Cloning and Characterization of Xenopus Rsk2, the Predominant p90 Rsk Isozyme in Oocytes and Eggs*

The 90-kDa ribosomal S6 kinases, the p90 Rsks, are a family of intracellular serine/threonine protein kinases distinguished by two distinct kinase domains. Rsks are activated downstream of the ERK1 (p44) and ERK2 (p42) mitogen-activated protein (MAP) kinases in diverse biological contexts, including progression through meiotic and mitotic M phases in Xenopus oocytes and cycling Xenopus egg extracts, and are critical for the M phase functions of Xenopus p42 MAPK. Here we report the cloning and biochemical characterization of Xenopus Rsk2.Xenopus Rsk1 and Rsk2 are specifically recognized by commercially available RSK1 and RSK2 antisera on immunoblots, but both Rsk1 and Rsk2 are immunoprecipitated by RSK1, RSK2, and RSK3 sera. Rsk2 is about 20-fold more abundant than the previously describedXenopus Rsk1 protein; their concentrations are approximately 120 and 5 nm, respectively. Rsk2, like Rsk1, forms a heteromeric complex with p42 MAP kinase. This interaction depends on sequences at the extreme C terminus of Rsk2 and can be disrupted by a synthetic peptide derived from the C-terminal 20 amino acids of Rsk2. Finally, we demonstrate that p42 MAP kinase can activate recombinant Rsk2 in vitro to a specific activity comparable to that found in Rsk2 that has been activated maximally in vivo. These findings underscore the importance of the Rsk2 isozyme in the M phase functions of p42 MAP kinase and provide tools for further examining Rsk2 function.

phosphorylate MEK1 and MEK2 on two serine residues in MEK's T-loop, thereby activating the MEK. Active MEK in turn can phosphorylate p42 or p44 at threonine and tyrosine residues within the sequence Thr-Glu-Tyr in MAPK's T-loop, thereby activating the MAPK.
p42 and p44 MAPK are probably best-known for their roles in mitogenesis and cell fate induction (4). However, these proteins have also been implicated in the regulation of meiotic and mitotic M phase. Much of the work on the M phase roles of MAPKs has been carried out in Xenopus laevis oocytes, eggs, and cycling egg extracts, where the relevant MAPK is p42 (ERK2) MAPK (5)(6)(7)(8).
The first indication that p42 MAPK might play a role in M phase regulation came from studies of Xenopus oocyte maturation. Maturation is a key step in the production of a fertilizable egg, and its study has a long history of providing important insights into the biochemistry of M phase regulation. Fully grown Xenopus oocytes are arrested in a G 2 -like state with an intact germinal vesicle (nucleus) and inactive Cdc2/cyclin B and p42 MAPK. Progesterone releases oocytes from this arrest state and brings about a resumption of meiosis I. Progesteronetreated oocytes undergo germinal vesicle breakdown, complete the first meiotic division, progress through interkinesis, enter meiosis II, and then spontaneously arrest in metaphase of meiosis II. Germinal vesicle breakdown is immediately preceded by the activation of the Mos/MEK1/p42 MAPK cascade and Cdc2/cyclin B. Interfering with the activation of any member of the MAPK cascade can delay or inhibit Cdc2 activation and oocyte maturation (9 -13). The MAPK requirement appears not to be absolute; many batches of oocytes ultimately do activate their Cdc2 and mature even when MAPK activation has been suppressed (14 -16). Nevertheless, it does appear that MAPK activation can promote Cdc2 activation (17)(18)(19)(20).
The p90 Rsk family of protein kinases (21,22) may be an important link between the activation of MAPK and the activation of Cdc2 (23). Rsks were originally identified as protein kinase activities that could phosphorylate the S6 protein of the 40 S subunit of the ribosome in vitro (24 -27). Rsks are likely to be responsible for phosphorylating S6 in Xenopus oocytes (27); however, in many other cell types, the p70 S6 kinase appears to be the more important regulator of S6 (28 -30). Like p42 MAPK, Rsks are activated in tissue culture cells in response to diverse mitogens and in oocytes in response to progesterone. Rsk activation depends on a series of phosphorylations carried out by MAPK, PDK1, and Rsk itself (22,(31)(32)(33). Rsks have been implicated in transcriptional regulation (34 -39), in cell survival (40 -42), and, as described below, in the M phase functions of p42 MAPK.
