CPEB Interacts with an Ovary-specific eIF4E and 4E-T in Early Xenopus Oocytes*

CPEB (cytoplasmic polyadenylation element-binding protein) is an important regulator of translation in oocytes and neurons. Although previous studies of CPEB in late Xenopus oocytes involve the eIF4E-binding protein maskin as the key factor for the repression of maternal mRNA, a second mechanism must exist, since maskin is absent earlier in oogenesis. Using co-immunoprecipitation and gel filtration assays, we show that CPEB specifically interacts, via protein/protein interactions, with the RNA helicase Xp54, the RNA-binding proteins P100(Pat1) and RAP55, the eIF4E-binding protein 4E-T, and an eIF4E protein. Remarkably, these CPEB complex proteins have been characterized, in one or more organism, as P-body, maternal, or neuronal granule components. We do not detect interactions with eIF4E1a, the canonical cap-binding factor, eIF4G, or eIF4A or with proteins expressed late in oogenesis, including maskin, PARN, and 4E-BP1. The eIF4E protein was identified as eIF4E1b, a close homolog of eIF4E1a, whose expression is restricted to oocytes and early embryos. Although eIF4E1b possesses all residues required for cap and eIF4G binding, it binds m7GTP weakly, and in pull-down assays, rather than binding eIF4G, it binds 4E-T, in a manner independent of the consensus eIF4E-binding site, YSKEELL. Wild type and Y-A mutant 4E-T (which binds eIF4E1b but not eIF4E1a), when tethered to a reporter mRNA, represses its translation in a cap-dependent manner, and injection of eIF4E1b antibody accelerates meiotic maturation. Altogether, our data suggest that CPEB, partnered with several highly conserved RNA-binding partners, inhibits protein synthesis in oocytes using a novel pairing of 4E-T and eIF4E1b.

Selective protein synthesis in oocytes, eggs, and early embryos of many organisms drives several critical aspects of early development, including meiotic maturation and entry into mitosis, establishment of embryonic axes, and cell fate determination. Protein synthesis is usually regulated at the initiation stage, mediated by the 5Ј m 7 GpppN mRNA cap structure bound by the translation initiation complex eIF4F, com-posed of eIF4E, the cap-binding protein, the RNA helicase eIF4A, and the large scaffold protein eIF4G, which has a consensus binding site YXXXXL for eIF4E, and additional sites for eIF3 and the poly(A)-binding protein. eIF3 recruits the small ribosomal subunit, whereas the eIF4E-eIF4G-poly(A)-binding protein relay results in the so-called "closed loop" model, responsible for the synergistic enhancement of translation by capped and polyadenylated mRNAs. A hallmark of translational control mechanisms is the role of mRNA-binding proteins, which recognize specific (and usually) 3Ј-UTR 2 cis-elements and influence the recruitment of the small ribosomal subunit to the 5Ј cap (reviewed in Refs. [1][2][3]. Probably, the best studied mRNA-binding protein is CPEB1 (cytoplasmic polyadenylation-binding protein 1), characterized in flies, worms, clams, Aplysia, Xenopus, and mammals, which in its conserved C terminus contains two tandem RNA recognition motif domains followed by two zinc finger domains, responsible for binding 3Ј-UTR cytoplasmic polyadenylation elements, consensus U 4 -6 A 1-3 U (reviewed in Refs. [1][2][3]. The related proteins CPEB2-CPEB4, found in mammals, have different RNA-binding preferences and function in neurons (4,5).
CPEB1 (hereafter CPEB) in clams and Xenopus performs a dual role; it represses cap-dependent translation in the oocyte and activates translation, via cytoplasmic polyadenylation, in meiotically maturing eggs and early embryos (6,7). Activated maternal mRNAs contain one or more cytoplasmic polyadenylation elements near the nuclear polyadenylation hexanucleotide, AAUAAA, whereas mRNAs lacking cytoplasmic polyadenylation elements are deadenylated upon meiotic maturation and concomitantly exit from polysomes. Unusually, deadenylated mRNAs are stable in eggs and early embryos, reflecting the absence of decapping activity. In Xenopus, cytoplasmic polyadenylation elements and hexanucleotide elements mediate poly(A) length control by a complex, including CPEB and the recently identified cytoplasmic poly(A) polymerase, GLD-2, as well as cleavage and polyadenylation specificity factor, symplekin, and the deadenylase PARN (8 -12). The complex between CPEB and cleavage and polyadenylation specificity factor/ GLD-2 is stabilized in response to progesterone-stimulated signaling during meiotic maturation, which leads to translational activation (12,13), and additional elements, including the polyadenylation response element and its trans-acting factor Musashi, control temporal aspects of translational activation of CPE-containing mRNAs (14). When CPEB functions as a repressor, on the other hand, it has been reported to interact with maskin (TACC3), an eIF4E-binding protein, an interaction that precludes the productive binding of eIF4E to eIF4G (15). Maskin is phosphorylated in response to progesterone, a modification that releases eIF4E (16). This paradigm example of a repressed "closed loop" form of mRNA may, however, be confined to late stage Xenopus laevis oocytes, since maskin is absent early in oogenesis (17,18) (this study), and its nonconsensus eIF4E-binding site is not conserved among vertebrates (2,15). Nevertheless, maskin serves as a prototype of a protein that bridges an RNA-binding protein bound to the 3Ј-UTR and eIF4E bound to the cap to inhibit translation initiation. Other examples include two regulators of posterior patterning in the Drosophila oocyte and embryo, Bruno and Smaug, which repress oskar and nanos mRNAs respectively, and the eIF4Ebinding protein Cup (19 -22). Alternatively, a repressive "closed loop" form of mRNA may form directly, without an intermediary protein such as maskin or Cup, using a homolog of the canonical eIF4E1a (class I) initiation factor. Thus caudal and hunchback mRNAs, which establish opposing morphogen gradients in Drosophila embryos, are repressed by Bicoid and the Nos-Pum-Brat NRE complex, respectively, in conjunction with d4E-HP, a class II eIF4E, which binds the cap but not eIF4G (23)(24)(25)(26). Intriguingly, recent studies indicate that 4E-HP has a considerably lower affinity for the cap than eIF4E1a (27,28). Given that Drosophila possesses seven genes encoding eight eIF4E-related proteins (26) and that vertebrates contain 4 -6 such proteins, spread among three distinct classes (25,29), which vary in their abilities to interact with the cap, eIF4G, and the 4E-BP family of proteins (which regulate general translation by competing for eIF4G using the shared eIF4E-binding consensus YXXXXL) (25,26), it seems likely that additional noncanonical eIF4E proteins will be found that regulate translation.
Previously, we showed that in clam and Xenopus oocytes, CPEB interacts with a DDX6 RNA helicase called Xp54 and that Xp54 tethered to the 3Ј-UTR of a reporter RNA represses its translation (30,31). A role in translation repression has also been reported for other members of this highly conserved helicase family, including Saccharomyces cerevisiae Dhh1 (32), Plasmodium DOZI (33), Drosophila Me31B (34), and mammalian RCK/p54 (35) (reviewed in Refs. 1 and 36). p54 helicases are found in P (processing)-bodies, distinct cytoplasmic foci that are sites of (reversible) RNA storage and of RNA decay in yeast and in mammalian cells and are composed of mRNA and factors mediating both RNA degradation and translational repression (37)(38)(39)(40)(41)(42), as well as in maternal and neuronal granules in flies and worms (43)(44)(45).
