Control Sites of Ribosomal S6 Kinase B and Persistent Activation through Tumor Necrosis Factor*

RSKB, a 90-kDa ribosomal S6 protein kinase family (RSK) member with two complete catalytic domains connected by a linker, is activated through p38- and ERK-mitogen-activated protein kinases. The N-terminal kinases of RSKs phosphorylate substrates; activation requires phosphorylation of linker and C-terminal kinase sites. Unlike other RSKs, the activation loop phosphorylation sites of both catalytic domains of RSKB, Ser196 and Thr568, were required for activity. RSKB activation depended on phosphorylation of linker Ser343 and Ser360 and associated with phosphorylation of nonconserved Ser347, but Ser347-deficient RSKB retained partial activity. The known protein kinase A and protein kinase C inhibitors, H89 and Ro31–8220, blocked RSKB activity. Treatment of HeLa cells with tumor necrosis factor, epidermal growth factor, phorbol 12-myristate 13-acetate, and ionomycin but not with insulin resulted in strong activation of endogenous RSKB. High RSKB activity and Ser347/Ser360 phosphorylation persisted for 3 h in tumor necrosis factor-treated cells, in contrast to the short bursts of p38, ERK, and RSK1–3 activities. In conclusion, a variety of stimuli induced phosphorylation and activation of RSKB through both p38 and ERK pathways; the persistence of activation indicated that RSKB selectively escaped cell mechanisms causing rapid deactivation of upstream p38 and ERK and other RSKs.

RSKB, a 90-kDa ribosomal S6 protein kinase family (RSK) member with two complete catalytic domains connected by a linker, is activated through p38-and ERK-mitogen-activated protein kinases. The N-terminal kinases of RSKs phosphorylate substrates; activation requires phosphorylation of linker and C-terminal kinase sites. Unlike other RSKs, the activation loop phosphorylation sites of both catalytic domains of RSKB, Ser 196 and Thr 568 , were required for activity. RSKB activation depended on phosphorylation of linker Ser 343 and Ser 360 and associated with phosphorylation of nonconserved Ser 347 , but Ser 347 -deficient RSKB retained partial activity. The known protein kinase A and protein kinase C inhibitors, H89 and Ro31-8220, blocked RSKB activity. Treatment of HeLa cells with tumor necrosis factor, epidermal growth factor, phorbol 12-myristate 13-acetate, and ionomycin but not with insulin resulted in strong activation of endogenous RSKB. High RSKB activity and Ser 347 /Ser 360 phosphorylation persisted for 3 h in tumor necrosis factor-treated cells, in contrast to the short bursts of p38, ERK, and RSK1-3 activities. In conclusion, a variety of stimuli induced phosphorylation and activation of RSKB through both p38 and ERK pathways; the persistence of activation indicated that RSKB selectively escaped cell mechanisms causing rapid deactivation of upstream p38 and ERK and other RSKs.
The three mitogen-activated protein kinase (MAPK) 1 cascades, denoted by p38, ERK, and c-Jun N-terminal kinase, are prominent pathways to transmit and process signals from the cell surface to the nucleus (for recent reviews see Refs. [1][2][3]. Despite similarity and convergence between MAPKs, all pathways often responding to one extracellular signal, selective connectivities control cell responses in a balance between ERK (associated with growth factor activation) and c-Jun N-terminal kinase and p38 pathways mediating responses to toxins, physical stress, and inflammatory cytokines (4 -20). MAPKs also control cell cycle responses through cyclin D1 and Cdk activities (21). More recent studies indicate important roles of MAPKs in brain-specific functions. c-Jun N-terminal kinase activities in brain subregions are higher than in peripheral organs (22). Differential MAPK activation in specific subregions was critical for the consolidation of long term spatial and taste memory (23,24), consistent with abnormal electrophysiologic responses in deficient mice (25). Transcription factors activated through MAPK pathways control memory tasks and sensitivity to drug addiction (26,27). Electrical stimulation of cortico-striatal pathways elicited ERK and coincident transcription factor activation in distinct subcellular locations (28 -30). While there remain many unanswered questions, these studies collectively indicate that MAPKs in addition to their role in cell growth and survival are essential components in activity-dependent modifications of synaptic transmission and neuronal plasticity. High functional diversification is documented by the presence of multiple genes and splice variants for all MAPKs, with tissue-specific expression of isoforms, selective pathway connectivities, and subcellular location (reviewed in Refs. 1 and 31).
