Tumor Necrosis Factor-α Activation of NF-κB Requires the Phosphorylation of Ser-471 in the Transactivation Domain of c-Rel*

Activation of the transcription factor NF-κB is controlled at two levels in resting T cells: an initial activation induced by the triggering of the TcR·CD3 complex and a second phase controlled by paracrine- or autocrine-secreted TNFα. The initial phase is regulated by p65 (RelA), whereas the second one is mainly dependent on c-Rel. We describe here a mutant clone, D6, derived from the parental T lymphoblastic line Jurkat that fails to activate NF-κB upon TNFα stimulation. This clone had no alteration in tumor necrosis factor-α (TNFα) signaling pathways nor in IκBα, -β, or -ε expression and degradation. However, TNFα induced an exacerbated apoptotic response in this clone compared with Jurkat cells. This mutant clone showed a defect in the intermediate-late translocation of c-Rel to the nucleus promoted by TNFα stimulation, whereas early translocation is not affected. Activation or translocation of p65-containing complexes was not altered in this mutant clone. Sequencing of the c-Rel gene from this clone revealed a mutation of Ser-471 to Asn in the transactivation domain. The mutant S471N transactivation domain fused to the Gal4 DNA binding domain could not be activated by TNFα, unlike the wild type. Moreover, the overexpression of the mutant protein c-Rel S471N into Jurkat cells abolished TNFα-induced NF-κB activity, thus demonstrating that this mutation is responsible for the failure of TNFα stimulation of NF-κB. Moreover, extracts from TNFα-stimulated Jurkat cells phosphorylated in vitro recombinant wild type GST-c-Rel 464–481 but not the GST-c-Rel mutant. Thus, TNFα-induced phosphorylation of Ser-471 seems to be absolutely necessary for TNFα activation of c-Rel.

the most important factors involved in this process of T cell activation are those belonging to the NF-B family, which regulates several of the most important genes induced during T cell activation (for review see Ref. 5). NF-B is rapidly activated by the TcR⅐CD3 complex, but at later phases of T cell activation, autocrine-or paracrine-secreted TNF␣ takes control of NF-B activation. The initial phase of NF-B activation depends on p65 translocation, whereas the later TNF␣-dependent phase is controlled by c-Rel (1).
The NF-B family of transcription factors is composed of homo-and heterodimers of a family of proteins that includes the Dorsal gene of Drosophila and the mammalian genes nfb1, nfb2, c-rel, relA (p65), and relB (for review see Ref. 6). All members share a conserved 300-amino acid region in their N-terminal portion that includes the dimerization, nuclear localization, and DNA binding regions. c-Rel, RelB, and RelA also have C-terminal transactivation domains that strongly activate transcription from NF-B sites. NF-B is regulated at least in part by its subcellular localization. Thus, functional NF-B complexes are held in the cytoplasm of resting T cells in an inactive state complexed with members of the IB family. In response to different activators (which include TcR⅐CD3 and TNF␣), IB is phosphorylated by IB kinases and subsequently degraded, liberating the active NF-B complex, which translocates to the nucleus and activates transcription (for review see Ref. 7).
Recently a second level of regulation of NF-B activation that involves the direct activation of transcriptional-competent NF-B family members has been emerging. Thus, the catalytic subunit of protein kinase A was shown to be bound to inactive NF-B complexes, and upon IB degradation, this catalytic subunit phosphorylated p65, resulting in an enhanced transcription-promoting activity (8). Moreover, it was reported that TNF␣ treatment of cells resulted in phosphorylation of Ser-529 in the transactivation domain of p65, resulting in the activation of the transcriptional activity of the protein (9). The small GTP-binding protein Ras enhanced p65/RelA transcriptional activity through a pathway that required the stress-activated protein kinase p38 or a related kinase (10), although it was not demonstrated whether this kinase was directly involved in activating NF-B or a transcriptional co-activator. The activity of Ras as well as the atypical protein kinase C has been also shown to be essential for the transcriptional activity of p65/ RelA in endothelial cells (11). This activation relies in the phosphorylation of the N-terminal Rel homology domain and not on the C-terminal transactivation domain. Other downstream effectors of Ras such as Raf/mitogen-activated protein kinase kinase (MEK), other small GTPases, phosphatidylinositol 3-kinase (PI3K), and the stress-activated protein kinase pathway were not involved in this Ras and protein kinase C-dependent Rel homology domain p65 phosphorylation. On the other hand, the IL-1-induced NF-B activity has been recently found to be dependent on PI3K activity (12). The acti-vation of PI3K triggers a signaling cascade that leads to the specific phosphorylation of p65/RelA subunit. This phosphorylation enhances p65-mediated transcription without affecting IB degradation, nuclear translocation of NF-B, or the ability of NF-B to bind to DNA. Nevertheless, the specific site(s) of phosphorylation was not identified.
In this work we report that a mutation in Ser-471 to Asn within the transactivation domain of c-Rel results in the abrogation of TNF␣-induced NF-B activity in T cells. This indicates the existence of a second level of regulation of c-Rel activity by TNF␣ through modulation of the activity of the transactivation domain.

