The 90-kDa Ribosomal S6 Kinase (pp90rsk) Phosphorylates the N-terminal Regulatory Domain of IκBα and Stimulates Its Degradation in Vitro *

Nuclear factor κB (NF-κB) is a eukaryotic member of the Rel family of transcription factors whose biological activity is post-translationally regulated by its assembly with various ankyrin-rich cytoplasmic inhibitors, including IκBα. Expression of NF-κB in the nucleus occurs after signal-induced phosphorylation, ubiquitination, and proteasome-mediated degradation of IκBα. The induced proteolysis of IκBα unmasks the nuclear localization signal within NF-κB, allowing its rapid migration into the nucleus, where it activates the transcription of many target genes. At present, the identity of the IκBα kinase(s) that triggers the first step in IκBα degradation remains unknown. We have investigated the potential function of the 90-kDa ribosomal S6 kinase, or pp90rsk, as a signal-inducible IκBα kinase. pp90rsk lies downstream of mitogen-activated protein (MAP) kinase in the well characterized Ras-Raf-MEK-MAP kinase pathway that is induced by various growth factors and phorbol ester. We now show that pp90rsk, but not pp70S6K or MAP kinase, phosphorylates the regulatory N terminus of IκBα principally on serine 32 and triggers effective IκBα degradation in vitro. When co-expressed in vivo in COS cells, IκBα and pp90rsk readily assemble into a complex that is immunoprecipitated with antibodies specific for either partner. While phorbol 12-myristate 13-acetate produced rapid activation of pp90rsk, in vivo, other potent NF-κB inducers, including tumor necrosis factor α and the Tax transactivator of human T-cell lymphotrophic virus, type I, failed to activate pp90rsk. These data suggest that more than a single IκBα kinase exists within the cell and that these IκBα kinases are differentially activated by different NF-κB inducers.

Nuclear factor B (NF-B) 1 is a transcription factor whose function is regulated by a family of cytoplasmic inhibitors termed the IBs (reviewed in Refs. 1 and 2). At present, nine IB family members have been identified (IB␣, IB␤, IB␥, IB␦, IB⑀, p105, p100, Bcl-3, and Cactus), each distinguished by the presence of multiple ankyrin repeats. The prototypic and best studied of the IBs is IB␣ (3), which binds to the heterodimeric NF-B complex (p50/Rel A) (4), masks the nuclear localization signal present in Rel A (5,6), and sequesters NF-B in the cytoplasm (4 -6). When appropriate inductive signals are delivered to the cell, phosphorylation of IB␣ ensues (7)(8)(9)(10), followed by the conjugation of multiple ubiquitin molecules and the degradation of the ubiquitinated IB␣ phosphoprotein by the 26 S proteasome complex (11)(12)(13). Of note, IB␣ degradation proceeds while the inhibitor is still physically associated with the NF-B heterodimer (10, 14 -17). However, the NF-B complex is ultimately liberated, allowing its rapid translocation into the nucleus, where it engages cognate enhancer elements and alters the transcriptional activity of various target genes.
Although phosphorylation of IB␣ is required for its proteolysis and the subsequent activation of NF-B, the nature of the cellular protein kinase(s) mediating this reaction remains unknown. Signal-induced phosphorylation involves two serine residues located at positions 32 and 36 near the N terminus of IB␣. Substitution of these serines with alanine residues generates a constitutively acting IB␣ repressor that readily binds to NF-B but fails to undergo signal-induced phosphorylation and degradation (7,8,10,18).
Studies in Drosophila have also yielded valuable insights into the biology of the Rel proteins and their control by the IBs. In the dorsal-ventral signal transduction pathway of Drosophila, dorsalizing signals mediated through the receptor Toll (an IL-1 receptor homologue) target Cactus (a member of the IB family) for degradation and result in activation of Dorsal (a Rel family member) (19). In this pathway, Pelle, a serine/threonine protein kinase, regulates the degradation of Cactus through phosphorylation, although it is unknown whether Pelle acts directly or indirectly on Cactus (20). Recently, a human IL-1 receptor-associated kinase has been cloned that is homologous to Pelle (21). This kinase appears to participate in the IL-1-induced signaling pathway leading to NF-B induction, but no evidence yet exists for its direct phosphorylation of IB␣.
Casein kinase II (CKII) also phosphorylates IB␣ in vivo (22)(23)(24). However, phosphopeptide mapping of phosphorylated IB␣ has shown that residues within the C-terminal PEST region, rather than the N-terminal serines, are targeted by CKII. Phosphorylation of the C terminus of IB␣ by CKII or other kinases may play a role in the constitutive degradation of uncomplexed IB␣. Additionally, CKII-mediated phosphorylation appears important for the accelerated turnover of IB␣ and the persistent induction of NF-B observed following HIV infection of macrophages (24). Immunodepletion of CKII from these cell extracts results in an inhibition of IB␣ degradation in vitro (24).
