Cytoplasmic CstF-77 Protein Belongs to a Masking Complex with Cytoplasmic Polyadenylation Element-binding Protein in Xenopus Oocytes*

Regulated mRNA translation is a hallmark of oocytes and early embryos, of which cytoplasmic polyadenylation is a major mechanism. This process involves multiple protein components, including the CPSF (cleavage and polyadenylation specificity factor), which is also required for nuclear polyadenylation. The CstF (cleavage stimulatory factor), with CPSF, is required for the pre-mRNA cleavage before nuclear polyadenylation. However, some evidence suggests that the CstF-77 subunit might have a function independent of nuclear polyadenylation, which could be related to the cell cycle. As such, we addressed the question whether CstF-77 might have a role in cytoplasmic polyadenylation. We investigated the function of the CstF-77 protein in Xenopus oocytes, and show that CstF-77 has indeed a role in the cytoplasm. The Xenopus CstF-77 protein (X77K) localizes mainly to the nucleus, but also in punctuate cytoplasmic foci. We show that X77K resides in a cytoplasmic complex with eIF4E, CPEB (cytoplasmic polyadenylation element-binding protein), CPSF-100 and XGLD2, but is not required for cytoplasmic polyadenylation per se. Impairment of X77K function in ovo leads to an acceleration of the G2/M transition, with a premature synthesis of Mos and AuroraA proteins. However, the kinetic of Mos mRNA polyadenylation is not modified. Furthermore, X77K represses mRNA translation in vitro. These results suggest that X77K could be involved in masking of mRNA prior to polyadenylation.

the cleavage reaction requires two sequences, the highly conserved hexanucleotide AAUAAA located 30 -20 nucleotides upstream of the cleavage site, and the G/U-rich sequence that lies downstream of the 3Ј-cleavage site (DSE for downstream element). These sequences are bound by different multisubunit factors. The cleavage and polyadenylation specificity factor (CPSF) 3 binds to the AAUAAA sequence via the CPSF-160 subunit, and CstF recognizes the DSE element via the CstF-64 protein (4,5). CPSF and CstF are known to interact cooperatively at 3Ј-end signals (6), and allow pre-mRNA cleavage with the cleavage factors I and II (CFI and CFII), the C-terminal domain of RNA polymerase II and the poly(A) polymerase (PAP). CPSF consists of the 160K, 100K, 73K, 30K, and hFip proteins (7)(8)(9), and CstF is an heterotrimeric structure composed of the 77K, 64K, and 50K proteins (10,11). CstF-77 is the subunit required for integrity of the CstF complex. CstF-77 binds to CstF-64, CstF-50 (12), and CPSF-160 (5) and therefore helps to hold together the much larger complete polyadenylation complex.
CPSF is also involved in cytoplasmic polyadenylation (13,14), a mechanism of controlling translation which is often used when transcription is quiescent, such as in oocytes or early embryos. In Xenopus oocytes, several dormant mRNAs are stored with short poly(A) tails, which are elongated when oocytes are stimulated by progesterone, provoking their translation.
Cytoplasmic polyadenylation in frog oocytes also requires two sequences on the 3Ј-UTR of target mRNAs, recognized by multiple protein components (reviewed in Ref. 15). CPEB directly binds to the CPE (cytoplasmic polyadenylation element) located upstream of the AAUAAA sequence, which is recognized by CPSF, as in nuclear polyadenylation (14). CPEB interacts with CPSF, which may help to stabilize its association with the AAUAAA element and a cytoplasmic poly(A) polymerase. GLD2, a poly(A) polymerase which is structurally distinguishable from nuclear PAP, was first identified in Caenorhabditis elegans (16). The Xenopus-related protein (XGLD2) was shown to bind to CPEB and CPSF and to participate in cytoplasmic polyadenylation (17,18). Sequences other than the CPE, the PRE sequences (polyadenylation response element), regulate early cytoplasmic polyadenylation of specific mRNAs including Mos (19,20). The CPEB protein has been shown to repress mRNA translation in immature oocytes and to direct cytoplasmic polyadenylation and translational activation in maturing oocytes. Indeed, mRNA translation repression is controlled by a complex composed of CPEB, Maskin, and eIF4E. Maskin interacts with CPEB and eIF4E and prevents eIF4G binding to eIF4E. After stimulation, CPEB is phosphorylated and cytoplasmic polyadenylation ensues, allowing disruption of Maskin-eIF4E binding and recruitment of eIF4G (21,22).
The yeast Saccharomyces cerevisiae and Drosophila melanogaster homologues of CstF-77 are, respectively, the Rna14 and Su(f) proteins. Several genetic and biochemical studies have shown a role for Rna14 in 3Ј-end processing (23)(24)(25), although other studies for Rna14 have suggested a function independent of nuclear polyadenylation. Indeed, rna14 mutants can be separated into two classes: the poly(A)-negative class, which contains mutants deficient in mRNA 3Ј-processing, and the poly(A)-positive class, which includes mutants that are not impaired in nuclear polyadenylation (26,27). Rna14 and Su(f) localize both to the nucleus and the cytoplasm. In yeast, this cytoplasmic localization is mainly in mitochondria (26,28). However, no mitochondrial function for Rna14 has been described so far. In D. melanogaster, the Su(f) protein has been implicated in cell cycle progression (29). In Xenopus and human cells, studies have shown that CstF-77, CstF-64, and CPSF-100 are concentrated in nuclear domains in Cajal or coiled bodies (30 -33) as well as various components required for transcription and processing of the three classes of nuclear transcript (reviewed in Ref. 34). However, nothing was reported concerning a possible cytoplasmic localization of the CstF subunits.