In immature oocytes, cyclin B is present and complexed with Cdc2; nevertheless, the Cdc2 is inactive as a result of inhibitory phosphorylation. The main Cdc2-inhibitory kinase present in oocytes is Myt1 (43,44). Xenopus Rsk(s) have been found to associate with the C terminus of Myt1 and can phosphorylate and inactivate Myt1 in vitro (23). Rsks in turn can be phosphorylated and activated by p42 MAPK (45). Thus, p42 MAPK can release Cdc2 from the negative effects of Myt1, through the intermediacy of Rsk.
The Mos/MEK/p42 MAPK cascade remains active throughout the remainder of maturation and in mature, unfertilized eggs. It plays a critical role in suppressing DNA replication and promoting Cdc2 activation between meiosis I and meiosis II (46,47). p90 Rsk may mediate this effect of p42 MAPK; expression of a constitutively active form of p90 Rsk restores many aspects of a normal meiosis I/II transition to oocytes treated with a pharmacological inhibitor of the MAPK cascade (15).
The MAPK cascade is also required for the establishment of the cytostatic factor arrest in unfertilized eggs. The cascade suppresses cyclin destruction and keeps unfertilized eggs arrested in metaphase of meiosis II until fertilization occurs (48 -50). Again, p90 Rsk appears to be a critical mediator of p42 MAPK. Depleting Rsk from a cycling Xenopus egg extract prevents the extract from undergoing mitotic arrest in response to p42 MAPK activation (51), and expression of constitutively active p90 Rsk causes embryos to arrest in M phase (52).
Given the importance of Rsk in the function of the MAPK cascade in Xenopus oocytes and eggs, we set out to determine what forms of Rsk were present in these systems. Four closely related Rsk cDNAs, RSK1, RSK2, RSK3, and RSK4, plus the more distantly related RSK-B and RLPK/MSK1 cDNAs, have been cloned from humans (53)(54)(55)(56). The only Xenopus Rsk cDNAs reported to date have been Rsk1 homologs (57). However, there is clear biochemical evidence for at least one more Rsk-like kinase in Xenopus oocytes; Xenopus Rsk-like kinase activities fractionate into two chromatographically discrete peaks, designated S6 kinase I and S6 kinase II, which copurify with related but distinct 90-to 92-kDa proteins (24,58,59). Therefore, we have carried out a degenerate PCR screen for Xenopus Rsk-like cDNAs. Here we present the isolation and characterization of a new Xenopus Rsk cDNA, the Xenopus homolog of Rsk2, which represents the major Rsk isozyme in Xenopus oocytes and eggs.

EXPERIMENTAL PROCEDURES
RT-PCR and Subcloning-RNA was typically obtained from 10 to 20 frozen Stage VI oocytes using RNAeasy kits (Qiagen) according to the manufacturer's instructions. RT-PCR was accomplished using the Superscript 2.0 kit according to the manufacturer's instructions (Life Technologies, Inc.). 5Ј and 3Ј RACE reactions were performed as recommended by the manufacturer's instructions (Life Technologies, Inc.). All PCR-amplified fragments were gel-purified and subcloned into pGEM-T according to the manufacturer's instructions (Promega).
Hemidegenerate RT-PCR was used to extend the open reading frame of Xenopus Rsk2. Hemidegenerate RT-PCR amplifies cDNA fragments by using a degenerate oligonucleotide derived from a family of sequences and specific oligonucleotide derived from a novel family member. In our case, a Rsk-derived degenerate oligonucleotide ensured that the amplified cDNA was a Rsk family member and a specific Rsk oligonucleotide ensured that the amplified cDNA was derived from our novel Rsk cDNA sequence.
