|Home | About | Journals | Submit | Contact Us | Français|
Germ cells divide and differentiate in a unique local microenvironment under the control of somatic cells. Signals released in this niche instruct oocyte reentry into the meiotic cell cycle. Once initiated, the progression through meiosis and the associated program of maternal mRNA translation are thought to be cell-autonomous. Here we show that translation of a subset of maternal mRNAs critical for embryo development is under the control of somatic cell inputs. Translation of specific maternal transcripts increases in oocytes cultured in association with somatic cells and is sensitive to EGF-like growth factors that act only on the somatic compartment. In mice deficient in amphiregulin, decreased fecundity and oocyte developmental competence is associated with defective translation of a subset of maternal mRNAs. These somatic cell signals that affect translation require activation of the PI3K/AKT/mTOR pathway. Thus, mRNA translation depends on somatic cell cues that are essential to reprogram the oocyte for embryo development.
Germ cell differentiation requires a unique microenvironment created by surrounding somatic cells. In gonads of adult Drosophila and C. elegans, this environment provides the “niche” that is key to the maintenance of the germ stem cell pool1. A similar niche is critical for spermatogonial stem cell replication and differentiation to maintain spermatogenesis in mammals2. In adult mammalian ovaries, the follicle microenvironment in which the oocyte develops is the conduit for bidirectional exchange of signals with surrounding somatic cells3. With the possible exception of transcription in model organisms, the molecular basis for somatic/germ cell interactions is still poorly understood.
Upon signals from the soma in the ovarian follicle, fully grown mammalian oocytes reenter into the meiotic cell cycle, complete the first meiotic division, and progress to metaphase II (MII) of the second division4. Although these transitions still occur if oocytes are freed from surrounding somatic cells and cultured in vitro, it is commonly accepted that, in the absence of somatic cell contacts, oocyte fertilization and embryo development are compromised5-7. These defects probably arise from disruptions in the poorly-defined molecular process by which the oocyte acquires developmental competence, termed cytoplasmic maturation8. Nuclear transfer experiments indeed show that the defects associated with oocyte denudation and in vitro culture reside in the cytoplasm6. Since cytoplasmic maturation of the oocyte and early embryo development proceed in the absence of transcription, competence to develop as an embryo must rely upon a genome-wide program of maternal mRNA translation and degradation.
Oocyte maturation and ovulation induced by the gonadotropin LH requires activation of paracrine/autocrine signals within the follicle. In addition to the release of prostaglandins and steroids, LH induces large increases in amphiregulin (Areg), epiregulin (Ereg), and betacellulin mRNAs within 1-3 hrs of stimulation in mural granulosa cells9, followed by an increase in cumulus cells10. These growth factors bind to the EGF receptor (EGFR) on granulosa cells, and their release mediates the LH-dependent transactivation of EGFR. Genetic and pharmacological data demonstrate that activation of this EGF network is essential to transmit the gonadotropin signal from the mural granulosa cells to the cumulus cells and the oocyte, to induce oocyte maturation, cumulus expansion and ovulation11-13.
Translation of maternal mRNAs during oocyte maturation requires polyadenylation directed by RNA binding proteins, the prototype being the cytoplasmic polyadenylation element binding (CPEB) protein. In Xenopus oocytes, CPEB-mediated translation is under the control of cell cycle regulators which function in a cell-autonomous fashion14. Limited information is available on whether translation during the meiotic cell cycle is affected by somatic cell signals. Here we have tested the hypothesis that the environment in which oocytes complete meiosis and signals from somatic cells control translation in the oocytes. This regulation is critical for mammalian oocyte competence to develop as an embryo.
