Spatial Redox Regulation of a Critical Cysteine Residue of NF-κB in Vivo

Reduction-oxidation (redox) regulation has been implicated in the activation of the transcription factor NF-κB. However, the significance and mechanism of the redox regulation remain elusive, mainly due to the technical limitations caused by rapid proton transfer in redox reactions and by the presence of many redox molecules within cells. Here we establish versatile methods for measuring redox states of proteins and their individual cysteine residues in vitro and in vivo, involving thiol-modifying reagents and LC-MS analysis. Using these methods, we demonstrate that the redox state of NF-κB is spatially regulated by its subcellular localization. While the p65 subunit and most cysteine residues of the p50 subunit are reduced similarly in the cytoplasm and in the nucleus, Cys-62 of p50 is highly oxidized in the cytoplasm and strongly reduced in the nucleus. The reduced form of Cys-62 is essential for the DNA binding activity of NF-κB. Several lines of evidence suggest that the redox factor Ref-1 is involved in Cys-62 reduction in the nucleus. We propose that the Ref-1-dependent reduction of p50 in the nucleus is a necessary step for NF-κB activation. This study also provides the first example of a drug that inhibits the redox reaction between two specific proteins.

The redox states of cysteine residues, which can change reversibly within cells, often greatly influence the various properties of proteins, such as protein stability, chaperone activity, enzymatic activity, and protein structure (1)(2)(3)(4)(5). It has also been suggested that several transcription factors bind to their cognate sites in a redox-regulated manner. Well characterized cases include the prokaryotic transcription factors SoxR and OxyR, which function as oxidative stress sensors, their DNA binding activated through oxidation of critical cysteine residues (6 -7). In most cases, however, the roles and mechanisms of redox regulation are not fully defined because it is difficult to monitor the alteration of redox states of proteins mainly due to the rapid proton transfer in redox reactions. A few have directly quantified the redox state of cysteine clustered with iron or amounts of oxidized cysteines using physicochemical or biochemical techniques (3, 8 -9), but these methods cannot describe the whole picture of redox states of a protein and are not widely applicable to other proteins. Therefore, most researchers have chosen an indirect way of using cysteine-substitution mutant proteins (3)(4)(5)7).
NF-B 1 is a eukaryotic transcription factor that regulates a wide variety of genes involved in immune function and development (10). NF-B is composed of two subunits, p50 and p65, both of which are members of the Rel family of transcription factors. NF-B normally exists in the cytoplasm, forming an inactive ternary complex with the inhibitor protein IB␣. Following the application of appropriate stimuli, NF-B is released from IB␣ and translocates into the nucleus, where it binds DNA and activates transcription of target genes. Mechanisms of NF-B activation have been extensively studied; however, it is largely unknown if, and how, the DNA binding step is activated in cells. Some reports have described that the DNA binding activity of NF-B is regulated by redox potential in vitro (11)(12)(13).
Recently, we reported an antiinflammatory drug that could allow us to solve this issue. The synthetic quinone derivative, (2E)-3-[5-(2,3-dimethoxy-6-methyl-1, 4-benzoquinoyl)]-2-nonyl-2-propenoic acid (E3330), is a novel anti-NF-B drug that specifically suppresses DNA binding activity of NF-B but not those of other inflammatory transcription factors, such as activator protein-1 (AP-1) and nuclear factor of activated T cell (NF-AT), in phorbol 12-myristate 13-acetate (PMA)-stimulated Jurkat cells (14). Interestingly, E3330 did not affect the DNA binding activity of purified NF-B or several steps of NF-B activation, including IB degradation, p65 phosphorylation, and nuclear translocation of NF-B (14). This led us to hypothesize that E3330 may target an unknown nuclear factor that stimulates the DNA binding activity of NF-B. Consistent with this assumption, we purified and identified redox factor-1 (Ref-1) from Jurkat nuclear extracts as an E3330-binding pro-tein that enhances DNA binding activity of NF-B in an E3330sensitive manner (15). Considering the specificity and affinity of the E3330-Ref-1 interaction (15), E3330 is likely a selective inhibitor of Ref-1 activity.