EXPERIMENTAL PROCEDURES
Cells-Jurkat cells were grown in Dulbecco's modified minimal essential medium (Life Technologies, Inc.) supplemented with 5% heatinactivated fetal bovine serum (Life Technologies, Inc.) and containing 100 g/ml streptomycin, 100 units/ml penicillin, 2 mM L-glutamine, plus nonessential amino acids at 37°C in 7% CO 2 . D6 Jurkat cells were obtained by subcloning Jurkat cells and screening of those clones for TNF␣ unresponsiveness.
Reagents-Recombinant human TNF␣ was purchased from Genzyme (Cambridge, MA). Phorbol myristate acetate (PMA) and calcium ionophore A23187 were purchased from Sigma. Cycloheximide was purchased from Roche Molecular Biochemicals. Casein was purchased from Sigma.
Plasmids-The pNF3TKLuc reporter plasmid contains a trimer of the NF-B binding motif of the H-2k gene upstream of the thymidine kinase minimal promoter and the luciferase reporter gene (13). The pNF3ConA Luc reporter contains three tandem repeats of the human immunodeficiency virus-1 long terminal repeat NF-B enhancer upstream of the conalbumin promoter and the luciferase reporter gene (kindly provided by Dr. J. Alcamí, Hospital 12 de Octubre, Madrid). Gal4 c-Rel transactivation domain 309 -588 wild type and 309 -588 S471N were made by cloning the corresponding PCR fragments into the XhoI-BglII site of the Gal4 c-Jun 1-166 plasmid, thus removing the c-Jun fragment. The templates for PCR reactions were pRc-hc-Rel (which consists of pRcCMV with c-Rel cDNA inserted in the HindIII-XbaI site) and pRc-hc-Rel S471N (site-directed mutagenesis of c-Rel nucleotide 1685 from G to A was accomplished using the Muta-Gene phagemid mutagenesis kit (Bio-Rad) with pRc-hc-Rel as template).
Apoptosis Measurement-Jurkat or D6 cells (5 ϫ 10 5 cells) were stimulated with or without TNF␣ (10 ng/ml) in the presence of 1 g/ml cycloheximide for 24 h before washing twice with ice-cold phosphatebuffered saline. Cells were then incubated with buffer staining (0.1% sodium citrate, 0.3% Nonidet P-40 and 0.05% propidium iodide (Sigma), and cell cycle analysis was immediately performed in a FACScalibur (Becton Dickinson, San Jose, CA) flow cytometer. Apoptosis was recorded as the percentage of cells with haplo-diploid content of DNA.
Electrophoretic Mobility Shift Assay-Preparation of nuclear extracts was described in detail elsewhere (14). The binding reaction consisted of 5 g of extracted nuclear protein and 5 g of poly(dI-dC) (Roche Molecular Biochemicals) in a reaction volume of 10 l containing 6 mM MgCl 2 . This mixture was then incubated at room temperature for 10 min, after which 2 l (50,000 cpm) of the NF-B consensus oligonucleotide (Promega, Madison, WI), end-labeled with [␥-32 P]ATP (specific activity ϭ 3,000 Ci/mmol; Amersham Pharmacia Biotech), was added. A control reaction mixture was always included in which a 100ϫ molar excess of non-radioactive NF-B oligonucleotide was added to verify the specificity of the binding reaction. After incubation in an ice bath for 15 min, the reaction mixtures were run on 5% PAGE. After drying, the gels were subjected to autoradiography.
Western Blots-Whole cell extracts (WCE) were made using TNT buffer as the lysis buffer (20 mM Tris-HCl, pH 7.6, 200 mM NaCl, 1% Triton X-100) supplemented with protease inhibitors (2 g/ml aproti-nin, 2 g/ml pepstatin, 2 g/ml leupeptin, 0.1 mM benzamidine, and 0.5 mM phenylmethylsulfonyl fluoride) and phosphatase inhibitors (5 mM NaF, 100 M sodium orthovanadate). Nuclear extracts were prepared as described in the preceding section. 10% of whole cell extracts or 5 g of nuclear extracts were run on a 10% SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Immobilon, Amersham Pharmacia Biotech). Rabbit anti-human p65, c-Rel, IB␣, IB␤, or IB⑀, as appropriate, was used as the first antibody, and goat anti-rabbit IgG peroxidase was used as the secondary antibody. The enhanced chemiluminescent (ECL) developing kit (Amersham Pharmacia Biotech) was used to identify the relevant band(s).
Immunoprecipitation-WCE made from 10 7 cells were immunoprecipitated in TNT buffer with 1 l of serum. Precipitates were collected on protein A-Sepharose (Amersham Pharmacia Biotech) and separated in a 10% SDS-PAGE and subsequently transferred to a polyvinylidene difluoride membrane. Membranes were analyzed by Western blot.
cDNA and Genomic DNA Sequencing-Total RNA from 10 7 Jurkat or D6 cells was extracted using Trizol (Life Technologies, Inc.) following the manufacturer's directions. 1 g of each sample was used for cDNA synthesis using the GeneAmp RNA PCR core kit (Perkin-Elmer) following kit instructions. The 5Ј sense primer for c-Rel cDNA amplification was CTCCTGACTGACTGACTGACTG, and the corresponding 3Ј antisense primer was ATGTCCAAAGTTGTATGC. Both primers were also used for sequencing in an automatic Perkin-Elmer sequencer. For PCR amplification of c-Rel exon VII (comprising the whole transactivation domain), the 3Ј antisense primer was the same as above, whereas the 5Ј sense primer was CACTTCCTTCTAATTCGC. The latter primer was also used for sequencing.
