NF-{kappa}B-Dependent Assembly of an Enhanceosome-Like Complex on the Promoter Region of Apoptosis Inhibitor Bfl-1/A1

Expression of the prosurvival Bcl-2 homologue Bfl-1/A1 is inducedby NF- ${kappa}$ B-activating stimuli, while B and T cells from c-rel knockoutmice show an absolute defect in bfl-1/a1 gene activation. Here,we demonstrate NF- ${kappa}$ B-dependent assembly of an enhanceosome-likecomplex on the promoter region of bfl-1. Binding of NF- ${kappa}$ B subunitc-Rel to DNA nucleated the concerted binding of transcriptionfactors AP-1 and C/EBPß to the 5'-regulatory regionof bfl-1. Optimal stability of the complex was dependent onproper orientation and phasing of the NF- ${kappa}$ B site. Chromatin immunoprecipitationanalyses demonstrated that T-cell activation triggers in vivobinding of endogenous c-Rel, c-Jun, C/EBPß, and HMG-ICto the bfl-1 regulatory region, coincident with selective recruitmentof coactivators TAFII250 and p300, SWI/SNF chromatin remodelingfactor component BRG-1, and basal transcription factors TATA-bindingprotein (TBP) and TFIIB, as well as hyperacetylation of histonesH3 and H4. These results highlight a critical role for NF- ${kappa}$ Bin bfl-1 transcription and point to the need for a complex andprecise regulatory network to control bfl-1 expression. To ourknowledge, this is the first demonstration of enhanceosome-mediatedregulation of a cell death inhibitor.

INTRODUCTION

Top

Introduction

Activation of the Rel/NF- ${kappa}$ B transcription factors was implicatedin inhibiting cell death triggered by a variety of stimuli,ranging from radiation to genotoxic agents (9, 10, 38, 61, 62,71). This survival response is dependent on the nuclear localizationof NF- ${kappa}$ B, because expression of a physiological inhibitor (I ${kappa}$ B ${alpha}$ M)suppressed its protective activity. Consistent with this observation,a number of Rel/NF- ${kappa}$ B target genes have been described as effectorsof its antiapoptotic effects (53, 64; reviewed in reference7). Among them, our group and others previously identified theprosurvival Bcl-2 family member Bfl-1/A1 as an important transcriptionaltarget of NF- ${kappa}$ B (23, 32, 63, 72).

bfl-1 gene expression is generally confined to immune cellsand tissues, in a pattern similar to that of NF- ${kappa}$ B itself, andis strongly induced by cytokine stimulation of leukemic, endothelial,and hemopoietic cells (27, 34, 42, 60). Constitutively elevatedlevels of bfl-1 transcripts are seen in mature neutrophils andare selectively induced in long-lived peripheral B cells (60).These observations suggest an important role for Bfl-1 in thesurvival and selection of distinct subsets of cells in the immunesystem. Consistent with this hypothesis, Bfl-1 can suppressapoptosis triggered by the proinflammatory cytokine tumor necrosisfactor ${alpha}$ (TNF- ${alpha}$ ), tumor suppressor p53, B-cell receptor aggregation,proapoptotic factors Bax and Bad, and chemotherapeutic agents(13, 15, 17, 23, 25, 26, 31, 44, 49, 60, 63). Importantly, Band T cells from c-rel^-/- mice exhibit an absolute defect inthe ability to express bfl-1/a1 in response to cell activation(23). These results emphasize the notion that bfl-1 gene expressionis strictly controlled and that NF- ${kappa}$ B must play a critical rolein this process.

The transcription of eukaryotic genes is highly regulated andrelies upon the binding of the correct constellation of factorsto particular sites within a defined gene locus. For a selectnumber of tightly regulated genes, transcription also dependsupon specific DNA-protein and protein-protein interactions ina highly ordered complex referred to as an enhanceosome. Thisprecisely orchestrated complex allows for a limited subset ofgenes, such as those regulated in a tissue-specific manner,to appropriately respond to a defined group of transcriptionfactors and the multistimulatory environment they inhabit. Inthis context, synergistic transcriptional activation derivesfrom the cooperative binding of transcription factors and architecturalfactors to recruit coactivators and basal transcription factorsthrough a novel activating surface (29, 39, 40). The enhanceosomealso provides a platform for the binding of covalent histonemodifiers to alter chromatin in a location-specific manner (35).The best-characterized enhanceosome-regulated genes are thoseencoding beta interferon (IFN-ß) and T-cell receptor ${alpha}$ (TCR ${alpha}$ ) (1, 21, 29, 35, 40, 59, 67, 68).

In prior studies, we showed that various NF- ${kappa}$ B-inducing stimuliled to upregulation of bfl-1 gene expression in different cells(72). For instance, bfl-1 transcripts were sharply elevatedin human Jurkat T cells activated with phorbol 12-myristate13-acetate (PMA) plus ionomycin. This activation was dependenton NF- ${kappa}$ B, because bfl-1 mRNA induction was substantially reducedin cells expressing I ${kappa}$ B ${alpha}$ M. Here, we have characterized how NF- ${kappa}$ Bregulates bfl-1 gene expression. We show that the sole NF- ${kappa}$ BDNA-binding site in the 5' regulatory region of bfl-1 promotesthe cooperative binding of C/EBPß and AP-1. In thiscontext, the c-Rel and p50 subunits of NF- ${kappa}$ B nucleated an enhanceosome-likecomplex containing NF- ${kappa}$ B, AP-1, and C/EBPß togetherwith chromatin architectural factor HMG-I and transcriptionalcoactivators. Chromatin immunoprecipitation (ChIP) analysesdemonstrated that T-cell activation triggers in vivo recruitmentof these factors to the endogenous bfl-1 locus, coincident withhistone acetylation. These results indicate that bfl-1 transcriptionis regulated by an NF- ${kappa}$ B-dependent enhanceosome-like complex,highlighting the need for a complex and precise regulatory networkto control its expression.

MATERIALS AND METHODS

Top

Materials and Methods

Plasmids and mutagenesis. The -1374/+83 bfl-1:luc reporter construct was generated bysubcloning the -1374/+83 region of bfl-1 from bfl-1:CAT (72)into the promoterless pGL3 · Basic reporter (Promega).The -1374/+83 bfl-1 region was cloned into pALTER-1 (Promega)and subjected to site-directed mutagenesis to inactivate thebinding sites for NF- ${kappa}$ B (-833 GTTTATTTACC), AP-1 (-864 GTCCTA),or C/EBPß (-927 TCGCT; Altered Sites Mutagenesis System;Promega) (mutated residues are underlined). Site-directed mutagenesiswas used to alter the orientation of the NF- ${kappa}$ B site in the -1374/+83bfl-1 region (r ${kappa}$ B; 5'-GGTAAATCCCC-3'), or its phasing by repositioningit +6 or +10 bp toward the transcription start site of bfl-1(QuickChange Site-Directed Mutagenesis kit; Stratagene). -933/-773:-100/+83bfl-1:lucwas generated by amplifying the -933/-773 region of bfl-1 withprimers 5'-CGACGCGTCTCCCGGGTTCAAGCAAT-3' and 5'-TCGTGCGTAGGCTGTGCGGGCGGATTGCCT-3'and amplifying the -100/+83 region with primers 5'-AGGCAATCCGCCCGCACAGCCTACGCACGA-3'and 5'-GAAGATCTGCTGCCTGGTGGAGAGCA-3'. Products from these reactionswere combined with overlap-anneal PCR and cloned in the MluIand BglII sites of pGL3-Basic. pcDNA3:C/EBPß (46),pCB6⁺:c-jun, pCB6⁺:c-fos (5), and CMVß:p300 (18) weregifts from D. E. Zhang (The Scripps Research Institute), T.Curran (St. Jude Children's Research Hospital), and D. Livingston(Dana Farber Cancer Institute and Harvard Medical School), respectively.pCMV-c-Rel was described previously (66). Where indicated, pcDNA3.1HisC(Invitrogen) was used as a control.