Recently, a novel ubiquitination-stimulated protein kinase has been identified that phosphorylates IB␣ in a serine 32/36dependent manner (25). This kinase resides in a large 700-kDa multiprotein complex, and ubiquitination of some component of the complex results in increased IB␣ phosphorylating activity. Whether ubiquitination directly activates the kinase or, alternatively, acts indirectly to alter another component of the complex remains unresolved.
We have investigated the potential role of a known intracellular protein kinase in the Ras-Raf-MEK-MAP kinase signaling pathway as an IB kinase. This enzyme, the 90-kDa ribosomal S6 kinase, or pp90 rsk , lies immediately downstream of MAP kinase in the phorbol ester and growth factor signaling pathway (26,27). We now show that pp90 rsk phosphorylates the N terminus of IB␣ principally on serine 32 and functionally induces IB␣ degradation in vitro. We further show that pp90 rsk and IB␣ can physically associate in vivo. Finally, we show that only a subset of the known NF-B-inducing signals leads to the activation of pp90 rsk . These findings suggest that rather than a single IB kinase, a family of IB kinases may exist within the cell that are differentially activated by different inducers of NF-B.

MATERIALS AND METHODS
cDNA Vectors and Expression of Proteins-The expression vector containing a hemagglutinin-epitope-tagged pp90 rsk cDNA (pMT2-HA-RSK1) was provided by Dr. Joseph Avruch (Harvard University and Massachusetts General Hospital, Boston, MA). Wild type IB␣ cDNA provided by Dr. Al Baldwin (University of North Carolina, Chapel Hill, NC) was cloned into the HindIII and XbaI sites of the pCMV4 eukaryotic expression vector provided by Dr. Mark Stinski (University of Iowa, Iowa City, IA). For in vitro translation, wild type IB␣ or mutant IB␣ containing alanine for serine substitutions at position 32 and/or 36 (4) cloned into the HindIII and XbaI or XmaI sites of the pBluescript SK(ϩ) vector (Stratagene). Biosynthetically radiolabeled IB␣ or its mutant analogues were synthesized by transcription-coupled in vitro translation in wheat germ extracts (Promega).
A bacterial expression plasmid encoding hexahistidine (His)-tagged IB␣ was constructed by cloning the IB␣ cDNA into the pTrcHisC vector (Invitrogen). The N-terminal ⌬1-36 IB␣ deletion mutant was cloned into the pRSETC vector (identical to the pTrcHis vector except in the promoter region). Wild type IB␣ and the S32/36A IB␣ mutant containing alanine for serine substitutions at both residues 32 and 36 were isolated from Escherichia coli lysates by purification on a nickel chelate column (Ni-NTA, Qiagen). Following an initial wash in high salt buffer (50 mM Tris-HCl, pH 7.5, 300 mM NaCl), successive washes were performed with elution buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl) containing increasing concentrations of imidazole (0, 10, 50, 100 mM). The His-tagged proteins were eluted in buffer containing 200 mM imidazole. The fractions containing the desired proteins were dialyzed overnight in 50 mM Tris-HCl, pH 7.5, and 2 mM DTT.
Cell Lines and Tissue Culture Conditions-Jurkat cells and Jurkat cells stably expressing either HTLV-I Tax or Tax antisense cDNA constructs and a neomycin resistance gene were maintained in RPMI 1640 supplemented with 10% fetal calf serum and penicillin/streptomycin at 37°C in 5% CO 2 ; 800 g/ml G418 was added to the Jurkat-Tax and anti-Tax cell culture media. Cells were treated with phorbol 12myristate 13-acetate (PMA) (50 ng/ml) or TNF-␣ (50 ng/ml) for various periods of time. Vehicle controls corresponding to the amounts of added Me 2 SO for PMA and water for TNF-␣ were performed in parallel. The cells were washed once with ice-cold phosphate-buffered saline and lysed in ELB buffer (50 mM HEPES, pH 7.4, 250 mM NaCl, 0.2% Nonidet P-40, 5 mM EDTA, 0.5 mM DTT, 1.0 mM phenylmethylsulfonyl fluoride, and protease inhibitor mixture containing 0.75 g/ml bestatin; 0.5 g/ml each of aprotinin, antipain, leupeptin, and trypsin inhibitor; 0.4 g/ml phosphoramidon; and 0.05 g/ml pepstatin). The cell lysates were clarified by centrifugation at 4°C for 15 min at 15,000 ϫ g, and the supernatant was used for immunoprecipitation as described below.