ERK and p38, in addition to transcription factors, activate downstream kinases such as MNK1/2, PRAK, and MAPKAPKs 2, 3, and 5 (9,(32)(33)(34)(35)(36). In addition, RSKs are important ERK and p38 substrates; ERK long was known to connect to RSK1-3, which in turn activate transcription factors such as CREB (37)(38)(39)(40)(41)(42) and are involved in cell cycle control and suppression of apoptotic cell death (reviewed in Ref. 43). Deficient RSK2 mutations associate with Coffin-Lowry syndrome (CLS), which presents complex manifestations including mental retardation (44 -46); histone H3 phosphorylation induced by EGF stimulation in CLS fibroblasts was deficient, suggesting that selective chromatin remodeling required RSK2 (47). The recently discovered RSK4 is commonly deleted in patients with X-linked mental retardation (XLMR) and may be a candidate XLMR gene (48). The general structure of RSKs contains complete N-terminal and C-terminal catalytic domains (called NTD and CTD, respectively), separated by an intervening sequence (referred to as linker). Activation requires phosphorylation in both NTD and CTD activation loops. In addition, RSKs contain multiple regulatory phosphorylation sites in the linker (49 -51) and C-terminal tail sequences (52,53). Known substrates of RSK1-3, such as CREB (39 -42), the transcriptional co-activators p300 and CREB-binding protein (54), and estrogen recep-tor (55), are phosphorylated by the NTD, whereas the linker and CTD have a role in NTD activation (49 -51, 56). Activation of RSK1 involved sequential phosphorylation of at least four sites, starting with phosphorylation through ERK of a threonine in the CTD activation loop and a serine in the linker, further phosphorylation of a linker serine through CTD, and phosphorylation of a serine in the NTD activation loop, leading to full kinase activity (50). Interestingly, in RSK2 activation PDK1 (and possibly other independent upstream kinases) rather than CTD phosphorylated the NTD activation loop serine, indicating that RSKs can integrate signals from ERK and independent upstream pathways (51). Similarly to the intermolecular associations between other kinases through dimerization (e.g. ERK (57,58) or scaffold proteins (59)), physical interactions of RSKs and MAPKs are facilitated through specific RSK C-terminal docking sites maintaining selective pathway connectivities and directing MAPKs to specific subcellular locations (53, 60 -62).
Recently, two novel members of the RSK family were discovered, MSK1 and RSKB (61,63,64), which in contrast to RSK1-3 were under dominant p38 control but also responded, albeit more weakly, to activation of the ERK pathway. Intriguingly, both MSK1 and RSKB appeared to phosphorylate CREB, thus indicating a symmetric downstream effector of ERK and p38 pathways (42,61,63). MSK1 is a potential histone H3 kinase and may phosphorylate HMG-14 protein (65). Similar to other RSKs (60), both MSK1 and RSKB located to the nucleus (61,63), indicating roles in specific subcellular compartments. Given the association of deficient RSK2 mutations with CLS (44 -46) and of RSK4 deletion with XLMR (48), it is intriguing that RSKB maps to the BBS1 locus (66), linked with Bardet-Biedl syndrome with manifestations quite similar to CLS including mental retardation (66 -68). Here we report a study on the control of RSKB activation through NTD, CTD, and linker phosphorylation sites. RSKB in cells stimulated by TNF treatment was activated within 10 min, and its activation persisted for prolonged periods, in contrast to the burst and rapid return to basal activity levels of the upstream MAPKs, p38 and ERK, and of the parallel RSK1-3, indicating that RSKB selectively escaped from a more common down-regulatory control of MAPKs and other RSKs.
Expression Constructs and Mutagenesis-Expression constructs for MAPKs and wtRSKB were described by Pierrat et al. (61). MEK1 was obtained from Stratagene (La Jolla, CA). RSKB point mutants were generated by site-directed mutagenesis using the Altered Sites in vitro mutagenesis system (Promega) according to the recommendations of the manufacturer.
Cell Culture, Transfection, and Extract Preparation-HeLa (ATCC CCL2) and HEK 293 (ATCC CRL 1573) cells were cultured in humidified air with 5% CO 2 at 37°C. HeLa cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 2 mM L-glutamine, 100 units/ml penicillin, 10 g/ml streptomycin, pH 7.4. HEK 293 cells were cultured in minimal essential medium supplemented with 10% fetal calf serum, 2 mM L-glutamine, 100 units/ml penicillin, 10 g/ml streptomycin, pH 7.4. RSKB-transfected cells activated by co-transfection with upstream kinases were cultured for 2 days including serum starvation for the last 16 h in 0.3% fetal calf serumcontaining medium. Kinase inhibitors SB202190 (10 M) and PD98059 (50 M) were added together with the starvation medium; a second dose of inhibitors (10 M SB202190 or 25 M PD98059) was added 1 h before harvesting the cells. Nontransfected HeLa cells were cultured for 2 days after splitting including a terminal 16-h serum starvation as above; typically, SB202190 (10 M) and PD98059 (50 M) were added 1 h before harvesting cells, and the activating reagents (arsenite, PMA, TNF, and similar reagents, as indicated) were added 30 min before harvesting cells, unless stated otherwise. After stimulation, cells were washed with ice-cold phosphate-buffered saline and lysed in buffer A containing 150 mM NaCl, 1% Triton X-100 (buffer A: 50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM EGTA, 10% glycerol, 500 M dithiothreitol, 5 mM NaPP i , 1 mM Na 3 VO 4 , 50 mM NaF, Complete TM protease inhibitor mixture (Roche Molecular Biochemicals)). Cell lysates were cleared by centrifugation at 14,000 rpm for 10 min at 4°C. Protein concentration was determined using the BCA reagents (Pierce).