TPX2 (Targeting Protein for the Xenopus kinesin xklp2) is a protein essential for spindle assembly and chromosome interaction with microtubules15, 16. It binds and activates Aurora A by promoting its autophosphorylation17. TPX2 level of expression is critical for spindle function, and altered expression is associated with aneuploidy and cancer18-20. In agreement with a previous report21, we show that TPX2 is undetectable in oocytes in prophase and accumulates during maturation to MII (Fig. 1). It has been proposed that the absence of TPX2 accumulation in prophase is due to protein degradation through APC/Cdh121. Indeed, little change in Tpx2 mRNA translation occurs during the early phases of oocyte maturation, but the late TPX2 accumulation is associated with an increased translation22. Surprisingly, we found that TPX2 protein accumulation is not only dependent on the stage of the meiotic cell cycle. Significant differences in TPX2 protein levels were observed when comparing MII oocytes matured in vivo with those matured in vitro in association with somatic cells, or those matured in vitro after being denuded. This initial finding suggests that TPX2 accumulation is sensitive to the environment in which the oocyte matures.
To investigate whether cumulus cells, the somatic cells surrounding the oocyte, play a role in the translation of maternal mRNAs and protein synthesis, we developed an in vitro model that preserves the somatic environment in which the oocyte matures. Translational reporters were constructed and injected into oocytes still surrounded by cumulus cells (cumulus cell - enclosed oocyte, CEO) (Fig. 2A,B). This model enables monitoring translation of selected maternal mRNA in oocytes that maintain contact with cumulus cells. Furthermore, translation rates in CEOs can be compared to those measured in denuded oocytes (DOs), which are no longer exposed to somatic signals.
Reporter constructs with luciferase ORFs under the control of 3’UTRs of Tpx2 or Dazl, an RNA binding protein essential for gametogenesis23, were injected into CEOs. Translation rates of these reporters increased as the oocytes progressed from GV to MII (Fig. 2C), consistent with our report of recruitment of the corresponding endogenous transcripts to the polysomes22. However, translation in CEO is further increased by supplementing the incubation medium with amphiregulin (AREG), an EGF-like growth factor that accumulates physiologically in the follicle during ovulation22, or EGF itself (Fig. 2C). Both ligands signal through EGF receptor (EGFR) on cumulus cells24 and are not expressed by oocytes in culture. Growth factor-induced effects were not detected when meiotic reentry was prevented with the phosphodiesterase inhibitor milrinone (Suppl. Fig 1), when a TPX2 reporter with truncated 3’UTR was injected (Suppl. Fig. 2), or when oocytes were denuded prior to stimulation (Fig. 2C). These findings demonstrate that the 3’ UTR , progression through meiosis, and somatic cells are all required for the growth factor-dependent stimulation. Exposure to AREG did not increase the stability of the reporter, confirming an effect on translation rate (Suppl. Fig. 1B).
Consistent with reporter translation, endogenous TPX2 and DAZL protein levels increased as oocytes mature (Fig. 2D,E). AREG further enhanced these protein levels, confirming that increased translation of the reporter reflects translation of the endogenous mRNA and accumulation of the encoded protein. When incubated with AREG and in contact with somatic cells and AREG, a rise in oocyte protein synthesis was independently confirmed by monitoring accumulation of IL-7, an oocyte secreted chemokine, in the medium (Fig. 2F).
To conclusively demonstrate that AREG does not directly stimulate the oocyte and that the signals are indirect and mediated by somatic cells, we used a genetic mouse model in which EGFR is down-regulated only in somatic cells. Mice carrying a null Egfr allele and a floxed allele (EgfrΔ/fl), and expressing a Cre recombinase under the control of the granulosa cell-specific CYP19A1 promoter25, display a 90% decrease in EGFR expression in granulosa cells13. Although LH-dependent in vivo oocyte maturation is impaired, in vitro spontaneous maturation progresses normally13. We used CEO derived from these mice to test whether Egfr gene inactivation in somatic cells ablates the AREG effects on oocyte translation in vitro. The experiment reported in Fig. 2G demonstrates that AREG does not significantly increase translation rate of TPX2 reporter when CEO from EgfrΔ/fl: CYP19A1-Cre mice are microinjected.