In Vitro Assay of Protein Kinase Activity-Phosphorylation of bacterially expressed IB␣ (0.5 g) was performed in 50-l reaction mixtures incubated for 15 min at room temperature. For Xenopus laevis eggderived MAP kinase, X. laevis egg-derived pp90 rsk , and mouse pp70 S6K , the reaction buffer contained 20 mM HEPES, pH 7.0, 10 mM MgCl 2 , 2 mM DTT, 100 M EGTA, 0.1 mg/ml bovine serum albumin, 100 M ATP, 25 Ci of [␥ 32 P]ATP (specific activity 3000 Ci/mol). Myelin basic protein (0.1 mg/ml, Sigma) or 40 S ribosomal subunits (0.05 mg/ml) generously provided by Dr. James Maller (Howard Hughes Medical Institute and University of Colorado Health Sciences Center, Denver, CO) were used as positive controls for these kinase reactions. Aliquots of the reactions were mixed in SDS-PAGE sample buffer, heated to 98°C, and microcentrifuged at 15,000 ϫ g for 2 min, followed by analysis of the supernatants by SDS-PAGE. Alternatively, IB␣ was immunoprecipitated from the reactions with peptide-specific polyclonal rabbit antisera (2.5 l) recognizing the C-terminus of IB␣. Immune complexes were reacted with formalin-fixed Staphylococcus aureus cells (Pansorbin A, 50 l, Calbiochem) and collected by centrifugation. The Protein Abound immune complexes were washed three times and similarly analyzed by SDS-PAGE and autoradiography. In the case of pp90 rsk , three independently purified preparations were tested. The specific activities of these pp90 rsk preparations were 4.6, 8.9, and 11.2 nmol of ATP incorporated/min/mg of protein assayed using Kemptide as substrate.
V8 Digestion of in Vitro Phosphorylated IB␣-Bacterially produced His-tagged IB␣ was phosphorylated with purified pp90 rsk as described above and subjected to mild V8 proteolysis at room temperature for 10 min. The protease inhibitor N ␣ -p-tosyl-L-lysine chloromethyl ketone was added to the reaction mixture together with 100 g/ml (final concentration) of bovine serum albumin to terminate proteolytic cleavage. IB␣ peptides containing the N-terminal His-epitope were immunoprecipitated from the digest using his-tag-specific antibodies. The immunoprecipitates were analyzed with Tris-Tricine gels by the method of Schä gger and von Jagow (28).
In Vitro Degradation of IB␣-In vitro synthesized [ 35 S]-radiolabeled IB␣ was phosphorylated with pp90 rsk in the presence of unlabeled ATP as described. The phosphorylated or control IB␣ proteins (25 l) were incubated in 100 l of reticulocyte lysates containing 5 mM DTT, 2.5 mM ATP, 1 mM creatine phosphate, and 200 g/ml creatine phosphokinase. Samples were removed at various times and quickly frozen in liquid nitrogen. The samples were then immunoprecipitated with C-terminalspecific anti-IB␣ antibodies and analyzed by SDS-PAGE followed by autoradiography.
Coimmunoprecipitation Assays-Monkey kidney COS7 cells, maintained in complete Dulbecco's modified Eagle's medium, were transfected with pMT2-HA-RSK1 (encoding pp90 rsk-1 ) or pCMV4-HA-PP2A (encoding the A regulatory subunit of protein phosphatase 2A) and CMV4-IB␣ using LipofectAMINE (Life Technologies, Inc.). After 48 h, the cells were starved for 1 h in methionine/cysteine-free Dulbecco's modified Eagle's medium and then metabolically radiolabeled with [ 35 S]methionine and [ 35 S]cysteine for 2 h. Whole-cell extracts were prepared by lysis in ELB buffer, followed by immunoprecipitation analyses as described above using either anti-HA-epitope antibody (BAbCo, Berkeley, CA) or anti-IB␣ antisera specific for the C terminus of this inhibitor. Nonradioactive COS cell lysates were also prepared and immunoprecipitated with the anti-HA or IB␣ antibodies followed by immunoblotting with anti-pp90 rsk -specific antibodies. These immunoblots were developed with a horseradish peroxidase-conjugated secondary antibody and enhanced chemoluminescense (ECL) as described by the manufacturer (Amersham).
Immune Complex-associated Protein Kinase Assay-Cell lysates derived from about 5 ϫ 10 6 Jurkat cells were precleared for 1 h and incubated with 10 l of anti-pp90 rsk antibody (Santa Cruz Biotechnology) at 4°C for 1 h. 30 l of Protein A-agarose (Boehringer Mannheim) was then added to the mixture, and the incubation was continued for an additional 1 h. The mixture was centrifuged at 4°C, followed by washing of the Protein A-agarose resin three times in ELB buffer. Immune complexes were washed three times with kinase buffer ( Phosphatase Treatment of Phosphorylated Proteins-Following the phosphorylation reactions, IB␣ was immunoprecipitated with anti-IB␣ antiserum specific for the C-terminus and Protein A-agarose. The bound immune complexes were washed three times in dephosphorylation buffer (50 mM Tris-HCl, pH 8.0, 1 mM EDTA) followed by the addition of 10 units of calf intestinal alkaline phosphatase (Boehringer Mannheim). The samples were then incubated at 37°C for 30 min and analyzed by SDS-PAGE and autoradiography.