Immunoprecipitation and Kinase Assays-Normalized amounts of cell lysates (equivalent to 200 g of total protein for transfected RSKB kinase assays; 1.0 and 0.050 mg of total protein equivalent with nontransfected cells for endogenous RSKB and RSK1/RSK2/RSK3 kinase assays, respectively) were precleared twice with 30 l of protein G-Sepharose slurry for 20 min at 4°C under constant agitation. Antibodies were added to precleared lysates and incubated overnight at 4°C on a rotary wheel. 30 l of protein G-Sepharose slurry were added. After a 1-h incubation, immune complexes were pelleted and washed twice with kinase (min) buffer (50 mM Tris-HCl, pH 7.5, 0.1 mM EGTA, 140 mM KCl) supplemented with 5 mM NaPP i . For in vitro kinase assays, beads were resuspended in 25 l of kinase (plus) buffer (33 M CREBtide, 30 M ATP, 10 mM MgCl 2 , 0.02 Ci/l [␥-33 P]ATP, in kinase (min) buffer) and incubated at 22°C for 30 min under constant agitation. Reactions were stopped by pipetting 20 l of the kinase reaction on a 1 ϫ 1-cm sheet of chromatography filter paper P81 (Whatman); filters were washed (four times) with 1 ml of 0.75% phosphoric acid, quickly immersed in acetone, and allowed to dry. Incorporated radioactivity was measured by liquid scintillation counting in a 1214-Rackbeta counter. Alternatively, for larger scale experiments, the reactions were stopped by the addition of 100 l of 0.75% phosphoric acid. 100 l of the resulting reaction were filtered in a 96-well Millipore phosphocellulose filter plate (Bedford, MA), washed (five times) with 100 l of 0.75% phosphoric acid, washed once with ethanol, and air-dried. Bound radioactivity was measured in a Packard (Meriden, CT) top counter, using 100 l of microscintillation mixture. For K m determinations, reaction rates were calculated from RSKB kinase assays run for 10 min at 30°C, in triplicate, using an estimated (by Coomassie staining) 40 ng of recombinant FLAG-RSKB from transiently transfected HEK 293 cells. K m was determined from Lineweaver-Burk plots by linear extrapolation. K m for ATP determined at 45 M CREBtide concentration was 35 M; K m for CREBtide at an ATP concentration of 100 M was 4 M. Kinase activity of ERK and p38 was measured using MBP and ATF-2 peptide, respectively, as previously reported (69). PhosphorImager readings of the incorporated radioactivities were not corrected for background incorporation in empty bead controls.
Purification of FLAG-RSKB-For large scale purification, the KpnI/ BamHI DNA fragment of pAlter-FLAG-RSKB (61), encoding FLAG-wtRSKB, was cloned into pREP7 (Invitrogen, Leek, The Netherlands), and a permanent HEK 293/pREP7-wtRSKB cell line harboring this episome was established by selection (200 g/ml hygromycin) and limiting dilution cloning. Stably transfected cells were grown in suspension spinner culture (500 ml) and treated with 50 mM sodium arsenite for 30 min (active RSKB), or arsenite was omitted (inactive RSKB). Cells were centrifuged (1500 rpm, 15 min, 4°C) and washed once with ice-cold phosphate-buffered saline. Pellet was resuspended in 20 ml of ice-cold 1% Triton X-100 in buffer A (30 min, 4°C, gentle agitation). NaCl was adjusted to 150 mM, and cell extract was centrifuged (10,000 ϫ g for 10 min). Extract was incubated with 5 ml of M2 affinity gel equilibrated in 150 mM NaCl in buffer A (1 h). Agarose slurry was settled in a column, washed with 20 bed volumes of 150 mM NaCl in buffer A and 6 bed volumes of buffer B (50 mM Tris-HCl, pH 7.5, 0.1 mM EGTA, 0.5 M dithiothreitol, 10% glycerol, 150 mM NaCl), and FLAG-RSKB was eluted with 0.1% Nonidet P-40, 300 g/ml FLAG peptide in buffer B. Fractions containing RSKB were determined by SDS-PAGE and pooled, and aliquots were frozen in N 2 and stored at Ϫ80°C. A typical yield was about 10,000 units/500 ml of culture (1 unit catalyzes the transfer of 1 pmol of phosphate/min to CREBtide).