The aim of our work was to study the cytoplasmic function of CstF-77, bearing in mind possible roles independent of nuclear polyadenylation that could be related to the cell cycle. We therefore addressed the question of whether CstF-77 could have a role in cytoplasmic polyadenylation. The Xenopus oocyte is an appropriate model to study other functions of CstF-77, as there is neither transcription nor nuclear polyadenylation during oocyte maturation (15). In this report, we identify the Xenopus CstF-77 protein (X77K) and show that it is partially localized in the cytoplasm. We demonstrate that X77K belongs to a cytoplasmic complex with eIF4E, CPEB, CPSF-100, and XGLD2. However, the protein is not required for cytoplasmic polyadenylation per se. Inhibition of X77K function in oocytes accelerates meiotic maturation and precociously induces Mos and AuroraA protein synthesis, without modifying the kinetic of Mos mRNA polyadenylation. Moreover, X77K inhibits in vitro mRNA translation in a dose-dependent manner. These results suggest that X77K could have a function in mRNA masking prior to cytoplasmic polyadenylation.

EXPERIMENTAL PROCEDURES
X77K Cloning and Protein Alignment-A Xenopus leavis oocyte cDNA library in gt11 from Clontech (ZL5000b) was screened with a 1.030-base pair (bp) probe obtained by RT-PCR from Xenopus RNA with the following degenerated oligonucleotides (Eurogentec): 5Ј-ATGGCYMARGCWTAYGAYT-TYGCACT-3Ј and 5Ј-ACKCKSGTRTTRTTRTCYTCRTT- Several clones containing the cDNA of the Xenopus CstF-77 homologue (X77K) were isolated and sequenced. The largest cDNA (about 2.300 bp) was subcloned at the EcoR1 site of the pBluescript II KS. It contains an open reading frame encoding a protein of 719 amino acids. This X77K amino acid sequence was aligned with those of Homo sapiens CstF-77 (NP001317), Mus musculus CstF-77 (AAH03241), D. melanogaster Su(f ) (P25991), C. elegans CstF-77 (AAA62311) and Saccharomyces cerevisiae RNA14 (NP013777) with the ClustalW program. The X77K sequence was submitted to GenBank TM with the following accession number: AM071387.
Xenopus Oocytes and Embryos-Stage VI oocytes were selected after surgical removal of ovaries from mature female X. laevis and treatment with collagenase (Sigma) at 1 mg/ml in OR-2 (5 mM Hepes pH 7.2, 82.5 mM NaCl, 2 mM KCl, 1 mM MgCl 2 ). All further manipulations of oocytes were performed in modified Ringer's Solution (MMR: 5 mM Hepes pH 7.8, 100 mM NaCl, 2 mM KCl, 1 mM MgSO 4 , 0.1 mM EDTA, 2 mM CaCl 2 ). Progesterone was used at a final concentration of 1 mM. For protein oocyte extracts, oocytes were homogenized at 5 l per oocyte in lysis buffer (20 mM Tris, pH 7.5, 50 mM NaCl, 50 mM NaF, 10 mM ␤-glycerophosphate, 5 mM Na 4 P 2 O 7 , 1 mM Na 3 VO 4 , 1 mM EDTA, 1 mM EGTA, 0.1 mM phenylmethylsulfonyl fluoride, plus 10 g/ml each of leupeptin, chymostatin, pepstatin, and aprotinin) at 4°C, centrifuged at 13,000 rpm for 3 min at 4°C, and the supernatant was used for further analysis. For manually enucleated oocytes, nucleus were collected and boiled in 5ϫ Laëmmli buffer, and the cytoplasm was homogenized at 5 l per cytoplasm in lysis buffer. In vitro fertilization and embryo cultivation was carried out as described (35). The quantification of proteins in oocyte and embryo extracts was done by Bio-Rad protein assay, with quantified BSA as standard. For microinjections, the usual injected volume for antibodies, purified protein, or RNA was 20 -40 nl per oocyte and the number of injected oocytes is 35 for each condition.
Antibodies and Western Blots-The anti-pTpY-MAPK was obtained from New England Biolabs (9106S). The anti-eIF4E was provided by S. Morley or was obtained from mouse antisera using the same peptide antigen. The anti-CPSF-100 and the anti-Pak5 were provided by E. Wahle and N. Morin's laboratories, respectively. The anti-␤ tubulin and anti-HA antibodies were obtained from E7 (Iowa hybridoma bank) and from the 12CA5 hybridomas, respectively. The anti-RPA and anti-Au-roraA were provided by J. M. Lemaître and C. Prigent, respectively. The anti-Mos, anti-ERK and IgG were obtained from Santa Cruz Biotechnology (sc-086, sc-94 and sc-2027, respectively). Other antibodies used against GST and Xenopus 77K, CPEB and GLD2 are rabbit polyclonal antisera and were affinity-purified. The 77K Cter antibody is directed against the 17 amino acids C-terminal of the human 77K protein: NH 2 -VPPVHDIYRARQQKRIR-COOH that was kindly provided by David Bentley, and the 573 antibody is directed against the following X77K peptide: NH 2 -LKDDVDRKPEYPKPD-COOH. The anti-CPEB antibody is directed against GST-CPEB fusion protein as described (36). The anti-XGLD2 antibody is directed against the GST-XGLD2 2/3 C-terminal fusion protein. For microinjections, dialyzed antibodies were used at 1 g/l. Western blots were probed with primary antibody at 50 ng/ml and the appropriate secondary antibody horseradish peroxidase conjugate diluted according to manufacturer's recommendations (Amersham Biosciences) and revealed by ECL (Amersham Biosciences).
RNA and Recombinant Proteins-The Mos 3Ј-UTR construct was obtained by inserting the SpeI/XbaI fragment from the pT7-G UK XMos plasmid kindly provided by N. Sagata (37) into SpeI/XbaI-digested pCS2 (Dave Tumer, University of Michigan). This 320-nucleotide fragment contains Mos PRE, CPE, and AAUAAA sequences. Mos 3Ј-UTR RNA were prepared by linearizing the pCS2-Mos320 with XbaI and carrying out transcription reaction according to Papin and Smith (38). The pCSH vector corresponds to the pCS2 vector with two HA tags inserted in the multiple cloning site. Capped mRNA encoding HA-X77K, HA-XGLD2 and HA-CPEB were prepared by cutting the pCSH-X77K ORF, pCSH-XGLD2 ORF, and pCSH-CPEB ORF, respectively with NotI and carrying out transcription reaction.