RESULTS
In Vitro Phosphorylation of IB␣ by pp90 rsk -Phorbol ester is both a potent inducer of NF-B and a known activator of the MAP kinase pathway. To determine whether kinases positioned along this or related pathways are capable of phosphorylating IB␣, the ability of MAP kinase, pp70 S6K , and pp90 rsk to phosphorylate IB␣ in vitro was examined. These kinases phosphorylate residues that lie within specific consensus sequences. For example, MAP kinase phosphorylates serines and threonines that precede proline residues (proline-directed kinase consensus sequence), while both pp90 rsk and pp70 S6K phosphorylate serines or threonines within the consensus sequence Arg-X-X-Ser/Thr (29).
Recombinant His-tagged IB␣ was used as the substrate in these in vitro kinase assays. Following the kinase reaction, IB␣ was immunoprecipitated from the reaction using polyclonal anti-IB␣ antibodies (30). Immunoprecipitation was performed to ensure that the phosphorylated protein was indeed recombinant IB␣. As shown in Fig. 1, pp90 rsk (lanes 7 and 8), but not pp70 S6K (lanes 3 and 4) or p42 MAPK (lanes 11 and 12), phosphorylated recombinant His-tagged IB␣ in vitro. The failure of the pp70 S6K and p42 MAPK preparations to phosphorylate IB␣ was not due to inactivation of these enzymes during their purification, as each capably phosphorylated known protein substrates, including the 40S ribosomal subunit for pp70 S6K (lanes 1 and 2) and myelin basic protein for p42 MAPK (lanes 9 and 10). Of note, while both pp90 rsk and pp70 S6K recognize the same consensus phosphoacceptor site and phosphorylate the S6 protein of the 40S ribosome, only pp90 rsk phosphorylated IB␣. Another kinase, Ca 2ϩ -calmodulin-dependent protein kinase II (CaMKII) also shares the same consensus phosphoacceptor site as pp90 rsk and pp70 S6K (31). However, like pp70 S6K , CaMKII failed to phosphorylate IB␣ in vitro, although it did phosphorylate one of its physiological substrates, synapsin II (data not shown). These data highlight the ability of pp90 rsk to utilize IB␣ as a substrate for phosphorylation in vitro.
Immunoprecipitation of a Phosphorylated N-terminal Fragment of IB␣-To ascertain whether pp90 rsk phosphorylates IB␣ at either of the two critical N-terminal serine residues located at positions 32 and 36, a 6xhis-IB␣ and a similarly epitope-tagged S32/36A IB␣ mutant in which both of these serines were substituted with alanine, was subjected to in vitro phosphorylation as described above. The protein was then subjected to mild V8 (endoprotease Glu C) proteolysis with a sequencing-grade protease. The cleaved proteins were immunoprecipitated with antiserum specific for the N-terminal hisepitope. This antiserum recognizes both wild type IB␣ and the S32/36A His-tagged mutants. Phosphorylation of IB␣ at either serine 32 or 36 should result in a fragment of 84/87 amino acids (including the epitope tag) when exposed to V8 protease, which cleaves at residue 48 or 51. Fig. 2A depicts the recombinant IB␣ protein showing the signal-dependent N-terminal phosphorylation sites in relation to the V8 cleavage sites. The smallest band in the immunoprecipitate shown in the leftmost two lanes of Fig. 2B, indicated by the arrow, migrates between the 8.16-and 10.6-kDa myoglobin fragment, consistent with the molecular size of the 84/87 N-terminal fragment of IB␣. In contrast, the S32/36A IB␣ protein failed to yield a similarly sized band (Fig. 2, fourth and fifth lanes). The even smaller band detected with these samples was also detected in the absence of added V8 protease. These results thus demonstrate that pp90 rsk is capable of phosphorylating the regulatory N terminus of IB␣ involving serine 32 and/or 36.
pp90 rsk -mediated Phosphorylation Promotes IB␣ Degradation in Vitro-Although a variety of kinases can phosphorylate IB␣ in vitro, the critical functional issue is whether these kinases promote IB␣ degradation. Using an in vitro degradation system, we studied whether pp90 rsk -mediated phosphorylation of IB␣ triggers its destruction. The reticulocyte lysate degradation system employed in these experiments contains all of the component proteins and macromolecules necessary for ubiquitin-dependent and -independent 26S proteasome-mediated degradation (31). A variety of proteins have been shown to be degraded in this reticulocyte lysate system, including ornithine decarboxylase, antizyme, and the transcription factors Fos, Jun, and Myc (32)(33)(34)(35)(36)(37)(38).