Both Catalytic Domains of RSKB Are Required for Effector
Function-To delineate the significance of phosphorylations in RSKB, mutations were introduced in the homologous activa-tion loop phosphorylation sites Ser 196 (NTD) and Thr 568 (CTD), S196A-RSKB, and T568A-RSKB (Fig. 1A) (61). Residue Ser 196 was also mutated to Glu (S196E-RSKB) to mimic the effect of negative charge associated with phosphorylation. These various mutants were introduced in HEK 293 cells and activated by co-transfection of mMKK6/p38 or MEK1/ERK upstream kinases in the presence and absence of SB202190 and PD98059. Precipitation kinase assays were performed with cell lysates, as indicated in the legend to Fig. 1B. Transfected wtRSKB responded to mMKK6/p38 in a series of experiments with a mean 27-fold activation (range 14 -30-fold). The response to MEK1/ERK in parallel experiments consistently was lower (mean 11-fold activation, range 8 -13-fold). In contrast to wtRSKB, all three activation loop mutants were nonresponsive to either mMKK6/p38 or MEK1/ERK (Fig. 1B) and to arsenite and PMA treatment in HEK293 cells transfected with these various RSKB constructs only (data not shown). This is consistent with the previous finding that mutations in either ATP binding site of NTD and CTD, K65A-RSKB and K440A-RSKB, resulted in inactive RSKB mutants (61). Interestingly, Dalby et al. (50) found that Ala mutations of the homologous residues in RSK1, Ser 222 (in the NTD) or Thr 574 (in the CTD), created mutants with partial activity in phorbol ester-treated COS1 cell transfectants. Furthermore, it is noted that the RSK2 mutant S227E-RSK2 (51), which is homologous to inactive S196E-RSKB, displayed strong activation by EGF treatment in COS7 cell transfectants, showing that here the Glu mutation in the NTD was permissive for activation of the whole enzyme. These divergent properties of RSK1, RSK2, and RSKB, which easily are explained by structural differences and experimental conditions, point to distinct control mechanisms among RSKs through distinct upstream kinases such as PDK1 in addition to MAPKs (51).
Control of RSKB through Linker Residues-The RSKB linker sequence contains three potential phosphorylation sites, Ser 343 , Ser 347 , and Ser 360 (Fig. 1A); of these, Ser 343 and Ser 347 fit the minimal MAPK consensus phosphorylation site, X(S/ T)P (where represents P or aliphatic residue) (32). Ser 343 and Ser 360 are conserved between RSKB, MSK1 (63), other RSKs (70), and p70 rsk (71). Previous studies of RSK1 and RSK2 have shown that kinase activation depended on sequential phosphorylation of such linker Ser/Thr residues (50,51). To investigate these residues in RSKB activation, Ala mutations in Ser 343 and Ser 360 were introduced (S343A-RSKB, S360A-RSKB). Ser 347 , which is not conserved in other RSKs (Fig. 1A), was also mutated to Ala (S347A-RSKB). Furthermore, Ser 360 was mutated to Glu to mimic the effect of phosphorylation. Each of these various mutants was individually introduced in HEK 293 cells and activated by co-transfection of mMKK6/p38 and MEK1/ ERK, and kinase assays were performed as above (Fig. 1B). S360A-RSKB and S343A-RSKB were found inactive and nonresponsive to p38 and ERK activation. In contrast, S360E-RSKB showed slightly elevated basal activity and, more importantly, partially responded to p38 and ERK activation; the inhibition through SB202190 and PD98059 confirmed these pathway connectivities (Fig. 1B). Intriguingly, the impact of the S360E mutation appears to be larger for the p38 pathway, suggesting a pathway-specific role of Ser 360 in RSKB activation. In contrast to Ser 343 , the Ala mutation of adjoining Ser 347 in S347A-RSKB resulted in partial responses to p38 and ERK activation (Fig. 1B).
These data both support and diverge from previous observations with RSK1 and RSK2 (50,51). The Ala mutation of Ser 386 in RSK2, homologous to S360A-RSKB, resulted in low basal activity and nonresponsiveness to EGF; however, a partial activation by an independent upstream kinase, PDK1, was preserved (51). Furthermore, S369A-RSK2 responded partially to EGF and was fully activated by PDK1 (51); the homology assignment of Ser 369 of RSK2 by sequence alignment points to Ser 343 rather than to Ser 347 of RSKB, but in the functional context the mutation of Ser 347 in S347A-RSKB is similar to S369A-RSK2. A further parallel is provided by the full or partial blocking effects of mutations in Ser 364 and Ser 381 of RSK1 (50), homologous to residues Ser 343 and Ser 360 , respectively, in RSKB.