Immunoprecipitation-Protein oocyte extracts corresponding to 20 oocytes were performed in extract buffer (100 mM KCl, 0.1 mM CaCL 2 , 1 mM MgCl 2 , 10 mM Hepes pH 7.6, 50 mM sucrose, plus 10 g/ml each of leupeptin, chymostatin, pepstatin, and aprotinin). For some experiments, the oocyte extracts were treated 10 min at room temperature with RNase A at 0,02 mg/ml. Oocyte extracts were incubated with 200 ng of antibodies over night at 4°C on wheel, supplemented with 20 l of protein A-Sepharose beads and mixed for another 45 min at 4°C on wheel. The immunoprecipitates were then washed three times in XRB A (0.2% Triton X-100, 10 mM Tris, pH 8, 150 mM NaCl, and 2 mM EDTA), three times in XRB B (0.2% Triton X-100, 10 mM Tris pH 8, 500 mM NaCl and 2 mM EDTA) and two times in 10 mM Tris, pH 8, eluted by boiling in 2ϫ Laëmmli buffer, separated by SDS-PAGE and analyzed by Western blotting.
For immunoprecipitation of radiolabeled X77K, 1 g of 77K antibody was incubated or not with 10 g of its antigen (77K Cter peptide coupled with thyroglobulin) before being applied to protein A-Sepharose beads for 30 min at 4°C. The beads were then washed three times with PBS, supplemented with [ 35 S]methionine X77K translated in rabbit reticulocyte lysates, and mixed 1 h on ice. The immunoprecipitates were then washed three times in XRB A, three times in XRB B, and two times in 10 mM Tris, pH 8, eluted by boiling in 2ϫ Laëmmli buffer, separated by SDS-PAGE, and analyzed by autoradiography.
Cap Column and GST Pull-down Assay-The cap column assay was performed as described (22). Briefly, 30 oocytes were homogenized in homogenization buffer (HB), and treated or not 10 min at room temperature with RNase A (0,02 mg/ml). The oocyte supernatants were supplemented with 0.1 mM GTP or 5 mM 7m GTP and applied in batch to a 7m GTP-Sepharose (cap) column (Amersham Biosciences) that had been equilibrated with HB plus bovine serum albumin (BSA, 0.1 mg/ml). Following mixing for 1 h at 4°C in Eppendorf tube, the cap column was extensively washed (100 bed volumes) with 0.1 M KCl, 50 mM Tris, pH 7.5. The bound material was then eluted with 5 mM 7m GTP on column. The eluates and a fraction of oocyte lysates collected before and after mixing with the cap column were separated by SDS-PAGE and analyzed by Western blotting.
The GST pull-down assay was performed as described (39). Briefly, HA-tagged CPEB and HA-tagged XGLD2 were in vitro translated in rabbit reticulocyte lysates (Promega) in the presence of unlabeled amino acids. GST-X77K expressed in Baculovirus and GST expressed in E. coli were purified with glutathione Sepharose (Amersham Biosciences). 5 g of purified GST-X77K or GST were bound to 20 l of glutathione beads in binding buffer. The beads were subsequently incubated 2h at room temperature in 400 l of binding buffer plus 5 g of the HA-CPEB or HA-XGLD2. The beads were then extensively washed with binding buffer and boiled in 2ϫ Laëmmli buffer. The eluates and inputs were separated by SDS-PAGE and analyzed by Western blotting.
For in vitro polyadenylation assays, egg extracts were prepared according to Murray and Kirschner (41). Egg extracts were adjusted to 1 mM ATP, 1 mM MgCl 2 , 7.5 mM creatine phosphate and 0.1 mM EGTA, pH 7.7. 30 ng of Mos 3Ј-UTR RNA was added to the egg extract and the polyadenylation assay was carried out at 23°C for 1 h, prior to RNA extraction as described in McGrew and Richter (42). Mos 3Ј-UTR polyadenylation analysis was carried out as described above (PAT assay). For depletion experiments, 2 g of antibodies were incubated 30 min at 4°C with 15 l of Dynabead protein A beads (Dynal). Beads were washed three times with PBS plus 100 g/ml BSA and incubated with 15 l of egg extract for 30 min at room temperature. After precipitation, depleted extract was collected, and used for polyadenylation assay and Western blot analysis. Bead precipitates were eluted by boiling in 2ϫ Laëmmli buffer and analyzed by Western blotting.
In Vitro Translation in Rabbit Reticulocyte Lysate (RRL)-The luciferase translation assay was performed according to the manufacturer's recommendations (Promega). 100 ng of Luciferase mRNA (provided in RRL kit) were added to 10 l of RRL in the presence of [ 35 S]methionine (2 l, 1000 Ci/mmol, methionine/cysteine mix, Amersham Biosciences) with or without the indicated quantities of GST-77K or GST proteins, for 1.5 h at 30°C. The equivalent of 20% of the reaction was used for autoradiography and Western blot analysis. The amount of proteins added to the reaction was analyzed by Coomassie staining. The luciferase RNA was analyzed by RT-PCR after the reaction. Reverse transcription was done with random primers and the amplification with the following oligonucleotides: 5Ј-AAGCCACCATGGAAGA-3Ј and 5Ј-CTCTAGAATTA-CACGGCGATC-3Ј.