Transfection and Stimulation of Jurkat T Cells-Jurkat cells were washed once and resuspended at 10 6 cells/ml in OPTIMEM medium (Life Technologies, Inc.). Cells were transfected with the Lipo-fectAMINE Plus reagent (Life Technologies, Inc.) in the preparation of the LipofectAMINE Plus-plasmid mixtures, in accordance with the manufacturer's instructions. The mixtures were incubated at 37°C in a 7% CO 2 incubator for 3 h before washing with fresh Dulbecco's modified minimal essential medium plus 5% fetal bovine serum and incubated further for 18 h. The cells were then washed once with, and resuspended at the same concentration in, Dulbecco's modified minimal essential medium plus 5% fetal bovine serum. Culture medium with or without 10 ng/ml of TNF␣ or PMA (10 ng/ml) plus calcium ionophore (1 M) was added to duplicate wells containing 0.5 ml of these cell suspensions, which were then incubated under the same conditions for 6 h. The cells were lysed with passive cell culture lysis reagent (Promega), microcentrifuged at full speed for 5 min at 4°C, and 20 l of each supernatant was used to determine luciferase activity in a Monolight 2010 luminometer (Analytical Luminescence Laboratory). The results were expressed as increases in luminescence relative to the value obtained with the non-stimulated control after normalization with respect to protein concentration determined by the bicinchoninic acid spectrophotometric method (Pierce).
Solid-phase in Vitro Phosphorylation Assay-c-Rel transactivation domain from position 464 to 481 (using as template pRc-hc-Rel wild type or pRc-hc-Rel S471N) was cloned into the BamHI-EcoRI site of plasmid pGEX2T (Amersham Pharmacia Biotech) to express recombinant GST-c-Rel fusion protein. These recombinant proteins were purified from Escherichia coli-induced cultures according to the manufacturer's instructions. 25 l of GSH-agarose-GST-c-Rel were used as substrate of an in vitro phosphorylation reaction in which whole cell extracts from non-stimulated or stimulated Jurkat cells were assayed. WCE were made from 10 6 Jurkat cells in 25 l, as described previously. The reaction mixture contained 20 mM Hepes, pH 7.6, 20 mM MgCl 2 , 20 mM ␤-glycerophosphate, 20 M ATP, and 1 Ci of [ 32 P]ATP (specific activity 3,000 Ci/mol). After 20 min at 30°C, the reaction was terminated by washing with TNT buffer. Phosphorylated protein was boiled in 25 l of Laemmli sample buffer and resolved in 10% SDS-PAGE, followed by autoradiography. For JNK and ␤-adrenergic receptor kinase activity assays, the corresponding kinase was immunoprecipitated as described from WCE, and the immunoprecipitate was subjected to an in vitro phosphorylation assay using GST-c-Jun (1-147) fusion protein and casein as substrates, respectively.
Orthophosphate Labeling-Jurkat cells were incubated with phosphate-free medium for 16 h before 3 mCi of [ 32 P]orthophosphate per million cells were added. After 6 h of incubation, replica plates were left untreated or stimulated with 10 ng/ml TNF␣ for 15 min. After that period of time, WCE were made as described, and resulting extracts were immunoprecipitated with anti-c-Rel antiserum 265, as described previously. Immunoprecipitates were resolved in a 10% SDS-PAGE and developed by autoradiography.

Characterization of TNF␣ Unresponsiveness of Jurkat Clone
D6 -TNF␣ is one of the most potent activators of NF-B in T cells (for review, see Ref. 7). Furthermore, neutralizing antibodies against TNF␣ block NF-B activation on primary T cells (1). Thus, as expected, TNF␣-induced NF-B reporter activity in Jurkat T cells (Fig. 1A), independent of the origin of the NF-B sites of the reporter plasmids (human immunodeficiency virus-1 long terminal repeat for NF3ConA or H-2k gene for NF3TK). We used these reporter assays to screen spontaneous mutant Jurkat T cell clones (obtained by limiting dilution of a parental Jurkat culture) defective in NF-B response to TNF␣. We were able to identify one clone, named D6, that had very little, if any, NF-B reporter inducibility when stimulated with TNF␣ (Fig. 1A). This failure of TNF␣ responsiveness was not due to deficient expression of TNF receptors, since clone D6 expressed the TNF receptors at levels comparable with Jurkat cells (data not shown). Moreover, stimulation with a potent T cell mitogenic stimulus, such as PMA plus calcium ionophore, led to strong NF-B reporter activity in clone D6 cells (Fig. 1A), indicating the specificity of the defect to TNF␣ stimulation.