Recombinant RSKB Activation Is Associated with Phosphorylation of Linker Sites, Ser 347 and Ser 360 -To investigate phosphorylation of the nonconserved site Ser 347 , and of Ser 360 , we performed immunoblot studies with anti-phosphopeptidespecific antibodies, PS347ab and PS360ab. These and a Cterminal peptide antibody, CTab, were extensively tested by precipitation and immunoblot analyses ( Fig. 2 and data not  shown). First, active RSKB, which had been purified from arsenite-stimulated stable transfectants and had a specific activity of 170 units/g and kinetic constants K m ϭ 35 M for ATP (at 45 M CREBtide) and K m ϭ 4 M for CREBtide (at 100 M ATP) (Fig. 2B, inset), was used to investigate the relation between phosphorylation, electrophoretic mobility, and kinase activity. Aliquots of active RSKB were treated in vitro with alkaline phosphatase (AP) in the presence and absence of the inhibitor sodium pyrophosphate (NaPP i ) and electrophoresed in SDS-PAGE, and immunoblot studies were performed. Parallel blots were first stained with PS347ab and PS360ab, respectively, stripped, and restained with CTab ( Fig. 2A). APtreated (20 units, 30 min), exhaustively dephosphorylated RSKB was identified in poststaining with CTab at a location consistent with its molecular mass (band a in Fig. 2A) but was nonreactive with PS347ab and barely reactive with PS360ab. The faint staining of dephosphorylated RSKB with PS360ab may be due to low level phosphorylation in Ser 360 possibly as a result of autophosphorylation (49) or to weak cross-reactivity with nonphosphorylated RSKB. RSKB treated with AP in the presence of NaPP i or treated with NaPP i only presented two bands with successive electrophoretic retardation (bands b and c in Fig. 2A) when probed with all PS360ab, PS347ab, and CTab. This confirms the phospho-Ser 347 specificity of PS347ab and supports at least a high degree of phospho-Ser 360 selectivity of PS360ab. Parallel kinase assays were carried out to determine the effect of AP treatment on RSKB activity (Fig.  2B). A low dose of AP inactivated RSKB at ambient temperature; in the presence of 5 mM NaPP i , the dose response shifted, and a clear AP dose-dependent RSKB inactivation became apparent. Intriguingly, in parallel control incubations at ambient temperature and 4°C in the absence of AP, substantially higher activities were found with incubation at 4°C, suggesting rapid decay of activity of purified RSKB with increasing temperature (Fig. 2B).
Similar albeit less well resolved band retardation patterns became apparent in studies with HEK 293 cells, which had been transfected with FLAG-wtRSKB with and without mMKK6/p38 co-transfection. Cell lysates were prepared and analyzed by immunoblotting with PS347ab, PS360ab, and anti-FLAG antibody M2ab (Fig. 2C). In the absence of mMKK6/p38 co-transfection, M2ab detected a band of nonactivated RSKB at the expected apparent molecular mass. This band was not stained by PS347ab, but PS360ab gave a strong signal; the findings with purified RSKB above suggest that this probably is due to Ser 360 phosphorylation of nonactivated RSKB. When RSKB was activated through the co-transfected p38 pathway, two poorly resolved retarded bands were seen. Interestingly, the most retarded of the bands was sensitive to the treatment of the cells with SB202190 (Fig. 2C, compare e.g. lanes 2 and 3 in the PS360ab blot). Since RSKB from cells co-transfected with p38/mMKK6 and treated with SB202190 was inactive (Fig. 1B), this most retarded band appeared to correlate with active enzyme.
p38 Phosphorylates and Activates RSKB in Vitro-Phosphorylation, electrophoretic retardation, and activity were investigated with active and inactive RSKB purified from stably transfected HEK 293 cells cultured in the presence or absence of arsenite, respectively (Fig. 2D). Aliquots of inactive RSKB (100 ng, 0.5 units) were incubated in kinase assay buffer containing [␥-33 P]ATP together with or without activated p38 that had been immunoprecipitated from p38/mMKK6 co-transfected cells; as control, active RSKB (100 ng, 8.5 units) was incubated in parallel. By SDS-PAGE, inactive RSKB presented a band at the expected molecular mass (Fig. 2D, left, lane 2), whereas active RSKB in addition showed retarded bands (Fig.  2D, left, lane 3). An aliquot of inactive RSKB that had been treated in vitro with activated p38 showed a band retardation pattern similar to that of active RSKB (Fig. 2D, left, lane 1). Kinase assays were then performed in vitro with aliquots of inactive RSKB (20 ng, 0.1 units) in the presence and absence of activated p38 immune complex and SB202190, as indicated, using CREBtide as substrate (Fig. 2D, right). The treatment with activated p38 elicited a strong activation of RSKB, which was fully abrogated in the presence of 10 M SB202190, showing that p38 was sufficient to activate RSKB in vitro.
Known Kinase Inhibitors Block RSKB Activity-To explore pharmacological tools, known kinase inhibitors of the isoquinoline-sulfonyl class, H89 (72), and the bisindolyl-maleimide class, Ro31-8220 (73), were studied in in vitro kinase assays using purified active RSKB and CREBtide as substrate; both compounds turned out to be potent RSKB inhibitors with IC 50 of about 100 nM (H89) and from 5 to 30 nM (Ro31-8220) (Fig. 3). H89 and Ro31-8220 originally had been reported as selective cAMP-dependent protein kinase and protein kinase C inhibitors, respectively (72, 73) but later were found to inhibit MSK1 and MAPKAP-K1 (63,65,74). It is noted that pyridinyl imidazole compounds such as SB202190 in parallel RSKB assays had substantially higher IC 50 values when compared with H89 and Ro31-8220, respectively.