Cell Culture and Immunofluorescence-Mouse NIH 3T3 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum in a 37°C incubator with 5% CO 2 . Xenopus XTC cells were cultured in 70% L15 Leibovitch's Media supplemented with 10% fetal bovine serum in hermetic tissue culture flask at 25°C. Cultured cells were fixed using 4% paraformaldehyde (in PBS) for 15 min at room temperature. The cells were then permeabilized in PBS containing 1% Triton-X100 for 10 min at room temperature and blocked in PBS containing 0.1% Tween and 3% BSA (blocking reagent). Antibody incubations were done 1 h at room temperature in blocking reagent followed by secondary Alexa-546-labeled goat anti-rabbit IgG (Molecular Probes) used in identical conditions. DNA was visualized by DAPI staining.

Identification of Xenopus CstF-77
Homologue-A X. laevis oocyte library was screened with a probe made by RT-PCR from Xenopus oocyte RNA and degenerate oligonucleotides. A cDNA encoding the Xenopus homologue of CstF-77 was isolated. This cDNA contains an ORF encoding a protein of 719 amino acids, with a predicted molecular mass of 83 kDa that we named X77K. We identified a domain organization similar to the known CstF-77 homologues, with 11 HAT (half a TPR) repeats, a bipartite nuclear localization signal (NLS) and a proline-rich domain at the C terminus (Fig. 1). Apart from the NLS, these domains are known to mediate protein-protein interactions, and the HAT repeats have been described to be specific to a family of proteins involved in RNA processing (11,43). For the human CstF-77 protein, the proline-rich domain has been shown to be necessary both for binding other CstF subunits and also for self association (12). It is noteworthy that the CstF-77 proteins are highly conserved, since the amino acid sequence of X77K shares 94% identity overall with the human and mouse CstF-77 proteins, 57% with D. melanogaster protein and 27% with S. cerevisiae protein (Fig. S1).
CstF-77 Is Localized Both in the Nucleus and Cytoplasm-To study the expression of X77K protein, we raised a polyclonal antibody against the human C-terminal peptide (77K antibody). By Western blotting, this antibody recognizes the HA-tagged X77K expressed in oocytes, and the purified Baculovirus GST-X77K protein ( Fig. 2A, lanes  2 and 5). It also detects endogenous Xenopus CstF-77 protein from oocytes or cell lysates ( Fig. 2A, lanes 1 and 3, respectively) and endogenous mouse (lane 4) and human protein (not shown). Moreover, the in vitro translated X77K protein is immunoprecipitated by the 77K antibody but not by the 77K antibody preincubated with the peptide antigen (Fig. 2B). These results show the specificity of the 77K antibody for the 77K protein.
A Western blot analysis on oocyte and embryo lysates shows that X77K is expressed at a constant level during Xenopus oogenesis and embryogenesis up to stage 40 (Fig. 3, A and B). In S. cerevisiae and D. melanogaster, the CstF-77 homologues Rna14 and Su(f) respectively localize both to the nucleus and cytoplasm (28,44). In Xenopus oocytes, the nuclear localization of X77K has been well determined, where it is present in Cajal bodies and lampbrush chromosomes (32). However, nothing has been reported concerning a possible cytoplasmic localization. To investigate whether X77K could be cytoplasmic in Xenopus laevis, lysates from nucleus, cytoplasm or total oocytes were analyzed by Western blotting with 77K antibody. Fig. 3C shows that X77K is mainly nuclear, but is also present in the cytoplasm. We verified that cytoplasmic extracts were not contaminated with nuclear extracts by probing the membrane for the nuclear protein RPA (replication protein A). This double localization was confirmed by immunofluorescence experi- ments on Xenopus XTC and mouse NIH3T3 cell lines (Fig. 3D): CstF-77 protein is mainly in the nucleus but is also localized in punctuated bodies distributed in the whole cytoplasm. These results show that the CstF-77 protein is localized both in the nucleus and cytoplasm in vertebrates, as in S. cerevisiae and Drosophila.
As X77K has a functional NLS but partially localizes to the cytoplasm, we investigated whether the protein shuttles between the GV (germinal vesicle) and the cytoplasm, even if it does not present a NES (nuclear export signal). To answer this question, oocytes were treated or not overnight with leptomycin B, a nuclear export inhibitor. They were then enucleated, to analyze whether the protein accumulated in the nucleus in the presence of leptomycin B. In this condition, no change in the relative levels of X77K between the nucleus and the cytoplasm was observed, showing that X77K does not shuttle between the two compartments (Fig. 3E). Surprisingly, the CPEB protein was sensitive to leptomycin B and accumulated in the nucleus (Fig. 3E). This result validates the leptomycin B treatment and implies that CPEB shuttles between the cytoplasm and the nucleus. This observation was not expected since CPEB was not detected in the nucleus under normal conditions (Ref. 35 and Fig. 3E) and suggests new functions for CPEB in the nucleus, or in nucleo-cytoplasmic transport.
X77K Is Part of a Cytoplasmic Complex with CPEB, eIF4E, CPSF-100, and XGLD2-As we asked whether X77K could have a function in cytoplasmic polyadenylation, we checked for X77K partners in the cytoplasm. In Xenopus oocytes, CPEB plays a key role in translational repression in immature oocytes, and in directing cytoplasmic polyadenylation during oocyte maturation. We investigated whether X77K could belong to a complex, which includes CPEB. We performed co-immunoprecipitations from total stage VI oocyte extracts, treated or not with RNase A, with antibodies directed against different proteins known to be in a cytoplasmic complex with CPEB. In this experiment, X77K, CPEB, eIF4E, and CPSF-100 proteins were co-immunoprecipitated with 77K, eIF4E, and CPEB antibodies but not with nonspecific IgG (Fig.  4A). As a negative control we used an antibody directed against the Pak5 (p21-activated kinase) protein. This antibody did not immunoprecipitate any of the proteins tested. Similarly, Pak5 protein was not immunoprecipitated by the 77K, eIF4E, or CPEB antibodies. These results show that X77K specifically interacts with eIF4E, CPEB, and CPSF-100 independently of the mRNA.
