Signal-dependent and -independent Degradation of Free and NF- k B-bound I k B a *

A family of inhibitory I k B molecules regulates the activation of the transcription factor NF- k B. One member of the I k B family, I k B a , plays a major role in the rapid signal-induced activation of NF- k B. I k B a itself is tran-scriptionally regulated by NF- k B allowing for a tight autoregulatory loop that is both sensitive to and rapidly influenced by NF- k B activating stimuli. For this pathway to remain primed both for rapid activation of NF- k B in the presence of signal and then to suppress NF- k B activation once that signal is removed, I k B a must be exquisitely regulated. The regulation of I k B a is mainly accomplished through phosphorylation, ubiquitination, and subsequent degradation. The mechanism(s) that regulate I k B a degradation needs to be able to target I k B a for degradation in both its NF- k B bound and free states in the cell. In this study, we utilize a full-length I k B a mutant that is unable to associate to RelA/p65. We show that the signal-induced I k B kinase (IKK) phosphorylation sites on I k B a can only significantly influence the regulation of signal-dependent but not signal-inde-pendent turnover of I k B a . We also demonstrate that the constitutive carboxyl-terminal casein kinase II phosphorylation sites are necessary for the proper regulation of both signal-dependent and -independent turnover of I k B a . These findings the cell. We also show that distinct phosphorylation sites can directly influence the efficiency of ubiquitination and subsequent degradation of free and NF- k B-associated I k B a in the presence or absence of stimuli. These results demonstrate the complexity of I k B a regulation necessary to ensure that NF- k B is rapidly and specifically activated by a diverse group of stimuli. and 5 m g of plasmid the vesicular stomatitis (VSVg) envelope downstream of the cytomegalovirus promotor-enhancer. 48 h, the was Polybrene added (8 m g/ml), and it was immediately used for infection. Each infection was performed by adding 3 ml of 1:1000 diluted virus supernatant to approximately 10 5 cells for 6 h. The cells were allowed to for 48 h and the infected cells were selected using 6 m g/ml puromycin (Calbiochem). In order to get equivalent levels of expression in the mutC and 3236mutC stable pools it was necessary to perform three rounds of infection and selection as described above. Blot B a / 2 MEF stable pools were with ng/ml TNF a for given some cases the MEFs were with m M b -lactone, for MEFs were also treated with b for the indicated the Cells were then washed 2 times with ice-cold phosphate-buffered saline and frozen on dry ice. Cytoplasmic extracts were made as described previously and separated on 10 or 12% SDS-polyacrylamide gels and transferred to 0.2- m m nitro- cellulose membranes (Schleicher & Membranes were probed with antibody, diluted 1:1,000 in 0.2% Tween-phosphate-buffered sa- line containing 5% nonfat milk, fo r 4 h at 4 °C.Horseradish peroxidase-conjugated donkey anti-rabbit sera (Amersham Pharmacia Biotech) 1:3,000 and incubated with the membranes fo r 1 h atroom temperature. Specific Life Products Inc.) by

The transcription factor NF-B is an important regulator of genes involved in immune and inflammatory responses, apoptosis, and cell proliferation (1,2). The NF-B/Rel family of transcription factors are a family of proteins that homo-and heterodimerize through a conserved Rel homology domain that consists of approximately 300 amino acids. The Rel homology domain of NF-B is responsible for homo-and heterodimerization, DNA binding activity, and nuclear localization (3)(4)(5). A large number of stimuli can cause NF-B to translocate from the cytoplasm to the nucleus, and activate target gene transcription. Stimuli that can activate NF-B include proinflammatory cytokines, bacterial lipopolysaccharide, phorbol esters, okadaic acid, and viral infection (2, 6 -8).
A group of inhibitory proteins belonging to the IB family regulate NF-B activation by sequestering NF-B in the cytoplasm. IB exerts its inhibitory effects by associating with the Rel homology domain of NF-B proteins, effectively masking their nuclear localization signal (9 -12). Although there are a number of IB proteins, IB␣ is the primary regulator of rapid signal induced activation of NF-B. Upon stimulation by a proinflammatory cytokine such as TNF␣, 1 a signaling cascade is initiated that results in the activation of the IB kinases IKK1 and IKK2 (13)(14)(15)(16)(17). This leads to the rapid phosphorylation of IB␣ at the signal-induced phosphorylation sites, serine 32 and serine 36 (18 -21). Once phosphorylated, IB␣ is polyubiquitinated by the Ubc5/E3RS IB ubiquitination enzyme pair (22,23) on lysine 21 and lysine 22 (24 -26). Polyubiquitinated IB␣ is degraded by the 26 S proteasome, thus exposing NF-B's nuclear localization signal and allowing NF-B to translocate to the nucleus (27)(28)(29). Once in the nucleus NF-B activates transcription of target genes including IB␣ (6,30). When the NF-B inducing signal is removed the newly synthesized IB␣ can suppress NF-B activity by preventing it from binding to the genomic DNA, and sequestering NF-B in the cytoplasm (31)(32)(33).