Diverse Stimuli Activate Endogenous RSKB-To investigate RSKB in nontransfected cells, a number of cell lines were first tested for expression of endogenous RSKB; consistent with previous RT-PCR analyses (61), endogenous RSKB protein was detected in HeLa, SK-N-SH, SK-N-MC, and HEK 293 cells (Fig. 4A). HeLa cells were then stimulated with various agents and the response of endogenous RSKB studied by precipitation kinase assays and immunoblotting (Fig. 4C). Control precipitation kinase assays with matched preimmune and immune rabbit CTab serum with activated and nonactivated cells confirmed the specificity of this antiserum (Fig. 4B). Treatment with TNF resulted in a strong activation; PMA and ionomycin stimulated RSKB to an even greater extent, whereas EGF FIG. 2. Retarded electrophoretic mobility associates with RSKB linker site phosphorylation and kinase activity. A, sequential RSKB phosphorylation associated with retarded electrophoretic mobility. Purified active RSKB aliquots were treated in vitro with AP (20 units, 30 min) in the presence or absence of phosphatase inhibitor NaPP i and with NaPP i alone, as indicated. Parallel immunoblots were stained with either PS347ab or PS360ab; both blots were stripped and reprobed with CTab to reveal protein loading, as indicated. Bands are labeled a, b, and c with increasing electrophoretic retardation. Blots revealed in autoradiograms. B, phosphatase treatment inactivates RSKB. Purified active RSKB was treated with various amounts of AP with and without added NaPP i for 30 min at ambient temperature; parallel controls without added AP were kept 30 min at ambient temperature (0), or on ice (0; 4°C), as indicated. Phosphocellulose filter kinase assays were performed using CREBtide as substrate. triggered a lesser response and insulin triggered no response (Fig. 4C, bottom). The activation of RSKB by the various protocols was associated with retarded migration in the immunoblots, but the broad retarded bands were never resolved in two bands comparable with recombinant RSKB (Fig. 4C, top). SB202190 and PD98059 were next used to dissect pathways upstream of RSKB (Fig. 4D). PD98059 in the conditions used only partially blocked MEK1, but the effect documented the ERK pathway connection. It appeared that TNF and ionomycin triggered RSKB through p38, but both p38 and ERK pathways operated in the RSKB response to EGF and PMA stimulation. This dual pathway connectivity of endogenous RSKB was confirmed by precipitation kinase assays with cells transfected with the upstream kinases mMKK6, p38, MEK1, and ERK1 (Fig. 4E). These findings confirm that p38-and ERK pathway signals converge on RSKB, as previously proposed from studies of transfected cells (61).
Transient p38 Activation Results in Persistent Activity of Endogenous RSKB-MAPK signals may result in transient and in long lasting cell responses. To investigate mechanisms translating transient signal input in persistent response, the relative kinetics of p38, ERK, RSK1, and RSKB activation were studied in HeLa cells treated with TNF (Fig. 5). The expected transient burst of p38 and ERK activity in the continued presence of TNF was seen with a peak about 10 min after TNF addition and a return to basal levels within 30 min, as shown by immunoblots with activation-specific antibodies to p38, ERK1, and RSK1 and kinase assays for p38 and ERK with ATF-2 and MBP peptide substrates, respectively (Fig. 5) (69,75). In sharp contrast, high RSKB activity was found in kinase assays even after 2-3 h, consistent with staining of retarded bands in immunoblots with PS347ab and PS360ab (Figs. 5 and 7).
To further investigate a potentially autonomously persistent activity of RSKB, HeLa cells were stimulated with TNF, SB202190 was added at various times up to 60 min after the onset of TNF treatment, and RSKB activities were determined 90 min after the start of TNF treatment in parallel (Fig. 6). Surprisingly, SB202190, even when added 60 min after TNF, abrogated RSKB activity as determined at 90 min, whereas RSKB activity persisted in the absence of SB202190 as before (Fig. 6). RSKB activity reproducibly had been found insensitive to SB202190 in cells stimulated through the ERK pathway (Fig. 4E), which strongly argues against a direct effect of SB202190 on RSKB in Fig. 6. Furthermore, the related compound SB202474, which is devoid of p38-inhibitory activity, did not interfere with the persistence of RSKB activation (not shown). While activities of SB202190 on some other cell component from experiences with the related compound SB203580 (76 -78) cannot be ruled out entirely, the findings in Fig. 6 most likely resulted from an effect of SB202190 on p38. This then suggests that RSKB, once activated by noninhibited p38, was maintained in the activated state by basal level p38 activity. To investigate whether the ERK-dependent RSKs, RSK1-3, showed similar persistent activation, HeLa cells again were stimulated with TNF for various times; RSKB, RSK1, RSK2, and RSK3 were precipitated in parallel; and kinase and immunoblot assays were performed. Interestingly, in contrast to the persistent activation of RSKB, the activation of all RSK1, RSK2, and RSK3 was transient and paralleled the ERK activation (Fig. 7, A and C). The sensitivity to the inhibitory activities of SB202190 and PD98059 treatment confirmed the respective p38 and ERK pathway connectivities of (a) RSKB and (b) RSK1-3 (Fig. 7B). These findings support the view that RSKB selectively escaped control mechanisms, which in general deactivated MAPKs and RSKs after a transient burst despite the continued presence of the original stimulus.