Cell transfection, luciferase, and CAT assays. The HtTA-1-derived cell line HtTA-CCR43 expressed c-Rel undertetracycline-regulated control (8). Jurkat E6.1 T-lymphocyticleukemia cells were obtained from the American Type CultureCollection. Endogenous NF- ${kappa}$ B activity was induced by cell treatmentwith PMA (50 ng/ml in dimethyl sulfoxide [DMSO]) plus ionomycin(1 µM in DMSO) overnight or with DMSO alone as a control.Jurkat T cells (2 x 10⁶) were transiently transfected with atotal of 4 µg of DNA with DMRIE-C reagent (4 µl;Invitrogen). Where indicated, cells were equally divided intotwo wells at 24 h posttransfection and treated with PMA plusionomycin or with DMSO as a control. Cells were harvested at48 h posttransfection and washed once with phosphate-bufferedsaline (PBS), and extracts were prepared with 1x passive lysisbuffer (Promega). Extracts were normalized for protein content(11) and assayed for luciferase activity by using the Promegaluciferase assay reagent on a TD-20/20 luminometer (Turner Designs).Cos-7 cells (1.4 x 10⁵) were transiently transfected with atotal of 6 µg of DNA by using calcium phosphate. Cellswere harvested 48 h later. Extracts were normalized for proteincontent, and chloramphenicol acetyltransferase (CAT) activitywas assayed on 50 µg of total protein for 2 h.

Gel retardation assays. Nuclear extracts from human Jurkat T cells treated with PMAplus ionomycin for 2 or 18 h, or with DMSO as a control, wereprepared as described previously (52). DNA probes consistedof 160-bp wild-type or mutant -933/-773 fragments correspondingto the 5' regulatory region of bfl-1 cloned into pBLCAT5 (AmericanType Culture Collection). Fragments were isolated from pBLCAT5by restriction with XbaI and SalI, ³²P labeled with Klenow enzyme,and used as probes in gel shift assays. Nuclear extracts (3µg) were incubated with DNA probes (4 x 10⁴ cpm) in amixture containing 10 mM Tris-HCl (pH 7.5), 50 mM NaCl, 7.5mM MgCl₂, 1 mM EDTA, 5% glycerol, 5% sucrose, 0.1% NP-40, 0.5µg of poly(dI-dC), 5 mM dithiothreitol (DTT), and 0.5mg of bovine serum albumin (BSA) per ml, and analyzed on 5%native polyacrylamide gels. Where indicated, DNA-protein complexeswere supershifted with anti-hc-Rel or anti-p50 antibodies (1136and 1263, respectively; a gift from N. Rice, ABL-National CancerInstitute, Frederick, Md.), anti-p65, anti-c-Jun, anti-C/EBPß(sc-109, -822, and -150; Santa Cruz Biotechnology), anti-p300(14991A; Pharmingen), or anti-HMGI-C (a gift from K. Chada,Robert Wood Johnson Medical School, Piscataway, N.J.) antibodies.Normal rabbit serum (NRS; Calbiochem) was used as a control.

In competition gel retardation assays, ³²P-labeled probes wereincubated with nuclear extracts as described above. After a30-min incubation, an aliquot of each binding reaction mixturewas loaded onto a nondenaturing gel, and a 250-fold molar excessof unlabeled double-stranded oligonucleotide probe containingthe C/EBPß binding site from the bfl-1 regulatoryregion (5'-CCCGGGTTCAAGCAATTCCCTCTCCTT-3') was added to thebinding reaction mixtures. Aliquots were taken at time intervalsand loaded onto a native 5% polyacrylamide gel, which was allowedto run between loadings. Bound complexes were quantified withImageQuant software (Molecular Dynamics).

Northern blot analysis. Total RNA (30 µg) was extracted with RNAzol B (TEL-TEST),fractionated on a 1% agarose-formaldehyde gel, and transferredonto a Hybond NX membrane (Amersham). The membrane was bakedfor 10 min at 80°C under vacuum and UV cross-linked witha Stratalinker (Stratagene). Probes were generated by randompriming with Klenow polymerase in the presence of [³²P]dCTPand [³²P]dGTP (19). The membrane was hybridized in Perfect HybPlus hybridization buffer (Sigma) at 68°C overnight. Themembrane was washed twice in a mixture containing 2x SSC (1xSSC is 0.15 M NaCl plus 0.015 M sodium citrate) and 0.05% sodiumdodecyl sulfate (SDS) at room temperature and twice in 0.1xSSC-0.1% SDS at 50°C, followed by autoradiography.

DNA-protein immunoprecipitation assays. Nuclear extracts (5 µg) from HtTA-CCR43 cells uninducedor induced to produce c-Rel upon removal of tetracycline for96 h were incubated with ³²P- and bromodeoxyuridine (BrdU)-labeledwild-type or m ${kappa}$ B mutant -933/-773 bfl-1 enhancer probes (50 fmol)in a mixture of 20 mM Tris-HCl (pH 7.8), 50 mM NaCl, 10 mM MgCl₂,0.1 mM EDTA, 15% glycerol, 2 mM DTT, 0.01% NP-40, 0.1 mM phenylmethylsulfonylfluoride (PMSF), 100 µg of BSA per ml, and 1 µgof poly(dI-dC) per 17 µl. Binding reactions were subjectedto UV cross-linking and immunoprecipitated with anti-c-Jun,anti-C/EBPß, anti-HMGI-C, or anti-c-Rel antibodies(22) or with normal rabbit immunoglobulin G (IgG) as a control(Calbiochem). Complexes were resolved on SDS-5% polyacrylamidegels.