Here we report an extensive analysis of proteins that interact with CPEB in Xenopus oocytes, enabled by a recently described monoclonal CPEB antibody (40). Co-immunoprecipitation and gel filtration analyses show that in early stage oocytes, CPEB interacts with Xp54, P100 (S. cerevisiae Pat1, Drosophila melanogaster HPat), RAP55B (S. cerevisiae Scd6, D. melanogaster Trailer Hitch, Caenorhabditis elegans CAR-1, and Lsm14), eIF4E-Transporter (4E-T, D. melanogaster Cup), and an eIF4E protein, all P-body components, in an RNA-independent manner. The eIF4E protein was identified as eIF4E1b, a close homolog of the canonical eIF4E1a cap-binding protein. Although eIF4E1b possesses all residues known to be required for capand eIF4G-binding, it binds m 7 GTP weakly, and rather than binding eIF4G, it binds 4E-T, in a YSKEELL-independent manner. Wild type and Y-A mutant 4E-T tethered to a reporter mRNA represses its translation in a cap-dependent manner, and injection of eIF4E1b antibody accelerates meiotic maturation. Our data point to a conserved translational complex operating in early Xenopus oocytes, which represses cap-dependent translation not by severing the eIF4E1a-eIF4G link but by using an alternative eIF4E-binding protein and an alternative eIF4E protein.

EXPERIMENTAL PROCEDURES
Plasmid Constructs-All plasmids, including IMAGE expressed sequence tag clones, subcloning, and mutagenesis are described in the supplemental materials.
Immunoprecipitation-Immunoprecipitation experiments using lysates prepared from oocytes injected with MS2 fusion protein mRNAs with mouse monoclonal MS2 antibodies and protein G-Sepharose were described previously (31). Immunoprecipitation of endogenous proteins was performed in a similar manner, using mouse monoclonal CPEB antibodies (40) or rabbit eIF4E1b antibody, 40 l of stage I-V lysate prepared in 1 ml of NET buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.5% Nonidet P-40, 1 mM EDTA, pH 8.0, 0.25% gelatin, 0.02% NaN 3 ) was incubated, with gentle mixing, with 1 l of CPEB antibody, 0.2 l control MS2 antibody, or 0.5 l of eIF4E1b/preimmune control antibodies for 2 h at 4°C prior to the addition of 10 l of protein G-Sepharose (mouse monoclonal antibodies) or protein A-Sepharose beads (rabbit polyclonal antibodies) for 2 h at 4°C. Beads were washed in NET buffer, and bound proteins were eluted in 20 l of protein sample buffer. Ten l of each sample was separated by SDS-PAGE, and proteins were detected by silver staining or Western blot. When RNase A-treated, Xenopus extracts were supplemented with RNase A to 20 pg/l extract and incubated at 20°C for 20 min prior to clarification by centrifugation at 10,000 ϫ g for 10 min at 4°C. Immunoprecipitation reactions for the isolation of proteins for peptide sequencing were performed essentially as described above by scaling up reaction volumes 10 -25-fold. RNA was extracted from scaled up immunoprecipitation reactions and aliquots of whole cell lysate as follows. Bound beads or extracts were incubated in 500 l of TNES (0.1 M Tris, pH 7.5, 0.3 M NaCl, 5 mM EDTA, 2% SDS) with the addition of 200 g/ml proteinase K at 50°C for 30 min with intermittent vortexing. RNA was purified by phenol/chloroform extraction followed by ethanol precipitation in the presence of tRNA and was resuspended in 12.5 l of H 2 O.
Reverse Transcription-PCR-Reverse transcription-PCRs were performed using 1 l of purified RNA and 0.5 l of random hexamers (Promega) in a 20-l reaction volume using Superscript II (Invitrogen) and following the manufacturer's instructions. PCRs were performed using 1 l of cDNA template in a 20-l reaction volume with the following oligonucleotide pairs: actin (1:2) and cyclin B1 (3:4; see supplemental Table  1) with Taq DNA polymerase.
Mass Spectrometry Analysis-Proteins within the gel-excised bands were sequenced by MS/MS in the Cambridge Centre for Proteomics, as described previously (31). Fragmentation data were used to search the National Center for Biotechnology Information data base using the MASCOT search engine (available on the World Wide Web). Probability-based MASCOT scores were used to evaluate identifications. Only matches with p Ͻ 0.05 for random occurrence were considered significant. 4E-T peptides were YDSDGVWDPEK, ATGR, and VISVDE-LEYR, and the P100(Pat1) peptides were WTDTVFLVAK, LSEEEFLGER, EEEPEALQPVK, VSTYATGQILEDK, AIDA-VSYAMPDEAIK, and VFLMFLEVEELAR; the eIF4E1a/b peptides are shown in Fig. 3.
Preparation of Antibodies-The P100(Pat1) antibody was raised against an N-terminal peptide DQESDEEPVKLEDD[C], and the eIF4E1b antibody was raised against an N-terminal peptide LSREKLDNEKRRKK[C] (Sigma). 4E-T antibody was raised against recombinant Xenopus 4E-T (Eurogentec). The X. laevis 4E-T expression plasmid is described in the supplemental materials. A culture of transformed BL21-CodonPlus-RIL cells was induced at 37°C using an Overnight Express autoinduction system (Novagen) following the manufacturer's instructions. Cells were harvested and resuspended in nitrilotriacetic acid buffer (300 mM NaCl, 1% (v/v) Triton-X-100, 50 mM sodium phosphate buffer, pH 7.8). Following French pressing, pellets were resuspended in nitrilotriacetic acid buffer, and aliquots were run on SDS-PAGE preparative gels. Coomassie-stained 4E-T containing gel slices were excised and used for immunization. Antiserum was subsequently purified against 10 g of 4E-T recombinant protein isolated by SDS-PAGE and Western blot analysis.
Protein Gel Electrophoresis and Western Blot Analysis-Protein samples were separated by SDS-PAGE on either 10 or 15% polyacrylamide gels and then used for Coomassie Blue staining, silver staining, or Western blot analysis using ECL (31,46). We used mouse monoclonal CPEB (1:5000 (40)), guinea pig FRGY2 (1:10,000 (47)), goat TIAR (1:2000, C-18; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), and goat Rsk1/p90 rsk (1:1000, C-21-G; Santa Cruz Biotechnology) antibodies and the following rabbit antibodies raised against Xenopus CPEB (1:12,000) (48) Complete EDTA-free protease inhibitors (Roche Applied Science)). The columns were incubated at 4°C for 1 h with agitation and then washed with HKE buffer. Proteins were eluted from the resin with 0.1 mM GTP and then with 70 M m 7 GpppG in HKE buffer and finally with 1ϫ SB (SDS). For peptide sequencing, 1 ml of stage III/IV lysate was applied to 100 l of cap-Sepharose beads in a 15-ml final volume. Columns were processed as above. Bound fractions were subjected to SDS-PAGE and Coomassie Blue staining, and the appropriate bands were excised for sequencing.
