Activation of Mos in response to Δ90-cyclin B.
Our previous work suggested that Mos activity could be regulated without changing the level of the Mos protein (44
). To test this hypothesis further, we prepared a cycloheximide-treated interphase Xenopus
egg extract, added 30 nM recombinant GST-Flag-Mos, and then incubated the extract for 60 min with either no added cyclin or sufficient Δ90-cyclin B to produce a sustained M-phase state. The Mos was pulled down with Flag antibody and subjected to a kinase assay using purified recombinant GST-MEK1 as a substrate and phospho-MEK1 immunoblotting to detect phosphorylation. As a control, we carried out Flag immunoprecipitations on extracts to which no GST-Flag-Mos had been added. As shown in Fig. , in the absence of added GST-Flag-Mos, the Mos activities of the Flag immunoprecipitates were undetectable (lanes 1 and 2). However, in the presence of added GST-Flag-Mos, there was substantial Mos activity (lane 3), which increased approximately threefold after incubation with Δ90-cyclin B without any apparent change in Mos levels (lanes 3 and 4). The activation of Mos was accompanied by a modest increase in its apparent molecular mass that could be detected on low-bis polyacrylamide gels (Fig. ) but not on standard Laemmli gels (Fig. ).
FIG. 1. Mitotic activation of Mos. (A) Activation of recombinant GST-Flag-Mos by Δ90-cyclin B-treated Xenopus egg extracts. The top blot shows Mos activity as indicated by an in vitro kinase assay using recombinant GST-MEK1 as a substrate and phospho-MEK1 (more ...) MS-MS analysis of Mos phosphorylation.
One straightforward mechanism for the cyclin-induced activation of Mos (Fig. ) and the shift in apparent molecular weight (Fig. ) would be a change in Mos phosphorylation. To test this idea and map the Mos phosphorylation sites, we treated GST-Flag-Mos with either interphase extract or M-phase extract and immunoprecipitated the GST-Flag-Mos with Flag antibody. The interphase-treated GST-Flag-Mos ran as a single band by SDS polyacrylamide gel electrophoresis, with a non-Mos band running just above it (Fig. ). The M-phase-treated GST-Flag-Mos ran as three bands, with the highest band running just below a non-Mos band (Fig. ). We excised the interphase GST-Flag-Mos band and the three M-phase Mos bands and subjected them to tryptic digestion and MS-MS analysis to look for tryptic phosphopeptides.
FIG. 2. Mos phosphorylation sites. (A) GST-Flag-Mos was treated with Δ90-cyclin B-treated Xenopus egg extracts or interphase extracts and subjected to immunoprecipitation with Flag antibody, followed by Coomassie blue staining (top) or immunoblotting (more ...)
In total, five phosphorylation sites were identified (Fig. ). Two sites (Ser 25 and Ser 105) were phosphorylated in interphase Mos (Fig. ). Ser 25 remained phosphorylated in M-phase Mos (Fig. ) while Ser 105 became dephosphorylated, and three new phosphorylations (Ser 3, Ser 16, and Ser 26) appeared. Four sites (Ser 3, Ser 16, Ser 25, and Ser 26) were situated N terminal to the Mos kinase domain (Fig. ). One site (Ser 105) was situated at or near the N-terminal end of the αC helix, a secondary structure element in the N-terminal lobe of the kinase domain that plays a key role in the regulation of cyclin-dependent kinases and Src family tyrosine kinases (17
Sites important for the electrophoretic mobility shift.
To see which of the phosphorylations might be responsible for the electrophoretic mobility shift seen in M-phase wild-type Mos, we expressed and purified Mos mutants with each of the five phosphorylation sites individually mutated to alanine, incubated them with interphase or M-phase extracts, and then examined the mobilities of the Mos proteins by immunoblotting. We also examined the effects of mutation of three other conserved serine residues not found to be phosphorylation sites (Ser 18, Ser 57, and Ser 102). As shown in Fig. , the wild-type Mos protein and four of the mutants (S18A, S57A, S102A, and S105A) exhibited normal shifts. One protein (S25A) showed a slightly diminished shift (Fig. ), two proteins (S16A and S26A) showed only a trace of a shift, and one protein (S3A) showed no shift. Thus, it appears that all four N-terminal phosphorylation sites (Ser 3, Ser 16, Ser 25, and Ser 26) contribute to the normal mitotic shift in Mos; mutation of any of the sites at least partially compromises the shift.