Ref-1 is a nuclear protein that was originally identified as a DNA repair enzyme with an apurinic (AP)-endonuclease activity (16 -18). Ref-1 has been reported to stimulate activities of AP-1 as well as other transcription factors including p53 and hypoxia inducible factor-1␣ (18Ϫ24). In case of AP-1, it was well investigated by using in vitro cysteine point mutagenesis analysis that DNA binding activity of AP-1 is stimulated by the reducing activity of Ref-1. We have recently shown that Ref-1 enhances the DNA binding activity of NF-B in vitro as well as NF-B-dependent transcriptional activation in vivo in an E3330-sensitive manner (15). However, few analyses have been performed to determine if these processes involve redox reaction, and how these processes occur in cells. The existence of many redox regulatory molecules, such as thioredoxin (Trx), thioredoxin-reductase (TrxR), glutaredoxin, nucleoredoxin, and glutathione, makes it difficult to explore the specific redox regulation between proteins in eukaryotic cells.
In order to overcome the difficulties associated with redox studies in vivo, we utilized E3330 as a useful tool because E3330 is a Ref-1-specific inhibitor that specifically inhibits NF-B activity but not the other transcription factors potentially regulated by Ref-1 as mentioned above. Using irreversible thiol-modifying reagents and LC-MS analysis, we evaluated here the redox states of NF-B during its activation step in vivo. We provide evidence that Cys-62 of NF-B p50 is selectively reduced by Ref-1 in the nucleus and that this reduction is a prerequisite for NF-B activation in vivo.
Recombinant Proteins-pET14b (Novagen)-based expression plasmids of full-length human p50 and Ref-1 have previously been described (15). An expression vector of full-length human Trx was prepared by inserting the entire open reading frame amplified by polymerase chain reaction into the NdeI-BamHI sites of pET14b (Novagen). Each His-tagged protein was expressed in Escherichia coli strain BL21(DE3) and purified under native conditions using Ni-NTAagarose beads (Qiagen) as instructed by the manufacturer. Purified proteins were dialyzed against dialysis buffer (HEPES pH 7.9, 100 mM KCl, 10% glycerol, 0.2 mM EDTA). For oxidation, p50 was further treated with 10 mM diamide for 30 min at 16°C. For reduction, Ref-1 and Trx were treated with 0.3 mM Tris-(2-carboxyethyl)phosphine-hydrochloride (TCEP) for 10 min at 37°C. Diamide and TCEP were removed by an additional Ni-NTA agarose chromatography, and purified proteins were dialyzed against dialysis buffer. Native TrxR purified from bovine (4 unit/mg) was purchased from American Diagnostica Inc. (Greenwich, Connecticut).
Redox Reaction in Vitro-Recombinant proteins were mixed in 100 l of reaction buffer (20 mM HEPES pH 7.9, 100 mM KCl, 10% glycerol, 0.2 mM EDTA). As a carrier protein, 100 g of carbonic anhydrase was included in the reactions to help stabilize the recombinant proteins and to normalize the difference in protein concentrations. After incubation for various times at 37°C, the reactions were used in electrophoretic mobility shift assay (EMSA) or fluorescence assays.
Fluorescence Assays-To label reduced cysteines in vitro, protein samples were incubated with 0.3 mM F5M for 5 min on ice. F5M labeling was stopped by the addition of 100 mM dithiothreitol (DTT). Proteins were acetone-precipitated, suspended in SDS sample buffer, and separated by SDS-PAGE. Fluorescence intensities of the proteins were quantified by scanning gels at an excitation wavelength of 490 nm using a fluorescence-image analyzer FLA2000 (Fuji Film). Data were analyzed with the Image Gauge software (Fuji Film).