Sequencing-DNA sequencing was performed either by cycle sequencing with Thermosequenase or by Sequenase 2.0 dideoxy chain termination (Amersham Pharmacia Biotech).
Determining Rsk1 and Rsk2 Abundances.-Concentrations of purified recombinant ⌬43-His 6 -Rsk1 and His 6 -Rsk2 were determined by Bradford assay and verified by comparing the Coomassie Blue staining intensities of the Rsk bands on SDS-PAGE to a series of albumin standards. The purified recombinant Rsk proteins were then used as standards on Rsk1 and Rsk2 immunoblots of oocyte lysates. Rsk amounts were quantified by phosphorimaging (Rsk1) or densitometry (Rsk2).
Heteromeric Complex Formation in Sf9 Lysates-Previously frozen Sf9 cell pellets expressing His 6 -Rsk2 or GST-MAPK were lysed by thawing at 4°C in 50 mM Tris-HCl (pH 8.5), 10 mM 2-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, 1% Nonidet P-40. Lysates were incubated at 4°C for 10 min and clarified by centrifugation for 5 min at 4°C in a microcentrifuge. Crude lysates were then mixed and incubated at 4°C for 15 min. Either protein G-agarose beads and anti-His 6 antibodies (CLONTECH, Palo Alto CA) or glutathione-coated agarose beads were next added to the lysates and allowed to incubate for an additional 75 min at 4°C. The beads were collected by brief centrifugation and washed two times with 20 volumes of XB (100 mM KCl, 10 mM HEPES, pH 7.7, 0.1 mM CaCl 2 , 1 mM MgCl 2 , and 50 mM sucrose) and resuspended in SDS-PAGE sample buffer.
Complex Formation in Oocyte and Egg Extracts with Purified His 6 -Rsk2 Proteins-Oocytes were isolated as described previously and lysed by trituration in lysis buffer (50 mM 2-glycerophosphate, 10 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and 1 mM Na 3 VO 4 ). Dilute cytosolic fractions were separated from lipids and cellular debris by a 2-min centrifugation at 4°C in a Beckman model E Microfuge with a right angle rotor. Egg extracts were prepared essentially as described previously (61). Typically, portions of extracts equivalent to either two oocytes or eggs diluted in lysis buffer were combined with 0.5 g of His 6 -Rsk2 protein or XB and incubated at 4°C for 20 min. Protein G-agarose beads were subsequently added in the absence or presence of anti-His 6 antibody and incubated for an additional 90 min at 4°C. The beads were then washed twice with XB and then resuspended in SDS-PAGE sample buffer. Peptide inhibition was performed by incubating diluted extracts and peptide for 10 min at room temperature prior to the addition of His 6 -Rsk2 protein.
Rsk2 Activation and S6 Kinase Assay-Rsk2 activity was determined by using S6 peptide as phosphoacceptor (Santa Cruz Biotechnology). Briefly, recombinant proteins (1.5 g of His 6 -Rsk2, 0.3 g of His 6 -MEK R4F, 0.2 g of His 6 -MAPK) were mixed and incubated in kinase buffer (20 mM HEPES, pH 7.6, 20 mM MgCl 2 , 0.1 mg/ml bovine serum albumin, and 1 mM ATP), final volume 25 l, for 60 min at 30°C. Next, 1% of the protein from above was combined with 10 g of S6 peptide and kinase buffer supplemented with [␥-32 P]ATP (67 Ci/ml; final volume 30 l). Samples were incubated for 10 min at 30°C, and the reactions were rapidly terminated on ice by the addition of 10 l of 0.5 M EDTA. Two equal portions of the S6 kinase assay were spotted onto P81 cellulose paper (Whatman) and then dried. The dry P81 cellulose paper was washed two times with 1% acetic acid, three times with water, and then allowed to dry. Radioactive incorporation was determined by Cerenkov counting of the individual samples.

