Hormones and growth factors induce the activation of a number of protein kinases that belong to the AGC subfamily, including isoforms of PKA, protein kinase B (also known as Akt), PKC, S6K p70 (ribosomal S6 kinase), RSK (p90 ribosomal S6 kinase) and MSK (mitogen- and stress-activated protein kinase), which then mediate many of the physiological processes that are regulated by these extracellular agonists. It can be difficult to assess the individual functions of each AGC kinase because their substrate specificities are similar. Here we describe the small molecule BI-D1870, which inhibits RSK1, RSK2, RSK3 and RSK4 in vitro with an IC50 of 10–30 nM, but does not signi-ficantly inhibit ten other AGC kinase members and over 40 other protein kinases tested at 100-fold higher concentrations. BI-D1870 is cell permeant and prevents the RSK-mediated phorbol ester- and EGF (epidermal growth factor)-induced phosphoryl-ation of glycogen synthase kinase-3β and LKB1 in human embry-onic kidney 293 cells and Rat-2 cells. In contrast, BI-D1870 does not affect the agonist-triggered phosphorylation of substrates for six other AGC kinases. Moreover, BI-D1870 does not suppress the phorbol ester- or EGF-induced phosphorylation of CREB (cAMP-response-element-binding protein), consistent with the genetic evidence indicating that MSK, and not RSK, isoforms mediate the mitogen-induced phosphorylation of this transcription factor.
Members of the AGC [PKA (cAMP-dependent protein kinase)/protein kinase G/PKC (protein kinase C)] subfamily of protein kinases, including isoforms of PKA, PKB (protein kinase B), S6K (p70 ribosomal S6 kinase), PKC, RSK (p90 ribosomal S6 kinase) and MSK (mitogen- and stress-activated protein kinase), mediate many of the cellular effects of extracellular agonists by phosphorylation of key regulatory proteins. These protein kinases possess catalytic domains that are approx. 40% identical with each other, but contain distinct non-catalytic domains, and participate in diverse signalling pathways. AGC kinases are also activated by different mechanisms. For example, agonists that stimulate adenylate cyclase induce the activation of PKA , whereas those that stimulate PI3K (phosphoinositide 3-kinase) induce the activation of PKB, S6K and certain isoforms of PKC [2–4]. RSK isoforms are activated by the MAPK (mitogen-activ-ated protein kinase) family members ERK1 (extracellular-signal-regulated kinase 1) and ERK2 in response to growth factors, phorbol esters and other agonists [5,6], whereas MSK isoforms are activated in vivo by two different MAPK family members, namely ERK1/ERK2 and the stress and cytokine-activated p38 MAP kinase . RSK and MSK isoforms are unusual in that they possess two catalytic domains in a single polypeptide. The N-terminal kinase domain is an AGC family member and catalyses the phosphorylation of all known substrates of these enzymes. The C-terminal kinase domain, which does not belong to the AGC family, is required for the activation of the N-terminal kinase domain (reviewed in ).
In order to study the physiological roles of AGC kinases, a commonly used approach has been to over-express the active forms in cells. However, due to the overlapping substrate specificities of many AGC kinases, it is likely that the over-expression of one member of this kinase subfamily will result in the phosphorylation of substrates that are normally phosphorylated by another AGC kinase. Another strategy has been to over-express catalytically inactive ‘dominant negative’ mutants of AGC kinases in cells. How-ever, such mutants are likely to interact with and inhibit the upstream protein kinase(s) that they are is activated by, and thus prevent the ‘upstream’ kinase(s) from phosphorylation of other cellular substrates. For example, a dominant negative RSK may interact with ERK1/ERK2 preventing the activation of MSK isoforms and hence the phosphorylation of CREB (cAMP-response-element-binding protein) . Furthermore, in Saccharomyces cerevisiae, over-expression of catalytically inactive Rck2p, a kinase that binds to and is activated by the Hog1P MAPK, sequestered the substrate-docking site of the Hog1P kinase, thereby preventing Hog1P from interacting with other substrates . Thus catalytically inactive Rck2P is acting as a dominant negative mutant of Hog1P and not Rck2P.
A few small cell-permeant molecules have been developed that inhibit the activation of AGC kinase members. For example, several compounds that prevent the activation of MKK-1 (MAPK kinase-1) have been described, which potently inhibit the activation of ERK1/ERK2 and hence the activation of RSK and MSK isoforms [11,12]. Inhibitors of PI3K have been employed to inhibit the activation of PKB and S6K , whereas rapamycin, a drug that inhibits the protein kinase mTOR (mammalian target of rapamycin), prevents the activation of S6K . Although these compounds have been valuable in providing evidence for or against a role of some AGC kinases in the regulation of particular cellular processes, the development of compounds that inhibit individual AGC kinase members specifically would be a major advance. However, thus far, mainly rather non-specific AGC kinase inhibitors have been deployed for this purpose. For example, the use of Ro318220 has been described in over 500 publications to inhibit conventional PKC isoforms, but it also inhibits RSK and S6K with similar potency , as well as other kinases that are not AGC-family members . H89, a compound originally identified as a PKA inhibitor, also inhibits MSK1, S6K and ROCK (Rho-dependent protein kinase) with a similar IC50 value to PKA [16,17]. Here, we demonstrate that BI-D1870 is a remarkably specific inhibitor of RSK isoforms, with a >500-fold greater selectivity over nine other AGC kinases tested. We also demonstrate that BI-D1870 inhibits RSK activity relatively specifically in cells.
