CPEB controls oocyte growth and follicle development in the mouse Free

Early animal development is programmed by maternally inherited mRNAs that are synthesized during oogenesis (Wickens et al., 2000; Richter,2000). These mRNAs are translated during oocyte maturation, when the cell enters the meiotic divisions, and during embryogenesis. Several dormant mRNAs in oocytes have short poly(A) tails, and when these tails are subsequently elongated, translation ensues. While probably all vertebrate oocytes contain maternal mRNA that is regulated by polyadenylation, most molecular details have been elucidated in Xenopus. mRNAs that undergo polyadenylation-induced translation contain two cis-acting sequences in the 3′ UTR: the cytoplasmic polyadenylation element (CPE) and the polyadenylation hexanucleotide AAUAAA. The CPE is bound by CPEB and the AAUAAA interacts with cleavage and polyadenylation specificity factor (CPSF)(Mendez and Richter, 2001). In response to progesterone signaling, the kinase Aurora A is activated by dissociating from glycogen synthase kinase 3 (GSK-3)(Sarkissian et al., 2004), and phosphorylates CPEB S174, which in turn activates the cytoplasmic polyadenylation machinery (Mendez et al.,2000a; Mendez et al.,2000b). This machinery includes not only CPEB and CPSF, but also Symplekin, which acts as a molecular scaffold, and GLD2, a poly(A) polymerase(Barnard et al., 2004). CPEB S174 phosphorylation stimulates the activity of GLD2, although the mechanism by which this is accomplished is unclear. Polyadenylation controls translation through maskin, a CPEB-associated protein(Stebbins-Boaz et al., 1999; Barnard et al., 2005). Maskin also binds the cap-binding factor eIF4E, and this interaction prevents the assembly of the eIF4F (eIF4E, eIF4G, eIF4A) initiation complex. Polyadenylation, however, causes the dissociation of maskin from eIF4E,thereby allowing the translation initiation to proceed(Cao and Richter, 2002; Richter and Sonenberg,2005).

In the maturing mouse oocytes, CPE-dependent polyadenylation also induces translation (Vassalli et al.,1989; Huarte et al.,1992; Gebauer et al.,1994; Oh et al.,2000), which is controlled by CPEB(Tay et al., 2000). A Cpeb knockout mouse (KO) was generated with the expectation that oocyte maturation would not proceed normally(Tay and Richter, 2001). Surprisingly, however, female mice contained no ovary; embryonic day (E) 18.5 female embryos contained ovaries but were mostly devoid of oocytes. E16.5 embryo ovaries contained oocytes, but their chromatin was morphologically abnormal and their development was arrested at pachytene, a stage when the synaptonemal complex promotes homologous recombination. In the KO mice,synaptonemal complex proteins SCP1 and SCP3 (SYCP1 and SYCP3 - Mouse Genome Informatics) were not synthesized, and the CPE-containing RNAs that encoded them were not polyadenylated (Tay and Richter, 2001). Thus, CPEB controls an earlier stage of meiosis than previously suspected. Spermatogenesis was also disrupted at pachytene in CPEB KO male mice, demonstrating that CPEB is a general regulator of meiosis.

Further investigations revealed that mouse oocyte CPEB T171 (equivalent to S174 in Xenopus) was phosphorylated at E16.5 (pachytene), underwent protein phosphatase 1 (PP1)-catalyzed dephosphorylation at E18.5, and again became phosphorylated during oocyte maturation(Tay et al., 2003). These results suggested that CPEB-mediated polyadenylation would probably next take place during maturation, when it promotes the translation of several mRNAs,including Mos and cyclin B1(Gebauer et al., 1994; Tay et al., 2000; Hodgman et al., 2001). The translation of Mos mRNA is particularly important for mouse development. MOS is a protein kinase that activates the MAP kinase cascade,which culminates in the activation of maturation promoting factor (MPF), a heterodimer of CDC2 and cyclin B1 that induces maturation; MOS thus has a mitogenic activity. MOS has a second activity; it is a component of the cytostatic factor (CSF) complex that prevents oocyte development beyond meiosis II (MII) until fertilization stimulates the resumption of cell division (O'Keefe et al.,1989; Sagata,1997; Gebauer and Richter,1997; Tunquist and Maller,2003). In Xenopus oocytes, MOS has both these activities,but in mouse oocytes it has retained only the CSF activity. Thus, oocytes derived from Mos knockout mice often fail to arrest at MII and undergo a high rate of parthenogenetic activation(Colledge et al., 1994; Hashimoto et al., 1994; Choi et al., 1996).

