Studies into the effect of the tyrosine kinase inhibitor herbimycin A on NF-kappa B activation in T lymphocytes. Evidence for covalent modification of the p50 subunit.

The tyrosine kinase inhibitor herbimycin A was found to block NF-κB stimulation in response to interleukin-1 and phorbol 12-myristate 13-acetate in EL4.NOB-1 thymoma cells and phorbol 12-myristate 13-acetate in Jurkat T lymphoma cells. The effect appeared not to involve inhibition of tyrosine kinase activation as neither interleukin-1 nor phorbol 12-myristate 13-acetate induced major changes in tyrosine phosphorylation in EL4.NOB-1 or Jurkat cells, respectively. Herbimycin A did not interfere with IκB-α degradation, and in unstimulated cells, it modified NF-κB prior to chemical dissociation with sodium deoxycholate. Because herbimycin A is thiol-reactive, we suspected that the target was the p50 subunit of NF-κB, which has a key thiol at cysteine 62. Herbimycin A inhibited DNA binding when added to nuclear extracts prepared from stimulated cells, which were shown to contain high levels of p50. Incubation of herbimycin A with 2-mercaptoethanol attenuated the effect. Herbimycin A was also shown to react directly with p50, blocking its ability to bind to the NF-κB consensus sequence. However, a mutant form of p50 in which cysteine 62 was mutated to serine was insensitive to herbimycin A. Finally, we demonstrated that the compound inhibited the expression of interleukin-2 (an NF-κB-regulated gene) in EL4.NOB-1 cells. These data therefore suggest that herbimycin A inhibits NF-κB by modifying the p50 subunit on cysteine 62 in the NF-κB complex, which blocks DNA binding and NF-κB-driven gene expression. The results urge caution in the use of herbimycin A as a specific tyrosine kinase inhibitor and suggest that the development of agents that selectively modify p50 may have potential as a means of inhibiting NF-κB-dependent gene transcription.

NF-B is a transcription factor that regulates the expression of genes involved in the immune and inflammatory responses, including many that code for cytokines, cell-surface receptors, adhesion molecules, and acute-phase proteins (1). It becomes activated in many cell types in response to viruses, bacteria, and stress factors as well as inflammatory cytokines such as interleukin-1 (IL-1) 1 (2) and nonphysiological stimuli such as the protein kinase C activator phorbol 12-myristate 13-acetate (PMA) (3). The DNA-binding subunits of NF-B currently comprise five members in mammals: p50, p65 (RelA), c-Rel, p52, and RelB (4 -6). RelA, RelB, and c-Rel are capable of transactivation, normally forming heterodimers with p50 or p52. This results in complexes with high DNA binding affinity. The predominant form of NF-B in resting cells, however, is a p50-RelA heterodimer, which is retained in the cytoplasm complexed to an inhibitor protein, IB. Upon stimulation with such agents as IL-1, IB dissociates from the NF-B heterodimer, which translocates to the nucleus, where it binds with high affinity to the NF-B consensus sequence in target genes, thereby modulating gene expression (1,3). Multiple forms of IB also occur (6 -10), with IB-␣ and IB-␤ being the two most important forms for NF-B activation.
A model for NF-B activation involving phosphorylation and proteolysis of IB-␣ has been proposed (11)(12)(13)(14). The evidence for proteolysis has come from studies demonstrating IB-␣ degradation during the activation process and from the observation that inhibitors of chymotrypsin-like proteases block NF-B activation in response to diverse stimuli (14). Recent evidence suggests that the multicatalytic cytosolic protease (proteosome) may be responsible for IB-␣ breakdown (15). A role for phosphorylation was indicated in studies demonstrating phosphorylation of IB-␣ in vitro by protein kinases A and C and heme-activated kinase, which resulted in dissociation of IB from NF-B (16 -18). Furthermore, a transient change in the electrophoretic mobility of IB was apparent in cytosolic extracts following exposure of cells to diverse stimuli. Treatment of these extracts with calf intestinal phosphatase or potato acidic phosphatase followed by Western blot analysis showed that this modified form was converted to the unmodified (nonphosphorylated) form (11). Phosphorylation of IB-␣ has not been shown directly, however, nor has the protein kinase(s) responsible for IB phosphorylation in intact cells been isolated. Recent evidence has suggested that IB-␣ phosphorylation may tag the protein for subsequent and rapid degradation by the chymotrypsin-like subunit of the proteosome, indicating that both phosphorylation and proteolysis are equally necessary for NF-B activation (15). Further complexity is suggested, however, from studies demonstrating that both p50 and RelA become phosphorylated upon activation of cells and require phosphorylation for DNA binding and transactivation (19).