To show that cytoplasmic X77K is involved in this complex, we enucleated oocytes and performed coimmunoprecipitation experiments from oocyte cytoplasm extracts treated with RNase A. The X77K, CPEB, and CPSF-100 proteins were co-immunoprecipitated by X77K, eIF4E, and CPEB antibodies (Fig. 4B). These results show that cytoplasmic X77K belongs to a complex with CPEB, eIF4E, and CPSF.
It has been reported that XGLD2, a divergent poly(A) polymerase, directly binds to CPEB, even in stage VI oocytes (18). We asked whether XGLD2 was also in the complex with X77K in stage VI oocytes. As XGLD2 is expressed at a low level in oocytes (18), immunoprecipitations were performed from oocytes overexpressing XGLD2. HA-tagged XGLD2 mRNA was injected into oocytes and, after 14 h of expression, the oocyte extracts were subjected to different antibody immunoprecipitations in the presence or absence of RNase A. In these experiments, HA-XGLD2, X77K, eIF4E, and CPEB proteins were co-immunoprecipitated with XGLD2, X77K, eIF4E, and CPEB antibodies (Fig. 4C), showing that XGLD2 is also part of the complex. It is interesting to note that two forms of eIF4E were immunoprecipitated by the eIF4E antibody but that only the lower form is mainly co-immunoprecipitated (Fig.  4, A and C).
To better characterize the complex and investigate whether the complex exists in metaphase II oocytes, we used another approach. Cap columns ( 7m GTP chromatography), which allow eIF4E recruitment, were used to identify proteins present in the complex. Stage VI and metaphase II extracts, treated (Fig.  4D, lanes 6 and 7) or not (lanes 2 and 3) with RNase, were supplemented with GTP to reduce nonspecific adsorption. As a negative control, some extracts were supplemented with free cap ( 7m GTP) (Fig. 4D, lanes 4 and 5). The extracts were then applied on a cap column and the bound material was eluted with free cap and detected by Western blot. A fraction of both X77K and CPEB from stage VI and metaphase II oocytes with or without RNase was retained on the column in the presence of GTP but not when 7m GTP was added in the extracts (Fig. 4D). This experiment was done using GTP at a lower concentration than 7m GTP in the extracts. To confirm the specificity of the binding, a control experiment was done either, using the same concentration of GTP and 7m GTP in the extracts, or, by eluting the bound proteins with GTP or 7m GTP (supplemental Fig. S2). We observed that GTP did not compete for eIF4E, 77K, and CPEB binding while 7m GTP did (supplemental Fig. S2, compare lane 6 with lane 8) and that GTP only weakly eluted bound proteins compare with 7m GTP (compare lane 4 with lane 2). A small amount of total X77K was retained on the resin (Figs. 4D and S2), which is coherent with the fact that only a small fraction of X77K is cytoplasmic and can bind to eIF4E. It is noteworthy that CPEB was also poorly retained on the resin (Figs. 4D and S2), even though it is abundant in the cytoplasm. However, this result could be explained by the fact that CPEB, known to be part of the masking complex, interacts only transiently with eIF4E (20). In metaphase II oocytes, the electro-phoretic mobility of CPEB is shifted and the protein is partially degraded (45). However, the same proportion of total CPEB was retained on the column in stage VI and metaphase II extracts. As negative controls, we looked for the binding of ERK and ␤-tubulin proteins, and observed, as expected, that they were not retained on the column (Fig. 4D). We failed to detect endogenous XGLD2 on the resin, probably because of its low expression level.
To assess whether the interactions were direct, we performed GST pull-down experiments with purified GST-X77K from Baculovirus and either HA-CPEB or HA-XGLD2 proteins translated in vitro. Fig. 4E shows that GST-X77K bound to HA-CPEB and HA-XGLD2, whereas GST alone did not. In the same kind of experiment, a direct interaction between GST-X77K and HA-eIF4E was not conclusive, because of a high background in the GST control (data not shown). Altogether, these results show that X77K belongs to a RNAindependent cytoplasmic complex with eIF4E, CPEB, CPSF-100, and XGLD2, and that X77K is capable of directly interacting with CPEB and XGLD2.
X77K Is Not Required for Cytoplasmic Polyadenylation-We showed that X77K binds to proteins required for cytoplasmic polyadenylation in oocytes. To address whether X77K is required for cytoplasmic polyadenylation, we performed in vitro polyadenylation tests in egg extracts that were subjected to 77K, or control IgG immunodepletions. As a control, we also did the experiment on egg extracts depleted for CPEB, a protein that is known to be required for cytoplasmic polyadenylation. Fig. 5A shows that immunodepletions of X77K and CPEB completely removed the endogenous proteins (left panel), which were retained on the antibody beads (right panel). Moreover, even if CPEB was apparently not depleted by 77K antibody, and vice versa, we observed a co-immunoprecipitation between X77K and CPEB in egg extracts (Fig. 5A, right panel) that confirmed our results obtained with oocyte lysates (Fig. 4).