Because Mos mRNA translation is regulated by cytoplasmic polyadenylation (Gebauer et al.,1994), we hypothesized that oocytes lacking CPEB would have reduced CSF activity and consequently undergo parthenogenetic activation. However, an investigation of this possibility required that oocytes contain CPEB at pachytene but not beyond this stage, thus obviating the use of the Cpeb KO mice. Consequently, we sought to destroy Cpeb mRNA after pachytene by generating transgenic (TG) mice expressing siRNA under the control of the zona pellucida 3 (Zp3) promoter. The Zp3promoter is expressed in growing dictyate stage oocytes at the end of prophase I (Epifano et al., 1995).

An analysis of several TG mouse lines demonstrated that Cpeb RNA was indeed substantially destroyed by the siRNA. While ovaries from pre-pubertal TG females were morphologically identical to those of wild-type animals, at puberty (6-8 weeks) and in older animals, a number of oocyte and ovarian abnormalities were observed: these included premature oocyte maturation and increased oocyte atresia, parthenogenetic activation, malformed meiotic spindles, precocious follicle activation and granulosa cell apoptosis. Fertility was substantially reduced.

CPEB interacts with a number of CPE-containing RNAs, including those encoding Mos and Gdf9, a growth factor synthesized in and secreted from oocytes that is important for coordinated oocyte-follicle development (Matzuk et al.,2002; Roy and Matzuk,2006). In the TG oocytes, both Mos and Gdf9 RNAs had aberrantly short poly(A) tails; GDF9 protein levels were also reduced. Some of the TG oocyte phenotypes resemble those of Mos(Colledge et al., 1994; Hashimoto et al., 1994; Choi et al., 1996) and Gdf9 KO mouse oocytes (Dong et al., 1996; Carabatsos et al.,1998). These results demonstrate that CPEB controls polyadenylation and translation during the dictyate stage, and that this regulation coordinates oocyte/follicle development, which in turn directly affects fertility.

The backbone Zp3 cassette-SV40 late poly(A), pRL-CMV (Promega)plasmid for the TgZ3GC1 construct is a modified version of pEGFP-N2 (Clontech)and pMoZP3 was kindly provided by Dr Richard Schultz (University of Pennsylvania) and is described in Svoboda et al.(Svoboda et al., 2001) and Stein et al. (Stein et al.,2003). A Cpeb dsRNA was constructed by ligating head to head the 645 and 670 nucleotide PCR amplification products encompassing the 612 nucleotides of the Cpeb 3′ UTR immediately after the stop codon (plus cloning sites and the loop structure). The 1.3 kb 3′ UTR inverted repeat was subsequently cloned into the pRL-CMV vector (see Fig. 2A). Transgenic mice were generated by pronuclear injection of purified DNA into fertilized embryos harvested from C57BL/6 inbred mice at the UMass Transgenic Animal Core. Transgenic founders were identified by Southern blots and PCR. Seven founder lines of mice, four males and three females, were obtained. These animals were bred to C57BL/6 animals; two males and one female produced offspring. These founders were used to establish three independent lines (i.e. Ln2f, Ln19m and Ln30m) of animals expressing dsRNA targeting CPEB in oocytes. Transgenic F1 males were mated to wild-type littermates and F2 progeny were tested for transgene transmission.

RNA from 10 GV stage oocytes as well as ovaries was isolated TRIzol Reagent(Invitrogen) and used for cDNA synthesis. RT-PCR was performed on 2.5- to 5-fold dilutions of first strand synthesis mixture. All primer sequences are listed in Table 1.

