RSK2 activity is regulated by its interaction with PEA-15

The ERK MAP (mitogen-activated protein) kinase cascade modulates many cellular processes including transcription, adhesion, growth, survival, and proliferation. One target substrate of ERK involved in regulating transcription is the p90 ribosomal S6 kinase (RSK) isozyme, RSK2. Here we demonstrate that a small death effector domain-containing protein called PEA-15 binds RSK2. RSK2 and PEA-15 (phosphoprotein enriched in astrocytes, 15 kDa) co-precipitated from cells and were colocalized in the cytoplasm. Furthermore, purified PEA-15 bound in vitro translated RSK2, suggesting that these proteins interact directly. PEA-15 does not bind to RSK1 and therefore exhibits some binding specificity. RSK2 binds the COOH terminus of PEA-15 and does not interact with its NH2-terminal death effector domain. We show that this interaction has functional consequences including the inhibition of RSK2-dependent CREB transcription. PEA-15 expression also blocks histone H3 phosphorylation, an RSK2-dependent event that may contribute to effects on gene expression. These results can be attributed to two effects of PEA-15 on RSK2. First, PEA-15 blocks nuclear accumulation of RSK2 after epidermal growth factor stimulation. Second, PEA-15 inhibits RSK2 kinase activity by 50%. A mutant of PEA-15 that binds RSK2 but is localized to the nucleus had no effect on RSK2-dependent transcription. Interestingly, this mutant also did not affect RSK2 kinase activity. This may indicate that cytoplasmic retention of RSK2 is also required for PEA-15 to impair kinase activity. PEA-15 does not alter ERK phosphorylation of RSK2 and is not itself a substrate of RSK2. Hence the effects of PEA-15 on RSK2 represent a novel mechanism for the regulation of RSK2-mediated signaling.


INTRODUCTION
X-linked disease marked by cognitive disabilities and skeletal malformations (14). In cells derived from CLS patients, EGF induced CREB phosphorylation and c-Fos transcription did not occur (15).
Hence RSK2 appears to be specifically required for these functions. Finally, RSK2 is also required for EGF induced phosphorylation of histone H3 (16) implicating RSK2 in the regulation of chromatin remodeling. The regulation of RSK activity presents a complex problem that is just beginning to be solved.
We have previously reported that the small death effector domain protein PEA-15 regulates the ERK MAP kinase pathway. PEA-15 is a 15 KDa protein that was originally identified as a major astrocytic phosphoprotein (17;18). The first 80 amino acids of PEA-15 correspond to the canonical Death Effector Domain (DED) sequence found in proteins that regulate apoptotic signaling pathways (19;20). The remaining 51 amino acids contain a serine (S104) that is phosphorylated by PKC (17) and a serine (S116) phosphorylated by calcium calmodulin kinase II (21). PEA-15 blocks an H-Ras to integrin signal (22). More recently we have shown that PEA-15 blocks ERK-dependent transcription and proliferation by binding ERKs and preventing their accumulation in the nucleus (23). PEA-15 contains a nuclear export sequence required for its capacity to maintain ERK in the cytoplasm. Genetic deletion of PEA-15 results in increased ERK nuclear localization with consequent increased c-Fos transcription and cell proliferation. Thus, PEA-15 can redirect the biological outcome of MAP kinase signaling by regulating the subcellular localization of ERK MAP kinase (23).