The regulation of IB␣ is carried out mainly through phosphorylation. Several phosphorylation sites have been identified on IB␣ and they include the signal-induced IKK phosphorylation sites located at serine 32 and serine 36 (18 -21), the constitutive CKII phosphorylation sites located in the carboxylterminal PEST domain (34,35), the protein kinase C site located in Ankyrin repeat 6 (ank6) (36,37), and a tyrosine phosphorylation site that can cause the dissociation but not the degradation of IB␣ from NF-B in Jurkat T cells (38). Although extensive analysis of the signal-dependent degradation of IB␣ has been done, little distinction has been made between the signal-dependent and -independent degradation of free and NF-B-bound IB␣. Therefore, we undertook a study to examine the role of various phosphorylation sites on IB␣ to determine their effects on both the signal-dependent and -independent degradation of free and NF-B-bound IB␣.
In this study, we characterize a full-length IB␣ mutant that is unable to associate with NF-B (designated as mutC) and exists as a free molecule in the cell. We also show that distinct phosphorylation sites can directly influence the efficiency of ubiquitination and subsequent degradation of free and NF-Bassociated IB␣ in the presence or absence of stimuli. These results demonstrate the complexity of IB␣ regulation necessary to ensure that NF-B is rapidly and specifically activated by a diverse group of stimuli. * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Cell Culture-HeLa cells, human embryonic kidney 293 cells, and IB␣ Ϫ/Ϫ mouse embryo fibroblasts (MEF) (39) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum in an atmosphere of 10% CO 2 , at 37°C. 293 cells stably expressing Moloney gag and pol (293gp) were maintained as above and selection was maintained with blastacidin (20 g/ml). TNF␣ and ␤-lactone were used at final concentrations of 10 ng/ml and 10 M, respectively, unless otherwise noted. Cyclohexamide was used at a final concentration of 75 g/ml.
In vitro transcription-translation was performed using the T7 promoter contained in the PCMX-PL1 vectors. Wild-type and mutant IB␣ proteins labeled with [ 35 S]methionine were produced using a wheat germ extract TNT kit as instructed by the manufacturer (Promega). 60 l of the TNT reaction was used for each sample that was tested in the in vitro ubiquitination assay.
Transfections-293 cell transfections were performed by the calcium phosphate method (42). Briefly, DNA was mixed in 0.625 M CaCl 2 and then added to an equal volume of 2 ϫ HEPES buffer saline solution (560 mM NaCl, 50 mM HEPES, 1.5 mM Na 2 HPO 4 , pH 7.1). The mixture was added to approximately 10 6 cells in 10 ml of medium and incubated at 37°C for 6 h (h) in 3% CO 2 . The medium was then changed and the cells were incubated for 48 h at 37°C, in 10% CO 2 .
Production of Virus and Stable Pools-Virus production for infection of the IB␣ Ϫ/Ϫ MEF cell line was performed by calcium phosphate co-transfection of the 293gp packaging cell line with 20 g of the pCLBabepuro retroviral vector, containing cDNAs for wild-type (wt) IB␣ or the various IB␣ mutants, and 5 g of the pMDG plasmid containing the vesicular stomatitis virus (VSVg) envelope downstream of the cytomegalovirus promotor-enhancer. After 48 h, the media was removed, filtered, Polybrene added (8 g/ml), and it was immediately used for infection. Each infection was performed by adding 3 ml of 1:1000 diluted virus supernatant to approximately 10 5 cells for 6 h. The cells were allowed to expand for 48 h and the infected cells were selected using 6 g/ml puromycin (Calbiochem). In order to get equivalent levels of expression in the mutC and 3236mutC stable pools it was necessary to perform three rounds of infection and selection as described above.
Cell Stimulation and Western Blot Analysis-The IB␣ Ϫ/Ϫ MEF stable pools were stimulated with 10 ng/ml TNF␣ for the given time points. In some cases the MEFs were pretreated with 10 M ␤-lactone, for 1 h, prior to stimulation. MEFs were also treated with ␤-lactone for the indicated times in the absence of signal. Cells were then washed 2 times with ice-cold phosphate-buffered saline and frozen on dry ice. Cytoplasmic extracts were made as described previously and separated on 10 or 12% SDS-polyacrylamide gels and transferred to 0.2-m nitrocellulose membranes (Schleicher & Schuell). Membranes were probed with antibody, diluted 1:1,000 in 0.2% Tween-phosphate-buffered saline containing 5% nonfat milk, for 4 h at 4°C. Horseradish peroxidaseconjugated donkey anti-rabbit sera (Amersham Pharmacia Biotech) was diluted 1:3,000 and incubated with the membranes for 1 h at room temperature. Specific bands were then resolved by using a Renaissance detection kit (NEN Life Science Products Inc.) as instructed by the manufacturer.
Electrophoretic Mobility Shift Assay-Nuclear extracts were made by the micropreparation technique (43), and gel shift analysis was performed as described previously (44). Briefly, 5 g of nuclear protein extract was incubated with 0.5 g of poly(dI-dC), on ice, for 20 min to block nonspecific DNA binding activity. A 32 P-labeled oligonucleotide containing the HIV-1 long terminal repeat B site was then added, and the mixture was incubated at room temperature for 30 min. The resulting complexes were then resolved on a 4% acrylamide gel, exposed to a PhosphorImager (Molecular Dynamics), and band intensities quantitated with ImageQuant software.