MATERIALS AND METHODS
BI-D1870 was synthesized as a racemate at Boehringer Ingelheim Pharma GmbH & Co. PD 184352 was obtained by custom synthesis, LY 294002 was from Calbiochem, EGF (epidermal growth factor) and microcystin-LR were from Life Technologies Inc (Paisley, Scotland, U.K.), FBS (foetal bovine serum) and other tissue culture reagents were from BioWhittaker, Complete™ protease-inhibitor cocktail tablets were from Roche. Forskolin, PMA, DMSO, antimycotic/antibiotic solution and dimethyl pimelimidate were from Sigma, while the precast 4–12% Bis/Tris gradient SDS/PAGE gels were from Invitrogen. GST (glutathione S-transferase)–ERK2 was expressed in Escherichia coli and activated with MKK1  and His6–PDK1 (phosphoinositide-depend-ent kinase 1) was expressed in Drosophila Sf21 cells . Apart from RSK3 (#14-462) and RSK4 (#14-702) that were purchased from Upstate, the protein kinases employed in Table 1 were gene-rated in the Division of Signal Transduction Therapy, University of Dundee. The protocols utilised to assay the 54 protein kinases listed in Table 1 are described in the Supplementary section (see http://www.BiochemJ.org/bj/401/bj4010029add.htm).
|10 μM||0.1 μM||10 nM|
|Protein kinase||Kinase family||% Remaining kinase activity|
|10 μM||0.1 μM||10 nM|
|Protein kinase||Kinase family||% Remaining kinase activity|
The following buffers were used: Lysis Buffer [50 mM Tris/HCl, pH 7.5, 1 mM EGTA, 1 mM EDTA, 1% (w/v) Triton-X 100, 1 mM sodium orthovanadate, 50 mM sodium fluoride, 5 mM sodium pyrophosphate, 0.27 M sucrose, 1 μM microcystin-LR, 0.1% 2-mercaptoethanol and Complete™ proteinase inhibitor cocktail (one tablet per 25 ml)], Buffer A (50 mM Tris/HCl, pH 7.5, 0.1 mM EGTA and 0.1% 2-mercaptoethanol), Buffer B (50 mM Tris/HCl, pH 7.5, and 0.1 mM EGTA) and SDS-sample buffer [50 mM Tris/HCl, pH 6.8, 2% (w/v) SDS, 10% (v/v) glycerol and 1% (v/v) 2-mercaptoethanol].
Expression, purification and activation of RSK1 and RSK2
cDNA clones encoding rat RSK1 were provided kindly by J. Avruch (Department of Molecular Biology, Massachusetts General Hospital, MA, U.S.A.) and the human RSK2 cDNA was described previously . The rat RSK1 sequence contained an HA-tag positioned N-terminal to the initiating methionine residue and differed from the Genbank entry M99169, at three residues E551G, S637N and G697A. The human RSK2 sequence differed from the Genbank entry NM_004586 at one residue, V45G. In order to generate transfer vectors for baculo-virus production, the RSK1 cDNA was cloned into the EcoRI site of pFastBAC-Hta vector (Life Technologies Inc), the resulting construct encodes RSK1 with an N-terminal His6-tag. The RSK2 cDNA was cloned into the BamHI site of a modified pFastBAC1 vector resulting in the residues MAHHHHHHGS being added N-terminally to Pro2 of the native sequence. The resulting constructs were then used to generate recombinant baculovirus using the Bac-to-Bac® system (Life Technologies Inc) following the manu-facturer's protocol. The resulting baculovirus were used to infect Sf21 cells (1.5×106/ml) at a multiplicity of infection of 5. The infected cells were harvested 72 h post infection and His6-tagged protein purified as described previously for His6–PDK1 . RSK1 and RSK2 were purified with yields of approx. 10 mg/l of infected cells, and were greater than 85% homogeneous as judged by densitometric scanning of Coomassie Blue-stained SDS/PAGE gels. The purified RSK1 and RSK2 isolated in this manner possessed a low basal activity of approx. 5 units/mg when assayed using Crosstide as a substrate , but could be activated by ERK2 to a specific activity of >300 units/mg . In order to activate the purified RSK1 and RSK2, a reaction was set up (2.0 ml total volume) in Buffer A containing RSK1 or RSK2 (0.2 mg/ml), 5 units/ml active GST–ERK2 and 3 μg/ml His6–PDK1, 10 mM magnesium acetate and 0.1 mM ATP. Following incubation for 1 h at 30 °C, the specific activity of the RSK1 and RSK2 enzymes had increased to 150 and 370 units/mg for RSK1 and RSK2 respectively. The reaction was terminated by adding EDTA to a concentration of 20 mM, and passed through a 0.5 ml glutathione–agarose column, equilibrated with Buffer A to remove GST–ERK2. To the eluate from this column, which contained RSK1 and RSK2, NaCl was added to a final concentration of 0.4 M and then the solution was passed over a 0.5 ml heparin–Sepharose column equilibrated in Buffer A containing 0.4 M NaCl to remove PDK1, which interacts with heparin–Sepharose under these conditions . The eluate containing activated RSK1 and RSK2 was dialy-sed into 50 mM Tris/HCl (pH 7.5), 270 mM sucrose, 150 mM NaCl, 0.1mM EGTA, 0.1% 2-mercaptoethanol, 0.03% Brij-35, 1 mM benzamidine and 0.2 mM PMSF, and aliquots were snap frozen in liquid nitrogen and stored at −80 °C prior to use.