We tested whether this lack of responsiveness to TNF␣ was specific for NF-B activation or whether it also extended to other TNF␣-mediated responses. The D6 clone showed an enhanced apoptotic response to TNF␣, as compared with parental Jurkat cells (Fig. 1C), indirectly indicative of a specific failure of TNF␣ to activate NF-B. We next tested whether this lack of responsiveness was due to a defect in the TNF␣ signaling mechanisms, not only in the NF-B pathway. Since AP-1 is another transcription factor activated by TNF␣, we tested the activity of JNK using in vitro phosphorylation assays with GST-c-Jun as a substrate. TNF␣ activated JNK activity in both Jurkat and clone D6 cells. As a control we immunoprecipitated ␤-adrenergic receptor kinase (GRK2) from the same extracts and tested its kinase activity using casein as substrate. We used ␤-adrenergic receptor kinase (GRK2) as a control because this kinase is highly expressed in T cells, but it is activated by G-protein-coupled receptors, not by TNF␣ receptors (15). As expected, GRK2 activity was not induced by TNF␣, whereas both Jurkat and clone D6 cells showed a similar basal activity (Fig. 1B). Therefore, although the TNF␣ induction of NF-B was abnormal in clone D6, other TNF␣-mediated pathways appeared normal.
Next, we tested whether this lack of activity resulted from the inability of TNF␣ to promote nuclear translocation of NF-B. As shown in Fig. 2A, gel-shift assays with a NF-B consensus oligonucleotide using nuclear extracts from TNF␣-treated Jurkat cells showed the presence of two NF-B binding complexes, labeled I and II, peaking from 30 min to 4 h and still present at 15 h. Surprisingly, there was also induction of NF-B binding complexes in clone D6 cells by TNF␣ stimulation, although this induction was less sustained over time than that observed in Jurkat cells, peaking at 30 min but then returning progressively to basal levels ( Fig. 2A). Furthermore, the kinetics of nuclear translocation of NF-B promoted by PMA plus ionophore treatment was similar in both Jurkat and clone D6 cells (Fig. 2B). To analyze the composition of the two complexes I and II, we used supershift assays where nuclear extracts were treated with antibodies directed against different NF-B family members before the addition of the probe. Both complexes I and II disappeared when the extracts were treated with anti-p50 antibody and also when anti-p65 and anti-c-Rel were used together (data not shown), indicating that these complexes were made from heterodimers of p50/p65 and p50/ c-Rel. Incubation with anti-p65 antibody alone revealed the presence of unshifted c-Rel-containing complexes (Fig. 2C,  lanes 2, 5, 8, and 11). Conversely, incubation with anti-c-Rel antibody revealed the presence of unshifted p65 complexes (Fig. 2C, lanes 3, 6, 9, and 12). Thus, complex I contained p65/p50 heterodimers, whereas complex II was composed of heterodimers of c-Rel/p50. Interestingly, this experiment showed a difference in the relative composition of NF-B complexes between Jurkat and clone D6. In control Jurkat cells, the relative amount of c-Rel in the DNA binding complexes (complex II) increased with time of TNF␣ treatment (Fig. 2C, compare the unshifted band in lanes 2 and 5). However, this increase was not seen in D6 cells (Fig. 2C, lanes 8 and 11). Both immunoprecipitates were subjected to an immunocomplex phosphorylation assay to test the activity of both kinases using casein and recombinant GST-c-Jun as substrates, respectively. C, induction of apoptosis. Jurkat or clone D6 cells were stimulated for 24 h with TNF␣ (10 ng/ml) in the presence of cycloheximide, and apoptosis was recorded by flow cytometry as the percentage of cells with haplo-diploid content of DNA.
We then directly analyzed the presence of c-Rel and p65 proteins in nuclear extracts from TNF␣-stimulated cells by Western blot. We found that in both Jurkat and clone D6 cells, TNF␣ induced translocation of p65 with the same kinetics (Fig.  3A); maximum translocation was observed at 30 min, with return to basal levels at 4 h. c-Rel was also translocated to the nucleus at 30 min but reached a maximum at 4 h and maintained this plateau at least for 15 h after stimulation in wildtype Jurkat cells. In clone D6 cells, this translocation also started at 30 min, but it was not sustained in time, returning to basal levels by 4 h. These results are indicative of a failure in clone D6 to correctly maintain activated c-Rel after TNF␣ stimulation. As occurred with the DNA binding complexes, the kinetics of PMA plus ionophore-induced translocation of p65 and c-Rel was similar in both cell types (Fig. 3B).
IkB Degradation Is Not Affected in the Jurkat D6 Clone-The previous results indicated a defect in c-Rel activation by TNF␣ in clone D6. One possibility was a differential association of c-Rel NF-B complexes with different members of the IB family, in such a way that c-Rel complexes in clone D6 cells could not be activated by TNF␣ as they were in wild-type Jurkat cells. To test this hypothesis, we immunoprecipitated IB␣, IB␤, and IB⑀ from non-stimulated Jurkat and clone D6 cells and blotted with anti-p65 and anti-c-Rel antibodies to assess the composition of the complexes bound to IB␣, IB␤, and IB⑀, respectively. There were no significant differences between Jurkat cells and clone D6. Both IB␤ and IB⑀ were equally bound to both c-Rel and p65, while IB␣ was bound predominantly to p65 (Fig. 4A). Since these experiments did not rule out the possibility that the kinetics of IB degradation were different, we stimulated Jurkat and clone D6 cells with TNF␣ for up to 120 min in the presence of cycloheximide to analyze the degradation of IB proteins (Fig. 4B). IB␣ was degraded after 30 min of TNF␣ stimulation in both Jurkat and clone D6 cells, whereas IB␤ and IB⑀ were not affected at all. Once again, there were no significant differences between clone D6 and Jurkat with respect to IB degradation kinetics upon TNF␣ stimulation. Indirectly, those results also indicate that no significant differences in the amount of c-Rel, p65, or IB proteins exist between both cell types. This has been confirmed by Western blots of cytoplasmic extracts (data not shown). Thus, the mechanisms leading to IB degradation remained unaltered in clone D6, explaining the observed TNF␣-induced translocation of NF-B complexes but not the inhibition of NF-B-dependent reporter activity.
