Differential Phosphorylation Controls Maskin Association with Eukaryotic Translation Initiation Factor 4E and Localization on the Mitotic Apparatus

Oocyte collection and injection.

Stage VI Xenopus oocytes were dispersed from ovaries by treatment with collagenase. They were injected with mos sense (ATCTAGTACAGTATCTCAATGTCCA) or antisense (complementary sequence) oligonucleotides (∼3 mg/ml, ∼70 nl/oocyte) as described by deMoor and Richter (3). Other oocytes were injected with mRNAs encoding wild-type (WT) or mutant Maskin (∼25 ng/oocyte) or Escherichia coli-expressed protein and incubated for 16 h at 20°C before Western blotting for Maskin and cyclin B1. Some of the oocytes were also cultured in the presence of progesterone for varying times before they were processed for Western blotting (probed for Maskin and cyclin B1) or two-dimensional phosphopeptide mapping. Some oocytes were also used to create extracts for H1 kinase assays, as described previously (3), and Maskin in vitro phosphorylation assays. Briefly, oocytes were lysed in H1 kinase buffer (80 mM β-glycerophosphate, 20 mM EGTA, 15 mM MgCl₂, 0.5 mM NaVaO₄, and protease inhibitors) at 20 μl per oocyte, centrifuged to remove insoluble material, and stored on ice.

Construction of Maskin mutants.

Full-length Maskin cDNA cloned into pET 30a (34) was used as a template for PCR-generated deletion mutants which were subsequently cloned into the BamHI and XhoI sites of the pET30a vector, with the exception of the S338-K754 mutant, which was created by digestion of the Maskin cDNA with HindIII and insertion into pET40. Point mutations of specific residues to alanine and aspartic acid were carried out with a Quick Change site-directed mutagenesis kit (Stratagene).

Two-dimensional phosphopeptide mapping.

To examine endogenous Maskin phosphorylation, ∼300 oocytes were incubated overnight in Barth's medium containing 0.1 mCi/ml [³²P]orthophosphate. The oocytes were further incubated with progesterone until maturation. They were lysed and centrifuged, and the supernatant was subjected to Maskin immunoprecipitation as previously described (34). Briefly, oocytes were washed in Barth's medium to remove free [³²P]orthophosphate, and then 100 labeled oocytes were resuspended in 500 μl phospoimmunoprecipiation (phospho-IP) buffer (100 mM KCl, 50 mM Tris [pH 7.5], 50 mM NaF, 80 mM β-glycerolphosphate, 5 mM EDTA, 5 mM Na pyrophosphate, 0.1 mM NaVaO₄, 0.1% NP-40, and protease inhibitors). The lysate was centrifuged to remove insoluble material and then cleared by incubation with a 100-μl bed volume of Sepharose at 4°C for 30 min. Fifteen microliters of anti-Maskin serum was added to the lysate with 100-μl bed volume protein A Sepharose and placed on a rocker at 4°C for 2 h. The beads were washed extensively with IP buffer (five times with 5 to 10× volume). The precipitated proteins were resuspended in sodium dodecyl sulfate (SDS) sample buffer, resolved by SDS-polyacrylamide gel electrophoresis (PAGE), and blotted onto polyvinylidene difluoride membrane. One side of the blot was subjected to immunoblotting to confirm that the labeled protein was indeed Maskin, while the majority of the labeled protein was digested with tosylsulfonyl phenylalanyl chloromethyl ketone-treated trypsin and the phosphopeptides were analyzed by thin-layer electrophoresis followed by thin-layer chromatography and phosphorimager analysis (23). H1 kinase extracts (see above) derived from noninjected or mos sense or antisense oligonucleotide-injected oocytes, in vitro-matured oocytes, or ovulated eggs were used for Maskin in vitro phosphorylation assays. The reactions were carried out in a final volume of 30 μl containing E. coli-expressed full-length or mutant histidine-tagged Maskin (1 to 3 μg), 8 μl extract, 50 mM Tris (pH 7.5), 10 mM MgCl₂, 1 mM EGTA, 1 mM dithiothreitol, 0.1 mM ATP, and 20 μCi [³²P-γ]ATP. Following incubation for 30 min at 30°C, Maskin was processed for two-dimensional phosphopeptide mapping as described above. Maskin was also phosphorylated in vitro with mitogen-activated protein (MAP) kinase (Calbiochem), cdk1/cyclin B (Calbiochem), or PKA (Sigma). In these reactions, oocyte extract was replaced with H1 kinase extract buffer. In some experiments, a PKA inhibitor, PKI (Calbiochem), was also added to the extracts. The resulting [³²P]phosphopeptides were processed as described above.