The GST–RSK21–389:S381E was expressed in E. coli and purified using glutathione–Sepharose as described previously . In order to activate this fragment of RSK2, a reaction was set up (1.0 ml total volume) in Buffer A containing 0.4 mg/ml GST–RSK21–389:S381E, 50 μg/ml His6–PDK1, 10 mM magnesium acetate and 0.1 mM ATP. Following incubation for 1 h at 30 °C the specific activity of the GST–RSK21–389:S381E was 54 units/mg. The reaction was terminated by addition of EDTA to 20 mM, and then 0.27 M sucrose was added, and aliquots of the activated GST–RSK21–389:S381E were snap frozen in liquid nitrogen and stored at −80 °C prior to use.
Antibodies recognizing LKB1 phosphorylated at Ser431 were raised in sheep . The antibodies that we raised previously against mouse LKB1 protein  did not immunoprecipitate hu-man LKB1 and we therefore raised a further antibody against human LKB1 expressed in E. coli with an N-terminal GST-tag, using a cDNA construct provided generously to us by Dr Nicoletta Resta (Medical Genetics Unit, Department of Biomedicine of Evolutive Age, University of Bari, Bari, Italy). The antibodies were affinity purified on CH–Sepharose coupled covalently to human GST–LKB1 and passed through a column of CH–Sepharose coupled to GST to remove anti-GST antibodies. The antibody that did not bind was selected. This antibody immunoprecip-itates human LKB1 efficiently and was used to immunoprecipitate LKB1 from HEK-293 (human embryonic kidney 293) cells. Antisera recognizing the phosphorylated forms of ribosomal S6 protein (Ser235; available from Upstate) were raised in sheep using the peptide A229KRRRLSpSLRASTS242, where Sp indicates phospho-serine. The antibody was affinity purified on CH–Sepharose that had been coupled covalently to the phosphorylated peptide, then passed over a column of CH–Sepharose coupled to the non-phosphorylated peptide. The flow-through fractions were collected.
The antibody used to immunoprecipitate and immunoblot RSK isoforms, was raised against the peptide RNQSPVLEPVGRS-TLAQRRGIKK (corresponding to residues 712 to 734 of human RSK2), whilst the antibody used to immunoprecipitate MSK1 was raised against the peptide F384KRNAAVIDPLQFHMGVER402. The following antibodies were from Cell Signalling Technology with the catalogue numbers indicated in parentheses: antibodies for immunoblotting ERK1/ERK2 (#9102), ribosomal S6 protein (#2212s) and CREB (#9192), as well as the phospho-specific antibodies recognising the phosphorylated forms of ERK1/ERK2 (#9106S) and GSK3α (glycogen synthase kinase 3α)/GSK3β (#9331S). The monoclonal antibody recognising the PH (pleck-strin homology) domain of PKB (#05-591), as well as the phos-pho-specific antibody recognising CREB phosphorylated at Ser133 (#06-519) and the phospho-specific antibody recognising the FKHR (forkhead in rhabdosarcoma transcription factor) phos-phorylated at Thr24 (#07-162), were from Upstate Inc. In order to immunoblot GSK3α and GSK3β, we combined antibodies raised against the C-terminal residues of rat GSK3α (Q471APDATP-TLTNSS483) and an antibody raised against the GSK3β protein. The antibody used to immunoprecipitate FKHR has been described previously . Secondary antibodies coupled to horse-radish peroxidase for use in immunoblotting were from Pierce.
Protein kinase assays
Purified His6–RSK1, His6–RSK2 or GST–RSK21–389:S381E (1–2 units/ml) were assayed for 10 min at 30 °C in a 50 μl assay mixture in Buffer A containing 30 μM substrate peptide (KEAKEKRQEQIAKRRRLSSLRASTSKSGGSQK), 10 mM magnesium acetate and 100 μM of [γ-32P]ATP. Reactions were terminated and analysed as described previously . The amount of enzyme that catalysed the phosphorylation of 1 nmol of substrate peptide in 1 min was termed one unit.