For binding of in vitro translated eIF4E1a or eIF4E1b, 100 l of translation mix was applied to 10 l of cap-Sepharose beads equilibrated with HKE buffer. After binding and elutions as above, the samples were analyzed by SDS-PAGE and autoradiography.
GST Binding Assays-GST-eIF4E1a-and GST-eIF4E1b-carrying plasmids were amplified in Escherichia coli and transformed into expression-competent BL21 CodonPlus-RIL cells. Proteins were expressed as described for recombinant X4E-T and purified from the insoluble fraction essentially following the protocol described by Joshi et al. (25). Control GST proteins were expressed as described above but were present in the soluble fraction following French pressure cell lysis and centrifugation.
GST proteins were purified by affinity chromatography with glutathione-Sepharose (GE Healthcare). 1-ml aliquots of beads were incubated with the GST or GST fusion proteins for 1 h at 4°C with gentle mixing. Beads were washed with buffer B (50 mM HEPES, pH 7.2, 100 mM NaCl, 1 mM EDTA, 2 mM dithiothreitol, 2 mM benzamidine, 1% Triton X-100) and finally in buffer C (50 mM Tris, pH 8.0, 100 mM NaCl, 2 mM dithiothreitol, 2 mM benzamidine) followed by elution in 1 ml of buffer C supplemented with 10 mM glutathione.
Three g of GST or 6 g GST-eIF4E1a or eIF4E1b was bound to 25 l of glutathione-Sepharose for 1 h at 4°C in buffer BB (20 mM Tris, pH 8.0, 200 mM NaCl, 1 mM EDTA, and 0.5% (v/v) Nonidet P-40). 50 l of reticulocyte lysate containing in vitro translated [ 35 S]Met-labeled protein was added and allowed to bind for 2 h at 25°C. Beads were washed in BB, and bound proteins were eluted in SDS-containing buffer before separation by SDS-PAGE and autoradiography.
Xenopus Oocyte Antibody Injection and Maturation-eIF4E1b and preimmune serum were purified by binding to protein A-Sepharose in 100 mM Tris (pH 8.0) for 1 h at 20°C. Beads were washed with 100 mM Tris (pH 8.0) and then with 10 mM Tris (pH 8.0) prior to elution with 100 mM glycine (pH 3.0) and dialysis against phosphate-buffered saline. Pools of 120 stage VI oocytes were injected with 50 nl of purified antibody (1 g/l IgG). Oocytes were incubated at 20°C in MBS with progesterone added to 0.1 g/ml as required and scored for germinal vesicle breakdown (GVBD) every 2 h. Random samples of 10 oocytes were taken at each time point for SDS-PAGE and Western blot analysis.

Characterization of the CPEB RNP Complex in Xenopus
Oocytes-Previously, we showed that MS2-tagged Xp54 helicase interacts with CPEB and eIF4E in oocytes, in the absence of RNA, and with endogenous Xp54, in an RNA-dependent manner (31) (Fig. 1A). To extend the identification of Xp54-binding proteins, further MS2 immunoprecipitations using lysates from stage VI oocytes expressing MS2-Xp54 mRNA were carried out. A prominent ϳ100-kDa protein that specifically interacted with MS2-Xp54 in an RNA-independent manner but not with MS2 protein (Fig. 1A) was sequenced. Two sets of peptides were obtained, both corresponding to ϳ100-kDa proteins. Three peptides corresponded to the Xenopus homolog of human eIF4E-Transporter (4E-T) (57), and seven peptides corresponded to the egg-specific protein P100 (58), a homolog of yeast Pat1 (32). Subsequently, the co-immunoprecipitation of 4E-T and P100(Pat1) with MS2-Xp54 was verified by Western blot analysis (Fig. 7F).
Since our Xp54 antibody did not precipitate efficiently, we used a mouse monoclonal antibody raised against human CPEB1 (40) to verify endogenous interactions. The mouse antibody detects a single band in Xenopus oocytes, corresponding in size to the 58-kDa CPEB (59) (Fig. 1B). The CPEB antibody, alongside a control mouse monoclonal antibody and a bead only control, was used to immunoprecipitate CPEB from mixed stage oocyte lysates. Prior to immunoprecipitation, the lysates were untreated or were digested with RNase that was sufficient to degrade the endogenous RNA ( Fig. 1, C and D). Protein G-Sepharose-bound proteins were analyzed by silver staining, which showed the co-precipitation of several prominent and specific proteins with CPEB, most of them in an RNA-inde-FIGURE 1. CPEB co-immunoprecipitates with Xp54, Pat1, RAP55, 4E-T, and a 24-kDa eIF4E in an RNA-independent manner. A, oocytes were injected with MS2 and MS2-Xp54 mRNA, lysates were prepared after protein expression, and portions were treated with RNase. Treated and untreated lysates were immunoprecipitated with MS2 antibodies, and the bead-bound proteins were analyzed by SDS-PAGE and silver staining. The thick arrow points to the ϳ100-kDa proteins sequenced, and thin arrows indicate MS2-Xp54 and endogenous Xp54. B, the human CPEB mouse monoclonal antibody recognizes CPEB specifically in Xenopus laevis (X.l.) oocytes. C, mixed stage oocyte lysate was treated (ϩ) or not (Ϫ) with RNase prior to immunoprecipitation with CPEB antibody (C), control monoclonal antibody (M), or beads only (Ϫ). Protein G-Sepharose-bound material was analyzed by SDS-PAGE and silver staining. D, aliquots of RNase-treated and control lysate were analyzed by agarose-gel electrophoresis and EtBr staining. E, protein G-Sepharose-bound material, as in B, was phenolextracted, and the recovered RNA was amplified by semiquantitative PCR, using actin and cyclin B1 mRNA primers. Total oocyte RNA was also amplified with these primers. F, two cell equivalents of staged oocytes (stages I-VI) and of progesterone-matured eggs (E) were analyzed by Western blot analysis with the indicated antibodies. G, mixed stage oocyte lysate, treated or not with RNase, was immunoprecipitated with CPEB antibody (C), control monoclonal antibody (M), or beads only (Ϫ). Immunoprecipitated samples were analyzed by Western blot as indicated, alongside 1 20 of input lysate (Total). IP, immunoprecipitation; mAb, monoclonal antibody. pendent manner (Fig. 1C). We also examined the mRNAs precipitated by CPEB and control antibodies by semiquantitative reverse transcription-PCR, using primers corresponding to actin and cyclin B1 mRNA. Although both transcripts were maternally expressed, the CPE-containing cyclin B1 mRNA was enriched in CPEB-bound material relative to actin CPElacking mRNA (Fig. 1E). These data confirm that the monoclonal antibody immunoprecipitates a specific CPEB RNP complex in Xenopus oocytes.