D6 Cells Expressed a Mutated Form of c-Rel-Another possibility that could account for the defect in c-Rel activation by TNF␣ in clone D6 cells was a defect in c-Rel itself. To test this hypothesis we sequenced the entire c-Rel gene in Jurkat and clone D6. We isolated total RNA from both cell types and used it as template for a reverse transcription-PCR reaction. Primers specific for c-Rel were used to amplify c-Rel cDNA, and the same primers were used for sequencing. Interestingly, we found a point mutation at position 1685 (G to A change) in c-Rel cDNA obtained from D6 cells. This resulted in a change of serine 471 to asparagine within the C-terminal region of c-Rel, which corresponds to the c-Rel transactivation domain (Fig.  5A). To confirm the previous results, we examined the c-Rel genomic DNA. Since the whole transactivation domain constitutes the last exon (exon VII) of the c-Rel gene (16), we used a primer specific for that exon to directly sequence from genomic DNA isolated from clone D6 cells. The result confirmed the point mutation that resulted in the change of Ser-471 to Asn found in clone D6 c-Rel cDNA. Moreover, genomic DNA form D6 cells contained 100% A at position 1685, indicating that both alleles were mutant.
The Ser-471 3 Asn c-Rel Transactivation Domain Is Not Activated by TNF␣-We hypothesized that the S471N mutation could be responsible of the defect in TNF␣-induced activation of c-Rel in the D6 clone. To confirm this hypothesis, we tested the transactivating capabilities of wild type versus the S471N mutant using the Gal4 "one-hybrid" technique for the study of transcription activation regions. To this end, we cloned the region of interest (c-Rel amino acids 309 to 588) downstream of the Gal4 DNA binding domain (Gal4-DBD), transfected the construct into Jurkat cells, and assayed for Gal4 directed reporter activity. This system has the advantage that the Gal4-transactivator fusion proteins are exclusively nuclear and are regulated independently of IB. Both wild type and S471N fusions showed substantial basal reporter activity by themselves, without any exogenous stimuli, whereas Gal4 DBD or Gal4 DBD c-Rel 309 -318 (which comprises only the first 10 amino acids of c-Rel transactivation domain) were inactive (Fig. 5B). However, only the wild type but not the S471N mutant further activated transcription of the reporter when the transfected cells were stimulated with TNF␣ (Fig.  5C). In contrast, PMA ϩ ionophore stimulated both the S471N mutant construct as well as the wild type. These results perfectly correlate with the defect observed in clone D6 upon TNF␣ stimulation using NF-B luciferase reporter assays and suggested that this defect was due to the mutant c-Rel.
To further corroborate this, we transfected the Gal4 c-Rel 309 -588 wild type and mutant S471N into clone D6 and assayed for TNF␣ or PMA plus ionophore stimulation of reporter activity. Both wild-type and mutant constructs induced a similar basal reporter activity, as in Jurkat wild-type cells. Interestingly, TNF␣ stimulation in clone D6 cells transfected with the wild-type construction was similar as the induction obtained in Jurkat cells, whereas Jurkat wild-type cells transfected with S471N mutant construction showed no induction upon TNF␣ stimulation (Fig. 5D). This result clearly indicated that the signaling mechanisms involved in c-Rel activation by TNF␣ were unaffected in clone D6, pointing at Ser-471 as a critical residue in c-Rel, responsible for TNF␣ activation. As expected, PMA plus ionophore stimulated both the wild-type and the S471N mutant constructions.
Overexpression of S471N Mutant Inhibited TNF␣ Stimulation of NF-B Activity-Since mutant S471N transactivation domain was not activated by TNF␣ stimulation, we tested whether this mutation was responsible for the phenotype observed in D6 cells. For this, we co-transfected c-Rel mutant S471N into wild-type Jurkat cells along with a NF-B luciferase reporter plasmid and assayed the activity induced by TNF␣ stimulation. As observed in Fig. 6, NF-B basal activity was not significantly altered by both mutant or wild-type overexpression. Interestingly, TNF␣ stimulation did not further increase NF-B activity in wild-type cells transfected with c-Rel mutant S471N, in contrast to empty vector transfected cells. Therefore, overexpression of c-Rel mutant S471N could reproduce the phenotype observed in D6 cells. Moreover, overexpression of wild-type c-Rel in D6 cells could significantly, although incompletely, increase TNF␣-induced NF-B activity in D6 cells, thus partially recovering the wild-type phenotype (Fig. 6). The lack of complete recovery could indicate that endogenous S471N protein was acting as a dominant negative mutant.