Analysis of eIF4E-Maskin interaction.

Cap (7mGTP) column chromatography has been described previously (2, 34). Briefly, mRNAs encoding Maskin (WT or A6 mutant) as well as eIF4E were translated in rabbit reticulocyte lysates in the presence of [³⁵S]methionine. Prior to chromatography, the lysates were supplemented with 0.25 mM GTP and in some cases with 0.25 mM 7mGTP as well. Some of the lysates also contained cdk1/cyclin B1 or PP2A (protein phosphatase 2A), the latter of which reduced or eliminated Maskin phosphorylation by lysate kinase(s). The protein that was retained on the cap columns was analyzed by SDS-PAGE and autoradiography.

Analysis of Maskin localization on spindles and centrosomes.

Xenopus egg extracts were primed with rhodamine-labeled tubulin as well as demembranated sperm DNA as described previously (4). The resulting spindles were fixed, centrifuged through a glycerol cushion onto glass slides (4), and immunostained with CPEB or Maskin antibody followed by fluorescein-conjugated secondary antibody. The DNA was stained with 4′,6′diamidino-2-phenylindole (DAPI). The spindles were visualized with a Nikon E600 Eclipse microscope. Centrosome-bound Maskin was also visualized in Xenopus XTC cells, which were immunostained with Maskin antibody (1:5,000) or nonimmune serum (1:500) as well as tubulin antibody. WT, S626A, S626D, and A6 Maskin molecules were fused to green fluorescent protein (GFP) and transfected using Lipofectamine (Invitrogen) into HeLa cells (18). GFP fluorescence demonstrated that WT, but not S626A Maskin, was associated with spindles and centrosomes. The GFP moiety was also used for Maskin coimmunoprecipitation experiments from transfected HeLa cells. Plates of transfected HeLa cells (60 mm) were washed with phosphate-buffered saline and then lysed in 0.333 ml IP buffer (100 mM KCl, 50 mM Tris [pH 7.5], 1 mM EDTA, 0.5% NP-40, 200 μg/ml RNase A, and protease inhibitors) for 30 min on ice. Lysates were centrifuged and cleared by incubation with nonimmune rabbit serum and protein A Sepharose for 1 h at 4°C. Cleared lysates were incubated with 2 μl anti-GFP (guinea pig) and protein A Sepharose for 2 h at 4°C and then washed extensively in IP buffer. The precipitates were immunoblotted for Maskin and MAP-215.

Several Maskin phosphorylation sites are developmentally regulated.

