| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
B Transcription Factor c-Rel1
* Division of Molecular Bioscience and
Division of Immunology and Genetics, John Curtin School of Medical Research, Australian National University, Canberra, Australia;
School of Molecular and Microbial Biosciences, University of Sydney, Sydney, Australia; and
Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
B family of transcription factors serves a major role in controlling gene expression in response to T cell activation, but the genome-wide roles of individual family members remain to be determined. c-Rel, a member of the NF-B family, appears to play a specific role in T cell function because T cells from c-Rel–/– animals are defective in their response to immune signals. We have used expression profiling to identify sets of genes that are affected by either deletion or overexpression of c-Rel in T cells. Very few of these genes exhibit a strong requirement for c-Rel; rather, c-Rel appears to modulate the expression of a large number of genes in these cells. The sets of c-Rel-affected genes are significantly enriched for genes containing consensus NF-B/Rel sites in their proximal promoter regions. In addition, their promoters contain a higher average density of NF-B/Rel sites compared with all genes represented on the microarrays. A transcriptional module comprised of two closely spaced c-Rel consensus sites is found with higher frequency in the c-Rel-affected gene sets and may represent an important control module for genes regulated by c-Rel or other NF-B family members. We confirmed the importance of these findings on a subgroup of genes by using quantitative PCR to monitor gene expression as well as in vitro c-Rel/DNA binding assays and luciferase reporter assays. The c-Rel-regulated genes identified here support a role for c-Rel in inflammatory responses as well as in the promotion of cell growth and survival. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
B family of transcription factors plays a major role in controlling inducible gene transcription in immune cells. The family consists of five members, RelA (p65), c-Rel, RelB, NF-B1 (p50), and NF-B2 (p52), which function as heterodimers or homodimers to regulate the transcription of a diverse array of genes (reviewed in Ref. 1). NF-B family members are characterized by a conserved N-terminal Rel homology domain that contains sequences for DNA binding, dimerization, IB binding, and nuclear localization. In addition, RelA, c-Rel, and RelB have distinct transactivation domains. NF-B proteins are sequestered in the cytoplasm by inhibitory IB proteins (1). In response to cell activation, signals transmitted via the IB kinase complex result in the phosphorylation of IB proteins, leading to their ubiquitination and degradation. NF-B complexes then translocate to the nucleus, bind to consensus DNA sequences (B sites) in gene promoters, and activate transcription.
The deletion of genes for individual NF-B family members in mice results in distinct physiological outcomes (reviewed in Ref. 2). Although RelA–/– animals die around embryonic day 15.5 due to hepatic TNF sensitivity, all other NF-B-deficient animals are viable but have immunological defects (2). In c-Rel–/– mice the immune system appears to develop normally (3) but the mice are immunodeficient and this phenotype is linked to effects on T cell, B cell, macrophage, and dendritic cell function (3, 4, 5). Analysis of c-Rel–/– mice in several disease models has revealed unique roles for c-Rel in the development of autoimmunity and allergy, susceptibility to pathogenic infection, and the ability to mount antiviral Ab responses (reviewed in Ref. 6). Perhaps one of the most intriguing effects that has been observed in c-Rel–/– animals to date, is that allogeneic transplants of either heart or pancreatic islet tissue survive for significantly longer periods than similar transplants in wild-type (wt)3 animals (7, 8). These and other effects have been attributed to defective Th1 responses.
The predominant effects of c-Rel loss are evident in lymphoid cell effector function where splenic T cells from c-Rel–/– animals show reduced ability to proliferate (3), a result of significantly less IL-2 production in response to TCR and costimulatory receptor activation (3, 9). The production of the inflammatory mediators IL-3 and GM-CSF is also impaired in c-Rel-deficient T cells, whereas the expression of other cytokines (IL-5, IFN-
, and TNF-
) is indirectly affected by the absence of IL-2 (10). Extensive in vitro and in vivo studies have shown that genes such as IL-12p40, IRF-4, c-myc, Bfl-1/A1, and Bcl2l1 (Bcl-xL) are dependent on c-Rel for correct expression (4, 11, 12, 13, 14). The promoters of many of these genes have potential c-Rel binding sites, implying a direct effect on gene transcription. We have recently shown that chromatin remodeling across the promoter regions of IL-2 and GM-CSF does not occur following the activation of CD4+ T cells from c-Rel–/– mice (9, 15). Moreover, the histone loss that accompanies this chromatin remodeling is not observed in c-Rel–/– T cells (16). In contrast, although c-Rel is required for transcription from the IL-12p40 promoter, it is not required for chromatin remodeling on this gene promoter in myeloid cells (17), suggesting that the function of c-Rel is highly tissue or gene specific.
Thus, c-Rel has several physiological and biochemical properties that are distinct from other NF-B family members and suggest a distinct role for this protein in gene transcription in T cells. It is unlikely, however, that the c-Rel target genes characterized to date fully explain the role of c-Rel in T cell activation and T cell-dependent immune responses.