For fluorescence labeling in vivo, J-50 cells cultured under different conditions were incubated in medium containing 10 mM F5M for 5 min at 37°C. All the subsequent steps were performed at 4°C. After washing twice with phosphate-buffered saline containing 10 mM F5M, cells were harvested by centrifugation, and pellets were resuspended in 8 packed cell volumes of hypotonic buffer containing 10 mM HEPES (pH 7.9), 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 0.5 mM phenylmethylsulfonyl fluoride, and 1 mM F5M. After a 10-min incubation period, Nonidet P-40 was added to a final concentration of 0.6%, and tubes were mixed vigorously for 10 s. Tubes were centrifuged for 5 s at 10,000 rpm, and supernatants (cytoplasmic fractions) were removed to separate vials. Nuclei were resuspended in 4 packed cell volumes of high salt buffer containing 20 mM HEPES (pH 7.9), 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 0.5 mM phenylmethylsulfonyl fluoride, and 1 mM F5M. After a 20-min incubation period, tubes were centrifuged for 2 min at 15,000 rpm, and supernatants (nuclear fractions) were saved. F5M labeling reactions were stopped by adding 100 mM DTT to the cytoplasmic and nuclear fractions. Following immunoprecipitation using an anti-FLAG M2 affinity gel, eluates were subjected to SDS-PAGE, and fluorescence intensities of proteins were analyzed as described above.
NEM Labeling and LC-MS Analysis-NEM labeling was performed similarly to F5M labeling, except that the experiments were performed with 1 mM (in vitro) and 20 mM (in vivo) NEM. Proteins were acetoneprecipitated, resuspended in a small volume of 8 M urea, and digested overnight at 37°C with LysC in digestion buffer containing 25 mM Tris-HCl (pH 9.0), 2 M urea, and 1 mM DTT. The high-performance liquid chromatography (HPLC)-coupled electron spray ionization timeof-flight MS system Mariner TK5000 (PerkinElmer) was used to analyze protein digests. Protein digests were initially separated on a protein/peptide ODS column with an inner diameter of 0.5 mm (PerkinElmer), using a linear gradient of 5-55% acetonitrile plus 1% formic acid at a flow rate of 5 m/min. Peptide masses were analyzed online, using instrument settings for positive ion polarity at 3800 V spray tip potential, 130 V nozzle potential, 11.5 V skimmer 1 potential, 7.6 V quadrupole DC potential, 0.6 V deflection voltage, Ϫ37.5 V einzel lens potential, 700 V quadrupole RF voltage, 140°C quadrupole temperature, 140°C nozzle temperature, 700 V push pulse potential, 278 V pull pulse potential, 5 V pull bias potential, 4000 V acceleration potential, 1400 V reflector potential, and 2150 V detector voltage. RNP is defined as (MI red )/[(MI red ) ϩ (MI ox )] ϫ 100 (%), where MI red and MI ox represent the mass intensities of NEM-labeled (MI red ) and -unlabeled (MI ox ) peptides. MI red partially comprises the mass intensity of hydrolytic-NEM-labeled peptide (ϳ10% total NEM). For a single peptide, a series of RNP values were determined with respect to the number of protons within the peptide (1-5H ϩ ). The RNP values obtained from peptides protonated differently were usually fairly similar. Figs. 2B and 4B show the averages of three or more RNP values.

RESULTS
Assessment of the Thiol-trapping System by F5M-To measure the redox state of proteins, F5M, a maleimide derivative with a fluorescein group was employed as a thiol-modifying reagent. Maleimides irreversibly modify reduced, but not oxidized, cysteines rapidly at 4°C at the physiological pH range (pH 6.5-8.0) where normal redox reactions can barely proceed. The thiol selectivity of maleimides is higher than those of other thiol-modifying reagents, such as iodoacetamides (25). Using F5M, the amount of proteins in the reduced state can be quantified as fluorescence intensity on an SDS-polyacrylamide gel (Fig. 1A). To assess this F5M-mediated thiol-trapping system, we initially examined a well characterized redox exchange reaction between Trx and TrxR in vitro (26). In the presence of NADPH, TrxR specifically reduces Trx at the two conserved cysteine residues, Cys-32 and Cys-35. A constant amount of oxidized Trx was incubated with increasing concentrations of TrxR for various times at 37°C, and reactions were processed for F5M labeling at 4°C. As shown in Fig. 1, B and C, the fluorescence intensity of Trx increased with time, depending on the concentration of TrxR. When both Cys-32 and Cys-35 of Trx were mutated to serine residues, fluorescence intensity did not increase to a detectable level over time. These results strongly suggest that the F5M labeling reaction is specific and permits the quantitative measurement of the protein redox states.