We then investigated several known and potential partners of CPEB by Western blot analysis. Each antibody was also used to probe a Western blot containing two cell equivalents of each of the six stages of oogenesis (I-VI), and of progesterone-matured eggs (E) to monitor the expression of each protein during oogenesis, and upon meiotic maturation (Fig. 1, F and G). Strong interactions, corresponding to ϳ5-10% of input, were observed in the case of Xp54, Pat1, 4E-T, and eIF4E, supporting data obtained with MS2-tagged Xp54 (31) (Figs. 1A and 7). An additional strong (Ͼ10% of input) interaction was observed between CPEB and RAP55B, detected by an antibody generated against the Drosophila Trailer Hitch (Tral) Sm domain that has been shown to recognize the maternal mouse paralog (54). (We note that both Xenopus RAP55A (ϳ62 kDa) and RAP55B proteins (ϳ54 kDa) (36) are detected by this antibody in oocytes and that the levels of RAP55A increase during oogenesis as reported previously (60), whereas those of RAP55B, which uniquely co-immunoprecipitates with CPEB (Fig. 1G), decline during this period.) These interactions did not depend on RNA, except in the case of Xp54, whose binding to CPEB was partially reduced by RNase treatment. This reduction is probably due to RNA-mediated Xp54 oligomerization (31). A second, apparently entirely, RNA-dependent and slightly weaker (ϳ2-3% of input) interaction was noted between CPEB and FRGY2 (Fig. 1G).
To examine further the significance of 4E-T binding to CPEB, we also probed Western blots with other eIF4E-binding protein antibodies but failed to observe significant binding of eIF4G or 4E-BP. Moreover, eIF4A, a component of eIF4F, similarly failed to efficiently co-precipitate with CPEB, attesting to the specificity of the interactions described above. We also probed with antibodies raised against Xenopus maskin (TACC3), obtained from three laboratories (15,51,52), and the deadenylase PARN (53) but did not observe significant binding (Fig. 1G) (see "Discussion").
All examined proteins were expressed during oogenesis, and with the exception of maskin, PARN and 4E-BP were present at high levels in early stages (Fig. 1F). In particular, eIF4E appears as three different sized proteins, of about 24, 26, and 30 kDa, as detected using a rabbit polyclonal antibody raised against the C-terminal peptide of human eIF4E, TATKSGSTTKNRFVV (49) (Fig. 3), conserved among vertebrate Class I eIF4E proteins (25,29). These proteins are differentially expressed during oogenesis, with the smallest protein being most abundant in early oogenesis and the two larger proteins only detectable from middle to late oogenesis and in eggs (Fig. 1F). Significantly, the smallest eIF4E protein was preferentially found in CPEB immunoprecipitates (Figs. 1 (F and G) and 4B), an observation also made recently using the same eIF4E antibody and an independent, polyclonal CPEB antibody (61). We conclude that in oocytes CPEB interacts with Xp54, Pat1, RAP55B, a 24-kDa eIF4E protein, and only one known eIF4E-binding protein, 4E-T. These CPEB partners as well as CPEB itself are more abundant in early stage oocytes in contrast to maskin and PARN, which accumulate late in oogenesis (Fig. 1F) (17,18,53) and may thus represent an early oocyte complex.
Supporting evidence for such an early CPEB RNP complex comes from Superose 6 HR 10/30 FPLC gel filtration of stage I/II oocyte lysates, in which Xp54, Pat1, RAP55B, FRGY2, 4E-T, and eIF4E 24 kDa all co-purified with CPEB in large ϳ3-MDa complexes ( Fig. 2A). With regard to RNA-binding proteins, we note that not all, as shown by TIA-1/TIAR, co-fractionate with CPEB ( Fig. 2A; also see Ref. 46). Strikingly, only the smallest 24-kDa eIF4E protein (and not the two larger forms) co-purifies with CPEB ( Fig. 2A), mirroring the immunoprecipitation data (Fig. 1G). Note too that 4E-T, but neither eIF4G, eIF4A, nor RAP55A, co-fractionates with CPEB, also in line with the immunoprecipitation data (Figs. 1G and 2A). Treating the lysate with RNase prior to gel filtration does not affect the copurification of Pat1, RAP55B, 4E-T, and eIF4E 24 kDa with CPEB, whereas Xp54 is partially released, and FRGY2 is completely released from the CPEB complex, as expected from the coimmunoprecipitation assays (Figs. 1G and 2B). Altogether, these data provide strong biochemical evidence that in early stage oocytes, CPEB specifically interacts, via direct or indirect protein/protein interactions, with Xp54, Pat1, RAP55B, 4E-T, and eIF4E 24 kDa and in an RNA-dependent manner with FRGY2. We went on to characterize eIF4E 24 kDa and 4E-T and their role in translational repression in oocytes.

CPEB Interacts with eIF4E1b, Not the Canonical eIF4E1a Cap-binding Protein-To identify maternal
Xenopus eIF4E proteins, m 7 GTP-Sepharose affinity chromatography was performed using lysates prepared from stage III/IV oocytes, in which the three proteins (Fig. 1F) are roughly equivalent in abundance. Following the binding of the lysate proteins to m 7 GTP-Sepharose, the beads were washed several times and then consecutively eluted with GTP-and cap analog-containing buffer, and any proteins remaining bound to the beads were removed with SDS-containing buffer. The samples were separated by SDS-PAGE, and one gel was stained with silver, whereas a duplicate gel was analyzed by Western blot to reveal the binding of the eIF4E proteins (Fig. 3, A and B). The silver-stained gel shows three prominent bands ranging from 20 to 30 kDa in the cap-analog elutions, corresponding in size to the eIF4E proteins detected in Western blots (Fig. 1). Comparing the load and flow-through fractions (Fig. 3B) shows that, although the two larger proteins quantitatively bound to the cap-Sepharose beads, the fastest migrating form bound more weakly, whereas the GTP and cap elutions show that all three proteins bound preferentially to m 7 GpppG (Fig.  3A). Mass spectrometry sequencing identified the 30-kDa capbinding protein as X. laevis eIF4E, with seven matching peptides (Fig. 3C). In particular, the N-terminal peptide ETGQEIENTNPQSTEEEK indicated the protein to be the long isoform of eIF4E, with a duplicated insert of 18 amino acids, presumably due to alternative splicing (62). With nine matching peptides, the 26 kDa band was identified as the eIF4E isoform that lacks the 18-amino acid insert. Single amino acid substitutions in some peptides indicated the existence of two allelic variants of this protein (Fig. 3C). Both the short and the long eIF4E proteins have been previously identified as members of the class I eIF4E group and have been assigned to the subgroup eIF4E1a (29). The least abundant, 24-kDa protein was assigned as eIF4E1b based on the sequences of four of its five peptides, which distinguish eIF4E1b from eIF4E1a, with the remaining peptide being common to both proteins (Fig. 3C) (29). eIF4E1b, a class I eIF4E, has been characterized in zebrafish oocytes (63) and is conserved in mammals (29,64). eIF4E1b proteins are highly related to the canonical cap-binding translation initiation factor; e.g. over the length of the whole  autoradiography (B, bottom). C, alignment of X. laevis eIF4E1a(L), eIF4E1a(S) (two allelic variants), and eIF4E1b. *, residues involved in cap binding; ϩ, residues involved in eIF4G binding (65,89). Peptides obtained by mass spectrometry sequencing of the indicated bands in A are shaded. The conserved C-terminal peptide used to generate eIF4E1 antibodies is boxed.