To do the polyadenylation test, Mos 3Ј-UTR RNA, bearing wild-type CPE and PRE elements, was incubated for 1 h with egg extracts or X77K, CPEB, or mock-depleted extracts. The total RNA was isolated at the beginning of the reaction and The ␤-tubulin, consistently expressed throughout oocyte maturation and embryogenesis, serves as a loading control. C, oocytes were manually enucleated and pools of nucleus and cytoplasms were collected separately. Proteins extracts were analyzed by Western blotting with 77K and RPA antibodies. The equivalent of one oocyte (StVI), one nucleus (N) or one cytoplasm (Cyto) was loaded. The RPA is exclusively expressed in nucleus and serves as a control of enucleation. D, immunofluorescence of NIH3T3 and XTC cells. Cells were fixed, stained with 77K antibody and with DAPI and observed by fluorescence microscopy. A nuclear and cytoplasmic expression of endogenous 77K is observed in both cell lines. E, X77K localization is not sensitive to leptomycin B. Oocytes were treated or not overnight with leptomycin B (100 nM final) and were manually enucleated. Pools of nucleus and cytoplasms were collected separately. Protein extracts were analyzed by Western blotting with 77K, CPEB, RPA, and ␤-tubulin antibodies. The equivalent of one oocyte (StVI), one cytoplasm (Cyt), or 0.5 nucleus (N) was loaded. The RPA and ␤-tubulin proteins are respectively expressed in nucleus and cytoplasm and serves as enucleation and loading controls. one hour later. Then, Mos 3Ј-UTR polyadenylation was checked by PAT (polyadenylation test) assay, to measure the poly(A) tail length. Fig. 5B shows that Mos 3Ј-UTR RNA was polyadenylated in control and X77K depleted extracts, while it was not polyadenylated in CPEB depleted extracts (com-pare lanes 2, 4, 6 with lane 8). These results show that depletion of X77K does not impair cytoplasmic polyadenylation in vitro.
To verify that this is also the case in vivo, we performed a polyadenylation test in oocytes. Oocytes were first injected with Mos 3Ј-UTR RNA. Some of them were subsequently injected with 77K, CPEB, or nonspecific IgG. 1-h later, the oocytes were induced, or not, to mature with progesterone, and harvested when 30% of the control oocytes have undergone GVBD (germinal vesicle break down). The total RNA was isolated and scored for Mos mRNA polyadenylation. In control oocytes, Mos 3Ј-UTR was polyadenylated only when the oocytes were submitted to progesterone (Fig. 5C compare lane 2 with  lane 3). The change in electrophoretic mobility observed was caused by the lengthening of the poly(A) tail, since the addition of oligo(dT) and RNaseH before the PAT assay suppressed the mobility shift (Fig. 5C, right panel). As in egg extracts, the impairment of X77K function did not modify the Mos 3Ј-UTR polyadenylation in the oocytes (Fig. 5C, compare lane 6  with lanes 3 and 8). In CPEB antibody injected oocytes, Mos 3Ј-UTR was polyadenylated to a lesser extent than in control oocytes (compare lane 10 with lanes 3 and 8). This short poly(A) tail of Mos 3Ј-UTR observed was probably due to a PRE-dependent polyadenylation. Indeed it has been shown that Mos 3Ј-UTR could be precociously polyadenylated independently of CPE/CPEB, but the length of the poly(A) tail was shorter than in the presence of CPEB (20). The above experiment shows that X77K is not required for polyadenylation of exogenous Mos mRNA.
To confirm this result, we performed a polyadenylation test on endogenous mRNAs from oocytes injected with IgG, 77K or CPEB antibodies, or non-injected (Fig. 5D). Cyclin B1 and AuroraA mRNAs were polyadenylated in control and 77K antibody-injected oocytes, which is in agreement with the result obtained with exogenous Mos mRNA. In CPEB antibody-injected oocytes, there is no polyadenylation of cyclinB1 mRNA that has FIGURE 4. X77K is part of a complex with eIF4E, CPEB, CPSF, and XGLD2. A, StVI oocyte extracts, treated or not with RNase A as indicated, were immunoprecipitated with 77K, eIF4E, CPEB, Pak5 antibodies, or with control IgG. The immunoprecipitates were analyzed by Western blotting with 77K, CPEB, eIF4E, CPSF-100, and Pak5 antibodies. B, oocytes were manually enucleated and cytoplasmic extracts were treated with RNase A and immunoprecipitated with 77K, eIF4E, CPEB, or control IgG. The immunoprecipitates were analyzed by Western blotting with 77K, CPEB, and CPSF-100 antibodies. C, oocytes were injected with mRNA encoding HA-tagged XGLD2 and, after 14 h of HA-XGLD2 expression, protein extracts were treated or not with RNase A and immunoprecipitated with different antibodies as indicated. The immunoprecipitates were analyzed by Western blotting with HA, 77K, eIF4E, and CPEB antibodies. The equivalent of 1 oocyte (A and C) and one cytoplasm (B) was loaded as input. D, cap column assay. Protein extracts from StVI oocytes or mature oocytes (MII) treated or not with RNase A were supplemented with GTP or 7m GTP and applied to 7m GTP-Sepharose beads (cap column). After beads washes, the bound proteins were eluted with 7m GTP. All of the eluted proteins (B, bound), the equivalent 1/10 of StVI oocytes or mature oocytes extracts (input St VI or input M II), one-eighth of the supernatant of protein extracts after cap column binding (U, unbound) were analyzed by Western blotting with eIF4E, 77K, CPEB, ERK, and ␤-tubulin antibodies. E, GST pull-down assay. 5 g of purified GST-X77K or GST were bound to glutathione beads that were subsequently incubated with in vitro translated HA-tagged CPEB (left panel) or HA-tagged XGLD2 (right panel). The bound proteins were boiled in SDS sample buffer. The eluates and the equivalent of one-tenth of the in vitro translated proteins (input) were analyzed by Western blotting with HA and GST antibodies.
no PRE sequence, while we observed a short poly(A) tail on AuroraA mRNA which bears a PRE element (19). The polyadenylation observed with our PAT assay is a specific reaction, as endogenous actin mRNA, which is not subjected to polyadenylation during oocyte maturation, is not polyadenylated in our test (Fig. 5D). Furthermore, we observed a weak deadenylation of actin mRNA from 6 h of progesterone addition, showing that the injected antibodies did not impair the deadenylation reaction during meiotic maturation. Altogether these results obtained in vitro and in ovo show that X77K is not required for cytoplasmic polyadenylation.