FIG. 3. Electrophoretic mobility shifts in Mos phosphorylation site mutants. Samples of wild-type GST-Flag-Mos (Mos-WT) and various Mos mutants were incubated with Δ90-cyclin B-treated Xenopus egg extracts or interphase extracts and were then immunoprecipitated (more ...) Sites important for Mos activation.
We next set out to determine whether phosphorylation affected the kinase activity of Mos, using either interphase or M-phase extracts to activate wild-type or mutant Mos proteins and using recombinant MEK1 and phospho-MEK1 immunoblotting to determine the activities of the Mos proteins. First, we compared the activities of extract-treated Mos proteins to their activities after incubation with λ phosphatase. As shown in Fig. , phosphatase treatment increased the activity of interphase Mos (P = 0.02 as calculated by a paired, one-tailed Student t test). Phosphatase treatment also possibly decreased the activity of M-phase Mos (Fig. ), although the statistical significance of the decrease was borderline (P = 0.09). This suggested that phosphorylation was capable of both positively and negatively regulating Mos activity.
FIG. 4. Effects of phosphorylation on Mos activity. (A) Effect of λ phosphatase treatment on Mos activity. GST-Flag-Mos was treated with Δ90-cyclin B-treated Xenopus egg extracts or interphase extracts and subjected to immunoprecipitation with (more ...)
Next, we examined the effects of alanine or glutamate substitutions on the activity of Mos. Substitution of alanine for either Ser 16 or Ser 26, two of the sites phosphorylated in mitotic Mos, had no effect on the basal activity or mitotic activation of Mos (Fig. ). Likewise, substitution of alanine for Ser 25 (a constitutive phosphorylation site) had no detectable effect on Mos activity (data not shown). However, Mos-S3A showed a markedly decreased activity, indicating that Ser 3 phosphorylation is required for Mos activity (Fig. ). Mos-S3E had a normal basal activity and normal activation in M phase. Taken together, these results suggest that the glutamate substitution successfully mimics the effects of phosphorylating Ser 3, but something in addition to the phosphorylation of Ser 3 contributes to the mitotic activation of Mos.
Mos-S105E exhibited a markedly decreased activity, consistent with the hypothesis that dephosphorylation of Ser 105 is necessary for mitotic activation of Mos (Fig. ). Mos-S105A had a normal basal activity and normal activation in M phase, suggesting that although Ser 105 dephosphorylation is required for the full activation of Mos, something other than the dephosphorylation of Ser 105, possibly the phosphorylation of Ser 3, is required as well.
We devised a second test of the kinase activities of the various Mos mutants. We depleted cycloheximide-treated, interphase Xenopus extracts of Mos, which abolished the extract's ability to activate p42 MAPK in response to treatment with Δ90-cyclin B (Fig. ). We then supplemented the extract with sufficient wild-type Mos to largely restore the p42 MAPK response (30 nM) (Fig. ) and asked whether the mutant Mos proteins could also restore p42 MAPK response. The mutant proteins were all nominally 30 nM, based on Coomassie blue staining, although there was some variation in the apparent protein concentrations by immunoblotting (Fig. , top). We found that the Mos-S16A and Mos-S26A proteins partially restored p42 MAPK responses (Fig. ), consistent with the data showing that these sites are not required for the mitotic activation of Mos (Fig. ). Mos-S3E also partially restored p42 MAPK activation, but Mos-S3A did not (Fig. ), even at concentrations of up to 120 nM (data not shown). This argues again that Ser 3 phosphorylation is required for the mitotic activation of Mos. Mos-S105A partially restored p42 MAPK activation, but Mos-S105E did not (Fig. ), even at concentrations of up to 120 nM (data not shown), again arguing that the dephosphorylation of Ser 105 is required for the mitotic activation of Mos.