[ 35 S]-Radiolabeled wild type IB␣ (Fig. 3, upper panel) and the S32/36A IB␣ mutant (lower panel) were synthesized in vitro and preincubated with Xenopus egg-derived pp90 rsk (lanes 1-4), mammalian pp70 S6K (lanes 5-8), or control kinase buffer lacking an added kinase (lanes 9 and 10). The IB␣ proteins were then incubated in a degradation-competent reticulocyte lysate that had specifically not been pretreated with RNase or hemin. Hemin is known to inhibit the 26S proteasome but is often added to reticulocyte lysates to improve translation since it prevents premature peptide chain termination (39). Degradation of [ 35 S]-IB␣ was monitored by immunoprecipitation with specific anti-IB␣ antisera over the course of a 120-min incubation conducted in the presence of an ATPregenerating system. As shown in lanes 1-4 of the upper panel, pp90 rsk -treated wild type IB␣ was significantly degraded during the 2-h incubation. However, IB␣ treated with pp70 S6K (lanes 5-8) or control buffer (lanes 9 and 10) was not degraded. In contrast to wild type IB␣, the S32/36A IB␣ mutant was not degraded when incubated with pp90 rsk (lower panel, lanes [1][2][3][4], indicating that IB␣ degradation in this in vitro system is dependent on the presence of these N-terminal regulatory serines, as it is in vivo. Together, these results suggest that pp90 rsk -mediated phosphorylation of IB␣ is biologically relevant since it leads to serine 32-and/or serine 36-dependent degradation of wild type IB␣ protein in vitro.  1-4) or pp90 rsk (lanes 5-8) were incubated with 40 S ribosomal subunits (40S, lanes 1, 2, 5, and 6) or wild type his-tagged IB␣ (lanes 3, 4, 7, and  8). Samples containing MAP kinase (lanes 9 -12) were similarly incubated with myelin basic protein (MBP, lanes 9 and 10) or wild type his-tagged IB␣ (lanes 11 and 12). In vitro kinase reactions were performed as described under "Materials and Methods" and analyzed by SDS-PAGE and autoradiography. Migration of the various protein substrates is indicated at the right margin. pp90 rsk Phosphorylates IB␣ by pp90 rsk Principally on Serine 32-Previous in vivo studies have shown that both S32 and S36 at the N terminus of IB␣ are required for signal-induced phosphorylation and degradation of this inhibitor (7)(8)(9)(10)18). Additionally, both of these serine residues are directly phosphorylated in vivo by an unknown kinase(s) following cellular stimulation with PMA or TNF-␣ (45). Phosphorylation at these sites in vivo results in a slower electrophoretic mobility for the IB␣ protein that is readily detectable on SDS-PAGE gels. In contrast, no mobility shift is observed when cells containing the S32/36A IB␣ mutant is similarly induced. To assess whether pp90 rsk mediates phosphorylation on S32, S36, or both sites, wild type or mutant IB␣ proteins altered at one or both of these N-terminal serines were used as substrates in the in vitro kinase reactions. Activated pp90 rsk was obtained by immunoprecipitating this kinase from PMA-stimulated HeLa cells. As shown in Fig. 4A, wild type IB␣ displayed a mobility shift when incubated with activated pp90 rsk (lane 2), while the S32/ 36A IB␣ mutant containing alanine substitutions at both positions 32 and 36 did not (lane 4). The S36A single-substitution mutant of IB␣ also exhibited a significant change in mobility in the presence of pp90 rsk (lane 8); however, the S32A IB␣ mutant displayed only a minimal change in migration (lane 6). These data suggest that the principal site of pp90 rskmediated phosphorylation in vitro corresponds to serine 32. Assuming that in vitro and in vivo degradation requirements are the same for IB␣, this finding suggests either that phosphorylation at this single site is sufficient for in vitro degradation or, alternatively, that a second kinase present in the reticulocyte lysate may act in concert with pp90 rsk to modify serine 36 and thus complete the phosphorylation requirements for degradation. Alternatively, pp90 rsk phosphorylation of serine 32 may facilitate subsequent modification of serine 36 by pp90 rsk .
Phosphatase Treatment of Phosphorylated IB␣ Reverses the Shift in Mobility on SDS-PAGE-To confirm that the observed mobility shifts reflect phosphorylation of IB␣, pp90 rsk -treated wild type and mutant S32/36A IB␣ proteins were incubated with calf intestinal alkaline phosphatase (Fig. 4B). The retarded mobility of pp90 rsk -treated wild type IB␣ (lane 3) was lost following treatment with phosphatase (lane 4). In contrast, phosphatase treatment of the S32/36A IB␣ mutant, which did not display a mobility shift in the presence of pp90 rsk , did not alter its electrophoretic mobility (compare lanes 7 and 8). These results confirm that pp90 rsk -mediated phosphorylation of IB␣ is responsible for the altered migration of the wild type IB␣ performed on pp90 rsk -phosphorylated and V8-proteolyzed wild type His-tagged IB␣ (duplicate samples in first and second lanes) or the IB␣ S32/36A mutant (fourth and fifth lanes). Proteolytic digestions were performed as described under "Materials and Methods." Undigested phosphorylated wild type and S32/36A IB␣ proteins are shown in the third and sixth lanes, respectively. The arrow indicates an appropriate-sized fragment for the N-terminal peptide uniquely generated with wild type IB␣ but not with the S32/36A mutant of IB␣. The even smaller band obtained with IB␣ S32/36A likely corresponds to a spontaneous degradation product, as it is also present in the untreated control sample.  (lanes 1-4), active pp70 S6K (lanes 5-8), or control buffer (lanes 9 and 10). Samples were removed from the reaction mixture at the indicated times and immunoprecipitated with anti-IB␣ antibodies, followed by SDS-PAGE and autoradiography.