To verify the difference in cap-binding between Xenopus eIF4E1a and eIF4E1b, their mRNAs were translated in rabbit reticulocyte lysate, and the products were analyzed by m 7 GTP-Sepharose chromatography (Fig. 3B). eIF4E1b binding to m 7 GTP was considerably weaker than that of eIF4E1a, when comparing both the amount bound (load and flow-through fractions) and the amount eluted with cap analog (m 7 GpppG). These experiments confirm the initial mass spectrometry data and indicate that Xenopus eIF4E1b, in contrast to eIF4E1a, binds m 7 GTP only weakly (Fig. 3, A and B), like its zebrafish counterpart (63).
Taking advantage of the unique basic N terminus of eIF4E1b, an antibody was raised against the peptide LSREKLD-NEKRRKK (Fig. 3C). The specificity of this antibody was verified by Western blot analysis of GST-eIF4E1a, GST-eIF4E1b, GST, and total oocyte lysate proteins. The eIF4E1b antibody specifically recognizes GST-eIF4E1b and the smallest, 24-kDa, eIF4E protein detected by the C-terminal peptide eIF4E1 antibody (Fig. 4A). Use of this eIF4E1b antibody confirms the identification of eIF4E1b as the eIF4E1 protein that co-immunoprecipitates with CPEB (Fig. 4B). Furthermore, in a reciprocal experiment, eIF4E1b antibody, but not preimmune antibody, immunoprecipitates CPEB in an RNA-independent manner (Fig. 4C). Altogether, these experiments provide evidence that CPEB interacts with eIF4E1b, rather than eIF4E1a, in line with the gel filtration data (Fig. 2).
XeIF4E1b Is Cytoplasmic and Its Expression Is Limited to Oocytes and Early Embryos-A distinguishing feature of eIF4E1b proteins is the presence of several tandem basic residues in their N termini, reminiscent of proteins that undergo nuclear import (29,64) (Fig. 3C). To ascertain to what extent eIF4E1b is cytoplasmic, Xenopus oocytes were fractionated into nuclear and cytoplasmic fractions. All three eIF4E1 proteins were exclusively cytoplasmic (Fig. 5A), with CBP80 attesting to the content of the nuclear fraction.
The levels of eIF4E1b, per oocyte, slowly decline during oogenesis, whereas the expression of eIF4E1a(L) and eIF4E1a(S) increases (Figs. 1F and S1). To examine their expression during embryogenesis, samples of equal numbers of staged embryos were analyzed by Western blot with the eIF4E1 antibody. The levels of both eIF4E1a(L) and eIF4E1a(S) remain approximately constant, at least until stage 42 (tadpole-like), whereas those of eIF4E1b decline to undetectable levels after stage 9, corresponding to midblastula (Fig. 5B). In the tested adult tissues, eIF4E1b is only detectable in the ovary (Fig. 5C). Xenopus eIF4E1b expression mirrors that of its zebrafish counterpart; D. rerio eIF4E1b mRNA is not detectable after the 10-somite stage, and the protein is largely restricted to the ovary, although low amounts are found in the testis and muscle (63). In mice, eIF4E1b (Eif4e1oo) mRNA is very abundant in the oocyte, in contrast to eIF4E1a mRNA, but undetectable at the late two-cell stage (64). Together, these data show  Fig. 7A for Coomassie Blue-stained recombinant proteins), and total oocyte lysate proteins (X.l.) were probed with the eIF4E1 antibody, which recognizes both eIF4E1a and eIF4E1b proteins (left) and the eIF4E1b peptidespecific antibody (right). B, immunoprecipitations (IP) with monoclonal CPEB antibody (C) or control monoclonal antibody (M) were carried out as in Fig. 1G and analyzed by Western blot with eIF4E1 antibody (left) or eIF4E1b-specific antibody (right). C, immunoprecipitations were carried out with preimmune (PI) or eIF4E1b peptide-specific antibody and mixed stage oocyte lysate (with or without prior RNase treatment) and analyzed by Western blot with the monoclonal CPEB antibody. FIGURE 5. eIF4E1b is cytoplasmic in oocytes, and its expression is confined to oocytes, eggs, and early embryos. A, whole stage VI oocyte lysates (T), cytoplasm (C), and nuclei (N and N3, 3 times the content of N) were analyzed by Western blot with eIF4E1 antibody. CBP80 antibody was used to assess the content of the nuclear fraction. B, two cell/embryo equivalents of two-cell embryos (2C) and stage 9 (midblastula), 12.5 (gastrula), 20 (neural fold closure), 25 (tail bud), and 42 (tadpole-like) embryos were analyzed by Western blot with the indicated antibodies. C, protein samples of the indicated adult tissues were analyzed by Western blot with the eIF4E1 antibody (top) and the eIF4E1b-specific antibody (bottom). that the expression of eIF4E1b is restricted to the oocyte, egg, and early embryo.
Characterization of Maternal Xenopus 4E-T-Human 4E-T was characterized as a large protein that binds eIF4E via an N-terminal peptide YTKEELL, very similar to eIF4E-binding sites in eIF4G and 4E-BP proteins (57). In HeLa cells, 4E-T is a largely cytoplasmic, shuttling protein, which can import eIF4E into nuclei via NLS and NES sequences (57). Xenopus 4E-T was characterized as described in detail in the legend to Fig. S2. Briefly, we noted the presence of two splice isoforms, differing in the inclusion or exclusion of a 546-nucleotide-long exon at codon 317. The X4E-T long isoform is 67% identical (79% similar) to human 4E-T and contains all three sequences that have been reported to promote nuclear import/export (57), whereas the short form lacks the N-terminal NES. Both the long and short forms of X4E-T contain a potential consensus eIF4Ebinding site YSKEELL at a very similar N-terminal location to the human protein. The X4E-T short isoform is more abundant in oocytes than the longer form, and both forms are cytoplasmic (Fig. S2). The full-length insert-lacking clone, used in most subsequent experiments, corresponded to NCBI accession number BC077338.