The immunoprecipitation of mRNP complexes was performed using ovaries from ten wild-type mice as described (Tenenbaum et al., 2000). The ovaries were homogenized in buffer (100 mmol/l KCl, 5 mmol/l MgCl₂, 10 mmol/l HEPES-KOH pH 7.0, 0.5% Nonidet P(NP)-40, 1 mmol/l DTT, 1 mmol/l Na vanadate, 10 mmol/l NaF, 80 mmol/lβ-glycero-phosphate, 0.2 mmol/l PMSF plus protein and RNaze inhibitors[0.2 mmol/l PMSF, 1 μg/ml Pepstatin A, 2 μg/ml Aprotinin, 2 μg/ml Leupeptin, 100 U/ml RNaseOUT (Invitrogen) and 0.2% vanadyl/ribonucleoside complex (GIBCO/BRL)] and frozen at -80°C. The lysate was pre-cleared for 10 minutes at 16,000 g (4°C). Magnetic beads were washed four times in 1 ml NT2 buffer (50 mmol/l Tris-HCl pH 7.4, 150 mmol/l KCl, 1 mmol/l MgCl₂, 0.05% NP-40) and then a final time when supplemented with 5% BSA. The beads were incubated with CPEB antibody or nonspecific IgG at 4°C overnight with tumbling. The beads were washed three times with NT2 buffer and re-suspended in 900 μl NT2 buffer supplemented with protease and RNase inhibitors. One hundred microliters of ovary extract was added, mixed briefly, and immediately centrifuged to collect a 100 μl sample of total RNA. The remaining sample was tumbled overnight at 4°C, then washed six times with cold NT2 buffer; the RNA was isolated using TRIZOL reagent.

To examine poly(A) tail length, we employed poly(U) Sepharose chromatography and thermal elution (Simon et al., 1996; Du and Richter,2005). Poly(U)-Sepharose (Sigma) was swollen in SB buffer (1 mol/l NaCl, 5 mmol/l Tris-HCl, pH 7.5, 10 mmol/l EDTA, 0.2% SDS), washed three times in EB buffer (90% formamide, 50 mmol/l HEPES, pH 7.5, 10 mmol/l EDTA, 0.2%SDS), and equilibrated in CSB buffer (25% formamide, 0.7 mol/l NaCl, 50 mmol/l Tris-HCl pH 7.5, 1 mmol/l EDTA). The beads (50 μl gravity packed) were mixed with 10 μg total RNA from whole ovaries of 3-week-old mice that was suspended in 50 μl of solution containing 1% SDS and 30 mmol/l EDTA,incubated for 5 minutes at 70°C and then diluted fivefold in CSB buffer. The beads and RNA were mixed with constant inversion for 1 hour at 25°C. The beads were briefly centrifuged, the supernatant was removed, and the beads washed three times with 1 ml each of LSB buffer (25% formamide, 0.1 mol/l NaCl, 50 mmol/l Tris-HCl, pH 7.5, 10 mmol/l EDTA), and then incubated in 0.3 ml LSB at 30°C for 3 minutes, pelleted by brief centrifugation, and the supernatant was removed and saved for analysis. This procedure was repeated for 45°C and 60°C followed by final incubation at 95°C. After each elution, the RNA was extracted and ethanol precipitated.

SDS-PAGE and protein blotting was performed on 10-20 μl of whole single ovary lysate corresponding approximately to 10-20% of the total tissue. The blots were probed with polyclonal CPEB primary antibody at 1:1000, monoclonal GDF9 primary antibody at 1:3000 (courtesy of M. M. Matzuk, Departments of Pathology, Molecular and Cellular Biology and Molecular and Human Genetics,Baylor College of Medicine).

Ovaries were washed briefly at room temperature in PBS, pH 7.2, blotted to remove excess liquid and fixed in Bouin's solution for 6-12 hours. The fixed tissue was washed extensively with 70% ethanol and paraffin embedded,sectioned and stained with hematoxylin and eosin. For TUNEL assays, paraffin sections (5 μm) were de-waxed and re-hydrated following standard procedures. The re-hydrated sections were boiled for 10 minutes in 100 mmol/l sodium citrate buffer, pH 6.0, left to cool at room temperature, and washed three times with PBS. The sections then were permeabilized for 2 minutes with 0.1% Triton X-100, 0.1% sodium citrate on ice, followed by three washes with PBS. Labeling was performed using In Situ Cell Death Detection Kit, POD(Roche) according to the manufacturer's instructions.