In this study, we have investigated the specific role of c-Rel in the regulation of genome-wide transcription in activated T lymphocytes. Using microarray expression profiling and quantitative PCR (QPCR) analysis of RNA from CD4+ T cells of c-Rel–/– mice and a T cell line overexpressing c-Rel, we have identified groups of genes whose expression is modulated by c-Rel. Although very few of these genes demonstrate a strong requirement for c-Rel in T cells, the sets of genes affected by c-Rel deletion or overexpression were found to be significantly enriched for genes containing NF-B/Rel binding sites and having a high density of these sites in their proximal promoter regions. The presence of a specific module comprised of two c-Rel binding sites is also enriched in these gene sets and may be a common feature of genes regulated by c-Rel and other NF-B family members. The genes identified as likely targets of c-Rel in this study support a role for c-Rel in inflammation and provide evidence of a prosurvival and growth promoting function for c-Rel in T cells.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
EL-4.IL-2 (EL-4) murine thymoma cells and Jurkat human leukemia T cells were maintained in RPMI 1640 medium supplemented with 10% FCS, 10 mM HEPES, 2 mM L-glutamine, and antibiotics. EL-4 cells were stimulated at 1 x 106 cells/ml with 10 ng/ml PMA (Sigma-Aldrich) and 1 µM calcium ionomycin A23187 (Sigma-Aldrich) and Jurkat T cells at 5 x 105 cells/ml with 20 ng/ml PMA and 1 µM ionomycin.
CD4+ T cell purification and stimulation
All mice were maintained in a specific pathogen-free facility. Spleens were isolated from wt C57BL/6 mice and c-Rel–/– mice backcrossed 10 generations onto C57BL/6 and CD4+ T cells purified as described previously (9). CD4+ T cells (1 x 106 cells/ml) were stimulated for the indicated times with an activating CD28 Ab and plate-bound CD3
Ab (BD Pharmingen) as described previously (9). For QPCR experiments cells were stimulated for 30 min to 24 h.
Plasmids
The human GM-CSF luciferase reporter construct (pGMLuc) contains a –620 to +37 bp promoter fragment (18) in the pXPG luciferase reporter (19). Mouse TNF-, ICAM-1, IL-2, and CREM luciferase reporters contain promoter regions of approximately –380 to +40 bp cloned into the pXPG luciferase reporter using KpnI/HindIII PCR products amplified from EL-4 genomic DNA. The human c-Rel cDNA in a pRc/CMV expression vector (Invitrogen Life Technologies) was subcloned into a pCDNA3 expression vector containing an internal ribosome entry site (IRES)/GFP insertion using a PCR product amplified with BamHI sites. Sequence analysis was used to confirm the integrity of all constructs.
RNA preparation, QPCR, and expression profiling
Total RNA was extracted from CD4+ T cells from wt C57BL/6 mice and c-Rel–/– mice and from transfected GFP+ EL-4 cells and reverse transcribed as described previously (9). SYBR Green real-time PCR was performed using an ABI PRISM 7700 sequence detection system (PerkinElmer/PE Biosystems) as described previously (9). To normalize for inefficiencies in cDNA synthesis and RNA input, PCRs for ubiquitin-conjugating enzyme (UBC) E2D 2 were conducted in parallel. The primer pairs used are listed in supplemental Table IA.4 Expression profiling analysis of RNA from wt and c-Rel–/– CD4+ T cells was performed on Affymetrix murine U74A version 2 set arrays and data were analyzed using the Affymetrix microarray suite version 4.0 software. Each array was normalized to user default settings with a target intensity of 150 across the entire chip. The scale factor between all chips was no greater than a factor of three with data discarded if the scale factor was larger. To find transcripts that were different between two samples, we selected for genes that were classified as increasing, with a difference of 2-fold or greater and with a significant p value score (<0.2). Transcripts were only included in the final lists if they fulfilled these selection criteria in two independent microarray experiments. Functional classification was conducted using a relational database program based on FileMaker Pro 5 developed for microarray analysis (G. S. Denyer, unpublished data). RNA from transfected vector- or c-Rel-overexpressing GFP+ EL-4 cells was analyzed on Affymetrix mouse expression set 430, version 2.0 arrays at the Biomolecular Resource Facility (John Curtin School of Medical Research, Australian National University, Canberra, Australia) according to standard protocols. Data were analyzed using the Affymetrix microarray suite version 5.1 software.
Immunoblot
Nuclear extracts were prepared from transfected GFP+ EL-4 cells as described previously (20). Cell extracts were resolved on denaturing SDS polyacrylamide gels, transferred to nitrocellulose membranes, and immunoblotted with anti-c-Rel (catalog no. sc-71, Santa Cruz Biotechnology) and anti-Sp1 (catalog no. sc-109, Santa Cruz Biotechnology) Abs. Proteins were detected using ECL Plus chemiluminescent substrate (GE Healthcare) and visualized using the Fuji Film luminescent image analyser (LAS-1000 Plus).
DNA binding assay
Recombinant c-Rel homodimer binding to predicted c-Rel binding sites was measured using a modified enzyme-linked immunoassay (21). Briefly, biotinylated, double-stranded, sense and antisense oligonucleotides (25-mer) corresponding to consensus, mutant, and gene-specific NF-B/Rel binding sequences were generated. NeutrAvidin-coated 96-well strip plates (Pierce) were incubated with 100 nM oligonucleotides followed by blocking with 3% skim milk in binding buffer (10 mM Tris (pH 7.5), 10 mM MgCl2, 5 mM EDTA, 10 mM DTT, 0.2% Nonidet P-40, 1% glycerol, 0.4% sucrose, and 0.5 mg/ml BSA). Purified recombinant human c-Rel protein produced in Escherichia coli was serially diluted (0.4–56 nM) in binding buffer containing 3% skim milk and 5 µg/ml poly(dI-dC) (GE Healthcare) and incubated for 1 h. c-Rel binding was detected with an anti-c-Rel Ab (catalog no. sc-71, Santa Cruz Biotechnology) and a secondary HRP-conjugated Ab. For development, a tetramethylbenzidine substrate (BD Pharmingen) was added, the reaction stopped with 1 M H2SO4 and color development was detected on a microplate reader (Molecular Devices) at 450 nm.