The involvement of a tyrosine kinase in NF-B activation has been indicated from the observation that tyrosine kinase inhibitors such as herbimycin A and genistein inhibit NF-B activation in response to IL-1 (20 -22). The target for these inhibitors has not been precisely determined, however. Both inhibitors have different mechanisms of action. Genistein is a competitive inhibitor for tyrosine kinases (23), while it has been suggested that herbimycin A, through its benzaquinone moiety, can directly modify a key thiol group on Src and Abl tyrosine kinases (24,25). The possible involvement of a Src-like tyrosine kinase was further suggested from studies demonstrating that expression of v-src in T cells correlated with NF-B activation, which was sensitive to herbimycin A (26).
Using herbimycin A, we have attempted to clarify the role of tyrosine kinases in the activation of NF-B in response to IL-1 and PMA in the murine thymoma cell line EL4.NOB-1 and PMA in the human T lymphoma line Jurkat E6.1. Our results suggest that the inhibitory effect is consistent with a model involving the direct covalent modification of the p50 subunit of NF-B, rather than inhibition of tyrosine kinase activity or other signals that lead to IB-␣ dissociation and degradation in these cells. The results further suggest that, using herbimycin A, it is possible to modify the p50 subunit of NF-B in intact cells and interfere with DNA binding and NF-B-driven gene expression. [␥-32 P]ATP (10 mCi/mmol), enhanced chemiluminescence (ECL) reagents, and anti-mouse IgG peroxidase conjugate were from Amersham International (Aylesbury, United Kingdom). The monoclonal antibody to phosphotyrosine was from Boehringer Mannheim (Mannheim, Germany). The affinity-purified rabbit polyclonal antibody raised against the amino-terminal domain of human IB-␣/MAD-3, the rabbit polyclonal antibody raised against the 15-amino acid sequence mapping at the nuclear localization sequence region of p50, and the rabbit polyclonal antibody raised against the carboxyl-terminal domain of human RelA and the amino terminus of c-Rel were supplied by Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Purified murine anti-IL-2, recombinant murine IL-2, biotinylated murine anti-IL-2, and avidin-peroxidase were supplied by Pharmingen (San Diego, CA).

Materials
Cell Culture-Human Jurkat E6.1 lymphoblast and murine EL4.NOB-1 thymoma cells were cultured in RPMI 1640 medium. All medium was supplemented with 100 units/ml penicillin, 100 g/ml streptomycin, and 10% (v/v) fetal calf serum. Cells were maintained at 37°C in a humidified atmosphere of 5% CO 2 . IL-1␣ and PMA stimulation was performed on cells in serum-containing medium at 37°C for all experiments unless otherwise stated.
Preparation of Subcellular Fractions-Subcellular fractions were prepared as described previously (27). Briefly EL4.NOB-1 or Jurkat cells (1-5 ϫ 10 6 /ml) were plated into 24-well plates (16-mm diameter) 20 min before stimulation. Cells were activated with IL-1␣ (10 ng/ml) or PMA (100 ng/ml) for different time periods. In some experiments, cells were pretreated with herbimycin A for 1 h before IL-1␣ or PMA treatment. Stimulation of cells was terminated by addition of ice-cold PBS, and nuclear and cytosolic extracts were prepared as described previously (27). Protein concentrations of nuclear extracts were determined by the method of Bradford (28), and the extracts were assayed immediately for NF-B activity or stored at Ϫ70°C until further use. All of the above steps were performed at 4°C unless otherwise stated.
Electrophoretic Mobility Shift Assay-Nuclear extracts (4 g of protein) were incubated with 10,000 cpm of a 22-base pair oligonucleotide containing the NF-B consensus sequence that previously had been labeled with [␥-32 P]ATP (10 mCi/mmol) by T4 polynucleotide kinase (29). Incubations were performed for 30 min at room temperature in the presence of 2 g of poly(dI⅐dC) and 100 mM Tris-HCl, pH 7.5, containing 100 mM NaCl, 10 mM EDTA, 50 mM DTT, 40% (w/v) glycerol, and 1 mg/ml nuclease-free bovine serum albumin. In some experiments, antibodies to p50, RelA, or c-Rel were added to the extracts before incubation with labeled oligonucleotide. Cell-free activation of latent NF-B activity required pretreatment of cytosolic extracts (4 g of protein) with 0.8% (w/v) sodium deoxycholate and 1.1% (w/v) Nonidet P-40 for 10 min on ice before incubation as described above with the labeled oligonucleotide probe. All incubations were subjected to electrophoresis on native 5% (w/v) polyacrylamide gels that were subsequently dried and autoradiographed.
In Vitro Experiments with Nuclear Extracts and p50 -Nuclear extracts (4 g) from IL-1␣-stimulated cells (10 ng/ml, 1 h) were incubated in vitro with herbimycin A or a mixture of herbimycin A and 2-mercaptoethanol (142 mM) for 10 min at room temperature. The mixture of herbimycin A and 2-mercaptoethanol was incubated for 10 min at room temperature prior to addition to extract. Extracts from IL-1-treated cells were also incubated with vehicle (dimethyl sulfoxide) or 2-mercaptoethanol alone. Binding buffer and poly(dI⅐dC) were then added, followed by a 30-min incubation with radiolabeled NF-B probe. Extracts were analyzed for NFB DNA binding activity by electrophoretic mobility shift assay. This protocol was also carried out using nuclear extracts prepared in DTT-free buffers.