In Vitro Ubiquitination Assay-HeLa cytoplasmic extracts, used in the in vitro ubiquitination assay, were made by lysis in hypotonic buffer (buffer A) containing 10 mM HEPES (pH 7.9), 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM dithiothreitol, and protease inhibitors (0.1 mM phenylmethylsulfonyl fluoride and 10 g/ml aprotinin). Lysates were cleared by centrifugation at 14,000 rpm, for 15 min, at 4°C. Lysates were then concentrated with a Centricon 3 concentrator (Amicon) as instructed by the manufacturer. Protein concentration was determined using the Bio-Rad protein assay system.
In vitro ubiquitination assays were carried out as described (29) with the following changes. Briefly, in vitro 35 S-labeled IB␣ was incubated with 5 mg/ml HeLa extract for 15 min at 4°C. A reaction mixture containing 50 mM Tris (pH 7.5), 5 mM MgCl 2 , 2 mM ATP, 10 mM creatine phosphate, 3.5 units/ml creatine kinase, 0.6 units/ml inorganic pyrophosphatase, 1 mg/ml ubiquitin, 3 M okadaic acid, 3 M ubiquitin aldehyde, and 10 M ␤-lactone was then added and the reaction was incubated at 37°C for 90 min. 50 mM Tris (pH 7.5) was substituted for the reaction mixture in control samples.
After the incubation, RIPA (100 mM NaCl, 20 mM Tris, pH 8.0, 0.5% Nonidet P-40) containing 5 mM N-ethylmaleimide was added and the samples were subjected to immunoprecipitation (IP) by antibodies directed against IB␣ or RelA/p65. IPs were carried out with 1.5 g of the appropriate antibody for 1 h at 4°C. Immunocomplexes were precipitated by incubation with protein A-Sepharose (Amersham Pharmacia Biotech) for 1 h at 4°C. The pellets were washed in RIPA buffer, SDS sample buffer was added, and the pellets were boiled for 5 min prior to SDS-polyacrylamide gel electrophoresis on 10% gels. Following electrophoresis, the gels were fixed in glacial acetic acid, amplified with 2,5-diphenyloxazole, rinsed with H 2 O, dried, and exposed to a Phos-phorImager for quantification.
Metabolic Labeling-6-cm plates of the IB␣ Ϫ/Ϫ MEF stable pools were washed twice in phosphate-buffered saline and incubated for 1 h in 2 ml of cystine-and methionine-free Dulbecco's modified Eagle's medium supplemented with 10% dialyzed fetal bovine serum (Life Technologies, Inc.). Labeling was performed for 3 h with 0.5 mCi of [ 35 S]methionine (NEN Life Science Products Inc.) per ml. Cells were then washed three times with complete medium and then chased for the given time points in complete medium. After each time point the cells were washed twice with ice-cold phosphate-buffered saline and cell pellets were frozen on dry ice for later manipulation. Cell pellets were thawed on ice and whole cell lysis was performed by adding 500 l of RIPA (20 mM Tris, pH 8.0, 100 mM NaCl, 0.2% sodium deoxycholate, 0.2% Nonidet P-40, 0.2% Triton X-100) containing the protease inhibitors aprotinin (10 g/ml; Sigma) and phenylmethylsulfonyl fluoride (1 mM; Sigma). DNA was sheared by 10 passes through a 20-gauge needle and lysates were cleared by centrifugation at 14,000 rpm for 15 min, at 4°C. The cleared lysates were measured for trichloroacetic acid-precipitable counts and the lysates were normalized for labeling efficiency. Normalized lysates were precleared with protein A-Sepharose for 1 h at 4°C and then incubated with RelA/p65 antiserum plus protein A-Sepharose for 4 h at 4°C. To reduce the background, coimmunoprecipitations that were done with the RelA/p65 antiserum were subjected to a second round of immunoprecipitation. The RelA/p65 antibody protein A-Sepharose pellets were boiled for 10 min in RIPA containing 100 g of bovine serum albumin/ml and 0.5% SDS. The SDS was diluted to 0.1% with RIPA buffer that contained 100 g of bovine serum albumin. A second IP was then done by adding IB␣ antiserum and protein A-Sepharose to isolate the RelA/p65-associated IB␣. To isolate the free IB␣, the supernatant from the original Rela/p65 co-IP was incubated with IB␣ antiserum and protein A-Sepharose. After immunoprecipitation, the pellets were washed in RIPA buffer, resuspended in SDS sample buffer, boiled for 5 min, and the eluted proteins were separated on a 12% SDS-polyacrylamide gel.

RESULTS
Signal-dependent Degradation of IB␣-To investigate the signal-dependent degradation of IB␣ in vivo, stable pools of IB␣ Ϫ/Ϫ MEFs (39) were generated by infecting naive IB␣ Ϫ/Ϫ MEFs with recombinant retroviral vectors containing either wild type (wt) murine IB␣, one of the IB␣ mutants ( Fig.  1 schematically describes IB␣ mutations), or GFP. The mutants described in Fig. 1 represent mutations of known and possible sites of serine and threonine phosphorylation. These mutants were chosen for this study because they would aid in the further elucidation of the role of serine and threonine phosphorylation in the signal-dependent and -independent degradation mechanisms of free and NF-B-bound IB␣. The IB␣ Ϫ/Ϫ MEF infections were carried out at a multiplicity of infection of much less than one to ensure that on average, after selection, each cell of a stable pool would have only a single copy of the integrated recombinant retroviral vector. This kept expression levels low and as close to endogenous expression levels as possible. Fig. 2A compares the relative expression levels of endogenous IB␣ in wt MEF cells to that of the transduced IB␣ Ϫ/Ϫ stable pools. It can be seen that the levels of the IB␣ proteins in the transduced cells are within 1-2-fold of the endogenous IB␣ (Fig. 2, lane 8, ϩ/ϩ).