We now show that PEA-15 binds RSK2 and alters its activity. PEA-15 does not bind to RSK1 and therefore appears to exhibit some binding specificity. PEA-15 expression impairs RSK2 kinase activity and prevents RSK2 accumulation in the nucleus. Moreover, PEA- 15  and the NES (I15A) have been described before (24). PEA-15 was subcloned from the pcDNA3 vector into the pcDNA3-HisC vector (Invitrogen) in frame with the His tag using BamHI and EcoRI restriction sites. Protein expression was verified using both a PEA-15 specific antibody (from Dr Hervé Chneiweiss) and a His-specific antibody (Novagen). RSK2 was cloned into pcDNA3 vector using forward primer 5'-GGATCCATGCCGCTGGCGCAG-CTGGCGGAC-3' that contains a BamHI restriction site and a reverse primer 5'-CTCGA-GTCACAGGGCTGTTGAGGTGATTTT-3' that contains a XhoI restriction site. RSK2, ERK1 and ERK2 antibodies were from Santa Cruz Biotechnologies, CA. A rabbit-RSK2 used for immunofluorescence and phospho-Histone H3 antibody were from Upstate Biotech. Phospho-RSK2 antibody was from Cell Signaling Technologies and the HA antibody 12CA5 was from Roche. EGF was obtained from Sigma. S 35 -Methionine labeling and GST pulldowns: CHO-K1 cells were cultured to 90% confluency in a 100 mm tissue culture dish. Cells were then cultured for at least 4 hours in the presence of S 35labeled Methionine. Cells were lysed in MLB buffer (25mM HEPES, 150mM NaCl, 1%NP-40, 0.25% sodium deoxycholate, 10% glycerol, 10mM MgCl 2 , 1mM EDTA and protease inhibitors) and then subjected to a pulldown assay using equal amounts of purified GST or GST-PEA-15 immobilized on agarose beads. The cell lysates were incubated with the purified proteins for at least 1 hour at 4 0 C. The beads were then washed extensively with MLB buffer and then once with buffer containing 50mM Tris-Cl (pH 7.5) and 150mM NaCl. The proteins were then cleaved with 1 unit of thrombin in a buffer containing 50mM Tris-Cl (pH 7.5), 150mM NaCl and 2.5mM of CaCl 2 . The beads were discarded and the supernatant was subjected to SDS-PAGE. The gels were dried and autoradiographed.
Immunoprecipitations and kinase assays: Cos-7 cells were transfected with pMT2 HA-RSK2 in 100mm dishes. The cells were serum starved for 24 hours prior to stimulation with 10ng/mL EGF for 10 minutes. Cells were washed in PBS and lysates were obtained by scraping the cells in cold lysis buffer containing 20mM Tris (pH 8), 137mM NaCl, 10% glycerol and 1% NP-40. Cell debris was discarded after centrifuging at 14,000 rpm for 20 minutes at 4 0 C. The supernatant was used to immunoprecipitate HA-RSK2 using the monoclonal HA-antibody 12CA5. Lysates were incubated with the antibody for 2-3 hours on ice. Immune complexes were precipitated with 50µl of Protein Asepharose (Amersham Pharmacia) beads for 1 hour at 4 0 C. The precipitates were washed extensively in lysis buffer supplemented with 1M NaCl. For kinase assays, the precipitates were resuspended in 40 µl of buffer containing 20mM MOPS (pH 7.2), 25mM β-glycerophosphate, 5mM EGTA, 1mM DTT and split into two equal aliquots. One aliquot was used in an S6 peptide kinase assay (Upstate Biotech) while the other aliquot was subjected to immunoblotting to verify equal RSK2 precipitation.
The kinase assay was carried out according to the manufacturer's protocol using an S6 peptide substrate to determine radiolabeled phosphate incorporation by scintillation counting. Alternatively, the S6 peptide was substituted with 10µg of myelin basic protein, GST-PEA15 or GST and the reaction was arrested with sample buffer after 10 minutes. Proteins were separated by SDS-PAGE, dried and visualized by autoradiography.
His-Tagged protein precipitations: 6X His-tagged constructs of PEA-15 (wt and mutants) were transfected into HeLa cells using Polyfect (QIAGEN) according to the manufacturer's protocol. The cells were cultured for 48 hours prior to lysing with 1mL of NP-40 lysis buffer per 100mm dish.
Lysates were obtained after centrifugation at 14000 rpm for 20 minutes at 4 0 C to remove cell debris.
For each sample, 50µl of Probond Ni-column (Invitrogen) was used after washing the column once with distilled water and with lysis buffer. Lysates were incubated with the columns for 45-60mins at 4 0 C. The pellets with the bound protein were extensively washed with lysis buffer and finally resuspended in 100µl of sample buffer. Proteins were subjected to SDS-PAGE and immunoblotting for endogenous RSK2.