ChIP assays and real-time PCR analysis. ChIP assays were performed as described (45, 47) with the followingmodifications. Jurkat T cells (5 x 10⁷) were treated with PMAplus ionomycin for 2 or 18 h or with DMSO alone as a control.After cross-linking with formaldehyde, chromatin was extracted,sonicated, and purified on CsCl gradients. Fractions containingchromatin were pooled and immunoprecipitated with antibodiesspecific for c-Rel (1136; N. Rice, ABL-National Cancer Institute),p65/RelA, c-Jun, C/EBPß, p/CAF, or TAFII250 (sc-109,-45, -150, -8999, and -735; Santa Cruz Biotechnology), HMGI-C(K. Chada, Robert Wood Johnson Medical School), p300 (14991A;Pharmingen), CBP (D. M. Livingston, Dana-Farber Cancer Institute,Boston, Mass.), BRG-1 (W. Wang, National Institute on Aging,National Institutes of Health), hTBP or hTFIIB (D. Reinberg,Robert Wood Johnson Medical School), or acetylated histone H3or H4 (06-599 and 06-866, respectively; Upstate Biotechnology).After washing, cross-linking was reversed, and the precipitatedchromatin was extracted with phenol-chloroform. After ethanolprecipitation, the DNA was resuspended in water (25 µl)and PCR amplified.

In standard PCRs, immunoprecipitated DNA (1 µl) was amplifiedfor 25 cycles with Amplitaq (Applied Biosystems). In real-timePCR analyses, immunoprecipitated DNA (1 µl) was amplifiedin 50-µl reaction mixtures containing SYBR Green mastermix (25 µl; Applied Biosystems) and oligonucleotide primersspecific for the 5' regulatory region of bfl-1 (300 µMeach; forward, 5'-TGGAGACAGAGTCTCGCTCTGTT-3'; reverse, 5'-GAACCCGGGAGACAGAAGTTG-3'),the proximal promoter region of bfl-1 (forward, 5'-GGATATTATAAAGTGATGCAAACAGAAATT-3';reverse, 5'-TGAGGCAATGTGCTGAGAATG-3'), or the 5' regulatoryregions of the I ${kappa}$ B ${alpha}$ gene (forward, 5'-GGCTCATCGCAGGGAGTTT-3';reverse, 5'-GAACTGGCTTCGTCCTCTGCTA-3') and the BLR1 chemokinereceptor gene (forward, 5'-TGAAATGCTTGACTAGCACTGATG-3'; reverse,5'-CGCGACCTGCCTCACAAC-3') as controls. Incorporation of SYBRGreen dye into PCR products was monitored in real time withan ABI PRISM 7700 sequence detection system (PE Applied Biosystems),allowing determination of the threshold cycle (C_T) at whichexponential amplification of PCR products begins. Each reactionwas performed in triplicate. The C_T values for PCR productscorresponding to individual loci were used to calculate theamount of chromatin immunoprecipitated in each reaction. Relativeenrichment over the no-antibody control was first determinedfor each sample. Relative factor recruitment in PMA- plus ionomycin-treatedcells was then normalized to the DMSO control for each antibody.The specificity of DNA amplification by primer pairs was verifiedby standard PCR (40 cycles) under the cycling conditions usedfor real-time analysis and evaluated by agarose gel electrophoresisand ethidium bromide staining. Absence of primer-dimer formationwas confirmed in amplification reactions without template.

RESULTS

Top

Results

T-cell activation induces binding of NF- ${kappa}$ B, AP-1, C/EBPß, HMGI-C, and p300 to the 5' bfl-1 regulatory region. Detailed analysis of a 1.4-kbp genomic fragment comprising the5' regulatory region of bfl-1 revealed consensus binding sitesfor transcription factors C/EBPß (-927) and AP-1 (-864)located within 100 bp of the sole NF- ${kappa}$ B DNA binding motif (-833)that we previously implicated in regulating bfl-1 transcription(72) (Fig. 1A). An AT-rich sequence corresponding to the recognitionmotif for chromatin modeling factor HMG-I (AATTTTTT) overlappedthe putative AP-1 binding site. The binding of endogenous NF- ${kappa}$ B,AP-1, and C/EBPß proteins to the 5' bfl-1 regulatoryregion was investigated with gel retardation assays. Incubationof nuclear extracts from human Jurkat T cells stimulated withPMA plus ionomycin with a 160-bp probe (-933/-773) derived fromthe 5' bfl-1 regulatory region induced formation of a large-molecular-weightcomplex compared to results with extracts from control cellstreated with DMSO (Fig. 1B, compare lanes 3 and 12 to lanes2 and 11). Parallel Northern blot analysis showed that thiswas correlated with a sharp induction of endogenous bfl-1 transcripts(Fig. 1C, compare lanes 2 and 3 to lane 1). Supershift analysesrevealed the presence of endogenous c-Rel, p50, c-Jun, and C/EBPßin this complex (Fig. 1B, lanes 4, 5, 7, and 8, respectively).An antibody specific for chromatin architectural factor HMGI-Csupershifted the complex, while an antibody for coactivatorp300 interrupted its formation (lanes 13 and 14). HMGI-C isrelated to HMG-I(Y), which is found in the IFN-ß enhanceosome,and enhances NF- ${kappa}$ B function in a manner similar to HMG-I(Y) (37,58, 67). In contrast, antibodies to RelA/p65 and HMG-I(Y) orNRS had no such effect (lanes 6 and 9) (data not shown). Theseresults demonstrate the binding of endogenous NF- ${kappa}$ B, c-Jun, C/EBPß,HMGI-C, and p300 proteins to the 5' regulatory region of bfl-1and indicate that the primary NF- ${kappa}$ B dimer regulating bfl-1 inactivated Jurkat T cells is a c-Rel/p50 heterodimer.

View larger version (41K):
[in this window]
[in a new window]
FIG. 1. T-cell activation causes binding of c-Rel, p50/NF- ${kappa}$ B1, c-Jun, C/EBPß, HMGI-C and p300 to the 5' regulatory region of bfl-1. (A) Schematic representation of the 5' regulatory region of the human bfl-1 gene. Binding sites for NF- ${kappa}$ B, AP-1, C/EBPß, and HMGI are indicated. The sequence of the 1.4-kbp 5'-regulatory region of bfl-1 is available in GenBank (accession no. AF479683). (B) Gel retardation analysis of factor binding to the 5' regulatory region of bfl-1. Jurkat T cells were treated with PMA plus ionomycin (Iono) for 18 h or with DMSO as a control. Nuclear extracts were incubated with a 160-bp probe derived from the -933/-773 region of bfl-1. Where indicated, antibodies to c-Rel, p50, p65, c-Jun, C/EBPß, p300, and HMGI-C or NRS was added to the binding reaction mixtures. Arrows indicate the DNA-protein complex, and brackets indicate supershifted complexes. (C) Northern blot analysis of endogenous bfl-1 transcripts in Jurkat T cells stimulated with PMA plus ionomycin for 2 or 18 h (lanes 2 and 3) or with DMSO as a control (lane 1). RNA was successively hybridized to probes specific for genes coding for Bfl-1 or GAPDH.