Experiments were also carried out with recombinant human p50 and a mutant form of this protein in which cysteine 62 was mutated to serine. Both proteins were in buffer comprising 10 mM Tris-HCl, pH 7.5, 1 M NaCl, 10 mM EDTA, 40% glycerol, 1 mg/ml nuclease-free bovine serum albumin, and 0.25 mM DTT. 11 ng of p50 or 200 ng of C62S mutant were incubated with herbimycin A (100 -0.1 M) for 2 h at 37°C prior to assessing DNA binding. Higher amounts of mutant were required to detect DNA binding, as described previously (30).
Anti-phosphotyrosine Immunoblot Analysis-Cells (5 ϫ 10 6 /ml) were preincubated in RPMI 1640 medium for 15 min at 37°C prior to treatment with 2 M herbimycin A for 1 h. Samples were then either unstimulated or stimulated with IL-1␣ (10 ng/ml) or PMA (100 ng/ml) for various time periods. Stimulation was terminated by addition of ice-cold PBS containing 10 mM NaF, 400 mM Na 3 VO 4 , and 5 mM EDTA. The samples were centrifuged at 150 ϫ g for 5 min, and the pellets were resuspended in lysis buffer containing 1% (v/v) Triton X-100, 10 mM Tris-HCl, pH 7.6, 5 mM EDTA, 50 mM NaCl, 30 mM sodium pyrophosphate, 50 mM NaF, 500 M Na 3 VO 4 , 2 g/ml aprotinin, 10 mM phenylmethylsulfonyl fluoride, and 100 mM leupeptin. Following a 30-min incubation at 4°C, the lysates were clarified by centrifugation at 14,000 ϫ g in a microcentrifuge for 15 min at 4°C. The supernatants were assayed for protein (28), and an aliquot containing 40 g of protein was mixed with 2 ϫ Laemmli sample buffer (31) and then boiled for 2 min. Equal amounts of protein were resolved by electrophoresis on an SDS-10% polyacrylamide gel (31). The proteins were then transferred onto nitrocellulose membrane and stained with Ponceau S, and the molecular mass standards were marked. The blots were blocked in PBS, 0.5% (v/v) Tween 20 containing 5% (w/v) nonfat milk for 1 h at room temperature, followed by three washes in PBS, 0.5% (v/v) Tween 20 (the two initial washes were 5 min each, whereas the third was 15 min). The blots were incubated with monoclonal antibody to phosphotyrosine for 1 h at room temperature and then washed three times as before. Finally, the blots were incubated with a 1:15,000 dilution of anti-mouse IgG peroxidase conjugate for 45 min, followed by five washes (four 5-min washes and one 15-min wash). The blots were developed by ECL Western blotting (Amersham International) according to the manufacturer's recommendations.
Anti-IB-␣ Immunoblot Analysis-Cells (5 ϫ 10 6 to 1 ϫ 10 7 ) were preincubated in RPMI 1640 medium for 15 min at 37°C prior to treatment with 2 M herbimycin A for 1 h. Samples were then either unstimulated or stimulated with IL-1␣ (10 ng/ml) or PMA (100 ng/ml) for various time periods. Stimulation was terminated by addition of ice-cold PBS, and the samples were centrifuged at 150 ϫ g for 5 min. The pellets were resuspended in ice-cold radioimmune precipitation assay buffer (PBS, 1% (v/v) Nonidet P-40, 0.5% (w/v) sodium deoxycholate, and 0.1% (w/v) SDS) containing 0.1 mg/ml phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, and 1 mM Na 3 VO 4 . Following gentle mixing, the samples were incubated for 30 min on ice. The cells were further disrupted by passage through 21-gauge needles (10 strokes). An additional 0.1 mg/ml phenylmethylsulfonyl fluoride was added to each sample, and then a 30-min incubation on ice followed. Finally, the samples were centrifuged at 14,000 ϫ g for 20 min at 4°C, and the supernatant constituted the total cell lysate. The supernatants were assayed for protein (28), and an aliquot containing 20 g of protein was mixed with 2 ϫ Laemmli sample buffer and boiled as described above. Equal amounts of protein were resolved by electrophoresis on an SDS-10% polyacrylamide gel. The proteins were transferred onto nitrocellu-lose membrane and then stained with Ponceau S, and the molecular mass standards were marked. The blots were blocked in PBS containing 5% (w/v) nonfat milk for 1 h at room temperature, followed by two 2-min washes in PBS, 0.5% (v/v) Tween 20. The blots were incubated in anti-IB antibody for 1 h at room temperature and then washed three times in PBS (the first two washes were 7 min in length, and the final wash was 3 min). A dilution of anti-rabbit IgG peroxidase conjugate (1:4000) prepared in PBS was incubated with the membrane for 45 min at room temperature. Finally, the membrane was washed four times in PBS (the first three washes were 5 min in length, and the final wash was 15 min). The blots were developed by ECL Western blotting (Amersham International) according to the manufacturer's recommendations.