Immunofluorescence assays: Cos-7 cells were co-transfected with HA-RSK2 and GFP or PEA-15 fused to GFP. Cells were grown for 24 hours and then serum-starved for 24 hours. Cells were then treated with 50ng/mL EGF for 0', 20' or 50' before fixing with 4% paraformaldehyde for 10'. Cells were then stained overnight at 4 0 C with anti-RSK2 antibody (Upstate Biotech) in 3% BSA and 0.2% Triton X-100 in PBS. The antibody was detected using anti-rabbit Texas-Red conjugated antibody from Molecular Probes. The cells were imaged in a Zeiss Axiovert 100M fluorescent microscope using a 100X oil-immersion objective.
Transcription assays: Transcription assays for RSK2 activity were carried out using the Stratagene path detect kit specific for CREB-mediated transcription. The strategy involves the use of a Gal4DBD-tagged CREB construct (pFA2-CREB) and a luciferase construct carrying the Gal 4 promoter region (pFR-Luc). Using this approach only the effect of transfected CREB on luciferase expression is measured. 250ng of pcDNA3 or PEA-15 (wt and mutants) were co-transfected with 500ng of pFR-Luc and 50ng of pFA2-CREB constructs. Cells were left serum starved for [16][17][18][19][20] hours prior to stimulation with 20% serum for one hour. Luciferase readings were obtained using the Promega assay kit according to the manufacturer's protocol. To verify the specificity of the effect for RSK2, activated RSK2 (Y707A) was co-transfected along with the other plasmids and cells were left serum starved prior to lysis. Luciferase readings were obtained in a luminometer for 5µg of total cellular protein according to the manufacturer's protocol.
In vitro transcription/translation: pcDNA3-RSK2 plasmid was used to transcribe RNA with a linked protein transcription-translation system (Ambion). The proteins were labeled using S 35 -Methionine according to the manufacturer's protocol. 1/10 th of the reaction was retained to load on the gel as input. The remaining 40 ul of the reaction was diluted to 500µl in buffer containing 25mM HEPES, 150mM NaCl, 1%NP-40, 0.25% sodium deoxycholate, 10% glycerol, 10mM MgCl 2 , 1mM EDTA and protease inhibitors. Purified GST-PEA-15 or GST immobilized on agarose beads was used as bait to carry out a protein pulldown. The translation mixture containing full-length proteins was first pre-cleared with GST for half an hour at 4 0 C. Equal amounts of GSTPEA-15 or GST were used for a pulldown experiment with the pre-cleared reaction mixtures. The proteins were incubated for 2-3 hours at 4 0 C before washing the pellet extensively with MLB buffer. The samples were resuspended in sample buffer and subjected to SDS-PAGE, drying and autoradiography.

PEA-15 binds RSK2
We previously reported that the small death effector domain containing protein, PEA-15, binds ERK1/2 and redirects traffic in the ERK MAP kinase pathway (23). To identify other proteins that interact with PEA-15 that might contribute to the effects on ERK signaling, we isolated PEA-15 interacting proteins from cell lysates by pull down with GST-PEA-15 fusion proteins. We labeled CHO-K1 cells with S 35 methionine, prepared lysates from these cells and ran the lysates over a glutathione agarose column coated with GST-PEA15 fusion protein containing full length PEA-15 fused to glutathione-S-transferase. We noted that in addition to bands at the size of ERK 1 and 2 we also saw a band at 90 kDa ( Figure 1A). The p90RSK proteins are targets of ERK1/2 (8) and regulate transcription and cell survival events. We therefore tested whether the 90KDa band was a member of the RSK family of proteins. Initial results showed that antibodies to RSK2 but not RSK1 reacted with a band in the GST-PEA15 pulldown at 90KDa (data not shown). To determine if the PEA-15 could precipitate RSK2, we transfected HeLa cells with a His tagged PEA-15 or the vector alone and isolated the overexpressed PEA-15 and any binding partners on a Ni +2 column. We observed that endogenous RSK2 co-precipitated with PEA-15 ( Figure 1B). Protein expression and pulldown were verified by Western blotting as indicated.
To determine if the interaction is direct we used a linked in vitro transcription and translation system to produce S 35 Methionine-labeled RSK2 and RSK1 proteins. We then mixed the labeled RSKs with agarose beads coated with purified GST-PEA-15. As controls we included GST coated agarose beads. PEA-15 binds RSK2 but not RSK1 or GST ( Figure 1C). Hence PEA-15 and RSK2 can interact directly as purified proteins.