Transient transfection of T cell lines and luciferase assays
EL-4 and Jurkat T cells (5 x 106 cells) were electroporated with a Bio-Rad Gene Pulser at 270 V and a capacitance of 975 microfarads. Fifteen micrograms of pCDNA3-IRES/GFP or pCDNA3-c-Rel-IRES/GFP plasmid was used in EL-4 transfections. Cells were recovered for 24 h before sorting for GFP expression (FACSVantage DiVa, Microscopy and Cytometry Resource Facility, John Curtin School of Medical Research, Australian National University, Canberra, Australia). For luciferase assays, Jurkat T cells were transfected in duplicate with 5 µg of empty pXPG plasmid or each the reporter plasmids described above either alone or in combination with 10, 20, or 25 µg of c-Rel expression plasmid. The total amount of plasmid added was equalized by the addition of the parent plasmid. Transfected cells were left unstimulated or stimulated for 8 h with PMA and ionomycin (P/I) and cell lysates were harvested and analyzed as described previously (22). Thirty micrograms of total protein was analyzed for luciferase activity using a Turner BioSystems microplate luminometer.
Computational promoter analysis and statistical analysis
Computational analyses of the promoter regions of c-Rel-responsive genes for NF-B/Rel binding site frequency and density were performed using the TELiS (Transcription Element Listening System) database (23). Exact binomial tests were used to calculate the statistical significance of $VNFKAPPAB_01 and $CREL_01 motif frequency in groups of c-Rel-responsive genes compared with the total population of genes on the arrays. Z tests were used to determine the statistical significance of the average number of these motifs in c-Rel-responsive genes compared with all of the genes on the arrays. Computational analyses of NF-B family motifs in selected c-Rel-responsive genes was performed using Gene2Promoter (release 4.2) and GEMS Launcher (release 4.3) within the GenomatixSuite (24). The V$NFKB transcription factor binding site (TFBS) family includes six matrices for NF-B/Rel family members. MatInspector (release 7.4.3) was used to identify a NF-B family TFBS that had a core similarity score of 1.0 and a matrix similarity score of 
0.8. FastM was used to design TFBS modules that contained two (15-mer) $VNFKAPPAB_01 motifs (matrix similarity score 0.89) or two $VCREL_01 motifs (matrix similarity score 0.91) separated by a minimum distance of 20 bp and maximum distances of 50–200 bp. ModelInspecter (release 5.4) was used to search a mouse Genomatix promoter database for FastM-defined NF-B and c-Rel TFBS modules. Exact binomial tests (two-sample tests for equality of proportions with continuity correction) were used to determine whether NF-B or c-Rel modules were enriched in the c-Rel-responsive gene sets compared with all of the genes on the arrays. Linear regression analysis was used to determine whether there was a significant relationship between the number of NF-B/Rel sites in the promoter regions and the percentage of genes responding to c-Rel deletion or overexpression.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
To gain a genome-wide view of the role of c-Rel in the gene expression program of T cells, we performed expression profiling on CD4+ T cells isolated from the spleens of wt and c-Rel–/– mice either in the resting state or stimulated for 2 h with Abs to CD3 and CD28. Of the genes whose expression was increased in response to CD3/CD28 activation in wt CD4+ T cells (S. Rao and M. F. Shannon, unpublished data), the expression of 
130 was attenuated to some degree in the absence of c-Rel (supplemental Table IB). The reductions were generally small, in the range of 50–60% (data not shown), and were consistent across three independent experiments. We did not find any genes whose expression was completely dependent on c-Rel, and this is likely due to redundancy with other NF-B/Rel family members, such as RelA, which are also translocated to the nucleus upon T cell activation. Gene Ontology classification indicated effects on cytokine, chemokine, and cell surface molecule expression as well as intracellular signal transduction, transcription, and nuclear functions (supplemental Table IB).
To monitor the effect of gain-of-function as well of loss-of-function of c-Rel and to determine which genes have the potential to be regulated by c-Rel, we next overexpressed c-Rel in EL-4 T cells and examined the effect on the gene expression program. Human c-Rel in a pCDNA3-IRES/GFP construct was transiently transfected into the EL-4 T cell line and the transfected cells were sorted on the basis of GFP expression. In cells transfected with an empty IRES/GFP plasmid, endogenous mouse c-Rel protein was not detected in the nucleus of nonstimulated (NS) cells but increased in amount with increasing time of P/I activation as expected (Fig. 1, upper panel). Exogenous human c-Rel expressed from the transfected c-Rel-IRES/GFP plasmid was detected in the nucleus of NS cells as well as in cells activated by P/I for 1–4 h (Fig. 1, lower panel). It should be noted that the mouse and human c-Rel proteins migrate at different sizes on SDS-PAGE and that the ratio of exogenous human to endogenous mouse c-Rel changes across the time course of activation mainly due to the large increase in endogenous c-Rel levels that occurs following P/I activation (Fig. 1, lower panel).