Interleukin-2 Assay-Cells (1 ϫ 10 6 /ml) were seeded in 24-well plates in RPMI 1640 medium containing 10% (v/v) fetal calf serum. Following a preincubation of 15 min at 37°C, cells were treated with varying concentrations of herbimycin A for 1 h. Cells were then either unstimulated or stimulated with IL-1␣ (5 ng/ml) or PMA (1 ng/ml) for 24 h at 37°C. The plates were centrifuged at 150 ϫ g for 10 min at room temperature. The supernatants were removed and stored at Ϫ20°C until assayed. IL-2 was measured using an enzyme-linked immunosorbent assay as described (32).
Statistical Analysis-Significance was evaluated by Student's t test for paired data. Fig. 1 demonstrates the effect of herbimycin A on NF-B activation by IL-1 and PMA in EL4.NOB-1 thymoma cells and PMA in Jurkat T lymphoma cells. Nuclear extracts from untreated EL4 cells contained trace levels of NF-B (Fig. 1, A and B, lane 1). Upon treatment with IL-1 (10 ng/ml) for 1 h, the level of this constitutive form of NF-B increased, and two additional complexes of lower electrophoretic mobility appeared in the nucleus (Fig. 1A, lane 2), as described previously (33,34). Pretreating cells with herbimycin A for 1 h resulted in a concentration-dependent inhibition of NF-B activation detectable in nuclear extracts (Fig. 1A, lanes 6 -8). Herbimycin A had no effect on its own (Fig. 1A, lanes 3-5). The activation of NF-B by PMA (100 ng/ml) in EL4 cells also proved sensitive to inhibition. As shown previously, EL4 cells require stimulation for up to 24 h with PMA before NF-B activation is detectable (Fig. 1B, lane 2) (34). As with IL-1, however, herbimycin A inhibited this response over the concentration range 0.02-2 M (Fig. 1B, lanes 6 -8). Herbimycin A did not have any effect on its own over this time course (Fig. 1B, lanes 3-5).

Herbimycin A Inhibits NF-B Activation Induced by IL-1 and PMA in EL4.NOB-1 Thymoma Cells and Jurkat T Lymphoma Cells-
The ability of herbimycin A to inhibit the activation of NF-B by PMA was in direct contrast to a previous study (20), which reported that PMA activation of NF-B in 70Z/3 pre-B cells was insensitive to the drug. 70Z/3 cells differ from EL4 cells in that the activation of NF-B by PMA in the cells is more rapid. We therefore next examined another cell type in which PMA causes rapid activation of NF-B, the human T lymphoma line Jurkat E6.1. A 1-h treatment of the cells with PMA resulted in a strong activation of NF-B (Fig. 1C, lane 2), which was sensitive to herbimycin A, with concentrations of 0.02-2 M inhibiting this activation (lanes 3-5). Furthermore, a similar pattern of inhibition was observed following a 24-h exposure to PMA (data not shown).
IL-1 and PMA Do Not Cause Major Changes in Tyrosine Phosphorylation in EL4.NOB-1 or Jurkat E6.1 Cells-We next attempted to monitor possible changes in tyrosine phosphorylation of proteins in cells exposed to IL-1 or PMA and to determine whether such changes, if any, were sensitive to herbimycin A. No dramatic changes in tyrosine phosphorylation were observed. IL-1 caused a weak increase in tyrosine phosphorylation of proteins in EL4 cells, ranging in molecular mass from 29 to 66 kDa ( Fig. 2A, lanes 2 and 3) 5 and 6). Control and IL-1-treated sample lanes incubated with primary antibody and excess phosphotyrosine ( Fig. 2A, lanes 7 and 8) or phosphoserine (lanes 9 and 10) are also shown, demonstrating that the primary antibody was specifically detecting phosphotyrosine. Similarly, in Jurkat T cells, PMA caused only a minor increase in tyrosine phosphorylation over 10 min (Fig. 2B,  lanes 1-3) in proteins between 45 and 66 kDa. No inhibition of these changes was seen when Jurkat cells were preincubated with herbimycin A (Fig. 2B, lanes 4 -6). Because of the disappointing changes observed, we carried out a positive control experiment in which Jurkat cells were treated with a combination of PMA and phytohemagglutinin (PHA), which are potent activators of T lymphocytes (35), with PHA being a strong stimulator of tyrosine phosphorylation (36). Treatment of Jurkat T cells for 5 min with PMA and PHA resulted in a prominent increase in tyrosine phosphorylation of proteins with molecular masses ranging from Ͻ29 to 66 kDa (Fig. 2C, lanes 2  and 3). Pretreatment with herbimycin A markedly inhibited these changes (Fig. 2C, lanes 4 and 5). We therefore concluded that neither IL-1 nor PMA dramatically affected tyrosine phosphorylation in EL4 or Jurkat cells, respectively, and that any changes observed were marginally sensitive to herbimycin A.
