed, deletion of this site virtually eliminates expression of IFN-γ in peripheral memory T-cells (13). In direct contrast in LPMC, Fig. 6 shows that deletion of this AP-1-binding site fails to abolish CD2 activation of LPMC. Likewise, although inducible nucleoprotein binding was detected in peripheral memory T cells binding to this site, EMSA analysis of nucleoprotein extracted from LPMC remained unchanged after CD2 activation.
Thus, it appears that there is a specific interplay between a complex of factors binding to numerous cis-regulatory sequences which may be regulated differently in LP T cells from those of PBL and T cell lines. The selective activation of these elements might play an important role in mediating cytokine expression in the intestine. CD2 signaling of LPMC or PBL results in functional activation of AP-1, suggesting that regulation of AP-1 binding may be essential for IFN-γ production. However, the previously identified AP-1-binding sites of the IFN-γ promoter including the proximal, distal, as well as BED region AP-1 site, are not the targets for CD2-directed transactivation in LPMC. In addition, these studies represent the first reports of transcriptional activation of the IFN-γ promoter in response to CD2 stimulation. The data presented in this study provide evidence indicating that regulation of IFN-γ production in LP T cells is complex, involving regulation of multiple cis-regulatory sequences within the IFN-γ promoter region that differ from those elements important for IFN-γ activation of peripheral T cells.
Acknowledgements
We thank Krystine Nguyen and Alice Chen for isolating human LPMC.
Footnotes
This work was supported by U.S. Public Health Service Grants DK-43211 and DK-46763 and Cedars Sinai Medical Center Inflammatory Bowel Disease Research Funds.
Abbreviations used in this paper: ATF, activating transcription factor; LP, lamina propria; LPMC, lamina propria mononuclear cells; PB, peripheral blood; PHA-L, PHA-leukoagglutinin.