NF- ${kappa}$ B binding to DNA promotes the binding of AP-1, C/EBPß, HMGI-C, and p300 to the -933/-773 bfl-1 probe. The binding of HMGI-C to the 5' regulatory region of bfl-1 andits close proximity to binding sites for NF- ${kappa}$ B, AP-1, and C/EBPßare somewhat reminiscent of the features of its homologue, HMG-I(Y),and transcription factors ATF2/c-Jun, IRF1, and NF- ${kappa}$ B in theenhanceosome-regulated IFN-ß-positive regulatory domains(36, 39, 43, 59). We thus explored whether individual mutationof the NF- ${kappa}$ B, AP-1, and C/EBPß sites in the -933/-773bfl-1 probe would affect the binding of the other factors inthe complex. The loss of transcription factor binding to theindividual mutant sites was confirmed in gel retardation assays(data not shown). In the context of the -933/-773 bfl-1 regulatoryprobe, mutation of the NF- ${kappa}$ B site alone suppressed inductionof the high-molecular-weight DNA-protein complex upon activationof Jurkat T cells with PMA plus ionomycin compared to the DMSOcontrol (m ${kappa}$ B; Fig. 2A, compare lanes 5 and 6 to lanes 2 and 3).Surprisingly, mutation of either the unique AP-1 (mAP-1) orC/EBPß binding sites (mC/EBPß) in the -933/-773probe still allowed formation of a DNA-protein complex witha mobility similar to that formed on the wild-type probe (lanes9 and 12). This raised the possibility that the binding of ac-Rel/p50 heterodimer to the bfl-1 regulatory region may helpto nucleate the binding of AP-1 and C/EBPß to theprobe. Alternatively, the compositions of the DNA-protein complexeson the wild-type and mutant bfl-1 probes may differ despitetheir apparently similar mobilities.

View larger version (48K):
[in this window]
[in a new window]
FIG. 2. The binding of NF- ${kappa}$ B to DNA promotes the binding of AP-1, C/EBPß, HMGI-C, and p300 to the -933/-773 bfl-1 probe. (A) Nuclear extracts from Jurkat T cells treated with PMA plus ionomycin (Iono), or with DMSO as control, were incubated with 160-bp probes derived from the -933/-773 region of bfl-1. The probes were either wild type (WT) or mutated at the NF- ${kappa}$ B, AP-1, or C/EBPß binding sites as indicated. The bracket highlights the large DNA-protein complex. (B) Supershift analysis of DNA-protein complexes formed on mutant AP-1 or C/EBPß probes with antibodies (Ab) specific to each factor. Arrows point to the DNA-protein complex, and brackets indicate supershifted complexes. (C) Supershift analysis of the complex bound to a -933/-773 probe in which the AP-1 and C/EBPß DNA sites were mutated simultaneously.

To discriminate between these two possibilities, the compositionof complexes formed on the mutant AP-1 and C/EBPßprobes was analyzed with supershift assays. Interestingly, c-Rel,p50, c-Jun, and C/EBPß were all found to be presentin complexes that formed on both the mAP-1 and mC/EBPß-933/-773 bfl-1 probes, despite the mutation of their respectiveDNA recognition sites (Fig. 2B, lanes 4 to 7 and 11 to 14).Similar results were obtained with a -933/-773 bfl-1 probe inwhich the C/EBPß and AP-1 sites were mutated simultaneously(mAP-1,C/EBPß). A high-molecular-weight DNA-proteincomplex still formed on the mutant probe with the NF- ${kappa}$ B DNA siteas the sole remaining recognition motif (Fig. 2C, lanes 3 and12). As with the wild-type probe, supershift analysis revealedthe presence of c-Rel, p50, c-Jun, C/EBPß, HMGI-C,and p300 in this complex (lanes 4, 5, 7, 8, 13, and 14, respectively).These data indicated that the multiprotein complexes that formedon the mutant bfl-1 probes contained the same transcriptionfactors as the complex bound to the wild-type probe. In conjunctionwith the data on the m ${kappa}$ B probe (Fig. 2A), these results suggestthat efficient complex formation on the 5' regulatory regionof bfl-1 is dependent on the binding of a c-Rel/p50 heterodimerto the NF- ${kappa}$ B site. Although we do not rule out the possibilitythat the high concentration of factors and probe available forin vitro binding may allow protein-protein interactions to sufficein gel retardation assays, it is also possible that the DNArecognition motifs for AP-1 and C/EBPß play a secondaryrole in vitro. In this scenario, the well-documented protein-proteininteractions between NF- ${kappa}$ B, AP-1, and C/EBPß are likelyto play an important role in stabilizing the complex (12, 16,51, 54-56, 70).

NF- ${kappa}$ B nucleates the binding of a multiprotein complex containing NF- ${kappa}$ B, AP-1, C/EBPß, and HMGI-C to the bfl-1 regulatory region. Since an intact NF- ${kappa}$ B DNA recognition motif was necessary topromote efficient binding of c-Jun and C/EBPß to the-933/-773 bfl-1 probe, we went on to test the hypothesis thatNF- ${kappa}$ B participates in recruiting these factors. To this end,we used nuclear extracts from HtTA-CCR43 cells, which conditionallyexpressed c-Rel under tetracycline-regulated control. Extractsfrom uninduced cells (+tet) or cells induced to express c-Rel(-tet) were tested for endogenous transcription factor bindingto the 5' regulatory region of bfl-1 in DNA-protein immunoprecipitationassays. After incubation with ³²P- and BrdU-labeled wild-typeor m ${kappa}$ B mutant -933/-773 bfl-1 probes, DNA-protein complexes weresubjected to UV cross-linking and immunoprecipitated with antibodiesspecific for c-Rel, c-Jun, C/EBPß, or HMGI-C. DNA-boundprotein complexes were resolved by SDS-polyacrylamide gel electrophoresisand revealed by autoradiography.

In the absence of c-Rel induction, no DNA-protein complex wasimmunoprecipitated with any of the antibodies specific for c-Rel,c-Jun, C/EBPß, or HMGI-C (Fig. 3A, lanes 2, 3, 5,and 6, respectively), despite the presence of endogenous c-Junand C/EBPß proteins in the cells as detected by Westernblotting (+tet; Fig. 3B). In sharp contrast, induction of c-Relexpression upon removal of tetracycline was sufficient to provokeimmunoprecipitation of very-large-molecular-mass DNA-proteincomplexes (>200 kDa) on the wild-type bfl-1 probe with anti-c-Rel,-c-Jun, -C/EBPß, or -HMGI-C antibodies (Fig. 3A, lanes8, 9, 11, and 12, respectively). This assay was specific, becauseno complex was immunoprecipitated with a normal rabbit IgG control(lanes 7 and 10). Moreover, mutation of the NF- ${kappa}$ B DNA site inthe -933/-773 bfl-1 probe abolished immunoprecipitation of DNA-proteincomplexes with all antibodies, further suggesting that the bindingof c-Rel to DNA is necessary to allow efficient binding of endogenousc-Jun, C/EBPß, and HMGI-C to this region (m ${kappa}$ B; Fig.3A, lanes 20, 21, 23, and 24). Altogether, these data indicatethat the binding of NF- ${kappa}$ B to the 5' regulatory region of bfl-1nucleates the cooperative binding of AP-1, C/EBPß,and HMGI-C. In conjunction with the experiments described abovesuggesting that NF- ${kappa}$ B may help to anchor this complex throughprotein-protein interactions, we postulate that bfl-1 gene expressionis regulated by an enhanceosome-like complex that is nucleatedby and dependent on NF- ${kappa}$ B DNA binding.