TNF␣ stimulation of the IB␣ Ϫ/Ϫ MEF stable pools in the presence of the translational inhibitor cyclohexamide show that, compared with wild type IB␣, the 3236 and M mutants were significantly more stable (Fig. 2B). The M mutant shows no change in expression over the time course while the 3236 mutant starts to show a slight reduction at the 60-min time point. The mutF mutation gives IB␣ partial resistance to TNF␣-dependent degradation but is less stable than the 3236 and M mutants. The mutC mutant shows no increase in stability and, in fact, is significantly less stable than wt IB␣. The 3236mutC mutation confers an increase in the stability of mutC containing IB␣, but is significantly more unstable than the 3236 mutation by itself.
The rate of IB␣ degradation shown in Fig. 2B was further examined by measuring the NF-B DNA binding activity following TNF␣ stimulation. The mutF mutation shows an approximate 50% reduction in DNA binding activity when compared with wt IB␣ (Fig. 2C). The 3236 and M mutations reduce the gel shift activity to what is observed in unstimulated control cells. Interestingly, cell pools containing either the mutC or 3236mutC mutations show no significant reduction in NF-B DNA binding activity. This is most likely due to the fact that the mutC mutation, which is a substitution of all five serine and threonine residues in the ank6 region of IB␣ to alanine, disrupts the association of IB␣ to NF-B. Fig. 3, A (in vitro) and B (in vivo), show that both mutC and 3236mutC do not associate with RelA/p65. It was recently shown that the ank6 region of IB␣ makes a critical contact to NF-B (11,12). Since the mutC mutation disrupts IB␣s association to NF-B and blocks any potential phosphorylation of ank6, it is possible that phosphorylation of ank6 plays a critical role in IB␣'s association to NF-B. We conclude that (i) IKK phosphorylation is necessary for signal-dependent degradation of both free and NF-B-associated IB␣, and (ii) that CKII phosphorylation is necessary for efficient signal-dependent degradation of NF-B-associated IB␣.
Next, in vitro ubiquitination assays were performed to determine the signal-induced ubiquitination profiles of both free and NF-B-associated IB␣ mutants. A representative gel is shown for each experiment. Quantitation was done on a Phos-phorImager, using ImageQuant software, and the resulting histograms display the percentage of the total labeled IB␣ in the reaction that was shifted. By graphing the percentage of the total IB␣ shifted we were able to correct for any loading differences between samples. Therefore, the histograms give a more accurate representation of the respective levels of ubiquitination on the various IB␣ molecules. When examining a pool of both RelA/p65-associated and free IB␣ it appears that IB␣ containing the mutF mutation is ubiquitinated slightly, but reproducibly, more efficiently than wt IB␣ (see Fig. 4A, lanes 1 and 2). The level of ubiquitination of the other IB␣ mutants in Fig. 4A is consistent with the rate of degradation and the gel shift results presented in Fig. 2, B and C. The 3236 and M mutants have levels of ubiquitination just above background (Fig. 4A, lanes 3 and 4).
The data with the mutC mutants (Fig. 4A, lanes 5 and 6) represents the levels of ubiquitination on free IB␣ molecules, since mutC does not associate with NF-B (Fig. 3, A and B). MutC alone is ubiquitinated very efficiently (lane 5), even better than wt (Fig. 4A, lane 5, compared with lane 1, respectively). The mutC3236 mutant shows a reduction in the level of ubiquitination compared with mutC alone, but the level of ubiquitination of 3236mutC is significantly higher than IB␣ containing just the 3236 mutation (Fig. 4A, lane 6, compared with lane 3, respectively). These data demonstrate that free IB␣ does undergo signal-induced phosphorylation and ubiquitination and that serine 32 and serine 36 play a significant role in this process. Fig. 4B displays the ubiquitination of the RelA/p65 associated pool of IB␣. The mutF mutation reduces the amount of signal-dependent ubiquitination below the level seen on wt IB␣ when it is in association with NF-B (Fig. 4B, lanes 1 and  2). This reduction in ubiquitination by mutF of only NF-Bassociated IB␣ explains why IB␣ is partially stabilized and why a reduction in the DNA binding activity of NF-B is observed in the IB␣ Ϫ/Ϫ MEF mutF stable pool (Fig. 2, B and  C).
In contrast to the NF-B-bound IB␣ (Fig. 4B, lanes 1 and 2), the signal-induced ubiquitination of only the free pool of IB␣ reveals that mutF is ubiquitinated slightly more efficiently than the wt (Fig. 4C, lanes 1 and 2). The 3236 and M mutants give large reductions in the amount of ubiquitination observed (Fig. 4C, lanes 4 and 5). Ubiquitination of mutC and 3236mutC, in the in vitro ubiquitination assay, could only be observed by direct anti-IB␣ immunoprecipitation (Fig. 4A, lanes 5 and 6) because these mutants do not associate to RelA (Fig. 3A). The in vitro ubiquitination assay data (Fig. 4, A-C) correlates well with the stability and gel shift observations made in the IB␣ Ϫ/Ϫ MEF stable pools (Fig. 2, B and C).