|
2-fold by c-Rel overexpression in NS cells, and at least 60 of these genes showed a 5-fold response (Table I and supplemental Table IC). This indicates that the expression of c-Rel alone can activate a program of gene expression in the absence of other exogenous signals in EL-4 T cells. Gene Ontology classification of the 5-fold gene set revealed effects on many genes encoding cytokines, chemokines, cell surface receptors, signal transduction molecules, and transcription factors (supplemental Table IC). The expression of a selection of genes of unknown function was also increased by c-Rel overexpression (supplemental Table IC), and these may be of interest in regard to the further understanding of c-Rel function in the immune system. In stimulated cells (at 0.5 and/or 2 h), c-Rel increased the expression of 700 genes by 2-fold and >80 by 5-fold (Table I). Indeed, 30–40 genes were increased in expression by 10-fold in either NS or P/I-stimulated cells, indicating that the level of c-Rel in the cell is limiting for the expression of certain genes in T cells. It should be noted that these c-Rel-affected genes were not preselected on the basis of their response to P/I and, thus, the list of genes includes inducible and constitutive genes as well as genes not normally expressed in the absence of exogenous c-Rel in these cells. Of the genes increased 5-fold by c-Rel, there were 35 genes in common between the NS and P/I-stimulated gene sets, with 23 genes responsive to c-Rel at all time points examined, 27 responsive in NS cells only, and 52 responsive in stimulated cells only (Table I and supplemental Table IC). There was only a relatively small overlap between the c-Rel-attenuated genes from the c-Rel knockout (KO) studies (KO gene set) and the c-Rel overexpression gene sets (<10% of the genes).
|
Genes controlled by c-Rel are enriched for NF-B/Rel sites in their proximal promoters
The c-Rel transcription factor binds to NF-B/Rel sites located in gene control regions to activate gene transcription, either as a homodimer or a heterodimer with other NF-B family members (reviewed in Refs. 1 and 25). It is likely that at least some of the genes identified above are direct targets of c-Rel and would be expected to have NF-B/Rel binding sites in their control regions. Consensus binding sites have been defined for the NF-B/Rel family and for specific family members from a combination of PCR selection experiments and computational approaches (26, 27, 28). It is unclear at this stage how selectivity for individual family members is generated or indeed whether there is in fact a high degree of redundancy in the ability of different family members to bind specific sites (reviewed in Ref. 29).
To determine whether the c-Rel-affected gene sets identified above were enriched for NF-B/Rel sites in their control regions, we used TELiS (23) to perform searches for NF-B/Rel motifs defined by the TRANSFAC version 3.2 database. The searches were performed on 300- or 600-bp regions upstream from the transcription start sites (TSS) of the genes or on a 1200-bp region, including 200 bp downstream and 1 kb upstream from the TSS. Two consensus motifs were used, V$NFKAPPAB_01 and V$CREL_01, representing a generic NF-B/Rel consensus site and a c-Rel consensus site, respectively (27, 30, 31), with either low stringency (0.8) or high stringency (0.9) matrix similarity score cutoffs. Statistically significant enrichment was calculated relative to the occurrence of these motifs in all genes represented on the arrays.
Using the low stringency matrix similarity score cutoff (0.8) and any length promoter region, there was no significant enrichment of the V$NFKAPPAB_01 or V$CREL_01 motifs in the gene sets affected by c-Rel overexpression in EL-4 cells (Table II and data not shown). However, with a high stringency (0.9) search for the V$NFKAPPAB_01 motif, the gene sets whose expression was increased either 2-fold or 5-fold in response to c-Rel were significantly enriched for genes containing this motif in the 300-bp proximal promoter regions (2- to 3-fold (p < 0.003) and 5-to 6-fold (p < 0.00007), respectively; Table II). This enrichment was also observed in the 600-bp regions (2- to 3-fold, p < 0.02) but became less significant either by increasing the size of the region searched or by lowering the c-Rel overexpression response threshold to 1.5-fold (Table II). When the V$CREL_01 motif was similarly analyzed, significant enrichment was only observed with a high stringency (0.9) search of the 300-bp promoter region in the 5-fold NS gene set and in the 2-fold P/I gene set (2-fold enrichment, p < 0.01; data not shown). Both high and low stringency searches of either the 300- or 600-bp regions for the KO gene set showed statistically significant enrichment of the V$NFKAPPAB_01 motif (2- to 3-fold, p < 0.002; Table II), whereas the V$CREL_01 motif was enriched in this gene set only for the high stringency search of the 300-bp promoter regions (2-fold, p = 0.0023; data not shown). However, it should be noted that the KO gene set was preselected on the basis of response to CD3/CD28 stimulation and, thus, the enrichment may be influenced by this parameter because the CD3/CD28- and P/I-responsive gene sets also show enrichment for NF-B/Rel sites (data not shown).
|
B/Rel sites with a higher matrix similarity score (0.9) were more likely to be affected by c-Rel overexpression or deletion. To determine whether sites with different matrix similarity scores had different abilities to bind recombinant c-Rel in vitro, modified ELISA-based DNA binding assays were performed using oligonucleotides spanning a selection of the predicted NF-B/Rel binding sites. Six sites with matrix similarity scores 0.9 (high stringency) (Fig. 2A) and six sites with matrix similarity scores between 0.9 and 0.8 (low stringency) (Fig. 2B) were tested for their ability to bind recombinant c-Rel. A c-Rel consensus and a mutant site, as well as a predicted site with a score of 0.79 (from the Nrf2 gene), were included as controls in these studies (Fig. 2B). All of the sequences with high stringency sites bound c-Rel homodimers in a concentration-dependent manner, similar to the consensus c-Rel site (Fig. 2A). The sites with the lower matrix similarity scores showed a wider range of abilities to bind c-Rel, with binding for two sites, c-myc (0.83) and Nfe2 (0.80), showing 50 and 40% of consensus site binding, respectively (Fig. 2B). There was no detectable binding to the Nrf2 site (0.79) or the c-Rel mutant site (Fig. 2). The sites with matrix similarity scores between 0.8 and 0.9 differ from the core consensus in that the 5' G pair is not conserved, with each of these sequences having a G in only one of these positions (Fig. 2B). The lack of a G in the second position of the G pair may be particularly critical, because the three sites with the least ability to bind c-Rel had a thymine or cytosine in this position (Fig. 2B). The G residue in this position has been shown to be an important contact residue in the crystal structure of c-Rel homodimers binding to the IL-2 CD28-responsive element (CD28RE) (30).