View larger version (24K):
[in this window]
[in a new window]
FIG. 3. NF- ${kappa}$ B nucleates formation of a DNA-protein complex containing c-Rel/p50, c-Jun, C/EBPß, and HMGI-C on the -933/-773 bfl-1 probe. (A) Nuclear extracts from uninduced (+tet, lanes 1 to 6 and 13 to 18) or induced (-tet, lanes 7 to 12 and 19 to 24) HtTA-CCR43 cells were incubated with a BrdU- and ³²P-labeled wild-type -933/-773 bfl-1 probe (lanes 1 to 12) or a mutant probe in which the NF- ${kappa}$ B site was inactivated (lanes 13 to 24). After UV cross-linking, DNA-protein complexes were immunoprecipitated (Ippt) with antibodies (Ab) specific for c-Rel, c-Jun, C/EBPß, and HMGI-C or control normal rabbit IgG, followed by autoradiography. (B) Western blot analysis of endogenous c-Rel, c-Jun, and C/EBPß proteins in nuclear extracts from HtTA-CCR43 cells uninduced (+tet) or induced to express c-Rel (-tet).

The orientation and phasing of the NF- ${kappa}$ B site affect the stability of the complex. The well-documented interactions between NF- ${kappa}$ B, AP-1, and C/EBPß(12, 16, 51, 54-56, 70) led us to assess possible effects ofalterations in the orientation and phasing of the NF- ${kappa}$ B bindingsite on the formation or stability of the complex. In the -933/-773bfl-1 probe, the center of the C/EBPß site is locatedsix helical turns away from the center of the AP-1 binding site.The center of the AP-1 binding site is three helical turns awayfrom the center of the NF- ${kappa}$ B site. To address the orientationrequirement, the NF- ${kappa}$ B site was placed on the opposite strandof the DNA, reversing its orientation with respect to the AP-1and C/EBPß sites in the -933/-773 probe (r ${kappa}$ B). Theneed for proper phasing was investigated by repositioning theNF- ${kappa}$ B site 6 or 10 bp away from the AP-1 and C/EBPßsites ( ${kappa}$ B+6 and ${kappa}$ B+10, respectively). This effectively moved theNF- ${kappa}$ B site one-half or a full helical turn away from the restof the bfl-1 complex (Fig. 4A).

View larger version (42K):
[in this window]
[in a new window]
FIG. 4. Phasing and orientation of the NF- ${kappa}$ B dimer affect the stability of the complex on the -933/-773 bfl-1 probe. (A) Schematic diagram of -933/-773 bfl-1 positional mutants. Arrows indicate the orientation of the NF- ${kappa}$ B dimer with respect to the rest of the 5' bfl-1 regulatory region. (B) Nuclear extracts from Jurkat T cells stimulated with PMA plus ionomycin were incubated with wild-type (WT) or mutant -933/-773 bfl-1 probes (r ${kappa}$ B, ${kappa}$ B+6, and ${kappa}$ B+10). Aliquots of the binding reaction mixtures were loaded onto a running nondenaturing gel at time zero and at intervals following addition of a 250-fold molar excess (250x) of an unlabeled competitor oligonucleotide. (C) Bound complexes were quantitated with ImageQuant software (Molecular Dynamics). Relative bound complex represents the ratio of DNA-bound complex remaining at indicated time points over the amount of bound complex at time zero. The data are representative of three independent experiments.

Competition gel retardation assays revealed significant differencesin the stability of complexes forming on the wild-type and ${kappa}$ B+10probes compared to those on the r ${kappa}$ B and ${kappa}$ B+6 probes, in whichthe NF- ${kappa}$ B binding site is located on the opposite face of thehelix. In these assays, DNA-binding reactions on the wild-type,r ${kappa}$ B, ${kappa}$ B+6, and ${kappa}$ B+10 probes were challenged with a 250-fold molarexcess of an unlabeled C/EBPß competitor oligonucleotide.Aliquots of the binding reaction mixtures were loaded onto arunning native gel at intervals following addition of the competitor(Fig. 4B). As plotted in Fig. 4C, the amount of remaining complexbound to the r ${kappa}$ B and ${kappa}$ B+6 probes following addition of competitorwas approximately 50% of that of the remaining complexes boundto the wild-type and ${kappa}$ B+10 probes. This indicated that improperphasing and orientation of the NF- ${kappa}$ B site with respect to theAP-1 and C/EBPß motifs adversely affect the stabilityof the complex (r ${kappa}$ B and ${kappa}$ B+6), in contrast to the wild-type and ${kappa}$ B+10 probes, in which proper phasing was restored. Similar resultswere obtained with gel retardation assays performed in the presenceof the anionic detergent Sarkosyl (data not shown). Altogether,these data support the premise that proper orientation and phasingof the NF- ${kappa}$ B dimer relative to the AP-1 and C/EBPßdimers are required for optimal stability of the complex.

Transcription coactivator p300 synergizes with NF- ${kappa}$ B, AP-1, and C/EBPß to activate the bfl-1 promoter. Consistent with their concerted binding to the -933/-773 bfl-1probe, NF- ${kappa}$ B, AP-1, and C/EBPß synergistically activatedexpression from the bfl-1 promoter (Fig. 5A). Jurkat T cellscotransfected with a -1374/+83bfl-1:luc reporter and small amountsof expression vectors for c-Rel or C/EBPß alone showedvirtually no activation under these conditions, whereas theuse of c-Fos and c-Jun (AP-1) led to a modest increase in luciferaseactivity. However, cotransfection of all three transcriptionfactors together led to a sharp 39-fold synergistic activationof the promoter. Moreover fusion of the -933/-773 fragment fromthe 5' bfl-1 regulatory region that encompasses only the C/EBPß,AP-1, NF- ${kappa}$ B, and HMG-I sites to the proximal promoter regionof bfl-1 was strongly activated in response to PMA plus ionomycin,and this was suppressed by a dominant I ${kappa}$ B ${alpha}$ M inhibitor of NF- ${kappa}$ B(-933/-773:-100/+83bfl-1:luc; Fig. 5B). Similar results wereobserved following CD3/CD28 stimulation (data not shown). Theseresults underscore an important role for the -933/-773 regionin the transcriptional regulation of bfl-1 and demonstrate thatthis region can function at a distance from or in close proximityto the bfl-1 promoter region to drive gene expression. Consistentwith the in vitro DNA-binding studies on mutant bfl-1 probes(Fig. 2), mutation of the NF- ${kappa}$ B recognition motif abolished reportergene activation, whereas mutation of the C/EBPß sitehad no significant effect (Fig. 5C). Interestingly, mutationof the AP-1 site led to a 50% decrease in activation, suggestingthat it is not completely dispensable in vivo. Altogether, theseresults further emphasize an important role for NF- ${kappa}$ B in allowingfactors to synergize on the bfl-1 promoter.