The in vitro ubiquitination results were confirmed by performing in vivo ubiquitination assays in the stably transduced IB␣ Ϫ/Ϫ MEF stable pools. The IB␣ Ϫ/Ϫ MEF stable pools were first treated with ␤-lactone, a potent and specific inhibitor of the proteasome, and then stimulated with TNF␣ (Fig. 4D). The ubiquitination patterns obtained in this experiment are similar to the corresponding in the in vitro assays. Wt, mutF, 3236, and M represent NF-B associated IB␣ (Fig. 4D, lanes  1-4, respectively) and give ubiquitination profiles that match those obtained for RelA/p65-associated IB␣ in the in vitro assay (Fig. 4B, lanes 1-4). Wt IB␣ is efficiently ubiquitinated in the presence of signal, mutF causes a decrease in ubiquiti-nation, and 3236 and M reduce ubiquitination to almost background levels (Fig. 4D, lanes 1, 2, 3, and 4, respectively).
MutC and 3236mutC represent free IB␣ in the IB␣ Ϫ/Ϫ MEF stable pools (Fig. 3B), and they also give ubiquitination patterns in vivo (Fig. 4D, lanes 5 and 6) that correlate with FIG. 2. Signal-dependent activation of IB␣ ؊/؊ MEF stable pools. A, IB␣ immunoblot (IB) of the relative expression levels of the various IB␣ Ϫ/Ϫ stable pools, an IB␣ Ϫ/Ϫ GFP stable pool, and endogenous IB␣ expression in IB␣ ϩ/ϩ MEFs. Quantitation of the relative expression level of IB␣ in each cell line was done using NIH Image 1.62 software. The results are given graphically in arbitrary units. B, IB␣ Ϫ/Ϫ MEFs were infected with a recombinant retroviral vector containing wild type murine IB␣ or one of the mutants. The stable pools were treated with cyclohexamide prior to being stimulated with TNF␣ for 0, 5, 15, or 60 min. Following stimulation the cells were harvested, cytoplasmic extracts made, and Western blot analysis was performed with anti-IB␣ sera. C, gel shift analysis of IB␣ Ϫ/Ϫ MEF stable pools stimulated with TNF␣ for 60 min. Three independent experiments were done. Quantification was done on a PhosphorImager, using Im-ageQuant software. The values obtained were averaged and the standard deviations were calculated.

FIG. 3. Association of wild type and mutC containing IB␣ mutants to RelA/p65 in vitro and in vivo.
A, in vitro association assay. 35 S-Labeled IB␣, mutC, or 3236mutC were mixed with uninduced cytoplasmic extract as described under "Experimental Procedures." The sample was then immunoprecipitated with anti-IB␣ or anti-RelA. The IB␣ IP shows input amount of labeled IB␣ species. The RelA IP shows the amount of the different IB␣ species that associated to the RelA/p65 present in the cytoplasmic extract. B, in vivo association assay. The wild type, mutC, and 3236mutC IB␣ Ϫ/Ϫ stable pools were labeled with [ 35 S]methionine for 3 h. The cells were washed and then lysed in RIPA buffer. A RelA IP was performed and then the precipitated material was subjected to a second round of IP with anti-IB␣ sera to reduce the background. This was undertaken to examine at RelA/p65-associated IB␣ in these cell lines. IB␣ IPs were also done to show that IB␣ is being expressed and labeled in the mutC and 3236mutC stable pools. those obtained in vitro (Fig. 4A, lanes 5 and 6). MutC alone is ubiquitinated very efficiently and the 3236mutC mutation gives a reduction in the amount of ubiquitination but the level of ubiquitination seen is significantly higher than that seen with just the 3236 mutant (Fig. 4D, compare lanes 3 and 6). These data show that free IB␣ is ubiquitinated in the presence of signal even when it is lacking its signal-dependent phosphorylation sites. It is likely that signal-independent degradation mechanisms that are involved in removing free IB␣ from the cell are contributing to the ubiquitination of the mutC containing mutants in these experiments.
From the in vitro and in vivo ubiquitination data we conclude that phosphorylation of serine 32 and serine 36 by the IKK is necessary for efficient signal-dependent ubiquitination and degradation of free and NF-B-associated IB␣. We also conclude that the constitutive CKII phosphorylation of the carboxyl terminus of IB␣ enhances the extent of ubiquitination of free IB␣ (Fig. 4C, lane 2), but reduces the ubiquitination of NF-B associated IB␣ (Fig. 4, B and D, lane 2).
Signal-independent Turnover of IB␣-To investigate whether or not the constitutive or signal-dependent phosphorylation sites are involved in the signal-independent turnover of NF-B-associated IB␣, pulse-chase experiments were performed on the wt, mutF, and 3236 IB␣ Ϫ/Ϫ MEF stable pools (Fig. 5, A-C). Compared with the half-life of wt IB␣, the mutF mutation gives a 2-fold increase in IB␣s half-life. The 3236 mutation has no effect on basal turnover, resulting in a half-life very similar to that of wt IB␣ (compare A and C). These results demonstrate that only the carboxyl-terminal CKII phosphorylation sites play a significant role in the basal turnover of NF-B-associated IB␣. The inducible phosphorylation sites (serine 32 and 36) have no effect on the basal turnover of NF-B-bound IB␣.