TNF␣-induced Phosphorylation of c-Rel Transactivation Domain-As previous work on v-Rel and avian c-Rel (5,8,9) showed that c-Rel is phosphorylated, we hypothesized that a phosphorylation event in Ser-471 was implicated in c-Rel activation by TNF␣. Metabolic labeling of Jurkat cells with [ 32 P]orthophosphate showed that c-Rel is a phosphoprotein and that its phosphorylation was increased by TNF␣ treatment (Fig. 7A). To test if the induced phosphorylation produced by TNF␣ specifically takes place in the c-Rel transactivation domain, we used GST-c-Rel 421-588 (which includes all the c-Rel transactivation sequences) 2 wild type and mutant S471N fusion proteins as substrates for a solid-phase kinase phosphorylation assay. Whole cell extracts from Jurkat cells and clone D6 cells stimulated or not with TNF␣ or PMA plus ionophore for 30 min were used for this phosphorylation assay. Both the wild type and mutant S471N c-Rel fusion proteins were strongly phosphorylated by one or more kinase activities present in unstimulated Jurkat cells. This phosphorylation was further increased in extracts from TNF␣ or PMA plus ionophore-treated cells in both wild type and mutant S471N (data not shown), demonstrating that TNF␣ (as well as PMA plus ionophore) stimulation promotes the phosphorylation of c-Rel 2 A. G. Martin and M. Fresno, manuscript in preparation.

FIG. 4. Analysis of IB expression and function in Jurkat and clone D6 cells. A, composition of NF-B⅐IB complexes in Jurkat and D6 cells. Whole cell extracts from non-stimulated cultures (10 7 cells)
were immunoprecipitated with anti-IB␣, anti-IB␤, or anti-IB⑀ antibody, as indicated. Immunoprecipitates (IP) were separated in a 10% PAGE, transferred to polyvinylidene difluoride, and incubated with p65-and c-Rel-specific antibodies. WB, Western blot. B, effects of TNF␣ on the degradation of IB␣, IB␤, and IB⑀ in Jurkat and D6 cells. Cell cultures were pre-incubated with cycloheximide (10 g/ml) for 15 min and stimulated with TNF␣ (10 ng/ml) for the indicated times. As a control, an aliquot of the culture was left untreated with cycloheximide (CHX; lane C). After stimulation, whole cell extracts were prepared using TNT as lysis buffer; 10% of those extracts were electrophoresed in 10% PAGE, transferred, and incubated with the indicated antibodies. transactivation domain. This suggested that Ser-471 would not be the only residue that is phosphorylated upon TNF␣ activation, although functional data reveal its critical role in TNF␣ activation of c-Rel. Thus, we then used a smaller construct comprising residues 464 to 481 to eliminate the background produced by other activation-dependent sites. The GST-c-Rel 464 -481 recombinant protein was used as a substrate for a similar solid phase in vitro phosphorylation assay. As shown in Fig. 7B, the recombinant wild-type protein was slightly phosphorylated by the extract from unstimulated cells, and this phosphorylation was strongly increased by extracts obtained after TNF␣ or PMA plus ionophore stimulation. In contrast, phosphorylation in the mutant Ser-471 c-Rel was not increased upon stimulation of cells, indicating that the Ser at position 471 is involved in the specific phosphorylation of c-Rel in a TNF␣-dependent manner. DISCUSSION NF-B activation in resting human T cells upon TcR activation has been shown to be controlled in a biphasic manner (1). Thus, nuclear translocation of p65/p50 and c-Rel/p50 heterodimers triggered by the TcR⅐CD3 complex stimulation was observed at early times, whereas at intermediate-late times the active NF-B complexes are mainly composed of c-Rel/p50 heterodimers. Interestingly, those intermediate-late events are controlled by autocrine or paracrine secretion of TNF␣ (1). Therefore, c-Rel emerges as the main NF-B family member stimulated by TNF␣ in the context of physiologic activation of resting T cells. This biphasic model of NF-B activation is in perfect agreement with the results found in the c-Rel knock-out mouse, which are deficient in primary T cell activation (17), (18). Those c-Rel null lymphocytes cannot proliferate to mitogenic stimuli, but they can be rescued by cytokines such as IL-2 to acquire the effector phenotype, thus demonstrating that c-Rel regulates the late phase of T cell activation. We report here the requirement of c-Rel for sustained NF-B activity Jurkat cells (10 6 /well) were cotransfected with 10 ng of Gal4 DBD fusion constructs and 100 ng of 5XGal4 luciferase reporter and cultured with medium in basal condition (white bars) or stimulated with TNF␣ (black bars) or PMA plus ionophore (gray bars). Reporter activity is expressed as fold induction above control. after TNF␣ stimulation. Thus, we have found a mutant Jurkat T cell clone, termed clone D6, that is defective in stimulated NF-B-dependent promoter activity when cells were activated by TNF␣ but not by PMA plus ionophore (Fig. 1A). This defect, however, does not involve down-regulation of TNF receptors or intracellular signaling, since it shows a normal JNK activation in response to TNF␣ stimulation (Fig. 1B). Furthermore, D6 cells showed an enhanced response to TNF␣-induced apoptosis, which restricted the defect to NF-B activation and correlates with the anti-apoptotic function described for NF-B (19,20). Moreover, as IB␣ is degraded and c-Rel is initially translocated upon TNF␣ stimulation with the same efficiency in Jurkat and clone D6, we postulated an intrinsic defect in c-Rel itself. Sequencing of the c-Rel gene present in clone D6 was evidence that clone D6 expressed a mutant version of c-Rel, characterized by a change of Ser residue at position 471 to an Asn residue (Fig. 5A). This position relies within the previously described transactivation domain of c-Rel (21,22), suggesting that this mutant version of c-Rel may be unable to promote TNF␣-dependent transcription. The experiments with Gal4DBD-c-Rel fusion proteins demonstrated that this mutation is indeed responsible for the TNF␣-enhanced c-Rel transcriptional activity. Moreover, the identical response of wild type Gal4-c-Rel fusion construct transfected either into mutant D6 or wild type Jurkat cells further confirmed that only c-Rel activation was impaired in clone D6. Furthermore, transfection of c-Rel wild-type into D6 cells partially recovered the wild-type phenotype, whereas overexpression of the c-Rel mutant S471N in Jurkat cells completely abolished TNF␣-induced NFB activity. Thus, this alteration in c-Rel was responsible for the lack of TNF␣ response observed in clone D6.