RESULTS
Cloning Xenopus Rsk2-We obtained RNA from stage VI Xenopus oocytes and performed RT-PCR with two sets of degenerate oligonucleotides derived from regions in the C-terminal kinase region of Rsk that are well conserved in Xenopus Rsk1 and human RSKs 1-4. The RT-PCR reactions yielded three distinct groups of DNA sequences. The first two groups of sequences were derived from Xenopus Rsk1␣ and Rsk1␤, two closely related Rsk1-like cDNAs that probably reflect the tetraploid ancestry of X. laevis (57). The third group of sequences were derived from a cDNA that was distinct from Rsk1␣ and Rsk1␤ but still clearly Rsk-related. We therefore completed the cDNA sequence of this new Rsk isozyme.
Through hemidegenerate RT-PCR we extended the coding sequence upstream and downstream, obtaining nearly 2 kb of what proved to be a 2.2-kb open reading frame. We completed the cloning of a cDNA of ϳ3 kb, similar in size to other Rsk cDNAs, through 5Ј and 3Ј RACE. To ensure that our final cDNA was derived from a single message, we re-isolated the entire cDNA by RT-PCR twice, using two pairs of specific oligonucleotides that hybridized to different sequences in the 5Ј-and 3Ј-untranslated regions. We sequenced these two independent clones and found that they both contained the sequences of the hemidegenerate RT-PCR and RACE fragments. We arrived at a consensus sequence by comparing the two long PCR products with each other and with the RT-PCR, hemidegenerate PCR, and RACE products.
The new Rsk-like sequence was identified as Xenopus Rsk2. The predicted amino acid sequence of Xenopus Rsk2 is 92% identical to human Rsk2, overall, and only 77-81% identical to human RSKs 1, 3, and 4, and Xenopus Rsk1␣ (Fig. 1, A and B). Xenopus Rsk2 has a predicted molecular mass of 83.3 kDa and a predicted isoelectric point of 6.3. Xenopus Rsk2 contains two protein kinase domains (Fig. 1, A and B, boxed and highlighted in gray), as do all of the Rsk proteins characterized to date. Each of the six phosphorylation sites identified in rat Rsk1 (31) is conserved in Xenopus Rsk2 and the other Rsk family members (Fig. 1A, asterisks), consistent with the observation that these kinases are regulated similarly. Xenopus Rsk2 lacks the N-terminal nuclear localization sequence that is characteristic of RSK3 (54). Xenopus Rsk2 is more distantly related to the recently cloned RSK-B and RLPK/MSK1 proteins (40% identical overall) (55,56).
We constructed phylogram trees for the Rsk N-terminal and C-terminal kinase domains (Fig. 2). Rsks 1-4 appear to have diverged from each other subsequent to the divergence of nematodes and insects from vertebrates.
Relationship of Xenopus Rsk2 to S6 Kinase II-Erikson and Maller (24,58,59) identified, purified, and characterized two S6 kinase activities from Xenopus egg extracts. They designated these activities S6 kinase I and S6 kinase II and obtained the sequences of seven tryptic peptides from S6 kinase II. As shown in Table I kinase II most likely represents Rsk2 rather than Rsk1. The Erikson and Maller S6 kinase II also apparently contains a second protein that gives rise to the 99-6 peptide, ADPSGFE-FLK, whose closest match in Rsk2 is ADPSQFELLK. This second protein might be a closely related Rsk2␤ protein or a more distantly related Rsk family member.
Phosphorylation of Rsk1 and Rsk2 during Oocyte Maturation-The extraordinarily close similarity between human and Xenopus Rsk2, and, to a lesser extent between human and Xenopus Rsk1, suggested that antibodies specific for human RSK1 and RSK2 might also be specific for Xenopus Rsk1 and Rsk2. We therefore subjected lysates from immature and mature oocytes to immunoblotting with commercially available RSK1 and RSK2 antisera. As shown in Fig. 3, the RSK1 serum recognized a closely spaced doublet with an apparent molecular mass of 74 -78 kDa in immature (G 2 phase) oocytes, which shifted to 94 -96 kDa in M phase. The RSK2 serum recognized a band at about 92 kDa in G 2 phase, which shifted to 94 -102 kDa in M phase. Because the RSK1-and RSK2-reactive bands are electrophoretically distinct, particularly in the G 2 phase samples, it follows that Xenopus oocytes express both Rsk1 and Rsk2 proteins, and that both proteins become hyperphosphorylated during oocyte maturation.
