Induction of Metaphase Arrest in Cleaving Xenopus Embryos by the Protein Kinase p90Rsk

See allHide authors and affiliations

Science  12 Nov 1999:
Vol. 286, Issue 5443, pp. 1365-1367
DOI: 10.1126/science.286.5443.1365


Before fertilization, vertebrate eggs are arrested in metaphase of meiosis II by cytostatic factor (CSF), an activity that requires activation of the mitogen-activated protein kinase (MAPK) pathway. To investigate whether CSF arrest is mediated by the protein kinase p90Rsk, which is phosphorylated and activated by MAPK, a constitutively activated (CA) form of Rsk was expressed inXenopus embryos. Expression of CA Rsk resulted in cleavage arrest, and cytological analysis showed that arrested blastomeres were in M phase with prominent spindles characteristic of meiotic metaphase. Thus, Rsk appears to be the mediator of MAPK-dependent CSF arrest in vertebrate unfertilized eggs.

The unfertilized eggs of most vertebrates are naturally arrested at metaphase of meiosis II. Upon fertilization, an increase in free calcium activates the anaphase-promoting complex, which drives exit from mitosis and entry into the first embryonic cell cycle. The enzymatic activity that causes metaphase arrest is CSF; it appears in meiosis II during oocyte maturation and disappears shortly after fertilization (1). Microinjection of CSF into blastomeres of cleaving embryos causes arrest in metaphase of the next cell cycle (CSF arrest).

CSF has not been purified, and therefore its molecular composition is unknown. However, the MAPK pathway is required for the generation of CSF activity. The c-mos proto-oncogene product (Mos), which is a MAPK kinase kinase, is synthesized during meiosis inXenopus and mice, and its expression is sufficient to produce CSF arrest in injected blastomeres (2, 3). CSF arrest can also be induced by a thiophosphorylated, phosphatase-resistant form of MAPK (4).

In Xenopus eggs and other systems, the 90-kD ribosomal protein S6 kinase (p90Rsk) is directly phosphorylated and activated by MAPK (5); during oocyte maturation, activation of p90Rsk closely parallels that of MAPK, and both enzymes are dephosphorylated and deactivated after fertilization, when CSF activity also disappears (3, 6). Cloning of p90Rsk inXenopus and mammalian systems revealed a structure with two distinct kinase domains (7). Three different isoforms of p90Rsk, termed Rsk1, Rsk2, and Rsk3, are present in mammalian cells, and all have the same basic two-domain structure (7, 8). Activation of p90rsk requires phosphorylation at two specific sites (Thr570and Ser362) by MAPK and autophosphorylation at a specific site (Ser378) by the COOH-terminal kinase domain (Fig. 1A) (9). Activity also requires phosphorylation by an unidentified kinase of a site in the middle of the activation loop of the NH2-terminal kinase domain (Ser220) (9).

Figure 1

A truncation mutant of p90Rsk1 is hyperactivated in vivo. (A) Primary structure of Xp90Rsk and deletion constructs (12). For Rsk, the two kinase domains and known phosphorylation sites (P) are indicated. ΔCT Rsk denotes a construct in which the entire COOH-terminal kinase domain (ΔCT) has been removed. CA Rsk is a construct identical to ΔCT except that 44 amino acids at the NH2-terminus were also deleted. All constructs have an identical NH2-terminal FLAG tag. (B) Hyperactivation of CA Rsk in vivo. Messenger RNAs encoding the FLAG-tagged constructs were injected into resting oocytes. After incubation for 18 hours, one group of oocytes was treated with progesterone to induce maturation, and after GVBD, immunoprecipitates of the activated enzymes were assessed for kinase activity or blotted with antibodies to FLAG to determine the amount of expression of each construct (13). Specific activities were calculated by normalizing total kinase activity to densitometric measurements of FLAG immunoreactivity and are shown relative to the activity of Rsk at GVBD.