In order to assay RSK and MSK1 in HEK-293 or Rat-2 cell lysates, these kinases were immunoprecipitated from the cell lysates (0.1 mg of lysate protein for RSK and 0.3 mg for MSK1) and assayed as described previously , except that for RSK assays the immunoprecipitates were washed twice with Buffer A containing 1 mM ATP and twice with Buffer A prior to the assay, as a precaution to ensure dissociation of BI-D1870 from the RSK isoforms.
Cell culture, stimulation and cell lysis
The rat embryo fibroblast cell line, Rat-2, was obtained from the European Collection of Cell Cultures. Rat-2 cells were cultured on 10 cm-diameter dishes in Dulbecco's Modified Eagle's medium supplemented with 10% (v/v) FBS. HEK-293 cells were cultured on 10 cm-diameter dishes in Dulbecco's Modified Eagle's me-dium supplemented with 10% FBS and 1×antimycotic/antibio-tic solution. Prior to stimulation, cells were cultured in the absence of serum for 16 h. Inhibitors were dissolved in DMSO at a 1000-fold higher concentration than they were used at. These inhibitors, or the equivalent volume of DMSO as a control, were added to the tissue culture medium 30 min prior to stimulation, unless indi-cated otherwise. The final concentration of DMSO in the culture medium was 0.1% and had no effect on agonist-induced activ-ation or phosphorylation of any of the substrates examined. The cells were stimulated with the indicated agonists and lysed in 1 ml of ice-cold Lysis Buffer, and centrifuged at 16000 g at 4 °C for 5 min. The supernatants were frozen in liquid nitrogen and stored at −80 °C until use. Protein concentrations were determined using the Bradford method with BSA as the standard.
Immunoprecipitation of LKB1 and FKHR
Polyclonal antibodies (1 mg) raised against human LKB1 or mouse LKB1 or FKHR were coupled covalently to protein G–Sepharose (1 ml) using dimethyl pimelimidate . Rat-2 (0.5 mg) or HEK-293 (1 mg) cell lysate proteins were incubated for 60 min at 4 °C with the anti-LKB1– or anti-FKHR–Protein G–Sepharose conjugates (5 μl). The immunoprecipitates were washed twice with 1 ml of Lysis Buffer containing 0.5 M NaCl, and washed twice with Buffer B. For immunoblot analysis, the beads were resuspended in SDS-Sample Buffer that did not contain 2-mercaptoethanol.
For blots of total cell lysates, 20 μg of lysate protein was used. All samples were subjected to SDS/PAGE and transferred on to nitrocellulose. For experiments in which LKB1, RSK2, FKHR, GSK3α/GSK3β isoforms and the phospho-ribosomal protein S6 and phospho-LKB1 were immunoblotted, the membranes were incubated in 50 mM Tris/HCl (pH 7.5), 0.15 M NaCl, 0.5% Tween 20 and 10% (w/v) dried skimmed milk for 7 h at 4 °C in the presence of 1 μg/ml antibody. Prior to immunoblotting with the phospho-Ser431 LKB1-specific antibody or the phospho-ribosomal protein S6 antibody, the antibodies were incubated with the non-phosphorylated forms of the peptide antigen (10 μg/ml) used to raise the antibodies. All the other antibodies obtained commercially were used a 1000-fold dilution of the stock and 5% (w/v) BSA was used in place of dried skimmed milk as the blocking agent. Proteins were detected using horseradish peroxidase-conjugated secondary antibodies and ECL® reagent (Amersham Pharmacia Biotech).
BI-D1870 is a potent inhibitor of RSK isoforms
BI-D1870, derived from a novel series of dihydropteridinones (Figure 1A), was identified in a kinase selectivity screen performed in the Division of Signal Transduction Therapy in the University of Dundee, as a potent inhibitor of RSK1. We found that BI-D1870 inhibited RSK1 and RSK2 with IC50 values of 10 nM (Figure 1B) and 20 nM (Figure 1C) respectively, when the kinase assays were performed with 100 μM ATP. When the assays were performed at a 10-fold lower ATP concentration, the IC50 of BI-D1870 was reduced to 5 nM for RSK1 and 10 nM for RSK2. These observations indicate that BI-D1870 is a potent ATP-competitive inhibitor of RSK isoforms. An active mutant of RSK2 lacking the C-terminal kinase catalytic domain (RSK21–389:S381E ), was inhibited by BI-D1870, with an IC50 of approx. 30 nM (Figure 1D), indicating that BI-D1870 inhibits the N-terminal AGC kinase domain. BI-D1870 also inhibited RSK3 and RSK4 with similar potency to RSK1 and RSK2 (Tables 1 and 2).