Abundance of Rsk1 and Rsk2-We carried out quantitative immunoblot analysis to determine the abundance of the oocyte Rsk1 and Rsk2 proteins, using purified recombinant Xenopus Rsk1␣ and Rsk2 as standards. As shown in Fig. 4 (A and B), Rsk1 was present at a concentration of about 0.8 ng per four oocytes, or 0.2 ng per oocyte. Taking the cytoplasmic volume of an oocyte to be 0.5 l and the molecular mass of Rsk1 to be 83 kDa, the concentration of Rsk1 in a Xenopus oocyte is approximately 5 nM. In a similar manner, we determined that an oocyte contains about 5 ng of Rsk2, corresponding to a concentration of approximately 120 nM (Fig. 4, C and D). Thus Rsk2 is the predominant Rsk isozyme in Xenopus oocytes.
Rsk Activity in Oocytes-Next we examined how much Rsk activity was present in oocytes and whether that activity could be accounted for by Rsk1 and Rsk2, given their estimated abundances. We subjected lysates from G 2 or M phase oocytes to immunoprecipitation using RSK1 and RSK2 antisera. We then subjected the immunoprecipitates to immunoblotting, to see how much Rsk was brought down, and to immune complex kinase assay. As shown in Fig. 5A, we found that, although the Rsk sera were fairly specific in immunoblots (see also Fig. 3), they were not specific for immunoprecipitation; both the RSK1 and RSK2 sera quantitatively brought down Rsk1 and Rsk2 proteins (Fig. 5A). Although we had no evidence for a Rsk3 isozyme in oocytes, we also tried RSK3 serum, and found that it brought down both Rsk1 and Rsk2 (Fig. 5A). The RSK1, RSK2, and RSK3 sera brought down similar amounts of Rsk1 and Rsk2 protein and Rsk activity (Fig. 5B), with the activities much higher in M phase lysates than G 2 phase lysates (Fig.  5B).
We used the results from each of the three RSK sera to calculate the total Rsk activity present in oocytes. We found an average of 11 pmol⅐min Ϫ1 of anti-Rsk immunoprecipitable S6 kinase activity per M phase oocyte or unfertilized egg ( Fig. 5B and data not shown). Given the estimated total amount of Rsk1 and Rsk2 protein (5.2 ng per oocyte) we calculate Rsk's specific S6 kinase activity to be 2.2 mol⅐min Ϫ1 ⅐mg Ϫ1 or, in molar terms, 179 mol of substrate phosphorylated per mol of Rsk per min. This specific activity is similar to that reported for purified rabbit ISPK (Rsk2) (2.0 mol⅐min Ϫ1 ⅐mg Ϫ1 ) and for Xenopus S6 kinase II (1.4 mol⅐min Ϫ1 ⅐mg Ϫ1 ), measured using Leu-Arg-Arg-Ala-Ser-Leu-Gly (Kemptide) as substrate (58,64,65). Thus the total Rsk activity present in Rsk immunoprecipitates can be accounted for by the estimated abundances of Rsk1 and Rsk2.
Rsk2 Forms a Heteromeric Complex with MAPK-We and others have previously reported that Rsk1 and p42 MAPK are present in complex in G 2 phase oocytes (62,66,67). This finding suggests that Rsk1 has a special role in MAPK signal transduction. We set out to determine whether Rsk2, like Rsk1, can form complexes with p42 MAPK. We expressed GST-tagged p42 MAPK and His 6 -Rsk2 separately in Sf9 cells, then mixed the lysates and carried out precipitations with glutathione beads or anti-His 6 antibodies. The anti-His 6 antibodies pulled down p42 MAPK when His 6 -Rsk2 was present, but not when His 6 -Rsk2 was absent (Fig. 6A). Conversely, the glutathione beads pulled down Rsk2 when GST-MAPK was present, but not when GST-MAPK was absent (Fig. 6B). Thus Rsk2 and p42 MAPK can form a stable complex, like Rsk1 and p42 MAPK.