Because Maskin is phosphorylated on several sites, we sought to determine which, if any, were maturation specific; this might indicate those that would be the most likely to regulate translation. Oocytes were injected with an antisense oligonucleotide directed against mos mRNA; the ablation of this mRNA prevents the progesterone-induced activation of the MAP kinase cascade and the resulting activation of MPF and serves as a convenient method for distinguishing early (pre-mos) and late (post-mos) events of maturation (3, 32). As expected, the mos antisense oligonucleotide prevented progesterone-induced MPF activation as determined by histone H1 phosphorylation. Compared to noninjected oocytes, the mos sense oligonucleotide had little effect, since in both groups of oocytes MPF activity was detected at 6 h after progesterone addition (Fig. 2A). Next, noninjected or mos antisense oligonucleotide-injected oocytes were cultured for 2 to 8 h with progesterone before extracts were prepared and supplemented with [³²P]ATP and E. coli-expressed Maskin. Extracts prepared from noninjected oocytes cultured for 2 h in progesterone yielded three obvious Maskin phosphopeptides (Fig. 2B, left panels, circled). As maturation proceeded, the three phosphopeptides remained but other phosphopeptides began to appear, particularly at 6 to 8 h when the oocytes contained robust MPF activity. In extracts derived from mos antisense oligonucleotide-injected oocytes, the same three phosphopeptides were observed at 2 or up to 8 h after progesterone addition (Fig. 2B, right panels, circled). These results show that Maskin undergoes a complex series of developmentally controlled phosphorylation events.

Identification of maturation-specific Maskin phosphorylation sites.

Because the maturation-specific phosphorylations of Maskin correlate with CPE-dependent mRNA translation (2, 34), we thought that they would be the most likely ones to influence eIF4E binding, and accordingly they were the first ones we sought to identify. Because our initial attempts to combine two-dimensional mapping and mass spectrometry or amino acid sequencing were not successful, we employed two other approaches. First, we made several deletions in Maskin and determined which phosphopeptides were lost when the E. coli-expressed proteins were phosphorylated in egg extracts. Figure 3A and B show that with this method, several phosphopeptides were tentatively mapped to various regions of Maskin. In combination with phosphoamino acid analysis that revealed a large amount of phosphoserine, some phosphothreonine, and no phosphotyrosine (data not shown), we were further able to ascertain which phosphopeptides (i.e., a-l) were likely to reside in certain regions of Maskin. However, because this approach could not identify specific phosphoamino acids, we addressed whether kinases that are known to become active during oocyte maturation would phosphorylate Maskin in vitro, and, if so, whether the phosphoresidues were the same as those observed when Maskin was phosphorylated in egg extracts. Because such kinases have preferred phosphorylation motifs, this would aid us in determining which amino acids were phosphorylated. The first kinase we used was cdk1 (cdc2), a component of MPF. This kinase, which usually recognizes SP and TP pairs, phosphorylated many of the same peptides in vitro as was detected in egg extracts (compare Fig. 3C with A). MAP kinase, which also recognizes SP and TP pairs, phosphorylated many fewer residues, but phosphopeptides a, c, d, and e were detected (Fig. 3D). Note that while cdk1 and MAP kinase both phosphorylated peptides a, c, d, and e, they did so to different extents. Taken together, these studies indicated the amino acids that were the most likely to be phosphorylated in egg extracts; indeed, when these residues were changed to alanine individually, specific phosphopeptides were lost (Fig. 3E to J, circled phosphopeptides were lost, compare with panel A). In combination, the substitution of six residues for alanine led to the loss of almost all the phosphopeptides (Fig. 3J, in this case the circled phosphopeptides were those that remained); some of the other peptides (Fig. 3J, circled and labeled “2”) were detected in immature oocytes (see Fig. 2 and 4). Figure 3K shows the assignment of several phosphoamino acids and Fig. 3L shows their relationship to the eIF4E binding domain and the coiled domain of Maskin.

We also note that differential trypsin digestion could yield multiple phosphopeptide spots that disappear with a single alanine substitution. For example, S638 is found in the following context: RES*PKK. Digestion with trypsin could give up to four peptides, since the E residue makes the first R a poor trypsin site, thus upstream K would sometimes be used. In addition, the tandem K residues could result in two possible carboxy end sites of the same phosphopeptide.

Protein kinase A (PKA) phosphorylates Maskin in oocyte extracts.