To determine whether this somatic effect on oocyte translation occurs also in vivo and whether global translation is affected, genetic mouse models perturbing the EGF network were investigated. We have reported that the EGF-like growth factors AREG, EREG and BTC are induced by LH at the time of ovulation and that transactivation of EGFR is indispensable for oocyte maturation and ovulation11, 13. Whereas inactivation of EGFR causes a block in oocyte maturation in vivo and defective ovulation11, Areg−/− mice ovulate and are fertile11 but litter size is significantly decreased (Fig. 3A). When fertilization rates were assessed in vitro, CEOs or denuded oocytes derived from Areg−/− mice fertilize at a rate significantly lower than wild type (WT) littermates, suggesting a defect in developmental competence (Fig. 3B and C). Identical to the Areg−/− follicle activated in vivo, isolated CEOs from wild type mice cultured in vitro are not exposed to AREG. Therefore, the role of AREG in promoting cytoplasmic maturation was further tested by adding exogenous AREG during WT CEO in vitro maturation. This treatment improved their fertilization rate (Fig.3D), a finding consistent with other studies demonstrating positive effects of EGF network on developmental competence26-28.
Defective spindle morphology was more frequently detected in MII oocytes derived from Areg−/− mice (Fig. 3E,F). Thus, inactivation of Areg in the somatic cells yields oocytes that mature but with signs of reduced developmental competence. Our model was then used to test whether the translational program executed during oocyte maturation is affected by the absence of this somatic signal. To capture actively translating mRNAs, polysomes were isolated by sucrose density gradient from WT and Areg−/− oocytes. Maternal transcripts associated with the polysomes were isolated and analyzed by microarray hybridization (Fig. 4). Consistent with our previous report, 3208 transcripts were recruited or released from the polysomes during oocyte maturation in wild type oocytes (Suppl. Files l and 2). A similar number of transcripts (3440) moved in and out of the polysome pool also in oocytes from Areg−/− mice, confirming that the translation program is qualitatively intact. However, quantitative analysis revealed significant changes in the level of transcripts associated with polysomes in mutant oocytes. When using a cutoff of p<0.05 (Fig. 4A), polysome association of approximately 200-300 transcripts was altered in the Areg−/− oocytes compared to WT (Fig. 4B). Transcripts normally recruited to the polysome during maturation were decreased in the Areg−/− oocytes (Fig.4C). The affected transcripts included the main functional categories of metabolism, embryonic development, cell cycle and RNA regulators (Fig. 4D). Analysis of the 3’UTRs affected in the Areg-/- oocytes did not reveal the presence of a common signature, with the possible exception of two consensus elements enriched in a subset of transcripts (Suppl. Fig. 3). We confirmed the decrease of selected transcripts in the polysome pool by qPCR (Fig. 4E). Protein levels of DAZL and TPX2 in MII oocytes of the Areg−/− mice were decreased (Fig. 4F,G), which is consistent with the decreased recruitment of these mRNAs to the polysome in vivo. The absence of somatic AREG affected only a subset of transcript translation. Polysome recruitment of Cyclin B1 and Tex19.1, two transcripts known to be highly regulated during oocyte maturation, were not affected in the Areg−/− oocyte and were insensitive to AREG stimulation in the in vitro translation assay (Fig.4H,I,J,K). Thus, somatic cell signals affect translation of a subset of maternal mRNAs in vivo and in vitro.