|Protein kinase||Kinase family||ATP concentration in assay (μM)||IC50 (μM)|
|Protein kinase||Kinase family||ATP concentration in assay (μM)||IC50 (μM)|
Effect of BI-D1870 on RSK activity in vitro
BI-D1870 is a specific inhibitor of RSK isoforms
In order to investigate the specificity of BI-D1870, we studied the effect of 10 μM, 0.1 μM and 0.01 μM BI-D1870 at ATP concentrations which approximate the Km constant for ATP, towards a panel of 54 protein kinases which included 14 AGC kinase members [RSK1, RSK2, RSK3, RSK4, MSK1, PKBα, PKBβ, S6K1, SGK1 (serum and glucocorticoid-induced kinase-1), PKCα, PDK1, PKA, PRK2 (PKC-related kinase 2) and ROCK-II], as well as three tyrosine kinases (Table 1). At 0.1 μM BI-D1870, a concentration that inhibited the four RSK isoforms by over 98%, most enzymes on the panel were not affected significantly, apart from PLK1 (polo-like kinase 1) whose activity was reduced by 83%, and Aurora B, DYRK1a (dual-specificity tyrosine-phosphorylated and -regulated kinase 1a), CDK2–A (cyclin-dependent kinase 2 complexed with cyclin A), Lck (lymphocyte kinase), CK1 (casein kinase-1) and GSK3β, whose activities were reduced by 40–70% (Table 1). PLK1 was inhibited by BI-D1870 with an IC50 of 100 nM, whilst the IC50 values for Aurora B, DYRK1a, CDK2–A, Lck, CK1 and GSK3β were 10- to 100-fold higher than that of the RSK isoforms (Table 2). Importantly, the activity of MSK1, the kinase most closely related to RSK, was not inhibited significantly by 0.1 μM BI-D1870.
Evidence that BI-D1870 inhibits RSK activity specifically in HEK-293 cells
To examine whether BI-D1870 could inhibit RSK activity in cells, we investigated the effect of BI-D1870 on the RSK-catalysed phosphorylation of its known substrates in HEK-293 cells, using PMA as an agonist to activate ERK1/ERK2 and RSK isoforms. Cells were incubated with 10 μM BI-D1870, for periods of 15 min for a total of 4 h prior to stimulation with PMA, the phosphoryl-ation of GSK3α at Ser21 and GSK3β at Ser9, which is mediated by RSK under these conditions , was investigated. These experi-ments demonstrated that incubation of cells with BI-D1870 greatly inhibited the PMA-induced phosphorylation of GSK3α and GSK3β (Figure 2A). In contrast, BI-D1870 had little effect at any time point on the PMA-induced activation of ERK1/ERK2, which is catalysed by Raf and MKK1, or the phosphorylation of CREB at Ser133, which is largely catalysed by MSK1 and MSK2 isoforms in fibroblasts . We next studied the effect of varying the concentration of BI-D1870 on the PMA-induced phosphoryl-ation of GSK3 isoforms and another RSK substrate, the protein kinase LKB1 (which is phosphorylated by RSK at Ser431 ) in HEK-293 cells. These experiments showed that the phosphoryl-ation of both proteins was inhibited in HEK-293 cells with an IC50 of approx. 1 μM (Figure 2B).
BI-D1870 inhibits RSK activity in vivo
Stimulation of HEK-293 cells with PMA also activates S6K, whose principal cellular substrate is thought to be the ribosomal protein S6 . Incubation of cells for up to 4 h with 10 μM BI-D1870 did not affect the PMA-induced phosphorylation of the ribosomal S6 protein (Figure 2), indicating that BI-D1870 does not inhibit the activation or activity of S6K in cells. In HEK-293 cells, PMA also induced a weak activation of p38 MAPK, which was also unaffected by BI-D1870 (Figure 2B).
In order to test whether BI-D1870 affected the activation of RSK isoforms, HEK-293 cells were stimulated with PMA in the absence or presence of BI-D1870. RSK1 and RSK2 isoforms were immunoprecipitated with an antibody that recognizes both forms of RSK, the immunoprecipitates were washed and assayed for activity in the absence of BI-D1870. Treatment of the cells with BI-D1870 did not markedly affect the large PMA-induced activation of ERK1/ERK2, RSK isoforms or the phosphorylation of RSK at Ser381 and Thr574, two phosphorylation sites required for the activation of RSK  (Figures 3A and 3B). BI-D1870 also did not affect the PMA-induced activation of MSK1 in HEK-293 cells (Figure 3C).
Effect of BI-D1870 on activation of RSK and MSK1
We also compared the effectiveness of BI-D1870 to inhibit the phosphorylation of GSK3 isoforms with that of PD 184352, a potent inhibitor of the activation of MKK1  (Figures 3A and 3B). BI-D1870 (10 μM) inhibited the PMA-induced phosphoryl-ation of GSK3 isoforms to the same extent as PD 184352 (2 μM). As expected, and in contrast with BI-D1870, PD 184352 also prevented the phosphorylation of ERK1/ERK2 and the phosphoryl-ation and activation of RSK and MSK isoforms as well as the phosphorylation of CREB.