Oocytes contain a kinase(s) that phosphorylates Maskin prior to and after the synthesis of mos (Fig. 2). As before, we considered what kinases(s) might be active before mos synthesis in progesterone-treated oocytes and even in immature (unstimulated) oocytes. The most obvious possibility was PKA, not only because Maskin contains a motif that is potentially recognized by this enzyme (RKQS⁶²⁶L) but also because Maller and Krebs (21) showed that it was active in oocytes. While this result has been questioned (33), a number of other studies strongly indicate that PKA is an important maturation-controlling enzyme in oocytes (6, 22, 38). To investigate the possible involvement of PKA, egg extracts were supplemented with PKI, a specific inhibitory peptide of this enzyme. Extracts were then supplemented with E. coli-expressed Maskin, which was then examined by two-dimensional phosphopeptide mapping. Figure 4A shows that the three phosphopeptides (circled) that were observed in extracts from eggs as well as from mos antisense oligonucleotide-injected oocytes (Fig. 2B) were eliminated, suggesting that they were products of PKA activity. Next, Maskin was phosphorylated in extracts derived from untreated oocytes; the same three phosphopeptides were observed (Fig. 4B). They were also observed when Maskin was phosphorylated in vitro with commercial PKA, which was confirmed by a mixing experiment (Fig. 4B). Maskin contains a PKA consensus site at S626. When this residue was mutated and the protein again phosphorylated in extracts, none of the three phosphopeptides was detected (Fig. 4C). Therefore, we surmise that Maskin is a substrate for PKA in oocytes prior to and after maturation and that the three phosphopeptides contain a single phosphorylated amino acid (S626); they most probably arose through partial trypsin digestion.

Maskin phosphorylation regulates eIF4E binding and mRNA translation.

During oocyte maturation, PABP-bound eIF4G displaces Maskin from eIF4E, thereby promoting the translation of cyclin B1 mRNA. Because the cdk1 phosphorylations of Maskin residues T58, S156, S311, S343, S453, and S638 also coincide with this event, we thought that they might contribute to the dissociation of this protein from eIF4E. To assess this possibility, mRNAs encoding wild-type (WT) or mutant Maskin proteins where the six residues noted above were changed to alanine (henceforth referred to as Maskin A6) as well as eIF4E were translated in a reticulocyte lysate in the presence of [³⁵S]methionine (note that while newly synthesized eIF4E contributes very little to the overall pool size of this protein in the lysate, the radiolabeling obviates the need for immunoblotting). The lysates were then supplemented with 0.2 mM GTP and in one case with 0.25 mM 7mGTP as well. Some of the lysates were further supplemented with the kinase cdk1/cyclin B1 (i.e., MPF) or phosphatase PP2A (to eliminate or reduce possible Maskin phosphorylation by reticulocyte kinases). The lysates were then applied to cap columns, and the material that was retained was analyzed by SDS-PAGE and phosphorimaging. The load fraction (1/20th the amount that was applied to the columns) shows that WT and A6 Maskin mRNAs were translated to the same extent and that the level of A6 Maskin was unaffected by cdk1 (Fig. 5A, lanes 6 to 8). While WT Maskin (plus PP2A) was retained on the cap column, WT Maskin that was phosphorylated by cdk1 was not; in both cases, however, eIF4E was retained on the column to the same extent (lanes 1 and 2). In contrast, A6 Maskin was retained on the column irrespective of whether cdk1 was added to the lysate (lanes 3 and 4); again in both cases, similar levels of eIF4E were retained. When free cap (7mGTP) was added to the lysate, neither Maskin nor eIF4E was retained on the cap column (lane 5). Thus, cdk1-mediated phosphorylation of Maskin controls its interaction with eIF4E.