Some salient features of oogenesis and folliculogenesis, as well as the stages during which CPEB has been shown to regulate translation, are depicted in Fig. 1. Primordial germ cells (PGCs) migrating to the genital ridge divide mitotically. In the embryonic ovary, follicle formation is preceded by incomplete synchronous mitotic division leading to the formation of the cystocytes (oocytes with interconnecting cytoplasmic bridges) and enter leptotene of prophase I(Pepling and Spradling, 1998; Pepling and Spradling, 2001). The oocytes contain CPEB at this stage but the protein is not phosphorylated and thus is inactive (Tay et al.,2003). At zygotene to pachytene, Aurora A phosphorylates CPEB T171, which induces the polyadenylation and translation of mRNAs encoding synaptonemal complex proteins 1 and 3 (SCP1, SCP3)(Tay and Richter, 2001). These proteins are essential components of the synaptonemal complex, a structure that promotes homologous recombination. As the oocytes exit pachytene, protein phosphatase 1 (PP1) dephosphorylates and inactivates CPEB. The oocytes then enter the dictyate stage (concomitant with perinatal cytocyst breakdown and primordial follicle formation), which can last from around birth to about 3 weeks; most oocytes, however, remain in the dictyate stage for the entire reproductive period (up 2 years). The oocytes that are activated enter a growth and development phase that takes about 3 weeks to complete; it culminates with re-entry into the meiotic divisions (maturation), when they undergo germinal vesicle breakdown (GVBD) and extrude the first polar body. During maturation, CPEB again is phosphorylated on T171, which promotes polyadenylation and translation of Mos, cyclin B1 and other mRNAs(Gebauer et al., 1994; Tay et al., 2000; Hodgman et al., 2001).

During the dictyate stage, the follicle undergoes striking growth and development. The zona pellucida is elaborated, antrum formation occurs, and in preovulatory follicles the oocyte becomes surrounded by several layers of cumulus granulosa cells. The zona pellucida is composed of three proteins, one of which, ZP3, begins to be expressed in growing oocytes (follicle stage 3a)at about day 6 post-partum. Because Cpeb KO oocytes do not develop beyond pachytene, we used the Zp3 promoter to express a palindromic RNA that, when processed by DICER, would generate multiple siRNAs and induce Cpeb RNA destruction (Fig. 2A). The 3′ UTR of Cpeb was targeted because it had no significant homology with other RNAs, which was not the case with the open reading frame (Mendez and Richter,2001; Mendez et al.,2002). Consequently, the 612 nucleotide inverted repeat RNA (1301 bases total) was appended to Gfp after the stop codon. A vector containing this sequence under the control of the Zp3 promoter was injected into pronuclear stage eggs from C57BL/6 mice; pups derived from these injected eggs were bred onto C57BL/6.

Three female and four male transgenic founders were analyzed; they appeared normal but two males and one female did not produce offspring. The male infertility was unexpected considering the normal appearance of their reproductive organs, and could be due to insertional inactivation of a gene required for sperm production. The infertile female was sacrificed at 10 months and had hypertrophic ovaries ∼4-fold larger than normal with no detectable follicular structure or oocytes; only a single primary stage 3a follicle was detected (see Fig. S1 in the supplementary material). Three lines of CPEB knockdown animals were retained for further analysis. F1 males from all the transgenic lines were fertile and produced offspring with the expected genotypic ratios. Transgenic F1 females from two transgenic lines did not produce offspring after repeated matings for 6 months; females from the other line produced offspring but only up to 4-6 months of age. In two independent experiments per line, single ovaries isolated from three to five F2 TG females contained 61% fewer GV oocytes than wild type (three females used) at 2 months of age, 88% fewer at 12 months of age. However, those few oocytes that were obtained matured in vitro to MII at a rate similar to wild-type oocytes (data not shown).