Several studies have been performed to evaluate the role of phosphorylation in the degradation of free IB␣ in the absence of stimulation (35,45,46). It has been shown that the mutF mutation or the removal of the carboxyl terminus of IB␣ will increase the stability of free IB␣ by approximately 2-fold (35,46). Here we look at the effects of the 3236 mutation on the basal turnover of free IB␣. Unstimulated IB␣ Ϫ/Ϫ MEF stable pools expressing either the mutC mutant or the 3236mutC mutant (Fig. 6, A  and B) were analyzed. As has been shown (Fig. 3B) mutC and 3236mutC do not associate to RelA/p65, thus all the IB␣ is in a free state in the IB␣ Ϫ/Ϫ MEF stable pools. Free IB␣ is at least 5 times more rapidly degraded in the absence of signal when compared with NF-B-bound IB␣ (compare Fig. 5A and 6A). Interestingly, the 3236 mutation in the mutC background increases the rate of signal-independent turnover of free IB␣ by about 3-fold (Fig. 6, A and B).
To determine whether or not ubiquitination plays a role in the basal turnover of NF-B-associated IB␣, the IB␣ Ϫ/Ϫ MEF stable pools were treated with the proteasome inhibitor ␤-lactone in the absence of signal. Unfortunately, the prolonged treatments with ␤-lactone necessary to see the accumulation of ubiquitinated IB␣ bound to NF-B in the absence of signal seems to activate the NF-B pathway (data not shown). This is FIG. 4. In vitro and in vivo signal-dependent ubiquitination of the IB␣ mutants. All experiments in this figure were done under stimulated conditions. The HeLa cytoplasmic extracts used in the in vitro ubiquitination assays were activated with a reaction mixture that contained 3 M okadaic acid. Control samples, designated by "C," were performed exactly like the wt sample except the reaction mixture was replaced by 50 mM Tris (pH 7.5). A, the in vitro signal-dependent ubiquitination assays were carried out as described under "Experimental Procedures." The IB␣ IP was carried out to examine total ubiquitination of both free and NF-B-associated IB␣. Quantitation was done on the PhosphorImager, using ImageQuant software. The percentage of labeled IB␣ shifted by ubiquitination was calculated by dividing the shifted counts by the total count. This calculation corrects for any differences in the input amounts of labeled IB␣. Three independent samplings were performed and the average of the three values obtained is depicted graphically. B, the RelA IP was done to isolate the labeled IB␣ that had become associated to NF-B present in the activated HeLa extract. Everything else is as was done for A. C, the supernatant from the RelA IP done in B was subjected to a second round of IP with anti-IB␣ sera. This isolated the free IB␣ in the in vitro ubiquitination assay. D, in vivo signal-dependent ubiquitination. The IB␣ Ϫ/Ϫ MEF stable pools were treated with ␤-lactone for 1 h and then stimulated with TNF␣ for 15 min. Cytoplasmic extracts were made and an IB␣ immunoblot was performed. Ub, ubiquitin. most likely caused by the tremendous stress that the cell experiences when its proteasomes are inactivated for prolonged periods of time. Consequently, no concrete conclusions could be made about the role of ubiquitination in the basal turnover of NF-B-associated IB␣.
The signal-independent ubiquitination of free IB␣ is quite robust (Fig. 7, lanes 2-4) which may explain the rapid basal turnover of free IB␣ when compared that of NF-B associated IB (110 and 550 min, respectively). Free IB␣ is very unstable in the absence of signal (Fig. 6A) and a detectable pool of ubiquitinated mutC and 3236mutC is observed in the absence of any stimulation and in the absence of a proteasome inhibitor (Fig. 7). The 3236 mutation in the mutC background does not reduce signal-independent ubiquitination of free IB␣ (Fig. 7,   lane 4). Therefore, phosphorylation of serine 32 and serine 36 is not necessary for efficient signal-independent ubiquitination and degradation of free IB␣.
We have previously shown that the carboxyl-terminal CKII sites are necessary for the efficient basal turnover of free IB␣ (35,46). Here we conclude (i) that efficient basal turnover of NF-B-associated IB␣ requires the carboxyl-terminal CKII phosphorylation and (ii) that IKK phosphorylation (serine 32 and 36) plays no role in the basal turnover of free or NF-B associated IB␣. DISCUSSION IB␣ functions as the primary regulator of NF-B in both stimulated and unstimulated cells. To accomplish this, IB␣ FIG. 6. Signal-independent turnover of free IB␣. A and B, everything was done the same as for Fig. 5 except these samples were directly IP with anti-IB␣ because they do not associate to RelA/p65 in these cell lines. A nonspecific (ns) band comes down in the IB␣ IP that migrates directly above the IB␣ band. The standard deviations for these samples were negligible and are not resolved at the given graphical resolution.

FIG. 5. Signal-independent turnover of NF-B-bound IB␣.