We next asked whether a combination of two mutations—S3E, presumably mimicking the phosphorylation of Ser 3, and S105A, mimicking the dephosphorylation of Ser 105—would suffice to produce mitotic levels of Mos activity in interphase extracts. As shown in Fig. , this was not the case; Mos-S3E/S105A had a normal basal activity and was activated normally in M-phase extracts. The same was true of the quadruple mutant Mos-S3E/S16E/S26E/S105A (Fig. ). In contrast, the double mutant Mos-S3A/S105E and the quadruple mutant Mos-S3A/S16A/S26A/S105E exhibited undetectably low activities in both interphase and M-phase extracts. This suggests that although Ser 3 phosphorylation and Ser 105 dephosphorylation are required for the mitotic activation of Mos, some third event is required as well.
Regulated binding of Mos to CK2β.
One possible mechanism to account for the activation of Mos-S3E/S105A in M-phase extracts (Fig. ) is provided by the work of Chen and coworkers (6
), who found that the regulatory (β) subunit of CK2 can bind to Mos and inhibit its activity. It seemed possible that the mitotic activation of Mos in Δ90-cyclin B-treated extracts could be due to decreased Mos-CK2β interaction, due either to changes in Mos, to changes in CK2β, or to both. To test these possibilities, we immunodepleted interphase extracts of CK2β (Fig. ) and assessed the ability of the extracts to activate Mos in the absence and presence of Δ90-cyclin B. As shown in Fig. , the phosphorylation of p42 MAPK was accelerated by approximately 10 min in the CK2β-depleted extracts, consistent with the hypothesis that CK2β is a physiologically relevant Mos inhibitor (6
). However, cyclin was still required for the phosphorylation of p42 MAPK in CK2β-depleted extracts (Fig. ), arguing that cyclin does not act solely by releasing Mos from CK2β.
FIG. 5. Regulated dissociation of CK2β from Mos. (A) Cyclin-dependent phosphorylation of p42 MAPK in CK2β-depleted extracts. Extracts were subjected to two rounds of immunodepletion with nonspecific immunoglobulin G or anti-CK2β. To one (more ...)
Adding recombinant CK2β back to CK2β-depleted extracts restored the normal timing of p42 MAPK phosphorylation (Fig. ). In addition, we supplemented the extracts with additional CK2β (Fig. ) and found that this increased the lag time between cyclin addition and p42 MAPK phosphorylation but did not render the extracts nonresponsive to cyclin (Fig. ).
Next, we examined whether the interaction between Mos and CK2β was regulated or constitutive. GST-Flag-tagged CK2β (20 nM) was incubated with GST-Flag-Mos (20 nM) in an interphase extract. Recombinant Δ90-cyclin B was added to an aliquot to drive it into M phase. We then immunoprecipitated the GST-Flag-Mos with Mos antibodies. Control experiments showed that equal amounts of GST-Flag-Mos were pulled down from interphase and cyclin-treated extracts (Fig. , lanes 7 versus 8, 9 versus 10, and 11 versus 12). The pulldowns were also probed for Mos-associated CK2β. As shown in Fig. small amount of CK2β came down with Mos from interphase extracts (lane 7) but not from M-phase extracts (lane 8). To determine whether the dissociation was due to changes in Mos phosphorylation, we carried out the same pulldown experiment with recombinant GST-Flag-Mos-S3A/S16A/S26A/S105E to mimic the interphase state of Mos and with GST-Flag-Mos-S3E/S3E/S26E/S105A to mimic the M-phase state of Mos. Both of these mutants still associated with CK2β during interphase and still dissociated from CK2β during M phase (Fig. , lanes 9 to 12). Thus, the dissociation of Mos from CK2β appears not to be triggered by a change in Mos phosphorylation; a change in CK2β (or possibly some other as-yet-unidentified factor) appears to be responsible for the dissociation.
Mos mutants as activators of oocyte maturation.
To test the biological relevance of the M-phase changes in Mos phosphorylation, we injected immature oocytes with various Mos proteins and determined their relative abilities to induce maturation. We used a concentration of Mos (3.5 ng/oocyte) that produced near-maximal maturation for wild-type GST-Flag-Mos and no maturation for the least active Mos protein (Mos-S3A/S16A/S26A/S105E) and injected ~50 oocytes with each of seven purified Mos mutants. As shown in Fig. , the biological potencies of the Mos proteins correlated well with their biochemical activities. Mos-S3E, Mos-S105A, and Mos-S3E/S16E/S26E/S105A were all similar to wild-type Mos in inducing maturation, whereas Mos-S3A, Mos-S105E, and Mos-S3A/S16A/S26A/S105E produced minimal levels of maturation. This is consistent with the hypothesis that the phosphorylation of Ser 3 and the dephosphorylation of Ser 105 are required for the activation of Mos as an inducer of meiotic maturation.