The F2 offspring from the TG lines and corresponding wild-type littermates were analyzed further; they expressed Gfp RNA(Fig. 2B), although a green fluorescent signal was too low to detect. However, in those lines containing Gfp RNA, there was almost no detectable Cpeb RNA. In line 2f, the efficiency of transgene expression was significantly lower compared to the other two lines, which might be expected to produce in a milder phenotype. Indeed, we did observe a range of phenotypes (see Table S1 in supplementary material). The levels of non-targeted Mos, Zp3 and α-tubulin RNAs were all unaffected by the expression of the transgene. We thus conclude that CPEB RNA was efficiently and specifically destroyed by siRNA.

The ovaries from both wild-type and TG animals were examined at intervals from 6 weeks to 10 months (Fig. 2C). While the 6-week-old wild-type and TG ovaries were similar and contained numerous follicles, the 2- and 4-month-old ovaries were strikingly different. At these times, the wild-type ovaries contained numerous follicles but the TG ovaries contained very few. By 10 months, almost no follicles were detected in the TG ovaries. The TG ovary displayed substantial granulosa cell apoptosis, as assessed by a TUNEL assay(Fig. 2D). The wild-type ovary was barely labeled by the TUNEL reagent, indicating that it contained few apoptotic cells.

Histological examination of TG ovaries at 6 weeks to 6 months revealed a number of oocyte phenotypes (Fig. 3). Oocytes often underwent precocious maturation(Fig. 3A), formed metaphase chromosomes assembled into multi-polar spindles(Fig. 3A,B, arrows) and improperly lacked the surrounding cumulus granulosa layer(Fig. 3A,B; see also Fig. 4A,C, stars). Other oocytes had chromosomes that were not properly aligned at the metaphase plate and/or underwent anaphase in which some chromosomes did not properly segregate(Fig. 3B, arrow). Some oocytes also had disorganized and mono-polar spindles(Fig. 3C), extruded polar bodies from opposite ends of the cell (Fig. 3D), or contained polar bodies with metaphase chromosomes(Fig. 3E, arrow) or pronucleus-like structures (Fig. 3F, arrow). Other oocytes had improper nuclear envelopes with disorganized chromatin (Fig. 3G, filled arrowheads), contained cytoplasmic structures that resembled nuage or perhaps nucleoli (Fig. 3H,I, empty arrowheads; see also Fig. 4I), or contained condensed DNA in the cytoplasm (Fig. 3I, bold arrow). Such anomalies were almost never observed in wild-type ovaries.

Several oocytes showed incipient signs of cell division, or indeed underwent multiple rounds of cell division while in the ovary. Fig. 4A shows an oocyte, having already extruded possibly a second polar body, about to undergo cell division. Fig. 4B-G show oocytes that have undergone parthenogenetic cell division, and, in Fig. 4D-F, up to 15 cells can be detected in each parthenote (notice metaphase cell in D, arrow). Several oocytes also contained multiple fragmented nuclei(Fig. 4G-H, arrows), and appeared to have nuage or perhaps even nucleoli in the cytoplasm(Fig. 4I, empty arrowheads).

We also noted that TG ovaries contained follicles that housed two oocytes(Fig. 5A). The TG ovaries contained a large number of follicles with collapsed oocytes that were aberrantly surrounded by either a single layer of squamous granulosa cells or several layers of highly compact small granulosa cells(Fig. 5B,C). The incidence of this phenotype increased with age, such that by 4 months almost half the follicles resembled those in Fig. 5B,C. Finally, cysts were evident on the TG ovaries(Fig. 5D,E). Four-month-old animals had on average one to three cysts per ovary, which doubled as the animals aged to 12 months. We could detect only a single cyst in ovaries from wild-type animals of the same age (data not shown; however, see Table S1 in the supplementary material).