77K Antibody Injected into Oocytes Accelerates the G 2 /M Transition-During the experiments of polyadenylation test in oocytes, we observed that X77K antibody injection accelerates oocyte maturation. The antibodies were injected into and maintained in the cytoplasm (Fig. 6A, insert), arguing that 77K antibody affected the function of the cytoplasmic X77K protein. Moreover, as maturation is accompanied with a cessation of transcription (15), an effect on the nuclear polyadenylation could be ruled out. We repeated the experiment several times and the 77K antibody injected oocytes always underwent 50% GVBD 1 h or more before the control oocytes (Fig. 6A), depending on the batch of oocytes. This was also the case when we injected another antibody directed against X77K (the 573 antibody) (data not shown). In agreement with these results, this acceleration was accompanied by a premature synthesis of Mos protein and activation of MAP kinase (Fig.  6B). In 77K antibody injected oocytes, Mos protein is well detected 2 h after progesterone induction, while in control oocytes (uninjected or IgG injected) it is clearly detected only 5 h after progesterone addition. The MAPK activation, revealed by the bi-phosphorylation of ERK, is also earlier in 77K antibody-injected oocytes (2 h) than in control oocytes (4 h). To investigate whether downstream events of maturation were accelerated, we checked for proteins where the synthesis is induced at GVBD. In 77K antibody injected oocytes, AuroraA synthesis increased 2 h after progesterone induction while, in control oocytes (IgG injected and uninjected oocytes), this increase is well observed at 5 h (Fig. 6B). This acceleration of protein synthesis also occurred for cyclin B1 (data not shown).
One supposition for this premature protein synthesis was that the impairment of X77K function in oocytes accelerates mRNA cytoplasmic polyadenylation. To test this hypothesis, we performed a time course of Mos 3Ј-UTR polyadenylation in 77K antibody-injected oocytes. The oocytes were first injected with Mos 3Ј-UTR RNA and then with 77K antibody or IgG. Oocytes were harvested at different times after progesterone induction, total RNA was extracted and checked for Mos mRNA polyadenylation. In control oocytes and 77K antibody injected oocytes, Mos mRNA polyadenylation occurred at the same time, showing that injection of 77K antibody does not modify the kinetics of cytoplasmic polyadenylation (Fig. 6C). CPEB is known to belong to a masking complex with eIF4E and FIGURE 6. Injection of 77K antibody accelerates oocyte maturation. A, oocytes were injected with 77K antibody or nonspecific IgG or left uninjected and maturation was induced by progesterone 1-h later. The percentage of GVBD was scored at the indicated times and graphed. For each treatment, 35 oocytes were injected. This graph is representative of six experiments. Inset, 5 h after 77K antibody injection, oocytes were manually enucleated and pools of cytoplasms and nucleus were collected separately. Proteins extracts were analyzed by Western blotting with 77K and RPA antibodies. The equivalent of one uninjected oocyte (St VI), one cytoplasm (cyto) and one nucleus (N) from injected oocytes was loaded. The RPA serves as a control of enucleation. An exclusive cytoplasmic localization of 77K IgG is observed. B, Western blot analysis of Mos and AuroraA synthesis, activated MAPK (ppERK) and ␤-tubulin in oocytes collected during an experiment similar to the one depicted in A. The ␤-tubulin serves as a loading control. C, oocytes were first injected with Mos 3Ј-UTR RNA and 30-min later were injected with 77K antibody or nonspecific IgG. 1-h later, maturation was induced with progesterone. Total RNA was extracted from pools of 5 oocytes collected at the indicated times and was submitted to Mos polyadenylation analysis. Fragment size is indicated on the left in base pair (bp).
Maskin. It has been reported that mRNA cytoplasmic polyadenylation allows the disruption of Maskin/eIF4E interaction by eIF4G recruitment, and translation ensues (22). This suggests that the role of cytoplasmic polyadenylation is to allow the binding of eIF4G to eIF4E. In 77K antibody-injected oocytes, Mos, AuroraA, and cyclin B1 proteins were precociously synthesized, while Mos mRNA polyadenylation was not accelerated. This phenotype, with the fact that X77K belongs to a complex with CPEB and eIF4E, suggests that X77K could have a role in mRNA masking before cytoplasmic polyadenylation.
X77K Represses mRNA Translation in Vitro-The above results document that X77K functions in vivo as part of a complex involved in mRNA masking, thereby preventing a premature translation of protein that must be synthesized at a precise time. To test a potential function of the X77K protein in mRNA translation inhibition, we performed an in vitro translation experiment. In this assay, we added increasing quantities of purified GST-77K or GST proteins in rabbit reticulocyte lysates programmed with a luciferase reporter, and checked for mRNA translation by measuring and quantifying [ 35 S]methionine incorporation. Fig. 7 shows that luciferase mRNA translation was inhibited by addition of recombinant X77K in a dose-dependent manner, while no translation inhibition was observed with the GST protein. Importantly, this effect is not because of destabilization of luciferase mRNA by the addition of the recombinant proteins. This result is in agreement with a role for X77K in mRNA masking.

DISCUSSION
In this study, we identified the Xenopus CstF-77 protein (X77K) and showed that the CstF-77 proteins are very well conserved in vertebrates (Figs. 1 and S1). X77K is expressed in oocytes and embryos at a constant level, and is localized mainly to the nucleus, but also in discrete cytoplasmic foci in vertebrate cells (Fig. 3). However, X77K does not shuttle between the two compartments (Fig. 3E). The cytoplasmic localization, together with other arguments (see Introduction), suggested a cytoplasmic role for CstF-77 protein. We established that X77K is part of a cytoplasmic complex with eIF4E, CPEB, CPSF, and XGLD2 ( Fig. 4) but is not required for cytoplasmic polyadenylation per se (Fig. 5). Nevertheless, impairment of X77K function in oocytes, accelerates oocyte maturation and protein synthesis without modifying the kinetic of Mos mRNA polyadenylation (Fig. 6). Moreover, X77K represses mRNA translation in vitro (Fig. 7). Altogether, these results support a role for X77K in mRNA masking.