Herbimycin A Does Not Prevent IL-1-or PMA-stimulated IB-␣ Degradation and Modifies NF-B Prior to Chemical Dissociation with Deoxycholate-We next determined whether herbimycin A could inhibit a key event in the signal transduction pathway to NF-B activation, the degradation of IB-␣ (14,15). Cells treated with IL-1 or PMA were probed for IB-␣ by Western blotting of cell extracts. Treatment of EL4 cells with IL-1 for 10 min caused a marked decrease in detectable IB-␣ (Fig. 3A, lanes 1-3). Pretreatment of the cells with 2 M herbimycin A for 1 h had no effect on this response and actually caused a marked potentiation of the degradation (Fig. 3A, lanes  4 -6). Similarly, PMA caused a decrease in IB-␣ in Jurkat cells after 30 min of stimulation (Fig. 3B, lanes 1 and 2), which again was not inhibited by prior incubation with herbimycin A (lanes  3 and 4). Herbimycin A again had a marked potentiating effect on the degradation of IB-␣. The results from these experiments suggested that the target for herbimycin A was unlikely to be the release and proteolysis of IB-␣.
This was further suggested from studies involving deoxycholate-treated cytosolic extracts, as shown in Fig. 3C. As has ml) were pretreated for 1 h with 2 M herbimycin A (HbA) or left untreated and then exposed to medium and IL-1␣ (10 ng/ml) (A), PMA (100 ng/ml) (B), or PMA (10 ng/ml) and PHA (10 g/ml) (C) for the indicated times. Cell lysates were prepared and subjected to SDS-polyacrylamide gel electrophoresis along with protein molecular mass markers as described under "Experimental Procedures." The proteins were transferred onto nitrocellulose membranes and stained with Ponceau S to ascertain equal loading and position of molecular mass markers. The blots were then probed with anti-phosphotyrosine antibody and processed as recommended using enhanced chemiluminescence. A shows a control extract and an IL-1-treated extract from EL4 cells probed using the primary antibody in the presence of 1.5 mM phosphotyrosine (lanes 7 and 8) or 1.5 mM phosphoserine (lanes 9 and 10). Molecular mass standards are indicated in kilodaltons. Arrowheads indicate the proteins that consistently showed enhanced phosphorylation in response to IL-1␣ (A), PMA (B), or PMA ϩ PHA (C).

FIG. 3. Herbimycin A does not prevent the IL-1␣-or PMAstimulated degradation of IB-␣ and modifies NF-B prior to chemical dissociation with deoxycholate.
Cell cultures (5 ϫ 10 6 / ml) were pretreated either with medium or with 2 M herbimycin A (HbA) for 1 h prior to stimulation for various periods with IL-1␣ (10 ng/ml) for EL4.NOB-1 cells (A) or with PMA (100 ng/ml) for Jurkat E6.1 cells (B). Cell lysates were prepared according to the method outlined under "Experimental Procedures" and subjected to SDS-polyacrylamide gel electrophoresis. Proteins were transferred to nitrocellulose membranes and probed with anti-IB-␣ antibody according to the manufacturer's recommendations (Santa Cruz Biotechnology, Inc.). Blots were developed as recommended for enhanced chemiluminescence (Amersham International). No other proteins other than those shown were detected. Corresponding molecular mass markers are shown in kilodaltons. In C, EL4.NOB-1 cells (5 ϫ 10 6 /ml) were left untreated (lanes 1 and 2) or were treated with herbimycin A (2 M) (lanes 3 and 4) for 1 h; cytosolic extracts were prepared and left untreated (lanes 1 and 3) or were treated with deoxycholate and Nonidet P-40 (lanes 2 and 4) as described under "Experimental Procedures." Samples were then assessed for NF-B. NF-B⅐DNA complexes are presented. been well documented, treatment of cytosolic extracts prepared from unstimulated cells with detergents such a deoxycholate and Nonidet P-40 will reveal latent NF-B by chemically dissociating IB from the NF-B complex (37). Deoxycholatetreated cytosolic extracts from EL4 cells incubated with herbimycin A showed decreased DNA binding activity compared with control cells that had not been exposed to herbimycin A (Fig. 3B, compare lanes 2 and 4). This result suggested that herbimycin A was able to interfere with NF-B while complexed with IB-␣ in the cytosol and that this impaired the DNA binding capability of the p50-RelA heterodimer.