Next we examined whether purified His 6 -Rsk2 protein was capable of forming a complex with endogenous oocyte MAPK. We added His 6 -Rsk2 to G 2 phase oocytes and immunoprecipitated the lysates with anti-His 6 -coated protein G beads. Endogenous MAPK was efficiently brought down with His 6 -Rsk2 but was not brought down by control beads or when no His 6 -Rsk2 was added (Fig. 6C).
The extreme C terminus of Rsk1 has been shown to be critical for its interaction with MAPK (66 -68). Consistent with these observations, we found that a synthetic 20-amino acid peptide derived from the C terminus of Rsk2 could inhibit the association of Rsk2 and p42 MAPK, with inhibition being halfmaximal at a peptide concentration of about 8 M (Fig. 6D). Note that both phosphorylated and non-phosphorylated p42 MAPK were pulled down with Rsk2 and that the C-terminal peptide was equally effective at inhibiting Rsk2 interaction with both forms of p42 MAPK (Fig. 6D). This finding is consistent with the observation that the phosphorylation of Rsk1, not the phosphorylation of p42 MAPK, allows Rsk1⅐p42 MAPK complexes to dissociate (66).
Activation of Rsk2 in Vitro-Next we investigated the requirements for the enzymatic activation of recombinant Rsk. Rsks are typically insoluble when expressed in E. coli, and efforts to extract and renature Rsk proteins from inclusion bodies result in low yields of enzymatically inactive protein. Therefore, we expressed and purified His 6 -Rsk2 protein from Sf9 cells.
Purified His 6 -Rsk2 could be resolved into two or three ϳ90-kDa bands on immunoblot analysis, suggesting that it had undergone some autophosphorylation or phosphorylation by insect cell kinases during its expression in Sf9 cells (Fig. 7A and data not shown). However, the recombinant His 6 -Rsk2 migrated faster than fully active Rsk2, and was low in activity (Fig. 7, A and B, and data not shown), indicating that it was not maximally phosphorylated. We incubated His 6 -Rsk2 with active MEK R4F and p42 MAPK in a linked kinase reaction to determine whether His 6 -Rsk2 could become phosphorylated and activated by these kinases in vitro. MEK R4F plus p42 MAPK caused His 6 -Rsk2 to become hyperphosphorylated (Fig.  7A, top panel) and activated (Fig. 7B). Neither MEK R4F nor p42 MAPK alone caused either hyperphosphorylation or activation ( Fig. 7A and data not shown). MEK R4F/p42 MAPKinduced Rsk2 phosphorylation was undiminished in His 6 -Rsk2 K97R, whose N-terminal kinase domain is non-functional (Fig.  7A, top panel). This observation is consistent with the hypothesis that Rsk phosphorylations are carried out by MAPK and the C-terminal Rsk kinase domain.