Although DNA binding of NF-B complexes was detected in the nucleus in clone D6 as well as in Jurkat T cells upon stimulation by TNF␣ ( Fig. 2A), there were striking differences related to the composition of the active NF-B complexes bound to the NF-B probe. In wild type cells, the NF-B complexes were composed of either p65/p50 or c-Rel/p50 heterodimers as revealed by anti-p65 plus anti-c-Rel antibody pretreatment of the extracts. Both complexes were detected in the nucleus as soon as 30 min after TNF␣ stimulation, but at late times (4 h) the factors bound to the NF-B probe were composed mainly of c-Rel in wild type Jurkat cells (Fig. 2C). The induction by TNF␣ of this c-Rel-containing complex at 4 h post-stimulation was absent in clone D6, indicating a failure of the later c-Rel induction upon TNF␣ stimulation. Once again, PMA plus ionophore stimulation triggered the activation of similar NF-B binding complexes in the nucleus of both Jurkat and clone D6 cells (Fig.  2B), restricting again the defect to TNF␣ stimulation. These results fit the biphasic model of NF-B activation by TNF␣ in T cells (1), in which there was an immediate nuclear translocation of p65 and c-Rel. p65 peaked immediately (30 min) but returned to basal levels after 4 h; in contrast, c-Rel was also induced by 30 min but was increasingly detected in the nucleus, reaching a plateau at 4 h that was maintained at least 15 h after stimulation. This progressive nuclear increase of c-Rel was absent in clone D6, whereas early p65 induction was similar than in Jurkat cells.
It has been described that different members of the IB family can bind to NF-B dimers and how these associations affect the NF-B functionality (23). However, the failure to retain c-Rel in the nucleus observed in clone D6, once activated with TNF␣, could not be attributed to a differential association of c-Rel with several IB family members in Jurkat and clone D6 cells. In agreement with previous reports (23), we found both p65 and c-Rel predominantly bound to IB␤ and IB⑀; however, immunocomplexes with IB␣ were also detected (Fig.  4A). Furthermore, IB␣ was degraded with the same kinetics in Jurkat and clone D6, with no degradation of IB␤ and IB⑀ (Fig. 4B). Those results are intriguing considering that only a c-Rel point mutation seems to be responsible for TNF␣ unresponsiveness. However, they can be explained considering that c-Rel but not p65 gene transcription requires a TNF␣-dependent NF-B activation (1). Altogether these results suggest that pre-existing c-Rel was translocated to the nucleus, but since it is inactive, it was degraded or removed from the nucleus. Because NF-B-dependent transcription is strongly depressed in the mutant, no new c-Rel is synthesized, so the overall result is that NF-B-dependent response fades away. This supports the model where TNF␣-induced NF-B activity, in the context of T cell activation, mainly relies in c-Rel activation. This activation is controlled by an autocrine loop: TNF␣ activation of c-Rel results in the transactivation of its own gene, further increasing NF-B-dependent transcription. Furthermore, our and S471N mutant (lanes 4 -6) proteins were used as substrates for in vitro solid-phase phosphorylation assays using whole cell extracts from non-stimulated cells or cells stimulated with TNF␣ or PMA plus ionophore (Io) for 30 min, as indicated. As controls, GST-c-Rel transactivation domain proteins were incubated in the absence of cellular extract, and GST alone was incubated with extract from TNF␣-stimulated cells; in both cases, the result was no band (not shown). The intensity of the bands was determined by densitometry of the film, and the optical density (O.D.) units are plotted. results point to c-Rel as a key mediator of TNF␣-induced NF-B-dependent activity. Thus, an inhibition of c-Rel abolishes this loop and results in a drastic overall effect of NF-B-dependent transcription and, consequently, in function (i.e. increased apoptosis). NF-B activation by TNF␣ involves the induced degradation of IB␣ through the activation of the IB kinase (24 -27). However, a second level of regulation of the transcriptional activity has been described for other NF-B family members, for example p65 (9,12). This second level of regulation of p65 activity involves the phosphorylation of Ser-529 residue within the transactivation domain upon TNF␣ stimulation, which enhances its transcriptional abilities. The drastic functional effect of c-Rel (S471N) mutation suggests that c-Rel is similarly subjected to a dual regulation by TNF␣. Thus, c-Rel activation at a first level involves IB degradation and subsequent translocation to the nucleus, whereas a second level involves the phosphorylation of the transactivation domain, as described here. It is noteworthy that the Ser residue at position 471 is strictly necessary for TNF␣ activation of c-Rel (Fig. 5A) but dispensable for PMA plus ionophore activation, as D6 cells respond to these stimuli as well as Jurkat cells. Similarly, the Gal4DBD-c-Rel 309 -588 is equally activated by PMA plus ionophore regardless of whether it carries the mutation, indicating that either the PMA plus ionophore-induced activation of c-Rel relies on the phosphorylation of other residues (Fig. 5C) or that the phosphorylation defect is compensated with other signals. The fact that PMA plus ionophore induced a stronger phosphorylation of the wild type molecule will suggest the first possibility.