Further evidence that BI-D1870 is a specific RSK inhibitor
We next assessed the effectiveness of BI-D1870 as an inhibitor of RSK activity in EGF-stimulated Rat-2 cells. As observed in HEK-293 cells, BI-D1870 inhibited the EGF-induced phosphorylation of LKB1 at Ser431 with an IC50 of approx. 1 μM. Furthermore, BI-D1870 did not affect the activation of ERK1/ERK2 and MSK1, nor did it inhibit the phosphorylation of CREB (Figure 4).
Effect of BI-D1870 on EGF-stimulated Rat-2 cells
In contrast with PMA, EGF not only activates the isoforms of RSK, MSK, S6K , but also triggers the activation of PKB as judged by using phospho-specific antibodies that recognize PKB phosphorylated at Thr308 and Ser473 (Figure 4). Since PKB phos-phorylates GSK3α and GSK3β at the same residues as RSK in vivo (reviewed in ), it would be expected that, as shown previously in Swiss 3T3 cells , blockade of PKB activation as well as blockade of RSK activity would be needed to prevent the EGF-induced phosphorylation of GSK3 isoforms. Consistent with this notion, treatment of Rat-2 cells with BI-D1870 alone or with the PI3K inhibitor LY294002 alone did not affect the EGF-induced phosphorylation of GSK3 isoforms, but when added to Rat-2 cells together, the EGF-induced phosphorylation of GSK3α or GSK3β was blocked completely (Figure 4 and Figure 5). Similarly, addition of both PD 184352 and LY 294002, to inhibit the activation of RSK and PKB respectively, prevented EGF-induced phosphorylation of GSK3, whereas either inhibitor alone or a combination of PD 184352 with BI-D1870 did not (Figure 5). As expected, LY 294002 inhibited the activation of PKB and the phosphorylation of two of its substrates, FKHR and FKHRL1, which are not phosphorylated by RSK . BI-D1870 did not affect the activation of PKB or the phosphorylation of the FKHR isoforms in EGF-stimulated cells (Figure 5), indicating that this compound does not affect the PKB pathway.
Effect of BI-D1870, PD 184352 and LY294002 on signalling pathways in Rat-2 cells
BI-D1870 does not inhibit PKA activity in vivo
PKA also phosphorylates GSK3α and GSK3β at Ser21 and Ser9 , and LKB1 at Ser431 [23,31], in the presence of agonists that elevate cAMP. In order to verify that BI-D1870 does not inhibit the activity of PKA in vivo, we stimulated HEK-293 and Rat-2 cells with forskolin, an activator of adenylate cyclase, which elevates the level of cAMP and hence the activity of PKA in the presence or absence of 10 μM BI-D1870. We then measured the phosphorylation of GSK3α, GSK3β and LKB1 at the residues that are phosphorylated by PKA. BI-D1870 (10 μM) did not inhibit the forskolin-induced phosphorylation of the GSK3 isoforms or LKB1 significantly (Figure 6).
BI-D1870 does not inhibit PKA in vivo
BI-D1870 induces the activation of ERK1/ERK2 in Rat-2 cells, but not in HEK-293 cells
Interestingly, the incubation of Rat-2 cells with BI-D1870 in the absence of EGF induced a time- (Figure 7A) and dose-dependent (Figure 7B) phosphorylation of ERK1/ERK2. ERK1/ERK2 phosphorylation was induced 5 min after treatment of cells with 10 μM BI-D1870, with maximal activation observed between 20 and 80 min, and a slow decline thereafter (Figure 7A). Incubation of Rat-2 cells with 10 μM BI-D1870 for 30 min resulted in the phosphorylation of ERK1 and ERK2 to approx. 50% of the level that is observed following stimulation with EGF (Figure 7B). Consistent with a significant activation of ERK and hence MSK1, BI-D1870 induced the phosphorylation of CREB at Ser133. In contrast, 10 μM BI-D1870 did not stimulate a detectable increase in ERK phosphorylation in HEK-293 cells (Figure 2B). These results are considered further in the Discussion.
BI-D1870 activates ERKs in Rat-2 cells
The results in the present study indicate that BI-D1870 is a highly specific cell-permeant inhibitor of RSK, which acts as an ATP-competitive inhibitor of the N-terminal AGC kinase domain of RSK (Figure 1). BI-D1870 is cell-permeant because at 10 μM it inhibits the phosphorylation of three known RSK substrates (GSK3α, GSK3β and LKB1) in response to agonists that induce the activation of RSK. BI-D1870 is stable in tissue culture medium as it inhibited the phosphorylation of RSK substrates efficiently even after 4 h, without affecting the PMA-induced phosphoryl-ation of ERK1/ERK2 (Figure 2A). BI-D1870 inhibited all RSK isoforms, with IC50 values of 10–40 nM in the presence of 20–100 μM ATP (Figure 1 and Tables 1 and 2). However, our in vivo results indicate that to inhibit the phosphorylation of RSK sub-strates completely in cells, a concentration of 10 μM BI-D1870 is required (Figures 2B and 4A). The requirement for higher com-pound concentration to suppress the activity of a protein kinase in vivo compared with those required in vitro is frequently observed with many ATP-competitive inhibitors of protein kinases. One reason is the far higher mM intracellular concentration of ATP in cells compared with the in vitro assay. A second reason is that there may be a barrier to the penetration of cells by BI-D1870. A third possibility is that RSK is a relatively abundant and active protein kinase, and inhibiting its activity partially may be insuf-ficient to prevent the phosphorylation of substrates. A fourth possibility is that the concentrations of PMA and EGF employed in this study induced maximal activation of ERK1/ERK2 and RSK isoforms; that is, lower levels of BI-D1870 may inhibit RSK more effectively after stimulation with lower agonist concentrations.