Incubation of CEO with AREG caused a rapid increase in AKT phosphorylation indicative of activation of the PI3K pathway (Fig. 5A,B). Surprisingly, we found that the time course of CEO activation is biphasic with a secondary increase in AKT phosphorylation detected at 120-180 min of incubation with AREG (Fig. 5A,B). Because the delayed increase in AKT phosphorylation was reminiscent of the AKT phosphorylation detected in the oocyte 2-3 hrs after LH/hCG stimulation29, 30, we investigated whether this secondary increase takes place in the oocyte. CEOs were incubated in the presence of AREG; at the end of the incubation, the oocytes were denuded and used for Western Blot analysis (Fig. 5C,D) or in situ detection by immunofluorescence (Suppl. Fig. 5). A delayed AKT phosphorylation was observed in oocytes exposed to AREG while still in complex with cumulus cells; conversely, this delayed phosphorylation could not be detected in denuded oocytes exposed to AREG (Fig. 5C,D and Suppl. Fig. 5). Variable AKT phosphorylation was observed in oocytes within the first 30 min whether cultured as CEO or denuded (Fig. 5D), a signal independent of AREG and likely caused by the mechanical denudation of the oocytes. These data suggest that AREG stimulation of CEOs causes a delayed transient increase in AKT phosphorylation in the oocyte. Pharmacological inhibition of PI3K with Wortmannin or LY 294002 blocked the AKT phosphorylation in oocytes when cultured in complex with cumulus cells in the presence of AREG for 2 hrs (Fig. 5E). Under these conditions, the AREG-dependent increased translation of the TPX2 reporter was abolished, whereas the cell cycle-dependent increase in reporter translation was not affected (Fig. 5F). Similar results were obtained with rapamycin (Fig. 5G and Suppl. Fig. 6), an mTOR inhibitor31.
It should be noted that, in the experiments reported above, the pharmacological inhibitors used block the PI3K/AKT/mTOR pathway both in somatic cells and in the oocyte. To determine whether the PI3K pathway function is required in the oocyte, we used a genetic model to test whether disruption of this pathway exclusively in the oocyte is sufficient to block the AREG-dependent increase in translation. To this aim, we used cumulus/oocyte complexes from Ptenfl/fl: ZP3-CRE mice. In this genetic model, the Cre recombinase is under the control of an oocyte-specific promoter and the Pten gene is ablated only in oocytes but not in cumulus cells. Previous data with this model show that Pten inactivation causes a constitutive increase in AKT phosphorylation32. In these oocytes, translation of the Tpx2 reporter is no longer dependent on AREG (Fig. 5H). This latter finding confirms that an intact PI3K pathway in the oocyte is necessary to translate the AREG signal from the soma.
Our findings demonstrate that translation of a subset of oocyte maternal mRNAs is under the control of somatic cell inputs acting through the PI3K/AKT/mTOR pathway. This regulation functions in concert with the translational control by meiotic cell cycle regulators and is involved in establishing the oocyte competence to successfully develop as embryos.
Our in vivo and in vitro data document that exposure of somatic cell/oocyte complexes to the EGF-like growth factor AREG causes an increase translation of a subset of maternal mRNAs in the oocyte. This activation requires cell-to-cell communication, as it is lost in denuded oocytes. The use of alleles affecting Egfr and PI3K signaling in the somatic and germ cell compartments respectively confirms that AREG action requires Egfr expression in the soma and an intact PI3K signaling in the oocyte. Thus, Egfr signaling in the somatic cellular compartment is translated into PI3K activation in the contiguous cell compartment, the oocyte. Given the critical role of EGFR signaling in tumor growth33, 34, it is possible that the signaling across cells described here is also necessary for communication of transformed cell with the surrounding cellular environment.