proteins.
Phosphorylation of IB␣ in the Presence or Absence of Rel A-Under basal conditions, IB␣ is normally complexed with NF-B in the cytosol. The Rel A subunit of NF-B is directly associated with IB␣ in this complex. Studies by Chen et al. (11) suggest that IB␣ is not only phosphorylated, but also ubiquitinated and degraded while still complexed to NF-B. To investigate whether the presence of Rel A affects the pattern of phosphorylation of IB␣, the pp90 rsk -mediated in vitro kinase assays were performed with IB␣ in both the presence and absence of in vitro cotranslated Rel A. These proteins assembled in vitro as indicated by the ability of anti-Rel A antibodies to coimmunoprecipitate IB␣ from the translation mix (data not shown). As shown in Fig. 4C, IB␣ displayed the same mobility shift when incubated with pp90 rsk in the presence or absence of Rel A (compare lanes 2 and 6). No changes in mobility were obtained when the S32/36A IB␣ mutant was incubated with pp90 rsk in the absence of Rel A, suggesting that Rel A did not occlude additional pp90 rsk phosphorylation sites whose modification would affect IB␣ mobility (data not shown).
pp90 rsk Physically Associates with IB␣ in Vivo-Proteins that participate in a linear pathway of signaling may some-times physically associate with each other. Some protein kinases associate with their substrates under circumstances where the kinase is inactive or where ATP is limiting. If the association between a protein kinase and its substrate is sufficiently avid, their interaction may be detected by coimmunoprecipitation of the two proteins. To assess whether IB␣ can physically associate with pp90 rsk in vivo, COS cells were cotransfected with expression vectors encoding HA-tagged pp90 rsk or control HA-tagged protein phosphatase-2A A regulatory subunit (HA-PP2A) and IB␣. Following transfection, proteins were metabolically radiolabeled with [ 35 S]methionine and cysteine, and the resultant cell lysates were immunoprecipitated with nonspecific preimmune, anti-HA, or anti-IB␣ antibodies. Immunoprecipitation of HA-PP2A and IB␣-transfected cells with anti-HA antibodies revealed a major band corresponding to HA-PP2A but no coimmunoprecipitation of IB␣ (Fig. 5A, compare lanes 1 and 2). Similarly, addition of anti-IB␣ antibody immunoprecipitated IB␣ but not PP2A (compare lane 3 to lanes 1 and 2). In contrast, cotransfection of cells with HA-pp90 rsk and IB␣ led to coimmunoprecipitation of both of these molecules using either the HA-or IB␣-specific antibodies ( lanes 5 and 6). The in vivo association of pp90 rsk and IB␣ was confirmed in experiments involving initial immunoprecipitation with anti-HA or anti-IB␣ followed by immunoblotting with an anti-pp90 rsk antibody (Fig. 5B). As shown in lanes 2 and 3, anti-IB␣ coimmunoprecipitated significant amounts of pp90 rsk in these cotransfected cells, while nonspecific preimmune sera yielded no detectable pp90 rsk (lane 1). Together, these results indicate that IB␣ can physically associate with pp90 rsk in vivo and thus provide further support for the possibility that pp90 rsk functions as a physiologically relevant IB␣ kinase.
Induction of pp90 rsk Kinase Activity by Phorbol Ester but Not by TNF-␣ or HTLV-I Tax-Studies were next performed to assess whether pp90 rsk is activated by various well known inducers of NF-B in vivo (Fig. 6). Activation of pp90 rsk was monitored either by a very small but perceptible change in its electrophoretic mobility, reflecting autophosphorylation (Fig.  6A), or by its ability to phosphorylate recombinant IB␣ when the latter was added as an exogenous substrate to an in vitro kinase assay performed with anti-pp90 rsk immunoprecipitates (Fig. 6B). PMA stimulation of Jurkat cells for 5 or 15 min resulted in the rapid activation of pp90 rsk (panel A, lanes 3 and 4) and phosphorylation of IB␣ (panel B, lanes 3 and 4). In contrast, two other recognized inducers of NF-B, TNF-␣ and HTLV-I Tax, did not significantly activate pp90 rsk autophosphorylation (panel A, lanes [5][6][7][8] or induce phosphorylation of IB␣ in the in vitro kinase assay (panel B, lanes 5-8). However, this particular preparation of TNF-␣ (see panel C, lanes 5-7) and HTLV-I Tax (data not shown) stimulated phosphorylation and degradation of IB␣. These findings indicate that only a subset of the known inducers NF-B leads to the activation of pp90 rsk , suggesting that other kinases are likely activated by different NF-B inducers, such as TNF-␣ and HTLV-I Tax. This result argues against the notion of a single cellular IB␣ kinase.