|
B/Rel sites in proximal promoter regions with higher matrix similarity scores in the gene sets affected by c-Rel deletion or overexpression. In general, sites with higher matrix similarity scores bind c-Rel homodimers more efficiently in in vitro binding studies.
The density of NF-B/Rel sites in proximal promoter regions is enriched in c-Rel-affected gene sets
We next examined whether the density of NF-B/Rel sites in the promoter regions of these c-Rel-responsive gene sets was higher than the average density of sites in the promoter regions of all genes on the arrays. There was a >2-fold statistically significant enrichment for the average number of V$NFKAPPAB_01 motifs using either low or high stringency searches of the 5-fold NS or P/I gene sets for all promoter lengths compared with the average density of sites in all promoters on the array (Table III). The highest significant enrichment was observed with the high stringency search of the 300-bp promoter fragments (6-fold, p = 1–10; Table III). The average density of V$NFKAPPAB_01 motifs was also significantly higher in the 2-fold gene sets, again most apparent in the 300-bp high stringency search (Table III). In the c-Rel KO gene set a similar pattern was observed, with the most significant enrichment for a high density of NF-B/Rel sites once again observed for the high stringency searches of either the 300- or 600-bp regions (3- to 4-fold, p = 1–10; Table III). When the density of V$CREL_01 motifs was examined, the high stringency cutoff for the 300-bp proximal promoter regions of the 5-fold gene set yielded the most significant enrichment (3-fold, p = 0.0009), and a higher density of c-Rel sites was also observed in the KO gene set (2-fold, p < 0.00003) (data not shown).
|
B/Rel binding sites in the gene promoter regions and the likelihood of response to c-Rel deletion or overexpression. All genes on the arrays were classified according to the number of V$NFKAPPAB_01 or V$CREL_01 motifs in their promoter regions using high or low stringency matrix similarity scores. The percentage of genes in each of the gene sets that responded to c-Rel overexpression or c-Rel deletion was then calculated and plotted against the site number. Analysis of the gene sets that responded 5- or 2-fold to c-Rel overexpression showed a positive relationship between the percentage of genes that responded to c-Rel overexpression and the number of high stringency V$NFKAPPAB_01 or V$CREL_01 motifs (Fig. 3A; data not shown). This relationship was statistically significant in the low stringency searches for both the V$NFKAPPAB_01 motif in the 300-bp promoter regions in the 5-fold gene sets (R2 > 0.8, p < 0.02) and for the V$CREL_01 motif in the 300-bp promoter regions in both the 5-fold and 2-fold gene sets (R2 > 0.8, p < 0.05). There were too few points in the high stringency searches for meaningful statistical analyses, but similar trends were observed (Fig. 3A). The significance of the correlation dropped with longer promoter regions (Fig. 3A) or by decreasing the cut-off of response to c-Rel overexpression to 1.5-fold (data not shown).
|
Because the data above suggested that the density of NF-B/Rel sites is enriched in the gene sets that respond to c-Rel overexpression, we tested a small group of promoters that contained different numbers and combinations of NF-B/Rel sites for their ability to respond to c-Rel overexpression in luciferase reporter assays. Three hundred to six hundred base pairs of the proximal promoter regions from the GM-CSF, ICAM-1, CREM, TNF-, and IL-2 genes were cloned into luciferase reporter constructs and transfected into Jurkat T cells together with a human c-Rel expression construct. The human GM-CSF promoter fragment used here contains seven predicted (nonoverlapping) NF-B/Rel sites (four c-Rel-specific sites), whereas ICAM-1 contains four (three c-Rel sites), TNF- has four (no c-Rel sites), and IL-2 and CREM have three (no c-Rel sites) and two (two c-Rel sites) NF-B/Rel sites, respectively (Fig. 4A). All of the promoter reporter constructs responded to P/I activation in transfected Jurkat T cells (data not shown). Only GM-CSF and ICAM-1 were transactivated in a dose-dependent manner by c-Rel in either NS (Fig. 4B) or P/I-stimulated cells (Fig. 4C). The distinguishing feature of these two responsive promoters was the presence of the greatest overall number of NF-B/Rel sites (Fig. 4A).
|
B/Rel sites in proximal promoter regions may be important in determining the response of genes to c-Rel deletion or overexpression. The occurrence of a dimeric c-Rel TFBS module in the promoters of genes controlled by c-Rel
Previous experimental evidence from a study of two chemokine genes has indicated that the occurrence of two NF-B/Rel sites located relatively close to one another within their control regions was essential for the responses of these genes to NF-B (32). We therefore asked whether there was any enrichment of a module containing two NF-B/Rel sites with a specific spacing arrangement in the c-Rel-affected gene sets. Using ModelInspecter, we searched a Genomatix promoter database of functional mouse promoter sequences (–500 to +100 bp relative to the first and last TSS, respectively) for genes that contained an NF-B or c-Rel TFBS module comprised of two V$NFKAPPAB_01 or two V$CREL_01 motifs, respectively, situated at a minimum distance of 20 bp from one another and a maximum distance of 200 bp. The module was designed such that the motifs could exist on either strand of DNA but was restricted to having a high stringency matrix similarity score.