Ref-1 Reduces p50, Leading to Activation of Its DNA Binding Activity in an E3330-sensitive Manner-We have previously shown using EMSA that Ref-1 stimulates DNA binding activity of p50/p50 homodimers and p50/p65 heterodimers, but not p65/p65 homodimers (15). These findings suggest that Ref-1 activates DNA binding of NF-B by affecting the p50 subunit. To directly determine whether Ref-1 reduces p50, we used the above F5M labeling method. Oxidized and reduced forms of p50 were prepared by incubating recombinant p50 with the oxidizing reagent diamide and the reducing reagent TCEP, as described under "Experimental Procedures." As shown in Fig Fig. 1B, after 1 min, p50 was reduced to ϳ50% of the level achieved after a 60-min incubation. Conversely, fluorescence intensity of Ref-1 was decreased to ϳ60% in 1 min. We thus concluded that Ref-1 directly and rapidly reduces p50 by a redox exchange reaction.
It has been suggested, but not proven, that inhibition of Ref-1 by E3330 occurs through the block of its redox activity (15). Consistent with the previous results, E3330 suppressed Ref-1-dependent activation of DNA binding by oxidized p50 in a dose-dependent manner (Fig. 2C, upper panel, lanes 1-6). Similarly, E3330 also inhibited the Ref-1-dependent reduction of oxidized p50 (lower panel). In contrast, E3330 did not affect redox states or DNA binding activity of the reduced form of p50. From these results, we conclude that E3330 inhibition is mediated by the block of Ref-1 redox activity.
Significant Reduction of the Cys-62 Residue of p50 by Ref. 1-The above method is a convenient way to measure changes in the redox status of proteins. However, it may not be suitable in cases where proteins contain multiple reduced cysteines of which only a subset is subject to redox regulation because F5M labeling of non-target cysteines may cause a high basal level of fluorescence.
Human p50 has seven cysteine residues at positions 62, 88, 119, 124, 162, 262, and 273. To identify cysteine residues of p50 that are reduced by Ref-1, we improved the above thiol-trapping method. The modified method comprises three steps: 1) irreversible labeling of reduced cysteines with another maleimide derivative, NEM, 2) digestion of NEM-labeled proteins by Lys-C endopeptidase, and 3) LC-MS analysis for the identi-FIG. 1. Redox monitoring system using maleimides. A, schematic representation of the thiol-trapping system for measuring the redox state of proteins (fluorescence assay). Since the thiol-modifying reagent, F5M, specifically labels reduced cysteine residues, the reduced state of proteins is measured as fluorescence intensity. B and C, visualization of the redox state of Trx after the reaction between Trx and TrxR. Oxidized Trx wild type and Trx C32, 35 S (40 pmol) were incubated with indicated amounts of TrxR for various times at 37°C in the presence of 0.2 mM NADPH, and reduced states of Trx were analyzed by fluorescence assays. The fluorescence intensities of Trx were measured and plotted against the reaction time (C). Fluorescence intensity of Trx treated with 2 pmol of TrxR for 2400 s was set as 100%.
fication of NEM-labeled (reduced) and -unlabeled (oxidized) peptides. NEM was employed because F5M-labeled peptides could not be fully recovered from liquid chromatography, probably due to the hydrophobicity as a result of F5M labeling (data not shown). From the mass intensities of NEM-labeled and -unlabeled peptides (MI red and MI ox ), the fraction of reduced cysteine residues was estimated using the equation: MI red / (MI red ϩ MI ox ) ϫ 100 (%) or the ratio of NEM-labeled peptide (RNP).