DISCUSSION
Phosphorylation and dephosphorylation represents a general strategy frequently employed for the dynamic regulation of eukaryotic transcription factor function. The enhancer-binding protein is often the specific target of such post-translational modifications that lead to an alteration in its DNA binding or functional activity. However, in the NF-B system, primary regulation is exerted through phosphorylation of IB␣, an ankyrin-rich inhibitor that sequesters the NF-B complex in the cytoplasm. Phosphorylation targets IB␣ for ubiquitination  2, 4, 6, and 8) or absence (lanes 1, 3, 5, and 7) of activated pp90 rsk immunoprecipitated from PMA-stimulated HeLa cells in the presence of unlabeled ATP (1 mM). Samples were analyzed for altered IB␣ electrophoretic mobility on SDS-PAGE gels. B, wild type IB␣ (lanes 1-4) or S32/36A IB␣ (lanes 5-8) was phosphorylated with activated pp90 rsk (lanes 3, 4, 7, and 8) or buffer control (lanes 1, 2, 5, and 6) as described for A and then treated with calf intestinal alkaline phosphatase (CIP) (even-numbered lanes) or buffer alone (odd-numbered lanes). C, [ 35 S]-labeled wild type IB␣ proteins were treated with activated pp90 rsk (even-numbered lanes) or kinase buffer alone (oddnumbered lanes) in the presence (lanes 5-8) or absence (lanes 1-4) of cotranslated Rel A. and subsequent degradation by the 26 S proteasome. Although a necessary step in the activation of transcription, phosphorylation alone does not result in the release of IB␣ from the NF-B complex and thus is insufficient for activation of NF-B-mediated transcription. Thus far, the identity of the kinase(s) responsible for coupling cellular activation to phosphorylation of IB␣ or other members of the IB family has remained elusive.
In the current paper, we have explored the possible function of pp90 rsk as a stimulus-coupled IB␣ kinase. In quiescent cells, inactive pp90 rsk resides in the cytoplasm and is partially complexed with its upstream regulator, p42/44 MAPK (41). Cellular activation mediated by various growth factors operating through the Ras-Raf-MEK-MAPK pathway or phorbol ester leads to the activation of MAP kinase, the phosphorylation and activation of pp90 rsk , and the import of these kinases into the nucleus. Activated forms of pp90 rsk have already been implicated in the regulation of various nuclear transcription factors, including c-Fos (40) and Nur77 (42,43). Recently, pp90 rsk has been reported to produce both positive and negative effects on another family of inducible transcription factors, the cyclic AMP response element-binding proteins (CREB). Specifically, FIG. 5. IB␣ and pp90 rsk physically associate in vivo. A, COS cells were cotransfected with expression vectors encoding IB␣ (all lanes) and HA-PP2A (lanes 1-3) or HA-pp90 rsk (lanes 4 -6) followed 48 h later by radiolabeling with [ 35 S]methionine/cysteine. Cell lysates were immunoprecipitated with nonspecific control antibodies (lanes 1 and 4), anti-HA antibodies (lanes 2 and 5), or anti-IB␣ antibodies (lanes 3 and 6). Samples were analyzed by SDS-PAGE and autoradiography. B, combined immunoprecipitation and anti-pp90 rsk immunoblotting were performed using nonradioactive COS cell lysates transfected as described above. Immunoprecipitations were performed using nonspecific (lane 1), anti-HA (lane 2), or anti-IB␣ (lane 3) antisera. Immunoblotting was performed using an anti-pp90 rsk -specific antibody and developed using a second antibody and ECL. pp90 rsk2 , one of three closely related rsk genes (rsk1, rsk2, rsk3), has been shown to function as a stimulus-coupled CREB kinase modulating CREB activity by phosphorylating this factor on a key regulatory serine located at position 133 (44). Conversely, pp90 rsk appears to oppose CREB action by inducibly but stably associating with the CREB-binding protein and blocking the interaction of this co-activator with CREB (45). Of note, the enzymatic function of pp90 rsk is not required for these latter inhibitory effects. Together, these various results provide an experimental precedent for the participation of pp90 rsk as a regulatory interface between the signals induced by the ligation of various growth factor receptors on the membrane and specific transcription factors that modulate the activity of target genes within the nucleus.