View larger version (71K):
[in this window]
[in a new window]
FIG. 5. NF- ${kappa}$ B, AP-1, and C/EBPß synergistically activate transcription from the bfl-1 regulatory region with coactivator p300. (A) Jurkat T cells were transiently cotransfected with a -1374/+83bfl-1:luc reporter construct (1.33 µg) and expression vectors for c-Rel, c-Jun, c-Fos, and C/EBPß (667 ng each), alone or together (3 TF). The total amount of transfected DNA was kept constant by addition of empty pcDNA3.1HisC vector. The average of three independent experiments is shown. (B) Transcriptional activity of the -933/-773 bfl-1 regulatory region. Jurkat T cells were transfected with 360 ng of -933/-773:-100/+83 bfl-1:luc or control -100/+83 bfl-1:luc reporter and 3.6 µg of pcDNA3 control or pcDNA3:I ${kappa}$ B ${alpha}$ M. Cells were treated with DMSO or stimulated with PMA plus ionomycin for 18 h prior to harvest. The results show the average fold activation of -933/-773:-100/+83 bfl-1:luc by treatment with PMA plus ionomycin from which background activity of the -100/+83 bfl-1:luc control was subtracted. (C) Cells were transfected with wild-type or m ${kappa}$ B, mAP-1, or mC/EBPß mutant -993/-773 bfl-1:CAT reporters (1.5 µg) and 1.5 µg each of CMV-c-Rel, CMV-c-Jun, and CMV-C/EBPß DNA or 4.5 µg of an empty pcDNA3.1HisC vector as a control. CAT activity was assayed at 48 h posttransfection. (D) Jurkat T cells were cotransfected with a total of 4 µg of DNA containing -1374/+81 bfl-1:luc (730 ng) together with expression vectors for c-Rel, c-Jun, c-Fos, and C/EBPß (360 ng each; TF mix), alone or together with p300 (1.8 µg). The average of three independent experiments is shown. (E) Assays were conducted essentially as in panel D, but with the -933/-773:-100/+83 bfl-1:luc reporter construct. The average of three independent experiments is shown.

NF- ${kappa}$ B, AP-1, and C/EBPß have all been shown to interactphysically and functionally with transcriptional coactivatorsp300 and CBP (3, 6, 20, 41, 50). In light of the critical roleof coactivators in regulating the IFN-ß enhanceosome(40), the recruitment of p300 to the -933/-773 probe raisedthe possibility of a functional role in the activation of bfl-1.To address this issue, an expression vector for p300 was cotransfectedinto Jurkat T cells along with expression vectors for c-Rel,c-Jun, c-Fos, C/EBPß, and the -1374/+83bfl-1:luc and-933/-773:-100/+83bfl-1:luc reporter constructs. As expected,transfection of p300 alone failed to increase transcriptionof the -1374/+83bfl-1:luc reporter (Fig. 5D). However it doubledactivation observed with the combination of NF- ${kappa}$ B, AP-1, andC/EBPß on either the -1374/+83bfl-1:luc or the -933/-773:-100/+83bfl-1:lucreporter (Fig. 5D and E). Combined with the gel shift data presentedabove, this demonstrates that p300 can synergistically activatetranscription from the bfl-1 promoter along with NF- ${kappa}$ B, AP-1,and C/EBPß.

T-cell activation induces complex assembly on the 5' regulatory region of bfl-1 in chromatin, leading to coactivator recruitment and histone hyperacetylation. To demonstrate their participation in a physiologically relevantcontext, in vivo assembly of endogenous factors on the 5' regulatoryregion of bfl-1 was investigated by ChIP. Human Jurkat T cellswere activated with PMA plus ionomycin for 2 or 18 h, or withDMSO as a control, and treated with formaldehyde to cross-linkendogenous protein-DNA and protein-protein complexes. Aftersonication, chromatin was immunoprecipitated with factor-specificantibodies followed by quantitative analysis by real-time PCRwith primers specific for the 5' regulatory region of bfl-1.A representative real-time PCR profile is shown in Fig. 6A.DNA enrichment in antibody-containing samples relative to theno-antibody controls was calculated by using the differencein the average threshold cycle (C_T) from three PCR amplificationsfor each immunoprecipitation.

View larger version (155K):
[in this window]
[in a new window]
FIG. 6. ChIP analysis of in vivo complex formation on the 5' regulatory region of bfl-1 upon T-cell activation. (A) Representative real-time PCR analysis of ChIP assays with an anti-c-Rel antibody (Ab). Accumulation of product in cells treated with PMA plus ionomycin or with DMSO as a control normalized to reporter fluorescence (R_N; y axis) is plotted as a function of cycle number (x axis). The threshold (dashed line) was calculated by multiplying the standard deviation of the R_N of the baseline (cycles 3 to 15) by 10. The average of three independent reactions is shown. (B) ChIP analysis of transcription factors c-Rel, p65/RelA, c-Jun, and C/EBPß and of architectural factor HMGI-C binding to the bfl-1 locus in human Jurkat T cells stimulated with PMA plus ionomycin (P+I) for 2 or 18 h or with a DMSO control. Factor binding was evaluated by determining the difference in the average number of PCR cycles at which amplification of the bfl-1 regulatory region reached the threshold between chromatin samples immunoprecipitated with transcription factor-specific antibodies and the no-antibody control (2^-CT). Standard deviations were calculated by using the formula s₁² + s₂² (ABI Prism 7700 Sequence Detection System, user bulletin 2, 1997. Applied Biosystems, Foster City, Calif.). (C) Control ChIP analysis showing p65/RelA recruitment to the 5' regulatory region of the NF- ${kappa}$ B target gene encoding I ${kappa}$ B ${alpha}$ in Jurkat T cells activated with PMA plus ionomycin and of c-Rel, c-Jun, and C/EBPß recruitment to the upstream regulatory region of the B-cell-specific chemokine receptor gene BLR1 as a negative control. (D) ChIP analysis of coactivator p300, TAFII250, CBP, and p/CAF recruitment in Jurkat T cells treated with DMSO or following activation with PMA plus ionomycin. (E) ChIP analysis of SWI/SNF chromatin remodeling factor component BRG-1 and basal transcription factors TBP and TFIIB recruitment to the proximal bfl-1 promoter region in human Jurkat T cells stimulated with PMA plus ionomycin or with a DMSO control. (F) ChIP analysis of histone H3 and H4 acetylation on the bfl-1 regulatory region before and after Jurkat T-cell activation with PMA plus ionomycin (top panel). ChIP samples from unstimulated or PMA-plus-ionomycin-treated Jurkat T cells were immunoprecipitated with an antibody for hyperacetylated histone H3 and subjected to 25 rounds of standard nonquantitative PCR. The products were resolved on an agarose gel and revealed by ethidium bromide staining (bottom panel).