DISCUSSION
The RSK family of kinases, RSK1-3, MSK1, and RSKB, share a number of common properties. They are composed of two complete catalytic kinase domains, with intervening linker and C-terminal tail sequences containing regulatory elements; their activation progresses by stepwise phosphorylation from CTD to NTD, the latter being the effector of downstream substrate phosphorylations (50,51,56,79). Furthermore, they typically locate to the nucleus to target nuclear substrates such as CREB (39 -42, 47, 61, 63, 65). The present studies show that RSKB is activated by various cell treatment protocols that also have been reported to activate other RSKs (2); TNF, ionomycin, PMA, and EGF elicited RSKB activation, whereas insulin had no effect. Despite these similarities, substantial evidence indicates a functional independence of the individual RSK family members. First, none of the RSKs complements deficient RSK2 in CLS (46) and possibly RSK4 in XLMR (48). Second, while RSK1-3 are activated through ERK, RSKB, as shown by its responses to MAPKs, effects of SB202190 and PD98059, and direct activation in vitro, mainly depended on p38 and to a lesser extent ERK (61,63). Third, RSK1 retained partial activation potential when either NTD or CTD was inactivated (50); in contrast, inactivating mutations of either NTD or CTD fully abolished RSKB activity. Furthermore, RSK2 integrated ERK and PDK1 signals, and the isolated NTD of RSK2 remained permissive for PDK1 activation (51); in contrast, at present there is no evidence for a similar signal integrative function of RSKB. The activation of RSKB by p38 in vitro shows that p38 alone, possibly combined with autophosphorylation, suffices to convert RSKB to the fully activated state. This supports the view that while ERK may substitute for p38 in activating RSKB, this presents a convergence point rather than a site that can integrate independent signals from two pathways.
Much has been learned about the role of RSK1-3 in growth factor stimulation, cell cycle, and survival control (for recent reviews, see Refs. 2 and 43). Less is known at present about the role of the more recently discovered MSK1 and RSKB (61,63,64). MSK1, through phosphorylation of histone H3 and HMG-14 protein, was reported to be a kinase functioning in chromatin remodeling associated with immediate early gene induction (65). RSKB and MSK1 as all RSKs, in parallel to Ca 2ϩ /calmodulin-and cAMP-dependent signaling, activate CREB (reviewed in Ref. 80); it will be an intriguing task to unravel the selective effects of CREB activation through the various pathways. In such studies, it will be important to remember that an inhibitor such as H89, previously thought to Lysates of HEK293, HeLa, SK-N-SH, and SK-N-MC cells were immunoprecipitated with rabbit CTab, SDS-PAGE was performed, and immunoblots were probed with mouse CTab, as indicated. The first lane shows rabbit preimmune serum control with HEK293 cell lysate. B, control immunoprecipitation of preimmune and immune rabbit CTab. Endogenous RSKB was precipitated from aliquots of HEK293 cell lysates with equal volumes of matched preimmune serum and serum after immunization with the C-terminal peptide. Kinase assays were performed in duplicates as described, using CREBtide as substrate (top). Aliquots of the CTab immunoprecipitates were analyzed in immunoblots using the independent mouse CTab as revealing agent (bottom). A representative (of n ϭ 2) experiments is shown. E, empty bead control in the precipitation; PS, preimmune serum; S, rabbit CTab immune serum; SϩSB, rabbit CTab precipitates from lysates of SB202190-treated HEK293 cells. C, electrophoretic retardation (top) and activation (bottom) of endogenous RSKB. HeLa cells were treated with TNF (200 ng/ml), EGF, insulin, and PMA (all 100 ng/ml) and 2 M ionomycin for 30 min, and kinase assays with rabbit CTab precipitates from cell lysates were performed with CREBtide as substrate, as indicated (bottom). Immunoblots of parallel precipitates were probed with mouse CTab, as indicated (top). D, effects of p38 and ERK pathway inhibition on endogenous RSKB activation. HeLa cells were stimulated with TNF, EGF, PMA, and ionomycin in the presence and absence of SB202190 and PD98059, and RSKB precipitation kinase assays were performed with cell lysates using CREBtide as substrate, as indicated. E, endogenous RSKB activated through p38 and ERK pathways. HEK293 cells were transfected with p38, mMKK6, ERK, and MEK1 and cultured with or without SB202190 and PD98059, and RSKB precipitation kinase assays were performed with lysates normalized to total protein content, as indicated.