Using a purified activated enzyme preparation in initial in vitro kinase assays, we found that pp90 rsk mediates phosphorylation of IB␣. Furthermore, based on V8 protease analysis, this phosphorylation involves the N-terminus of IB␣, where two critical residues for signal-induced phosphorylation, Ser-32 and Ser-36, reside. pp90 rsk -mediated phosphorylation of IB␣ proved biologically relevant since this post-translational modification stimulated proteasome-dependent degradation of IB␣. In contrast, the S32/36A double-substitution mutant of IB␣ was not degraded in the presence of activated pp90 rsk . The stoichiometry of IB␣ phosphorylation by pp90 rsk appeared quite high, as assessed by the ability of pp90 rsk to elicit a gel mobility shift for IB␣. By this criterion, one-third to one-half of the IB␣ molecules displayed an altered mobility in the presence of pp90 rsk . Interestingly, the closely related S6 kinase, pp70 S6K , which recognizes the same consensus phosphoacceptor site, Arg-X-X-Ser/Thr, as pp90 rsk is incapable of phosphorylating IB␣. Likewise, CaMKII, a calmodulindependent kinase that also phosphorylates within the same consensus phosphoacceptor site as pp90 rsk and pp70 S6K , fails to phosphorylate IB␣. These findings indicate that the recognition of IB␣ by these protein kinases involves determinants beyond the consensus phosphoacceptor site. It is likely that the overall three-dimensional structures of IB␣ and the kinase play a pivotal role in the effectiveness of this protein-protein interaction.
Since a "purified" pp90 rsk preparation was used in these in vitro studies, we considered the possibility that the IB␣ phosphorylation might be due to a contaminating kinase. However this possibility is unlikely because: 1) each of three independently purified pp90 rsk preparations displayed the same IB␣ kinase activity in at least three assays; 2) overloading of an SDS-PAGE gel with the pp90 rsk preparation followed by silver staining revealed only pp90 rsk and no other bands; and 3) autophosphorylation reactions with the pp90 rsk preparation revealed no other bands. The observed IB␣ kinase activity in the pp90 rsk preparations thus appears to reflect the activity of pp90 rsk rather than a contaminant.
Since serines 32 and 36 located near the N terminus of IB␣ are key regulatory sites that must be directly phosphorylated to trigger subsequent ubiquitination and degradation of this inhibitor (7-10, 18, 46), potential phosphorylation of these sites by pp90 rsk was studied. Serine 32 is embedded within a sequence that conforms to the consensus phosphoacceptor site for phosphorylation by pp90 rsk ; however, serine 36 is not. With wild type IB␣ or the S32A and S36A single-substitution mutants of this inhibitor as substrates, pp90 rsk was shown to readily phosphorylate IB␣ proteins containing serine 32. In contrast, as judged by mobility shift, serine 36 functioned as a very poor substrate for pp90 rsk . Since both serine 32 and serine 36 must be directly phosphorylated for subsequent degradation (46), these findings suggest that the action of pp90 rsk alone may not be sufficient to trigger the subsequent ubiquitination and degradation of IB␣. It is possible that a second, yet unidentified, kinase present in the reticulocyte lysate mediates phosphorylation at serine 36, thus promoting IB␣ degradation. Although unproven, it is intriguing to consider the possibility that phosphorylation in vivo at the first serine site by one kinase may facilitate sequential phosphorylation at the second serine site by a different kinase. Alternatively, pp90 rsk phosphorylation at serine 32 may enhance its activity at serine 36, producing a polarity to the sequence of modifications. Precedence for regulation by such sequential phosphorylation is found in the case of both pp90 rsk and pp70 S6K phosphorylation of S6 peptide (47). Finally, degradation of IB␣ in vitro may proceed with phosphorylation at serine 32 only.
To test the potential in vivo relevance of pp90 rsk as an IB␣ kinase, we investigated whether these proteins can physically associate within a cell. Using COS cells for cotransfection with HA epitope-tagged pp90 rsk and IB␣ expression vectors, we demonstrated that these two proteins, but not similarly epitope-tagged control proteins, are coimmunoprecipitated using either anti-HA or anti-IB␣-specific antibodies. Using mutants of IB␣ to study this interaction further, our preliminary results indicate that the N terminus of IB␣ spanning the regulatory serines at positions 32 and 36 is not required for IB␣ binding to pp90 rsk . In contrast, deletion of ankyrin repeats 1-5 of IB␣ severely impairs the interaction of IB␣ with pp90 rsk .
Our final series of studies explored what NF-B-inducing signals operate through pp90 rsk -mediated phosphorylation of IB␣. These studies revealed that PMA induced activation of pp90 rsk , phosphorylation of IB␣, and induction of NF-B. In sharp contrast, neither TNF-␣ nor the Tax trans-activator protein of HTLV-I activated pp90 rsk . However, both of the latter inducers potently stimulated phosphorylation and degradation of IB␣ and activated nuclear NF-B expression. This result clearly indicates that not all NF-B inducers operate through pp90 rsk activation. These results predict the likely existence of multiple IB kinases that are differentially coupled to various signaling pathways. Thus, the critical kinase(s) ultimately responsible for phosphorylating serines 32 and 36 may vary, depending on the particular NF-B-inducing signal. We propose that pp90 rsk represents one such kinase in a larger set of enzymes that regulate IB␣ phosphorylation.