Next we examined the status of Ser-383 phosphorylation in His 6 -Rsk2. Ser-383 is situated in the sequence Phe-Ser-Phe within the linker that connects N-terminal and C-terminal kinase domains (Fig. 1). The Ser-383 site is notably conserved among the AGC protein kinases, which include protein kinase A (PKA), Akt/protein kinase B, protein kinase C, and p70 S6k (31,32). In some AGC kinases, the site corresponding to Ser-383 appears to be phosphorylated by PDK1 or PDK2; in others it appears to be autophosphorylated. We found that purified His 6 -Rsk2 had a low level of Ser-383 phosphorylation, as detected by a phospho-Ser-383 antibody, and underwent an increase in its Ser-383 phosphorylation in response to MEK R4F and p42 MAPK (Fig. 7A). This indicates that PDK1 and PDK2 are dispensable for Ser-383 phosphorylation and suggests that Ser-383 undergoes MAPK-stimulated autophosphorylation. MAPK-stimulated Ser-383 phosphorylation was undiminished in His 6 -Rsk2 K97R (Fig. 7A). Thus, Ser-383 phosphorylation appears to be carried out by the C-terminal kinase domain. The basal activity of His 6 -Rsk2 was approximately 0.2 mol⅐min Ϫ1 ⅐mg Ϫ1 (Fig. 7B). Incubation with activated MAPK caused the activity to increase to 1.8 mol⅐min Ϫ1 ⅐mg Ϫ1 , similar to the specific activity found for Rsks immunoprecipitated from M phase oocytes (2.2 mol⅐min Ϫ1 ⅐mg Ϫ1 ; see above). Despite the fact that Rsks and PKAs have similar substrate specificities, His 6 -Rsk2 kinase activity was insensitive to the PKA pseudosubstrate inhibitor PKI (Fig. 7B), in agreement with a previous report for purified S6 kinase II (69). DISCUSSION We have cloned the Xenopus homolog of human RSK2. Xenopus Rsk2 appears to be the main Rsk isozyme in Xenopus oocytes and eggs; its concentration is approximately 120 nM, compared with approximately 5 nM for Rsk1. The predicted sequence of Rsk2 indicates that this protein, rather than Rsk1, represents the principal component of S6 kinase II. Thus, in a happy coincidence of nomenclature, the numbers of the cDNA (Rsk2) and the biochemical activity (S6 kinase II) correspond. It will be of interest to determine whether Rsk1 represents S6 kinase I.
The abundances of Rsk2 and Rsk1 are sufficient to account for the total Rsk-like activity in mature oocytes. We cannot rule out some contribution from a Rsk3 or Rsk4 isozyme, although we did not detect Rsk3-or Rsk4-like sequences by RT-PCR.
Xenopus Rsk2 is very closely related to its human homolog RSK2. Much of the sequence difference between these two proteins lies near the N terminus, between amino acids 13 and 45. The Rsk1 cDNAs also differ from Rsk2 and from each other in this region, and the RSK3 cDNA has its distinctive nuclear localization sequence in this region. There is a stretch of sequence in the interkinase linker region (amino acids 387-405) that is well conserved between the two Rsk2 s but differs in sequence from Rsks 1, 3, and 4; if there are Rsk2-specific regulators or substrates, this region may be involved. Otherwise, Rsks 1-4 are very closely related to each other.
Although both protein kinase domains are very well conserved among Xenopus Rsk2 and the other Rsk family members, there is somewhat better conservation of the N-terminal   (15,23,51,52), cell survival functions (40,41), and transcriptional changes (34,36,37,39,72). The Rsk2⅐p42 MAPK complex can be disrupted by peptides from the Rsk2 C terminus. These findings suggest a potential way of specifically interfer- FIG. 5. Immunoprecipitation of Rsk1 and Rsk2 by various RSK antisera. A, RSK1, RSK2, and RSK3 sera bring down similar amounts of Rsk1 and Rsk2. Lysates from G 2 phase Xenopus oocytes were subjected to immunoprecipitation using RSK1, RSK2, or RSK3 antiserum, or protein G beads alone. The immunoprecipitates were subjected to 10.5% SDS-PAGE, followed by immunoblot analysis, as indicated above. The 68-kDa prestained molecular mass marker was run until 1 cm from bottom of gel to enhance separation of Rsk1 from Rsk2. B, kinase activities of RSK immunoprecipitates. Lysates from G 2 -or M phase oocytes were subjected to immunoprecipitation using RSK1, RSK2, or RSK3 antiserum, or protein G beads alone. The immunoprecipitates were washed and subjected to S6 kinase assay with an S6 peptide substrate and [␥-32 P]ATP. Results were quantified by Cerenkov counting.