To confirm the interaction we used an additional in vivo assay to isolate the protein complex.
We co-transfected His tagged PEA-15 with HA-tagged RSK2 in Cos-7 cells and allowed expression for 48 hours. We then immunoprecipitated RSK2 using an anti-HA antibody (Cell Signaling Technology) and immunoblotted for PEA-15. We found a significant amount of PEA-15 coimmunoprecipitated with RSK2 ( Figure 1D). Additionally, we were able to observe endogenous PEA-15 that had also been precipitated in a complex with RSK2 (data not shown). Thus we have shown that PEA-15 and RSK2 interact with each other and that the interaction is direct.
To determine the regions of PEA-15 that may be necessary for this interaction we used point Since PEA-15 binds RSK2 we determined whether the two proteins co-localize in the cell.
We found that while PEA-15 appeared to be strongly co-localized with RSK2, control cells transfected with GFP did not show similar co-localization patterns (Figure 2). Neither PEA-15 nor RSK2 appeared to be confined to any particular compartment of the cytoplasm, but were instead diffusely distributed throughout it under these conditions. These co-localization experiments provide

PEA-15 affects RSK2 dependent transcription
Having found that PEA-15 binds to RSK2, we sought to determine if PEA-15 expression affects RSK2 function. One known function of RSK2 is the activation of the transcription factor CREB (15;25). To examine the functional consequences of PEA-15 binding to RSK2 we carried out in vivo transcription assays using a CREB driven luciferase construct in the Pathdetect system. Cells were treated with serum or not and then lysed and assayed for luciferase activity as a measure of We then sought to exclude the possibility that PEA-15 blocks CREB activation by an RSK2 independent mechanism. We therefore examined if PEA-15 expression could alter activation of CREB by an activated mutant of RSK2 (Y707A, Figure 4). PEA-15 was co-transfected with the activated RSK2 construct. Cells transfected with wt PEA-15 strongly inhibited RSK2 dependent CREB mediated transcription (Figure 4). Furthermore a non-ERK binding tail mutant (L123R) and the NES mutant, I15A, did not inhibit transcription (Figure 4). Activated RSK2 expression was verified by blotting for RSK2 in the lysates ( Figure 4). Hence, PEA-15 is able to impair RSK2 activation of CREB.

Overexpression of PEA-15 decreases phosphorylation of histone H3
RSK2 has been reported to be required for the phosphorylation of Histone H3 (16). To further determine the functional consequences of PEA-15 expression on RSK2 kinase, we examined the effect of PEA-15 on the RSK2 specific nuclear substrate histone H3. We co-transfected Cos-7 cells with RSK2 and PEA-15 or vector control. The MEK inhibitor was used to remove the high background phosphorylation of histone H3 at Ser10 that was observable even upon serum starvation (data not shown and previous reports). Cells were then stimulated with EGF. We observed that PEA-15 expression blocked H3 phosphorylation compared to the vector control ( Figure 5) in response to EGF stimulation.

PEA-15 affects RSK2 kinase activity
One mechanism by which PEA-15 could block RSK2 activation of CREB and histone H3 phosphorylation is by blocking the kinase activity of RSK2. To address this we looked at PEA-15 effects on RSK2 kinase activity using an S6 peptide based kinase assay. In this assay PEA-15 or control vector alone were co-transfected with HA-tagged RSK2 into cells followed by immunoprecipitation of the RSK2 and determination of its ability to phosphorylate S6 peptide. We found that PEA-15 inhibits RSK2 kinase activity by 50% compared to a vector control ( Figure 6A). Surprisingly, the nuclear localized PEA-15 mutant (I15A) did not appear to inhibit RSK2 kinase activity ( Figure 6A To determine if PEA-15 expression alters phosphorylation at this site, cells stimulated with EGF in the presence or absence of overexpressed PEA-15 were immunoblotted with a Ser386 phospho epitope antibody. PEA-15 did not affect phosphorylation at Ser386 ( Figure 6C). Therefore PEA-15 does not affect RSK2 activity by altering its activation state.