From the results shown in Fig. 2A, we presumed that residues reduced by TCEP and Ref-1 should be functionally important. Recombinant p50 previously oxidized with diamide was incubated with TCEP or Ref-1 and processed as described above. Lys-C endopeptidase digests of p50 include three peptides containing a single cysteine residue and two with two cysteine residues (Table I). Fig. 3A shows representative mass spectra of the Cys-62-containing peptide. When oxidized p50 was left untreated, a large fraction of Cys-62 appeared to be fully oxidized because only the unlabeled peptide (53-77) was detected. Prior incubation with TCEP or Ref-1 resulted in the NEM-labeled peptide (53-77-NEM), indicating that Cys-62 was reduced by these reagents. The RNPs, obtained from their mass intensities, suggest that treatment of p50 with TCEP or Ref-1 resulted in ϳ60 or ϳ50% reduction of Cys-62 (Fig. 3B).
Similarly, all cysteine-containing peptides were identified by LC-MS analysis. RNP values for these peptides are presented in Fig. 3B. For peptides containing two cysteines, RNPs were calculated for singly and doubly labeled peptides, respectively. All the cysteine residues were reduced by Ref-1 by varying degrees; however, the reduction of Cys-62 by Ref-1 was significant and highly sensitive to E3330 inhibition. For comparison, EMSAs for an identical set of reactions were performed (Fig.  3C). The DNA binding activity of p50 correlated well with the redox states of Cys-62, indicating that Cys-62 plays a determinant role in DNA binding. Unexpectedly, TCEP did not reduce all the cysteine residues of p50. TCEP may be capable of reducing only the exposed cysteine residues under the non-denaturing condition. Consistent with this hypothesis, structural data show that thiol groups of Cys-88 and Cys-162, which were not reduced by TCEP in our experiments, are buried in the p50 protein (27).
The Reduced Form of p50 Cys-62 Is Essential for Its DNA Binding Activity-The above results do not exclude the possibility that, in addition to Cys-62, other cysteines may play some role in redox regulation of p50. To address this issue, we carried out EMSA and fluorescence assays for point mutants of p50, whose cysteine residues at positions 62, 119, 262, and 273 were  individually changed to serines (C62S, C119S, C262S, and C273S). Serine was chosen because it may mimic a reduced form of cysteine. Similar sets of the point mutants were previously described and analyzed by EMSA (12,28). As observed for wild type p50, DNA binding activities of C119S, C262S, and C273S were not detectable under oxidized conditions and were strongly stimulated by TCEP in a concentration-dependent manner (Fig. 4). In sharp contrast, C62S showed strong DNA binding even in its oxidized form and was not affected by TCEP, i.e. by the redox states of the other cysteine residues. The results of EMSA are consistent with those of earlier reports (12,28,29). Taken together, it is very likely that Cys-62 alone is responsible for Ref-1-mediated reduction and activation of p50.
Reduction of p50 by Ref-1 in the Nucleus-Using these redoxmonitoring methods, we investigated the possibility that the redox state of NF-B is regulated during the activation process in vivo. We established a Jurkat cell line expressing p50 fused to FLAG and 6ϫ His tags at the N terminus (J-50 cells) to obtain a sufficient quantity of homogenous NF-B complex from cells. FLAG-His-p50 (FLAG-p50) behaved similarly to endogenous p50, i.e. it formed a ternary complex with p65 and IB␣ in the cytoplasm, was translocated into the nucleus on PMA treatment concomitant with IB degradation, and bound DNA as a heterodimer with p65 in a sequence-specific manner (data not shown).