T-cell activation led to a strong increase in endogenous c-Relbinding to the 5' regulatory region of bfl-1 at the 2-h timepoint compared to that in the DMSO control (23-fold; Fig. 6B),consistent with the sharp induction in endogenous bfl-1 transcriptsin Jurkat T cells treated with PMA plus ionomycin for 2 h (Fig.1C) (72). In agreement with the in vitro gel retardation analyses,no binding of p65/RelA to the 5' regulatory region of bfl-1was detected under these conditions (Fig. 6B). In contrast,p65 was efficiently recruited to the 5' regulatory region ofthe NF- ${kappa}$ B target gene encoding I ${kappa}$ B ${alpha}$ in control ChIP assays (Fig.6C). Akin to c-Rel, binding of endogenous c-Jun and C/EBPßto the bfl-1 regulatory region was also sharply induced uponT-cell activation compared to that in DMSO-treated cells ( $~$ 20-fold;Fig. 6B). While detection of c-Rel and c-Jun was reduced atthe 18-h time point, that of C/EBPß remained strong.Since efficient binding of c-Rel to DNA was detected at 2 and18 h poststimulation in in vitro gel retardation assays, coincidentwith bfl-1 transcript expression (Fig. 1B and C) (data not shown),it is possible that in the context of chromatin, some epitopesin c-Rel may become masked at later time points, thereby precludingits efficient detection by ChIP at 18 h (Fig. 6B). The factthat a different anti-c-Rel antibody could detect sevenfoldmore c-Rel bound at the bfl-1 locus in cells activated for 18h compared to the level in the DMSO control is consistent withthis hypothesis (data not shown). Concomitant with the recruitmentof c-Rel, c-Jun, and C/EBPß, recruitment of architecturalfactor HMGI-C was observed at 2 h and was further increasedat 18 h poststimulation (Fig. 6B). The specificity of theseanalyses was confirmed in control ChIP assays examining factorrecruitment to the 5' regulatory region of the gene for chemokinereceptor BLR1, a B-cell-specific NF- ${kappa}$ B target that is not expressedin Jurkat T cells (2, 65). As anticipated, little or no recruitmentof c-Rel, c-Jun, or C/EBPß to the BLR1 locus was observedfollowing T-cell stimulation (Fig. 6C). Combined, these resultsdemonstrate efficient in vivo recruitment of endogenous c-Rel,c-Jun, C/EBPß, and HMGI-C proteins to the 5' regulatoryregion of bfl-1 in response to T-cell activation.

In vivo assembly of c-Rel, c-Jun, C/EBPß, and HMGI-Cat the bfl-1 locus coincided with preferential recruitment oftranscriptional coactivators TAFII250 and p300, but not theclosely related coactivator CBP or associated factor P/CAF (Fig.6D). Moreover, coactivator binding to the complex was correlatedwith transient recruitment of the BRG-1 ATPase component ofthe SWI/SNF remodeling complex and general transcription factorsTBP and TFIIB at the proximal promoter region of bfl-1 (Fig.6E). Importantly, the presence of acetyltransferases p300 andTAFII250 in the complex was correlated with hyperacetylationof histones H3 and H4 (Fig. 6F). As with the IFN-ßenhanceosome, acetylation of histone H4 was transient and wasdramatically reduced at the 18-h time point when TAFII250 wasno longer present. In contrast, association of coactivator p300with the bfl-1 locus was sharply elevated at 18 h poststimulationand was rather correlated with hyperacetylation of histone H3.Altogether, these studies in the context of intact chromatindemonstrate that T-cell activation provokes the in vivo bindingof NF- ${kappa}$ B, AP-1, C/EBPß, and HMGI to the bfl-1 regulatoryregion and recruitment of histone acetyltransferases, chromatinremodeling complex SWI/SNF, and basal transcription factors.In conjunction with our gel retardation, DNA coimmunoprecipitation,and transactivation assays, these findings support a model inwhich NF- ${kappa}$ B promotes assembly of an enhanceosome-like complexto acetylate histones and induce bfl-1 gene expression uponlymphoid cell activation (Fig. 7).

View larger version (21K):
[in this window]
[in a new window]
FIG. 7. Model for induction of bfl-1 gene transcription by an NF- ${kappa}$ B-dependent enhanceosome-like complex. (A) In unstimulated cells, histones located in the 5' regulatory region of bfl-1 are hypoacetylated and the bfl-1 gene is not expressed. (B) Upon cell activation, NF- ${kappa}$ B enters the nucleus and promotes assembly of transcription factors AP-1, C/EBPß, and architectural factor HMGI on the 5' regulatory region of bfl-1. (C) This enhanceosome-like complex recruits coactivators TAFII250 and p300 and the SWI/SNF chromatin-remodeling complex, leading to acetylation of histone tails and recruitment of basal transcription factors to drive bfl-1 gene expression.

DISCUSSION

Top

Discussion

The transcriptional regulation of eukaryotic genes is an intricateprocess. Nucleosomes, ATP-dependent chromatin remodeling factors,covalent chromatin modifiers, and sequence-specific DNA-bindingproteins play key roles in determining whether a particulartranscript is expressed or not (reviewed in reference 33). Thiscomplexity is reflected in the enhanceosome mechanism, describedin detail for the IFN-ß gene (40, 58, 59, 67, 68).Our description here of the assembly of an NF- ${kappa}$ B-dependent enhanceosome-likecomplex on the bfl-1 promoter is the first to suggest an enhanceosomemechanism to control transcription of a cell death inhibitor.The NF- ${kappa}$ B-mediated regulation of bfl-1 can be defined as enhanceosome-likebased on (i) the cooperativity in transcription factor bindingobserved in gel retardation, protein-DNA immunoprecipitation,and ChIP assays; (ii) extensive protein-protein interactionshighlighted by the ability of NF- ${kappa}$ B to recruit and anchor AP-1,C/EBPß, HMGI-C, and p300 on the DNA; and (iii) synergyin transactivation between these transcription factors and withtranscriptional coactivators.

The IFN-ß enhanceosome complex is dependent on theorientation of individual transcription factor binding sites(29, 59). Here, we have examined the requirements for formationof the bfl-1 enhanceosome-like complex and the subsequent eventsthat occur in vivo. While the IFN-ß enhanceosome isproximal to the transcription start site, the bfl-1 enhanceosomeis located distally. However, repositioning the -933/-773 regulatoryregion in proximity to the core bfl-1 promoter retained 40 to50% activity in response to PMA plus ionomycin compared to thewild type -1374/+83 bfl-1 reporter. Supershift and gel retardationassays revealed a dependence on proper orientation and phasingof the NF- ${kappa}$ B site in the bfl-1 regulatory region for optimalstability of the complex. Relocation of this site to the oppositeface of the helix, by reversing its orientation or moving itby half of a helical turn, destabilized the complex. The factthat complexes could assemble on mutant r ${kappa}$ B and ${kappa}$ B+6 probes wasnot entirely unexpected, since gel retardation assays indicatedthat an intact NF- ${kappa}$ B site was sufficient to allow efficient complexformation on -933/-773 probes in vitro. These findings are reminiscentof the transcriptionally active enhanceosome-like complex thatassembles on the nur77 promoter, which depends on intact DNAbinding of MEF2D but does not require the DNA-binding activityof the synergistically acting transcription factor NFATp. Inthis context, NFATp acted as a coactivator for MEF2D (69). AlthoughDNA contact with AP-1 and C/EBPß was not absolutelyrequired for in vitro formation of the complex on the -933/-773bfl-1 probe when NF- ${kappa}$ B was present, our transactivation assayswith the mAP-1 mutant reporter suggest that in a natural context,cooperative protein-protein and protein-DNA interactions involvingthe NF- ${kappa}$ B and AP-1 binding sites are likely to be involved inefficient formation of the complex and optimal activation ofbfl-1 gene expression in vivo.