To evaluate a possible role for p90Rsk in mediating CSF arrest, we generated a constitutively active form of the enzyme. Inasmuch as the phosphorylation of exogenous substrates by p90Rsk is mediated by the NH2-terminal kinase domain (10), we generated a construct in which the COOH-terminal domain was deleted completely with or without a truncation of the NH2-terminal domain similar to that which causes constitutive activation of MAPK kinase (11, 12). To evaluate their activation in vivo, we injected mRNAs encoding FLAG-tagged versions of these two constructs or the wild-type full-length enzyme (Fig. 1A) into resting stage VI oocytes. Some of the oocytes were then treated with progesterone to cause meiotic maturation, which activates endogenous Rsk to 75% of its maximum activity by the time of germinal vesicle breakdown (GVBD) (6). Rsk proteins were isolated from resting and GVBD oocytes with antibody to FLAG coupled to agarose beads and assayed for kinase activity in vitro (13). The NH2- and COOH-terminally truncated enzyme (CA Rsk) was activated by progesterone treatment to a specific activity 4 times that of wild-type Rsk and 12 times that of Rsk with only the COOH-terminal truncation (ΔCT Rsk) (Fig. 1B).

To assess whether MAPK could activate Rsk in vitro, we incubated with MAPK the FLAG-tagged wild-type protein recovered from a resting oocyte (13). MAPK activated the enzyme 70-fold and a Ser220 → Ala220 mutant was not activated at all (Fig. 2A). The sequence around Ser220 is very closely related to a phosphorylation site in the p70S6K that is phosphorylated by 3-phosphatidylinositol–dependent kinase–1 (PDK-1) (14). When recombinant PDK-1 was incubated with FLAG-tagged Rsk from resting oocytes, activity increased minimally, by about twofold. However, when both PDK-1 and MAPK were incubated with Rsk, activity increased nearly 100-fold (Fig. 2). In contrast, the presence of both MAPK and PDK-1 caused only about a fivefold activation of CA Rsk and ΔCT Rsk, which lack the MAPK phosphorylation site at Thr570. Recombinant Rsk lacking Ser220 was not activated. These results indicate that the kinase mediating the required phosphorylation at Ser220 in Rsk may be PDK-1 (9) and this site is highly phosphorylated even in resting oocytes. For a comparison of the activation in vitro with that in vivo, oocytes expressing these constructs that had undergone GVBD were assayed for Rsk activity, which was compared to that in resting oocytes (Fig. 2B). Rsk was activated 55-fold whereas ΔCT Rsk and CA Rsk were activated almost 25-fold despite their different specific activities (Fig. 1B). The activation of Rsk in vivo at GVBD may be less than that in vitro because only 50 to 75% of endogenous Rsk is activated at GVBD (6).

Figure 2

Activation of p90Rsk by MAPK and PDK-1 in vitro or at GVBD. (A) In vitro activation of p90Rsk1. Messenger RNAs encoding FLAG-tagged full-length wild-type Rsk, ΔCT Rsk, CA Rsk, or Ser220 → Ala220 Rsk (Ser/Ala220) were injected into resting oocytes. After incubation for 18 hours, FLAG-tagged protein was immunoprecipitated and phosphorylated with either MAPK or PDK-1, or both, as indicated. Subsequently, the immunoprecipitates were washed and assayed for S6 peptide kinase activity (13). To control for differences in expression of each construct, we expressed activation as fold activation over activity present in the absence of phosphorylation by either MAPK or PDK-1. (B) In vivo activation at GVBD. Messenger RNAs encoding the indicated constructs were injected, and Rsk proteins were isolated as immune precipitates from resting and GVBD oocytes. Subsequently, kinase assays were performed, and fold activation is expressed in relation to that in resting oocytes.

Because CA Rsk had achieved a higher specific activity than the other Rsk variants in oocytes undergoing maturation, we evaluated its ability to cause CSF arrest after injection of mRNA encoding FLAG-tagged CA Rsk into one blastomere of a two-cell embryo (15). Injected blastomeres ceased dividing after two or three divisions, whereas the uninjected side continued to divide (Fig. 3). Expression of β-galactosidase (β-Gal), Rsk, or ΔCT-Rsk did not cause cleavage arrest. The arrest caused by CA Rsk was similar to that seen after injection of mRNA encoding Mos in both timing and appearance (Fig. 3) (2).

Figure 3

Cleavage arrest in blastomeres caused by CA Rsk. One blastomere of a two-cell embryo was injected with mRNA encoding either β-Gal, Rsk, ΔCT Rsk, CA Rsk, or Mos, as indicated (15). Development was monitored with a dissecting microscope and photographed when control embryos had reached stage 7. Similar results were obtained in six independent experiments.