FIG. 6. Complex formation between Rsk2 and p42 MAPK. A, p42 MAPK is pulled down with Rsk2. Lysates from Sf9 cells expressing either His 6 -Rsk2 or GST⅐MAPK were combined and incubated at 4°C for 90 min with protein G-agarose beads and anti-His 6 antibodies, both individually and together. His 6 -Rsk2 and any Rsk2-associated proteins were pulled down with anti-His 6 antibodies. The immunoprecipitates were then immunoblotted for Rsk2 and p42 MAPK. B, Rsk2 was pulled down with p42 MAPK. Lysates from Sf9 cells expressing His 6 -Rsk2 or GST⅐MAPK were combined with glutathione-agarose beads both individually and together and processed as in A. C, binding of endogenous Xenopus p42 MAPK to His 6 -Rsk2. G 2 phase Xenopus oocyte lysates were incubated for 20 min at 4°C in the presence or absence of purified Xenopus His 6 -Rsk2 protein. These lysates were then incubated with protein G-agarose beads in the presence and absence of anti-His 6 antibody and incubated for 90 min at 4°C. The first lane represents the total lysate used in each immunoprecipitation. After centrifugation and washing, the beads were resuspended in sample buffer and resolved by 10.5% SDS-PAGE, followed by p42 MAPK immunoblotting. D, Xenopus egg extracts containing both phosphorylated and unphosphorylated MAPK were treated similarly to C except that lysates were preincubated in the presence or absence of a Rsk2 C-terminal peptide for 10 min at 22°C prior to adding His 6 -Rsk2 protein. Beads were collected by centrifugation, washed, and processed as in C.
ing with Rsk activation in vivo.
Purified recombinant Xenopus Rsk2 can be activated to high specific activity by incubation with purified recombinant MEK R4F and p42 MAPK. This activation process involves phosphorylation by p42 MAPK as well as MAPK-induced autophosphorylation of Ser-383 by the C-terminal Rsk kinase domain. In addition, the protein kinase PDK1 has been implicated in Rsk activation, and PDK1Ϫ/Ϫ embryonic stem cells are defective in activating p90 Rsks (32). There are two candidate PDK1 sites in Rsk2; Ser-383 and Ser-224. The latter site lies in the sequence Tyr-Ser-Phe in the T-loop of the N-terminal kinase domain. In view of our finding that Rsk2 can undergo Ser-383 phosphorylation in the absence of PDK1, PDK1 is most likely essential for the phosphorylation of Ser-224.
Frödin and coworkers have recently proposed a sequential model for the activation of Rsks by MAPKs, based on their studies of various deletion mutants of rat Rsk2 expressed in COS7 cells (33) and other studies of Rsk activation in vitro and in transfected cells (31,70,71,73). The first step is the docking of MAPK at the C terminus of Rsk and the phosphorylation of Rsk in the C-terminal kinase domain activation loop and the interkinase linker region by MAPK. The second step is the autophosphorylation of Ser-383 (using the Xenopus Rsk2 numbering; in rat Rsk2 the corresponding residue is Ser-386) by the C-terminal kinase domain. The third step is the docking of PDK1 at the phosphorylated Ser-383, which results in activation of PDK1. The final step is the phosphorylation of Ser-224 (Ser-227 in rat Rsk2) in the N-terminal kinase domain by PDK1. At first it might seem difficult to reconcile the final step of this model with our finding that purified Rsk2 can be fully activated by MEK R4F plus p42 MAPK (with no added PDK1). We suspect that Xenopus Rsk2 has a substantial level of Ser-224 phosphorylation when isolated from insect cells. Similarly, Dalby et al. (31) have reported that inactive rat Rsk1 is par-tially phosphorylated at Ser-222 (the corresponding residue) when isolated from unstimulated COS-1 cells, even though the MAPK phosphorylation sites and MAPK-dependent autophosphorylation sites are essentially non-phosphorylated. Frödin et al. (33) have hypothesized that the phosphorylation of Ser-224 persists longer than the phosphorylation of the MAPK sites and autophosphorylation sites. This hypothesis would account for our observation that purified recombinant Rsk2 can be fully activated by MEK R4F and p42 MAPK, and Dalby's observation that Ser-222 is partially phosphorylated when Rsk1 is inactive. The hypothesis remains to be tested directly.