The distinct and slow deactivation kinetics of RSKB, when compared with the upstream MAPKs and the parallel ERK-dependent RSKs, further support a selective RSKB function(s). In many cell activation protocols, the early burst of MAPK and RSK activities is rapidly reduced to basal levels despite the continued presence of the original stimulus, due to the activities of specific phosphatases, to receptor modulation, and to potential further mechanisms (82,83). In other instances, long lasting activations have been reported. EGF and NGF both activated the ERK pathway in PC12 cells, but a striking difference was that only NGF was able to promote differentiation correlating with persistent ERK cascade activation, whereas EGF-induced activation was rapidly down-regulated at a site upstream of MAPK kinase (84 -86). Persistent activation of a RSK, which was identified by co-immunoprecipitation with ERK1, correlated with PC12 differentiation (86). An intriguing example of a mechanism that allows translation of transient signal input into persistent response is provided by Ca 2ϩ /calmodulin-dependent protein kinase II, where an initial activating phosphorylation step interferes with autoinhibitory function, leaving the kinase in an autonomously activated state (87)(88)(89)(90)(91). In our studies, RSKB presented a different type of signal processing leading to persistent activation; high RSKB FIG. 5. Transient p38 and ERK signals induce persistent activation of endogenous RSKB. HeLa cells were stimulated with TNF (200 ng/ml) for 0, 2, 10, 30, and 120 min. Cell lysates normalized for total protein content were immunoprecipitated in parallel with rabbit CTab, p38, ERK, and RSK1 antibodies for kinase assays and immunoblots. p38 and ERK kinase substrates ATF-2 and MBP peptides, respectively, were analyzed by SDS-PAGE and PhosphorImager analysis, as indicated (top row, ATF-2 and MBP); numbers under gel bands represent activation factors calculated from PhosphorImager counts. Whole cell lysates normalized for total protein content were analyzed in parallel immunoblots with activation-specific antibodies to p38 (P-p38ab), ERK1 (P-Erkab), and RSK1 (P-p90rskab), as indicated (middle rows, left and right). Parallel RSKB CTab precipitates from normalized lysates were analyzed in immunoblots stained with mouse PS347ab and PS360ab antibodies, as indicated (middle rows, left). RSKB activation at various times was determined in CTab precipitation kinase assays from normalized lysates, as indicated (left bottom).
FIG. 6. Persistent RSKB activation is sensitive to treatment with p38 inhibitor, SB202190. HeLa cells were stimulated by TNF (200 ng/ml) treatment for 90 min. In a series of parallel assay cultures, SB202190 was added either 30 min before TNF (Ϫ30), or 15 min (ϩ15), 30 min (ϩ30), 45 min (ϩ45), and 60 min (ϩ60) after TNF. Controls were left untreated (NS) or treated only with TNF (TNF 1.5 h). RSKB precipitation kinase assays with lysates normalized to total protein content were performed with CREBtide as substrate in parallel 90 min after the TNF addition. All activities normalized to the nonstimulated control (NS) as indicated. A representative series of parallel experiments (of n ϭ 2) is shown. FIG. 7. Differential deactivation kinetics of RSKB and RSK1, RSK2, and RSK3 in response to p38 and ERK activation. A, HeLa cells were stimulated with TNF (200 ng/ml) for various times, and RSKB and RSK1-3 activities were determined in precipitation kinase assays of lysates normalized to total protein content, using CREBtide as substrate, as indicated. A representative series of parallel experiments (of n ϭ 2) is shown. B, HeLa cells were stimulated by TNF treatment for 15 min in the presence and absence of SB202190 and PD98059, and immunoprecipitation kinase assays were performed with lysates normalized to total protein content, using CREBtide as substrate, as indicated. Kinase activities normalized to the response at 15 min for each RSKB, RSK1, RSK2, and RSK3, respectively. A representative series of parallel experiments (of n ϭ 2) is shown. C, endogenous RSKB activation at various times correlates with electrophoretic retardation and Ser 347 /Ser 360 phosphorylation. HeLa cells were treated with TNF and ionomycin for various times in the presence or absence of SB202190 and PD98059, and immunoblots with lysates normalized to total protein content were performed with PS347ab and PS360ab, as indicated. Blots revealed in autoradiogram. A representative series of parallel experiments (of n ϭ 2) is shown. activation persisted even after 2-3 h despite the rapid deactivation of p38, ERK, and RSK1-3. However, rather than being autonomous, this persistent RSKB activity depended on a cell component that was sensitive to the presence of SB202190. The insensitivity of RSKB to SB202190 treatment of cells when activated through the ERK pathway strongly argues against a direct effect of the inhibitor on RSKB. While the possibility cannot be ruled out entirely that SB202190 modulated some unknown cell component, its most likely target was p38. This suggests that while the initial activation of RSKB required high p38 activity associated with the stimulated burst response, substantially lower p38 activities sufficed to maintain RSKB in an activated state and allowed it to escape mechanisms that efficiently deactivated p38, ERK, and RSK1-3. Several explanatory models may be considered. The persistent RSKB activation might be linked to the presence of the nonconserved Ser 347 and possibly other phosphorylation site(s), and phosphorylation-associated molecular conformation(s). Activated RSKB might be inaccessible to phosphatases, or specific phosphatase(s) might be lacking. Furthermore, upon activation, RSKB might translocate to a subcellular compartment(s) together with an activated p38 subpopulation not detected in kinase assays, which of necessity average total cellular p38; a strong association with p38 is consistent with the initial discovery of RSKB by intracellular interaction trap screen and with co-immunoprecipitation (61). Further studies are required to dissect these mechanisms.