PEA-15 alters RSK2 localization
A mutant of PEA-15 that accumulates in the nucleus (I15A) did not impair the kinase activity of RSK2 ( Figure 6A). In addition, PEA-15 has been shown to bind ERK and sequester it in the cytoplasm (23). This suggests that PEA-15 might regulate RSK2 similarly by altering its subcellular localization. To test this hypothesis, we investigated whether PEA-15 regulates the translocation of  (26). However, we were unable to detect this in our system for RSK2 even at very early time points post EGF stimulation (data not shown). Our results therefore suggest that PEA-15 changes the subcellular localization of RSK2. This may contribute to the effect of PEA-15 on CREB mediated transcription and appears to be required for PEA-15 inhibition of RSK2 kinase activity.

DISCUSSION
We have shown that PEA-15 binds directly to RSK2 and alters both its subcellular localization and its kinase activity. In particular PEA-15 expression prevents RSK2 translocation into the nucleus and results in decreased RSK2 kinase activity. This has the functional outcome of inhibiting RSK2 activation of the transcription factor CREB and RSK2 phosphorylation of Histone H3. Both PEA-15 and RSK2 are reported to be expressed in many of the same cell types including astrocytes (17;27). Hence we have identified a novel RSK2 binding partner that may play a significant role in regulating its signaling activity.
PEA-15 is reported to affect ERK MAP kinase signaling and apoptosis (23;28-30). Perhaps some of these effects are due to PEA-15's interaction with RSK2. For example, c-fos activation in PEA-15 null mice was shown to be twice that of wild type mice. RSK2 has been shown to be required for activation of c-Fos (31). Hence, PEA-15 binding to RSK2 may normally play a role in limiting the ability of RSK2 to access nuclear targets like c-fos, CREB, and histone H3. Cells lacking PEA-15 may then have abnormally high levels of these activities of RSK2 leading to effects on the cell cycle. Interestingly PEA-15 null cells are also reported to sustain an increased rate of proliferation (23).
We have also shown that PEA-15 binds ERK1/2 (23;24). Here we show that RSK2 binds PEA-15 in a distinct manner from that of ERK. Specifically, RSK2 binds to a mutant PEA-15 that cannot bind ERK (D74A). This suggests that the two binding sites are different and that PEA-15 may be able to bind to both ERK and RSK2 at the same time, perhaps preventing or promoting their interaction. However the fact that neither RSK2 nor ERK could bind the PEA-15 tail mutant (L123R) indicates that the binding sites on PEA-15, while distinct, may overlap. The precise nature of these interactions remains to be determined.
RSK2 phosphorylates histone H3 in response to EGF stimulation (16). The phosphorylation of histone H3 by RSK2 was hypothesized by Sassone-Corsi and colleagues to be important in MAP kinase regulation of transcription. Here we show that expression of PEA-15 inhibits RSK2 phosphorylation of histone H3. This is likely due to perturbation of RSK2 translocation to the nucleus. Rapid phosphorylation of histone H3 correlates with EGF-induced expression of immediate-early response genes such as c-fos (32). We previously reported that cells lacking PEA-PEA-15 negatively regulates RSK2 activity.
It is interesting that PEA-15 alters both the subcellular localization of RSK2 and its activity.
We have found that a mutant of PEA-15 that binds RSK2 but is localized to the nucleus does not block RSK2 mediated activation of transcription (I15A, Figure 4). Moreover, the same mutant does not block RSK2 kinase activity ( Figure 6A). We propose that this is evidence that effects on transcription and RSK2 activity in the cell are predominantly due to the altered localization of RSK2.
However it remains possible that binding of the I15A mutant of PEA-15 is physically different than that of wild type and therefore is incapable of blocking RSK2 kinase activity. Alternatively, retention of RSK2 in the cytoplasm may decrease RSK2 activity by an as yet undetermined mechanism.
PEA-15 is one of the first RSK binding proteins suggested to negatively regulate RSK function. It is the first that binds specifically to RSK2. Recently the adapter protein 14-3-3β was reported to bind specifically to RSK1 and negatively regulate it by interfering with RSK1 phosphorylation (33). Together these data suggest that there may be several RSK binding partners that can maintain signal specificity and modulate the growth factor response by different mechanisms. As such, this work provides important new insight for unraveling the complex regulation of RSKs.