To visualize the overall redox state of NF-B in cells, we modified the F5M labeling protocol as follows: J-50 cells were maintained with or without PMA for 30 min, labeled in situ with membranepermeable F5M, lysed, and fractionated into cytoplasmic and nuclear fractions. From these fractions, tagged NF-B complexes were immunoprecipitated with anti-FLAG antibody and separated on SDS gels. Protein amounts and fluorescence intensities of FLAG-p50 and p65 were measured by silver staining and fluorescent imaging of the gels (Fig. 5). Relative fluorescence intensity was calculated by normalizing the fluorescence intensity to its protein amount, and values of cytoplasmic FLAG-p50 and p65 in unstimulated cells were set as 100%. Because a very small amount of FLAG-p50 was present in the nucleus of unstimulated cells, the corresponding sample was concentrated 6-fold before analysis. F5M labeled the FLAG-p50 protein more strongly in the nucleus than in the cytoplasm of untreated cells (Fig. 5A, lanes 3 and 9). In contrast, the relative fluorescence intensity of p65 was similar in both the cytoplasm and nucleus. PMA stimulation of cells increased nuclear p50 and p65 protein amounts, but did not significantly affect their overall redox states, regardless of subcellular location (Fig. 5, A and B, lanes 3, 4, 9, and 10). These results indicate that p50 is reduced upon translocation into the nucleus and that this reduction occurs independently of extracellular stimuli. To our knowledge, this is the first report demonstrating different redox states of a protein in the nucleus and cytoplasm. What is the protein(s) responsible for the observed reduction of p50 in the nucleus? Among the various redox regulatory proteins, Trx and Ref-1 have been implicated in the modulation of NF-B activity (12,13,30). We therefore employed E3330 to investigate the possible role of Ref-1 in the reduction of p50 in the nucleus. As shown in Fig. 5A, treatment of cells with increasing concentrations of E3330 prior to PMA stimulation inhibited p50 reduction in the nucleus, but not in the cytoplasm. In contrast, E3330 had little effect on the redox states of p65 (Fig. 5B). These results indicate that Ref-1 is involved in p50 reduction in the nucleus but not in the reduction of cytoplasmic p50/p65 or nuclear p65.
Redox Regulation of Cys-62 in Vivo-We next examined the redox states of individual cysteine residues of p50 using a combination of in vivo NEM labeling and LC-MS analysis. Following sequential treatment with E3330 and PMA, J-50 cells were labeled in situ with NEM and separated into cytoplasmic and nuclear fractions. These extracts were sequentially subjected to anti-FLAG affinity chromatography and nickel chelate chromatography under denaturing conditions to yield homogenous FLAG-p50 (Fig. 6B, top left panel). Very small amounts of FLAG-p50 were obtained from the nuclear fraction without PMA treatment, and therefore no further analyses were performed on this sample. Purified FLAG-p50 was digested with Lys-C endopeptidase and subjected to LC-MS, and the data were processed similarly to Fig. 3. Fig. 6A shows the representative mass spectra of the Cys-62-containing peptide. In the cytoplasm of untreated or PMA-treated cells, a large fraction of p50 Cys-62 was oxidized, as observed from the mass intensity of unlabeled peptide (53-77), which was higher than that of the NEM-labeled peptide (53-77-NEM). In sharp contrast, p50 in the PMA-treated cell nucleus contained significantly more reduced Cys-62, evident from the relative mass intensity of NEM-labeled and unlabeled peptides. All the cysteine-containing peptides were similarly identified by LC-MS analyses, and RNP values for these peptides are presented in Fig. 6B. These results indicate that half or more of the Cys-88, Cys-119, Cys-124, and Cys-162 residues are reduced constitutively, regardless of the subcellular location of NF-B or the presence of PMA or E3330. In sharp contrast, the level of Cys-62 reduction appears to be low (ϳ19%) in the cytoplasm and high (ϳ80%) in the nucleus. Similarly, Cys-262/ 273 reduction is low-to-moderate (3-57%) in the cytoplasm and moderate-to-high (30 -80%) in the nucleus. These results indicate that nucleus-specific reduced state of p50 observed in Fig.  4A is due to the reduction of these three cysteine residues. Strikingly, the increased reduction of Cys-62 was severely impaired by prior treatment of the cells with E3330, resulting in a reduction level far below that found in the cytoplasm (Fig. 6,  A and B). In contrast, the reduction of Cys-262/273 in the nucleus was not substantially affected by E3330. It is therefore likely that Ref-1 plays a major role in the reduction of Cys-62 in FIG. 4. The reduced form of Cys-62 is critical for DNA binding by p50. Oxidized forms of p50 wild type (WT) and its mutants were incubated with increasing concentrations of TCEP for 1 h at 37°C. Their DNA binding activities and redox states were analyzed by EMSA and fluorescence assays. C62S, C119S, C262S, and C273S represent p50 mutants carrying cysteine to serine substitutions at positions 62, 119, 262, and 273.