The kinase involved in this TNF␣-induced phosphorylation of Ser-471 is presently unclear. Several NF-B family members, including IB␣, p50, p65, and c-Rel, have been shown to be phosphorylated, indicating a role of phosphorylation in functional activity. Thus, nuclear import of Drosophila protein Dorsal is regulated by phosphorylation (28). Moreover, recent results demonstrated that p65 transcriptional activity is augmented by a TNF␣-induced phosphorylation of Ser-529 (9). Phosphorylation of p65 also regulates its interaction with transcriptional machinery (8). On the other hand, the Ras and protein kinase C-dependent phosphorylation of p65 in the N-terminal Rel homology domain was found to be essential for its transcriptional activity (11). Thus, protein kinase C may have a dual role in NF-B activation, as it has been also demonstrated to be a IB kinase inducer (29). A Ras-dependent mechanism for activation of p65 that requires p38 MAPK activity independent of PI3K activation has been also described (10). In contrast, the p65 phosphorylation and subsequent activation induced by IL-1 has been recently shown to be dependent on PI3K activation (12). Moreover, the Ser/Thr kinase Akt (also known as protein kinase B), activated through lipid products of PI3K enzymatic activity, has been implicated in the regulation of NF-B in Jurkat cells (30). However, it is unlikely that PI3K directly phosphorylates p65, since Akt stimulates IB kinase activity and subsequent degradation of IB␣, so it is not the PI3K-stimulated activity that directly phosphorylates p65. On the other hand, c-Rel was demonstrated to be specifically phosphorylated in Ser residues (5). The MEKK1/JNK1 pathway has been shown to be implicated in NF-B activity, and JNK1 has been found to be associated to c-Rel in vivo (31); however, JNK1 did not phosphorylated c-Rel in vitro, indicating a different kind of regulation and discarding JNK1 as a direct activating kinase.
We have shown that c-Rel is already phosphorylated in Jurkat cells, but TNF␣ further induces its phosphorylation (Fig.  7A). Solid phase assays using GST-c-Rel 421-588 as the sub-strate for phosphorylation by WCE demonstrated a strong basal phosphorylating activity in non-stimulated cells. However, phosphorylation of c-Rel was further increased after TNF␣ treatment. When a shorter recombinant protein spanning the residues 464 to 481 was used, only very low levels of basal phosphorylation could be detected, but a strong induced phosphorylation upon TNF␣ or PMA plus ionophore stimulation of the wild type protein was detected. This TNF␣-induced phosphorylation did not take place when the mutant protein was used, thus suggesting an essential role of Ser-471 phosphorylation for TNF␣-mediated activation of c-Rel (Fig. 7B). Although Ser-471 could be phosphorylated, our results do not exclude completely the possibility that it could be necessary for the binding of a specific kinase without being itself phosphorylated. PMA plus ionophore stimulation also induced a strong phosphorylation of the wild-type 464 -481 fusion protein, suggesting that it was also able to activate the same or a different kinase capable of phosphorylating Ser-471. However, as mentioned above, the PMA plus ionophore-induced phosphorylation does not take place in mutant D6 c-Rel, albeit c-Rel was fully active in these cells. Although Ser-471 is essential for TNF␣induced c-Rel activity and seems to be phosphorylated, we have detected some TNF␣-induced phosphorylation in the transactivation domain of the S471N mutant protein. This result suggests that TNF␣ activates one or several kinases that can phosphorylate c-Rel transactivation domain in vitro at different positions, not only in Ser-471. We have not succeeded in the identification of the kinase(s) involved in c-Rel phosphorylation by "in-gel" kinase assays (data not shown), which suggests that the kinase involved is not active in this kind of experiment.
On the other hand, it has been described that c-Rel and v-Rel interact with the basal transcriptional machinery factors TATA-binding protein (TBP) and transcription factor IIB through their transactivation domains (32). However, other reports show that TBP interaction relies exclusively on the first 50 amino acids of c-Rel, within the Rel homology domain (33). Mutation S471N abolishes a phosphorylation site in the transactivation domain that may be necessary for the interaction between co-factors required for TNF␣ activation of c-Rel. It will be of interest to test whether Ser-471 is implicated in the interaction of c-Rel with the transcriptional machinery to understand the mechanism of TNF␣ activation of c-Rel. Studies are in progress to elucidate this hypothesis.