eIF4E1b Binds 4E-T but Not eIF4G in Vitro-To examine the binding of the eIF4E1 proteins to eIF4E-binding proteins, GSTtagged eIF4E1a and eIF4E1b proteins, as well as GST alone, were purified from E. coli, and pull downs were performed with equimolar amounts of recombinant proteins and in vitro translated eIF4G, 4E-T, and maskin (Fig. 6). In the case of 4E-T, the wild type protein was compared with a mutant protein, in which Tyr 28 alone or in combination with Leu 33 and Leu 34 residues of the eIF4E-binding site YSKEELL was mutated to alanine to abrogate eIF4E binding (57). As shown in Fig. 6B, eIF4E1a bound 4E-T and eIF4G with high affinity and specificity, and the binding to 4E-T was largely YSKEELL-dependent, as expected from previous studies (57), both single and triple mutations behaving the same (data not shown). In contrast, eIF4E1b showed significantly reduced binding to eIF4G, the same as background binding by GST, although it possessed all residues known to be necessary for eIF4G binding (Fig. 3C) (65). In similar assays, zebrafish eIF4E1b, in contrast to its eIF4E1a paralog, also did not bind eIF4G or 4E-BP (63). However, X. laevis eIF4E1b bound 4E-T to nearly the same extent as eIF4E1a and had approximately the same affinity for both wild type and mutant proteins, indicating that eIF4E1b probably binds 4E-T at a different site from eIF4E1a. Binding of the in vitro translated X4E-T long isoform to eIF4E1a and eIF4E1b proteins was indistinguishable from that of the X4E-T short isoform, demonstrating that the alternative 182-amino acid insert (Fig. S2) did not influence binding to eIF4E proteins, at least in vitro (data not shown). Last, we noted that neither eIF4E protein had very significant affinity in vitro for maskin, in line with the observation that in the yeast two-hybrid system, maskin and eIF4E interact weakly (15). We conclude that in vitro, Xenopus eIF4E1b binds preferentially to 4E-T, rather than eIF4G, in agreement with immunoprecipitation and gel filtration data ( Figs. 1 and 2). Moreover we note that this binding was insensitive to mutation of the eIF4E1a-binding site, YSKEELL, consistent with the absence of significant binding to eIF4G and 4E-BP1 (Fig. 6B) (63).
Tethered 4E-T Represses Translation-To assess its possible role in translational regulation, 4E-T was tethered to the 3Ј-UTR of firefly luciferase mRNA via MS2 binding sites, as previously described for Xp54 (30,31). mRNAs encoding MS2 fusion proteins were injected into stage VI oocytes first, followed by a second injection of (nonadenylated) firefly luciferase (Fluc) mRNA alongside an internal control Renilla mRNA. X4E-T was observed to repress firefly luciferase expression in four experiments ϳ4-fold (Figs. 7A and S3), in a manner requiring MS2 binding sites (Fig. 7B) and independently of similar changes in Fluc mRNA levels (Fig. 7D). The degree of repression by 4E-T was comparable with that seen with full-length Xp54, whereas a truncated version of Xp54 (Xp54N), comprising the N-terminal domain (residues 1-302) alone, was inactive (Fig. 7A).
To assess the dependence of the repression mechanism on the 5Ј cap structure, we first compared the level of translation of (nonadenylated) Fluc RNAs bearing m 7 GpppG or ApppG caps or the CSFV IRES (56). The IRES reporter RNA was capped with ApppG to avoid degradation. Translation of the CSFV IRES-led reporter RNA in oocytes was found to be as robust as that mediated by the m 7 GpppG cap, whereas A-capped RNA was very poorly translated (Fig. 7, A and C) (data not shown). Translational repression by 4E-T required the reporter RNA translation initiation to be cap-dependent, rather than mediated by the CSFV IRES, which does not require eIF4F or eIF3 (66).
Binding to MS2-tagged proteins expressed in oocytes was determined using a monoclonal MS2 antibody, as previously described (30,31). According to the silver-stained gel of immunoprecipitated samples, both MS2-Xp54 proteins are efficiently synthesized in oocytes, whereas the two 4E-T proteins are less well expressed (Fig. 7E). As shown in Fig. 7F, no significant binding was observed to the two negative controls, MS2 alone and MS2-Xp54N. MS2-Xp54 co-immunoprecipitated with Pat1 and 4E-T, in line with the previously described mass spectrometry peptide data ( Fig. 1A; see "Experimental Procedures"). Moreover, MS2-Xp54, like endogenous CPEB, co-immunoprecipitated with eIF4E1b rather than eIF4E1a. (Indeed, when we originally described the binding of eIF4E to MS2-Xp54 (31), using the same eIF4E antibody, we did not appreciate the heterogeneity of maternal eIF4E1 proteins and that only the smallest eIF4E, now identified as eIF4E1b, binds Xp54). In contrast, ectopically expressed 4E-T binds both eIF4E1a and eIF4E1b, unlike endogenous 4E-T, which is found in complexes and gel filtration fractions that only contain eIF4E1b (Figs. 1  and 2). Binding of eIF4E1a to MS2-4E-T could be eliminated by mutation of the eIF4E-binding site, Y28A (Fig. 7F). Such a mutation in MS2-4E-T only partially reduced binding of eIF4E1b, in line with the in vitro binding data (Fig. 6), and only partially alleviated its translational repression (Fig. 7A). Both in vitro and in vivo binding experiments (Figs. 6 and 7) indicate that eIF4E1b binds 4E-T at a separate site to eIF4E1a. Preliminary attempts to delineate this second site suggest that it also lies in the N terminus, corresponding to 4E-T (residues 1-380), which retains full binding of eIF4E1b and full translational repression (data not shown). We conclude that MS2-Xp54 and MS2-4E-T(Y-A) repress translation of reporter RNA while specifically interacting with eIF4E1b, rather than eIF4E1a. This finding is suggestive of a role for eIF4E1b as a co-repressor of cap-dependent translation, for which more direct evidence is given in Fig. 8.

Injection of eIF4E1b Antibody Accelerates
Meiotic Maturation-The addition of progesterone to stage VI oocytes, arrested in prophase of meiosis I, triggers their completion of meiosis I, marked by nuclear breakdown (or GVBD), and entry into meiosis II, where they arrest in metaphase awaiting fertilization. GVBD results from the translational activation of CPE-containing mRNAs, including c-mos mRNAs. Synthesis of c-mos, a MAPKKK, leads to the activation of MAPK, one of whose targets is p90 rsk (reviewed in Ref. 67). To assess eIF4E1b function in this process, protein A-Sepharose-purified antibody was injected into stage VI oocytes to neutralize protein function. Similarly purified preimmune serum was also injected into a separate batch of oocytes, and all were stimulated with progesterone to undergo meiotic maturation. In contrast to preimmune control antibody-injected oocytes, eIF4E1b antibody-injected oocytes showed accelerated maturation, as determined by timing of germinal vesicle breakdown (percentage of GVBD), MAPK/ERK kinase, and p90 rsk activation (Fig. 8).
Acceleration of meiotic maturation by injection of eIF4E1b was relatively modest although consistently noted in three separate experiments (data not shown). It was not detectable in the absence of progesterone, showing that eIF4E1b neutralization is not sufficient to cause GVBD and that other changes in the messenger RNP complex are required and are promoted by In each case, lysates from five pools of 5-10 oocytes each were prepared and assayed by the dual reporter luciferase system. In A-C, results are from one experiment; data in A were typical of four independent experiments, whereas data shown in B and C were typical of two independent experiments. D, the stability of 33 P-labeled Luc mRNAs was assessed 16 h after injection by agarose gel electrophoresis and autoradiography. E and F, immunoprecipitations were carried out with MS2 antibody, and lysates were prepared from oocytes injected with MS2 and MS2 fusion proteins. E, silver-stained gel of immunoprecipitations. *, MS2 fusion proteins. F, Western blots of immunoprecipitations with the indicated antibodies. A, in Drosophila, oskar mRNA is (in part) repressed by the binding of a 3Ј-UTR-binding protein Bruno to Cup (4E-T), which associates with eIF4E, excluding eIF4G. The helicase Me31B (Xp54) also participates in repression (5,6). C and D, 4E-HP, a weak cap-binding class II eIF4E, represses caudal and hunchback mRNAs by directly interacting with Bicoid (7,8), whereas Ago2, also a weak cap-binding protein, is responsible for repression of translation initiation by micro-RNAs (9). B, the repression of CPE-mRNAs in Xenopus oocytes by the eIF4E1b-4E-T-CPEB complex may rely upon the combination of conserved P-body components, alongside a weak cap-binding eIF4E protein, which can act as a co-repressor when tethered to the 3Ј-UTR.
progesterone signaling (Fig. 8). Acceleration of GVBD by antibody neutralization demonstrates that normally eIF4E1b actively represses translation in the oocyte.