Herbimycin A Reacts Directly in Vitro with IL-1-activated NF-B-Our failure to clearly demonstrate herbimycin A-sensitive tyrosine phosphorylation changes in response to IL-1 and PMA, coupled with the lack of effect of the drug on IB-␣ degradation, suggested that signal transduction processes were not the target for herbimycin A. As herbimycin A contains a thiol-reactive benzaquinone moiety (38), we therefore determined its effect on NF-B in vitro as the p50 subunit of NF-B contains a key thiol group on cysteine 62 that, if modified substantially, decreases DNA binding (30). Initial experiments demonstrated that higher concentrations of herbimycin A were needed to inhibit DNA binding in vitro than those seen in intact cells, as shown in Fig. 4A. Incubation of extracts from IL-1treated cells with 250 M herbimycin A reduced DNA binding dramatically (Fig. 4A, lane 3). Interestingly, when 1 l of 10% 2-mercaptoethanol was incubated with herbimycin A prior to addition to the NF-B extract/DNA probe mixture, the effect of the herbimycin A was abolished, as indicated by the ability of NF-B to bind DNA as normal (Fig. 4A, lane 4). A similar effect was observed with excess DTT (data not shown). 2-Mercaptoethanol alone or a vehicle control for herbimycin A had little effect (Fig. 4A, lanes 5 and 6). A possible interpretation of this result was that a covalent reaction had occurred with a thiol group on NF-B, most probably on the p50 subunit. The neutralization of herbimycin A by thiol-containing compounds such as 2-mercaptoethanol was most likely due to a reaction with the benzaquinone moiety on herbimycin A.
We suspected that the presence of 50 mM DTT in the DNA binding reaction buffer might react with herbimycin A, thus decreasing the concentration of herbimycin A available to react with thiol group(s) on NF-B. Therefore, we probed this possibility as this could account for the higher concentration of herbimycin A required in vitro to observe an effect. This involved carrying out the binding reaction in the absence of DTT as well as using nuclear extracts prepared in DTT-free buffers. As DTT aids DNA binding, the overall binding capability of IL-1-activated NF-B was slightly diminished (Fig. 4B, lane 1) Much lower concentrations of herbimycin A were capable of inhibiting DNA binding in the absence of DTT, however (Fig.  4B, lanes 2-5). Prior incubation of herbimycin A with 2-mercaptoethanol again prevented herbimycin A from affecting NF-B (Fig. 4B, lanes 6 -9). The presence of 2-mercaptoethanol generally increased DNA binding in all samples, as expected (Fig. 4B, lanes 6 -9), with DTT having a similar effect (data not shown). High concentrations of vehicle (dimethyl sulfoxide) equivalent to those used in experiments with 250 M herbimycin A had a modest inhibitory effect on the fastest migrating complex (Fig. 4B, lane 10).
Herbimycin A Reacts with Human Recombinant p50, but Not with the C62S Mutant-As we suspected that p50 was a target for herbimycin A in the NF-B complex, we next carried out in vitro experiments with recombinant human p50 and attempted to demonstrate the presence of p50 in nuclear extracts from IL-1-and PMA-treated cells. Fig. 5A (lanes 2 and 3) demonstrates that the ability of p50 to bind the NF-B consensus sequence was blocked by herbimycin A at concentrations from 10 to 100 M, indicating that herbimycin A could directly modify p50. We next tested its effect on a p50 mutant in which cysteine 62 was substituted with serine. This is the only cysteine in p50 that is important for DNA binding (30). The mutant protein has a much lower affinity for the B site (30), and 200 ng of C62S mutant were used. Herbimycin A did not inhibit the ability of this mutant to bind DNA at any concentration tested (Fig. 5B).
We next demonstrated the presence of p50 in IL-1-and PMA-treated cell extracts. Fig. 6A (lane 4) shows that nuclear extracts from IL-1-stimulated EL4.NOB-1 cells contained large amounts of p50, as indicated by specific antibodies to p50 causing a supershift in the DNA probe. p50 was also detected in untreated samples (Fig. 6A, lane 3). Much lower amounts of RelA were evident in the samples (Fig. 6A, lane 6), the gels requiring prolonged exposure to reveal this subunit, hence the high levels of NF-B apparent in samples prepared from unstimulated cells shown in this figure. No evidence for c-Rel in . Binding buffer and poly(dI⅐dC) were then added, followed by a 30-min incubation with radiolabeled NF-B probe. Extracts were analyzed for NF-B DNA binding activity as described under "Experimental Procedures." A control extract from cells not treated with IL-1 is also shown (lane 1). B, NF-B⅐DNA complexes are presented. The protocol employed in A was repeated with slight modifications using nuclear extracts from EL4.NOB-1 cells treated with IL-1 (10 ng/ml, 1 h) prepared in DTT-free buffers. Lane 1 shows samples from IL-1-treated cells subsequently left untreated. The indicated concentrations of herbimycin A (lanes 2-5) or of the mixture of herbimycin A and 2-mercaptoethanol (lanes 6 -9) were added to extracts as indicated and left for 30 min at room temperature. Binding buffer with poly(dI⅐dC) but without DTT was then added and assayed for NF-B. the DNA-binding complexes was obtained in nuclear extracts of either untreated or IL-1-treated EL4 cells (Fig. 6A, lanes 7 and  8). Similarly, nuclear extracts from EL4 cells treated with PMA for 24 h showed strong supershifts with anti-p50 and anti-RelA antibodies (Fig. 6B, lanes 4 and 6, respectively). As in Fig. 5A, untreated EL4 cells contained p50 (Fig. 6B, lane 3). Interestingly, samples from PMA-stimulated EL4 cells contained much higher levels of RelA (Fig. 6B, lane 6) than IL-1-treated samples, indicating that PMA probably induced RelA expression in these cells. Using nuclear extracts from Jurkat cells treated for 1 h with PMA, it again appeared that the cells contained high levels of p50 (Fig. 6C, compare lanes 1 and 2). RelA was also detected (Fig. 6C, lane 3), and the decrease in binding observed in the presence of anti-c-Rel (lane 4) suggested that c-Rel was also present. The marked effect of the antibodies on the extracts indicted that the major protein complexes binding to the NF-B motif were p50 and, to a lesser extent, RelA. Taken together, these results indicated that herbimycin A could directly modify p50 most probably at cysteine 62, and the presence of large amounts of p50 in nuclear extracts suggested that this would be a likely target for herbimycin A in cells.