A remarkable feature of BI-D1870 is its very high specificity for RSK isoforms. BI-D1870 inhibited RSK isoforms >500-fold more potently than the ten other related AGC kinases that were tested in vitro. We have also verified that BI-D1870 is a highly specific RSK inhibitor in vivo, because treatment of cells with 10 μM BI-D1870 did not significantly inhibit the phosphorylation of the ribosomal S6 protein by S6K, the phosphorylation of GSK3 isoforms or FKHR1/FKHRL1 by PKB, the phosphorylation of CREB by MSK1, the phosphorylation of GSK3 isoforms or LKB1 by PKA, the PDK1-mediated activation of S6K1 and PKBα, or the PKC-mediated PMA-induced activation of ERK1/ERK2. It will be important in future work to perform a detailed structural analysis of BI-D1870 bound to RSK to understand the basis for the high specificity of BI-D1870. Understanding how BI-D1870 inhibits RSK activity at the structural level may provide important insights into how specific inhibitors of other AGC kinases could be developed.
It has been reported recently, that an acetylated flavonol glyco-side isolated from the tropical plant Forsteronia refracta termed SL0101 inhibited RSK2 with an IC50 of 90 nM . The select-ivity of this compound is uncertain, as it has not yet been tested against a large panel of other kinases, but was reported not to inhibit MSK1, S6K1, PKA and the FAK (focal adhesion kinase) kinase significantly. Employing an elegant structure-based design strategy, Taunton et al.  have synthesized an inhibitor, fmk, which interacts covalently with and modifies covalently the C-terminal kinase domain of RSK isoforms. In cell extracts, fmk formed a specific covalent interaction with RSK isoforms and also inhibited the phosphorylation of histone H3 at Ser10 in cells that over-express RSK2. However, the specificity of fmk towards a panel of protein kinases was not assessed  and whether it is really a specific inhibitor of RSK has yet to be rigorously evaluated.
As discussed in the Introduction, it was originally shown that a dominant negative mutant of RSK inhibited the growth factor-induced phosphorylation of the transcription factor CREB at Ser133 and that the over-expression of RSK in cells induced the phos-phorylation of CREB at this residue . This was originally interpreted to indicate that RSK was catalysing the phosphoryl-ation and, hence, the activation of CREB. However, the finding that CREB was a vastly more efficient substrate for MSK1 than RSK in vitro  and that the compound H89, at concentrations that inhibit MSK isoforms but not RSK, prevents growth-factor-induced CREB phosphorylation [23,35], indicated that MSK1, rather than RSK, may mediate the phosphorylation of CREB in cells. The finding that the growth factor and phorbol ester induced phosphorylation of CREB was greatly reduced, whereas phosphorylation of GSK3 or LKB1  mediated by RSK was unaffected in MES (mouse embryonic stem) cells lacking MSK1  and mouse embryonic fibroblast cells lacking both MSK1 and MSK2 , strongly support the conclusion that MSK1/MSK2 mediate the phosphorylation of CREB. Interestingly, trace residual growth-factor-induced phosphorylation of CREB that occurs in the fibroblast cell lines lacking MSK1 and MSK2 is not prevented by 10 μM BI-D1870, indicating that this is not catalysed by RSK isoforms . The finding in the present study, that the RSK inhib-itor BI-D1870 does not inhibit the phosphorylation of CREB detectably in cells, but completely inhibits the phosphorylation of GSK3 isoforms and LKB1 (Figures 2 and 4), provides further evidence that RSK is not rate limiting for the phosphorylation of CREB in the cells, under the conditions that we have examined. Moreover, the observation that addition of BI-D1870 to Rat-2 cells (in the absence of any other agonist) stimulates ERK1/ERK2 phos-phorylation, resulting in the activation of MSK1 and CREB phosphorylation, but not the phosphorylation of the GSK3 iso-forms (Figure 7), also strongly supports the conclusion that MSK1/MSK2 (rather than RSK isoforms) mediates the phos-phorylation of CREB.