From a total number of 89,204 promoter sequences searched, we identified 1,064 gene promoter sequences with unique names that contained the NF-B TFBS module and 1,686 that contained the c-Rel TFBS module within their 600-bp proximal promoter regions. Although the NF-B module showed no significant enrichment, the c-Rel module appeared to be enriched in the c-Rel overexpression-affected gene sets (Table IV). Although only 5.5% of the genes on the array contained the c-Rel module, 13% of the genes in the 5-fold NS or P/I gene sets contained the module, an enrichment of 2-fold (p < 0.05; Table IV). When the percentage of genes on the array that were affected by c-Rel overexpression was compared with the percentage of genes with the c-Rel module that were affected by c-Rel overexpression, there was also a significant enrichment observed (2- to 3-fold, p < 0.03; Table IV). The c-Rel and NF-B modules were both enriched 2-fold in the KO gene set with the c-Rel module showing greater statistical significance (Table IV), but the caveat of preselection of these genes for response to CD3/CD28 activation as discussed earlier also applies here.
|
Confirmation of the response of a subgroup of genes to c-Rel deletion or overexpression
We selected a subgroup of 33 genes to confirm the data obtained from expression profiling studies and to cross-check the c-Rel overexpression and deletion microarray data. The subgroup was selected as a combination of genes from both deletion and overexpression arrays and, most importantly, the genes were selected with a range of NF-B/Rel site densities ranging from 0 to 6 sites in their 300-bp proximal promoter regions (see Fig. 7 for summary). Promoter sequences (600 bp) were extracted from the murine Genomatix promoter database using the Gene2Promoter software and analyzed for the presence of all NF-B family sites and, specifically, for c-Rel binding sites. The MatInspector program was used to identify NF-B/Rel family TFBS with a core similarity score of 1.0 and a matrix similarity score of 0.8. Overlapping NF-B/Rel TFBS on opposite strands of the DNA, due to the palindromic nature of many NF-B/Rel sites, were counted as a single site but are shown as overlapping sites in the schematic in Fig. 7. The subset of NF-B/Rel sites that were classified as c-Rel sites are also shown in Fig. 7 and the genes that contained the dimeric c-Rel or NF-B TFBS modules are indicated.
|
|
|
B/Rel sites in their 300-bp proximal promoter regions. In addition, four of these genes contain a dimeric c-Rel TFBS module. A number of other genes were consistently affected in either the deletion or the overexpression experiments, thus suggesting that these genes are also likely c-Rel target genes. These genes include PD1, TER1, and MAP3K8 (c-Rel deletion) and Tnfaip3, Gadd45
, MIG, MyD116, and IRF-4 (c-Rel overexpression). |
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
Microarray analysis led to the identification of >100 genes where expression was attenuated by c-Rel deletion, but the confirmation rate by QPCR (30%) suggested that many of the effects were relatively minor. For the genes where an effect was confirmed by QPCR it is also clear that although c-Rel may play a role in the expression of these genes, it is not essential. In agreement with these results, a similar analysis of the gene expression program in c-Rel–/– macrophages has also shown a relatively minor impact, with only 65 genes showing reduced expression and only four genes being unequivocally shown to be significantly affected by c-Rel deletion (33). The most likely explanation for the relatively subtle effects of c-Rel deletion is that other NF-B/Rel factors also play a role in the expression of these genes and that by reducing the overall level of NF-B proteins in the cell (through the absence of one family member) the expression of the genes is reduced but not eliminated. Significant functional compensation between members of the NF-B/Rel family has previously been shown for a small group of genes in response to TNF- in fibroblasts lacking various NF-B family members (34). In addition, there is significant overlap both in the functional groupings and in the specific genes affected by c-Rel deletion and those affected in other expression profiling studies where NF-B activation was globally inhibited (35, 36, 37). Small scale studies of genes affected by RelA deletion in RelA–/– mouse embryonic fibroblasts (34) or RelA RNA interference treatment in HeLa cells (38) also identified genes such as IP-10, Bcl2l1, and IB as RelA targets, suggesting a common gene expression program executed by the NF-B/Rel family of transcription factors.