Herbimycin A Inhibits IL-1-and PMA-induced IL-2 Production in EL4.NOB-1 Cells-Finally, we determined the effect of herbimycin A on IL-2 production in EL4.NOB-1 cells. NF-B has recently been shown to be the sole transcription factor required for increased expression of the murine IL-2 gene in response to IL-1 (39). Pretreatment of EL4 cells with increasing concentrations of herbimycin A for 1 h dramatically inhibited the induction of IL-2 in response to IL-1 (Fig. 7A). A concentration of 2 M decreased the IL-1 response from 0.9 Ϯ 0.2 to Ͻ0.1 ng/ml. The ability of PMA to induce IL-2 was similarly inhibited, as shown in Fig. 7B. PMA is a much more powerful inducer of IL-2 than IL-1 at these concentrations, causing ϳ50-fold more IL-2. In spite of this, herbimycin A could also block the PMA response, reducing IL-2 production from 50 Ϯ 10 ng/ml with PMA alone to 3.25 Ϯ 0.4 ng/ml in the presence of 2 M herbimycin A. These data suggest that modification of NF-B by herbimycin A is likely to have consequences for NF-B-driven gene expression. DISCUSSION The initial aim of this study was to explore the involvement of tyrosine kinase(s) in the activation of NF-B by IL-1 and PMA in T lymphocytes using the tyrosine kinase inhibitor herbimycin A. We demonstrated that herbimycin A inhibited the activation of NF-B by both IL-1 and PMA in EL4.NOB-1 thymoma cells and PMA in Jurkat E6.1 lymphoma cells. The target for herbimycin A, however, appeared not to be tyrosine kinases or indeed signals leading to IB-␣ dissociation and degradation, but the NF-B complex itself. Our results are consistent with a model involving covalent modification by herbimycin A of cysteine 62 on the p50 subunit of NF-B.
We initially suspected that herbimycin A was affecting something other than tyrosine kinases because of our observation that only a 1-h treatment with herbimycin A was required to observe inhibition. Previous studies have shown that optimal inhibition of Src kinase by the drug requires prolonged exposure as part of the mechanism of inhibition involves enhanced degradation of the enzyme (40,41). In addition, the observation that herbimycin A inhibited the activation of NF-B by PMA was unexpected as PMA is known to have weak effects on tyrosine phosphorylation in cells (42)(43)(44), its major cellular target being protein kinase C. We clearly demonstrated that herbimycin A could inhibit the activation of NF-B by PMA, suggesting either that tyrosine kinases were important for PMA action or, alternatively, that herbimycin A was inhibiting something other than tyrosine kinases. Only minor changes in tyrosine phosphorylation were observed in response to IL1 or PMA, with herbimycin A having a negligible inhibitory effect. Our failure to detect major changes in tyrosine phosphorylation in EL4 cells in response to IL-1 is in contrast to other reports showing such changes in K562 cells (45), human A375-C6 melanoma cells (21), and Th2 cells (22). The major change in tyrosine phosphorylation demonstrated in these studies occurred in the molecular mass range 40 -45 kDa. It is likely that these corresponded to p42/p44 MAP kinases as, in another study, it was demonstrated that IL-1 increased the phosphorylation of p42/p44 MAP kinases on tyrosine, threonine, and serine residues (46). The phosphorylations were likely to be due to the activation of MAP kinase kinase, which is a dual specificity tyrosine/serine-threonine kinase that phosphorylates MAP kinase (47) and is activated by IL-1 in fibroblasts and KB epidermal cells (48). IL-1 has been shown to be a poor activator of MAP kinases in EL4 cells (46), which is consistent with our failure to detect major changes in tyrosine phosphorylation in the molecular mass range for MAP kinases in these cells. The area of IL-1 signal transduction is controversial, with no clear pathway emerging despite intense effort (49 -51). A consensus on the activation of serine/threonine kinases belonging to the MAP kinase family has recently emerged, however (52,53); and any tyrosine kinase changes that occur in response to IL-1 are likely to be on kinases in these pathways rather than there being a general increase in tyrosine phosphorylation. The involvement of MAP kinases in NF-B activation is ill defined.