The observation that BI-D1870 alone significantly activated the ERK1/ERK2 pathway in Rat-2 cells (Figure 7) indicates that RSK activity controls a negative-feedback loop that regulates ERK1/ERK2 activation in certain cell types. This conclusion is also supported by the finding that, in MES cells lacking PDK1 and in which RSK is therefore inactive, ERK1/ERK2 phosphorylation in unstimulated cells is elevated significantly . It is therefore possible that in unstimulated Rat-2 cells, basal levels of RSK activity suppress the activity of a protein that stimulates the ERK1/ERK2 pathway. The inhibition of RSK that occurs when cells are incubated with BI-D1870 may switch off this negative-feedback loop. Douville and Downward  reported that in PC12 cells, RSK phosphorylates and inactivates mSOS, the guanine nucleotide exchange factor for Ras. More recently RSK has been shown to inhibit ERK/MAPK signalling during Drosophila devel-opment, although the precise mechanism by which RSK inhibited ERK in this system was not defined . The role of RSK in regulating the activity of ERK1/ERK2 is likely to be cell specific, as the addition of 10 μM BI-D1870 to unstimulated HEK-293 cells did not induce detectable activation of ERK1/ERK2, p38/MAPK (Figure 2B) or JNK (c-Jun N-terminal kinase; results not shown). RSK is thought to play an important role in stimulating cell growth and promoting cell survival [8,40] and is therefore a potential target for an anti-cancer drug. However, the observation that inhibition of RSK in some cells promotes the activation of ERK1/ERK2, which might therefore stimulate rather than inhibit cell growth, is potentially an adverse side effect. Furthermore, the discovery that mutation of the RSK2 gene in humans results in Coffin–Lowry syndrome, characterized by mental retardation, facial dysmorphisms and skeletal deformations , also indic-ates that drugs that inhibit RSK isoforms could have undesired effects.
RSK has been suggested to phosphorylate a large number of cellular substrates and to play an important role in promoting cell growth and survival (reviewed in [8,40]). However, the evidence for an involvement of RSK in regulating these processes in many cases is based largely on the over-expression of constitutively active and dominant negative RSK mutants in cells, which, as dis-cussed above, can lead to erroneous conclusions. The activation of RSK also requires its phosphorylation at one site by PDK1. Thus a dominant negative mutant of RSK could also interact with PDK1, preventing PDK1 from activating a number of other AGC kinase members. Small cell-permeant inhibitors of RSK, such as BI-D1870, should not suffer from these problems and have the potential to facilitate the investigation of the cellular roles of RSK in the same way that inhibitors of MKK-1 activation (PD 98059, U0126 and PD 184352) and the p38 MAPK inhibitor (SB 203580) have enabled many roles of particular MAPK pathways to be defined. In order to determine whether a substrate is phos-phorylated by RSK in cells, it will be necessary in the future to demonstrate that it is inhibited by BI-D1870, as well as by inhibitors of the activation of MKK-1. Furthermore, the phosphorylation of the substrate should not occur in cells that lack PDK1, which also lack RSK activity . Thus far, the only RSK substrates that have met all of these criteria are GSK3α, GSK3β and LKB1. It will be important to deploy BI-D1870 to re-evaluate the role that RSK plays in mediating phosphorylation of the myriad of other substrates [8,40] that have been suggested for this protein kinase.
We thank Andreas Schnapp (Boehringer Ingelheim Pharma GmbH & Co. KG) for insightful discussions, Dr Nicoletta Resta for providing us with cDNA encoding human LKB1, our colleagues at the University of Dundee for their help in the following analyses, Matthew Elliot for help with performing the kinase specificity screen of BI-D1870 shown in Table 1, Carla Baillie for help with the expression of RSK isoforms and the protein production and antibody purification teams (Division of Signal Transduction Therapy) co-ordinated by Hilary McLauchlan and James Hastie for generation and purification of kinases and antibodies, the Dundee DNA sequencing service, and Agnieszka Kieloch for assistance with tissue culture. G.P.S. was supported by a Diabetes U.K. Studentship and P.C. by a Royal Society Research Professorship and an EU Framework 6 programme grant (Protein kinases-novel drug targets of the post genomic era). We also acknowledge the support of the AICR (Association for International Cancer Research), Diabetes U.K. and the U.K. MRC (Medical Research Council).
cAMP-dependent protein kinase/protein kinase G/protein kinase C
cyclin-dependent kinase 2 complexed with cyclin A
dual-specificity tyrosine-phosphorylated and -regulated kinase 1a
epidermal growth factor
extracellular-signal-regulated kinase 1
foetal bovine serum
forkhead in rhabdosarcoma transcription factor
glycogen synthase kinase 3α
human embryonic kidney 293
c-Jun N-terminal kinase
mitogen-activated protein kinase
mouse embryonic stem
mitogen- and stress-activated protein kinase
phosphoinositide-dependent kinase 1
polo-like kinase 1
cAMP-dependent protein kinase A
protein kinase B
protein kinase C
Rho-dependent protein kinase
p90 ribosomal S6 kinase
p70 ribosomal S6 kinase
serum- and glucocorticoid-induced kinase-1