To examine the molecular foundation that could be responsible for the phenotypes noted above, we sought to identify oocyte mRNAs that were bound by CPEB. Extracts from wild-type ovaries from 3-week-old mice were mixed with RNase and protease inhibitors plus CPEB antibody (see Tay and Richter, 2001; Tay et al., 2003), or as a control, non-specific IgG pre-bound to magnetic beads. The CPEB-containing ribonucleoproteins were then immuno-selected and the RNA was extracted and used for RT-PCR with primers that were specific to CPE-containing RNAs. As controls, RNAs without CPEs were also analyzed(Fig. 6A). Because CPEB is present only in oocytes and not in ovarian somatic cells(Gebauer and Richter, 1996),the RNAs we detected would also be oocyte-derived. CPEB interacted with a wide range of RNAs including those that encode OBOX1(Rajkovic et al., 2002) and SMAD family members (Pan et al.,2005), which are especially abundant in early oocytes (stages 2 to 4), H1FOO (Tanaka et al.,2005) and GDF9, which are abundant in primary to fully grown oocytes (stages 3b to 8), and MOS and SPIN, which are detected in fully developed oocytes (Mutter and Wolgemuth,1987; Iwaoki et al.,1993; Oh et al.,1997). Based on these results, we think that CPEB is probably present throughout oogenesis.

CPEB interacted with mRNAs that encode proteins involved in signal transduction (e.g. Smad1 and Smad5), cell cycle control(e.g. spindlin, Bub1b, Mos), transcription (e.g. Obox1),chromatin remodeling and regulation of gene expression (e.g. H1foo),maintenance of methylation patterns important for imprinting [e.g. Dnmt1o (Dnmt1 - Mouse Genome Informatics)], and protein modification (TiParp, Trim61). However, we were particularly interested in Mos and Gdf9 RNAs because the phenotypes associated with mice lacking these genes resemble some of the phenotypes we observed with our TG animals. For example, oocytes from Mos KO animals undergo parthenogenesis, as do oocytes from our TG animals. Gdf9 KO animals exhibit granulosa cell apoptosis (when inhibinα levels are also reduced) (Wu et al., 2004) and aborted follicle development(Elvin et al., 2000); similar features were observed in our TG animals as well. We therefore examined whether the poly(A) tails of these RNAs were affected in the oocytes of TG versus wild-type animals. We initially employed a PAT [poly(A) test] described by Salles and Strickland (Salles and Strickland, 1995). This PCR-based assay was inadequate for our analysis, perhaps because of the paucity of starting material. Consequently,we employed chromatography on poly(U) Sepharose followed by thermal elution and RNA detection by RT-PCR, which has been used successfully to analyze polyadenylation of small amounts of RNA (Du and Richter, 2005). Total RNA from ovaries of 6- to 8-week-old mice (wild type and TG) was applied to the poly(U) beads, which was followed by washing and elution at increasing temperatures; RNAs with the longest tails should elute at the higher temperatures(Fig. 6B, top panel). The majority of Mos, H1foo (an oocyte-specific histone H1 variant), Gdf9 and Obox1 RNAs eluted at 45°C for both wild-type and TG ovaries (Fig. 6B, bottom panel); some RNA also eluted at 60°C for the wild type. These results indicate that a subpopulation of these RNAs have poly(A) tails that are shorter in the TG versus wild-type ovaries. The RNAs encoding Mater(Nalp5 - Mouse Genome Informatics) and α-tubulin, which have no CPE, eluted at both temperatures in wild-type and TG ovaries. To assess whether the reduced poly(A) tail length resulted in decreased translation, a western blot for GDF9 was performed (Fig. 6C). Compared with wild type, Gdf9 was substantially reduced in the TG ovaries. Therefore, CPEB controls the polyadenylation and translation of Gdf9 RNA.

This study demonstrates that CPEB regulates oocyte development during the dictyate stage, and that this regulation has a profound influence on folliculogenesis as well. Because CPEB is required for meiotic progression beyond pachytene (Tay and Richter,2001), we sought to knockdown this protein after this stage, and consequently generated Zp3-regulated siRNA against Cpeb RNA; Zp3 is expressed only at the end of prophase I. The follicles from these transgenic animals had a range of phenotypes including premature oocyte maturation, parthenogenesis and granulosa cell apoptosis. CPEB interacted with several CPE containing RNAs, and the two encoding Mos and Gdf9 had aberrantly short poly(A) tails in the TG compared with wild-type oocytes. Finally, GDF9 protein levels were markedly reduced in the TG animals, suggesting that some of the phenotypes were probably due to the paucity of this crucial growth factor.