The relatively subtle effect of c-Rel deletion on the gene expression program of T cells is also consistent with the physiological effects of c-Rel deletion. In many models of immunity, c-Rel–/– mice have been shown to have a defective but not totally ineffective immune response (reviewed in Refs. 2 and 6). For example, c-Rel–/– mice are less efficient but not unable to clear an influenza virus or Toxoplasma gondii infection (39, 40). Similarly, c-Rel–/– mice have a slower but not completely abrogated response in allograft rejection (7, 8).
c-Rel overexpression in EL-4 T cells led to a more dramatic effect on gene expression, with 30–40 genes showing at least a 10-fold increase in expression. These experiments have identified genes where c-Rel (or total NF-B/Rel) protein levels are limiting for expression as well as genes with the potential to be c-Rel targets in T cells or in other cell types. Indeed, some of the genes that responded to c-Rel overexpression (e.g., MIG, CD40, and Iigp1) are generally not expressed in T cells at high levels but are important products of other cells of the immune system. Redundancy within the NF-B transcription factor family as discussed above may be one reason for the relatively minor effects of deleting a single family member but the more pronounced effects in overexpression experiments. The ability of gene control regions to assemble distinct sets of transcription factors to activate gene expression in different circumstances (41) may also account for the relatively weak effect of deletion but larger effects of overexpression. Indeed, deletion and overexpression studies of a group of 55 transcription factors in yeast have shown that in most cases deletion showed markedly fewer changes in the expression profile compared with overexpression and that in many cases there was little concordance in the gene sets affected in each scenario (42). We also found in this study that the overlap between the gene sets responding to c-Rel overexpression and the KO gene set was small (<10%). Further experimentation in the yeast system argued that overexpression identified bona fide targets of the transcription factor (42). Thus, the genes identified here from c-Rel overexpression studies are likely to be bona fide targets either in T cells or in other cells of the immune system where c-Rel is expressed. Together, these genome-wide effects of c-Rel loss-of-function and gain-of-function suggest that, at least in T cells, c-Rel may play a role in amplifying or sustaining the transcriptional responses of genes that are redundantly targeted by all NF-B proteins.
QPCR analysis of a subset of genes identified a group of genes with a high probability of being c-Rel targets. These genes include GM-CSF, ICAM-1, CD40, IP-10, Bcl2l1, and, with lesser certainty, IL-2, IL-13, Iigp1, and IL-4. Several of these genes, GM-CSF, IL-2, ICAM-1, IP-10, Bcl2l1, and CD40, have previously been described as NF-B target genes either in single genes studies or expression profiling studies (9, 12, 20, 43, 44, 45). We have previously shown that GM-CSF and IL-2 are c-Rel target genes where c-Rel binds to and activates transcription through the CD28RE of the promoter (9, 15, 18). c-Rel-containing NF-B complexes have also been shown to bind to the ICAM-1, Bcl2l1, and IL-4 promoters in specific cell types (12, 43, 46). However, the involvement of c-Rel in the regulation of the IP-10, IL-13, Iigp1, and CD40 genes has not previously been reported. In support of some of these genes as direct targets of c-Rel, a recently published chromatin immunoprecipitation (ChIP)-on-chip study of NF-B/Rel proteins binding to LPS-induced genes in macrophages identified ICAM-1 and Bcl2l1 as direct binding targets of c-Rel (47). Similarly, Tnfaip3 and Gadd45, shown in this study to be induced by c-Rel overexpression, were also bound by c-Rel in LPS-stimulated macrophages (47). A number of other genes in the deletion or overexpression gene sets, which were not among the genes tested by QPCR, were also bound by c-Rel in the above study (47), but c-Rel chromatin immunoprecipitation experiments in T cells will be necessary to confirm these data. The genes identified in this study as likely targets of c-Rel encode factors that have important roles in immune cell signaling and leukocyte migration (e.g., ICAM-1, MIG, and IP-10), as well as molecules important for the control of cell proliferation and survival (e.g., Bcl2l1, Tnfaip3, CD40, and Gadd45) that are consistent with the known function of c-Rel (reviewed in2, 6). In addition, this group of target genes may help to further elucidate the specific role of c-Rel in T cell-dependent immune responses, in particular in Th1-mediated inflammation and transplant survival (7, 48, 49, 50) and as a prosurvival factor in human lymphomas and animal models of tumorigenesis (51, 52, 53).
The c-Rel-affected gene sets determined from deletion or overexpression experiments showed significant enrichment for genes with NF-B/Rel sites in their promoters, implying that many of the genes are likely to be direct c-Rel targets. These effects were particularly striking in the 300-bp proximal promoter regions of the genes, using a search for high stringency consensus sites in the 5-fold or 2-fold response gene sets. Results were similar irrespective of whether an NF-B or a c-Rel consensus motif was analyzed, again suggesting redundancy within the family for NF-B/Rel proteins and binding sites. In support of the stronger correlation observed when a high stringency matrix similarity score was used for the searches, in vitro binding studies indicated that there was a tendency for the sites with lower matrix similarity scores to bind c-Rel less efficiently than sites with higher matrix similarity scores.
In addition to the simple presence of NF-B/Rel sites, the density of NF-B/Rel sites was also significantly enriched in these promoters. In agreement with the statistically significant relationship between c-Rel effects and NF-B/Rel site density, the two gene promoters that responded to c-Rel overexpression in luciferase reporter assays had a higher density of NF-B/Rel sites than the three promoters that did not respond. It is notable that the IL-2 promoter did not respond to c-Rel overexpression even though there is significant evidence in the literature that IL-2 is a c-Rel target gene (9, 54), and we have shown in this study that the endogenous IL-2 gene responded to c-Rel overexpression. In fact, the response of the IL-2 promoter to c-Rel overexpression has not previously been shown, although antisense c-Rel lowers promoter activity and c-Rel can bind at least in vitro to the IL-2 CD28RE (18, 54). It is possible that c-Rel levels are not limiting for IL-2 promoter function, at least in Jurkat cells, and that the effect seen on the endogenous gene in EL-4 cells represents a cell-specific difference in endogenous c-Rel levels or is mediated through other control regions or mRNA stability control.