The known function of nuclear CstF-77 protein is to enhance cleavage by stabilizing the interaction between CPSF and AAUAAA in the precleavage complexes (46). We envisage that cytoplasmic X77K could have the same function of complex stabilization on cytoplasmic mRNA. This could explain why cytoplasmic X77K is not required for cytoplasmic polyadenylation per se, because its nuclear counterpart is required for pre-mRNA cleavage but not for the subsequent poly(A) addition step of the reaction. In the nucleus, CstF-77 interacts with CPSF via its CPSF-160 subunit (5). In the cytoplasm, X77K also binds to CPSF and CPEB (Fig. 4B). Moreover, X77K directly interacts with XGLD2 and CPEB in vitro, showing that X77K, CPSF, CPEB, and XGLD2 are part of the same cytoplasmic complex. CstF-77 has no RNA binding domain, thus, one or several binding partners must recruit X77K to increase the affinity of the polyadenylation complexes (nuclear or cytoplasmic) on the target RNA. In the nucleus, CstF-64 is the RNAbinding protein that targets, with CPSF-160, CstF-77 on the pre-mRNA. In the cytoplasm, CPEB could replace CstF-64 for this function. Moreover, CstF-77 has a functional NLS, leading to its intranuclear targeting (31), 4 and does not shuttle between the nucleus and the cytoplasm (Fig. 3E), which could explain the low level of its cytoplasmic counterpart. This suggests that CstF-77 can stay into the cytoplasm only in interaction with other proteins, and that a fraction of X77K newly synthesized can be recruited by CPEB, XGLD2, or CPSF-160 and probably other proteins (see below).
CstF-77 is not the only protein that belongs to both nuclear and cytoplasmic polyadenylation complexes: besides CPSF, shown to be involved in cytoplasmic polyadenylation (14), Symplekin was recently identified as part of the cytoplasmic polyadenylation machinery (18). Symplekin is a dual location protein that has been localized to the cytoplasmic plaques of tight junctions, and in the nucleus where it interacts with CPSF, 4 C. Rouget and E. Mandart, unpublished data. CstF-77, and CstF-64 (12,47). However, Symplekin does not interact with the fully assembled CstF, suggesting that Symplekin and CstF-77 can compete for the same site in CstF-64 (12). CPEB binds both X77K (this study) and Symplekin (18), strengthening the idea that CPEB could have the same function in the cytoplasm than CstF-64 in the nucleus.
A cytoplasmic localization for the other CstF subunits has not been described in vertebrates, and the S. cerevisiae homologue RNA15 is only nuclear (28). However, we cannot rule out a possible cytoplasmic function for the CstF-64 or CsF-50 subunits, as gametes express different protein isoforms from those present in somatic cells. Indeed, a second form of CstF-64 protein (tau CstF-64), expressed in meiotic and post-meiotic male germ cells, has been described (48). Searches for specific CstF-64 or CstF-50 isoforms in oocytes, and their possible functions in the cytoplasm, have not been tested, because of a lack of purified antibodies recognizing the corresponding Xenopus proteins.
We observed an acceleration of the G 2 /M transition in 77K antibody-injected oocytes submitted to progesterone. This phenotype is probably because of premature Mos protein synthesis, that activates the MAPK pathway and MPF (maturation promoting factor). However, the Mos mRNA polyadenylation kinetic is not accelerated, suggesting that impairment of X77K function allows Mos mRNA recruitment on the ribosome independently of its polyadenylation state. It has been suggested that the function of cytoplasmic polyadenylation was to displace eIF4E from a masking complex (22). One possibility is that 77K antibody, by interacting with the X77K protein, disrupts the masking complex, allowing the binding of eIF4E to eIF4G, and the recruitment of the small ribosomal subunit by the mRNA. Thus, X77K could play a role in mRNA masking prior to cytoplasmic polyadenylation.
However, as X77K represses CPE-independent mRNA translation (Fig. 7), it probably has a more general function in mRNA translation inhibition. In that case, X77K would be targeted on the mRNA by one or several proteins other than CPEB. CPSF can recruit X77K on the mRNA by itself. Nevertheless, other RNA-binding proteins, in cooperation with CPSF, could also address X77K on the messenger RNA. Xp54, a ubiquitous translational repressor that also interacts with CPEB (49), could be one of them. XPumilio (50,51) and XDAZL (52), other translational regulators independent of CPE-dependent cytoplasmic polyadenylation, are also potential candidates. Thus, CPSF and an X77K-containing complex could interact cooperatively to stabilize a RNA masking complex, until it is disrupted or modified by a signal allowing mRNA translation.
CstF-77 protein is localized in discrete cytoplasmic foci in NIH3T3 and XTC cells (Fig. 3D). In oocytes, Xp54 and p47, the Xenopus and Spisula homologues of the yeast decapping activator Dhh1, are localized in stored mRNP and bind to CPEB (49,53,54). Moreover, it has been reported that human CPEB1 expression induces the assembly of stress granules, that in turn, recruit the processing bodies (PB) (55), and that eIF4E is localized in PB (56,57). Thus, proteins initially involved in mRNA translation can be localized in PB and, inversely, decapping activator can be stored in mRNP in oocytes. Recently, Brengues et al. (58) have shown that mRNA molecules move from PB to polysomes and vice versa. These data show that stress granules, PB and stored mRNP are dynamically linked sites of mRNP remodeling.
The fact that: 1) X77K belongs to a complex with eIF4E, CPEB, and XGLD2, 2) the impairment of 77K function leads to a premature protein synthesis in oocytes, 3) recombinant X77K represses mRNA translation in vitro, and 4) we observed CstF-77 in NIH3T3 and XTC cytoplasmic foci, suggest that cytoplasmic CstF-77 could have a ubiquitous function in mRNP remodeling events, for the transition of mRNA from inactive to active translation.