The necessity of CPEB for oocyte growth and follicle development was an unexpected finding, which was based primarily on the CPEB phosphorylation pattern during oogenesis. CPEB T171 phosphorylation is required for cytoplasmic polyadenylation (Mendez et al., 2000a; Hodgman et al.,2001), and an analysis of mouse oocytes in prophase I showed that while this residue was phosphorylated at E16.5 (generally in pachytene), it was not at E18.5 (generally in diplotene)(Tay et al., 2003). Maturation was the next time that CPEB activity was known to occur, and at this stage it again was T171 phosphorylated. These results suggested that CPEB was not essential between pachytene and maturation. Therefore, we surmised that a knockdown of Cpeb after pachytene would result in maturation defects,particularly those related to Mos (see Introduction), the mRNA of which is translationally controlled by CPEB(Gebauer et al., 1994). However, CPEB T171 phosphorylation was not examined during the dictyate stage,so there was no a priori reason why polyadenylation-induced translation would not occur during this time. Our findings that CPEB is essential for this time of oogenesis, that it is necessary for polyadenylation and translation of (at least) Gdf9 mRNA, suggests that it is phosphorylated during this period. We thus propose that although the phosphatase PP1 dephosphorylates CPEB at the end of prophase I, CPEB might undergo an additional phosphorylation-dephosphorylation cycle before maturation. It is also important to point out that all oocytes do not grow synchronously, but instead there are cohorts of cells at different phases of oogenesis. Thus, an analysis of ovarian oocytes from adult animals would contain oocytes at several developmental stages.

The range of phenotypes we observed is unlikely to be attributable to the mis-regulation of one or only a few mRNAs. Indeed, the data in Fig. 6 show that CPEB is co-immunoprecipitated with many CPE-containing RNAs. The CPE is a U-rich structure and because 3′ UTRs are often AU-rich, one might surmise that the polyadenylation and translation of many mRNAs are regulated by CPEB. An in silico analysis of 3′ UTRs revealed that up to one-third contain CPE-like sequences (Oh et al.,2000). However, 3′ UTRs can also be highly structured, which could obviate an interaction with CPEB, or contain other sequences that could inhibit polyadenylation (Simon et al.,1992). It was therefore not surprising when only about 7% of cellular RNAs were found to actually undergo cytoplasmic polyadenylation;although all the RNAs that were polyadenylated contained CPEs, some RNAs that were not polyadenylated also contained CPE-like elements(Du and Richter, 2005). While this still represents a large number of sequences, our focus on Gdf9mRNA as a substrate for CPEB activity during oocyte growth was suggested not only by the oocyte phenotypes in the TG animals, but by the follicle phenotypes as well. GDF9, a member of the transforming growth factor-β(TGFβ) superfamily, is synthesized in and secreted from oocytes throughout the dictyate stage (Dong et al.,1996; Elvin et al.,1999a), and is one of several proteins that coordinate oocyte and follicle development (Soyal et al.,2000; Matzuk et al.,2002). Oocytes from Gdf9 knockout mice have an enhanced growth rate, have few surrounding granulosa cells and eventually undergo degradation/collapse/apoptosis, but occasionally undergo parthenogenetic activation in vivo and in vitro (Carabatsos et al., 1998). We have observed enhanced oocyte growth in our TG CPEB knockdown animals; these oocytes also are irregularly shaped(Fig. 4I). Several oocytes were detached from their granulosa cells and had undergone parthenogenetic activation within the ovary (Figs 3 and 4). Follicle development in Gdf9 knockout mice is also disrupted; primary stage 3a follicles lack a thecal layer (Dong et al.,1996), and in later stage follicles granulosa cell expansion is inhibited (Elvin et al.,1999b). In the absence of GDF9 and inhibin-α, granulosa cell apoptosis occurs (Wu et al.,2004). We also detect thecal cell abnormalities and substantial granulosa cell apoptosis (Fig. 2), so it is possible that the translation of additional RNAs is reduced in the TG. Based on these observations and the fact that the poly(A)tail length of Gdf9 RNA is reduced in the TG animals, it is not surprising that little GDF9 protein was produced. Thus, it seems probable that the lack of GDF9 (and probably others) contributes to the oocyte and follicle phenotypes in the TG animals.