DISCUSSION
Our major findings are first that in Xenopus oocytes, CPEB is an integral component of a very large RNP complex with Xp54, Pat1, RAP55B, FRGY2, 4E-T, and eIF4E. Evidence for the complex comes from in vivo pull downs using MS2-Xp54 and MS2-4E-T ( Figs. 1 and 7), co-immunoprecipitation of endogenous proteins with a monoclonal CPEB antibody (Fig. 1), and gel filtration (Fig. 2). Second, the eIF4E protein in this complex was identified as the ovary-specific eIF4E1b, using cap-Sepharose chromatography and mass spectrometry sequencing and an eIF4E1b-specific antibody. eIF4E1b binds the cap weakly, in contrast to the canonical initiation factor eIF4E1a, and interacts with 4E-T in a YSKEELL-independent manner (Figs. 3-6). Third, tethered MS2-4E-T(Y-A), which bound eIF4E1b but not eIF4E1a, was shown to repress translation, and injection of antibodies to eIF4E1b accelerated meiotic maturation progression ( Figs. 7 and 8).
The CPEB complex we describe is most abundant in early stage oocytes. Notably, the levels of RAP55B and eIF4E1b peak in stage I/II oocytes; the levels of other components are at their highest during stages II-IV/V, and decline in stage VI. During meiotic maturation, several components undergo modification, including phosphorylation and proteolysis, in particular CPEB, Pat1, and 4E-T (Figs. 1 and S1).
We have not detected significant interactions between CPEB and the previously described partners maskin and PARN, identified as such in late stage oocytes, when they are most abundant (11,12,15,17,18,53) (Fig. 1). This may reflect the use of different antibodies, which could preferentially target CPEB in separate complexes differentially expressed during oogenesis. In this study, we have exploited, for the first time, the availability of mouse monoclonal CPEB antibodies (40), which recognize the Xenopus protein specifically in oocyte lysates and enable the ready detection of co-immunoprecipitating proteins using rabbit antibodies by Western blot analysis. The existence of two CPEB-containing complexes in stage VI Xenopus oocytes, reflecting (at least in part) its dual role as a translational repressor and a cytoplasmic polyadenylation factor, has been previously described (11). The early oocyte CPEB complex we describe remains to be fully characterized, both in terms of protein and RNA components. However, our data provide a possible mechanism for the repression of CPEB target mRNAs early in oogenesis, by the use of an alternative eIF4E-binding protein, 4E-T, in combination with eIF4E1b and a set of conserved RNA-binding proteins.
We cannot exclude the possibility that this complex functions also later in oogenesis, since we continue to detect interactions between CPEB, Xp54, 4E-T, and eIF4E1b in stage VI oocytes (Figs. 7F and S4), 3 whereas we fail to observe binding of maskin to CPEB (Fig. S4). Others have also noted some of the interactions reported in this study in stage VI Xenopus oocytes. Tanaka et al. reported that FLAG-tagged FRGY2 interacts in an RNA-dependent manner with Xp54, whereas Xp54 binds RAP55A in a largely RNA-independent manner. Moreover, tethered RAP55A represses translation in oocytes, mediated by the Lsm domain shared with RAP55B (60). Our data extend these observations considerably by showing that endogenous RNP complexes, abundant in early oocytes, contain CPEB, in addition to Xp54, RAP55B, Pat1, 4E-T, and an ovary-specific eIF4E1b, all interacting via protein-protein interactions, and that tethered 4E-T represses translation in conjunction with eIF4E1b. We also find the association between CPEB and FRGY2 to require RNA. The two studies differ in the identification of the RAP55 protein that binds Xp54 in stage VI oocytes (RAP55A) 4 and the one that co-purifies with CPEB in early oocytes (RAP55B) (Figs. 1 and 2). We find that the ratio between RAP55A and -B proteins changes dramatically during oogenesis (Fig. 1), possibly at least partly explaining the varying results, although the existence of two Xp54-RAP55 complexes cannot be excluded.
Co-immunoprecipitation and pull-down assays have revealed interactions between paralogs of p54 and RAP55 proteins in C. elegans embryos (Cgh-1 and CAR-1) (44,68) and Drosophila oocytes (Me31B and Tral) (69). The Xp54-P100/ Pat1 interaction is conserved in yeast (Dhh1 and Pat1) (70) and Drosophila Schneider cells (Me31B and Hpat) (71). Xenopus P100 is the first vertebrate member of the Pat1 family to be characterized. The sequences of P100 and its highly conserved mammalian counterparts do not provide any hint regarding function. P100 probably binds RNA, it is present in a P-body like complex in oocytes, and its expression is restricted to the ovary (58) (this study). Interestingly, data base searches suggest the presence of two related Pat1 proteins in vertebrates, one of Preimmune and eIF4E1b antibodies were purified on protein A-Sepharose as described under "Experimental Procedures" prior to microinjection into stage VI oocytes, stimulated to mature with 0.1 g/ml progesterone. The timing of meiotic maturation was assessed by determining germinal vesicle breakdown (% GVBD) (A) and by Western blot with MAPK/ERK and p90 rsk antibodies (B) over 12 h following antibody injection and the addition of progesterone. Acceleration of meiotic maturation by injection of eIF4E1b antibody was not detectable in the absence of progesterone or with 10 g/ml progesterone (data not shown) but only with the low, presumably suboptimal, dose of 0.1 g/ml.
Moreover, the identified CPEB RNP proteins have been reported to be enriched in P bodies, maternal "germinal granules," and neuronal granules; these various forms of visible RNP granules contain translationally inactive mRNA, which may be destined for eventual decay, transport to a particular location, or translational activation. Yeast Dhh1, Pat1, Scd6, and eIF4E and human CPEB, RCK/p54, RAP55, YB-1, 4E-T, and eIF4E are found in P-bodies, Drosophila Orb, Me31B, Tral, HPat, Yps, Cup, and eIF4E in P bodies in tissue culture cells and in germinal granules and C. elegans Cgh-1, CAR-1, and Cey proteins in embryo P granules (reviewed in Refs. 41, 42, and 71). For example, Xp54 (Dhh1, Me31B, Cgh-1, and RCK) and RAP55 (Scd6, Tral, and CAR-1) proteins localize to P bodies in yeast and mammalian cells (37,40,74), in Drosophila S2 cells (71), and neuronal granules (45), in germinal granules in Drosophila oocytes (69,75), and in P granules in C. elegans embryos (44,68,76). In Drosophila, Orb, Me31B, Tral, and Cup localize within the future oocyte in developing regions 2 and 3 of the germarium, composed of germ line cysts, and are associated with the Balbiani body (69,75,77). The Balbiani body is a collection of organelles, including mitochondria, endoplasmic reticulum, and granulofibrillar material, located adjacent to the nucleus in young oocytes of diverse species. In Xenopus, granulofibrillar material has been shown to contain germinal granule proteins and RNAs that are incorporated into germ cells. Molecular studies of germinal granules strongly implicate them in regulated utilization of mRNA, but their precise composition and role of components remains to be determined (reviewed in Refs. 78 and 79). Although a systematic assessment of the localization of CPEB RNP proteins requires extensive future investigation, we note that Xp54 and RAP55 proteins are found in the Balbiani body in early Xenopus oocytes and in mouse oocytes within germ cell cysts, respectively (43,54). It is tempting to speculate on the basis of these studies that the CPEB complex functions in the differentiation of the oocyte in the Xenopus germ line cyst. Such a hypothesis is supported by the report that adult female CPEB knock-out mice contained vestigial ovaries devoid of oocytes, and ovaries from midgestation embryos contained oocytes arrested at the pachytene stage (80).