ChIP analysis of endogenous factor recruitment to the bfl-1regulatory region in intact chromatin demonstrated inductionof c-Rel, c-Jun, and C/EBPß binding to the bfl-1 locusupon T-cell activation, in agreement with our in vitro data.These studies also revealed selective recruitment of coactivatorsp300 and TAFII250 and a sharp induction in histone acetylationat the bfl-1 locus. While further studies are needed to establishthe precise timing of factor recruitment to the bfl-1 regulatoryregion, it is intriguing to note the transient binding of TAFII250,BRG-1 (SWI/SNF), TBP, and TFIIB. Whereas binding of these factorswas sharply elevated at the 2-h time point, their detectionhad declined to basal levels by 18 h. Although it is not possibleat this time to distinguish between departure of the factorsand epitope masking at the later time point, these data maysuggest interesting differences between coactivators TAFII250and p300 in regulating factor recruitment, histone modification,and bfl-1 gene expression in response to cell activation. Together,our data highlight interplay between the factors that regulatebfl-1 through specific protein-protein and protein-DNA interactionsand support a model in which NF- ${kappa}$ B promotes assembly of an enhanceosome-likecomplex to acetylate histones and induce chromatin remodelingto promote bfl-1 expression upon lymphoid cell activation (Fig.7).

Our finding that bfl-1 gene expression may involve an NF- ${kappa}$ B-dependentenhanceosome-like complex is physiologically relevant, becauseit is consistent with and provides an explanation for the failureof B and T cells from c-Rel knockout mice to express endogenousbfl-1/a1 upon cell activation (23). In this regard, similarbinding sites for AP-1, C/EBPß, and HMG-I are foundin close proximity to the unique NF- ${kappa}$ B recognition motif in the5' regulatory region of the mouse a1 gene. Our findings arealso consistent with prior studies from our group and othersshowing that stable expression of a dominant-negative form ofI ${kappa}$ B ${alpha}$ repressed endogenous bfl-1/a1 transcription (15, 31, 44,63, 72). Our in vitro and in vivo analyses indicating that themajor NF- ${kappa}$ B dimer that controls bfl-1 gene expression in activatedJurkat T cells is a c-Rel/p50 heterodimer agree with prior studiesindicating basal expression of endogenous bfl-1/a1 in mouseWEHI-231 B cells, in which endogenous nuclear NF- ${kappa}$ B complexesconsist of c-Rel/p50 heterodimers (72). However, it is possiblethat RelA might contribute to bfl-1 gene expression in othercell types or in response to particular stimuli. Although wedo not rule out the possibility that other transcription factorsbinding to sites that lie outside of the -933/-773 bfl-1 regulatoryregion may also participate in regulating bfl-1, our studiesrevealed a critical role for NF- ${kappa}$ B in this process.

The coordinated involvement of AP-1, C/EBPß, and NF- ${kappa}$ Bin bfl-1 gene expression implies an important role for Bfl-1in rescuing cells from apoptosis in the immune system. Indeed,pairwise combinations of these factors act synergistically withone another in regulating expression of many genes involvedin the acute phase of inflammation. Their combination is alsoresponsible for activation of the inflammatory cytokines interleukin1 (IL-1), IL-6, and TNF- ${alpha}$ (24, 30, 57, 70). Architectural factorHMGI-C was implicated in enhancing NF- ${kappa}$ B-mediated transcriptionand is rearranged in mesodermally derived tumors, such as B-and T-cell leukemias (4, 37). The strong induction of bfl-1transcripts by proinflammatory cytokines in endothelial, leukemic,and hemopoietic cells is consistent with a role in protectingthem from apoptotic extinction (14, 27, 34, 42). The observationthat Bfl-1/A1 gene expression is also sharply induced in matureB cells that have entered the long-lived pool highlights itspossible contribution to cell survival during negative selection(60). In this context, Bfl-1 might be viewed as a key determinantof cell fate for immature peripheral B cells.

Elevated bfl-1 gene expression was reported in several humancancers (28, 48). Moreover, cell treatment with the chemotherapeuticdrug etoposide elevated bfl-1 transcripts in HT1080 fibrosarcomacells, and this elevation was in turn responsible for theirresistance to the killing effects of this drug (49, 63). Ourfindings that bfl-1 gene regulation is a multicomponent affairsuggest that the engagement of signaling pathways leading toactivation of NF- ${kappa}$ B, AP-1, and C/EBPß may need to bemonitored to avoid desensitizing oncogenic cells to chemotherapeutictreatment. We have presented here the first evidence of an enhanceosome-likemechanism regulating expression of an antiapoptotic gene. Itis not unreasonable to propose that other cell death regulatorsmay be controlled in this manner, given the potentially devastatingconsequences associated with their inappropriate expression.Since NF- ${kappa}$ B regulates a number of antiapoptotic genes and isinvolved in several different enhanceosome complexes, NF- ${kappa}$ B-dependentenhanceosome control of other anti- or proapoptotic genes maycome to light. Future studies will help to elucidate how theirtranscriptional regulation is controlled in normal cells andin cancer cells.

ACKNOWLEDGMENTS

This work was supported by Public Health Service grants (CA83937and CA54999) from the National Cancer Institute and by a grantfrom The Charlotte Geyer Foundation to C.G. We acknowledge partialsupport from an NIH predoctoral training grant in biotechnology(GM08339 [L.E.]) and an NIH predoctoral training grant in biochemistryand molecular biology (GM08360 [L.E., L.L., and M.S.]).

We are grateful to K. Chada, T. Curran, D. Livingston, D. Reinberg,N. Rice, W. Wang, and D.E. Zhang for the kind gift of reagents.We are indebted to S. Lomvardas and D. Thanos for the ChIP protocoland to E. White for sharing her real-time PCR instrument. Wethank C. Abate-Shen, A. B. Rabson, and J. Suh for modified gelretardation assay procedures and M. Hampsey, A. B. Rabson, R.Steward, and members of the Gélinas laboratory for fruitfuldiscussions and suggestions during the course of this work.We also thank M. Hampsey, A. Rabson, and B. Rayet for helpfulcomments on the manuscript.

FOOTNOTES

* Corresponding author. Mailing address: CABM, 679 Hoes Lane, Piscataway, NJ 08854-5638. Phone: (732) 235-5035. Fax: (732) 235-5289. E-mail: gelinas{at}cabm.rutgers.edu.

Present address: Department of Pathology, Children's Hospital,Harvard Medical School, Boston, MA 02115.

REFERENCES

Top