FIG. 5. Visualization of the redox states of NF-B in vivo.
A and B, J-50 (2 ϫ 10 7 ) cells were incubated with the indicated concentrations of E3330 for 2 h, followed by 50 g/ml PMA for 30 min where specified. Cells were labeled with F5M and fractionated as described under "Experimental Procedures." FLAG-NF-B complexes were immunoprecipitated from the cytoplasmic and nuclear fractions using an anti-FLAG M2 affinity gel, eluted with FLAG peptide (500 M), resolved by SDS-PAGE, and visualized by fluorescence scanning or silver staining. The relative fluorescence intensities normalized to protein amounts are shown at the bottom. Immunoprecipitation from control Jurkat cells resulted in undetectable fluorescence (lanes 1 and 2). the nucleus but not of the other cysteine residues of p50. These findings, together with the observation that E3330 inhibits NF-B-dependent transcriptional activation (14), strongly suggest that Ref-1-mediated reduction of p50 Cys-62 in the nucleus is a prerequisite for NF-B activation.

Ref-1 Activates NF-B through Reduction of p50 -We
showed here that Ref-1 specifically reduced Cys-62 of p50, which led to activation of NF-B DNA binding in vitro and in vivo. We also showed that Trx and TrxR had no effect on the redox status of p50. Previous studies have implicated Trx in the redox regulation of NF-B (12)(13)30). However, several lines of evidence suggest that Ref-1 plays a major role in p50 reduction in vivo. In vitro, a 2.5-fold molar excess of Ref-1, but not of Trx, reduced and activated p50 in 1 min ( Fig. 2A), suggesting that Ref-1 possesses a higher reducing potential than Trx. Previous studies used a 1000-fold molar excess of Trx and a 15or 30-min incubation to see activation of NF-B (12,30). In addition, Trx predominantly localizes in the cytoplasm (31). Quantitative Western blotting suggests that Ref-1 exists at 0.5ϳ1 ϫ 10 6 molecules per cell or 100-to 200-fold molar excess to p50, whereas Trx exists at 0.1ϳ0.2 ϫ 10 6 molecules per cell or 20-to 40-fold molar excess in the nucleus of PMA-stimulated Jurkat or J-50 cells. 2 It was previously reported that cytoplasmic Trx translocates into the nucleus upon stimulation of cells with PMA or UV light (31)(32); however, we did not observe any detectable change in Trx localization even 2-h post-PMA stimulation (data not shown). Because p50 in the nucleus became reduced within 30-min post-PMA stimulation and because this reduction was strongly inhibited by E3330 (Figs. 5A and 6B), it seems unlikely that Trx directly reduces p50 in the nucleus, at least under our experimental conditions. Trx may contribute to NF-B activation indirectly by enhancing Ref-1 redox activity (31,33) or by regulating IB degradation in the cytoplasm (34). Our data are consistent with a recent report showing that induction of Ref-1 expression coincides with the activation of DNA binding by NF-B, whereas inhibition of Ref-1 expression by an antisense oligonucleotide abolishes DNA binding activity of NF-B in granule neurons (35).
Ref-1 targets p50 but not its heterodimeric partner p65. p50 (436 amino acids) and p65 (552 amino acids) show significant sequence similarity over the Rel homology region (RHR) spanning ϳ300 amino acids at their N termini (ϳ40% identity and ϳ50% similarity). The RHR, which encodes two functions, DNA 2 T. Nishi and H. Handa, unpublished observations. creased by improving the sensitivity of LC-MS analysis. That said, these methods are the versatile ways of monitoring the redox states of cysteines in vitro and in vivo and provide an additional opportunity for the study of redox regulation.