A-C, pulse-chase analysis was performed to determine the half-life of NF-B-associated IB␣ under signal-independent conditions. The stable pools were metabolically labeled with [ 35 S]methionine for 3 h, washed, and then chased for 0, 60, 120, or 240 min. A RelA IP was performed and then the precipitated material was subjected to a second round of IP with anti-IB␣ sera to reduce the background. This isolated the RelA/p65-associated IB␣ in these cell lines. Quantitation was done on the PhosphorImager, using ImageQuant software. The data were converted into percent relative intensity by assigning the zero time point of each time course as 100% and assigning the remaining time points a percentage that correlates to the fraction of its signal. The percent relative intensities were then graphed against time, in minutes, and half-lives were calculated.
itself must be a tightly regulated protein. NF-B activity is positively and negatively regulated through IB␣ phosphorylation and NF-B-dependent synthesis of IB␣, respectively. Mechanisms that ensure proper NF-B activity must exist to regulate signal-dependent degradation of free and NF-B-associated IB␣ as well as the signal-independent turnover of free and NF-B-associated IB␣. These mechanisms also must allow newly synthesized IB␣ the opportunity to enter the nucleus in order to remove NF-B from the DNA and/or inhibit further activation upon the removal of signal (33). One way that the cell can easily and efficiently regulate the multiple states and fates of IB␣ is through phosphorylation. Phosphorylation seems to be involved in almost all aspects of IB␣s regulation. We demonstrate when and how some of the different phosphorylation sites on IB␣ can influence ubiquitination, degradation, and the overall stability of free and NF-B-associated IB␣ in the presence or absence of NF-B inducing stimuli. Our conclusions about the role of phosphorylation and ubiquitination in the regulation of signal-dependent and -independent degradation of free and NF-B-associated IB␣ are summarized in Fig. 8.
Signal-dependent Degradation of NF-B-associated IB␣-Several groups have shown that Ser-32 and Ser-36 undergo signal-dependent phosphorylation and that mutating these sites stabilizes IB␣ in the presence of stimuli (18 -21). There are many different views on the role of the carboxyl terminus and the CKII phosphorylation sites in signal-dependent degradation of IB␣ (19, 21, 29, 35, 46 -49). Using a variety of phosphorylation site mutations stably expressed in an IB␣ Ϫ/Ϫ MEF cell line, we have confirmed that by mutating serine 32 and serine 36 to alanine, IB␣ is stabilized in the presence of TNF␣ stimulation (18 -21). We have also shown that the mutation of the carboxyl-terminal CKII sites (mutF) causes a significant stabilization of IB␣ during TNF␣ stimulation but to a lesser extent than the 3236 mutation. Although work shown here was carried out using transduced IB␣ Ϫ/Ϫ MEF cells, similar results were also obtained using transduced HeLa cell lines (data not shown). From these results we conclude that both IKK and CKII phosphorylation play an important role in the signal-dependent degradation of NF-B-associated IB␣.
Experiments were also carried out to study the critical step of IB␣ polyubiquitination during signal-dependent degradation of NF-B-associated IB␣. In vitro and in vivo systems clearly showed that IB␣ stability was increased in the 3236, M, and mutF mutants due to decreased levels of ubiquitination. The 3236 mutant simply blocks the signaling cascade from continuing to the ubiquitination step, while the mutF mutation is only able to decrease the efficiency of the reactions that lead to IB␣ degradation. Thus, in agreement with others we conclude that both the amino-terminal IKK sites and the carboxylterminal CKII sites play significant roles in the signal-dependent degradation of NF-B-associated IB␣. We also conclude that both amino-terminal IKK and carboxyl-terminal CKII phosphorylation is necessary for efficient signal-induced ubiquitination of NF-B associated IB␣.
Signal-dependent Degradation of Free IB␣-To maintain a sensitive and rapidly responding NF-B pathway the cell must be relatively free of unbound IB␣. If newly synthesized IB␣ does not associate with NF-B, the cell must clear this free population of IB␣ before proper activation of NF-B can take place. There is evidence that free IB␣ is degraded in a signaldependent manner and that this event is mainly regulated by the amino terminus of IB␣ (45,46). There are also reports that large carboxyl-terminal deletions that stretch into the sixth ankyrin repeat, and are believed to disrupt NF-B association, can stabilize IB␣ in the presence of stimuli (21,47,48,50).
Here we investigated the role of the amino-and carboxylterminal phosphorylation sites in the context of a full-length IB␣ molecule using both in vitro and in vivo systems. We were able to identify a mutant, mutC, that is degraded upon stimulation but does not associate to NF-B. The mutC mutant contains five serine and threonine to alanine mutations in the ank6 region. The 3236 mutation in the mutC background is able to partially stabilize IB␣ in the presence of TNF␣ without is necessary for efficient basal degradation of both free and NF-Bassociated IB␣. Ubiquitination is involved in the basal turnover of free IB␣ but it is still unclear whether or not ubiquitination plays a role in the basal turnover of NF-B-associated IB␣. Signal-dependent degradation of free IB␣ only requires IKK phosphorylation (IKK P) while efficient signal-dependent degradation of NF-B-associated IB␣ requires both CKII P and IKK P. Signal-dependent degradation of free and NF-B-associated IB␣ takes place in a ubiquitin-dependent manner.
having any effect on NF-B DNA binding activity. The 3236mutC mutant also shows reduced signal-induced ubiquitination in vivo compared with the wild type levels seen in the presence of mutC alone (Fig. 4). A reduction in signal-dependent ubiquitination of free IB␣ containing the 3236 mutation was also observed in vitro. Therefore the amino-terminal IKK phosphorylation sites play a critical role in signal-induced ubiquitination and degradation of free IB␣.