To determine whether the cdk1 phosphorylations of Maskin influence translation, mRNA encoding WT or A6 Maskin proteins was injected into oocytes, some of which were then incubated with progesterone. At several time points, the oocytes were harvested and the extracted protein was probed on Western blots for Maskin and cyclin B1. All the oocytes synthesized about the same amount of Maskin, which was ∼8-fold greater than endogenous Maskin. As expected, cyclin B1 levels accumulated in response to progesterone even when excess WT Maskin mRNA was injected (Fig. 5B). In contrast, the accumulation of cyclin B1 was somewhat slower when the A6 Maskin mRNA was injected. These results were confirmed in Fig. 5C, where E. coli-expressed WT and A6 Maskin proteins were injected. In a more dramatic fashion, the A6 Maskin protein prevented cyclin B1 accumulation compared to WT Maskin (overall [³⁵S]methionine incorporation was unaffected, indicating that general cap-dependent translation is not regulated by Maskin; data not shown). We infer from these results that the A6 Maskin protein, which is not phosphorylated by cdk1 and thus not readily dissociated from eIF4E, acted as a dominant-negative mutant protein and inhibited cyclin B1 mRNA translation.

S626 phosphorylation controls Maskin association with the mitotic apparatus.

We could detect no change in cyclin B1 mRNA translation in oocytes injected with Maskin S626A (data not shown). To assess the possible function of this phosphoresidue, we considered another aspect of Maskin, which is its association with the mitotic apparatus (13). Maskin contains a coiled-coil domain that resembles those in the TACC (transforming acidic coiled-coil) family of proteins (7, 8, 9). Because these proteins are often associated with centrosomes and spindle microtubules and contain conserved putative PKA phosphorylation sites in the same region as Maskin (e.g., human TACC3, Drosophila melanogaster DTACC; see below), we thought that the S626 phosphorylation might mediate Maskin localization on the mitotic apparatus. To first confirm the data of Groisman et al. (13) that Maskin indeed is associated with spindles and centrosomes, rhodamine-labeled tubulin and demembranated sperm DNA were added to egg extracts; the resulting spindles were immunostained for Maskin as well as for CPEB and then collected on glass coverslips by centrifugation and visualized by fluorescence microscopy. Figure 6 demonstrates that both of these proteins were associated with the mitotic apparatus; Maskin was detected on spindles but was particularly concentrated at centrosomes. CPEB, on the other hand, was not particularly concentrated at centrosomes but appeared as puncta on the spindles. We next added E. coli-expressed Maskin (WT and S626A mutant) to egg extracts together with rhodamine-tubulin to assess whether the two proteins would be differentially localized to the mitotic apparatus. Unfortunately, neither protein was detected on spindles or centrosomes, perhaps denoting a foible of the extract (data not shown). We therefore turned to living cells to examine Maskin localization. In XTC cells, a Xenopus somatic cell line, Maskin again was detected strongly at centrosomes (Fig. 7A). To investigate the function of the S626 phosphorylation, we generated GFP-Maskin fusion proteins to distinguish between endogenous and exogenous Maskin. However, because these cells were transfected with a very low efficiency, we could not use them to test the function of S626 phosphorylation. Consequently, HeLa cells, which contain no protein that reacts with our anti-Xenopus Maskin antibody, were efficiently transfected with constructs encoding GFP-Maskin fusion proteins, in one case containing an S626A alteration. While WT Maskin strongly localized to the mitotic apparatus, the GFP-S626A Maskin remained dispersed throughout the cytoplasm (Fig. 7A). Conversely, cdk1 phosphorylation of Maskin appears to have no effect on localization as GFP-Maskin A6 was detected on spindles and centrosomes (data not shown).

TACC family member proteins are linked to microtubules through an association with MAP-215 (7, 8, 9, 28). Because Maskin has several structural similarities to vertebrate and invertebrate TACC proteins (34), including a conserved serine corresponding to S626 (Fig. 7B), we thought that the phosphorylation of this residue might mediate an interaction between Maskin and MAP-215. To test this possibility, HeLa cells transfected with WT, S626A, or A6 Maskin proteins fused to GFP were subjected to immunoprecipitation with GFP antibody and immunoblotted for Maskin and MAP-215 (Fig. 7C). Nearly equal amounts of MAP-215 were coimmunoprecipitated with GFP antibody irrespective of whether the Maskin moiety was WT or contained amino acid substitutions. Thus, Maskin S626 phosphorylation does not affect binding to MAP-215.