Previous workers have suggested a role for tyrosine kinases in the activation of NF-B by IL-1. The evidence, however, has been largely circumstantial, with data being presented for tyrosine kinase changes in response to IL-1 that were inhibited by herbimycin A or another tyrosine kinase inhibitor, genistein (20,21). These results were then used as evidence for tyrosine kinase involvement in the activation of NF-B, which also proved susceptible to inhibition by the compounds. The precise identity of the tyrosine kinase involved or indeed its substrate were not determined, although recent evidence demonstrating that overexpression of v-src leads to NF-B activation has suggested a role for a Src-like tyrosine kinase (26). Both of the inhibitor studies suggested, however, that the putative tyrosine kinase involved in IL-1 action must occur upstream of IB release and degradation. Our failure to demonstrate inhibition of IB degradation, coupled with the effect on latent NF-B, questioned signal transduction processes being the target for herbimycin A and indicated that NF-B complexed to IB was susceptible to herbimycin A inhibition.
Because of these findings, our efforts next turned to NF-B itself. The mechanism of action of herbimycin A with regard to tyrosine kinases involves the covalent modification of a thiol group on target kinases such as pp60 v-src (24,25,38). The evidence for this initially came from the observation that thiolcontaining compounds such as 2-mercaptoethanol prevented the inhibitory effect of herbimycin A on pp60 v-src (25) and, more recently, from a direct demonstration that herbimycin A covalently modifies both Src and Abl (24). The chemical moiety on herbimycin A that has been implicated in this modification is a benzaquinone group that is highly thiol-reactive (38). Because the p50 subunit of NF-B has been shown to contain a key thiol on cysteine 62 that, if mutated to serine or oxidized, greatly reduces DNA binding (30), we suspected that this may have been the target for herbimycin A. This was supported by experiments in which herbimycin A proved inhibitory when added directly to nuclear extracts from activated cells, an effect blocked by first treating herbimycin A with 2-mercaptoethanol or DTT. In these in vitro experiments, higher concentrations of herbimycin A were needed than in intact cells. Similar differences have been shown by other workers, whereby experiments demonstrating a direct effect on tyrosine kinases in vitro re-  (24,25,38), whereas inhibition of tyrosine kinases in intact cells generally involves submicromolar concentrations (40,54). It is possible that herbimycin A becomes concentrated inside cells. Comparing in vitro experiments with those in vivo is somewhat difficult, however, as conditions inside the cell under which herbimycin A reacts with thiol groups will differ from those pertaining in vitro. This has also been postulated by Fukazawa et al. (24) as an explanation for concentration differences with regard to tyrosine kinases. They suggest that the target kinases in their native environment may be more accessible to the drug than in vitro. This may also apply to p50.
We also found that the ability of recombinant human p50 to bind DNA probe containing the NF-B consensus sequence was blocked, while the C62S mutant was unaffected. p50 has three cysteines at positions 62, 119, and 273, which would be potential targets for herbimycin A. Mutagenesis studies have shown that only cysteine 62 is important for DNA binding since the C62S mutation decreased the affinity for the B site 10-fold (30). We found that the C62S mutant of p50 was insensitive to herbimycin A. These in vitro data, coupled with the detection of large amounts of p50 in EL4 and Jurkat cells, led us to conclude that the most likely target for herbimycin A in EL4 and Jurkat cells was cysteine 62 on p50. We were unable to carry out experiments that would demonstrate adduct formation between herbimycin A and p50 in intact cells. Previous experiments on herbimycin A and tyrosine kinases have used metabolically labeled herbimycin A of low specific activity and immune complexes from v-src-transformed NIH3T3 fibroblasts or K562 cells, which express high levels of p210 bcr-abl (25). Analogous experiments on p50 in EL4 or Jurkat cells would be difficult to perform. The amount of p50 in the cells is likely to be very low compared with the tyrosine kinases in the aforementioned study, and the low specific activity of herbimycin A would further lower the detection limit of this approach.
Finally, we found that herbimycin A blocked IL-1-and PMAinduced IL-2 in EL4 cells over a concentration range identical to that which inhibited NF-B activation. The ability of herbimycin A to interfere with p50 is therefore likely to have consequences for NF-B-driven gene expression as NF-B has been shown to have a central role in the induction of IL-2 by IL-1 (39).
In conclusion, the results presented here therefore urge caution in the use of herbimycin A as a specific tyrosine kinase inhibitor. They further suggest that the ability to selectively modify p50 may have potential as a means of inhibiting NF-B-dependent gene transcription.