In addition to Gdf9, Mos mRNA polyadenylation was also reduced in the TG mice; however, we were unable to detect MOS by immunoblotting, probably because the protein is not abundant and because of the paucity of oocytes that could be obtained from the TG ovaries. While we speculate that MOS levels may also be reduced, those few oocytes that we did obtain and culture in vitro matured normally to MII (data not shown); we did not observe any oocytes that underwent parthenogenetic activation in vitro, which was observed in the Mos knockout mouse (e.g. Colledge et al., 1994). It should be noted that our Cpeb knockdown strategy almost certainly would not result in the complete loss of CPEB in all oocytes. Indeed, some oocytes may even contain relatively normal levels of CPEB, and those may be the ones that progress normally to MII in vitro(compare Fig. 2B and Fig. S2 in the supplementary material).

CPEB regulation of translation is essential at least three times during oogenesis: at pachytene, during oocyte growth and during meiotic maturation. While it is not surprising that CPEB controls translation during maturation when a number of cell cycle control proteins must be synthesized, why it regulates translation at pachytene and during oocyte growth is less clear. At pachytene, CPEB regulates the polyadenylation-induced translation of Scp1 and Scp3 mRNAs (Tay and Richter, 2001). These mRNAs presumably are present but repressed at zygotene or even earlier; perhaps synaptonemal complex formation occurs when transcription decreases; thus, the requirement for translational activation of dormant mRNAs. Some of the target mRNAs are especially abundant in early stage oocytes (e.g. Smad5), while others are more abundant in fully grown oocytes (e.g. Mos). CPEB might orchestrate the oocyte-stage-specific translation of the class of CPE-containing mRNAs necessary for the oocyte to progress from one stage of development to the next(e.g. Gdf9). This succession of translation activation itself would have to be regulated by cellular cues most likely conferred by subsets of signaling cascades leading to CPEB activation/inactivation. Why mRNAs encoding a protein such as Gdf9 would need to be regulated by CPEB during the dictyate stage is not obvious. Perhaps hormonal or other signaling cues start the cytoplasmic polyadenylation engine so that several RNAs are translated simultaneously to coordinate oocyte growth and folliculogenesis.

Because the CPE and CPEB are involved in mRNA repression as well as activation (Stutz et al.,1997; Stutz et al.,1998; de Moor and Richter,1999; Stebbins-Boaz et al.,1999; Cao and Richter,2002), some mRNAs in the TG oocytes might be translated preciously. Indeed, in mouse embryo fibroblasts (MEFs) derived from CPEB knockout mice, myc mRNA is translated at an elevated rate, and is involved in the bypass of cellular senescence(Groisman et al., 2006). Thus,some of the phenotypes observed in the TG oocytes could be an amalgam of underexpression and overexpression of specific mRNAs.

Finally, several of the phenotypes we observe in the Cpeb TG follicles resemble those of the human premature ovarian failure (POF)syndrome. POF is caused by a variety of factors (e.g. viral infections,autoimmunity, environmental toxins), including mutations in the Gdf9gene (Laissue et al., 2006). Because CPEB is necessary for Gdf9 expression, it is possible that additional genetic lesions that lead to POF will map to the CPEB gene. Thus,the CPEB RNAi TG mouse described here may serve as a useful tool to elucidate some of the etiology of human POF.

We thank Richard Schultz (University of Pennsylvania), Martin Matzuk(Baylor College of Medicine) and David Garlick (Charles River Laboratories)for reagents and advice. This work was supported by a grant from the NIH(HD37267). Additional core support from the Diabetes Endocrinology Research Center Program Project (DK32520) is gratefully acknowledged.