The remarkable retention in multicellular organisms of RNP complexes composed of a set of RNA-binding proteins (CPEB, Xp54, RAP55, Pat1, and FRGY2) as well as eIF4E and the eIF4Ebinding protein 4E-T to repress mRNAs demonstrates their fundamental roles in translational control. And yet, we barely understand how these proteins effect their control, providing considerable scope for future investigations. In somatic cells, in addition to P bodies, under certain conditions, stress granules are generated that also contain untranslated mRNA, RNAbinding proteins, and initiation factors. Although some components of P bodies and stress granules are common or at least can be detected in close proximity, by and large they have distinct components. For example, stress granules contain eIF4G, eIF4A, and eIF3, absent from P bodies, as are the proteins TIAR and ribosomal proteins, found in stress granules. Common to both are untranslated RNAs, which can reenter polysomes (reviewed in Refs. 41 and 81). Although some reports have shown that micro-RNA-mediated repression occurs in P bodies (35,82) (reviewed in Ref. 83), a recent study demonstrates that P body formation is a consequence rather than a cause of repression (35,71). Determination of the precise architecture of the RNP complexes and the contribution of individual components to translational repression is a major future goal. It will be of particular interest to understand the relationship between P bodies in somatic cells, with a rapidly changing mRNA population transcribed in response to cellular needs, and where some mRNAs decay whereas others return to translation, with that of the P-body-like CPEB RNP in Xenopus oocytes, containing stably stored maternal mRNAs, destined only for translational activation.
This study reports on Xenopus 4E-T for the first time. In HeLa cells, 4E-T (where "T" represents transporter) was initially characterized as an eIF4E-binding protein that is capable of importing eIF4E into nuclei in the presence of leptomycin B (57). At steady state, human 4E-T localizes to P bodies. To date, 4E-T is the only eIF4E-binding protein detected in P bodies; eIF4G (and eIF4A) is distributed homogenously in the cytoplasm in unstressed cells (39,84). 4E-T is required for P body formation and for the localization of eIF4E in P bodies (39,84). When overexpressed, human 4E-T represses cap-dependent reporter mRNA translation in a YTKEELL-dependent manner (84). In Drosophila, the characterized paralog of 4E-T is Cup, which binds eIF4E and mediates translational repression by the 3Ј-UTR-binding proteins Bruno and Smaug (19 -22). Besides the eIF4E-binding sites, vertebrate 4E-T and Cup are most similar in a short 25-amino acid region, unique to this family of proteins, whose role remains to be determined. The mouse paralog, Clast4, is maternally expressed and is phosphorylated during meiotic maturation (85). Interestingly, phosphorylation of human 4E-T during mitosis releases eIF4E (86).
We show that Xenopus 4E-T, but not eIF4G or 4E-BP1, is found in the CPEB RNP complex in early oocytes and that the only eIF4E1 protein in this complex is eIF4E1b, rather than the canonical cap-binding factor eIF4E1a. eIF4E1b binds the cap very weakly and interacts with 4E-T, rather than eIF4G, at a separate site from eIF4E1a. Preliminary experiments indicate that the potential separate eIF4E1b binding site lies within the first 329 amino acids of X4E-T but does not require the 25amino acid-long "Cup-homology" domain ( Fig. S2) (data not shown). Although the existence of an alternative binding site is supported by the observation that Cup has two distinct binding sites for eIF4E (21) (Fig. S2), we found that mutation of the sequence in X4E-T resembling the second site in Cup had no effect on eIF4E1a or eIF4E1b binding (data not shown).
Paralogs of eIF4E1b are conserved in vertebrates but are not discernible in flies or worms. These invertebrates contain 5-8 eIF4E-like proteins, and possibly one may fulfill the role of eIF4E1b. eIF4E1b expression is confined to oocytes, eggs, and early embryos in Xenopus, zebrafish, and mice. The participation of a weak cap-binding protein in the CPEB complex is at first sight perplexing in view of its proposed role in translational repression. Nevertheless, alongside recently described parallels, it is possible to propose a model for CPEB. 4E-HP, a class II eIF4E, represses caudal and hunchback mRNAs by binding the cap and Bicoid and Nanos/Pum/Brat, respectively, which interact with the mRNAs directly via particular 3Ј-UTR cis-elements (23,24). Subsequent studies of 4E-HP have shown that it binds the m 7 GpppG 30-fold less well than eIF4E1a (27,28). Interestingly, it has also been shown that Argonaute 2, tethered to the target mRNA 3Ј-UTR by micro-RNA (83), also possesses weak cap-binding activity, responsible for repression of translation initiation by micro-RNAs (87). A similar scenario may underlie the repression of CPE-RNAs in Xenopus oocytes by the eIF4E1b-4E-T-CPEB complex. According to this model, weak cap binding by 3ЈUTR-tethered proteins would be sufficient for inhibition of translation of target mRNAs, could in principle be readily reversed by altering 3Ј-UTR RNP dynamics, and would avoid inhibition of general protein synthesis (Fig. 9).
The structure of XeIF4E1b, with and without a cap, was homology-modeled using the respective solved structures of mouse eIF4E1a. 5 The structures of mouse and Xenopus pro-teins were largely superimposable, not revealing very obvious explanations for lack of cap-binding by eIF4E1b. The most plausible causal difference lies in the N terminus, part of the region that interacts with eIF4G in eIF4E1a (88), possibly preventing eIF4G binding to eIF4E1b, reducing cap affinity. However, removal of the basic N-terminal region in XeIF4E1b does not enhance cap binding (data not shown) and does not promote complementation by the zebrafish eIF4E1b in yeast lacking its own eIF4E protein (63). It is therefore not yet clear why this class of eIF4E1 proteins fails to bind the cap robustly.
In summary, our data suggest that in early Xenopus oocyte CPEB functions in complex with conserved RNA-binding proteins, which inhibit translation using an unconventional pairing of 4E-T and eIF4E1b, potentially providing a second example, in addition to 4E-HP, of a situation in which a poor cap-binding eIF4E protein mediates mRNA-specific repression.