Confocal microscopy of arrested embryos stained for DNA and α-tubulin revealed metaphase spindles that exhibited a meiotic-like morphology in that they lacked prominent asters and exhibited the classical “barrel” morphology (Fig. 4) (16–18). In some cases, two spindles were seen in one large blastomere, perhaps reflecting failure of cytokinesis as previously reported for Mos overexpression in somatic cells (18). The spindles present in blastomeres arrested after the injection of Mos mRNA were very similar in morphology to those in blastomeres injected with CA Rsk (Fig. 4). These results indicate that arrested blastomeres are in metaphase and that Rsk may be a mediator of CSF activity.

Figure 4

Metaphase spindles in blastomeres arrested by CA Rsk or Mos. Blastomeres arrested after injection of mRNA for either CA Rsk or Mos (Fig. 3) were fixed 5 hours after injection and analyzed by confocal microscopy for spindles as described (16). Red indicates α-tubulin, green indicates DNA, and yellow indicates an overlapping signal. Scale bar, 10 μm.

During meiosis, MAPK is part of a feedback loop that promotes synthesis of Mos, thereby strengthening signal transduction through the pathway (3, 19). In arrested blastomeres expressing CA Rsk, Mos was undetectable, and MAPK and endogenous Rsk were not activated, whereas both MAPK and Rsk were activated in Mos-expressing blastomeres (Fig. 5A). This suggests that expression of CA Rsk did not induce expression of Mos to activate the MAPK pathway and produce CSF arrest but rather is the immediate downstream component of this pathway mediating the arrest. In CA Rsk- and Mos-arrested blastomeres, an increased fraction of cyclin B2 was present in a slower migrating form, characteristic of M phase (Fig. 5A) (3).

Figure 5

CSF arrest induced by CA Rsk without activation of the endogenous MAPK pathway. (A) Biochemical analysis of CSF arrest. Both blastomeres of a two-cell embryo were injected with FLAG-tagged mRNA encoding β-Gal, Rsk, ΔCT Rsk, CA Rsk, or Mos, as indicated. Five hours after CSF arrest induced by either CA Rsk or Mos, arrested blastomeres were isolated and analyzed for the indicated individual components by protein immunoblotting. pMAPK, phosphorylated (active) MAPK; pRsk, Rsk phosphorylated on Ser362, which denotes the activated state of endogenous Rsk (9, 21). (B) Total Rsk activity in eggs and embryos. Extracts from embryos expressing β-Gal, Rsk, ΔCT Rsk, CA Rsk, or Mos as in (A) were assayed for S6 peptide kinase activity. (C) S6 peptide kinase activity of FLAG immunoprecipitates from embryos expressing β-Gal, Rsk, ΔCT Rsk, and CA Rsk. The same extracts as in (B) were also immunoprecipitated with antibodies to FLAG coupled to agarose and assayed for S6 peptide kinase activity.

To compare the level of total Rsk activity in embryos expressing β-Gal, Rsk, ΔCT Rsk, CA Rsk, and Mos, we assayed extracts of these embryos for total S6 kinase activity in vitro. CA Rsk expression resulted in a level of Rsk activity equivalent to 50% of that caused by the expression of Mos (Fig. 5B). Kinase assays of FLAG immunoprecipitates from these extracts revealed that only CA Rsk produced substantial kinase activity (Fig. 5C), demonstrating the constitutive activation of CA Rsk in blastomeres. Whether altered signaling, changes in localization, or some other mechanism accounts for the high CA Rsk activity in blastomeres versus that in resting oocytes remains to be determined.

These results indicate that the meiotically activated Mos/MAPK pathway causes CSF arrest through the MAPK-dependent activation of p90Rsk. Because Rsk expression did not activate the endogenous MAPK pathway, MAPK requires no other substrate for induction of CSF arrest. The importance of Rsk in CSF activity is supported by a study by Bhatt and Ferrell (20), who show that Rsk is necessary for expression of CSF activity in egg extracts. Our data indicate that CSF arrest is not a consequence of direct regulation of the spindle assembly checkpoint or the anaphase-promoting complex by MAPK.

  • * To whom correspondence should be addressed. E-mail: jim.maller{at}


Stay Connected to Science

Navigate This Article