From the QPCR analysis of endogenous gene expression in either c-Rel–/– CD4+ T cells or EL-4 cells overexpressing c-Rel, it was also evident that genes with a higher density of sites were more likely to be confirmed as bona fide c-Rel targets. Previously, the number of binding motifs for a particular transcription factor has been successfully used to predict the response to a transcription factor and to predict gene expression patterns during the yeast cell cycle (55). A computational approach to identify NF-B-regulated genes has also shown that, among a group of immune genes, the region immediately upstream of the TSS (230 bp) contains the highest density of NF-B sites compared with regions further upstream or with the promoters of a random group of genes (56). Thus, a high density of higher affinity NF-B/Rel sites in proximal promoter regions may be an important criterion for the NF-B/Rel dependence of a gene.
However, although there may be a general correlation between a high density of NF-B/Rel sites and the ability to respond to c-Rel, the degree of response of a gene may not always be predicted by site density. For example, Iigp1 was one of the most highly responsive genes to c-Rel overexpression in EL-4 cells (60- to 300-fold) but contained only two NF-B/Rel sites within 600 bp of the TSS. It is possible that some of these very highly responsive genes are activated by a feed-forward loop involving c-Rel and other transcription factors such as IRF and STAT, which were identified here as putative c-Rel target genes and could cooperate with c-Rel to activate genes such as Iigp1 and MIG. Indeed, searches of immune gene sets for the co-occurrence of TFBS has identified NF-B/Rel, IRF, and STAT factors as the most likely set of transcription factors that coregulate immune genes (57, 58).
In addition to simple TFBS density, more complex arrangements also seem to exist in higher eukaryotes, and studies of Dorsal function in Drosophila have shown that both the quality and arrangement of sites is critical in the readout of the Dorsal gradient during embryo development (59). A c-Rel TFBS module, but strikingly not an NF-B module, containing two consensus c-Rel sites within 200 bp of one another in the proximal promoter region was significantly enriched in the c-Rel-affected gene sets. This module also occurs in eight of the genes that we confirmed as c-Rel targets either in deletion, overexpression studies, or in both, as well as in the two gene promoters that responded to c-Rel overexpression in luciferase assays. Leung and colleagues have shown that activation of the IP-10 gene in a lentiviral reporter-based assay required the presence of two NF-B sites within the proximal promoter of the gene (32). These studies also showed that NF-B site sequence specificity and the arrangement of multiple gene-specific sites is important for determining NF-B dimer selection and the binding of coactivators for gene-specific transcription (32). It is possible that this dimeric c-Rel module confers a degree of c-Rel selectivity and specific coactivator requirements on these genes. A group of 1600 genes was identified with this c-Rel module from database searches, and further investigation of the relationship of these genes to c-Rel response and c-Rel binding in different cell types is warranted.
In summary, our analysis of the role of c-Rel in T cell transcriptional responses suggests that c-Rel is a modulator of gene transcription in T cells. Moreover, we have shown that the transcriptional responses of c-Rel target genes may be dictated by the quality, number, and/or the specific arrangement of NF-B/Rel binding sites within their proximal promoters. These findings are supported by experimental evidence on specific genes. The c-Rel target genes identified here implicate c-Rel in leukocyte trafficking, particularly that related to Th1-mediated inflammation, and in cell growth control and antiapoptotic functions.
|
Acknowledgments |
|---|
|
Disclosures |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
|
Footnotes |
|---|
1 This work is supported by grants from the National Health and Medical Research Council of Australia and Diabetes Australia (to M.F.S.) and an Australian Postgraduate Award Ph.D. scholarship received by K.B. 
2 Address correspondence and reprint requests to Dr. M. Frances Shannon, Division of Molecular Bioscience, John Curtin School of Medical Research, Australian National University, Canberra, Australia. E-mail address: frances.shannon{at}anu.edu.au
3 Abbreviations used in this paper: wt, wild type; CD28RE, CD28 responsive element; IRES, internal ribosome entry site; KO, knockout; NS, nonstimulated; P/I, PMA/ionomycin; QPCR, quantitative PCR; TFBS, transcription factor binding site; TSS, transcription start site; UBC, ubiquitin-conjugating enzyme.
4 The online version of this article contains supplemental material.
Received for publication July 18, 2006. Accepted for publication March 16, 2007.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
B and Rel proteins: evolutionarily conserved mediators of immune responses. Annu. Rev. Immunol. 16: 225-260. [Medline]B signalling pathway using mouse knockout and transgenic models. Oncogene 25: 6781-6799. [Medline]B proteins in immunity and disease. Bioessays 25: 767-780. [Medline]B family of transcription factors during alloimmunity. J. Leukocyte Biol. 72: 1054-1062. B-dependent assembly of an enhanceosome-like complex on the promoter region of apoptosis inhibitor Bfl-1/A1. Mol. Cell. Biol. 23: 2749-2761. B family directly activates expression of the apoptosis inhibitor Bcl-xL. Mol. Cell. Biol. 20: 2687-2695. B-dependent c-myc expression. Immunity 21: 19-30. 
, P/I-stimulated cells in A and CD3/CD28-stimulated cells in B.
) and c-Rel sites (
) relative to the TSS. B and C, These reporter plasmids were cotransfected in Jurkat T cells with increasing amounts of c-Rel expression plasmid (10, 20, and 25 µg) and analyzed for luciferase activity in NS (B) or P/I-activated cells (C). Data are the fold changes in activity compared with empty vector-transfected cells and are the average of two transfection experiments with duplicate luciferase measurements ± SD.