Several groups have reported on the effects of deleting or mutating the carboxyl-terminal PEST domain. Five groups report that the PEST domain does play a role in signal-dependent degradation (19,21,29,45,48), while four other groups report that it does not (46,47,49,50). Here we examine free full-length IB␣ containing five serine and threonine to alanine mutations in its PEST domain (mutF). In our in vitro ubiquitination system the mutF mutation caused a slight increase in the extent of signal-induced ubiquitination of free IB␣ when compared with wild type IB␣. This is in contrast to the decrease in ubiquitination that was seen for mutF while in association with NF-B. In the presence of NF-B inducing stimuli, the differential regulation of IB␣ stability by the carboxylterminal CKII sites indicates that a very interesting and sophisticated mechanism is utilized to regulate the different populations of IB␣ during signaling. We conclude that CKII phosphorylation is necessary for efficient signal-dependent degradation of NF-B-associated IB␣, but it is not necessary for efficient signal-dependent degradation of free IB␣ (46). In fact, the absence of CKII phosphorylation on free IB␣, under stimulated conditions, may potentiate the rate at which it is ubiquitinated and subsequently degraded by the 26 S proteasome. This ensures that NF-B activation is maintained until stimuli is removed and IKK activation is down-regulated.
Signal-independent Turnover of NF-B-associated IB␣-IB␣ is a more stable molecule while associated to NF-B than in its free state (45,51,52). However, in the associated state, IB␣ is not totally stabilized because NF-B activation and IB␣ degradation can be observed within a couple of hours after treating cells with cyclohexamide to block new protein synthesis (53,54). This activation is due to the signal-independent turnover of the NF-B-associated IB␣. We show that mutating the IKK phosphorylation sites in IB␣ has no effect on the basal degradation rate of NF-B-associated IB␣ (Fig. 5, A and C). On the other hand, mutating the CKII sites caused IB␣s half-life to double (Fig. 5B). We were unable to determine if ubiquitination plays a significant role in the basal turnover of NF-B-associated IB␣ because long term treatment (2-4 h) of the cells with ␤-lactone seems to activate the NF-B pathway. A role for ubiquitination in the basal turnover of NF-B-associated IB␣ may be observed in Fig. 7 (comparing  lanes 1 and 2). It appears that there may be a detectable pool of ubiquitinated IB␣ even in the absence of an NF-B inducing signal. Therefore, we conclude that only the constitutive CKII phosphorylation sites are important for signal-independent degradation of NF-B-associated IB␣.
Basal Turnover of Free IB␣-We have previously shown that both a carboxyl-terminal deletion of 39 amino acids and that the mutF mutation can increase the stability of free IB␣ in the absence of signal (35,46,49). In this report we go on to show that free IB␣ is very efficiently ubiquitinated in the absence of signal, and that polyubiquitinated IB␣ can be detected even in the absence of a proteasome inhibitor (Fig. 7,  lane 3). We also demonstrate that the 3236 mutation is unable to block ubiquitination of free IB␣ in the absence of signal (Fig. 7, lane 4). We propose that only the CKII sites are significant regulators of the signal-independent degradation of free IB␣, and that this degradation is mediated by ubiquitination and the 26 S proteasome.
An important component of IB␣ regulation/degradation that still needs to be identified is the signal-independent ubiquitin conjugation-ligase complex. We have shown that free and possibly NF-B-bound IB␣ undergo signal-independent ubiquitination. It has also been shown that the IB-ligase complex, that was recently identified (22,23,55,56), only binds to and ubiquitinates IB␣ when it is phosphorylated at its signalinduced phosphorylation sites (22,23,(55)(56)(57). This suggests that a signal-independent IB␣ ubiquitin-ligase complex must exist. It is possible that this signal-independent ubiquitin ligase recognizes the same residues as the signal-dependent ligase but only when serine 32 and serine 36 are unphosphorylated. When IB␣ is bound to NF-B the signal-independent ligase may only be able to associate to IB␣ with a weak affinity due to steric hindrance caused by IB␣s association to NF-B. When IB␣ is free in the cell the steric hindrance may be relieved allowing the signal-independent ubiquitin ligase to bind to IB␣ with a much higher affinity. This would explain the differences in the rate of ubiquitination and basal turnover between free and NF-B-bound IB␣. This could also explain why we observed an increase in the basal turnover of free IB␣ containing the 3236 mutation (Fig. 6B). By changing serine 32 and serine 36 to alanine we may have created a molecule with an even higher affinity for the signal-independent ubiquitinligase complex. An increase in the affinity of this ligase for its target would most likely result in an increase in the rate of its ubiquitination and subsequent degradation. Identification of the signal-independent ubiquitin ligase would allow further dissection of the mechanisms involved in IB␣ regulation.
Although our results are internally consistent and supported by previous biochemical and structural studies it is always necessary to consider the possibility that introducing mutations into any given protein may have unintended effects. It is for this reason that we utilized point mutations rather than gross deletions as has been done in other studies. It was our hope that by minimizing the changes made to IB␣ and by utilizing full-length proteins that we would be able to obtain the most meaningful results.
In summary, we have investigated the role of phosphorylation in the regulation of IB␣ stability. We have demonstrated that specific phosphorylation events can have very different implications depending on the state of the cell and of IB␣. The complexity of the regulation of IB␣ is a reflection of its importance to the cell. It will be interesting and of much importance to further explore the effects that other components of the IB kinase complex and degradation machinery have on the phosphorylation, ubiquitnation, and degradation of IB␣.