Mos function in the spindle assembly checkpoint.
The demonstration that p42 MAPK function is important for the spindle assembly checkpoint in Xenopus
egg extracts (24
), together with more-recent evidence implicating Mos in the M-phase activation of p42 MAPK in extracts (44
), implied that Mos might be important for the spindle assembly checkpoint. To test this hypothesis, we incubated CSF-arrested Xenopus
egg extracts with a high concentration of sperm chromatin (15,000 per μl) in the absence or presence of nocodazole (10 ng/μl) and then added calcium. As shown in Fig. , histone H1 kinase activity fell in the absence of nocodazole but was maintained at high levels in the presence of nocodazole, corroborating previous work (24
). Moreover, depleting Mos abrogated the extract's ability to maintain high levels of H1 kinase activity in the presence of nocodazole, and adding back recombinant wild-type Mos restored the H1 kinase activity to M-phase levels (Fig. ). This demonstrates that Mos is required for the spindle assembly checkpoint in extracts.
These findings also provide us with a second context in which to assess the biological activities of Mos mutants. As shown in Fig. , neither Mos-S3A nor Mos-S105E was able to maintain high levels of H1 kinase activity in Mos-depleted extracts treated with sperm chromatin, nocodazole, and calcium. Thus, both oocyte maturation (Fig. ) and the spindle assembly checkpoint (Fig. ) require the presence of an activable form of Mos.
Phosphorylation of Ser 3 and Ser 105 in oocytes and extracts.
We next set out to determine whether the phosphorylation of Ser 3 and Ser 105 was regulated in vivo. We obtained phospho-specific antibodies against phosphoserine 3 (pS3-Mos) and phosphoserine 105 (pS105-Mos). As shown in Fig. , the pS3-Mos antibody recognized M-phase wild-type Mos but not interphase wild-type Mos. The pS3-Mos antibody did not recognize Mos-S3A but did recognize M-phase Mos-S105A (Fig. ). Conversely, the pS105-Mos antibody recognized interphase wild-type Mos but not M-phase wild-type Mos and recognized interphase Mos-S3A but not interphase Mos-S105A (Fig. ). This establishes the specificities of the two antibodies.
FIG. 7. Phosphorylation of Mos at Ser 3 and Ser 105 during oocyte maturation and after egg activation. (A) Specificities of the pS3-Mos and pS105-Mos antibodies. Wild-type (WT) or mutant GST-Flag-Mos was treated with Δ90-cyclin B-treated Xenopus egg extracts (more ...)
We then looked for changes in Mos levels and Mos phosphorylation during Xenopus
oocyte maturation. In the experiment whose results are shown in Fig. , H1 kinase activity first appeared about 2 h after progesterone treatment. Mos was first detectable 1 h after progesterone treatment and reached maximal levels by 4 to 5 h (Fig. ), consistent with previous reports (2
). The time course of Ser 3 phosphorylation was similar, reaching maximal levels by 4 to 5 h, but there was no detectable pS105-Mos signal (Fig. ). These findings indicate that fairly early on in maturation, the balance of kinase and phosphatase activities favors the phosphorylation of Ser 3 and the dephosphorylation of Ser 105 (Fig. and ).
We also followed Mos levels and phosphorylation after treating dejellied eggs with calcium ionophore to parthenogenetically activate them. As shown in Fig. , Mos levels began to fall within 20 min of ionophore treatment and continued to fall through 60 min. This gradual decrease in Mos levels resulted in a more temporally abrupt decrease in p42 MAPK phosphorylation, which appeared to be nearly complete within 30 min of ionophore treatment. This suggested that something in addition to the gradual decrease in Mos levels contributed to the inactivation of the p42 MAPK cascade, and indeed the inactivation of p42 MAPK corresponded temporally to the dephosphorylation of Mos at Ser 3 and the phosphorylation of Mos at Ser 105 (Fig. ). Thus, it appears that the inactivation of the p42 MAPK cascade after egg activation results not only from a decrease in Mos levels but also from a decrease in Ser 3 phosphorylation and an increase in Ser 105 phosphorylation.