Growth factor regulation of translation via the PI3K/AKT/mTOR pathway is established for somatic cells35, 36. However, the molecular mechanism underlying the specific regulation of a subset of transcripts in growth factor-target cells remains controversial. Through target of rapamycin complex 1 (mTORC) the PI3K pathway converges on regulation of S6 ribosomal protein and eukaryotic initiation factor 4E (eIF4E), a key rate-limiting initiation factor for cap-dependent translation35, 37. 4E-binding protein 1 (4EBP1), one of the three 4EBP binding proteins expressed in mammals, is phosphorylated by mTORC1 causing the dissociation from eIF4E and the formation of an active initiation complex38. Transcripts sensitive to mTOR regulation have signature sequences at the 5’ UTR, including a polypyrimide (5’TOP)39 or PRTE sequence40, but it is not clear whether these sequences are indispensable. The mechanism of translational regulation we describe in the oocyte is clearly distinct. 5’TOP or PRTE sequences are not readily discernible in the 5’ UTR of the prototype transcripts we have investigated in the oocyte. A selective activation is retained when different luciferase reporters are injected in oocytes, even though they have an identical 5’ UTR sequence. Conversely, the 3’ UTR of these mRNAs is essential because TPX2 3’UTR reporter recapitulates this regulation whereas the 3’UTR of, for instance, cyclin B does not support this regulation. Elements present in the 3’ UTR are necessary because a truncated TPX2 3’UTR is not sensitive to somatic cell signals. An additional feature that distinguishes the effects we describe in the oocyte from those in somatic cells is that none of the oocyte targets of regulation identified are ribosomal proteins, whereas many of the growth factor targets in somatic cells are ribosomal proteins41. On the contrary, the regulation we describe takes place on a background of widespread destabilization and decreased translation of mRNAs coding for ribosomal proteins22, 42. The pharmacological and genetic analyses performed indicate that the AKT/mTOR pathway is required for the AREG-depended translational regulation in the oocyte. We propose that mTOR signaling in oocytes controls translation via unique mechanisms that target the 3’UTR (Fig. 6). The possible role of the 5’UTR in this growth factor-dependent regulation remains to be determined.
Correct mRNA translation and accumulation of corresponding proteins in the oocyte is critical for efficient meiotic cell cycle progression, fertilization, and embryo development and necessary to attain full developmental potential. Our findings provide a molecular explanation for the widely reported observation that fully grown oocytes dissociated from the surrounding somatic cells may complete nuclear maturation but are defective in fertilization and embryo development4, 6, 26. Our findings support the idea that components required for maturation, fertilization and embryo development reach optimal levels when oocyte-somatic cell communication is preserved. In the absence of somatic cells, these components accumulate at levels sufficient for cell cycle progression but robustness and stability of the oocyte machinery is compromised, with meiosis becoming prone to errors. This conclusion is supported by our observation that TPX2, a key component in spindle assembly21, accumulates significantly less in the absence of somatic input, and spindle defects become more frequent in MII oocytes. In the same vein, decreased accumulation of Dazl, which may function upstream of TPX222, likely destabilizes the network of RNA binding protein required for maturation and embryo development. Thus, correct execution of the program of maternal mRNA translation and somatic inputs are essential for oocyte developmental competence. As an extension of this concept, biomarkers monitoring this program, such as secreted proteins, should be of prognostic value for the “fitness” of an oocyte to develop into an embryo that successfully implants and sustains pregnancy to term.
We thank Drs. Diana Laird, Davide Ruggero, Todd Nystul, and Robert Belloch at UCSF for advice during the studies and for critical reading of the manuscript, and Andrej Susor for assisting with the oocyte confocal analysis. This work was supported by Eunice Kennedy Shriver National Institute of Child Health and Human Development/NIH cooperative agreement 1U54HD055764-06, as part of the Specialized Cooperative Centers Program in Reproduction and Infertility Research and RO1-GM097165 to MC.
JC developed the COC translation assay and performed the microarray experiments and some of the AKT assays; ST performed the characterization of the Areg null phenotype; FX contributed with Western blot studies and oocyte isolation for microinjection; CJL helped with immunostaining experiments and the microinjections in cumulus oocyte complexes; HC performed the experiments on protein secretion and contributed to the writing of the manuscript. FF performed microinjections in CEO, KH contributed with the preparation of the translational luciferase reporters, CO and JS performed the bioinformatic analysis of the microarray data. MIC advised on data analysis and discussed results; MRS provided reagents and constructs and advised in the interpretation of the data; MC conceived the project, designed the experiments, analyzed the data and wrote the paper.
COMPETING FINANCIAL INTERESTS
The authors declare no competing interests